UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Gene profiling in enteroviral heart disease Yanagawa, Bobby 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-902897.pdf [ 31.71MB ]
Metadata
JSON: 831-1.0099788.json
JSON-LD: 831-1.0099788-ld.json
RDF/XML (Pretty): 831-1.0099788-rdf.xml
RDF/JSON: 831-1.0099788-rdf.json
Turtle: 831-1.0099788-turtle.txt
N-Triples: 831-1.0099788-rdf-ntriples.txt
Original Record: 831-1.0099788-source.json
Full Text
831-1.0099788-fulltext.txt
Citation
831-1.0099788.ris

Full Text

Gene Profiling in Enteroviral Heart Disease By Bobby Yanagawa B.Sc , The University of British Columbia A THESIS SUBMITTED FOR THE PARTIAL F U L F I L M E N T OF THE REQUIREMENTS FOR THE DOCTOR OF PHILOSOPHY In THE F A C U L T Y OF MEDICINE Department of Pathology and Laboratory Medicine We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A April 2004 © Bobby Yanagawa, 2004 Abstract The application of high-throughput genomical strategies and bioinformatical tools coupled with established molecular techniques allows researchers to gain new insights into the pathogenesis of infectious diseases. In humans, coxsackievirus B3 (CVB3) is the primary etiological agent of viral myocarditis, an inflammatory disease process involving the heart muscle. Specific therapy is currently unavailable. Viral myocarditis is a complex multiphasic infectious-inflammatory-reparative process. To address this temporal dimensionality, microarrays and non-array-based molecular techniques and histological and functional assays were used to further define enteroviral pathogenesis and its relation to heart failure. A re-analysed cDNA array-based dataset and an Affymetrix GeneChip®-based dataset from the murine heart during acute viremic, inflammatory and reparative stages were mined and compared. Global decreases in expression of metabolic and mitochondrial genes, increases in signalling, matrix and contractile protein genes, and distinctive patterns for genes in other functional groups are described. The transcript and protein profiles of heat shock protein 27, Muscle L I M protein and cathepsin L were confirmed using RT-PCR and immunohistochemistry, respectively. To focus on direct virus and host cell interactions, the transcriptional profile of CVB3-infected HeLa cells was investigated. Notably, increased expression of transcription factor c-fos downstream of extracellular signal-related kinase was found, a pathway our laboratory has previously shown to be important for virus replication and pathogenesis. The in vitro array experiment included a M E K l inhibitor treatment group. Commercial and in-house bioinformatical filtration, cluster analysis and visualization tools were used to determine potential gene targets downstream in the E R K pathway in the setting of CVB3 infection. Marked upregulation of oncogenes, serpins and matrix metalloproteinase genes, among others, were found in infected HeLa cells and in the heart. Together, our animal and cell culture array studies have contributed insight into host responses to enteroviral infections from which new testable hypotheses have been generated. Table of Contents Abstract ii Table of Contents i i i List of Tables v List of Figures vi List of Abbreviations x Dedication xiii Acknowledgements xiv CHAPTER I Introduction to Viral Myocarditis 1.1 Overview of Viral Myocarditis 1 1.2 Coxsackievirus B3 17 1.3 Cell Signalling 21 1.4 Cell Death , 24 1.5 Immune Response 28 1.6 Experimental Treatments 31 1.7 Conclusion 37 CHAPTER II Introduction to Gene Profiling and Bioinformatics 2.1 Introduction 38 2.2 Microarray Platforms 40 2.3 Data Mining Algorithms 45 CHAPTER III Gene Profiling In Vivo 3.1 Characterization of the Mouse Model of Viral Myocarditis 54 3.2 cDNA-Based Bioinformatical Analysis of Mouse Heart Genes 68 3.3 GeneChip® Array-Based Approach to Gene Profiling in Mouse Hearts 80 3.4 A Comparative Gene Profiling Approach in CVB3-Infected Hearts 89 3.5 Discussion 114 iii CHAPTER IV Gene Profiling In Vitro 4.1 GeneChip® Array-Based Profiling of CVB3-Infected HeLa cells 134 4.2 Triplicate Gene Profiling of CVB3-Infected HeLa cells with M E K l Inhibition 147 4.3 Discussion 169 CHAPTER V Conclusion and Future Work 5.1 Conclusion 182 5.2 Future Work 186 Glossary 192 Materials and Methods 194 Bibliography 207 iv List of Tables Table 1: Most differentially regulated genes in CVB3-infected mouse hearts using cDNA microarrays 75 Table 2: Small functional gene group profiles in CVB3-infected hearts 78 Table 3: Up-regulated genes at 3 and 9 day hearts post infection 87 Table 4: Down-regulated genes at 3 and 9 day hearts post infection 88 Table 5: Genes of interest in CVB3-infected HeLa cells averaged over all timepoints 139 Table 6: Comparative assessment of surrogate models systems for human enteroviral heart disease 170 Table 7: Primer sequences for RT-PCR 205 V List of Figures Figure 1: Histological example of human myocarditis 3 Figure 2: Clinical diagnosis of myocarditis 5 Figure 3: Viral myocarditis: A triphasic disease 11 Figure 4: Biological processes during coxsackievirus B3 infection 14 Figure 5: Myocardial heterogeneity in myocarditis 16 Figure 6: The coxsackievirus B3 lifecycle .19 Figure 7: Viral protease cleavage events 20 Figure 8: Coxsackievirus B3-triggered signalling pathways 22 Figure 9: Molecular pathways of apoptosis 26 Figure 10: Inhibition of Rhabdomyosacroma (RD) cell lysis by DAF-Fc and CAR-Fc 32 Figure 11: DAF-Fc attenuates myocarditic pathogenesis and virus replication 33 Figure 12: CAR-Fc protects again myocarditis 34 Figure 13: CAR-Fc reduces virus replication and infectious virion in CVB3-infected mouse hearts 35 Figure 14: CAR-Fc protects again pancreatitis 36 Figure 15: An example of Affymetrix GeneChip® image acquisition to image quantitation 42 Figure 16: Kaplan-Meier survival curve for CVB3-infected mice 56 Figure 17: Histological characterization of viremic phase CVB3 infection of mouse hearts 57 Figure 18: Histological characterization of inflammatory phase CVB3 infection of mouse hearts 58 Figure 19: Histological characterization of reclamative phase CVB3 infection of mouse hearts 59 Figure 20: Semi-quantitative histological grades of murine myocarditis 62 Figure 21: Plaque assay for infectious virus in myocarditic tissue 63 Figure 22: Two-dimensional echocardiography of sham-injected mouse hearts 64 Figure 23: Two-dimensional echocardiography of CVB3-infected mouse hearts.... 65 vi Figure 24: Posterior wall dimension measurements using 2-dimensional echocardiography of normal and myocarditic hearts 66 Figure 25: Ejection fraction using 2-dimensional echocardiography of normal and myocarditic hearts 67 Figure 26: Experimental design for microarray studies 69 Figure 27: Gene profde overview and functional classification in CVB3-infected mouse hearts 74 Figure 28: Hierarchical clustering of differential gene expression in CVB3-infected hearts 77 Figure 29: GeneChip® expression profiles in CVB3-infected murine hearts 84 Figure 30: Temporal display of gene expression in CVB3-infected mouse hearts... 86 Figure 31: Natiuretic peptide gene profile in CVB3-infected hearts 91 Figure 32: Chemokine and interferon gene profile in CVB3-infected hearts 92 Figure 33: Ubiquitin/proteasome gene profile in CVB3-infected hearts 95 Figure 34: Collagen gene profile in CVB3-infected hearts 96 Figure 35: Cell death, inflammation and early fibrosis during the inflammatory stage of myocarditis 97 Figure 36: Fibrosis and calcification during the reclamative stage of myocarditis... 98 Figure 37: Cyclin gene profile in CVB3-infected hearts 99 Figure 38: C A R gene expression in CVB3-infected hearts 100 Figure 39: SI 00 gene profile in CVB3-infected hearts 102 Figure 40: Complement gene profile in CVB3-infected hearts 103 Figure 41: Muscle L I M protein gene profile in CVB3-infected hearts 104 Figure 42: Cathepsin L gene profile in CVB3-infected hearts 106 Figure 43: Cathepsin L protein expression in CVB3-infected hearts 107 Figure 44: Peripheral-type benzodiazepine receptor gene profile in CVB3-infected hearts 108 Figure 45: Peripheral-type benzodiazepine (PBR) receptor protein expression in CVB3-infected hearts 109 Figure 46: Heat shock protein genes in CVB3-infected hearts 111 Figure 47: Heat shock protein 27 (HSP27) protein expression vii in CVB3-infected hearts 112 Figure 48: Myofibrillar gene profile in CVB3-infected hearts 113 Figure 49: The mitochondrial injury hypothesis 119 Figure 50: The cathepsin L anti-viral response hypothesis 127 Figure 51: Overview of gene expression profiles in CVB3-infected HeLa cells 138 Figure 52: Temporal display of gene expression in CVB3-infected HeLa cells 140 Figure 53: Clustering gene expression data from CVB3-infected cells 141 Figure 54: Function-based classification map 144 Figure 55: Biological system transcriptional up regulation in CVB3-infection in vitro 145 Figure 56: C-fos up regulation in CVB3-infected HeLa cells and mouse hearts 146 Figure 57: Temporal display of genes in CVB3-infected cells +/-U0126 150 Figure 58: Down regulation of intracellular genes in CVB3-infected HeLa cells 151 Figure 59: Differential expression of virus-related genes in CVB3-infected HeLa cells 152 Figure 60: Signal transduction gene up regulation in CVB3-infected HeLa cells 155 Figure 61: Functional distribution of up-regulated genes at 9 hours in CVB-infected HeLa cells 156 Figure 62: Functional distribution of E R K downstream genes at 9 hours in CVB-infected HeLa cells 158 Figure 63: Map of known M A P K - E R K downstream genes at 9 hours in CVB-infected HeLa cells 160 Figure 64: Serpin gene profile in CVB3-infected HeLa cells 163 Figure 65: Serpin gene profile in CVB3-infected mouse hearts 164 Figure 66: Metalloproteinase gene profile in CVB3-infected HeLa cells 165 Figure 67: CYP gene profile in CVB3-infected HeLa cells 167 viii Figure 68: The extracellular matrix alteration hypothesis 181 Figure 69: The experimental strategy from gene profiling to functional understanding 188 ix List of Abbreviations The following is a list of abbreviations in alphabetical order: AP-1 Activating protein-1 (transcription factors) ANP Atrial natriuretic peptide Apaf-1 Apoptosis activating factor-1 ATF3 Activating transcription factor 3 BNP Brain (b-type) natriuretic peptide C A R Coxsackievirus and Adenovirus Receptor Cdkn Cyclin-dependent kinase inhibitor Chemokine Chemoattractant cytokine CHF Congestive heart failure CNP Cardiac (c-type) natriuretic peptide CTGF Connective tissue growth factor Cts Cathepsin CVB3 Coxsackievirus B3 Cy3/5 Cyanine 3/5 C Y P Cytochrome P450 Cytc Cytochrome c D A F Decay Accelerating Factor (CD55) D C M Dilated cardiomyopathy D M E M Dulbecco's modified Eagle's media DMSO Dimethyl sulfoxide D N A Deoxyribonucleic acid E. Coli Escherichia coli E C M Extracellular matrix EIF4y Eukaryotic initiation factor-4y E M B x Endomyocardial biopsy E R K Extracellular signal-related kinase GeneChip® A.k.a. oligonucleotide array/Affymetrix array GRO Growth-related oncogene HeLa Henrietta Lacks; human cervical carcinoma cell line HIV Human immunodeficiency virus HL-1 Atrial cardiac muscle cell line HSP Heat shock protein Ig Immunoglobulin IGTP Interferon-y-inducible guanosine triphosphatase IFN Interferon INOS Inducible NO synthase IRES Internal ribosomal entry sequence ISH In situ hybridization Jak Janus kinase L V Left ventricle/ventricular M A C Membrane attack complex M A G E - M L Microarray Gene Expression-Markup Language M A P K Mitogen activated protein kinase M E K l M A P K kinase-1 M H C major histocompatibility complex M I A M E Minimal information about a microarray experiment MIP Macrophage inflammatory protein M L P Muscle L I M Protein M M Mismatch M M P Matrix metalloproteinase MOI Multiplicity of infection Nip21 Nineteen-kilodalton interacting protein 21 N F K B Nuclear factor KB N K Natural killer NO Nitric oxide PBR Peripheral-type benzodiazepine receptor PBS Phosphate-buffered saline Pfu Plaque-forming units Pi Post-infection xi P13K Phosphatidyl inosito 1-3-kinase P M Perfect match R N A Ribonucleic acid RT-PCR Reverse transcriptase-polymerization chain reaction SAGE Serial analysis of gene expression SCID Severe combined immunodeficiency Serpin Serine protease SOM Self-organizing map SR Sarcoplasmic reticulum Stat Signal transducers and activators of transcription TGF-p Transforming growth factor- P U0126 A small molecule M E K l inhibitor U P G M A Unweighted pair-group method using arithmetic averages U R L Unique reference locator UTR Untranslated region xii Dedication I would like to dedicate this thesis to my mother Junko, father Keiju and sister Kana Yanagawa for their love and support. I would also like dedicate this work to my mentors Drs. Bruce M . McManus and Decheng Yang. X l l l Acknowledgements I would like to thank the numerous people for their great support, their advice and technical expertise and most importantly their friendship, without which this work could not have been performed. It has been a great privilege to work under the supervision of Drs. Bruce McManus and Decheng Yang and I thank them both. I would like to thank all members of the Cardiovascular Research Laboratory, The James Hogg iCAPTURE Centre for Cardiovascular and Pulmonary Research, St. Paul's Hospital, especially Dr. Honglin Luo, John Zhang, Caroline Cheung, Jane Yuan, Mary Zhang, Mitra Esfandiarei, Nana Rezai, Paul Cheung and Agripina Suarez. I would also like to thank all the following investigators for their collaborative spirit: Dr. Tracy Deisher, Stephanie Bonigut and Dr. Tom Daniel from Amgen Corporation, Dr. Raymond Ng and Zsuzsanna Hollander from the Department of Computer Science, University of British Columbia; Dr. Rong Zhu, Department of Statistics, University of British Columbia; Dr. Timothy Triche, Dr. Jonathan Buckley and Ms. Betty Schuab from the Children's Hospital Los Angeles; Dr. George Schreiner at SCIOS, Inc.; and Dr. O. Brad Spiller from the University College of Medicine, Wales, UK. Generous funding has been provided by the Canadian Institutes of Health Research, the Heart and Stroke Foundation of Canada, the Michael Smith Foundation for Health Research, David Hardwick and the Department of Pathology and Laboratory Medicine. Our research is also supported by the Providence Health Care, the St. Paul's Hospital Foundation and the University of British Columbia. XIV Chapter I Introduction to Viral Myocarditis The major focus of this thesis is the high-throughput genomical study of enteroviral heart disease. The overall aim is to increase our understanding of the role of infection and immunity in viral myocarditis. In order to provide a clinical context for the research, it is necessary to understand the current understanding of human myocarditis including diagnosis, common causes, outcomes, treatments and the triphasic nature of disease progression. Coxsackieviral pathogenesis is integral to the subsequent inflammation, so an introduction is provided on the current knowledge of the viral genome and lifecycle, as well as host cell signalling, cell death and immune response to infection. Together, these events influence host gene regulation following coxsackievirus B3 (CVB3) infection. Finally, as this work may present new opportunities for therapeutic intervention, I present current work on experimental treatments for viral myocarditis. 1.1 Overview of Myocarditis Myocarditis, both infectious and non-infectious, is a major cause of sudden unexpected death in patients less than 40 years old, and contributes significantly to the societal burden of heart failure [1]. It is a non-ischemic inflammatory disease of the myocardium [2] . The diagnostic hallmarks of myocarditis are inflammatory infiltration and myocyte death in endomyocardial biopsy (EMBx) specimens [3]. The causes of inflammation may or may not be known. Figure 1 provides a histological example of human myocarditis from an autopsy specimen. Diagnosis The clinical presentation of myocarditis is often unpredictable and may be associated with diverse, nonspecific and variable signs and symptoms including shortness of breath, chest pain, life-threatening arrhythmias and cardiogenic shock [4]. Particularly if viral in origin, patients may also experience severe flu-like symptoms. Currently, right ventricular endomyocardial biopsy (EMBx) remains the definitive diagnostic procedure for myocarditis. Mason et al [5] has contributed much to our l understanding of the E M B x as a diagnostic tool for myocarditis. Pathological review should include a thorough assessment of myocyte cell death, inflammation and fibrosis following guidelines set out in the Dallas Criteria [3]. Acute myocyte injury associated with myocarditis is characterized by sarcoplasmic vacuolization as well as coagulative cell death and contraction band disruption. Chronically, myocarditis may be associated with myocyte hypertrophy and heterogeneity, and interstitial fibrosis and focal mononuclear inflammation may continue. Initial biopsy is necessary first to establish the diagnosis of myocarditis [6] and then with follow-up biopsies to determine the status as ongoing, resolving or resolved. The clinical features and diagnostic strategies for myocarditis cases are summarized in Figure 2. A major drawback to E M B x is false negative results, thus several (5-10) biopsy specimens are often recommended [7]. Other clinical diagnostic approaches may include peripheral blood leukocyte count, chest radiography, electrocardiogram, echocardiography, viral serology and fecal isolation of virus or genome [4]. The use of biomarkers for cardiac injury has been investigated previously but has thus far rarely assisted in diagnosis except for patients who present acutely [8, 9]. In the Myocarditis Treatment Trial, mean serum creatine kinase C K - M B levels were not found to be greater in patients with myocarditis as compared to normal controls [5]. Troponin I (cTnl) was increased in 34% of myocarditic patients and the greater increments of change, prolonged expression [10] and specificity to the heart [11] makes this biological substance a more suitable blood marker for cardiac injury. One aim of the work presented here is the identification of genes whose proteins may be exploited as potential markers, and thus prognostic indicators for inflammatory cardiac injury. 2 3 Figure 1: Histological example of human myocarditis. A 54-year-old woman endured 5 days of "flu-like" symptoms, was hospitalized for shortness of breath and inability to sleep, and was diagnosed with left ventricular failure, cardiogenic shock and marked hypoxemia. She developed uncontrollable cardiac arrhythmias, hypotension and ventricular tachycardia with death occurring on the same day. A. Histological sections stained with hemotoxylin and eosin (H&E) are shown. There is prominent myocardial cellular infiltrates (arrows) as compared to (B) control myocardial sections. C. Immunohistochemical staining for CD3 (T-cell marker) and CD68 (marker for macrophages, monocytes, neutrophils, basophils and lymphocytes) revealed focal areas of positivity. 4 A. Signs and Symptoms Flu, shortness of breath with or without chest pain, heart failure, arrhythmias, hypotension, other syndromes B. Primary Clinical Diagnostic Tools Leukocyte count, chest radiography, EKG, echocardiography, CPK, troponin, serology, virus culture from in fecal samples, coronary angiography C. EMBx: 1 S T Biopsy • Myocarditis with Borderline or without fibrosis myocarditis No myocarditis 2 n d B i o p s ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ ^ Ongoing Resolving myocarditis myocarditis Resolved myocarditis E. Molecular Techniques ISH/RT-PCR/IHC 5 Figure 2: Clinical diagnosis of myocarditis. Myocarditis may present with diverse, nonspecific and variable manifestations. A. A list of common signs and symptoms are listed. B. If a diagnosis of myocarditis is suspected, there are a range primary clinical diagnostic tools to detect virus presence and myocardial injury at the disposal of the clinician. C. Of these, the endomyocardial biopsy (EMBx), (E) at times combined with in situ hybridization (ISH), reverse polymerase-polymerase chain reaction (RT-PCR) and immunohistochemistry (IHC), is the gold standard. The first E M B x is used to distinguish presence and degree of myocarditis. D. Subsequent biopsies are necessary to establish disease progression based on comparisons with the previous biopsy(ies). 6 Causes of myocarditis Myocarditis may be caused by infection by viruses, bacteria, rickettsia and fungi, as well as toxic and allergic effects, hypersensitivity responses, and endocrine and immune disorders [1]. The most common viruses associated with myocarditis include picornaviruses, influenza viruses, human immunodeficiency virus (HIV), herpesviruses such as Epstein-Barr virus and cytomegalovirus, and adenoviruses [12]. Among the prevalent enteroviruses that cause myocarditis are the coxsackieviruses B (CVB) 1-6, and serotype 3 (CVB3) is specifically notable. Coxsackieviruses were first isolated during a paralytic outbreak in 1948 [13] and C V B was isolated in two patients with aseptic meningitis the following year [14]. From the mid-1950s CVB3 association with human myocarditis has been widely demonstrated [15] and recently reiterated [1, 16]. Martino and colleagues [17] performed an extensive study of the literature to find that 38-50% of myocarditis cases are CVB3 positive as determined by serological assessment of IgM to enterovirus or neutralizing antibodies. Ten-year World Health Organization global surveillance data confirmed that C V B has the highest cardiovascular diagnoses (20-30 per 1000 cases) of all viral infections [18]. In young populations, coxsackievirus infection is particularly pernicious. During a nine-year period, the Nassau County Medical Center (NCMC) tested 602 culture specimens that were positive for group B coxsackievirus among 153,250 live births. Of the 77 infants less than three months of age, aseptic meningitis was the most common clinical syndrome with six deaths resulting from overwhelming coxsackievirus infection [19]. However, as in adult cases, the true incidence of viral infections in heart disease is unknown due to a combination of difficulties in establishing definitive clinical diagnosis, demonstrating virus particles in the heart, establishing a clear association between infection and injury to the heart, false negative results associated with biopsies and the lack of autopsies performed after sudden death [20]. 7 Outcome Suspicions of an infectious-immune cause for dilated cardiomyopathy (DCM) [21] have recently been supported by both animal models and clinical studies [17]. Although the association between viral myocarditis and D C M is now accepted, the extent of enteroviral involvement in human D C M cases is equivocal. Satoh et al [22], using nested reverse transcriptase polymerase chain reaction (nRT-PCR), detected enteroviral R N A in 17 of 35 patients (49%) with D C M . Giacca et al [23] also used nRT-PCR for enterovirus genome on EMBx samples from 53 patients with dilated cardiomyopathy and found only four positive samples. A recent meta-analysis of 17 studies showed that 23% of patients with D C M were enterovirus positive versus 7% of control with an odds ratio of 3.8 (as defined as odds for positivity in D C M / odds in control tissue) [24]. As enterovirus-positive patients have been shown to experience rapid progression to heart failure (<12 months), the detection of viral genome has significant prognostic implications. Thus, the progression from myocarditis to D C M is well supported, but the mechanism(s) are still unclear. Experimental evidence from cell culture and animal models has shown that viral 2A protease cleavage of dystrophin may be one possible mechanistic explanation for a viral role in D C M . Dystrophin connects the cytoskeletal actin-binding site to the p-dystroglycan extracellular matrix anchor, thus its cleavage completely disrupts the cytoskeletal architecture [25, 26]. Dystrophin-glycoprotein complex disruption is the primary cause of Duchenne muscular dystrophy [27, 28] and has been identified in patients with familial D C M and viral myocarditis [29]. Increased myocardial stiffness is another possible mechanism for chamber dilation as cardiac fibrosis and remodelling occur up to one year following CVB3 infection in mice [30]. In this thesis, I utilized 2-dimensional echocardiography to show chronic decreases in ejection fraction in CVB3-infected mouse hearts and provide potential molecular explanations for long-term functional deterioration. Treatments Despite much research and prospective patient trials with myocarditis treatments, current management of active myocarditis is still based on supportive therapy for systolic dysfunction [31]. The Myocarditis Treatment Trial Investigators determined that immune 8 suppression with a combination of prednisone and cyclosporine or azathioprine did not improve cardiac function in patients with myocarditis and their data suggested that acute inflammation may even reduce overall disease severity [5]. Recent immunomodulatory approaches which target the interferon (IFN) pathways have led to some success in blocking immune enhancement and microbial infection. Interferons are pleiotropic cytokines that exhibit immunomodulatory effects [32]. Interferon pathways are divided into IFN-a, -P and -co, which binds the type I interferon receptor complex and IFN-y which binds the type II interferon receptor complex. Binding of both subtypes of IFNs activate downstream Janus kinases [33]. Patients with myocardial enteroviral or adenoviral persistence and left ventricular (LV) dysfunction who were treated with IFN-p showed elimination of viral genome and improved L V function [32]. Two cases of enterovirus-linked myocarditis have been reported in which IFN-a therapy has led to clinical and hemodynamic improvement and virus clearance at secondary biopsy [34]. To date, a randomized placebo-controlled clinical trial has not been performed with IFN-a therapy for viral myocarditis. A Triphasic Disease Myocarditic injury related to viruses was, until recently, experimentally studied as predominantly an inflammatory disease [35-37]. Myocarditis has since been re-conceptualized in the following phases: (1) early myocyte death associated with direct viral infection, followed by (2) inflammation with further damage, and later by (3) reparative and ultimate remodeling (Figure 3). In fact, recognition of the calculus of this disease is not new, as the temporal complexity of human myocarditis was captured in an early report by Woodruff: "Early in the illness scattered hypereosinophilic myofibers, widespread edema and only a few inflammatory cells are present. Later, myofibers exhibit loss of striations, clumping of cytoplasm, fragmentation and eventually dissolution or dropout. The degenerating or partially necrotic myofibers are usually surrounded by mononuclear cells, such as 9 lymphocytes, plasma cells, and macrophages. These mononuclear cells are commonly seen invading muscle fibers, some of which were broken down completely" (in [1], Section Pathology, Adolescent and Adult Lesions, p 432; italicizes do not exist in the original text and were included by this author) 10 Myocyte Dropout and Structural Breakdown 11 Figure 3: Viral myocarditis: A triphasic disease. Viral myocarditis can be conceptualized in three continuous but distinct phases [135]. A. The first stage is characterized by direct virus infection of cardiac myocytes depicted here as a duel between the pink cardiac myocyte (eosinophilic when stained with hematoxylin and eosin) and the blue/green viral capsid derived from the electron microscopic image and reflects the icosahedral shape of this non-enveloped 28nm virion. The results are early myocyte death, associated viremia and triggering of a humoral immune response. Injury and the cytokine milieu propagates (B) further inflammation and damage as illustrated here by the small blue cellular inflammatory cells which reflect the approximate size and deep blue nuclear hematoxylin stain when stained with hematoxylin and eosin. C. Chronically, there can be any number of the following: disruption of the ordered myocyte architecture, heterogeneity in myocyte size, myocyte multinucleation and fibrosis (as illustrated by the yellow tissue deposits). Depending on the extent of myocyte loss the process leads to various degrees of reclamation of the tissue structure, and ultimate restoration of function or cardiac dilatation. 12 Our laboratory has been interested in direct virus-induced damage as an important precursor to inflammatory infiltration and a major determinant of long-term consequences. The importance of virus-induced myocytolysis is supported by the observation that widespread coagulative and contraction band cell death are often unattended by an evident local inflammatory response, even late in disease [38]. In fact, prominent cytopathic alterations often co-localize to viral replication as identified by in situ hybridization of both positive and negative strand viral RNA [39]. Chow et al [38, 40] showed extensive CVB3-induced myocarditis and high subsequent mortality in severe combined immunodeficient (SCID) mice (lacking functional B and T lymphocytes), as well as in double mutant SCID/beige mice (lacking functional B and T lymphocytes and natural killer cells). We and others have since shown that enterovirus infection injures target cells by disruption of normal protein degradation [41], translation [42], cytokine milieu, cell cycle [43], micro vasculature and extracellular collagen matrix [44], and by protease cleavage of host proteins [25](Figure 4). In response to microbial infection and injury to the heart, the innate and humoral immune responses are initiated. The first wave of inflammatory processes is initiated by an influx of macrophages, natural killer (NK) cells and a humoral neutralizing antibody response. The next wave includes the antigen-specific T-lymphocytes and antibodies [45, 46]. There is still debate, as to whether the specific arms of the immune system are always beneficial or at times, destructive in the development of myocarditis [47]. The end result of direct viral replication and protracted inflammatory infiltration is a multifaceted myocyte phenotype (Figure 5). Focal myocardial injury accompanied by reparative fibrosis leads to various degrees of myocyte heterogeneity, as evidenced histologically. The ultimate outcome such events is either restoration of heart function or dilation and congestive heart failure. 13 (1) Microvascular Injury (8) Matrix Modulation Extracellular Matrix (4) Protein Degradation (6) Protease Cleavage of Host Cell Proteins (7) Cytokine Release (5) Cell Cycle Alterations I 4 Figure 4: Biological processes during coxsackievirus B3 infection. Experiments in cell culture and in animal models have shown evidence of myocardial injury by several mechanisms. The cell is diagramatically represented here as a pink myocyte, the virus is represented as a blue/green icosahedron and the intracellular nucleus, mitochondria and lysosome are labeled. 1. Observed microvascular spasm following CVB3 infection may be the result of virus leaving the microcirculation and entering the heart tissue. Following target cell infection, (2) apoptotic [114] and (3) pro-survival signalling [73] take place. 4. Coxsackieviruses use the ubiquitin/proteasome protein degradation pathway to degrade host proteins such as cyclin D [357], which is partially responsible for (5) cell cycle disruption [45]. 6. Viral proteases cleave host proteins including dystrophin of the cytoskeletal apparatus [29]. 7. Cellular injury triggers a pro-inflammatory cytokine response [45] which inturn causes cleavage of extracellular matrix by activation of matrix metalloproteinases [342]. The combination of these and other virus-activated events determines the scope of direct viral injury and subsequent immune infdtration. 15 Homogeneity c "3 o E o Of Early stress response Myocyte dropout-necrosis/apoptosis Changes in endocrine function Cytoskeletal dysorganization Migration-myocyte slippage Fetal gene expression Changes in connective tissue matrix Hypertrophy and Atrophy Pro-inflammatory reaction Change in metabolism/energetics Hyperplasia-atrial>ventricular (in the very young) Heterogeneity Figure 5: Myocardial heterogeneity in myocarditis. Following viral injury in the normal heart as depicted histologically in the top image, a combination of compensatory and deleterious remodeling processes listed at the right, together, lead to myocardial heterogeneity, as represented histologically in the bottom image. Such heterogeneity impinges on heart function and ultimately contributes to the observed dilated cardiomyopathy phenotype. 16 1.2 Coxsackievirus B3 Coxsackieviruses are the most common causative agent of myocarditis, viral or otherwise [18]. The in vivo experimental model for myocarditis used in this thesis is the coxsackieviral infection of myocarditis susceptible mice. Therefore, this section contains a description of the current knowledge of the CVB3 genetics, genomics and lifecycle. Coxsackievirus is an enterovirus of the family picornaviridae. It is a positive strand R N A virus. Kandolf et al [48] first cloned and sequenced the viral 7.4 kB genome which contains a single open reading frame flanked by a 3' and 5' untranslated region. Instead of a cap structure 7-methyl guanosine triphosphate group, the genome contains a small viral protein (Vpg) attached to the 5' untranslated region (UTR) end and a polyadenylated tail at the 3' end. The lengthy 5' UTR of coxsackievirus B3 also forms a highly ordered secondary structure and contains the internal ribosomal entry sequence (IRES) which allows for viral cap-independent R N A transcription. Mutational studies by Yang et al [49] within the 5' UTR demonstrate the importance of the stem loops G and H within the IRES, the secondary and tertiary R N A structures and also putative protein-RNA interactions for viral replication. The coxsackievirus proteins are translated as a single polyprotein which is subsequently post-translationally cleaved into functional viral proteases and structural proteins. An early and important biological stage of the virus infectious cycle is host cell receptor attachment. Coxsackievirus B3 can recognize both the coxsackievirus and adenovirus receptor (CAR) [50] and decay-accelerating factor (DAF) co-receptor [51] proteins to facilitate infection. C A R (46 kDa) is a member of the immunoglobulin (Ig) superfamily [52] and has been shown to interact with Zonula-Occludens-1, an epithelial tight junctional complex protein [53]. D A F (CD55) has four short consensus repeat (SCR) sequences, a serine/threonine rich region and a glycosylphosphatidylinositol anchor to the outer leaflet of the cell membrane [54]. Attachment to the SCR 2-3 region of the D A F receptor likely facilitates initial virus attachment and increases CAR-binding efficiency to the canyon of the virus protein coat [55, 56]. As a membrane-bound regulator of complement activation, 17 D A F affords resistance to complement-mediated host cell damage by dissociation of C3 and C5 convertases that blocks both classical and alternative complement pathways [54, 57]. However, influence of CVB3-receptor interactions on immune or complement responses is yet unknown [58]. Coxsackieviruses have a rapid replicative cycle as illustrated in Figure 6. Following virus entry and uncoating, genomic R N A acts as a template for both translation and transcription of minus-strand RNA. The incoming positive strand R N A directs synthesis of the polyprotein via a cap-independent IRES mechanisms using confiscated host ribosomes and other translational machinery [49]. This polyprotein is subsequently processed into individual structural (Vpl-Vp4) and non-structural proteins following cleavage by viral protease 2A, 3C and conceivably 3CD. Viral enzymes for RNA transcription then synthesize minus-strand RNA which acts as an essential template for transcription of complementary viral progeny plus-strand R N A genomes. Structural proteins are formed by partial cleavage of the PI precursor which are packaged with positive strand RNA molecules to form progeny virions. Viral proteases are also known to cleave host proteins causing host cellular injury (Figure 7). Klingel et al [39] used in situ hybridization to show viral persistence and ongoing inflammation in mouse hearts up to 30 days post-CVB3 infection. Although coxsackieviruses do not have a well-characterized latent phase of infection, recent evidence suggests that CVB3 R N A may persist in target cells in a double-stranded intermediate confirmation [59]. Such persistent low-level CVB3 is cleared slower during the post-infectious inflammatory phase of myocarditis than during fulminant infection [60]. 18 (1) Virus attachment (2) Virus uncoating CAR' (6) Releasing progeny virus „ . . . . . Release genomic RNA (+ve sense)! Capsid proteins and • non-structural proteases A (3) Translat ion by host apparatus! A \ * \ / \ —IffSS Infectious vir ions (5) Vir ion assembly f T l - A A A Non-infectious H) R N A replication by viral 3D polymerase p r o v i r i o n S i and host machinery 0 II JIIIIIfflinHIIlirAAA +ve sense progeny ••r""> -ve sense template UUU 5 UUU Smooth ER Figure 6: The coxsackievirus B3 lifecycle. (1) CVB3 binds to the coxsackievirus and adenovirus receptor (CAR) and decay-accelerating factor (DAF) host proteins for (2) uncoating and cell entry. Upon entry, the genomic R N A acts as a template for both (3) polyprotein translation and (4) negative-strand RNA transcription. (5) The polyprotein is processed into individual structural (Vpl-Vp4) and non-structural proteins, which together with progeny RNA, forms the non-infectious provirion. (6) The provirion undergoes a critical cleavage step and is released from the cell by lysis. 19 CVB3 CAR receptor DAR Viral R N A Translation Progeny RNA Capsid Proteins Progeny virions VPgH VPO| VP3 | VP1 J 2A| 2BJ 2C | 3A 13B Protease cleavages [2A AAA. (^fipases) r^^trophjp)(] [ P A B P ) \ Q [ M A P ^ ) ( J [polyjT T Structural Breakdown ( J J C R E B ; (flfPoiy" 0 [ C B T I Decreased Transcription Figure 7: Viral protease cleavage events. Viral proteases 2A, 3C and 3CD are responsible for cleavage of the polyprotein but can also cleave host cell proteins. Viral protease 2A can cleave dystrophin, poly-A binding protein (PABP), eukaryotic initiation factor 4 gamma (eIF4G) and potentially caspases. Viral protease 3C can cleave cyclic-AMP response element binding protein (CREB/ATF), the transcription factors octomer-binding transcription factor (Oct-1) and structural protein microtubule-associated protein-4 (MAP-4), among other proteins. Protease 2B increases plasma membrane permeability. 20 1.3 Cell Signalling Although little is yet known concerning the host genomic response to enterovirus infection, a growing body of knowledge characterizing host signalling pathways invoked post infection exists. Such knowledge has comes from experiments in transformed and primary cell cultures as well as animal models which include genetically-modified mice. The major consequence of cell signalling is the activation of transcription factors which subsequently activate or suppress gene transcription in host cells. Thus, an understanding of the signalling pathways induced by CVB3 is important to better understand the context of host cell transcription. The terminally-differentiated state of cardiac myocytes begs the need for cellular mechanisms to protect itself against ischemic, anoxic, pressure overload, volume overload and viral insults. Protective responses include fetal gene re-expression, immediate early stress responses, cytokine induction, cytoskeletal remodeling and repair [61]. Substantial published evidence suggests that host cells, including cardiomyocytes, actively utilize anti-viral protective mechanisms in response to CVB3 infections (Figure 8) [62]. Our laboratory and others have shown that CVB3 can takeover 'cardioprotective' signalling pathways during the course of its lifecycle and that selective modulation of such signalling pathways can profoundly effect virus replication and host cell/organ viability. Mitogen-activated protein kinases (MAPKs) are intracellular signal transduction serine/threonine protein kinases activated in response to a wide variety of extracellular stimuli, including viral infection [63]. Our laboratory has shown that CVB3 infection induces a biphasic activation of the M A P K extracellular signal-regulated kinase (ERK) [64]. Selective inhibition of M A P K kinase-1 ( M E K l ; upstream of ERK) blocks CVB3 progeny release, decreases virus protein production and circumvents apoptosis [65]. Opavsky et al [66] further showed E R K activation in isolated cardiac myocytes and in hearts of myocarditis-susceptible A/J mice. Taken together, these data suggest that CVB3 triggers E R K signalling in cultured cells, freshly isolated cardiomyocytes, and in the heart 2 1 Figure 8: Coxsackievirus B3-triggered signalling pathways. Following infection of susceptible cells, our laboratory has shown activation of the extracellular signal-regulated kinase (ERK) pathway, phosphatidylinositol 3-kinase (PI3K), apoptotic signaling and cell cycle signaling. Others have shown tyrosine kinase, Jak/Stat and NFkB pathways, among others, to be important in viral infections. Differential gene transcription is an important end result of the above pathways. 22 to augment viral replication. The mechanism of pro-viral and pro-apoptotic functions of E R K are currently unknown. Thus, identification of differential transcription of target genes in the E R K pathway, to provide insights into the mechanism of E R K activity, is one major focus of this thesis work. The sarcoma family kinase Lck (p561ck) is required for CVB3 replication in T-cell lines and for viral replication and persistence in vivo [67]. Since Lck has been shown to directly activate E R K in T-cell lines [68], the potential intersection of these pathways in viral pathogenesis is of great interest. In this regard, D A F has been localized to lipid raft caveolae, cholesterol-rich membrane invaginations known to aggregate raft-associated signalling proteins [69]. Recently, C A R has also been shown to be necessary for M A P K signalling following infection [70]. Here, I show early differential host cell transcription 30 minutes post CVB3-infection, suggesting a viral receptor-triggered mechanism. Other intracellular signalling pathways are also involved in enterovirus infection. Recently, inhibition of janus kinase (Jak) signalling has been shown to block cardiac myocyte antiviral defense, increase viral replication and induce cardiomyopathy and mortality in CVB3-infected mice [71]. Nuclear factor K B ( N F K B ) is a ubiquitous transcriptional regulator for functions ranging from immune reactions to growth control [72]. N F K B knockout mice survive lethal encephalomyocarditis virus challenge likely due to premature death in target cells [73]. Transfection of an N F K B cis element decoy oligonucleotide, which competes for the cis NFkB binding site in the promoter region of several cytokine genes, reduced expression of pro-inflammatory cytokines ICAM-1, IL-2, TNF, and iNOS, and inflammation related to experimental autoimmune myocarditis [74]. Together, these findings suggest that N F K B regulates target cell apoptosis, virus replication and subsequent inflammation. By studying the genomic signature in CVB3-infected hearts and cells, I can potentially gain a greater understanding of such upstream signalling processes and the molecular mechanisms of downstream cellular consequences. 23 1.4 Cell Death It was once thought that infection by viruses simply involves overwhelming the cellular transcriptional and translational machinery and compromise of membrane integrity [75]. Indeed, acute infection by certain viruses can lead to overt cell death [76, 77]. Still other viruses take over apoptotic apparati to induce or circumvent cell death [78]. Viruses can manipulate host death machinery [79, 80]; in fact, many cellular anti- or pro-apoptotic genes were first discovered by virologists [79-83]. Coxsackieviruses have been shown to cause apoptosis in transformed cells and cell death resembling a mixed apoptotic and necrotic phenotype in vivo [84, 85]. Immune cytotoxicity in vivo generally causes a more overt cell death characterized histologically as coagulative cell death. The cell death phenotype in the heart caused by viral pathogenesis and immune infiltration, together, can influence host genomic adaptive, reparative, and degenerative processes. The following section is an introduction to the two broad phenotypes of cell death and the current knowledge of cell death in CVB3 infection. Necrosis, the equivalent of accidental cell death, is associated with loss of cellular homeostasis characterized by rupture of mitochondrial or plasma membranes. Cellular swelling due to water and electrolyte uptake, rounding up, loss of plasma and nuclear membrane integrity, and release of contents into surrounding tissue space are classical features of necrosis [86, 87]. Releasing intracellular contents often induces an inflammatory response that propagates the necrotic process [88-90]. Apoptosis, on the other hand, is a regulated and energy-consuming cell death process involved in physiological and pathological processes [91-93]. Apoptosis can be considered in terms of morphological consequences or biochemical mechanisms. Morphological changes during apoptosis include cell and nuclear shrinkage, nuclear chromatin condensation and breakdown into nucleosomal fragments, vesicle formation, budding of apoptotic bodies, and phagocytosis by neighboring macrophages and parenchymal cells [89, 90, 94]. Biochemically, apoptosis is typically regulated by a family of death proteases, 24 known as caspases. Caspases target the following structural and signalling proteins and degradation of such may be responsible for the cellular morphological changes that are observed during apoptosis: actin [95], a-fodrin [96], lamin A [97], lamin B [98], nuclear mitotic-associated protein (NuMA), Rabaptin-5 [99], keratin 88 [100], gelsolin [101] and the integrin-interacting focal adhesion kinase [102]. Apoptosis can be broadly described as primarily a receptor- or mitochondrial-mediated mechanism (Figure 9). Receptor mediated pathways are induced by binding of death-inducing ligands such as Fas-L, TNF-a and TRAIL to their respective receptors [103] and activation of caspase-8, followed by activation of down-stream caspases [104]. Alternatively, exposure to cellular insults can release cytochrome c (cyt c) from the mitochondria, which may associate with apoptosis activating factor-1 (Apaf-1) and pro-caspase-9, and in the presence of dATP, activate caspase-9 [105]. Caspase-9 can then cleave and activate caspase-3, leading to apoptosis through downstream events similar to those that occur in receptor-mediated processes. 25 Figure 9: Molecular pathways of apoptosis. Binding of TNF or Fas to their respective receptors results in the recruitment of adaptor proteins to form a death-inducing signaling complex (DISC) thereby inducing dimerization and activation of caspase-8, which activates downstream 'executioner' caspases. Granzyme B (GraB) can induce cleavage of caspase-10, or caspase-3 and -7 to a lesser degree, followed by activation of downstream caspases. Mitochondrial activation can release cytochrome c (cyt c) via pro-apoptotic Bcl-2 family proteins such as Bid and facilitates Apaf-1-mediated caspase-9 activation, which then cleaves caspase-3. Caspase-3 induces D N A fragmentation by cleaving ICAD/DFF, which then allows translocation of C A D into the nucleus or may cleave other structural or repair proteins or other caspases, such as caspase-6, which can cleave nuclear lamins. 26 The contribution of apoptosis to myocyte death in myocarditis has been the subject of debate [106, 107] but mounting histological and ultrastructural evidence and specific assays for key apoptosis markers suggest that apoptosis does participate [108, 109]. Apoptosis may be stimulated by virus-induced mitochondrial alterations [84], viral protease-based disruption of cellular homeostasis [84, 85, 110], and differential host transcription and translation. Our laboratory has previously shown caspase cleavage in CVB3-infected HeLa cells [84, 85], isolated cardiomyocytes and in mouse hearts (Yanagawa et al, unpublished observations). Apoptosis is triggered by several picornaviruses [111-115]. Considering the short replicative cycle of picornaviruses, prolonged host cell survival is of less value than to more complex viruses. On the other hand, premature cell death may prevent viral dissemination as has been shown by the cardioprotective effects of anti-apoptotic molecules in vivo [66, 71, 73]. Preliminary investigations with general caspase inhibition had no significant effect on myocardial apoptosis and inflammation when administered concurrent with virus infection (Yanagawa et al, unpublished observations). If apoptosis does contribute to myocardial injury following CVB3 infection, then anti-apoptotic strategies may be used in this settting to preserve myocardial integrity. 27 1.5 Immune Response Much work has been performed to characterize cellular immune infiltration in human and experimental models of myocarditis. In my array-based studies, the transcriptional profile of the heart likely includes contributions from infiltrating cells, particularly during stages of acute inflammation. Thus, to understand the outcome of altered transcription, it is important to understand which infdtrating cells are present and active in this setting. This section characterizes the makeup and describes the contribution of innate and adaptive immunity in myocarditis. Innate Immunity The innate immune response is one of the first lines of defense in the clearance of cardiotropic viruses. Infection of target cells and activation of macrophages trigger a pro-inflammatory cytokines which further activates macrophages, T cells and N K cells induce nitric oxide (NO) production, depress myocardial contractility and induce myocyte hypertrophy [116, 117]. Experimentally, early stage increases in interleukin-1 (IL-1), IL-6, tumor necrosis factor-a (TNF-a) and IFN-y are seen in myocarditic hearts [118]. A second wave of cytokines include IL-2, IL-4 and IL-10 and further increases in TNF-a, IFN-y and IL-lp, accompanied by infiltrating cells. Patient studies have revealed increased TNF-a and IL-1 in hearts with myocarditis [119, 120]. These cytokines are predominantly of the THI-proinflammatory type [118]. Specific cytokine modulation may improve outcome such as pretreatment with anti-TNF-a monoclonal antibodies which prolonged survival of virus infected mice whereas recombinant TNF-a increased virus replication and worsened myocardial necrosis [121,122]. IFNs bind to specific receptors to subsequently activate intracellular signalling necessary for activation and expression of IFN-responsive genes [123]. IFNs of the a/p subtype are referred to as type I IFNs and IFN-y is the only type IIIFN. Treatment of IFN-a with virus infection has been shown to reduce virus replication and myocardial injury [110]. Similar protective effects were shown with IFN-p. Transgenic mice over-expressing IFN-y in 28 pancreatic (3 cells protect mice from lethal CVB3 infection and subsequent myocarditis [124]. We previously showed up-regulation of the anti-viral and pro-survival interferon-gamma-inducible guanosine triphosphatase in CVB3-infected mouse hearts also suggesting the importance of IFN signalling [125, 126]. Cytokine expression also activates inducible NO synthase (iNOS) production of NO which signals through cGMP to trigger a cascade of biological events [127]. Nitric oxide has negative inotropic properties (causes contractile force decreases) primarily acting through S-nitrosylation of L-type calcium channels in the sarcoplasmic reticulum [125]. There are also direct antiviral actions of NO which may be due, in part, to direct inhibition of viral R N A synthesis [128]. The importance of the anti-viral role of NO is highlighted by high mortality in IRF-1 knockout mouse, which lack iNOS [129] and by increased CVB3 titres and mortality with NO synthase inhibitor treatment [130]. Adaptive Immunity The adaptive or antigen-specific immune response involves T and B cells which are committed to recognizing a specific antigen. Developmental progression of the acquired immune system may be considered triphasic beginning with lymphocyte-antigen recognition, followed by an activation phase for clonal expansion and differentiation, and an effector phase for elimination of antigen by activated lymphocytes. Antigen presentation and recognition involves the major histocompatibility complex (MHC) molecules. Target cell M H C class I binds endogenously-synthesized peptide antigens and are primarily recognized by cytotoxic T lymphocytes (Tc) via cell surface CD8. M H C class II on the surface of antigen-presenting cells recognize endocytosed peptides for presentation to helper T cells (TH) via CD4, facilitated by co-stimulator molecules such as CD28. These antigen presenting cells may include macrophages, B lymphocytes, dendritic cells and endothelial cells. 2 9 Human and animal models of acute viral myocarditis have T R I cytokine profiles [118]. Activated T H (CD4+) precursor cells differentiate into either T H I or T H 2 cells. T H I cells activate the cellular immune response, in particular macrophages, antibody production by B lymphocytes and T-cell clonal expansion, through release of IL-2, IFN-y and TNF-p. Thus clearance of virus and virus-infected cells in myocarditis are mediated primarily by macrophage phagocytosis and cellular myocytolytic clearance, as opposed to an antibody response. In human lymphocytic myocarditis, there is a predominance of T-lymphocytes, cytotoxic CD8+ Tc and CD4+ T H cells, macrophages and a lack of B lymphocytes by histological and histochemical assessment [3,131]. 30 1.6 Experimental Treatments The understanding of pathogenesis gained from gene expression profiling can potentially reveal novel therapeutic targets for myocarditis. Recently, the focus on viral injury causing early myocardial damage has inspired therapies that directly target the virus particle or genome itself. To this point, pleconaril, a picornavirus protein-coat binding agent, showed a clinical attenuation in 26 of 38 (78%) human enteroviral infections, including 12 of 16 patients with viral meningoencephalitis [132]. Antiviral treatments such as soluble receptor therapies and antisense oligonucleotide therapy [133] can block target organ injury and virus replication. Soluble CVB3-receptors, D A F and C A R , synthesized as IgGl-Fc fusion proteins (DAF-Fc and CAR-Fc, respectively) have been particularly effective. I found that DAF-Fc inhibits complement activity and that both DAF-Fc and CAR-Fc block CVB3 infection in vitro (Figure 10). Treatment with DAF-Fc in vivo preceding (PreDAF group) or concurrent (CoDAF group) with CVB3 infection resulted in the following: a significant decrease in lesion area and cell death (Figure 11) and reduced virus replication in mouse hearts as compared to delayed treatment (Post-DAF group) or virus alone (Virus group) [134]. Similar CAR-Fc treatment in vivo completely ablated lesion area, inflammation and virus replication in CVB3-infected hearts (Figure 12, 13) and pancreas (Yanagawa et al, In press Journal of Infectious Diseases) (Figure 14). Such therapies may be efficacious in treatment of fulminant and potentially life-threatening infections for which an enteroviral etiology can be readily identified. 31 Virus Dilution: Gauntt Kandolf Evans No Virus Gauntt Kandolf Evans No Virus » v f y « j v <o «o ^ <b °» <S <S cS cS cS cS cS cS c^S <6 K. n a i v t , N N . N » t > v » , v t « v t < N » > « . ^ > . None ... — .-, * a # o • • • • •V. • •» « « c c c t t t t f i cjlkv. ••••s. * * • , * * J** c O ro o < Q m B o L i . i a: < o 10-1 10-2 10-3 10-4 10-5 10-6 10"7None 2.5 mM 0.5 mM C^:#-#-## (Gauntt strain CVB3) Figure 10: Inhibition of Rhabdomyosacroma (RD) cell lysis by DAF-Fc and CAR-Fc. RD cells were preincubated with (A) DAF-Fc and (B) increasing concentrations of CAR-Fc and infected with various strains of CVB3 at increasing virus dilutions. Low concentrations of CAR-Fc and DAF-Fc inhibited virus infection in vitro (CAR-Fc inhibition of Kandolf strain was similar, data not shown). 32 Sham PreDAF L • CoDAF PostDAF Virus /n S/fu Hybridization ***** , • *» • *m -L • * \\\\\\\\m Figure 11 : DAF-Fc attenuates myocarditic pathogenesis and virus replication. Mice were infected with CVB3 in the presence or absence of DAF-Fc 3 days pre-, co-, and 3 days pi, or sham-infected (8 mice/group), and sacrificed on day 7. Cell death is prominent in postDAF tissue (arrow), which closely resembles virus group tissue containing intracellular calcification (arrowhead) and vacuolization (long arrow) localized to areas of in situ positivity. PreDAF and coDAF tissue is similar to sham tissue with a preservation of myocardial integrity and few areas of in situ positivity (320X). 33 Figure 12: CAR-Fc protects against myocarditis. Mice were infected with CVB3 in the presence or absence of CAR-Fc 3 days pre-, co-, and 3 days pi, or sham-infected (8 mice/group), and sacrificed on day 7. Heart tissue from PreCAR and CoCAR groups exhibit myocardial preservation, absence of viral RNA, and closely resembles sham tissue. PostCAR tissue exhibits cell death (arrows), presence of CVB3 RNA (arrow heads) and inflammation (broken arrow), although there is a reduction in both as compared to Virus tissue. 34 10 > n </) 6 o a. 5? 4 B o 5 0 . § a o 40 ~ >< = | 30 O) E Z 20 Jr t: O <0 ® "= 10 §• E Q- O) 0 Sham PreCAR CoCAR PostCAR Virus Figure 13: CAR-Fc reduces virus replication and infectious virion in CVB3-infected mouse hearts. A. Heart sections were processed for positive strand viral RNA in situ hybridization (see Fig. 12). Total area of positive staining was quantified and standardized using ImagePro-Plus® and expressed as a percentage of total cross-sectional area of the heart. Sham, PreCAR and CoCAR heart tissues did not contain virus RNA, PostCAR expressed reduced virus RNA as compared to Virus. B. Monolayers of HeLa cells were infected with heart homogenate, overlayed with agar/medium, and incubated for 72 hours. Viral plaques were fixed, stained and counted. PreCAR and CoCAR treatment completely eliminated myocardial virus and PostCAR reduced virus titres as compared to Virus tissue (Student's /-test, * pO.OOOl, # p<0.01). 35 Figure 14: CAR-Fc protects against pancreatitis. Mice were infected with C V B 3 in the presence or absence of CAR-Fc 3 days pre-, co-, and 3 days pi, or sham-infected (8 mice/group), and sacrificed on day 7. Exocrine pancreas tissue from PreCAR and CoCAR groups exhibit complete tissue preservation, absence of viral RNA, and closely resembles Sham tissue. PostCAR tissue exhibits cell death, presence of C V B 3 RNA and inflammation, although there is a reduction in both as compared to Virus tissue. Arrow heads show preservation of endocrine pancreas. 36 1.7 Conclusion Over the past 50 years since coxsackieviruses were first identified, we have gained much insights into the viral lifecycle and pathogenesis, the resulting inflammation and sequelae in the heart, and how to treat patients who present with myocarditis. The careful characterization of the phases of viral myocarditis provides a knowledge base from which to conceptualize in vivo genomic changes across these time points. An understanding of the pro-survival and pro-apoptotic signalling pathways triggered by CVB3 infection helps to analyse both causative upstream and resultant downstream genomic changes. An understanding of the makeup of infiltrating cells helps to understand the gene changes during acute and chronic inflammatory stages and potential cells of origin. Therefore, an understanding of these aspects of viral myocarditis is necessary to study the overall genomic profile in this disease. 37 Chapter II Introduction to Gene Profiling and Bioinformatics High-throughput assessment of gene transcript levels through the application of microarray technologies is central to this thesis. Multiple types of microarray technology for expression profiling were applied for this research program. Each will be briefly introduced. The groundwork for this thesis was laid in previous investigations from our laboratory, including studies of viral myocarditis using differential mRNA display [126] and cDNA microarrays [135]. The findings from these efforts will be summarized. The bioinformatical programs, including the relevant functions, will be described. Introduction This thesis is focused on the expression of genes in one cell type or group of cells within a tissue. As such, it is important to understand genome dynamics - the intricate control of expression balanced by R N A transcription and also degradation. Selinger et al [136] utilized oligonucleotides to show the average RNA half-life in the Escherichia coli (E. coli) MG1655 model to be 6.5 min. The half-life in mammalian cells is comparable although there is a broader range of long-lived and short-lived transcripts [137]. Therefore, measurable changes in gene expression following viral infection may occur on the order of minutes. Once synthesized, the mRNA molecule or gene is processed and translated into a protein polypeptide sequence of amino acids in the cytoplasm and these proteins control or regulate nearly every cellular biological process. Although the relationship is non-linear, investigations into gene transcriptional patterns may reveal insights into cellular protein expression and function [138]. Thus, by studying gene changes, I intend to gain an understanding of which proteins are potentially being differentially regulated. Although the coxsackieviral lifecycle has been elucidated and some major interactions with host cell proteins discovered, the effects of virus infection on host cellular gene 38 transcription patterns are largely unknown. Genomic studies using differential mRNA display (reviewed in [139]) in our laboratory showed 28 differential gene expression events following C V B 3 infection [126]. Select genes were partially sequenced and expression was confirmed for previously uncharacterized genes such as Nip21 and IGTPase by Northern blot. Using Tet-On-inducible HeLa cells, the Bcl-2 family member Nip21 was shown to reduce CVB3 replication by triggering mitochondria-mediated apoptosis of target cells [140]. More recently, we showed a protective role for IGTPase in CVB3-infected HeLa cells through PI3K activation and caspase inhibition [125]. These studies illustrate the overall approach of our laboratory to utilize gene profiling as a hypothesis-generating tool leading to targeted functional validation of the proteins encoded by the differentially-transcribed genes in enteroviral heart disease. Our next transcriptional investigations relied on cDNA microarrays to investigate, on a larger scale, temporal host gene responses [135]. Pooled R N A from myocardium of 14 infected and non-infected mouse groups were hybridized onto two identical custom cDNA microarrays. Of 7000 clones initially screened, 169 known genes had average expression levels that were significantly different (greater than log2 fold change of 1.8 in signal intensity) in experiment versus sham-infected control samples at one or more time points. Hughes et al [141] showed biologically relevant gene expression changes of log2 fold change of 1.5, with highly optimized protocols, but a slightly more conservative cut-off was chosen for these studies. The significance threshold for gene expression changes was chosen based on the expression level, therefore, fixed fold changes may underestimate significance at high levels and underestimate them at low levels [142-144]. As a drawback to this study, there was no statistical analysis performed to determine the statistical threshold for significance. Rather a blanket threshold was assigned and together with knowledge of the biological function, confirmatory hypotheses were pursued. Known genes were sorted according to their functional groups and interpreted in the context of viremic, inflammatory and healing phases of the myocarditic process. The re-interpretation and confirmation of this dataset will be one focus of this thesis. 39 2.2 Microarray Platforms Two microarray platforms were used for experimentation in this thesis, the cDNA array and the Affymetrix GeneChip®. Here, I provide an introduction to microarray technology and the three broad categories of platforms which are currently available. Microarrays are ordered sets of nucleotide probes which provide a medium for matching known and unknown D N A or RNA samples based on complementary base-pairing. These powerful research tools are broadly divided into cDNA arrays, Affymetrix oligonucleotide (oligo) arrays (also known as GeneChip®s) and long oligo arrays. Spotted cDNA arrays typically consist of immobilized D N A (500-5,000 nucleotides long) on a solid glass surface or nylon. Typically, glass substrates have covalently-coated reactive amines, aldehydes or epoxide groups to allow for stable attachment of cDNA and oligonucleotide molecules. GeneChips® are made up of 25mer oligo probes which include a "perfect match" (PM) and "mismatch" (MM) probe pair with 11-20 probe pairs per gene synthesized in situ by photolithography at Affymetrix [174]. Long oligonucleotide arrays contain 50-70 base oligonucleotide molecules spotted onto glass slides [145, 146]. A l l microarrays are exposed to a set of sample targets, either separately or in a mixture, and submitted for hybridization and detection. The following is a description of cDNA and GeneChip® microarray platforms: cDNA Arrays D N A microarrays were first developed at Stanford, California [147]. As a point of clarification, the probe refers to the nucleotide on the array and the target refers to the transcript in the sample. Currently, high-speed robotic spotters are utilized to spot cDNA clone probes with known identities on glass (or nylon) substrates for determination of complementary binding, allowing massively parallel gene expression studies. For dual-dye experiments, fluorescently-labeled experimental and control cDNA molecules are generated by mRNA reverse transcription in the presence of cyanine 3 (Cy3) or cyanine 5 40 (Cy5) dCTP. Recently, bias in dye fluorescence intensity due to physical properties of the dyes (e.g. heat and light sensitivity, relative half-life), dye incorporation, during scanner settings image acquisition have been recognized. Systematic variability in fluorescent dyes Cy3 and Cy5 indicate that dye swap labelling of samples and local regression normalization can limit this source of bias [148]. Dye swap is the process of using two fluorescent tags, on control and experimental target sample RNA, and subsequently switching the fluorescent tag on the two samples. The fluorescent signals from each configuration of dye labels are used as technical replicates. Each pair of fluorescently labelled cDNA samples (experiment and control) are hybridized competitively onto the microarray. The amount of hybridized material is measured by sequential excitation of the two fluorophores with a scanning laser in the range of emission spectra, known as image acquisition (Figure 15). The image is then converted to a ratio of intensities for experiment over control in a process known as image quantification. The ability to customize cDNA arrays allows individual researchers to readily perform multiple replicates. Array customization is favourable when studying organisms for which microarrays are not commercially available or those studying unique sequences. Two-sample analyses automatically yield paired expression ratios for analysis. Technical parameters, including sample concentration, brightness, dye labelling, exposure time, hybridization, washing stringencies and scanning camera sensitivity, must be optimized. However, this process has potential for error during PCR amplification of the cDNA probes or sample, which may compromise data quality. 4 1 Descriptions 2rtJ IS-9-2J U9WW2 c Signal Detection 34G39_at 74.4 A Cluster Incl. M57732:Human hepatic nuclear factor 1 (TCF1J mRNA, corr 34640_al 87.2 A Cluster Incl. X7134G:Homo sapiens HNF1-B mRNA /cds=UN KNOWN /g 34641_at 68.3 A Cluster Inci. X54380:Human mRNA for pregnancy zone protein /cds=(29 34642_at 8431.1 P Cluster Incl. U28964:Homo sapiens 14-3-3 protein mRNA, complete cds < 34643_at 58066.8 P Cluster Incl. M53458:Human ribosornal protein S4 (RPS4X) isoform mRW 34644_at 26301.6 P Cluster Incl. AB021288:Hcmo sapiens mRNA for beta 2-microglobulin, co 34645_at 63026.3 P Cluster Incl. X55715:Human Hums3 mRNA (or 40S ribosornal protein s3 / 34646_at 27884.1 P Cluster Incl. Z25749:H.sapiens gene tor ribosornal protein S7 A;ds={81 ,Gt 34647_at 6171.3 P Cluster Incl. X52104:Human mRNA lor pG8 protein /cds=[175,2019] Jqb--34648_at 4221.0 P Cluster Incl. Z12830:H.sapiens mRNA for SSR alpha subunit /cds=(29,8f 34G49_at 2940.3 P Cluster Incl. M14219:Humanchondroitin/dermatan sulfate proteoglycan ( 34650_al 719.1 P Cluster Incl. U3G798:Homo sapiens platelet cGI-PDE mRNA, complete a 34651_at 4883.2 P Cluster Incl. M58525:Horno sapiens catechol-O-methyltransf erase (COMT 34652_at 53.0 A Cluster Incl. U779G8:Human neuronal PAS1 [NPAS1] mRNA, complete c 34G53_at 93.8 A Cluster Incl. Z33905:H.sapiens gene for 43kD acetylcholine receptor-ass 35014_at 137.6 A Cluster Incl. X58401 :Human L2-9 transcript of unrearranged immunoglobi 35015_al 140.4 A Cluster Incl. M27533:Human Ig rearianged B7 protein mRNA VC1 -region. 35016_at 85.5 A Cluster Incl. M13560:Human la-associated invariant gamma-chain gene / 3501?J_at 1609.0 P Cluster Incl. M804G9:Human MHC class I HLA-J gene, exons 1 -8 and coi 35018_al 322.2 A Cluster Incl. UG1538:Human calcium-binding protein chp mRNA, complet 35019_at 68.1 A Cluster Incl. AF054180:Homo sapiens hematopoietic cell derived sine fine 35020_at 127.4 A Cluster Incl. D82344:Homo sapiens mRNA for NBPhox, complete cds /a 35021 _al 153.2 A Cluster Incl. U8932G:Homo sapiens bone morphogenetic protein receptor 35022_at 80.8 A Cluster Incl. S 83308: S0X5=Sry-related HMG box gene {alternatively splic ^t-^^ i __ . A . Figure 15: An example of Affymetrix GeneChip 1 image acquisition to quantitation. Individual total RNA samples were processed into labeled and fragmented cRNA for GeneChip® hybridization. GeneChips®s containing 16-20 probe pairs for >12, 000 genes were laser excited and images were acquired. The full array is visualized in (A) higher power view with spot grid is shown in B. C. Image quantitation was performed and reported (from left to right) as a unique Affymetrix identification, raw signal intensity, present/absent call and sequence annotation. 42 Affymetrix Oligonucleotide Arrays (GeneChip s) The second microarray platform used in this thesis is the Affymetrix oligonucleotide array, also known as a GeneChip®. New generation GeneChip®s for both human and rodents contains >33,000 annotated genes which are approaching true pan-genome scale studies. These arrays are designed with a set of 11-20 complementary sequence short oligonucleotides (known as a perfect match; 25mer) and a set of oligos that contain a single base mismatch to the predicted sequence (mismatch), for every probe on the array [149]. The probe pairs, described earlier, are technical replicates distributed along the gene complementary to specified 25mer sequences along the gene. The use of the probe set compensates for the lack of specificity of short oligos and enables estimation of variability among the precision of the measurement at the probe level. A detailed description of probe design can be found at the Affymetrix website (http://www.affymetrix.com/). Briefly, expressed sequences from databases-including GenBank, RefSeq, and dbEST, are collected and subdivided into clusters representing distinct transcripts. Sequence alignment to the human genome reveals splicing and polyadenylation variants, as well as distinguishing low and high quality sequences for probe design. Splice variants can be probed using the Affymetrix platform using all of sequences corresponding to the splice variants of a gene as input for the Probe Match tool in Microarray Suite 5.1. However, the total number of probes matching the query sequence should constitute at least 70% of the total number of probes in the probe set. Recently, more sophisticated methods of measuring the relative contributions of splice variants have been reported by Affymetrix [150]. Tissue or cell samples are collected and processed for hybridization. Typically, 15-40 ug of total R N A or 0.2 to 5 ug of mRNA is required. Total RNA samples are isolated from cells or tissues and double stranded cDNA is prepared using an oligo(dT)-T7 RNA polymerase promoter-containing primer. cDNA is transcribed in vitro into cRNA using biotin-labeled dNTP, fragmented by partial hydrolysis and hybridized to the GeneChip®. Each array is then submitted for excitation by a laser tuned to the excitation spectra. Following excitation, the dyes decay to a lower energy state, emitting fluorescence which is scanned using a fluorescent scanner optimized for detecting phycoerythrin-streptavidin 43 emission. The image is digitally analyzed and the signal intensity for each perfect match/mismatch pair is determined. In our experience, GeneChip®s have greater reproducibility given the standardization from chip synthesis to hybridization and from washing to detection. There are however, also reports to show that well calibrated cDNA array platforms can have greater reproducibility than Affymetrix arrays [151]. Ultimately, the data quality may reflect operator skill or laboratory expertise. Recently, innovations in oligonucleotide array technologies, specifically the use of a virtual photolithographic mask known as a Maskless Array Synthesizer (MAS), have contributed to the availability of custom oligonucleotide arrays (www.nimblegen.com; [152, 153]. In this system, U V light is shone on the M A S , a computer-controlled digital micromirror array, which then transmits a pattern of U V light onto the photosensitive groups covalently attached to the array surface. Long Oligonucleotide Arrays Long oligonucleotide arrays consist of 50-70mer oligonucleotides covalently attached to an array surface. Unique oligonucleotide probe sequences complementary to the target sequence are first synthesized. These oligonucleotides are printed onto glass slides (often in replicate spots) and crosslinked using U V irradiation. Target R N A is reverse transcribed into cDNA, Cy3- and Cy5-labeled (often with dye swapping where dye labels are alternated to control for any bias in dye coupling or emission efficiency of Cy dyes), and hybridized to the array. Wang et al [185] showed that the performance of 70mer long oligonucleotide arrays was comparable to cDNA arrays with a correlation coefficient of 0.80 and confirmed differential expression of sixty of 65 transcripts by real time-PCR. The spotted cDNA and GeneChip®s are the two microarray platforms utilized in this thesis work. There is also a need for informatical tools to process and analyze the resulting biological information. The synergy of high-throughput assays with tools to intelligently analyze the information can significantly accelerate exploratory research. 44 2.3 Data Mining Algorithms As a single genomic or proteomic experiment can produce tens of thousands of data points, software is necessary to extract, store, analyze, visualize and conceptualize useful biological information. Early post-array analyses were limited to lists of genes that were up and down-regulated [135, 154]. Recently, bioinformatics has emerged from traditional informatics to meet the needs for microarray design and printing, image processing, data storage and data analysis. Data analysis tools can be grouped into fdtering tools, used to dissect out interesting expression events; clustering tools, to rationalize gene expression patterns into groups of coordinately expressed genes; and visualization tools to better see patterns of gene expression. When used in an integrative manner, these programs can help dissect out genes encoding potentially important proteins, signalling pathways and signalling networks. The following is an introduction to computational biological programs and algorithms which were used to generate the data in this thesis. Here, I discuss the programs which were used for specific informatics objectives in analyzing the datasets in this thesis. This is neither a complete collection of all of the programs available nor of the entire functionality of each program. Rather, we describe our precise informatic objectives and how these were satisfied by the programs used. The bioinformatical programs used were Microarray Suite 5.1 (MAS 5.1; not to be confused with Maskless Array Synthesizer [MAS]), GeneSpring 5.0.1™, GenMAPP 1.0, and Genetrix (August 2002 release version). I also discuss the progress in standardization in reporting microarray data. Finally, I present an introduction to statistical analysis and the concerns related to interpretation of large-scale microarray datasets. Together, the combination of informatics and statistical analysis with the aforementioned programs has provided a foundation from which to develop hypotheses for further validation. Microarray Suite 5.1 In these experiments, M A S 5.1 provided instrument control and recorded experimental parameters relevant to the washing and scanning of the array. Specifically, M A S 5.1 45 controlled the Affymetrix Fluidics Station 450 including the temperatures and timing of each wash and stain step. This program also controlled the operation of the HP Agilent GeneArray Scanner used to scan the array. It controlled such parameters as the number of scans performed for each measurement, the pixel value and the wavelength of the laser and detector. M A S 5.1 then transformed the raw scanned images acquired from the GeneChip® first to a data image file. From that image, background correction and probe cell average intensities were measured to give the cell file. A single chip analysis was then performed to determine transcript signal intensities, detection calls and a measure of confidence, which was saved as a chip file. The text file is a spreadsheet format of the chip file which includes the following: raw signal intensity value, unique Affymetrix identification, detection call (presence/marginal/absent call based on the number of probe pairs for which the signal intensity for the PM>MM), the detection p-value and the gene description. M A S 5.1 also provided a report for image acquisition and interpretation parameters, quality control gene spikes and percent present calls (a measure of perfect match over mismatch intensities) which was saved for future reference. For my experiments, transformation with M A S 5.1 was a necessary first step in processing raw GeneChip® data. I limited the use of M A S 5.1 to control of array washing and hybridization steps and generation of text files which were submitted for subsequent analysis. For these tasks, there was no alternative to the use of M A S 5.1. Text files exported from M A S 5.1 were then submitted to GeneSpring™, GenMAPP and Genetrix for further analysis. I will now discuss the use of these tools to meet our experimental requirements. GeneSpring 5.0.1™ GeneSpring 5.0.1™ is a visualization and analysis platform designed for use with genomic expression data (Silicon Genetics, www.silicongenetics.com). I utilized GeneSpring for the following functions: to provide gene annotation, to view the temporal expression of all genes and subsets of genes, to search genes based on keywords which correspond to their description, to search genes based on membership in predetermined functional groups and 46 to perform expression-based clustering. GeneSpring 5.0.1 functions beyond those listed above were not used in this thesis. GeneSpring 5.0.1™ analysis was carried out in these experiments in the following manner. The Affymetrix U95A dataset which contained annotation information was downloaded from the Affymetrix website (www.silicongenetics.com). Affymetrix text files from M A S 5.1 were loaded into GeneSpring 5.0.1™ for each individual experiment. Time was treated as a continuous experimental parameter and treatment a non-continuous parameter. When applicable, triplicate data were assigned as biological replicates. Subsequent analyses were carried out in the 'Graph' view. Genes were first selected based on 'interesting' expression patterns, e.g. genes which exhibit an increase in expression during the inflammatory stage of myocarditis. The potential role of such genes in viral myocarditis was decided by their known biological function and/or known roles in myocarditis or related disease settings. The expression of functional gene groups was also investigated in GeneSpring 5.0.1™. The vocabulary and functional group ontologies utilized by GeneSpring 5.0.1™ were adapted from the Gene Ontology (GO) Annotation Project (European Bioinformatics Institute; [155]. Here, genes are first divided into three major ontologies, molecular function, biological process and cellular compartment, which are further subdivided into more specialized groups. Just based on the expression of a single gene, the relevance of that gene may not be immediately obvious. By adding functional group information to the genes, one can add more background biological knowledge at a higher granularity level. For example, by combining groups of genes, based on biological functions, certain functional groups may stand out for further investigation. The clustering function in GeneSpring 5.0.1™ was utilized as one strategy to reduce the complexity of expression data. Clustering is an unbiased approach to pattern discovery and data mining, broadly classified as bottom up or agglomerative (cluster joining) and top down or divisive (cluster breaking)[156]. 47 Hierarchical clustering is a type of divisive approach based on a set of dissimilarities of the distance metric for the objects being clustered. Initially, each object is assigned to its own cluster and the algorithm proceeds iteratively, at each stage joining the two most similar clusters as measured by the mean distances between the items that the clusters contain, continuing until there is just a single cluster. The distance measure between signal intensity patterns is considered a vector with both distance and direction. As a caveat with agglomerative algorithms a 'poor' decision made early during the tree construction will ultimately affect all subsequent clusters. Partitional clustering algorithms are constructed so that the behaviour in each group is distinct from any of the other groups. For instance, the k-means clustering algorithm divides genes into a user-defined number (k) of equal-sized groups. Cluster numbers are initially inputed based on prior knowledge of the biological system but are ultimately best optimized empirically. The number of clusters chosen will affect the results as large numbers of clusters will cause suboptimal groups whereas small numbers of clusters will results in grouping of genes with an insufficient similarity measure. Centroids are created in expression space at the average location of each group of genes. After all genes have been assigned a cluster, the locations of the centroids are recalculated and the process is repeated until the maximum number of iterations has been reached. K-means clustering has been exploited to mine transcriptionally up-regulated cellular systems in HIV-infected resting CD4+ T-cells [157]. Since the average property of all genes in a cluster determines the overall cluster characteristics, a poor choice of k (predetermined number of clusters) may cause important subsets of genes to be overlooked [158]. Too many clusters would separate genes with relatively similar expression profiles whereas too few clusters would combine genes with relatively dissimilar profiles. In my experiments, GeneSpring 5.0.1™ provided gene annotation and visualization tools which helped to appreciate significant changes in gene expression in infected as compared to non-infected samples. Alternatives to GeneSpring 5.0.1™ exist such as TIGR tools which also provides clustering and a similar visualization tool. 48 GenMAPPl.O GenMAPP 1.0 (Gene MicroArray Pathway Profiler) is an application designed to quickly and easily visualize gene expression data on maps representing known biological pathways [159]. A MAPP is a graphical display that shows biological relationships between genes or gene products. Hundreds of MAPPs representing diverse biological pathways and functionally related genes can be downloaded from a public database. These MAPPs, created and submitted to GenMAPP 1.0 by the scientific community, are then reviewed. Custom MAPPs for hypothesis testing may be drawn with the graphics tools provided by the GenMAPP 1.0 program submitted for public use. In my experiments, GenMAPP 1.0 was exploited to visualize the differential expression of function-based gene groups and to create a specific custom MAPP. First, GeneChip® text files were converted in Excel to ratios of intensity of experiment / control (virus / sham). Data from each experiment was loaded and saved in GenMAPP 1.0 as individual experiments and analysis criteria were selected in the Expression Dataset Manager. Since GenMAPP 1.0 cannot manage time-dependent changes or multiple variables (e.g. multiple treatments) in the analysis, one time point and experimental treatment was chosen per experiment. For each experiment, the criteria builder was used to set the ratio fold cut-offs which illustrates signal intensity changes as a color parameter. Here, colors were selected as red for up regulation, yellow for no change, blue for down-regulation and grey as not reported. Human GO MAPPs related to genes in cellular component, biological process and molecular function were downloaded from the GenMAPP website (www.genmapp.org). Data was visualized for all MAPPs in these groups and images were saved with the Print Screen function as image quality was superior to the image export function. I also created one custom MAPP to illustrate all known genes downstream of the M A P K -E R K signalling pathway. Drafting tools within M A P P Builder version 1.0 were utilized to build a custom MAPP of Downstream E R K genes based on the published literature. 49 GenMAPP 1.0 was a unique and valuable tool to assess gene changes in a holistic pathway and functional group approach. Genetrix (August 2002 release version) Genetrix is a genomic data analysis program created by Dr. Jonathan Buckley, University of Southern California, CA. Genetrix (August, 2002 version) was used to find average intensity values in a time course experiment over all time points to identify genes with differential expression values over several time points. GeneChip® text files were loaded into Genetrix as a single experiment with an experiment and control group. Gene expression over all time points was analysed as replicate data and mean signal intensities were calculated and plotted on a signal intensity vs expression difference plot. Those genes with relatively high expression differences for mean signal intensities was highlighted and labeled with a description. There is often overlap in the functionality in data analysis programs but subtle differences in normalization and visualization features allows the user to view the data from different perspectives. The use of different programs, strategies and array platforms has given me greater confidence in my results. The data in this thesis has been submitted to the above programs and the most useful output will be discussed. Microarray Standardization To facilitate comparison and integration of microarray data in the life sciences, the Microarray Gene Expression Data (MGED) Society (http://www.mged.org/index.html) organized the creation of microarray databases and promotes the sharing of high quality, well-annotated data. Two important contributions of the M G E D Society are the Minimum Information About a Microarray Experiment (MIAME), a set of guidelines that outlines the minimum information required to interpret microarray data [160], and the Microarray Gene Expression Markup Language (MAGE-ML) , an XML-based language to communicate information about microarray based experiments [161]. I have adopted 50 the M I A M E criteria for the descriptions of microarray experiments in the Results section of this thesis. Statistical Issues The optimum method for statistical analysis of microarray data is a subject of debate and will be briefly addressed here. As with any experiment, to determine whether a change in gene expression is significant, I defined the population parameter to test, formulated the null and alternative hypotheses, computed the appropriate test statistic and compared the calculated p-value (a measure of probability that a difference between groups during an experiment happened by chance) to the pre-specified false positive level. There are two major issues related to statistical analysis with microarray experiments: the consideration of multiple testing and the use of an appropriate statistical test based on the sample distribution. First, the issue of multiple testing will be discussed. Typical biological experiments have a 95% confidence interval. However, multiple comparisons can inflate the type I error rate (incorrectly rejecting a true null hypothesis). The simplest adjustment is called the Bonferroni which reduces the significance level inversely proportional to the number of independent tests performed (Bonferroni CE, 1936). However, this test is much too conservative as an original p-value of 0.05 for 7000 genes by this method becomes 7.1X10"6 (0.05/7000 tests). In such a case, legitimate significant results will likely fail to be detected. Thus, better statistical approaches are needed. The Bonferroni step-down adjusted p value affords improved power over single-step procedures while maintaining strong control of the error rate (defined as the probability of making a type I error). In these experiments, the Bonferroni-Holm method was utilized [162]. The Bonferroni-Holm method is described mathematically as the following: For Hypothesis H, where j=l..k ; P is the unadjusted p-value which are ordered such as | Pi | < | P 2 1 < I P3 I ... I Pk I (i.e. from smallest to largest). Then, Pi is compared to ak. If Pi < ak then reject Hi and continue, otherwise stop and fail to reject any hypothesis. 51 Then P2 is compared to ci(k-i). If P2 ^ a(k-i) then reject H 2 and continue, otherwise stop and fail to reject any hypothesis H 2 , . . .,Hk. Continue until a stop is encountered or until all hypotheses have been rejected. This test does not provide a confidence interval and does not require model or distribution-related assumptions which make it applicable to any family of pairwise comparisons. It is also less conservative than the standard Bonferroni test. Next, statistical testing will be addressed. A /-test was used to determine whether the mean signal intensity difference between samples from two groups was statistically meaningful. The /-test is defined as a ratio of the difference between two groups over the variability, also known as the standard error of the difference (between the two groups). There are two categories of statistical tests based on the sample distribution: parametric and non-parametric. Parametric inferential statistical methods are mathematical procedures for statistical hypothesis testing which assume that the signal intensity distribution has well-defined characteristics. Non-parametric statistical tests do not rely on the estimation of parameters describing the gene distribution on the array but may be too conservative. I have used the Welch's /-test in GeneSpring™ 5.0.1, a parametric statistical test, for analysis of differences in signal intensities. For instance, in the in vitro HeLa cell GeneChip® experiments in the Results, the observations are considered independent as the cells used were randomly placed into control or experimental groups. The measure of signal intensity is considered interval or exact-number data. The majority of signal intensities measured were unchanged and the distributions were symmetric but the variance in intensities in experiment and control samples were considered unequal. Based on these features of the intensity data, the distribution is considered independent, of interval character, symmetric and nonhomogeneous. Therefore, the Welch's /-test was utilized. Both the Welch's and the more common Student's /-test (first developed by W. S. Gossett) are used to measure 'small' samples. However, these tests differ in that the 52 Student's t-test is used to determine in paired groups with equal variances. The Student's Mest can be of several types: the one sample, used to compare an inferred population mean from a hypothetical population (e.g. from the literature); the two independent sample, used to compare two inferred means from populations for which the measured values are independent; and two dependent sample type, used to compare two inferred means from populations for which the measured values are dependent of each other. If the variances in the intensity data for the control and experimental groups described above were considered equal, then the appropriate statistical test would be the two independent sample Student's t-test There are still other important matters including data normalization and, for GeneChip®s, the optimal interpretation of probe pair signal intensities. Data normalization is the process of removing non-biological, systematic technical differences from intensity measurements. Affymetrix normalizes arrays by scaling the average values for each array (Affymetrix M A S v5.1 Manual). Two important assumptions are that signal intensities have a linear relationship between arrays and that most of the genes are unchanged on any given chip. Local regression techniques such as Lowess (locally weighted scatterplot smoothing) are more robust local data fit normalization methods [148, 163]. Another important issue specifically related to GeneChip® data analysis is the differences in hybridization between the 11-20 oligonucleotide probes for a given gene on the GeneChip®. Differences in hybridization may result from differences in target binding thermodynamics (due to differences in GC content), location on the chip or non-specificity of the oligo sequence. Affymetrix M A S 5.1 calculates an average difference value but more robust MultiChip Average algorithms have recently been developed [164]. Other important issues are the optimum method to address chip noise, probe hybridization variability, housekeeping genes and spike genes into the analysis, which are currently being investigated. 53 CHAPTER III Gene Profiling In Vivo 3.1 Characterization of the Mouse Model of Viral Myocarditis Rationale I sought to characterize histological and functional heart muscle injury and virus replication in CVB3-infected adolescent A/J mouse hearts. This myocarditis-susceptible mouse will be used as an in vivo model to profile gene expression changes. Experimental Design Forty adolescent A/J mice were randomly placed into two groups: 28 mice infected intraperitoneally with 1X10 ? plaque-forming units (pfu) of Gauntt strain CVB3 or 12 mice sham-infected (injected with sterile PBS). Gauntt strain is originally a Nancy strain virus isolated by Woodruff et al [165] and previously characterized by our laboratory [166]. This virus strain was utilized because it causes more severe and prevalent histological lesions in mouse hearts as compared to other strains [166]. At 3, 9 and 30 days post-infection (pi) mice were submitted for two-dimensional echocardiography, and heart (mid short-axis section), spleen, pancreas, kidney and liver were harvested for subsequent analysis. Tissues were formalin-fixed and paraffin-embedded, then submitted for routine histology including H & E and Masson's trichrome stain, and in situ hybridization (ISH) for virus RNA (+ and - strands). Heart base sections were stored at -80°C. This tissue was subsequently homogenized in PBS and infectious virus particles were measured by plaque overlay assay. Contribution by the Author This author was responsible for experiment planning, animal husbandry, propagation of virus, infection, animal monitoring, sacrifice, tissue harvesting and processing, plaque assay and image acquisition. Routine histology and ISH were performed by clinical histology laboratory, St. Paul's Hospital and technical specialist Agripina Suarez, respectively. Echocardiography was performed jointly by this author and Stephanie 54 Bonigut (Technologist, Amgen) using an Agilent Sonos 5500 Echocardiography System used in collaboration with Dr. Chris Thompson, Clinical Cardiology, St. Paul's Hospital. Results L Overall survival The overall mortality rate of adolescent A/J mice (N=28) following CVB3-infection was 43.5% by 30 days post infection (pi) with the greatest mortality occurring between 7 and 10 days post-infection with deaths (Figure 16). Regardless of survival, all infected animals exhibited significant morbid signs, particularly between 7-14 days post-infection, which included any number of the following: decreased activity, roughness in fur, loss of fur, loss of appetite, weight loss, conjunctivitis, and skittish behavioural changes and self-mutilation. The severity of morbidity and high mortality rates in CVB3-infected mice is analogous to pernicious enterovirus infection seen in young human populations [19]. it Histological injury and virus replication Viral myocarditis exhibits distinct histological features during the stages of viremia, inflammation and reclamation. By day 3 pi, hearts exhibited scattered foci of cell vacuolization localized to replicating virus (Figure 17). Hearts then exhibited extensive myocyte death, coagulative changes and contraction band necrosis, large areas of in situ calcified myocytes and mononuclear cell infiltration by day 9 (Figure 18). Injury was accompanied by in situ hybridization positivity (CVB3 + strand RNA) co-localized to injured and dead myocytes. Cardiac myocytes are the major cell target to support coxsackievirus infection in the heart. Endothelial cells have been shown to be susceptible to coxsackievirus infection, although to a much lesser extent [167]. Lesions were accompanied by cellular infiltrate previously characterized as mononuclear cells, lymphocytes and macrophage subsets [166]. The pattern of lesion area was punctuate but spread throughout the mid-ventricular short axis section including R V and L V , septum and even papillary muscles. By day 30, widespread fibrosis was seen in areas of myocyte dropout localized to myocyte disruption and sparse to no remaining viral burden (Figure 55 1 - p n n m n n n r 0.8 0.6 0.4 0 "t—1—I I I I I—I I I I I — I— !— I — TH — I I—I I I I I I I—I I I 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 Days Post-CVB3 Infection Figure 16: Survival curve for CVB3-infected mice. Forty male 4-5 week old A/J mice were intraperitoneal^ infected with 105 plaque forming units of CVB3 (Gauntt strain) or PBS. Mice were observed daily for morbidity and mortality. Overall survival at day 30 was 56.5%. Mortality occurred primarily in two stages: during the inflammatory phase of infection from days 7-9 and less so chronically about day 20 post infection. 56 a m 1 ^ s e f t ' : > . i * »* 03 N — I-j X3 o o '5b _o c 3 CD '  V « t' J f " ^ £ 8 <u 03 ** _c & » ;—; co 03 O -S t <a a CI, 03 -4—» 1> .5 2 CD -g > 03 ? CO 93 3 g j £ 03 — — -CO i 3 03 CO cu I o 03 r j 00 t> c ^ a u x .£ c co o D H co I I O . cU o J= C o ° c o O O e g > s © rt ^ 8 O <u 1:1 CO c a c c « N 3 co co 03 si => . s o c 57 ' V ^ * s N TJ £ 3 1 L •is i' • c cu I) in 4 CO ce •— CO O J= 09 as o S o c g CQ > u CU co cd 43 o CO CO C CU VH Ji <2 E c ^ o s "s 0 ~ ••c .a ° -3 ,<L> . » C XI 1 >> co as < 8 « X G S 3 <u I £ • co CU m jo CQ | > — U g o «> •1 o B to 3 2 CO — co cu -= st o En co C c '£ '3 CO O P H cd co •S? > Z o o ** C CO O U ^ O cu c a cu 43 o IS - o m•S cS • s o ® c« co co CU CD 3 y O " o C cu u u •Is 'co ,—, 1 O a> X - UJ co CO ) H s 153 < •5 2 .2 £ 70 -o E <U CO O c3 SH C3 Xi O cd u '5b _g o t o , C U CD CO CU C3 a 5 O X S "2 cu C cT S !> u * c d o i S eg « ^ b o x 'CO e C N CO — fa 2 co 3 > OS T3 cu CO c a O C 'C CO CO co 5 1 CU _ Z .2 '— >> s 3 o s § X ^3 c « H CO ° >, B > a ' 5 T3 o h O H ^ o ^ *-S 2 § o "2 o S o x O H .g co _ , 6 8 8 ^ OQ X rfi ^ H | | co o >,c2 § M 2 § H ^ o I ( N cU tn bo X * E o a o •2 * cu > CO t> co cu cu •2 S s > co o O H C — • cu l-1 x cu x 3 ^ 2 S « o S 65 O 58 co co O O -c ft .5 59 19). Appreciable myocyte heterogeneity likely from disregulated atrophy, hypertrophy and cell death was observed. For a semi-quantitative assessment of pathology, hearts were graded without knowledge of the experimental groups (B.M.M.) for histological features of injury (Figure 20). A l l grades are on a 0-5 scale. There were subtle but significant foci of cell death as shown by an average semi-quantitative grade of 0.6 by day 3. By day 9, there were grades of 4.5 for myocardial lesion area, 4.5 for cell death and 3.5 for calcification and 2.1 for inflammation. By day 30, grades of 2.6 for fibrosis, 3.0 for extensive lesion area and 1.6 for calcification with variable chronic inflammation were noted (N=4 for all groups but day 30-virus, N=3). Epicarditis is normal in A/J mice and not part of the virus-induced injury [166] To further confirm myocardial infection, plaque assay was performed to detect infectious virus particles. An increase in virus particle release at 3 and 9 days pi and no recoverable infectious virus particles by day 30 was shown (Figure 21). Thus there was overall consistency between presence of virus RNA by ISH and production of myocardial infectious by plaque assay. iiu Functional changes Two-dimensional echocardiography was used to determine if the mouse model of myocarditis functionally resembles infection in humans. Mice were anesthetized and hearts were imaged in parasternal short (at the mitral valve and papillary muscles) and long axis planes using echocardiography (Figure 22, parasternal short axis section at the level of the papillary muscles are shown). Heart function was quantitated as ejection fraction defined as the fraction of blood in the left ventricle pumped out of the heart during systole (e.g. with each heart beat). At day 3 pi, no differences in L V wall thickness or ejection fraction in virus-infected mouse hearts were seen (Figure 23, 24). At day 9, hearts exhibited 80.5% diastolic and 50.3% systolic L V wall thickening, an accompanying reduction in inner chamber dimensions and a non-significant trend of increasing ejection fraction (Figure 24). At day 30, virus-infected hearts experienced a 19.9% reduction in ejection fraction as 60 compared to age-matched control mice (Figure 25). Sham-infected mice exhibited no significant changes but a slight decreasing trend in ejection fraction from 5 to 9 weeks of age. The cardiac dysfunction seen in patients with human myocarditis including decrease in ejection fraction and cardiac arrhythmias (heart conduction abnormalities) has been well characterized and recently described [5]. Taken together, the CVB3-infected A/J mouse histologically and functionally resembles chronic human viral myocarditis as described. This data supports the use of mouse chronic myocarditis as a model for infectious dilated cardiomyopathy using a measurable functional endpoint. 61 5 Cell Death I I Lesion Area I I Calcification • M M ™ ™ Inflammation t M Fibrosis Figure 20: Semi-quantitative histological grades of murine myocarditis. Forty male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At days 3, 9 and 30 post-infection, heart were harvested, processed for histology and stained with H & E and graded without knowledge of the experimental groups (B.M.M.) for salient histological features of heart injury. Subtle but significant foci of cell death were present by day 3. By day 9, high grades of myocardial pathology with modest inflammation was observed. By day 30, fibrosis and extensive lesion area and calcification with variable chronic inflammation were observed. Baseline hearts represent uninfected mice at day 0, and did not exhibit any appreciable histopathology (N=4 for all groups but day 30-virus, N=3). 62 2.0E+08 E (0 a w *-> 'E a) c 5 o u_ a> 3 D" (TJ t: a> x ^ ex** ^ c jc^ .<y=> or> .<3T «ft* Figure 21: Plaque assay for infectious virus in myocarditic tissue. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At 3, 9 and 30 days following CVB3 or sham (PBS) infection, apical portions of mouse hearts were flash frozen, homogenized and submitted for plaque assay analysis (N=4 for all groups but day 30-virus, N=3). Abundant virus particles were isolated from heart tissue at 3 and 9 days pi and no recoverable virus particles at day 30. 63 Day 3 P I : e . * Car YANAGA 2d 21 FES 32 2e: Irs: 33 pficc B'3'<'F5 SFh I - = I ; H I N E S3 2-1 <3-.'-"> d*yl Uil CCKF '3 3CM i i l l - Z 1P53PM 3-V-4 End Systolic dimonsion SAX. Papillary muscle Day 9 S 12 Day 30 5 12 JO-S-1 End Sytt«Uc dlm*w**on SAX. Papillary muscle Figure 22: Two-dimensional echocardiography of sham-injected mouse hearts. Male 4-5 week old A/J control mice were intraperitoneally infected with sterile PBS. At days 3, 9 and 30, mice were submitted for 2D-echocardiography. Images represent cardiac short axis views at the level of the papillary muscles in end systole (as shown by 3-lead EKG) and the outline (red) represents the width of the left and right ventricular myocardium. No appreciable differences were observed in gross heart images at 3, 9 and 30 days (N=4 for all groups but day 30 Virus, N=3). 64 Day 3 Day 9 Day 30 Figure 23: Two-dimensional echocardiography of CVB3-infected mouse hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At days 3, 9 and 30, mice were submitted for 2D-echocardiography. Images are papillary muscle short axis views in end systole (as shown by 3-lead EKG) and the outline (red) represents the width of the left and right ventricular myocardium. Increased L V wall thickness and deceased L V chamber size was found at day 9 pi. Chamber dilation and increased overall chamber size was found at day 30 pi (N=4 for all groups but day 30 Virus, M=3). 65 co 120 09 = 100 o Days Post Infection — Virus, Diastole Virus, Systole ~ ~ Sham, Diastole ~ - ~ Sham, Systole Figure 24: Posterior wall dimension measurements using 2-dimensional echocardiography of normal and myocarditic hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain) or PBS. At day 0 and then 3, 9 and 30 days following CVB3 or sham (PBS) infection, mouse hearts were submitted for 2-dimensional echocardiography (Sonos 5000, Philips). Systolic and diastolic parasternal short axis measurements were taken at the level of the papillary muscles (N=5) in the dorsal recumbency position. There was increased diastolic (red) and systolic (orange) posterior wall thickness, particularly at 9 days pi as compared to diastolic (light blue) and systolic (dark blue) wall dimensions for sham-infected mice (N=4 for all groups but day 30 Virus, N=3). There was also a trend in decrease in diastolic posterior wall thickness at day 3. 66 0) O) £ C (0 <0 Q. 51 vo O «—-15 10 O O (0 .2 © o > o ro u T Q c 2 -10 p -15 -20 -25 •30 Days Post Infection O Control • Virus Figure 25: Ejection fraction using 2-dimensional echocardiography of normal and myocarditic hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain) or PBS. At day 0 and then 3, 9 and 30 days following CVB3 or sham (PBS) infection, mouse hearts were submitted for 2-dimensional echocardiography (Sonos 5000, Philips). Systolic and diastolic parasternal long axis measurements and parasternal short axis measurements were taken at the level of the mitral valve and papillary muscles (N=5). Measurements were used to calculate ejection fraction using a modified Simpson's algorithm (N=4 for all groups but day 30-virus, N=3). Infected mice showed a trend in functional increase at day 9, and a marked decrease at day 30 (filled square). Sham infected animals exhibited a trend in decrease in function over time (unfilled diamond). 67 3.2 cDNA-Based Bioinformatical Analysis of Mouse Heart Genes Biological Question What are the differentially transcribed genes and functional gene groups in CVB3-infected hearts? Rationale To understand the host transcriptional profde during the viremic, inflammatory and reparative stages in myocarditic hearts, we previously performed a cDNA array time course experiment on CVB3-infected hearts at 3, 9, and 30 days pi (N=14 mice per sample, N=2 arrays per timepoint, Figure 26). Known genes with a signal intensity ratio greater than a log2 fold increase of 1.8 in experimental/control sample at any one time point were grouped according to functional categories [135]. The differential gene expression cutoff was determined from a survey of previous published studies and from the cutoff used in the original published study [135, 168-170]. To extend these observations, the original dataset of approximately 7000 genes and expressed sequence tags (ESTs) was re-interpreted against the GenBank database release 129, using B L A S T N version 2.2.4, and further informatical analysis was performed. Transcriptional information from this previously acquired cDNA array database was mined. MIAME Criteria: Experimental Design and Quality Control Experimental design: Adolescent A/J mice were randomly placed into two groups of mice infected intraperitoneally with 1X10 5 pfu of Gauntt strain CVB3 or sham-infected (PBS). At 3, 9 and 30 days pi, heart tissues were flash frozen (N=14 mice/group), pooled R N A was extracted, processed and hybridized to custom spotted cDNA arrays (Incyte). Array Design: Custom cDNA arrays contained RS7000 C D N A clones randomly collected from a normalized male Wistar rat heart cDNA library [171]. Microarray fabrication, sample hybridization and image interpretation using a proprietary in-house analysis program, GemTools, were performed at Incyte Pharmaceuticals (Palo Alto CA) [172]. 68 Treatment Timepoints 10 5pfu 3, 9, 30 days pi Sham 3, 9, 30 days pi Array cDNA Array 105 pfu (N=2; 7000 genes and ESTs) 3, 9, 30 days pi Sham 3, 9, 30 days pi M O 1 1 0 GeneChip* (N=1; 25,204 genes) • 30 m, 1,3, 7, 9 h Sham • 0, 30 m, 1, 3, 7, 9 h GeneChip' (N=1; 12,627 genes) DMSO + U0126 + +- 30 m, 3, 9 h * 9 h DMSO 0, 30 m, 3, 9 h GeneChip® (N=3; 12, 627 genes) 69 Figure 26: Experimental design for microarray studies. Diagrammatic representation of mouse- (A,B) and HeLa cell-based (C,D) array experiments including timepoints, virus dosages, treatments and hybridization array-type. Male, adolescent A/J mice were infected with 105 plaque forming units (pfu, as measured by plaque overlay assay) of CVB3 and sacrificed on 3, 9 and 30 days post infection. Hearts were harvested and R N A was isolated and processed for hybridization (A) to duplicate cDNA arrays (N=15 mice/time point; ratio of intensity values are an average of N=2 chips/time point; Incyte Pharmaceuticals). For experiment B, RNA from CVB3- and sham-infected mice were pooled (N=4 animals but N=3 for day 30 virus) and hybridized to custom mouse GeneChip® arrays (N=l). C. To study enterovirus infection in an in vitro model, HeLa cells were infected with a multiplicity of infection of 10 (MOI 10, ie. 10 pfu virus/cell) and at time points from 0 to 9 hours post infection, RNA was hybridized to GeneChip® arrays. D. In the final microarray experiment, all cells were preincubated in DMSO or DMSO and U0126 and infected with CVB3 or sham. Triplicate biological samples were taken at 0, 30min, 3h and 9 h post infection and RNA was hybridized to GeneChip® arrays. 70 cDNA inserts were generated by PCR amplification with primers derived from flanking vector sequences [173]. PCR products were arrayed from 96-well microtiter plates onto silanated microscope slides in an area of 1.8 cm 2 using print tips driven by high-speed robotics. Printed arrays were incubated for 4 hours in a humid chamber and rinsed once in 0.2% SDS (1 min), twice in H 2 0 (1 min), and once in sodium borohydride solution (1.9 g of N a B H 4 dissolved in 300 mL of PBS and 100 mL of 100% ethanol; 5 min). The arrays were submerged in H 2 0 (2 min) at 95°C, transferred quickly into 0.2% SDS (1 min), rinsed twice in H2O, air dried, and stored in the dark at 25°C [168]. Samples: Mouse ventricular heart portions from each experimental group (N=14 mice/group) were pooled, flash-frozen in liquid nitrogen and stored at -70°C. Heart tissue was homogenized using a rotor-stator homogenizer in lysis buffer and sample R N A was isolated using the QIAGEN (Valencia, CA) RNeasy® isolation kit as per the manufacturer's instructions for isolation of mRNA from animal tissues. R N A was twice selected by oligo(dT) chromatography. Fluorescently-labeled cDNA probes were generated by reverse transcription of poly(A)+ mRNA in the presence of Cy3 or Cy5 dCTP (Amersham). Measurements: Degree of competitive hybridization was quantified by sequential excitation of the 2 fluorophores (Cy3 and Cy5) with a scanning laser. Expression data were omitted i f the signal was derived from <40% of the area of the printed spot as determined by the GemTools in-house program (Incyte Pharmaceuticals). A l l intensity data was stored in oracle databases and accessed using a proprietary application called Clone Navigator (SCIOS Inc.). Normalization Controls: Expression data were omitted by GemTools if the signal intensity was <2.5 fold greater than local background signal. Signal intensity values represent an average intensity from two separate cDNA arrays and are reported as a log2 fold change in experimental over sham samples. Signal intensities are log transformed to provide a symmetric view of up and down-regulation. Contribution by the Author Coxsackievirus infection of mouse, tissue collection, RNA isolation and array hybridization were performed jointly by iCAPTURE Centre and SCIOS researchers as 71 described in [135]. Re-interpretation of the data set including B L A S T N search and hierarchical clustering was performed by Ann Kapoun (Technologist, SCIOS Inc.). This author was responsible for function-based visualization of data, small functional gene set analysis and identification of interesting genes based on biological function. Results Male adolescent A/J mice were infected with CVB3 and degree of infection was confirmed by histological assessment, in situ hybridization for virus positive-strand RNA detection and plaque assay for infectious virus particles. Whole hearts, including left and right ventricular but not atrial tissue, from myocarditis-susceptible mice were harvested. Atrial myocardial tissue was excluded as atrial myocytes have been shown to differ from ventricular myocytes in their susceptibility to infection by CVB3. Total RNA was grouped from hearts, which artificially reduces the biological variability of transcriptional events. Pooled R N A samples were hybridized to two identical custom spotted cDNA arrays, which are considered technical replicates. Intensity values were reported as averages of the two technical replicate arrays. Function-based grouping In a function-based approach, all annotated genes were categorized according to functional groups (segmented by color) at the three time points and graphed based on their fold change in expression intensity (values are averages of N=2 arrays) at days 3, 9 and 30 (Figure 27). Functional annotation was provided by Gem Tools, a proprietary informatical program (SCIOS Inc.). These results are consistent with our previous observation that functional groups of genes have similar expression patterns and extends these findings by including the magnitude of expression ratio values [135]. Transcription in virus-infected hearts is most perturbed during inflammation. This may reflect transcriptional events in activated cellular infiltrates which are present histologically (Figure 18). Across all time points, there was an increase (in order of decreasing intensities) in the signal intensity of genes involved in cell division, the cytoskeleton, the extracellular matrix, metabolic transport, stress responses and immunology. There was also a decrease in transcript signal 72 intensity (in order of decreasing intensity) involved with contraction, metabolism, mitochondrial function, and channel and transport proteins. Transcriptional decrease in contractile and metabolic genes may contribute to the observed chronic reduction in ejection fraction. The number of genes with a similar intensity pattern within functional groups suggests that these are 'real' transcriptional events in myocarditic hearts. The fifteen most highly regulated genes from the cDNA array database were chosen based on expression alone and their functional categories and intensities are show in Table 1. The chip-to-chip variation for these gene intensities can be considered representative of all the variation on the array. Marked up-regulation of heat shock protein 27 during viremia and M H C class I during inflammation, known biological events, can be used to help validate this gene subset. Based on known biological function, genes such as SlOO-related protein, muscle L I M protein and cathepsin L were noted as potentially interesting but not previously been considered in the setting of myocarditis. The ability to identify such novel genes is one of the strengths of this exploratory investigation and as such, it will be focus of my discussion of hypothesis generation. Expression-based grouping In an expression-based analysis approach, signal intensities were submitted for hierarchical clustering. Data was Z-score normalized prior to clustering (as described in Materials and Methods). The Z-score method transforms data as a unit of the standard deviation [174]. Genes were clustered into eight unique groups according to normalized (Z score) intensities using a U P G M A (unweighted pair-group method using arithmetic averages) clustering algorithm. Genes clustered into 8 groups (red represents up-regulation; green represents down-regulation; black represents no change) and sample expression patterns were then plotted for the three time points (Figure 28). The two most populous clusters expressed marked gene up- (group 8, 36.8% of total) and down-regulation (group 2, 19.8% of total) at 9 day pi. This finding suggests that either cellular infiltration is highly transcriptionally active, that infiltration induces transcription in endogenous cells, or both. The mitochondrial genes primarily belong to groups 1 and 2, immediate early response genes 73 Communication Cell Defence Gene/Protein Expression Figure 27: Gene profile overview and functional classification in CVB3-infected mouse hearts. The experimental protocol is illustrated and described in Figure 26 A. The data set was then re-interpreted. Line graphs are shown for gene signal intensity ratios at days 3, 9 and 30 post-CVB3 infection. On the X-axis, Genes were grouped by color according to annotated biological function and on the Y-axis, fold-change in ratios of signal intensity (CVB3-infected over sham-infected hearts) is plotted. Genes which belong to functional groups exhibited similar temporal expression patterns. Global increases in the transcript signal intensities related to cell division, the cytoskeleton, the extracellular matrix, metabolic transport, stress responses and immunology are shown. Decreases in transcript signal intensities related to contraction, metabolism, mitochondrial function, and channel and transport proteins are shown. Statistical analysis was performed for individual gene expression events but is not represented in this figure. 74 # Funct ional Category Descr ipt ion Day 3 Day 9 Day 30 1 Cell Division: Cell cycle: cyc l in G -1.25 (-1.3,-1.2) -2.5 (-2.8,-2.2) -1.15 (-1.2.-1.1) 2 Cell Signaling/Communication: Cell adhesion: galectin-3 ; aka Mac-2, and IgE binding protein 1.2 (1.2,1.2) 5.75 (9.5,2) 3.25 (4.3,2.2) 3 Cell Signaling/Communication: Effectors/modulators: Serca 0 (1.6-1.5) -1.8 (-13,-2.3) 0(1.7,-1) 4 Cell Signaling/Communication: Metabolism: MIPP65 -1.15 (-11,-12) -2.3 (-2,-2.6) -1.3 (-1.1,-1.5) 5/13 Cell Signaling/Communication: Receptors: perlpherai-type benzodiazepine receptor 2.25 (2.1,2.4) 5.75 (5.5,6) 1.5 (1.6,1.4) 6 Cell Signaling/Communication: Miscellaneous: S-100 related protein 0(1.1,-1.2) 4.2 (5.3,3.1) 1.3 (1.3,1.3) 7 Cell Structure/Motility: General: LIM protein (FLH1) OR Cak receptor (Cak/DDR) 5 (5.1,4.9) 8.3 (8.2,8.4) 2 (2.1,1.9) 8 Cell/Organism Defense: Homeostasis, stress response: hsp 27 4.05 (3.9,4.2) 2.7 (2.5,2.9) 1.25 (1.3,1.2) 9 Cell/Organism Defense: Immunology: MHC class I antigen 4.45 (4.6,4.3) 7.25 (8.2,6.3) 3.35 (3.3,3.4) 10 Cell/Organism Defense: Immunology: MHC class I RT1.Au heavy chain 6.05 (6.3.5.8) 8.55 (8.9,8.2) 3.35 (3.9,2.8) 11 Gene/Protein Expression: Protein synthesis/turnover: cathepsin L 3.2 (3.8,2.6) 6.2(6.1,6.3) 2.25 (2.5,2) 12 Metabolism: Transport: beta-giobin -2.2 (-2.1,-2.3) -4.75 (-4.1,-5.4) -2.2 (-1.8,-2.6) 14 Miscellaneous: metal loth ionein (mt-1) 7.75 (6.4,9.1) 4.45 (4.5,4.4) 1.85 (1.8,1.9) 15 Miscellaneous: hist id ine-r ich calc ium b ind ing protein (HRC) -1.15 (-1.1,-1.2) -3.95 (-3.9,-4) 0 (-1.7,1.2) Table 1: Top differentially regulated in CVB3-infected mouse hearts using cDNA microarrays. As described in Figure 26, male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain) or PBS. At days 3, 9 and 30 post-infection heart tissue harvested, RNA was extracted and processed for hybridization to duplicate cDNA arrays (N=15 mice/time point; ratio of intensity values are an average of N=2 chips/time point; Incyte Pharmaceuticals). Based on expression and biological function, average signal intensity values (log2) for fifteen genes of interest (individual chip intensity values) shown in Figure 27 are listed. 75 primarily belong to groups 5 and 6, and immune-related gene clustered to groups 7 and 8 (Figure 28). Such findings further confirm consistent gene expression patterns between functionally-related genes. Small gene set analysis Initially, genes were grouped based on signalling pathways to understand which signalling pathways are transcriptionally regulated. However, the lack of signalling genes present on the array did not permit this analysis. Therefore, grouping was applied to dissect out transcriptionally activated discrete biological processes and genes were grouped into small (3-5 genes) functional sets. The top average intensities and standard deviations were calculated and shown in Table 2. Groups of genes with a large average differential expression and small standard deviation were considered consistently up-regulated and more likely to be involved in the disease process. As a 'proof of principle', the natriuretic factor value was log2 fold increased by 2.3 (0.5 standard deviation) and heat shock response average was log2 fold increased by 2 (0.5 standard deviation), both at day 9 pi. Other notable differentially regulated groups include up-regulation of ubiquitin/proteasome, collagen and extracellular matrix genes, and down-regulation of calcium transport genes. Concerted up- and down-regulation of discrete groups of genes affords greater confidence in data as compared to single gene differential expression. The cDNA array database was reduced in complexity using both function-based and expression cluster strategies to identify potentially important patterns of gene expression. A combination of analyses at the level of several hundred genes in broad functions or several genes in discrete functions has provided insights and potential targets for further investigations. 76 3 day 9 day 30 day Hierarchical clustering Total no of records: 651 Data normalization: Z Score Clustering method: UPGMA (unweighted average) Similarity measure: Euclidean distance Ordering function: Average value Figure 28: Hierarchical clustering of differential gene expression in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A. Gene profdes in infected hearts (compared to sham controls) were submitted for normalization (Z score) and UPGMA (unweighted pair-group method using arithmetic averages) clustering using Euclidean distance for the similarity measure. Based on these parameters, genes cluster into eight unique groups according to expression patterns of signal intensity (red represents upregulation and green represents downregulation). 77 B ^ ^ - ^ S t d Dev Ave rage ^ " " " " ^ ^ LO Hl(+/-) LO Not participating Potential interest Hl(+/-) Set of interest Potential interest Group Group Member Genes Ave (days pi) St Dev (days pi) 3 9 30 3 9 30 Natriuretic Factors brain natriuretic factor natriuretic peptide receptor 2 natriuretic peptide precursor A 1.8 2.3 0.7 0.6 0.5 1.8 Interferon Response interferon-related dev regulator 1 IFN gamma-inducing factor binding IFN-inducible protein variant 10 3.9 3.0 1.1 1.5 1.4 0.6 Heat Shock Response heat shockprotein 27 heat shock protein, 86 kDal heat shockprotein 70-2 heat shock 70kD protein 8 2.9 2 -0.2 1.4 0.5 1.3 Cytokine Receptor TGF-b type II receptor TNF receptor 1 precursor TNF intracellular-interacting FGF receptor-1 2.2 2.3 1.8 0.9 0.3 0.5 Extracellular Matrix Extracellular matrix protein 2 dermatan sulfateproteoglycan-ll fibronectin 1 (Fn1) 1.4 3.7 1.7 1.3 2.4 0.3 Collagen alpha-1 type IV collagen collagen alpha 2 type V procollagen, type I, alpha 2 collagen type I receptor 0.1 2.6 0.6 1.6 0.9 1.7 Calcium Transport Ca++ transport, slow twitch 2 brain Ca+2-ATPase ryanodine receptor 2, cardiac -0.4 -2.0 -0.1 1.4 0.8 1.8 Ubiquitin/ Proteasome ubiquitin-like protein/rps30 deubiquitinating enzyme Ubp69 protease 28 subunit, beta 1.0 2.9 1.8 2.0 1.5 0.6 78 Table 2: Small functional gene group profiles in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A. Fold change values in infected heart samples (log2; compared to sham controls) were grouped based on function (3-5 genes), for which the average and standard deviation expression were calculated in Excel and groups of interest were identified according to (A). B. Gene sets of interest include natriuretic factors, interferon response, heat shock and calcium transport genes, which have high differential intensity averages (+/-) and low standard deviations. 79 3.3 GeneChip -Based Approach to Gene Profiling in Mouse Hearts Biological Question What are the differentially transcribed genes and gene groups in CVB3-infected hearts? Rationale We previously utilized differential mRNA display and cDNA microarrays to investigate differential transcription in CVB3-infected mouse hearts. Here, gene profding was performed using an Affymetrix platform to generate a data set for comparison with previous results and to extend the previous finding to include a greater number of genes during the viremic, inflammatory and reparative stages of CVB3 infection. MIAME Criteria: Experimental Design and Quality Control Experimental design: Forty adolescent A/J mice were randomly placed into two groups of 28 mice infected intraperitoneally with If/ pfu of Gauntt strain CVB3 (4 groups) or 12 mice sham-infected (PBS). At 3, 9 and 30 days pi heart tissues were flash frozen (N=4 mice/group), pooled R N A was extracted, processed and hybridized to Affymetrix GeneChip®s. Signal intensities were normalized in Rosetta Resolver® and analyzed. A unique reference locator (URL) to a supplemental database with gene identification and signal intensity values will be made publicly available shortly following publication of results. Array Design: Custom Mouse GeneChips® (Amgen) containing 25, 204 annotated genes were manufactured by Affymetrix (Santa Clara, CA) for Amgen Inc. (Thousand Oaks, CA; please refer to Affymetrix website [www.affymetrix.com] for a detailed explanation of GeneChip® synthesis). Samples: Mouse heart apical ventricular portions from each experimental group (4 mice, except day 30 virus sample from 2 mice) were pooled, flash-frozen in liquid nitrogen, and stored at -70°C. Heart tissue was homogenized using a rotor-stator homogenizer in lysis buffer and sample R N A was isolated using the QIAGEN (Valencia, CA) RNeasy® isolation kit as per the manufacturer's instructions for isolation of animal tissues total R N A (see 80 Materials and Methods for detailed a procedure). Total R N A concentration was measured using spectophotometric analysis (1 OD at 260nm equals 40ug R N A per mL) and concentration and purity were checked by measurement of absorbance at 260 and 280nm (A260/A280 was between 1.9-2.1). Synthesis of double-stranded cDNA from total RNA was performed using a T7-(dT)24 primer (Gibco B R L Life Technologies). Biotin-labeled cRNA was synthesized by in vitro transcription and labeled using an ENZO Bioarray HighYield R N A Transcript Labeling kit as per the manufacturer's instructions. The biotin-labeled cRNA product was fragmented and stored in -20°C until hybridization (see Materials and Methods section for a more detailed account). Measurements: The fluorescently-labeled cRNA fragments were hybridized to the oligonucleotide probes on the array for 30 min at45°C, after which the array was washed, the probes were laser-induced and the image was scanned with a GeneChip® Scanner 50. Normalization Controls: GeneChip® output fdes were visually inspected for artifacts and proper array grid alignment. Initial analyses of array data were performed using Microarray M A S software (version 5.1). Non-degraded sample cRNA yield 375' intensity ratios between 1.0 and 3.0 (as degradation preferentially occurs at the 5' end). The minimum accepted signal intensity value was 115 intensity units as recommended by Affymetrix (reliability of expression data is considered high at 500.0, medium at 150.0 and low at 50.0 intensity units). Each 25mer probe consists of a perfect match (PM) probe with a complimentary sequence to the target and a mismatch (MM) probe with a single mismatch at the 13 t h nucleotide. The probe set is positive if signal intensity for perfect match is greater than mismatch. Those genes with either a present or marginal call (based on intensity difference P M / M M and number of positive probe pairs) were included in the analysis and absent-call genes were discarded. Therefore, only those genes with a high number of positive probe pairs would be considered for analysis. The statistical difference threshold (intensity of P M / M M probes) was set at 1.5 and 40% probe positivity for the array. Background fluorescence was less than 100 units and the presence of hybridization controls lOOpM CreX (Bacteriophage PI ere recombinase gene), 25 p M BioD (E. coli) 81 gene dethiobiotin), 5 p M BioC (E coli gene) and 1.5 p M BioB (Affymetrix GeneChip1* Eukaryotic Hybridization Control Kit, 900299) in the hybridization cocktail were confirmed. Each gene was probed by 16-20 sets of P M / M M pairs and the difference between these pairs was averaged to give the average difference (Avg Diff; Affymetrix, Inc., Santa Clara, CA). Genes with negative Avg Diffs across all samples were removed. Signal intensities are reported as a logio fold change in experimental over baseline samples. Logio was chosen over the standard log2 to initially focus on large fold changes. The sham 30 day control samples were lost due to experimental error. Thus experimental intensity values were compared against baseline controls. However, at the time of writing this thesis, all samples were rerun and signal intensities for sham 3, 9, and 30 sham samples were unchanged greater than a logio change of 0.5. Statistical analysis Intensity values were exported to Rosetta Resolver® for further analysis. However, a cross-gene error model approach was used to obtain an estimate of the variability in signal intensity for the pooled R N A was sample (http://www.rosettabio.com/home.html.). According to this model, genes with similar expression levels from the same chip were "borrowed" to construct a pseudo sample as the basis to estimate the "global" or "local" errors for each individual gene. This algorithm is. based on two assumptions: (1) the measurement of the expression of a gene on one single chip can be used to estimate the true population mean expression of that gene and (2) genes with similar measurement values "share" the same variance within the treatment group (proprietary algorithm, Rosetta Resolver®) Contribution by the Author This author was responsible for experimental planning and communication between the iCAPTURE Centre, Vascular Biology, Amgen investigators and the University Industry Liaison Office, University of British Columbia. This author was also responsible for virus 82 propagation, storage and titre, animal husbandry, infection, monitoring, sacrifice and tissue harvesting. Extraction of RNA, processing and hybridization to Affymetrix GeneChip®s was performed at the Microarray Facility, Amgen Inc. Computational analyses including data normalization, clustering and filtering were performed using Rosetta Resolver® by this author and Dr. Tracy Deisher, Vascular Biology, Amgen Inc. Literature research surrounding the function of genes and their previously published disease involvement was done by this author. Results To confirm genes of interest in the cDNA database and to extend our analysis to a larger number of genes, an in vivo experiment was performed using Affymetrix GeneChip®s (Figure 26 B). Genes were first filtered according to expression intensity. There were 548 genes differentially increased >logi0 fold change of 0.48 ( p « 0 . 0 0 1 ) at any one time point. Those expressed greater than logio fold change of 1.0 (230 genes) in at least one time point ( p « 0 . 0 0 1 ; 0.91% of all genes) were plotted as a hierarchical cluster (Figure 29) and in a temporal manner (Figure 30). Sham gene expression variability at 3 and 9 days, ranged from logio fold change of 0.0-0.5 above and below the baseline values. Consistent with the cDNA array database, differential gene expression was most prevalent at day 9. Of the genes markedly increased at 3 and 9 days pi, the most interesting based on degree of differential signal intensity and known biological function have been catalogued (Table 3). GenBank accession numbers for genes associated with infectious disease are colored blue (18 or 26) and heart disease (8 of 26) are colored red, 3 of 26 genes are involved in both infectious and heart diseases, and three genes have not previously been studied in either disease setting. Similarly, of the down-regulated genes, 6 of 26 were infectious disease-related, 5 of 26 were heart disease-related, and 17 genes were novel (Table 4). 83 -1.0 0 Log(Ratio) Thumbnai l V iew 1.0 I I OiU)U><<< _T IT IT ZJ' =V _i' a> a> A) c c c 3 3 3 <w w <o «<<•< < < < W ( O W W ( O W o g to EP-Slc4a1 [tr-beta-1-globin alpha-globin Hba-a1 —Amy2 j-Gsta4 LSir13 rCes3 1rTcea3 Tnntl Vtn Hcn2 I r-Ces3 Lr-Ckmm LrDbp ^-Dbp — C d k n t b i—lgk-V28 1 — I g M r C i s h l Casp11 Casp11 Mthfd2 Ms4a6b Msrl rritgb2 f-MmpS f-Gp49a Ccr5 tSerpina3g MIC-1 Ptprc j-Ccl3 | LGp49b FcgM tMmp3 Gp49a Piral j -Basp1 f-FLAP LrClecsfB l|LPira1 rSlc11a1 VCXCI5 Ccr5 ||rSerpina3g wm>i i l 1- Ptprc rCcl3 f-Gp49b J—Fcgrl 1rMmp3 f-Gp49a LPira1 j-Basp1 LFLAP rClecsfB L Piral rSlc11a1 VCxcl5 LTnc j-Psat-pending U/-CCI3 L Ly6a r C t s s &-Cd53 L Plek -f—atrial natriuretic factor LrLtbp2 Mtgb2 j—Tnc Tj—Cxcl5 ^—Chi3l3 j -Ccl6 mm B^-Chi3l3 1—Ccr1 |—Col3a1 0sf2-pending ' — E n d J-Hp LAldh1a2 -Lox j-Gzmb 111 rn |f—Sprrl a Lox j-onzin "LrlL-1rn ' Hrg1 j-Timp1 Lf-Timpl l _Fcgr1 rlfi205 rfGbpl L Gbp1 rSerpina3n fGbp2 Llfi47 -Cxcl9 Higp-pending j H g t p 0 rlfl204 Figure 29: GeneChip expression profiles in CVB3-infected murine hearts. The experimental protocol is illustrated and described in Figure 26 B. A l l genes (25, 204) were hierarchically clustered according to their fold change (logio; virus or sham/baseline, green represents upregulation, red represents downregulation). 84 "-Ptx3 -Cxcnrj -Cyp1b1 -THBS1 -CKCHO rPsmb8 LGtpi-pending -H2-L -MHC Q2-k -Tgtp "Tgtp -ligp-pending -C2 -Slfnl -CisM -MIP-1b j-Trim30 h§ka L-H2-BT (Llfi202b |Jrlfi204 T-CCI12 LrCish3 ^Ppicap — M s 4 a 4 c j -Ppicap L O a s l 2 f-Vig1 -pending - l f-Cdkn1a Hrf7 -ma -Tyki -Stat2 -Cxcl9 LrStatl ^-Statl t-Apod r[1-|fi202a f l rGbp3 WO) (/)<<< ZT ZT ZT =;• ^- =;• Q) Q) W C C C 3 3 3 tit » co Q.Q. Q. 0.0. a. < < < • < < ' < W ( O W w t o u w c o w < < < 3" 3" 3" — ^ ~ 0> 0) 01 c C C 3 3 3 co co co o o o Q. Q- Q. W O y W ( O W I LPdcd1lg1 1—Isg15 j -Ccl2 n-cci7 I—Slfn4 rlfit2 jHfit l f-Saa3 LrSM3 Hrt7 I r-mti LhCcIS L Usp18 Ccl5 J-Lgals3 r-r— Lgals3 H Spp1 1 Spp1 r~Mx2 Tor3a — H 2 - A a r-H2-Eb1 rP—lQh-4 " — I g kappa chain — I g a i-lgh-4 H g kappa-chain j-lgA V-D-J-heavy chain f -B cell antigen receptor Vlgk-V28 Hoi IC2a-21432_at Mmp12 lgk-V28 Mmp12 H2-K2 Figure 29: Continued 85 Sham (Days pi) CVB3 (Days pi) -2.0 Figure 30: Temporal display of gene expression in CVB3-infected mouse hearts. The experimental protocol is illustrated and described in Figure 26 B. Data was analysed by Resolver and only genes (230) differentially regulated >1.0 fold (logio scale; p « 0 . 0 0 1 ) over baseline control hearts are shown. Both control and experimental groups share common baseline transcriptional values. 86 Name Description GenBank Day 3 Day 9 Day 30 Saa3 serum amyloid A (SAA) 3 protein X03505 2 2 0.06 Slfn4 schlafen4 (Slfn4) gene AF099977 1.35 2 0.29 Usp18 ubiquitin specific protease 18 AW 047653 2 1.88 0.83 Ccl2 PDGF-inducible protein (JE) gene M19681 1.37 1.82 0.06 Lgals3 lectin, galactose binding, soluble 3 X16834 0.53 1.79 0.85 onzin onzin AA790307 1.03 1.64 0.49 Tlmpl tissue inhibitor of metalloproteinase 1 V00755 0.38 1.61 0.3 Apod apolipoprotein D X82648 1.55 1.57 0.55 Adam8 disintegrin and metalloprotease dom 8 X13335 0.24 1.57 0.11 Ccl5 chemokine (C-C motif) ligand 5 AF065947 2 1.56 0.93 Ifi47 interferon gamma ind pro 47 kDa M63630 1.27 1.53 0.8 | Serpina3n serine proteinase inhibitor, clade A M64086 1.07 1.51 0.8 Gzmb granzyme B M12302 0.67 1.51 0.31 A1b interferon inducible protein 1 U19119 1.34 1.49 0.96 Cyp1b1 cytochrome P450, 1b1 X78445 0.84 1.47 0.31 Cxcl9 chemokine (C-X-C motif) ligand 9 M34815 1.3 1.37 0.78 ligp-pending interferon-inducible GTPase AJ007971 1.28 1.31 1 Ccl6 chemokine (C-C motif) ligand 6 M58004 0.41 1.29 0.59 Ccl8 chemokine (C-C motif) ligand 8 AB023418 0.12 1.28 1.18 Vcaml vascular cell adhesion molecule 1 U12884 1.11 1.26 0.52 Itgb2 integrin beta 2 M31039 0.36 1.26 0.67 Col3a1 procollagen, type III, alpha 1 AA655199 -0.22 1.25 0.34 Il7r interleukin 7 receptor M29697 0.06 1.24 0.55 0sf2-pending osteoblast specific factor 2 D13664 -0.2 1.24 0.85 Serpina3g serine proteinase inhibitor M64085 0.62 1.18 0.62 Csf2rb2 colony stimulating factor 2 receptor M29855 0.49 1.18 0.49 Blue, genes involved infection; Red, genes involved in heart disease Table 3: Top upregulated genes at 3 and 9 day hearts post infection. The experimental protocol is illustrated and described in Figure 26 B. The top upregulated genes at days 3 and 9 and their fold change (logio, compared to baseline controls) are listed. Then, based on a search of the published literature, genes known to be involved in heart disease or infectious diseases were labeled red or blue, respectively (boxes denote involvement in both processes). 87 Name Descr ipt ion GenBank Day 3 Day 9 Day 30 Tcea3 transcription elongation factor A (Sll), 3 AM 32239 -0.14 -1.48 -0.56 Hbb-b2 Mouse gene for beta-1-globin. V00722 -0.95 -1.35 -0.54 Vtn vitronectin M77123 -0.17 -1.14 -0.81 Gsta4 glutathione S-transferase, alpha 4 L06047 -0.06 -1.12 -0.52 RD MHC class III region RD gene AF109906 -0.59 -0.92 -0.85 Spr sepiapterin reductase Al 530375 -0.16 -0.91 -0.35 Ctse cathepsin E gene AJ009840 -0.81 -0.88 -0.85 Ptprb protein tyrosine phosphatase, rec type B X58289 -0.34 -0.85 -0.47 Mapt microtubule-associated protein tau M18775 -0.21 -0.83 -0.34 Cideb cell death-inducing DNA frag factor AF041377 -0.43 -0.82 -0.38 Z r f p l zinc ring finger protein 1 AA763950 -0.3 -0.71 -0.6 Dbt dihydrolipoamide chain transacylase E2 L42996 0.09 -0.67 -0.23 Lrba LPS-responsive beige-like anchor AW 122230 -0.28 -0.66 -0.27 Cxcl12 chemokine (C-X-C motif) ligand 12 L12030 -0.47 -0.61 -0.18 S!c29a2 solute carrier family 29 X86682 -0.31 -0.6 -0.62 Srprb signal recognition particle receptor U17343 -0.27 -0.58 -0.35 Fxyd2 FXYD dom-containing ion transport reg 2 X70060 -0.29 -0.56 -0.5 Sgcd sarcoglycan, delta AB024923 -0.36 -0.55 -0.43 M e e d methylcrotonoyl-coenzyme A carboxylasel AW 123316 0.02 -0.54 -0.39 Mapkapk5 MAP kinase-activated protein kinase 5 AF039840 -0.14 -0.51 -0.47 Tie1 tyrosine kinase receptor 1 X80764 -0.18 -0.45 -0.03 Thra thyroid hormone receptor alpha U09504 0.4 -0.44 0.53 Zfp1 zinc finger protein 1 X16493 -0.22 -0.38 -0.54 Pdpk l 3-phosphoinositide dep protein kinase-1 AF079535 -0.29 -0.34 -0.59 igj immunoglobulin joining chain M90766 -0.29 -0.31 0.8 Csnk2a1 casein kinase II, alpha 1 AM 94248 -0.05 -0.29 -0.68 Blue, genes involved infection; Red, genes involved in heart disease Table 4: Top downregulated genes at 3 and 9 day hearts post infection. The experimental protocol is illustrated and described in Figure 26 B. The top downregulated genes and their fold change (logio, compared to baseline controls) at days 3 and 9 are listed. Then, based on a search of the published literature, genes known to be involved in heart disease or infectious diseases are labeled red or blue, respectively (boxes denote involvement in both processes). 88 3.4 A Comparative Gene Profiling Approach in CVB3-Infected Hearts Biological Question What are consistent differential transcription events in CVB3-infected hearts using cDNA arrays, GeneChip®s, semi-quantitative reverse transcriptase-polymerization chain reaction (RT-PCR), and immunohistochemical protein expression approaches? Rationale Two microarray-based gene datasets were compared to find gene consistent expression patterns. Consistent transcriptional events between two array-based datasets were identified for genes of potential biological interest. For genes of interest, expression was confirmed using RT-PCR and protein expression was demonstrated using immunohistochemistry. Experimental Design Data from cDNA and GeneChip® experiments were directly compared for consistent and inconsistent gene expression events. Genes of interest based on known biological function were confirmed using semi-quantitative RT-PCR. A complete description of array design, samples, measurements and normalization controls is presented in Sections 3.2 and 3.3. Contribution by the Author This author was responsible for all comparative analyses between cDNA and Affymetrix GeneChip®-based datasets, extensive gene function literature searches and generation of hypotheses. This author also performed tissue harvest, RNA extraction, primer design and RT-PCR. Tissue processing, immunohistochemistry and image acquisition were performed by this author with the exception of paraffin-embedding and sectioning steps which were performed by the Clinical Histology Laboratory (St. Paul's Hospital). 89 Results In the following sections, the expression of genes and functional gene groups of interest were compared from both cDNA and GeneChip® databases. Expression events are discussed in the context of (i) proof-of-principle confirmation of well-characterized events in myocardial inflammatory or infectious processes, (ii) gene identification of previously known processes, (iii) unexpected or negative events, or (iv) novel emerging hypotheses. L Proof-of-principle The microarray data was validated by confirmation of well-characterized transcriptional events in this disease process. Increased expression of natriuretic factors occurs following myocardial injury and IFN and chemokine response genes are increased in infectious diseases. The p-values generated from the cross-gene error model for GeneChip® data should be considered statistical evidence for transcriptional events of interest. Increases in natriuretic peptides at the transcript and protein levels are well-studied in cardiac disorders including myocarditis and are utilized as clinical markers for heart injury [175, 176]. GeneChip® and cDNA array data were internally consistent in BNP gene with a logio fold change of 0.51 (p«0 .001) at 9 days pi and in return to baseline by 30 days pi (Figure 31). Similarly, ANP gene expression was up-regulated at a login fold change of 1.01 (p«0) at day 9 and at a log, 0 fold change of 0.75 (p«0 .001) at day 30. No significant changes in CNP levels (p<0.96) were seen in the GeneChip® data (CNP was not included on the cDNA array). Thus, there was internal consistency between both array datasets and from published accounts of natriuretic peptide activity. Signal intensities for interferon and chemokine genes were markedly increased by up to a logio fold change of 2 at 3 and 9 days pi compared with baseline hearts. The induction of these responses is consistent with published findings [177, 178] (Figure 32). Interferons consist of type 1 and 2 responses [179, 180], which act via Jak/Stat [181, 182] to up regulate an IFN-inducible host-protective response. Chemokines have many functions such as recruitment and activation of immune-related cell types into sites of inflammation. 90 Figure 31: Natiuretic peptide gene profde in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. Brain natiuretic peptide (BNP) and atrial natiuretic peptide (ANP), but not c-type natiuretic peptide (CNP), are upregulated following CVB3-infection in (A) GeneChip®s (compared to baseline controls; as shown with all genes [blue] from Figure 30) and (B) cDNA arrays (compared to sham controls). There is also a late increase in ANP-C (Atrial natiuretic peptide clearance receptor precursor) but not ANP-A (atrial natiuretic peptide receptor-A precursor). 91 CVB3 (Days pi) IGTP IGTP-Pend IGTP-Pend IGIP47 Ifitl IP30 IP30 Isg15 IFNgr2 Isg15 CCL5 CCL5 CXCL1 CXCL1 CXC 3,4,5 CXCL 12,6,1 CXCL1 -2.0 Figure 32: Chemokine and interferon gene profde in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 B. To validate the GeneChip® array transcriptional results, global increases in expression of genes related to (A) interferon (compared to baseline controls; as shown with all genes [blue] from Figure 30) and (B) chemokine responses are shown in CVB3-infected mouse hearts (compared to baseline controls). 92 ii. Confirmation of known processes The following genes are involved in biological processes which have been shown to be important in enterovirus infection. Here, genes related to the ubiqitin/proteasome pathway, extracellular matrix and cell cycle will be discussed. We have previously shown increased ubiquitin-dependent proteolysis of key regulatory proteins such as cyclin D l in CVB3-infected HeLa cells [43]. I showed logio fold increases of 1.97 and 1.41 in expression of ubiquitin-specific protease 18 (USP18) at day 3 and 9 pi, respectively (Figure 33). By cDNA array, there was a log2 fold increase of 4.6 in signal intensity of psme2 at day 9, an IFN-gamma-inducible PA28 activator complex to the 20S proteasome; log 2 fold increase of 2.2 for ubiquitin-like protease 3 by day 9; and log2 fold increase of 2.5 for ubiquitin-specific processing protease 69 (Ubp69) at day 30. This data suggests involvement of several arms of the ubiquitin proteasome pathway in CVB3 infection. Collagen, specifically types I and III, are the most abundant extracellular matrix (ECM) components on the heart. The accumulation of collagen is required for cardiac repair but excessive fibrosis can lead to maladaptive remodeling and contractile dysfunction. Collagen genes are up-regulated including a logio fold increase of 1.25 for thin fibrillar and a logio fold increase of 0.84 for thick fibrillar collagen in GeneChip®, and comparable increases in cDNA arrays, both in mouse hearts at day 9 pi (Figure 34). Histologically, Masson's trichrome stain revealed collagen deposition by day 9 (Figure 35). By day 30, fibroblast proliferation and prominent areas of collagenous fibrotic replacement tissue were evident (Figure 36). Since superfluous collagen deposition affects heart function, modulation of this reparative response may lead to improved longterm survival. We have previously shown that CVB3 infection of HeLa cells in vitro can lead to cell cycle arrest [43]. I show logio fold increases of 1.37 and 1.30 for cyclin dependent kinase inhibitor 1A (CDKN1A; p21), a critical downstream mediator of wild-type p53 and a cell cycle regulator at day 3 and 9 pi (Figure 37). By cDNA array, decreased expression of 93 cyclins B2 and G l was seen. Taken together, the increase in cell cycle inhibitors and decrease in cyclin gene expression may contribute to dysregulation of the cell cycle apparatus. Since latent CVB3 infection of host cells in arrest may provide a virus reservoir, cell cycle alterations in target cells may be critical for chronic infections [183]. iii An unexpected result Expression of C A R can be considered unexpected based on previous findings. Several groups have now shown increased transcript and protein expression of C A R in human dilated cardiomyopathy (DCM) [184], in experimental models of autoimmune myocarditis [185] and myocardial infarction [186]. No significant change in gene expression of the viral receptor C A R was seen by GeneChip® analysis (Figure 38). The lack of differential expression can also provide important molecular information. 94 Sham (Days pi) CVB3 (Days pi) Usp18 \ Usp18 Ubc m Figure 33: Ubiquitin/proteosome gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. CVB3-infected mouse hearts show acute upregulation of ubiqutin/proteosome-related genes by (A) GeneChip® (compared to baseline controls) and (B) cDNA arrays (compared to sham controls). 95 C V B 3 (Days pi) -2.0 B eg o CD c CO sz o o 5 4 3 2 1 0 -1 -2 -ColV / , -" C o l IV / I / \ I ^"•^V Col I 0 3 9 30 d Figure 34: Collagen gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. CVB3-infected mouse hearts show a concerted increase in expression of several collagen subtype genes including a logio increase of 1.25 fold increase for thin fibrillar and a logio increase of 0.84 for thick fibrillar collagen by (A) GeneChip® (compared to baseline controls). These genes are also increased by (B) cDNA arrays (compared to sham controls). 96 H&E Masson's t r ichrome Figure 35: Cell death, inflammation and early fibrosis during the inflammatory stage of myocarditis. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At day 9 post-infection, mice were sacrificed and hearts were harvested and processed for histological assessment by H & E and Masson's trichrome stain. Inflammation (arrows) and tissue necrosis (arrow heads) are evident by H&E (left panel) and accompanied by collagenous deposition (light green component) by Masson's trichrome stain (right panel; 80X low and 320X high power images). 97 H&E Masson's t r ichrome Figure 36: Fibrosis and calcification during the reclamative stage of myocarditis. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At day 30 post-infection, mice were sacrificed and hearts were harvested and processed for histological assessment by H & E and Masson's trichrome stain. Fibroblasts proliferation as shown by the dark staining nuclei (arrows) and collagenous replacement fibrosis (light green component of Masson's trichrome stain) clearly demarcated from intact myocardium (arrowheads; 80X low and 320X high power images). 98 Sham (Days pi) CVB3 (Days pi) 2.0 1.0 CT O m CT c to sz O | -1 .0 -2.0 B cy CT O 0 -0.5 -1 • -15 CT ' - ° C ro •c -2 O 2 o -2.5 -3 Cdkn 1a Cdkn 1a Cdkn 1c Cdkn 1a Cdkn 1b 1 S i 1 1 Cyclin B2 \ / C y c l i n G1 30 d Figure 37: Cyclin gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. CVB3-infected mouse hearts show consistent increases in expression of cyclin-dependent kinase inhibitor genes by (A) GeneChip* (compared to baseline controls) and a decrease in cyclin genes by (B) cDNA arrays (compared to sham controls). 99 Sham (Days pi) CVB3 (Days pi) CAR Figure 38: C A R gene expression in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 B. CVB3-infected mouse hearts show no significant change in C A R gene expression using GeneChip® arrays (compared to baseline controls). 100 iv. Novel genes The following genes have known biological roles in cardiac injury, infectious diseases or both, but their contribution to myocarditis is unclear. I will limit this section to the most interesting findings based on the known functional role of the proteins encoded by these genes. The SI00 binding proteins regulate intracellular Ca2+-dependent signalling. The murine S100A subtypes 6, 10 and 11 binding protein transcript expressions peaked at day 9 with logio fold increases of 0.74, 0.53 and 0.55, respectively, in virus-infected hearts by GeneChip® analysis (Figure 39). S100A10 showed a log2 fold increase of 5.6 in intensity ratio at day 9 during CVB3-infection by cDNA array. Confirmation by RT-PCR showed, in a different set of mice, S100A6 (AF140232.1) gene increases at day 9 pi. Increases in SI00A gene expression during the inflammatory stage of infection in the heart may have several host-beneficial roles. Complement is an important arm of the innate immune response. The gene profile of complement pathway components revealed all members (present on the GeneChip® array) to be up-regulated. Complement component factor B (H2-Bf), an essential serine protease of the alternative pathway of complement activation, was logio fold increased 1.15 (p^O) at 3 and 1.30 (p=0) at 9 days pi (Figure 40). The only complement-related gene on the cDNA array was H2-Bf and expression of which was logio fold increased 7.2 at day 3 and 7.1 (p«0) at day 9. RT-PCR analysis showed a chronic increase of H2-Bf (NM 008198.1) expression at day 30 pi. Muscle L I M proteins (MLP) are L I M protein family members that promote myogenesis and have been associated with congestive heart failure. A peak in expression was shown at day 9 for genes encoding Four and a half L I M domain 1 (FHL1), which were logio fold increased 0.85 at day 3 and 0.50 (p«0 .001) day 9 pi in GeneChip® and increased log2 fold by 6.8 by cDNA arrays (Figure 41). By RT-PCR, FHL1 (AF134773.1) gene expression 101 A Sham (Days pi) CVB3 (Days pi) 2.0 of 1.0 o _ i CD ? 0.0 co sz o o -1.0 -2.0 B S100A11 S100A6 S100A6 S100A10 CN O ) o c CD SZ O o 30 d Sham Day 3 Day 9 Day 30 S100A10 p- Act in • • • • Figure 39: SI00 gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. GeneChip® (A) (compared to baseline controls) and cDNA array (B) (compared to sham controls) approaches to gene profiling show a strong expression peak of several S100A members at day 9 in CVB3-infected mouse hearts. C. S100A10 (NM031114.1) expression in vivo was confirmed using RT-PCR in a separate experiment (values are compared to P-actin; see Materials and Methods for primer sequences). 102 Sham (Days pi) CVB3 (Days pi) C3 C1qa,a,b H2-Bf pfc 30 d c H2-Bf Sham Day 3 Day 9 Day 3C H I P-Actin Figure 40: Complement gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. GeneChip® data shows global increases in complement-related genes and in particular, complement component factor B (H2-Bf) expression increases 1.15 and 1.3 fold changes (logio; A) (compared to baseline controls), cDNA array shows 7.4 and 7.1 fold changes (log2; B) at 3 and 9 days (compared to sham controls), respectively. Complement component factor B (NM 008198.1) expression was confirmed using RT-PCR in a separate experiment but showed a different pattern of expression at the 30 day time point (compared to P-actin). 103 2.0 1.0 o o> 0.0 O) c ro o IL -2.0 Sham (Days pi) C V B 3 (Days pi) Lim11 B CM o u> c ro JZ O o 0 3 9 30 d Sham Day 3 Day 9 Day 30 MLP | P M W i l l i •Hi P-Actin 21 E3EZZ3 Figure 41: Muscle LIM protein gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. A. CVB3-infected mouse hearts show an increase in LIM protein gene expression by GeneChip® (logio, compared to baseline controls) and (B) 6.8 fold (logi) by cDNA array (compared to sham controls). C. LIM gene expression (AF134773.1) was confirmed using RT-PCR in a separate experiment. There is increased in expression at all time points (compared to p-actin). 104 peaked at 9 days pi and was maintained at 30 days pi, a pattern of expression consistent with the array-based data. The importance of dystrophin cleavage has been highlighted in viral myocarditis but alterations in other cytoskeletal proteins have not been studied. The lysosomal cellular compartment includes cysteine-peptidases such as the cathepsin family of proteins. Cathepsins also have extra-lysosomal functions involved in physiological and pathological processes such as M H C class II-mediated antigen presentation. Cathepsin S, K, Z, C, P and L were transcriptionally up-regulated at day 9 and 30 pi (by GeneChip®). Cathepsin L (CTSL) is the most potent lysosomal cysteine and aspartic protease and was up-regulated at all time points and peaked at a log2 fold increase of 6.1 at day 9 by cDNA array analysis (Figure 42). RT-PCR confirmation further showed up-regulation of Cathepsin L (NM013156.1) following infection. Cathepsin L protein was strongly expressed in ventricular cardiac myocytes adjacent to myocarditic lesions by immunohistochemistry 9 days pi (Figure 43). The peripheral-type benzodiazepine receptor (PBR) interacts with the voltage-dependent anion channel and regulates permeability on the outer mitochondrial membrane. PBR was increased at all time points and peaked at a logio fold increase of 0.59 by GeneChip® data and a log2 fold increase of 0.59 by cDNA array, both at day 9 (Figure 44). By RT-PCR, there was also increased expression of PBR which was most prominent at day 3 and less so at day 9 and 30 pi following CVB3 infection. Protein expression was also increased at all stages of myocarditis. At 3 days pi, PBR positivity was present in few endothelial cells and myocytes (Figure 45). Intense myocyte-localized PBR positivity was seen at day 9 directly adjacent to necrotic areas. At day 30, clusters of myocytes adjacent to fibrotic tissue remained positive. Considering the importance of mitochondria in CVB3-mediated cell death, PBR up-regulation may be an important homeostatic check in response to apoptosis signalling in infected host cells and cells next to necrotic and active inflammatory and fibrotic events. 105 Sham (Days pi) CVB3 (Days pi) B CM O) 8 o _ J CD 6 O) c ha 4 o T3 2 O U_ o -30 d Sham Day 3 Day 9 Day 30 Ctsl p-Actin • M M • n i Figure 42: Cathepsin gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. A. CVB3-infected mouse hearts show an increase in several cathepsin genes by GeneChip® (logio, compared to baseline controls) and (B) 6.1 fold (log2) increase in cathepsin L b y cDNA array (compared to sham controls). C. Cathepsin L (NM013156.1) was confirmed using RT-PCR in a separate experiment (compared to P-actin). 106 Sham CVB3 Figure 43: Cathepsin L protein expression in CVB3-infected hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At day 3, 9 and 30 post-infection, mice were sacrificed and hearts were harvested and processed for immunohistochemical staining with an anti-CTSL antibody (Santa Cruz). CVB3-infected hearts exhibit no positivity at day 3, strong inflammatory cell and myocyte -localized signal intensity at day 9, and occasional punctate signal at day 30 (X320, darkfield low power inset). Little to no background signal is observed in sham hearts (X80). 107 A Sham (Days pi) CVB3 (Days pi) 2.0 d> 1.0 o 0 j? 0.0 re sz O T3 O -1.0 -2.0 B o 03 U) e CO x: O o PBR 8 6 4 2 0 N=5 probes PBR v 9 30 d Sham Day 3 Day 9 Day 30 PBR 121 B-actin Figure 44: Peripheral-type benzodiazepine (PBR) receptor gene profde in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. A . CVB3-infected mouse hearts show an increase in PBR gene expression by GeneChip® (logio, compared to baseline controls) and (B) 5.2 fold increase (log2) by cDNA array (compared to sham controls). C. PBR gene expression (NM012515.1) was confirmed using RT-PCR in a separate experiment (compared to P-actin). 108 Sham CVB3 Figure 45: Peripheral-type benzodiazepine (PBR) receptor protein expression in CVB3-infected hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At day 3, 9 and 30 post-infection, mice were sacrificed and hearts were harvested and processed for immunohistochemical staining with an anti-PBR antibody (Trevigen). CVB3-infected hearts exhibit occasional PBR signal at day 3, strong myocyte-localized signal intensity at day 9, and punctate signal at day 30 (X320, darkfield low power inset). Weak background signal is observed in sham hearts (X80). 109 Heat shock proteins (HSPs) are molecular chaperones whose increased expression has been shown to be a protective response in ischemia reperfusion injury. Several HSPs were differentially-regulated in myocarditis including a log2 fold increase of 3.5 in HSP27 at day 3 by cDNA array. Logio fold increases of 0.68 and 0.64 were seen for HSP70 at 3 and 9 days pi. HSP 86 intensity was increased acutely and HSP27 chronically at a logio fold increase of 0.70 (p~0) by GeneChip® (Figure 46). Early up-regulation of HSP27 was confirmed using RT-PCR and immunohistochemistry. Although gene expression of HSP27 was markedly increased at days 3 and 9, there was little increase in the protein level at day 3 as compared to day 9 (Figure 47). At day 3, even in or near myocytes with active virus replication (by ISH for CVB3 + strand RNA) HSP27 protein is not expressed. There is strong and abundant HSP27 expression in myocyte clusters adjacent to necrotic foci at day 9. At day 30, there were few positively stained myocytes (10-20 per short axis section) next to areas of fibrosis. I show here that HSP27 and likely several other heat shock protein genes are expressed in CVB3-infected hearts. Distinct expression patterns for cytoskeletal gene subsets including P-actin, a-actin and a-actinin, and cardiac actin and actinin were found. P-actin was used as a transcriptional housekeeping control gene for RT-PCR in the mouse heart. Several probes for the P-actin gene ( M l 2481; GeneChip®) showed no significant change in transcription from 3 to 9 days pi (Figure 48). Interestingly, there was marked increases in expression of a-actin as well as Z-band genes such as a-actinin and M L P as previously discussed. Gene expression increased to logio fold of 1.5 for skeletal and 0.9 fold for smooth muscle a-actin, 1.14 fold for a-actinin, and 1.52 fold for y-actin at day 30 by GeneChip®s (Figure 48). Cardiac actin and actinin expression peaked to a logio fold increases of 2.0 and 1.56 at day 9 in GeneChip®s. 110 2.0 ~ | 10 o a> 0.0 c ra 1-1.0 o -2.0 Sham (Days pi) CVB3 (Days pi) 9 3 30 HSP25 HSP70 B • HSP27 • HSP86 • HSPa8 • HSP47 • HSP60 • HSPel Sham Day 3 Day 9 Day 30 HSP27 , — p-actin Figure 46: Heat shock protein (HSP) gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 A and B. Several HSP genes are found to be up regulated in CVB3-infected HeLa cells using (A) GeneChip® arrays (logio, compared to baseline controls) and (B) HSP27 by cDNA arrays (compared to sham controls). C. HSP27 (NM 012620.1) expression was confirmed using RT-PCR in a separate experiment (compared to P-actin). Sham ISH(CVB3+) CVB3 Figure 47: Heat shock protein 27 (HSP27) protein expression in CVB3-infected hearts. Male 4-5 week old A/J mice were intraperitoneally infected with 105 plaque forming units of CVB3 (Gauntt strain). At day 3, 9 and 30 post-infection, mice were sacrificed and hearts were harvested and processed for immunohistochemical staining using an anti-HSP27 antibody (Santa Cruz). No HSP27 positivity was seen at day 3, even near ISH-confirmed virus replication. There are areas of strong myocyte positivity at day 9, and the occasional positive myocyte at day 30 in CVB3-infected hearts (X320, darkfield low power inset). No signal is observed in sham hearts (X80). 112 CVB3 (Days pi) B o O) c ro sz O o 2.0 1.0 0.0 -1.0 -2.0 30 CVB3 (Days pi) 9 CVB3 (Days pi) o cu O) c ro sz O •u o 2.0 1.0 0.0 -1.0 -2.0 p-actin 30 (/.-cardiac and skeletal actin, actinin 30 — ' cc-smooth muscle actin, actinin Figure 48: Myofibrillar gene profile in CVB3-infected hearts. The experimental protocol is illustrated and described in Figure 26 B. A . There is no change in P-actin genes with and without CVB3 infection. There are acute increases in cardiac-specific actin and actinin (B) and chronic increases in smooth muscle-specific actin and actinin (C) (logio, compared to baseline controls). 113 3.5 Discussion Characterization of the mouse model of viral myocarditis The mouse model has many positive attributes as a surrogate for human heart disease and is commonly used to study viral myocarditis [166, 187]. The genetics of several inbred mice are well characterized and large numbers of experimental animals can be used to obtain ample tissue and replicates for statistical power. Comparative sequence analysis between the draft sequence of the mouse genome and humans genome revealed roughly 85% similarity in the coding sequences [188-190]. This suggests the possibility of comparative genomic studies between human and murine responses to CVB3 infection. Orthology between specific mouse and human sequences can be searched in the Mouse Genome Informatics database (The Jackson Laboratory, Bar Harbor, Maine, http ://www. informatics.j ax. org). The A/J mouse-CVB3 infection model exhibits three distinct stages of viral myocarditis: viremic injury, immune infiltration and reclamation. The first phase characterized by virus infection and direct damage of target myocardial cells. The second phase characterized by mobilization of innate and adaptive defenses including mononuclear inflammatory cell infiltration of target organs. The third phase is characterized by tissue healing with various degrees of fibrosis and cardiac dilation. Data shown here verifies that enterovirus infection in mice alters heart function. Chronically, where large portions of the myocardium are replaced by fibrosis, ejection fraction is reduction by 19.9%. Myocardial enlargement and dilation (by histological inspection) and consistency in ventricular free wall thickness suggest that the myocardium was hypertrophied. We noted a trend in increase in ejection fraction during peak inflammation. This may be attributed to the hypermetabolic state during acute infection but is surprising considering elevated levels of NO during myocarditis, which is negatively inotropic [191]. These findings suggest that this animal model may be used as a surrogate for infectious dilated cardiomyopathy studies. 114 There are, however, important differences between mice and humans related to coxsackieviral pathogenesis. For instance, A/J mice have epicarditis regardless of infection, CVB3 binds human but not rodent D A F for cell entry, and there are differences in rodent and human gene variants. These and yet unidentified factors may have an influence on viral pathogenesis. cDNA-based bioinformatical analysis of mouse heart genes Genomic studies of infectious diseases have already catalogued transcriptional and proteomic events triggered by a wide range of agents including human immunodeficiency virus [192], cytomegalovirus [193], herpes simplex virus-1 [194] and hepatitis C virus [195]. Such studies are primarily focused on one fundamental issue in infectious diseases: how to comprehensively and integratively grasp the interactions between microbial pathogens and host genomics. In my pursuit of such studies, I focused on coxsackievirus B3 (CVB3) pathogenesis in the setting of viral infection of the heart during injury, inflammation and repair. For these and other exploratory studies, there is no governing hypothesis, rather a biological question focused on the identification of differentially expressed genes following CVB3 infection. In this report, I re-interpreted an existing cDNA dataset to mine useful information. Considering the transitory nature of transcription and the distinct stages of myocarditis, the temporal manner in which this data was accumulated is a significant strength. In-house analysis methods such as a small gene set analysis for calculating intensity averages and standard deviations for small functional gene groups were developed. Hierarchal clustering, which grouped both functionally-related and seemingly unrelated genes into expression clusters, and function-based grouping to find trends in gene expression within functional groups provided much insight. In my experimental approach using GeneChip®s and cDNA arrays, RNA was pooled. There are two types of pooling which can be performed [196]. 'Complete pooling' refers 115 to an experiment where all samples from one group are pooled and hybridized onto one array. The alternative, 'sub-pooling' refers to an experiment where subsets of samples are chosen for pooling onto more than one chip. For our cDNA experiments, sub-pooling was performed as sample R N A from 14 mice were pooled onto two arrays. For the in vivo Affymetrix array experiment, complete pooling was performed as R N A from from four mouse hearts were pooled and hybridized onto a single GeneChip®. Since age-matched inbred mice exhibit a range of phenotypes as demonstrated by the variability in myocarditis grades (Figure 20), these studies cannot address mouse to mouse biological variability. However, the present datasets afford greater assurance of data validity than that generated from a single heart. In my analysis, genes that exhibit high copy numbers and large changes in signal intensity were a focus in this thesis. However, it is possible that genes which exhibit relatively low signal intensity (known as low copy number genes) and which have small transcriptional changes following virus challenge may have considerable biological importance. Such genes may be transcription factors, and upstream signalling kinases and phosphatases whose subtle but significant changes may result in dramatic phenotypic alterations. At the time of this cDNA array experiment, dye labeling variability was not appreciated. Infected heart samples were labeled with Cy3 and sham samples with Cy5. Since Cy3 (red) intensity is often lower than Cy5 (green) intensity, differential expression would likely underestimated transcriptional increases and overestimate decreases. This is purely speculation and would require experimental verification. Also, because these arrays were printed with single, not duplicate, spots, two separate arrays were used and mean signal intensity values were reported. If this experiment were repeated, heart samples would be labeled and dyes subsequently swapped, both sets of labeled samples would be hybridized to multiple cDNA arrays with duplicate spots, and submitted for appropriate normalization. Both expression- and function-based clustering showed that functionally related genes have similar expression patterns (Figure 27 and 28). Brown et al [147] similarly showed in 78 116 different experiments with Saccharomyces cerevisiae that gene clustering based on eight different physiological conditions was similar within functional groups. There were global shifts in transcription of distinct functional groups, many of which have previously not been characterized in viral myocarditis. Increases in expression of cell cycle, cytoskeletal, extracellular matrix, stress response and immunological gene were shown. Decreases in expression of contractile, metabolic and mitochondrial genes were noted. Such protracted transcriptional patterns were seen chronically which may not be surprising considering that Klingel et al [39] detected ongoing viral presence and counted chronic cellular inflammation in mouse hearts 30 days post-CVB3 infection. The rate of residual viral RNA clearance from the myocardium is significantly slower than clearance during acute infection, despite a prolonged inflammatory response [60]. These results were compared with previously published transcriptional profiles from dilated and hypertrophic cardiomyopathic patients [197, 198]. Array analyses of human D C M tissues revealed a striking down-regulation of transcripts involved in C a 2 + signalling and homeostasis such as inositol 1,4,5-triphosphate (ITP) receptor, ITP 3-kinase, sarcoendoplasmic reticulum Ca 2 +-ATPase 3, and Ca27calmodulin-dependent protein kinase kinase 2 [197] and up-regulation of sarcomeric and cytoskeletal genes, including a-myosin heavy chain (a-MyHC), cardiac troponin I and tropomyosin. Using small functional group analysis, I show down-regulation of calcium transport genes clusters (Table 2) and up-regulation of several cytoskeletal genes. Hypertrophic and dilated human hearts also show marked increases in ANF, the ubiquitin/proteasome pathway, calcium signalling and down-regulation of cyclins in D C M [197]. Hypertrophic hearts exhibit increased metabolic, ribosornal and heat shock genes and decrease of extracellular matrix (collagen III, elastin) and metabolic (acetoacetyl-coenzyme A thiolase, phosphoglucomutase 1 and thymidylate synthase) genes. Interestingly, there is clear commonality in up-regulation of genes involved in ubiquitin/proteasome, ribosornal and heat shock genes and down-regulation of metabolic genes shown here during myocarditis. 117 Observed decreases in metabolic transcripts are potentially important considering the central energy production role of mitochondria in contractile cells and contribution of mitochondria to apoptotic cell death. As discussed earlier, these experiments do not identify the cellular transcriptional source. However, mitochondria, constitute one third of total cardiomyocyte volume which strongly suggests that myocytes are a cell of origin [199]. Metabolic gene alterations which impaired redox state or oxidative damage in myocytes could stimultaneously accelerate cardiac injury [200]. In fact, abnormalities in cardiac mitochondrial respiratory enzymes have been linked to heart failure in pediatric dilated and hypertrophic cardiomyopathies [201-203]. One possible mechanism of mitochondrial injury may be a switch in chief energy utilization from fatty acid P-oxidation to glucose and lactate [204]. Human D C M hearts are metabolically abnormal with a down-regulation of enzymes in the mitochondrial oxidative phosphorylation of fatty acids [205]. On the other hand, mitochondrial transcript down-regulation in myocarditic hearts may be part of a host protective response against overload and oxidative damage from electron leak and the accumulation of reactive oxygen species [206, 207]. Impaired mitochondrial function could result in decreased availability of ATP, reduced transmembrane potential, contractile defects and apoptosis [208]. The molecular alterations in mitochondrial function may be a major determinant in the balance between life and death signalling (Figure 49). We have previously linked mitochondrial alterations to cytochrome c release and activation of downstream caspases in cultured cells infected with CVB3 [84, 85]. In addition to regulating metabolic homeostasis, mitochondrial genes may also regulate host cell death in viral injury. Down-regulation of mitochondrial genes in myocarditis may beget metabolic abnormalities and accelerate progression to heart failure. These genomic studies were limited by the genes present on the array which represents a portion of the entire mouse genome. Since the probe sequences originate from public EST and gene sequence databases, their selection are also biased in that they represent 118 Hypothesis: Decreased expression of metabolic transcripts impairs mitochondrial function and triggers reactive oxygen species (ROS) release which alters the balance between apoptotic and ERK signalling in CVB3-infected mouse hearts. DAF CAR Ras GTP I Raf mm Decreased Metabolic Transcripts t f \ Impaired oxidative phosphorylation MEK I Erk1/2 I ROS* Pro-survival Gene Expression Figure 49: The mitochondrial injury hypothesis. Based on gene profding investigations in C V B 3 infection of mouse hearts, we hypothesize that global decreased expression of metabolic transcripts impairs mitochondrial function which alters the balance between myocyte apoptotic and survival signalling. 119 previously-identified genes. Whether metabolic gene decreases would be consistent for unbiased or pan-genomic investigations are unknown. Unbiased transcriptional studies can be performed using serial analysis of gene expression (SAGE), differential mRNA display and comparative EST sequencing of cDNA libraries. For these high-throughput transcriptional approaches, a prior knowledge of the sequences to be analysed are not required, facilitating discovery of novel genes. Of these, SAGE is the most comprehensive and sensitive for complimentary analysis to microarray data sets [209]. Differential display and ESTs are also useful but lack the sensitivity and throughput of SAGE and microarrays. At the time of writing this thesis, there are two other published reports of microarray analysis in heart infection models. Garg et al [169] used arrays to show a deficiency of mitochondrial oxidative phosphorylation during Chagasic cardiomyopathy caused by infection with the parasitic protozoa Trypanosoma cruzi. Decreases in genes involved in mitochondrial oxidative phosphorylation complexes I (NADH-ubiquinone oxidoreductase) and IV (cytochrome c oxidase) are consistent with my findings. Peng and colleagues [210] examined the transcriptional profiles of CVB3-infected mouse hearts but limited their report to up-regulation of Bag-1, a Bcl-2 family protein, at day 7 post-infection. The lack of transcriptional information on any other genes prevents any comparisons between this study and my own. GeneChip®-Based Approach to Gene Profiling in Mouse Hearts This experiment was then replicated using an Affymetrix platform to compare the transcriptional profile in a different set of CVB3-infected mice. Interestingly, the majority of highly up-regulated genes had previously been associated with either heart disease or infectious disease. However, most down-regulated genes were previously unstudied in either setting. This highlights the bias in studying gene up-regulation as compared to down-regulation, a potentially important area of genomic study. 120 The most pronounced alterations in gene expression, both in intensity and in number of genes, occurs during peak inflammation. Two likely explanations for this observation are discussed: (1) Inflammatory cells have a significant cardiac presence at day 9 compared to day 3 and 30. As such cells are transcriptionally active and unique from resident cells, the arrays would detect a different transcriptional profile. (2) Considering the elegance of the CVB3 lifecycle, the virus may have adapted to cause a subtle cellular transcriptional response. Thus the transcriptional response of myocytes in response to inflammation may be more pronounced than during direct virus infection. A Comparative Gene Profiling Approach in CVB3-Infected Hearts Three different gene-level assays and one protein-level assay were used to investigate the transcriptional response of the myocardium, in different sets of mice, to enterovirus infection. A distinct advantage of this work is the ability to compare between two array databases [211]. As a 'proof-of-principle' for these assays, the transcriptional profiles of natriuretic factors, chemokine and IFN response genes were confirmed. These genes are known to be expressed during heart injury and infectious diseases settings. The heart undergoes fetal gene re-expression in response to stress, including natriuretic factors in response to stress from an increased hemodynamic load [212]. The discovery of atrial natriuretic peptide in 1981 by de Bold et al [213, 214] expanded the perception of the heart from simply a mechanical pump to that of a neuroendocrine organ. There are three main natriuretic peptides: atrial natriuretic peptide (ANP) and B-type/brain natriuretic peptide (BNP) are of myocardial cell origin, while C-type/cardiac natriuretic peptide (CNP) is of endothelial origin (the fourth, dendroaspsis natriuretic peptide (DNP) has been reported to be present in human plasma and atrial myocardium). In response to atrial and ventricular stretch, the heart releases ANP and BNP, respectively. Together, they induce vasorelaxation, inhibition of aldosterone secretion in the adrenal 121 cortex, and inhibition of renin secretion in the kidney. BNP is synthesized in the heart and released directly proportional to ventricular volume expansion and pressure overload. ANP is normally present within atrial myocardial cells and absent from ventricular myocardium, but can be detected in human and experimental models of cardiomyopathy [215]. Pressure overloaded ANP-/- mice have cardiac hypertrophy and increased expression of E C M genes, suggesting that ANP protects against cardiac remodelling by regulation of E C M components [216]. My results show increased expression of ANP and BNP but not CNP. The magnitude and pattern of expression between cDNA and GeneChip® results is similar. Confirmation of natriuretic factor mRNA expression lends credibility to identification of novel genes newly associated in these datasets. In reaction to a broad range of infectious diseases, including viral infection, IFNs are activated as part of an immediate early response as confirmed in the microarray datasets. In viral myocarditis, type I and possibly type II signalling is important for the prevention of early cell death due to CVB3 infection [123]. Using differential mRNA display, our laboratory has previously shown up-regulation of IFN-gamma-inducible guanosine triphosphatase (IGTPase) in CVB3-infected mouse hearts [126]. Zhang et al [125] further showed IGTPase over expression inhibits viral replication and delays CVB3-induced apoptosis. Here, activation of an IFN gene program in myocarditis, including several IGTPase subsets and IFN-inducible genes, was shown. Chemokines are cytokines that activate or chemoattract leukocytes. Chemokines such as MIP-1 [217], MIP-2 [218] and monocyte chemotactic protein-1 (MCP-1) [219] have already been identified to play critical roles in the recruitment of inflammatory cells in myocarditic lesions. Up-regulation of several chemokines including CCL5 chemokine RANTES and several C X C L chemokines, growth regulated proteins including growth-regulated oncogene (GRO) and macrophage inflammatory protein-2 (MIP-2; the murine counterpart of human IL-8) were shown. Up-regulation was most prominent during the inflammatory stage of myocarditis and again, supports the validity of this dataset. 122 To illustrate the consistency between the gene profiling techniques utilitized in this laboratory, I will highlight the P-globin gene. Our laboratory originally identified the mouse P-globin gene to be markedly downregulated using differential mRNA display in acute myocarditic hearts [126]. This gene was then shown here to be highly down-regulated by cDNA array and GeneChip®. Our laboratory previously assessed P-globin major protein levels to show a decrease in cardiac myoglobin concentration in CVB3-infected hearts. Thus, there is a consistency between differential display, cDNA array and Gene Chip® data. Interestingly, P-globin transcript decreases may relate to cardiac myoglobin deficiency and impinge on cardiac metabolism during viral myocarditis. I have also presented differential gene expression in the following areas which support our laboratory's previous findings: (1) ubiquitin/proteasome inhibition, (2) extracellular matrix metabolism and (3) cell cycle regulation. 1. Increased transcription of ubiquitin-specific protease 18 proteasome activators PA28-P genes may be triggered by damaged and misfolded intracellular proteins to facilitate M H C class I antigen presentation. Increased transcription of cyclin dependent kinase inhibitors and decreases in cyclins B2 and G l together likely disrupts cell cycle progression and may induce aberrant cyclin-triggered signalling. Cell cycle status influences CVB3 replication as virus replication is reduced in quiescent host cells (GO phase) but active in dividing cells [183]. Whether the ubiquitin proteasome pathway activation is linked to cell cycle disruption, antigen presentation or has a more direct effect on virus replication, such as in virus persistence, remains an intriguing possibility. 2. Post-myocarditic scar formation occurs in both humans and in experimental systems [1]. I show a peak in expression of collagen I and III, the predominant subtypes in the heart. The makeup and extent of collagenous replacement tissue influences the cardiac repair and dysfunction shown here in the mouse model (Figure 35 and 36). Histologically, collagen deposition is punctate at day 9 and when gene expression is high but drops off at day 30 when collagen scars have developed. 123 3. We have shown cell cycle inhibition in CVB3-infected HeLa cells and increased ubiquitination and proteasome degradation as the mechanism of reduction in key proteins such as cyclin D [43]. CVB3 infection disrupts host cell homeostasis by blocking the cell cycle at the boundary between the Gap 1/Synthesis phases. Cell cycle gene down-regulation may be another method that CVB3 disrupts host cellular homeostasis. The main goal of this thesis work is to develop novel and testable hypotheses for future investigations. To this point, one of the most exciting aspects of exploratory research is finding genes which are markedly differentially regulated with potentially-interesting biological functions but have not previously been associated with enterovirus infections or myocarditis. I will now discuss the novel genes found, the rationale behind their potential functional role in viral myocarditis, and the exciting questions that they tempt: Emerging Questions I: Does SlOO-binding protein alter calcium signalling in CVB3-infected hearts? The SI00 proteins are important regulators of transient intracellular Ca -dependent signalling mechanisms, first identified in the nervous system [220, 221]. They primarily act intracellularly by regulating effector proteins in a Ca2+-dependent manner [222], but may also act extracellularly [223]. S100A and S100B proteins are widely distributed, with highest levels in cardiomyocytes and glial cells, respectively [220]. Several S100 proteins have been circumstantially linked to roles in heart disease [224]. S100B is normally not expressed in the heart, but is induced after myocardial infarction and has a putatively cardioprotective [225, 226]. S100A1 is reduced in end-stage heart failure [227] and S100A1 null mice have dramatically reduced responses to hemodynamic stress possibly by reduced cardiac C a 2 + sensitivity [224]. Mechanisms of S100A1 activity include inhibition of phosphorylation of cytoskeletal proteins (glial fibrillary acidic protein and tau) and transcription factors (MyoD), regulation of signalling and transcription factors 124 [220, 228, 229], and stimulation of Ca 2 release [230]. Murine SlOO-related protein subtypes A6, 10 and 11 were up-regulated in enterovirus-infected myocardium particularly during the inflammatory stage. Expression of SI00 genes may help to restore calcium homeostasis [231]. These findings on SI00 mRNA expression provide a basis for more focused investigation in viral myocarditis. Emerging Questions II: Do alterations in muscle LIM proteins influences cardiac structure or function following enterovirus infection? Muscle L I M protein (MLP) is a member of the L I M domain protein family that binds a-actinin at the Z disk in striated muscle [232, 233]. By linking the P-spectrin network to myofibrillar actin filaments, M L P stabilizes the association of the contractile apparatus with the sarcolemma [234]. It also acts to regulate muscle-specific gene expression and is thus necessary and sufficient for cardiac hypertrophic growth and sarcomere organization [232, 235]. Stypmann et al [236] and Arber et al [237] have shown that cathepsin L-deficient mice develop DCM-like illness. A defect in the Z-disc MLP/T-cap complex, a key component of the cardiomyocyte stretch sensor machinery, can lead to D C M [238]. This is the first report to my knowledge that MLPs are increased chronically in enterovirus-infected hearts. If the expression is indeed of myocyte origin, it may be a host-adaptive response to viral injury of cytoskeletal apparatus. M L P up-regulation, despite global down-regulation of cytoskeletal genes found using cDNA arrays, suggests that this is a unique pattern. Considering the importance of M L P to the cytoskeleton and cardiac contraction, modulation of these genes may influence chronic heart function. Emerging Questions III: Does the complement pathway contribute to immune activation in enteroviral heart disease? Complement, is an integral arm of the innate immune system. It is central to the recruitment of the cellular immune system and enhances the production of antibodies, as well as directly destroying pathogens and pathogen-infected cells through formation of 125 lytic pores [239-241]. Alternative complement pathway activity is necessary for virus protein retention in splenic germinal centers and B-cell association in secondary follicle germinal centres [242]. But the role of complement in propagation of myocardial inflammation has not been addressed. Increased expression of several complement genes were found in CVB3-infected mouse hearts including Clqa,b, C2, C3, C4, H2-Bf and Pfc. These proteins participate in both classical and alternative complement pathways. Huang et al [197] showed increased CD59 expression in hearts of hypertrophic and dilated cardiomyopathic patients. I found that addition of CD59, a membrane attack complex of complement (MAC) inhibitor, during CVB3 infection reduced inflammation but increased virus replication and myocardial injury (Yanagawa et al, unpublished observations). Whether complement is anti-viral or propagates an injurious immune response needs to be fully elucidated. Emerging Questions IV: Do cathepsins have a protective role in enterovirus infection? Cathepsins contribute to the terminal degradation of proteins in the lysosome but have important extra-lysosomal functions as well. Cathepsins can degrade myofibrillar proteins leading to architectural degradation and cathepsin L is the most powerful lysosomal protease that degrades collagen and elastin [243]. Cathepsins are also involved in restricted antigen proteolysis for M H C II antigen presentation by macrophages and lymphocytes [244]. Thus observed up-regulation of cathepsin genes may be part of a host protective response to inhibit virus infection. Overall, the increase in cathepsin genes, and in particular cathepsin L, may be a host response to facilitate microbial presentation on the M H C II molecule and increased protein turnover during viral infection (Figure 50). Interestingly, a detailed characterization of cathepsin L distribution in rat tissues found that protein expression is four times higher in kidney than in liver, spleen, lungs and brain, and lowest in the heart, skeletal muscle, and gastrointestinal tract [245]. Total cathepsin L activity is higher in atria as compared to ventricles of hearts [246]. Together, these studies show high cathepsin expression in tissues which are also resistant to CVB3 infection. It is 126 Hypothesis: Increased expression of Cathepsin L is a host adaptive response to facilitate protein turnover, virus clearance and other putative roles in CVB3-infected mouse hearts. DAF Antigen processing Lysosome 1 MHC II presentation (Macrophages and potentially myocytes) h r CAR Increased J ^ * . Cathepsin L rfWtyfiOft Transcripts Cathepsin L Approx. 10% ECM • break Extracellular down alternative angiotensin-generating enzyme Figure 50: The cathepsin L anti-viral response hypothesis. Based on gene profding investigations in CVB3 infection of mouse hearts, we hypothesize that increased expression of Cathepsin L is a host adaptive response to facilitate protein turnover, virus clearance and other putative roles in CVB3-infected mouse hearts including involvement angiotensiogen processing and E C M metabolism. 127 tempting to speculate that these proteases may have a critical host protective role following virus infection and may influence viral tissue tropism. Emerging questions V: Are HSPs cardioprotective in enteroviral challenge? Heat shock proteins (HSPs) are ubiquitous molecules expressed in response to stress in all living organisms. Coxsackieviruses depression cellular R N A and protein synthesis and induce production of oxygen free radicals which are known to increase HSP expression. Coxsackievirus infection can induce HSP70 synthesis in cultured cardiac myocytes [247]. Cardioprotective heat shock responses have been shown in ischemia-reperfusion injury [248] and hearts overexpressing HSP70 and 27 have reduced infarct size and improved functional recovery [249, 250]. I show acute increases in early HSP27, 70 and 86 genes and chronic increases in HSP25. Interestingly, early HSP27 up-regulation was not accompanied by protein expression, even in degenerating myocytes. Virus-induced translational suppression may limit early protein expression regardless of target cell HSP 27 gene status. However, a necrotic and inflammatory or reparative milieu likely triggers the increases in both HSP27 transcript and protein levels. Increases in HSPs may limit injurious reactive oxygen species. Several heat shock proteins including 70, 27 and 60 can directly inhibit specific checkpoints in the apoptosis cascade (as summarized in [251]). Together, the roles of HSPs as molecular chaperones, in inhibition of reactive oxygen species, and direct blockade of apoptosis signals suggest that HSP gene expression is a protective response to enterovirus-induced injury. Emerging questions VI: What are the role of the peripheral-type benzodiazepine receptor in cardiomyocytes and leukocytes? First described in peripheral tissues, the 18kDa peripheral-type benzodiazepine receptor is abundant in mitochrondrial outer membranes (reviewed in [252]). Reported functions of PBR include steroid biosynthesis, cell growth and differentiation, calcium channel activity and anion transport. Here, CVB3-infected hearts showed increased peak of expression of 128 PBR during acute inflammation. PBR transcription may be induced in cardiac myocytes to preserve mitochondrial integrity for the following reasons: PBR protein expression was localized to myocytes; early increases suggest an endogenous cellular origin; the importance of mitochondria in cardiac myocytes and the association between PBR, the voltage-dependent anion channel and the adenine nucleotide carrier at the mitochondria [253]; and we have shown mitochondrial release of cyt c triggers CVB3-induced apoptosis in vitro (Carthy et al, Virology, in press). The sharp increase in PBR during inflammation suggests that it may have important roles in target cell apoptosis and immune-related viral clearance. Emerging questions VII: Do alterations in myofibrillar genes contribute to the decreased heart function? I showed the gene signatures for three groups of contractile proteins: a-actins, actinins and P-actins. P-actin is commonly used as a housekeeping gene. In the mouse model of myocarditis, no changes in the expression of genes encoding p-actin were seen, despite marked changes in both a actin and several actinin isoforms. Transcription of P-actin was also unchanged using RT-PCR. Thus, in these experiments, P-actin is a good control gene in the mouse model of viral myocarditis. Marked increase in cardiac actin and actinin genes were seen at day 9 pi. We also showed that M L P is increased significantly at this timepoint. Together, L I M and actinin are joined at the Z-line of striated muscle and distribute contractile force through the sarcomeric matrix [254]. As alluded to earlier, viral proteases readily cleave dystrophin which joins F-actin to the dystroglycan complex [25, 64, 255]. Acute increases in Z-line associated proteins may be a cellular response to maintain cytoskeletal integrity. Increases in oc-smooth muscle actin occurred chronically. Activated fibroblasts can differentiate into oc-smooth muscle-expression myofibroblasts, responsible for matrix deposition and scar formation. Thus, the major cellular source for the a-smooth muscle 129 actin transcripts are likely activated myofibroblasts involved in the fibrotic reparative process. The differential expression of myofibrillar genes may alter the contractile and matrix makeup of the heart and impinge on heart function. Caveats Much excitement exists surrounding novel high-throughput techniques and bioinformatical tools. But, as with any experimental technique, there are important caveats associated with these approaches. As these technologies are relatively new, I will provide an extensive discussion of caveats associated with (i) microarray technology, (ii) computational analyses and (iii) environmental factors as related to these studies. (i) Microarray Technology Microarray-based genomic studies are based on the premise that there is a correlation between, target-probe signal intensity gene expression and protein function. However, this relationship is likely highly non-liner. Post-transcriptional alternative splicing and post-translational modifications, such as proper folding, covalent phosphorylation and nitrosylation, cleavage/activation and cellular localization, may affect protein function. The coxsackievirus genome codes for viral proteases which selectively cleave host cell translational proteins, such as eukaryotic initiation 4-y and poly-(A)-binding protein [256], [257]. As such, relating gene transcription to protein expression and function in cells and tissues infected with CVB3 is particularly difficult. The IRES database (http://ifr31 w3.toulouse.inserm.fr/IRESdatabase) provides information on cellular and viral genes transcribed by a cap-independent mechanism. These genes are one focus of ongoing confirmatory studies. Validatory studies are therefore necessary to show protein expression and its contribution to pathogenesis. At best, the presence of a differentially expressed gene can only imply involvement, while the absence of gene alterations simply suggests inactivity [258]. There are technical aspects of microarray design that warrant discussion. Firstly, probe-target hybridization is affected by base GC content which varies from probe to probe and 130 influences the thermodynamic properties of hybridization. The resulting disparity between probe-target bindings makes signal intensity unrelatable to transcript number between genes. Currently, a 5-10% variation in signal intensities among replicate spots on the same cDNA arrays and 30-50% variation among corresponding elements on different arrays [259]. Affymetrix researchers typically experience -10% variability in intensity data when the same sample is hybridized to multiple oligonucleotide arrays (Dr. Mark Reimers, Center for Genomics and Bioinformatics, Karolinska Institute, unpublished observations). Such errors can be attributed to: (1) uneven fluid flow, (2) proximity of heating elements in the hybridization chamber, (3) presence of dust particles on the chip and (4) scratches on the chip. Such technical issues are important when interpreting raw microarray data. Custom-made cDNA array chips are further affected by greater variation in spot uniformity and dye label variation. In my in vivo profding studies, hearts were homogenized and RNA was isolated and processed for hybridization. Thus, no attempt was made to isolate myocardial cell types. Therefore, the cell type from which a signal originated cannot be determined. Although the majority of the myocardium is composed of myocytes by mass, endothelial cells, smooth muscle cells, fibroblasts and resident progenitor cells are present in a complex spatial arrangement. Following infection, infiltrating leukocytes, infected and bystander myocyte cells and activated myofibroblasts may also contribute to the transcriptional profile. Such heterogeneity could potentially dilute an important cell-specific or spatial-specific signal to below the detection limit. Thus, differential gene expression in cellular subtypes may be underestimated. Dissection of specific events/signalling pathways to cell types would require cellular isolation techniques, such as laser-based methods or traditional purification of cell populations using fluorescent markers. On this note, Kamme et al [260] successfully performed laser capture microdissection and microarray analysis on single neuron population subsets and Eberwine et al [261] measured transcript abundances in single cells themselves. 131 The day 30 sham samples were faulty and as a result, the transcriptional profiles of C V B 3 -infected animals were compared to baseline animals and not age-matched controls. At the time of thesis submission, these samples were replicated and no significant changes in gene expression in day 30 sham-infected hearts were found (preliminary unpublished observations). Computational Analyses As supported by the growing literature, the use of different bioinformatical tools can yield very different end results [262] and it remains to be seen which technique(s) will emerge as the most informative. The choice of clustering algorithm and criteria were optimized based on dataset and system properties and can significantly affect the resulting memberships. Hierarchical clustering with six different criteria (centroid, median, single, ward, complete, mcquitty) using five different clustering algorithms were performed (data not shown). I found by visual inspection that altering the criteria but not the clustering algorithm itself, significantly affected membership of genes. Quantitative algorithms to validate clustering methods are also available [263], but such analysis was not performed here. There is also much debate over the optimum method for data normalization and standardization of microarray experiments [144]. Environmental factors These genomic studies will eventually lend insights to why otherwise healthy adults are infected with a relatively common virus and develop pernicious inflammation of the heart. However, like most other diseases, both infectious and non-infectious, a genomic investigation is incomplete without considering environmental factors which affect disease pathogenesis [264]. During coxsackieviral infection, host nutritional status has been shown to not only alter susceptibility to infection, possibly through reduced monocyte chemo-attractant protein-1, but also directly affect virus genomic regulation [265, 266]. Mice deficient in selenium and infected with a non-virulent strain of CVB3 support a specific 6-132 nucleotide mutation to a virulence phenotype in the virus RNA. In rural areas of southwestern China, selenium deficiencies are associated with high incidence of endemic cardiomyopathy, termed Keshan disease, from which various serotypes of coxsackievirus have been isolated [267]. Thus, in studying genomic and eventually genetic aspects of myocarditis pathogenesis, environmental factors should also be considered. Summary Gene profiling studies reported here and elsewhere, have unraveled new and exciting transcriptional events of individual genes as well as groups of genes. The expression of these genes has led to insights to the transcriptional regulatory events and the involvement of downstream cellular processes. With diligent experimental planning, proper use of informatical analysis and intelligent choice of confirmatory tools, one can overcome or at least minimize many of the abovementioned caveats. This exploratory approach has already yielded significant emerging questions which have already catapulted the laboratory's efforts to understand enterovirus infections and inflammatory heart diseases. 133 CHAPTER IV Gene Profiling In Vitro 4.1 GeneChip® Array-Based Profding of CVB3-Infected HeLa cells Biological Question What are the differentially transcribed genes and gene groups in CVB3-infected HeLa cells? Rationale Many cell types, both endogenous and infiltrating, are involved in CVB3 infection of the heart. To dissect out early transcriptional events in host cells during CVB3 infection, I used an in vitro HeLa cell infection model. MIAME Criteria Experimental design: Human cervical carcinoma epithelial (HeLa) cells were serum starved overnight and either infected with C V B (multiplicity of infection=10) or sham-infected (PBS) for 45 min under serum-free conditions. Kandolf strain CVB3, originally identified as a Nancy strain, has been previously characterized, cloned and sequenced, and was used in the in vitro experiments [24]. This virus strain was utilized here based on pathogenicity in the HeLa cell model and to directly compare our results with concurrent molecular studies [64]. Following infection, CVB3-containing media was removed and fresh media with 10% fetal bovine serum was added for the remainder of the infectious process. At 30 minutes, 1, 3, 7 and 9 hours (h) post-infection (N=l), cells were harvested; RNA was isolated, processed and hybridized to GeneChip®s (U95A). A unique reference locator (URL) to a supplemental database of gene identification and signal intensities will be made publicly available following publication of results. Array Design: Human U95A GeneChips® (12, 627 annotated genes) which were manufactured using in situ photolithography were purchased from Affymetrix (Santa Clara, CA). 134 Samples: HeLa cells were collected using a cell scraper in 1 ml ice cold PBS and spun at 1000 rpm for 10 minutes. Supernatant was quickly removed and cell lysates were flash-frozen in liquid nitrogen, and stored at -80°C. RNA was isolated using the QIAGEN (Valencia, CA) RNeasy® isolation kit as per the manufacturer's instructions for isolation of cellular mRNA (see Materials and Methods for detailed procedures). Total RNA concentration was measured using spectophotometry analysis (1 OD at 260nm equals 40ug RNA per mL) and concentration and purity were checked by measurement of absorbance at 260 and 280nm. Synthesis of fragmented biotin-labeled cRNA product was performed as described previously (see Materials and Methods section for a more detailed account). Measurements: Fluorescently-labeled cRNA fragments were hybridized to the array probes for 30 min at 45°C, after which the probe array was washed and scanned with the GeneChip® Scanner 50 with a laser that excites the fluorescent label. Normalization Controls: GeneChip® initial quality control analyses were performed using M A S software (version 5.1) as described previously. Subsequent normalization and analysis was performed in GeneSpring 5.0.1™ (Silicon Genetics) with time and infection as continuous and color-coded independent variables, respectively. The 50th percentile for all signal intensities was used as a positive control for each sample; each measurement was divided by this synthetic positive control value, assuming that this was at least 10. The per-chip normalization is useful to help control for minor chip-to-chip differences in probe preparation, hybridization conditions, etc. Only genes marked as marginal or present were accepted for further analysis. The bottom 10 t h percentile was used as a test to correct for background subtraction. This value was never less than the negative of the synthetic positive control. Per-gene normalization is useful to compare gene profde patterns when normalized values are at different intensity levels. Each gene was normalized to itself by making a synthetic positive control for that gene, and dividing all measurements for that gene by this positive control, assuming it was at least 0.01. This synthetic control was the median of the gene's expression values over all the samples. Reliability of expression data is considered high at 500.0, medium at 150.0 and low at 50.0 intensity units. Genes with signal intensity above 115 units were 135 considered for analysis. Normalized values below 0 were set to 0. Signal intensities are reported as log2 fold changes of normalized intensity values for virus-infected over sham control samples. Normalized intensity values were given considering the relatively small fold changes reported for some genes. Statistical analysis Sample HeLa cell R N A was isolated and processed one plate of cells per GeneChip®. A cross-gene error model was used to obtain an estimate of the variability in signal intensity. Bonferroni step-down adjusted p value multiple testing corrections were used in GeneSpring 5.0.1™. Contribution by the Author This author was responsible for experimental planning and communication between investigators at the iCAPTURE Centre and Inphogene Biocommunications. The author was also responsible for propagation and maintenance of virus stocks, cell culture, infection of HeLa cells and cell harvesting. Extraction and processing of RNA for array hybridization and initial data conversion in M A S 5.1 were performed at Inphogene Biocomm. (Vancouver, B.C.) with assistance from the author. Subsequent normalization and bioinformatical analysis was performed in GeneSpring 5.0.1™ by this author. Analysis in Genetrix (August, 2002 version) was performed by this author, Nana Rezai and Dr. Jonathan Buckley at the University of Southern California. Results Since the discovery of the importance of direct virus-mediated injury in enteroviral heart disease [268], the interaction between virus and host cells has been the fociis of intense investigation. The heart is a complex organ which includes diverse cell types and infection involves many simultaneous cellular processes. Thus, in an isolated HeLa cell system, the transcriptional outcome of virus-host interactions from initial infection to virus release was studied. HeLa cells have been 'workhorse cells' used as a model to study viral pathogenesis. In fact, identification of cell surface receptors, architectural injury and 136 induction of survival signals, among other important findings were first made in this model, only later to be confirmed in the intact heart [25, 64, 255]. Therefore, HeLa cells are a valid initial model to study the molecular mechanisms of enteroviral infection. A l l findings of interest will be confirmed in cultured cardiomyocytes and in myocarditis-susceptible mice. L CVB3-infection time course experiment Cultured HeLa cells were infected with CVB3 and at 0, 30m, 1, 3, 7, and 9h post-infection, cells were collected and total R N A was isolated. Double stranded cDNA was synthesized, in vitro transcribed into cRNA, fragmented by partial hydrolysis and hybridized to GeneChip®s. Signal intensities were converted into a text format in M A S and submitted to Genetrix (August, 2002 version). GeneChip® expression intensities for all genes were averaged over all time points and plotted according to signal intensity. As expected, differential expression and signal intensity are inversely related where genes with high signal intensities were mostly unchanged whereas, low copy number genes showed a deviation from 1 (Figure 51). Genes outside the normalized intensity ratio confidence intervals 0.5 and 2 were highlighted and labeled. Genes with high signal intensity and large intensity ratio over all time points are potential genes of interest and listed in Table 5. Genes were then plotted temporally from 0 to 9h pi (Figure 52). a. Expression-based clustering K-means clustering (k=20 clusters) analysis was utilized to identify those genes up-regulated at 30m, 1, 7 and 9h pi (Figure 53). These time points correspond to virus-receptor binding, the start of virus transcription, progeny virus release and host cell apoptosis, respectively. Unlike genes up-regulated in vivo, the majority of genes up-regulated in vitro were signalling-related and differential expression was most prominent at the 7-9h time point. Cluster analysis revealed differential expression of transcription factors TR3 orphan receptor with log2 fold increase of 3.39 (6.9/0.66 normalized intensity units for virus/sham) 137 Correlation -a033 |p <0.00001J Signal Intensity (Normalized Units) Figure 51: Overview of gene expression profiles in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 C. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to Genetrix (August, 2002 version). Intensities over all timepoints were averaged and plotted as ratios of experimental/control intensities vs. magnitude of signal intensity. Genes with both strong signal intensity and high average differential expression were annotated as potential genes of interest. 138 Up Regulation Down Regulation Glycoprotein lb (platelet), alpha polypeptide Protein phosphatase 2 (formerly 2A) Neuronal She adaptor homolog Alkylglycerone phosphate synthase Fibroblast growth factor receptor 1 Zinc finger protein 262 Rho guanine nucleotide exchange factor (GEF) 5 RAB6A, member RAS oncogene family Chondroitin sulfate proteoglycan 5 (neuroglycan C) Lipase A, lysosomal acid Putative chemokine receptor; GTP-binding protein Collagen, type XI, alpha 2 Transforming growth factor, beta 1 Signal transducer and activator of transcription 1, 91kD Protein C receptor, endothelial (EPCR) Laminin, beta 2 (laminin S) Cytochrome P450, subfamily XXVIA, polypeptide 1 Prostaglandin E receptor 3 (subtype EP3) Myosin binding protein C, slow type Collagen, type VI, alpha 3 Dual specificity phosphatase 5 Low density lipoprotein receptor-related protein 5 Hemoglobin, epsilon 1 Tumor protein p53 binding protein, 1 SHP2 interacting transmembrane adaptor Mitochondrial ribosornal protein L33 Dystrophia myotonica-protein kinase Lysosomal-associated membrane protein 2 V-fos FBJ murine osteosarcoma viral oncogene homolog Tumor necrosis factor (ligand) superfamily, member 14 Platelet-derived growth factor receptor, alpha polypeptide Neurotrophic tyrosine kinase, receptor, type 3 Connective tissue growth factor Glycerol-3-phosphate dehydrogenase 2 (mitochondrial) Immunoglobulin heavy constant gamma 3 (G3m marker) Transforming growth factor, beta receptor II (70-80kD) Calcium/calmodulin-dependent protein kinase kinase 2, beta Janus kinase 1 (a protein tyrosine kinase) Integrin, beta 1 (fibronectin receptor, beta polypeptide) Cyclin-dependent kinase 5 Glycogen synthase kinase 3 beta Vascular cell adhesion molecule 1 MHC class I region ORF Heat shock 70kD protein 1B Glutathione S-transferase A2 Microsomal glutathione S-transferase 3 Tissue inhibitor of metalloproteinase 4 PAI-1 mRNA-binding protein Glycine receptor, alpha 1 Heat shock 70kD protein 1A Table 5: Genes of interest in CVB3-infected HeLa cells averaged over all timepoints. The experimental protocol is illustrated and described in Figure 26 C. Gene expression intensities were averaged over all time points and top up- and down-regulated genes are shown. 139 Figure 52: Temporal display of gene expression patterns in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 C. Signal intensity data was converted into text format in Microarray Suite 5.1 and inputted into GeneSpring 5.0.1TM. Gene intensity values (12, 626 genes) were normalized by gene (to the median value) and by chip (to the 50th percentile). 140 c 0 1 3 7 9 h p i CTGF 18s Figure 53: Clustering gene expression data from CVB3-infected cells. The experimental protocol is illustrated and described in Figure 26 C. (A) A l l normalized intensities were submitted for K-means clustering (20 cluster sets; 100 iterations; standard correlation; GeneSpring 5.0.1TM). (B) Set 10 (235 genes) clustered transcripts up regulated 7-9 hours pi. (C) Connective tissue growth factor (CTGF) upregulation was further confirmed using RT-PCR. CTGF is also increased >10 fold acutely in CVB3-infected mouse hearts (data not shown). 141 at 7h and log2 fold increase of 3.41 (5.3/0.5 normalized intensity units for virus/sham) at 9h pi. AP-1 transcription factors such as c-myc showed log2 fold increase of 0.87 (1.7/0.98 normalized intensity units for virus/sham) at 7h and log2 fold increase of 0.86 (2.0/1.1 normalized intensity units for virus/sham) and phosphatases were up-regulated (Figure 53). Connective tissue growth factor (CTGF), another gene of interest, was inceased 5.8 and 3.9 normalized intensity units at 7 and 9h pi (sham below detection limit) [269]. Confirmatory RT-PCR analysis has confirmed up-regulation of CTGF transcript 9h pi (Figure 53). CTGF is a potent trigger of cellular processes underlying fibrosis and as such its expression is of great interest as a virus-induced pro-fibrotic factor in viral myocarditis. b. Function-based gene profiling A gene function map was created based on the Swiss-Prot database (Figure 54) and genes with known functions were assigned to one or more groups. Global trends in gene expression (>15 groups) were identified and four functional groups were selected. These events, these groups of genes were plotted adjacent to the stages of the virus life cycle (Figure 55). Due to high background, immediate early transcriptional changes were difficult to identify but collagens appear to be down-regulated. By 3h pi, signal transduction genes were up-regulated while collagen and heat shock genes were down-regulated. At 7 and 9h pi, these trends continued and several cellular oncogenes were up-regulated. Up-regulation of AP-1 transcription factors was observed including log2 fold increase of 1.81 (3.5/1.0 normalized intensity units for virus/sham) for the cellular oncogene c-fos at 7h and log2 fold increase of 1.06 (2.3/01.1 normalized intensity units for virus/sham) at 9h pi (Figure 56). C-fos transcript was also increased in CVB3-infected mouse hearts. This analysis and others in ouhr laboratory led to the finding by Luo et al [64] that the M A P K -E R K pathway is triggered in a biphasic manner in CVB3-infected HeLa cells. These findings led to further efforts to elucidate the downstream transcriptional targets of the M A P K - E R K pathway in the next section. 142 This initial time course data provided insights into trends in signalling genes, phosphatases, kinases and transcription factors that are up-regulated following C V B 3 infection. However, the large variability in control HeLa cells limited the analysis to large differential expression events. Therefore, this experiment was repeated with triplicate samples hybridized to separate GeneChip®s for greater confidence to detect subtle but potentially important transcriptional changes. 143 Carboxsylases CaA enzyme Dehydrogenase Demethylase Heiicase Ligase Phosphodiesterase Protease Tyrosine Phosphatase Serine threonine Tyrosine Adhesion cGMP-dep Cyclin-dep GSK MAPK MAPKK MAPKKK MHC antigen Complement Acute phase response Antimicrobial Blood coagulation Immunoglobulin Opsonin Anticoagulant Antioxidant Cell adhesion Elongation factor Lysin Ubiquitin Microtubular Dynamics Immunity, Proteiru Enzyme \ ONA Binding Ri bo n ucle op rotei n Translation Factor Death Growth factor IL-10 IL-1 IL-8 PDGF Other Chaperone PI3K PKC Cell Cycle Regulator Apoptosis Regulator Motor Nucleic Acid Binding RNA Cytokine Growth factor Hormone Seven less Ligand Cancer Oncogene Tumor suppressor Molecular Function All Genes Receptor Signal Transducer Structural Protein Storage Transport Fibrinogen Interstitial matrix ECM-Cell Growth and Maintenance Cell ^ Cell A d h e s i o n / Communication \ A - " Biological Process Cellular Component Extracellular / Cilium C n r : ~ ; o s ? n a * Endosome Cytoplasm Cell Death Signal Transduction 7\ Mitochondrion / Ribosome Cell death Necrosis Senescence Cell Surface-Linked Signal Transduction Developmental Processes k Physiological Processes Intracellular Signalling Vacuole! ER Intracellular--* \ Lysosome Membrane 1 Ca++ dep receptor signall ing Enzyme linked G-protein Dependent Integrin receptor signall ing Wnt receptor signall ing Nitric oxide Phosphorelay-MAPKKK Protein kinase cascade Voltage-dep signall ing NFkB Stat Golgi Apparatus Basement Golgi Integral Lysosomal Nuclear Peroxisomal Plasma Basement membrane Collagen Elastin ECM M icrof ibrils Figure 54: Function-based classification map. A gene function map was created based on the functional classifications in GeneSpringTM, adapted from the Swiss-Prot database. This map was used as a knowledge base from which in-house functional clusters were created to investigate the gene expression patterns of discrete functional groups. 144 Figure 55: Biological system transcriptional up regulation in CVB3-infection in vitro. The experimental protocol is illustrated and described in Figure 26 C. Signal intensities were converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5 . 0 . 1 ™ . Function-based data mining approaches for normalized intensities (see Figure 54 for gene function map) revealed transcriptionally activated biological systems in CVB3-infected HeLa cells. Gene systems are listed in parallel to the CVB3 lifecycle. 145 2.0 cfl.O c?0.0 re O 2-1.0 o -2.0 9 3 E ro CVB3 co 10m 30m 1 3 5 7 9 h pi P-ERK1/2 (From Luo et al, 2002, J Virol, 76:3365) Figure 56: C-fos up regulation in CVB3-infected HeLa cells and mouse hearts. The experimental protocol is illustrated and described in Figure 26 B and C. A. Normalized intensities for C-fos (V01512) gene expression was increased at 7 to 9 hours using GeneChip^s. (B) C-Fos was also increased acutely in CVB3-infected hearts at 3 and 9 day post infection (data not shown). C. The upstream singalling protein M A P K - E R K was activated at 10 min, and 7 to 9 hours by Western blot (From Luo et al, 2002, J Virol, 76:3365). 146 4.2 Triplicate Gene Profiling of CVB3-Infected HeLa cells with M E K l Inhibition Biological Questions What are the differentially transcribed genes and gene groups in CVB3-infected HeLa cells? What are the downstream transcriptional targets of the E R K signalling pathway? Rationale Viral infection of mouse hearts involves a multifaceted, temporally infectious process in a complex tissue. To understand the differential transcriptional events from virus-host interaction, we used an in vitro HeLa cell model of CVB3 infection. Since previous in vitro experiments resulted in large gene variability, I repeated the time course with triplicate biological replicate samples. Also, to focus on the elucidation of downstream transcriptional targets of the M A P K - E R K pathway, a specific M E K l inhibitor was utilized as a comparator group. MIAME Criteria: Experimental Design and Quality Control Experimental Design: HeLa cells were serum starved and either pre-incubated with dimethyl sulfoxide (DMSO; vehicle), pre-incubated with U0126 (lOuM in DMSO) and infected with CVB3 (Kandolf strain, multiplicity of infection [MOI] 10), or pre-incubated with DMSO (vehicle) and infected with PBS (sham). DMSO treatment was utilized as a negative infection control as DMSO may cause apoptosis in cell culture. Following infection, virus was removed and fresh media containing 10% fetal bovine serum was added for the remainder of the infectious process. At 0, 30 minutes, 3 and 9h following CVB3 infection (N=3 plates per time point), RNA was isolated, processed and hybridized to GeneChip®s (N=3 per sample; Probe Array Lot: 2002284). A U R L to a supplemental database will be made available shortly following publication of results. M I A M E descriptions of array design, sample preparation, measurement conditions and normalization controls are identical as Section 4.1. Signal intensities are reported as log2 fold changes of normalized intensity values for virus infected over sham control samples. 147 Normalized intensity values were given considering the relatively small fold changes reported for some genes. Statistical analysis Each plate of HeLa cells were infected, RNA was isolated and processed per GeneChip® (N=3). These samples were considered true biological replicates. Array quality control measures were identical as in Section 3.3 and statistical analysis was performed in GeneSpring 5.0.1™ for triplicate samples. Parametric statistical analysis was performed using the Welch Mest for expression intensities with unequal variances. This method was performed on sample replicates and with the assumption that variability between replicates is similar for all genes with similar expression intensities. Bonferroni step-down adjusted p value multiple testing corrections were used in GeneSpring 5.0.1™. Contribution by the Author Experimental planning and communication between iCAPTURE Centre and University of Southern California investigators was executed by this author. HeLa cell culture, infection and collection were performed by this author, John Zhang and Dr. Honglin Luo at the iCAPTURE Centre. Sample processing including RNA extraction, cDNA and labeled-cRNA synthesis were performed by the author and Nana Rezai. Sample hybridization was performed at the Genome Core Facility, Children's Hospital Los Angeles by technical specialist Betty Schaub, the author and Nana Rezai. Bioinformatical analysis including transformation in M A S 5.1, normalization and examination in GeneSpring 5.0.1™ and GenMAPP 1.0 were performed by this author. Sample processing and RT-PCR were performed by this author. Results L Triplicate CVB3'-infection time course experiment with MEKl Inhibitor Treatment Based on the initial investigations in Section 4.1, the CVB3-HeLa cells time course was repeated to target specific time points in the CVB3 lifecycle with triplicate GeneChip® arrays with three separate biological samples for increased statistical confidence. Initial 148 analysis led to the finding that inhibition of E R K blocks virus replication and delays host cell death [64]. An additional experimental group of HeLa cells preincubated with a M E K l inhibitor (U0126, Promega) followed by infection with CVB3 was added. The concentration of lOuM was chosen based on a dose response experiment to determine inhibition of ERK1/2 activation and virus replication in CVB3-infected HeLa cells [64]. U0126 selective inhibits E R K but also has weak inhibitory activity on the MAPK-p38 pathway [270]. However, p38 is not activated in CVB3-infected HeLa cells (Luo et al, unpublished observations). Gene profiling was performed at 30m, 3 and 9h pi (Figure 52). For increased statistical power, triplicate arrays were run per time point. A l l genes were plotted based on treatment parameters and expression intensity (Figure 57). The expression of relatively few genes was significantly changed following CVB3 infection which occurs mostly at the 9h time point. a. Function-based gene profiling Functional groups of genes were investigated for general trends in gene behavior following CVB3 infection. The Swiss-Prot-based functional classification, as described previously and illustrated in Figure 53, was utilized. Global down-regulation of intracellular genes, including mitochondrial, lysosomal and cytoplasmic, were observed at 9h pi (Figure 58). Although many of the signal intensities of the genes in these groups were below the threshold of 1.8 fold changes in intensity, the global nature of these trends suggests that this is an important phenomenon. These changes were largely unaffected by blockage of the E R K pathway. Marked alterations were seen in several 'virus-related' transcripts, virus-encoded genes that exist in mammalian genomes. Previously-mentioned c-fos was log2 fold increased by 2.50 (3.4/0.60 normalized intensity units for virus/sham) and partially restored with U0126 treatment to a log2 fold increase of 1.66 at 9h pi (Figure 59). Similarly, c-jun (an AP-1 family transcription factor) was log2 fold increased to 1.49 (3.1/1.1 normalized intensity units for virus/sham) and log2 fold increased to 1.13 with CVB3 and U0126 at 9h pi. Abl 149 Figure 57: Temporal display of genes in CVB3-infected cells +/-U0126. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5.0.1™ Genes were plotted with time on the x-axis and normalized intensity values on the y-axis. In A , the color parameter is a categorical treatment value and in B, continuous values of signal intensity as shown on the right side of the figure. 150 Lysosomal Sham Virus Virus+U0126 ATP synthase, H+ transport P r o c a r b o x y p e p t i d a s e Figure 58: Down regulation of intracellular genes in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5.0.1™. Patterns of gene transcription are shown (see Figure 54 for gene function map). A down regulation of normalized intensity of intracellular genes, including mitochondrial, lysosomal and cytoplasmic, is seen 7-9 h pi. Specific genes of potential interest are illustrated. 151 Virus related genes Sham Vjun avian sarcoma virus 11 oncogene homolog Virus+U0126 V abl Abelson murine leukemia viral oncogene homolog 1 isoform a Figure 59: Differential expression of virus-related genes in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5 . 0 . 1 T M . Patterns of gene signal intensities are shown (see Figure 54 for gene function map). Differential expression of host cellular gene normalized intensity related to virus life cycles, are shown post CVB3 infection. Jun, abl and fos oncogenes and inhibitor of apoptosis (IAP) have reported pro-survival functions. 152 oncogene and inhibitor of apoptosis are anti-apoptotic proteins with log2 fold down-regulated to -1.08 (0.43/0.91 normalized intensity units for virus/sham) and log2 fold up-regulated to 0.58 (2.1/1.4 normalized intensity units for virus/sham) at 9h, respectively, but partially returned to baseline with U0126. Inhibitor treatment at concentrations, which completely blocks E R K activation, did not completely block these transcriptional events, suggesting that other pathways or upstream crosstalk events trigger their expression. Differentially regulated genes in CVB3-infected HeLa cells are mainly signal transduction-related. Notably, expression of CTGF showed log2 fold increase of 1.74 (2.4/0.72 normalized intensity units for virus/sham) and insulin-like growth factor-2 (IGF-2) log2 fold increase of 1.14 (2.1/0.95 normalized intensity units for virus/sham) at 9h. This up-regulation was partially restored with U0126 treatment to a log2 fold increase of -0.35 (1.1/1.4 normalized intensity units for virus/sham) for CTGF and log2 fold increase of 0.72 (1.56/0.95 normalized intensity units for virus/sham) for IGF-2, respectively, suggesting that E R K signalling may have some contribution. CTGF transcript was also increased 5.1 fold at 30 min and 2.4 fold at 9h (Figure 60). CTGF, which is induced by transforming growth factor-P (TGF-P), mediates stimulatory actions of TGF-P E C M synthesis [271]. Based on its important role in fibrosis and excessive scarring, wound repair, neoangiogenesis and apoptosis in myocardial infarction, we are interested in the potential role of CTGF in fibrosis following viral myocardial injury is of interest [272]. b. Expression-based clustering Genes were also identified based on expression patterns alone. I clustered genes according to expression patterns based on a three-stage grouping (-, 0 , +) at 30 minutes, 3h, 9h and 9h with U0126 treatment pi. In-house clustering was performed as follows: a table containing the genes and their cluster numbers and another table containing the genes and their functional groups were sorted based on their Affymetrix unique identification. The elements unique to each table were then combined in both tables, sorted and positioned side by side on an Excel worksheet. 153 Script was written to generate two new tables based on clusters. For each gene cluster, the different functional groups were assigned and pie charts of pie charts were created to illustrate the functional makeup of expression-based gene clusters. Genes up-regulated at 9h pi but not any other time point were fdtered and the functional make up of these genes was determined (Figure 61). Several genes were involved in signal transduction, enzyme function and those broadly characterized as intracellular components. These include transcription factors (ATF3, c-myc, jun B, TFIID, ETR101, IIB), kinases (serine threonine kinase, EHK1 receptor tyrosine kinase ligand), phosphatases (MAP kinase phosphatase, dual-specificity protein phosphatase protein phosphatase-1 inhibitor) and cellular receptors (Nerve growth factor receptor, Mitogen induced nuclear orphan receptor). Interestingly, the signal-relatedness of up-regulated gene functions in vitro is very different than the makeup of up-regulated genes in vivo. 154 Sham Virus Virus+U0126 Receptor Figure 60: Signal transduction gene up regulation in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 5.1 and submitted to GeneSpring 5.0.1TM. Patterns of gene transcription are shown (see Figure 54 for gene function map). Normalized intensities for signal transduction ligand genes are increased in CVB3-infected HeLa cells. Marked increases in normalized intensities for CTGF and insulin-like growth factor-2 (IGF-2) post CVB3 infection are shown. 155 Nerve growth factor receptor C-type lectin precursor Monocarboxylate transporter homologue M C T 6 Na+-dependent phosphate cotransporter Sodium/glucose cotransporter Serine/threonine kinase Nerve growth factor receptor Sodium/glucose cotransporter-like protein Calcium-binding protein chp Glutamate receptor type 4 • Cancer • Cell Cycle Regulator • Enzyme • Immunity Protein • Nucleic Acid Binding o Signal Transducer • Transport • Other Groups EHK.1 receptor tyrosine kinase ligand M A P kinase phosphatase Dual-specificity protein phosphatase protein phosphatase-1 inhibitor R N A polymerase II elongation factor E L L 2 NADH-Ubiquinone Oxidoreductase Transcription factor TFIID Activating transcription factor 3 Transcription factor ETR101 General transcription factor IXB Cytochrome b5 ( C Y B 5 ) Jun-B gene HnRNPcore protein A l Mitogen induced nuclear orphan receptor Figure 61: Functional distribution of up regulated genes at 9 hours in CVB-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5.0.1™. A l l genes were fdtered using a custom algorithm to isolate genes which with normalized intensity up regulation at 9 hours pi (and not at any other timepoint) and these genes were categorized according to their functional roles. The majority of up regulated genes are related to intracellular functions, particularly signal transduction and transcription factors. Genes of biological interest are listed (Note: genes may appear in more than one functional category). 156 it Downstream genes of MAPK-ERK Another focus of this set of experiments was to elucidate potential transcriptional targets of the M A P K - E R K . Function- and expression-based approaches to isolate trends and potential genes downstream of this important pathway were investigated. a. Expression-based clustering In a similar cluster analysis as above, genes down-regulated at 9h pi with U0126-treatment were fdtered and analyzed in an expression-based manner (Figure 62). The in-house clustering was performed and for each gene cluster, the functional groups were assigned and pie-of-pie charts were created to illustrate the functional makeup of expression-based gene clusters. A l l genes were clustered according to expression patterns based on a simple three-stage grouping (-, 0 , +) at 30 minutes, 3 hours, 9h and 9h with U0126 treatment pi. Thus, there were 81 (34) possible clusters but all genes grouped into 71 actual clusters. The genes up-regulated at 9h pi but then reduced with U0126 treatment were fdtered and the functional makeup of these genes was analyzed. Most genes are involved in 'molecular function' and fewer genes are categorized as 'biological function' and 'cellular components' (Figure 62). The classification is adapted from gene annotation in the Swiss-Prot database as illustrated in Figure 53. Genes involved in molecular function include those involved in MAPK-related signalling (MKK7, M K P , p38), cyclins (H and G l ) a cancer-related (c-fos, jun B, abl) genes, suggesting that they may play important functions in ERK-CVB3 interactions. Genes involved in biological processes include the voltage-dependent calcium channel beta-1, N F K B 2 and several growth factors (Insulin-like growth factor binding protein 5, FGF-6, V E G F C). Cellular component genes include four cytochrome P450 (CYP) and microtubule-associated protein IB (MAP1B). A related protein to MAP1B, MAP5, is a known target of picoraavirus 3C protease cleavage suggesting a link between protease cleavage and 157 Insulin-like growth factor binding protein 5 Eotaxin Insulin Homology to T N F and C D 4 0 ligand Sodium phosphate transporter (NPT3) Na+/H+ exchanger Sialophorin (CD43) Alpha-2,8-polysialyltransferase (PST) Insulin-like growth factor binding protein 5 F G F - 6 V E G F C Growth hormone-releasing hormone receptor G R O - b e t a A b l oncogene C-fos oncogene Jun-B Integrin alpha subunit Soxl N F k B 2 Voltage-dependent calcium channel beta-1 I Cell Comunicat ion I Signal Transduction Coagulation factor V • A p o p t o s i s Regulator • Cancel • Cell Cycle Regulator • Enzyme • Immunity Protein • Micro! ubular Dynamics • Nucleic Ac id Binding • Transport • Signal Transducer • Storage • St iuclural Protein • Other Groups c G M P phosphodiesterase p21-activated kinase 3 Mitogen-activated protein kinase kinase 7 M A P kinase phosphatase p38 M A P kinase Protein tyrosine phosphatase delta • Extracellular • Intracellular • Membrane Collagen type IV alpha 3 E C M protein 1 BH3-containing Bcl-2 protein C Y P 4 5 0 c l 7 , P C N 3 , H F L a , 2A13 C y c l i n G l interacting protein Microtubule-associated protein I B A l k a l i myosin light chain 1 Figure 62: Functional distribution of E R K downstream genes at 9 hours in CVB-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5.0.1™. Genes were fdtered using a custom algorithm for normalized intensity down regulation following U0126 treatment at 9 h pi and were categorized according to their functional roles. The majority of up regulated genes are related to intracellular functions, particularly signal transduction and enzymes. Genes of biological interest are listed (Note: genes may appear in more than one functional category). 158 transcriptional regulation [273]. Verification of these apparent E R K downstream genes and their potential pro-viral and pro-apoptotic role remains to be determined. b. Function-based gene profiling The published literature was searched for transcription factors, signalling, cytoskeletal and other genes, which are known direct downstream transcriptional targets of the E R K pathway in any cellular system and triggered by any stimulus. I utilized the GenMAPP 1.0 visualization tool to show the transcriptional result of U0126 treatment at 9h pi as the following ratio (normalized intensity of CVB3+U0126: normalized intensity of CVB3; Figure 63). Of the known downstream M A P K - E R K genes, few were down-regulated by U0126 (blue) and 7/8 genes were transcription factors (Ets translocation variant 1, NF-IL-6p, EPH-related receptor tyrosine kinase ligand 6, c-fos, c-jun and jun B). Function and expression approaches found many overlapping downstream E R K genes but many genes were unique to one approach. In my experience, the utilization of a variety of approaches has been helpful to find different patterns and as an internal confirmation for my analyses. iii. Interesting genes and gene groups Using GeneSpring 5.0.1™-based functional clustering and GenMAPP 1.0 visualization and combining both in vitro and in combination with in vivo data, the following novel genes and gene groups differentially regulated during CVB3 infection, were identified: (i) serine protease inhibitors (serpins), (ii) matrix metalloproteinases (MMPs) and (iii) cytochrome P450s (CYPs). L Serine protease inhibitors Serpins regulate a diverse array of serine and cysteine proteinases associated with essential biological processes such as fibrinolysis, coagulation, inflammation and apoptosis. Activation of select serpins was found in CVB3-infected HeLa cells and mouse hearts (Figures 64 and 65). GenMAPP 1.0 was used to visualize a snapshot of the ratio of serpin 159 Author. Bobby Yanagawa Maintained by: Bobby Yanagawa E-mail: byanagawa@mrl.jbc.ca Last modified: December 2, 20Q2 Expression Dataset Name: HeLa Cell New Legend Downstream Genes of ERK 1/2 Transcription Factors Upregulation Downregul ati on No change No oriteh a met Not found |Ets transit cation variant 1 i ' * | ELK-1 | i 1717 Elk1 | i . i 2 i s ELK4: iou ELK4ii.oi4 PROTEIN TYROSINE PHOSPHATASE SAP-1 0.829 PPAR binding protein 0.9674 Signaling Proteins Son ofsevenless protein homolog 1J2 9421 :Protein kinase C inhibitor protein-1 ; i .s22s :Ras related protein Rab4AB 9423 Protein tyrosine phosphatase^ .9237 Cytoskeletal Proteins |PPAR-gamma|i Q323 Estrogen receptor-related proteink 4762 Estrogen receptor beta;0.7568 Golgi associated microtubule binding protein 11 0563 :Caldesmon2 3s36 | Synapsin I |o.s982 Cellular oncogene fos ellular oncogene fos ;o 5243 : Microtubule-associated protein tau 11.0877 J0.7966 B.389S 04718 10.4718 :Myc proto-oncogene proteinjo 7 i e e N-myc proto-oncogene protein : i . iS64 Endothelial transcription factor GATAi ; i 5113 1448 05283 Other Targets Tyrosine 3-hydroxylase | i 3053 |o .3358 Stathmin;o.522i Topoisomerase II alpha| 1 5788 Nuclear factor NF II 6 beta lated receptor tyrosine kinase Ii |E74-like factor 4 j T - : ; ; iOncogene AML-1:0SG81 Figure 63: Map of known M A P K - E R K downstream genes at 9 hours in CVB-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and submitted to GeneSpring 5.0.1™ for normalization. Ratios of normalized signal intensities between CVB3 + U0126/CVB3 treatment groups at 9 hours post-infection were calculated in Microsoft Excel and submitted to GenMAPP 1.0. Based on an extensive review of the literature, a MAPP of downstream genes of E R K was created and normalized signal intensity ratios were submitted for visualization. Ratio cut-off was arbitrarily set at <=0.5 for downregulation (blue), >0.5 and <1.5 for no change (yellow) and >1.5 for upregulation (red). Transcription of several E R K downstream transcription factors are down regulated following U0126 treatment 160 gene expression in virus-infected cells at 9h pi, notably A6, G l and F2 (red represents up-regulation; Figure 64). Up-regulation of serpin 3c,n acutely and serpin gl and 2 chronically were observed using GeneChip®s. Acute log2 fold increases of (B) 3.6 and 3.1 were shown for serpin el at days 3 and 9 pi, respectively, by cDNA array (C) and RT-PCR (D; NM_012620.1; Figure 65). Increased serpin expression may limit inflammatory infdtration as seen following myocardial infarction [274] or directly inhibit myocyte apoptosis [275]. ii Matrix metalloproteinases Matrix metalloproteinases are zymogens which are involved in extracellular matrix metabolism. The increased signal intensity of MMP-related genes was seen in vitro and in vivo following CVB3 infection (Figure 66). HeLa cells challenged with CVB3 up-regulated MMP-3, 7 and 9. There was increased signal intensity of MMP-3 and -12; their endogenous inhibitor TIMP-1 (tissue inhibitor of MMP-1); and Adam8, a type I transmembrane proteins with metalloproteinase and disintegrin-containing extracellular domains, in CVB3-infected mouse hearts (Figure 66). Transcript expression for 2 of 3 probe sets for the macrophage elastase MMP-12 increased logio fold of 1.65 (p«0 .001) at day 9 to logio fold of 1.83 at day 30. However, MMP-12 expression was unchanged in CVB3-infected HeLa cells in vitro. Taken together, this data suggests that endogenous tissue macrophages and potentially infdtrating monocytes but not target cells are responsible for MMP-12 expression. Conversely, differential up-regulation of MMP-3 in vivo and in vitro suggests that infected host cells are responsible for such expression. Expression of an M M P program likely allows for inflammatory infdtration but also contributes to chronic dysfunction and cardiac dilation. iii Cytochrome P450 Cytochrome P450 (CYP) is a ubiquitous family of enzymes responsible for the metabolism of a wide variety of drugs and their metabolites. There was up-regulation of numerous P450 genes, including CYP3A4, 5, 7 and CYP2A7, 2C8 and 2J2 in CVB3-infected HeLa cells (Figure 67). A logio fold increases of 1.47 in signal intensity was seen for C Y P l b l in 161 vivo during the inflammatory stage of myocarditis. Other cardiac CYPs including 2D6, 2J2 and 2C9 were not present on the arrays. The direct result of C Y P expression in the heart is unknown, but CYPs may modulate vascular homeostasis. 162 Hs_serpin Autttor. Adapted from Gene Ontology Maintained toy: GenMAPP.org E-mait genmapp@gladstone.ucsf.edu LastmodiSed: 3*2/2002 Right click here for Notes. SERPIN HUR7 SCCA1 SCCA2 SERPINA1 SERPINA10 SERPINA2 SERPINA3 SERPINA4 SERPINAS SERPINA6 SERPINA7 SERPINAS SERPINB1 SERPINB2 SERPINB3 SERPINB4 1.37*8 0.7301 1.4848 1.3031 |4.8976 1.8506 D4433 I0.834S 1.1373 1.9707 0.7898 0.4834 SERPINB6 0.8S78 SERPINC1 1.9376 SERPIND1 0.99 SERPINE1 2.2976 SERPINE2 SERPINF1 SERPINF2 2.8556 SERPING1 3.1071 SERPINH1 1.6335 SERPINH2;o.es96 SERPINI1 0 .4626 0.9527 1 Expression Dataset N a m e : HeLa Cell N e w Legend BU p r e g u l a t i o n D o w n r e g u l a t i o n No c h a n g e No c r i t e r i a met Not f o u n d Figure 64: Serpin gene profile in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 D. Signal intensity data was converted into a text format in Microarray Suite 5.1 and normalized in GeneSpring 5.0.1™ and submitted to GenMAPP 1.0. Signal intensity ratio for virus/sham cut-off was arbitrarily set at <=0.5 for downregulation (blue), >0.5 and <1.5 for no change (yellow) and >1.5 for upregulation (red). Gene labels are Swiss-Prot designations. Serpin genes were up regulated at 9 hours following CVB3 infection in HeLa cells were probed using GeneChip® arrays and visualized by GenMAPP. 163 CVB3 (Days pi) 2.0 g) 1.0 o ? 0.0 (0 sz O 1-1.0 •2.0 30 1 1 Serpin g1 Serpin 2 Pi" 11 }in t l )in 3c eriin 3g >erpin 3n Serpin 2 Serpin la Ser Jin 2 Ser >in 3c Serpin b4 B 30 d c Sham Day 3 Day 9 Day 30 Serpin e1 p-actin 1 mm Figure 65: Serpin gene profile in CVB3-infected mouse hearts. The experimental protocol is illustrated and described in Figure 26 A and B. Serpin genes are found to be up regulated in CVB3-infected HeLa cells using (A) GeneChip® arrays (compared to baseline controls), in mouse hearts using (B) cDNA arrays (compared to sham controls) and using (C) RT-PCR (N M O 12620.1), all in separate experiments (compared to P-actin). 164 ALLOPEPTIOASE ADAMTS7 ACY1 0.7376 ADAMTS6 FLJ00064 ADAMTS9 P A P P A 1.3427 AFG3L1 RCE1A 0.9799 AFG3L2 0.6846 TLL 0.8944 ANFEP 1.3616 TLL2 CPA1 1.792 XCE CP A3 1.6936 LOENOOPEPTIOASE CPB1 2.1771 ADAM 10 0.S703 • a m 3.0434 ADAM 11 1.508 0.1649 ADAM12 3.4953 2.2781 ADAM 15 0.8197 DCP1 :ADAM17 0.5123 DCP1 1.6976 ADAM? 0.6871 D.1783F3.? ADAM 20 1.5014 DJ964r7.1 ADAM21 5.0374 DPEP1 D.37S1 : ADAM22 0.777 DPP3 ADAM22 1 4 2 1 7 ENPEP 1.7427 3 13?S IFERTILIN BETA] ADAM29 KIAA1226 ADAM30 LTA4H 0 7884 ADAMS 1 2 2 5 8 MBTPS2 0.6605 ADAMS 0.7569 M1FR-2 ADAMTS1 MIPEP 0.808 ADAMTS3 1 2 1 MMP14 1.3983 ADAMTS3 MMP15 1.1473 ADAMTS4 MMP1G 1.245 ADAMTS5 MMP17 1.1303 ADAMTS6 MMP20 1.4309 MMP20 MMP21/22 MMP21/22 MMP21/22 MMP ?3 MMP24 MMP25 MMP28 MP1 PGN 1.0581 1.9164 0.5952 0.3433 T H 0 P 1 TRHDE UQCRC1 U0CRC2 0.6941 0.8121 ZMPSTE24 0.842 BMP1 1.1001 MEP1A 0.8781 MEP1B 0.7696 |ADAMTS2j i 9248 | MMP10 11.4702 I 1523 10.4277 IT7T1 MMP26 MMP3 MMP7 2.7538 3.S575 wsa INTERSTITIAL COLLAGENASE 1 W M P 1 ~ | 1 5 2 3 5 NEUTROPHIL COLLAGENASE I MMP8 |o.35d3 Author Adapted from Oene Ontology Maintained by. GenMAPP.org E-maif: genmapp@gladstone ucsf edu Last modified: 3/2/2002 Right click here for Notes. Expression Dataset Nome: HeLa Cell Ncvv Legend ENDOTHE LIN-CONVERTING ENZYME ' 11.2427 ECE1 ECE2 GELATINASE A I MMP2 11.4157 GELATINASE B | MMP9 11 5439 MLN 62. CART1 1 3499 TLL 0 39 TLL2 INSULYSIN I IDE |o.7915 ;2 .1631 1.7047 TRAF3 TRAF4 TRAF6 MACROPHAGE ELASTA6E | E I A 2 | l 6139 I MMP12 j l 0235 MATRILYS1N COLLAGENASE MMP7 MEP1A 0.8781 Upregulat ion Oownregulat ion No change No cri tar i a mat Not found CARB0XYPEPTIDA5E A I ! 79 ; I CPA1 CPA3 CP AH CPB1 CPD CPE CPM CPM 2.1771 0 4002 3.0434 0.1649 CPN1 2 2731 DJ860F19.3 B Sham (Days pi) 9 3 CVB3 (Days pi) 30 165 Figure 66: Metalloproteinase gene profile in CVB3-infected HeLa cells. The experimental protocol is illustrated and described in Figure 26 B and D. A. Signal intensity data was converted into a text format in Microarray Suite 5.1 and normalized in GeneSpring 5.0.1™ and submitted to GenMAPP 1.0. Signal intensity ratio for virus/sham cut-off was arbitrarily set at <=0.5 for downregulation (blue), >0.5 and <1.5 for no change (yellow) and >1.5 for upregulation (red). Gene labels are Swiss-Prot designations. Metalloproteinases and their inhibitors are transcriptionally up regulated in CVB3-infected HeLa cells 9 hours post-infection. (B) There is an increase in metalloproteinase genes during the inflammatory and chronic stages of infection in CVB3-infected mouse hearts (logio, compared to baseline controls). 166 Hs_cytochrome P450 AtttMor: Adapted from Gene Ontology Maintained by: GenMAPP org H-iirmH: genrnapp@glacist0ne-ucsf.edu Last modified: 3/2/2002 Right click here for Notes. CYTOCHROME P450 |CYP11A 1.1143 :CYP1A1 1 0061 JCYP1A2 0.846? JCYP1B1 0.558 | CYP24 1.0206 £YP2A13 1.2051 CYP2A7 1.6344 CYP26 CYP2B6 0.7236 CYP2C10 CYP2C18 0.6849 CYP2C19 1.3083 CYP2C8 2.9075 CYP2C9 CYP2D6 CYP2E1 2.2066 CTP2F1 1 8944 |CYP2G1 HOMOLOG] CYP2S1 CYP3A3 1.6419 CYP3A4 3.2732 CYP3A5 3.9393 CYP3A7 7.9692 CYP1B1 1.4362 ARACHI0ONIC ACID EPQXYGENy •"OrPiC 12.3741 AROMATASE ; CYP19 j 1-2787 ra\ r i n i n i i . x n t j n n i f v n F N t ' Expression Dataset Name: HeLa Call New Legend TBXAS1 1.2725 JCYP27B1JH31 CHOLESTEROL 7-ALPHA-MONO OXYGENASE | CYP7A1 J0.533 COUMARIH 7-HYDROXYLASE fCYPZAfi -1.0357 DOPAMINE-BETA-MONO OXYGENASE | DBH |i.*7u FATTY ACID (OMEGA-1J-HYDR0XYLASE Upregulation Dcwnregula l ion No change No cr i ter ia met Not found t Y P l A 1 l b 7405 LAN0STER0L 14-ALPHA-DEMETHYLASE | CYP51 |0.8213 0XYSTER0L 7-ALPHA-HYDR0XYLASE I CYP7B1 |0.7459 PROSTAGLANDIN-I SYNTHASE RETINOIC ACIO 4-HYDROXYLASE 1.3912 |CYP: 26A1 -STEROID 11-8ETA-MQNQOX YGENASE icypiJBip.2809 |CYP11B2|2.9773 STEROID 17-ALPHA-M ONG OXYGENASE ; CYP17 ;7.164.3 STEROID 21-MQNOOXYOENASE CYP21 ;CYP21AZ;2.2416 STEROL 12-ALP HA-HYDROXYLASE [ C Y P 8 B 1 | THROMBOXANES SYNTHASE l f W i * l I 167 Figure 67: CYP gene profile in CVB3-infected HeLa cells. A . The experimental protocol is illustrated and described in Figure 26 B and D. Signal intensity data was converted into a text format in Microarray Suite 5T and normalized in GeneSpring 5.0.1TM and submitted to GenMAPP 1.0. Signal intensity ratio for virus/sham cut-off was arbitrarily set at <=0.5 for downregulation (blue), >0.5 and <1.5 for no change (yellow) and >1.5 for upregulation (red). Several cytochrome P450 genes are transcriptionally up regulated in CVB3-infected HeLa cells at 9 hours. B. Hearts from CVB3- and sham-infected mice were harvested and pooled at 3, 9 and 30 post-infection (N=4 animals but N=3 for day 30 virus) and hybridized to custom mouse GeneChip® arrays as described in Figure 26. There is an increase in C Y P l b l gene during the inflammatory and chronic stages of CVB3 infection (loglO, compared to sham controls). 168 4.3 Discussion L GeneChip® array-based profiling of CVB3-infected HeLa cells Recently, early direct virus injury to target myocytes in viral myocarditis has been appreciated [85, 276] . To better understand the molecular mechanisms of host cellular responses to viral injury, a well-characterized in vitro (human cervical carcinoma) HeLa cell model of C V B 3 infection was utilized. Insights into coxsackievirus pathogenesis have come from human studies, cellular and animal models, as well as from comparator viruses such as the closely related poliovirus. The ease of manipulation and rapidity of growth of HeLa cells make them a useful model to study host signalling mechanisms in C V B 3 infection. In fact, key discoveries such as isolation of the C A R receptor [50, 277] ( D A F was discovered in RD cells [51]) and viral cleavage of dystrophin [25, 26] were originally made in HeLa cells with subsequent verification in vivo. With any experimental model there are caveats in translating to human disease. Table 6 outlines important characteristics and considerations that were used to decide on experimental models. Notably, recent evidence suggests comparable signalling responses to C V B 3 infection between experimental models. As alluded to earlier, E R K signalling was shown in C V B 3 infection of HeLa cells [64] , isolated cardiac myocytes, and in hearts of myocarditis-susceptible A / J mice [66]. Triggering of N F K B translocation occurs in infected HeLa cells (Luo et al, unpublished observations) and inhibition in vivo can block virus infection and limit tissue injury [73]. Similar apoptotic signalling has been well-characterized in HeLa cells [85], transformed atrial myocytes [84] and in the heart (Yanagawa et al, unpublished observations) during C V B 3 infection. There are enough similarities in cellular responses between HeLa cells, atrial myocyte cell lines, T-cell lines, isolated cardiac myocytes, in vivo mouse models and human myocarditis warrant their use as surrogate models. 169 Properties Human Intact Mouse (+/- genetic modification) Murine Myocyte Surrogate (HeLa) "Humanoid" +++++ +++ +++ ++++ Integrative Pathogenesis +++++ +++++ ++ ++ Ability to Isolate Host Cell and Virus Interplay ++ ++ +++++ +++++ Feasibility of Pathological Studies + +++++ +++++ ++++ Receptor Expression: CAR +++++ +++++ +++++ +++++ DAF +++++ +* +* +++++ Evidence of Comparability 9 • ERK activation ERK activation ERK activation Table 6: Assessment of surrogate models systems for human enteroviral heart disease. I provide a critical analysis of models of enteroviral infection including human samples, mouse, cultured myocyte and transformed cell cultures. Salient properties are graded from + to +++++. The consistent finding of E R K activation in CVB3 infection of mouse hearts, myocyte and HeLa cells provides evidence for their comparability (ERK activation in human tissue has not been studied). * Coxsackievirus B3 is unable to bind murine. 170 Initially, the transcriptional profile in CVB3-infected HeLa cells was probed from 0-9h post-infection. Beginning at initial virus-receptor attachment to caspase-3 activation, substrate cleavage to the beginning of host cell apoptosis (see Introduction for a review of apoptosis) [85]. The transient nature of most differential gene expression suggests that specific biological stages in the virus lifecycle may be responsible for triggering these events. In the virus lifecycle, as early as one hour pi, viral proteases can selectively cleave host proteins [85] which may contribute to early differential transcription. Expression of collagen genes is decreased as early as l h pi, which suggests that either virus-receptor interactions or viral protease cleavage events may be responsible. From 3h pi active virus replication with shut-off of host translation occurs. Finally, at 7 to 9h, molecular checkpoints of host cell apoptosis are triggered and cells begin to exhibit cytopathic effects. Most differential transcriptional alterations occurred at 9h pi. Expression of oncogenes and pro-survival signal transduction genes may represent a host response to maintain cell viability during peak virus replication. Studying differential transcriptional events in the context of viral lifecycle stages is useful to create rational hypotheses about which specific viral events are responsible for host transcriptional changes. Overall, due to considerable noise in control expression data, the expression intensities were averaged over all time points prior to analysis. This set of genes with differential regulation during the course of CVB3 infection afforded greater confidence. Genes with both high signal intensity and large differential expression changes were highlighted as likely real events. Such interesting genes include the protease, MMP11; signalling molecules, Ras-like protein Tc21 and MAPK-activating protein kinase 2; and the growth factor, CTGF. CTGF, a cysteine-rich C C N family protein induced by TGF-P originally isolated from human vein umbilical endothelial cells, is of particular interest [269]. Emerging Questions 1: What is the impact of CTGF signalling in CVB3-infected hosts? CTGF triggers cellular processes including fibrosis, cell proliferation, adhesion, migration, and the synthesis of extracellular matrix. I show up-regulation of CTGF in both C V B 3 -171 infected HeLa cells and in mouse hearts (data not shown). Confirmation by two array methods and one non-array-based method affords confidence in the validity of increased CTGF expression (Figures 51, 53, 60). Increased CTGF transcript and protein is found in myocytes, fibroblasts and myofibroblasts in marginal zone infarcts where it mediates the TGF-P fibrotic response in the heart [271, 272]. Recently, Coppola et al used human Affymetrix muscle array to show log2 fold increase of 2.9 in CTGF in canine pacing-induced heart failure as shown on the HOPGENE website (supported by the NHLBI , NIH Programs for Genomic Applications [PGA]; http://www.hopkins-genomics.org). The interaction of CTGF and MMPs provides a connection to matrix synthesis and metabolism [278]. CTGF can orchestrate many cellular processes and understanding the downstream targets will greatly increase our understanding of viral pathogenesis. CVB3-infected HeLa cell samples exhibit up-regulation of c-fos transcripts at early and late time points, as shown previously with Echovirus-1 infection [279]. The AP-1 transcription factors control cellular proliferation, survival and death. AP-1 proteins are homo and heterodimers composed of basic leucine zipper motifs and include c-fos, c-jun, the ATF protein and the inhibitory jun B, among others. These genes are up-regulated by the M A P K pathways, Jak/Stat, among others. Infection of HeLa cells with CVB3 induced both biphasic activation of the M A P K - E R K and PI3K/Akt pathways [64] (Luo et al, unpublished observations). The importance of E R K and Jak/Stat to promote CVB3 replication and pathogenesis in host cells has been demonstrated but the mechanisms are unknown. One major downstream target of AP-1 proteins is cyclin D l although there was no change in cyclin D l mRNA levels in the dataset and in our previous results following infection [43]. Studies into groups, pathways and networks of genes have provided important insights into biological processes involved in viral heart disease. I showed altered expression of genes involving the cell cycle, integrins, M A P K signalling pathways and oncogenes, among many others up-regulated in CVB3-infected HeLa cells. Up-regulation of serine-threonine kinases and integrin-linked receptors suggests their involvement in host cell signalling. 172 The increase in oncogenes may reflect, on a molecular level, the failing heart as a culmination of balanced hypertrophic fetal gene re-expression, apoptotic responses and 'tumor-like growth' target cells common to cancer biology [280]. Our laboratory and others have shown re-expression of fetal genes, including IFN-inducible protein-10 and IGTPase [125], which are characteristic of tumorigenesis. There are differences between genes expressed in hearts and HeLa cells infected with CVB3. Such differences are expected for the following reasons: (1) Myocarditic hearts include many different cell types including cardiac myocytes, vascular endothelial cells, fibroblasts and activated myofibroblasts, smooth muscle cells and infiltrating cells. (2) Heart tissue undergoes biological processes such as vasospasm, inflammatory infiltration, tissue repair, hemodynamic stress, as well as direct infection where as CVB3-infected HeLa cells represent immediate host response to infection and possibly paracrine-like cytokine effects. (3) The in vivo time course investigates transcriptional changes during viremic (day 3), inflammatory (day 9) and reparative (day 30) stages of disease whereas the in vitro time course focuses on the early interaction from 0 to 9h pi. (4) The datasets are not easily comparable as in vivo experiments utilized GeneChip®s with 25, 204 mouse genes whereas the in vitro experiments utilized U95A GeneChip®s with 12, 627 human genes. (5) Cultured cells exist in very different cellular environments than their in vivo counterparts. In particular, tissue cells have ready access to nutrients and oxygen, live in an uninhibited two-dimensional environment with complete freedom to grow and exist in a unicellular milieu. This is clearly not the case in vivo, even in a relatively well-perfused organ such as the heart. (6) Differences in genes represented between the data sets exist from lack of mouse and human gene homology and also gene selection bias. For these reasons, observed differences between the in vivo and in vitro data are not surprising. Future comparative genomic analyses can provide greater insights to these differences. Genomic researchers have realized that with the biological variability associated with the transcriptome and technical noise inherent in microarrays that some form of replication is desirable, as recently reviewed by Churchill [281]. Replication is often necessary to probe 173 low copy number genes. These tightly-controlled transcripts may code for key regulator proteins. Therefore, I repeated the CVB3 infection time course with three biological replicate GeneChip®s at key viral time points and with an additional focus on elucidation of the transcriptional targets of the M A P K - E R K pathway. it Triplicate gene profiling of CVB3-infected HeLa cells with MEKl inhibition The in vitro time course was repeated with triplicate arrays from separate experimental samples. To my knowledge, this study is the only triplicate time course Affymetrix GeneChip® time course analysis of any disease process infectious or otherwise. A graph of all genes confirms our previous observation that most differential expression events occur at the 9h time point. Results from a) function-based filtering and b) expression-based clustering approaches will be discussed. Davies et al [270] have performed an extensive study to determine the selectivity and efficacy of 28 commercially-available inhibitor compounds against a large panel of protein kinases. The compound U0126 is a commonly used E R K pathway inhibitor [282] which at lOpM inhibits M E K l by 44±1% and SAPK2a/p38 by 25±1% (as compared to untreated control) but not other serine/threonine-specific protein kinases. Work from our laboratory has shown that p38 is not activated in CVB3-infected HeLa cells (unpublished observations). The recently developed compound PD184352 (lOuM, ParkDavis) blocked M E K l by 95%±1% without inhibiting other relevant kinase pathways. These experiments were carried out with U0126 based on previously published work using this particular compound, but it would be interesting to comparison with the transcriptional results from PD184352 and other M A P K - E R K inhibitory molecules would be interesting. a. Function-based gene profiling Many genomic trends were studied based on their biological role but I will limit my discussion to the following three function-based transcriptional trends: (1) down-regulation of 'intracellular genes' and differential regulation of (2) 'virus-related' and (3) 'signal transduction' genes. 174 1. Down-regulation of intracellular genes, including mitochondrial, lysosomal and cytoplasmic, occur late during the infectious process. A specific example is provided by the H+-ATP synthase, the main enzyme responsible for the formation of ATP in mammalian aerobic cells. Down-regulation of key metabolic enzymes is observed both in CVB3-infected HeLa cells and mouse hearts. Such transcriptional changes may alter substrate specificity and other metabolic deficiencies previously shown in cardiomyopathic tissues [204]. 2. Late differential gene expression is seen in virus-encoded genes following CVB3 infection. These genes include v-jun, v-abl and v-fos oncogenes and the baculovirus encoded inhibitor of apoptosis (IAP) gene. As alluded to earlier, there is in vitro and in vivo data to suggest increased presence of AP-1 transcription factors during CVB3 infection. The functional role of these transcription factors are likely complex and may target transcription of downstream genes, some of which may be involved in apoptosis. The Abl proto-oncogene encodes a non-receptor tyrosine kinase positively regulated by D N A damage signals to regulate cell cycle checkpoint, D N A repair and apoptosis [283]. A similar human oncoprotein, Bcr-Abl, is critical in the pathogenesis of chronic myelogenous leukemia [284]. Abl can activate proteasome-dependent protein degradation [285], induce PI3K activation [286] and associate directly with Jak to induce Jak/Stat signalling [287], all observed during CVB3 infection. It will be interesting to determine what contribution, if any, Abl has to these and other processes in the setting of CVB3 infection. The IAP response is induced during caspase-dependent apoptotic [288]. These genes were first identified in Cydia pomonella granulosis virus, a baculovirus that can subvert apoptosis in infected cells [289]. IAPs are also involved in N F K B signal transduction [290], cell cycle regulation [291] and putatively protein ubiquitination [292]. A 5'-IRES motif regulates XIAP-mediated cytoprotection, thus this gene is likely to be translated in the setting of cap-dependent translational shutdown as seen during CVB3 infection [293]. 175 3. Signal transduction ligand expression is up-regulated in CVB3-infected HeLa cells. There is marked increases in CTGF and IGF-2. Although IGF-2 has not been studied in coxsackievirus infections, there is considerable in vitro and in vivo evidence to support a role for the IGF-1 and -2 in promoting myocyte survival and improving myocardial function during ischemic injury [294, 295]. b. Expression-based clustering Gene clustering was utilized to fdter genes that were up-regulated at 9h pi but not at 30m or 3h. Genes were analysed to determine functional trends in gene expression and identify specific genes. These genes encode mainly signalling proteins such as M A K P , serine/threonine kinases and dual-specificity protein phosphatase and transcription factors such as TFIID, ATF3, ETR101 and General TFIIB. I will focus my discussion to the ubiquitous stress response element, (activating transcription factor 3) ATF3, which is up-regulated at 9h pi [296]. This gene is transcriptionally up-regulated in most stress responses [297]. ATF3 is an early cellular stress response to virus infection, U V and IR ionizing radiation, proteasome inhibition, and growth factor deprivation [298]. This gene can be transcriptionally repressive or active when bound to c-jun and has a self-repressing, negative feedback mechanism partially explaining its transient expression [299]. The interesting possibility that ATF3 and other genes are part of a common response to many external stimuli supports the need for databases of gene profiles. The analysis performed here was useful to isolate a group of genes based on expression profile, then to perform a functional breakdown and illustrate key genes of interest. The back-end clustering tool (described in Materials and Methods) and function-based clustering scripts as well as the pie-chart visualization tool was developed for this work in collaboration with researchers at the Faculty of Computer Science, UBC. These programs can now be used for the analysis of any set of microarray data. 176 Downstream genes of MAPK-ERK Interest in M A P K - E R K as a central pathway that orchestrates a wide range of host cell responses begs the question: what are the downstream target genes transcribed post-CVB3 infection? To address this question, cells were treated with a specific M E K l inhibitor (upstream kinase of ERK) prior to CVB3 infection. By uncovering downstream genes which are expressed following infection but repressed with the M E K l inhibitor, I identified target genes and increased our molecular understanding of the apparent pro-virus effect of E R K signalling in target cells and organs. a. Expression-based clustering Expression-based clustering helped to identify genes which were up-regulated during virus infection and then suppressed by U0126. Few genes in Figure 62 are known direct downstream gene targets of ERK. It is likely that many genes are transcribed downstream transcription factor thus triggered indirectly by ERK. There are four C Y P genes which are down-regulated with U0126 pi and will be discussed in detail later. A cancer-related gene, GROp, is down-regulated by U0126 treatment during CVB3 infection. As alluded to earlier, host response to several infectious agents, both viral and bacterial, have been subjected to microarray analysis [262, 300, 301]. Such studies have shown that GROp (M57731), IL-8 (M28130) and leukemia inhibitory factor (X13967) are upregulated, suggesting that they may be part of a stereotypic transcriptional response to diverse infectious agents including Salmonella, Listeria monocytogenes and CVB3 [300, 301]. Since GROP is released in response to inflammation as a chemoattractants for several leukocytes it may have important roles following CVB3 infection. b. Function-based gene profiling To identity genes responsible for the 'pro-viral' effect of ERK, the differential gene profiles of known E R K downstream genes were probed in the presence an absence of M E K l inhibition after virus infection. Genes were grouped into transcription factors, signalling proteins and cytoskeletal proteins, among which primarily transcription factors 177 experienced U0126-induced down-regulation at 9h pi. As regulatory proteins which can influence the transcription of many genes within its regulon, transcription factors are particularly important targets of E R K signalling. The up-regulated E R K transcription factors include Ets translocation variant 1, NF-IL6-(3, c-jun, c-fos, jun B and EPH-related receptor tyrosine kinase ligand 6 pi. Interestingly, c-fos and jun B are both transcription factors which were originally discovered as virally encoded genes. Chicken c-jun message can be translated through a cap-independent IRES process which increases the likelihood that this protein is translated cap-independently [302]. NF-IL6 is a key nuclear factor that activates gene expression in response to IL-6, which is induced in viral myocarditis, and controls HIV-1 gene expression [303, 304]. Despite differences in viral strategies (RNA virus versus retrovirus), the involvement of NF-IL6 with HIV-1 transcription suggests that it may also regulate enterovirus replication. This combination of hypothesis-driven exploratory strategy is a powerful approach to investigate a specific aspect of viral pathogenesis. The following section describes genes and gene sets of interest in CVB3-infected HeLa cells as targets for emerging questions in CVB3 pathogenesis. Emerging Questions II: What is the role of serpins in CVB3 infection? The serpin superfamily of serine proteinase inhibitors has a central role in controlling proteinases in many biological pathways. Myocardial infection with encephalomyocarditis virus induces 150% increase in myocardial serine elastase activity [305]. Treatment with a serine elastase inhibitor reduced microvascular constriction, inflammation, tissue necrosis and reduced collagen deposition in myocarditis [305]. A recombinant serpin attenuates ischemia reperfusion injury by inhibiting neutrophil-accumulation into the ischemic-reperfused myocardium and by inactivating cytotoxic metabolites (proteases and superoxide radicals) released from neutrophils [274]. During inflammation, I show increased serpin expression both in vitro and in vivo. As such it may be an endogenous mechanism to counter inflammatory and scar tissue formation. There is also direct evidence for the serpin CrmA in inhibition of mitochondrial apoptosis in ventricular 178 myocytes during hypoxic injury [275]. Coxsackievirus B3 has been shown to induce mitochondrial release of cytochrome c and serpins may have a similar anti-apoptotic function in this model. Emerging Questions III: How do extracellular matrix and MMP alterations contribute to chronic cardiac dysfunction in CVBS-infected hearts? The cellular myocardium is surrounded by a gel-like extracellular matrix (ECM) made up of collagen, glycosaminoglycans, glycoproteins, basement membranes and other molecules. The collagen matrix provides structural integrity and myocyte alignment, both of which are critically important for pressure distribution, electrical conductivity, cytokine and growth factor distribution and micro vasculature support [306-308]. Collagen can be broadly grouped into type I thick fibrillar collagen [309] and type III thin collagen, the abundance, organization and ratio of which may shift in disease states. Perturbations in E C M protein architecture and alterations in the collagen ratio are implicated in myocardial remodeling with hypertension, hypertrophy, ventricular tachycardia, ischemia, dilation and senescence [310-314]. Extracellular matrix alterations during myocarditic injury have been well documented and are important to the deterioration of heart function. I have shown increased collagen mRNA in myocarditic hearts and histological collagen accumulation in former necrotic zones (Figures 34, 35, 36). The other arm of the E C M is in degradation. Matrix metalloproteinases (MMPs) are a family of zinc endopeptidases which metabolize specific E C M substrates. Direct evidence for M M P contribution in myocardial disease comes from MMP-1 overexpressing transgenic mice, which exhibit hypertrophy, increases in collagen III transcription, wholesale collagen degradation and loss of systolic and diastolic function [315]. Experimental models of ischemic [316-319] and non-ischemic cardiomyopathy [320-322] have also shown M M P activity. I show an increase in the MMP-12, -3, Adam-8 and TIMP-1 gene expression in vivo and several MMP-1 , -3, -7 and -9, Adam-12, 21 and 28 in vitro. Specifically, a significant and sustained chronic increase in macrophage metalloelastase MMP-12 was shown, suggesting chronic macrophage activation. There is growing interest in the role of myocardial MMP-12 in atherosclerotic 179 lesions [323]. Recently, L i and colleagues [324] showed the presence of several collagen subtypes during acute viral myocarditis. We have also recently confirmed increased activation of MMP-3 and -9 and decreased T1MP-3 protein in CVB3-infected hearts (Cheung et al, unpublished observations). In this regard, I hypothesize that extracellular proteases matrix and concurrent production of E C M production leads chronically to interstitial fibrous tissue accumulation which can impede heart function (Figure 68). Altered regulation of MMPs and their inhibitors in myocarditic hearts suggests that they may be targeted therapeutically to improve heart function during the wound-healing response. Emerging Questions IV: What are the roles of cytochrome P4'50s in CVB3 infection? Cytochrome P450 (CYP) was first discovered in 1958, so named because the unique 450-nm optical absorption peak and the enzyme's hemoprotein nature [325]. This superfamily of enzymes is crucial for the oxidative, peroxidative, and reductive metabolism of steroids, fatty acids, prostaglandins, and most therapeutic drugs and environmental pollutants. Little published work exists on C Y P activity in the heart, although Granville et al [326] showed that inhibition could attenuate ischemia/reperfusion injury. In the mouse heart, C Y P l b l is found to be up-regulated particularly during the inflammatory stage of enterovirus infection. C Y P l b l is a newly identified C Y P gene family that is found to be expressed in several normal human tissues including the heart [327]. In CVB3-infected HeLa cells, several C Y P genes including CYP3A7, 17, 11B1 and 11B2 were increased. These proteins may be released systemically to affect liver metabolism of drugs, foreign chemicals, arachidonic acid and eicosanoids, cholesterol, or have yet undiscovered functions. Few groups have studied C Y P expression in the injured heart, virally or otherwise, thus they represent an understudied and potentially important metabolic process in heart disease. 180 Hypothesis: Altered expression of matrix metalloproteinases and ECM molecules leads to chronic myocardial stiffness in CVB3-infected mouse hearts. MMP-3,-7,-9, -12 cathepsins degrade ECM ECM breakdown products induces ECM Degradat ion Fibroblast collagen production Observed increase in interstitial fibrosis and decrease in heart function ECM Production: collagen fibronectin ECM protein 2 ECM Product ion Dilated Card iomyopathy Figure 68: The extracellular matrix alteration hypothesis. Based on gene profding investigations in CVB3 infection of mouse hearts and HeLa cells, we hypothesize that altered expression of matrix metalloproteinases and specific E C M molecules leads to chronic myocardial stiffness in CVB3-infected mouse hearts. 181 CHAPTER V: Conclusion and Future Work 5.1 Conclusion Microarrays are broadly used for gene discovery, disease prognosis, disease diagnosis, drug discovery and fundamental pathogenesis studies. This work describes the profdes of gene expression during enterovirus infection of cultured cells and heart tissue, through the combination of array-based and gene-specific profiling methods. To facilitate the analysis, I applied established bioinformatical tools and developed an analysis and visualization program to dissect out time-dependent events at different levels of functional annotation. I will review the most important questions which have arisen from this work and from which targeted biological confirmatory investigations already underway. (1) Following CVB3 infection, I identified acute increases in the expression of murine S100A subtypes 6, 10 and 11 binding protein transcripts in mouse hearts. These genes encode proteins important in regulation of transient intracellular Ca2+-dependent signalling mechanisms and have been circumstantially linked to roles in heart disease [224, 227]. Increased expression of SI00 genes may help to restore calcium homeostasis during viral infection [231]. (2) There are also increases in muscle L I M protein genes, which suggest that they may be involved in alterations of cardiac structure or function following enterovirus infection. MLPs bind the contractile apparatus to junction proteins in the cellular membrane [232, 233]. Alterations in MLPs have been associated with cardiomyopathies [235-238]. Considering the importance of M L P to the cytoskeleton and cardiac contraction, modulation of these genes may influence chronic heart function. To this point, I characterized for the first time, using two-dimensional echocardiography, heart function as measured by ejection fraction and wall thickness in CVB3-infected hearts. During the immune stage of infection, hearts exhibited systolic and diastolic L V wall thickening, and chronically, a significant reduction in ejection fraction. To my knowledge, 182 this is the first time that functional changes have been described in this experimental myocarditis model. Together with histological characterization performed, the C V B 3 -infected A/J mouse histologically and functionally resembles human viral myocarditis [5]. These findings support the use of this model for infectious dilated cardiomyopathy using a measurable functional endpoint. (3) I show up-regulation of complement genes Clqa,b, C2, C3, C4, H2-Bf and Pfc, in CVB3-infected mouse hearts, which suggests that they may contribute to immune activation in inflammatory heart disease. Complement component factor B increases were further confirmed by cDNA and RT-PCR. Complement triggers the cellular immune system and enhances antibody production, as well as directly destroying pathogens and pathogen-infected cells [5, 239-241]. Increased expression of classical and alternative complement pathways likely play an important role in triggering antibody-mediated immunity in CVB3-infected mouse hearts. (4) Increased expression of cathepsin S, K, Z, C, P and L genes was observed in myocarditic hearts, which could have protective roles in enterovirus infection. The potent lysosomal cysteine and aspartic protease cathepsin L (CTSL) protein was also increased and co-localized to myocytes and inflammatory cells. Considering that cathepsins degrade collagen and elastin [243] and participate in antigen proteolysis for M H C II presentations [244], the observed elevation of expression may be part of a host protective response to inhibit virus infection. (5) I show that gene expression of heart shock proteins (HSPs) 27, 70, 86 is up-regulated acutely. Strong and early up-regulation of HSP27 was confirmed using RT-PCR and immunohistochemistry. HSPs are expressed in response to stress and are cardioprotective in myocardial ischemia-reperfiision injury [248-250]. In enterovirus myocarditis, increases in HSPs may act as molecular chaperones, reduce reactive oxygen species which are injurious to protein folding and D N A integrity, and prevent apoptosis [251]. 183 (6) The peripheral-type benzodiazepine receptor (PBR) interacts with the voltage-dependent anion channel and regulates permeability on the outer mitochondrial membrane. In enterovirus-infected mice, PBR gene and protein expression peaked at day 9. It is possible that PBR transcription was induced in cardiac myocytes to preserve mitochondrial integrity considering PBR protein expression was localized to myocytes, and the interaction of PBR, the voltage-dependent anion channel and the adenine nucleotide carrier [253]. (7) Global changes in myofibrillar genes are observed in major three trends: acute increases in a-actin and a-actinin, chronic increases in smooth muscle a-actin and y-actin, and no change in P-actin. The differential expression of myofibrillar genes may alter the contractile and matrix makeup of the heart and impinge on ventricular pump function. Our in vitro Affymetrix GeneChip® experiments revealed target cell genomic responses to CVB3 infection outside the complexity of the in vivo approach. In this cell culture system, I also show novel differential transcriptional events from which the following novel hypotheses were developed. (8) Several serpins including types g l , 2, f l , 3c, 3g, 3n were markedly up-regulated in CVB3-infected HeLa cells and myocarditic tissues. The serpin superfamily of serine proteinase inhibitors have a central role in controlling proteinases in many biological pathways. Coxsackievirus B3-induced up regulation of serpins may inhibit mitochondrial apoptosis in ventricular myocytes [275]. (9) I show increases in extracellular matrix and MMPs in enterovirus-infected HeLa cells which may contribute to chronic cardiac dysfunction in CVB3-infected hearts. I have also shown increased collagen mRNA in myocarditic hearts and histological collagen accumulation in former necrotic zones. Collagen perturbations in E C M protein architecture and alterations in the collagen ratio are implicated in myocardial remodeling with hypertension, hypertrophy, ventricular tachycardia, ischemia, dilation and senescence [275, 310-314]. The other arm of E C M metabolism is in degradation by MMPs. I show an 184 increase in the MMP-12, -3, Adam-8 and TIMP-1 gene expression in vivo and several MMP-1 , -3, -7 and -9, Adam-12, 21 and 28 in vitro. There is strong evidence for M M P contribution in myocardial disease [316-322, 335]. It is likely that extracellular proteases matrix and concurrent production of E C M production leads chronically to interstitial fibrous tissue accumulation which can impede heart function. (10) Finally, I show increased gene expression of cytochrome P450 3A7, 17, 11B1 and 11B2 in CVB3-infected HeLa cells. This superfamily of enzymes is crucial for the oxidative, peroxidative, and reductive metabolism of steroids, fatty acids, prostaglandins, and most therapeutic drugs and environmental pollutants. C Y P l b l is also up-regulated in vivo particularly during the inflammatory stage of enterovirus infection. Few groups have studied C Y P expression in the injured heart, virally or otherwise, thus they represent an understudied and potentially important metabolic process in heart disease. In this thesis, I used two microarray platforms as well as non-array assays to gain a greater understanding of the underlying host protective and remodeling responses in enteroviral heart disease. For the first time, I characterize functional deterioration in CVB3-infected mouse hearts which reflects the phenotype of human myocarditis and provide potential molecular explanations. In my experience, microarrays and bioinformatical tools have been very useful in generating rational and testable emerging questions. 185 5.2 Future Work The future directions are divided into (1) direct extensions of this work and (2) a look towards the field of genomics and proteomics in studies of enteroviral infections. Direct extensions Our findings suggest that the A/J model closely resembles the human disease functionally. As echocardiography is non-invasive, serial echo measurements on mice infected with CVB3 would extend my initial observations of chronic functional deterioration. Determination of the functional endpoint of these animals by extending our original study to 60 and even 120 days pi would be an important extension of this work. Such studies are crucial to fully characterize the CVB3-A/J mouse model for infectious dilated cardiomyopathy. This model could be used to test the efficacy of therapeutic compounds to restore chronic myocardial function following acute myocarditis. This thesis work is part of a larger investigation in our laboratory to dissect out the mechanisms of viral injury in host cells. My work has revealed differentially regulated genes in two surrogate experimental models of human myocarditis and has initiated new investigations into the most promising leads to confirm and to understand their functional outcomes. The transcriptional events outlined here will be investigated in a three step manner: first transcript increases will be verified using non-array-based approaches; then protein expression must be characterized; and, finally, functional outcome in viral heart disease will be investigated (Figure 69). The functional studies will be tailored to the specific gene/protein of interest and are facilitated by the availability of powerful gene and protein over-expression and exclusion technologies. Transgenic mice expressing target genes using an MHC-promoter can be exploited to express proteins in vivo and Tet-On expression systems can be used to transiently express proteins in cell culture. Recently, high efficiency protein transduction has been achieved in cell culture and in ex-vivo hearts by linking the 11-amino-acid transduction domain of HIV TAT to target proteins [336]. To exclude genes, small 186 interfering RNA-mediated gene silencing is used in vitro and knockout animals can be developed for in vivo studies. In this regard, Baygenomics (http://baygenomics.ucsf.edu) investigators (supported by the NHLBI , NIH PGA) are utilizing a gene-trap technology for large-scale creation of mutant mouse embryonic stem (ES) cells for the purpose of generating knockout mice. There are currently 2530 ES cell lines including murine v-Abl and the following ubiquitin/proteasome-related genes of interest to this thesis work: (Proteasome [prosome, macropain] 26S subunit) Psmdl, 12, 14, ubiquitin protein ligase, ubiquitin specific protease and ubiquitin-associated protein (latest access date: August 18 th, 187 M i c r o a r r a y A n a l y s i s B i o i n f o r m a t i c s : G e n e ID G e n e C o n f i r m a t i o n : C T S L RT-PCR Northern Blot Targeted Array Day 9 pi P r o t e i n C o n f i r m a t i o n : C T S L Immunohistochemist ry ELISA Western Blot s V Day 9 pi F u n c t i o n a l s t u d y o f C T S L in C V B 3 I n f e c t i o n In vitro studies: - Transfection of HeLa cells, cardiomyocyte cell lines and freshly-isolated cardiomyocytes - Tet-On overexpressing HeLa cells In vivo studies: - Genetically modified mouse models (KO and trangenic) - Small interfering RNA Cathepsin L transcriptional inhibition 188 Figure 69: The experimental strategy from gene profiling to functional understanding. Gene profiling has can lead to new experimental leads using the following approach. Microarrays can be used to study thousands of gene expression events following CVB3 infection. Bioinformatical tools are helpful to highlight genes with expression patterns of interest such as the cathepsin L in CVB3-infected mouse hearts. Based on the potential biological interest of such a gene, non-array based assays can then be used to confirm expression at the gene and the protein levels. Positive confirmation may suggest importance and lead to the investigation of protein localization, activation and functional role in the disease of interest. Cathepsin L gene and protein expression were shown to be increased in CVB3-infected mouse hearts and experiments to dissect out the functional role in viral infection are planned. 189 2003). I hope that the emerging questions outlined in this thesis inspire others to take up these exciting investigations where I have left off. Future directions for genomics In pursuit of a greater understanding of coxsackievirus pathogenesis and myocarditis, future well-planned microarray experiments can answer important questions such as: what are the differences in transcriptional signatures of two related viruses, such as CVB3 and CVB4, in a cell system related to differences in pathogenesis in vivo? What are the transcriptional events in cells which are susceptible to infection but range in capacity to promote viral replication? What are the transcriptional profdes in hearts of patients with myocarditis and how do these compare with myocarditis from mouse models? A better understanding of CVB3-induced viral myocarditis may include a list of all virus and host genes, their spatial and temporal expression during the various stages of infection, and their interactions in all target organs. To this point, a true virus chip which incorporates both host and virus genes to see the relative expression of both sets of genes together would be useful [262]. Such studies would enhance our understanding of how various viral strains induce differences in pathogenesis and help explain which genes are responsible for resistance to disease as shown in select populations [337]. A detailed understanding of coxsackievirus pathogenesis could catapult efforts to synthesize a non-pathogenic virus as a vehicle for targeted therapeutic drug delivery to target organs. Moreover, if the host gene response to CVB3 is specific, tissue microarrays combined with E M B x may be used as a diagnostic marker for infection by front-line physicians. Molecular re-classification of myocarditis based on gene profiles could be used for prognostic prediction or to tailor treatment regiments. Currently, the majority of genomic data is stored in individual public websites, in on-line publications which can be accessed directly from journal websites, and some data is stored in the NCBI database. The data generated from the four microarray experiments discussed here will be available on-line at the iCAPTURE Centre website 190 (http://wwwicapmre.ubc.ca/home.shtm following publication of these studies. In this regard, the creation of a public infection database as a repository for microarray data would facilitate cross-comparison microbial studies between cells under different challenges and different cells challenged by similar infectious agents. Major database and informatics hurdles would be to ensure uniformity in storing array-based information and consistency in data interpretation between diverse experimental conditions and proficiencies in array facilities. Proteomics as the study of the protein complement of the genome, was coined in 1995 [338] and has since grown to over 800 entries in the Web of Science database (Dec. 2001, http://www.isinet.com/isi/products/citation/wos/index.html). Genomic data has recently been complemented by proteomics approaches which are increasing in throughput and robustness. Currently, two-dimensional polyacrylamide gel electrophoresis and peptide mass fingerprinting method is the workhorse for protein identification [339]. Here, each stained protein spots is excised from a gel, digested, and applied to a mass spectrometer, such as a matrix-assisted laser desorption ionization (MALDl)-time-of-flight (TOF) mass spectrometer, and searched against a protein sequence database. Another recent development is the use of high pressure liquid chromatography, subsequent to 2D gel separation, to distinguish molecules with much lower masses (as reviewed in [340]). At the time of writing this thesis, a fore-runner technology is the isotope-coded affinity tag (iCAT) system that allows better relative quantification and concurrent identification [341]. With the refinement of such technologies, high-throughput translational investigation into enteroviral infection will significantly advance the findings presented here. 191 Glossary BLAST: y?asic Zocal Alignment Search 7bol. A heuristic sequence comparison algorithm used to search sequence databases for optimal local alignments to a query [328]. Bioinformatics: The application of computational techniques to analyze the information associated with biological problems [329]. Clustering: To apply unsupervised methods for organizing multivariate data into groups with roughly similar patterns. Clustering can be broadly separated into top-down agglomerative/hierarchical approaches which look for pattern similarities, or divisive bottom-up approaches, which look for pattern differences [330]. Data Mining: To sort through large volume data for identification of meaningful patterns and relationships. Data mining can be considered descriptive or predictive. The following are common analysis functions: Association: To look for patterns where one event is connected to another event Sequence or path analysis: To look for patterns where one event leads to another Classification: To look for new patterns Clustering: See above Forecasting: To discover patterns in data that can lead to reasonable predictions about the future DNA microarray: A microarray platform based on spotted cDNA probes first developed at the laboratory of Dr. Pat Brown, Stanford, CA [331]. Expressed Sequence Tag (EST): A partial sequence from the 5' or 3' end of a cDNA clone, obtained by performing a single raw sequence read from a random cDNA clone. The major source of publicly available ESTs is the EST database at the National Center for Biotechnology Information (NCBI) [332]. 192 GenBank: The National Institutes of Health (NIH) genetic sequence database containing an annotated collection of all publicly available DNA sequences and descriptions available to search [333] (http://www.ncbi.nlrn.nih.gov/). Gene Expression: A series of events by which the biological information carried by a gene is released and made available to the cell. Expression of an mRNA transcript may then be translated and further processed into a functional protein. GeneChip®: A High-throughput microarray platform based on oligonucleotide probes, also known as oligonucleotide (oligo) arrays and Affymetrix (Affy) arrays (Affymetrix, CA) [149]. Microarray: A low- or high-density set of cDNA or oligonucleotide molecules used for parallel hybridization analysis [168]. Normalization: Data transformation needed to identify and remove systematic sources of variation [144]. Transcriptome: The complement of genes which are expressed in a cell or collection of cells [334]. 193 Materials and Methods Virus Propagation Stock CVB3 was generously provided by Dr. Charles Gauntt (University of Texas Health Sciences Center, San Antonio, TX). Gauntt strain CVB3 is originally a Nancy strain virus isolated by JF Woodruff et al [165] previously characterized by our laboratory [166]. The full length Kandolf strain virus, also originally identified as a Nancy strain, has been previously characterized, cloned and sequenced [21]. Comparative infectious analysis by our laboratory revealed that Gauntt strain was particularly cardiotropic and was used for the in vivo studies. Kandolf strain was used previously by our laboratory for differential mRNA display experiments [126] and in vitro signalling assays [41, 43, 64, 125, 140] and as such was used for in vitro transcriptional studies. Both strains of virus were propagated in HeLa cells purchased from the American Type Culture Collection, stored at -80°C and titres were routinely determined at the beginning of all experiments. Virus Infection In Vivo Male, adolescent A/J mice were infected (intraperitoneal) with 105 plaque-forming units of a myopathic variant of CVB3 (Gauntt strain). Three levels of protection were used for protection of virus-infected animals and the user: the room air in the level 2 containment facility was HEPA filtered, animals were kept in a laminar flow shelf in a separate housing room, and cages were fitted with a microisolator top. Mice were observed daily for signs of morbidity and for any mortality. Animals that exhibited severe morbid signs including roughness of fur, skittish behavior, unresponsiveness to stimulation and self-mutilation were euthanized with CO2. At given time points, mice were euthanized with CO2 and the following target organs were removed in a rapid manner: heart, liver, pancreas, kidney and spleen. Tissues were either fixed in 10% formalin overnight for histological assays or snap frozen in liquid nitrogen and stored in -80°C. Unused animal tissues were autoclaved and deposited in the dead animal holding refrigerator for incineration by the Animal Research Facility staff, St. Paul's Hospital. A l l animal experiments have been approved by the University of British Columbia-Committee on Animal Care (Protocol Number: A0-0131). 194 Virus Infection In Vitro HeLa cells (American Type Culture Collection) were grown and maintained in Dulbecco's modified Eagle's media (DMEM) supplemented with 10% heat-inactivated fetal calf serum. CVB3 (Kandolf strain) was propagated in HeLa cells and stored at -80°C. Virus titer was routinely determined prior to infection by a plaque assay of HeLa cell monolayers as described below. HeLa cells were grown in complete medium, and upon reaching 80% confluence, cells were serum starved by incubation in serum-free D M E M and treated with DMSO or U0126 in DMSO for 24 h. For viral infection, growth-arrested HeLa cells were infected at a multiplicity of infection (MOI) of 10 with PBS and DMSO, CVB3 and DMSO or CVB3 and U0126 in DMSO for 40 min. Cells were washed with PBS and cultured in serum-free D M E M and treated with fresh DMSO or U0126 in DMSO. At given time points, medium was quickly removed and cells were washed in 1ml PBS. Cells and PBS were scraped into a 2ml screw cap tube, snap frozen in liquid nitrogen and stored in -80°C. Histology and Semi-Quantitative Histological Analysis Mid ventricular portions of heart specimens, as well as liver, pancreas, kidney and spleen, were formalin-fixed and embedded in paraffin, and 4um sections were cut and stained with either H&E or Masson's trichrome stain at the Clinical Histological Laboratory, St. Paul's Hospital. H&E sections were graded for the extent of myocarditis based on intensity and character of injury and inflammatory infiltration, as previously described [166]. Myocardial lesion area, cell death, calcification, and epicarditis were graded without knowledge of the experimental groups (BMM). Grades were based on the following scale: 0, no or questionable presence; 1, limited focal distribution; 2-3, intermediate severity; 4-5, coalescent and extensive foci over the entirety of the transversely-sectioned ventricular tissue. For a more quantitative evaluation of the infection, ImagePro-Plus® version 4.0 software was used to quantify the percent of total cross-sectional area containing CVB3 RNA. A segmentation file that recognized positive ISH staining on the basis of hue, saturation and intensity was developed and the total area of staining measured and expressed as a percent of total cross sectional area. Al l statistical analysis was carried out using paired Student's /-tests and by analysis of variants (ANOVA). 195 Plaque Assay The amount of CVB3 in cell supernatant or cytoplasmic extracts was determined on monolayers of HeLa cells by agar overlay plaque assay method as previously described [242]. Briefly, cardiac tissue (apex) was homogenized and solubilized in sterile PBS and stored at -80°C. Samples were serially diluted 10-fold in sterile PBS and overlaid on 90-95% confluent monolayers of HeLa cells in 6-well plates (Corning Costar) at 80% confluency. Cells were incubated for 1 hour (5% CO2, 37°C) with manual rocking every 10 min. Media containing non-bound virus was removed and warm complete M E M containing 0.75% agar was overlaid in each well. The plates were incubated for 36 to 48 hours (5% C 0 2 , 37°C), fixed with Carnoy's fixative (95% EtOH, acetic acid (3:1)) and stained with 1% crystal violet. The plaques were counted and the viral concentration was calculated for 7-8 mice per group as pfu per milliliter of culture supernatant. Al l plaque assay experimentation was performed in a level 2 containment facility and type IIA/B biological safely cabinet. In Situ Hybridization In situ hybridization was performed according to the modified method described previously [242]. Briefly, 4pm paraffin-embedded tissue sections were dewaxed, rehydrated and according to the following protocols: xyleneX2, 100% ethanol X2, 90% ethanol, 70% ethanol all for 5 min. Slides were then permeabilized with proteinase K. Afterwards, the tissues were acetylated, dehydrated, air dried and then hybridized overnight at 55°C using digoxigenin-labeled CVB3 antisense riboprobes or CVB3 sense riboprobe. The CVB3 riboprobes were prepared from the full length CVB3 cDNA using an in vitro transcription kit according to the manufacturer's (Promega) instructions. Infected and uninfected mouse tissues (hearts, pancreas, kidney, liver, spleen) were used as positive and negative controls, respectively. Following a stringent washing in 1:1 formamide/2x SSC (300 mM NaCl, 30 mM sodium citrate, pH 7.0), 2x SSC and 0.2x SSC, detection of hybridization was done incubation in Strept Avidin-Biotin Complex/Alkaline Phosphatase (Dako, Mississauga, ON) at room temperature for 20 min. Slides were incubated in the Vector Red (Vector Laboratories Inc.) and washed 3X5min with TBS-T and for 5 min with dH20. Slides were counterstained with hemotoxylin for 2 min, washed with Lithium Chloride for 1 min, and rehydrated with 196 increasing concentrations of ethanol, 70%, 90%, 100% and xyleneX2, each for 1 minute. Immunohistochemistry Immunohistochemical staining was performed on formalin-fixed, paraffin-embedded, sections for target proteins using a streptavidin-biotin amplification method. Briefly, sections were dewaxed and rehydrated in xylene and graded ethanols according to the following protocols: xyleneX2, 100% ethanol X2, 90% ethanol, 70% ethanol all for 5 min. Slides were treated with an Avidin/Biotin block, serum blocked for 1 hour, washed 3X5 min using Tris-buffered saline (TBS-T; pH 7.6) and incubated with primary antibody overnight at 4°C. Slides were washed 3X5min with TBS-T and exposed to biotinylated IgG antibodies (goat or rabbit according to the primary antibody used; Vector Laboratories, Inc., Burlingame, CA) for 45 min (room temperature), washed 3X5min with TBS and incubated in Strept Avidin-Biotin Complex/Alkaline Phosphatase (Dako, Mississauga, ON) at room temperature for 20 min. Slides were incubated in the chromagen Vector Red (Vector Laboratories Inc.) and washed 3X5min with TBS-T and for 5 min with dl-LoO. Slides were counterstained with hemotoxylin for 2 min, washed with Lithium Chloride for 1 min, and rehydrated with increasing concentrations of ethanol, 70%, 90%, 100% and xyleneX2, each for 1 min. Primary rabbit polyclonal antibody for PBR (1:50 dilution) was purchased from Trevigen (Gaithersburg, MD) and polyclonal rabbit and goat antibodies for HSP27, CTSL and MLP (both 1:100 dilution), respectively, were purchased from Santa Cruz (Santa Cruz, CA). Microscopy Histological images were taken in brightfield with a Nikon Eclipse TE 300 fluorescent microscope at 10X, 20X and 40X objective lenses in lightfield and darkfield microscopy. Images were recorded using the SPOT Enhanced digital camera and processed using the SPOT advanced software, version 3.4.2 (both from Diagnostic Instruments Inc.). The SPOT program was to calculate white balance correction in the red, blue and green wavelengths (recalculated for each session), to set exposure auto gain limit at 8 (0-16), for fine focus and for image acquisition. Images were then exported in uncompressed TIFF formats to crop and fine adjust brightness for presentation. 197 Gene Expression The following procedure is based on the Affymetrix GeneChip® Expression Analysis Technical Manual (©1999, Affymetrix Inc., 701021 Rev.4). Isolation of Total RNA from Cell Lysates Cells were washed and collected in PBS, spun at 300 X g for 5 minutes and frozen at -80°C. Cell lysates were thawed and total RNA was isolated using the QIAGEN (Valencia, CA) RNeasy® isolation kit as per the manufacturer's instructions (RNeasy® MicroHandbook) for isolation of animal cells. For 1X105 cells, 75ul Buffer RLT was added to the cell pellet which was pipetted into a QIAshredder spin column and spun at 14, 000 rpm for 10 min. Then, 75ul of 70% ethanol was added to sample which supports selective RNA binding to the RNeasy membrane, and sample was transferred to an RNeasy MinElute Spin Column and spun briefly at 14, 000 rpm. Buffer RW1 (350ul), buffer RPE (500ul) and 80% ethanol (500ul) were added consecutively to the sample and briefly spun at 14, 000 rpm. RNA is then eluted with 14ul RNase-free water. Total RNA concentration was measured using spectophotometry analysis using the convention that 1 OD at 260nm equals 40ug RNA per mL. Sample concentration and purity were checked by measurement of absorbance at 260 and 280nm. The ratio of A 2 6 0 / A 2 8 0 was between 1.9-2.1. Isolation of Total RNA from Mouse Heart Tissue Mouse heart apex was thawed and total RNA was isolated using the QIAGEN (Valencia, CA) RNeasy® isolation kit as per the manufacturer's instructions (RNeasy® Micro HandBook) for isolation of animal tissues and described in detail above. Heart tissue was disrupted and homogenized using a rotor-stator homogenizer in PBS and stored at -80°C. Sample concentration and purity were checked by measurement of absorbance at 260 and 280nm. Synthesis of Double-Stranded cDNA From Total RNA First strand cDNA synthesis was performed with 8-16.0ug total RNA. For primer hybridization, lOug sample RNA, 2ul T7-(dT)24 primer (Gibco BRL Life Technologies, see below for sequence) and DEPC-treated water (made up to 20ul) were incubated at 70°C for 10 minutes. Then 4ul 5X first strand buffer (Invitrogen), 2ul 0.1M DTT, and 198 lul lOnM dNTP mix and 2ul superscript II reverse transcriptase (200U/ul) were added and the mixture was incubated at 42°C for one hour. T7-(dT)24 primer (Gibco BRL Life Technologies) 5 '-GGCCAGTGAATTGTAATACGACTCACTATAGGGAGGCGG-(dT) 2 4-3 ' The cDNA sample was kept on ice and second strand cDNA was synthesized by adding the following reagents to the finished first strand mixture: 9lul DEPC-treated water, 30ul 5X second strand reaction buffer, 3ul lOnM dNTP, lul E. coli DNA ligase (lOU/ul), 4 ul E. coli DNA polymerase I (10U/ ul) and E. coli RNase H (10U7 ul). The mixture was cooled at 16°C for two hours, 2ul T4 DNA polymerase (10U) was then added and re-cooled for 5 minutes followed by addition of lOul 0.5M EDTA. The cDNA was then stored at -20°C and submitted for further purification using an ethanol precipitation procedure. Absolute (100%) ethanol, ammonium acetate and sample cDNA were vortexed and centrifuged at 12,000 X g at room temperature for 20 minutes. Supernatant was removed and washed with 80% ethanol (stored at -20°C). The sample was centrifuged at 12,000 X g for 5 minutes, supernatant was carefully removed, pellet was air dried, then resuspended in RNase-free water and stored. Synthesis of Biotin-Labeled cRNA and Fragmentation In vitro transcription was performed using an ENZO Bioarray™ HighYield™ RNA Transcript Labeling kit as per the manufacturer's instructions. Briefly, the reaction mixture contained 4ul of 10X HY Reaction Buffer, 4ul of biotin-labeled ribonucleotides, 4ul of 10X DTT, 4ul of 10X RNase inhibitor mix, 2ul of 20X T7 RNA polymerase, lug of template cDNA and dl-bO for a final volume of 40ul. The reaction mixture is incubated at 37°C for 4 hours and the product is stored at -80°C. The amount of the cRNA was measured by spectrophotometry, as previously described, and quality was measured by running the sample on a 1% agarose gel. The biotin-labeled cRNA product was fragmented by incubation with 10ul 5X fragmentation buffer and 40ul of sample in RNase free H 2 0 at 94°C for 35 minutes and stored in -20°C until hybridization. 199 GeneChip Hybridization Hybridization refers to the annealing of probe and target nucleic acid strands following base-pairing rules. Hybridization was performed in an Affymetrix Hybridization Oven 640. A eukaryotic hybridization cocktail (fragmented cRNA 15ug, 20X eukaryotic hybridization control [bioB, bioC, bioD, ere] herring sperm DNA [lOmg/ml] acetylated BSA [50mg/ml], 2X hybridization buffer 150ul, dH 2 0 up to 300ul) was heated to 99°C, then to 45°C, each for 5 min in a heat block. The 2X hybridization buffer contained 12X MES stock (70.4g MES-free acid monohydrate, 193.3g MES sodium salt, 800ml dH 2 0 up to 1L at pH 6.6 and 0.22pm fdtered), 17.7ml 5M NaCl, 4.0ml 0.5M EDTA, 0.1ml 10% Tween-20 and 19.9ml dH 20. The probe array was fdled with 200ul hybridization buffer and incubated at 45°C for 10 min. Hybridization cocktail was spun at 14, 000 rpm for 5 min, the probe array was injected with the hybridization cocktail and placed in the Hybridization Oven 650 for 16h at 60 rpm. GeneChip® Washing and Staining Following probe-target hybridization, arrays were washed in the Affymetrix Fluidics Station 450. Relevant experiment information was entered into MAS 5.1. The GeneChip® was fdled with Non-stringent wash A buffer (300ml 20X SSPE, 1.0ml 10%) Tween-20, 699ml H 2 0 and 0.2um fdtered) and submitted for the following wash protocol: 10 cycles of 2 mixes/cycle with wash buffer A at 25°C, 4 cycles of 15 mixes/cycle with wash buffer B (83.3ml 12XMES stock buffer, 5.2ml 5M NaCl, 1.0ml 10% Tween-20, 910.5ml dH 20, 0.2pm fdtered) at 15°C, and 10 cycles of 4 mixes/cycle with wash buffer A at 25°C. The GeneChip® was fdled with non-stringent wash buffer A and submitted for the following staining protocol: 10 cycles of 2 mixes/cycle with wash buffer A, 4 cycles of 15 mixes/cycle of wash buffer B at 50°C, streptavidin phycoerythrin solution (300ul 2X MES stain buffer, 24.0pl acetylated BSA [50mg/ml], 6.0ul streptavidin phycoerythrin [lmg/ml], 270pl dH 20) at 25°C for 30 min, and 10 cycles of 4 mixes/cycle with wash buffer A at 25°C. GeneChip® Scanning Following wash and stain steps, the GeneChip® was scanned on a HP Agilent GeneArray® Scanner. The scanner was turned on for 10 min prior to use. The 200 Number of Scans was set to 2X image scan and in the Options, pixel value of 3um and wavelength (of laser) was set at 570nm. The GeneChip® was placed into the holder and scanned. cDNA arrays Microarrays contained approximately 7000 cDNA clones randomly collected from a normalized male Wistar rat heart cDNA library [171]. cDNA inserts were generated by PCR amplification with primers derived from flanking vector sequences [173]. PCR products were arrayed from 96-well microtiter plates onto silanated microscope slides in an area of 1.8 cm2 using print tips driven by high-speed robotics. Printed arrays were incubated for 4 hours in a humid chamber and rinsed once in 0.2% SDS (1 minute), twice in H2O (1 minute), and once in sodium borohydride solution (1.9 g of NaBH 4 dissolved in 300 mL of PBS and 100 mL of 100% ethanol; 5 minutes). The arrays were submerged in H2O (2 minutes) at 95°C, transferred quickly into 0.2% SDS (1 minute), rinsed twice in H 2 0 , air dried, and stored in the dark at 25°C. Excised sample hearts were flash frozen in liquid nitrogen and stored at -80°C. Hearts were then transferred into sterile PBS solution and homogenized. Total RNA was isolated using an RNeasy isolation kit as per the manufacturer's instructions (Qiagen). Al l RNA samples were stored at -80°C until further use. Each pair of fluorescently labeled cDNA sample probes was applied to the microarray and allowed to hybridize competitively to the 7000 elements. The array was sequentially excitated on the 2 fluorophores with a scanning laser. The scanned image was acquired and degree of hybridization was quantified. Microarray florescent image analysis was performed using the Incyte Pharmaceutical in-house proprietary software Gem Tools. Expression data were omitted if the signal was <2.5-fold over local background or derived from <40% of the area of the printed spot. A l l data was stored in oracle databases and accessed using an in-house proprietary application called Clone Navigator (SCIOS Inc.). Differential expression values were represented as ratios of intensities between experiment and control. Array production, sample hybridization and image acquisition and quantification were performed at Incyte Pharmaceuticals and further information regarding labeling and hybridization buffer concentrations and image analysis parameters are considered proprietary. 201 Custom Clustering Analysis (+,0,-) Clustering Clustering was performed utilizing C++ script programmed in-house and based on the following simple algorithm: For gene intensity ratio values at 0, 30 min, 3h, 9h, 9h+U0126, labels of'+', '0', or '-' were assigned from the following criteria: +, if ratio > 1.8 0, if 1/1.8 <= ratio <= 1.8 -, if ratio < 1/1.8 ratio = virus/sham Al l genes were placed in one of 81 ([3 labels]4 t i m e P ° i n t s ) possible combinations of clusters. These genes and were saved in Excel (Microsoft Office 2000) spreadsheets for further analysis. Mapping Clusters to Functional Groups An Excel (Microsoft Office 2000) file was created with genes of interest sorted based on their unique Affymetrix ID. Functional annotation from the Swiss-Prot database (http://www.ebi.ac.uk/swissprot) was inputted into the same Excel file and mapping of clusters with functional groups was achieved by putting the two sorted lists side by side. Another C++ script was used to select specific subsets of genes e.g. those that had virus/sham ratios decreased by at least 1.5 when U0126 inhibitor was added. Code was written in C++ to find the number of genes which belong to each functional group. A series of 'pies of pie charts' were generated to show functional breakdown of different sets of genes and to identify their individual members. Briefly, clusters derived using the custom method described above were mapped with the functional annotation from the Swiss-Prot database (http://www.ebi.ac.uk/swissprot). The mapping' was done as follows: table 1 (containing the genes and their cluster numbers) and table 2 (containing the genes and their functional groups) were sorted based on their Affymetrix ID. Unique elements to each table were added to the other table. The two tables were sorted and positioned side-by-side on an Excel worksheet (Microsoft Office 2000). Scripts were then 202 written to create two new tables. Tables were generated based on clusters, with each cluster providing the number of different functional groups at the first three functional levels/divisions (given that the functional groups had multiple levels; tables not shown). Pies of pie charts that show the distribution of functional groups within the list of genes in the respective clusters were created for groups of potential importance (Figure 61). For the same groups line graphs were created, using the Excel chart function, to show the magnitude of the expression of these genes in a specific cluster. GeneSpring 5.0.1® GeneChip® data files (*.dat image file), cell files (*.cel) and chip files (*.chp) were generated and saved using Affymetrix MAS 5.0. Data was then converted to a standard ASCII text (*.txt a file which can be spreadsheet analysis programs) file for download into GeneSpring 5.0.1™ (Silicon Genetics). For analysis in GeneSpring 5.0.1™, the parameters time and infection were used as continuous and color-coded independent variables, respectively. Each gene signal was normalized to itself such that the median of all measurements taken from that gene was set to 1. In the case for signals which are anomalously low, the lower median cut-off value was set to 10.0 expression units, under which signals will not be used for calculations. Reliability of expression data was set at 500.0 for high, 150.0 for medium and 50.0 for low. Analysis results were saved either as screen images or Excel spreadsheet format. Z score normalization Z normalization is a scaling method to set the mean value to 0.0 and the maximum and minimum values of the scaled values to +3.0 to -3.0 (standard deviations). This requires data distribution to be roughly Gaussian (Z score of logio values can be used for non-Gaussian data). If C is a gene in hybridized sample P, then it is computed as: Zscore(P,C) = (C - mean(P))/St Dev(P) BLAST Search Basic local alignment search was performed using BLASTN version 2.2.4 against GenBank release 129 [328]. Al l data was stored in oracle databases and accessed using an in-house proprietary application called "Clone Navigator" (SCIOS Inc.). GenMAPP 1.0 203 GeneChip data was converted in Excel to ratios of intensity of experiment / control (virus / sham) and visualized with both standard and my custom-made MAPPs. Data was downloaded into GenMAPP 1.0 and analysis criteria were selected in the Expression Dataset Manager. Criteria builder was used to set the ratio fold cut-offs as the following: up-regulation was set to <=1.6 (red); up-regulation trend was >1.6 and <1.0 (light red); down-regulation trend was >=1.0 and >0.6 (light blue); down-regulation was >=0.6 (blue). Human GO MAPPs related to cellular component, biological process and molecular function were downloaded on March 4 t h, 2002. Data was visualized for all MAPPs in these groups and images were saved with the Print Screen function as image quality was superior than the image export function. Tools within MAPP Builder version 1.0 were utilized to build a custom MAPP of Downstream ERK genes based on the published literature. RT-PCR To validate the microarray results, we examined the differential expression of selected genes by RT-PCR. HeLa cell and heart tissue mRNA was isolated and used for reverse transcriptase-polymerase chain reaction (RT-PCR). Quality of sample RNA and concentration was determined by spectroscopy. Briefly, an oligo-dT primer was used with the murine leukemia virus RT for first-strand synthesis. Amplification of desired cDNA was performed with sense and antisense primers designed independently of the sequence used on either microarrays. Primers were designed manually with the following parameters: length of 20bp, 50% GC content, far-removed from start and stop codons, no repetitive nucleotides (ie.>3), and with a target amplicon of 200bp. The cDNA was submitted for 94°C incubation for 10 min and 35 PCR cycles with the following parameters: denaturation at 94°C for 30 seconds, annealing at 52°C for 2 minutes and extension at 72°C for 30 seconds. This was followed by a final extension step at 72°C for 10 min. RT-PCR was performed on a Perkin Elmer Gene Amp PCR System 9700. Individual amplicons were visualized using a 1.5% DNA agarose gels containing ethidium bromide (3ug/ul) for visualization. Gels were visualized by Eagle Eye instmmentation. Primers were designed de novo to amplify products roughly 200 bp in length. Images were acquired using an Eagle Eye II Image System (Stratagene) using a CCD camera under ultraviolet light. 204 The primer sequences used are given below: Gene Name Primer 1 (5'-3') Primer 2 (5'-3') Serpin AAG CAG AGG ATC CGT ATG GC CAC ATG CCC ATC CTG ATA CT SI 00 ACC ATG ATG CTT ACA TTT CAC CTT CTT CTG CTT CAT GTG TAC T PBR CTC CGC TGG TAT GCT AGC TT TCG CCG ACC AGA GTT ATC AC PAI-1 AGG CCA CCA ACT TCG GAG TAA CTT GTG GAA CAG GCG CTG GT MT-1 TCC TGC GGC TGC AAG AAC TG CAC AGC ACG TGC ACT TGT CC LIM CAC TGC TGC CTG AGG TGC TT GGC TGA TCC TGG TAA GTG AT HSP27 CCG CCT CTT CGA TCA AGC TT CGC TGA TTG TGT GAC TGC TT H2-Bf AGA TCC ACA GGC TCC TGG AG CAT GAA GGA ATC TTG GCA GG Cathepsin L TTA ATG CAC AGT GGC ACC AG GTG AGA ACA GTC CAC AGG GT P-Actin GGA CTC CTA TGT GGG TGA CGA GG GGG AGA GCA TAG CCC TCG TAG Al Table 7: Primer sequences for RT-PCR. Primer sequences for nine genes found to be differentially regulated by cDNA and GeneChip arrays and (3-actin control gene are listed. 205 Ech ocardiography Male adolescent (25-30 g) A/J mice were either CVB3 or sham infected and anesthetized with 0.15ml IP (calculated for a 20g mouse) of a Ketamine and Xylazine mixture (Ketamine (lOOmg/ml) 1.5ml, Xylazine (20mg/ml) 0.5ml, PBS/water 8.0ml) for a total concentration of 0.45 mg/kg Ketamine and 0.03 mg/kg Xylazine. Anesthesia was tested by toe-pinch. Mice were shaved in the left chest area, Nair (Nair Products) was applied for lmin and hair was wiped/removed. Mice placed in the dorsal recumbency position on a bed of agarose gel (1%) and heart echocardiography was measured by placing an echo probe directly against the gel for 5 min. Two-dimensional echocardiography was performed according to the standards of the American Society of Echocardiography using an Agilent Sonos 5500 Echocardiography System (Philips) and S12 transducer (12MHz) [342]. To obtain a stable baseline hemodynamic state, mouse subjects were anesthesized for 5-10 min before measurements were taken. Parasternal long axis measurements were taken from the aorta to the apical inner LV wall in end systole and end diastole as an average of five consecutive heart beats and a mean value was calculated. Parasternal short axis widths were taken at the level of the mitral valve and papillary muscles in end systole and end diastole as an average of five consecutive heart beats and a mean value was calculated. Averaged values were used to calculate ejection fraction using the modified Simpson's algorithm. Al l values were acquired onto both VHS and electronic formats. Following heart function analysis, mice were sacrificed and organs were harvested. 206 Bibliography [I] Woodruff, J.F., Am J Pathol, 1980. 101(2):425-84. [2] Aretz, H.T., Hum Pathol, 1987. 18(6):619-24. [3] Chow, L.H., etal., Arch Pathol Lab Med, 1989. 113(12):1357-62. [4] Winters, G.L. and B.M. McManus, Myocarditis, in Cardiovascular Pathology, 3rd Edition, M . Silver, A.I. Gotlieb, and F.J. Schoen, Editors. 2001, Churchill Livingstone: London. [5] Mason, J.W., et a l , N Engl J Med, 1995. 333(5)269-75. [6] Veinot, J.P., Can J Cardiol, 2002. 18(l):55-65. [7] Hauck, A.J., D.L. Kearney, and W.D. Edwards, Mayo Clin Proc, 1989. 64(10):1235-45. [8] Smith, S.C., et al., Circulation, 1997. 95(1): 163-8. [9] Franz, W.M., et a l , Clin Chem, 1996. 42(2):340-l. [10] Jaffe, A S . , et al., Clin Chem, 1996. 42(11): 1770-6. [II] Adams, J.E., 3rd, et al., Circulation, 1993. 88(l):101-6. [12] Kim K-S, et al., 77ze primary viruses of myocarditis, in Myocarditis: From Bench to Bedside, C L . Jr., Editor. 2002, Mayo Academic Press: Rochester, MN. [13] Dalldorf, G. and G.M. Sickles, Science, 1948. 108:61-2. [14] Melnick, J.L., E.W. Shaw, and E.C. Curnen, Proc Soc Exp Biol Med, 1949. 71:344-9. [15] Verlinde, J., H. Van Tongeren, and A. Kret, Acta Pediatr, 1956. 187:113-8. [16] Fujioka, S., et al., Am Heart J, 1996.131(4):760-5. [17] Martino, T.A., P. Liu, and M.J. Sole, Enteroviral myocarditis and dilated cardiomyopathy: A review of clinical and experimental studies. Human enterovirus infections, ed. H.A. Rotbart. 1995, Washington, D.C.: A S M Press. 291-351. [18] Grist, N.R. and D. Reid, Epidemiology of viral infections of the heart, in Viral infections of the heart, J.E. Banatvala, Editor. 1993, Eward Arnold: London. [19] Kaplan, M.H., etal., Rev Infect D is, 1983. 5(6): 1019-32. [20] Martino, T.A., P. Liu, and M.J. Sole, Circ Res, 1994. 74(2): 182-8. [21] Kopecky, S.L. and B.J. Gersh, Curr Probl Cardiol, 1987. 12(10):569-647. [22] Satoh, M. , et a l , Eur Heart J, 1994. 15(7):934-9. 207 [23] Giacca, M. , et al., J Am Coll Cardiol, 1994. 24(4) :1033-40. [24] Baboonian, C. and T. Treasure, Heart, 1997. 78(6):539-43. [25] Badorff, C , et al., Nat Med, 1999. 5(3):320-6. [26] Badorff, C , et al., J Biol Chem, 2000. 275(15): 11191-7. [27] Towbin, J.A., Curr Opin Cell Biol, 1998.10(1):131-9. [28] Towbin, J.A., K.R. Bowles, and N.E. Bowles, Nat Med, 1999. 5(3):266-7. [29] Badorff, C , G.H. Lee, and K.U. Knowlton, Herz, 2000. 25(3):227-32. [30] Kawai, C , et al., Jpn CircJ, 1978. 42(l):43-7. [31] Batra, A.S. and A.B. Lewis, Curr Opin Pediatr, 2001. 13(3):234-9. [32] Kuhl, U., et al., Circulation, 2003. 107(22):2793-8. [33] Rane, S.G. and E.P. Reddy, Oncogene, 2000.19(49):5662-79. [34] Daliento, L., et al., J Heart Lung Transplant, 2003. 22(2):214-7. [35] Blay, R., et al., Am J Pathol, 1989.135(5):899-907. [36] Leslie, K., et a l , Clin Microbiol Rev, 1989. 2(2): 191 -203. [37] Lodge, P.A., et al., Am J Pathol, 1987. 128(3):455-63. [38] Chow, L.H., K.W. Beisel, and B.M. McManus, Lab Invest, 1992. 66(1)24-31. [39] Klingel, K., et al., Lab Invest, 1998. 78(10): 1227-37. [40] Chow, L.H., LabAnim Sci, 1993. 43(2): 133-5. [41] Luo, H., et al., Am J Pathol, 2003. 163(2):381-5. [42] Etchison, D., et al., JBiol Chem, 1982. 257(24): 14806-10. [43] Luo, H., et al., J Virol, 2003. 77(1): 1-9. [44] Kishimoto, C , et al., Clin Immunol Immunopathol, 1997. 85(l):47-55. [45] Weremeichik, H., et al., Eur Heart J, 1991. 12 Suppl D:154-7. [46] Beck, M.A., etal., Am JPathol, 1990. 136(3):669-81. [47] Liu, P., et al., Can J Cardiol, 1996.12(10):935-43. [48] Kandolf, R. and P.H. Hofschneider, Proc Natl Acad Sci USA, 1985. 82(14):4818-22. [49] Yang, D., et al., Virology, 1997. 228(1 ):63-73. [50] Bergelson, J.M., et a l , J Virol, 1998. 72(l):415-9. [51] Bergelson, J.M., et al., J Virol, 1995. 69(3):1903-6. [52] Carson, S.D. and N.M. Chapman, Biochemistry, 2001. 40(48):14324-9. [53] Cohen, C.J., etal., Proc Natl Acad Sci USA, 2001. 98(26): 15191-6. [54] Nicholson-Weller, A. and C.E. Wang, J Lab Clin Med, 1994. 123(4):485-91. [55] Rossmann, M.G., etal., Nature, 1985. 317(6033):145-53. 208 [56] Shieh, J.T. and J.M. Bergelson, J Virol, 2002. 76(18):9474-80. [57] Nicholson-Weller, A., et al., J Immunol, 1982. 129(1):184-9. [58] Ito, M. , et al., Circ Res, 2000. 86(3):275-80. [59] Tam, P.E. and R.P. Messner, J Virol, 1999. 73(12):10113-21. [60] Reetoo, K.N., et al., J Gen Virol, 2000. 81(Pt 11):2755-62. [61] Hunter, J.J. and K.R. Chien, N EnglJ Med, 1999. 341(17):1276-83. [62] Yanagawa, B., et al., Life and death signaling pathways in CVB3-induced myocarditis, in Myocarditis: From bench to bedside, L.T. Cooper, Editor. 2001, Humana Press: Totowa. p. 161-96. [63] Brunet, A. and J. Pouyssegur, Essays Biochem, 1997. 32:1-16. [64] Luo, H., et a l , J Virol, 2002. 76(7):3365-73. [65] Huber, M. , et al., J Virol, 1999. 73(5):3587-94. [66] Opavsky, M.A., etal., JClin Invest, 2002. 109(12):1561-9. [67] Liu, P., et al., Nat Med, 2000. 6(4):429-34. [68] Ettehadieh, E., et al., Science, 1992. 255(5046):853-5. [69] Simons, K. and E. Ikonen, Nature, 1997. 387(6633):569-72. [70] Cunningham, K.A., N .M. Chapman, and S.D. Carson, Virus Res, 2003. 92(2): 179-86. [71] Yasukawa, H., et al., J Clin Invest, 2003. lll(4):469-78. [72] Gilmore, T.D., Oncogene, 1999. 18(49):6842-4. [73] Schwarz, E . M , et al., J Virol, 1998. 72(7):5654-60. [74] Yokoseki, O., et al., Circ Res, 2001. 89(10):899-906. [75] Voiculescu, M. , et al., Virology, 1979. 30(3):207-16. [76] Leroy, E.M., et al., Lancet, 2000. 355(9222):2210-5. [77] Berger, T.M., J.H. Caduff, and J.O. Gebbers, PedInfect Dis J, 2000. 19(7):653-6. [78] Granville, D.J., et a l , Cell Death Diff, 1998. 5(8):653-9. [79] Yew, P.R. and A.J. Berk, Nature, 1992. 357(6373):82-5. [80] Boyd, Cell, 1994. 79(6):1121. [81] Duckett, C.S., etal., EMBO J, 1996.15(11)2685-94. [82] Rapp, U.R., etal., Proc Natl Acad Sci USA, 1983. 80(14):4218-22. [83] Stehelin, D., et a l , Nature, 1976. 260(5547): 170-3. [84] Carthy, C M . , et al., Lab Invest, 1999. 79(8):953-65. [85] Carthy, C M . , et al., J Virol, 1998. 72(9):7669-75. 209 86] Trump, B.F., et a l , Toxicol Pathol, 1997. 25(l):82-8. 87] Majno, G. and I. Joris, Am J Pathol, 1995. 146(1):3-15. 88] Saraste, A. and K. Pulkki, Cardiovasc Res, 2000. 45(3):528-37. 89] Kerr, J.F., C M . Winterford, and B.V. Harmon, Cancer, 1994. 73(8):2013-26. 90] Fiers, W., et al., Oncogene, 1999. 18(54):7719-30. 91] Wyllie, A.H. , Curr Opin Genet Dev, 1995. 5(1):97-104. 92] Wyllie, A.H. , J.F. Kerr, and A.R. Currie, Int Rev Cytol, 1980. 68:251-306. 93] Kerr, J.F., A.H. Wyllie, and A.R. Currie, Br J Cancer, 1972. 26(4):239-57. 94] Thornberry, N.A. and Y. Lazebnik, Science, 1998. 281(5381): 1312-6. 95] Kayalar, C , et al., Proc Natl Acad Sci USA, 1996. 93(5):2234-8. 96] Cryns, V.L. , et al., J Biol Chem, 1996. 271(49):31277-82. 97] Takahashi, A., et al., Proc Natl Acad Sci USA, 1996. 93(16):8395-400. 98] Neamati, N. , et al., J Immunol, 1995.154(8):3788-95. 99] Cosulich, S.C., et al., EMBOJ, 1997.16(20):6182-91. 100] Caulin, C , G.S. Salvesen, and R.G. Oshima, J Cell Biol, 1997. 138(6): 1379-94. 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 Kothakota, S., et a l , Science, 1997. 278(5336):294-8. Wen, L.P., etal., JBiol Chem, 1997. 272(41):26056-61. Schulze-Osthoff, K , etal, Eur J Biochem, 1998. 254(3):439-59. Muzio, M. , et al., J Biol Chem, 1998. 273(5):2926-30. Li , P., et al., Cell, 1997. 91(4):479-89. Anversa, P., et al., Lab Invest, 1998. 78(7):767-86. Anversa, P. and J. Kajstura, Circ Res, 1998. 82(11):1231-3. Yamada, T., etal., Heart Vessels, 1999. 14(l):29-37. Colston, J.T., B. Chandrasekar, and G.L. Freeman, Cardiovasc Res, 1998. 38(l):158-68. Matsumori, A., C.S. Crumpacker, and W.H. Abelmann, J Am Coll Cardiol, 1987.9(6):1320-5. Tolskaya, E.A., et al., J Virol, 1995. 69(2): 1181-9. Tyler, K.L., et al., J Virol, 1995. 69(11):6972-9. Jelachich, M.L. and H.L. Lipton, J Virol, 1996. 70(10):6856-61. Anderson, R., et al., Brain Res, 2000. 868(2):259-67. Ammendolia, M.G., etal., JMed Virol, 1999. 59(l):122-9. Robinson, N.M. , et a l , Eur J Clin Invest, 1999. 29(8):700-7. 210 117] Gluck, B., et al., Herz, 2000. 25(3):255-60. 118] Seko, Y., et al., J Pathol, 1997.183(l):105-8. 119] Matsumori, A., et al., Br Heart J, 1994. 72(6):561-6. 120] Levine, B., etal., TVEngl JMed, 1990. 323(4):236-41. 121] Yamada, T., A. Matsumori, and S. Sasayama, Circulation, 1994. 89(2):846-51. 122] Matsumori, A., T. Yamada, and C. Kawai, Eur Heart J, 1991. 12(Suppl D):203-5. 123] Wessely, R., et al., Circulation, 2001.103(5):756-61. 124] Horwitz, M.S., et al., Nat Med, 2000. 6(6):693-7. 125] Zhang, H.M., et al., J Biol Chem-In Press, 2003. 20:20. 126] Yang, D., et al., Circ Res, 1999. 84(6):704-12. 127] Moncada, S. and A. Higgs, N Engl J Med, 1993. 329(27):2002-12. 128] Zaragoza, C , et al., J Clin Invest, 1997. 100(7): 1760-7. 129] Liu, P., et al., Eur Heart J, 1995. 16(Suppl 0):25-7. 130] Lowenstein, C.J., et al,JClin Invest, 1996. 97(8): 1837-43. 131] Seko, Y., et al., Circulation, 1991. 84(2):788-95. 132] Rotbart, H.A. and A.D. Webster, Clin Infect Dis, 2001. 32(2):228-35. 133] Wang, A., et al., Antimicrob Agents Chemother, 2001. 45(4): 1043-52. 134] Yanagawa, B., et al., Lab Invest, 2003. 83(l):75-85. 135] Taylor, L.A., et al., Circ Res, 2000. 87(4):328-34. 136] Selinger, D.W., et al., Genome Res, 2003.13(2):216-23. 137] Raghavan, A., et al., Nucleic Acids Res, 2002. 30(24):5529-38. 138] Jansen, R., D. Greenbaum, and M . Gerstein, Genome Res, 2002. 12(l):37-46. 139] Zhang, J.S., et al., Mol Biotechnol, 1998.10(2):155-65. 140] Zhang, H.M., etal, Circ Res, 2002. 90(12):1251-8. 141] Hughes, T.R., etal., Cell, 2000.102(1): 109-26. 142] Newton, M.A., et al., J Comput Biol, 2001. 8(l):37-52. 143] Tseng, G .C , etal., Nucleic Acids Res, 2001. 29(12):2549-57. 144] Quackenbush, J., Nat Genet, 2002. 32(Suppl):496-501. 145] Zhao, X., et al., Nucleic Acids Res, 2001. 29(4):955-9. 146] Call, D.R., D.P. Chandler, and F. Brockman, Biotechniques, 2001. 30(2):368-72, 74, 76 passim. [147] Brown, A.J., etal., EmboJ, 2001. 20( 12) :3177-86. 211 [148] Yang, Y.H. , et a l , Nucleic Acids Res, 2002. 30(4):el 5. U49] Pease, A.C., et al., Proc Natl Acad Sci USA, 1994. 91(11):5022-6. 150] Wang, H., et al., Bioinformatics, 2003. 19(Suppl l):i315-22. 151] Yuen, T., et al., Nucleic Acids Res, 2002. 30(10):e48. [152] Nuwaysir, E.F., etal, Genome Res, 2002. 12(11): 1749-55. 153] Singh-Gasson, S., et al., Nat Biotechnol, 1999. 17(10):974-8. [154] Chambers, J., et al., J Virol, 1999. 73(7):5757-66. [155] Consortium, G.O., Genome Res, 2001. ll(8):1425-33. [156] Azuaje, F., Brief Bioinform, 2003. 4(1):31-42. [157] Chun, T.W., et a l , Proc Natl Acad Sci U S A, 2003. 100(4): 1908-13. [158] Sherlock, G., Curr Opin Immunol, 2000. 12(2):201-5. [159] Dahlquist, K.D., et al., Nat Genet, 2002. 31(l):19-20. [160] Brazma, A., et al., Nat Genet, 2001. 29(4):365-71. [161] Spellman, P.T., et a l , Genome Biol, 2002. 3(9):RESEARCH0046. [162] Holm, S., Scand J Statist, 1979. 6:65-70. [163] Quackenbush, J., Nat Rev Genet, 2001. 2(6):418-27. [164] Bolstad, B.M., et a l , Bioinformatics, 2003. 19(2): 185-93. 165] Woodruff, J.F., J Infect Dis, 1970. 121(2): 164-81. [166] Chow, L.H., C.J. Gauntt, and B.M. McManus, Lab Invest, 1991. 64(l):55-64. [167] Zanone, M.M. , et al., J Immunol, 2003. 171(l):438-46. [168] Schena, M. , et al., Proc Natl Acad Sci USA, 1996. 93(20): 10614-9. 169] Garg, N., V.L. Popov, and J. Papaconstantinou, Biochim Biophys Acta, 2003. 1638(2): 106-20. [170] Peng, C.F., et al., Physiol Genomics, 2002. 9(3):145-55. [171] Stanton, L.W., et al., Circ Res, 2000. 86(9):939-45. [172] Iyer, V.R., et al., Science, 1999. 283(5398):83-7. 173] Airman, T.J., et al., Nat Genet, 1999. 21(1 ):76-83. [174] Cheadle, C , et a l , J Mol Diagn, 2003. 5(2):73-81. [175] Takemura, G., et al., IntJ Cardiol, 1995. 52(3):213-22. 176] Sigurdsson, A. and K. Swedberg, Am Heart J, 1996. 132(1 Pt 2 Su):229-34. [177] Sherry, B., Viral Immunol, 2002. 15(l):17-28, [178] Matsumori, A., Cytokines in experimental myocarditis, in Myocarditis: From bench to bedside, L.T. Cooper, Editor. 2003, Humana Press: Totowa. p. 109-34. 212 [179] Pestka, S., et al , Annu Rev Biochem, 1987. 56:727-77. [180] Uze, G., G. Lutfalla, and K.E. Mogensen, J Interferon Cytokine Res, 1995. 15(l):3-26. [181] Constantinescu, S.N., etal., Proc Natl Acad Sci USA, 1994. 91(20):9602-6. [182] Haque, S.J. and B.R. Williams, Semin Oncol, 1998. 25(1 Suppl 1): 14-22. [183] Feuer, R., et al., J Virol, 2002. 76(9):4430-40. [184] Noutsias, M. , et al., Circulation, 2001. 104(3):275-80. [185] Ito, M. , et al., Circ Res, 2000. 86(3):275-80. [186] Fechner, H., et al., Circulation, 2003.107(6):876-82. [187] Paigen, K., Nat Med, 1995. l(3):215-20. [188] Waterston, R.H., et al., Nature, 2002. 420(6915):520-62. [189] Maniatis, T. and B. Tasic, Nature, 2002. 418(6894):236-43. [190] Roberts, G.C. and C.W. Smith, Curr Opin Chem Biol, 2002. 6(3):375-83. [191] Hare, J.M. and W.S. Colucci, Prog Cardiovasc Dis, 1995. 38(2):155-66. [192] Geiss, G.K., et al., Virology, 2000. 266(1):8-16. [193] Zhu, H., et al., Proc Natl Acad Sci USA, 1998. 95(24):14470-5. [194] Stingley, S.W., etal., J Virol, 2000. 74(21):9916-27. [195] Bigger, C.B., K . M . Brasky, and R.E. Lanford, J Virol, 2001. 75(15):7059-66. [196] Peng, X. , et a l , BMCBioinformatics, 2003. 4(1):26. [197] Hwang, J.J., et al., Physiol Genomics, 2002. 10(1 ):31 -44. [198] Barrans, J.D., et al., Am J Pathol, 2002. 160(6):2035-43. [199] Palmer, J.W., B. Tandler, and C L . Hoppel, J Biol Chem, 1917. 252(23):8731-9. [200] Wallace, D.C., Am Heart J, 2000.139(2 Pt 3):S70-85. [201] Kelly, D.P. and A.W. Strauss, N Engl J Med, 1994. 330(13):913-9. [202] Marin-Garcia, J. and M.J. Goldenthal, Cardiovasc Res, 1994. 28(4):456-63. [203] Wallace, D .C , Annu Rev Biochem, 1992. 61:1175-212. [204] Sambandam, N. , etal., Heart Fail Rev, 2002. 7(2):161-73. [205] Carvajal, K. and R. Moreno-Sanchez, Arch Med Res, 2003. 34(2):89-99. [206] Vogt, A . M . and W. Kubler, Basic Res Cardiol, 1998. 93(1):1-10. [207] Ide, T., et al., Circ Res, 2001. 88(5):529-35. [208] Green, D.R., Cell, 1998. 94(6):695-8. [209] Patino, W.D., O.Y. Mian, and P.M. Hwang, Circ Res, 2002. 91(7):565-9. [210] Peng, T., et al., Cardiovasc Res, 2001. 50(l):46-55. 213 [211] Aach, J., W. Rindone, and G.M. Church, Genome Res, 2000.10(4):431-45. [212] Chien, K.R., etal, FasebJ, 1991. 5(15):3037-46. [213] deBold, A.J., et al, Life Sci, 1981. 28(l):89-94. [214] de Bold, A.J., Science, 1985. 230(4727):767-70. [215] Edwards, B.S., etal. ,/ Clin Invest, 1988. 81(l):82-6. [216] Wang, D., et al., Hypertension, 2003. 42(l):88-95. [217] Fuse, K., et a l , Clin Exp Immunol, 2001. 124(3):346-52. [218] Kishimoto, C , et al., J Mol Cell Cardiol, 2000. 32(4):631-8. [219] Kobayashi, Y., et al., Autoimmunity, 2002. 35(2):97-104. [220] Donato, R., Biochim Biophys Acta, 1999.1450(3): 191-231. [221] Moore, B.W., Biochem Biophys Res Commun, 1965. 19(6):739-44. [222] Heierhorst, J., et al., Nature, 1996. 380(6575):636-9. [223] Hofmann, M.A., et a l , Cell, 1999. 97(7):889-901. [224] Du, X.J., et al., Mol Cell Biol, 2002. 22(8):2821-9. [225] Tsoporis, J.N., etal., J Clin Invest, 1998. 102(8): 1609-16. [226] Tsoporis, J.N., et al., J Biol Chem, 1997. 272(50):31915-21. [227] Remppis, A., et al., Biochim Biophys Acta, 1996. 1313(3):253-7. [228] Baudier, J., et al., Biochemistry, 1995. 34(24):7834-46. [229] Schafer, B.W. and C.W. Heizmann, Trends Biochem Sci, 1996. 21(4):134-40. [230] Fano, G., et a l , FEBSLett, 1989. 255(2):381-4. [231] Hasenfuss, G., et al., Circ Res, 1992. 70(6): 1225-32. [232] Arber, S., G. Haider, and P. Caroni, Cell, 1994. 79(2):221-31. [233] Arber, S. and P. Caroni, Genes Dev, 1996. 10(3):289-300. [234] Flick, M.J. and S.F. Konieczny, J Cell Sci, 2000. 113(Pt 9): 1553-64. [235] Heineke,J., etal., Circulation, 2003.107(10):1424-32. [236] Stypmann, J., et al., Proc Natl Acad Sci USA, 2002. 99(9):6234-9. [237] Arber, S., et al., Cell, 1997. 88(3):393-403. [238] Knoll, R., et al., Cell, 2002. lll(7):943-55. [239] Walport, M.J., N Engl J Med, 2001. 344(15): 1140-4. [240] Walport, M.J., N Engl J Med, 2001. 344(14): 1058-66. [241] Barrington, R., etal., Immunol Rev, 2001.180:5-15. [242] Anderson, D.R., et al., J Virol, 1996. 70(7):4632-45. [243] Kirschke, H., et al., Biochem J, 1982. 201(2):367-72. [244] Watts, C , Annu Rev Immunol, 1997. 15:821-50. 214 [245] Bando, Y., E. Kominami, and N. Katunuma, J Biochem (Tokyo), 1986. 100(l):35-42. [246] Sugden, P.H. and A. Lazou, Cardiovasc Res, 1984. 18(8):483-5. [247] Huber, S.A., Lab Invest, 1992. 67(2):218-24. [248] Donnelly, T.J., et al., Circulation, 1992. 85(2):769-78. [249] Marber, M.S., et al., J Clin Invest, 1995. 95(4): 1446-56. [250] Vander Heide, R.S., Am J Physiol Heart Circ Physiol, 2002. 282(3):H935-41. [251 ] Delogu, G., et al., Curr Opin Crit Care, 2002. 8(5):411 -6. [252] Woods, M.J. and D.C. Williams, Biochem Pharmacol, 1996. 52(12):1805-14. [253] McEnery, M.W., et al., Proc Natl Acad Sci USA, 1992. 89(8):3170-4. [254] Wang, K., et al., Biophys J, 1993. 64(4):1161-77. [255] Bergelson, J.M., et al., J Infect Dis, 1997. 175(3):697-700. [256] Lamphear, B.J., et al., J Biol Chem, 1995. 270(37):21975-83. [257] Kerekatte, V., et al., J Virol, 1999. 73(1):709-17. [258] Blackwell, J.M., Trends Mol Med, 2001. 7(11):521-6. [259] Yue, H., et al., Nucleic Acids Res, 2001. 29(8):E41-1. [260] Kamme, F., et al., JNeurosci, 2003. 23(9):3607-l 5. [261] Eberwine, J., et al., Proc Natl Acad Sci USA, 1992. 89(7):3010-4. [262] Kellam, P., Genome Biol, 2000. 1 (2):RE VIEWS 1009. [263] Yeung, K.Y. , D.R. Haynor, and W.L. Ruzzo, Bioinformatics, 2001. 17(4):309-18. [264] Willett, W.C., Science, 2002. 296(5568):695-8. [265] Beck, M.A., et al., Nat Med, 1995. l(5):433-6. [266] Beck, M.A. and C.C. Matthews, Proc Nutr Soc, 2000. 59(4):581 -5. [267] Peng, T., et al., J Clin Microbiol, 2000. 38(10):3538-43. [268] McManus, B.M., et al., Clin Immunol Immunopathol, 1993. 68(2):159-69. [269] Bradham, D.M., et al., J Cell Biol, 1991.114(6): 1285-94. [270] Davies, S.P., et al., Biochem J, 2000. 351(Pt 1):95-105. [271] Chen, M.M., et al., J Mol Cell Cardiol, 2000. 32(10): 1805-19. [272] Ohnishi, H., et al., J Mol Cell Cardiol, 1998. 30(11):2411-22. [273] Joachims, M. , K.S. Harris, and D. Etchison, Virology, 1995. 211(2):451-61. [274] Murohara, T., J.P. Guo, and A . M . Lefer, J Pharmacol Exp Ther, 1995. 274(3): 1246-53. 215 [275] Gurevich, R.M., K .M. Regula, and L.A. Kirshenbaum, Circulation, 2001. 103(15): 1984-91. [276] Kanda, T., et al., IntJ Cardiol, 1999. 68(l):13-22. [277] Krah, D.L. and R.L. Crowell, Virology, 1982. 118(1): 148-56. [278] Hashimoto, G., et al., J Biol Chem, 2002. 277(39):36288-95. [279] Huttunen, P., et al., Virology, 1998. 250(l):85-93. [280] Hoshijima, M . and K.R. Chien, J Clin Invest, 2002. 109(7):849-55. [281] Churchill, G.A., Nat Genet, 2002. 32(Suppl):490-5. [282] DeSilva, D.R., et al., J Immunol, 1998. 160(9):4175-81. [283] Wang, J.Y., Oncogene, 2000. 19(49):5643-50. [284] Gishizky, M.L., Cytokines Mol Ther, 1996. 2(4):251 -61. [285] Dai, Z., etal, Genes Dev, 1998. 12(10):1415-24. [286] Varticovski, L., et a l , Mol Cell Biol, 1991. 11(2):1107-13. [287] Danial, N.N., et al , Mol Cell Biol, 1998.18(11):6795-804. [288] Vaux, D.L. and J. Silke, Biochem Biophys Res Commun, 2003. 304(3):499-504. [289] Crook, N.E., R.J. Clem, and L.K. Miller, J Virol, 1993. 67(4):2168-74. [290] Chu, Z.L., et a l , Proc Natl Acad Sci USA, 1997. 94(19):10057-62. [291] Li , F., et al., Nature, 1998. 396(6711):580-4. [292] Yang, Y. , et al., Science, 2000. 288(5467):874-7. [293] Holcik, M . , et al., Nat Cell Biol, 1999. l(3).190-2. [294] Buerke, M. , et al., Proc Natl Acad Sci USA, 1995. 92(17):8031-5. [295] Kotlyar, A. A., et al., Heart, 2001. 86(6):693-700. [296] Liang, G., et al., J Biol Chem, 1996. 271(3): 1695-701. [297] Hai, T., et a l , Gene Expr, 1999. 7(4-6):321-35. [298] Manger, I.D. and D.A. Relman, Curr Opin Immunol, 2000. 12(2):215-8. [299] Wolfgang, C D . , et al., J Biol Chem, 2000. 275(22): 16865-70. [300] Eckmann, L., et al., J Biol Chem, 2000. 275(19): 14084-94. [301] Cohen, P., et al., J Biol Chem, 2000. 275(15):11181-90. [302] Sehgal, A., et al., Oncogene, 2000.19(24):2836-45. [303] Tesmer, V .M. , etal, Proc Natl Acad Sci USA, 1993. 90(15):7298-302. [304] Buckner, A.E., V . M . Tesmer, and M. Bina, Virus Res, 2002. 89(l):53-63. [305] Lee, J.K., et al., Nat Med, 1998. 4(12): 1383-91. 216 [306] Gerdes, A . M . , S.E. Campbell, and D.R. Hilbelink, Lab Invest, 1988. 59(6):857-61. 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324; 325 326 327 328 329 330 331 332 333 334 Hatt, P.Y., et a\.,JMol Cell Cardiol, 1979. ll(10):989-98. Robinson, T.F., L. Cohen-Gould, and S.M. Factor, Lab Invest, 1983. 49(4):482-98. Weber, K.T., J Am Coll Cardiol, 1989. 13(7):1637-52. Weber, K.T., Cardiovasc Res, 2000. 46(2):211-3. Zwolinski, R.J., C R . Hamlin, and R.R. Kohn, Proc Soc Exp Biol Med, 1976. 152(3):362-5. Caulfield, J.B. and P.E. Wolkowicz, Toxicol Pathol, 1990. 18(4 Pt l):488-96. Schubert, A., et a l , Basic Res Cardiol, 2001. 96(4):381-7. Li , Y.Y. and A . M . Feldman, Drugs, 2001. 61(9): 1239-52. Kim, H.E., et al., J Clin Invest, 2000.106(7):857-66. Ducharme, A., et a\.,JClin Invest, 2000. 106(l):55-62. Peterson, J.T., et al., Cardiovasc Res, 2000. 46(2):307-15. Romanic, A .M. , et al., Life Sci, 2001. 68(7):799-814. Tyagi, S.C.,etal.,y Cell Physiol, 1996. 167(l):137-47. Peterson, J.T., et al., Circulation, 2001. 103(18):2303-9. Tyagi, S.C., et al., Mol Cell Biochem, 1996. 155( 1): 13-21. Weber, K.T., etal., Circulation, 1990. 82(4): 1387-401. Curci, J.A., Qtal,JClin Invest, 1998. 102(11):1900-10. Li , J., etal., Cardiovasc Res, 2002. 56(2):235-47. Klinberg, M., Arch Biochem Biophys, 1958. 75:376-86. Granville, D.J., et al., Proc Natl Acad Sci USA, 2004.101(5):1321-6. Epub 2004 Jan 20. Sutter, T.R., et al., J Biol Chem, 1994. 269(18): 13092-9. Altschul, S.F., etal., J Mol Biol, 1990. 215(3):403-10. Franklin, J., Ann N YAcad Sci, 1993. 700:145-52. Slonim, D.K., Nat Genet, 2002. 32(Suppl):502-8. Schena, M. , et a l , Science, 1995. 270(5235):467-70. Boguski, M.S., T.M. Lowe, and C M . Tolstoshev, Nat Genet, 1993. 4(4):332-3. Benson, D.A., et a l , Nucleic Acids Res, 1998. 26(l):l-7. Velculescu, V.E., etal, Cell, 1997. 88(2):243-51. 2 1 7 [335] Kawai, S., etal., Jpn CircJ, 1987. 51(12):1385-92. [336] Gustafsson, A.B., et a l , Circulation, 2002. 106(6):735-9. [337] Hill, A. V., Br Med Bull, 1999. 55(2):401-13. [338] Wasinger, V.C., et al., Electrophoresis, 1995. 16(7):1090-4. [339] Gorg, A., et a l , Electrophoresis, 2000. 21(6): 1037-53. [340] Issaq, H.J., Electrophoresis, 2001. 22(17):3629-38. [341] Han, D.K., et al., Nat Biotechnol, 2001. 19(10):946-51. [342] Schiller, N.B., et al., J Am Soc Echocardiogr, 1989. 2(5):358-67. 218 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
https://iiif.library.ubc.ca/presentation/dsp.831.1-0099788/manifest

Comment

Related Items