Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cell death dynamics monitoring using Raman micro-spectroscopy Karimbabanezhadmamaghani, Pooya 2015

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

Item Metadata

Download

Media
24-ubc_2015_may_karimbabanezhadmamaghani_pooya.pdf [ 5.47MB ]
Metadata
JSON: 24-1.0167694.json
JSON-LD: 24-1.0167694-ld.json
RDF/XML (Pretty): 24-1.0167694-rdf.xml
RDF/JSON: 24-1.0167694-rdf.json
Turtle: 24-1.0167694-turtle.txt
N-Triples: 24-1.0167694-rdf-ntriples.txt
Original Record: 24-1.0167694-source.json
Full Text
24-1.0167694-fulltext.txt
Citation
24-1.0167694.ris

Full Text

CELL DEATH DYNAMICS MONITORING USING  RAMAN MICRO-SPECTROSCOPY   by Pooya Karimbabanezhadmamaghani   B.Sc., Sharif University of Technology, 2012  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Physics and Astronomy)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  February 2015  © Pooya Karimbabanezhadmamaghani, 2015 ii Abstract  Biopharmaceuticals play a crucial role in curing diseases like Cancer and diabetes. Bioreactors are the heart of the industry. Cell losses due to cell death such as apoptosis and necrosis in the bioreactor decreases production efficiency and subsequently increases the cost of production. Furthermore, the study of apoptosis and necrosis cell death mechanisms has a great scientific and clinical importance in cancer therapy. In this project, Raman micro-spectroscopy is used to study apoptosis and necrosis in Chinese’s Hamster Ovary (CHO) cells that are one the main host cell lines used in the production of biopharmaceuticals.  Apoptosis and Necrosis were induced in CHO cells using camptothecin and oxygen and glucose deprivation. The changes in the chemical composition of these enriched apoptotic and necrotic cell cultures were then analyzed using Raman spectroscopy which revealed novel biological concepts of the cell death process. Moreover, highly distinguishing Raman characteristics were identified for each death mode. These observations made by Raman spectroscopy were confirmed using a broad range of conventional and advanced biological assays in the field ranging from FACS analysis and fluorescent dyes to fluorescence microscopy. Studying Raman Spectra gave a clear image about DNA, RNA and Protein level changes during the process of apoptosis and necrosis in CHO cells. Using Principle Components Analysis (PCR) enabled viable, necrotic, early and late apoptotic populations to be clearly distinguished. This technology may provide the basis for the development of a non-invasive probe to monitor and predict cell death in bioreactor cultures in real-time and possibly allow cultures to avoid  entering the cell death phase. In addition, the vast majority of cancer treatment methods involve cell death and apoptosis and, therefore, improving our knowledge about the biology of cell death will help support and advance research and treatment in this area.    iii Preface  I was the major contributor of this thesis together with my supervisor Professor James Piret. I designed and performed most of the experiments. Dr. Stanislav Konorov and Dr. Hans George Schulze helped me with developing the experimental setup and the data processing methods.                       iv Table of Contents  Abstract ................................................................................................................................................................. ii Preface .................................................................................................................................................................. iii Table of Contents .............................................................................................................................................. iv List of Tables ...................................................................................................................................................... vii List of Figures ................................................................................................................................................... viii List of Abbreviations ......................................................................................................................................... xi Acknowledgements .......................................................................................................................................... xiii Dedication .......................................................................................................................................................... xiv Chapter 1: Introduction ....................................................................................................................................... 1 1.1  Classes of cell death .................................................................................................................. 1 1.1.1    Apoptosis ............................................................................................................................................ 1 1.1.2    Necrosis ............................................................................................................................................... 3 1.1.3    Autophagy .......................................................................................................................................... 5 1.2  Raman spectroscopy ................................................................................................................ 6 1.3  Biomedical application of Raman spectroscopy ......................................................... 12 1.3.1    Sub-cellular level ........................................................................................................................... 12 1.3.2    Cellular level .................................................................................................................................... 13 1.3.3    Tissue level ...................................................................................................................................... 14 1.4  Study of cell death using Raman micro-spectroscopy ............................................... 15 Chapetr2: Materials and methods ................................................................................................................. 22 2.1  Cell line ....................................................................................................................................... 22  v 2.2  Cell culture ................................................................................................................................ 22 2.2.1    Maintenance medium and cultures ....................................................................................... 22 2.2.2    Fed-batch culture .......................................................................................................................... 23 2.2.3    Cell banking ..................................................................................................................................... 23 2.3  Cell death induction ............................................................................................................... 24 2.3.1    Necrosis induction ........................................................................................................................ 24 2.3.2    Apoptosis induction ..................................................................................................................... 25 2.4  Flow cytometry to detect apoptosis and necrosis ...................................................... 26 2.5  ATP measurement .................................................................................................................. 26 2.6  Protein measurement ........................................................................................................... 27 2.7  DNA measurement ................................................................................................................. 28 2.8  RNA measurement ................................................................................................................. 29 2.9  Raman spectroscopy ............................................................................................................. 30 2.9.1    Sample preparation ...................................................................................................................... 30 2.9.2    Cell selection method .................................................................................................................. 30 2.9.3    Single cell analysis ........................................................................................................................ 33 2.9.4    Raman microscopy equipment ................................................................................................ 33 2.10  Data analysis of Raman spectra ...................................................................................... 36 Chapter3: Results and discussion .................................................................................................................. 37 3.1  Apoptosis induced cells ........................................................................................................ 37 3.1.1    PCA classification of apoptotic cells ...................................................................................... 51 3.2  Necrosis induced cells........................................................................................................... 54 3.2.1    PCA classification of necrotic cells ......................................................................................... 64 Chapter 4: Conclusions and future prospects .......................................................................................... 66  vi Bibliography ...................................................................................................................................................... 69 Appendices..........................................................................................................................................................76 Appendix A: custom written software for spectra data analysis ................................... 76                        vii List of Tables   Table 1-1: Literature review of apoptosis study using Raman micro-spectroscopy. .............................17 Table 1-2: Literature review of necrosis study using Raman micro-spectroscopy. ............................. 20 Table 3-1: Raman peak assignments of interest for this project. ............................................................. 40      viii List of Figures  Figure 1-1: Diagram of different cell death modes in-vitro. It shows different possible pathways from a viable cell to a necrotic dead cell. ................................................................................................................... 5 Figure 1-2: Different scattering modes. ................................................................................................................ 7 Figure 1-3: Energy level diagram of Raman and Rayleigh scattering. ........................................................9 Figure 1-4: Raman spectrum from the detected photon intensity and their Raman Shifts from viable CHO cells. ............................................................................................................................................................... 10 Figure 1-5: Schematic diagram of spontaneous confocal Raman micro-spectroscopy. ........................ 11 Figure 2-1: Oxygen-Glucose deprivation to induce necrosis. ..................................................................... 24 Figure 2-2: Ternary complex of DNA, topo I and CPT (Yikrazuul, 2009). ........................................... 25  Figure 2-3: Methanol fixed cells’ configuration on the gold mirror. ...................................................... 31 Figure 2-4: Pixel configuration of cells’ bulk under Raman microscope. .............................................. 32 Figure 2-5: Single cell spectrum acquisition setup. ...................................................................................... 33 Figure 2-6: Calibration spectrum of Raman microscope using silicon wafer. ..................................... 34 Figure 2-7: Laser intensity vs. time after turning on the laser on........................................................... 35 Figure 2-8: Laser wavelength fluctuation vs. time measured by silicon wafer. ................................... 36 Figure 3-1: Flow cytometric analysis of Annexin V/PI stained cells of viable and 6,12 and 24h Camptothecin treated CHO cells. ............................................................................................................................... 37 Figure 3-2: 100 raw spectra of CHO cells treated with Camptothecin for 24 hours. ....................... 38 Figure 3-3: 100 processed (background subtracted and base line corrected) spectra of CHO cells treated with Camptothecin for 24 hours................................................................................................................. 39 Figure 3-4: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1). .............................................................................................................................. 41 Figure 3-5: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 600 cm-1 to 980 cm-1. ................................................................. 42  ix Figure 3-6: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1020 cm-1 to 1400 cm-1. ............................................................... 42 Figure 3-7: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1400 cm-1 to 1800 cm-1. .............................................................. 43 Figure 3-8: Loading of PC 1 depicts specific peak changes between viable/6h, viable/12h,  viable /24h populations. ............................................................................................................................................................. 43 Figure 3-9: Loading of PC 1 depicts specific peak changes between 6h/12h, 6h/24h,  12h /24h population. ......................................................................................................................................................................... 44 Figure 3-10: RNA peak (809 cm-1) change in apoptosis induced culture. ............................................. 45 Figure 3-11: Nucleic Acid peaks change in apoptosis induced culture. ................................................... 46 Figure 3-12: Lipid peaks change in apoptosis induced culture.................................................................. 47 Figure 3-13: Protein peaks change in apoptosis induced culture. ............................................................ 49 Figure 3-14: Chemical components changes in apoptosis induced culture measured by biochemical assays. .......................................................................................................................................................... 50 Figure 3-15: Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable, 6h, 12h, 24 h CPT treated populations. ....................................................................................................... 51 Figure 3-16: Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable and 6h CPT treated populations. ................................................................................................................... 52 Figure 3-17: Score plots of Principle Components Analysis (PCA), PC 4 vs. PC 1, comparing 12h and 24h CPT treated populations. ............................................................................................................................. 53 Figure 3-18: Flow cytometric analysis of Annexin V/PI stained cells of viable and 24h oxygen-glucose deprived CHO cells. ......................................................................................................................................... 54 Figure 3-19: 100 raw spectra of 24 hours oxygen-glucose deprived CHO cells................................... 55 Figure 3-20: 100 processed (background subtracted and base line corrected) spectra of 24 hours oxygen-glucose deprived CHO cells ........................................................................................................................... 56  x Figure 3-21: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 600 cm-1 to 980 cm-1 .................................................................. 57 Figure 3-22: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1020 cm-1 to 1400 cm-1. ............................................................... 57 Figure 3-23: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1400 cm-1 to 1800 cm-1. .............................................................. 58 Figure 3-24: Loading of PC 1 depicts specific peak changes between viable and 24 hours oxygen-glucose deprived populations....................................................................................................................................... 59 Figure 3-25: RNA peak (809 cm-1) changes between viable and 24 hours oxygen-glucose deprived populations......................................................................................................................................................................... 60 Figure 3-26: Nucleic Acid peaks change in necrosis induced culture. .................................................... 61 Figure 3-27: Protein peaks change in necrosis induced culture. .............................................................. 62 Figure 3-28: Lipid peaks change in necrosis induced culture. .................................................................. 62 Figure 3-29: Chemical components change in necrosis induced culture measured by biochemical assays. .................................................................................................................................................................................. 63 Figure 3-30: Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable and 24h oxygen-glucose populations. ........................................................................................................... 65   xi List of Abbreviations        A                          Adenine      ADP                       Adenosine diphosphate      ATG                      AuTophaGy-related      ATP                      Adenosine Triphosphate       BCA                      Bicinchoninic Acid      CAD                     Caspase-activated DNase       CARS                    Coherent anti-Stokes Raman spectroscopy CCD                     Charge-Coupled Device   CFDA                   Carboxyfluorescein diacetate (Fluorescent Dye) CHO    Chinese Hamster Ovary CPT                     Camptothecin DAPI                    4',6-diamidino-2-phenylindole (Fluorescent Dye) DMSO             Dimethyl Sulfoxide ES cells                  Embryonic Stem cells FACS                     Fluorescence-Activated Cell Sorting FT-Raman               Fourier Transform based Raman spectroscopy G                          Guanine  GFP     Green Fluorescence Protein HMGB1                   High-mobility group protein B1 IDC                       Infiltrating Ductal Carcinoma IF                         Immunofluorescence IGF                       Insulin-like Growth Factor IR                         Infrared  LC3                       Light Chain 3 (Protein)  xii LDH                      lactate dehydrogenase LMP                      Lysosomal Membrane Permeabilization MCF-7                   Michigan Cancer Foundation MES cells               Mouse Embryonic Stem cells MMP                    Mitochondrial Membrane Permeabilization NMR                    Nuclear Magnetic Resonance  NAD                     Nicotinamide adenine dinucleotide NCCD                   Nomenclature Committee on Cell Death NIR                       Near Infrared  PBS                      Phosphate buffered saline PC                       Principle Component  PCA                     Principle Component Analysis  PCD                     Programmed Cell Death  PI                        Propidium Iodide  PS                       Phosphatidylserine RR                       Resonance Raman ROS                     Reactive Oxygen Species  SERS                    Surface-Enhanced Raman Spectroscopy SRS                      Stimulated Raman Spectroscopy SS-DNA                 Single Strand DNA SNV                     Standard Normal Variate  SVM                     Support Vector Machine  TB                        Trypan-Blue  tPA              Tissue Plasminogen Activator TUNEL                  Terminal deoxynucleotidyl transferase dUTP Nick End Labeling UV                        Ultra Violet   xiii Acknowledgements  First, I am grateful to the Almighty God, who has bestowed his love and light upon me since the first moment in my life and my family members that have supported me in good and bad times. There are many people who without their help, I would have never accomplished this project and I would like to express my sincere thanks to all of them. Professor James Piret (supervisor), Professor Robin Turner (collaborative supervisor), Professor Stefan Reinsberg (co-supervisor and thesis reader), Dr. Stanislav Konorov (research advisor at Raman spectroscopy lab and thesis advisor), Dr. Hans George Schulze (advisor for data processing), Dr. Malcolm Kennard (thesis reader and editor), Chris Sherwood (research assistance at cell culture lab), Andy Johnson (research assistant at FACS facility) and all staff of the Michael Smith Laboratories and the Biomedical Research Centre.            xiv      Dedication    To my late grandfather, Who taught me lessons for entire life…       1 Chapter 1: Introduction  1.1  Classes of cell death       Living cells can die in various ways. Different classes of cell death can be defined based on morphological, enzymological, and functional changes that may occur during cell death. Also, immunological changes during death are an important factor in the classification of cell death types. Some types of cell death may induce an immune system response (immunogenic), whereas others do not (non-immunogenic)(Melino 2001). Since 2005, the Nomenclature Committee on Cell Death (NCCD) have made recommendations to unify definitions and to provide guidance to distinguish between the various types or classes of cell death (Kroemer et al. 2005). Although, in the literature, there have been many definitions and categorizations of cell death, four classes are now accepted as the main types of cell death: Apoptosis, Necrosis, Autophagy and Cornification (Kroemer et al. 2009).  1.1.1    Apoptosis       The term Apoptosis comes from a Greek route: “Apo: from/off/without and ptosis: falling” (Kerr 1965).  It was initially discovered in the 18th Century and later the work by Sydney Brenner, Robert Horvitz and John. E Sulston resulted in the 2002 Noble prize in medicine. When a cell goes through apoptosis, changes occur in both the morphological characteristics of cell and the chemical composition of cells. Morphological changes are used as the 'gold standard' for distinguishing apoptosis. Distinct changes to the internal organelles occur in final stages of apoptosis such as rounding up of the cell, reduction in cellular volume, nuclear fragmentation, condensation of chromatin, and blebbing of the plasma membrane. A bleb is a bulge or protrusion of cell’s plasma membrane that resulted from localized decoupling of cell’s cytoskeleton from plasma membrane (Fackler & Grosse 2008).  In addition, for cells in complex organisms, dying cells can trigger an immune response where they become engulfed by resident  2 phagocytes (Kroemer et al. 2009). In cell culture or in tissue, apoptotic cells are more often seen as single cells, whereas necrotic cells are more often seen in cluster-form (Darzynkiewicz et al. 1997). In some literature, it is assumed that Programmed Cell Death (PCD) and apoptosis are synonymous, where PCD naturally occurs and governs the normal life span of a cell. However, this assumption is not totally correct, since apoptosis usually requires induction by an external stimulus or during normal physiological development, for example in the development of fingers, which can show non-apoptotic features (Roach & Clarke 2000). During apoptosis, cysteine aspartate-specific proteases, enzymes that degrade proteins inside a cell are converted into their active forms and termed caspases (Alnemri et al. 1996). Caspase activation is one of classical hallmarks of apoptosis (Wlodkowic et al. 2011). Caspase activation can be detected by analysis of the active caspases or cleaved caspase substrates using various methods such as colorimetric or fluorogenic substrate-based assays, fluorescence-activated cell sorting (FACS) accompanied by antibody based immunofluorescence (IF) microscopy quantification, and immunoblotting (Kroemer et al. 2009). DNA fragmentation can be also used as another indication of apoptosis and can be detected using methods such as DNA ladders, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assays, FACS quantification of hypodiploid cells or Raman single cell imaging. Other less common techniques can also be used to study apoptosis based on physiological changes within the cell such as dissipation of mitochondrial transmembrane potential (ΔΨm), activation of proapoptotic Bcl-2 family proteins, mitochondrial membrane permeabilization (MMP), exposure of phosphatidylserine (PS) residues that produce ‘eat me’ signals for normal neighboring cells, Reactive Oxygen Species (ROS) over generation and single strand DNA accumulation (ssDNA)(Kroemer et al. 2009; Green & Kroemer 1998; Yang et al. 2002; Green & Kroemer 2004). Although caspase activation plays a pivotal role in apoptosis and is used as the main method to detect apoptosis, in order to confidently identify apoptosis, it is  3 advised to accompany the detection of caspase activation with one of the other detection methods described above (Wlodkowic et al. 2011).  1.1.2    Necrosis      Necrosis term is Greek word that means “death” (Proskuryakov et al. 2003). Historically, necrosis was considered as accidental and uncontrolled cell death as opposed to PCD that was caused by extreme environmental conditions such as heat, severe mechanical trauma, toxins and ischemia (Majno & Joris 1995). However, recently increasing evidence has concluded that necrosis is the result of well-orchestrated form of cell death regulated by metabolic pathways or transduction pathways signal sets (Festjens et al. 2006). The cell undergoes distinct morphological changes during necrosis including increasing cell volume, plasma membrane rupture, loss of intracellular organelles and contents, and moderate chromatin condensation. In addition, several other changes happen within the cell including mitochondrial membrane permeabilization (MMP) (controlled by cyclophilin D), lysosomal membrane permeabilization (LMP), ROS over production by Fenton reactions in lysosomes, activation of phospholipases, lipoxygenases and sphingomyelinases (causing a decrease intracellular lipid levels), increasing cytosolic concentrations of calcium ion (Ca2+) (causing mitochondrial overload and activation of non-caspase proteases (calpains and cathepsins)), uncoupling or reactive oxygen species production change in mitochondria (ROS, nitroxidative stress with nitric oxide or analogous compounds), decreased ATP levels, and release of extracellular cytosolic enzymes such as cytoplasmic lactate dehydrogenase(LDH) (Festjens et al. 2006; Nicotera et al. 1999; Kroemer et al. 2009). There are many detection methods that can be used to identify necrosis such as colorimetric/fluorogenic substrate-based assays that indicate activation of calpain and cathepsin, luminometric assessments of ATP/ADP ratio to detect drop in ATP level, FACS quantification with lysomorphotropic probes to show lysosomal membrane permeabilization (LMP). Plasma membrane rupture can occur in late apoptosis and late autophagy (see Section 1.1.3), but early  4 detection of plasma membrane rupture or permeabilization usually indicates that the death mechanism is necrosis. Plasma membrane rapture can be detected by conventional trypan-blue staining techniques that are common to most biology-based laboratories. Other methods to detect necrosis include colorimetric/fluorogenic substrate based assays of culture supernatant to detect the secretion of extracellular cytosolic enzymes, and FACS quantification by ROS-sensitive probes to detect ROS over generation in mitochondria (Kroemer et al. 2009). Recently, an innovative and more accurate method to detect necrosis has been developed based on the high-mobility group B1 protein (HMGB1) (Golstein & Kroemer 2007). In normal cells, HMGB1 resides in the nucleus that can attach to DNA and promote the assembly of proteins in target section of DNA (Ito et al. 2006). However, in the cells undergoing necrosis, HMGB1is passively released into the extracellular medium (Scaffidi et al. 2002). By immunoblotting of culture medium with HMGB-1-specific antibodies, necrosis can be accurately detected (Ito et al. 2006). Apoptosis and autophagy, in their final stages, show necrosis death like characteristics, and is sometimes termed 'secondary necrosis' (Kanduc et al. 2002). To complicate matters further, often cells or tissues experience the various forms of cell death simultaneously. Therefore, distinguishing between late apoptosis, secondary necrosis, late autophagy and primary necrosis in a cell culture is not straightforward issue. Figure 1-1 shows the relationship between the different types of cell death. One important advantage of the HGMB-1 assay is that the method can distinguish between these various types of cell death and detect primary necrosis (Krysko et al. 2008). Despite there being many methods to detect necrosis, the methods are not totally reliable and necrosis is often identified once the absence apoptotic or autophagic indications has been determined (Golstein & Kroemer 2007). More recently, efforts have focused on identifying changes to levels of various biochemical compounds and markers in cells undergoing necrosis using Raman confocal micro-spectroscopy. In studies where necrosis is induced by both oxygen and glucose deprivation, Raman analysis has shown that lipid and RNA levels decrease in necrotic cells compared to normal cells, whereas protein content increases. Furthermore, Raman  5 spectroscopy can also detect some conformational changes in both proteins and nucleic acids in necrotic cells (Kunapareddy et al. 2008).  Figure 1-1: Diagram of different cell death modes in-vitro. It shows different possible pathways from a viable cell to a necrotic dead cell.  1.1.3    Autophagy       Autophagy is a term that has Greek root. (Auto “self” and phagein “ to eat”) (Lin et al. 2013). Autophagy is an intracellular degradation mechanism that digests cytoplasmic components within lysosomes (Levine & Klionsky 2004). Autophagy is usually induced by nutrient starvation or mild environmental stresses such as lack of essential nutrients, nitrogen or carbon deprivation, withdrawal of essential amino acids, low dose UV radiation, mild physical stress, and hypoxia. Emerging results suggest that autophagy is a survival mechanism for cells to escape cell death. This ability has been observed to help cancer cells survive chemotherapeutic drugs, and treatments such as hypoxia and nutrient deprivation in un-vascularized solid tumors or tumor centers (Eisenberg-Lerner et al. 2009). When a mammalian cell undergoes autophagy,  6 intracellular material is sequestered within a double-membrane vesicle called an autophagosome. Autophagosomes mature and then fuse with lysosomes, where their contents are degraded to simpler components such as amino acids or nucleotides (Mizushima et al. 2010). It is also observed that morphological changes can occur in cells undergoing autophagy such as vacuolization of the cytoplasm and the formation of double–membrane autophagic vacuoles (Baehrecke 2005). Autophagosome formation involves the microtubule associated protein 1 light chain 3, or MAP1-LC3, commonly known as LC3 (Matsushita et al. 2007). LC3 has two forms inside the cell, the soluble form (LC-I), and the lipidated form of LC3-II, which is found attached to the outer membrane of the autophagosome. Measurement of LC3-II is a good indicator of the accumulation of autophagasomes and the induction of autophagy. Thus the ratio of the different forms of LC3 or the conversion of LC3-I to LC3-II, which can easily be measured using Western blotting or immunofluorescence microscopy (Kroemer et al. 2009), can be used to detect changes within cells undergoing autophagy compared to normal cells (Konorov et al. 2012). To measure LC3, it is also possible to use a fluorescent protein like GFP that binds to LC3 and produces fluorescent punctate spots called GFP-LC3-positive puncta. Counting of these punctate spots can give a numerical estimation about accumulation of autophagosomes and the degree of autophagy (Klionsky et al. 2008). There is another family of genes that are also involved in autophagosome formation. Generally, these genes are called AuTophaGy-related (ATG genes) and the corresponding products of these genes are referred to as the ‘core’ autophagy machinery. These genes and their products regulate autophagosome formation (Xie & Klionsky 2007). Measurement of the gene products using genetic assays such as RNA interference is another method to detect autophagy (Kroemer et al. 2009).  1.2  Raman spectroscopy      Discovered by Sir C.V. Raman in 1928, Raman spectroscopy is a laser based technique that can monitor vibrational, rotational and low-frequency modes of molecules. This relies on an  7 inelastic scattering that is known as Raman scattering (Schäfer & Schmidt 2012). When a monochromatic laser light collides with a molecular structure, the vast majority of the incident photons scatter without any change in their energy and frequency. This type of scattering is called Rayleigh scattering. In fact during Rayleigh scattering, the sample absorbs one photon and releases another photon with same energy and frequency (an elastic interaction) but most likely in a different direction from the incident photon (Figure 1-2). Because in this interaction, the sample molecule does not gain energy, it returns to its initial energy level after jumping to a higher energy level during the interaction (Figure 1-3). Approximately one of every 106 -108 photons, scatters with a changed energy and frequency (an inelastic scattering). These photons are Raman scattered photons. Most Raman scattered photons lose energy and this is called Stokes Raman scattering and the frequency change of the incoming photon is the ‘Stokes frequency shift’. In rare cases, the sample scatters photons with higher energy. This type of scattering is called 'Anti-Stokes' Raman scattering and the frequency increase is known as the ‘Anti-Stokes frequency shift’.  Figure 1-2: Different scattering modes.      Sample    Incident  beam Rayleigh Scattering Stokes Raman Scattering Anti-Stokes Raman Scattering  8      The shift in frequency (Raman Shift) of scattered photon is commonly expressed in terms of cm-1 wave-numbers. If we suppose that the wavelength of an incident photon is λ0 and the wavelength of a scattered photon is λ1, then wave-number (Δ𝜔) can be obtained with the following formula: Δ𝜔 =1𝜆0−1𝜆1         Equation.1      Since usually, the wavelength of a scattered photon is longer than the incident photon, the wave-number is usually positive. We can use the Einstein formula for the energy of a photon and derive the following relation between changes in energy (ΔE) and the frequency (Δυ) change for the scattered photon (where c is the light speed 3×108 m/s; h is the Plank constant 6.62× 10-34 m2kg/s): υ=c/λ    ⇛     Δυ= ν0 – ν1 = cΔω       Equation.2 E=hν     ⇛     ΔE= E0-E1=hcΔω       Equation.3      As the conservation of energy law predicts, the energy gain or loss of scattered photons is equal to the energy loss or gain of the sample molecule and this is equal to the change in the vibrational ground states before and after the interaction with the photon (Figure 1-3). The vibrational ground states are unique for each chemical bond, such that the energy differences between incident and scattered photon acts as a fingerprint for each chemical bond.   9  Figure 1-3: Energy level diagram of Raman and Rayleigh scattering.       As Figure 1-3 shows, by monitoring the frequency of scattered photons and counting them for every wave-number (proportional to the Raman frequency shift), a spectrum is generated that can give a unique and precise indication of the chemical composition of the sample. The intensity of photons with certain Raman shifts is an indicator of the level of the corresponding chemical bond. For abundant biological components such as DNA, RNA or proteins, the Raman spectrum and dominant wave-numbers are evident. As a result, we can monitor the abundant components of biological samples. Figure 1-4 shows a spectrum that has been taken from CHO  10 cells indicating the dominant peaks that correspond to biological components.  Figure 1-4: Raman spectrum from the detected photon intensity and their Raman Shifts from viable CHO cells.       Raman spectroscopy uses a laser as a source of monochromic light, with near ultraviolet and near infrared commonly used in Raman microscopy. For high-resolution, confocal Raman micro-spectroscopy, a helium-neon laser with a wavelength of 632.8 nm in the red part of the visible spectrum is a usual choice. For most biological application, a near infrared diode laser (785 nm) or a Nd:YAG laser (1064nm) are common (Ellis & Goodacre 2006).   A band-pass optical filter known as maxline laser transmitting filter (see Figure 1-5) narrows down laser light around its peak wavelength. A Dichromic beam splitter, long wave pass filter reflects the laser beam and is transparent for scattered photons guides laser beam to objective lens, which focuses laser beam on the point of interest on sample and collects scattered photons from sample. The photons with longer wavelength than the incident photons go through long wave pass filter and are focused by an optical lens to a narrow slit (~25µm) spectrometer. The  11 frequency of the collected photons is measured by a spectrometer and converted to a digital signal by a CCD camera. A computer saves and analyses the recorded data.     Figure 1-5: Schematic diagram of spontaneous confocal Raman micro-spectroscopy.        There are several variations of Raman spectroscopy that can be used to obtain specific resolution or to acquire specific information of sample. Major variations of Raman spectroscopy  12 are Spontaneous Raman, Stimulated Raman Spectroscopy (SRS), Resonance Raman (RR), Non-resonance Raman, and Coherent anti-Stokes Raman spectroscopy (CARS).  1.3  Biomedical application of Raman spectroscopy       Raman scattering was discovered in 1928, but it took more than half a century until the invention and large-scale production of high quality and affordable lasers that allowed the development of and use of Raman spectroscopy in both scientific and industrial fields. It was in early 1990s that Raman spectroscopy became used in biomedical studies with the number of papers involving Raman Spectroscopy published in biomedical journals increasing from ~1400 in 1995 to over 6500 in 2013 (Web of knowledge). There are several critical characteristics about Raman spectroscopy that gives it significant advantage over other methods of analysis. The most important advantage of Raman spectroscopy is the fact that samples and cells etc. can be studied non-invasively and non-destructively and as a result, Raman Spectroscopy is ideally suited for the long term monitoring of biological processes and phenomena. In the case of studying cell death, most conventional methods either destroy the sample in the process of measurement or require sampling that can perturb the cell death process. In addition, water does not interfere with the Raman signal, unlike infrared spectroscopy, making this method ideal for studying of cells in their aquatic environment and having a broad range of application in life sciences. The large range of spatial resolution for Raman Spectroscopy allows it to be applied at sub-cellular, cellular and tissue scales.  1.3.1    Sub-cellular level       Raman spectroscopy has widely been used for study of sub-cellular organelles. Applying immuno-fluorescence images as a reference and using specific software, it has been possible to reconstruct an image of lipid droplets, nucleolus, nucleoli and the endoplasmic reticulum in human pancreatic cancer MIA PaCa-2 (CRL-1420) and colorectal adenocarcinoma HT29 (HTB- 13 38) cells using confocal Coherent anti-Stokes Raman spectroscopy (CARS) spectra. Resolution achieved by this technique was between 200-300nm, which gave an unprecedented and precise chemical map of the sub-cellular components (El-Mashtoly et al. 2014). Certain peaks in Raman spectra contain critical information about the chemical bonds and backbone structures of biological molecules. As a result, Raman spectroscopy provides a powerful method to study the secondary structure of proteins and the process of protein folding, as well as giving a clear image of spatial configuration of nucleic acids inside cells. In addition, there are many applications for study of biopolymers using this technique (Peticolas 1975). Using Confocal Raman with resolution of 1µm and specific software, has allowed the study of cytoplasm and Cytochrome C and mitochondrial fingerprints as well as the dynamic pattern of Centrosome C and mitochondria inside apoptotic MCF-7 cells. Thus Raman spectroscopy has provided a Caspase independent pathway to detect apoptosis by observing the gradual release of Cytochrome C from mitochondrial clusters into the cytoplasm and has been used to study the effect of the anticancer drug (paclitaxel) on cancer cells (Salehi et al. 2013). Raman spectroscopy opens the way to real time monitoring of anticancer drugs effects remotely in the process of treatment without need to take sample cells or tissue from patient. In addition, DNA fragmentation and nuclear condensation have been reportedly used for detecting of apoptosis and necrosis using Raman Spectroscopy (Brauchle et al. 2014; Krafft et al. 2006; Zoladek et al. 2011; Ong et al. 2012; Owen et al. 2006; Notingher et al. 2004).  1.3.2    Cellular level       Because of its non-invasive nature, Raman spectroscopy also has a big advantage over conventional methods like fluoroscopy imaging to monitor and characterize the differentiation process of stem cells. CARS micro-spectroscopy of stem cells has been used to show that DNA and RNA content of stem cells decreases as they differentiate into specific cell types. Raman micro-spectroscopy has also been used to distinguished between certain cell types that are  14 differentiated from stem cells by monitoring the presence of hydroxyapatite as an indicator of the formation of bone cells and lipid as a marker for adipocytes cells (Downes et al. 2011). The study of mouse embryonic stem cells (mESC) has shown that differentiated cells have a relatively higher tryptophan ring-breathing mode near 760 cm-1 and trypsin ring-breathing mode at 854 cm-1. In addition, it was noted that differentiated cells had a higher Protein/RNA ratio. Raman images with 1µm resolution from individual cells, shows distinguished morphological distribution between differentiated and undifferentiated mESC at 780 cm-1 (O-P-O stretch, DNA) and 811 cm-1 (O-P-O stretch, RNA) peaks (Konorov et al. 2007).       Raman spectroscopy also has a broad range of applications for in vitro cellular cultures. In a novel technique, human neuroblastoma cells were treated with nuclear-targeted gold nanoparticles (AuNPs) that acted as intercellular probes. The spectra of whole cell area as well as nucleus were collected using surface-enhanced Raman spectroscopy (SERS). Principal Component Analysis (PCA) of these spectra clearly distinguished between progenitor and differentiated types of these cells. In addition, analysing spectra from the nucleus showed a higher DNA/RNA ratio and level of proteins among differentiated cells (Huefner et al. 2013).       Raman spectroscopy has also been used for bacteria identification. In a study, 30 colonies of four strains of bacteria including Staphylococcus epidermidis (1457 and 9142) and Escherichia coli were successfully differentiate using PCA analysis (Almarashi et al. 2012). Surface Raman Spectroscopy (SERS) gave a novel and fast method to discriminate between gram-positive (Enterococcus faecalis and Streptococcus pyogenes) and gram-negative (Acinetobacter baumannii and Klebsiella pneumoniae) bacteria genera (Prucek et al. 2012).  1.3.3    Tissue level  The fact that Raman spectroscopy is non-destructive and needs only a very small amount of sample (micrograms), make it very promising alternative to conventional methods for tissue identification. A variety of tissues, including glioblastoma, meningioma, and normal bladder  15 tissue, has been precisely discriminated using Raman micro-spectroscopy. Glioblastoma, a most aggressive form of brain tumour, shows a higher content of DNA compared to necrotic brain tumour tissue. Comparison between spectra of these two types of tissue gives 100% differentiation accuracy (Koljenović et al. 2005). This is a very promising result and may lead to the development of techniques that do not require biopsies and harmful tissue sampling for brain tumour diagnostics. Normal fibroadenoma and infiltrating ductal carcinoma (IDC) human breast tissue can be differentiated without need for lengthy sample preparation using FT-Raman (Fourier Transform based Raman spectroscopy) spectra. Spectra analysis showed that normal breast tissue has a higher level of fat (83% of all chemical composition) compared to fibroadenoma (26%) and IDC tissues (2%). In contrast, IDC tissue had a higher level of collagen (40%)and cholesterol (19%) compared to normal (1%, 10%) and Fibroadenoma (16%, 15%) tissues (Bitar Carter et al. 2004). Results like these are showing that Raman-based diagnostics probes can be future of biopsy-free cancer diagnostics that will give faster, easier and less harmful diagnostics possibilities.  1.4  Study of cell death using Raman micro-spectroscopy The most common mode of cell death that has been investigated by Raman micro-spectroscopy is apoptosis, most likely because apoptosis is the regular class of cell death that is induced by chemotherapeutics agents for cancer treatment. In a vast majority of these studies chemicals were used to induce apoptosis on a group of cells and afterwards, using Raman Spectroscopy, different peaks corresponding to lipids, proteins, nucleic acids (DNA and RNA) in spectra of apoptotic cells were compared to normal (control) cells (Ladiwala et al. 2013; Brauchle et al. 2014; Ong et al. 2012; Owen et al. 2006; Notingher et al. 2004). In other studies high spatial resolution Raman micro-spectroscopy (~1µm) was used to monitor the morphological and distribution evolution of essential cell components like DNA, protein or lipid among apoptotic cells studies (Krafft et al. 2006; Zoladek et al. 2011). In one case (Salehi et al. 2013), the dynamic  16 pattern of Cytochrome C, a more specific protein involved in apoptosis, was probed using high resolution Raman spectroscopy. Table 1 and Table 2 summarize the most significant literature on the study of apoptosis and necrosis using Raman micro-spectroscopy.   17 Table 1-1: Literature review of apoptosis study using Raman micro-spectroscopy. Cell Type Induction method and Spectra normalization methods Endorsement by Conventional methods Raman spectra analysis and Features Reference Saos-2 SW-1353   7 days incubation in room temp.  Vector normalization  FACS Annexin V and PI Caspase 3&6 activation detection   Significant PCA differences reflect the modality and the stages of cell death. >90% Classification of single cell Raman spectra to predefined cell death modalities by Support Vector Machine (SVM) technique. Saos-2: Increasing Raman bands (795, 1375 cm-1) and decreasing signals (1003, 1658 cm-1) for early apoptotic cells. Reduction of further signals (1047 cm-1) for late apoptotic. SW-1353: decreasing signals at 1047 cm-1, but an increasing band at 1375 cm-1 for early apoptotic. Decreasing band at 786 cm-1 and an increasing signal at 1437cm-1 for late apoptotic.  Brauchle et al. 2014  Hippocampal progenitors stem cells isolated from Wistar rats. TNF-α and IFN-γ (20 ng/ml) treatment over a 48 hours course. No mentioned normalization. DNA Fragmentation detection using TUNEL assay. Cell viability (Cell metabolism) assessment using  MTT assay. Increase in Raman band located in the region 775~875 cm-1, as a sign of nuclear condensation seen in the early stages of apoptosis.  Ladiwala et al. 2013 MCF-7                 100 µM Paclitaxel for 30 minutes  No normalization        No conventional biological assay                  Dynamic pattern of Cytochrome C and mitochondrion clusters have been mapped inside individual cells using visualized correlation between spectra of cell and reference spectra of cytochrome C and mitochondria. Apoptosis via a Caspase independent pathway has been detected by spotting gradual release of Cytochrome C from mitochondrial clusters into cytoplasm in paclitaxel treated cells.            Salehi et al. 2013         18 Cell Type Induction method and Spectra normalization methods Endorsement by Conventional methods Raman spectra analysis and Features Reference Human embryonic lung fibroblast cells (strain L132)   1 mM Glyoxal treatment for 24 hours Normalized to the phenylalanine peak at  1003 cm-1 No conventional biological assay  Raman microphotographs of individual cells with the resolution of 1µm have been drawn to show changing shape of nuclei, microtubule, cytoplasmic inclusions, vesicles and cytoplasm, peripheral membrane during apoptosis induction process.  Cell nucleus shrinkage (Pykonosis) has been spotted in early stages of apoptosis. Decreasing intensity of nucleic acid bands in cluster-averaged Raman spectra of the nucleus and cytoplasm potentially indicates degradation and conformational changes of DNA and RNA. Rounding of cells, a further intensity decrease of nucleic acids bands, fragmentation of the nucleus, disappearance of lipid bodies, and formation of blisters at the cell surface have been spotted during late stages of apoptosis.  Higher protein content with nucleic acids incorporated in blisters compared to peripheral membrane of untreated cells. Krafft et al. 2006 Human breast cancer cells (MDA-MB-231)  300µM Etoposide treatment for 6 hours with sampling in 2 hours intervals Normalization to Phenylalanine (1005 cm-1). Wild field Fluorescence microscopy of cells stained by Annexin V, 6-CFDA and DAPI.  Increasing in DNA band intensity (788 cm-1) in apoptotic cells (~1.5 folds at 6h), as an indicator of DNA condensation.  Increasing in lipid band intensity (1659, 1449, 1301 cm-1),  (~2 folds at 6 h) , as an indicator of lipid bodies accumulation  Raman microphotographs of individual cells with the resolution of 1µm shows:  1-Spreading of DNA content (788cm-1) from nuclei into the cytoplasm as apoptosis proceeds. As an indicator of DNA fragmentation. 2-Formation and enlarging of lipid (1659cm-1) clusters inside cytoplasm as apoptosis proceeds.  Zoladek et al. 2011 Human chronic myelogenous leukemia cells (K562 cell line)        72 hours treatment with Cytosine arabinoside (300 µM) Normalization to max of each spectra for pure biological components No normalization for cell spectra 100µM Etoposide treatment for 48 hours with sampling in 24 hours intervals      DNA fragmentation and nuclear condensation confirmed among apoptotic cells by nuclear acids staining with Hoechst-33258    Increase in membranous lipids (734 cm−1) DNA/RNA related peak reduction (794 cm−1 and the region around 1098 cm−1) Slight decrease at protein peak (1011 cm−1) Increase in Protein/lipid peak (1462, 1672 cm−1) Raman spectra of pure Actin (protein), Albumin (protein), Triolein (lipid), Phosphatidylcholine (lipid), DNA (nucleic acid), RNA (nucleic acid) and Glycogen (Polysaccharide) have been acquired and spectra of apoptotic and control cells fitted to these spectra. Fitting coefficient shows change in contribution of each component to whole spectra of cells.  Protein decreased (from 61.0% to 52.8%) DNA decreased (from 5.1% to 1.7%) RNA increased (from 7.2% to 8.2%) Glycogen decreased (from 2.7% to 1.9%) Lipid increased (from 24.1% to 35.3%) Significant PCA differences reflect the modality and the stages of cell death (Apoptosis) Ong et al. 2012      19 Cell Type  Induction method and Spectra normalization methods Endorsement by Conventional methods Raman spectra analysis and Features Reference Human epithelial-like lung carcinoma cell line (A549) cells (ATCC)  Normalized using the standard normal variate (SNV) transformation method  Exposure to 0.5-1000 µM sulfur-mustard For 1h at 37 °C (200 and 500 µM have been used for apoptosis analysis) Exposure to 10 nM Ricin (1mg/ml in PBS solution) for 24h MTT to assess cell metabolic activity (cell viability). Western blotting analysis of apoptosis regulator protein p53 to detect DNA fragmentation Raman spectra of pure primary cellular components including: DNA (Calf thymus), RNA (Baker's yeast), Lipid (Phosphatidyl choline and Cholesterol), High α-helix content protein (Human serum albumin), High β-pleated sheet content protein (chymotrypsin), Actin, collagen and glycogen have been acquired. Spectra of apoptotic and control cells fitted to these spectra using classical least square (CLS) method. CLS fitting parameters have been used to determine the relative concentrations of the main cellular components showing following changes: DNA decreased by: 27.5%(24h) and 87.0%(48h) Significant decrease in RNA level  Significant increase of Lipid level: 43%(48h) Negligible decrease of Protein level PCA analysis successfully discriminated between Etoposide treated and untreated cells.  Exposure of the cell to higher dose of Etoposide  (250 µM) for 6h shows insignificant reductions in DNA or RNA concentration. Owen et al. 2006 Human alveolar epithelial cell line A549   Normalized using the standard normal variate (SNV) transformation method  (MTT) to assess cell viability. Fluorescence microscopy using (Annexin V, PI and Hoechst 33258 to  distinguish between apoptosis and necrosis. Raman spectra of pure primary cellular components including: DNA, RNA, Lipid (Phosphatidyl choline), Protein (Human serum albumin) have been acquired as a source for peak change evaluation. Decrease in protein peaks (1005 cm-1 and dominant peaks that can be seen in Human Serum Albumin spectra) Decrease in DNA peaks (782cm−1, 788 cm−1 , and dominant peaks that can be seen in pure DNA spectra) Increase in Lipid peaks (1301 cm−1, 1449 cm−1, 1660 cm−1)   I Notingher et al. 2004  20 Table 1-2: Literature review of necrosis study using Raman micro-spectroscopy. Cell Type  Induction method and  Spectra normalization methods Endorsement by Conventional methods Raman spectra analysis and Features Reference Saos-2 SW-1353  Incubation in 55°C-heated water bath for 90 minutes. Vector Normalization FACS (Annexin V and  PI staining)  Significant PCA differences reflect the modality and the stages of cell death. >95% Classification of single cell Raman spectra to predefined necrotic modality by Support Vector Machine (SVM) technique. Saos-2: Increase in 1375 cm-1 peak associated with release of HMGB1that alters chromatin architecture.  SW-1353: Increase in 1003, 1244, 1375 cm-1 peaks Decrease in 1437 and 1658 cm-1  peaks  Brauchle et al. 2014 MEL-28  Oxygen-glucose deprivation for necrosis induction. Normalized to 2993 cm-1 (strong peak primarily contributed from protein, less lipid and glycogen) Flow Cytometry by Propidium iodide (PI). Viability assessment by trypan-blue exclusion assay  Necrotic Induction effects:  Decrease in the relative amount of:  Lipid (717, 935, 1300, ~1315, 2854, 2896, 1448 cm-1) RNA (990, 1080 ~1100 cm-1)  Increase in relative amount of: disordered protein structure (1231 cm−1),  Protein structures (642, 827, 851, 875, 877, 1617cm-1) DNA (1421 cm-1)  Kunapareddy et al. 2008  21 Cell Type  Induction method and  Spectra normalization methods Endorsement by Conventional methods Raman spectra analysis and Features Reference Human chronic myelogenous leukemia cells (K562 cell line)   72 hours treatment with Triton X-100 (100 µM) Normalization to maximum intensity of each spectrum for pure biological components No normalization for spectra of cells. Induction of necrosis has been confirmed by trypan- blue exclusion assay and examination of the stained cell suspension on a hemocytometer slide. Decrease in membranous lipids (734 cm−1).  Significant reduction in both DNA and RNA concentration (794, 1098, 1356, 1592 cm−1 ) Notably decrease in protein peak (1011 cm−1) Drop at Protein/lipid peaks (1462, 1672 cm−1) Raman spectra of pure Actin (protein), Albumin (protein), Triolein (lipid), Phosphatidylcholine (lipid), DNA (nucleic acid), RNA (nucleic acid) and Glycogen (Polysaccharide) have been acquired and spectra of necrotic and control cells fitted to these spectra. Fitting coefficient shows change in contribution of each component to whole spectra of cells.  Protein increased (from 61.0% to 83.6%) DNA decreased (from 5.1% to 0.7%) RNA decreased (from 7.2% to 3.7%) Glycogen decreased (from 2.7% to 0.4%) Lipid decreased (from 24.1% to 11.6%) Significant PCA differences reflect the modality and the stages of cell death (necrosis) Ong et al. 2012 Human lung cells (A549 cell line)    7 hours treatment with Triton X-100 (250 µM) with sampling every 30 minutes. Normalized using the standard normal variate (SNV) transformation method Normalized using the standard normal variate (SNV) transformation method  (MTT) assay to assess cell viability. Western blotting analysis of poly (ADP ribose) polymerase (PARP) to detect DNA fragmentation. Decrease protein peaks (1322 cm−1 (40%), 1342 cm−1 (50%), 1005cm−1 (60%)) 25% decrease in the width of the Amide I peak (~1655–1680 cm−1) ~80-90% decrease in decrease the 786 cm−1 peak as a confirmation of DNA fragmentation that has been seen by western blot assay. 40% decrease and a red shift (from 1094 cm−1 initially to 1090 cm−1 after 360 min) of the PO2 peak (1095 cm−1) Slight increase in phospholipid peak (719 cm−1) Ioan Notingher et al. 2004  22 Chapetr2: Materials and methods   2.1  Cell line      A model Chinese Hamster Ovary (CHO) cell line CHO-M was used in this study that expressed a monoclonal antibody. CHO-M previously named ChK2 437.89.56. Derived from Lonza’s CHO-K1SV cell line producing human monoclonal anti-interleukin 1 β IgG1 antibody (Kennard et al. 2009).   2.2  Cell culture      We used 25, 125 and 500 mL sterile shake flasks (Corning and Nalgene) for culturing the cells. Cell cultures were performed in a 5% CO2 humidified incubator (Adolf Kuhner AG, Basel, Switzerland) at the shaking speed of 140 rpm and the temperature of 37ºC.  2.2.1    Maintenance medium and cultures      1 mL vials were used for the cell storage. The cells were stored at ~10×106 cell/mL and -150ºC. The Maintenance medium for cells consisted of CD CHO (Gibco-Invitrogen, Grand Island, NY) supplemented with 6 mM glutamine (Gibco), 4.5µg/mL bleocin (Calbiochem, La Jolla, CA) and 100 µg/mL hygromycin B (Gibco). The reason for introducing these antibiotics was maintaining selection pressure during thawing and maintenance of cells.       First, the vials of cells and maintenance medium were thawed in a 37ºC water bath. Afterwards, the 1 mL of thawed cells was transferred to a 125 mL flask containing 20 mL of fresh maintenance medium. The maintenance medium was warmed avoid thermal shock. At least 3 cell passages were performed before using newly thawed cells in experiments. The passaging was carried out when cell concentration reached ~3×106 cells/mL; the cells were then diluted with fresh maintenance medium to a density of 1×106 cells/mL.   23      The viability, cell size and cell density were measured by a Cedex automatic cell counting machine (Innovatis, Bielefeld, Germany). This machine stains cells with trypan-blue dye (Sigma) and uses an image-processing algorithm to detect stained and non-stained cells. This algorithm measures cell concentration and sizes. Furthermore, it gives information about cell aggregation formation. Samples were diluted with 0.25% trypsin-EDTA solution and incubated for 15 minutes at 37ºC in order to remove cell aggregates and to obtain a more accurate cell concentration. The growth curve of the cells was monitored to ensure that the cells were in exponential growth phase prior to starting experiments.    2.2.2    Fed-batch culture      The fed-batch protocol was established based on a former protocol developed for a CHO cell line expressing tPA (Jardon et al. 2012a) and adapted to another CHO cell lines that expressed a monoclonal antibody (Nasseri et al. 2013). Fed-batch cultures were carried out in a 125 mL conical shake flask containing 20 mL of fresh maintenance medium. Maintenance medium (See Section 2.2.1) was supplemented with 25 ng/mL insulin-like growth factor (IGF), 4 mM glutamine and 4x anti-clumping agent for fed-batch cultures. Fed-batch cultures were seeded at the cell density of 1×106 cells/mL and cell passaging performed according to Section 2.2.1.  2.2.3    Cell banking      The CHO-M cells used in this study had originally been prepared by a former PhD student (Jardon et al. 2012) and were stored in 1 mL vials at 10×106 cells/mL in the cryopreservation medium consisting of 85% CD CHO with 6 mM glutamine and 15% DMSO (Sigma). For banking, cells in exponential growth were recovered from a fed-batch culture and centrifuged at 800 rpm for 5 minutes. The supernatant was removed and the cells re-suspended in the cryopreservation medium at 10×106 cells/mL. The cells were then transferred to 1 mL labeled cryovials and placed in special freezing containers filled with 100% isopropanol at - 24 80ºC. After 12 hours, the vials were transformed to cryostorage boxes at -150 ºC under liquid nitrogen.  2.3  Cell death induction 2.3.1    Necrosis induction      Numerous studies have shown that oxygen and glucose deprivation separately can induce apoptosis and autophagy, but Hlalky (Hlatky et al. 1988), showed that combined oxygen-glucose induces necrosis. About 2x106 of cells (cell concentration: ~ 1x106 cells/mL) were pelleted by centrifugation in a 15 mL tube at the speed of 800 rpm. Supernatant was carefully removed and 10mL of PBS added gently without disturbing the pellet (Figure 2-1). The PBS solution was sufficient to deprive the cells of oxygen and glucose simultaneously. It is possible, however, that only the inner cells within the pellet were completely deprived of oxygen.   Figure 2-1: Oxygen-Glucose deprivation to induce necrosis.  25 2.3.2    Apoptosis induction Cells where incubated in 20 µM camptothecin (CPT) for 6, 12 and 24 hours in 25 mL flasks to induce apoptosis. CPT (C20H16N2O4) is a cytotoxic quinolone alkaloid, which was originally isolated from the bark and stem of the camptotheca tree that is native to China and Tibet where locals call it the happy tree, tree of life or cancer tree since its bark is used for cancer treatment. CPT inhibits the DNA enzyme topoisomerase I (topo I)(Efferth et al. 2007), which is a class of enzymes that cuts one strand of double stranded DNA, relaxes the strand and then re-anneals the strands. The topo I class of enzymes play a key role in removing DNA supercoils during the DNA replication and transcription processes, disentangling intertwined DNA during mitosis, and DNA strand breaking during recombination and chromosome condensation(Wang 2002). CPT binds with hydrogen bonds to both topo I and covalently complexes with DNA to create a stable ternary complex. This inhibits DNA re-ligation and, therefore, induces apoptosis by damaging DNA (Adams et al. 2006).  Figure 2-2 shows molecular structure of ternary complex of DNA, topo I and CPT. The toxic effect of CPT occurs mostly during S phase of growth as a result of the collision between the replication fork and the ternary complex created by DNA, CPT and topo I (Pommier 2003).  Figure 2-2: Ternary complex of DNA, topo I and CPT (Yikrazuul, 2009).  26 2.4  Flow cytometry to detect apoptosis and necrosis       Cells (~1×105 cells) were stained using the FITC Annexin V/Dead Cell Apoptosis Kit (V13242, Invitrogen). A sample of approximately 250,000 cells (cell concentration: ~1x106 cells/mL) were recovered from the culture, centrifuged in a 1.5 mL micro-centrifuge tube at 800 rpm for 5 minutes and supernatant removed. The cells were then washed with 1mL PBS solution, centrifuged again at 1000 rpm for 5 minutes and PBS removed.  The washed cells were re-suspended in the Annexin V binding buffer and subsequently stained with Annexin V and PI dyes according to the manufacturer's instructions. Cells were incubated in an opaque 1.5 mL micro-centrifuge tube for 15 minutes at room temperature to give sufficient time for attachment of the stains. The stained cell samples were stored on ice prior to analysis by FACS. BD FACS Caliber flow cytometer (BD Biosciences, Franklin Lakes, NJ), using 488 and 633 nm lasers, were used to analyze the stained cells. The FACS data were analyzed with FlowJo (Tree Star, Ashland, OR) software.   2.5  ATP measurement      Levels of cellular ATP were measured using CellTiter-Glo Luminescent Cell Viability kit (Promega, TB288). Cell samples were removed from the culture (cell concentration: ~ 1x106 cells/mL), centrifuged in 1.5 mL micro-centrifuge tubes at 800rpm for 5 minutes and supernatant removed. Approximately 400 to 40,000 cells were re-suspended in 1mL PBS solution. Desirable numbers of cells were added to the wells of a 96-well opaque walled plate (CORFW96) and CellTiter-Glo reagent (Promega, Madison, WI) that contains 100 µL protocol provided cell lysis buffer and a recombinant luciferase enzyme that emits light (luminescence) when it reacts with ATP. The intensity of this emitted light is proportional to the concentration of ATP in the sample solution. A standard curve of luminescence versus ATP concentration was derived from assay provided ATP standard solutions at six different ATP concentrations (0.1, 0.4, 1, 2, 4, 10 M).  In addition, three wells in the plate were loaded only with CellTiter-Glo reagent  27 to act as blanks and to determine the background luminescence. All unnecessary lights in the laboratory were turned off during sample preparation to minimize the luciferase reaction with ATP. After filling the wells, the plate was covered with a thick aluminum foil and plate was gently shaken for 5 minutes on a shaker table (250 rev/min). The plate was left for 10 minutes at room temperature after shaking to stabilize luminescent signal. The luminosity activity of the samples was measured using Tecan M200 plate reader (Männedorf, Switzerland). Data were analyzed with i-control software (Männedorf, Switzerland).   2.6  Protein measurement       The intercellular protein levels were measured with PierceTM BCA Protein Assay Kit (Thermo Scientific, 23227) according to the manufacturer’s instructions. Cu+2 is reduced to Cu+ by the protein in the presence of an alkaline medium. This reaction absorbs light at 562 nm and the absorbance is proportional to the protein concentration of the sample solution. Although, intercellular proteins have different structure and may behave slightly differently in this reaction, overall this assay can give a good estimate of the total intercellular protein concentration. Cells were centrifuged at 800 rpm for 5 minutes and re-suspended in the fresh growth medium. Afterwards, they were disrupted by two freeze thaw cycles at -70ºC and 37ºC in order to lyse the cells. Desirable numbers of cells were added to the wells of a white opaque 96-well micro-plate (Corning, CLS3600). 200 µL of working reagent of PierceTM BCA Protein Assay was added to each well and shaken for 30 seconds at room temperature on a plate shaker at speed of 250 rpm for 5 minutes. A standard curve was derived from BCA assay provided standard solutions at eight different concentrations (25, 125, 250, 500, 750, 1000, 1500, 2000 g/mL). In addition, three wells in the plate were loaded only with working reagent to act as blanks and to determine the background absorbance. The micro-plate was then covered with a thick aluminum foil to avoid any reaction with environmental light and incubated for 30 minutes at 37ºC and cooled to room temperature for a further 10 minutes. The absorbance of samples was recorded  28 at 562 nm using Tecan M200 plate reader (Männedorf, Switzerland). Data were analyzed with i-control software (Männedorf, Switzerland).   2.7  DNA measurement      The intercellular DNA levels were measured with CyQUANT Cell Proliferation Assay kit (Life technologies, C35007) according to the manufacturer’s instructions. This assay is based on CyQUANT GR dye, which is a green fluorescence dye that binds to cellular nucleic acids. Since the dye cannot differentiate between RNA and DNA, RNAase enzyme is added prior to the assay to remove all RNA from the sample. The fluorescence intensity emitted by a sample is proportional to the DNA concentration in the sample solution. Approximately 105 cells were collected from a given cell culture and centrifuged for 5 minutes at 200×g. The supernatant was removed and discarded carefully without disturbing cell pellet. Cells were centrifuged at 800 rpm for 5 minutes and re-suspended in the fresh growth medium. Afterwards, cells were disrupted by two freeze thaw cycles at -70ºC and 37ºC in order to lyse the cells. The cell pellet was then re-suspended in the non-fluorescent cell lysis buffer supplemented with 180 mM NaCl, 1 mM EDTA and RNAase enzyme (final RNAase concentration: 1.35 kunitz units/mL). The RNAase treated samples were incubated for 1 hour in room temperature. Calculated volumes of the solution were added to the wells of a black 96-well micro-plate (Corning, CLS3650). 200 µL CyQUANT reagent that contains cell lysis buffer and CyQUANT GR dye was added to wells and samples were treated for 5 minutes at room temperature. A standard curve was derived from assay provided DNA standard solutions (λ DNA) at eight different concentrations (10, 50, 100, 200, 400, 600, 800, 1000 ng/mL) by diluting provided 1µg/mL λ DNA sample with CyQUANT GR/cell-lysis buffer.  In addition, three wells in the plate were loaded only with CyQUANT lysis buffer to act as blanks and to determine the background fluorescence. All the unnecessary lights in the laboratory were turned off during sample preparation to avoid fluorescence bleaching. The fluorescence of the samples was recorded using Tecan M200 (Männedorf, Switzerland)  29 plate reader with filters appropriate for ~480 nm excitation and ~520 nm emission maxima. Data were analyzed with i-control software (Männedorf, Switzerland).    2.8  RNA measurement      Intercellular RNA content was measured with Quant-iT TM RNA Assay Kit (Invitrogen, 33140) according to the manufacturer’s instructions. The assay is based on a fluorescence dye that is highly selective for single stranded RNA and doesn’t bind to double strand DNA. As a result, there is no need for pretreatment with DNAase to remove DNA from the samples. The fluorescence intensity emitted by a sample is proportional to the RNA concentration in the sample solution. Approximately 105 cells were collected from a given cell culture and centrifuged for 5 minutes at 200×g. The supernatant was removed and discarded carefully without disturbing cell pellet. Cells were centrifuged at 800 rpm for 5 minutes and re-suspended in the fresh growth medium. Afterwards, cells were disrupted by two freeze thaw cycles at -70ºC and 37ºC order to lyse the cells. The cell pellet was re-suspended again in fresh medium to obtain a homogenous mixture of cells. Calculated volumes of the sample (for desirable numbers of cells) were added to the wells of a 96-well black micro-plate (Corning, CLS3650). 200 µL Quant-iT working solution that contained the fluorescent dye was added to the wells. A standard curve was derived from assay provided RNA standard solutions (rRNA derived from E.Coli bacteria) at eight different concentrations (2, 5, 7, 10, 15, 20, 25, 50 ng/mL). In addition, three wells in the plate were loaded only with Quant-iT working solution to act as blanks and to determine the background fluorescence. All the unnecessary lights in the laboratory were turned off during sample preparation to avoid fluorescence interaction with the samples. The micro-plate was covered with a thick aluminum foil to protect it from environmental lights. The fluorescence of the samples were recorded using Tecan M200 (Männedorf, Switzerland) plate reader with filters appropriate for ~644 nm excitation and ~673 nm emission maxima. Data were analyzed with i-control software (Männedorf, Switzerland).  30  2.9  Raman spectroscopy  2.9.1    Sample preparation      Approximately 2×105 cells were collected from a given culture. The cells were centrifuged for 5 minutes at 800 rpm and the supernatant was carefully removed and discarded. The cell pellet was re-suspended in 5 mL of PBS solution and shaken gently either with a vortexer or by hand (the shaking should gentle to avoid any cell death by mechanical shear). The cells were re-centrifuged for 5 minutes at 1000 rpm and the PBS removed carefully without disturbing cell pellet (it is important to remove all the PBS to minimize interference with the Raman spectra). The cell pellet was re-suspended in 5ml cold methanol (100%) at -20ºC. The suspension was gently shaken and left at -20ºC for 30 minutes. The solution was then centrifuged for 5 minutes at 800 rpm and the methanol removed and discarded. Approximately 100 L of wet cells were collected with pipette and placed on a Round Protected Gold mirror (Thor labs, Newton, NJ) and left until the methanol evaporated from mirror surface. The cell samples on the gold carrier mirrors were the covered and stored at 4 ºC until analysed.  2.9.2    Cell selection method      As Figure 2-3 shows, microscopic analysis of the methanol fixed cells on the gold carrier mirrors revealed there were two groups cells, bulk cells and single cells.   31  Figure 2-3: Methanol fixed cells’ configuration on the gold mirror.       The region of 'Bulk cells' was used to acquire data on apoptotic and necrotic cells. Figure 2-4 shows an example of a region of Bulk Cells that was completely covered with cells with no spaces on it chosen for Raman analysis where one spectrum was taken from each pixel. Using this method, a large number of replicates can be analyzed from the same sample setting. Furthermore, this method may simulate bioreactor conditions, where a Raman probe targets millions of cells in one single spectrum.  32  Figure 2-4: Pixel configuration of cells’ bulk under Raman microscope.       For samples that were sorted by the FACS machine, there were fewer numbers of cells on the gold carrier mirror with almost no visible Bulk Cells. As a result, single cells were used to take Raman spectra. Furthermore, high resolution Raman microscopy was used to study intercellular chemical maps of single cells. To avoid biased cell selection by, for example, size or shape, the following algorithm was applied to cell selection: 1- Move the cursor of microscope randomly over an area where there is a chance for finding single cells (In this stage, 5X optical zoom was applied). 2- Change optical zoom to 20X to find a candidate area and tune the microscope cursor to the vicinity of that area. 3- Change optical zoom 50X and find the closest single cell regardless its morphological shape or size and acquire spectrum. 4- Change the optical zoom to 5X and return to step one.   33 2.9.3    Single cell analysis      FACS sorted cells were selected for analysis according to the single cell selection algorithm (see Section 2.9.2). As Figure 2-5 shows, a single target cell was divided into approximately 6 regions with virtual pixels (~3µm×15µm). Raman spectra were acquired from each region with the final cell spectrum averaged from these 6 regions. The spectra were averaged since spectra across the cell can vary due to the presence of the nucleolus. In addition, using this averaging method, the variation due to the presence of the cell organelles could also be cancelled out, resulting in a spectrum that would be more representative of the chemical components of cell.   Figure 2-5: Single cell spectrum acquisition setup.  2.9.4    Raman microscopy equipment         A custom build Raman micro-spectroscopy system (InVia, Renishaw, Gloucestershire, UK) was used for acquisition of all Raman measurements. Spectra acquisition was performed at room temperature and samples were kept in 4 ºC while they were not being analyzed under microscope. A 50X objective lens (Olympus Life Science, Hamburg, Germany) was used for spectra gathering. Raman scattering was generated by a laser with 785 nm (visible red light) wavelength and 300 mW power. Radiation time (exposure time) was different for apoptotic and necrotic cell samples. The exposure time was set to give a maximum signal/noise ratio without  34 saturation of CCD camera. For FACS sorted cells, exposure times were set to 15 seconds. The spectral collection range was set from 300 cm-1 to 1800 cm-1 in order to optimize between spectra acquisition time and deriving the maximum possible biochemical information from a specimen. The cell samples were all fixed on gold-coated glass mirrors, since the mirror surface increases Raman signal 4x compared to a glass surface such as a common Petri dish. Because, in the presence of mirror, both forward and backward scattered photons could be collected from a two-pass beam path (Schulze et al. 2013). As Figure 2-6 shows, the calibration of the Raman microscope prior to the start of data collection with standard silicon based wafers designed for calibration purposes. Scattered Raman signal from silicon is centered at ~520 cm-1 Raman shift. All the tuning knobs were set to get maximum laser intensity in every experiment.    Figure 2-6: Calibration spectrum of Raman microscope using silicon wafer.       The laser that was used for generating Raman scattering needed a given time to warm up and stabilize and as can be seen in Figure 2-7 the laser intensity takes approximately 90 minutes  35 to stabilize. Based on this, all the spectra acquisition began about 90 minutes after turning on the laser to ensure stable laser intensity during the experiment.  Figure 2-7: Laser intensity vs. time after turning on the laser.        Furthermore, the laser has a natural fluctuation of central wavelength. Figure 2-8 shows the result of several silicon tests that have been run on an identical spot on the silicon wafer under an identical optical setup. As shown, the central wavelength of laser can have a variance of up to one to two wave-numbers in each test. This is important in the interpretation of the spectra peak shifts, where one or two wave-numbers peak shifts may be due to laser fluctuations rather than biological differences between specimens. To relate a peak shift to the presence of a biological entity, the peak shift must be observed over the analysis of several replicates.    36  Figure 2-8: Laser wavelength fluctuation vs. time measured by silicon wafer.  2.10  Data analysis of Raman spectra      Matlab 12.0 (The MathWorks, Natick, MA) was used for spectra processing and data analysis. Background spectra were collected from an empty site on mirror close to the data collection region and subtracted from original spectra. Afterwards, all spectra was processed by automated cosmic spike removal, baseline flattening using a moving average, peak stripping and automated smoothing steps. Principal component analysis (PCA) was performed to monitor major differences among different samples and creating PCA images. High-resolution chemical maps for specific peaks were generated using custom written software (See Appendix A). No normalization has been performed in the single cell data collection mode. However, for spectra that were collected in the bulk cell mode, various normalization methods including normalization to phenylalanine peak (1003 cm-1), nucleic acid peak (780 cm-1) and vector normalization were applied.   37 Chapter3: Results and discussion  3.1  Apoptosis induced cells       Over a course of 24 hours, highly viable CHO cells (96.6% viable based on trypan-blue assay) were collected from an exponential culture that had been treated with 20 µM CPT. Initially the vast majority of cells did not stain positive for either Annexin V or PI (Annexin V-, PI-). However, after 6 hours of induction, a group of cells tested positive for Annexin V, although not for PI (Annexin V+, PI-), These cells represent early apoptotic features. Longer induction results in cells taking up PI dye as they enter into a 'late apoptotic phase' (McIntyre et al, 2014). Finally, after 24 hours of treatment most cells have become highly stained with PI (Figure 3-1).    Figure  3-1: Flow cytometric analysis of Annexin V/PI stained cells of viable and 6,12 and 24h Camptothecin treated CHO cells.   38      Cells were fixed with methanol and put on the gold mirrors. A region with a bulk population of cells was selected on the mirror and 100 spectra were collected (exposure time of the laser was set at about 30 sec per spectrum). Because of heterogeneity of this region, the intensity of collected spectra had a considerable inter-population variance (Figure 3-2).  Figure 3-2: 100 raw spectra of CHO cells treated with Camptothecin for 24 hours.       The inter-population variance was first reduced by normalization to the phenylalanine peak and averaging. Then using a custom built software (See Appendix A), the background was subtracted and the baselines removed. In addition, cosmic spikes were removed and several smoothing filters were applied to the spectra. These processing steps considerably reduced the inter-population variability among spectra. Usually, peaks with Raman shift bigger than  1000 cm-1 have more standard deviation and are broader. Furthermore, in some spectra there were peaks that may come from dust or crystals on the mirror. These peaks were effectively removed by averaging over a large number of spectra (Figure 3-3).  39  Figure 3-3: 100 processed (background subtracted and base line corrected) spectra of CHO cells treated with Camptothecin for 24 hours.        Table 3 depicts most common and informative peaks that can be seen in a cellular spectrum. In particular, these peaks are highly important in analyzing cell death process and have been chosen from a review of numerous studies on cell death using Raman spectroscopy. Because of difference in lasers and also due to the natural fluctuation in laser wavelength, Raman peaks can shift by up to two or three wave-numbers. This should be considered in data analysis.           40 Table 3-1: Raman peak assignments of interest for this project (Movasaghi et al. 2007). Raman peak Assignment 596 cm-1 Phosphatidylinositol 666 cm-1        G, T (ring breathing modes in the DNA bases)-tyrosine-G backbone in RNA 718 cm-1 Lipids (phospholipid C-N stretch) 724 cm-1 DNA/RNA (adenine ring breathing) 757 cm-1 Proteins (tryptophan symmetric ring breathing) 784 cm-1 782 cm-1: DNA/RNA (pyrimidine ring breathing); 788 cm-1: DNA (backbone O-P-O stretching) 809 cm-1 RNA (backbone O-P-O stretching); proteins (C-C stretching of proline and hydroxyproline, e.g. in collagen) 827 cm-1 Proteins (proline, hydroxyproline, out-of-plane ring breathing in tyrosine); DNA/RNA (asymmetric O-P-O stretching) 853 cm-1 Proteins (proline C-C stretch in collagen; also tyrosine ring breathing in other proteins); carbohydrates (glycogen, polysaccharides C-O-C stretching) 877 cm-1 C-C-N+symmetric stretching (lipids), C-O-C ring (carbohydrate) 937 cm-1 Proteins (collagen type I C-C stretching, -helix C-C stretching); carbohydrates (glycogen), Proline (collagen type I), Amino acid side chain vibrations of proline and hydroxyproline, as well as a (C-C) vibration of the collagen backbone
C-C backbone (collagen assignment) 959 cm-1 Hydroxyapatite 1003 cm-1 Proteins (symmetric ring breathing in phenylalanine) 1031 cm-1  Proteins (collagen, keratin, C-N stretching in proteins, C-H in-plane bending of phenylalanine); lipids (phospholipids); carbohydrates (polysaccharides) 1066 cm-1 Proline (collagen assignment) 1078 cm-1 C-C or C-O stretching mode of phospholipids 1100 cm-1 Several bands of moderate intensity, belonging to amide III and other groups (proteins) 1124 cm-1        ν(C-C) skeletal of acyl backbone in lipid (trans-conformation) 1235 cm-1       Amide III 1485 cm-1        G, A (ring breathing modes in the DNA bases) Nucleotide acid purine bases (guanine and adenine) 1573 cm-1 Guanine, adenine, TRP (protein) 1669 cm-1     Amide I, Carbonyl stretch (C=O), Cholesterol ester       Processed spectra on viable, 6h, 12h and 24h CPT treated cells were normalized to the tallest peak, phenylalanine. Single cell spectra and biochemical assays have shown that protein (peaks) are most stable, with phenylalanine being the most stable peak (1003 cm-1). Normalization was performed to cancel out the effect of laser fluctuation and heterogeneity of sample configuration on the mirror. Due to the stability of this peak, normalization of spectra to the phenylalanine peak is common in Raman spectroscopy of apoptotic cells (Laporte et al. 1997; Rodríguez- 41 Casado et al. 2001; Zoladek et al. 2011; Krafft et al. 2006).  Figure 3-4 shows mean normalized spectra of viable, 6,12 and 24 h cells treated with CPT.  Figure 3-4: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1).        To visualize the differences between these four populations, the spectra were divided to three regions, 600 cm-1 to 980 cm-1, 1020 cm-1 to 1400 cm-1, 1400 cm-1 to 1800 cm-1 (Figures 3-5, 3-6, 3-7). Since the phenylalanine 1003 cm-1 peak was used for normalization and doesn’t carry any information, it was removed to allow smaller peaks to be visualized clearer.   42  Figure 3-5: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 600 cm-1 to 980 cm-1.   Figure 3-6: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1020 cm-1 to 1400 cm-1.   43  Figure 3-7: Mean spectra of viable, 6h, 12h and 24h Camptothecin treated CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1400 cm-1 to 1800 cm-1.        Loading spectra of first PCA component, PC1, illustrates main differences between two populations (Figures 3-8 and 3-9). In these figures, the increase of a particular peak is shown by a new positive peak and the decrease of a particular peak is shown by a new negative peak. Analysis of these changes show the main peaks that should be investigated in the induction of apoptosis. Figure 3-8: Loading of PC 1 depicts specific peak changes between viable/6h, viable/12h,  viable /24h populations.   44  Figure 3-9: Loading of PC 1 depicts specific peak changes between 6h/12h, 6h/24h,  12h /24h population.        The RNA peak at 809 cm-1 has special importance since it is the only pure RNA peak in the spectrum range without interference from DNA backbone structures. Figure 3-10 shows changes in this peak. Amounts are normalized to control sample (0 hours CPT treated) that is 96.6% viable based on trypan-blue exclusion of live cells using a hemocytometer. Changes in this peak indicate a significant drop in the level of RNA over the first 12 hours after CPT treatment. Intercellular RNA measurement by the Quant-iT TM RNA Assay also shows the same trend of decreasing RNA in the first 12 hours after CPT treatment. Degradation of mRNA has been observed in early apoptosis (Del Prete et al. 2002) and FACS analysis showed that the cells are mostly in early apoptotic phase (Figure 3-1) in these first 12 hours. As a result, during this period there is a sharp decrease in both mRNA and rRNA levels and subsequently a sharp reduction in the Raman spectra RNA peak. The decrease in RNA levels could be due to the inhibition of rRNA synthesis and defective pre-mRNA maturation during the process of apoptosis (Fraschini et al. 2005) or it could be a result of active RNA cleavage by the enzyme RNase L. (Díaz-Guerra et al. 1997).    45  Figure 3-10: RNA peak (809 cm-1) change in apoptosis induced culture.       There are also several other nucleic acids peaks in the spectrum. The majority of these peaks have contributions from both DNA and RNA and some of them such as 827 cm-1 and 1573 cm-1 also have contributions from protein structure. The information from these peaks can be evaluated together with other peaks to create a clear image about actual intercellular changes. Figure 3-11 shows 6 major nucleic acids peaks in the spectrum. All the peaks were normalized to the control (0h CPT treated cells) population for a clear comparison. As can be seen there is a common trend of nucleic acid degradation in all peaks, with the decrease in the 827 cm-1 peak being negligible compared to the decline in the other peaks. This comes from the fact that part of this peak comes from stable protein structures. Decline in the  1485 cm -1 peak is another indicator of the degradation of RNA, where the 1485 cm-1 peak corresponds to the nucleotide acid purine bases (guanine and adenine). Thus, a drop in RNA levels will increase the decline of this 1485 cm-1 peak compared to other DNA based peaks. 00.20.40.60.810 6h 12h 24hRelative Intensity Hours  46  Figure 3-11: Nucleic Acid peaks change in apoptosis induced culture.       Caspase-activated DNase (CAD) is the enzyme responsible for DNA cleavage during apoptosis and activation of this enzyme is one of the hallmarks of apoptosis (Enari et al. 1998). Cleaved DNA fragments are roughly about 180 base pairs (Wyllie 1980) and can leak out of the cell into the cell culture during apoptosis induction as well as during sample preparation. The role of lipids in apoptosis is very important and complicated. Phosphatidylserine (PS), a phospholipid found in the inner leaflet of eukaryotic cellular membranes, translocates to outer leaflet and acts as a “eat me” signal for recognition and clearance of apoptotic cells by phagocytes. In addition, there is a huge change in mitochondrial phospholipids during apoptosis. The displacement of the phospholipid cardiolipin from the inner mitochondrial membrane is another marker of early stage apoptosis (Chaurio et al. 2009). In a review of 32 publications on mitochondrial lipid changes in eukaryotic cells during apoptosis, it was noted that 11 types of lipids decreased and 21 types of lipid increased (Crimi and Esposti 2011). Notably, the level of cholesterol-rich lipid rafts were significantly increased in apoptotic cells compared to viable cells (Li et al. 2006). Proton NMR spectroscopy has shown lipid accumulation along with degradation of proteins and DNA in apoptotic muscle cells (Astrakas et al. 2005). Most Raman studies of 00.20.40.60.811.20 6h 12h 24hRelative Intensity Time 784 cm-1 666 cm-1 724cm-1 827 cm-1 1485cm-1 1573 cm-1  47 apoptotic cells have reported increases in lipid peaks (Ong et al. 2012; Owen et al. 2006; Notingher et al. 2004; Zoladek et al. 2011) whereas, in a few cases, decreases in lipid peaks have been noted as resulting from the loss of lipid bodies inside cytoplasm (Krafft et al. 2006). In the present study (Figure 3-12), it can be seen that there is a slight decrease in the lipid/protein peaks (1124 cm-1, 1031 cm-1, 596 cm-1) in late apoptotic cells (12h, 24h). Unfortunately, the pure lipid peak ( 718cm-1) is very small in CHO cells spectra and is on the shoulder of a large nucleic acid peak (724 cm-1). As a result, it was not possible to verify lipid changes confidently from this peak. The peaks shown in Figure 3-12 also have a contribution from protein structures and closer analysis of these peaks show that their reduction is almost insignificant and mainly come from protein structures. It can, therefore, be concluded that the level of lipids are either constant or slightly increasing in CPT treated cells. In addition, non-normalized spectra from early and late apoptotic FACS sorted cells show a clear increase in the lipid peaks (1076 cm-1 and 596cm-1) compared to viable cells, which may support the conclusion that lipid levels are increasing during apoptosis.   Figure 3-12: Lipid peaks change in apoptosis induced culture.  00.20.40.60.811.2 0 6h 12h 24h1124 cm-1 1031 cm-1 596 cm-1 48    The Raman spectra were normalized to phenylalanine peak (1003 cm-1), which is a α-amino acid and changes in this peak are strongly correlated with changes in other protein peaks. Although, there is a considerable amount of crucial information embedded in these protein peaks (especially when these peaks are associated with other cellular biochemical components), phenylalanine normalized Raman spectra, unfortunately, cannot be used to study intercellular protein changes. Figure 3-13 shows changes in protein-associated peaks for cells undergoing apoptosis. Most of these peaks change little during apoptosis, although the BCA protein assay shows an overall drop of about 30% in the intercellular protein level 24 hours after CPT treatment (Figure 3-14).  This difference between Raman and BCA assay is directly related to normalization of the peaks to phenylalanine. The peak at 827 cm-1 has contribution from both proteins and nucleic acids and shows more of a drop compared to protein dominant peaks such as 1031 cm-1 phenylalanine peak. The peak at 1235 cm-1 corresponds to Amide III, which belong to amide group that have weaker bonds compared to amines (such as amino acids)(Kemnitz and Loewen 2007). Amides can be rapidly hydrolyzed during apoptosis (Maccarrone and Finazzi-Agró 2003) and as a result, the amide peak would be expected to fall more than the amino acid peaks (phenylalanine in this case).    49   Figure 3-13: Protein peaks change in apoptosis induced culture.       The 1669 cm-1 peak, which corresponds to Amide I and cholesterol ester, shows a smaller drop compared to the 1235 cm-1 Amide III peak, which is expected due to lipid accumulation and the stability of the ester bonds during apoptosis (Maccarrone and Finazzi-Agró 2003). The peak at 1031 cm-1 corresponds mainly to phenylalanine, but also contains some phospholipid component and shows a slight increase during apoptosis. This slight increase can only come from the lipid component where the spectra are normalized to phenylalanine. 00.20.40.60.811.2 0 6h 12h 24h853 cm-1  827 cm-1  937 cm-1  1031 cm-1  1235 cm-1 50  Figure 3-14: Chemical components changes in apoptosis induced culture measured by biochemical assays.         Intercellular ATP is required for apoptotic cells and the apoptosis machinery will shut down without sufficient energy derived from ATP. Dying cells, therefore, follow different pathways depending on level of intercellular ATP. In other words, presence of intercellular ATP can cause the cell undergo apoptosis under certain stresses and the removal of ATP will cause necrotic induction under same stress factors (Eguchi et al. 1997). In fact, this intercellular ATP level is thought to control the apoptosis process (Richter et al. 1996).  As Figure 3-14 shows, ATP levels increased 6 hours after CPT treatment. This increase in ATP is well known among eukaryotic cells in the early stages of apoptosis and before activation of caspase-3 and internucleosomal DNA fragmentation in late apoptosis (Zamaraeva et al. 2005). The maintenance of ATP levels during apoptosis in CHO cells has also been frequently reported for other mammalian cells (Garland and Halestrap 1997).  51 3.1.1    PCA classification of apoptotic cells       One of the main advantages of Raman spectroscopy is its non-invasive ability to study and characterize apoptotic cells and cultures. This is very important in the case of pharmaceutical bioreactors where the expensive content of these bioreactors may be lost if some factor induces cell death in the culture. An early detection monitoring system such as Raman microscopy could prevent huge economic losses. Furthermore, distinguishing between viable, necrotic and apoptotic cells in their different stages of apoptosis could also play a key role in tumor diagnosis and cancer therapy. Score plots of the principal components of Raman spectra can successfully classify mammalian cell populations. Plotting lowest principal components, PC2 versus PC1, (Figure 3-15) can easily distinguish between three populations: i) viable; ii) early apoptotic (6h CPT treated) and mixture of late apoptotic (12h CPT treated); and iii) secondary necrosis (24h CPT treated).   Figure 3-15: Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable, 6h, 12h, 24 h CPT treated populations.   52      There is an obvious difference between the second principal component (PC2) of a mixture of viable and early apoptotic cells and a mixture of late apoptotic cells and secondary necrotic cells. The similarity of viable cells and early apoptotic cells and the similarity of secondary necrotic cells and late apoptotic cells are obvious by studying the mean Raman spectra (Figures 3-4, 3-5, 3-6, and 3-7). PC2 compares changes in absolute values of Raman peaks while PC1 is correlated with the spectra. However, a plot of PC2 versus PC1 is able to distinguish between viable and early apoptotic cells. Figure 3-16 and Figure 3-1 show that there are some early apoptotic cells in 0h CPT populations and there are some viable cells amongst the 6h CPT cells. These cells cross the boundary line between two clusters in Figure 3-16.  Figure 3-16:  Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable and 6h CPT treated populations.    53      For classification of 12h CPT and 24h CPT, higher principle components are required. The reason for this is due to inter-population similarities between these two populations. This similarity can be easily seen in FACS gating (Figure 3-1) and Raman Shifts (Figures 3-4, 3-5, 3-6, and 3-7). However, plotting PC4 versus PC1 can distinguish between 12h and 24 CPT treated CHO cells (Figure 3-17). Combining the two score plots (Figure 3-16 and 3-17) it is possible to clearly classify the apoptotic cell populations based on Raman spectra. Potentially, this technology could be used monitor cell culture conditions in a bioreactor and providing a warning if the culture inside bioreactor undergoes a possible apoptotic inducer stress. This technology may also provide a non-invasive, fast and easy handling alternative to FACS analysis that could open new doors to further cell death related studies.  Figure 3-17: Score plots of Principle Components Analysis (PCA), PC 4 vs. PC 1, comparing 12h and 24h CPT treated populations.   54 3.2  Necrosis induced cells       Figure 3-18 shows FACS analysis of viable and cells that have been deprived of oxygen and glucose for 24 hours. It has been shown that deprivation from either oxygen or glucose can induce apoptosis. However, deprivation from both oxygen and glucose can inhibit the metabolic activity required for apoptosis and induce necrosis (Hlatky et al. 1988). Viable cells are mostly (PI-, AnnexinV-) with a small percentage of apoptotic cells that can be found in any cell culture due to limitation of medium components or mechanical sheer generated within a shake flask culture. In Figure 3-18 it can be seen that nearly all the 24h Oxygen-Glucose deprived cells take up PI whereas about half uptake AnnexinV.  (PI+, AnnexinV-) are termed primary necrotic cells while (PI+, AnnexinV+) are termed secondary necrotic cells. Primary necrosis is an immediate cell death that appears under severe stress or death stimuli, while secondary necrosis is follows an apoptotic process (Janko et al. 2011).    Figure 3-18:  Flow cytometric analysis of Annexin V/PI stained cells of viable and 24h oxygen-glucose deprived CHO cells.        PI+, Annexin- cells show a ruptured plasma membrane that is an indicator of primary necrotic cell death. PI+, Annexin+ cells show phospholipid phosphatidylserine (PS) exposure and membrane permeability that can only happen in late apoptotic and secondary necrosis cells  55 (Carmona-Gutierrez et al. 2010). The Oxygen-Glucose deprivation method induces necrosis extremely rapidly and effectively. Trypan-blue data shows that after 6h of Oxygen-   Glucose deprivation more than 80% of cells were dead. As a result, the majority of cells did not have enough time and ATP to undergo apoptotis and became primary necrotic.  Figure 3-19 shows 100 spectra from a bulk of oxygen-glucose deprived cells fixed by methanol on a gold-coated mirror. The heterogeneity in thickness and spatial configuration under microscope results in a different background for each spectrum (Figure 3-19). The spectrum acquisition was carried out on a 'bulk' sample as discussed in Chapter 2 Section 2.9.2 using an exposure time of 40 seconds for each spectrum.  Figure 3-19: 100 raw spectra of 24 hours oxygen-glucose deprived CHO cells.       Subtracting the background spectrum and applying a smoothing function on the spectra eliminated a large part of the inter-population variance originating from the heterogeneity in thickness and spatial configuration of the cells on the mirror (Figure 3-20). Remaining variance may be due to different exposure times and/or biological variance among cells.  56  Figure 3-20: 100 processed (background subtracted and base line corrected) spectra of 24 hours oxygen-glucose deprived CHO cells        To eliminate the effect of different exposure time and to achieve a reasonable comparison with control cells (viable) that have different bulk cell configurations and thickness on the mirrors, the spectrum needs to be normalized to an internal standard. Biochemical assays show that the intercellular levels of proteins remain fairly stable. In addition, single cell spectral analysis has shown an almost stable peak of phenylalanine at 1003 cm-1 among viable, early apoptotic, late apoptotic and necrotic cells (Chapter 3, Section 3.2). As a result, all spectral peaks were normalized to this phenylalanine peak. To aid in distinguishing peaks the spectra were divided to three regions, 600 cm-1 to 980 cm-1, 1020 cm-1 to 1400 cm-1, 1400 cm-1 to  1800 cm-1. Figures 3-21, 3-22 and 3-23 depict mean spectra of viable and 24h oxygen-Glucose deprived populations.  57  Figure 3-21: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 600 cm-1 to 980 cm-1   Figure 3-22: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1020 cm-1 to 1400 cm-1.   58  Figure 3-23: Mean spectra of viable and 24 hours oxygen-glucose deprived CHO cells normalized to Phenylalanine peak (1003 cm-1)- between 1400 cm-1 to 1800 cm-1.     Loading principle component analysis (PC 1) of viable and 24h oxygen-glucose deprived cells illustrates the main differences in the Raman spectra between these two populations (Figure 3-24). In this figure, the increase of a particular peak is shown by a new positive peak and the decrease of a particular peak is shown by a new negative peak. Analysis of these changes shows the main peaks that should be investigated in the induction of necrosis.    59  Figure 3-24: Loading of PC 1 depicts specific peak changes between viable and 24 hours oxygen-glucose deprived populations.        The peak at 809 cm-1, which is corresponds to RNA, shows a significant drop (70%) in necrotic cells (24h oxygen-glucose deprived cells) compared to viable cells (control) (Figure 3-25). RNA measurement by Quant-iT RNA kit also shows about 60% percent drop in intercellular RNA level (Figure 3-29). Significant RNA level reduction has been reported in previous Raman studies of necrotic cells (Kunapareddy et al. 2008; Ong et al. 2012). This reduction may be due to ruptured cell membranes that allows RNA molecules to exit the cell or from active RNA degradation during early stage of necrosis induction when the cell still has sufficient ATP to run the apoptotic pathway.    60  Figure 3-25: RNA peak (809 cm-1) changes between viable and 24 hours oxygen-glucose deprived populations.       The nucleic acid peaks at 666, 724, 1485 and 1573 cm-1 decreased significantly in necrotic cells compared to viable cells. The peak at 827 cm-1 differs from the other peaks since it is also derived from proline, hydroxyproline, tyrosine which are amino acids and represent the protein levels. Decrease in nucleic acid peaks have also been reported by a majority of Raman studies on necrotic cell death (Ong et al. 2012; Notingher et al. 2004). DNA measurement by Cyquant DNA assay confirms that intercellular DNA falls in necrotic cells (Figure 3-29). The peak at 1485 cm-1 corresponds to the purine bases (A and G) and shows a sharp drop as a result of the decrease in both RNA and DNA levels. The decrease in nucleic acid peaks can be correlated with chromatin condensation and the changing architecture of chromatin caused by release of HGMB1 in primary necrotic cells (Bianchi and Manfredi 2004) and internucleosomal DNA fragmentation in secondary necrotic cells (Mizuta et al. 2013). Chromatin condensation can effectively decrease the nucleus size and the resulting Raman scattering.   00.20.40.60.811.2Viable NecroticRelative Intensity 809 cm-1 61  Figure 3-26: Nucleic Acid peaks change in necrosis induced culture.       Peaks at 827cm-1,853 cm-1 and 937 cm-1, which  correspond to proline, all show significant increases. In addition, in necrotic cells a new peak at 1066 cm-1 has appeared that was not detected in viable cells. This peak corresponds to both proline and hydroxyproline that can be found in collagen as in the 853 cm-1 and 937 cm-1 peaks. Proline and hydroxyproline are the main components of the secondary protein structure of collagen (Szpak 2011). Increasing peaks related to secondary protein structure is a sign of protein denaturation that has been described as a feature of necrotic cell death (OPIE 1962). 00.20.40.60.811.21.4Viable NecroticRelative Intensity 784 cm-1666 cm-1724 cm-1827 cm-11485 cm-11573 cm-1 62  Figure 3-27: Protein peaks change in necrosis induced culture.       Peaks at 1235 cm-1 and 1669 cm-1 correspond to Amide I and Amide III respectively and indicate an increase in necrotic cells compared to viable cells. The shift in the Amide I 1235 cm-1 peak has been noted in other studies of the necrotic cell death as a hallmark of the necrotic cell death. The change is due to the deformation of protein structures during necrosis (Kunapareddy et al. 2008; Brauchle et al. 2014; PUPPELS 1991).   Figure 3-28: Lipid peaks change in necrosis induced culture.  00.20.40.60.811.21.4Viable NecroticRelative Intensity 827 cm-1853 cm-1937 cm-11031 cm-11669 cm-11235 cm-100.20.40.60.811.21 21124 cm-11031 cm-1596 cm-1 63      The peak at 1031 cm-1 is mainly corresponds to phenylalanine, but it also contains some phospholipid components and shows a slight decrease during necrosis. This slight decrease can only come from the lipid component since the spectra are normalized to phenylalanine. Peaks at 596 cm-1 and 1124 cm-1 that correspond to phospholipids show about 60% percent decline in necrotic cells (Figure 3-28). Phospholipid degradation in a necrotic cell may result from rupture and degradation of phospholipid plasma membrane during necrosis. Decrease in lipid metabolism and components in necrotic cells, measured by biochemical precise assays, has also been reported by others (Roszczenko et al. 2013).      As previously mentioned in Chapter 3, Section 3.1, depletion of intercellular ATP is a determining factor that forces cells to undergo necrosis (Tsujimoto 1997). Figure 3-29 shows that the intercellular ATP level had almost been reduced to zero in 24h oxygen-glucose deprived CHO cells, further confirming that both oxygen and glucose deprivation both induce necrosis.   Figure 3-29: Chemical components change in necrosis induced culture measured by biochemical assays.   64 3.2.1    PCA classification of necrotic cells       As previously mentioned in the PCA classification of apoptotic cells (Section 3.1.1.), it would be very useful in industrial, clinical and research settings to be able to non-invasively distinguish between viable, early apoptotic, late apoptotic and necrotic cells. This is even more important in the case of CHO cells that are one of the main host cell lines used in the industrial manufacture of many biopharmaceuticals. Any stress or unwanted cell death induction in an industrial bioreactor could result in a huge economic loses. Many factors such as risks of contamination and long sampling times make it difficult to monitor bioreactors for cell death using conventional biochemical methods or FACS analysis. Thus, the development of a non-invasive and real time probe based on Raman spectroscopy may potentially have a wide application in the pharmaceutical industry. Many of the drugs used in cancer therapy induce apoptosis or necrosis in tumor cells and distinguishing cell death type is crucial to determining the effectiveness of these therapies and drugs. This Raman spectroscopic technique combined with a proper confocal Raman probe may eliminate the need for current harmful, painful and time consuming sampling methods such as biopsy. The Score plot (Figure 3-30) of first principle component (PC1) versus second principle component (PC2) clearly distinguishes between the 24h oxygen-glucose deprived and viable cell populations. This was expected based on the clear differences between the mean spectra of viable and necrotic cells (Figures 3-21, 3-22 and 3-23).  65  Figure 3-30: Score plots of Principle Components Analysis (PCA), PC 2 vs. PC 1, comparing viable and 24h oxygen-glucose populations.    66 Chapter 4: Conclusions and future prospects   Chinese Hamster Ovary (CHO) cells are among the main host cell lines that are being used for the production of biopharmaceuticals. At any one time, throughout the world, there are a large number of fermenters that contain million dollar investments of CHO cells. Perturbations in the environmental conditions of these fermenters may induce cell death and may result in huge economical losses. Furthermore, science behind cell death is extremely important for the study of mechanisms of cancer therapy methods and aging processes. In this study, two major classes of cell death, apoptosis and necrosis, were investigated using Raman micro-spectroscopy, a laser based non-invasive tool that can be used to monitor changes in intercellular chemical composition.  Apoptosis was induced in CHO cells by a 24 hour treatment with 20 µM Camptothecin (CPT) that inhibits the DNA enzyme topoisomerase I resulting in apoptosis by preventing DNA synthesis. Apoptosis induction was verified by FACS analysis (Annexin V and PI staining). ATP measurement and trypan-blue haemocytometry were used as negative controls for necrosis induction. Raman spectroscopy of apoptotic cells showed a significant decrease in the RNA associated peaks in the first 12 hours after treatment. Spectral data also revealed a considerable drop in the DNA-associated peaks. This confirmed reduced DNA synthesis inhibited by CPT and the classical “DNA cleavage” feature of the apoptotic cell death. These results were consistent with the reduction in total intercellular RNA and DNA as measured by fluorescence based biochemical RNA and DNA assays. It was also noted that the peak corresponding to Amide III decreased compared to other cellular proteins in cells undergoing apoptosis. Although, direct measurement of total intercellular protein level by BCA protein assay revealed 30% percent drop. Analysis of peaks corresponding to lipid-protein indicated a phospholipid accumulation in the early stages of apoptosis.   67 To induce necrosis, CHO cells were deprived of oxygen and glucose for 24 hours. Necrotic cell death was verified by trypan-blue hemocytometry, measurement of intercellular ATP level and FACS analysis (Annexin V and PI staining). Raman spectroscopy of necrotic cells showed a significant decrease in the peaks corresponding to RNA and DNA levels. Although, the drop in RNA was more significant. These drops were verified by conventional biochemical RNA and DNA measurement assays. The levels of Amide I and Amide III increased significantly compared to phenylalanine. This shift is the sign of protein degradation during the necrotic cell death. However, direct measurement of the total intercellular protein level by BCA protein assay revealed 20% percent drop. Spectral data revealed a significant drop in the peaks corresponding to levels of phospholipid, which is a clear sign of plasma membrane rupture and degradation during necrosis.  Principle component analysis (PCA) was used to identify the peaks that vary between apoptotic, necrotic and viable cells. Using PCA score plots showed that viable, early apoptotic, late apoptotic, primary and secondary necrotic cells could be clearly distinguished. This analytical technique based on Raman spectroscopy could potentially be used for early and non-invasive detection of cell death induction in bioreactors. In addition, it could be developed into a highly precise alternative method to studying apoptosis and necrosis compared to conventional biochemical methods that are usually time consuming and lab intensive.  The ultimate objective of this study will be to develop an in-vitro, non-invasive probe to study bioreactor cell cultures in both industrial and research settings that will eliminate the need for sampling and fixing of cells to monitor the health of cell cultures. Further, the idea of cooling apoptotic or necrotic cells to a temperature close to zero to stop the progression of cell death and taking Raman spectra from cooled cells may be a very promising approach to accomplish this aim. Studying different strains of CHO cells and other host cell lines can broaden our knowledge about cell death and can help to identify and study the general features of apoptotic and necrotic cell death. FACS sorting can be used to isolate pure early apoptotic,  68 late apoptotic, primary and secondary necrotic populations. Study of these well-defined populations in longer temporal courses can give a deeper understating of different stages of cell death. High resolution Raman spectroscopy can be used to monitor intracellular changes of RNA, DNA, proteins and different organelles during cell death induction, which can give a detailed image of cell death machinery at the sub-cellular level.                      69 Bibliography Adams, D.J. et al., 2006. Camptothecin analogs with enhanced activity against human breast cancer cells. II. Impact of the tumor pH gradient. Cancer chemotherapy and pharmacology, 57(2), pp.145–54. Almarashi, J.F.M. et al., 2012. Raman Spectroscopy of Bacterial Species and Strains Cultivated under Reproducible Conditions. Spectroscopy: An International Journal, 27(27), pp.361–365. Alnemri, E.S. et al., 1996. Human ICE/CED-3 protease nomenclature. Cell, 87(2), p.171. Astrakas, L.G. et al., 2005. Proton NMR spectroscopy shows lipids accumulate in skeletal muscle in response to burn trauma-induced apoptosis. FASEB journal : official publication of the Federation of American Societies for Experimental Biology, 19(11), pp.1431–40. Baehrecke, E.H., 2005. Autophagy: dual roles in life and death? Nature reviews. Molecular cell biology, 6(6), pp.505–10. Bianchi, M.E. & Manfredi, A., 2004. Chromatin and cell death. Biochimica et biophysica acta, 1677(1-3), pp.181–6. Bitar Carter, R.A. et al., 2004. <title>FT-Raman spectroscopy study of human breast tissue</title>. In A. Mahadevan-Jansen et al., eds. Biomedical Optics 2004. International Society for Optics and Photonics, pp. 190–197. Brauchle, E. et al., 2014. Cell death stages in single apoptotic and necrotic cells monitored by Raman microspectroscopy. Scientific reports, 4, p.4698. Carmeliet, P. et al., 1998. Role of HIF-1alpha in hypoxia-mediated apoptosis, cell proliferation and tumour angiogenesis. Nature, 394(6692), pp.485–90. Carmona-Gutierrez, D. et al., 2010. Apoptosis in yeast: triggers, pathways, subroutines. Cell death and differentiation, 17(5), pp.763–73. Chaurio, R.A. et al., 2009. Phospholipids: key players in apoptosis and immune regulation. Molecules (Basel, Switzerland), 14(12), pp.4892–914. Crimi, M. & Esposti, M.D., 2011. Apoptosis-induced changes in mitochondrial lipids. Biochimica et biophysica acta, 1813(4), pp.551–7. Darzynkiewicz, Z. et al., 1997. Cytometry in cell necrobiology: analysis of apoptosis and accidental cell death (necrosis). Cytometry, 27(1), pp.1–20. Díaz-Guerra, M., Rivas, C. & Esteban, M., 1997. Activation of the IFN-inducible enzyme RNase L causes apoptosis of animal cells. Virology, 236(2), pp.354–63.  70 Downes, A. et al., 2011. Raman spectroscopy and CARS microscopy of stem cells and their derivatives. Journal of Raman spectroscopy : JRS, 42(10), pp.1864–1870. Efferth, T. et al., 2007. Molecular target-guided tumor therapy with natural products derived from traditional Chinese medicine. Current medicinal chemistry, 14(19), pp.2024–32. Eguchi, Y., Shimizu, S. & Tsujimoto, Y., 1997. Intracellular ATP levels determine cell death fate by apoptosis or necrosis. Cancer research, 57(10), pp.1835–40. Eisenberg-Lerner, A. et al., 2009. Life and death partners: apoptosis, autophagy and the cross-talk between them. Cell death and differentiation, 16(7), pp.966–75. El-Mashtoly, S.F. et al., 2014. Automated identification of subcellular organelles by coherent anti-stokes Raman scattering. Biophysical journal, 106(9), pp.1910–20. Ellis, D.I. & Goodacre, R., 2006. Metabolic fingerprinting in disease diagnosis: biomedical applications of infrared and Raman spectroscopy. The Analyst, 131(8), pp.875–85. Enari, M. et al., 1998. A caspase-activated DNase that degrades DNA during apoptosis, and its inhibitor ICAD. Nature, 391(6662), pp.43–50. Fackler, O.T. & Grosse, R., 2008. Cell motility through plasma membrane blebbing. The Journal of cell biology, 181(6), pp.879–84. Festjens, N., Vanden Berghe, T. & Vandenabeele, P., Necrosis, a well-orchestrated form of cell demise: signalling cascades, important mediators and concomitant immune response. Biochimica et biophysica acta, 1757(9-10), pp.1371–87. Fraschini, A. et al., 2005. Changes in extranucleolar transcription during actinomycin D-induced apoptosis. Histology and histopathology, 20(1), pp.107–17. Garland, J.M. & Halestrap, A., 1997. Energy Metabolism during Apoptosis: bcl-2 PROMOTES SURVIVAL IN HEMATOPOIETIC CELLS INDUCED TO APOPTOSE BY GROWTH FACTOR WITHDRAWAL BY STABILIZING A FORM OF METABOLIC ARREST. Journal of Biological Chemistry, 272(8), pp.4680–4688. Golstein, P. & Kroemer, G., 2007. Cell death by necrosis: towards a molecular definition. Trends in biochemical sciences, 32(1), pp.37–43. Green, D. & Kroemer, G., 1998. The central executioners of apoptosis: caspases or mitochondria? Trends in cell biology, 8(7), pp.267–71. Green, D.R. & Kroemer, G., 2004. The pathophysiology of mitochondrial cell death. Science (New York, N.Y.), 305(5684), pp.626–9. Hlatky, L., Sachs, R.K. & Alpen, E.L., 1988. Joint oxygen-glucose deprivation as the cause of necrosis in a tumor analog. Journal of cellular physiology, 134(2), pp.167–78.  71 Huefner, A. et al., 2013. Intracellular SERS nanoprobes for distinction of different neuronal cell types. Nano letters, 13(6), pp.2463–70. Ito, N., DeMarco, R. & Lotze, M., 2006. Sa.63. Successful Identification of Necrotic Cell Death at the Single Cell Level: Flow Cytometric Analysis of High Mobility Group B1 and Histone H1. Clinical Immunology, 119(119), p.S127. Janko, C. et al., 2011. CRP discriminates primary from secondary necrosis. Annals of the Rheumatic Diseases, 70(Suppl 2), pp.A8–A8. Jardon, M.A. et al., 2012. Inhibition of glutamine-dependent autophagy increases t-PA production in CHO cell fed-batch processes. Biotechnology and bioengineering, 109(5), pp.1228–38. Jardon, M.A., 2012. Modulating autophagy and glutamine metabolism in CHO cells to increase fed-batch process performance. Kanduc, D. et al., 2002. Cell death: apoptosis versus necrosis (review). International journal of oncology, 21(1), pp.165–70. Kemnitz, C.R. & Loewen, M.J., 2007. “Amide resonance” correlates with a breadth of C-N rotation barriers. Journal of the American Chemical Society, 129(9), pp.2521–8. Kennard, M.L. et al., 2009. Auditioning of CHO host cell lines using the artificial chromosome expression (ACE) technology. Biotechnology and bioengineering, 104(3), pp.526–39. Kerr, J.F.R., 1965. A histochemical study of hypertrophy and ischaemic injury of rat liver with special reference to changes in lysosomes. The Journal of Pathology and Bacteriology, 90(2), pp.419–435. Klionsky, D.J. et al., 2008. Guidelines for the use and interpretation of assays for monitoring autophagy in higher eukaryotes. Autophagy, 4(2), pp.151–75. Koljenović, S. et al., Tissue characterization using high wave number Raman spectroscopy. Journal of biomedical optics, 10(3), p.031116. Konorov, S.O. et al., 2007. In situ analysis of living embryonic stem cells by coherent anti-stokes Raman microscopy. Analytical chemistry, 79(18), pp.7221–5. Konorov, S.O. et al., 2012. Raman microspectroscopy of live cells under autophagy-inducing conditions. The Analyst, 137(20), pp.4662–8. Krafft, C. et al., 2006. Studies on stress-induced changes at the subcellular level by Raman microspectroscopic mapping. Analytical chemistry, 78(13), pp.4424–9. Kroemer, G. et al., 2009. Classification of cell death: recommendations of the Nomenclature Committee on Cell Death 2009. Cell death and differentiation, 16(1), pp.3–11. Kroemer, G. et al., 2005. Classification of cell death: recommendations of the Nomenclature Committee on Cell Death. Cell death and differentiation, 12 Suppl 2, pp.1463–7.  72 Krysko, D. V et al., 2008. Methods for distinguishing apoptotic from necrotic cells and measuring their clearance. Methods in enzymology, 442, pp.307–41. Kunapareddy, N., Freyer, J.P. & Mourant, J.R., Raman spectroscopic characterization of necrotic cell death. Journal of biomedical optics, 13(5), p.054002. Ladiwala, U., Bankapur, A. & Thakur, B., Raman spectroscopic detection of rapid , reversible , early-stage inflammatory cytokine-induced apoptosis of adult hippocampal progenitors / stem cells. , pp.1–23. Laporte, L., Stultz, J. & Thomas, G.J., 1997. Solution conformations and interactions of alpha and beta subunits of the Oxytricha nova telomere binding protein: investigation by Raman spectroscopy. Biochemistry, 36(26), pp.8053–9. Levine, B. & Klionsky, D.J., 2004. Development by self-digestion: molecular mechanisms and biological functions of autophagy. Developmental cell, 6(4), pp.463–77. Li, Y.C. et al., 2006. Elevated levels of cholesterol-rich lipid rafts in cancer cells are correlated with apoptosis sensitivity induced by cholesterol-depleting agents. The American journal of pathology, 168(4), pp.1107–18; quiz 1404–5. Lin, N.-Y. et al., 2013. Autophagy regulates TNFα-mediated joint destruction in experimental arthritis. Annals of the rheumatic diseases, 72(5), pp.761–8. Maccarrone, M. & Finazzi-Agró, A., 2003. The endocannabinoid system, anandamide and the regulation of mammalian cell apoptosis. Cell death and differentiation, 10(9), pp.946–55. Majno, G. & Joris, I., 1995. Apoptosis, oncosis, and necrosis. An overview of cell death. The American journal of pathology, 146(1), pp.3–15. Matsushita, M. et al., 2007. Structure of Atg5.Atg16, a complex essential for autophagy. The Journal of biological chemistry, 282(9), pp.6763–72. McIntyre, Catherine ;Mittar, D., Detection of Apoptosis Using the BD Annexin V FITC Assay on the BD FACSVerseTM System. Available at: http://www.bdbiosciences.com/documents/BD_FACSVerse_Apoptosis_Detection_AppNote.pdf [Accessed December 8, 2014]. Melino, G., 2001. The Sirens’ song. Nature, 412(6842), p.23. Mizushima, N., Yoshimori, T. & Levine, B., 2010. Methods in mammalian autophagy research. Cell, 140(3), pp.313–26. Mizuta, R. et al., 2013. DNase γ is the effector endonuclease for internucleosomal DNA fragmentation in necrosis. H. Nakano, ed. PloS one, 8(12), p.e80223. Movasaghi, Z., Rehman, S. & Rehman, I.U., 2007. Raman Spectroscopy of Biological Tissues. Applied Spectroscopy Reviews, 42(5), pp.493–541.  73 Nasseri, S.S., 2013. Autophagy inhibition and hypertonicity increased monoclonal antibody production in chinese hamster ovary fed-batch cultures. Nicotera, P., Bernassola, F. & Melino, G., 1999. Nitric oxide (NO), a signaling molecule with a killer soul. Cell death and differentiation, 6(10), pp.931–3. Notingher, I. et al., 2004. Discrimination between ricin and sulphur mustard toxicity in vitro using Raman spectroscopy. Journal of the Royal Society, Interface / the Royal Society, 1(1), pp.79–90. Notingher, I., Selvakumaran, J. & Hench, L.L., 2004. New detection system for toxic agents based on continuous spectroscopic monitoring of living cells. Biosensors & bioelectronics, 20(4), pp.780–9. Ong, Y.H., Lim, M. & Liu, Q., 2012. Comparison of principal component analysis and biochemical component analysis in Raman spectroscopy for the discrimination of apoptosis and necrosis in K562 leukemia cells. Optics express, 20(20), pp.22158–71. OPIE, E.L., 1962. On the relation of necrosis and inflammation to denaturation of proteins. The Journal of experimental medicine, 115, pp.597–608. Owen, C.A. et al., 2006. In vitro toxicology evaluation of pharmaceuticals using Raman micro-spectroscopy. Journal of cellular biochemistry, 99(1), pp.178–86. Peticolas, W.L., 1975. Application of Raman spectroscopy to biological macromolecules. Biochimie, 57(4), pp.417–28. Pommier, Y., 2003. Repair of and checkpoint response to topoisomerase I-mediated DNA damage. Mutation Research/Fundamental and Molecular Mechanisms of Mutagenesis, 532(1-2), pp.173–203. Del Prete, M.J. et al., 2002. Degradation of cellular mRNA is a general early apoptosis-induced event. FASEB journal : official publication of the Federation of American Societies for Experimental Biology, 16(14), pp.2003–5. Proskuryakov, S.Y., Konoplyannikov, A.G. & Gabai, V.L., 2003. Necrosis: a specific form of programmed cell death? Experimental cell research, 283(1), pp.1–16. Prucek, R. et al., 2012. Reproducible discrimination between gram-positive and gram-negative bacteria using surface enhanced Raman spectroscopy with infrared excitation. The Analyst, 137(12), pp.2866–70. PUPPELS, G., 1991. Laser irradiation and Raman spectroscopy of single living cells and chromosomes: Sample degradation occurs with 514.5 nm but not with 660 nm laser light*1. Experimental Cell Research, 195(2), pp.361–367. Richter, C. et al., 1996. Control of apoptosis by the cellular ATP level. FEBS Letters, 378(2), pp.107–110.  74 Roach, H.I. & Clarke, N.M., 2000. Physiological cell death of chondrocytes in vivo is not confined to apoptosis. New observations on the mammalian growth plate. The Journal of bone and joint surgery. British volume, 82(4), pp.601–13. Rodríguez-Casado, A. et al., 2001. Structure of Bacteriophage P22 Portal Protein in Relation to Assembly:  Investigation by Raman Spectroscopy †. Biochemistry, 40(45), pp.13583–13591. Roszczenko, A. et al., 2013. The effect of exposure to chlorfenvinphos on lipid metabolism and apoptotic and necrotic cells death in the brain of rats. Experimental and toxicologic pathology : official journal of the Gesellschaft für Toxikologische Pathologie, 65(5), pp.531–9. Salehi, H. et al., 2013. Confocal Raman data analysis enables identifying apoptosis of MCF-7 cells caused by anticancer drug paclitaxel. Journal of biomedical optics, 18(5), p.56010. Scaffidi, P., Misteli, T. & Bianchi, M.E., 2002. Release of chromatin protein HMGB1 by necrotic cells triggers inflammation. Nature, 418(6894), pp.191–5. Schäfer, R. & Schmidt, P.C. eds., 2012. Methods in Physical Chemistry, Weinheim, Germany: Wiley-VCH Verlag GmbH & Co. KGaA. Schulze, H.G. et al., 2013. Label-free imaging of mammalian cell nucleoli by Raman microspectroscopy. The Analyst, 138(12), pp.3416–23. Szpak, P., 2011. Fish bone chemistry and ultrastructure: implications for taphonomy and stable isotope analysis. Journal of Archaeological Science, 38(12), pp.3358–3372. Tsujimoto, Y., 1997. Apoptosis and necrosis: intracellular ATP level as a determinant for cell death modes. Cell death and differentiation, 4(6), pp.429–34. Wang, J.C., 2002. Cellular roles of DNA topoisomerases: a molecular perspective. Nature reviews. Molecular cell biology, 3(6), pp.430–40. Wlodkowic, D. et al., 2011. Apoptosis and beyond: cytometry in studies of programmed cell death. Methods in cell biology, 103, pp.55–98. Wyllie, A.H., 1980. Glucocorticoid-induced thymocyte apoptosis is associated with endogenous endonuclease activation. Nature, 284(5756), pp.555–6. Xie, Z. & Klionsky, D.J., 2007. Autophagosome formation: core machinery and adaptations. Nature cell biology, 9(10), pp.1102–9. Yikrazuul, May 2009, Binding of CPT to topo I and DNA, viewed 3 August 2014, <http://en.wikipedia.org/wiki/Camptothecin#mediaviewer/File:Camptothecin_binding.svg >. Yang, M.-Y. et al., 2002. Reversible phosphatidylserine expression on blood granulocytes related to membrane perturbation but not DNA strand breaks. Journal of leukocyte biology, 71(2), pp.231–7.  75 Zamaraeva, M. V et al., 2005. Cells die with increased cytosolic ATP during apoptosis: a bioluminescence study with intracellular luciferase. Cell death and differentiation, 12(11), pp.1390–7. Zoladek, A. et al., 2011. Non-invasive time-course imaging of apoptotic cells by confocal Raman micro-spectroscopy. Journal of Raman Spectroscopy, 42(3), pp.251–258.                        76 Appendices   Appendix A: custom written software for spectra data analysis  We used this program for background subtraction, baseline correction (flattening), spectra smoothing, automated cosmic spike removal, peak stripping, automated peak detection and spectra normalization. The code has been created and executed using Matlab 12.0 (The MathWorks, Natick, MA). Program’s Code: % Processing sequence % (Converts sequence to code that calls functions at each of numbered points; user input required only at point 1) %  % Processing % 1. Load a sequence of files or load a map and parse (determined automatically) and get key input information % 2. Load background, if any, then load a sequence of files or load a map and parse. Subtract background % 3. Do baseline flattening (blf)  % 4. Remove spikes  % 5. Get standard deviation and SNR of data set from subsample % 6. Smooth (determine best method) % 7. Scale to counter smoothing effects % 8. Normalize (select method: spectral max, internal standard, vector normalization, probability quotient vector normalization, volume exclusion) % 9. QC:compare smoothed originals to smoothed finals using normalized derivatives and PCA scores   77 %  % Analysis % 10. PCA and check for clustering/uniformity (alert user if clustering present); ID and provide principal components % 11. 2DCOS and provide synchronous and asynchronous maps % 12. Statistics?   % Code   %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% % Specify core results name, intermediate results will be appended to this core name % Example % input filename:    '542ThreeWeeks - Raw Data - 1 - Liquid + coverslip - 1 acc 40 sec acq x20.txt  % core results name: 'D542liq_ 3week' % Intermediate results matrices, as successive processing steps are performed, will have suffixes appended to this core name.  % E.g. , after loading raw data it becomes 'D542liq_ 3week_raw',  % after subsequent 15-iteration baseline flattening it becomes 'D542liq_ 3week_raw _blf15', etc. % Note: uipickfiles may be better, download from Matlabcentral if needed.   clear all close all pauseTime = 0; processNum = 0; suffixes = {};  78 warningstrings = {}; warningNumber  = 0;  % 1. Load a sequence of files or load a map and parse (determined automatically) and get key input information resultsName = 'Temp_'; loadBoxTitle = 'Select the analyte file(s) to load'; [pathName, CellofFileNames, formatDataLoaded, generatedvarname, suffix]  = f_loadDatafile _s _v2(resultsName, loadBoxTitle, pauseTime); processNum  = processNum + 1; suffixes{processNum} = suffix; useData = eval(generatedvarname); clear(generatedvarname)  origStr = pathName(length(pathName)-12:length(pathName)); modifiedStr = strrep(origStr, '/', 'x'); modifiedStr = strrep(modifiedStr, '_', ''); resultsName = strrep(modifiedStr, ' ', ''); prompt = {'Enter a descriptive core name (e.g. ID_054_liq _ 24h_ 3mW) for your data results, for intermediate results, suffixes will be appended to this core name (e.g. ID_054_liq _ 24h_ 3mW_raw _blf15 etc.). A suggestion is supplied below. ',...           'Are there background spectra to load and subtract from the analyte spectra? Y/N : ',...           'The number of iterations to use for baseline flattening can be determined automatically if the default is accepted. Otherwise, specify the number of iterations to use : ',...     'Smoothing of spectra uses a Savitzky-Golay filter, the default is a zero-order filter. A second-order filter could be used if desired, resulting in generally less smoothing but  79 better peak retention. Higher orders are not recommended. Enter a different order (e.g. 2) if desired: ',...     'Smoothing of spectra uses the specified Savitzky-Golay filter iteratively and the number of iterations can be determined automatically. If fewer iterations are required, specify them now: '...           'Specify the time to pause between processing steps. This will activate the plotting of intermediate results. Useful if troubleshooting unsatisfactory processing, otherwse use 0.'...           'Specify the saturation level of the current detector (i.e. the maximum spectral value that can be obtained). Spectra with such values, along with other pathological spectra, will be culled before analysis.'};  dlg_title = 'Input for semi-automated spectral data processing'; num_lines = [2, 100]; def = {resultsName,'N','Auto','0','Auto','0','156175'}; useranswer = inputdlg(prompt,dlg_title,num_lines,def); if isempty(useranswer)     % No input, return to keyboard     clear all     close all  return else     for n = 1:length(useranswer);useranswer{n} = strtrim(useranswer{n});end; end  resultsName = [useranswer {1} '_']; figName = [useranswer {1} ' '];  80 pauseTime = str2num(useranswer{6}); % useranswer{2} assessed below if isempty(str2num(useranswer{3}))     Max_j = 1e128; else     Max_j = str2num(useranswer{3}); end polyorder = str2num(useranswer{4}); if isempty(str2num(useranswer{5}))     iterations  =  1e128; else     iterations = str2num(useranswer{5}); end saturationLevel = 0.99*str2num(useranswer{6});  % Specify normalization methods     optionsStr = {'1. Spectral maximum method (generally destroys correlation structure of data, follow-up data analyses of limited use)',...                   '2. Internal standard method (preserves correlation structure of data, follow-up data analyses useful)',...                   '3. Vector normalization method (generally destroys correlation structure of data, follow-up data analyses of limited use)',...                   '4. Probability quotient vector normalization method (NOT good where numerous peaks change, but preserves correlation structure of data, follow-up data analyses possibly useful)',...                   '5. Volume exclusion method not available in this arrangement',...  81                   '6. None (when no normalization is desired or initially when unsure about the appropriate normalization approach).'};     [normMethod,responseMade] = listdlg('PromptString','Select a normalization method:',...                                         'SelectionMode','single',...                                         'ListString',optionsStr,'ListSize',[999 100], 'InitialValue', 6);     if (normMethod(1) == 2 || normMethod(1) == 5)         prompt = 'What is the peak position of the internal standard or the volume exclusion reference? Specify in terms of x-values, e.g. if the x-values are in pixels, specify approximate pixel number, if in wavenumbers, specify approximate wavenumber: ';         def = {'0'};         temp = inputdlg(prompt,'Provide reference peak position',2,def);         peakPosition = str2num(temp{1});     else         peakPosition = 0;      end generatedvarname = genvarname([resultsName suffix]); BGname = [resultsName  'bg']; BGgeneratedvarname = [resultsName  'bg']; assignin('base', generatedvarname, useData); % Check to see if spectra actually contain data indices = useData(:,2:end) > 0; if isempty(indices)     Message = ['Spectra all zeros or negative. Please check your data. Processing aborted.'] ;     Title = 'Data integrity';     h = msgbox(Message, Title, 'warn', 'modal');   82     return; end disp([formatDataLoaded ' into variable:    ' generatedvarname]); pause(pauseTime) % Load background, if any. if isempty(useranswer{2}) || useranswer{2} == 'Y' || useranswer{2} == 'y'     loadBoxTitle = 'Select the background file(s) to load';  [pathName, CellofBGFileNames, formatDataLoaded, BGgeneratedvarname, suffix]  = f_loadDatafile _s _v2(BGgeneratedvarname, loadBoxTitle, pauseTime);  bckground = eval(BGgeneratedvarname);          if pauseTime > 0 % Plot loaded data for inspection          figHandles = f_plotRaman(2, [1,2], figName); figure(figHandles(1)); plot(bckground(:,1), bckground(:,2:end)); axis tight                 if size(bckground,2) > 2            figure(figHandles(2)); surf(bckground(:,1), 1:size(bckground,2)-1, bckground (:,2:end)'); shading interp; axis tight;         else             close(figHandles(2))         end     end     % Check to see if background spectra actually contain data     indices = bckground(:,2:end) > 0;     if isempty(indices)         Message = ['Background all zeros or negative. Please check your data. Processing aborted.'] ;  83         Title = 'Data integrity';         h = msgbox(Message, Title, 'warn', 'modal');          return;     end     disp([formatDataLoaded ' into variable:    ' BGgeneratedvarname]) end Message = ['Data loaded. Working on step ' num2str (processNum+1) '...'] ; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']); %%%%%%%%%%%%  -----   Begin processing  -----  %%%%%%%%%%%%%%%%%%%%%%%%%% % 2. Subtract background, if any close all; if isempty(useranswer{2}) || useranswer{2} == 'Y' || useranswer{2} == 'y'     BGtemp = eval(BGgeneratedvarname);     if size(BGtemp,2) > 10         % can use f_spikeRemoval here if f_spikeRemoval _v4 results unsatisfactory        [BGgeneratedvarname, suffix, indicesCombi] = f_spikeRemoval _v4(BGtemp, BGgeneratedvarname, pauseTime);     else % cannot use f_spikeRemoval for fewer than 4 spectra (i.e. below), not recommended for fewer than 10         [BGgeneratedvarname, suffix] = f_deSpike _v2(BGtemp, BGgeneratedvarname, pauseTime);         end  84     disp(['Background after spike removal in variable:    ' BGgeneratedvarname])     pause(pauseTime)     close all;          % Subtract background     average_Noise = 0; % or estimate with: "average_Noise = f_spectralStddev_v2(useData(:,2));"     [generatedvarname, suffix]  = f_incremBackgroundCorrection _v2(useData, bckground, generatedvarname, pauseTime);     processNum  = processNum + 1;     suffixes{processNum} = suffix;     useData = eval(generatedvarname);     disp(['Spectra after background subtraction in variable:    ' generatedvarname])     pause(pauseTime) else     processNum  = processNum + 1;     suffixes{processNum} = 'nobgs_';       generatedvarname = genvarname([generatedvarname suffixes{processNum}]);     assignin('base', generatedvarname, useData);     useData = eval(generatedvarname); end Message = ['Background removal, if any, completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  % input(['Number ' num2str (processNum) ' finished']);  85  % 3. Do baseline flattening (blf)  close all suffix2 = ''; if Max_j == 1e128      % Determine number of iterations to use automatically on a subsample of 20 spectra     if size(useData,2) > 20         subsample(:, 1)   = useData(:,1);         subsample(:,2:21) = useData(:,[randi([2,size(useData,2)],1,20)]);     else         subsample  = useData;     end     pixelwidth = useData(2,1) - useData(1,1);     [blf_iterations _used, tempgeneratedvarname, suffix2] = f_autobaselineRemoval _SWiMA _v2(subsample, Max_j, generatedvarname, pauseTime);     if 2*blf_iterations _used*pixelwidth > 50         warningNumber  = warningNumber + 1;         blf_iterations _used = ceil(50/(2*pixelwidth));         new_warningstrings{warningNumber} = ['Too many baseline flattening iterations, window size possibly too large. Iterations limited to ' num2str (blf_iterations _used) '.'];         warningstrings = {warningstrings{:}, new_warningstrings{:}};     end     generatedvarname = tempgeneratedvarname;     BGgeneratedvarname = [BGgeneratedvarname suffix2];     Max_j = blf_iterations _used;     clear('subsample');  86     close all     end % Use same parameter settings for background and analyte spectral baseline flattening [blf_iterations _used, generatedvarname, suffix] = f_autobaselineRemoval _SWiMA _v2(useData, Max_j, generatedvarname, pauseTime); useData = eval(generatedvarname); processNum  = processNum + 1; suffixes{processNum} = [suffix2 suffix]; disp(['Baseline flattened data in variable:    ' generatedvarname]) Message = ['Baseline flattened. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']);  % 4. Remove spikes  close all indicesCombi = []; if size(useData,2) > 10    [generatedvarname, suffix, indicesCombi] = f_spikeRemoval _v4(useData, generatedvarname, pauseTime); else % cannot use f_spikeRemoval for fewer than 4 spectra (i.e. below), not recommended for fewer than 10     [generatedvarname, suffix] = f_deSpike _v2(useData, generatedvarname, pauseTime); end if sum(sum(indicesCombi)) > 0  87     warningNumber  = warningNumber + 1;             new_warningstrings{warningNumber} = ['Problem spikes may exist, see indicesCombi for locations. Number of problems: ' num2str (sum(sum(indicesCombi))) '.'];     warningstrings = {warningstrings{:}, new_warningstrings{:}}; end processNum  = processNum + 1; suffixes{processNum} = suffix; useData = eval(generatedvarname); disp(['Data after spike removal in variable:    ' generatedvarname]) Message = ['Cosmic spikes removed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']);  % 5. Get standard deviation and SNR of data set from subsample close all; specStddevs  = []; specMaxPeaks = []; SNRs         = []; % Get noise of original data origData = eval(genvarname([resultsName suffixes{1}])); % Raw data for n = 2:size(origData,2)  specStddevs(n)  = f_spectralStddev _v2(origData(:,n)); end % Get max peak values of baseline-flattened and despiked data  88 for n = 2:size(origData,2)  specMaxPeaks(n) = max(origData(:,n));  SNRs(n)         = specMaxPeaks (n)/specStddevs(n); end average_SNR   = mean(SNRs(2:end)); average_Noise = mean(specStddevs(2:end)); processNum  = processNum + 1; suffixes{processNum} = ''; disp(['In ' genvarname([resultsName suffixes{1}]) ]) disp(['the estimated mean SNR is:   ' num2str (average_SNR) ' and ']) disp(['the estimated mean spectral noise is:   '  num2str(average_Noise)]) Message = ['Basic spectral attributes determined. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']);  %  6. Smooth  close all; [generatedvarname, suffix, smoothingIters_used] = f_smoothAutomated _v2(useData, generatedvarname, iterations, polyorder, pauseTime); processNum  = processNum + 1; suffixes{processNum} = suffix; useData = eval(generatedvarname); disp(['Data after smoothing in variable:    ' generatedvarname])  89 Message = ['Smoothing completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']); % 7. Scale smoothed results close all; noisyData = eval(genvarname([resultsName suffixes{1:processNum}]));  [generatedvarname, suffix] = f_weightNscale _smoothedData(noisyData, useData, generatedvarname, pauseTime); processNum  = processNum + 1; suffixes{processNum} = suffix; useData = eval(generatedvarname); disp(['Data after scaling in variable:    ' generatedvarname]) Message = ['Scaling completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) % input(['Number ' num2str (processNum) ' finished']); % 8. Normalize (select method: spectral max, internal standard, vector normalization, probability quotient vector normalization, volume exclusion) close all; if responseMade == 0  % No normalization was selected, further analyses skipped. Return to keyboard  return else   90  switch normMethod     case 1 % Maximum peak        [generatedvarname, suffix] = f_normInternalStd(useData, generatedvarname, peakPosition, pauseTime);              processNum  = processNum + 1;              suffixes{processNum} = suffix;     case 2 % Internal standard         [generatedvarname, suffix] = f_normInternalStd(useData, generatedvarname, peakPosition, pauseTime);              processNum  = processNum + 1;              suffixes{processNum} = suffix;     case 3 % Vector normalization (really: constant sum)         [generatedvarname, suffix] = f_normConstSum(useData, generatedvarname, pauseTime);              processNum  = processNum + 1;              suffixes{processNum} = suffix;     case 4 % Probability quotient vector normalization             [generatedvarname, suffix, new_warningstrings] = f_normProbQuotient(useData, generatedvarname, specStddevs, pauseTime);             warningstrings = {warningstrings{:}, new_warningstrings{:}};             processNum  = processNum + 1;             suffixes{processNum} = suffix;     case 5 % Volume exclusion normalization (variant of internal standard)           dsp('Not available in this arrangement, normalized to maximum peak')       [generatedvarname, suffix] = f_normInternalStd(useData, generatedvarname, peakPosition, pauseTime);  91             processNum  = processNum + 1;             suffixes{processNum} = suffix;     case 6 % No normalization             suffix = 'notnormd_' ;                         generatedvarname = [generatedvarname suffix];             assignin('base', generatedvarname, useData);             processNum  = processNum + 1;             suffixes{processNum} = suffix;  end end useData = eval(generatedvarname);  disp(['Data after normalization in variable:    ' generatedvarname]) Message = ['Normalization completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  pause(pauseTime) save([pathName '/' resultsName '.mat']); disp(['Intermediate results saved in: ' resultsName '.mat']) % return; % input(['Number ' num2str (processNum) ' finished']);  % 9. QC:compare smoothed originals to smoothed finals using normalized derivatives and PCA scores  pauseTime = 1; if size(useData,2) <= 2    % Diagnostic plots  92     close all     set(0,'Units','pixels');     scrsz = get(0,'ScreenSize');         figHandles = f_plotRaman(2, [1,1], figName);         figure(figHandles(1));tempdata = evalin('base','genvarname ([resultsName suffixes{1}])');plot(evalin('base',[tempdata '(:,1)']), evalin('base',[tempdata '(:,2:end)']));axis tight         title([figName 'Raw spectra']);         figure(figHandles(2));         for n = 1:size(useData,2)-1             plot(useData(:,1), useData(:,n+1),'color', [(0.75 + 0.25*(-1)^n)*n/size(useData,2) (size(useData,2)-n)/size(useData,2) (1 - 0.5*(-1)^n)*n/(2*n)],'linewidth',2);axis tight                 end         title([figName 'Processed spectra']);          save([pathName '/' resultsName '.mat']);         Message = 'Processing and analyses completed.';         Title = 'Processing status';         h = msgbox(Message,Title,'warn','modal');     return end  iterations = mode(smoothingIters_used); [generatedvarname, suffix, smoothingIters_used] = f_smoothAutomated _v2(origData, genvarname([resultsName suffixes{1}]), iterations, polyorder, pauseTime); QCraw = eval(generatedvarname); QCprocessed = useData; clear(generatedvarname);  93 [QCgeneratedvarname_first, QCgeneratedvarname_last, outliers, new_warningstrings] = f_QualityControl _v4(QCprocessed, QCraw, pauseTime, figName); warningstrings = {warningstrings{:}, new_warningstrings{:}}; disp(['Quality control data in variables:    ' QCgeneratedvarname_first ' and ' QCgeneratedvarname_last]) processNum  = processNum + 1; suffixes{processNum} = ''; if pauseTime > 0 || ~isempty(warningstrings)      Message = ['Quality control data acquired. Working on step ' num2str (processNum+1) '...'];     Title = 'Processing status';     h = msgbox(Message,Title,'warn','modal');  else     Message = ['Quality control data acquired. Working on step ' num2str (processNum+1) '...'];     Title = 'Processing status';     h = msgbox(Message,Title,'warn','modal');  end for n = 1:6     switch n                  case 1          saveas(n, [pathName '/' resultsName suffixes {1} '.fig']);      case 2          saveas(n, [pathName '/' resultsName suffixes {1:processNum} '.fig']);      case 3          saveas(n, [pathName '/' resultsName suffixes {1:processNum} 'surf.fig']);  94      case 4          saveas(n, [pathName '/' resultsName  'compare_ 1deriv.fig']);      case 5          saveas(n, [pathName '/' resultsName  'compare_ 2deriv.fig']);      case 6          saveas(n, [pathName '/' resultsName suffixes {processNum} 'clustering.fig']);      otherwise          saveas(n, [pathName '/' resultsName 'unspecified.fig']);             end end % input(['Number ' num2str (processNum) ' finished']); return  % 10. PCA:  ID and provide principal components % close all; [PCAldgs, PCAscrs, PCAeigs] = princomp(useData (:,2:end)'); PCAeigs = PCAeigs/sum(PCAeigs); sumeigs = 0; PCs_major = 0; while sumeigs < 0.95 && PCs_major < 5     PCs_major = PCs_major + 1;     sumeigs = sumeigs + PCAeigs(PCs_major); end figHandles = f_plotRaman(3, [3,4,5], figName);  figure(figHandles(1));  for n = 1:PCs_major  95     plot(useData(:,1), PCAldgs(:,n),'color', [(0.75 + 0.25*(-1)^n)*n/size(useData,2) (size(useData,2)-n)/size(useData,2) (1 - 0.5*(-1)^n)*n/(2*n)],'linewidth',1);axis tight; %     pause(pauseTime) end  axis tight; saveas(figHandles(1), [pathName '/' resultsName 'PCAldgs.fig']); figure(figHandles(2)); for n = 1:PCs_major      plot(PCAscrs(:,n), PCAscrs(:,n+1),'*','color', [(0.75 + 0.25*(-1)^n)*n/size(useData,2) (size(useData,2)-n)/size(useData,2) (1 - 0.5*(-1)^n)*n/(2*n)],'linewidth',2); axis tight;  %      pause(pauseTime) end axis tight; saveas(figHandles(2), [pathName '/' resultsName 'PCAscrs.fig']); figure(figHandles(3));  for n = 1:2*PCs_major     plot(100*PCAeigs(1:2*PCs_major),'ok','linewidth',2); end axis tight; saveas(figHandles(3), [pathName '/' resultsName 'PCAeigs.fig']); disp(['Principal component analysis results in :    PCAldgs, PCAscrs, and PCAeigs.']) processNum  = processNum + 1; suffixes{processNum} = ''; Message = ['Principal component analysis completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status';  96 h = msgbox(Message,Title,'warn','modal');  % input(['Number ' num2str (processNum) ' finished']); % 11. 2DCOS and provide synchronous and asynchronous maps [TDCOV, PDDCOV, TDCOR, PDDCOR, TDASN, PDDASN, TDDSR, PDDDSR] = f_ 2DCOS_ 2DDRS(useData(:,1), 1:size(useData,2)-1, useData(:,2:end)); figHandles = f_plotRaman(2, [6,6], figName); % 2DCOS maps figure(figHandles(1)); fig_axes = findobj(figHandles(1)); axes(fig_axes(max(size(fig_axes)))); surf(TDCOR(2:end,1), TDCOR(1,2:end), TDCOR(2:end,2:end)); shading interp; view(2)     % View azimuth = 0; elevation = 90 --> view from top contour3(TDCOV(2:end,1), TDCOV(1,2:end), TDCOV(2:end,2:end)+max(max(TDCOV(2:end,2:end)))+1, 'k');    axis tight box on hold on; axes(fig_axes(max(size(fig_axes)-1))); plot(useData(:,1), useData(:,2),'k'); set(gca,'XTickLabel',[]); set(gca,'YTickLabel',[]); axis tight box on title([figName '2DCOS maps'],'FontName','Times','FontSize',24); axes(fig_axes(max(size(fig_axes)-2))); plot(useData(:,end),useData(:,1),'k'); set(gca,'XTickLabel',[]); set(gca,'YTickLabel',[]); axis tight  97 box on drawnow; saveas(figHandles(1), [pathName '/' resultsName '2DCOS.fig']); % Perturbation domain decompositions figure(figHandles(2));           fig_axes = findobj(figHandles(2)); axes(fig_axes(max(size(fig_axes)))); surf(PDDCOR(2:end,1), PDDCOR(1,2:end), PDDCOR (2:end,2:end)'); ylabel('Perturbation sequence','FontName','Times','FontSize',24); view(2);   % View azimuth = 0; elevation = 90 --> view from top box on shading interp;     hold on; contour3(PDDCOV(2:end,1), PDDCOV(1,2:end), PDDCOV (2:end,2:end)'+max(max(PDDCOV(2:end,2:end)))+1, 'k'); axis tight axes(fig_axes(max(size(fig_axes)-1))); plot(useData(:,1), useData(:,2),'k'); set(gca,'XTickLabel',[]); set(gca,'YTickLabel',[]); title([figName '2D perturbation domain maps'],'FontName','Times','FontSize',24); axis tight box on axes(fig_axes(max(size(fig_axes)-2))); plot(1:size(useData,2)-1,'k'); set(gca,'XTickLabel',[]); set(gca,'YTickLabel',[]); axis tight box on drawnow;   98 saveas(figHandles(2), [pathName '/' resultsName '2DPDD.fig']) disp('2D correlation spectroscopy analyses results in :  TDCOV, TDCOR, PDDCOV, and PDDCOR.') processNum  = processNum + 1; suffixes{processNum} = ''; Message = ['2D correlation spectroscopy analyses completed. Working on step ' num2str (processNum+1) '...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal'); % input(['Number ' num2str (processNum) ' finished']); % 12. Statistics (load and track 2 or more groups of spectra?) % close all figHandles = f_plotRaman(2, [1,1], figName);  figure(figHandles(1)); plot(useData(:,1), mean(useData (:,2:end)'),'k','linewidth',2);title([figName 'Mean of spectra'],'FontName','Times','FontSize',24);box on;axis tight; saveas(figHandles(1), [pathName '/' resultsName 'mean.fig']); figure(figHandles(2)); plot(useData(:,1), std(useData (:,2:end)'),'r','linewidth',2);title([figName 'Standard deviation of spectra'],'FontName','Times','FontSize',24);box on;axis tight; saveas(figHandles(2), [pathName '/' resultsName 'STD.fig']); disp('Basic statistic results in :  *******.') processNum  = processNum + 1; suffixes{processNum} = ''; Message = ['Basic statistics completed. Housekeeping...']; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal');  99 % input(['Number ' num2str (processNum) ' finished']); % 13. Bring QC figures to front. Clean up and save workspace  for n = 6:-1:1     figure(n) end figure(2)  save([pathName '/' resultsName '.mat']); disp(['Processing and analyses completed. Results save in: ' resultsName '.mat']) processNum  = processNum + 1; suffixes{processNum} = ''; Message = 'Processing and analyses completed.'; Title = 'Processing status'; h = msgbox(Message,Title,'warn','modal'); if ~(isempty(warningstrings))     dlgname = 'Alert';     for n = 1:max(size(warningstrings))         warningstrings{n} = [num2str (n) ' .' warningstrings{n}];     end     hw = warndlg(warningstrings, dlgname); end  clear('Diff1_dsp','Max_j','def','dlg_title','figHandles','formatDataLoaded','generatedvarname','iterations','loadBoxTitle','modifiedStr','n','normMethod','num_lines'); clear('optionsStr','origData','origStr','pauseTime','peakPosition','polyorder','processNum','prompt','responseMade','subsample','suffix','temp','useData','fig_axes');  100 clear('Message','Title','h','pauseTime','sumeigs','n','Latest_diff1','Diff1dspSG_','QCgeneratedvarname_first','QCgeneratedvarname_last','Raw_diff1','BGgeneratedvarname','BGname'); clear('specMaxPeaks','specStddevs','noisyData','BGnoisyData','tempgeneratedvarname','BGData_blf','BGuseData','Data_blf','bckground','indices','new_warningstrings','pixelwidth','suffix2','figName'); clear('resultsName','hw','dlgname');                   

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:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0167694/manifest

Comment

Related Items