Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

The role of Pannexin 2 in mitochondrial functions and cell death Le Vasseur, Maxence 2016

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

Item Metadata


24-ubc_2016_september_levasseur_maxence.pdf [ 21.05MB ]
JSON: 24-1.0305028.json
JSON-LD: 24-1.0305028-ld.json
RDF/XML (Pretty): 24-1.0305028-rdf.xml
RDF/JSON: 24-1.0305028-rdf.json
Turtle: 24-1.0305028-turtle.txt
N-Triples: 24-1.0305028-rdf-ntriples.txt
Original Record: 24-1.0305028-source.json
Full Text

Full Text

The role of Pannexin 2 in mitochondrial functionsand cell deathbyMaxence Le VasseurB.Sc Biomedical Sciences, Université de Montréal, 2003M.Sc Neurociences, Université de Montréal, 2006a thesis submitted in partial fulfillmentof the requirements for the degree ofDoctor of Philosophyinthe faculty of graduate and postdoctoralstudies(Neuroscience)The University of British Columbia(Vancouver)June 2016c© Maxence Le Vasseur, 2016AbstractPannexins constitute a small family of membrane channels homologous tothe invertebrate gap junction proteins. Three distinct isoforms, called Panx1,Panx2, and Panx3, are expressed alongside connexins in chordates. Un-like connexins, pannexins do not connect the cytoplasm of adjacent cellsbut function as unitary channels regulating the exchange of ions and smallmolecules between the cytoplasm and extracellular milieu. The biochem-ical properties and functions of Panx1 and Panx3 channels have been in-vestigated, but our understanding on Panx2 channels has progressed at amuch slower pace. The first objective of this thesis was to comprehen-sively map the expression and distribution of Panx2 protein in mammaliantissues. Prior to this work, Panx2 distribution had been studied exclu-sively by analyzing the expression of its transcript which was found tobe largely restricted to the central nervous system (CNS). In this thesis,Panx2 messenger RNA (mRNA) and protein levels were analyzed in differ-ent tissues and Panx2 transcriptional activity was found to poorly predictPanx2 protein abundance. Panx2 protein levels were lower in nervous tis-sues although transcriptional analysis showed disproportionately high Panx2mRNA levels in the CNS. Furthermore, endogenous Panx2 channels werefound to be sequestered within the endomembrane system of the cell. Thesecond objective focused on characterizing the subcellular localization andbiological function of Panx2 channels. Using subcellular fractionation, itwas determined that Panx2 co–fractionates with mitochondrial and endo-plasmic reticulum (ER) markers thereby suggesting that Panx2–containingcompartments can associate, at least transiently, with the ER and mito-iichondria. The analysis of Panx2 dynamics in living cells combined withimmunogold electron microscopy confirmed that Panx2 is not randomly dis-tributed within the cytoplasm but preferentially localizes at membrane con-tact sites called mitochondria–associated ER membranes (MAMs) whichphysically and functionally tether the ER to mitochondria. Finally, the bi-ological function of Panx2 was partially elucidated by showing that Panx2expression is modulated by the energetic requirements of the cell and canregulate apoptosis.iiiPrefaceA version of Chapter 2 has been published in a peer reviewed journal:M. Le Vasseur, J. Lelowski, J. F. Bechberger, W.C. Sin, and C. C.Naus. Pannexin 2 protein expression is not restricted to the CNS.Frontiers in Cellular Neuroscience, 8:392, 2014.Chapter 3 is a manuscript currently in preparation and will be submittedin spring 2016:M. Le Vasseur, K. Huang, V. C. Chen, W. A. Vogl and C. C. Naus.Pannexin 2 localizes at endoplasmic reticulum–mitochondria contactsites and regulates apoptosis. In preparation.Co–authors have contributed either by performing experiments, collect-ing samples/data, or by assisting with the interpretation of the data.• Jonathan Lelowski was a visiting student from the Université de Poitiersin France who worked under my supervision from May 2013 to August2013. Together we acquired some of the Western blots and immunoflu-orescent images shown in Chapter 2.• Our laboratory manager John F. Bechberger helped with the collectionand processing of the tissues and provided insights for the optimizationof Panx2 immunostaining in Chapter 2.• Kate Huang was student from the British Columbia Institute of Tech-nology (BCIT) who did an internship under my guidance from Mayiv2015 to August 2015. Together we studied changes in Panx2 expressionin cells grown in regular and glucose–free media in Chapter 3.• Dr. Vincent C. Chen was a postdoctoral fellow in our laboratory withwhom I explored various strategies to investigate Panx2 subcellularlocalization. He helped me with the design of the experiment and theinterpretation of the data in Chapter 3 and encouraged me to developtools to study Panx2 interaction with mitochondria.• None of the electron microscopy data shown in Chapter 3 would havebeen possible without the help of Dr. Wayne A. Vogl from the Uni-versity of British Columbia. Dr. Vogl taught me how to perform im-munogold electron microscopy and helped me with the interpretationof those data in Chapter 3.• Dr. Christian C. Naus was the supervisor author and helped carryboth projects to fruition by securing funding, providing advice andreviewing the manuscripts.I designed and performed each experiment presented in this thesis withthe aforementioned help or individually. I also entirely wrote both manuscripts.This work was covered by the Animal Care Committee from the UBC Re-search Ethics Board under the Animal Protocol A11-0169 ("Neuronal andastrocytic gap junctional involvement in the development in the developmentof Alzheimer disease").vTable of ContentsAbstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iiPreface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ivTable of Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . viList of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ixList of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xGlossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiiAcknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . .xviii1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Gap junctions . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1.1 Connexins . . . . . . . . . . . . . . . . . . . . . . . . . 21.1.2 Innexins . . . . . . . . . . . . . . . . . . . . . . . . . . 31.1.3 Pannexins . . . . . . . . . . . . . . . . . . . . . . . . . 31.1.4 Structural topology and assembly . . . . . . . . . . . . 51.1.5 Intercellular and non–junctional channels . . . . . . . 91.1.6 Glycosylation and trafficking . . . . . . . . . . . . . . 111.1.7 Conservation of pannexin genes . . . . . . . . . . . . . 121.1.8 Alternative splicing in pannexin genes . . . . . . . . . 121.1.9 Pannexin channel properties . . . . . . . . . . . . . . . 171.1.10 Pharmacological blockers . . . . . . . . . . . . . . . . 20vi1.1.11 Expression and distribution . . . . . . . . . . . . . . . 211.1.12 Physiological and pathological functions of pannexinchannels . . . . . . . . . . . . . . . . . . . . . . . . . . 231.1.13 Intracellular pannexin channels . . . . . . . . . . . . . 281.2 Mitochondria–associated ER membranes . . . . . . . . . . . . 291.2.1 The ERMES complex in yeast . . . . . . . . . . . . . 311.2.2 ER–mitochondria contact sites in chordates . . . . . . 321.2.3 Functional importance of MAMs . . . . . . . . . . . . 351.3 Motivation, objectives, and highlights . . . . . . . . . . . . . 452 Pannexin 2 protein expression is not restricted to the CNS 482.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 482.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . 502.2.1 Animal care . . . . . . . . . . . . . . . . . . . . . . . . 502.2.2 Antibodies . . . . . . . . . . . . . . . . . . . . . . . . 502.2.3 Cell culture . . . . . . . . . . . . . . . . . . . . . . . . 502.2.4 Epitope mapping . . . . . . . . . . . . . . . . . . . . . 512.2.5 RNA isolation and real–time quantitative PCR . . . . 522.2.6 Protein isolation and Western blotting . . . . . . . . . 532.2.7 Immunofluorescence . . . . . . . . . . . . . . . . . . . 542.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 552.3.1 Characterization of two novel monoclonal antibodiesspecific for Panx2 . . . . . . . . . . . . . . . . . . . . . 552.3.2 Panx2 has a ubiquitous protein expression profile . . . 562.3.3 Transcript levels do not predict Panx2 protein levels . 592.3.4 Panx2 protein is localized to cytoplasmic compartments 612.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 682.4.1 Absence of correlation between Panx2 mRNA and pro-tein levels . . . . . . . . . . . . . . . . . . . . . . . . . 682.4.2 Panx2: a cytoplasmic unusual suspect . . . . . . . . . 703 The gap junction protein pannexin 2 localizes at mito-chondria contact sites and sensitizes cells to apoptosis . . 73vii3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 733.2 Materials and methods . . . . . . . . . . . . . . . . . . . . . . 753.2.1 Animal care . . . . . . . . . . . . . . . . . . . . . . . . 753.2.2 Antibodies . . . . . . . . . . . . . . . . . . . . . . . . 753.2.3 Plasmid construction . . . . . . . . . . . . . . . . . . . 763.2.4 Cell culture . . . . . . . . . . . . . . . . . . . . . . . . 773.2.5 Live–cell imaging and Panx2 trajectory analysis . . . . 783.2.6 Immunofluorescence and co–localization analysis . . . 783.2.7 Pre–embedding immunoelectron microscopy . . . . . . 793.2.8 Subcellular fractionation and MAM isolation . . . . . 803.2.9 Western blotting . . . . . . . . . . . . . . . . . . . . . 813.2.10 Apoptotic DNA fragmentation analysis . . . . . . . . 823.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 823.3.1 Panx2 forms microdomains on ER membranes . . . . 823.3.2 Panx2 localizes at ER–mitochondria contact sites . . . 853.3.3 Panx2 expression is modulated by the mitochondrialrespiratory status . . . . . . . . . . . . . . . . . . . . . 923.3.4 Panx2 sensitizes cell to apoptosis . . . . . . . . . . . . 943.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 994 General discussion and concluding remarks . . . . . . . . . 1034.1 Beyond the transcript: What controls Panx2 protein levels? . 1044.2 What controls the gating of Panx2 channel and what goesthrough the pore? . . . . . . . . . . . . . . . . . . . . . . . . 1084.3 Functional implication of Panx2 at MAMs . . . . . . . . . . . 1114.4 The role of Panx2 in mitochondria–mediated cell death . . . 1154.5 Implication of Panx2 in human diseases . . . . . . . . . . . . 1164.5.1 Panx2 and Alzheimer’s disease . . . . . . . . . . . . . 1164.5.2 Panx2 and cerebral ischemia . . . . . . . . . . . . . . 118Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120viiiList of TablesTable 1.1 Connexin nomenclature . . . . . . . . . . . . . . . . . . . . 4ixList of FiguresFigure 1.1 Connexin and pannexin topology . . . . . . . . . . . . . . 8Figure 1.2 Conservation of pannexin amino acid sequences betweenspecies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13Figure 1.3 Ratio of substitution rates at non-synonymous and syn-onymous sites in pannexin paralogs . . . . . . . . . . . . . 14Figure 1.4 Multiple sequence alignment of Panx2 orthologs . . . . . 17Figure 1.5 Panx2 gene expression is down–regulated in human braintumors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26Figure 1.6 Calcium regulates the activity of rate–limiting mitochon-drial dehydrogenases . . . . . . . . . . . . . . . . . . . . . 41Figure 1.7 Calcium stimulates oxidative phosphorylation and increasesmitochondrial energy conversion potential . . . . . . . . . 43Figure 2.1 Specificity of two novel anti–Panx2 monoclonal antibodies 57Figure 2.2 Clone N121A/1 recognizes an epitope located within thelast 15 amino acids of Panx2 C–terminal . . . . . . . . . . 58Figure 2.3 The two anti–Panx2 antibodies detect endogenous Panx2 59Figure 2.4 Panx2 protein is ubiquitously expressed . . . . . . . . . . 60Figure 2.5 Panx2 transcript levels do not correlate with protein levels 61Figure 2.6 Expression of Panx1 and Panx2 proteins during brain de-velopment . . . . . . . . . . . . . . . . . . . . . . . . . . . 62Figure 2.7 Panx2 protein is expressed in the gastrointestinal tract . . 63Figure 2.8 Cuboidal cells from kidney tubule express Panx2 protein . 64Figure 2.9 Germ cells from mouse testis express Panx2 protein . . . 65xFigure 2.10 Panx2 protein is expressed in the photoreceptor inner seg-ment and outer plexiform layer of the mouse retina . . . . 66Figure 2.11 Panx2 protein localizes in the cytoplasm of CNS neurons 67Figure 3.1 Panx2 does not co–localize with common organelle markers 85Figure 3.2 Panx2 clusters in ER microdomains . . . . . . . . . . . . 87Figure 3.3 Panx2 does not associate with peroxisomes . . . . . . . . 87Figure 3.4 Panx2 puncta associate with mitochondria . . . . . . . . 88Figure 3.5 Panx2 localizes in ER microdomains associated with mi-tochondria in vivo . . . . . . . . . . . . . . . . . . . . . . 90Figure 3.6 Absence of cross–reactivity between subclass specific anti–mouse secondary antibodies . . . . . . . . . . . . . . . . . 91Figure 3.7 Panx2 is enriched in purified MAMs . . . . . . . . . . . . 92Figure 3.8 Panx2 localizes at ER–mitochondria contact sites . . . . . 94Figure 3.9 Mitochondrial respiration influences Panx2 expression . . 95Figure 3.10 Panx2 expression increases upon inhibition of mitochon-drial respiration . . . . . . . . . . . . . . . . . . . . . . . 96Figure 3.11 Panx2 sensitizes cells to apoptosis . . . . . . . . . . . . . 98Figure 4.1 Proposed model of Panx2 regulation at MAMs . . . . . . 113Figure 4.2 Cerebral ischemia stimulates Panx2 expression . . . . . . 119xiGlossaryACSL4 long-chain acyl–CoA synthetase 4AD Alzheimer’s diseaseADP adenosine diphosphateApaf–1 apoptosis—protease activating factor 1ASC apoptosis–associated speck–like protein containing a CARDATP adenosine triphosphateCBX carbenoxolonebp basepairBCA bicinchoninic acidBif1 Bax–interacting factor 1BSA bovine serum albuminCaMKIIα Ca2+/calmodulin–dependent protein kinase II αcDNA complementary DNACNS central nervous systemCOX cytochrome c oxidaseCx connexinxiiDa daltonDAPI 4’,6-diamidino-2-phenylindoleDEPC diethylpyrocarbonateDMEM Dulbecco’s Modified Eagle MediumDMSO dimethyl sulfoxideDNA deoxyribonucleic acidDrp1 dynamin–related protein 1EDTA ethylenediaminetetraacetic acidEGTA ethylene glycol tetraacetic acidEMC ER membrane protein complexER endoplasmic reticulumERMES ER–mitochondria encounter structureEST expressed sequence tagETC electron transport chainFBS fetal bovine serumGFP green fluorescent proteinGCL ganglion cell layerGRP75 75 kDa glucose–regulated proteinHRPO horseradish peroxidaseIF immunofluorescenceIgG immunoglobulinIL-1β interleukin-1βxiiiIL-18 interleukin-18IMM inner mitochondrial membraneINL inner nuclear layerInx innexinIPL inner plexiform layerIP3R inositol 1,4,5-trisphosphate–gated receptorIP3R1 type I inositol 1,4,5–trisphosphate–gated receptorIP3R3 type III inositol 1,4,5–trisphosphate–gated receptorIS photoreceptor inner segmentK+ potassium ionMAM mitochondria–associated ER membraneMCA middle cerebral arteryMCU mitochondrial Ca2+ uniporterMDM12 mitochondrial distribution and morphology protein 12MDM34 mitochondrial distribution and morphology protein 34MFN1 mitofusin–1MFN2 mitofusin–2miRNA micro RNAMRB mitochondrial resuspension buffermRNA messenger RNANADH nicotinamide adenine dinucleotideNCL nitrocellulosexivNMDAR N-methyl-d-aspartate receptorNLRP1 NACHT, LRR and PYD domains-containing protein 1OMM outer mitochondrial membraneOMP25 outer membrane protein 25ONL outer nuclear layerOPL outer plexiform layerORF open reading frameOS photoreceptor outer segmentOxPhos oxidative phosphorylationPanx pannexinPACS–2 phosphofurin acidic cluster sorting protein 2PAR–1 protease-activated receptor–1PB phosphate bufferPBS phosphate buffer salinePCR polymerase chain reactionPDC pyruvate dehydrogenase complexPDF probability density functionPDH pyruvate dehydrogenasePDK pyruvate dehydrogenase kinasePE phosphatidylethanolaminePFA paraformaldehydePS phosphatidylserinexvPSD phosphatidylserine decarboxylasePVDF polyvinylidene fluorideqPCR quantitative real-time PCRRIPA radioimmunoprecipitation assayRNA ribonucleic acidRNP ribonucleoprotein particleROS reactive oxygen speciesRyR ryanodine receptorsigma–1R sigma-1 receptorSDS sodium dodecyl sulfateSDS-PAGE sodium dodecyl sulfate polyacrylamide gel electrophoresisSH3 Src homology 3shRNA small hairpin RNAsiRNA silencing RNASSTR5 somatostatin receptor 5STS staurosporineTCA tricarboxylic acidTCE 2,2,2-trichloroethanolTEM transmission electron microscopeTNF–α tumor necrosis factor αtrkB tyrosine receptor kinase BUTR untranslated regionxviUV ultravioletVDAC1 voltage-dependent anion channel 1WB Western blotXIAP X-linked inhibitor of apoptosis proteinxviiAcknowledgmentsThis has most certainly been a PhD of great and stimulating challenges.First of all, I would like to thank Dr. Christian Naus for giving me the op-portunity to join his group. Together, we definitely navigated in unchartedwaters when I stubbornly decided to tackle the intriguing but complicatedPanx2 problem. The sea has been rough at times, but you never let medrown and guided me to the harbor. I also want to express my warmestappreciation to Dr. Wayne Vogl without whom an important part of thisproject would have been possible. To explore the infinitely small guided byyour infinite wisdom and charisma has been one of the most beautiful para-doxes and most memorable experiences of my PhD. I would also like to thankDr. Vincent Chen for being such an incredible mentor. I truly appreciatedyour admirable wit and enthusiasm. Your passion was contagious and work-ing on proteomics projects with you has been a remarkable and fortunateexperience. Joining the Naus lab is also joining the Bechberger family and Iwould like to thank Lynne and John for being such great ’lab parents’. Youboth provided a nurturing (and nutritious!) environment and I appreciatedall the social events you organized and the gestures you made to keep ushappy. I wouldn’t have had nearly the same pleasure working without you.A special thank to Dr. Wun Chey Sin, your stern but righteous criticismsforced me to better myself. Also, your dedication to fitness often motivatedme to leave the bench and trade my lab coat for running shoes which has al-ways turned out to be a great thing to do. To Dr. Moises Freitas–Andrade,what a pleasure it has been to share the countless obstacles of the pannexinendeavor with you! I truly enjoyed our discussions about life, Montréal,xviiiscience and look forward to visiting your own lab one day. Also thank youfor introducing me to the Hendrick’s Gin and tonic with cucumber. As Iam typing this I am actually sipping... water... and it doesn’t taste nearlyas good. I also want to thank all the graduate students from the Naus lab(Charles Lai, Cima Cina, Mike Kozoriz, Annie Aftab, and Steve Bond) fortheir friendship and for creating a stimulating and cooperative environment.To Pascal St-Pierre, Annie Aftab, Jason Barker and John Abramyan, thankyou very much for providing me with the moral support and friendship Ineeded when everything else was drifting. Several undergraduate or visitingstudents have worked in our lab and helped in a countless number of waysand I am very grateful to all of you. More specifically, I would like to thankKate Huang and Jonathan Lelowski who I had the privilege to supervisepersonally. It has been an honor to work with both of you. I also wantto acknowledge the financial support of the Natural Sciences and Engineer-ing Research Council of Canada (NSERC). I would also like to thank Dr.Vanessa Auld, Dr. Christopher Loewen, and Dr. Timothy Murphy for theirconstructive criticisms as members of my advisory committee. Last but notleast, I would like to thank my whole family for their unwavering and lovingsupport.xixChapter 1Introduction1.1 Gap junctionsIn 1959, Edwin Furshpan and David Potter published a seminal paper de-scribing some peculiarities of the crayfish giant motoneuron synapse. Theauthors reported a type of electrical transmission at the synapse that couldnot be explained by the conventional unidirectional chemical transmittermodel prevailing at the time [96]. They recorded post–synaptic poten-tials with strikingly shorter synaptic delays than those observed at chemicalsynapses and demonstrated that hyperpolarizing currents applied to a post–synaptic fiber propagated antidromically to the pre–synaptic fiber [96]. Inorder to explain their observations, Furshpan and Potter proposed a modelin which synapses had "low synaptic resistance to current flow" and as-sumed "that there was no separation between the component membranes"forming the synapse [96]. Furshpan and Potter assumptions were supportedby electron microscopy studies showing regions of very close juxtapositionbetween pre– and post–synaptic membranes [249, 250] thereby suggestingthat the cytoplasm of adjacent cells could be physically coupled. This hy-pothesis was later supported by dye coupling assays showing that small(< 1000Da) hydrophilic membrane–impermeable dyes injected in one cellcould freely diffuse to neighboring cells [140, 141, 184]. This observationfurther reinforced the idea that specialized membrane domains allowed cy-1toplasmic continuity between cells. Overall, these studies eventually led tothe discovery of cellular structures, later called gap junctions, which phys-ically and functionally link the cytoplasm of adjacent cells. Gap junctionsare formed by the oligomerization of monomeric proteins into large mem-brane channels. Channels from adjacent cells assemble together into a largedouble–membrane spanning molecular complex that facilitates intercellularcommunication by physically joining the cytoplasm of each cell. Two fam-ilies of membrane channels independently acquired the ability to form gapjunctions in Metazoans; the connexins (Cxs) which are restricted exclusivelyto chordates and the innexins (Inxs) which are present in all eumetazoans[215, 230, 267, 328] except the echinoderms in which intercellular couplinghas not been reported [263, 267].1.1.1 ConnexinsAlthough the electrophysiological studies that originally reported intercellu-lar communication were performed in invertebrates, the molecular composi-tion of gap junctions was first identified in vertebrates. The first connexincomplementary DNA (cDNA) was cloned in 1986 from liver samples [220];nearly three decades after the original observation of electrotonic couplingby Furshpan and Potter. Several additional connexin isoforms have sincebeen identified, largely thanks to advancements in new technologies such ashigh-throughput sequencing. Although new connexin genes are likely to beidentified as more genomes are being sequenced and curated every year, thecompleteness and accuracy of the mouse and human genomes suggest thatall connexin genes have likely been identified in those species. A total of 20and 21 genes constitute the large multi-member connexin family in mouseand human respectively [272].NomenclatureTwo distinct nomenclature systems has been adopted to identify connexinproteins. The most commonly used nomenclature is based on the molecularweight of individual monomers (e.g. Cx26 for the 26 kDa isoform) and is the2system that will be adopted throughout this thesis. The other system dividesconnexin isoforms into subgroups based on their evolutionary relationshipand numbers elements of each subgroup chronologically according to theirorder of discovery (e.g. Gjb1 for Cx32, and Gjb2 for Cx26). A side-by-sidecomparison of these two nomenclatures is presented in Table 1.1 for humanand mouse connexin proteins.1.1.2 InnexinsDespite early observations supporting functional intercellular coupling ininvertebrates, numerous attempts to clone connexin genes in invertebrateswere unsuccessful. No connexin homologues were identified in invertebrates,even in model organisms with fully sequenced genomes such as D. melano-gaster and C. elegans. It was eventually correctly hypothesized that inver-tebrate gap junctions were assembled from proteins unrelated to connexins.The first invertebrate gap junctions were cloned and molecularly identifiedin the early 1990s [157, 279, 318]. The protein family was initially namedOPUS in reference to the first genes identified (ogre, passover, unc-7 andshaking-B) [17]. However, this terminology created confusion since passoverand shaking-B were later shown to be allelic variations of a unique gene[60] and the name opus already referred to a transposable element in thefruit fly[146]. It was therefore decided that a different terminology shouldbe adopted and the name innexin, which stands for invertebrate analog ofconnexins, was selected because it better reflected the function of the proteinfamily [231].1.1.3 PannexinsAs previously mentioned, connexins are restricted to the chordate lineageand are not found in the genome of invertebrates. However, it appears thatinnexins are not exclusive to the invertebrate phyla but are also expressedalongside connexins in different taxonomic groups including chordates. In2000, using degenerate primers, Panchin and colleagues were able to clonesequences homologous to innexins from mollusk and flatworm cDNA [216].3Table 1.1. Two different nomenclature styles exist for naming connexins. Con-nexins can be classified according to their evolutionary relationships or accordingto their approximate molecular weight. A side-by-side comparison of these twonomenclatures is presented for human and mouse connexins.Human MouseGJA1 CX43 GJA1 CX43GJA3 CX46 GJA3 CX46GJA4 CX37 GJA4 CX37GJA5 CX40 GJA5 CX40GJA6 CX33GJA8 CX50 GJA8 CX50GJA9 CX59, CX58GJA10 CX62 GJA10 CX57GJB1 CX32 GJB1 CX32GJB2 CX26 GJB2 CX26GJB3 CX31 GJB3 CX31GJB4 CX30.3 GJB4 CX30.3GJB5 CX31.1 GJB5 CX31.1GJB6 CX30 GJB6 CX30GJB7 CX25GJC1 CX45 GJC1 CX45GJC2 CX47, CX46.6 GJC2 CX47GJC3 CX30.2 GJC3 CX29GJD2 CX36 GJD2 CX36GJD3 CX31.9 GJD3 CX30.2GJD4 CX40.1 GJD4 CX39GJE1 CX23 GJE1 CX234These sequences were then subjected to BLAST for homology matches andunambiguously aligned with similar sequences from species representativeof the Platyhelminthes, Nematoda, Arthropoda, Mollusca and even theChordata phyla thereby confirming that this protein family had radiatedthroughout the Metazoa [216]. A few years later, genes encoding proteinshomologous to the invertebrate innexins were successfully cloned from rat[42] and human [14]. Because this protein family was no longer exclusive tothe invertebrates but almost ubiquitous to the animal kingdom, the terminnexin appeared inappropriate and it was suggested that the name pan-nexin (Panx) should be used instead [216]. The term ’pannexin’ was deemedmore representative of the reality because the Latin prefix pan means "all" or"throughout"; a suitable prefix for a family of proteins expressed throughoutthe Metazoa.There are only three pannexin paralogs in chordates which have beendesignated Panx1, Panx2 and Panx3. These genes resulted from duplica-tion events that predated the radiation of all higher vertebrates and areconsequently orthologous across species (i.e. cat Panx1 corresponds to hu-man Panx1) [328]. Interestingly, there are at least four pannexin genespresent in the teleost lineage as a result of an ancestral whole genome dupli-cation [36, 160]. Two panx1 genes, named panx1a and panx1b, were retainedfollowing the third major whole genome duplication event which occurredin the teleost lineage. Both panx1 genes were shown to produce proteinsthat diverged functionally in term of their expression pattern, subcellulardistribution and physiological properties such as gating and conductance[36].1.1.4 Structural topology and assemblyDespite the absence of evolution relationship and sequence homology, con-nexin and pannexin proteins share a similar structural topology and molec-ular organization acquired through functional convergence. Both proteinfamilies have four hydrophobic transmembrane domains linked by two ex-tracellular loops and one cytoplasmic loop (Figure 1.1A,B). Although it5has been shown that Cx26 and Cx43 could insert into membranes post-translationally in cell-free transcription and translation system [4, 5, 337],connexins are generally co-translationally inserted in the endoplasmic reticu-lum (ER) and transit through the Golgi before being trafficked to the plasmamembrane. Connexins and pannexins do not have N–terminal signal pep-tides but use internal hydrophobic transmembrane segments for binding thesignal recognition particle and consequently their N– and C-termini facethe cytoplasm. During their transit to the plasma membrane, six connexinmonomers oligomerize into a channel called ’connexon’ (Figure 1.1). Vari-ous studies on membrane proteins support the idea that subunits of integralmembrane proteins are oligomerized into multisubunit complexes in the ER[102, 128, 252] and the assembly of most connexin monomers into multimericchannels seem to adhere to this principle. However, it has been suggestedthat the assembly of Cx43 and Cx32 monomers into multisubunit channelsoccurs after their exit from the ER, most probably in the trans-Golgi network[5, 204]. Once inserted in the plasma membrane, two connexons provided byadjacent cells dock together to form a mature double–membrane spanningchannel that connects the cytoplasm of neighboring cells (Figure 1.1B).Many cell types express more than one connexin isoform and numerousstudies investigating the oligomerization behavior of connexin monomersdemonstrated that this heterogeneity greatly increases the diversity andcomplexity of gap junction channels. Monomers from a single connexinisoform not only assemble together into homomeric connexons but can alsooligomerize with connexins from different isoforms and form heteromericconnexons. For example, it has been shown that Cx43 can oligomerize withCx37 [40], Cx40 [56, 120, 302], Cx56 [25], and Cx46 [66] but not with Cx32[66, 90]. Similarly, Cx46 can oligomerize with Cx50 [135] and Cx32 canoligomerize with Cx26 [181, 280]. Interestingly, HeLa cells, alveolar epithe-lial cells, and osteoblastic cells all express both Cx43 and Cx46 isoforms butCx43/Cx46 heteromeric connexons have only been detected in HeLa andalveolar epithelial cells, but not in osteoblastic cells [66]. Therefore, theability to form heteromeric connexons or to differentially sort each connexinisoform can be cell type specific. The molecular diversity of gap junctions is6further increased by the composition of apposing connexons and the signalsthat regulate docking compatibility. Apposed connexons from neighboringcells can either have the same (homotypic) or different molecular composi-tion (heterotypic) [155].Soon after their discovery, it was predicted that pannexins could alsooligomerize into multimeric complexes which were named ’pannexons’ inanalogy to connexon hemichannels formed by connexins (Figure 1.1D). Ele-gant electron microscopy studies on membranes isolated from cells express-ing Panx1 or Panx2 identified channel-like structures supporting the exis-tence of pannexons [7]. The existence of oligomeric pannexons was also de-termined using a chemical cross–linking technique similar to the approachused to study the multimeric arrangement of connexin proteins [308]. Us-ing this approach, Boassa and colleagues were the first to demonstrate thatcross–linked Panx1 proteins resolved by sodium dodecyl sulfate polyacry-lamide gel electrophoresis (SDS-PAGE) produce a banding pattern consis-tent with a hexameric state [29]. Monomeric Panx1 had an expected sizeof ∼ 49kDa while the largest complex was ∼ 290kDa which correspond to ahexameric arrangement. Rather surprisingly, using a similar approach thesame group later showed that cross–linked Panx2 migrated well above theexpected size of a hexamer (∼ 420kDa) on a SDS-PAGE gel [7]. In a subse-quent series of experiment using Panx2 protein, the authors estimated thatPanx2 most likely adopt an octameric arrangement where 8 rather than 6Panx2 monomers oligomerize into a channel [7]. However, it should be notedthat because of the large size of Panx2 monomers (∼ 70kDa) the exact ar-rangement of the Panx2 channel could not be resolved on a regular proteingel. Therefore, the octameric arrangement was predicted by using a trun-cated form of Panx2 (∼ 41kDa) which lacked most of its C–terminal tail[7]. Therefore, although the octameric arrangement seems to be the mostplausible scenario, it is still unclear whether endogenous Panx2 also formsoctamers. The oligomerization arrangement of Panx3 channels yet remainsto be determined but is presumed to be similar to the Panx1 multimericstate. Panx1 and Panx3 channels can form functional Ca2+ channels in theER [133, 306] and it is therefore reasonable to assume that the oligomeriza-7tion of pannexin monomers into multimeric subunit channels occurs in theER.Figure 1.1. Connexin and pannexin topology. Connexin and pannexin sharea common structural topology despite the absence of sequence homology. Bothare transmembrane proteins with four transmembrane domains, two extracellularloops, one cytoplasmic loop, and cytoplasmic N– and C–terminal domains (A,C).Connexin and pannexin monomers both oligomerize to form a functional ’connexon’and ’pannexon’ channel respectively. However, only connexin channels can assembleinto a gap junction that allows intercellular communication. Pannexin are glyco-sylated and large glycan residues prevent the docking of pannexin channels fromadjacent cells (B,D). The amino acid sequences in A and C represent human Cx43and Panx2 respectively. (PTMs = post-translational modifications)Several studies have reported that pannexins can form heteromeric ar-rangements. Immunoprecipitation experiments have shown that Panx1 can8interact with Panx2 or Panx3 whereas Panx2 and Panx3 cannot interact[7, 43, 225]. Interestingly, the co–expression of Panx1 and Panx2 proteinsincreased Panx2 localization at the cell surface [225]. However, it shouldbe noted that Panx1/Panx2 heteromeric channels are highly unstable [7]and exhibit reduced current amplitudes, modified voltage gating kinetics,and reduced dye uptake relative to homomeric Panx1 channels [42, 43, 225].Therefore, although co–expression of Panx1 and Panx2 have been reportedin several tissues, the occurrence of functional Panx1/Panx2 channels isconsidered unlikely. In fact, as we will see in Chapter 2, cells co–expressingPanx1 and Panx2 proteins have the ability to sort both channels to differentcellular compartments.1.1.5 Intercellular and non–junctional channelsBecause the predicted topology of pannexins is very similar to that of con-nexins, it was originally assumed that pannexons from neighboring cellscould also dock to each other and form double–membrane spanning intercel-lular channels. This idea was supported by a few groups reporting functionalcell–cell coupling in cells expressing Panx1 or Panx3 channels [42, 133, 164,256, 306]. Roberto Bruzzone first reported functional pannexin–mediatedintercellular coupling by recording large transjunctional currents in pairedXenopus oocytes expressing Panx1 homomeric or Panx1/Panx2 heteromericchannels but not in oocytes expressing homomeric Panx2 or Panx3 channels[42]. Interestingly, in contrast to connexins, Panx1 and Panx1/Panx2 inter-cellular channels exhibited a very weak sensitivity to transjunctional voltage(Vj) [42]. Another group recently replicated these results and demonstratedthe existence of transjunctional currents in HeLa cells over–expressing Panx1or Panx3 channels [256]. The intercellular movement of small membrane–impermeable dyes has also been used to study pannexin gap junctional cou-pling. For example, the over–expression of Panx1 in rat C6 glioma cellshas been shown to increase gap junctional coupling and to allow the trans-fer of sulforhodamine 101 dye between cells [164]. Similarly, HeLa cellsover–expressing Panx1 or Panx3 have higher intercellular permeability to9the anonic dye AlexaFluor 350 [256]. The over–expression of Panx1 inLNCaP human prostate cancer cells or Panx3 in the osteoprogenitor cellline C2C12 has been shown to facilitate intercellular Ca2+ transfer betweenneighboring cells [133, 306]. However, all observations supporting the con-cept of pannexin–mediated cell–cell coupling are limited to heterologous orover-expression systems and the idea that pannexins form intercellular gapjunction–like channels has been disputed [276]. A major point of contentionarose from the fact that unlike connexins all three pannexins are glycosylatedat their extracellular loops [225] with carbohydrate moieties that stericallyhinder the docking of pannexons from adjacent cells [29]. The glycosylationpattern of pannexin channels seems to be cell type specific. Sahu and col-leagues have shown that the glycosylation of Panx1 is cell type dependentand that cells with greater glycosylation fail to form pannexin intercellu-lar junctions [256]. Enzymatic deglycosylation has been shown to enhancejunctional conductance in oocytes over–expressing Panx1 protein [30] butthis treatment is highly artificial and could not be replicated by anothergroup [256]. Therefore, the ability of endogenous pannexins to form cell–cell junctional channels under physiological conditions remains controversial.A careful examination of the subcellular localization of pannexin proteinsdoes not support the idea that pannexins form intercellular coupling chan-nels [276]. For example, several studies have shown that unlike connexins,Panx1 channels fail to form clusters at areas of cell–cell contacts and oftenlocalize on unapposed membrane in polarized cells [29, 239]. Furthermore,as will we discussed in greater details in Chapter 2 and Chapter 3, endoge-nous Panx2 is not expressed at the plasma membrane but remains restrictedto the endomembrane system therefore making Panx2–mediated intercel-lular coupling virtually impossible. Consequently, it is generally assumedthat endogenous pannexons from neighboring cells do not significantly con-tribute to direct cell–cell gap junctional communication but rather primarilyform non–junctional membrane channels controlling the exchange of ions andsmall molecules between the cytoplasm and extracellular space.101.1.6 Glycosylation and traffickingSeveral groups have investigated the trafficking and life cycle of pannexinproteins. Like many other integral membrane proteins, pannexins are co–translationally inserted into the ER [26, 30] where they get N-glycosylatedinto high mannose species by the attachment of oligosaccharides to a ni-trogen atom of an asparagine residue [29, 224, 225]. The glycan structure,consisting of 14 sugar residues (Glc3Man9GlcNAc2), is transferred en blocand covalently linked to residue N254 located on the second extracellularloop of rat or mouse Panx1, residue N86 located on the first extracellularloop of mouse Panx2 and residue N71 located on the first extracellular loopof mouse Panx3 [29, 225]. The oligosaccharides are then trimmed by glycosi-dases in the ER before being trafficked to the Golgi. In the Golgi, the glyco-proteins are further processed by the attachment of additional sugar residueswhich forms complex N–glycoproteins. The addition of oligosaccharide moi-eties induces a noticeable gain in molecular weight and the un–glycosylated(GLY0), high mannose (GLY1) and complex N-glycoprotein (GLY2) formsof Panx proteins can be successfully resolved by SDS-PAGE [29, 224, 225].Panx1 and Panx3 proteins are trafficked to the Golgi via Sar1–dependentCOPII vesicles [26] where they are processed into complex N–glycoproteinsbefore being secreted to the plasma membrane [29, 224, 225]. Interest-ingly however, Panx2 is N–glycosylated into a high mannose form but is notprocessed into complex N–glycoproteins [225] and is not trafficked to theplasma membrane [165, 225] thereby suggesting that Panx2 does not transitthrough the Golgi apparatus but most likely remains within the ER. In-terestingly, palmitoylation has been proposed to influence Panx2 dynamicsand trafficking in neural progenitors. It has been suggested that palmitoyla-tion sequesters Panx2 in Golgi and ER membranes in hippocampal neuronprogenitors while mature neurons express a depalmitoylated form of Panx2that can be localized to the plasma membrane [285].111.1.7 Conservation of pannexin genesA comparison of the percentage of amino acid conversation between pan-nexin orthologs indicates that Panx2 is the most conserved protein amongthe three vertebrate paralogs. The percentage of identical residues betweenaligned sequences is substantially higher for Panx2 than Panx1 and Panx3paralogs, especially in evolutionary distant species (Figure 1.2). This highdegree of conservation indicates that Panx2 is more constrained and sub-jected to higher selective pressure than other Panx genes. Other statisticalmethods have been developed to investigate molecular evolution and selec-tion pressure acting on protein-coding genes [327]. One of such method isthe calculation of the dN/dS ratio which measures the substitution ratesat non-synonymous (non-silent) and synonymous (silent) sites. If an aminoacid change is neutral and has no impact on molecular evolution, the rate ofsubstitutions at silent and non-silent sites should be similar and the dN/dSratio should approach 1. However, if the amino acid change is deleterious,purifying selection will constrain substitutions at non-silent sites and thedN/dS ratio will be much lower than 1. A comparison of the dN/dS ratiosbetween pannexin paralogs indicates that dN/dS values are lower for Panx2therefore indicating that Panx2 is subjected to higher evolutionary pressurethan other pannexin genes (Figure 1.3). These results indicate that Panx2is subjected to a strong purifying selection that maintains Panx2 structureand function by preventing deleterious non-silent substitutions. This strongconservation suggests that Panx2 serves a fundamental role in living sys-tems.1.1.8 Alternative splicing in pannexin genesPanx1 and Panx2 transcripts have been shown to undergo alternative splic-ing thereby increasing the number of putative protein isoforms that can beencoded by a single gene. Two shorter Panx1 isoforms (Panx1c and Panx1d)have been identified in the rat pituitary gland and epididymis in addition tofull-length Panx1 (Panx1a) [177, 301]. The Panx1c isoform is characterizedby the total lost of exon 3 which corresponds to the second transmembrane12Figure 1.2. Conservation of pannexin amino acid sequences between species. Thepercentage of identical amino acids was calculated from multiple pairwise globalalignments between human pannexins and their corresponding orthologs. Panx2is the most conserved of the three pannexins with sequence identity values thatremain high even in evolutionary distant species. In comparison, Panx1 and Panx3identity values drop substantially in phylogenetically distant species. For example,human (H. sapiens) and zebrafish (D. rerio) Panx2 proteins are 71.4% identicalover their aligned amino acid sequences while Panx1 and Panx3 respectfully share57.6% and 52.4% identity only between human and zebrafish. Multiple pairwiseglobal alignments were performed using the R package Biostrings with a gap openpenalty of 10, a gap extension penalty of 0.5 and the Blosum62 cost matrix. The se-quence identity represents the ratio obtained by dividing the number of exact matchresidues over the length of the entire aligned sequence (expressed in percentage).13Figure 1.3. Ratio of substitution rates at non-synonymous and synonymous sitesin pannexin paralogs. The dN/dS ratio compares the substitution rates at non-synonymous and synonymous sites. These ratios are much lower for Panx2 therebyindicating that the selective pressure against deleterious non-silent substitutions isstronger for Panx2 than other pannexin paralogs. This purifying selection helpedmaintain Panx2 structure and function. The dN/dS ratios were retrieved fromEnsembl Release 82 (September 2015).14domain and a portion of the cytoplasmic loop while Panx1d has a partialdeletion of exon 4 producing a truncated C–terminal tail [177, 301]. Inter-estingly, over-expression studies show that Panx1a is located at the plasmamembrane while the shorter Panx1 isoforms are confined to the cytoplasm[177]. This difference in subcellular localization suggests that the differentPanx1 splice variants are not redundant but serve different physiologicalfunctions. Two Panx1 splice variants, hPanx1a and hPanx1b, have alsobeen cloned in human [14, 187]. The alternative splicing of exon 5 canlead to the addition of a four–amino acid insert (GMNI) at position 400 inhPanx1b [187]. Although the hPanx1b variant seems to be preferentiallyexpressed in endogenous systems, both proteins showed similar localizationand functional properties when over–expressed in cells [187]. Therefore,the biological relevance of this alternative splicing event remains unknown.Three different Panx2 transcripts identified as Panx2a, Panx2b and Panx2cwere cloned from the zebrafish retina and are believed to encode proteins of580, 604, and 652 amino acids respectively [338]. However, as of EnsemblRelease 82 (September 2015), the genome database reports only two humanPanx2 splice variants encoding proteins of 643 and 677 amino acids and twomouse Panx2 splice variants encoding protein of 651 and 677 amino acids.The shorter and longer splice variants differ exclusively in their C–terminalsequences. It is uncertain whether the occurrence of an additional Panx2splice variant is unique to fishes or constitute a feature more widely spreadacross different taxa. Furthermore, the expression and distribution of eachPanx2 protein isoform has not been thoroughly investigated and it is cur-rently unknown whether both isoforms are expressed and whether they havedifferent functional characteristics.It should also be noted that the 5’ end of the Panx2 open readingframe (ORF) contains two in-frame ATG codons that are located only 30 bpapart and which are both identified as putative translation initiation sites. Ithas been suggested that the second ATG codon is the most probable trans-lation initiation site because it is located in a more optimal Kozak contextthan the first ATG codon [14]. I retrieved the human Panx2 cDNA sequence(ENST00000395842) flanked with 100 basepair (bp) of 5’ untranslated re-15gion (UTR) DNA sequence from the Ensembl Release 82 (September 2015)and interrogated this sequence using the translation initiation site predictionsoftware ATGpr_sim [211]. This approach confirmed that the second ATGcodon has the highest reliability score (0.69) while the first ATG codon didnot provide a significant reliability score. Another observation supports theidea that the second ATG codon is the most probable translation initiationsite. A multiple pairwise alignment of Panx2 amino acid sequences from over40 different species retrieved from the Ensembl Release 82 (September 2015)indicates that the 10 amino acid sequence encoded by the 30 bp upstreamof the second ATG codon is poorly conserved across taxa and even oftenmissing in many species. In contrast, the amino acid sequence starting atthe second ATG codon is extremely well conserved Figure 1.4. Therefore, al-though we cannot exclude the possibility that alternative initiation increasesPanx2 isoform diversity in certain conditions, it is probable that the first 30bp of Panx2 transcript obtained from most databases do not get translated.Even if this short mRNA sequence gets translated, it is unclear whether theadditional 10 amino acids in Panx2 N–terminal have any functional signif-icance. While we and others have used Panx2 constructs starting at thesecond ATG codon in over–expressing systems [42, 165, 170], several groupshave used constructs starting at the first ATG codon [7, 31, 225, 322]. In-terestingly, while Bruzzone and colleagues did not observe Panx2-mediatedcurrents in oocytes expressing the shorter Panx2 construct starting at thesecond ATG codon [42], another group has detected Panx2–mediated cur-rents in oocytes expressing the longer construct [7]. It is possible that bothconstructs differ in term of their abilities to be trafficked at the plasma mem-brane but Ambrosi and colleagues briefly mentioned that they did not findany differences in structure or trafficking between the longer and shorterPanx2 constructs [7]. However, this represents an anecdotal remark andsupporting evidence on the expression, trafficking, and functionality of bothconstructs are needed to substantiate this claim.16Figure 1.4. Multiple sequence alignment of Panx2 orthologs. Several Panx2 or-thologs have two putative translation initiation codons resulting in two N–terminalmethionine residues separated by 10 amino acids. The first 10 putative amino acidsare poorly conserved across species while the N–terminal portion starting at thesecond methionine residue is well conserved. Multiple sequence alignment was per-formed using the R package msa with the ClustalW algorithm, a gap open penaltyof 10, a gap extension penalty of 0.5, and the Blosum substitution matrix.1.1.9 Pannexin channel propertiesIn 2003, Roberto Bruzzone demonstrated that pannexons form functionalchannels at non–junctional plasma membrane by recording robust macro-scopic membrane currents in Xenopus oocytes expressing Panx1 proteins[42]. This pioneer work was later corroborated by other groups studyingthe electrophysiological properties of Panx1 channels [13, 36, 182]. Panx1channels are characterized by very large unitary conductances (∼ 275 pS to>500 pS) [13, 36, 182]. Interestingly, the Panx1b ohnolog found in teleostsis characterized by a lower unitary conductance (∼ 123 pS) indicative of its17functional divergence from Panx1a channel [36].Panx1 channels have no less than 4 subconductance states (5%, 25%,30% and 90% of maximal conductance) in addition to the fully open state[13]. Although Panx1 channels are normally fully closed at resting mem-brane potentials, they can nonetheless open under physiological conditions.Several groups have shown that the opening of Panx1 channels is pro-moted by depolarizing conditions [13, 36, 182]. Panx1 channels are alsomechanosensitive and can open at resting potential upon mild mechanicalstress causing membrane deformation [13, 182] or following osmotic stress inairway epithelial cells [239]. The open state of Panx1 channels is increasedfollowing an elevation in cytoplasmic Ca2+ [183], or following NMDA recep-tor activation in hippocampal neurons [296]. Activation of the purinergic re-ceptors P2Y1, P2Y2 or P2X7 by extracellular adenosine triphosphate (ATP)also induces large Panx1-mediated currents, possibly through an increase incytoplasmic Ca2+ concentration [130, 183].Several groups have shown that Panx1 channels are highly permeable toATP and reported Panx1–mediated ATP release in oocytes [13] and in a va-riety of cell types expressing endogenous Panx1 such as erythrocytes [183],endothelial cells [108], astrocytes [130], airway epithelial cells [239], stressedcardiac myocytes [81], and apoptotic Jurkat cells [50]. The permeabilityof Panx1 channel to ATP was further confirmed by studies that showedthat ATP release was deficient in astrocytes and erythrocytes from Panx1knockout mice [236, 284]. The idea that Panx1 forms a large ATP releasechannel has however been challenged. A few groups have suggested thatPanx1 channels have a relatively small unitary conductance (∼ 70 pS) andare preferentially permeable to anions smaller than 250Da with high chloridepermeability but no ATP permeability [188, 251]. A recent study addressedthis discrepancy in Panx1 channel conductance and showed that Panx1 chan-nels have two open channel conformations with different conductances andpermeabilities [317]. Upon depolarization by elevated extracellular potas-sium ions, Panx1 formed high–conductance channels permeable to ATP asreported by most groups. In contrast, when the cells were depolarized inthe absence of potassium ions, Panx1 channels had a low conductance state18(∼ 50 pS) and were impermeable to ATP [317].The biophysical properties of the other pannexin paralogs have beenmuch less documented. Surprisingly, the initial study by Bruzzone whichreported robust Panx1 membrane currents failed to record Panx2– or Panx3–mediated currents in oocytes [42]. As we will explore in greater details in thefollowing chapters, Panx2 is largely restricted to the endomembrane systemand thus the intracellular localization of Panx2 channels might have pre-cluded the recording of Panx2–mediated current at the plasma membrane.However, other groups have subsequently shown that Panx2 and Panx3 formfunctional membrane channels. In 2010, Cinzia Ambrosi used a liposome-based permeability assay to study Panx1 and Panx2 channels [7] therebyovercoming the problem induced by the absence of Panx2 channels at theplasma membrane. Purified Panx1 or Panx2 pannexons were reconstitutedin liposomal vesicles loaded with cytochrome c. Ascorbate was subsequentlyadded and ascorbate permeability was measured using cytochrome c reduc-tion as a readout. A time–dependent reduction of liposomal cytochrome cwas observed in both Panx1 and Panx2 proteoliposomes thereby confirm-ing that Panx1 and Panx2 homomeric channels are both functional [7]. Inthe same study, Ambrosi and colleagues also detected functional pannexinchannels in oocytes over–expressing Panx2. They showed that homomericPanx2 channels opened and generated a current with a maximal amplitudeof ∼ 2.1 µA when a voltage ramp from -100 mV to 100 mV was applied tooocytes [7]. The amplitude of Panx2-mediated current was very similar toPanx1 current generated under the same conditions (∼ 2.0 µA) [7]. Anothergroup independently showed that Panx2–expressing oocytes displayed anoutwardly rectifying membrane conductance at positive membrane poten-tials [117]. Interestingly, currents measured at a holding potential of +60mVwere much smaller for oocytes expressing Panx2 than Panx1 (∼ 0.37 µA vs∼ 1.13 µA) [117]. However, this difference might have originated from lowerexpression of Panx2 at the plasma membrane as the study did not recordsingle channel conductance. Therefore, it is still unclear whether Panx2channels have a lower conductance than Panx1 channels. Small membrane—impermeable dyes have also been used by several groups to study pannexin19channels. Cells over–expressing Panx1, Panx2 or Panx3 proteins but notwild-type cells were shown to be permeable to the sulforhodamine B dyethereby indicating that all pannexin proteins can assemble into functionalmembrane channels at the cell surface upon over–expression [48, 224, 225].1.1.10 Pharmacological blockersSeveral pharmacological compounds have been shown to affect the conduc-tance of pannexin channels. However, it should be noted that high affinityblockers specifically inhibiting pannexin channels have not been found yetand most drugs affect a considerable array of channels. However, severalblockers have been successful at discriminating conductances from connexinhemichannels and pannexins based on different affinity for both channel fam-ilies. One of the most commonly used chemicals is carbenoxolone (CBX), aderivative of glycyrrhetinic acid which is found in the licorice plant. CBXhas been widely used as a reasonably effective blocker of connexin channelsover the past two decades [67] and was later shown be a reversible and evenmore potent blocker of Panx1 channels with an IC50 value of 5 µm [43].Panx1 channels are more sensitive to CBX than connexin hemichannels andconsequently low doses of CBX seem to preferentially inhibit Panx1 activity.The anti–malarial drug mefloquine is another chemical that preferentially in-hibits Panx1 over connexin channels at low doses. While Panx1 channels areinhibited with an IC50 of 50 nm [131], many connexin isoforms require over10 µm to reach a similar effect [61]. Probenecid, a drug used for the treat-ment of gout symptoms, has recently gained popularity as a Panx1 channelblocker. Probenecid blocks Panx1 channels with an IC50 of approximately150 µm without significant inhibitory effect on connexin channels even atdoses in the millimolar range [268]. More recently, the food dye BrilliantBlue FCF, found in blue Gatorade and blue M&Ms, was shown to be a fairlyselective inhibitor of Panx1 channels with an IC50 of 270 nm with no effecton either Cx46 or the chimeric connexin Cx32E143 [316]. A Panx1 mimeticinhibitory peptide called 10Panx1 has also been used to block Panx1 channelactivity. This 10 amino acid peptide is directed against the first extracel-20lular loop of Panx1 and has been shown to potently inhibit dye uptake,Panx1–mediated ionic currents, and Panx1–mediated interleukin-1β (IL-1β)release [222, 315]. This mimetic peptide also blocks Panx1–mediated activa-tion of N-methyl-d-aspartate receptor (NMDAR) secondary currents [296].However, it is worth noting that 10Panx1 can also inhibit some connexinisoforms and that the effects of 10Panx1 on Panx1 channels have been par-tially reproduced using polyethylene-glycol with a molecular weight of 1500dalton (Da) [315]. This result suggests that 10Panx1 might not work byinterfering with the extracellular loop of Panx1 but simply by blocking thepore of channels with a suitable size [65, 315].It is important to keep in mind that most pharmacological studies haveprimarily focused on Panx1 channels and that the effects of most compoundson Panx2 channels remain ambiguous. For example, using a liposome-based permeability assay, Ambrosi and colleagues reported that reconsti-tuted Panx1 and Panx2 channels had a similar sensitivity to CBX withan IC50 of 14 ±4 µm and 15 ±4 µm for Panx1 and Panx2 respectively [7].However, when expressed in oocytes, Panx2 channel activity was unaffectedeven by large concentration of CBX (100 µm) or probenecid (1mm) whilePanx1 channels were efficiently blocked [7]. It is possible that factors notpresent in the more artificial liposome-based assay can modify Panx2 suscep-tibility to inhibitors. Therefore, it is currently unknown whether endogenousPanx2 activity is inhibited by pharmacological blockers commonly used toinhibit Panx1 channel activity.1.1.11 Expression and distributionPanx1 has been shown to be expressed in a wide variety of cell and tissuetypes since its discovery and a thorough discussion on Panx1 expressionand distribution has been published elsewhere [34]. While Panx1 is widelyexpressed, the distribution profile of Panx2 initially appeared to be more re-stricted. Original studies exploring Panx2 expression by Northern blottingreported a distribution profile largely or exclusively limited to the centralnervous system (CNS) [14, 42]. By performing Northern blot analysis on rat21polyadenylated messenger RNA (mRNA) obtained from a panel of 16 differ-ent tissues, Roberto Bruzzone showed that Panx2 transcripts were mostlyrestricted to the spinal cord, brain and eye with very limited expression inthe thyroid, prostate, kidney, and liver and virtually no detectable expres-sion in any other tissues [42]. Using the same Northern hybridization ap-proach but on human tissues, Ancha Baranova later showed that Panx2 wasstrongly expressed in the brain but absent from all 15 other tissues tested[14]. Different groups subsequently confirmed the massive enrichment ofPanx2 transcripts in the CNS by using quantitative real-time PCR (qPCR)on zebrafish tissues [36, 338]. Since all gene expression profiling studies re-ported that Panx2 transcriptional activity was largely restricted to the CNS,Panx2 was rapidly labeled as a brain–specific protein by the gap junctioncommunity.Panx1 and Panx2 transcriptional activity has been shown to be tempo-rally but inversely regulated during brain development [312]. Panx1 mRNAexpression decreases substantially while Panx2 mRNA expression increasesnoticeably from early developmental stages to adulthood [312]. Overall,these observations strongly suggested that Panx2 is a CNS–specific proteininvolved in the development of the nervous system. However, because of theabsence of suitable antibodies, most of our knowledge on Panx2 has beenlimited to the analysis of its transcript and until recently, very few studieshad investigated Panx2 at the protein level. However, as will be discussed ingreater details in Chapter 2, variations in Panx2 mRNA abundance do notpredict changes in Panx2 protein levels and therefore our assumptions onPanx2 expression and distribution need to be re–evaluated. Very recently,Panx2 was shown to be expressed in the myenteric and submucosal gangliaof the human colon, particularly in the cytoplasm of neurons from the en-teric system [78]. Panx2 was also found in the cytoplasm of melanotrophsand neurons in the pituitary gland [176].221.1.12 Physiological and pathological functions of pannexinchannelsSeveral studies have investigated the roles of Panx1, and to a lesser extentPanx2 and Panx3 channels, under physiological and pathological conditions.Ischemic injuryCerebral ischemia induces massive disruption of the electrochemical gradi-ent across the plasma membrane and leads to a dramatic anoxic depolar-ization of neurons (reviewed in [294]). This massive depolarization causesexcessive release of neurotransmitters, such as glutamate and ATP, whichinduces excitotoxicity and utimately cell death. Roger Thompson was thefirst researcher to propose that the opening of Panx1 channels could beresponsible for this anoxic depolarization and lead to neuronal excitotoxi-city and necrosis under ischemic conditions [295]. He recorded large anddeleterious inward currents inhibited by gap junction blockers in isolatedhippocampal neurons exposed to ischemic conditions mimicked by oxygenand glucose deprivation [295]. The tools for studying Panx1 were limitedat that time but other groups later corroborated the involvement of Panx1channels in ischemia by showing that blocking or knocking out Panx1 re-duced the anoxic depolarization and procured a neuroprotective advantageunder ischemic conditions [52, 82, 84, 320]. Interestingly, one group failed toreport any neuroprotection in a Panx1 knockout mouse model but reporteda modest neuroprotective effect when Panx1/Panx2 double knockout micewere subjected to stroke [16]. These double knockout mice presented smallerstrokes and less severe neurological deficits than wild type mice after occlu-sion of the middle cerebral artery by electrocoagulation [16]. This resultsuggests that compensatory mechanisms might be at play in single knock-out mice but the specific contribution of Panx2 channel in ischemia is stillcurrently unknown.Ischemia involves the dysregulation of several ion channels. For example,excessive activation of NMDARs occurs prior to Panx1 activation and hasthe ability to activate several downstream events leading to neuronal death23(reviewed in [294]). Interestingly, in a subsequent study, Thompson showedthat aberrant activation of NMDARs induces the activation of Panx1 chan-nel and could contribute to epileptiform seizure activity in the hippocampus[296]. The activation of Panx1 channels by NMDAR was shown to involvethe phosphorylation of Panx1 residue Y308 by Src family kinases [321].InflammationIt has been proposed that the neuroprotection effect observed upon silencingPanx1 channels might result from a reduction in inflammation as Panx1 wasshown to directly interact and regulate the activation of the inflammasomeprotein complex in neurons and astrocytes [70, 84, 269]. The inflammasomeis a component of the innate immune system which promotes the maturationof the pro–inflammatory cytokines interleukin-1β (IL-1β) and interleukin-18 (IL-18). The oligomerization and activation of the inflammasome complextriggers the activation of caspase–1 which then processes IL-1β and IL-18precursors into their bioactive form prior to their release in the surroundingtissue [79, 192]. The composition of the inflammasome complex depends onthe signal which triggers its assembly but co–immunoprecipitation experi-ments have shown that upon brain injury or neuronal insult, Panx1 inter-acts with several components of the multiprotein inflammasome complex,including the P2X7 purinergic receptor, NACHT, LRR and PYD domains-containing protein 1 (NLRP1), apoptosis–associated speck–like protein con-taining a CARD (ASC), caspase–1, caspase–11, and X-linked inhibitor ofapoptosis protein (XIAP) [70, 222, 269]. Inhibiting Panx1 with silencingRNA (siRNA) or the 10Panx1mimetic inhibitory peptide has initially shownto block caspase–1 activation, IL-1β and IL-18 processing and their releasefrom mouse and human macrophages [222, 223]. Furthermore, high ex-tracellular potassium ion (K+) concentrations increase Panx1 channel per-meability and activate inflammasome assembly and caspase–1 activation inneurons and astrocytes, an effect that was inhibited by the Panx1 channelblocker probenecid or by silencing Panx1 expression [269]. More recently,Panx1 knockout mice were shown to have reduced inflammasome–mediated24activation of caspase–1 and production of IL-1β in a retinal ischemia model[84]. However, these results have been challenged by other studies whichdid not report any deficit in inflammasome assembly, activation of caspase–1 and the release of IL-1β or IL-18 in Panx1 knockout mice [16, 237] and aconsensus on the exact role of Panx1 in inflammation has not been reachedyet. On the other hand, Panx2 is unlikely to be involved in inflammasomeassembly and activation as IL-1β processing and release was unaffected inPanx2 knockout mice [16].Panx1 serves another important role in inflammation as it mediates therelease of the ATP ’find-me’ signal by apoptotic cells [50]. Apoptotic cellsrelease an ATP ’find-me’ signal during the earliest stages of cell death topromote phagocytic clearance [87]. This signal is abrogated by pharmaco-logical inhibition of Panx1 channels, siRNA–mediated Panx1 knockdown, orPanx1 knockout [50, 237]. The activation of Panx1 channels is regulated bycaspase–3 and caspase–7 in apoptotic cells [50]. The Panx1 C–terminal tailcontains specific caspase–cleavage sites which are essential for the opening ofPanx1 channels upon apoptosis [50]. Interestingly, caspase–3 and caspase–7cleavage sites have also been identified in the Panx2 C–terminal tail [227].However, this result was obtained in a cell–free assay and the physiologi-cal relevance of cleavage of Panx2 by caspases has not been explored yet.Panx1 has also recently been shown to regulate leukocytes extravasationduring acute inflammation. Lohman et al. [185] showed that tumor necrosisfactor α (TNF–α) induced Panx1–mediated ATP release by endothelial cells,thereby promoting leukocyte adhesion and emigration through the venouswall. This effect was abrogated upon silencing Panx1 activation with siRNAor pharmacologically with 10Panx1. Interestingly, this effect was dependenton the recruitment of Src family kinases (SFK) and SFK–dependent phos-phorylation of Panx1 [185].CancerPublicly available gene expression databases identified a positive correla-tion between Panx2 expression levels and post diagnostic survival in brain25cancer patients [180]. This is particularly important as Panx2 expressionis frequently down–regulated in brain tumors. For example, the expressionof Panx2, but not Panx1, was shown to be substantially down–regulated ina dataset of human brain tumors of various histogenesis retrieved from theNCBI Gene Expression Omnibus database Figure 1.5.Figure 1.5. Panx2 gene expression is down–regulated in human brain tumors.The dataset GDS1813 comprising 48 human brain tumors of various histogenesisand 4 normal brain samples was retrieved from the NCBI Gene Expression Omnibusdatabase and queried for Panx1 and Panx2 gene expression. The expresssion of bothpannexins is regulated differently in brain tumors. While Panx2 gene expressionis severely reduced in brain tumors, Panx1 mRNA levels are unchanged. Thearithmetic mean is represented by dark red squares.Soon after this early report indicating that glioma patients express-ing higher Panx2 levels experienced a better outcome [180], Charles Lai,a former graduate student in our laboratory, showed that Panx2 reducesin vitro and in vivo oncogenicity properties [165]. Over–expressing Panx226protein in C6 rat glioma cells reduced cell proliferation and saturation den-sity, suppressed anchorage–independent growth and decreased in vivo tu-mor growth [165]. Surprisingly, despite the fact that Panx1 transcript levelsremains largely unchanged in brain tumors (Figure 1.5), over–expressingPanx1 nonetheless dramatically reduced in vitro and in vivo tumorigenicityof C6 glioma cells [164]. However, the anti–tumorigenic properties of Panx1appear to be cell–type specific as another group has recently reported thatsilencing Panx1 expression in melanoma cell lines reduces melanoma progres-sion and promotes their reversion towards a melanocytic phenotype [226].Surprisingly, Furlow et al. [95] have identified a mutation encoding a trun-cated form of Panx1 in highly metastatic human breast cancer cells. Thisnonsense mutation (C268T) was shown to substitute a premature termina-tion codon at the amino acid position 90, resulting in a truncated Panx1called Panx11-89. Interestingly, Panx11-89 was shown to stimulate the activ-ity of Panx1 channel and enhance Panx1–mediated ATP release [95]. Thisaugmentation in extracellular ATP was shown to promote metastatic cellsurvival during traumatic deformation in the microvasculature, thereby in-creasing metastatic efficiency [95]. Overall, these results raise the possibilitythat Panx1 activity might promote or reduce tumor progression dependingon its context of expression and activity level.Cellular proliferation and differentiationSeveral studies from independent groups have suggested that pannexin chan-nels influence cellular proliferation and differentiation. In 2010, Leigh AnneSwayne suggested that Panx2 could influence cellular differentiation. Swayneand colleagues found that Panx2 is down–regulated is neuronal progeni-tors and immature neurons but re–expressed in mature neurons [285]. In-terestingly, silencing Panx2 expression accelerated the morphological andbiochemical differentiation of N2a mouse neuroblastoma cells into matureneurons [285]. Steve Bond, a former PhD student in our group, and oth-ers have proposed that Panx3 might also play a role in the regulation ofchondrocyte and osteoblast proliferation and differentiation [35, 133, 134].27Panx3 is expressed in chondrocytes and osteoblasts prior terminal differen-tiation [35, 134] and was shown to enhance maturation of both cell typesto a terminally differentiated state [133, 134]. Interestingly, very recently,Panx3 knockout mice were shown to have more robust femora and humerithan wild–type mice, thereby further supporting the idea that Panx3 playsa role in long bone development [46]. Panx1 and Panx3 channels have alsobeen shown to regulate myoblast proliferation and differentiation in skeletalmuscle [167]. Panx1 protein expression increases dramatically during dif-ferentiation of primary skeletal muscle by serum deprivation and was foundto promote myoblast differentiation in over–expressing studies [167]. Onthe contrary, pharmacological inhibition of pannexin channels inhibits my-oblast differentiation [167]. Interestingly, the glycosylation pattern of bothchannels changes subtantially in human fetal vs adult muscle tissues therebyindicating that pannexin glycosylation might play a role in skeletal muscledevelopment [167].1.1.13 Intracellular pannexin channelsGap junction proteins are predominantly expressed at the plasma mem-brane where they either facilitate the exchange of ions and small moleculesbetween the cytoplasm and extracellular space or in the case of connex-ins, establish intercellular coupling allowing direct communication betweenneighboring cells. However, this dogma has been challenged recently asstudies describing intracellular pannexin channels have emerged. AlthoughPanx1 channels are primarily expressed at the plasma membrane, they havealso been sporadically observed in the cytoplasm. For example, the pituitarygland expresses several Panx1 isoforms that are targeted to different cellu-lar compartments [176, 177]. The longest isoform is expressed in the plasmamembrane while two shorter isoforms are confined to the cytoplasm [177].The panx1b gene found in the teleost lineage also localizes in intracellularvesicles while the panx1a is primarily expressed at the plasma membrane[36]. Surprisingly, while comparing antibodies recognizing different Panx1epitopes, Cone and colleagues reported different subcellular localizations for28Panx1 in the rat brain [53]. Although they could not prove that all the an-tibodies were specific, their results nonetheless suggest that Panx1 might bespecifically targeted to different cellular compartments. Panx1 and Panx3channels have also been shown to localize in the ER in addition to theplasma membrane [133, 306]. More importantly, as we will see in furtherdetails in the following chapters, Panx2 has a unique subcellular distribu-tion for a gap junction protein and is primarily found in the endomembranesystem of the cell and rarely, if ever, observed at the plasma membrane[31, 165, 285, 332]. Our group has reported that Panx2 staining overlappedwith ER membranes [165] while other groups have suggested that Panx2 isfound in endolysosome compartments [31, 322]. However, these studies usedover–expressing systems which might have induced mistrafficking towardsthe lysosomal compartment and the subcellular localization of endogenousPanx2 has not been identified yet. The function of this cytoplasmic pool ofpannexin channels remains elusive and constitutes an active source of dis-cussion within the gap junction community. Panx1 and Panx3 have beenproposed to form Ca2+ leak channels in the ER thereby affecting the load ofER Ca2+ stocks [133, 306]. However, those results were primarily obtainedusing over–expressing systems and it remains unclear whether endogenouspannexins can also form Ca2+–permeable ion channels in the ER. As wewill see in great detail in Chapter 3, this thesis presents solid evidence show-ing that endogenous Panx2 is expressed in specific ER microdomains thatlocalize in the vicinity of mitochondria and uncovers that Panx2 modulatescell activity through non–traditional routes for a gap junction protein.1.2 Mitochondria–associated ER membranesMitochondria are double membrane-bound organelles often described as the"powerhouse of the cells" because they produce most of the cellular energy.These organelles generate the high energy molecule ATP by utilizing theoxygen we breathe to oxidize the calories derived from our diet through ametabolic pathway called oxidative phosphorylation. Although ATP pro-duction represents one of their most important function, mitochondria are29nonetheless involved in several other cellular processes such as lipid pro-cessing, calcium signaling, regulation of programmed cell death and cellularsignaling through mitochondrial reactive oxygen species (ROS). Mitochon-dria are comparmentalized by a double–membrane but must nonethelessphysically interact with other cellular components for proper functioning.Several independent groups reported a close morphological relationship be-tween the mitochondria and ER in the 1960s [54, 253]. Unique specializa-tions of the ER membrane, known as mitochondria–associated ER mem-branes (MAMs), physically recruit mitochondria and functionally coupleboth organelles. In HeLa cells, up to 20% of mitochondria establish suchphysical contacts with ER membranes [246]. ER–mitochondria contact siteshave an average distance of 25 nm between both organelles and can span sev-eral hundred nanometers in length, thereby covering an important area of themitochondria [64]. It is well-established that the cytoskeleton is importantfor maintaining the shape and structural integrity of organelles. However,it has been shown that the ER–mitochondria association is retained follow-ing microtubules or intermediate filaments depolimerization [189, 274] or inhighly dynamic organelles [198]. These observations indicate that the asso-ciation of the ER with the outer mitochondrial membrane (OMM) must re-quire tethers that physically couple the two organelles. Electron microscopystudies have reported the existence of such proteinaceous structures phys-ically tethering mitochondria to ER membranes [64]. This proteinaceouslinkage spans approximately 12% of the mitochondrial outer surface [64].MAMs have been shown to regulate mitochondrial energy production, lipidbiosynthesis, intracellular Ca2+ signaling and apoptosis [69, 119, 171, 248].Furthermore, MAMs are increasingly recognized for their roles in patholog-ical conditions such as obesity, Alzheimer’s disease, stroke and Parkinson’sdisease [9, 11, 112, 214, 261]. Owing to their ability to modulate mitochon-drial behavior and regulate physiological and pathological functions, mucheffort have been dedicated to understanding the molecular composition, or-ganization, and function of MAMs. As we will see in the following sections,the molecular composition of these tethers is slowly emerging and a few pro-tein complexes facilitating the coupling of ER and mitochondria have been30identified in yeast. Although the nature of these tethers is less documentedand more debated in higher organisms, several proteins were nonethelessfound to be enriched in MAMs in mammalian cells.1.2.1 The ERMES complex in yeastThe molecular composition of a protein complex linking the ER to mito-chondria has recently been identified by Kornmann and colleagues througha genetic screening of the yeast Saccharomyces cerevisiae [153]. The authorsreasoned that if ER–mitochondria contact sites are important, mutations ofproteins tethering both organelles would be detrimental but should be res-cued by expressing a chimeric protein artificially linking the two organelles[153]. This approach was based on a previous study performed in a ratleukemia cell line (RBL-2H3) which used a similar synthetic protein [64].In both studies, a chimeric construct comprising a fluorescent protein fusedto an OMM–targeting sequence at the N–terminus and an ER–targetingsequence at the C–terminus was used to artificially tighten the ER and mi-tochondria. By screening for yeast mutants that were unable to grow inthe absence of this chimeric linker Kornmann and colleagues identified twogenes: mitochondrial distribution and morphology protein 12 (MDM12) andmitochondrial distribution and morphology protein 34 (MDM34). Mdm12,a peripheral OMM protein and Mdm34, an integral OMM protein, wereshown to form a complex with two other proteins: the integral ER glycopro-tein Mmm1, and Mdm10, another OMM protein [32, 330]. The compositionof this protein complex was further confirmed by showing that the chimericlinker also rescued Mmm1 and Mdm10 mutants. This multiprotein complexwas named ER–mitochondria encounter structure (ERMES) and was shownto physically connect the ER to mitochondria [153]. In addition to these fourproteins, the conserved Ca2+-binding GTPase Gem1 has also recently beenshown to constitute a regulatory subunit of the ERMES complex [154, 283].More recently, another genetic screen performed in yeast identified a sec-ond protein complex tethering ER to mitochondria [163]. This conservedcomplex, named ER membrane protein complex (EMC), is made of six sub-31units, Emc1-6, which have been shown to localize to the ER and to interactwith Tom5, an OMM protein [163]. Lahiri and colleagues found that mu-tants missing EMC subunits had reduced ER–mitochondria contacts despitehaving a fully functional ERMES complex [163]. These results suggest thatEMC and ERMES complexes can work independently in coupling the ERand mitochondria. As will be shown in the following sections, the ER-MES complex does not merely provide a tethering force between the ERand mitochondria but is also directly or indirectly involved in importantmitochondrial functions.1.2.2 ER–mitochondria contact sites in chordatesAlthough the precise molecular composition of MAMs in higher eukaryotesremains uncertain, several proteins have been shown to preferentially local-ize or to be enriched at ER–mitochondria contact sites. For example, the ERCa2+ channel type III inositol 1,4,5–trisphosphate–gated receptor (IP3R3)has been found to localize in the proximity of mitochondria and to influ-ence mitochondrial Ca2+ uptake in mammalian cells [197]. Interestingly,another IP3 receptor, IP3R1, has been shown to indirectly interact with theOMM channel voltage-dependent anion channel 1 (VDAC1) through themitochondrial matrix chaperone 75 kDa glucose–regulated protein (GRP75)which forms a physical bridge between IP3R and VDAC1 [288]. There-fore, it has been suggested that the IP3R1–GRP75–VDAC1 complex mightconstitute the molecular tether coupling the ER to mitochondria in chor-dates. However, ER–mitochondria contact sites from IP3R triple knockoutcells (IP3R1-/-, IP3R2-/-, and IP3R3-/-) were undistinguishable from thoseof wild-type cells when examined by electron microscopy [64], indicatingthat IP3R–independent tethers must also couple ER and mitochondria.A mitochondrial dynamin–related protein called mitofusin–2 (MFN2)has been shown to tether ER to mitochondria in mouse embryonic fibrob-lasts and HeLa cells [68]. MFN2 located on the ER was shown to tether thetwo organelles by directly interacting with mitofusin–1 (MFN1) and MFN2located on the OMM [68]. However, mitochondrial Ca2+ uptake was re-32duced by only ∼ 20% in MFN2-/- cells following IP3–generated cytosolicCa2+ transients by ATP stimulation [68]. Moreover, the ER–mitochondriajuxtaposition was only reduced by approximately 40% in Mfn2-/- cells [68]thereby indicating that other complexes must also form a bridge betweenmitochondria and ER. However, the idea that MFN2 tethers the ER tomitochondria was severely challenged by additional studies reporting thatthe loss of MFN2 does not reduce but rather increase the number of ER–mitochondria contact sites and the strength of Ca2+ transfer between or-ganelles [55, 91]. In light of these new findings, it has been proposed thatMFN2 works as an ER–mitochondria tether antagonist preventing excessiveproximity between both organelles [91].The multifunctional sorting protein phosphofurin acidic cluster sortingprotein 2 (PACS–2) was also shown to control the apposition of the ER tomitochondria as PACS–2 depletion caused mitochondria to uncouple fromthe ER [270]. Moreover, the localization of PACS–2 at ER–mitochondriacontact sites influences the induction of apoptosis. Apoptotic signals inducethe redistribution of PACS–2 from the bulk of the ER to regions of closeapposition with mitochondria [270]. Following the induction of apoptoticpathways, PACS–2 was shown to bind to Bid and to promote its translo-cation to mitochondria [270]. Following its translocation to mitochondria,Bid is cleaved by caspase–8 to form the potently apoptotic truncated Bid(tBid) which promotes cytochrome C release through its interaction withBax [111, 319]. Silencing PACS–2 expression increased cell survival follow-ing staurosporine (STS) treatment by preventing the translocation of Bidto mitochondria and its processing by caspase–8 [270].The ER protein sigma-1 receptor (sigma–1R) has been shown to forma Ca2+-sensitive chaperone at MAM sites [118]. Under physiological Ca2+concentrations, sigma–1R associates with BiP, another ER chaperone, atER–mitochondria contact sites. However, upon depletion of ER Ca2+ stocks,the sigma-1R/BiP complex quickly disassembles and sigma–1R binds to theIP3R3 to prevent IP3R3 degradation by the proteasome, thereby sustainingCa2+ signaling to mitochondria [118].In addition to the aforementioned studies which looked at the localiza-33tion of individual proteins at MAMs, two recent shotgun proteomic analysesof MAMs isolated from mouse brain [234] or human fibroblasts infected witha human cytomegalovirus [335] have been published. Both proteomic stud-ies used differential centrifugation followed by subcellular fractionation ona Percoll gradient to strip mitochondria from their associated membranesand isolate a so–called "pure" MAM fraction [323]. By following this enrich-ment approach, 991 [335] and 1212 [234] proteins were identified in MAMsby Zhang and Poston respectively. However, these numbers must be inter-preted cautiously as most of the proteins identified were not restricted tothe MAMs but were also found in the bulk of the ER and in various otherorganelles. For example, 24% of the 1212 proteins identified by Poston werefound to reside in the plasma membrane [234]. This high ratio can be ex-plained by the impossibility to isolate pure organelles or cellular domainsdevoid of contaminants by fractionation. Proteomic mapping using APEXtagging constitute an interesting alternative to subcellular fractionation asit bypasses the need for organelle purification [127, 241]. The ascorbateperoxidase, or APEX, is a genetic tag that was originally engineered forelectron microscopy [191]. In addition to catalyzing the polymerization ofdiaminobenzidine into precipitate that are resolvable by electron microscopy,APEX also oxidizes numerous phenol derivatives to phenoxyl radicals whichcovalently binds to proteins [241]. These radicals are short–lived, have asmall labeling radius (<20 nm), and can be used for proteomic labeling byadding biotin–phenol to the medium [241]. Following the labeling, biotiny-lated proteins can be isolated and identified by mass spectrometry. In 2014,Hung and colleagues successfully used this approach to map the proteomeof the mitochondrial intermembrane space and identified nine new mito-chondrial proteins that were subsequently confirmed using other approachessuch as microscopy and Western blotting [127]. Studies using APEX toinvestigate the composition of molecular complexes at MAMs yet remainto be published but Dr Alice Y. Ting from the Massachusetts Institute ofTechnology is currently following this strategy [298].341.2.3 Functional importance of MAMsThe recruitment of mitochondria on ER membranes does not only physicallycouple both organelles but also form functional microdomains facilitating theflow of information between the ER and mitochondria. These contact sitesconstitute important signaling hubs and the number of cellular processesthe MAMs have been shown to regulate has increased considerably over thelast 10 years or so. In the following sections, we will discuss some of thebiological processes that have been shown to be regulated by the MAMs ineukaryotes.Lipid biosynthesis and traffickingRecently, new vesicular transport routes connecting mitochondria to per-oxisomes and lysosomes have been discovered [208, 277]. However, thoseroutes are unidirectional and the endomembrane system does not estab-lish vesicular traffic routes towards mitochondria. Therefore the mode oflipids delivery to mitochondrial membranes has long remained enigmatic. Inmammalian cells, most phospholipids are synthesized in the ER but mito-chondria nonetheless supply all the cardiolipin and an important proportionof phosphatidylethanolamine (PE). Therefore, the exchange of phospho-lipids between the ER and mitochondria is not only essential to maintainthe integrity of mitochondrial membranes but also to ensure proper lipidmetabolism at the cellular level. Mitochondria produce PE by convert-ing phosphatidylserine (PS), a phospholipid synthesized in ER membranes,through a decarboxylation step catalyzed by the enzyme phosphatidylserinedecarboxylase (PSD) (reviewed in [305]). Whereas yeasts contain two dis-tinct PSDs (Psd1 located in mitochondria and Psd2 in the Golgi/vacuole),mammalian cells contain only one PSD protein located in the inner mito-chondrial membrane [228, 299, 300, 333]. The idea that ER–mitochondriacontact sites could be involved in the exchange of lipids between both or-ganelles originated in the early 1990s [254, 303] and was supported by studiesdemonstrating that MAMs are several–fold enriched in enzymes involved inlipid biosynthesis such as phosphatidylserine synthase, acyl-CoA:cholesterol35acyltransferase and diacylglycerol acyltransferase [254, 281, 282]. Althoughsome points of contention still remain, several studies have further substan-tiated the idea that MAMs are involved in lipid exchange between the ERand mitochondria over the last decade.Yeast missing the EMC or ERMES complexes showed partial defects inlipid exchange between the ER and mitochondria [153, 163] thereby rein-forcing the functional importance of ER–mitochondria contact sites in lipidprocessing and biosynthesis. For example, yeast mutants missing multiplecomponents of the EMC had decreased PS transfer from the ER to mito-chondria as shown by a reduction in mitochondrial PS and PE levels [163].Although the ERMES complex was also originally shown to be involved inthe transfer of PS between the ER and mitochondria [153], subsequent stud-ies failed to reproduce these results [209, 314]. It is possible that the ERMESand EMC complexes compensate for each other under certain conditions andcreate redundant routes between the ER and mitochondria. Alternative lipiddelivery routes might also be used in yeast which could help traffic lipidsto mitochondria in ERMES mutants. For example, other contact sites havebeen observed between mitochondria and the yeast lysosome-like vacuole(called vCLAMP) which could be an alternative route for lipids delivery tomitochondria [85, 125]. Interestingly, ERMES and vCLAMP are at leastpartially redundant as the absence of one complex induces a compensatoryincrease of the other while the elimination of both complexes is lethal [85].Mitochondrial divisionER–mitochondria contact sites have also been shown to be important formitochondrial fission. The distribution of mitochondria within the cyto-plasm is not only regulated by motility but also requires mitochondrial di-vision and fusion events. Mitochondrial division is catalysed by the highlyconserved dynamin–related protein 1 (Drp1) in mammals (Dnm1 in yeast).Both proteins self–assemble into spirals characterized by a diameter muchsmaller than the diameter of unconstricted mitochondria, indicating thatDrp1/Dnm1–independent mitochondrial constriction events must occur prior36to the recruitment of Drp1/Dnm1 [94, 132, 161]. In 2011, Friedman et al.[94] discovered that ER–mitochondria contact sites play an important rolein defining the sites of mitochondrial division by marking the points of con-striction before the fission event. Electron microscopy and live–cell confocalmicroscopy showed that ER tubules wrap around mitochondria and initiatethe constriction prior to the recruitment of Drp1 [94].Mitochondrial calcium homeostasisEarly reports in the 1960s clearly demonstrated that energized mitochon-dria can selectively uptake Ca2+ [73] but the question of how Ca2+ entersmitochondria has nonetheless constituted a challenging puzzle over the last50 years. Extensive work from several groups has however changed the land-scape and much details have emerged over the last decade on mitochondrialCa2+ uptake and regulation. Ca2+ import across the OMM occurs throughthe VDAC1 [240]. The OMM has traditionally been considered freely per-meable to small molecules and ions but it should be noted that selectivepermeabilization of the OMM has been shown to increase the magnitude ofmitochondrial Ca2+ [63], thereby suggesting that a certain selectivity mighttake place at the OMM. Although biochemical, pharmacological, electro-physiological and live-cell imaging studies had reported solid and converg-ing evidence for Ca2+ uptake by the mitochondrial matrix, the molecularnature of the Ca2+ channel at the inner mitochondrial membrane (IMM)remained a mystery until very recently. In 2011, using in silico approaches,two groups independently identified the previously uncharacterized trans-membrane protein CCDC109A as the inner membrane mitochondrial Ca2+uniporter (MCU) [18, 71]. Silencing MCU severely abrogated mitochon-drial Ca2+ uptake while overexpressing MCU had the opposite effect andmarkedly increased mitochondrial Ca2+ uptake [18, 71].The MCU was characterized as a highly selective, inwardly rectifyingCa2+ channel [151], but earlier studies had reported that mitochondrial Ca2+uptake machinery had a low affinity for Ca2+. The Km value representingthe concentration required for half maximal rate of Ca2+ uptake by isolated37mitochondria was found to range between 5 µm to 13 µm [45]. These valuesare more than an order of magnitude higher than cytosolic Ca2+ concen-tration which is maintained around 100 nm in resting conditions. Owing tothis very low affinity for Ca2+, it was initially assumed that cytosolic Ca2+would never significantly contribute to Ca2+ uptake by mitochondria underphysiological conditions. Transient cytosolic Ca2+ peaks of approximately2 µm had been observed under physiological conditions [6], but these valueswere still negligible in comparision to the concentration required to elicit asubstantial mitochondrial Ca2+ uptake. It was therefore predicted that theeffect of Ca2+ on mitochondria was only marginal and could only regulatemitochondrial activity under pathological conditions inducing cellular Ca2+overload. However, the development of targetable and more accurate Ca2+indicators has revealed rapid and transient increases of mitochondrial Ca2+following elevations of cytosolic Ca2+ [244, 245]. These results by Rizzutoand colleagues were quite intriguing because they highlighted an apparentcontradiction between the rapid response of mitochondria to fluctuationsof cytosolic Ca2+ concentrations and the low affinity of their transportersfor Ca2+. It was not until the late 1990s that this contradiction was re-solved when Rizzuto and others showed that despite at low affinity for Ca2+,mitochondria do nonetheless efficiently uptake Ca2+ because of their closejuxtaposition with ER membranes [62, 92, 246]. ER–mitochondria contactsites are stable enough to generate very localized microdomains with highCa2+ concentration that allow the rapid accumulation of large amount ofCa2+ by mitochondria [62, 92]. Furthermore, the Ca2+ coupling between ERand mitochondria can be weakened or solidified by respectively disruptingor strengthening the proteinaceous tethers maintaining the integrity of theMAMs [64]. Therefore, it is now largely assumed that the rapid accumu-lation of Ca2+ in the mitochondrial matrix is dependent on MAMs whichposition mitochondria in close proximity to ER Ca2+ release sites.Several groups have independently demonstrated that Ca2+ uptake bymitochondria influences mitochondrial metabolism and cellular functions.For example, mitochondrial Ca2+ uptake is essential to maintain a metabol-ically energized pool of mitochondria in living cells [139]. As will be dis-38cussed in the following sections, early work on purified enzymes and subse-quent studies on isolated mitochondria or intact cells have shown that sev-eral enzymes of the tricarboxylic acid (TCA) cycle are strongly stimulatedby Ca2+ (reviewed in [105]). More recently, Ca2+ has also been found toregulate energy conversion potential by influencing mitochondrial oxidativephosphorylation (OxPhos) [59, 107, 293].Mitochondrial calcium regulates the TCA cycle. It has been known for along time that the activity of three rate-limiting mitochondrial dehydroge-nases involved in the TCA cycle, and consequently ATP production, aredependent on Ca2+ [44, 194]. The effects of Ca2+ on the TCA cycle aresummarized in Figure 1.6. The activity of pyruvate dehydrogenase (PDH)has been shown to be indirectly stimulated by Ca2+ [74, 76]. PDH is amajor component forming the pyruvate dehydrogenase complex (PDC) andcontributes to the conversion of pyruvate into acetyl-CoA. The activity ofPDH is tightly controled through a phosphorylation-dephosphorylation cy-cle [178, 179]. The phosphorylation of PDH by the pyruvate dehydrogenasekinase (PDK) inactives PDH and inhibits the conversion of pyruvate intoacetyl-CoA [178, 179]. However, the phosphorylation of PDH is reversedby a Ca2+-dependent pyruvate dehydrogenase phosphatase. The activityof the phosphatase, and consequently of PDH, is markedly increased by anaugmentation of Ca2+ concentrations [74, 76]. For example, the conversionrate of pyruvate into acetyl-CoA by PDH increased by sevenfold when Ca2+was raised from 10 nm to 9 µm [74].As shown in (Figure 1.6), the activity of two other key enzymes of theTCA cycle is also modulated by Ca2+: the isocitrate dehydrogenase, whichcatalyzes the oxidative decarboxylation of isocitrate into α–ketoglutarate,and the α–ketoglutarate dehydrogenase (or oxoglutarate dehydrogenase) re-sponsible for the conversion of α–ketoglutarate (or oxoglutarate) into suc-cinyl CoA. The main effect of mitochondrial Ca2+ is to greatly reduce theMichaelis constant Km of both enzymes [75, 76, 97, 168, 193]. Therefore, inthe presence of Ca2+, less substrate is needed to achieve a conversion rateequivalent to half of the maximal enzymatic velocity. Consequently, in the39presence of Ca2+, the production of α–ketoglutarate and succinyl CoA isincreased at lower concentration of isocitrate and oxoglutarate respectively.It is believed that Ca2+ directly stimulates the activity of isocitrate de-hydrogenase and α–ketoglutarate dehydrogenase but in contrast to the wellcharacterized effect of Ca2+ on PDH, the exact mechanism by which Ca2+ in-creases the substrate affinity of isocitrate dehydrogenase and α–ketoglutaratedehydrogenase is still poorly understood.Mitochondrial calcium regulates oxidative phosphorylation complexes. Aswe have seen mitochondrial Ca2+ increases the efficiency of the TCA cy-cle by stimulating the conversion rate of three rate-limiting dehydrogenases.Interestingly, the maximum respiratory rate of heart mitochondria increasesin a near linear fashion with increasing concentrations of nicotinamide ade-nine dinucleotide (NADH) [200, 202]. Therefore, upon moderate increasein mitochondrial Ca2+ the TCA cycle produces more NADH which is de-livered to the electron transport chain (ETC) and as a result, cellular res-piration and ATP synthesis increase. However, it has been proposed thatCa2+ can also modulate cellular respiration independently of its action onthe TCA cyle [100, 202]. For example, by studying the effect of Ca2+ onmitochondrial oxygen consumption under conditions where the metabolicdriving forces (NADH) were uncoupled from ATP production rates, Territoand colleagues demonstrated that the increase in oxidative phosphoryla-tion observed in the presence of Ca2+ could not be solely attributable to asimple increase in NADH production but involved the direct activation ofATP synthase [293]. The exact mechanism by which mitochondrial Ca2+directly stimulates oxidative phosphorylation is still debated. However, ithas been suggested that Ca2+ might stimulate mitochondrial respiration byinfluencing the phosphorylation state of ETC complexes [258]. All mam-malian oxidative phosphorylation (OxPhos) complexes have been shown tobe phosphorylated in vivo [8, 58, 129]. Although the functional implicationof phosphorylation has not been addressed for all complexes, the phospho-rylation of cytochrome c (Cyt c) and cytochrome c oxidase (complex IV orCOX) has been investigated in more details.40Figure 1.6. Calcium regulates the activity of rate–limiting mitochondrial dehydrogenases. The tricarboxylic acid cycle(TCA cycle) is a series of enzymatic reactions taking place in the mitochondrial matrix and is an important step of aerobicrespiration in living cells. A) When Ca2+ concentration in the mitochondrial matrix is low, three enzymatic reactionslimit the velocity of the TCA cycle and mitochondria are poorly energized. B) An increase in mitochondrial matrix Ca2+augments the activity of three mitochondrial dehydrogenases: the pyruvate dehydrogenase, which converts pyruvate intoacetyl-CoA, the isocitrate dehydrogenase, involved in the conversion of isocitration into α–ketoglutarate, and finally theα–ketoglutarate dehydrogenase which converts α–ketoglutarate into succinyl CoA. As a consequence, the TCA cycle is moreefficient, mitochondria are fully energized and cellular respiration is increased.41Cyt c is a multifunctional protein located in the mitochondrial inter-membrane space and as part of the ETC, it shuttles electrons from thebc1 complex (or complex III) to cytochrome c oxidase (COX). COX is thelast enzyme of the ETC and transfers the electrons it receives from Cytc to molecular oxygen which is reduced to water. The activity of Cyt cwas shown to be inhibited by phosphorylation on tyrosine residues 48 or97 [173, 221, 331]. Phosphorylation of Cyt c decreases cellular respirationas indicated by reduced oxygen consumption by COX and a two-fold in-crease in its Km value [173, 221, 331]. Importantly, cellular respiration isalso directly inhibited by phosphorylation of COX [172]. The maximal ve-locity of COX was shown to decrease by approximately 30% and its Kmvalue to increase by more than two–fold following cAMP-dependent tyro-sine phosphorylation of its subunit I [172]. Moreover, the activity of COX istightly regulated by allosteric inhibition by high mitochondrial ATP/adeno-sine diphosphate (ADP) ratios, a phenomenon known to adjust ATP pro-duction to the physiological demands [10, 205]. Importantly, this allostericinhibition is controlled by COX phosphorylation and has been shown to bereversed by Ca2+–induced dephosphorylation [22]. Ca2+–dependent phos-phatases have been shown to dephosphorylate several mitochondrial proteins[129] and as shown in Figure 1.7 it has been proposed that mitochondrialCa2+ can regulate cellular respiration by influencing the phosphorylationstatus of the OxPhos complexes [22, 258].Regulation of intrinsic apoptotic pathwaysOwing to their ability to regulate Ca2+ exchange between ER and mito-chondria, MAMs are also increasingly recognized as important functionalmicrodomains involved in the induction of apoptosis [219, 232, 247, 304].Apoptosis is a genetically regulated and finely tuned process of programmedcell death that plays an important role in the development and maintenanceof tissues as well as in the removal of harmful cells (reviewed in [292]). Dys-regulation of programmed cell death has dire consequences on cellular andtissue homeostasis and unsurprisingly, the study of apoptosis has received42Figure 1.7. Calcium stimulates oxidative phosphorylation and increases mito-chondrial energy conversion potential. A) The transport of electrons through theETC is initiated with the oxidation of NADH by complex I (NADH dehydrogenase)or succinate by complex II (not shown). The electrons are then successively trans-ferred to coenzyme Q, complex III (or bc1), cytochrome C (Cyt c) and complexIV (COX or cytochrome c oxidase). Complex IV is the final enzyme of the ETCand transfers electrons to molecular oxygen thereby producing water. The energyreleased by this electron transfer is coupled to the movement of protons across themitochondrial inner membrane, thereby creating a transmembrane electrochemicalgradient which ultimately drives ATP synthesis by complex V (or ATP synthase).Phosphorylation has been shown to limit the conversion rate of several enzymaticcomplexes along the electron transport chain. B) Mitochondrial Ca2+ uptake ac-tivates Ca2+–dependent phosphatases which dephosphorylate OxPhos complexesand increase energy conversion potential and mitochondrial ATP production.43much attention in the context of diseases such as cancer [41].As the transport of Ca2+ across the MCU channel is driven by the con-siderable force of the transmembrane potential (−140mV to −180mV), themitochondrial matrix can rapidly accumulate substantial amounts of Ca2+.For example, under basal conditions mitochondrial Ca2+ is maintained atlow concentrations of approximately 100 nm to 500 nm but can dramaticallyincrease up to 10 µm to 20µm upon physiological stimulation by IP3 [62, 255].As we have reviewed in the previous sections, MAMs help to maintain anoverall energized pool of mitochondria by controlling mitochondrial Ca2+uptake which ultimately regulates the TCA cycle and oxidative phospho-rylation. However, excessive Ca2+ release at MAM sites can overload mi-tochondria and induce apoptosis through the opening of the mitochondrialpermeability transition pore (mPTP) and the associated release of apoptoticfactors such as Cyt c (reviewed in [158]).The association of the ER and mitochondria is dynamically regulated.For example, proapoptotic stimuli such as serum starvation and tunamycintreatment, have been shown to increase the frequency of ER–mitochondriaassociation and decrease the gap between both organelles [64]. It has beenproposed that an increase in the strength of this association, i.e. a tight-ening of the coupling between both organelles, can increase the sensitivityof the cells to apoptotic challenges [64]. In support to this idea, the artifi-cial tigthening of ER and mitochondria with a chimeric linker was found toincrease mitochondrial Ca2+ overloading and permeabilization upon stimu-lation with thapsigargin [64]. These observations indicate that a remodelingof ER–mitochondria contact sites might be an important step in the induc-tion of apoptosis. This idea is supported by results presented in Section 1.2.2and which showed that PACS–2 is redistributed from the bulk of the ERto MAMs in apoptotic cells [270]. Several proapoptotic challenges such asH2O2, C2-ceramide, or arachidonic acid have been shown to facilitate theexchange of Ca2+ between ER and mitochondria [37, 103]. The inositol1,4,5-trisphosphate–gated receptor (IP3R) channels are particularly impor-tant in that regard. IP3R3 channels have been shown to selectively transmitapoptotic Ca2+ to mitochondria while IP3R1 channels predominantly me-44diated cytosolic Ca2+ mobilization [197]. The preferential implication ofIP3R3 in apoptosis is also supported by knockdown studies showing thatsilencing IP3R3, but not IP3R1, improved the viability of chick dorsal rootganglion neurons following nerve growth factor withdrawal [28]. However,silencing the expression of IP3R3 by 97% reduced ER–mitochondria Ca2+transfert by only about 30-40% and did not totally abolished bile acid–induced apoptosis in HEK or CHO cells [197]. These observations suggestthat other additional routes of ER–mitochondria Ca2+ are likely possible inthe context of apoptosis.1.3 Motivation, objectives, and highlightsThe properties and functions of Panx1 channels, and to some extent Panx3channels, have been well documented since the discovery of pannexins in2000. On the contrary, our understanding on Panx2 channels has progressedat a much slower pace and still remains superficial to this day. Severalreasons can explain this unbalanced distribution of knowledge among thepannexin channels. Soon after their discovery, Panx1 was found to be ubiq-uitously expressed and generated enthusiasm from very different fields ofresearch. On the contrary, Panx2 expression was thought to be limited tothe central nervous system and therefore generated a more circumscribedinterest. Furthermore, initial studies on pannexins suggested that Panx2channels were restricted to the cytoplasm; a much peculiar location for a gapjunction protein. This observation generated confusion among the commu-nity and was even suspected to be artifactual as early studies relied primarilyon over–expressing systems to address Panx2 localization. Finally, the lackof reliable antibodies has long hampered the study of Panx2. The primarygoal of this thesis was to increase our knowledge on Panx2. More specifi-cally, the aim was to characterize the expression, localization and functionof Panx2 protein. This goal was divided into two interconnected parts.Chapter 2 characterizes the expression and distribution of Panx2 proteinin mammalian tissues. Prior to this work, several gene expression studieshad found that Panx2 transcripts were largely restricted to the CNS and45Panx2 was consequently assumed to be a CNS–specific protein. However,the lack of suitable antibodies prevented the creation of a comprehensivemap of Panx2 protein expression and Panx2 protein localization profile wasmostly inferred from the distribution of its transcript. Since numerous in-dependent studies have shown that transcript and protein levels are poorlycorrelated, I hypothesized that Panx2 protein levels might not correlatewith its transcript levels and might therefore be expressed outside the CNS.Chapter 2 starts with the characterization of two novel anti–Panx2 anti-bodies which were subsequently used for the analysis of Panx2 protein ina large panel of mouse tissues. I show that Panx2 transcriptional activityis a poor predictor of Panx2 protein abundance and is completely uncorre-lated with Panx2 protein levels. In fact, despite showing disproportionatelyhigh transcript levels, the CNS expressed less Panx2 protein than any othertissues analyzed by Western blot. Panx2 protein was found to be ubiqui-tous and its expression was readily detected in every tissue examined, evenwhen transcriptional analysis predicted very low Panx2 protein expression.Furthermore, the immunofluorescence staining in mouse tissues proved thatendogenous Panx2 channels are indeed sequestered within the endomem-brane system of the cell. Overall, this work shows that Panx2 is a gapjunction channel ubiquitously expressed and confirms that Panx2 is uncon-ventional as it remains inside the cell; a much interesting localization for agap junction protein.Chapter 3 explores in further details the subcellular localization andbiological functions of Panx2 channels. Previous studies had shown thatPanx1 and Panx3 proteins form complex glycoprotein species requiring post–translational modifications occurring in the Golgi. However, the glycosyla-tion of Panx2 was shown to be limited to the addition of a high mannosegroup which does not involve transition through the Golgi apparatus. Con-sequently, I hypothesized that Panx2 would primarily distributes in the ER.Chapter 3 starts by presenting data obtained by subcellular fractionationand by analyzing Panx2 dynamics in live cells. The data strongly sug-gest that Panx2–containing compartments can associate, at least transiently,with the ER and mitochondria. This idea is corroborated by high resolution46immunogold electron microscopy which proves that Panx2 localizes at ER–mitochondria contact sites in tissues. Quite interestingly, the expression ofPanx2 protein is shown to be influenced by the energetic requirements of thecell. Panx2 is shown to be substantially up–regulated under conditions thatforce the cell to increase mitochondrial ATP production. However, I showthat Panx2 can sensitize cells to programmed cell death and therefore Panx2expression at ER–mitochondria contact sites must be tightly regulated.47Chapter 2Pannexin 2 proteinexpression is not restrictedto the CNS2.1 IntroductionGap junction proteins are traditionally described as aqueous plasma mem-brane channels which allow rapid cell–to–cell communication by directlyconnecting the cytoplasm of adjacent cells. In chordates, connexins arethe canonical gap junction proteins while gap junctions in invertebrates areformed exclusively by the evolutionarily unrelated innexin family. In 2000,another small gene family named pannexin was identified based on sequencehomology with the innexin family and was found to be expressed along-side connexins in chordates [216]. Three distinct pannexin paralogs (Panx1,Panx2 and Panx3) were initially identified in vertebrates [15, 215, 216]but recent studies showed that Panx1 has been retained as two indepen-dent ohnologs in teleost as a result of an ancestral whole genome dupli-cation [36, 160]. Despite the lack of sequence similarity between innex-ins/pannexins and connexins, both families share structural resemblance.Connexins and pannexins both have a predicted topology consisting of four48membrane–spanning domains, two extracellular loops, a cytoplasmic loop,and cytoplasmic N– and C–termini [215]. Despite sharing structural resem-blance with connexins, the ability of pannexin channels to form gap junc-tional coupling remains controversial. A few groups reported that Panx1 andPanx3 can form intercellular junctional channels [42, 133, 164, 256, 306] buttheir observations were limited to heterologous or over–expression systemand undisputable evidence supporting Panx–based coupling is still lacking.In contrast to connexins, all three pannexins are glycosylated at their ex-tracellular loops [225] with carbohydrate moieties that sterically hinder thedocking of channels from adjacent cells [29]. Therefore, it is largely acceptedthat under physiological conditions, pannexin channels primarily form non–junctional membrane channels controlling the exchange of ions and smallmolecules between the cytoplasm and extracellular space and do not signifi-cantly contribute to direct cell–to–cell gap junctional communication [276].Several gene expression profiling studies reported that Panx2 transcrip-tional activity is largely restricted to the CNS in human [14], rat [42] andzebrafish [36, 338]. Minimal Panx2 mRNA levels have also been detectedin some non–neural tissues such as the eye, thyroid, prostate, kidney, liver,heart and olfactory epithelium [36, 42, 83, 336] but given the much largerPanx2 mRNA levels found in the CNS, Panx2 transcript and correspond-ing protein are largely assumed to be primarily expressed in the CNS. Inthe healthy brain, Panx2 protein was shown to have a complex distributionpattern and is expressed in pyramidal cells and interneurons alike [332]. In-terestingly, Panx2 protein was also detected in astrocytes following ischemiain the rat but not in the healthy brain [332]. Panx2 protein is also presentin hippocampal neural progenitors and mature neurons both in vitro andin vivo [285]. However, because Panx2 is believed to be primarily CNS–specific, the mapping of Panx2 protein distribution in other tissues has notbeen undertaken.In this study, we compared Panx2 gene transcription and protein expres-sion profiles in mouse tissues using a combination of qPCR, Western blotand immunofluorescence. Our results reveal that Panx2 mRNA and proteinlevels are not correlated and demonstrate that Panx2 protein expression is49more ubiquitous than initially predicted.2.2 Materials and methods2.2.1 Animal careAll experiments were performed in accordance with the guidelines estab-lished by the Canadian Council on Animal Care and were approved by theUniversity of British Columbia Animal Care Committee (protocol numberA11-0169).2.2.2 AntibodiesThe two Panx2 mouse monoclonal antibodies (clone N121A/1 and cloneN121A/31) were generated by UC Davis/NIH NeuroMab Facility (Davis,CA, USA) using an immunogen made of the entire rat Panx2 protein se-quence (accession number P60571) minus the first 10 amino acids. Bothclones were used at 20 µgmL−1 for immunofluorescence and 5 µgmL−1 forWestern immunoblotting or dot blotting. The rabbit anti–Panx1 polyclonalantibody was generously provided by Dr. Dale Laird from the Universityof Western Ontario (London, ON, Canada) and was used at 2 µgmL−1 forimmunofluorescence and 0.4 µgmL−1 for Western immunoblotting. The rab-bit anti–GFAP (Sigma, St. Louis, MO, USA) was used at 1:500. Purifiedimmunoglobulin from non–immunized mouse was obtained from JacksonImmunoresearch (cat# 015-000-003; West Grove, PA, USA) and was usedat the same concentration as the anti–Panx2 antibodies. AlexaFluor– andhorseradish peroxidase (HRPO)–conjugated goat secondary antibodies wereobtained from Invitrogen (Carlsbad, CA, USA) and Sigma (St. Louis, MO,USA) respectively.2.2.3 Cell cultureWild–type C6 glioma cells as well as C6–Panx1GFP, C6–Panx2 and C6–Panx2GFP stable transfectants were cultured as previously described [164,165]. Briefly, cells were grown in low glucose Dulbecco’s Modified Eagle50Medium (DMEM) (Sigma–Aldrich, St. Louis, MO, USA) containing 10%fetal bovine serum, 10units/mL penicillin, and 10µgmL−1 at 37 ◦C and 5%CO2. Primary cultures of astrocytes were prepared as previously described[169]. Briefly, cortices were dissected from early postnatal (P0–P1) mousepups, freed of meninges, minced and mechanically triturated in DMEM.The cell suspension was then strained through a 70 µm filter and seededinto T75 flasks (2 cortices/flask). Cells were cultured in DMEM (Sigma–Aldrich, St. Louis, MO, USA) containing 10% fetal bovine serum (FBS), 10units/mL penicillin, and 10µgmL−1 streptomycin at 37 ◦C and 5% CO2 andthe medium was initially replaced 3 days after plating and every other daysubsequently. After 7–8 days, the flasks were vigorously shaken to removeloosely attached cells and primary astrocytes were harvested with trypsin–EDTA (Invitrogen, Carlsbad, CA, USA) and frozen in DMEM, 10% FBS,and 8% dimethyl sulfoxide (DMSO). Frozen astrocytes were thawed andplated on glass coverslips coated with poly–L–ornithine (0.01% solution,Sigma–Aldrich, St. Louis, MO, USA). Cultures were maintained for 5, 10or 15 days prior to staining. The percentage of Panx2–positive astrocyteswas defined as the number of cells that stained positively for Panx2 dividedby the number of nuclei. For each time point, that ratio was calculated byaveraging the values obtained from three coverslips with 10 field of views(168 µm×225 µm) per coverslip.2.2.4 Epitope mappingA library of 70 overlapping peptides covering the entire sequence of the ratPanx2 (Uniprot accession number P60571) minus the four transmembranedomains was obtained from Genscript (Piscataway, NJ, USA). Peptides were15 amino acids in length with a 7 amino acids overlap. A total of 100 µg ofpeptide was pre–incubated overnight at 4 ◦C in dot–blot buffer (8mol L−1urea, 100mmol L−1 NaH2PO4, 10mmol L−1 Tris, pH 8.0) containing 10µgbovine serum albumin. Peptides were then dot–blotted on nitrocellulose(NCL) membrane (Bio–Rad, Hercules, CA, USA), washed with phosphatebuffer saline (PBS) (pH 7.4), dried at 37 ◦C and blocked for 1h in milk51solution (4% nonfat milk, 20mmol L−1 Tris, 150mmol L−1 NaCl pH7.4).The membrane was then immunoprobed for 2h at room temperature withthe primary antibody followed by HRPO–conjugated secondary antibodies(Sigma, St. Louis, MO, USA) for 1h at room temperature.2.2.5 RNA isolation and real–time quantitative PCRThree 3 to 5 month old mice were deeply anesthetized by intraperitonealinjection of sodium pentobarbital and perfused transcardially with 10mL to15mL of PBS (pH 7.4) followed by 10mL to 15mL of aqueous ammoniumsulfate solution (5.3mol L−1 ammonium sulfate, 25mmol L−1 sodium cit-rate, 10mmol L−1 ethylenediaminetetraacetic acid (EDTA), pH5.2) to pre-cipitate degenerative RNases. Organs were rapidly harvested and storedat −80 ◦C in the same solution. Total ribonucleic acid (RNA) was har-vested from 50mg of tissue using Trizol (Invitrogen, Carlsbad, CA, USA)according to the manufacturer’s directions. Air–dried RNA samples were re–solubilized in diethylpyrocarbonate (DEPC)–treated double distilled H2Oand RNA quantity and purity was assessed using a NanoDrop 1000 Spec-trophotometer (Thermo Scientific, Waltham, MA, USA). All samples hadA260/280 and A260/230 ratios above 1.9 and 2.3 respectively. A total of500 ng per sample was reverse transcribed into cDNA in a 10 µL reaction vol-ume using qScript (Quanta Biosciences, Gaithersburg, MD, USA) accordingto the manufacturer’s instructions. Real–time qPCR was performed in 18 µLreaction volume containing 45 ng of cDNA and 0.4 µmol L−1 primers dilutedin 2X Fast Plus EvaGreen R© qPCR Master Mix (Biotium Inc., Hayward,CA, USA). The following primer pairs were used to amplify Panx2 cDNA(forward 5’–AGAAGGCCAAGACTGAGGCG–3’ and reverse 5’–GGAGCATCTTTGGTGGGTGC–3’) and the reference gene DNA–directed RNApolymerase II subunit RPB1 (Polr2a) cDNA (forward 5’–AGCTGGTCCTTCGAATCCGC–3’ and reverse 5’–TGGACTCAATGCATCGCAGGA–3’).Primers were designed to span an exon junction to prevent amplification ofgenomic deoxyribonucleic acid (DNA). Samples were amplified in duplicateusing the CFX96 Real–Time PCR Detection System (Bio–Rad, Hercules,52CA, USA) with the following cycling scheme: 2 min at 95 ◦C followed by 50cycles consisting of 5s denaturation at 95 ◦C, 5s annealing at 60 ◦C and 25selongation at 72 ◦C. Raw data were exported as text files and analyzed withthe qpcR package for R [243]. Amplification efficiencies and Cy0 crossingpoints [113] were calculated from a 5–parameter log–logistic model fittedto the raw fluorescence data to accommodate asymmetrical amplificationcurves [278]. Values from duplicated qPCR runs were averaged. Expressionratios were calculated from three biological replicates by the Pfaffl methodto correct for variation in polymerase chain reaction (PCR) efficiency [229]and normalized against the reference gene Polr2a. Spinal cord mRNA levelswere used as baseline to compare Panx2 expression across tissues. Prop-agation of error was estimated by a Monte Carlo simulation with 10000iterations as described in the qpcR package documentation.2.2.6 Protein isolation and Western blottingOrgans were quickly collected after transcardial perfusion with PBS, flashfrozen in liquid nitrogen and stored at −80 ◦C until needed. Tissues orcells were homogenized in radioimmunoprecipitation assay (RIPA) buffer(150mmol L−1 NaCl, 50mmol L−1 Tris–HCl pH 8.0, 0.5% Sarkosyl, 1%IGEPAL, 0.1% sodium dodecyl sulfate (SDS)) containing protease inhibitors(Pierce, Rockford, IL, USA) and phosphatase inhibitors (Sigma, St. Louis,MO, USA). Protein concentration was determined using a bicinchoninicacid (BCA) assay kit (Pierce, Rockford, IL, USA) and 50µg was separatedon 10% Tris–glycine SDS-PAGE gels containing 0.5% 2,2,2-trichloroethanol(TCE) (Sigma, St. Louis, MO, USA). Upon electrophoresis completion, pro-tein bands were visualized at 300 nm on an AlphaImager 3400 transillumina-tor (AlphaInnotech, San Leandro, CA, USA) as previously described [162]and electroblotted on NCL membrane (Bio–Rad, Hercules, CA, USA). Pro-tein bands on NCL were re–visualized under ultraviolet (UV) for quantifica-tion and total protein normalization [114]. For analysis which did not requirequantification, TCE was omitted and polyvinylidene fluoride (PVDF) mem-branes were used (Bio–Rad, Hercules, CA, USA). Membranes were blocked53at room temperature for 1h in milk solution (4% nonfat milk, 20mmol L−1Tris, 150mmol L−1 NaCl, pH7.4) and probed with primary antibodies at4 ◦C overnight followed by HRPO–conjugated secondary antibodies (Sigma,St. Louis, MO, USA) for 1h at room temperature. All antibodies were di-luted in blocking solution. HRPO activity was visualized with AmershamECL Prime Western Blotting Detection Reagent (GE Healthcare Life Sci-ences, Pittsburgh, PA, USA) or SuperSignal West Femto ChemiluminescentSubstrate (Thermo Scientific, Waltham, MA USA) and exposed on BioflexEcono films (Clonex, Markham, Ontario, Canada). Image acquisition forWestern blot quantification was done as previously described [99]. Briefly,film images were acquired on an AlphaImager 3400 (AlphaInnotech, San Le-andro, CA, USA) under stable transillumination and fitted with CCD cam-era lacking automatic gain control. Final 16–bit 1392 x 1040 pixel imageswere corrected for shading to compensate for non–homogenous illuminationand densitometry analysis was performed using the Image Studio Lite soft-ware (LI–COR, Lincoln, NE, USA). Panx2 protein ratios were calculated bydividing the band density of each tissue by the band density of the spinalcord. A tissue lysate from spinal cord was resolved on each gel to permitbetween gel comparisons. Panx2 protein ratios were calculated from threeindependent biological replicates.2.2.7 ImmunofluorescenceThree 3 to 5 month old mice were transcardially perfused with PBS followedby 10mL to 15mL 10% formalin (Fisher) or 4% paraformaldehyde (PFA)in PBS. Tissues were rapidly harvested and postfixed overnight at 4 ◦Cin the same fixative. Tissues were equilibrated at 4 ◦C in 30% sucrosein PBS containing 0.05% sodium azide for cryoprotection, embedded inTissue–Tek O.C.T. compound (Sakura Finetek, Torrance, CA, USA), frozen,cryosectioned at 10 µm and air–dried. Tissue sections were rehydrated in0.1mol L−1 phosphate buffer (PB) (pH 7.4), post–fixed with 4% PFA for10 min and washed twice with PB. Antigen retrieval was performed by in-cubating the sections in 10mmol L−1 sodium citrate (pH 8.5) at 80 ◦C for5430min. Sections were cooled to room temperature, washed with PB andtreated with 1% sodium borohydride in PB for 30 min. After several washesin PB, samples were blocked for 1h at room temperature with 5% goatserum in PB containing 0.2% Triton X-100 and incubated overnight at 4 ◦Cin primary antibody diluted in 1% goat serum in PB. Sections were thenwashed in PB and incubated for 1h at room temperature with the appro-priate AlexaFluor 488 or 568 secondary antibodies (Invitrogen, Carlsbad,CA, USA), followed by mounting in ProLong Gold antifade reagent with4’,6-diamidino-2-phenylindole (DAPI) (Invitrogen). Cells grown on cover-slips were simply fixed with 4% PFA for 15-20 min prior blocking and im-munostaining. Imaging was performed on a Leica TCS SP5 confocal micro-scope (Leica, Mannheim, Germany). When indicated, images were furtherprocessed with an iterative Lucy–Richardson deconvolution algorithm [313].All images were displayed as individual optical sections taken from a z–stackand not as maximal projection. All image acquisition and post–acquisitionprocessing steps were kept constant when comparing sections stained withanti–Panx2 antibodies and immunoglobulins from non–immunized animal.2.3 Results2.3.1 Characterization of two novel monoclonal antibodiesspecific for Panx2We initially characterized the selectivity of two novel anti–Panx2 mono-clonal antibodies (N121A/1 and N121A/31) by showing that both clonesidentified a single band of the proper size in C6 cells over–expressing ratPanx2 (Figure 2.1A). Importantly, none of the antibody cross–reacted withPanx1 (Figure 2.1A) which has been shown to be co–expressed with Panx2in the CNS [312]. Selectivity was further supported by an electrophoreticmobility assay showing that the band identified by both clones shifted by27 kDa when Panx2 C–terminal tail was tagged with green fluorescent pro-tein (GFP) (Figure 2.1A). Intriguingly, tagging Panx2 with GFP decreasedthe avidity of clone N121A/1. Using a library of overlapping peptides cov-55ering the amino acid sequence of Panx2, we mapped the epitope of cloneN121A/1 to the last 15 amino acids of Panx2 C–terminal tail (Figure 2.2).It is therefore likely that the addition of a GFP tag adjacent to the epitopecaused steric hindrance and reduced the avidity of clone N121A/1. We alsotested the antibodies by immunofluorescence and showed that the labellingfrom both clones overlapped with the GFP fluorescence signal emitted byPanx2–GFP but not by Panx1–GFP (Figure 2.1B).To test whether the novel antibodies could also be used on samples ex-pressing endogenous Panx2 protein we immunoprobed protein lysates pre-pared from eight tissues and separated by SDS-PAGE (Figure 2.3). Ourresults show that both clones recognized a band of approximately 70kDa cor-responding to the expected size of endogenous Panx2 as previously reported[332] (Figure 2.3). Overall our results identified two different commercialmonoclonal antibodies specific for Panx2. Both clones recognize rat andmouse Panx2 and we also successfully used clone N121A/1 to detect humanPanx2 (see Chapter 3). To avoid redundancy and because both clones wereequally specific we selected clone N121A/1 for subsequent analysis.2.3.2 Panx2 has a ubiquitous protein expression profileInterestingly our initial results showed substantial Panx2 protein amountoutside the nervous system (Figure 2.3). To further compare the Panx2 pro-tein expression profile of different tissues, we carried out semi–quantitativedensitometry analysis on protein lysates obtained from 16 tissues, separatedby SDS-PAGE and immunoprobed for Panx2 (Figure 2.4A). Because the ex-pression of common reference proteins was subjected to important fluctua-tions across tissues, a stain–free total protein normalization strategy was em-ployed to control for even loading as previously described [114, 162]. Briefly,we incorporated TCE in the gel formulation which, upon UV irradiation,catalyzes a covalent reaction with tryptophan residues. This reaction emitsfluorescence that can be imaged and documented in gel and following pro-tein transfer on membranes [114, 162]. The Panx2 staining density for eachtissue was then normalized against the intensity of the TCE–fluorescence56Figure 2.1. Specificity of two novel anti–Panx2 monoclonal antibodies. Twoanti–Panx2 monoclonal antibodies (clones N121A/1 and N121A/31) showed highspecificity both in Western blot (A, C) and immunolabeling (B). (A) Both clonesdetected a single band of the proper size in C6 cells over–expressing Panx2 but didnot cross–react with Panx1 or identify unspecific protein bands in wild–type C6cells. The molecular weight of the protein identified by both clones shifted by 27kDawhen Panx2 C–terminal was tagged with GFP (black arrowheads). Interestingly,the avidity of clone N121A/1 was drastically reduced when Panx2 was tagged withGFP. This is explained by the fact that clone N121A/1 recognizes an epitopeimmediately adjacent to the GFP tag (see Figure 2.2). (B) Immunolabeling signalfrom both clones co–localized with GFP fluorescence emitted from Panx2GFP butnot Panx1GFP when expressed in GFP cells. Immunoglobulin from non–immunizedmouse did not give any signal thereby ruling out the possibility of unspecific bindingof mouse IgG. Scale bars, 20 µm.57Figure 2.2. Clone N121A/1 recognizes an epitope located within the last 15amino acids of Panx2 C–terminal. (A) A library of 70 overlapping peptides span-ning the entire Panx2 amino acid sequence minus the four transmembrane domainswas dot blotted (A1 to J6) and immunoprobed with the N121A/1 anti–Panx2 an-tibody. A total of 25 µg of protein lysate from wild–type C6 and C6Panx2 gliomacells (K6 and L6 respectively) was also dot blotted alongside peptides to providenegative and positive control respectively. The clone N121A/1 specifically recog-nized C6Panx2 protein lysate and peptide J6 corresponding to the last 15 aminoacids of Panx2 C–terminal (TFEEPRTVVSTVEF). (B) Densitometry analysis ofthe dot blot staining shown in A. Only peptide J6 showed signal above threshold.Values were normalized against the density of the C6Panx2 sample. (C) The epi-tope identity was further confirmed by a blocking peptide assay. Pre–absorbing theN121A/1 anti–Panx2 antibody with a 200– or 1000–fold molar excess of peptideJ6 dramatically reduced immunolabeling while pre–absorption with a peptide ran-domly selected along the Panx2 amino acid sequence (peptide A3) did not alter thelabeling intensity. Scale bars = 20 µm.58Figure 2.3. The two anti–Panx2 antibodies detect endogenous Panx2. Bothclones were equally specific in Western blot performed with various tissue lysatesand identified a protein of ∼ 70kDa. The occasional detection of a ∼ 50kDa band insome tissues suggest that the antibodies might recognize Panx2 degradation prod-ucts or alternatively that the anti–mouse secondary antibody detects endogenousimmunoglobulin heavy chains.measured for the entire lane after protein transfer on NCL membrane (Fig-ure 2.4A). Our data indicate that Panx2 protein is present at substantiallevels in every tissue studied (Figure 2.4A,B). More surprisingly, in contra-diction with current predictions, Panx2 protein was lower in the nervoussystem than in any other tissues (Figure 2.4B).2.3.3 Transcript levels do not predict Panx2 protein levelsGene profiling studies have reported that Panx2 mRNA expression is largelyrestricted to the central nervous system in human [14], rat [42, 83] andzebrafish [36, 338] but a similar profiling study has not been completed inmouse. To determine whether our observations on Panx2 protein expressioncould be explained by species–specific Panx2 transcriptional activity, wecompared the Panx2 transcription profile using RNA isolated from 16 mousetissues, reverse–transcribed into cDNA and analyzed by qPCR. Althoughprimers were designed to span an exon junction, end–point PCR was initiallyperformed using non–transcribed RNA as template to confirm the absenceof genomic DNA amplification. We also tested the specificity of our primerpair by visualizing the amplification product by gel electrophoresis and byanalyzing the amplicon’s melting curve. Our results are in accordance with59Figure 2.4. Panx2 protein is ubiquitously expressed. (A) Panx2 protein lev-els were semi–quantified in 16 tissues using stain–free total protein quantificationto normalize protein levels across samples. Following exposure to TCE and UV,protein bands electroblotted on NCL were visualized by fluorescence (TCE, secondpanel) and loading normalization (LN) was performed by dividing the fluorescencedensity from an entire individual lane by the total fluorescent density measuredfrom the spinal cord lane on the corresponding membrane (LN, third panel). Nor-malized Panx2 protein ratios were expressed relative to Panx2 levels found in thespinal cord (fourth panel). (B) Panx2 protein ratios were calculated from 3 differ-ent mice (relative to spinal cord, red dots represent mean values). Panx2 proteinlevels were lower in the CNS than any other tissues.previous studies [14, 36, 42, 83, 338] and showed that Panx2 transcriptionalactivity largely predominates in the CNS (Figure 2.5A). Panx2 transcriptlevels detected in non–neural tissues were several orders of magnitude lowerthan in the CNS (Figure 2.5A). Furthermore, we demonstrated that thereis no significant correlation between Panx2 transcript and correspondingprotein levels (Figure 2.5B).Our results showed that Panx2 mRNA and protein levels are not corre-lated when compared across different tissues but does not exclude a possiblecorrelation within a specific tissue as opposed to between different tissues.This scenario appears unlikely however as we have shown that Panx2 pro-tein levels remain surprisingly constant in the brain over a developmentalperiod during which Panx2 mRNA levels have been shown to be tempo-rally up–regulated (Figure 2.6) [312]. Overall, these results suggest that60Figure 2.5. Panx2 transcriptional activity does not correlate with protein lev-els. (A) Panx2 mRNA levels were measured by qPCR in 16 tissues and expressedrelative to Panx2 mRNA levels found in the spinal cord. Error propagation wasestimated by a Monte Carlo simulation and data distribution represented using acombination of violin and box plots (red dots represent mean values). Data wereplotted on a logarithmic scale. Panx2 mRNA levels were several orders of magni-tude higher in the CNS than in any other tissues. (B) Panx2 protein and mRNAlevels were not correlated. No significant correlation was observed even after valuesfrom the CNS were eliminated (inset).regulatory mechanisms unrelated to transcriptional activity must also con-trol Panx2 protein levels and indicate that Panx2 protein levels cannot bedirectly inferred from the quantification of its transcript levels.2.3.4 Panx2 protein is localized to cytoplasmiccompartmentsWe next characterized the expression and distribution of Panx2 in differ-ent tissues by immunofluorescence. In the gastrointestinal tract, an im-portant population of glandular and epithelial cells displayed strong Panx2immunoreactivity (Figure 2.7). Parietal cells, which secrete gastric acid,and the apical surface of epithelial cells of the stomach were strongly re-active for Panx2 (Figure 2.7A). In the small and large intestine, a popula-tion of columnar epithelial cells were also strongly reactive for Panx2 (Fig-ure 2.7B,C). As Panx1 has also been shown to be expressed in the columnarepithelial cells of the human colon [77], we tested whether Panx1 and Panx261Figure 2.6. Expression of Panx1 and Panx2 proteins during brain development.The expression of Panx1 and Panx2 proteins is differently regulated during mousebrain development. While Panx1 expression decreases substantially approximatelytwo weeks after birth, Panx2 expression remains constant.could co–localize in these cell types. Interestingly, Panx1 and Panx2 didnot co–localize but showed quite different subcellular distribution patterns(Figure 2.7D). Panx1 expression was largely restricted to the plasma mem-brane between the epithelial cells. In opposition, Panx2 was not discernibleat the plasma membrane but remained largely confined to the cytoplasmicarea (Figure 2.7D).In the kidney, cuboidal cells forming the single layered epithelium oftubules were strongly labelled (Figure 2.8) whereas glomeruli cells were not(data not shown). Panx2 staining was predominantly cytoplasmic and couldnot be detected at the plasma membrane. Similarly, germ cells from testisseminiferous tubules showed abundant perinuclear and cytoplasmic but noplasma membrane staining (Figure 2.9).Panx2 immunoreactivity displayed a distinct pattern in the mouse retina(Figure 2.10). Photoreceptor inner segments protruding into the subretinal62Figure 2.7. Panx2 protein is expressed in the gastrointestinal tract. Sec-tions of mouse stomach (A), small intestine (B) and colon (C) were stainedwith an immunoglobulin from non–immunized mouse (Mouse IgG panels) or theN121A/1 monoclonal anti–Panx2 antibody (anti–Panx2 and Deconvolved/DICpanels). Panx2 protein was heavily expressed in the parietal (asterisk) and epithe-lial (arrow) cells of the stomach (A) and the epithelial cells of the small (B) and large(C) intestine. Panx2 distribution was primarily perinuclear and cytoplasmic. Nospecific staining was observed with the IgG from non–immunized mouse. Putativelymphocytes expressing endogenous immunoglobulins were occasionally labelled bythe anti–mouse secondary antibody (arrowheads). (D) Panx1 and Panx2 showeddistinct subcellular distribution in the colon. Scale bars: 20 µm.63Figure 2.8. Cuboidal cells from kidney tubule express Panx2 protein. Sections ofmouse kidney labelled with an immunoglobulin from non–immunized mouse (MouseIgG left panel) or the N121A/1 monoclonal anti–Panx2 antibody (middle and rightpanels). Panx2 protein was distributed in the cytoplasm of the epithelial cells liningthe lumen of renal tubule but was not discernible at the plasma membrane. Scalebars: 20 µ were densely decorated with Panx2–labelled aggregates (Figure 2.10).Only sparse immunoreactivity was observed in the outer nuclear layer whichforms the compact layer containing photoreceptor cell bodies (Figure 2.10).Substantial staining was also observed in the outer plexiform layer whichcomprises a dense network of neuronal synapses formed between photore-ceptors and bipolar and horizontal cell dendrites (Figure 2.10).Despite showing lower Panx2 protein levels than any other tissues (Fig-ure 2.4), Panx2 immunoreactivity was easily distinguishable by immunofluo-64Figure 2.9. Germ cells from mouse testis express Panx2 protein. Sections ofmouse testis were labelled with an immunoglobulin from non–immunized mouse(Mouse IgG panel) or the N121A/1 monoclonal anti–Panx2 antibody (anti–Panx2and Deconvolved/DIC panels). Panx2 protein was localized in the cytoplasm ofgerm cells in the seminiferous epithelium. Smooth muscle cells (arrow) and inter-stitial tissue (asterisk) showed high autofluorescence levels. Scale bars: 20 µm.65Figure 2.10. Panx2 protein is expressed in the photoreceptor inner segment andouter plexiform layer of the mouse retina. Sections from mouse retina were la-belled with an immunoglobulin from non–immunized mouse (Mouse IgG panel) orthe N121A/1 monoclonal anti–Panx2 antibody (anti–Panx2 and Deconvolved/DICpanels). Panx2 protein was primarily expressed in the photoreceptor IS (inset a)or the OPL (inset b). Panx2 clustered in small to large aggregates which ap-peared to be primarily cytoplasmic. However, because of the high degree of cellularcompaction we could not reach definitive conclusions regarding Panx2 subcellulardistribution in the retina. photoreceptor inner segment (IS), photoreceptor outersegment (OS), outer plexiform layer (OPL), outer nuclear layer (ONL), inner nu-clear layer (INL), inner plexiform layer (IPL), ganglion cell layer (GCL). Scalebars: 20 µm.rescence in the CNS (Figure 2.11) and showed a complex expression patternas previously reported [332]. Panx2 was widely distributed in the cytoplasmof neurons throughout the CNS but was not readily detected in astrocytesin vivo (Figure 2.11A,B,C). Interestingly, we showed that the majority ofprimary astrocytes (63.8 ±0.9%) expressed cytoplasmic Panx2 at 5 days invitro (Figure 2.11D). However, the percentage of Panx2–positive astrocytesrapidly declined after 10 and 15 days in vitro (6.9 ±1.2% and 7.3 ±1.1%)hereby suggesting that Panx2 is expressed by immature but not mature as-66Figure 2.11. Panx2 protein localizes in the cytoplasm of CNS neurons. Panx2heavily labelled the cytoplasm, but not the plasma membrane, of cortical neurons(a), Purkinje cells (b) and spinal cord motoneurons (c). Panx2 was not detected inGFAP–positive astrocytes in the brain or the spinal cord (c) but was detected inover 60% of primary astrocytes cultured for 5 days (d). Scale bars: 20 µm.trocytes. That observation could explain the up–regulation of Panx2 expres-sion seen in astrocytes following ischemia [332] as ischemia is characterizedby astrocyte proliferation.672.4 DiscussionThis study reveals that Panx2 protein is more ubiquitous than initially pre-dicted. By performing qPCR and semi–quantitative Western blot analysison a panel of mouse tissues, we showed that fluctuations in Panx2 mRNAabundance do not predict changes in Panx2 protein levels. We showed thatPanx2 protein levels are surprisingly more abundant in non–neural tissuesthan in the CNS; an observation opposite to Panx2 transcriptional activitywhich is weak in non–neural tissues and largely predominant in the CNS.The ubiquitous expression of Panx2 protein suggests a more fundamentalfunction than the CNS–specific role which was originally proposed. Al-though the exact function of Panx2 remains elusive, based on our in vivoimmunofluorescence results we hypothesize that Panx2 channels do not sig-nificantly contribute to communication exchange between the intracellularand extracellular spaces but rather control intracellular signaling throughcytoplasmic compartments.2.4.1 Absence of correlation between Panx2 mRNA andprotein levelsThe initial Panx2 gene expression profiles were obtained from Northern blotsusing commercial rat [42] and human [14] mRNA. Although notable dif-ferences exist between the two studies, both groups reported that Panx2mRNA is largely predominant in the CNS; an observation that was subse-quently confirmed in zebrafish using qPCR [36, 338]. Our results show thatPanx2 mRNA follows a similar expression profile in the mouse since Panx2transcript levels are 40 to over 1600 times higher in the CNS than in othertissues. Because of this dramatic disparity in Panx2 mRNA expression it haslong been assumed that Panx2 protein was preferentially, if not exclusively,expressed in the CNS.However, in almost every organism steady–state transcript concentra-tions only partially correlate with protein expression levels [1] and the as-sumption that transcripts can predict protein abundances has been heavilychallenged. Post–transcriptional regulatory mechanisms have overwhelming68influence on changes observed at the proteome level [93] and protein lev-els cannot be accurately extrapolated from transcript levels because severalfactors unrelated to transcriptional control also directly influence proteinlevels. For example, protein degradation rate has been shown to influencethe correlation between transcripts and corresponding protein levels as sta-ble proteins are less affected by perturbations in mRNA levels than proteinswith high turnover rates [238]. Hence, the long half–life of Panx proteins[224] could efficiently buffer important fluctuations in mRNA levels and de-crease the impact of Panx2 transcripts on Panx2 protein levels. Moreover,mRNAs can be stored in large ribonucleoprotein particles (RNPs) that areoften referred to as RNA granules [147]. These RNPs contain translationallyrepressed mRNAs and are essential for the transport of mRNAs in highlypolarized cells such as neurons [147]. Therefore, the discrepancy betweenPanx2 mRNA and protein levels could be explained by the fact that Panx2mRNAs are primarily contained non–translatable pools in neurons. Bearingthis information in mind, it is safe to affirm that variations of Panx2 tran-script levels should be interpreted restrictively, without assuming equivalentchanges at the protein level.Mass–spectrometry–based proteomics can perform large–scale unbiasedanalyses of biological systems and examine which genes are translated intoproteins in specific tissues. Recently, two groups assembled and publishedmass–spectrometry–based drafts of the human proteome into databases avail-able online for real–time analysis [150, 325]. Interestingly, unique Panx2peptides were identified in the ileum, colon and ovary [325] as well as thegut, spinal cord, urinary bladder, liver, ovary, testis and prostate [150]. Al-though substantial improvements are still needed to achieve a complete andquantitative proteome coverage, these independent studies nonetheless cor-roborate our results and demonstrate that Panx2 protein expression is notrestricted to the CNS.It is important to note that Panx2 protein ratios showed high variabilityin some tissues (Figure 2.4B). This is more likely attributable to the limiteddynamic range of the chemiluminescence technique that was used for thequantification of Panx2 protein expression. A better alternative would have69been to use a ratiometric analysis based on infrared detection of proteinbands to increase the linear detection range and reproducibility [334].2.4.2 Panx2: a cytoplasmic unusual suspectTechnical reasons such as prevalent autofluorescence prevented the analy-sis of certain tissues by immunofluorescence. Nonetheless, our study showsthat Panx2 protein was heavily distributed in the cytoplasmic compartmentand could not be readily detected at the plasma membrane in all tissuesanalyzed by immunofluorescence (9 out of 16) or in cultured primary astro-cytes expressing endogenous Panx2 protein. Previous studies had reportedcytoplasmic Panx2 in transfected overexpression system [26, 165, 225] or inneurons and neural progenitor cells [285, 332] but we are the first group toidentify endogenous cytoplasmic Panx2 in such a large variety of tissues.The unique intracellular distribution of Panx2 protein is in striking con-trast with Panx1 and Panx3 proteins which are primarily localized at theplasma membrane [26, 224, 225]. The cellular localization of Panx proteinsis influenced by glycosylation [225]. All three Panx paralogs are glycosy-lated to a high mannose form in the endoplasmic reticulum (ER) [26, 225]but interestingly only Panx1 and Panx3 proteins form complex glycopro-tein species requiring post–translational modifications occurring in the Golgi[26, 225]. Panx1 and Panx3 proteins follow a COPII–dependent endoplasmicreticulum to Golgi secretory pathway prior to being trafficked to the plasmamembrane [26]. In contrast, the absence of complex glycosylated Panx2 sug-gests Panx2 protein follows a different trafficking pathway which does notinvolve transition through the Golgi and subsequent trafficking to the plasmamembrane. Intriguingly, Panx2 has been shown to co–localize with the en-dolysosomal enriched mannose-6-phosphate receptor in N2a neuroblastomacells expressing Panx2 tagged with GFP [322]. However, the study usedtransient overexpression of tagged Panx2 protein which might have resultedin the accumulation of misfolded or misassembled proteins and increased thelikelihood of artifactual missorting in endomembrane compartments. There-fore, additional localization studies detecting endogenous Panx2 protein us-70ing a combination of different approaches are still needed for the accurateidentification of Panx2–positive cytoplasmic compartments.Others have detected putative Panx2 at the plasma membrane of matureprimary hippocampal neurons [285] or in cell types over–expressing Panx2[7]. The ectopic expression of Panx1 and Panx2 in NRK cells has also beenshown to increase Panx2 trafficking to the plasma membrane [225]. How-ever, the physiological relevance of this increase in Panx2 at the plasmamembrane is unclear because Panx1/Panx2 heteromeric channels are un-stable [7]. Although we cannot exclude the possibility that undetectablelevels of Panx2 are distributed at the plasma membrane in some cell types,we conclude that under physiological conditions Panx2 protein is primarilylocalized in the cytoplasmic compartment in most, if not all, tissues.Gap junctions have traditionally been described as plasma membranechannels connecting the cytoplasm of adjacent cells or controlling the ex-change of small molecules between the intracellular and extracellular spaces.Our results suggest that a different model must apply to Panx2 becauseits range of action seems to be restricted to the cytoplasmic milieu. Con-sequently, we hypothesize that Panx2 can modulate cell activity throughnon–conventional routes and novel intracellular signaling pathways. In thataspect, it is interesting to note that over–expression of Panx1 and Panx3 canform calcium permeable channels in the ER [133, 306]. As over–expressionof Panx2 in C6 cells showed a prominent signal overlap with the ER [165]it is possible that Panx2 can also modulate ER calcium signaling. More-over, although an essential property of gap junction proteins is their abilityto oligomerize to form transmembrane channels, it should be emphasizedthat gap junctions also have channel–independent functions. For example,Cx43 has recently been shown to control the biogenesis of autophagosomesthrough the sequestration of several autophagy–related proteins [21]; a func-tion independent of Cx43 channel activity. Moreover, our group and othershave shown that Cx43 is required for proper neuronal migration during thedevelopment of the neocortex and demonstrated that this role was indepen-dent of Cx43 channel activity [51, 86]. Since Panx2 has a long C–terminaltail (361 amino acids) it is reasonable to suggest that protein–protein inter-71actions involving its C–terminus are likely to play an important role in thefunction of Panx2. However, as the exact nature of Panx2 subcellular com-partment remains unknown, formulating hypothesis regarding the functionof Panx2 remains rather difficult.Our understanding of Panx2 protein currently assumes that Panx2 func-tion can be extrapolated from our knowledge of the other pannexin proteins.More precisely, Panx2 is often perceived as a CNS–specific protein assuminga role complementary, if not redundant, to the function of Panx1 channel.However, our study shows that this assumption is misleading and unlikelyto increase our knowledge on any of the pannexin channels. Prior to ourwork, several studies investigating the role of pannexin channels outside ofthe CNS focused exclusively on Panx1 and Panx3 but completely neglectedthe potential implication of Panx2. As our study shows that Panx2 proteinexpression is more ubiquitous than initially predicted it would be interest-ing to revisit these original studies while taking into account the presence ofPanx2. This is especially important in the context of Panx1 knockout micesince the deletion of Panx1 could have compensatory effects by altering theexpression level of Panx2. Another cautionary note needs to be highlightedregarding the techniques that are currently used to assay Panx2 functional-ity. Several studies use patch–clamping of the plasma membrane to addressthe functionality of Panx2 channels [16, 233]. Although we cannot totallyexclude the presence of Panx2 at the plasma membrane, our study nonethe-less shows that Panx2 protein is predominantly cytoplasmic. Consequently,Panx2 channel properties cannot be investigated through electrophysiologi-cal recordings at the plasma membrane.72Chapter 3The gap junction proteinpannexin 2 localizes atmitochondria contact sitesand sensitizes cells toapoptosis3.1 IntroductionMitochondria are organelles of endosymbiotic origin best known to producethe high energy content molecule ATP through respiration. Despite be-ing highly compartmentalized by a double–membrane, mitochondria haveacquired the ability to interact with other organelles during evolution andas a consequence, their behavior within eukaryotic cells has expanded wellbeyond energy production. Mitochondria are physically and functionallyassociated with the ER at contact sites called MAMs which are formedby the very close juxtaposition of ER membranes with the outer mito-chondrial membrane (OMM). These contact sites are important for theregulation of various cellular processes such as lipid synthesis, mitochon-73drial Ca2+ homeostasis, cellular respiration, and the initiation of cell death[69, 119, 171, 248, 304]. The very close apposition of mitochondria withER membranes allows efficient mitochondrial Ca2+ uptake despite the lowaffinity of mitochondrial Ca2+ transport machinery [246]. By being locatedin the immediate vicinity of ER Ca2+ channels, mitochondria are exposedto Ca2+ microdomains that largely exceed average cytosolic concentrationsthereby allowing substantial Ca2+ influx across mitochondrial membranes.Moderate Ca2+ influx increases mitochondria metabolism by stimulatingthe activity of mitochondrial enzymes and oxidative phosphorylation pro-tein complexes [106, 137, 139, 194, 217]. However, excessive ER Ca2+ releasechallenges mitochondrial homeostasis and initiates a cascade of events lead-ing to cell death [310]. Owing to their ability to control such crucial cellularprocesses, MAMs have been proposed to be involved in pathological con-ditions such as obesity, Alzheimer’s disease, stroke and Parkinson’s disease[9, 11, 112, 214, 261].Although a few proteins such as PACS–2, sigma–1R, MFN2, VDAC1,and IP3R have been shown to play important roles at ER–mitochondriacontact sites [68, 118, 270, 288], the molecular landscape of these contactsites still remains largely uncharacterized in mammalian cells. We recentlyshowed that the gap junction protein Panx2 is exclusively restricted tounidentified compartments of the endomembrane system in mammalian cells[170] (see also Chapter 2). Pannexins constitute a small family of membranechannels (Panx1, Panx2 and Panx3) homologous to the innexins, the inver-tebrate gap junction proteins [216]. However, unlike innexins and connexins,pannexins do not participate in direct cell–cell coupling but primarily formnon–junctional membrane channels [276]. Interestingly, Panx1 and Panx3channels has been shown to form Ca2+ leak channels in the ER [133, 306]thereby suggesting that Panx2 could also localize in the ER.In this study, we employed a combination of subcellular fractionation,particle tracking in live–cell, and immunogold electron microscopy and foundthat the gap junction protein Panx2 is enriched at contact sites linking ERand mitochondria. We demonstrate that Panx2 expression increases underconditions that stimulate cellular respiration. Furthermore, we also show74that over–expressing Panx2 in cancer cells increases their sensitivity to apop-totic signals while silencing it renders the cells more resistant to apoptosis.Overall, these findings identify a novel channel at the ER–mitochondria in-terface and a new player in the regulation of apoptosis.3.2 Materials and methods3.2.1 Animal careAll experiments were performed in accordance with the guidelines estab-lished by the Canadian Council on Animal Care and were approved by theUniversity of British Columbia Animal Care Committee3.2.2 AntibodiesThe anti–Panx2 mouse monoclonal antibody (cat# 75-212, clone N121A/1)was from UC Davis/NIH NeuroMab Facility (Davis, CA, USA) and wasused at 1 µgmL−1 to 2 µgmL−1 for immunofluorescence (IF), 2 µgmL−1 to5 µgmL−1 for immunogold and 2 µgmL−1 to 4 µgmL−1 for Western blot(WB). Purified immunoglobulin from non–immunized mouse was obtainedfrom Jackson Immunoresearch (cat# 015-000-003; West Grove, PA, USA)and was used at the same concentration as the anti–Panx2 antibody. Mousemonoclonal antibodies against EEA1 (cat# 610456, IF: 1/250, WB: 1/5000),p47a (cat# 610890, IF: 1/250), Rab4 (cat# 610888, IF: 1/250) and GM130(cat# 610822, IF: 1/250, WB: 1/1000) were all from BD Biosciences (Frank-lin Lakes, NJ, USA). The monoclonal antibodies against ATP synthase sub-unit α (cat# A21350, IF: 1/400, WB: 1/1000) and mitochondrial Hsp70(cat# MA3-028, IF: 1/400) were from Thermo Fisher Scientific (Rock-ford, IL, USA). The anti–sodium potassium ATPase antibody (cat# ab7671,WB: 1/5000) was from Abcam (Cambridge, United Kingdom). The anti–cleaved caspase–3 antibody was from Cell Signaling Technology (Danvers,MA, USA) (cat# 9661, WB: 1/1000). The anti–SKL antibody was kindlyprovided by Dr Richard Rachubinski from the University of Alberta andwas used at 1:250 for immunofluorescence. The anti–sec16 antibody was75kindly provided by Dr Ivan Robert Nabi from the University of BritishColumbia. The anti–GRP78/BiP antibody (cat# G8918, IF: 1/200, WB:1/3000), the anti–calnexin (cat# C4731, IF: 1/200, WB: 1/2000), andHRPO–conjugated goat secondary antibodies were obtained from Sigma (St.Louis, MO, USA). AlexaFluor488–conjugated goat anti–mouse IgG2a (cat#A21131), AlexaFluor 594–conjugated goat anti–mouse IgG3 (cat# A21155)and other AlexaFluor–conjugated secondary antibodies were from ThermoFisher Scientific and were used at 1/400. CF594–conjugated goat anti–mouse IgG2b secondary antibody (cat# 20269, IF: 1/400) was from Bi-otium (Hayward, CA, USA). The nanogold anti–mouse Fab’ (cat# 2002)secondary antibody was from Nanoprobes (Yaphank, NY, USA) and wasused at 1/ Plasmid constructionAn expression vector encoding a mitochondrial matrix–targeted mCherry(mito–mCherry) was generated by modifying a DsRed–mito plasmid [207]by restriction free cloning. The mCherry sequence was initially subclonedfrom a mCherry–miniSOG–N1 plasmid kindly provided by Dr Roger Tsienfrom the University of California, San Diego using the following forwardand reverse primers respectively: 5’–AGTCCAGAGTCAAGTACAGCTGGGATCCATGGTGAGCAAGGGCGAG–3’ and 5’–GGCCCTCTAGAGCGGCCGCTTACTTGTACAGCTCGTCCATGC–3’. The reaction productwas purified using the QIAquick gel extraction kit (Qiagen, Venlo, Limburg,Netherlands) and used in a secondary PCR reaction with the DsRed–mitoplasmid as template, thereby replacing the DsRed sequence by the mCherrygene. The mito–Turqoise construct was designed similarly. Forward 5’–CCAGAGTCAAGTACAGCTGGGATCCATGGTGAGCAAGGGCGAG–3’ and reverse 5’–TAGGGCCCTCTAGAGCGGCCGCTTACTTGTACAGCTCGTCCATGC–3’ primers were used to amplify the Turquoise2 se-quence from the pmTurquoise2–Golgi plasmid obtained from Addgene (cat#36205; Cambridge, MA, USA). The amplification product was subsequentlyused in a second PCR reaction with our mito–mCherry plasmid as tem-76plate thereby replacing the mCherry gene with the Turquoise2 sequence.The ER–targeted mCherry expression vector was also engineered by re-striction free cloning. A set of primers were used to amplify mCherry(forward: 5’–CTGGGCGCCGCCGCCGACATGGTGAGCAAGGGCGAG–3’, reverse: 5’–GATGGATATCTGCAGAATTCTTACAGCTCGTCCTTCTTGTACAGCTCGTCCATGCC–3’) and the reaction product wasused in a second PCR with the pcDNA3–D1ER plasmid from Dr RogerTsien as template. These steps essentially swapped the cameleon calciumsensor sequence for the mCherry gene while keeping the calreticulin ER–targeting and KDEL ER–retention signals. To label peroxisomes, a SKLtargeting signal was fused 3’ of the mCherry sequence by inverse PCR us-ing the following primers (forward: 5’–GCGACTCTAGATCATAATCAGCC–3’, reverse: 5’–TTATTATAATTTTGACTTGTACAGCTCGTCCATGC–3’) and the mCherry–miniSOG–N1 plasmid as template. The inversePCR reaction replaced the miniSOG sequence by the SKL targeting signal.All PCR reactions were carried out using Phusion high fidelity DNA poly-merase (Thermo Fisher Scientific). The two small hairpin RNA (shRNA)pLKO.1 delivery vectors against human Panx2 were obtained from Dharma-con (Lafayette, CO, USA) (TRCN0000150720, mature antisense 5’–AAATAAACTCTTCACCTCAGG–3’ and clone ID TRCN0000155858, mature anti-sense 5’–ATGAAGTTGCTCTTGCGGAAG–3’). Clones TRCN0000150720and TRCN0000155858 correspond to shRNA1 and shRNA5 respectively.The non–targeting shRNA pLKO.1 construct was from Addgene (cat# 1864).3.2.4 Cell cultureUnless otherwise stated, rat C6 glioma, C6 Panx1–EGFP, C6 Panx2–EGFP,SK–N–SH human neuroblastoma, A549 human lung adenocarcinoma, A172human glioblastoma and rat NRK cells were grown in DMEM (cat# D6429,Sigma) supplemented with 10% fetal bovine serum (FBS) and cultured at37oC and 5% CO2. To promote mitochondrial respiration, cells were grownin glucose–free DMEM (cat# 11966, Thermo Fisher Scientific) containing10mm D–(+)–galactose (cat# G0750, Sigma) and supplemented with 1mm77sodium pyruvate (cat# 11360-070, Thermo Fisher Scientific) and 10% dia-lyzed FBS (cat# 26400, Thermo Fisher Scientific). When comparing cellsgrown in regular and glucose–free conditions, all media were supplementedwith dialyzed FBS.3.2.5 Live–cell imaging and Panx2 trajectory analysisThe day prior imaging, C6 Panx2–EGFP cells plated on 35mm (No 1.5)glass bottom dishes (MakTek Corporation, Ashland, MA, USA) were trans-fected with ER–mCherry, mito–mCherry or mCherry–SKL constructs usingLipofectamine 2000 or Lipofectamine 3000 (Thermo Fisher Scientific) ac-cording to the manufacturer’s instructions. A few minutes before imaging,cells were labeled for 5 min with CellMask Deep Red plasma membranestain at 2.5 µgmL−1 to 5 µgmL−1 (Thermo Fisher Scientific). The GFPand mCherry channels from transfected cells were acquired simultaneouslyfor 1 min with a frame interval of 0.188s using a Leica TCS SP5 confocalmicroscope (Leica, Mannheim, Germany). The far red channel from thesame optical section was then captured to record the plasma membrane anddelimit the cytoplasm boundaries. Time–lapse images were processed withan iterative Tikhonov–Miller deconvolution algorithm [313] and the move-ment of Panx2 puncta was analyzed using the particle tracker plugin fromthe MosaicSuite for ImageJ [259]. In house software written in R was usedto randomize the distribution of Panx2 trajectories within the cytoplasmand to compute the minimal distance between the centroid of individualPanx2 puncta and mCherry–labeled organelles. The source codes can bedownloaded from GitHub ( Immunofluorescence and co–localization analysisAdult mice were transcardially perfused with 5mL of heparinized saline fol-lowed by 30mL of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1mPB. Organs were sliced in 2mm to 3mm thick sections, postfixed for 4hin 4% paraformaldehyde at 4 ◦C, equilibrated overnight in 30% sucrose in20mm PBS, embedded in Tissue–Tek O.C.T. (Sakura Finetek, Torrance,78CA, USA), frozen and cryosectioned at 10 µm. Cells grown on coverslipswere simply fixed with 4% paraformaldehyde and 0.1% glutaraldehyde in0.1m PB for 20 min at room temperature. Samples were washed in 20mmPBS and antigen retrieval was performed by incubation in 10mm sodiumcitrate (pH 8.5) pre–warmed at 80 ◦C for 30s (cells) or 2.5 minutes (tissuesections). Samples were then washed with PBS, treated with 0.1% sodiumborohydride in PBS for 30 min and further washed with PBS. Sampleswere blocked for 1h at room temperature with 5% goat serum or 5% bovineserum albumin (BSA) and incubated overnight at 4 ◦C in primary antibodydiluted in blocking solution. The samples were then washed in PBS andincubated for 1h at room temperature with secondary antibodies diluted inblocking solution, followed by mounting in ProLong Gold or ProLong Dia-mond antifade reagent with DAPI (Thermo Fisher Scientific). Imaging wasperformed on a Leica TCS SP5 confocal microscope. Images were furtherprocessed with an iterative Lucy–Richardson deconvolution algorithm [313].All images were displayed as individual optical sections taken from a z–stackand not as maximal projection. The Mander’s coefficients were calculatedusing the JACoP ImageJ plugin as previously described [33].3.2.7 Pre–embedding immunoelectron microscopyFor pre–embedding immunogold tissues were processed as indicated abovefor conventional immunofluorescence but sections were stored at −20 ◦C in30% glycerol, 30% ethylene glycol in 20mm PB following cryosectioning.Pre–embedding immunogold staining was performed as previously described[196]. Briefly, following the antigen retrieval and sodium borohydride treat-ments, sections were blocked for 1h in 5% BSA in 20mm PBS, incubatedwith the primary antibody diluted in blocking buffer for 1h at room tem-perature and washed with PBS. Sections were then blocked with 5% BSAfor an additional 30 min before incubating them with goat anti–mouse Fab’fragments conjugated with 1.4 nm gold particles diluted in blocking bufferfor 1h at room temperature. Sections were then washed in 1% BSA in20mm PBS followed by three additional washes in PBS. Sections were then79postfixed in 2% glutaraldehyde for 10 min and washed thoroughly in dis-tilled water. Gold particles were silver enhanced in the dark for 10 minusing the HQ Silver kit (Nanoprobes) according to the manufacturer’s in-structions. Sections were then washed thoroughly with distilled water andimmersed in 5% sodium thiosulfate for 3 min. Sections were then postfixedin 1% osmium tetraoxide : 1.5% potassium ferricyanide in 0.1m sodiumcacodylate buffer for 10 min and thoroughly washed with distilled water.Sections were then stained with 1% aqueous uranyl acetate for 5 min be-fore dehydration through an ascending concentration ethanol series and em-bedding in EMBED 812 resin (Electron Microscopy Sciences, Hatfield, PA,USA). Thin sections were cut on a Leica EM UC7 ultramicrotome (LeicaMicrosystems), collected on 200 mesh copper grids (Electron MicroscopySciences) and stained with uranyl acetate and lead citrate. Finally, sectionswere imaged on a Tecnai G2 Spirit electron microscope (FEI North AmericaNanoPort) operated at 120 kV.3.2.8 Subcellular fractionation and MAM isolationThe subcellular fractionation was performed as followed. Five 15cm con-fluent plates of C6 Panx2–EGFP cells were homogenized in 2mL of 0.25msucrose, 1mm EDTA and 10mm Hepes (pH 7.4) supplemented with proteaseinhibitors (Pierce, Rockford, IL, USA). The homogenate was centrifuged at1500g for 10min and the supernatant was brought to a final concentration of35% OptiPrep (Sigma Aldrich) using a 60% stock solution (60% OptiPrep,0.25m sucrose, 1mm EDTA and 10mm Hepes). Three milliters were loadedat the bottom of a tube and overlaid with 1ml of 30%, 20%, 17.5%, 15%,12.5%, 10%, 7.5%, 5%, and 2.5% of OptiPrep. The sample was then cen-trifuged at 200000g for 2.5h and 0.5mL fractions were collected and frozenuntil ready to use.MAMs and pure mitochondria stripped from their associated membraneswere isolated by subcellular fractionation as previously described [323]. Ap-proximately 0.5 g of mouse liver was homogenized in 225mm mannitol,75mm sucrose, 0.5% BSA, 0.5mm ethylene glycol tetraacetic acid (EGTA)80and 30mm Tris-HCl pH 7.4 and a crude mitochondrial pellet was obtained bydifferential centrifugation. The crude mitochondrial pellet was resuspendedin mitochondrial resuspension buffer (MRB) buffer (250mmmannitol, 5mm,HEPES (pH 7.4) and 0.5mm EGTA), overlayed on top of a Percoll medium(225mm mannitol, 25mm HEPES pH 7.4, 1mm EGTA, 30% Percoll) andcentrifuged at 95,000g at 4 ◦C for 30 min. Following centrifugation, theMAM fraction was identified as a diffuse white bands located above the mi-tochondrial pellet. The diffuse white bands and mitochondrial pellet werecollected separately, diluted ten times with MRB buffer, and centrifuged at6,300g for 10 min. The supernatant of the mitochondrial pellet was dis-carded and the pellet containing pure mitochondria collected. The MAMsupernatant was collected and centrifuged at 100,000g to obtain a MAM–enriched pellet.3.2.9 Western blottingTissues or cells were homogenized in RIPA buffer (150mm NaCl, 25mmTris–HCl pH 8.0, 0.5% Sarkosyl, 1% IGEPAL, 0.1% SDS) containing pro-tease inhibitors (Pierce, Rockford, IL, USA) and phosphatase inhibitors(Sigma, St. Louis, MO, USA). Protein concentration was determined us-ing a BCA assay kit (Pierce, Rockford, IL, USA) and 15 µg to 50µg wasseparated on 10% Tris–glycine SDS-PAGE gels before electroblotting onnitrocellulose or PVDF membrane (Bio–Rad, Hercules, CA, USA). Mem-branes were blocked in milk solution (4% nonfat milk, 20mm Tris, 150mmNaCl, pH7.4) and probed with primary antibodies at 4 ◦C overnight fol-lowed by HRPO–conjugated secondary antibodies (Sigma, St. Louis, MO,USA). All antibodies were diluted in blocking solution. HRPO activity wasvisualized with Amersham ECL Prime Western Blotting Detection Reagent(GE Healthcare Life Sciences, Pittsburgh, PA, USA) or SuperSignal WestFemto Chemiluminescent Substrate (Thermo Fisher Scientific) and exposedon Bioflex Econo films (Clonex, Markham, Ontario, Canada). Image ac-quisition for Western blot quantification was done as previously described[99]. Briefly, film images were acquired on an AlphaImager 3400 (AlphaIn-81notech, San Leandro, CA, USA) under stable transillumination and fittedwith CCD camera lacking automatic gain control. Final 16–bit 1392 x 1040pixel images were corrected for shading to compensate for non–homogenousillumination and densitometry analysis was performed using the Image Stu-dio Lite software (LI–COR, Lincoln, NE, USA).3.2.10 Apoptotic DNA fragmentation analysisOne million of cells were plated on 60mm dishes the day prior to the assay.Cells were then treated with DMSO (control) or 1 µm STS (cat# ab120056,Abcam) for 2 to 24h. At the end of the treatment, apoptotic DNA wasisolated as previously described [122]. Briefly, cells were resuspended in50 µL of lysis buffer (1% Igepal CA-630, 20mm EDTA, and 50mm Tris, pH7.5) and pelleted at 1600g for 5 min. The supernatant containing apoptoticDNA was collected and the extraction was repeated one more time. Thesupernatants were combined, brought to 1% SDS and treated with RNase A(final concentration 5 µg µL−1) at 56 ◦C for 2h. Samples were then treatedovernight with proteinase K (final concentration 2.5 µg µL−1) at 37 ◦C. Sam-ples were then acidified with 3m sodium acetate (pH 5.2)(final concentration0.3m) and apoptotic DNA fragments precipitated by adding 0.7 volume ofisopropanol. Finally, DNA fragments were separated by electrophoresis ona 1% agarose gel containing Sybr safe (Thermo Fisher Scientific) and visu-alized under UV on an AlphaImager 3400.3.3 Results3.3.1 Panx2 forms microdomains on ER membranesWe have previously reported that Panx2 is not expressed at the plasmamembrane like other gap junction proteins but distributes as puncta inmembrane–enclosed cytoplasmic compartments [165, 170] (see also Chap-ter 2). To study the subcellular distribution of Panx2 we initially per-formed a series of co–staining experiments but did not observe substantialco–localization with the markers tested (Figure 3.1). To gain additional82insight on Panx2 localization, we performed a protein profiling of a post-nuclear supernatant fractionated by centrifugation on a density gradient.Surprisingly, the distribution profile of Panx2 was positively correlated withthe profiles of the mitochondrial protein cytochrome C and the ER proteinGrp78/BiP (Figure 3.2A,B), indicating that Panx2 might interact, at leasttransiently, with these organelles.To test whether Panx2 associates with the ER, we analyzed Panx2 andER dynamics in C6 Panx2–EGFP cells transfected with an ER–targetedmCherry construct (ER–mCherry). We found that most Panx2 puncta lo-calized on the surface of the ER network (Figure 3.2C), thereby suggestingthat Panx2 forms microdomains on ER membranes. To further analyze theassociation between Panx2 puncta and the ER, we tracked the trajectoryof individual Panx2 punctum, measured the minimal distance between thecentroid of each Panx2 punctum and the surrounding ER network, and cal-culated the probability distribution of the nearest distances between Panx2puncta and the surrounding ER (Figure 3.2D). The bimodal distribution in-dicates that the majority of Panx2 puncta were directly associated with theER, thereby further reinforcing the idea that Panx2 localizes in ER microdo-mains. To determine whether this observation was simply a stochastic eventattributable to a densely packed environment, we compared the probabilitydensity function (PDF) of Panx2 puncta before and after the randomiza-tion of their trajectories within the cytoplasm. A much larger proportion ofPanx2 puncta was associated with the ER before randomization which con-firms the non–stochastic nature of the interaction (Figure 3.2D). We testedthe robustness of our interaction analysis by showing that the distributionof Panx2 puncta in relation to peroxisomes, an organelle that is not pre-dicted to interact with Panx2, did not substantially differ from a stochasticdistribution (Figure 3.3). The non–stochasticity of Panx2–ER associationwas also validated by comparing the cumulative distribution function beforeand after randomization of Panx2 trajectories. More Panx2 puncta werelocated within 100 nm of the ER prior randomization of their trajectoriesthan after randomization (70.82 ± 5.60 % prior randomization vs 46.95 ±6.75 % after randomization, p–value = 1.005 x 10-09, paired t-test, n = 13).83Figure 3.1. Panx2 does not co–localize with common organelle markers(continued)84Figure 3.1. Panx2 does not co–localize with common organelle markers (contin-ued). A: C6 cells expressing Panx2–EGFP were stained for various organelles butnone of the markers co–localized with Panx2. B: The Manders’ co–localization co-efficient was calculated in triplicate for each marker. A minimal coefficient value of0 corresponds to non–overlapping images while a maximal value of 1 represents per-fect overlap between both images. All Manders’ coefficient calculated were below0.1, indicating that Panx2 does not substantially co–localize with endosomal vesi-cles (EEA1 or Rab4), AP3–coated vesicles (p47A), mitochondria (ATP synthase α,mtHsp70), ER exit sites (sec16), peroxisomes (SKL) or the Golgi (GM130). Scalebars: 10 µm and 5 µm (insets).Finally, we showed that Panx2–ER association was very stable as Panx2microdomains and ER remained associated despite the high dynamism ofER tubules (Figure 3.2E). Overall, these results clearly indicate that Panx2form microdomains on ER membranes.3.3.2 Panx2 localizes at ER–mitochondria contact sitesAs Panx2 also co–fractionated with mitochondrial markers (Figure 3.2A,B), we transfected C6 Panx2–EGFP with a vector encoding a mitochondrialmatrix–targeted mCherry (mito–mCherry) to analyze Panx2 and mitochon-drial dynamics in live cells. We noticed that Panx2 puncta were foundin close proximity to mitochondria (Figure 3.4A) and confirmed the non–stochasticity of Panx2–mitochondria association by showing that a largerproportion of Panx2 puncta was associated with the mitochondrial networkprior randomization of Panx2 trajectories (Figure 3.4B). The association ofPanx2 with mitochondria was further confirmed by calculating the cumu-lative distribution function before and after randomization. Panx2 punctahad a significantly higher probability to be localized within 100 nm of a mi-tochondrion before randomization of their trajectories (21.74 ± 8.08 % priorrandomization vs 13.44 ± 3.58 % after randomization, p–value = 0.0008251,paired t-test, n = 17). Moreover, once established, the association betweenPanx2 puncta and mitochondria was very stable even though the mitochon-drial network was highly dynamic (Figure 3.4C).Mitochondria and ER are physically and functionally associated at con-85Figure 3.2. Panx2 clusters in ER microdomains. A: Panx2 co–fractionates withER and mitochondrial markers. Post–nuclear supernatant from C6 Panx2–EGFPcells was fractionated by ultracentrifugation on a discontinous iodixanol gradi-ent and twenty–four fractions were collected and immunoprobed for various or-ganelle markers. A substantial amount of Panx2 co–fractionated with mitochon-drial (VDAC1) and ER (Grp78/BiP, calreticulin) markers. B: Three additionalfractionations were performed but fractions were paired to allow the probing of allfractions on a single membrane for quantification. Panx2 distribution positivelycorrelated (Pearson’s correlation) with the mitochondrial marker cytochrome C (r= 0.633, p = 0.0273) and the ER protein Grp78/BiP (r = 0.76, p = 0.0041) butnot with the Golgi marker GM130 (r = -0.407, p = 0.1892). C: C6 Panx2–EGFPcells were transfected with an ER–targeted mCherry construct and trajectories ofPanx2 puncta were tracked in living cells. A large majority of Panx2 puncta werelocated on the ER network as indicated by the white arrowheads (inset). Scalebars: 10 µm and 5 µm (inset).(continued)86Figure 3.2. Panx2 clusters in ER microdomains (continued). D: A PDF of thenearest distances between the centroid of Panx2 puncta and the ER was computed(orange curve) and compared to the PDF computed after randomization of Panx2trajectories within the cytoplasm (purple curve). The PDFs, calculated from 13cells, show that a larger proportion of Panx2 puncta were localized on the ER net-work before randomization. E: Frame sequence showing that Panx2–ER associationwas stable for over 28s despite the high mobility of the ER network. Panx2–ERassociation was even maintained in the last two frames (e11 and e12) despite sub-stantial movements of the ER tubule. For reference, the initial position of thePanx2 puncta in frame e1 is also shown in all frames (white arrowheads). Scalebar, 2 µm.Figure 3.3. Panx2 does not associate with peroxisomes. A: C6 Panx2–EGFPwere transfected with a vector encoding a peroxisome–targeted mCherry (mCherry–SKL). A small number of Panx2 puncta can be found in the proximity of peroxi-somes (arrowheads in inset) but this association was shown to be random. Scalebars: 10 µm and 5 µm (inset). B: The trajectory of individual Panx2 punctumwas used to calculate a probability density function showing the distribution ofthe nearest distances between Panx2 puncta and the surrounding peroxisomes (or-ange curve). This distribution was compared to the probability density functionobtained after randomizing Panx2 puncta trajectories (purple curve). This com-parison shows that the probability distribution of Panx2 puncta within 100 nm ofa peroxisome does not substantially differ from the random distribution, indicatingthat the association of some Panx2 puncta with peroxisomes is stochastic.tact sites called mitochondria–associated ER membranes (MAMs). SincePanx2 forms microdomains on ER membranes and because these microdo-mains are located near mitochondria, we naturally tested whether Panx2was found at ER–mitochondria contact sites. We initially confirmed thatendogenous Panx2 also clustered on ER microdomains in vivo (Figure 3.5A).Additionally, we showed that most Panx2 microdomains partially overlappedwith mitochondria staining, indicating that Panx2 is strongly associated87Figure 3.4. Panx2 puncta associate with mitochondria. A: The dynamic of Panx2puncta was analyzed in C6 Panx2–EGFP cells transfected with a mitochondria–targeted mCherry construct. A large number of Panx2 puncta were closely associ-ated with the mitochondrial network as indicated by the white arrowheads (inset).Scale bar, 10 µm. B: A PDF of the nearest distances between Panx2 puncta andmitochondria was calculated before and after randomization of Panx2 trajectories(orange and purple curves respectively). The PDFs, computed from 17 cells, showthat Panx2 puncta were not randomly distributed within the cytoplasm but pref-erentially localized in the proximity of mitochondria. C: Frame sequence showingthat Panx2 association with mitochondria is stable even though the mitochondrialnetwork was highly dynamic. A Panx2 puncta approaching a mitochondrion inframe c1 established contact with the organelle in frames c2 and c3. The associ-ation was briefly lost (frame c4) but stably re-established for over 18s from framec5 onward. For reference, the initial position of the Panx2 puncta in frame c1 isshown in all frames (white arrowheads). Scale bar, 2 µm.88with mitochondria in vivo (Figure 3.5B). This overlap was not caused bycross–reactivity (Figure 3.6). We then triple stained mouse brain sectionsfor Panx2, ER, and mitochondria and showed that an important numberof Panx2 puncta simultaneously contacted the ER and mitochondrial net-works in neurons (Figure 3.5C), indicating that Panx2 localizes at the ER–mitochondria interface in vivo. We also confirmed that endogenous Panx2localized at ER–mitochondria contact sites in cell culture (Figure 3.5C).To further confirm the presence of Panx2 at ER–mitochondria contactsites, we isolated MAMs from mouse liver as previously described [323]. Weopted for this organ because it is an abundant and homogeneous sourceof mitochondria and because we have previously reported high Panx2 ex-pression in mouse liver [170] (see also Figure 2.4 in Chapter 2). Panx2 waspelleted alongside mitochondria in the crude mitochondrial fraction and wasfurther enriched in the MAM fraction (Figure 3.7A,B). The ER chaperoneGrp78/BiP and the long-chain acyl–CoA synthetase 4 (ACSL4) protein, anenzyme involved in lipid biosynthesis, were also enriched in the MAM frac-tion as previously reported [118, 323]. Interestingly, the outer mitochondrialmembrane (OMM) protein VDAC1 was also detected at low levels in theMAMs (Figure 3.7A) while the inner mitochondrial membrane protein ATPsynthase α was absent from the isolated MAMs (Figure 3.7B). As VDAC1is physically linked to IP3Rs located on ER membranes [288], we suggestthat the molecular complexes coupling both organelles remain intact duringthe isolation procedure and strip some of the outer membrane from mito-chondria. Panx2 was not found in pure mitochondria stripped from theirassociated membranes (Figure 3.7), thereby confirming that Panx2 is foundon ER membranes that associate with mitochondria but is not a mitochon-drial protein.The observation of mouse brain by regular transmission electron mi-croscopy demonstrates that MAMs are formed by the very close appositionof the ER membrane and the outer mitochondrial membrane (OMM) (Fig-ure 3.8A). Since the intermembrane distance is below the resolution of con-ventional confocal microscopy we used pre–embedded immunogold electronmicroscopy to study the localization of Panx2 at the ultrastructural level89Figure 3.5. Panx2 localizes in ER microdomains associated with mitochondriain vivo. A: Brain sections were stained for Panx2 and the ER marker calnexin.Panx2 formed discrete puncta in neurons that primarily clustered in ER microdo-mains as indicated by the white arrowheads (inset). B: Brain sections were stainedfor Panx2 and ATP synthase α, a subunit of mitochondrial complex V. A largeproportion of Panx2 staining overlapped with brain mitochondria in neurons as in-dicated by the white arrowheads (inset). C: Mouse brain sections were co–stainedfor Panx2, the ER marker calnexin and the mitochondrial protein ATP synthase α.Panx2 is found on ER microdomains that are closely associated with mitochondriain vivo as indicated by the white arrowheads. Scale bars: 10 µm. D: Human neurob-lastoma SK-N-SH cells were co-transfected with ER–mCherry and mito–Turquoiseconstructs and stained for Panx2. An important proportion of Panx2 clusters weresimultaneously associated with the ER and mitochondrial networks as indicated bythe white arrowheads. Scale bar: 10 µm90Figure 3.6. Absence of cross–reactivity between subclass specific anti–mousesecondary antibodies. Mouse brain sections were co–stained for Panx2 and themitochondrial protein ATP synthase α (A, top panel) or for Panx2 and the mi-tochondrial protein mtHsp70 (B, top panel). Omission of the primary antibodyagainst mitochondrial proteins shows that the anti–mouse IgG2b (A, middle panel)and the anti–mouse IgG3 (B, middle panel) secondary antibodies do not label themouse IgG2a anti–Panx2 primary antibody. Omission of the primary antibodyagainst Panx2 shows that the anti–mouse IgG2a secondary antibody does not la-bel the mouse IgG2b anti–ATP synthase α (A, bottom panel) or the mouse IgG3anti–mtHsp70 (B, bottom panel) primary antibodies. Scale bars: 10 µmin mouse brain (Figure 3.8B–D). Our staining shows that the majority ofgold clusters were found in very close apposition with the OMM at ER–mitochondria interfaces (Figure 3.8B). Because the length of the primaryantibody–Fab’ nanogold complex (8 + 6 = 14 nm) can be longer than thegap that separates ER and mitochondrial membranes (Figure 3.8A), goldparticles sometimes appeared to be positioned on mitochondrial membranes(Figure 3.8B). However, when ER and mitochondrial membranes could besuccessfully resolved, the gold particles were shown to be located in thenarrow cytoplasmic cleft separating both organelles (Figure 3.8C). More-over, consistent with our confocal images in Figure 3.2, gold clusters not91Figure 3.7. Panx2 is enriched in purified MAMs. A: MAMs and pure mitochon-dria were isolated from mouse liver by ultracentrifugation on a Percoll gradientand immunoprobed with different markers. Panx2 is found in the crude mitochon-drial and MAM fractions but not the pure mitochondrial fraction. Purified MAMswere also enriched in Grp78/BiP and ACSL4. B: Isolated MAMs showing a strongenrichment in Panx2 are devoid of inner mitochondrial membrane contaminant asshown by the absence of ATP synthase α.juxtaposed to mitochondria were located on ER membranes (Figure 3.8D),confirming that Panx2 is located in ER microdomains and not on mitochon-drial membranes.3.3.3 Panx2 expression is modulated by the mitochondrialrespiratory statusMitochondrial respiratory activity can modulate ER–mitochondria interac-tion. For example, MAMs associate more strongly with energized than non-energized mitochondria [199]. As cancer cells grown on glucose rely pri-marily on aerobic glycolysis for energy production even in conditions thatsupport mitochondrial respiratory chain [307], it is reasonable to assumethat the association of MAMs with mitochondria can be reduced in cancercells. Interestingly, Panx2 expression is also substantially reduced in cancercells when compared to non–transformed cells [165] (see also Figure 1.5 inChapter 1). As Panx2 localizes in MAMs, we predicted that Panx2 expres-sion would increase upon enhancing mitochondrial respiration status. Totest this hypothesis, we maximized mitochondrial respiration by growingcells on galactose, a carbon source that requires mitochondrial respiration92Figure 3.8. Panx2 localizes at ER–mitochondria contact sites(continued)93Figure 3.8. Panx2 localizes at ER–mitochondria contact sites (continued). A:Representative electron micrograph of a mouse brain showing that the gap betweenER and mitochondria is often imperceptible at ER–mitochondria contact site. B:Representative TEM micrograph of a mouse brain cortex showing the localizationof Panx2 by immunogold. Most gold clusters are found in close association with mi-tochondria (open arrowheads). Gold clusters not located near mitochondria (solidarrowheads) are located on ER membranes as seen in D. C: Representative exam-ple of an immunogold staining showing the presence of Panx2 at ER–mitochondriacontact sites (solid red arrowhead in left panel). Gold particles are located in thenarrow gap which defines the cytoplasmic region between ER and mitochondrialmembranes. D: Representative TEM image of a mouse brain cortex showing thelocalization of Panx2 (red arrowhead) on ER membranes by immunogold staining.The primary antibody used recognizes an epitope within the last 15 amino acidsof Panx2 C–terminal tail. The gold particles are located in the cytoplasm and notinside the ER lumen which is consistent with the predicted localization of Panx2 C–terminal tail. E-F: Representative electron micrographs illustrating the absence ofgold particles when the anti–Panx2 primary antibody was replaced by an antibodyfrom a non–immunized mouse. The localization of the ER membranes at MAMs isindicated by the open red arrowheads in C and F. ER tubules were pseudocoloredin cyan (A, D) and mitochondria in purple(A,B,D,E). Scale bars: 500nm (A), 1 µm(B,E), 100 nm (C,F) and 200 nm (D).for energy production [3, 144, 190]. As predicted, substituting glucose forgalactose induced a dramatic increase in Panx2 protein expression in allcancer cell types tested (Figure 3.9). These results indicate that an increasein mitochondria energy conversion potential requires more Panx2 channels,most probably as a result of higher recruitment of MAMs to mitochondria.Surprisingly, Panx2 expression is also up–regulated upon inhibition of theETC (Figure 3.10). Prolonged inhibition of the ETC complex I or com-plex IV with rotenone or cyanide respectively induced a dose–dependentincrease in Panx2 protein expression thereby suggesting that inhibiting mi-tochondrial respiration induces an ATP deficit and triggers a compensatoryup–regulation of Panx2.3.3.4 Panx2 sensitizes cell to apoptosisIn addition to regulating mitochondrial respiration, ER–mitochondria con-tact sites control apoptotic cell death [64]. To determine whether Panx294Figure 3.9. Mitochondrial respiration influences Panx2 expression. A: SK-N-SH,A549, A172 and NRK cells were grown in regular DMEM (GLU: 25mm glucose)or glucose–free medium (GAL: 10mm galactose). Panx2 expression was minimalin high glucose medium but increased dramatically in galactose conditions. B:Growing A549 cells in galactose medium induced a significant 3.5 fold increase inPanx2 protein expression when compared to cells grown in glucose medium (n =4, p = 0.03321). Error bars represent the standard error of the mean.95Figure 3.10. Panx2 expression increases upon inhibition of mitochondrial respi-ration. A: Rotenone and KCN respectively inhibit complex I and IV of the ETCin the mitochondria. B: Human SK-N-SH cells grown in regular DMEM (25mmglucose) were treated with increasing doses of KCN for 24h. KCN induced a dose–dependent increase in Panx2 protein expression. C: Similarly, rotenone also induceda dose–dependent increase in Panx2 protein expression. Error bars represent thestandard error of the mean.96might be involved in the induction of apoptosis, we tested the sensitivity ofcells to STS–induced apoptosis following manipulation of Panx2 expression(Figure 3.11). Interestingly, C6 rat glioma cells over–expressing Panx2 werehighly susceptible to apoptosis and showed activation of caspase–3, a lateexecutioner of apoptotic program, and DNA fragmentation starting 4h afterSTS treatment. In comparison, C6 wildtype or C6-Panx1EGFP cells, whichdo not express significant amount of endogenous Panx2 (see Figure 2.1 inChapter 2), did not show substantial caspase–3 activation or DNA fragmen-tation within the first 12h of STS treatment (Figure 3.11A,B). Interestingly,while C6 wildtype cells started to show sign of DNA fragmentation 24h fol-lowing STS treatment, C6-Panx1EGFP cells did not show any obvious signof apoptosis during the entire course of the experiment (Figure 3.11A,B). Itshould be pointed out however that most C6-Panx1EGFP cells were roundedup and floating towards the end of the experiment and therefore presumablydead. It is therefore possible that C6-Panx1EGFP cells preferentially diedfrom necrosis rather than apoptosis following STS treatment.To test whether the up–regulation of Panx2 expression following an in-crease in mitochondrial respiration could sensitize cells to apoptosis, wetreated A549 cells grew on galactose to STS treatment. A549 cells were ei-ther transfected with an anti–Panx2 shRNA delivery vector which reducedPanx2 expression by up to 90% or with a non–targeting shRNA construct(Figure 3.11C). Silencing Panx2 expression substantially decreased A549cells’ sensitivity to apoptosis (Figure 3.11D,E), thereby reinforcing the ideathat Panx2 localization at MAMs is involved in the induction of programmedcell death.97Figure 3.11. Panx2 sensitizes cells to apoptosis. A: C6 wildtype (WT), C6–Panx1EGFP and C6 Panx2–EGFP glioma cells were treated for 2h to 12h with 1 µmSTS and cell lysates analyzed by immunoblotting for cleaved caspase–3. C6 Panx2–EGFP, but not C6WT and C6–Panx1EGFP, showed strong activation of caspase–3starting 4h after STS treatment. B: Concomitant to the activation of caspase–3,cells expressing Panx2 also showed evidence of DNA fragmentation as soon as 4hafter STS treatment. In contrast, C6WT did not show sign of DNA fragmentationuntil 24h after treatment. C6–Panx1EGFP did not show noticeable sign of DNAfragmentation during the period surveyed. C: A549 cells were transfected withnon–targeting or anti–Panx2 shRNA constructs. shRNA1 and shRNA5 silencedPanx2 expression by approximately 70% and 90% respectively. Silencing Panx2expression protected cells from STS–induced apoptosis as shown by the absenceof caspase–3 activation (D) and DNA fragmentation (E) in A549 cells transfectedwith anti–Panx2 shRNA5.983.4 DiscussionIn the present study, using a combination of subcellular fractionation, par-ticle tracking in live–cell and immunogold labeling in tissues we report thatthe ubiquitous gap junction protein Panx2 localizes on ER membranes andclusters in microdomains that are tightly associated with mitochondria. Weshow that Panx2 protein expression is modulated by mitochondrial respira-tion activity and increases upon mitochondrial energization. Moreover, thismodulation of Panx2 expression by mitochondrial activity reveals an unex-pected but important role for Panx2 in the induction of apoptotic programsfollowing cellular challenges.Panx2 localization at MAMsSince its discovery in 2000, a number of studies have demonstrated the atyp-ical distribution of Panx2 channels within the cell. Our group and severalothers have reported that Panx2 does not localize at the plasma membranelike other gap junction proteins but primarily localizes in membrane–boundcompartments within the cytoplasm [31, 165, 170, 322, 332]. It has beensuggested that palmitoylation sequesters Panx2 within the Golgi and theER in immature neurons while the depalmitoylated form of Panx2 found inmature neurons is allowed to traffic to the plasma membrane [285]. However,our results do not support this idea. We studied the subcellular localizationof endogenous Panx2 at nanometer resolution in the adult brain and didnot observe Panx2 staining at the plasma membrane. On the contrary, ourresults show that Panx2 distributes in clusters located on ER membranesand that most of these clusters are directly associated with mitochondria inmature and fully differentiated neurons.More recently, endolysosomal targeting sequences and endocytic recog-nition sequences have been identified in all three pannexin paralogs usingbioinformatics tools designed to predict eukaryotic linear motifs [38]. Al-though the biological relevance of these sequences have yet to be explored,a few reports have nonetheless supported these in silico data and suggestedthat Panx2 can localize in endolysosomal vesicles in over–expressing sys-99tems [31, 322]. For example, Boassa and colleagues have reported signif-icant overlap between exogenous Panx2 and the early endosomal proteinsp47A and Rab4 [31]. However, we could not reproduce these observationsin our laboratory (see Figure 3.1). The adapter protein p47A is a sub-unit of the adaptor protein complex AP–3, which facilitates the buddingof vesicles from Golgi membranes and their sorting to the endolysosomalsystem [72, 212]. However, Panx2 localization in p47A–containing vesiclesis improbable under normal conditions since Panx2 is unlikely to transitthrough the Golgi. Although all three pannexin paralogs get N–glycosylatedinto high mannose species in the ER [29, 224, 225], only Panx1 and Panx3proteins are processed into complex N–glycoproteins within the Golgi be-fore being secreted to the plasma membrane [29, 224, 225]. In contrast toPanx1 and Panx3, Panx2 is not processed into complex N–glycoproteins[225], suggesting that it does not transit through the Golgi apparatus butmost likely remains within the ER. Boassa and colleagues also performedelectron microscopy on cells over–expressing Panx2 C-terminally tagged witheither miniSOG or tetracysteine residues to study the localization of Panx2by electron microscopy [31]. Their staining revealed intracellular localiza-tion of Panx2 in ’tubulo–vesicular membranous compartments’ that wereidentified as part of the endolysosomal system [31]. However, the authorsadmittedly recognized the existence of some discrepancies between their im-munofluorescence and electron microscopy results as double immunolabelingrevealed significant overlap between calnexin and Panx2 stainings, suggest-ing that Panx2 localized in the ER [31]. Our results clearly demonstratesthat Panx2 primarily forms microdomains on ER membranes that are closelyassociated with mitochondria. As autophagosomes have been shown to format ER–mitochondria contact sites [116], we cannot exclude the possibilitythat some Panx2–containing membranes will eventually be trafficked to theendolysosomal systems. However, we have not detected Panx2 in endolyso-somal compartments by immunogold labeling of endogenous Panx2 in tissueswhich suggests that this event would be rare under physiological conditions.Therefore, the prominent localization of exogenous Panx2 in endolysosomalcompartments as reported by Boassa and colleagues might have been caused100by missorting owing to the over–expression of C–terminally tagged Panx2[31].Biological relevance of Panx2 at MAMsThe localization of Panx2 channels at MAMs might facilitate communica-tion and promote the exchange of ions and small molecules between ER andmitochondria. It is well established that the close apposition of ER andmitochondrial membranes at MAMs allows the propagation of Ca2+ signalsfrom the ER to mitochondria [64, 246]. Under physiological conditions, amoderate uptake of Ca2+ by mitochondria increases respiration by stimu-lating the activity of mitochondrial enzymes and oxidative phosphorylationprotein complexes [106, 137, 139, 194, 217]. However, cellular challenges caninduce excessive ER Ca2+ release which compromises mitochondrial Ca2+homeostasis and initiate apoptotic programs [64, 310]. The IP3Rs and ryan-odine receptors (RyRs) are two major ER Ca2+ channels which have beenshown to influence mitochondrial Ca2+ uptake [152, 197]. However, whenco–expressed in the same cell, both receptors segregate into different ER do-mains [149, 265]. This segregation organizes ER Ca2+ stores into spatiallydistinct compartments functioning as discrete units [109]. This physicaland functional segregation of ER Ca2+ channels suggests that the regu-lation of Ca2+ dynamics at ER–mitochondria contact sites is complex andinvolves different ER microdomains containing distinct Ca2+ channels. Thisidea is supported by studies showing that silencing IP3Rs in cells that donot express RyRs only partially reduced mitochondrial Ca2+ uptake [197],indicating that additional channels must also regulate Ca2+ exchange atthe ER–mitochondria interface. Interestingly, previous reports have demon-strated that Panx1 and Panx3 can form Ca2+–permeable ER channels inover–expressing systems [133, 306], thereby suggesting that Panx2 channelsmight be involved in Ca2+ exchange between the ER and mitochondria. Ourresults showing an up–regulation of Panx2 expression upon mitochondrialenergization and enhanced resistance to apoptotic challenges after silencingPanx2 expression support this idea.101However, we cannot exclude the possibility that Panx2 might also regu-late apoptotic programs independently of its channel activity. The associa-tion of the ER with mitochondria is dynamically regulated and can influencethe cellular response to apoptotic signals. Molecular tethers that couple ERand mitochondria display some diversity in length which influences the dis-tance between both organelles [64]. A reduction of the interorganellar junc-tion has been shown to increase the functional association between the ERand mitochondria and increase mitochondrial Ca2+ uptake [64]. Under nor-mal conditions, the coupling between both organelles stimulates mitochon-drial respiration by allowing moderate increase in Ca2+ uptake. However,cellular challenges can induce a tightening of ER–mitochondria coupling andreduce the gap between both organelles. This can lead to excessive mito-chondrial Ca2+ uptake, mitochondrial permeabilization and apoptosis [64].Therefore, Panx2 could facilitate the induction of apoptotic programs byreducing the gap and enhancing the coupling between the ER and mito-chondria independently of its channel activity. Panx2 is characterized by along and highly disordered C–terminal tail (> 360 amino acids) and it is notunreasonable to assume that Panx2 could directly or indirectly interact withmitochondrial proteins and promote the tethering of ER to mitochondria.We are currently testing this idea by studying ER–mitochondria coupling inPanx2 knockdown cells.102Chapter 4General discussion andconcluding remarksWhile the number of studies investigating the biological properties of Panx1and Panx3 paralogs have accumulated considerably since the discovery ofpannexin genes in 2000, technical limitations and conceptual challenges havelimited our understanding of the Panx2 protein. The work discussed in thisthesis overcame some of these limitations and challenges and discoverednovel findings about the expression, localization and biological function ofPanx2. We now know that Panx2 protein is ubiquitously expressed andnot limited to the nervous system as previously suggested by transcriptionalexpression analysis. We can also confidently assert that unlike other gapjunctions, the Panx2 channel is primarily, if not exclusively, localized in cy-toplasmic compartments. We have strong evidence indicating that Panx2is not expressed at the plasma membrane but is found in ER microdo-mains localized at contact sites formed by the close juxtaposition of ERand mitochondrial membranes. Finally, we have a fuller understanding ofthe biological function of Panx2 as we now know that Panx2 expression ismodulated by mitochondrial respiratory activity and can regulate apoptosis.These findings have been presented in much details in the preceding chap-ters but some important questions have not been directly addressed yet. Inthis concluding chapter, the main findings will be critically revisited while103focusing our attention on those questions and leading the discussion towardstentative answers.4.1 Beyond the transcript: What controls Panx2protein levels?As we have seen in Chapter 2, Panx2 transcriptional activity is a poor predic-tor of Panx2 protein abundance and does not correlate with Panx2 proteinlevels. Albeit interesting in its own right, this observation is not uniqueto Panx2. Since 2004, several independent studies have strongly refutedthe assumption that protein levels can be deducted by measuring mRNAconcentrations [186, 262, 297, 311, 325]. In fact, most studies reportedthat only about 40% of the variance in protein levels can be explained bymRNA concentrations in mammalian cells or tissues. Interestingly, five outof six RNA–processing genes analyzed by Tian and colleagues even showeda negative correlation between mRNA and protein levels, suggesting thatposttranscriptional regulation could constitute a conserved mechanism thatregulates the expression of specific class of genes [297]. Changes in mRNAlevels induced by diseases have also been shown to poorly correlate with pro-tein levels. For example, Taquet and colleagues reported a > 410% increasein somatostatin receptor 5 (SSTR5) mRNA expression without a significantchange in protein levels in the peripheral blood mononuclear cells of Crohn’sdisease patients when compared to healthy controls [291].It is well established that protein abundance can be controlled at thetranscriptional, post–transcriptional, translational and post–translationallevels. Interestingly, it was demonstrated that translational rate constantsare the determinant factors controlling protein levels [262], indicating thatprotein abundance is predominantly regulated at the ribosomal level. Schwan-häusser et al. [262] showed that the correlation between mRNA and proteinlevels could be improved from 40% to 95% simply by taking into accounttranslation rate constants. Since the same transcript might not necessarilyhave the same translation rate constant when expressed in different cell typesor microenvironments, it is possible that the absence of correlation between104Panx2 mRNA and protein levels could be explained by variations in transla-tion efficiency. For example, a slower translation rate of Panx2 transcripts inneurons could explain the high Panx2 mRNA levels but low Panx2 proteinlevels measured in the CNS. In this regard, micro RNAs (miRNAs) couldhave a major influence on translation rate constants. MiRNAs are about22–nucleotide long noncoding RNAs that downregulate target mRNA ex-pression. The mechanism of action of miRNA and the functional outcomeof the miRNA–target mRNA association has been strongly debated overthe last 20 years. It was initially believed that miRNAs could downregulateprotein expression by translational repression without significantly reduc-ing transcript levels [174, 264, 324]. However, subsequent studies have sug-gested that miRNAs acted predominantly by promoting mRNA degradation[12, 104] or by initially eliciting translational repression followed by targetmRNA degradation [19, 80]. As we have seen in Chapter 2, despite showingthe lowest levels of Panx2 protein, Panx2 transcript levels are much higherin the CNS than in any other tissue. Therefore, if gene silencing by miRNAssolely worked by eliciting mRNA degradation, it would be difficult to pro-pose a model by which miRNAs could control Panx2 translation within theCNS. However, a recent study evaluated the relative contribution of trans-lation repression and mRNA degradation by miRNAs in mutant mice [136].This study was quite exhaustive and included a total of 159 target genesfrom 77 miRNA mutant mice. The analysis showed that 48% of the tar-get genes were primarily regulated by translation repression (76/159), while29% were mainly controlled by mRNA degradation (46/159), and 23% wereregulated by both processes (37/159) [136]. Interestingly, the expressionof almost all target genes identified in developing cells or tissues was pre-dominantly regulated by mRNA degradation while target genes identifiedin terminally differentiated cells were primarily controlled by translation re-pression [136]. It is therefore very likely that miRNA would regulate Panx2expression through translational repression without reducing transcript lev-els within the CNS since neurons are terminally differentiated cells. Whilegene silencing by mRNA degradation is non–reversible (increasing proteinoutput after mRNA degradation requires an increase in transcriptional ac-105tivity and de novo synthesis of mRNA), translation repression is immediate,transient and reversible. We have seen in Chapter 3 that Panx2 proteinlevels increase dramatically in conditions that require higher mitochondrialrespiratory activity. Therefore, regulating Panx2 expression by translationrepression could represent a biological advantage by allowing the cell torapidly modulate its pool of Panx2 channels. For example, neuronal RNAgranules have been shown to store mRNAs under translational arrest andto release them to the translatable pool upon certain stimulation paradigmsuch as depolarization[156]. It would be interesting to determine whether alarge proportion of Panx2 mRNAs are stored as non–translatable mRNAsinto such granules in neurons and other cell types. Proteins and mRNAscan be easily separated from RNA granules isolated by fractionation on asucrose gradient [156]. By combining this approach to qPCR, it would be rel-atively straightforward to determine the percentage of Panx2 mRNAs thatis stored as translationally repressed mRNAs in RNA granules as opposedto translatable mRNAs in the cytoplasm. Using this approach, Krichevskyand Kosik [156] have shown that several mRNAs, including the somatoden-dritic mRNAs coding for NMDAR1, Ca2+/calmodulin–dependent proteinkinase II α (CaMKIIα), and tyrosine receptor kinase B (trkB), translocatefrom the non–translatable RNA granules to translationally active polysomesfollowing neuronal depolarization.As was briefly explained in Chapter 2, protein degradation can also in-fluence the correlation between mRNA and protein levels. Raj et al. [238]have shown that proteins with long half–life are less correlated with mRNAlevels because stable proteins can buffer more fluctuations in mRNA con-centrations than proteins with high turnover rates [238]. For example, Rajand colleagues estimated that the correlation between mRNA and proteinlevels decreased from 92% to 43% when the protein half–life increased from∼ 1.5h to 25h in their system [238]. Although the exact half–life of Panx2has not been measured, it has been estimated that Panx1 and Panx3 arestable proteins with a half–life longer than 6h in over–expressing systems[224], suggesting that a long half–life might have contributed to the lackof correlation between Panx2 mRNA and protein levels. As Panx2 pro-106tein levels are much lower in the CNS than in any other tissues, one mightwonder whether Panx2 has a higher turnover in the CNS. Neurons haveadapted their proteolytic processes to cope with the challenges imposed bytheir unique morphology [290] and it is possible that Panx2 gets degradedmore rapidly as a result. Albeit attractive, this hypothesis is neverthelessunlikely to explain the low abundance of Panx2 protein in the CNS. Quiteunexpectedly, Schwanhäusser et al. [262] have shown that protein degrada-tion had a rather small impact on protein abundance when compared totranslational regulation. Therefore, even if Panx2 turnover rate was foundto be higher in neurons than in any other cell types, it is still uncertainwhether this increase would automatically translate into much lower Panx2protein levels in the CNS.As of Ensembl Release 82 (September 2015), two Panx2 splice variantshave been identified in the mouse genome and we cannot exclude the pos-sibility that tissue–specific splicing might have influenced the correlationbetween Panx2 mRNA and protein levels. The pair of primers employedin Chapter 2 could anneal to both splice variants and would have amplifiedboth transcripts indistinguishably. On the contrary, the epitope of one ofthe monoclonal antibody used in Chapter 2 was mapped to a region of theC–terminal that differs between both Panx2 isoforms. Consequently, thisantibody recognized only the longest, and most prevalent, splice variant.As the epitope of the second antibody has not been mapped, it is possiblethat both antibodies recognized exclusively the longest isoform. Therefore,if the CNS primarily expressed the shorter Panx2 isoform, it would havebeen readily detected at the transcript level, but not at the protein level,thereby causing a severe discrepancy between Panx2 mRNA and proteinlevels. However, this scenario seems unlikely for several reasons: 1) Asshown in Figure 2.5, Panx2 mRNA and protein levels were still severelyuncorrelated after the removal of the CNS samples. Differential expressionof Panx2 splice variants cannot explain a situation where low levels of bothtranscripts results in high protein levels of one of the isoform. 2) As of En-sembl Release 82 (September 2015), most expressed sequence tag (EST) andcDNA sequence evidence supports the existence of the longest Panx2 isoform107in mouse while evidence supporting the alternative splice variant is almostnon–existent. 3) Despite the fact that several protein isoforms might begenerated by alternative splicing, RNAseq data and MS–based proteomicshave shown that the majority of protein–encoding genes express a singledominant isoform, irrespective of tissue or cell type [89, 110, 210]. Althoughit appears unlikely that the CNS expresses a different Panx2 isoform, thisscenario should nonetheless be tested experimentally in order to reach adefinitive conclusion. Transcripts coding for both isoforms should be quan-tified individually using isoform–specific primers and antibodies recognizingboth Panx2 isoforms should be generated.Overall, the work presented in Chapter 2 suggests that Panx2 proteinexpression is actively repressed in the CNS, most probably at the trans-lational level. We have seen in Chapter 3 that Panx2 sensitizes cells toprogrammed cell death. As neurons are terminally differentiated and can-not proliferate, maintaining low levels of Panx2 protein in this cell typemight confer an advantage by minimizing the induction of apoptosis whenexposed to challenges. Interestingly, Sertoli cells, which are the ’nurse’ cellssupporting spermatogenesis in testis, are also terminally differentiated andnon–proliferative in adults [266]. The immunofluorescent staining presentedin Chapter 2 shows that Panx2 is primarily expressed in germ cells and notin Sertoli cells in testis, thereby supporting the idea that Panx2 might bekept at lower levels in cell types that do not easily regenerate.4.2 What controls the gating of Panx2 channeland what goes through the pore?In all cell types, ionized Ca2+ constitutes a universal and indispensableintracellular signaling molecule. Spatiotemporal oscillations in Ca2+ con-centration regulate a wide variety of functions such as neuronal excitability,cell proliferation and differentiation, muscle contraction, secretion and celldeath [24]. Because of their large electrochemical gradient (usually between−150mV to −180mV), mitochondria can accumulate a considerable amountof Ca2+ and influence both the amplitude and the spatiotemporal propaga-108tion of Ca2+ signals [218]. For example, mitochondrial Ca2+ handling hasbeen shown to modulate synaptic transmission in neurons [27, 195, 198].However, mitochondrial Ca2+ transporters have a low affinity with a Kmestimated to be around 10 µm in permeabilized cells [23]. Therefore, mito-chondria need to be exposed to high Ca2+ concentrations to significantlyinfluence cellular Ca2+ homeostasis and Ca2+–mediated processes. Whileextracellular Ca2+ levels range between 1mm to 2mm, cytosolic Ca2+ lev-els are tightly controlled and maintained at a concentration around 100 nmunder normal conditions [138, 271]; a concentration much lower than theaffinity of mitochondrial Ca2+ transporters. Therefore, mitochondria needto be located in the immediate vicinity of Ca2+ release sites in order to beexposed to ’hotspots’ where Ca2+ reaches concentrations that are sufficientto allow mitochondrial uptake.It is tempting to propose that Panx2 can form Ca2+ channels at theMAMs and thereby influence mitochondrial metabolism. This is a reason-able assumption since Panx1 and Panx3 have both been shown to form Ca2+leak channels in the ER [133, 306]. Resting Ca2+ concentrations in the lumenof the ER ([Ca2+]L) has been shown to fluctuate between 50 µm to 300 µm indorsal root ganglia neurons [273] and can reach up to 500 µm in fibroblasts[123] and 630 µm in hepatocytes [49]. These high concentrations would besufficient to inhibit the opening of several connexin channels [20, 175, 257]but unlike most connexins, pannexin channels are insensitive to extracel-lular Ca2+. For example, changing extracellular Ca2+ concentrations fromCa2+–free to 10mm did not affect Panx1– or Panx1/Panx2–mediated cur-rents [43, 187]. Therefore, the opening of Panx2 channels is unlikely to beinhibited even if the [Ca2+]L was unusually high.Pannexins form large pore channels with poor selectivity which are gen-erally kept closed under normal conditions to prevent dissipating the electro-chemical gradient across the membrane in which they are inserted. The sig-nals that control the opening of Panx2 channels at the MAMs are currentlyunknown but several stimuli have been identified that modulate the gatingof pannexin channels even at resting membrane potentials. For example,the gating of Panx1 channels is strongly regulated by K+ [269, 317]. Panx1109forms high–conductance channels (∼ 500 pS) characterized by a strong per-meability to ATP in response to an increase in extracellular K+, even at hy-perpolarized potentials [269, 317]. In contrast, the same channel shows a lowconductance (∼ 50 pS) with negligeable ATP permeability when activated byvoltage in the absence of K+ [317]. However, it is improbable that K+ everregulates the gating of Panx2 channels at MAMs as K+ concentrations aresimilar in the cytosol and the ER lumen [275]. Furthermore, Panx2 channelswere shown to be insensitive to extracellular K+ when exogenously expressedin oocytes [7]. Voltage is also an important factor that controls the gatingof pannexins. Pannexin channels are normally closed at negative potentialsbut readily open upon depolarization [7, 42, 43, 269]. For example, Panx2channels induced large outward currents at membrane potentials exceeding+75mV when exogenously expressed in oocytes [7]. However, voltage gatingis also unlikely to significantly modulate Panx2 opening and gating at MAMsince the ER transmembrane potential is negligible under resting conditionsand only minimally depolarized upon Ca2+ release [98, 166]. Like other bi-ological membranes, the transmembrane potential of the ER is establishedby electrochemical forces. Large K+–selective conductance constitutes themajor ionic conductances of the ER and in the absence of a significant K+gradient between the cytosol and ER lumen [275], the ER cannot develop asignificant membrane potential under physiological conditions [98]. Intracel-lular Ca2+ might play an important role in the gating of Panx2 at MAMs.While pannexin channels are insensitive to important fluctuations in ex-tracellular Ca2+ [43, 187], Panx1 channel readily opens at normal restingmembrane potentials following an increase in intracellular Ca2+ [183, 235].Several stimuli have been shown to augment Panx1 channel activity by in-creasing intracellular Ca2+. For example, caffeine–induced Ca2+ release bystimulation of RyRs induces the opening of large conductance channels (∼300 pS) identified as Panx1 channels in myocytes [148]. Activation of thehistamine receptor or thrombin activation of protease-activated receptor–1 (PAR–1) have also been shown to mobilize cytosolic Ca2+ and activatePanx1 channels in endothelial cells [108]. Moreover, the increase in cytoso-lic Ca2+ observed following the activation of several purinergic receptors110has also been shown to influence the activity of Panx1 channel [183, 222].It is currently unknown whether Panx2 is also activated by a raise in intra-cellular Ca2+ but the possibility that Panx2 channel might be involved inCa2+–induced Ca2+ release at ER–mitochondria contact sites is interestingand deserves some attention.Interestingly, caspase 3 and caspase 7 cleavage sites have been identifiedin the Panx2 C–terminal tail using a cell–free assay [227]. The electrophys-iological properties of the truncated form of Panx2 have not been examinedyet, but similar cleavage sites have been identified and studied in more depthin Panx1 channel. The proteolytic cleavage of the Panx1 C–terminal tail bycaspase 3 and 7 results in a constitutively open channel that is responsiblefor the release of the ATP ’find me’ signal in apoptic cells [50, 237]. Wehave seen in Chapter 3 that Panx2 sensitizes cells to apoptosis. Therefore,it would be interesting to test whether the proteolytic cleavage of the Panx2C–terminal tail can increase the opening probability of Panx2 channels andreinforce the apoptotic process by releasing more Ca2+ at MAMs. In or-der to test whether caspase–dependent cleavage of Panx2 is required forthe induction of apoptosis, sensibility to apoptotic stimuli should be com-pared in cells expressing wildtype Panx2 and cells expressing a mutatedform of Panx2 lacking predicted caspase cleavage sites. The precise positionof Panx2 caspase cleavage site is currently unknown but cleavage of Panx2C–terminal by caspase–3 was shown to produce a ∼ 36 kDa fragment ina cell–free assay [227]. This result suggests that the cleavage site is mostlikely located within the first part of the C–tail. Interestingly, predictedcaspase cleavage sites were identified within that region using the databaseCaspDB [159]. However, in order to identify the genuine caspase cleavagesites, the candidate sites should be systematically mutated by site–directedmutagenesis.4.3 Functional implication of Panx2 at MAMsWe have seen in Chapter 3, that Panx2 protein levels are low in cells grownon glucose but markedly increase in glucose–free conditions which force the111cells to rely on mitochondrial respiration for energy production [3, 144, 190].This result suggests that Panx2 is important for the regulation of mitochon-drial bioenergetic processes. Interestingly, an increase in ER–mitochondriacontact sites has been shown to stimulate oxygen consumption and ATPproduction in mammalian cells [39]. These effects were shown to be de-pendent on an increase in mitochondrial Ca2+ [39]. As we discussed inChapter 1, mitochondrial Ca2+ increases the activity of matrix dehydroge-nases and oxidative phosphorylation protein complexes which are requiredfor mitochondrial respiration and ATP production. Therefore, the resultspresented in Chapter 3 suggest that the increase in Panx2 expression canstimulate mitochondrial Ca2+ uptake and ATP production upon conditionsthat increase mitochondrial respiratory activity (Figure 4.1A,B). In orderto test this idea, mitochondrial Ca2+ levels, oxygen consumption rate andATP levels would need to be monitored in wild type cells and cells in whichPanx2 has been silenced (by shRNA or siRNA for example). If the proposedhypothesis is validated, we should expect lower levels of mitochondrial Ca2+and lower mitochondrial respiratory activity following Panx2 silencing.Swayne et al. [285] have reported that Panx2 levels are down–regulatedin neuronal progenitors and immature neurons but increased in mature neu-rons. Interestingly, several groups have recently reported that cellular dif-ferentiation is characterized by mitochondrial maturation and a transitionfrom glycolysis to oxidative phosphorylation as the main ATP productionpathway [2, 309]. For example, it was shown that dividing progenitor cellsmostly depend on glycolysis for energy production in the embryonic retinawhile cells in the fully differentiatied retina primarily use oxidative phos-phorylation [2]. The transition from anaerobic to aerobic metabolism occursduring the early stages of embryonic development in mammals. Using cul-tured rat embryos, Morriss and New [203] showed that aerobic metabolismis responsible for only ∼ 5% of energy production before gestation day 9but accounts for ∼ 95% of energy production after gestation day 11. As wehave seen that Panx2 expression increases substantially in conditions thatstimulate oxidative phosphorylation, it would be interesting to test whetherPanx2 protein levels also increase during this shift. In Chapter 2, we have112Figure 4.1. Proposed model of Panx2 regulation at MAMs. A: In cells with highrate of glycolysis such as cancer cells, mitochondria are poorly energized and mostof the ATP is derived from glycolysis. Panx2 levels are maintained low, Panx2–mediated Ca2+ release is minimal at MAMs and mitochondrial Ca2+ uptake islow. B: In conditions of high mitochondrial respiratory activity, Panx2 expressionis up–regulated and the increase in Panx2–mediated Ca2+ release leads to highermitochondrial Ca2+ uptake. As a result, the energy conversion potential increasesand mitochondria are fully energized. C: Under pathological conditions, the over–expression or over activation of Panx2 channels induces excessive Ca2+ release atMAMs and triggers mitochondrial Ca2+ overloading and ultimately apoptosis.seen that Panx2 protein levels are fairly constant in the developing mousebrain. However, all samples were taken after the transition period (gestationday 11 to postnatal day 30) and we have not measured Panx2 levels priorthis transition yet.The model presented in Figure 4.1 suggests that Panx2–mediated Ca2+release at MAMs induces a moderate increase in mitochondrial Ca2+ whichstimulates cellular respiration and ATP production. However, it is welldocumented that excessive mitochondrial Ca2+ uptake can result in perme-ability transition and activation of intrinsic apoptotic pathways [115, 287].Interestingly, Charles Lai, a former PhD student in our laboratory, showedthat Panx2 had strong anti–tumorigenic effects when over–expressed in C6glioma cells [165]. In vitro and in vivo tumor growth was severely compro-mised in C6 cells over–expressing Panx2 compared to wild type cells whichdid not express Panx2. The results presented in Chapter 3 complement theseobservations by demonstrating that Panx2 anti–tumorigenic properties are113most probably explained by an increase susceptibility to apoptosis. There-fore, according to our model, the expression or gating of Panx2 channels donot only regulate mitochondrial respiratory activity but also influence cellsurvival under physiological or pathological situations (Figure 4.1C).As we have briefly mentioned in Chapter 3, we cannot exclude the possi-bility that the role of Panx2 at ER–mitochondria contact sites is independentof mitochondrial Ca2+ uptake. The distance separating both organelles isalso a key determinant in the regulation of ER–mitochondria contact sites[64] and this distance could potentially be modulated by Panx2. Inter-estingly, ER Ca2+ release reduces mitochondria motility and promotes thedocking of mitochondria to ER membranes [329]. Mitochondria immobiliza-tion can be induced by the stimulation of the IP3R or RyR and this effectis independent of mitochondrial Ca2+ uptake [329]. As pannexin channelscan form Ca2+ channels in the ER [133, 306], the effect of Panx2 activityon mitochondria movement should also be investigated. Panx2 could alsodirectly promote the docking of mitochondria on ER membranes throughphysical interaction with mitochondrial proteins. Highly conserved proteininteraction sites were discovered on Panx2 C–terminal tail by scanning thePanx2 amino acid sequence with bioinformatics tools (Eukaryotic LinearMotifs, For example, the last six amino acids of Panx2C–terminal tail (VSTVEF) have been identified as a conserved PDZ bind-ing motif. Interestingly, the outer membrane protein 25 (OMP25) is a mi-tochondrial protein anchored to the OMM that contains a PDZ domainfacing the cytoplasm [206]. OMP25 has been shown to recruit the inosi-tol 5’-phosphatase synaptojanin 2A [206] and the stress–activated proteinkinase–3 (SAPK3) to mitochondria [57]. It would be interesting to deter-mine whether Panx2 could promote the recruitment of mitochondria to ERmembranes through its interaction with OMP25. Panx2 C–terminal tail alsocontains one highly conserved (DGGPRLP, amino acids 639–645) and 8 lessconserved Src homology 3 (SH3) binding motifs which could potentially bindto Bax–interacting factor 1 (Bif1). Bif1 is a multifunctional protein with aC–terminal SH3 domain involved in the regulation of apoptosis, mitochon-drial morphology, and autophagy. It has been shown to transit between114the cytosol and the OMM but its association with mitochondria increasesduring apoptosis [143]. Interestingly, other proteins present in MAMs alsocontain proline–rich motifs which could bind to Bif1 SH3 domain. For exam-ple, Drp1 contains the PXXP motif in amino acids 610–618 (PIPIMPASP)and MFN2 contains two putative SH3 binding motifs (amino acids 523–526,PLLP, and amino acids 595–609, PIPLTPANPSMPPLP) [142].4.4 The role of Panx2 in mitochondria–mediatedcell deathThe results presented in Chapter 3 clearly demonstrate that the expressionof Panx2 protein sensitizes cells to cell death by apoptosis. Over–expressingPanx2 protein precipitated the initiation of the apoptotic program in cellssubjected to apoptotic stimuli whereas suppressing Panx2 expression inhib-ited the initiation of apoptosis. However, only two hallmarks of apoptosiswere investigated in the current study: the activation of caspase–3, and thefragmentation of nucleic DNA. Although these two events constitute funda-mental hallmarks of apoptosis, they both appear during the later stages ofthe apoptotic program. To further characterize the role of Panx2 in apop-tosis, it would be interesting to determine whether Panx2 also affects theearlier stages of the apoptotic process. Based on the hypothesis presented inthis thesis and which suggests that Panx2 expression at MAM sites regulatesmitochondrial Ca2+ uptake, one would expect that Panx2 also stimulatesthe earlier stages of the apoptotic process. Mitochondria constitute impor-tant checkpoints of the apoptotic program. Upon excessive Ca2+ uptake,the integrity of mitochondrial membranes is compromised and several mi-tochondrial proteins are released into the cytosol. One of these protein,cytochrome c, plays a major role in the initiation of the apoptotic program.The release of cytochrome c is a two–step process which involves the dis-ruption of cytochrome c association with the IMM and permeabilization ofthe OMM by Bax [213]. Once released in the cytoplasm, cytochrome c trig-gers the assembly of the so–called apoptosome by binding to the proteinapoptosis—protease activating factor 1 (Apaf–1) [242]. Once fully assem-115bled, the apoptosome forms a functional platform that successfully recruitsand activates caspase 9 [242]. Therefore, in order to further characterize therole of Panx2 in mitochondria–mediated apoptosis, it would be importantto determine whether Panx2 promotes the assembly of the mitochondrialpermeability transition pore complex and the release of cytochrome c in thecytoplasm. Furthermore, similarly to the effect of Panx2 on caspase 3 acti-vation, it would interesting to determine whether Panx2 also promotes theactivation of caspase 9 following the assembly of the apoptosome.4.5 Implication of Panx2 in human diseasesAs we have seen in the previous chapters, ER–mitochondria contact sitesregulate several cellular processes such as lipid processing, mitochondrialCa2+ homeostasis, cellular respiration, and the initiation of cell death [69,119, 171, 248, 304]. As they regulate important cellular processes, MAMshave also been studied in the context of diseases and have been found tobe involved in various pathological conditions such as obesity, Alzheimer’sdisease (AD), stroke and Parkinson’s disease [9, 11, 112, 214, 261]. In the fol-lowing sections, the localization of Panx2 at the ER–mitochondria interfacewill be reexamined in the context of human diseases.4.5.1 Panx2 and Alzheimer’s diseaseAD is a progressive, irreversible and fatal brain disorder causing neuronaldeath and slowly destroying memory and thinking skills. At the microscopiclevel, AD is characterized by the abnormal accumulation of extracellularplaques containing beta amyloid peptides (Aβ) and intra-neuronal tanglesof hyper-phosphorylated tau proteins [126]. It has been proposed that theaccumulation of Aβ plaques causes and drives AD by directly or indirectlykilling neurons. However, this hypothesis has recently been challenged andtherapies aiming to reduce the production of Aβ have offered very limitedclinical advantages [47]. These observations suggest that alternative path-ways must be involved in the development and evolution of the disease.Accumulating evidence have shown that mitochondria are involved in the116initiation and progression of AD and mitochondrial dysfunctions often pre-cede the accumulation of Aβ plaques and intra–neuronal tangles [201]. Forexample, cultured neurons treated with inhibitors of the mitochondrial ETChave increased levels of phosphorylated tau protein [88] and rats treated withETC inhibitors such as sodium azide or rotenone have an abnormal accu-mulation of hyper–phosphorylated tau protein, present signs of neurode-generation and experience cognitive impairments that recapitulate severalsymptoms of AD [124, 289]. In addition, it has been shown that AD patientshave abnormally high caspase 3 activity which increases the production ofAβ peptides through caspase 3–mediated cleavage of the Aβ precursor pro-tein [101]. Interestingly, cytoplasmic hydrid (cybrid) cell lines transplantedwith mitochondria isolated from platelets of AD patients showed increasedoxidative stress, elevated caspase 3 activity and an overproduction of Aβpeptides [145], thereby further reinforcing the link between mitochondrialdysfunctions and the progression of AD. Overall, these results have led tothe emergence of the mitochondrial cascade hypothesis which proposes thata decline in mitochondrial functions is responsible, at least partially, for thecellular changes observed in AD [286].In Chapter 3, it was demonstrated that Panx2 is intrinsically linked tomitochondrial functions. Panx2 protein levels were shown to be stronglyinfluenced by mitochondria energy conversion potential. Panx2 expressionwas significantly increased when cells were grown in conditions preventingefficient glycolysis but requiring optimal mitochondrial respiration. Sur-prisingly, I also recently discovered that Panx2 protein expression is sub-stantially up–regulated in human neuroblastoma cells treated with ETCinhibitors known to produce AD–like symptoms (Figure 3.10) [124]. Wehave also seen in Chapter 3 that Panx2 increases caspase 3 activation andsensitizes cells to apoptosis under cellular stress. Therefore, it seems thatPanx2 can recapitulate some of the cellular characteristics of AD. Interest-ingly, ER–mitochondria contact sites have been shown to be a major siteof Aβ production in the brain [261] and to play an important role in thedevelopment of AD pathology [121, 260]. Proteins previously shown to beenriched at ER–mitochondria contact sites were found to be up–regulated in117human AD brain and AD mouse model [121]. Interestingly, in the AD mousemodel this up–regulation was observed prior to the appearance of amyloidplaques [121] thereby indicating that perturbations in the composition orfunctions of MAMs occur early in the pathogenesis of the disease. We can-not ignore the evidence suggesting that Panx2 plays a pivotal role in cellularpathways shown to be directly affected by AD and it is therefore temptingto speculate that Panx2 might be involved in the pathogenesis of AD. Itwould be particularly interesting to determine whether Panx2 expression orsubcellular localization is altered in AD brains. As shown in Chapter 3, anup–regulation of Panx2 expression would render neurons more susceptible tocell death and it would therefore be important to explore whether blockingPanx2 up–regulation or activity would prevent neuronal death and alleviatesome of the symptoms.4.5.2 Panx2 and cerebral ischemiaMolecular oxygen is essential for the transport of electrons along the ETCand mitochondrial respiration is rapidly interrupted upon oxygen depri-vation. Because of their high energetic demand, neurons rely heavily onoxidative phosphorylation for ATP production and are not as resilient asother cell types when subjected to oxygen deprivation. We have seen inChapter 1 that Panx1 and Panx2 channels influence how neurons handleischemic challenges. Panx1/Panx2 double knockout mice were shown tohave smaller strokes and more modest neurological deficits than wild typemice after cerebral ischemia [16]. Although Bargiotas et al. [16] did not re-port any neuroprotection in Panx1-/- or Panx2-/- single mutants, our groupand other have demonstrated a neuroprotective effect in Panx1-/- mice orupon pharmacological inhibition of Panx1 channel [52, 320, 326]. These re-sults indicate that Panx1 and Panx2 channels might work independently inthe pathogenesis of ischemia. Interestingly, we have recently reported thatPanx2 levels are substantially up–regulated by cerebral ischemia (Figure 4.2)and preliminary results suggest that this up–regulation starts soon after theocclusion of the middle cerebral artery (MCA). Since we have shown in118Figure 4.2. Cerebral ischemia stimulates Panx2 expression. A: Three differentmice (Stroke 1 to 3) were subjected to cerebral ischemia by electrocoagulation ofthe MCA. One mouse was subjected to the same surgical procedure but withoutpermanent occlusion of the MCA (control). The ischemic region was sampled 4days after the surgery and Panx2 expression was assessed by Western blot. Panx2is up–regulated in the ischemic brain. B-C: Panx2 immunofluorescent labeling inthe healthy (B) and ischemic cortex (C). Those data were generated by Dr MoisesFreitas–Andrade, a postdoctoral fellow in our laboratory.Chapter 3 that Panx2 sensitizes cells to apoptosis, we propose that reducingPanx2 expression or activation might have a neuroprotective advantage forneurons subjected to ischemic conditions.119Bibliography[1] R. d. S. Abreu, L. O. Penalva, E. M. Marcotte, and C. Vogel. Globalsignatures of protein and mRNA expression levels. MolecularbioSystems, 5(12):1512–1526, Dec. 2009. → pages 68[2] M. Agathocleous, N. K. Love, O. Randlett, J. J. Harris, J. Liu, A. J.Murray, and W. A. Harris. Metabolic differentiation in theembryonic retina. Nature Cell Biology, 14(8):859–864, Aug. 2012. →pages 112[3] C. Aguer, D. Gambarotta, R. J. Mailloux, C. Moffat, R. Dent,R. McPherson, and M.-E. Harper. Galactose Enhances OxidativeMetabolism and Reveals Mitochondrial Dysfunction in HumanPrimary Muscle Cells. PLoS ONE, 6(12):e28536, Dec. 2011. → pages94, 112[4] S. Ahmad and W. H. Evans. Post-translational integration andoligomerization of connexin 26 in plasma membranes and evidence offormation of membrane pores: implications for the assembly of gapjunctions. The Biochemical Journal, 365(Pt 3):693–699, Aug. 2002.→ pages 6[5] S. Ahmad, J. A. Diez, C. H. George, and W. H. Evans. Synthesisand assembly of connexins in vitro into homomeric and heteromericfunctional gap junction hemichannels. The Biochemical Journal, 339( Pt 2):247–253, Apr. 1999. → pages 6[6] D. G. Allen and J. R. Blinks. Calcium transients in aequorin-injectedfrog cardiac muscle. Nature, 273(5663):509–513, June 1978. → pages38[7] C. Ambrosi, O. Gassmann, J. N. Pranskevich, D. Boassa, A. Smock,J. Wang, G. Dahl, C. Steinem, and G. E. Sosinsky. Pannexin1 and120Pannexin2 Channels Show Quaternary Similarities to Connexons andDifferent Oligomerization Numbers from Each Other. Journal ofBiological Chemistry, 285(32):24420–24431, June 2010. → pages 7, 9,16, 19, 21, 71, 110[8] A. M. Aponte, D. Phillips, R. K. Hopper, D. T. Johnson, R. A.Harris, K. Blinova, E. S. Boja, S. French, and R. S. Balaban. Use of(32)p to study dynamics of the mitochondrial phosphoproteome.Journal of Proteome Research, 8(6):2679–2695, June 2009. → pages40[9] E. Area-Gomez, M. Del Carmen Lara Castillo, M. D. Tambini,C. Guardia-Laguarta, A. J. C. de Groof, M. Madra, J. Ikenouchi,M. Umeda, T. D. Bird, S. L. Sturley, and E. A. Schon. Upregulatedfunction of mitochondria-associated er membranes in alzheimerdisease. The EMBO journal, 31(21):4106–4123, Nov. 2012. → pages30, 74, 116[10] S. Arnold and B. Kadenbach. The intramitochondrial atp/adp-ratiocontrols cytochrome c oxidase activity allosterically. FEBS letters,443(2):105–108, Jan. 1999. → pages 42[11] A. P. Arruda, B. M. Pers, G. Parlakgül, E. Güney, K. Inouye, andG. S. Hotamisligil. Chronic enrichment of hepatic endoplasmicreticulum-mitochondria contact leads to mitochondrial dysfunctionin obesity. Nature Medicine, 20(12):1427–1435, Dec. 2014. → pages30, 74, 116[12] S. Bagga, J. Bracht, S. Hunter, K. Massirer, J. Holtz, R. Eachus, andA. E. Pasquinelli. Regulation by let-7 and lin-4 mirnas results intarget mrna degradation. Cell, 122(4):553–563, Aug. 2005. → pages105[13] L. Bao, S. Locovei, and G. Dahl. Pannexin membrane channels aremechanosensitive conduits for atp. FEBS letters, 572(1-3):65–68,Aug. 2004. → pages 17, 18[14] A. Baranova, D. Ivanov, N. Petrash, A. Pestova, M. Skoblov,I. Kelmanson, D. Shagin, S. Nazarenko, E. Geraymovych, O. Litvin,A. Tiunova, T. L. Born, N. Usman, D. Staroverov, S. Lukyanov, andY. Panchin. The mammalian pannexin family is homologous to theinvertebrate innexin gap junction proteins. Genomics, 83(4):706–716,Apr. 2004. → pages 5, 15, 21, 22, 49, 59, 60, 68121[15] M. T. Barbe. Cell-Cell Communication Beyond Connexins: ThePannexin Channels. Physiology, 21(2):103–114, Apr. 2006. → pages48[16] P. Bargiotas, A. Krenz, S. G. Hormuzdi, D. A. Ridder, A. Herb,W. Barakat, S. Penuela, J. von Engelhardt, H. Monyer, andM. Schwaninger. Pannexins in ischemia-induced neurodegeneration.Proceedings of the National Academy of Sciences, 108(51):20772–20777, Dec. 2011. → pages 23, 25, 72, 118[17] T. M. Barnes. Opus: a growing family of gap junction proteins?Trends in genetics: TIG, 10(9):303–305, Sept. 1994. → pages 3[18] J. M. Baughman, F. Perocchi, H. S. Girgis, M. Plovanich, C. A.Belcher-Timme, Y. Sancak, X. R. Bao, L. Strittmatter,O. Goldberger, R. L. Bogorad, V. Koteliansky, and V. K. Mootha.Integrative genomics identifies mcu as an essential component of themitochondrial calcium uniporter. Nature, 476(7360):341–345, Aug.2011. → pages 37[19] A. A. Bazzini, M. T. Lee, and A. J. Giraldez. Ribosome profilingshows that mir-430 reduces translation before causing mrna decay inzebrafish. Science (New York, N.Y.), 336(6078):233–237, Apr. 2012.→ pages 105[20] D. L. Beahm and J. E. Hall. Hemichannel and junctional propertiesof connexin 50. Biophysical Journal, 82(4):2016–2031, Apr. 2002. →pages 109[21] E. Bejarano, A. Yuste, B. Patel, R. F. Stout Jr, D. C. Spray, andA. M. Cuervo. Connexins modulate autophagosome biogenesis.Nature Cell Biology, 16(5):401–414, May 2014. → pages 71[22] E. Bender and B. Kadenbach. The allosteric atp-inhibition ofcytochrome c oxidase activity is reversibly switched on bycamp-dependent phosphorylation. FEBS letters, 466(1):130–134,Jan. 2000. → pages 42[23] P. Bernardi. Mitochondrial transport of cations: channels,exchangers, and permeability transition. Physiological Reviews, 79(4):1127–1155, Oct. 1999. → pages 109[24] M. J. Berridge, M. D. Bootman, and P. Lipp. Calcium–a life anddeath signal. Nature, 395(6703):645–648, Oct. 1998. → pages 108122[25] V. M. Berthoud, E. A. Montegna, N. Atal, N. H. Aithal, P. R. Brink,and E. C. Beyer. Heteromeric connexons formed by the lensconnexins, connexin43 and connexin56. European Journal of CellBiology, 80(1):11–19, Jan. 2001. → pages 6[26] R. Bhalla-Gehi, S. Penuela, J. M. Churko, Q. Shao, and D. W. Laird.Pannexin1 and Pannexin3 Delivery, Cell Surface Dynamics, andCytoskeletal Interactions. Journal of Biological Chemistry, 285(12):9147–9160, Jan. 2010. → pages 11, 70[27] B. Billups and I. D. Forsythe. Presynaptic mitochondrial calciumsequestration influences transmission at mammalian centralsynapses. The Journal of Neuroscience: The Official Journal of theSociety for Neuroscience, 22(14):5840–5847, July 2002. → pages 109[28] S. Blackshaw, A. Sawa, A. H. Sharp, C. A. Ross, S. H. Snyder, andA. A. Khan. Type 3 inositol 1,4,5-trisphosphate receptor modulatescell death. FASEB journal: official publication of the Federation ofAmerican Societies for Experimental Biology, 14(10):1375–1379, July2000. → pages 45[29] D. Boassa, C. Ambrosi, F. Qiu, G. Dahl, G. Gaietta, andG. Sosinsky. Pannexin1 Channels Contain a Glycosylation Site ThatTargets the Hexamer to the Plasma Membrane. Journal of BiologicalChemistry, 282(43):31733–31743, Sept. 2007. → pages 7, 10, 11, 49,100[30] D. Boassa, F. Qiu, G. Dahl, and G. Sosinsky. Trafficking Dynamicsof Glycosylated Pannexin1 Proteins. Cell Communication andAdhesion, 15(1-2):119–132, Jan. 2008. → pages 10, 11[31] D. Boassa, P. Nguyen, J. Hu, M. H. Ellisman, and G. E. Sosinsky.Pannexin2 oligomers localize in the membranes of endosomal vesiclesin mammalian cells while pannexin1 channels traffic to the plasmamembrane. Frontiers in Cellular Neuroscience, 8:468, 2014. → pages16, 29, 99, 100, 101[32] I. R. Boldogh, D. W. Nowakowski, H.-C. Yang, H. Chung,S. Karmon, P. Royes, and L. A. Pon. A protein complex containingmdm10p, mdm12p, and mmm1p links mitochondrial membranes anddna to the cytoskeleton-based segregation machinery. MolecularBiology of the Cell, 14(11):4618–4627, Nov. 2003. → pages 31123[33] S. Bolte and F. P. Cordelières. A guided tour into subcellularcolocalization analysis in light microscopy. Journal of Microscopy,224(3):213–232, Dec. 2006. → pages 79[34] S. R. Bond and C. C. Naus. The pannexins: past and present.Frontiers in Physiology, 5:58, 2014. → pages 21[35] S. R. Bond, A. Lau, S. Penuela, A. V. Sampaio, T. M. Underhill,D. W. Laird, and C. C. Naus. Pannexin 3 is a novel target for runx2,expressed by osteoblasts and mature growth plate chondrocytes.Journal of bone and mineral research: the official journal of theAmerican Society for Bone and Mineral Research, 26(12):2911–2922,Dec. 2011. → pages 27, 28[36] S. R. Bond, N. Wang, L. Leybaert, and C. C. Naus. Pannexin 1ohnologs in the teleost lineage. The Journal of membrane biology, 245(8):483–493, Aug. 2012. → pages 5, 17, 18, 22, 28, 48, 49, 59, 60, 68[37] A. Bononi, M. Bonora, S. Marchi, S. Missiroli, F. Poletti, C. Giorgi,P. P. Pandolfi, and P. Pinton. Identification of pten at the er andmams and its regulation of ca(2+) signaling and apoptosis in aprotein phosphatase-dependent manner. Cell Death andDifferentiation, 20(12):1631–1643, Dec. 2013. → pages 44[38] A. K. Boyce, R. T. Prager, L. E. Wicki-Stordeur, and L. A. Swayne.Pore positioning. Channels, 8(2):110–117, Mar. 2014. → pages 99[39] R. Bravo, J. M. Vicencio, V. Parra, R. Troncoso, J. P. Munoz,M. Bui, C. Quiroga, A. E. Rodriguez, H. E. Verdejo, J. Ferreira,M. Iglewski, M. Chiong, T. Simmen, A. Zorzano, J. A. Hill, B. A.Rothermel, G. Szabadkai, and S. Lavandero. Increaseder-mitochondrial coupling promotes mitochondrial respiration andbioenergetics during early phases of er stress. Journal of CellScience, 124(Pt 13):2143–2152, July 2011. → pages 112[40] P. R. Brink, K. Cronin, K. Banach, E. Peterson, E. M. Westphale,K. H. Seul, S. V. Ramanan, and E. C. Beyer. Evidence forheteromeric gap junction channels formed from rat connexin43 andhuman connexin37. The American Journal of Physiology, 273(4 Pt1):C1386–1396, Oct. 1997. → pages 6124[41] J. M. Brown and L. D. Attardi. The role of apoptosis in cancerdevelopment and treatment response. Nature Reviews Cancer, 5(3):231–237, Mar. 2005. → pages 44[42] R. Bruzzone, S. G. Hormuzdi, M. T. Barbe, A. Herb, and H. Monyer.Pannexins, a family of gap junction proteins expressed in brain.Proceedings of the National Academy of Sciences of the United Statesof America, 100(23):13644–13649, Nov. 2003. → pages 5, 9, 16, 17,19, 21, 22, 49, 59, 60, 68, 110[43] R. Bruzzone, M. T. Barbe, N. J. Jakob, and H. Monyer.Pharmacological properties of homomeric and heteromeric pannexinhemichannels expressed in Xenopus oocytes. Journal ofNeurochemistry, 92(5):1033–1043, Mar. 2005. → pages 9, 20, 109, 110[44] E. Carafoli. Intracellular calcium homeostasis. Annual Review ofBiochemistry, 56:395–433, 1987. → pages 39[45] E. Carafoli and M. Crompton. The regulation of intracellularcalcium by mitochondria. Annals of the New York Academy ofSciences, 307:269–284, Apr. 1978. → pages 38[46] D. Caskenette, S. Penuela, V. Lee, K. Barr, F. Beier, D. W. Laird,and K. E. Willmore. Global deletion of panx3 produces multiplephenotypic effects in mouse humeri and femora. Journal of Anatomy,Jan. 2016. → pages 28[47] M. A. Castello, J. D. Jeppson, and S. Soriano. Moving beyondanti-amyloid therapy for the prevention and treatment of Alzheimer’sdisease. BMC Neurology, 14(1):169, Sept. 2014. → pages 116[48] S. J. Celetti, K. N. Cowan, S. Penuela, Q. Shao, J. Churko, andD. W. Laird. Implications of pannexin 1 and pannexin 3 forkeratinocyte differentiation. Journal of Cell Science, 123(Pt 8):1363–1372, Apr. 2010. → pages 20[49] J. Y. Chatton, H. Liu, and J. W. Stucki. Simultaneousmeasurements of ca2+ in the intracellular stores and the cytosol ofhepatocytes during hormone-induced ca2+ oscillations. FEBSletters, 368(1):165–168, July 1995. → pages 109[50] F. Chekeni, M. Elliott, J. Sandilos, S. Walk, J. Kinchen,E. Lazarowski, A. Armstrong, S. Penuela, D. Laird, G. Salvesen, and125others. Pannexin 1 channels mediate/find-me/’signal release andmembrane permeability during apoptosis. Nature, 467(7317):863–867, 2010. → pages 18, 25, 111[51] C. Cina, K. Maass, M. Theis, K. Willecke, J. F. Bechberger, andC. C. Naus. Involvement of the cytoplasmic c-terminal domain ofconnexin43 in neuronal migration. The Journal of Neuroscience: TheOfficial Journal of the Society for Neuroscience, 29(7):2009–2021,Feb. 2009. → pages 71[52] A. Cisneros-Mejorado, M. Gottlieb, F. Cavaliere, T. Magnus,F. Koch-Nolte, E. Scemes, A. Pérez-Samartín, and C. Matute.Blockade of p2x7 receptors or pannexin-1 channels similarlyattenuates postischemic damage. Journal of Cerebral Blood Flow andMetabolism: Official Journal of the International Society of CerebralBlood Flow and Metabolism, 35(5):843–850, May 2015. → pages 23,118[53] A. C. Cone, C. Ambrosi, E. Scemes, M. E. Martone, and G. E.Sosinsky. A comparative antibody analysis of pannexin1 expressionin four rat brain regions reveals varying subcellular localizations.Frontiers in Pharmacology, 4:6, 2013. → pages 29[54] D. E. Copeland and A. J. Dalton. An association betweenmitochondria and the endoplasmic reticulum in cells of thepseudobranch gland of a teleost. The Journal of Biophysical andBiochemical Cytology, 5(3):393–396, May 1959. → pages 30[55] P. Cosson, A. Marchetti, M. Ravazzola, and L. Orci. Mitofusin-2independent juxtaposition of endoplasmic reticulum andmitochondria: an ultrastructural study. PloS One, 7(9):e46293, 2012.→ pages 33[56] G. T. Cottrell and J. M. Burt. Heterotypic gap junction channelformation between heteromeric and homomeric cx40 and cx43connexons. American Journal of Physiology. Cell Physiology, 281(5):C1559–1567, Nov. 2001. → pages 6[57] N. W. Court, E. Ingley, S. P. Klinken, and M. A. Bogoyevitch. Outermembrane protein 25-a mitochondrial anchor and inhibitor ofstress-activated protein kinase-3. Biochimica Et Biophysica Acta,1744(1):68–75, May 2005. → pages 114126[58] R. Covian and R. S. Balaban. Cardiac mitochondrial matrix andrespiratory complex protein phosphorylation. American Journal ofPhysiology - Heart and Circulatory Physiology, 303(8):H940–H966,Oct. 2012. → pages 40[59] R. Covian, S. French, H. Kusnetz, and R. S. Balaban. Stimulation ofoxidative phosphorylation by calcium in cardiac mitochondria is notinfluenced by camp and pka activity. Biochimica Et Biophysica Acta,1837(12):1913–1921, Dec. 2014. → pages 39[60] D. Crompton, M. Todman, M. Wilkin, S. Ji, and J. Davies. Essentialand Neural Transcripts from the Drosophila shaking-B Locus AreDifferentially Expressed in the Embryonic Mesoderm and PupalNervous System. Developmental Biology, 170(1):142–158, July 1995.→ pages 3[61] S. J. Cruikshank, M. Hopperstad, M. Younger, B. W. Connors, D. C.Spray, and M. Srinivas. Potent block of cx36 and cx50 gap junctionchannels by mefloquine. Proceedings of the National Academy ofSciences of the United States of America, 101(33):12364–12369, Aug.2004. → pages 20[62] G. Csordás, A. P. Thomas, and G. Hajnóczky. Quasi-synapticcalcium signal transmission between endoplasmic reticulum andmitochondria. The EMBO journal, 18(1):96–108, Jan. 1999. → pages38, 44[63] G. Csordás, M. Madesh, B. Antonsson, and G. Hajnóczky. tcbidpromotes ca(2+) signal propagation to the mitochondria: control ofca(2+) permeation through the outer mitochondrial membrane. TheEMBO journal, 21(9):2198–2206, May 2002. → pages 37[64] G. Csordás, C. Renken, P. Várnai, L. Walter, D. Weaver, K. F.Buttle, T. Balla, C. A. Mannella, and G. Hajnóczky. Structural andfunctional features and significance of the physical linkage betweener and mitochondria. The Journal of Cell Biology, 174(7):915–921,Sept. 2006. → pages 30, 31, 32, 38, 44, 94, 101, 102, 114[65] G. Dahl. Gap junction-mimetic peptides do work, but in unexpectedways. Cell Communication & Adhesion, 14(6):259–264, 2007Nov-Dec. → pages 21127[66] J. Das Sarma, R. A. Meyer, F. Wang, V. Abraham, C. W. Lo, andM. Koval. Multimeric connexin interactions prior to the trans-golginetwork. Journal of Cell Science, 114(Pt 22):4013–4024, Nov. 2001.→ pages 6[67] J. S. Davidson and I. M. Baumgarten. Glycyrrhetinic acidderivatives: a novel class of inhibitors of gap-junctional intercellularcommunication. structure-activity relationships. The Journal ofPharmacology and Experimental Therapeutics, 246(3):1104–1107,Sept. 1988. → pages 20[68] O. M. de Brito and L. Scorrano. Mitofusin 2 tethers endoplasmicreticulum to mitochondria. Nature, 456(7222):605–610, Dec. 2008. →pages 32, 33, 74[69] O. M. de Brito and L. Scorrano. An intimate liaison: spatialorganization of the endoplasmic reticulum-mitochondria relationship.The EMBO Journal, 29(16):2715–2723, Aug. 2010. → pages 30, 74,116[70] J. P. de Rivero Vaccari, G. Lotocki, O. F. Alonso, H. M. Bramlett,W. D. Dietrich, and R. W. Keane. Therapeutic neutralization of thenlrp1 inflammasome reduces the innate immune response andimproves histopathology after traumatic brain injury. Journal ofCerebral Blood Flow and Metabolism: Official Journal of theInternational Society of Cerebral Blood Flow and Metabolism, 29(7):1251–1261, July 2009. → pages 24[71] D. De Stefani, A. Raffaello, E. Teardo, I. Szabò, and R. Rizzuto. Aforty-kilodalton protein of the inner membrane is the mitochondrialcalcium uniporter. Nature, 476(7360):336–340, Aug. 2011. → pages37[72] E. C. Dell’Angelica, V. Shotelersuk, R. C. Aguilar, W. A. Gahl, andJ. S. Bonifacino. Altered trafficking of lysosomal proteins inhermansky-pudlak syndrome due to mutations in the beta 3a subunitof the ap-3 adaptor. Molecular Cell, 3(1):11–21, Jan. 1999. → pages100[73] H. F. Deluca and G. W. Engstrom. Calcium uptake by rat kidneymitochondria. Proceedings of the National Academy of Sciences ofthe United States of America, 47:1744–1750, Nov. 1961. → pages 37128[74] R. M. Denton, P. J. Randle, and B. R. Martin. Stimulation bycalcium ions of pyruvate dehydrogenase phosphate phosphatase. TheBiochemical Journal, 128(1):161–163, June 1972. → pages 39[75] R. M. Denton, D. A. Richards, and J. G. Chin. Calcium ions and theregulation of nad+-linked isocitrate dehydrogenase from themitochondria of rat heart and other tissues. The BiochemicalJournal, 176(3):899–906, Dec. 1978. → pages 39[76] R. M. Denton, J. G. McCormack, and N. J. Edgell. Role of calciumions in the regulation of intramitochondrial metabolism. effects ofna+, mg2+ and ruthenium red on the ca2+-stimulated oxidation ofoxoglutarate and on pyruvate dehydrogenase activity in intact ratheart mitochondria. The Biochemical Journal, 190(1):107–117, July1980. → pages 39[77] E. F. Diezmos, S. L. Sandow, I. Markus, D. Shevy Perera, D. Z.Lubowski, D. W. King, P. P. Bertrand, and L. Liu. Expression andlocalization of pannexin-1 hemichannels in human colon in healthand disease. Neurogastroenterology & Motility, 25(6):e395–e405, June2013. → pages 61[78] E. F. Diezmos, S. L. Sandow, D. S. Perera, D. W. King, P. P.Bertrand, and L. Liu. Pannexin-2 is expressed in the human colonwith extensive localization in the enteric nervous system.Neurogastroenterology and Motility: The Official Journal of theEuropean Gastrointestinal Motility Society, 27(5):672–683, May 2015.→ pages 22[79] C. A. Dinarello, D. Novick, S. Kim, and G. Kaplanski. Interleukin-18and IL-18 Binding Protein. Frontiers in Immunology, 4, Oct. 2013.→ pages 24[80] S. Djuranovic, A. Nahvi, and R. Green. mirna-mediated genesilencing by translational repression followed by mrna deadenylationand decay. Science (New York, N.Y.), 336(6078):237–240, Apr. 2012.→ pages 105[81] E. Dolmatova, G. Spagnol, D. Boassa, J. R. Baum, K. Keith,C. Ambrosi, M. I. Kontaridis, P. L. Sorgen, G. E. Sosinsky, and H. S.Duffy. Cardiomyocyte atp release through pannexin 1 aids in earlyfibroblast activation. American Journal of Physiology. Heart andCirculatory Physiology, 303(10):H1208–1218, Nov. 2012. → pages 18129[82] M. Domercq, A. Perez-Samartin, D. Aparicio, E. Alberdi,O. Pampliega, and C. Matute. P2x7 receptors mediate ischemicdamage to oligodendrocytes. Glia, 58(6):730–740, Apr. 2010. →pages 23[83] G. Dvoriantchikova, D. Ivanov, Y. Panchin, and V. I. Shestopalov.Expression of pannexin family of proteins in the retina. FEBSLetters, 580(9):2178–2182, Apr. 2006. → pages 49, 59, 60[84] G. Dvoriantchikova, D. Ivanov, D. Barakat, A. Grinberg, R. Wen,V. Z. Slepak, and V. I. Shestopalov. Genetic ablation of pannexin1protects retinal neurons from ischemic injury. PloS One, 7(2):e31991,2012. → pages 23, 24, 25[85] Y. Elbaz-Alon, E. Rosenfeld-Gur, V. Shinder, A. H. Futerman,T. Geiger, and M. Schuldiner. A dynamic interface between vacuolesand mitochondria in yeast. Developmental Cell, 30(1):95–102, July2014. → pages 36[86] L. A. B. Elias, D. D. Wang, and A. R. Kriegstein. Gap junctionadhesion is necessary for radial migration in the neocortex. Nature,448(7156):901–907, Aug. 2007. → pages 71[87] M. R. Elliott, F. B. Chekeni, P. C. Trampont, E. R. Lazarowski,A. Kadl, S. F. Walk, D. Park, R. I. Woodson, M. Ostankovich,P. Sharma, J. J. Lysiak, T. K. Harden, N. Leitinger, and K. S.Ravichandran. Nucleotides released by apoptotic cells act as afind-me signal to promote phagocytic clearance. Nature, 461(7261):282–286, Sept. 2009. → pages 25[88] M. Escobar-Khondiker, M. Höllerhage, M.-P. Muriel, P. Champy,A. Bach, C. Depienne, G. Respondek, E. S. Yamada, A. Lannuzel,T. Yagi, E. C. Hirsch, W. H. Oertel, R. Jacob, P. P. Michel,M. Ruberg, and G. U. Höglinger. Annonacin, a naturalmitochondrial complex i inhibitor, causes tau pathology in culturedneurons. The Journal of Neuroscience: The Official Journal of theSociety for Neuroscience, 27(29):7827–7837, July 2007. → pages 117[89] I. Ezkurdia, J. M. Rodriguez, E. Carrillo-de Santa Pau, J. Vázquez,A. Valencia, and M. L. Tress. Most highly expressed protein-codinggenes have a single dominant isoform. Journal of Proteome Research,14(4):1880–1887, Apr. 2015. → pages 108130[90] M. M. Falk, L. K. Buehler, N. M. Kumar, and N. B. Gilula. Cell-freesynthesis and assembly of connexins into functional gap junctionmembrane channels. The EMBO journal, 16(10):2703–2716, May1997. → pages 6[91] R. Filadi, E. Greotti, G. Turacchio, A. Luini, T. Pozzan, andP. Pizzo. Mitofusin 2 ablation increases endoplasmicreticulum–mitochondria coupling. Proceedings of the NationalAcademy of Sciences, 112(17):E2174–E2181, Apr. 2015. → pages 33[92] L. Filippin, P. J. Magalhães, G. D. Benedetto, M. Colella, andT. Pozzan. Stable Interactions between Mitochondria andEndoplasmic Reticulum Allow Rapid Accumulation of Calcium in aSubpopulation of Mitochondria. Journal of Biological Chemistry, 278(40):39224–39234, Mar. 2003. → pages 38[93] E. J. Foss, D. Radulovic, S. A. Shaffer, D. R. Goodlett, L. Kruglyak,and A. Bedalov. Genetic Variation Shapes Protein Networks Mainlythrough Non-transcriptional Mechanisms. PLoS Biol, 9(9):e1001144,Sept. 2011. → pages 69[94] J. R. Friedman, L. L. Lackner, M. West, J. R. DiBenedetto,J. Nunnari, and G. K. Voeltz. Er tubules mark sites of mitochondrialdivision. Science (New York, N.Y.), 334(6054):358–362, Oct. 2011.→ pages 37[95] P. W. Furlow, S. Zhang, T. D. Soong, N. Halberg, H. Goodarzi,C. Mangrum, Y. G. Wu, O. Elemento, and S. F. Tavazoie.Mechanosensitive pannexin-1 channels mediate microvascularmetastatic cell survival. Nature Cell Biology, 17(7):943–952, July2015. → pages 27[96] E. J. Furshpan and D. D. Potter. Transmission at the giant motorsynapses of the crayfish. The Journal of Physiology, 145(2):289–325,Mar. 1959. → pages 1[97] J. L. Gabriel and G. W. Plaut. Inhibition of bovine heart nad-specificisocitrate dehydrogenase by reduced pyridine nucleotides:modulation of inhibition by adp, nad+, ca2+, citrate, and isocitrate.Biochemistry, 23(12):2773–2778, June 1984. → pages 39131[98] A. M. Garcia and C. Miller. Channel-mediated monovalent cationfluxes in isolated sarcoplasmic reticulum vesicles. The Journal ofGeneral Physiology, 83(6):819–839, June 1984. → pages 110[99] M. Gassmann, B. Grenacher, B. Rohde, and J. Vogel. Quantifyingwestern blots: pitfalls of densitometry. Electrophoresis, 30(11):1845–1855, June 2009. → pages 54, 81[100] F. N. Gellerich, Z. Gizatullina, S. Trumbeckaite, H. P. Nguyen,T. Pallas, O. Arandarcikaite, S. Vielhaber, E. Seppet, andF. Striggow. The regulation of oxphos by extramitochondrialcalcium. Biochimica Et Biophysica Acta, 1797(6-7):1018–1027, 2010Jun-Jul. → pages 40[101] F. G. Gervais, D. Xu, G. S. Robertson, J. P. Vaillancourt, Y. Zhu,J. Huang, A. LeBlanc, D. Smith, M. Rigby, M. S. Shearman, E. E.Clarke, H. Zheng, L. H. T. Van Der Ploeg, S. C. Ruffolo, N. A.Thornberry, S. Xanthoudakis, R. J. Zamboni, S. Roy, and D. W.Nicholson. Involvement of Caspases in Proteolytic Cleavage ofAlzheimer’s Amyloid-β Precursor Protein and Amyloidogenic AβPeptide Formation. Cell, 97(3):395–406, Apr. 1999. → pages 117[102] M. J. Gething, K. McCammon, and J. Sambrook. Expression ofwild-type and mutant forms of influenza hemagglutinin: the role offolding in intracellular transport. Cell, 46(6):939–950, Sept. 1986. →pages 6[103] C. Giorgi, K. Ito, H.-K. Lin, C. Santangelo, M. R. Wieckowski,M. Lebiedzinska, A. Bononi, M. Bonora, J. Duszynski, R. Bernardi,R. Rizzuto, C. Tacchetti, P. Pinton, and P. P. Pandolfi. Pmlregulates apoptosis at endoplasmic reticulum by modulating calciumrelease. Science (New York, N.Y.), 330(6008):1247–1251, Nov. 2010.→ pages 44[104] A. J. Giraldez, Y. Mishima, J. Rihel, R. J. Grocock, S. Van Dongen,K. Inoue, A. J. Enright, and A. F. Schier. Zebrafish mir-430promotes deadenylation and clearance of maternal mrnas. Science(New York, N.Y.), 312(5770):75–79, Apr. 2006. → pages 105[105] B. Glancy and R. S. Balaban. Role of mitochondrial ca2+ in theregulation of cellular energetics. Biochemistry, 51(14):2959–2973,Apr. 2012. → pages 39132[106] B. Glancy, W. T. Willis, D. J. Chess, and R. S. Balaban. Effect ofcalcium on the oxidative phosphorylation cascade in skeletal musclemitochondria. Biochemistry, 52(16):2793–2809, Apr. 2013. → pages74, 101[107] B. Glancy, W. T. Willis, D. J. Chess, and R. S. Balaban. Effect ofcalcium on the oxidative phosphorylation cascade in skeletal musclemitochondria. Biochemistry, 52(16):2793–2809, Apr. 2013. → pages39[108] S. Gödecke, C. Roderigo, C. R. Rose, B. H. Rauch, A. Gödecke, andJ. Schrader. Thrombin-induced atp release from human umbilicalvein endothelial cells. American Journal of Physiology. CellPhysiology, 302(6):C915–923, Mar. 2012. → pages 18, 110[109] V. A. Golovina and M. P. Blaustein. Spatially and FunctionallyDistinct Ca2+ Stores in Sarcoplasmic and Endoplasmic Reticulum.Science, 275(5306):1643–1648, Mar. 1997. → pages 101[110] M. Gonzàlez-Porta, A. Frankish, J. Rung, J. Harrow, andA. Brazma. Transcriptome analysis of human tissues and cell linesreveals one dominant transcript per gene. Genome Biology, 14(7):R70, 2013. → pages 108[111] A. Gross, X. M. Yin, K. Wang, M. C. Wei, J. Jockel, C. Milliman,H. Erdjument-Bromage, P. Tempst, and S. J. Korsmeyer. Caspasecleaved bid targets mitochondria and is required for cytochrome crelease, while bcl-xl prevents this release but not tumor necrosisfactor-r1/fas death. The Journal of Biological Chemistry, 274(2):1156–1163, Jan. 1999. → pages 33[112] C. Guardia-Laguarta, E. Area-Gomez, C. Rüb, Y. Liu, J. Magrané,D. Becker, W. Voos, E. A. Schon, and S. Przedborski. α-synuclein islocalized to mitochondria-associated er membranes. The Journal ofNeuroscience: The Official Journal of the Society for Neuroscience,34(1):249–259, Jan. 2014. → pages 30, 74, 116[113] M. Guescini, D. Sisti, M. B. L. Rocchi, L. Stocchi, and V. Stocchi. Anew real-time pcr method to overcome significant quantitativeinaccuracy due to slight amplification inhibition. BMCbioinformatics, 9:326, 2008. → pages 53133[114] A. Gürtler, N. Kunz, M. Gomolka, S. Hornhardt, A. A. Friedl,K. McDonald, J. E. Kohn, and A. Posch. Stain-Free technology as anormalization tool in Western blot analysis. Analytical Biochemistry,433(2):105–111, Feb. 2013. → pages 53, 56[115] G. Hajnóczky, G. Csordás, S. Das, C. Garcia-Perez, M. Saotome,S. Sinha Roy, and M. Yi. Mitochondrial calcium signalling and celldeath: approaches for assessing the role of mitochondrial ca2+uptake in apoptosis. Cell Calcium, 40(5-6):553–560, 2006 Nov-Dec.→ pages 113[116] M. Hamasaki, N. Furuta, A. Matsuda, A. Nezu, A. Yamamoto,N. Fujita, H. Oomori, T. Noda, T. Haraguchi, Y. Hiraoka, A. Amano,and T. Yoshimori. Autophagosomes form at ER-mitochondriacontact sites. Nature, 495(7441):389–393, Mar. 2013. → pages 100[117] D. B. Hansen, Z.-C. Ye, K. Calloe, T. H. Braunstein, J. P. Hofgaard,B. R. Ransom, M. S. Nielsen, and N. MacAulay. Activation,permeability, and inhibition of astrocytic and neuronal large pore(hemi)channels. The Journal of Biological Chemistry, 289(38):26058–26073, Sept. 2014. → pages 19[118] T. Hayashi and T.-P. Su. Sigma-1 receptor chaperones at theer-mitochondrion interface regulate ca(2+) signaling and cellsurvival. Cell, 131(3):596–610, Nov. 2007. → pages 33, 74, 89[119] T. Hayashi, R. Rizzuto, G. Hajnoczky, and T.-P. Su. MAM: morethan just a housekeeper. Trends in Cell Biology, 19(2):81–88, Feb.2009. → pages 30, 74, 116[120] D. S. He, J. X. Jiang, S. M. Taffet, and J. M. Burt. Formation ofheteromeric gap junction channels by connexins 40 and 43 invascular smooth muscle cells. Proceedings of the National Academyof Sciences of the United States of America, 96(11):6495–6500, May1999. → pages 6[121] L. Hedskog, C. M. Pinho, R. Filadi, A. Rönnbäck, L. Hertwig,B. Wiehager, P. Larssen, S. Gellhaar, A. Sandebring, M. Westerlund,C. Graff, B. Winblad, D. Galter, H. Behbahani, P. Pizzo, E. Glaser,and M. Ankarcrona. Modulation of the endoplasmicreticulum–mitochondria interface in Alzheimer’s disease and relatedmodels. Proceedings of the National Academy of Sciences, 110(19):7916–7921, July 2013. → pages 117, 118134[122] M. Herrmann, H. M. Lorenz, R. Voll, M. Grünke, W. Woith, andJ. R. Kalden. A rapid and simple method for the isolation ofapoptotic DNA fragments. Nucleic Acids Research, 22(24):5506–5507, Dec. 1994. → pages 82[123] A. M. Hofer and I. Schulz. Quantification of intraluminal free [ca] inthe agonist-sensitive internal calcium store using compartmentalizedfluorescent indicators: some considerations. Cell Calcium, 20(3):235–242, Sept. 1996. → pages 109[124] G. U. Höglinger, A. Lannuzel, M. E. Khondiker, P. P. Michel,C. Duyckaerts, J. Féger, P. Champy, A. Prigent, F. Medja,A. Lombes, W. H. Oertel, M. Ruberg, and E. C. Hirsch. Themitochondrial complex i inhibitor rotenone triggers a cerebraltauopathy. Journal of Neurochemistry, 95(4):930–939, Nov. 2005. →pages 117[125] C. Hönscher, M. Mari, K. Auffarth, M. Bohnert, J. Griffith,W. Geerts, M. van der Laan, M. Cabrera, F. Reggiori, andC. Ungermann. Cellular metabolism regulates contact sites betweenvacuoles and mitochondria. Developmental Cell, 30(1):86–94, July2014. → pages 36[126] Y. Huang and L. Mucke. Alzheimer mechanisms and therapeuticstrategies. Cell, 148(6):1204–1222, Mar. 2012. → pages 116[127] V. Hung, P. Zou, H.-W. Rhee, N. D. Udeshi, V. Cracan, T. Svinkina,S. A. Carr, V. K. Mootha, and A. Y. Ting. Proteomic mapping ofthe human mitochondrial intermembrane space in live cells viaratiometric apex tagging. Molecular Cell, 55(2):332–341, July 2014.→ pages 34[128] S. M. Hurtley and A. Helenius. Protein oligomerization in theendoplasmic reticulum. Annual Review of Cell Biology, 5:277–307,1989. → pages 6[129] M. Hüttemann, I. Lee, L. Samavati, H. Yu, and J. W. Doan.Regulation of mitochondrial oxidative phosphorylation through cellsignaling. Biochimica Et Biophysica Acta, 1773(12):1701–1720, Dec.2007. → pages 40, 42[130] R. Iglesias, G. Dahl, F. Qiu, D. C. Spray, and E. Scemes. Pannexin1: the molecular substrate of astrocyte "hemichannels". The Journal135of Neuroscience: The Official Journal of the Society forNeuroscience, 29(21):7092–7097, May 2009. → pages 18[131] R. Iglesias, D. C. Spray, and E. Scemes. Mefloquine blockade ofpannexin1 currents: resolution of a conflict. Cell Communication &Adhesion, 16(5-6):131–137, Dec. 2009. → pages 20[132] E. Ingerman, E. M. Perkins, M. Marino, J. A. Mears, J. M.McCaffery, J. E. Hinshaw, and J. Nunnari. Dnm1 forms spirals thatare structurally tailored to fit mitochondria. The Journal of CellBiology, 170(7):1021–1027, Sept. 2005. → pages 37[133] M. Ishikawa, T. Iwamoto, T. Nakamura, A. Doyle, S. Fukumoto, andY. Yamada. Pannexin 3 functions as an ER Ca(2+) channel,hemichannel, and gap junction to promote osteoblast differentiation.The Journal of Cell Biology, 193(7):1257–1274, June 2011. → pages7, 9, 10, 27, 28, 29, 49, 71, 74, 101, 109, 114[134] T. Iwamoto, T. Nakamura, A. Doyle, M. Ishikawa, S. de Vega,S. Fukumoto, and Y. Yamada. Pannexin 3 regulates intracellularatp/camp levels and promotes chondrocyte differentiation. TheJournal of Biological Chemistry, 285(24):18948–18958, June 2010. →pages 27, 28[135] J. X. Jiang and D. A. Goodenough. Heteromeric connexons in lensgap junction channels. Proceedings of the National Academy ofSciences of the United States of America, 93(3):1287–1291, Feb.1996. → pages 6[136] H. Y. Jin and C. Xiao. Microrna mechanisms of action: What havewe learned from mice? Frontiers in Genetics, 6:328, 2015. → pages105[137] J. D. Johnston and M. D. Brand. Stimulation of the respiration rateof rat liver mitochondria by sub-micromolar concentrations ofextramitochondrial ca2+. The Biochemical Journal, 245(1):217–222,July 1987. → pages 74, 101[138] H. C. Jones and R. F. Keep. Brain fluid calcium concentration andresponse to acute hypercalcaemia during development in the rat.The Journal of Physiology, 402:579–593, Aug. 1988. → pages 109136[139] L. S. Jouaville, P. Pinton, C. Bastianutto, G. A. Rutter, andR. Rizzuto. Regulation of mitochondrial atp synthesis by calcium:evidence for a long-term metabolic priming. Proceedings of theNational Academy of Sciences of the United States of America, 96(24):13807–13812, Nov. 1999. → pages 38, 74, 101[140] Y. Kanno and W. R. Loewenstein. Intercellular diffusion. Science(New York, N.Y.), 143(3609):959–960, Feb. 1964. → pages 1[141] Y. Kanno and W. R. Loewenstein. Cell-to-cell passage of largemolecules. Nature, 212(5062):629–630, Nov. 1966. → pages 1[142] M. Karbowski and R. J. Youle. Dynamics of mitochondrialmorphology in healthy cells and during apoptosis. Cell Death &Differentiation, 10(8):870–880, 2003. → pages 115[143] M. Karbowski, S.-Y. Jeong, and R. J. Youle. Endophilin B1 isrequired for the maintenance of mitochondrial morphology. TheJournal of Cell Biology, 166(7):1027–1039, Sept. 2004. → pages 115[144] E. T. Kase, N. Nikolić, S. S. Bakke, K. K. Bogen, V. Aas, G. H.Thoresen, and A. C. Rustan. Remodeling of oxidative energymetabolism by galactose improves glucose handling and metabolicswitching in human skeletal muscle cells. PloS One, 8(4):e59972,2013. → pages 94, 112[145] S. M. Khan, D. S. Cassarino, N. N. Abramova, P. M. Keeney, M. K.Borland, P. A. Trimmer, C. T. Krebs, J. C. Bennett, J. K. Parks,R. H. Swerdlow, W. D. Parker, and J. P. Bennett. Alzheimer’sdisease cybrids replicate beta-amyloid abnormalities through celldeath pathways. Annals of Neurology, 48(2):148–155, Aug. 2000. →pages 117[146] S. Kidd and M. W. Young. Transposon-dependent mutantphenotypes at the notch locus of drosophila. Nature, 323(6083):89–91, 1986 Sep 4-10. → pages 3[147] M. A. Kiebler and G. J. Bassell. Neuronal RNA Granules: Moversand Makers. Neuron, 51(6):685–690, Sept. 2006. → pages 69[148] M.-C. Kienitz, K. Bender, R. Dermietzel, L. Pott, and G. Zoidl.Pannexin 1 constitutes the large conductance cation channel ofcardiac myocytes. The Journal of Biological Chemistry, 286(1):290–298, Jan. 2011. → pages 110137[149] Y. Kijima, A. Saito, T. L. Jetton, M. A. Magnuson, and S. Fleischer.Different intracellular localization of inositol 1,4,5-trisphosphate andryanodine receptors in cardiomyocytes. The Journal of BiologicalChemistry, 268(5):3499–3506, Feb. 1993. → pages 101[150] M.-S. Kim, S. M. Pinto, D. Getnet, R. S. Nirujogi, S. S. Manda,R. Chaerkady, A. K. Madugundu, D. S. Kelkar, R. Isserlin, S. Jain,J. K. Thomas, B. Muthusamy, P. Leal-Rojas, P. Kumar, N. A.Sahasrabuddhe, L. Balakrishnan, J. Advani, B. George, S. Renuse,L. D. N. Selvan, A. H. Patil, V. Nanjappa, A. Radhakrishnan,S. Prasad, T. Subbannayya, R. Raju, M. Kumar, S. K.Sreenivasamurthy, A. Marimuthu, G. J. Sathe, S. Chavan, K. K.Datta, Y. Subbannayya, A. Sahu, S. D. Yelamanchi, S. Jayaram,P. Rajagopalan, J. Sharma, K. R. Murthy, N. Syed, R. Goel, A. A.Khan, S. Ahmad, G. Dey, K. Mudgal, A. Chatterjee, T.-C. Huang,J. Zhong, X. Wu, P. G. Shaw, D. Freed, M. S. Zahari, K. K.Mukherjee, S. Shankar, A. Mahadevan, H. Lam, C. J. Mitchell, S. K.Shankar, P. Satishchandra, J. T. Schroeder, R. Sirdeshmukh,A. Maitra, S. D. Leach, C. G. Drake, M. K. Halushka, T. S. K.Prasad, R. H. Hruban, C. L. Kerr, G. D. Bader, C. A.Iacobuzio-Donahue, H. Gowda, and A. Pandey. A draft map of thehuman proteome. Nature, 509(7502):575–581, May 2014. → pages 69[151] Y. Kirichok, G. Krapivinsky, and D. E. Clapham. The mitochondrialcalcium uniporter is a highly selective ion channel. Nature, 427(6972):360–364, Jan. 2004. → pages 37[152] O. Kopach, I. Kruglikov, T. Pivneva, N. Voitenko, and N. Fedirko.Functional coupling between ryanodine receptors, mitochondria andca(2+) atpases in rat submandibular acinar cells. Cell Calcium, 43(5):469–481, May 2008. → pages 101[153] B. Kornmann, E. Currie, S. R. Collins, M. Schuldiner, J. Nunnari,J. S. Weissman, and P. Walter. An ER-Mitochondria TetheringComplex Revealed by a Synthetic Biology Screen. Science (NewYork, N.Y.), 325(5939):477–481, July 2009. → pages 31, 36[154] B. Kornmann, C. Osman, and P. Walter. The conserved GTPaseGem1 regulates endoplasmic reticulum–mitochondria connections.Proceedings of the National Academy of Sciences, 108(34):14151–14156, Aug. 2011. → pages 31138[155] M. Koval, S. A. Molina, and J. M. Burt. Mix and match:investigating heteromeric and heterotypic gap junction channels inmodel systems and native tissues. FEBS letters, 588(8):1193–1204,Apr. 2014. → pages 7[156] A. M. Krichevsky and K. S. Kosik. Neuronal RNA Granules: A Linkbetween RNA Localization and Stimulation-Dependent Translation.Neuron, 32(4):683–696, Nov. 2001. → pages 106[157] S. N. Krishnan, E. Frei, G. P. Swain, and R. J. Wyman. Passover: agene required for synaptic connectivity in the giant fiber system ofdrosophila. Cell, 73(5):967–977, June 1993. → pages 3[158] G. Kroemer, L. Galluzzi, and C. Brenner. Mitochondrial membranepermeabilization in cell death. Physiological Reviews, 87(1):99–163,Jan. 2007. → pages 44[159] S. Kumar, B. J. van Raam, G. S. Salvesen, and P. Cieplak. Caspasecleavage sites in the human proteome: Caspdb, a database ofpredicted substrates. PloS One, 9(10):e110539, 2014. → pages 111[160] S. Kurtenbach, N. Prochnow, S. Kurtenbach, J. Klooster, C. Zoidl,R. Dermietzel, M. Kamermans, and G. Zoidl. Pannexin1 channelproteins in the zebrafish retina have shared and unique properties.PloS One, 8(10):e77722, 2013. → pages 5, 48[161] A. M. Labrousse, M. D. Zappaterra, D. A. Rube, and A. M. van derBliek. C. elegans dynamin-related protein drp-1 controls severing ofthe mitochondrial outer membrane. Molecular Cell, 4(5):815–826,Nov. 1999. → pages 37[162] C. L. Ladner, J. Yang, R. J. Turner, and R. A. Edwards. Visiblefluorescent detection of proteins in polyacrylamide gels withoutstaining. Analytical biochemistry, 326(1):13–20, Mar. 2004. → pages53, 56[163] S. Lahiri, J. T. Chao, S. Tavassoli, A. K. O. Wong, V. Choudhary,B. P. Young, C. J. R. Loewen, and W. A. Prinz. A ConservedEndoplasmic Reticulum Membrane Protein Complex (EMC)Facilitates Phospholipid Transfer from the ER to Mitochondria.PLoS Biol, 12(10):e1001969, Oct. 2014. → pages 31, 32, 36139[164] C. Lai, J. Bechberger, R. Thompson, B. MacVicar, R. Bruzzone, andC. Naus. Tumor-suppressive effects of pannexin 1 in C6 glioma cells.Cancer research, 67(4):1545–1554, 2007. → pages 9, 27, 49, 50[165] C. Lai, J. Bechberger, and C. Naus. Pannexin2 as a novel growthregulator in C6 glioma cells. Oncogene, 28(49):4402–4408, 2009. →pages 11, 16, 26, 27, 29, 50, 70, 71, 82, 92, 99, 113[166] A. K. M. Lam and A. Galione. The endoplasmic reticulum andjunctional membrane communication during calcium signaling.Biochimica et Biophysica Acta (BBA) - Molecular Cell Research,1833(11):2542–2559, Nov. 2013. → pages 110[167] S. Langlois, X. Xiang, K. Young, B. J. Cowan, S. Penuela, and K. N.Cowan. Pannexin 1 and pannexin 3 channels regulate skeletal musclemyoblast proliferation and differentiation. The Journal of BiologicalChemistry, 289(44):30717–30731, Oct. 2014. → pages 28[168] V. B. Lawlis and T. E. Roche. Regulation of bovine kidneyalpha-ketoglutarate dehydrogenase complex by calcium ion andadenine nucleotides. effects on s0.5 for alpha-ketoglutarate.Biochemistry, 20(9):2512–2518, Apr. 1981. → pages 39[169] H. T. Le, W. C. Sin, S. Lozinsky, J. Bechberger, J. L. Vega, X. Q.Guo, J. C. Sáez, and C. C. Naus. Gap junction intercellularcommunication mediated by connexin43 in astrocytes is essential fortheir resistance to oxidative stress. The Journal of BiologicalChemistry, 289(3):1345–1354, Jan. 2014. → pages 51[170] M. Le Vasseur, J. Lelowski, J. F. Bechberger, W.-C. Sin, and C. C.Naus. Pannexin 2 protein expression is not restricted to the cns.Frontiers in Cellular Neuroscience, 8:392, 2014. → pages 16, 74, 82,89, 99[171] M. Lebiedzinska, G. Szabadkai, A. W. E. Jones, J. Duszynski, andM. R. Wieckowski. Interactions between the endoplasmic reticulum,mitochondria, plasma membrane and other subcellular organelles.The International Journal of Biochemistry & Cell Biology, 41(10):1805–1816, Oct. 2009. → pages 30, 74, 116[172] I. Lee, A. R. Salomon, S. Ficarro, I. Mathes, F. Lottspeich, L. I.Grossman, and M. Hüttemann. camp-dependent tyrosinephosphorylation of subunit i inhibits cytochrome c oxidase activity.140The Journal of Biological Chemistry, 280(7):6094–6100, Feb. 2005.→ pages 42[173] I. Lee, A. R. Salomon, K. Yu, J. W. Doan, L. I. Grossman, andM. Hüttemann. New Prospects for an Old Enzyme: MammalianCytochrome c Is Tyrosine-Phosphorylated in Vivo†. Biochemistry, 45(30):9121–9128, Aug. 2006. → pages 42[174] R. C. Lee, R. L. Feinbaum, and V. Ambros. The c. elegansheterochronic gene lin-4 encodes small rnas with antisensecomplementarity to lin-14. Cell, 75(5):843–854, Dec. 1993. → pages105[175] H. Li, T. F. Liu, A. Lazrak, C. Peracchia, G. S. Goldberg, P. D.Lampe, and R. G. Johnson. Properties and regulation of gapjunctional hemichannels in the plasma membranes of cultured cells.The Journal of Cell Biology, 134(4):1019–1030, Aug. 1996. → pages109[176] S. Li, I. Bjelobaba, Z. Yan, M. Kucka, M. Tomic, and S. S.Stojilkovic. Expression and roles of pannexins in atp release in thepituitary gland. Endocrinology, 152(6):2342–2352, June 2011. →pages 22, 28[177] S. Li, M. Tomić, and S. S. Stojilkovic. Characterization of NovelPannexin 1 Isoforms from Rat Pituitary Cells and their Associationwith ATP-gated P2X Channels. General and comparativeendocrinology, 174(2):202–210, Nov. 2011. → pages 12, 15, 28[178] T. C. Linn, F. H. Pettit, F. Hucho, and L. J. Reed. Alpha-keto aciddehydrogenase complexes. xi. comparative studies of regulatoryproperties of the pyruvate dehydrogenase complexes from kidney,heart, and liver mitochondria. Proceedings of the National Academyof Sciences of the United States of America, 64(1):227–234, Sept.1969. → pages 39[179] T. C. Linn, F. H. Pettit, and L. J. Reed. Alpha-keto aciddehydrogenase complexes. x. regulation of the activity of thepyruvate dehydrogenase complex from beef kidney mitochondria byphosphorylation and dephosphorylation. Proceedings of the NationalAcademy of Sciences of the United States of America, 62(1):234–241,Jan. 1969. → pages 39141[180] O. Litvin, A. Tiunova, Y. Connell-Alberts, Y. Panchin, andA. Baranova. What is hidden in the pannexin treasure trove: thesneak peek and the guesswork. Journal of cellular and molecularmedicine, 10(3):613–634, 2006. → pages 26[181] D. Locke, N. Perusinghe, T. Newman, H. Jayatilake, W. H. Evans,and P. Monaghan. Developmental expression and assembly ofconnexins into homomeric and heteromeric gap junctionhemichannels in the mouse mammary gland. Journal of CellularPhysiology, 183(2):228–237, May 2000. → pages 6[182] S. Locovei, L. Bao, and G. Dahl. Pannexin 1 in erythrocytes:function without a gap. Proceedings of the National Academy ofSciences of the United States of America, 103(20):7655–7659, May2006. → pages 17, 18[183] S. Locovei, J. Wang, and G. Dahl. Activation of pannexin 1 channelsby ATP through P2Y receptors and by cytoplasmic calcium. FEBSLetters, 580(1):239–244, Jan. 2006. → pages 18, 110, 111[184] W. R. Loewenstein and Y. Kanno. Studies on an epithelial (gland)cell junction. i. modifications of surface membrane permeability. TheJournal of Cell Biology, 22:565–586, Sept. 1964. → pages 1[185] A. W. Lohman, I. L. Leskov, J. T. Butcher, S. R. Johnstone, T. A.Stokes, D. Begandt, L. J. DeLalio, A. K. Best, S. Penuela,N. Leitinger, K. S. Ravichandran, K. Y. Stokes, and B. E. Isakson.Pannexin 1 channels regulate leukocyte emigration through thevenous endothelium during acute inflammation. NatureCommunications, 6:7965, 2015. → pages 25[186] E. Lundberg, L. Fagerberg, D. Klevebring, I. Matic, T. Geiger,J. Cox, C. Algenäs, J. Lundeberg, M. Mann, and M. Uhlen. Definingthe transcriptome and proteome in three functionally different humancell lines. Molecular Systems Biology, 6:450, Dec. 2010. → pages 104[187] W. Ma, H. Hui, P. Pelegrin, and A. Surprenant. Pharmacologicalcharacterization of pannexin-1 currents expressed in mammaliancells. The Journal of Pharmacology and Experimental Therapeutics,328(2):409–418, Feb. 2009. → pages 15, 109, 110[188] W. Ma, V. Compan, W. Zheng, E. Martin, R. A. North,A. Verkhratsky, and A. Surprenant. Pannexin 1 forms an142anion-selective channel. Pflügers Archiv: European Journal ofPhysiology, 463(4):585–592, Apr. 2012. → pages 18[189] C. A. Mannella, M. Marko, P. Penczek, D. Barnard, and J. Frank.The internal compartmentation of rat-liver mitochondria:tomographic study using the high-voltage transmission electronmicroscope. Microscopy Research and Technique, 27(4):278–283,Mar. 1994. → pages 30[190] L. D. Marroquin, J. Hynes, J. A. Dykens, J. D. Jamieson, andY. Will. Circumventing the crabtree effect: replacing media glucosewith galactose increases susceptibility of hepg2 cells to mitochondrialtoxicants. Toxicological Sciences: An Official Journal of the Societyof Toxicology, 97(2):539–547, June 2007. → pages 94, 112[191] J. D. Martell, T. J. Deerinck, Y. Sancak, T. L. Poulos, V. K.Mootha, G. E. Sosinsky, M. H. Ellisman, and A. Y. Ting.Engineered ascorbate peroxidase as a genetically encoded reporterfor electron microscopy. Nature Biotechnology, 30(11):1143–1148,Nov. 2012. → pages 34[192] F. Martinon, K. Burns, and J. Tschopp. The Inflammasome.Molecular Cell, 10(2):417–426, Aug. 2002. → pages 24[193] J. G. McCormack and R. M. Denton. The effects of calcium ions andadenine nucleotides on the activity of pig heart 2-oxoglutaratedehydrogenase complex. Biochemical Journal, 180(3):533–544, June1979. → pages 39[194] J. G. McCormack, A. P. Halestrap, and R. M. Denton. Role ofcalcium ions in regulation of mammalian intramitochondrialmetabolism. Physiological Reviews, 70(2):391–425, Apr. 1990. →pages 39, 74, 101[195] K. Medler and E. L. Gleason. Mitochondrial Ca2+ BufferingRegulates Synaptic Transmission Between Retinal Amacrine Cells.Journal of Neurophysiology, 87(3):1426–1439, Mar. 2002. → pages109[196] R. C. N. Melo, E. Morgan, R. Monahan-Earley, A. M. Dvorak, andP. F. Weller. Pre-embedding immunogold labeling to optimizeprotein localization at subcellular compartments and membrane143microdomains of leukocytes. Nature Protocols, 9(10):2382–2394, Oct.2014. → pages 79[197] C. C. P. Mendes, D. A. Gomes, M. Thompson, N. C. Souto, T. S.Goes, A. M. Goes, M. A. Rodrigues, M. V. Gomez, M. H.Nathanson, and M. F. Leite. The Type III Inositol1,4,5-Trisphosphate Receptor Preferentially Transmits ApoptoticCa2+ Signals into Mitochondria. Journal of Biological Chemistry,280(49):40892–40900, Sept. 2005. → pages 32, 45, 101[198] S. L. Mironov and N. Symonchuk. Er vesicles and mitochondriamove and communicate at synapses. Journal of Cell Science, 119(Pt23):4926–4934, Dec. 2006. → pages 30, 109[199] M. Monni, L. Corazzi, G. Migliorati, and R. Roberti. Respiratorystate and phosphatidylserine import in brain mitochondria in vitro.The Journal of Membrane Biology, 173(2):97–105, Jan. 2000. →pages 92[200] V. K. Mootha, A. E. Arai, and R. S. Balaban. Maximum oxidativephosphorylation capacity of the mammalian heart. AmericanJournal of Physiology - Heart and Circulatory Physiology, 272(2):H769–H775, Feb. 1997. → pages 40[201] P. I. Moreira, L. M. Sayre, X. Zhu, A. Nunomura, M. A. Smith, andG. Perry. Detection and localization of markers of oxidative stress byin situ methods: application in the study of alzheimer disease.Methods in Molecular Biology (Clifton, N.J.), 610:419–434, 2010. →pages 117[202] R. Moreno-Sánchez, B. A. Hogue, and R. G. Hansford. Influence ofnad-linked dehydrogenase activity on flux through oxidativephosphorylation. The Biochemical Journal, 268(2):421–428, June1990. → pages 40[203] G. M. Morriss and D. A. New. Effect of oxygen concentration onmorphogenesis of cranial neural folds and neural crest in cultured ratembryos. Journal of Embryology and Experimental Morphology, 54:17–35, Dec. 1979. → pages 112[204] L. S. Musil and D. A. Goodenough. Multisubunit assembly of anintegral plasma membrane channel protein, gap junction connexin43,144occurs after exit from the er. Cell, 74(6):1065–1077, Sept. 1993. →pages 6[205] J. Napiwotzki, K. Shinzawa-Itoh, S. Yoshikawa, and B. Kadenbach.Atp and adp bind to cytochrome c oxidase and regulate its activity.Biological Chemistry, 378(9):1013–1021, Sept. 1997. → pages 42[206] Y. Nemoto and P. De Camilli. Recruitment of an alternativelyspliced form of synaptojanin 2 to mitochondria by the interactionwith the pdz domain of a mitochondrial outer membrane protein.The EMBO journal, 18(11):2991–3006, June 1999. → pages 114[207] M. Neuspiel, R. Zunino, S. Gangaraju, P. Rippstein, andH. McBride. Activated Mitofusin 2 Signals Mitochondrial Fusion,Interferes with Bax Activation, and Reduces Susceptibility toRadical Induced Depolarization. Journal of Biological Chemistry,280(26):25060–25070, Jan. 2005. → pages 76[208] M. Neuspiel, A. C. Schauss, E. Braschi, R. Zunino, P. Rippstein,R. A. Rachubinski, M. A. Andrade-Navarro, and H. M. McBride.Cargo-selected transport from the mitochondria to peroxisomes ismediated by vesicular carriers. Current biology: CB, 18(2):102–108,Jan. 2008. → pages 35[209] T. T. Nguyen, A. Lewandowska, J.-Y. Choi, D. F. Markgraf,M. Junker, M. Bilgin, C. S. Ejsing, D. R. Voelker, T. A. Rapoport,and J. M. Shaw. Gem1 and ermes do not directly affectphosphatidylserine transport from er to mitochondria ormitochondrial inheritance. Traffic (Copenhagen, Denmark), 13(6):880–890, June 2012. → pages 36[210] K. Ning and A. I. Nesvizhskii. The utility of massspectrometry-based proteomic data for validation of novel alternativesplice forms reconstructed from rna-seq data: a preliminaryassessment. BMC bioinformatics, 11 Suppl 11:S14, 2010. → pages108[211] T. Nishikawa, T. Ota, and T. Isogai. Prediction whether a humancdna sequence contains initiation codon by combining statisticalinformation and similarity with protein sequences. Bioinformatics(Oxford, England), 16(11):960–967, Nov. 2000. → pages 16145[212] G. Odorizzi, C. R. Cowles, and S. D. Emr. The ap-3 complex: a coatof many colours. Trends in Cell Biology, 8(7):282–288, July 1998. →pages 100[213] M. Ott, J. D. Robertson, V. Gogvadze, B. Zhivotovsky, andS. Orrenius. Cytochrome c release from mitochondria proceeds by atwo-step process. Proceedings of the National Academy of Sciences ofthe United States of America, 99(3):1259–1263, Feb. 2002. → pages115[214] Y.-B. Ouyang and R. G. Giffard. ER-Mitochondria Crosstalk duringCerebral Ischemia: Molecular Chaperones and ER-MitochondrialCalcium Transfer. International Journal of Cell Biology, 2012, 2012.→ pages 30, 74, 116[215] Y. Panchin. Evolution of gap junction proteins–the pannexinalternative. Journal of experimental biology, 208(8):1415–1419, 2005.→ pages 2, 48, 49[216] Y. Panchin, I. Kelmanson, M. Matz, K. Lukyanov, N. Usman, andS. Lukyanov. A ubiquitous family of putative gap junctionmolecules. Current Biology: CB, 10(13):R473–474, June 2000. →pages 3, 5, 48, 74[217] A. V. Panov and R. C. Scaduto. Substrate specific effects of calciumon metabolism of rat heart mitochondria. The American Journal ofPhysiology, 270(4 Pt 2):H1398–1406, Apr. 1996. → pages 74, 101[218] S. Patergnani, J. M. Suski, C. Agnoletto, A. Bononi, M. Bonora,E. De Marchi, C. Giorgi, S. Marchi, S. Missiroli, F. Poletti,A. Rimessi, J. Duszynski, M. R. Wieckowski, and P. Pinton. Calciumsignaling around Mitochondria Associated Membranes (MAMs). CellCommunication and Signaling : CCS, 9:19, Sept. 2011. → pages 109[219] S. Patergnani, C. Giorgi, S. Maniero, S. Missiroli, P. Maniscalco,I. Bononi, F. Martini, G. Cavallesco, M. Tognon, and P. Pinton. Theendoplasmic reticulum mitochondrial calcium cross talk isdownregulated in malignant pleural mesothelioma cells and plays acritical role in apoptosis inhibition. Oncotarget, 6(27):23427–23444,Sept. 2015. → pages 42146[220] D. L. Paul. Molecular cloning of cdna for rat liver gap junctionprotein. The Journal of Cell Biology, 103(1):123–134, July 1986. →pages 2[221] P. Pecina, G. G. Borisenko, N. A. Belikova, Y. Y. Tyurina,A. Pecinova, I. Lee, A. K. Samhan-Arias, K. Przyklenk, V. E.Kagan, and M. Hüttemann. Phosphomimetic substitution ofcytochrome c tyrosine 48 decreases respiration and binding tocardiolipin and abolishes ability to trigger downstream caspaseactivation. Biochemistry, 49(31):6705–6714, Aug. 2010. → pages 42[222] P. Pelegrin and A. Surprenant. Pannexin-1 mediates large poreformation and interleukin-1beta release by the atp-gated p2x7receptor. The EMBO journal, 25(21):5071–5082, Nov. 2006. → pages21, 24, 111[223] P. Pelegrin, C. Barroso-Gutierrez, and A. Surprenant. P2x7 receptordifferentially couples to distinct release pathways for il-1beta inmouse macrophage. Journal of Immunology (Baltimore, Md.: 1950),180(11):7147–7157, June 2008. → pages 24[224] S. Penuela, R. Bhalla, X.-Q. Gong, K. N. Cowan, S. J. Celetti, B. J.Cowan, D. Bai, Q. Shao, and D. W. Laird. Pannexin 1 and pannexin3 are glycoproteins that exhibit many distinct characteristics fromthe connexin family of gap junction proteins. Journal of Cell Science,120(Pt 21):3772–3783, Nov. 2007. → pages 11, 20, 69, 70, 100, 106[225] S. Penuela, R. Bhalla, K. Nag, and D. W. Laird. Glycosylationregulates pannexin intermixing and cellular localization. MolecularBiology of the Cell, 20(20):4313–4323, Oct. 2009. → pages 9, 10, 11,16, 20, 49, 70, 71, 100[226] S. Penuela, L. Gyenis, A. Ablack, J. M. Churko, A. C. Berger, D. W.Litchfield, J. D. Lewis, and D. W. Laird. Loss of pannexin 1attenuates melanoma progression by reversion to a melanocyticphenotype. The Journal of Biological Chemistry, 287(34):29184–29193, Aug. 2012. → pages 27[227] S. Penuela, A. W. Lohman, W. Lai, L. Gyenis, D. W. Litchfield,B. E. Isakson, and D. W. Laird. Diverse post-translationalmodifications of the pannexin family of channel-forming proteins.Channels (Austin, Tex.), 8(2):124–130, 2014. → pages 25, 111147[228] A. K. Percy, J. F. Moore, M. A. Carson, and C. J. Waechter.Characterization of brain phosphatidylserine decarboxylase:localization in the mitochondrial inner membrane. Archives ofBiochemistry and Biophysics, 223(2):484–494, June 1983. → pages 35[229] M. W. Pfaffl. A new mathematical model for relative quantificationin real-time rt-pcr. Nucleic acids research, 29(9):e45, May 2001. →pages 53[230] P. Phelan and T. A. Starich. Innexins get into the gap. BioEssays:News and Reviews in Molecular, Cellular and DevelopmentalBiology, 23(5):388–396, May 2001. → pages 2[231] P. Phelan, J. P. Bacon, J. A. Davies, L. A. Stebbings, M. G.Todman, L. Avery, R. A. Baines, T. M. Barnes, C. Ford, S. Hekimi,R. Lee, J. E. Shaw, T. A. Starich, K. D. Curtin, Y. A. Sun, and R. J.Wyman. Innexins: a family of invertebrate gap-junction proteins.Trends in genetics: TIG, 14(9):348–349, Sept. 1998. → pages 3[232] P. Pinton, C. Giorgi, R. Siviero, E. Zecchini, and R. Rizzuto.Calcium and apoptosis: ER-mitochondria Ca2+ transfer in thecontrol of apoptosis. Oncogene, 27(50):6407–6418, Oct. 2008. →pages 42[233] I. K. H. Poon, Y.-H. Chiu, A. J. Armstrong, J. M. Kinchen, I. J.Juncadella, D. A. Bayliss, and K. S. Ravichandran. Unexpected linkbetween an antibiotic, pannexin channels and apoptosis. Nature, 507(7492):329–334, Mar. 2014. → pages 72[234] C. N. Poston, S. C. Krishnan, and C. R. Bazemore-Walker. In-depthproteomic analysis of mammalian mitochondria-associatedmembranes (mam). Journal of Proteomics, 79:219–230, Feb. 2013.→ pages 34[235] N. Prochnow, S. Hoffmann, R. Vroman, J. Klooster, S. Bunse,M. Kamermans, R. Dermietzel, and G. Zoidl. Pannexin1 in the outerretina of the zebrafish, danio rerio. Neuroscience, 162(4):1039–1054,Sept. 2009. → pages 110[236] F. Qiu, J. Wang, D. C. Spray, E. Scemes, and G. Dahl. Twonon-vesicular atp release pathways in the mouse erythrocytemembrane. FEBS letters, 585(21):3430–3435, Nov. 2011. → pages 18148[237] Y. Qu, S. Misaghi, K. Newton, L. L. Gilmour, S. Louie, J. E. Cupp,G. R. Dubyak, D. Hackos, and V. M. Dixit. Pannexin-1 is requiredfor atp release during apoptosis but not for inflammasome activation.Journal of Immunology (Baltimore, Md.: 1950), 186(11):6553–6561,June 2011. → pages 25, 111[238] A. Raj, C. S. Peskin, D. Tranchina, D. Y. Vargas, and S. Tyagi.Stochastic mRNA Synthesis in Mammalian Cells. PLoS Biology, 4(10):e309, 2006. → pages 69, 106[239] G. A. Ransford, N. Fregien, F. Qiu, G. Dahl, G. E. Conner, andM. Salathe. Pannexin 1 contributes to atp release in airway epithelia.American Journal of Respiratory Cell and Molecular Biology, 41(5):525–534, Nov. 2009. → pages 10, 18[240] E. Rapizzi, P. Pinton, G. Szabadkai, M. R. Wieckowski,G. Vandecasteele, G. Baird, R. A. Tuft, K. E. Fogarty, andR. Rizzuto. Recombinant expression of the voltage-dependent anionchannel enhances the transfer of ca2+ microdomains tomitochondria. The Journal of Cell Biology, 159(4):613–624, Nov.2002. → pages 37[241] H.-W. Rhee, P. Zou, N. D. Udeshi, J. D. Martell, V. K. Mootha,S. A. Carr, and A. Y. Ting. Proteomic mapping of mitochondria inliving cells via spatially restricted enzymatic tagging. Science (NewYork, N.Y.), 339(6125):1328–1331, Mar. 2013. → pages 34[242] S. J. Riedl and G. S. Salvesen. The apoptosome: signalling platformof cell death. Nature Reviews. Molecular Cell Biology, 8(5):405–413,May 2007. → pages 115, 116[243] C. Ritz and A.-N. Spiess. qpcR: an R package for sigmoidal modelselection in quantitative real-time polymerase chain reactionanalysis. Bioinformatics, 24(13):1549–1551, Jan. 2008. → pages 53[244] R. Rizzuto, A. W. Simpson, M. Brini, and T. Pozzan. Rapid changesof mitochondrial ca2+ revealed by specifically targeted recombinantaequorin. Nature, 358(6384):325–327, July 1992. → pages 38[245] R. Rizzuto, M. Brini, M. Murgia, and T. Pozzan. Microdomains withhigh ca2+ close to ip3-sensitive channels that are sensed byneighboring mitochondria. Science (New York, N.Y.), 262(5134):744–747, Oct. 1993. → pages 38149[246] R. Rizzuto, P. Pinton, W. Carrington, F. S. Fay, K. E. Fogarty,L. M. Lifshitz, R. A. Tuft, and T. Pozzan. Close contacts with theendoplasmic reticulum as determinants of mitochondrial Ca2+responses. Science (New York, N.Y.), 280(5370):1763–1766, June1998. → pages 30, 38, 74, 101[247] R. Rizzuto, P. Pinton, D. Ferrari, M. Chami, G. Szabadkai, P. J.Magalhães, F. Di Virgilio, and T. Pozzan. Calcium and apoptosis:facts and hypotheses. Oncogene, 22(53):8619–8627, Nov. 2003. →pages 42[248] R. Rizzuto, S. Marchi, M. Bonora, P. Aguiari, A. Bononi, D. DeStefani, C. Giorgi, S. Leo, A. Rimessi, R. Siviero, E. Zecchini, andP. Pinton. Ca(2+) transfer from the ER to mitochondria: when, howand why. Biochimica Et Biophysica Acta, 1787(11):1342–1351, Nov.2009. → pages 30, 74, 116[249] J. D. Robertson. Ultrastructure of two invertebrate synapses.Proceedings of the Society for Experimental Biology and Medicine.Society for Experimental Biology and Medicine (New York, N.Y.), 82(2):219–223, Feb. 1953. → pages 1[250] J. D. Robertson. Recent electron microscope observations on theultrastructure of the crayfish median-to-motor giant synapse.Experimental Cell Research, 8(1):226–229, 1955. → pages 1[251] R. A. Romanov, M. F. Bystrova, O. A. Rogachevskaya, V. B.Sadovnikov, V. I. Shestopalov, and S. S. Kolesnikov. The ATPpermeability of pannexin 1 channels in a heterologous system and inmammalian taste cells is dispensable. Journal of Cell Science, 125(22):5514–5523, Nov. 2012. → pages 18[252] J. K. Rose and R. W. Doms. Regulation of protein export from theendoplasmic reticulum. Annual Review of Cell Biology, 4:257–288,1988. → pages 6[253] J. R. Ruby, R. F. Dyer, and R. G. Skalko. Continuities betweenmitochondria and endoplasmic reticulum in the mammalian ovary.Zeitschrift für Zellforschung und Mikroskopische Anatomie, 97(1):30–37, Mar. 1969. → pages 30[254] A. E. Rusiñol, Z. Cui, M. H. Chen, and J. E. Vance. A uniquemitochondria-associated membrane fraction from rat liver has a high150capacity for lipid synthesis and contains pre-golgi secretory proteinsincluding nascent lipoproteins. The Journal of Biological Chemistry,269(44):27494–27502, Nov. 1994. → pages 35, 36[255] G. A. Rutter, P. Burnett, R. Rizzuto, M. Brini, M. Murgia,T. Pozzan, J. M. Tavaré, and R. M. Denton. Subcellular imaging ofintramitochondrial Ca2+ with recombinant targeted aequorin:significance for the regulation of pyruvate dehydrogenase activity.Proceedings of the National Academy of Sciences of the United Statesof America, 93(11):5489–5494, May 1996. → pages 44[256] G. Sahu, S. Sukumaran, and A. K. Bera. Pannexins form gapjunctions with electrophysiological and pharmacological propertiesdistinct from connexins. Scientific Reports, 4:4955, 2014. → pages 9,10, 49[257] H. A. Sánchez, G. Meşe, M. Srinivas, T. W. White, and V. K.Verselis. Differentially altered Ca2+ regulation and Ca2+permeability in Cx26 hemichannels formed by the A40V and G45Emutations that cause keratitis ichthyosis deafness syndrome. TheJournal of General Physiology, 136(1):47–62, July 2010. → pages 109[258] T. H. Sanderson, C. A. Reynolds, R. Kumar, K. Przyklenk, andM. Hüttemann. Molecular mechanisms of ischemia-reperfusion injuryin brain: pivotal role of the mitochondrial membrane potential inreactive oxygen species generation. Molecular Neurobiology, 47(1):9–23, Feb. 2013. → pages 40, 42[259] I. F. Sbalzarini and P. Koumoutsakos. Feature point tracking andtrajectory analysis for video imaging in cell biology. Journal ofStructural Biology, 151(2):182–195, Aug. 2005. → pages 78[260] E. A. Schon and S. Przedborski. Mitochondria: the next(neurode)generation. Neuron, 70(6):1033–1053, June 2011. → pages117[261] B. Schreiner, L. Hedskog, B. Wiehager, and M. Ankarcrona.Amyloid-β peptides are generated in mitochondria-associatedendoplasmic reticulum membranes. Journal of Alzheimer’s disease:JAD, 43(2):369–374, 2015. → pages 30, 74, 116, 117[262] B. Schwanhäusser, D. Busse, N. Li, G. Dittmar, J. Schuchhardt,J. Wolf, W. Chen, and M. Selbach. Global quantification of151mammalian gene expression control. Nature, 473(7347):337–342,May 2011. → pages 104, 107[263] Sea Urchin Genome Sequencing Consortium, E. Sodergren, G. M.Weinstock, E. H. Davidson, R. A. Cameron, R. A. Gibbs, R. C.Angerer, L. M. Angerer, M. I. Arnone, D. R. Burgess, R. D. Burke,J. A. Coffman, M. Dean, M. R. Elphick, C. A. Ettensohn, K. R.Foltz, A. Hamdoun, R. O. Hynes, W. H. Klein, W. Marzluff, D. R.McClay, R. L. Morris, A. Mushegian, J. P. Rast, L. C. Smith, M. C.Thorndyke, V. D. Vacquier, G. M. Wessel, G. Wray, L. Zhang, C. G.Elsik, O. Ermolaeva, W. Hlavina, G. Hofmann, P. Kitts, M. J.Landrum, A. J. Mackey, D. Maglott, G. Panopoulou, A. J. Poustka,K. Pruitt, V. Sapojnikov, X. Song, A. Souvorov, V. Solovyev, Z. Wei,C. A. Whittaker, K. Worley, K. J. Durbin, Y. Shen, O. Fedrigo,D. Garfield, R. Haygood, A. Primus, R. Satija, T. Severson, M. L.Gonzalez-Garay, A. R. Jackson, A. Milosavljevic, M. Tong, C. E.Killian, B. T. Livingston, F. H. Wilt, N. Adams, R. Bellé,S. Carbonneau, R. Cheung, P. Cormier, B. Cosson, J. Croce,A. Fernandez-Guerra, A.-M. Genevière, M. Goel, H. Kelkar,J. Morales, O. Mulner-Lorillon, A. J. Robertson, J. V. Goldstone,B. Cole, D. Epel, B. Gold, M. E. Hahn, M. Howard-Ashby,M. Scally, J. J. Stegeman, E. L. Allgood, J. Cool, K. M. Judkins,S. S. McCafferty, A. M. Musante, R. A. Obar, A. P. Rawson, B. J.Rossetti, I. R. Gibbons, M. P. Hoffman, A. Leone, S. Istrail, S. C.Materna, M. P. Samanta, V. Stolc, W. Tongprasit, Q. Tu, K.-F.Bergeron, B. P. Brandhorst, J. Whittle, K. Berney, D. J. Bottjer,C. Calestani, K. Peterson, E. Chow, Q. A. Yuan, E. Elhaik,D. Graur, J. T. Reese, I. Bosdet, S. Heesun, M. A. Marra, J. Schein,M. K. Anderson, V. Brockton, K. M. Buckley, A. H. Cohen, S. D.Fugmann, T. Hibino, M. Loza-Coll, A. J. Majeske, C. Messier, S. V.Nair, Z. Pancer, D. P. Terwilliger, C. Agca, E. Arboleda, N. Chen,A. M. Churcher, F. Hallböök, G. W. Humphrey, M. M. Idris,T. Kiyama, S. Liang, D. Mellott, X. Mu, G. Murray, R. P. Olinski,F. Raible, M. Rowe, J. S. Taylor, K. Tessmar-Raible, D. Wang, K. H.Wilson, S. Yaguchi, T. Gaasterland, B. E. Galindo, H. J. Gunaratne,C. Juliano, M. Kinukawa, G. W. Moy, A. T. Neill, M. Nomura,M. Raisch, A. Reade, M. M. Roux, J. L. Song, Y.-H. Su, I. K.Townley, E. Voronina, J. L. Wong, G. Amore, M. Branno, E. R.Brown, V. Cavalieri, V. Duboc, L. Duloquin, C. Flytzanis, C. Gache,F. Lapraz, T. Lepage, A. Locascio, P. Martinez, G. Matassi,152V. Matranga, R. Range, F. Rizzo, E. Röttinger, W. Beane,C. Bradham, C. Byrum, T. Glenn, S. Hussain, G. Manning,E. Miranda, R. Thomason, K. Walton, A. Wikramanayke, S.-Y. Wu,R. Xu, C. T. Brown, L. Chen, R. F. Gray, P. Y. Lee, J. Nam,P. Oliveri, J. Smith, D. Muzny, S. Bell, J. Chacko, A. Cree, S. Curry,C. Davis, H. Dinh, S. Dugan-Rocha, J. Fowler, R. Gill, C. Hamilton,J. Hernandez, S. Hines, J. Hume, L. Jackson, A. Jolivet, C. Kovar,S. Lee, L. Lewis, G. Miner, M. Morgan, L. V. Nazareth,G. Okwuonu, D. Parker, L.-L. Pu, R. Thorn, and R. Wright. Thegenome of the sea urchin strongylocentrotus purpuratus. Science(New York, N.Y.), 314(5801):941–952, Nov. 2006. → pages 2[264] K. Seggerson, L. Tang, and E. G. Moss. Two genetic circuits repressthe caenorhabditis elegans heterochronic gene lin-28 after translationinitiation. Developmental Biology, 243(2):215–225, Mar. 2002. →pages 105[265] K. J. Seymour-Laurent and M. E. Barish. Inositol 1,4,5-trisphosphateand ryanodine receptor distributions and patterns of acetylcholine-and caffeine-induced calcium release in cultured mouse hippocampalneurons. The Journal of Neuroscience: The Official Journal of theSociety for Neuroscience, 15(4):2592–2608, Apr. 1995. → pages 101[266] R. M. Sharpe, C. McKinnell, C. Kivlin, and J. S. Fisher.Proliferation and functional maturation of Sertoli cells, and theirrelevance to disorders of testis function in adulthood. Reproduction,125(6):769–784, Jan. 2003. → pages 108[267] V. I. Shestopalov and Y. Panchin. Pannexins and gap junctionprotein diversity. Cellular and molecular life sciences: CMLS, 65(3):376–394, Feb. 2008. → pages 2[268] W. Silverman, S. Locovei, and G. Dahl. Probenecid, a gout remedy,inhibits pannexin 1 channels. American Journal of Physiology. CellPhysiology, 295(3):C761–767, Sept. 2008. → pages 20[269] W. R. Silverman, J. P. de Rivero Vaccari, S. Locovei, F. Qiu, S. K.Carlsson, E. Scemes, R. W. Keane, and G. Dahl. The pannexin 1channel activates the inflammasome in neurons and astrocytes. TheJournal of Biological Chemistry, 284(27):18143–18151, July 2009. →pages 24, 109, 110153[270] T. Simmen, J. E. Aslan, A. D. Blagoveshchenskaya, L. Thomas,L. Wan, Y. Xiang, S. F. Feliciangeli, C.-H. Hung, C. M. Crump, andG. Thomas. Pacs-2 controls endoplasmic reticulum-mitochondriacommunication and bid-mediated apoptosis. The EMBO journal, 24(4):717–729, Feb. 2005. → pages 33, 44, 74[271] T. J. B. Simons. Calcium and neuronal function. NeurosurgicalReview, 11(2):119–129, June 1988. → pages 109[272] G. Söhl and K. Willecke. An update on connexin genes and theirnomenclature in mouse and man. Cell Communication & Adhesion,10(4-6):173–180, 2003 Jul-Dec. → pages 2[273] N. Solovyova, N. Veselovsky, E. Toescu, and A. Verkhratsky. Ca2+dynamics in the lumen of the endoplasmic reticulum in sensoryneurons: direct visualization of Ca2+-induced Ca2+ release triggeredby physiological Ca2+ entry. The EMBO Journal, 21(4):622–630,Feb. 2002. → pages 109[274] B. J. Soltys and R. S. Gupta. Interrelationships of endoplasmicreticulum, mitochondria, intermediate filaments, and microtubules–aquadruple fluorescence labeling study. Biochemistry and Cell Biology= Biochimie Et Biologie Cellulaire, 70(10-11):1174–1186, 1992Oct-Nov. → pages 30[275] A. V. Somlyo, H. G. Gonzalez-Serratos, H. Shuman, G. McClellan,and A. P. Somlyo. Calcium release and ionic changes in thesarcoplasmic reticulum of tetanized muscle: an electron-probe study.The Journal of Cell Biology, 90(3):577–594, Sept. 1981. → pages 110[276] G. E. Sosinsky, D. Boassa, R. Dermietzel, H. S. Duffy, D. W. Laird,B. MacVicar, C. C. Naus, S. Penuela, E. Scemes, D. C. Spray, R. J.Thompson, H.-B. Zhao, and G. Dahl. Pannexin channels are not gapjunction hemichannels. Channels (Austin, Tex.), 5(3):193–197, 2011May-Jun. → pages 10, 49, 74[277] V. Soubannier, G.-L. McLelland, R. Zunino, E. Braschi,P. Rippstein, E. A. Fon, and H. M. McBride. A vesicular transportpathway shuttles cargo from mitochondria to lysosomes. Currentbiology: CB, 22(2):135–141, Jan. 2012. → pages 35154[278] A.-N. Spiess, C. Feig, and C. Ritz. Highly accurate sigmoidal fittingof real-time PCR data by introducing a parameter for asymmetry.BMC Bioinformatics, 9(1):221, Apr. 2008. → pages 53[279] T. A. Starich, R. K. Herman, and J. E. Shaw. Molecular and geneticanalysis of unc-7, a caenorhabditis elegans gene required forcoordinated locomotion. Genetics, 133(3):527–541, Mar. 1993. →pages 3[280] K. A. Stauffer. The gap junction proteins beta 1-connexin(connexin-32) and beta 2-connexin (connexin-26) can formheteromeric hemichannels. The Journal of Biological Chemistry, 270(12):6768–6772, Mar. 1995. → pages 6[281] S. J. Stone and J. E. Vance. Phosphatidylserine synthase-1 and -2are localized to mitochondria-associated membranes. The Journal ofBiological Chemistry, 275(44):34534–34540, Nov. 2000. → pages 36[282] S. J. Stone, M. C. Levin, P. Zhou, J. Han, T. C. Walther, and R. V.Farese. The endoplasmic reticulum enzyme dgat2 is found inmitochondria-associated membranes and has a mitochondrialtargeting signal that promotes its association with mitochondria.The Journal of Biological Chemistry, 284(8):5352–5361, Feb. 2009.→ pages 36[283] D. A. Stroud, S. Oeljeklaus, S. Wiese, M. Bohnert, U. Lewandrowski,A. Sickmann, B. Guiard, M. van der Laan, B. Warscheid, andN. Wiedemann. Composition and Topology of the EndoplasmicReticulum–Mitochondria Encounter Structure. Journal of MolecularBiology, 413(4):743–750, Nov. 2011. → pages 31[284] S. O. Suadicani, R. Iglesias, J. Wang, G. Dahl, D. C. Spray, andE. Scemes. ATP signaling is deficient in cultured Pannexin1-nullmouse astrocytes. Glia, 60(7):1106–1116, July 2012. → pages 18[285] L. A. Swayne, C. D. Sorbara, and S. A. L. Bennett. Pannexin 2 IsExpressed by Postnatal Hippocampal Neural Progenitors andModulates Neuronal Commitment. Journal of Biological Chemistry,285(32):24977–24986, June 2010. → pages 11, 27, 29, 49, 70, 71, 99,112155[286] R. H. Swerdlow, J. M. Burns, and S. M. Khan. The Alzheimer’sDisease Mitochondrial Cascade Hypothesis. Journal of Alzheimer’sdisease : JAD, 20(Suppl 2):265–279, 2010. → pages 117[287] G. Szabadkai and R. Rizzuto. Participation of endoplasmicreticulum and mitochondrial calcium handling in apoptosis: morethan just neighborhood? FEBS letters, 567(1):111–115, June 2004.→ pages 113[288] G. Szabadkai, K. Bianchi, P. Várnai, D. De Stefani, M. R.Wieckowski, D. Cavagna, A. I. Nagy, T. Balla, and R. Rizzuto.Chaperone-mediated coupling of endoplasmic reticulum andmitochondrial Ca2+ channels. The Journal of Cell Biology, 175(6):901–911, Dec. 2006. → pages 32, 74, 89[289] T. Szabados, C. Dul, K. Majtényi, J. Hargitai, Z. Pénzes, andR. Urbanics. A chronic alzheimer’s model evoked by mitochondrialpoison sodium azide for pharmacological investigations. BehaviouralBrain Research, 154(1):31–40, Sept. 2004. → pages 117[290] H.-C. Tai and E. M. Schuman. Ubiquitin, the proteasome andprotein degradation in neuronal function and dysfunction. NatureReviews Neuroscience, 9(11):826–838, Nov. 2008. → pages 107[291] N. Taquet, S. Dumont, J.-L. Vonesch, D. Hentsch, J.-M. Reimund,and C. D. Muller. Differential between protein and mrna expressionof ccr7 and sstr5 receptors in crohn’s disease patients. Mediators ofInflammation, 2009:285812, 2009. → pages 104[292] R. C. Taylor, S. P. Cullen, and S. J. Martin. Apoptosis: controlleddemolition at the cellular level. Nature Reviews. Molecular CellBiology, 9(3):231–241, Mar. 2008. → pages 42[293] P. R. Territo, V. K. Mootha, S. A. French, and R. S. Balaban.Ca(2+) activation of heart mitochondrial oxidative phosphorylation:role of the f(0)/f(1)-atpase. American Journal of Physiology. CellPhysiology, 278(2):C423–435, Feb. 2000. → pages 39, 40[294] R. J. Thompson. Pannexin channels and ischaemia. The Journal ofPhysiology, 593(16):3463–3470, Aug. 2015. → pages 23, 24[295] R. J. Thompson, N. Zhou, and B. A. MacVicar. Ischemia opensneuronal gap junction hemichannels. Science (New York, N.Y.), 312(5775):924–927, May 2006. → pages 23156[296] R. J. Thompson, M. F. Jackson, M. E. Olah, R. L. Rungta, D. J.Hines, M. A. Beazely, J. F. MacDonald, and B. A. MacVicar.Activation of pannexin-1 hemichannels augments aberrant burstingin the hippocampus. Science (New York, N.Y.), 322(5907):1555–1559, Dec. 2008. → pages 18, 21, 24[297] Q. Tian, S. B. Stepaniants, M. Mao, L. Weng, M. C. Feetham, M. J.Doyle, E. C. Yi, H. Dai, V. Thorsson, J. Eng, D. Goodlett, J. P.Berger, B. Gunter, P. S. Linseley, R. B. Stoughton, R. Aebersold,S. J. Collins, W. A. Hanlon, and L. E. Hood. Integrated genomic andproteomic analyses of gene expression in mammalian cells. Molecular& cellular proteomics: MCP, 3(10):960–969, Oct. 2004. → pages 104[298] A. Y. Ting. Discovery and characterization of novel synaptic andmitochondrial proteins via peroxidase- mediated live cell proteomicmapping. Molecular Biology of the Cell, 26(25):M145, Dec. 2015. →pages 34[299] P. J. Trotter and D. R. Voelker. Identification of anon-mitochondrial phosphatidylserine decarboxylase activity (psd2)in the yeast saccharomyces cerevisiae. The Journal of BiologicalChemistry, 270(11):6062–6070, Mar. 1995. → pages 35[300] P. J. Trotter, J. Pedretti, and D. R. Voelker. Phosphatidylserinedecarboxylase from Saccharomyces cerevisiae. Isolation of mutants,cloning of the gene, and creation of a null allele. Journal ofBiological Chemistry, 268(28):21416–21424, May 1993. → pages 35[301] P. Turmel, J. Dufresne, L. Hermo, C. E. Smith, S. Penuela, D. W.Laird, and D. G. Cyr. Characterization of pannexin1 and pannexin3and their regulation by androgens in the male reproductive tract ofthe adult rat. Molecular Reproduction and Development, 78(2):124–138, Feb. 2011. → pages 12, 15[302] V. Valiunas, J. Gemel, P. R. Brink, and E. C. Beyer. Gap junctionchannels formed by coexpressed connexin40 and connexin43.American Journal of Physiology. Heart and Circulatory Physiology,281(4):H1675–1689, Oct. 2001. → pages 6[303] J. E. Vance. Phospholipid synthesis in a membrane fractionassociated with mitochondria. The Journal of Biological Chemistry,265(13):7248–7256, May 1990. → pages 35157[304] J. E. Vance. MAM (mitochondria-associated membranes) inmammalian cells: Lipids and beyond. Biochimica et Biophysica Acta(BBA) - Molecular and Cell Biology of Lipids, 1841(4):595–609, Apr.2014. → pages 42, 74, 116[305] J. E. Vance and G. Tasseva. Formation and function ofphosphatidylserine and phosphatidylethanolamine in mammaliancells. Biochimica Et Biophysica Acta, 1831(3):543–554, Mar. 2013.→ pages 35[306] F. Vanden Abeele, G. Bidaux, D. Gordienko, B. Beck, Y. V. Panchin,A. V. Baranova, D. V. Ivanov, R. Skryma, and N. Prevarskaya.Functional implications of calcium permeability of the channelformed by pannexin 1. The Journal of Cell Biology, 174(4):535–546,Aug. 2006. → pages 7, 9, 10, 29, 49, 71, 74, 101, 109, 114[307] M. G. Vander Heiden, L. C. Cantley, and C. B. Thompson.Understanding the warburg effect: the metabolic requirements of cellproliferation. Science (New York, N.Y.), 324(5930):1029–1033, May2009. → pages 92[308] J. K. VanSlyke and L. S. Musil. Analysis of connexin intracellulartransport and assembly. Methods (San Diego, Calif.), 20(2):156–164,Feb. 2000. → pages 7[309] S. Varum, A. S. Rodrigues, M. B. Moura, O. Momcilovic, C. A.Easley, IV, J. Ramalho-Santos, B. Van Houten, and G. Schatten.Energy Metabolism in Human Pluripotent Stem Cells and TheirDifferentiated Counterparts. PLoS ONE, 6(6):e20914, June 2011. →pages 112[310] R. Villegas, N. W. Martinez, J. Lillo, P. Pihan, D. Hernandez, J. L.Twiss, and F. A. Court. Calcium release from intra-axonalendoplasmic reticulum leads to axon degeneration throughmitochondrial dysfunction. The Journal of Neuroscience: TheOfficial Journal of the Society for Neuroscience, 34(21):7179–7189,May 2014. → pages 74, 101[311] C. Vogel, R. d. S. Abreu, D. Ko, S.-Y. Le, B. A. Shapiro, S. C.Burns, D. Sandhu, D. R. Boutz, E. M. Marcotte, and L. O. Penalva.Sequence signatures and mrna concentration can explain two-thirdsof protein abundance variation in a human cell line. MolecularSystems Biology, 6:400, Aug. 2010. → pages 104158[312] A. Vogt, S. G. Hormuzdi, and H. Monyer. Pannexin1 and Pannexin2expression in the developing and mature rat brain. Brain Research.Molecular Brain Research, 141(1):113–120, Nov. 2005. → pages 22,55, 60[313] C. Vonesch and M. Unser. A fast thresholded landweber algorithmfor wavelet-regularized multidimensional deconvolution. IEEEtransactions on image processing: a publication of the IEEE SignalProcessing Society, 17(4):539–549, Apr. 2008. → pages 55, 78, 79[314] C. Voss, S. Lahiri, B. P. Young, C. J. Loewen, and W. A. Prinz.Er-shaping proteins facilitate lipid exchange between the er andmitochondria in s. cerevisiae. Journal of Cell Science, 125(Pt 20):4791–4799, Oct. 2012. → pages 36[315] J. Wang, M. Ma, S. Locovei, R. W. Keane, and G. Dahl. Modulationof membrane channel currents by gap junction protein mimeticpeptides: size matters. American Journal of Physiology. CellPhysiology, 293(3):C1112–1119, Sept. 2007. → pages 21[316] J. Wang, D. G. Jackson, and G. Dahl. The food dye FD&c Blue No.1 is a selective inhibitor of the ATP release channel Panx1. TheJournal of General Physiology, 141(5):649–656, Jan. 2013. → pages20[317] J. Wang, C. Ambrosi, F. Qiu, D. G. Jackson, G. Sosinsky, andG. Dahl. The membrane protein pannexin1 forms two open-channelconformations depending on the mode of activation. ScienceSignaling, 7(335):ra69, July 2014. → pages 18, 19, 109, 110[318] T. Watanabe and D. R. Kankel. Molecular cloning and analysis ofl(1)ogre, a locus of drosophila melanogaster with prominent effectson the postembryonic development of the central nervous system.Genetics, 126(4):1033–1044, Dec. 1990. → pages 3[319] M. C. Wei, W. X. Zong, E. H. Cheng, T. Lindsten,V. Panoutsakopoulou, A. J. Ross, K. A. Roth, G. R. MacGregor,C. B. Thompson, and S. J. Korsmeyer. Proapoptotic bax and bak: arequisite gateway to mitochondrial dysfunction and death. Science(New York, N.Y.), 292(5517):727–730, Apr. 2001. → pages 33[320] R. Wei, J. Wang, Y. Xu, B. Yin, F. He, Y. Du, G. Peng, and B. Luo.Probenecid protects against cerebral ischemia/reperfusion injury by159inhibiting lysosomal and inflammatory damage in rats. Neuroscience,301:168–177, Aug. 2015. → pages 23, 118[321] N. L. Weilinger, P. L. Tang, and R. J. Thompson. Anoxia-inducednmda receptor activation opens pannexin channels via src familykinases. The Journal of Neuroscience: The Official Journal of theSociety for Neuroscience, 32(36):12579–12588, Sept. 2012. → pages24[322] L. E. Wicki-Stordeur, A. K. J. Boyce, and L. A. Swayne. Analysis ofa pannexin 2-pannexin 1 chimeric protein supports divergent rolesfor pannexin c-termini in cellular localization. Cell communication &adhesion, 20(3-4):73–79, Aug. 2013. → pages 16, 29, 70, 99, 100[323] M. R. Wieckowski, C. Giorgi, M. Lebiedzinska, J. Duszynski, andP. Pinton. Isolation of mitochondria-associated membranes andmitochondria from animal tissues and cells. Nature Protocols, 4(11):1582–1590, 2009. → pages 34, 80, 89[324] B. Wightman, I. Ha, and G. Ruvkun. Posttranscriptional regulationof the heterochronic gene lin-14 by lin-4 mediates temporal patternformation in c. elegans. Cell, 75(5):855–862, Dec. 1993. → pages 105[325] M. Wilhelm, J. Schlegl, H. Hahne, A. M. Gholami, M. Lieberenz,M. M. Savitski, E. Ziegler, L. Butzmann, S. Gessulat, H. Marx,T. Mathieson, S. Lemeer, K. Schnatbaum, U. Reimer, H. Wenschuh,M. Mollenhauer, J. Slotta-Huspenina, J.-H. Boese, M. Bantscheff,A. Gerstmair, F. Faerber, and B. Kuster. Mass-spectrometry-baseddraft of the human proteome. Nature, 509(7502):582–587, May 2014.→ pages 69, 104[326] X.-X. Xiong, L.-J. Gu, J. Shen, X.-H. Kang, Y.-Y. Zheng, S.-B. Yue,and S.-M. Zhu. Probenecid protects against transient focal cerebralischemic injury by inhibiting hmgb1 release and attenuating aqp4expression in mice. Neurochemical Research, 39(1):216–224, Jan.2014. → pages 118[327] n. Yang and n. Bielawski. Statistical methods for detectingmolecular adaptation. Trends in Ecology & Evolution, 15(12):496–503, Dec. 2000. → pages 12160[328] M. R. Yen and M. H. Saier. Gap junctional proteins of animals: theinnexin/pannexin superfamily. Progress in Biophysics and MolecularBiology, 94(1-2):5–14, 2007 May-Jun. → pages 2, 5[329] M. Yi, D. Weaver, and G. Hajnóczky. Control of mitochondrialmotility and distribution by the calcium signal: a homeostaticcircuit. The Journal of Cell Biology, 167(4):661–672, Nov. 2004. →pages 114[330] M. J. Youngman, A. E. A. Hobbs, S. M. Burgess, M. Srinivasan, andR. E. Jensen. Mmm2p, a mitochondrial outer membrane proteinrequired for yeast mitochondrial shape and maintenance of mtDNAnucleoids. The Journal of Cell Biology, 164(5):677–688, Mar. 2004.→ pages 31[331] H. Yu, I. Lee, A. R. Salomon, K. Yu, and M. Hüttemann.Mammalian liver cytochrome c is tyrosine-48 phosphorylated in vivo,inhibiting mitochondrial respiration. Biochimica Et Biophysica Acta,1777(7-8):1066–1071, 2008 Jul-Aug. → pages 42[332] A. Zappalà, G. Li Volti, M. F. Serapide, R. Pellitteri, M. Falchi,F. La Delia, V. Cicirata, and F. Cicirata. Expression of pannexin2protein in healthy and ischemized brain of adult rats. Neuroscience,148(3):653–667, Sept. 2007. → pages 29, 49, 56, 66, 67, 70, 99[333] J. Zborowski, A. Dygas, and L. Wojtczak. Phosphatidylserinedecarboxylase is located on the external side of the innermitochondrial membrane. FEBS letters, 157(1):179–182, June 1983.→ pages 35[334] M. Zellner, R. Babeluk, M. Diestinger, P. Pirchegger, S. Skeledzic,and R. Oehler. Fluorescence-based Western blotting for quantitationof protein biomarkers in clinical samples. ELECTROPHORESIS, 29(17):3621–3627, Aug. 2008. → pages 70[335] A. Zhang, C. D. Williamson, D. S. Wong, M. D. Bullough, K. J.Brown, Y. Hathout, and A. M. Colberg-Poley. Quantitativeproteomic analyses of human cytomegalovirus-induced restructuringof endoplasmic reticulum-mitochondrial contacts at late times ofinfection. Molecular & cellular proteomics: MCP, 10(10):M111.009936, Oct. 2011. → pages 34161[336] H. Zhang, Y. Chen, and C. Zhang. Patterns of heterogeneousexpression of pannexin 1 and pannexin 2 transcripts in the olfactoryepithelium and olfactory bulb. Journal of molecular histology, 43(6):651–660, Dec. 2012. → pages 49[337] J. T. Zhang, M. Chen, C. I. Foote, and B. J. Nicholson. Membraneintegration of in vitro-translated gap junctional proteins: co- andpost-translational mechanisms. Molecular Biology of the Cell, 7(3):471–482, Mar. 1996. → pages 6[338] G. Zoidl, M. Kremer, C. Zoidl, S. Bunse, and R. Dermietzel.Molecular Diversity of Connexin and Pannexin Genes in the Retinaof the Zebrafish danio rerio. Cell Communication and Adhesion, 15(1-2):169–183, Jan. 2008. → pages 15, 22, 49, 59, 60, 68162


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



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"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items