The regulation and maintenance of iron homeostasis is critical to human health. As a constituent of hemoglobin, iron is essential for oxygen transport and significant iron deficiency leads to anemia. Eukaryotic cells require iron for survival and proliferation. Iron is part of hemoproteins, iron-sulfur (Fe-S) proteins, and other proteins with functional groups that require iron as a cofactor. At the cellular level, iron uptake, utilization, storage, and export are regulated at different molecular levels (transcriptional, mRNA stability, translational, and posttranslational). Iron regulatory proteins (IRPs) 1 and 2 post-transcriptionally control mammalian iron homeostasis by binding to iron-responsive elements (IREs), conserved RNA stem-loop structures located in the 5’- or 3‘- untranslated regions of genes involved in iron metabolism (e.g. FTH1, FTL, and TFRC). To identify novel IRE-containing mRNAs, we integrated biochemical, biocomputational, and microarray-based experimental approaches. Gene expression studies greatly contribute to our understanding of complex relationships in gene regulatory networks. However, the complexity of array design, production and manipulations are limiting factors, affecting data quality. The use of customized DNA microarrays improves overall data quality in many situations, however, only if for these specifically designed microarrays analysis tools are available. Methods In this project response to the iron treatment was examined under different conditions using bioinformatical methods. This would improve our understanding of an iron regulatory network. For these purposes we used microarray gene expression data. To identify novel IRE-containing mRNAs biochemical, biocomputational, and microarray-based experimental approaches were integrated. IRP/IRE messenger ribonucleoproteins were immunoselected and their mRNA composition was analysed using an IronChip microarray enriched for genes predicted computationally to contain IRE-like motifs. Analysis of IronChip microarray data requires specialized tool which can use all advantages of a customized microarray platform. Novel decision-tree based algorithm was implemented using Perl in IronChip Evaluation Package (ICEP). Results IRE-like motifs were identified from genomic nucleic acid databases by an algorithm combining primary nucleic acid sequence and RNA structural criteria. Depending on the choice of constraining criteria, such computational screens tend to generate a large number of false positives. To refine the search and reduce the number of false positive hits, additional constraints were introduced. The refined screen yielded 15 IRE-like motifs. A second approach made use of a reported list of 230 IRE-like sequences obtained from screening UTR databases. We selected 6 out of these 230 entries based on the ability of the lower IRE stem to form at least 6 out of 7 bp. Corresponding ESTs were spotted onto the human or mouse versions of the IronChip and the results were analysed using ICEP. Our data show that the immunoselection/microarray strategy is a feasible approach for screening bioinformatically predicted IRE genes and the detection of novel IRE-containing mRNAs. In addition, we identified a novel IRE-containing gene CDC14A (Sanchez M, et al. 2006). The IronChip Evaluation Package (ICEP) is a collection of Perl utilities and an easy to use data evaluation pipeline for the analysis of microarray data with a focus on data quality of custom-designed microarrays. The package has been developed for the statistical and bioinformatical analysis of the custom cDNA microarray IronChip, but can be easily adapted for other cDNA or oligonucleotide-based designed microarray platforms. ICEP uses decision tree-based algorithms to assign quality flags and performs robust analysis based on chip design properties regarding multiple repetitions, ratio cut-off, background and negative controls (Vainshtein Y, et al., 2010). / Die Regulierung und Aufrechterhaltung der Eisen-Homeostase ist bedeutend für die menschliche Gesundheit. Als Bestandteil des Hämoglobins ist es wichtig für den Transport von Sauerstoff, ein Mangel führt zu Blutarmut. Eukaryotische Zellen benötigen Eisen zum Überleben und zum Proliferieren. Eisen ist am Aufbau von Hämo- und Eisenschwefelproteinen (Fe-S) beteiligt und kann als Kofaktor dienen. Die Aufnahme, Nutzung, Speicherung und der Export von Eisen ist zellulär auf verschiedenen molekularen Ebenen reguliert (Transkription, mRNA-Level, Translation, Protein-Level). Die iron regulatory proteins (IRPs) 1 und 2 kontrollieren die Eisen-Homeostase in Säugetieren posttranslational durch die Bindung an Iron-responsive elements (IREs). IREs sind konservierte RNA stem-loop Strukturen in den 5' oder 3' untranslatierten Bereichen von Genen, die im Eisenmetabolismus involviert sind (z.B. FTH1, FTL und TFRC). In dieser Arbeit wurden biochemische und bioinformatische Methoden mit Microarray-Experimenten kombiniert, um neue mRNAs mit IREs zu identifizieren. Genexpressionsstudien verbessern unser Verständnis über die komplexen Zusammenhänge in genregulatorischen Netzwerken. Das komplexe Design von Microarrays, deren Produktion und Manipulation sind dabei die limitierenden Faktoren bezüglich der Datenqualität. Die Verwendung von angepassten DNA Microarrays verbessert häufig die Datenqualität, falls entsprechende Analysemöglichkeiten für diese Arrays existieren. Methoden Um unser Verständnis von eisenregulierten Netzwerken zu verbessern, wurde im Rahmen dieses Projektes die Auswirkung einer Behandlung mit Eisen bzw. von Knockout Mutation unter verschiedenen Bedingungen mittels bioinformatischer Methoden untersucht. Hierfür nutzen wir Expressionsdaten aus Microarray-Experimenten. Durch die Verknüpfung von biochemischen, bioinformatischen und Microarray Ansätzen können neue Proteine mit IREs identifiziert werden. IRP/IRE messenger Ribonucleoproteine wurden immunpräzipitiert. Die Zusammensetzung der enthaltenen mRNAs wurde mittels einem IronChip Microarray analysiert: Für diesen Chip wurden bioinformatisch Gene vorhergesagt, die IRE-like Motive aufweisen. Der Chip wurde mit solchen Oligonucleotiden beschichtet und durch Hybridisierung überprüft, ob die präzipitierten mRNA sich hieran binden. Die Analyse der erhaltenen Daten erfordert ein spezialisiertes Werkzeug um von allen Vorteilen der angepassten Microarrays zu profitieren. Ein neuer Entscheidungsbaum-basierter Algorithmus wurde in Perl im IronChip Evaluation Package (ICEP) implementiert. Ergebnisse Aus großen Sequenz-Datenbanken wurden IRE-like Motive identifiziert. Dazu kombiniert der Algorithmus, insbesondere RNA-Primärsequenz und RNA-Strukturdaten. Solche Datenbankanalysen tendieren dazu, eine große Anzahl falsch positiver Treffer zu generieren. Daher wurden zusätzliche Bedingungen formuliert, um die Suche zu verfeinern und die Anzahl an falsch positiven Treffer zu reduzieren. Die angepassten Suchkriterien ergaben 15 IRE-like Motive. In einem weiteren Ansatz verwendeten wir eine Liste von 230 IRE-like Sequenzen aus UTR-Datenbanken. Daraus wurden 6 Sequenzen ausgewählt, die auch im unteren Teil stabil sind (untere Helix über 6 bp stabil). Die korrespondierenden Expressed Sequence Tags (ESTs) wurden auf die humane oder murine Version des IronChips aufgetragen. Die Microarray Ergebnisse wurden mit dem ICEP Programm ausgewertet. Unsere Ergebnisse zeigen, dass die Immunpräzipitation mit anschließender Microarrayanalyse ein nützlicher Ansatz ist, um bioinformatisch vorhergesagte IRE-Gene zu identifizieren. Darüber hinaus ermöglicht uns dieser Ansatz die Detektion neuer mRNAs, die IREs enthalten, wie das von uns gefundene Gen CDC14A (Sanchez et al., 2006). ICEP ist ein optimiertes Programmpaket aus Perl Programmen (Vainshtein et al., BMC Bioinformatics, 2010). Es ermöglicht die einfache Auswertung von Microarray Daten mit dem Fokus auf selbst entwickelten Microarray Designs. ICEP diente für die statistische und bioinformatische Analyse von selbst entwickelten IronChips, kann aber auch leicht an die Analyse von oligonucleotidbasierten oder cDNA Microarrays adaptiert werden. ICEP nutzt einen Entscheidungsbaum-basierten Algorithmus um die Qualität zu bewerten und führt eine robuste Analyse basierend auf Chipeigenschaften, wie mehrfachen Wiederholungen, Signal/Rausch Verhältnis, Hintergrund und Negativkontrollen durch.

Osteoarthritis (OA) is a chronic, degenerative condition of articular joints. The prevalence of OA is high in many mammalian populations, though our understanding of the disease is limited, with the initiating factors and the early phenotype of the disease being poorly characterised. Clinically, the early-stage of OA is rarely identified, precluding the identification and treatment of affected individuals. Consequently, in vitro models of OA typically reflect the later stages of the disease, and are rarely validated against the naturally-occurring disease. This project utilised tissue from a naturally-occurring canine disease (medial coronoid process disease) to characterise the transcriptome of early-stage OA, and inform different in vitro models, to try and refine the model conditions. Medial coronoid processes from affected dogs were removed and graded histologically, both manually and through the development of a semi-automated assessment. Early-stage OA was characterised by a decrease in the chondrocyte density, an increase in the thickness of the articular cartilage and a loss of proteoglycan. No histological changes in bone morphology were noted in early-stage OA. A transcriptomic approach was adopted, in which the transcriptome of earlystage canine OA was assessed in the coronoid process samples. The canine data generated were meta-analysed alongside published datasets from in vivo models of early-stage OA. These data were from rodent models of the disease. A panel of genes were identified as being associated with the early stage of the disease across multiple datasets. By immunoassay, synovial fluid was screened for pro-inflammatory cytokines and in affected canine joints, interleukin 8 was found to be increased. Three in vitro models (cytokine stimulation of monolayer cell cultures, cyclic compression of agarose embedded cells and impact loading of osteochondral cores) were refined through modification of their stimuli. An identified panel of differentially expressed genes were used to screen each model under different parameters. Hierarchical clustering analysis was used to cluster the panel of conditions so that those which most closely reflected the naturally occurring disease were selected for more detailed transcriptomic analysis by microarray. Chondrocytes and osteoblasts were stimulated with a range of cytokine conditions, using IL-1β and IL-8 based on use in the literature and immunoassay findings. Monolayers were stimulated for a range of times and conentrations with either a single stimulus or multiple cytokines in the medium. The cells responded differently to the cytokine stimulus, requiring different stimuli to most closely replicate the transcriptomic profile of the natural disease. Microarray profiling revealed that cytokine stimulation enriched genes associated with the extracellular matrix and the extracellular region in both cells types. For the cyclic compression model, cells were embedded in an agarose gel matrix and cyclically compressed for various time periods followed by various incubation periods after compression. Both chondrocytes and osteoblasts responded in a similar manner to the cyclic compression stimulus when a post loading incubation step was included to replicate the transcriptomic profile of the natural disease. Cyclic compression enriched gene clusters associated with response to oxidative stress and the extracellular matrix When osteochondral cores were harvested from joints and impacted to represent a traumatic injury, the model could not replicate the transcriptomic model of the natural disease, although increased sGAG release nitric oxide (NO) production was observed. Degradation of mRNA in both tissues was a feature of this model regardless of the loading condition, which precluded further analysis by microarray, but highlighted the significant limitations that were associated with this model. None of the three models tested could accurately reflect the transcriptomic changes of the early-stage OA phenotype in cartilage or bone. A unified model, combining cytokine stimulation with cyclic compression drove cells towards the diseased phenotype in bone. Inflammatory pathways were activated as well as the proteases MMP3 and MMP13. However, chondrocytes were seemingly unresponsive to the multifactorial model, and this will require further analysis. The chronic nature of OA makes it difficult to match in vitro models to the transcriptomic phenotype identified in naturally occurring OA, particularly with respect to the differential expression of structural genes which were identified in the naturally occurring disease but not the models. This work highlights the limitations of existing models, but proposes a validation process which can be used to direct invitro models towards the naturally occurring phenotype.

In 2003, Pomeroy et al. published a research study that described a gene expression based prediction of central nervous system embryonal tumour (CNS) outcome. Over a half of decade, many models and approaches have been developed based on experimental data consisting of 99 samples with 7,129 genes. The way, how meaningful knowledge from these models can be extracted, and how this knowledge for further research is still a hot topic. This thesis addresses this and has developed an information method that includes modelling of interactive patterns, important genes discovery and visualisation of the obtained knowledge. The major goal of this thesis is to discover important genes responsible for CNS tumour and import these genes into a well structured knowledge framework system, called Brain-Gene-Ontology. In this thesis, we take the first step towards finding the most accurate model for analysing the CNS tumour by offering a comparative study of global, local and personalised modelling. Five traditional modelling approaches and a new personalised method – WWKNN (weighted distance, weighted variables K-nearest neighbours) – are investigated. To increase the classification accuracy and one-vs.-all based signal to- noise ratio is also developed for pre-processing experimental data. For the knowledge discovery, CNS-based ontology system is developed. Through ontology analysis, 21 discriminate genes are found to be relevant for different CNS tumour classes, medulloblastoma tumour subclass and medulloblastoma treatment outcome. All the findings in this thesis contribute for expanding the information space of the BGO framework.

Tendon injury is painful and often debilitating, and is a one of the most prevalent soft tissue injuries encountered in the clinic. While common, the underlying molecular and genetic processes of tendon damage and repair remain poorly understood. The work described herein used genome-wide expression analyses to investigate tendon injury and healing from three perspectives. The first identified novel gene expression in tendon fibroblasts following their stimulation with nitric oxide (NO). Of particular relevance to tendon healing was the observation that stimulated fibroblasts express a number of extracellular matrix (ECM) genes in response to NO in a dose-dependent manner, and that NO significantly affects cellular adhesion, a critical process during tendon repair. These findings will be of use when optimising dosages of NO delivery in future work investigating NO as potential treatment for tendon injuries. The second study examined gene expression in an acute tendon injury model in the rat at 1, 7, and 21 days post injury, roughly representing the inflammation, proliferation, and remodelling phase of wound repair. Several novel genes and pathways were found to be differentially expressed at each stage of healing. Of particular interest were the presence of a significant number of genes related to glutamate signaling, a method of cellular communication that has not previously been shown to exist in tendon. Also upregulated were a number of genes which have previously only been associated with embryonic development. Finally, gene expression in a supraspinatus tendinopathy model in the rat was investigated. Several genetic pathways were identified in tendinopathic tendons which have not previously been associated with the disease, and, analogous to the acute injury model study, glutamate signaling gene overexpression was also prevalent. Further in vitro studies showed that the expression of these genes in tendon fibroblasts were stimulated by glutamate treatment, which in turn upregulated pro-apoptotic pathways causing cell death. This may prove to be an important causative factor in the tendon degeneration seen in tendinopathy, in which apoptosis has been identified as playing a significant role. The results of these studies contribute to a better understanding of the aetiology of several extremely common pathologies of this soft tissue, and may help to develop more targeted therapies for increasing the efficacy of tendon healing in future.

De nombreux travaux ont démontré la modulation, par les tumeurs, de certaines fonctions des cellules du système immunitaire. Dans le cadre de notre travail, nous avons étudié les lymphocytes T CD4+, cellules clefs de la réponse immune spécifique, chez des patientes atteintes d’un cancer du sein. Sur base de l’établissement des profils d’expression génique des lymphocytes T infiltrant les tumeurs, nous avons dérivé la « tumor-infiltrating CD4+ signature » (TICD4S) composée de 61 gènes immuns et qui reflète l’état d’activation immunitaire. Cette signature présente une valeur prédictive chez les patientes porteuses de tumeurs ERBB2-positives et ER-négative/PR-négative/ERBB2-négative: une plus forte expression de ces gènes est associée à une meilleure survie. Nous avons également étudié conjointement les profils géniques établis au départ des lymphocytes T CD4+ de la tumeur, du ganglion axillaire et du sang de dix patientes. Nous avons constaté que ces profils d’expression génique des TIL CD4+ diffèrent selon le statut ER de la tumeur qu’ils infiltrent. Les lymphocytes T ganglionnaires CD4+ subissent également les effets de la masse tumorale et, tout comme les TIL, sont moins activés chez les patientes porteuses de tumeurs ER-négatives. Par contre, les lymphocytes T sanguins semblent subir dans une moindre mesure les effets de la tumeur et peu de différences ont été notées par rapport à leurs homologues isolés chez des donneuses saines. Notre étude a contribué à la caractérisation des lymphocytes T CD4+ chez les patientes atteintes d’un cancer du sein et offre matière à réflexion pour des investigations futures où davantage de patientes pourraient être incluses et où des analyses pourraient être réalisées également au niveau protéique. Les cellules immunes semblent jouer un rôle dans l’évolution de la pathologie cancéreuse et sont susceptibles d’offrir une valeur prédictive sur l’évolution clinique des patientes porteuses de tumeur ER-négative et ERBB2-positive.

Testicular function is fundamental to male fertility, since testicular cells act in collaboration with each other to signal sex differentiation, the initiation of puberty and spermatogenesis. Complications that can be influenced by many factors will affect sperm number, morphology, motility, chromatin quality and acrosomal integrity. The purpose of these studies was to analyze the changes in gene expression in the developing testes and analyze the seasonal changes in gene products in sperm of mature bucks. In the first experiment, testes were harvested from five Alpine bucks at 0, 2, 4, 6, and 8 months of age. Northern and in situ hybridization indicated that the largest change in gene expression occurred during the first 4 months of goat testes development. Sex determining region Y-box 9 (SOX9) and Heat Shock protein A8 (HSPA8) peaked at 2 months of age, and were expressed in Sertoli cells and spermatogonium, respectively. At 4 months, expression of Stimulated by Retinoic Acid gene 8 (STRA8), Protamine1 (PRM1) and Outer Dense Fiber protein 2 (ODF2) was strongly up-regulated in early and maturing germ cells, respectively. In the second experiment, RNA from ejaculated spermatozoa collected from mature Alpine bucks in peak (October) and non-peak (April) breeding season were analyzed on a 4 x 44K Agilent bovine microarray. One thousand three hundred and eighteen gene products were differentially expressed 2-fold or more (p ≤ 0.05 ) was expressed in mature goat sperm collected October and April. To eliminate the likelihood of false positives, the cut off was set to fold change of 3 or more at p ≤ 0.01 which narrowed the list of genes to 50 transcripts. Real time PCR results confirmed the expression of Sperm Adhesion Molecule 1 (SPAM1) in April, and the expression of Glycerol kinase 2(GK2) and Myc Binding Protein 2 (MYCBP2) in October. Based on the results from both experiments, it can be concluded that: SOX9 and HSPA8 expression play an important role in tubular formation and germ cell maintenance; two months after SOX9 and HSPA8 expression, genes that are associated with spermatogenesis initiation and completion are upregulated; and validation of the seasonal changes in sperm mRNA levels may provide additional insight to testicular events as they relate to breeding and non-breeding season.

There is an inherent conflict over land use between humans and wildlife. Human activities can alter habitat, creating pressure on North American large carnivore populations. Traditional wildlife techniques can be slow to show population declines, especially in long lived species with slow reproduction rates and high mortality of young, such as grizzly bears (Ursus arctos), which leads to delayed information for land managers trying to find the balance between human use of land and preservation of wildlife. Concern about population health of grizzlies in Western Alberta, Canada has lead to investigation of the impacts of current land use within grizzly bear habitat. The objective of this work was to develop a protein microarray that could detect patterns of physiological stress in a rapid manner with small samples of grizzly bear tissue. Sampling from four regions in the foothills of the Rocky Mountains in Alberta resulted in the capture of 133 bears. During the developmental phase, proteins involved with mitochondrial function were found, using two dimensional gel electrophoresis, to be altered in situations of increased stress. Limited cross-reactivity was found when evaluating grizzly bear stress protein expression using commercially available protein microarrays. The protein microarray developed in this thesis consists of 31commercial antibodies validated for grizzly bears. These antibodies recognize proteins associated with different aspects of the stress response, including the hypothalamic-pituitary-adrenal axis, apoptosis/cell cycle, cellular stress, and oxidative stress and inflammation. Skin was selected as the tissue for evaluation of protein expression. Strong correlations were found between many of the proteins within functional categories. Model selection for the protein categories revealed variation that corresponded with region, serum markers of stress (total cortisol and hsp60), growth, the density of roads in the habitat and the amount of anthropogenic change in the bears home range. Regional trends of expression found bears in Swan Hills and bears from North highway 16 having elevated expression of the proteins measured by the microarray. The protein microarray was thus able to detect expression patterns reflecting physiological and environmental markers. The array shows great promise for future use in detection of potential distress in wildlife populations due to alterations of their habitat.

Streptococcus pneumoniae is a gram-positive bacterium that can be found in both healthy carriers as well as in people that have developed disease. One of the major virulence factors of pneumococci is their polysaccharide capsule. Based on the capsule that surrounds the bacteria, pneumococci are divided into at least 90 different serotypes. Some serotypes seem to be more related to virulence than others. I have with comparative genome hybridization microarray technique, studied gene differences between 18 epidemiological well-characterised pneumococcal strains with different potential to cause disease. A microarray chip based on two sequenced pneumococcal genomes, R6 and TIGR4, has already been designed. According to Hierarchical clustering, both the serotype and the genetic type as assessed by MLST (sequence type or ST) seem to have impact on the relationship of clinical isolates. Most clinical isolates of the same serotype are clustered together except for serotype 14 isolates that seem to be more divergent. Further more the number of genes that are divergent between clinical isolates compared to R6 and TIGR4 differ from 65 to 289. Preliminary results indicate that although there is diversity among clinical isolates some are more closely related to each other then others. Absent genes seem to be evenly distributed among all 18 clinical isolates tested but hypothetical genes and genes for cell envelope are two groups of role categories that are absent to the largest extent in all isolates. According to results from microarray analysis, a gene region, spr0112-spr1015- is present in all type 9V isolates and absent in many isolates of serotype 14, 19F and 7F. These results have been confirmed by polymerase chain reaction (PCR). Conserved genes in a region around the capsule genes have been sequenced to identify marker genes for a capsulular switch between serotype 9V and 14. Preliminary results of the sequencing showed that as much as 750kb might have been transferred in the event of capsular switch.

There is an inherent conflict over land use between humans and wildlife. Human activities can alter habitat, creating pressure on North American large carnivore populations. Traditional wildlife techniques can be slow to show population declines, especially in long lived species with slow reproduction rates and high mortality of young, such as grizzly bears (Ursus arctos), which leads to delayed information for land managers trying to find the balance between human use of land and preservation of wildlife. Concern about population health of grizzlies in Western Alberta, Canada has lead to investigation of the impacts of current land use within grizzly bear habitat. The objective of this work was to develop a protein microarray that could detect patterns of physiological stress in a rapid manner with small samples of grizzly bear tissue. Sampling from four regions in the foothills of the Rocky Mountains in Alberta resulted in the capture of 133 bears. During the developmental phase, proteins involved with mitochondrial function were found, using two dimensional gel electrophoresis, to be altered in situations of increased stress. Limited cross-reactivity was found when evaluating grizzly bear stress protein expression using commercially available protein microarrays. The protein microarray developed in this thesis consists of 31commercial antibodies validated for grizzly bears. These antibodies recognize proteins associated with different aspects of the stress response, including the hypothalamic-pituitary-adrenal axis, apoptosis/cell cycle, cellular stress, and oxidative stress and inflammation. Skin was selected as the tissue for evaluation of protein expression. Strong correlations were found between many of the proteins within functional categories. Model selection for the protein categories revealed variation that corresponded with region, serum markers of stress (total cortisol and hsp60), growth, the density of roads in the habitat and the amount of anthropogenic change in the bears home range. Regional trends of expression found bears in Swan Hills and bears from North highway 16 having elevated expression of the proteins measured by the microarray. The protein microarray was thus able to detect expression patterns reflecting physiological and environmental markers. The array shows great promise for future use in detection of potential distress in wildlife populations due to alterations of their habitat.

Coral reefs are one of the world s most valuable ecosystems but are declining at an accelerating rate. Common stressors impacting coral health include elevated temperatures, changes in light intensity, sedimentation, and increased exposure to pollutants. Traditionally, physiological responses have been measured to assess coral health but usually do not identify the stressor or the underlying mechanisms causing a response. In addition, coral may be stressed beyond recovery by the time a physiological response is observed. Changes in gene expression are key elements of the stress response, usually occur before physiological damage is evident, and can be directly related to the causative agent of stress. My research focuses on detecting sublethal responses to stress in Scleractinian coral using genetic biomarkers and gene expression profiling. Through the application of molecular technology, I have developed a coral stress gene microarray to investigate the responses of coral to various stressors. Results from controlled laboratory exposures provide evidence for unique gene expression profiles associated with specific stressors. Results from field studies reveal the feasibility of using array technology to investigate changes in gene expression of natural coral populations across time and between sites. For example, the array has been used to detect stress in coral populations related to seasonal events, such as precipitation as well as point source stress, such as xenobiotics. The temporal and spatial regulation of specific genes within a genome determines the metabolic activity of an organism and can be used to identify changes in cellular responses to various stimuli. These cellular events precede population-level changes and could be useful biomarkers if linked to specific physiological or ecological events. This research is important because it identifies stress at a sub-lethal level and can aid resource managers in decision making by prioritizing the stressors impacting coral reefs.