Philosophiae Doctor - PhD / Breast cancer is the second most common cancer amongst South African women. Despite ongoing efforts to combat breast cancer, current prognostic and/or therapeutic monitoring methods are limited since very little improvement, in the rate of long term recurrence of breast cancer, has been observed. Considering this, developing novel strategies to detect breast cancer recurrence – at an early onset – is crucial for monitoring the disease and potentially preventing disease progression. Methods currently used for the detection of BC are costly and can also be very uncomfortable for the patient. These methods are also too costly to use as a routine test, following surgery or treatment to assess disease progression. Thus, developing a cost-effective detection method appears to be an appealing alternative. Serum/blood-based biomarkers are ideal targets for the development of low cost detection assays. Two candidate biomarkers, unique ligand binding protein 2 (ULBP2) and glial cell line-derived neurotrophic factor family receptor alpha 1 (GFR1) were identified using bioinformatics and proteomics, respectively. These biomarkers have demonstrated to be useful prognostic biomarkers for breast cancer. The selection of aptamers against these biomarkers can facilitate the development of cost-effective detection methods. Aptamers are short DNA or RNA oligonucleotides that have very high affinity and specificity for its targets and can potentially replace antibodies as tools for molecular recognition in detection systems, such as the enzyme-linked immunosorbent assay (ELISA), lateral flow assays and electrochemical biosensors.

Landscape ecotoxicology is the study of dose-response relationships to toxicants and integrating environmental factors across a defined landscape. In this thesis, I contributed new knowledge to the field of landscape ecotoxicology by adapting analytical methods to assess spatial patterns of chemical exposure among different wildlife keystone species, quantify the relationships between contaminant sources and exposures, and quantify dose-response relationships across large landscapes. Currently, there are few landscape ecotoxicology tools available for quantifying geospatial patterns of environmental toxicology data. To address this gap, I adapted spatial and statistical methods and demonstrated how they can be used to 1) integrate data and assess spatial patterns of contaminant exposure; 2) assess spatial patterns of exposure to complex mixtures; and 3) examine dose-response patterns across landscapes. I developed fur Hg as a biomarker medium as a non-invasive biomonitoring tool in river otter (Lontra canadensis) and mink (Neovison vison) by developing conversion factors that can be used to estimate internal organ Hg from fur Hg, using a meta-regression approach. Based on these results, I suggest that the fur Hg screening guideline be reduced from 20 ug/g to 15 ug/g to be more conservative. I also quantified how the distribution of fur Hg changes across the pelt of river otters. Results from this study indicate that topcoat should be used for biomonitoring as it is less variable than the undercoat and samples should be taken from the forebody (head and legs) for the most accurate organ Hg estimation. Using biomarkers of exposure, I quantified the relationship between sources of Hg and factors that promote Hg bioaccumulation with dietary Hg from stomach contents and fur Hg to establish fur as a proxy for bioavailability of environmental Hg. I also assessed spatial dose-response patterns between fur Hg and fur cortisol using a geographically weighted regression (GWR). Based on these results I use my proposed fur screening guideline of 15 µg/g to categorize fur Hg exposures and demonstrate that at low exposures (<15 µg/g) in fur, Hg has a positive relationship with cortisol. Conversely, at high exposures (>15 µg/g) in fur, Hg has a negative relationship with cortisol. This research provides a field example of heterogeneous dose-response relationships. Finally, I assessed spatial patterns of complex metal exposures in a variety of biomonitoring datasets. I used normalization and transformation techniques to effectively combine datasets comprised of different species and life stages. I then used a spatial principal components analysis (sPCA) to exemplify clusters of complex exposures associated with oil and gas development in regions of Alberta, Canada. These advancements in the field of landscape ecotoxicology will help advance evidence-based long-term ecological monitoring programs.

MSc (Med) Molecular Medicine and Haematology, Faculty of Health Sciences, University of the Witwatersrand, 2009 / This is the third phase of a project to identify, confirm, and operationalise biomarkers for crystalline silica dust exposure that could be used for surveillance of dust exposure levels in South African mines. The first phase of the project involved a comprehensive review of the relevant literature [Gulumian et al., 2006] from which ten potential biomarkers of effect were identified as being worthy of further investigation. The second phase of the project examined the ten identified biomarkers in silica dust-exposed and unexposed black male subjects [Murray et al., 2006]. Two of the ten short listed biomarkers, namely erythrocyte glutathione peroxidase (GPx) and serum Clara cell protein 16 (CC16), were found to have significantly reduced levels in the silica dust-exposed versus unexposed subjects. In addition, the biomarkers were found to be unaffected by HIV sero-status, smoking, age and the presence of silicosis. As a result, this third phase of the project aimed to confirm the levels of and further analyze GPx and CC16 in miners exposed to crystalline silica dust. This third phase involved the measurement of the levels of erythrocyte GPx and serum CC16 in 80 adult male gold miners upon their return from leave and then again two to six months after they had returned to work (involving exposure to crystalline silica). Before the field work was conducted, however, the optimal operational parameters for the biomarkers (namely storage temperature, delay in time between blood collection and separation, laboratory temperature and storage duration) were established. The results of these optimization experiments were used to develop Standard Operating Procedures (SOPs) for biomarker specimen handling and storage under field conditions, and for laboratory assays. In this phase, the findings of the second phase were confirmed in that the levels of GPx and CC16 were lowered in miners exposed to crystalline silica dust and were unaffected by age, race and cigarette smoking. In addition, while CC16 was unaffected by the presence of radiological silicosis, GPx may have been affected. Finally, the decrease in the levels of GPx activity and CC16 concentration observed in the study were unaffected by the level of silica dust exposure (high or low) as determined by job category or by the duration of crystalline silica exposure. Regarding the levels of GPx activity, the results suggested that GPx levels decrease after two to six months of chronic exposure to crystalline silica dust and remain decreased (throughout the working week and over a weekend) and then increase or even recover to normal levels during a period of leave. It was therefore concluded that GPx activity levels rise and fall, in response to silica dust exposure, gradually and over periods of some time, possibly months. The CC16 results were, however, less promising. After two to six months of chronic exposure to crystalline silica dust there was a significant change in CC16 on a Wednesday afternoon following an 8-hour shift and during the duration of a shift. In addition, there is the possibility that the observed changes were due to a time-dependent diurnal variation in the CC16 levels. It was concluded that the results of the current phase warrant further research into the use of erythrocyte GPx and serum CC16 as biomarkers of early effect from crystalline silica exposure.

L’objectif de cette thèse est de définir des biomarqueurs relatifs à l’immunothérapie allergénique (ITA). Il peut s’agir de biomarqueurs prédictifs d’une réponse au traitement qui vont permettre aux cliniciens d’adapter les schémas thérapeutiques ou de biomarqueurs d’efficacité facilitant le suivi clinique des patients au cours du traitement. La stratégie de recherche est basée sur une hypothèse qui consiste à dire que les cellules dendritiques (DCs) sont impliquées dans le succès de l’immunothérapie. En particulier, nous supposons que le traitement induit une baisse des DCs effectrices et une augmentation des DCs tolérogènes.Dans une première partie, un criblage de molécules biologiques et pharmacologiques a été entrepris sur les DCs dérivées des monocytes afin de générer in vitro des DCs effectrices de type DC1 etDC17 et des DCs régulatrices. Quatre molécules ont ainsi été identifiées pour leurs propriétés polarisantes. En particulier, les protéases d’Aspergillus oryzae se sont révélées être des inducteurs forts de tolérance. Le phénotype des DCs régulatrices obtenu a été étudié en détail ainsi que la polarisation et la fonctionnalité des lymphocytes T générés après cocultures.Dans une deuxième partie, deux approches de protéomique quantitative (la 2D-DIGE et la LCMS/MS sans marquage) ont été utilisées pour comparer les protéomes des DCs régulatrices et desDCs effectrices. Le différentiel d’expression des protéines les plus pertinentes a été validé au niveau transcriptionnel et protéique dans différents modèles. Le suivi des marqueurs dans des cellules du sang de patients traités ou non par ITA lors d’une étude clinique randomisée, contrôlée, en double aveugle, a permis de définir deux nouveaux biomarqueurs d’efficacité précoce de l’immunothérapie. Ces marqueurs pourront être suivis lors des traitements de désensibilisation pour distinguer les patients répondeurs des non-répondeurs. Par ailleurs, le suivi de ces biomarqueurs pourrait être essentiel dans d’autres pathologies comme les maladies auto-immunes ou encore la transplantation. / The aim of this thesis is to define biomarkers of allergen-specific immunotherapy (SIT).These biomarkers can be predictive of a clinical response or could be efficacy biomarkersable to discriminate responders versus non responder patients. The research strategy is based on the following hypothesis: if immunotherapy works, effector DCs are decreased where as regulatory DCs are increased locally or in the peripheral blood.First, we screened several biological or pharmacological agents to identify effector orregulatory DCs polarization agents. Four distinct molecules lead to the generation of eitherDC1, DC17 or regulatory DCs. In particular, proteases from Aspergillus Oryzae were clearinducer of tolerogenic DCs. The phenotype of those cells and the CD4+ T cell polarization induced after coculture were characterized extensively.In a second part, two proteomic approaches were used to compare the whole cell proteome of generated DCs. Most pertinent markers of polarization were validated in several cellular models. Markers were also followed in a randomized, double blind, placebo controlled clinical trial testing the efficacy of grass pollen tablets. Two markers were up regulated in patients who responded to the treatment pointing to a potential role of these proteins as early efficacy biomarkers. These markers are of crucial interest in the follow up of patients after SIT and could also be used in other diseases like autoimmune diseases or transplantation.

Glycosaminoglycans (GAGs) are linear polysaccharides found on most animal cell surfaces and in extracellular matrices. Their key biological roles include cell signalling, cell-to-cell recognition, bacterial and viral adhesion, and antibody production. They are composed of disaccharide repeating units, containing uronic acid and amino sugar residues, and may be highly heterogeneously sulfated. Protein-GAG complexes are thought to play an important role in a number of aspects of cancer development, but are an under-studied area due to a lack of enabling tools to facilitate their analysis as discussed in Chapter 1. To obtain homogenous GAG structures, a range of conditions for the separation of GAG oligosaccharides by Zwitterionic Hydrophilic Interaction Liquid Chromatography (ZIC-HILIC) were developed using commercial chondroitin sulfate standards; these are discussed in Chapter 2. Mixed-modal separation mechanisms were explored across a different buffer compositions and elution programs to optimise the conditions to suit individual classes of native and modified glycosaminoglycans. These methods were then applied to the separation of chondroitin sulfate mixtures and heparin oligosaccharides. ZIC-HILIC may also be coupled to Mass Spectrometry to produce an online analytical method for GAG mixtures. A range of optimised conditions for the analysis of low molecular weight glycosaminoglycan oligosaccharides by Electrospray Mass Spectrometry were developed using commercial chondroitin sulfate disaccharide standards; as discussed in Chapter 3. These conditions were then employed in the tandem mass spectrometry (MS/MS) of chondroitin sulfate to identify diagnostic fragmentation patterns and this was applied in the characterisation of chondroitin sulfate disaccharides derived from enzymatic cleavage. The position of ring substituents can also be identified using this methodology, which is desirable in the analysis of chemically-labelled GAG structures. To allow for high resolution EPR and NMR studies of protein-GAG interactions, synthetic procedures for the incorporation of paramagnetic centres into GAG oligosaccharides and proteins were developed and these are discussed in Chapter 4. A propargyl-modified lysine was synthesised for recombinant expression into myoglobin. Copper-Catalysed Azide-Alkyne Cycloaddition (CuAAC) was employed in the spin labelling of myoglobin, however traditional experimental conditions were found to reduce the TEMPO-based spin labels used. Alternative conditions were developed using glutathione to stabilise the copper (II) catalyst without reducing the spin label. Spin labelling of the modified lysine was monitored by EPR. Modification of the non-reducing end was explored for the spin labelling of GAG oligosaccharides, to avoid the ring opening that typically occurs in reducing end labelling, and make use of the unsaturated uronic acid that is derived when obtaining GAGs through enzymatic depolymerisation. A hydrazide labelling of uronic acids was initially adopted, however due to concerns regarding selectivity and availability of hydrazide spin labels, an alternative Thiol-Ene Click (TEC) method was developed. TEC labelling of monosaccharides and unnatural amino acids was found to proceed efficiently under aqueous conditions and was monitored by NMR and HPLC. Preliminary studies in the TEC labelling of chondroitin sulfate disaccharides proved promising, however further optimisation is required for this method to be utilised in the study of protein-GAG interactions by NMR.

Research in the field of neuroendocrine tumors (NETs) has increased over the last decade, including studies focused on biochemical markers (biomarkers) of the disease. There is also growing interest in how NETs impact patients' quality of life (QOL). Consequently, there is a paucity of information about whether the expression of the specific disease biomarkers affects QOL as well as whether the primary tumor site impacts QOL. Using the explanatory model of health promotion and quality of life in chronic disabling conditions as the theoretical framework and data collected with the Norfolk QOL-NET instrument, this study's purpose was to fill that gap in knowledge through research questions addressing the relationship between the primary tumor site and NET patients' total QOL score as well as the effect of specific NET biomarkers on NET patients' total QOL score. Data were analyzed using descriptive statistics, one-way analysis of variance (ANOVA), regression analysis, and post hoc tests to determine significance. Results from an ANOVA showed that abnormal NET biomarkers affected total QOL (p = 0.011). In the analyses of whether the independent biomarker variables affected the dependent total QOL variable, only the result for Serotonin Normal was significant (p = 0.002). The presence of abnormal biomarker measurements also affected two of the Norfolk QOL-NET domains significantly, gastrointestinal and physical functioning (p = 0.005 and p = 0.030, respectively). By understanding the relationship between NETs and patient QOL, the potential positive social change implications are helping NET patients assess the severity of their condition, determining what affects their well-being, and using this information to help monitor their treatment/progress.

Petroleum hydrocarbons are emitted into the environment via natural and anthropogenic activities. Once emitted, these hydrocarbons can be transported globally, persisting and accumulating in aquatic ecosystems. In the Alberta oil sands region (AOSR), mining activities have significantly altered and polluted the surrounding aquatic and terrestrial environments with heavy metals and various petroleum hydrocarbons including polycyclic aromatic hydrocarbons (PACs). Though PACs have been tracked through time using dated lake sediment cores, separating natural and anthropogenic PACs can be difficult. In the Peace Athabasca Delta (PAD) this task is especially difficult as this region has been receiving annual inputs of naturally eroded bitumen throughout history. Petroleum biomarkers are unique petrogenic compounds (i.e. derived from petroleum) which may provide a secondary proxy to track mining impacts. This thesis investigated the impacts of mining activities on the AOSR and the PAD using two different proxies, PAC and petroleum biomarkers. These two regions were compared to reference lakes to the south and northwest of the Athabasca oil sands formation, in order to provide a natural signal, with minimal oil sands mining contamination. Historically deposited PACs and petroleum biomarkers were analysed in radiometrically dated lake sediment cores from the AOSR and the PAD, Alberta. Sediment profiles in the AOSR (Saline Lake) showed increases in PAC fluxes for both alkylated and parent compounds coeval with mining activities. Alkylated PAC fluxes in reference lakes (Mariana Lake and BM11) increased at the height of oil sands development (1990s). PAD lakes showed no statistical increase in PAC flux through time due to high levels of naturally eroded bitumen entering the system. Parent PAC diagnostic ratios, however, showed clear shifts from pyrogenic (primarily wood burning) in pre-development sediments to petrogenically derived PACs in modern sediments, in both AOSR and PAD lakes, coeval with oil sands development. Petroleum biomarker diagnostic ratios in Saline Lake and PAD lakes remained stable through time, indicating a clear current and historical petroleum signal originating from the AOSR. Reference lakes (Mariana Lake and BM11) showed the greatest change in petroleum biomarkers. Historically, these lakes had signatures uncommon of petroleum sources, however, in recent years petroleum inputs from mining development were revealed by these petroleum biomarkers. This study compared the historical trends of several petroleum hydrocarbons in lake sediment to the historical emissions of these petroleum hydrocarbons from oil sands mining operations. Notably, we show the potential for petroleum biomarkers to trace petroleum hydrocarbon contamination in the environment.

Most environmental studies concerning the environmental health of the Swan- Canning River system have focussed on nutrient inputs from both rural and urban catchments that are the cause of algal blooms. On occasions these algal blooms have resulted in fish deaths attributed to oxygen starvation. Relatively few studies have examined whether non-nutrient contamination is affecting the health of the riverine environment. Those studies that have, have concentrated on measuring the levels of heavy metals, organochlorines, organophosphates, and hydrocarbons in the sediments and water of the river system, and in the flesh of the biota. However, chemical analysis often fails to detect chemicals of concern due to high laboratory detection limits. In addition, analysis of the body burden of contaminants within biota does not necessarily convey if exposure is inducing adverse effects at the individual or ecosystem levels. The use of biochemical markers as a tool for the assessment of the health of the Swan-Canning River system was examined under a collaborative research project with the Waters and Rivers Commission, established in response to the recognition of the paucity of information from chemical analyses. The present study focussed on the estuarine portion of the Swan-Canning River system, using the black bream (Acanthopagrus butcheri), an estuarine dependent fish species, as a biomonitoring tool. Prior to the commencement of this study it had been determined that the black bream was a suitable fish species for use as a biomonitoring tool when using mixed function oxygenase (MFO) activity induction under laboratory conditions. / Biopsies taken from feral black bream collected from eight sites during the period 2000 to 2002 from the estuary confirmed that the use of MFO induction in this fish species as a biomarker of exposure to organic contaminants is a reliable biomarker. Fish gender was a confounding factor in the interpretation of MFO induction when using the enzyme ethoxyresorufin-O-deethylase (EROD) as EROD activity was suppressed in both pre- and post-spawning female black bream. No such suppression was identified when using the MFO enzyme ethoxycoumarin-O-deethylase (ECOD). However, due to differences in the pattern and intensity of the induction of EROD and ECOD activities it was concluded that ECOD activity was not a substitute for EROD activity to detect certain chemical as ECOD activity represents a different cytochrome P450 pattern to EROD activity. No spatial, seasonal or interannual differences in the level of the enzyme sorbitol dehydrogenase (SDH) in the blood of the black bream were measured indicating that the interpretation of MFO activity induction was not compromised by hepatocellular damage. This study has shown that the black bream in the Swan-Canning Estuary are exposed to, and are metabolising polycyclic aromatic hydrocarbons (PAHs), notwithstanding that the chemical analysis of the contaminant load of these substances in the estuarine waters is consistently below laboratory detection limits. In addition, biomarker responses such as ECOD activity indicate that various other organic pollutants are present and are being metabolised by the black bream. / The measurement of biliary metabolites clearly show that, under winter conditions, the comprehensive drainage system of the Swan Coastal Plain contributes PAHs from pyrogenic sources such as burnt fuels into the estuary although the onset and intensity of rainfall events notably impacts on the volume of stormwater inflow. During the summer months, when freshwater flow is minimal, petrogenic sources of PAHs are dominant. Metabolic enzyme analysis points to the black bream being challenged in their aerobic capacities during summer, and that gill tissue was the most suitable tissue to evaluate the aerobic and anaerobic capacity of this fish species. Furthermore, there was a significant negative correlation between stress protein (hsp70) expression and DNA integrity in field-collected fish suggesting that the black bream within the estuary are highly stressed. No gradient of response in biomarker levels was identified in the Swan-Canning Estuary under either winter or summer conditions indicating there are multiple sources of inputs of potential pollutants along the length of the estuary. Stormwater and road runoff are the primary source of pollutant input into the estuary in the winter months, while summer biomarker levels, particularly PAH, appear to reflect the high usage of the estuary for recreational purposes and runoff from poorly irrigated parks and gardens. Significant rainfall events at any time of the year have the potential to adversely impact the biota of the estuary, particularly when these events result in a flush of water from the drains following long dry periods. / The study shows that the black bream is a suitable fish species to use under field conditions to detect the presence of bioavailable non-nutrient contamination within the Swan-Canning Estuary. A suite of biomarkers in black bream have been tested seasonally and annually but only a small number of biomarkers have proven suitable for routine monitoring of the health of the Swan-Canning Estuary. This treatise concludes with several recommendations for further investigations into biomarkers of fish health for the purpose of increasing our understanding on the sources and type of contamination entering the estuary, and potential effects on the aquatic biota of the Swan-Canning River system. These recommendations include, but are not limited to: (1) the need to determine baseline levels for the different biomarkers investigated in this study, (2) the examination of the Moore River or the Warren River estuaries as potential reference sites for biomarker studies in the Swan- Canning Estuary, (3) the advantage of identifying a second estuarine-dependent indigenous fish as a biomonitoring tool, (4) the requirement for a targeted study aimed at clarifying the relationship between major drain discharges, biomarker levels and impacts on river biota, and (5) a study of estuarine waters utilising SPMDs be undertaken in tandem with biomarker analysis of field captured fish would be beneficial.

Pollution from human and animal faecal waste is a major cause of deteriorating water quality and increased nutrient loads in coastal and inland waterways. Management of this problem depends on knowing which sources of faecal matter are the cause and what is the degree and extent of the pollution. Bacterial indicator organisms have long been the principal method used to test water samples for faecal contamination. However, none of the currently used bacterial indicators on their own are source specific enough to distinguish different sources of faecal matter. The use of faecal sterol biomarkers in conjunction with existing bacterial indicators offers a new way to distinguish sources of faecal contamination. This study investigates the sources of faecal sterols, the relationship of coprostanol to existing bacterial indicators of faecal pollution, the degradation of faecal sterols and the problem of determining the sources of faecal contamination and the distribution of faecal contamination using faecal sterol biomarkers. 5p-Stanols (i.e. faecal sterols) were found to be significant constituents of human, herbivore (i.e. cows, sheep etc.) and pig and cat faeces. Human faeces contained 73 ± 4% coprostanol in relation to the sum of coprostanol and 24-ethylcoprostanol and primary treated effluent contained 86 ± 0.4% coprostanol. Herbivore faeces contained 38 ± 4% coprostanol and 62 ± 4% 24-ethylcoprostanol whereas pig faeces contained 50 � 5% of each compound. Both birds and dogs faeces contained either trace amounts of 5B-stanols or they could not be detected. Notable differences were observed in the abundance of Closthdium perfringens spores between the faeces of birds and domestic pets such as cats and dogs. The above differences were subsequently exploited to distinguish faecal contamination in Lake Tuggerah. An examination of the relationships between coprostanol and bacterial indicator concentrations from several environments revealed that 60 and 400 ng L of coprostanol corresponded to currently defined primary and secondary contact limits for bacteria measured as either thermotolerant coliforms or enterococci in the environment. Four degradation experiments showed faecal sterols and related sterols such as cholesterol decay at similar rates. An induction period was observed in all experiments which meant that simple exponential equations to describe the rate of decay of coprostanol were inadequate; a complimentary log - log transformation of the data was used and the equation: Y = l-Exp(-Exp(time x -0.01 + temp x -0.158 + 3.33)) x 100 was derived where Y equals the predicted percentage of coprostanol remaining over time at a given temperature. In terms of persistence in the environment, Clostridium perfringens spores > coprostanol > enterococci > thermotolerant coliforms. Two field studies were undertaken to highlight the use of faecal sterols. In the Lake Tuggerah study, the results indicated that faecal contamination of receiving waters in the Tuggerah Lakes during rain events was significant, but was not derived from human faecal matter; rather it appears to be principally derived from native birds and, to a lesser extent, domestic pets. In the Derwent Estuary study, based on the distribution of the faecal biomarker coprostanol, the mid estuary and parts of the upper estuary (from Newtown Bay to Taroona), were found to be severely contaminated by sewage. In summary, the use of faecal sterols to trace faecal contamination were found to be an invaluable addition to the tools water managers use to investigate faecal pollution.

La découverte, la caractérisation et le dosage de biomarqueurs ont permis douvrir de nouveaux horizons tant en thérapeutique que dans létablissement dune médecine personnalisée. Dans un contexte danalyse en milieu complexe, lapplication de techniques fiables, sensibles, exactes et de haute résolution est de rigueur. Dans ce but, le couplage de la spectrométrie de masse à une technique séparative telle que la chromatographie liquide (LC-MS) ou lélectrophorèse capillaire (CE-MS) offre des potentialités intéressantes. Une méthode de dosage de la cysdopa, un marqueur du mélanome, dans le plasma par LC-MS a été mise au point. Une étape préliminaire dextraction en phase solide sur un support hybride alliant des interactions de type hydrophobe, hydrophile et échangeurs de cations a conduit à lobtention dexcellents taux de récupération. Les conditions chromatographiques assurant la séparation de la cysdopa des composés endogènes susceptibles dinterférer et les paramètres de la détection par spectrométrie de masse en tandem ont été optimisés. La présence déventuels effets de la matrice a été investiguée en détails. Une étude de la stabilité de la cysdopa dans les différentes conditions opératoires a été réalisée répondant ainsi aux normes préscrites par la FDA. Ensuite, une validation complète de la méthode en milieu plasmatique selon lapproche faisant appel aux profils dexactitude sur une gamme de concentration allant de 1,6 à 200 ng/ml a été menée avec succès démontrant ladéquation de la méthode. Enfin, une étude pilote ainsi quune validation en cours détude ont permis de tester cette méthode de dosage. La seconde technique mise en oeuvre consiste en lapplication du couplage CE-MS à lanalyse de lhepcidine, un petit peptide régulant le métabolisme du fer. Les conditions électrophorétiques assurant la séparation de ce peptide des constituants dun mélange type ont été déterminées. Le traitement de la paroi interne du capillaire et laddition dune cyclodextrine nont pas permis dans la présente application daméliorer la sélectivité et lefficacité des pics. Lutilisation dun électrolyte de haute force ionique associé à laddition dun modificateur organique a par contre offert les meilleures performances. Une approche multivariée en deux temps (criblage puis modélisation) a été appliquée pour loptimisation des paramètres CE-MS. Les effets principaux du voltage du capillaire et de la pression du gaz nébulisant se sont révélés être significatifs tout comme leffet quadratique de ce dernier ainsi que les interactions entre la concentration en électrolyte et le voltage du capillaire dune part et la pression du gaz nébulisant dautre part. Finalement, la sensibilité en terme de rapport signal/bruit a été comparée en CE-UV et CE-MS établissant les potentialités du couplage CE-MS pour lanalyse de peptides. Discovery, caracterization and determination of biomarkers provide new perspectives in therapeutics and establishment of personnal medecine. Analysis in complex matrix requires reliable, sensitive, accurate and high resolution techniques. To achieve this, coupling of mass spectrometry with separative techniques such as liquid chromatography (LC-MS) or capillary electrophoresis (CE-MS) offer interesting potentialities. A method for the determination of cysdopa, a melanoma biomarker, in human plasma was developed using LC-MS. Solid phase extraction in mixed mode, combining hydrophobic, hydrophilic and cation exchange interactions led to excellent recovery. The chromatographic conditions ensuring separation of cysdopa from potential by interfering endogenous compounds and tandem mass spectrometry detection parameters were optimised. The presence of matrix effect was investigated in detail. A full stability study was then performed according to FDA requirements. Finally, a complete validation of the method in plasma following the approach of accuracy profiles over a concentration range from 1,6 to 200 ng/ml was successfully performed, demonstrating adequacy of the developed method. A pilot study and an in-study validation were finally perfomed to test this assay method. The second implemented technique consists in the application of CE-MS coupling to hepcidin analysis, a small peptide regulating iron metabolism. Electrophoretic conditions ensuring separation of hepcidin from model peptides have been determined. Resolution and/or efficiency were not improved by using coated capillaries or by adding a cyclodextrin. A high ionic strength electrolyte associated to the addition of an organic modifier provided the best performance. A multivariate approach (screening and modeling) was applied to the optimisation of CE-MS parameters. The principal effet of capillary voltage and nebulizing gas pressure were significant as well as the quadratic effet of the last parameter and the interactions of electrolyte concentration and, for one part, capillary voltage and, for another part, nebulizing gas pressure. Finally, sensitivity with respect to signal/noise ratio was compared between CE-UV and CE-MS, assessing potentialities of CE-MS for peptide analysis.