La présence de sucres complexes constitue une source nutritive importante pour le microbiote qui assure leur dégradation via des CAZymes. Dans le cadre de cette thèse, nous avons construit in silico un modèle de type minimicrobiome contenant 177 génomes représentatifs des communautés bactériennes dans un microbiote intestinal conventionnel. L’analyse du contenu de ce minimicrobiome nous a permis d’estimer leur abondance et leur diversité. De plus, la comparaison du contenu CAZymes par groupe bactérien de type « phylum » a révélé une variabilité inter-phylum, notamment une diversité de familles CAZymes et une abondance en gènes bien plus élevées chez les Bacteroidetes. Dans un deuxième temps, nous avons développé une puce à ADN sur laquelle nous avons greffé des sondes non redondantes ciblant plus de 6500 gènes codant des CAZymes. Nous avons ensuite testé la "puce CAZyme" par hybridation d’ADN bactérien extrait d’échantillons de selles. Nos résultats suggèrent que cette méthode serait plus sensible dans la détection de CAZymes provenant de bactéries rares par rapport à la métagénomique. Ainsi, il est intéressant de noter qu’en utilisant la puce CAZyme, nous avons pu détecter un gène codant pour une famille GH6, alors que les études métagénomiques n’ont jamais réussi à détecter ce gène dans le microbiome intestinal humain et animal. Enfin, l’examen de huit échantillons de selles a permis l’identification d’un noyau CAZome contenant 46 familles de GHs et PLs, ce qui suggérerait que le microbiote intestinal est caractérisé par une stabilité fonctionnelle en dépit de variations taxonomiques importantes entre les individus testés et indépendamment de leur état de santé. / The bacterial communities that inhabit our gut ensure their growth and survival by extracting their carbon source from the food that transits through the intestines. The complex carbohydrates included in the human diet are almost exclusively degraded by the gut microbiota using CAZymes. We built a minimicrobiome model using 177 genomes associated to gut microbiota. The CAZyme content analysis revealed their huge diversity and abundance in our minimicrobiome model. At the phylum level, the Bacteroidetes genomes showed the greatest CAZyme diversity and abundance. Interestingly, as most of CAZymes found in Bacteroidetes genomes contain a signal peptide allowing their secretion in the intestinal lumen and/or in periplasmic space, members of this phylum are suggested to be the primary degraders of complex carbohydrates. Further, we developed a microarray containing probes to target more than 6,500 CAZyme genes. We then validated the CAZyme microarray by the hybridization of bacterial DNA extracted from the stool samples of individuals. Our results suggest that a microarray-based study can detect genes from low-abundance bacteria better than metagenomic-based studies. A striking example was the detection of gene encoding a GH6-family in all subjects examined, whereas metagenomic studies have consistently failed to detect this gene in both human and animal gut microbiomes. In addition, an examination of eight stool samples allowed the identification of a corresponding core CAZome containing 46 CAZymes families that suggests a functional stability of the gut microbiota despite large taxonomical variations between individuals and independently of health state.

Cette thèse a été réalisée dans le cadre du projet Futurol, un projet national français qui vise à produire du bioéthanol à partir de biomasses végétales telles que le bois ou la paille de céréale. Pour cela, la biomasse doit être prétraitée puis digérée enzymatiquement pour libérer des sucres fermentescibles. Ma contribution dans ce projet a été de découvrir des enzymes originales pour l’hydrolyse de l’hémicellulose, un hétéropolysaccharide, constituant majeur de la paroi cellulaire des cellules végétales. Afin de rechercher de nouveaux biocatalyseurs, une approche de métagénomique a été adoptée afin de sonder les intestins de deux espèces de termites : N. corniger, un termite xylophage, et T. hispaniolae un termite humivore / xylophage. 30 000 clones métagénomiques ont été criblés sur 10 substrats cellulosiques et hémicellulosique, et 660 hits ont été obtenus. La comparaison phénotypique a montré une différence claire entre ces deux banques, probablement liée au régime alimentaire des deux espèces de termite. Le séquençage de 45 clones N. corniger a révélé 120 séquences codant pour des enzymes originales, de nombreuses étant multimodulaires et / ou organisées en cluster de gènes. Dans un second temps, une approche à haut-débit a été adoptée pour le clonage, l’expression et la caractérisation légère de 104 enzymes entières ou formes tronquées. 45 protéines recombinantes ont été produites de manière soluble, et les activités de 19 enzymes et de 12 modules enzymatiques ont été montrées, permettant la mise au point d’une boite à outil hemicellulolytique. Dans certains cas, l’activité de modules classés « Inconnus » a pu être déterminée. Cette approche a été particulièrement pertinente dans le cas de Pm69, une enzyme multimodulaire GH3-UNK-CBM48-CE1 montrant les 3 activités glucosidase, xylosidase and estérase. Cette étude a permis de poser les bases d’un brevet sur cette enzyme. D’un autre côté, les enzymes ayant montré une activité xylanase ou féruloyle-estérase se sont révélées complémentaires d’un cocktail cellulolytique durant la dégradation de paille de blé prétraitée. Enfin, dans une troisième partie, nous avons étudié un fragment d’ADN provenant la banque P. militaris, codant pour 19 ORFs et appartenant à une espèce du genre Bacteroides. La caractérisation biochimique d’Abn43A, Abn43B, Abf51A et Abf51B-trunc a montré que ces 4 enzymes portent des actions complémentaires sur l’hydrolyse de l’arabinane, et qu’elles peuvent agir de manière synergique pour la dégradation de ce polymère pectique. Enfin, l’étude détaillée des 19 ORFs codées sur ce fragment d’ADN nous a permis de proposer un schéma global de détection, d’hydrolyse et de métabolisation de l’arabinane par cette espèce du genre Bacteroides. / This thesis was performed in the context of the Futurol project, a French national project that aims at producing bioethanol from plant biomass such as wood and cereal straw. To reach that goal, the biomass must be pretreated, and enzymatically degraded to release fermentable simple sugar. My implication in that project was to discover original enzymes that can hydrolyze the hemicellulose, a major heteropolysaccharide found in plant cell wall.To mine for new biocatalysts, the gut microbial communities of two species of termite were investigated by a metagenomic approach : Nasutitermes corniger, a wood-feeder termite, and Termes hispaniolae supposed to be a soil-wood feeder. 30 000 metagenomic clones were screened on an array of 10 cellulosic and hemicellulosic substrates and 660 hits were obtained. Phenotypic comparison showed clear differences between both environments, probably related to the diet of the termite. The sequence of 45 N. corniger metagenomic inserts revealed 120 original sequences encoding for putative enzymes of interest. Original sequences encoding for multimodular enzymes were revealed and many ORFs were organized in clusters, suggesting that these enzymes are encoded on Polysaccharides Utilization Locus. In a second part, a high-throughput approach was used for the cloning, the expression and the slight characterization of 104 full-size and truncated enzymes. Forty five recombinant proteins were produced soluble, and their investigation revealed the activity of 19 enzymes and of 12 enzymatic modules, representing a hemicellulolytic tool-box for endo- and exo-type activities. In some cases, the implication of “Unkown” domains in the activity of multimodular enzymes was demonstrated. This approach was particularly efficient for the study of the GH3-UNKCBM48-CE1 Pm69, and this study triggered the patent process for this multiactive glucosidase, xylosidase and esterase. The xylanases and the feruloyl esterases were shown to be particularly efficient to supplement cellulolytic cocktails on pretreated wheat straw. In a third part, we investigated a DNA fragment belonging to a species of the genus Bacteroides and that encoded 19 ORFs. The biochemical characterization of Abn43A, Abn43B, Abf51A and Abf51B-trunc showed that these four enzymes harbored complementary actions for the hydrolysis of the arabinan, and that they can act synergistically for the hydrolysis of this pectic polymer. We also revealed that Abn43B had an original mode of action that we classified as exo-arabinanase. Finally, the in-depth study of the 19 ORFs allowed us to propose the entire scheme for arabinan detection, hydrolysis and utilization by the Bacteroides species carrying this DNA sequence

Wood formation or xylogenesis is a fundamental process for so diverse issues as industry, shelter and a sustainable environment. Wood is comprised of secondary xylem, rigid large cells with thick cell walls that are lignified. The basis for the sturdy cells is an advanced composite made up of cellulose fibers cross-linked by hemicelluloses and finally embedded in lignin. This fiber-composite is the secondary cell walls of woody plants. Cell division and differentiation is regulated by switching on and off genes. Proteins encoded by these genes execute the major functions in the cells. They steer the entire machinery operating the structure and function of the cells, maintaining growth and synthesising essential products such as the cell wall carbohydrates.   Here we describe the investigation of genes and proteins involved in xylan formation as well as the development of a model system that will aid the functional analysis of wood formation. Xylan is the main hemicellulose or cross linking glycan in dicot wood and thereby one of the most abundant carbohydrates on earth. We demonstrate that hybrid aspen cell suspension cultures can be used as a model system for secondary cell wall formation. We have also examined glycosyltransferases from CAZy family 43 that play a part in secondary cell wall formation. We have focused on one of these, Pt×tGT43A, a likely ortholog of Arabidopsis IRX9, which plays a crucial role in xylan formation. The protein was transiently expressed in Nicotiana benthamiana and its function and localization is described. Also, we investigate a glycoside hydrolase, Pt×tXyn10A, involved in wood formation. Its role is not clear but it most likely modifies xylan as it gets incorporated into the secondary cell wall after secretion from the Golgi. This influences the interaction between cellulose, xylan and lignin in the finished wood cell. We have also cloned a transcription factor, Pt×tMYB021, a likely ortholog of Arabidopsis MYB46 and we show that it activates GT43A, GT43B and Xyn10A. By analysis of the promoter sequences we identify a CA-rich motif putatively important for xylem-specific genes.   By mastering proteins involved in xylogenesis we will acquire the tools to improve and develop the wood product market. Xylan is an immense unexploited source of renewable carbohydrate. New products envisioned include e.g. faster growing trees, changed fiber characteristics, optimised utilization of wood carbohydrates for biofuels and biomaterials as well as invention of intelligent materials by biomimetic engineering. / Vedbildning, eller xylogenes, är en grundläggande mekanism för så skilda områden som industri, boende och en hållbar miljö. Ved består av sekundärt xylem som är starka, stora celler med tjocka cellväggar som är lignifierade. Grunden för de starka cellerna är en avancerad komposit bestående av cellulosafibrer tvärbundna av hemicellulosa och slutligen ingjutet i lignin. Denna fiberkomposit är den sekundära cellväggen i vedartade växter. Celldelning och differentiering regleras genom att sätta igång och stänga av gener. Proteiner som kodas av dessa gener utför de viktigaste funktionerna i cellerna. De styr hela maskineriet som upprätthåller cellernas struktur och funktion, underhåller tillväxt samt tillverkar nödvändiga produkter såsom cellväggskolhydraterna. Här beskriver vi utforskningen av gener och proteiner som är inblandade i xylanbildning liksom utvecklandet av ett modellsystem som kommer vara en hjälp i den funktionella analysen av vedbildning. Xylan är den vanligaste hemicellulosan, eller korsbindande glykanen, i lövträd och därför en av de vanligaste kolhydraterna på jorden. Vi demonstrerar att hybridaspcellkulturer i suspension kan användas som ett modellsystem för sekundär cellväggsbildning. Vi har också undersökt glykosyltransferaser från CAZy-familj 43 som tycks spela en viktig roll i bildandet av sekundär cellvägg. Vi har fokuserat på en av dessa, Pt×tGT43A, en trolig ortolog till Arabidopsis IRX9 som spelar en viktig roll i xylanbildning. Proteinet har uttryckts övergående i Nicotiana benthamiana och dess funktion och lokalisering beskrivs. Dessutom undersöker vi ett glykosidhydrolas, Pt×tXyn10A, involverad i vedbildning. Dess roll är oklar men högst sannolikt modifierar det xylan medan det inkorporeras i sekundära cellväggen efter sekretion från Golgi. Detta influerar interaktionen mellan cellulosa, hemicellulosa och lignin i den slutliga vedcellen. Vi har också klonat en transkriptionsfaktor, Pt×tMYB021, en trolig ortolog till Arabidopsis MYB46 och vi visar att den aktiverar GT43A, GT43B och Xyn10A. Genom analys av promotorsekvenserna har vi identifierat ett CA-rikt motiv förmodat viktigt för xylemspecifika gener.Genom att bemästra proteinerna som är ansvariga för vedbildning får vi verktyg att utveckla skogsproduktsmarknaden. Xylan är en ofantligt stor outnyttjad källa till förnyelsebara kolhydrater. En vision är nya produkter som till exempel snabbväxande träd, ändrade fiberegenskaper, optimerat användande av vedkolhydrater för biobränsle och biomaterial såväl som utvecklandet av intelligenta material genom biomimetisk ingenjörskonst. / QC20100730

Arabinogalactan proteins (AGPs) are proteoglycans heavily substituted by arabinogalactan polysaccharides. These are composed of arabinose and galactose, and minor sugars such as glucuronic acid (GlcA), fucose and xylose. The arabinogalactan polysaccharides do not decorate classical AGPs exclusively, but they can also be found decorating a wide range of proteins. Arabinogalactan proteins have been implicated in many processes of plant development. Recently, AGPs were proposed to bind and store calcium at the plasma membrane. They are extracellular, and are localised mainly at the plasma membrane via a GPI-anchor. They can also be soluble in the apoplast. Their low abundance, chemical similarity and high functional redundancy have hindered their study. My strategy to overcome these difficulties was to study knock-out Arabidopsis thaliana plants of glycosyltransferases that transfer sugars specifically onto AG-polysaccharides. Glucuronic acid makes up about 10% of the arabinogalactan polysaccharide structure in Arabidopsis thaliana cell culture AGPs. Previously, the glucuronic acid transferase A TGLCA T14A, a member of the CAZy Glycosyl Transferase 14 family, was shown to transfer GlcA specifically onto AGPs, and knock-out Arabidopsis plants showed a 30% reduction in [Me]GlcA substitution in AGP-enriched preparations. However, no clear growth phenotype was observed. The characterisation of knock-out plants of other GT14 family members and combinations thereof is described here. Based on previous studies (Lamport and Várnai, 2013), I assayed in vitro the calcium binding capacity of AGP extracts from WT and knock-out plants. The results showed that AGP extracts from knock-out plants can hold less calcium than WT plants in vitro. A wide range of plant growth phenotypes were identified. Growth phenotypes can be explained by changes in the cytoskeleton and deficiencies in calcium signaling. Our evidence suggests links between structural deficiencies of extracellular proteoglycans to extracellular calcium and cytoskeleton. This research has the potential to create a new model system for the study of molecular mechanisms dependent on calcium that drive cell expansion, division and differentiation in plants.