Chromosomal toxin-antitoxin (TA) modules have been proposed to function as regulators of cell growth in response to environmental perturbations. The objective of this study was to characterize the MazEF TA system of the human pathogen Streptococcus mutans. Our data showed that the mazEF genes form a bicistronic operon. MazF toxin had a toxic effect on cells and this effect can be neutralized by coexpression of its cognate antitoxin MazE. Furthermore, we demonstrated that MazE and MazF proteins interact with each other in vivo, confirming the nature of this TA as a type II addiction system. We also demonstrated that MazF is a toxic nuclease arresting cell growth through the mechanism of RNA cleavage and that MazE inhibits the RNase activity of MazF by forming a protein complex. Our results suggest that the MazEF TA might represent a cell growth modulator facilitating the persistence of S. mutans in the oral cavity.

Chromosomal toxin-antitoxin (TA) modules have been proposed to function as regulators of cell growth in response to environmental perturbations. The objective of this study was to characterize the MazEF TA system of the human pathogen Streptococcus mutans. Our data showed that the mazEF genes form a bicistronic operon. MazF toxin had a toxic effect on cells and this effect can be neutralized by coexpression of its cognate antitoxin MazE. Furthermore, we demonstrated that MazE and MazF proteins interact with each other in vivo, confirming the nature of this TA as a type II addiction system. We also demonstrated that MazF is a toxic nuclease arresting cell growth through the mechanism of RNA cleavage and that MazE inhibits the RNase activity of MazF by forming a protein complex. Our results suggest that the MazEF TA might represent a cell growth modulator facilitating the persistence of S. mutans in the oral cavity.

A major challenge in biology is to understand and predict the effect of mutations on protein structure, stability and function. Chapter 1 provides a general introduction on protein sequence-structure relationships and use of the CcdAB toxin-antitoxin system as a model to study molecular determinants of mutant phenotypes. In Chapter 2, we describe the use of saturation mutagenesis combined with deep sequencing to determine phenotypes for 1664 single-site mutants of the E. coli cytotoxin, CcdB. We examined multiple expression levels, effects of multiple chaperones and proteases and employed extensive in vitro characterization to understand how mutations affect these phenotypes. While general substitution preferences are known, eg polar residues preferred at exposed positions and non-polar ones at buried positions, we show that depth from the surface is important and that there are distinctly different energetic penalties for each specific polar, charged and aromatic amino acid introduced at buried positions. We also show that over-expression of ATP independent chaperones can rescue mutant phenotypes. Other studies have primarily looked at effects of ATP dependent chaperone expression on phenotype, where it is not possible to say whether mutational effects on folding kinetics or thermodynamic stability are the primary determinant of altered phenotypes, since there is energy input with these chaperones. The data suggest that mutational effects on folding rather than stability determine the in vivo phenotype of CcdB mutants. This has important implications for efforts to predict phenotypic effects of mutations and in protein design. While looking at the mutational landscape of a given gene from an evolutionary perspective, it is important to establish the genotype-phenotype relationships under physiologically relevant conditions. At the molecular level, the relationship between gene sequence and fitness has implications for understanding both evolutionary processes and functional constraints on the encoded proteins. Chapter 3 describes a methodology to test the fitness of individual CcdB mutants in E.coli over several generations by monitoring the rate of plasmid loss. We also propose a methodology for high throughput analysis of a pool of CcdB mutants using deep sequencing to quantitate the relative population of each mutant in a population of E.coli cells, grown for several generations and build the fitness landscape. While the F-plasmid based CcdAB system is known to be involved in plasmid maintenance through post-segregational killing, recent identification of ccdAB homologs on the chromosome, including in pathogenic strains of E.coli and other bacteria, has led to speculations on their functional role on the chromosome. In Chapter 4, we show that both the native ccd operon of the E.coli O157 strain as well as the ccd operon from the F- plasmid when inserted on the E.coli chromosome lead to protection from cell death under multiple antibiotic stress conditions through formation of persisters. Both the ccdF and ccdO157 operons may share common mechanisms for activation under stress conditions and also display weak cross activation. The chromosomal toxin shows weaker activity as compared to the plasmidic counterpart and is therefore less efficient in causing cell death. This has important implications in generation of potential therapeutics that target these TA systems. Chapter 5 describes the use of site-saturation mutagenesis coupled with deep sequencing to infer mutational sensitivity for the intrinsically disordered antitoxin, CcdA. The data allows us to make comparisons between overall as well as residue specific mutational sensitivity patterns with that of globular proteins, like CcdB (described in Chapter 2) and study toxin- antitoxin interaction and regulation through saturation suppressor mutagenesis. Interestingly, we found several examples of synonymous point mutations in CcdA that lead to loss of its activity. In Chapter 6 we attempt to explore the molecular bases for some of these synonymous mutations. In most cases the mutated codon has a similar overall codon preference to the WT one. Initial findings suggest a change in mRNA structure leading to change in CcdB: CcdA ratio, thereby causing cell death. These observations have important implications, because TA systems are ubiquitous, highly regulated and are known to be involved in multiple functions including drug tolerance. However a role for RNA structure in their regulation has not been shown previously. Appendix–I lists the mutational sensitivity scores for the CcdB mutants. Phenotypes for CcdA mutants obtained through deep sequencing have been tabulated in Appendix-II. Overall, we provide extensive datasets for mutational sensitivities of a globular (CcdB) and an intrinsically disordered protein (CcdA). Exploration of the molecular determinants of these mutant phenotypes not only provides interesting insights into CcdAB operon function but is also useful in understanding various aspects of protein stability, folding and activity as well as regulation of gene expression in bacteria.

Staphylococcus aureus, est un pathogène humain majeur responsable d’infections nosocomiales et communautaires. Avec l’utilisation excessive des antibactériens, l’incidence et l’émergence de souches de S. aureus multi-résistantes aux antibiotiques ont augmenté rapidement depuis plusieurs années et constituent un véritable problème de santé publique. Le succès de S. aureus en tant que pathogène est lié à sa capacité à s’adapter rapidement à un nouvel environnement et à produire un arsenal de facteurs de virulence dont l’expression fait intervenir des protéines mais également des ARN régulateurs (ARNrég). Au cours de cette thèse, nous avons montré que les ARNrég sprG1 et SprF1 constitue un système toxine-antitoxine (STA) de type I fonctionnel où sprG1 code pour deux peptides toxiques. En condition normale de croissance, l’expression de la toxine est régulée par l’antitoxine SprF1 au niveau transcriptionnel et/ou post-transcriptionnel et traductionnel, permettant au pathogène S. aureus de croître normalement. En revanche, lorsque la bactérie est confrontée à une carence nutritive globale, l’expression de l’antitoxine est réprimée, laissant ainsi la toxine sprG1 s’accumuler dans la cellule et traduire les peptides toxiques PepG144 et PepG131, responsables de la stase bactérienne. Les deux peptides sécrétés sont capables de lyser les bactéries compétitrices présentes dans le milieu et les érythrocytes humains. Nous avons également montré, qu’en condition de stress hyperosmotique, SprF1 fixe directement les ribosomes, probablement par l’intermédiaire d’un ou de deux sites de fixation aux ribosomes, afin de réguler la synthèse protéique globale et de favoriser la persistance de S. aureus. Ces résultats montrent que SprF1 appartient à une nouvelle classe émergente d’ARNrég régulant la traduction par fixation directe sur le ribosome. Le STA sprG1/SprF1 est le premier exemple de STA de type I où l’antitoxine est le principal acteur de la fonction biologique. / Staphylococcus aureus is a major human pathogen responsible for nosocomial and community-acquired diseases. With the excessive use of antibiotics, incidence and emergence of multidrug-resistant strains of S. aureus have rapidly increased over the last decade and constitute a serious public health concern. The success of S. aureus as a pathogen is due to its ability to adapt quickly to new environment and to produce an arsenal of virulence factors whose expressions are regulated by proteins and regulatory RNA (regRNA). During my PhD thesis, we showed that RNAs sprG1 and SprF1 constitute a functional type I toxin-antitoxin system (TAS) where sprG1 encodes two toxic peptides. During normal growth conditions, toxin expression is regulated by the antitoxin SprF1 at transcriptional and/or post-transcriptional and translational level, allowing the pathogen to grow. Conversaly, when bacteria are confronted to global nutritive starvation, the antitoxin expression is repressed. This allows accumulation of the sprG1 toxin in cell and translation of both toxic peptides, PepG144 and PepG131, responsible for bacterial stasis. Interestingly, both secreted peptides are able to lyse competitor bacteria in the medium and human erythrocytes. We also showed that upon hyperosmotic stress, SprF1 directly binds ribosomes, probably though one or two ribosome-binding sites, to regulate overall protein synthesis and promote S. aureus persistence. These results suggest that SprF1 belongs to the new emerging class of regRNA regulating translation by direct ribosome-binding. The sprG1/SprF1 TAS is the first example of type I TAS where antitoxin is the leading player of the biological function.

京都大学 / 0048 / 新制・課程博士 / 博士(医科学) / 甲第19631号 / 医科博第69号 / 新制||医科||5(附属図書館) / 32667 / 京都大学大学院医学研究科医科学専攻 / (主査)教授 中川 一路, 教授 岩井 一宏, 教授 西渕 光昭 / 学位規則第4条第1項該当 / Doctor of Medical Science / Kyoto University / DFAM

Les métallo-b-lactamases (MBL) acquises constituent une menace sanitaire par risqué d’impasse thérapeutique dans les infections causées par les bacilles à Gram négatif, en particulier lorsque ces bactéries produisent une b-lactamase à spectre étendu (BLSE). La MBL VIM-1 est une carbapénémase qui hydrolyse toutes les β-lactamines, à l’exception des monobactames. Cette enzyme a émergé en Grèce où elle est désormais endémique chez les entérobactéries. Six souches de bacilles à Gram négatif produisant une carbapénémase ont été isolées chez un même patient qui avait été rapatrié de Grèce. Trois de ces souches, Providencia stuartii (Ps), Proteus mirabilis (Pm) et Escherichia coli (Ec), produisaient la MBL VIM-1 et la BLSE SHV-5. Dans chacune de ces trois souches, les gènes blaVIM-1 et blaSHV-5 étaient portés par un plasmide transférable par conjugaison in vitro. Les plasmides extraits des transconjugants présentaient le même profil de restriction et portaient un intégron de classe 1 identique dans lequel le gène blaVIM-1 était intégré. Nous avons émis l’hypothèse qu’un plasmide codant pour VIM-1 et SHV-5 avait été transféré de la souche Ps vers les souches Pm et Ec dans le tube digestif du patient et avons reproduit ce transfert in vivo dans un modèle de souris gnotoxéniques. Au cours de cette expérience, le plasmide codant pour VIM-1 et SHV-5 a été transféré avec succès de la souche Ps vers la souche réceptrice E. coli J53, en dehors de toute pression de sélection par les antibiotiques. Nous avons ensuite analysé la séquence complète du plasmide pTC2 extrait d’un transconjugant obtenu par conjugaion in vivo. Ce plasmide de type co-intégrat (IncA/C, IncR) de 180kb possédait un squelette de type IncA/C et une région de multirésistance (MDR1) au sein de laquelle était intégré un fragment de type IncR de 13kb. L’analyse de cette séquence nous a permis d’identifier un système de transfert de type IncA/C complet et intact et différents types de systèmes de maintien, à la fois au sein du squelette IncA/C et au sein du fragment IncR. La région mosaïque MDR1 contenait neuf séquences d’insertion (sept copies de l’IS26, une IS1 et une IS6100), 10 gènes de résistance aux antibiotiques et l’opéron mer de résistance au mercure qui étaient intégrés dans des transposons unitaires, des transposons composites ou des intégrons. Le plasmide pTC2 cumule des propriétés qui font de lui un véhicule performant de la résistance aux antibiotiques : un large spectre d’hôte, de bonnes capacités de transfert, de bonnes capacités de maintien dans une population bactérienne, une grande plasticité de sa région MDR1 et une grande variété de gènes de résistance. / Acquired metallo-b-lactamases (MBLs) represent a threat for the treatment of infections caused by Gram-negative bacteria, particularly by enterobacteria that already produce extended-spectrum b-lactamases (ESBL). VIM-1 MBL, which is a carbapenemase that can hydrolyze all classes of β-lactam antibiotics except monobactams, has emerged in Greece and is now commonly found in Enterobacteriaeae. Six carbapenemase-producing strains of Gram-negative bacilli were isolated from a unique patient transferred from Greece to a French hospital. Three of these strains, Providencia stuartii (Ps), Proteus mirabilis (Pm) and Escherichia coli (Ec) strains, were shown to produce the MBL VIM-1 and the ESBL SHV-5. In each of these three strains, the blaVIM-1 gene was carried by a plasmid transferable by in vitro conjugation. The plasmids extracted from the transconjugants displayed a unique restriction profile and harboured identical VIM-1-containing class 1 integrons. Considering the hypothesis that this VIM-1 plasmid had probably been transferred from the Ps strain to the Ec and Pm strains, we performed in vivo conjugation assays in the digestive tract of gnotobiotic mice colonized with E. coli J53, to demonstrate that the VIM-1 plasmid harboured by strain Ps was transferable in vivo, in absence of antibiotic pressure. We determined the complete nucleotide sequence of the VIM-1-encoding plasmid pTC2, which was isolated in a Greek Providencia stuartii multiresistant strain. This 180-kb plasmid was found to be a multireplicon plasmid (IncA/C, IncR), with a large IncA/C backbone and a mosaic multidrug resistance (MDR1) region, in which was inserted a 13-kb IncR fragment. A CD-search-based annotation of the plasmid allowed the identification of a complete IncA/C-type transfer system and of several putative maintenance modules, either on the IncA/C backbone, and on the IncR fragment. The complex MDR1 region contained nine insertion sequences (seven copies of the IS26, one IS1 and one IS6100), 10 resistance genes and a mercury resistance operon integrated either into unit transposons, composite transposons or integrons. The broad-host range, the transfer capacities, the stability, the high plasticity of the MDR1 region combined to the variety of resistance genes make pTC2 a superspreader of resistance determinants.