Thèse De Doctorat

AIX-MARSEILLE UNIVERSITE

FACULTE DE MÉDECINE DE MARSEILLE

ECOLE DOCTORALE DES SCIENCES DE LA VIE ET DE LA SANTE

Thèse de Doctorat

Présentée par

Monsieur Matthieu MILLION

En vue de l’obtention du grade de Docteur d’Aix -Marseille Université Spécialité Maladies Infectieuses

Caractérisation des altérations du microbiote digestif associées à l’obésité et rôle de la manipulation du microbiote digestif dans l’obésité

Soutenue le 15 mai 2013

Composition du Jury :

Mr le Professeur Jean-Louis Mège Président du jury Mr le Professeur Antoine Andremont Rapporteur Mr le Professeur Gilbert Greub Rapporteur Mr le Professeur Didier Raoult Directeur de thèse

Unité de recherche sur les maladies infectieuses et tropicales émergentes, UMR CNRS 7278 Directeur : Pr. Didier Raoult

SOMMAIRE

AVANT PROPOS 1

RESUME 2

SUMMARY 4

INTRODUCTION 6

Partie I: Identification des altérations du microbiote digestif associées à l’obésité 9

Article I : REVIEW - Gut bacterial microbiota and obesity 12

Article II : REVIEW - The relationship between gut microbiota and weight gain in humans 22

Article III : Obesity-associated gut microbiota is enriched in

Lactobacillus reuteri and depleted in Bifidobacterium animalis and Methanobrevibacter smithii 42

Article IV : Correlation between body mass index and gut concentrations of Lactobacillus reuteri , Bifidobacterium animalis ,

Methanobrevibacter smithii and Escherichia coli 52

Partie II : Le rôle de la manipulation du microbiote digestif dans l’obésité 60

Article V : REVIEW - The role of the manipulation of the gut microbiota in obesity 66

Article VI : Comparative meta-analysis of the effect of

Lactobacillus species on weight gain in humans and animals 73 Article VII : Species and strain specificity of Lactobacillus

probiotics effect on weight regulation 83

Article VIII : Publication biases in probiotics 86

Article IX : Lactobacillus rhamnosus bacteremia: an emerging clinical entity 89

Article X : Occam's razor and probiotics activity on

Listeria monocytogenes 102

Article XI : REVIEW - Human gut microbiota: repertoire and variations 104

Article XII : Vancomycin-associated gut microbiota alteration and weight gain in human adults 124

CONCLUSIONS ET PERSPECTIVES 151

REFERENCES 153

ANNEXES 154

Article XIII : Microbial culturomics: Paradigm shift in the human gut microbiome study 155

Article XIV : Non contiguous finished genome sequence and description of Bacillus timonensis sp. nov. 165

Article XV : Rapid and accurate bacterial identification in probiotics and yoghurts by MALDI-TOF mass spectrometry 176

REMERCIEMENTS 182

AVANT-PROPOS

Le format de présentation de cette thèse correspond à une recommandation de la spécialité Maladies Infectieuses et Microbiologie, à l’intérieur du Master des

Sciences de la Vie et de la Santé qui dépend de l’Ecole Doctorale des Sciences de la Vie de Marseille.

Le candidat est amené à respecter des règles qui lui sont imposées et qui comportent un format de thèse utilisé dans le Nord de l’Europe et qui permet un meilleur rangement que les thèses traditionnelles. Par ailleurs, la partie introduction et bibliographie est remplacée par une revue envoyée dans un journal afin de permettre une évaluation extérieure de la qualité de la revue et de permettre à l’étudiant de commence r le plus tôt possible une bibliographie exhaustive sur le domaine de cette thèse Par ailleurs, la thèse est présentée sur article publié, accepté ou soumis associé d’un bref commentaire donnant le sens général du travail. Cette forme de présentation a paru plus en adéquation avec les exigences de la compétition internationale et permet de se concentrer sur des travaux qui bénéficieront d’une diffusion internationale.

Professeur Didier RAOULT

RESUME

L’avènement des méthodes de séquençage moléculaire à l arge échelle a permis l’identification d’altérations du microbiote digestif spécifiquement associés à l’obésité notamment un ratio Bacteroidetes/Firmicutes diminué chez les obèses. Depuis, de nombreux travaux ont décrit de nouvelles altérations associées à l’obésité, notamment une augmentation des représentants du genre Lactobacillus mais l’ensemble de ces résultats sont souvent l’objet de controverses. Afin de clarifier si le genre Lactobacillus était associé à l’obésité, nous avons réalisé deux études cas témoins (la deuxième étant le prolongement de la première avec un effectif de 263 individus) qui nous ont permis d’identifier que les altérations du microbiote digestif sont plus reproductibles au niveau de l’espèce. A ce titre nous avons retrouvé une plu s grande concentration de Lactobacillus reuteri dans le microbiote digestif de sujets obèses alors que les concentrations de Bifidobacterium animalis , Methanobrevibacter smithii et

Escherichia coli étaient diminuées. Nous avons pu établir une relation dose-dépendante entre la concentration de Lactobacillus reuteri et l’indice de masse corporelle. Par ailleurs, nous avons réalisé une méta-analyse sur les résultats des études publiées et avons retrouvé une association entre les genres Bifidobacterium (6 études, 348 individus) et Methanobrevibacter

(3 études, 195 individus) avec l’absence d’obésité.

La manipulation du microbiote digestif par les antibiotiques et les probiotiques, principalement des Lactobacillus , a été utilisée depuis plus de 50 ans dans l’agriculture pour un effet promoteur de croissance. Afin de clarifier l’effet des probiotiques contenant des

Lactobacillus sur le poids, nous avons effectué une méta-analyse incluant 17 essais randomis ées chez l’homme, 51 études chez l’animal et 14 études sur des modèles experimentaux. Lactobacillus acidophilus , Lactobacillus ingluviei et Lactobacillus fermentum

étaient associés à une prise de poids significative chez les animaux. Lactobacillus plantarum

était associé à une perte de poids chez des animaux obèses et Lactobacillus gasseri était

2 associé avec une perte de poids à la fois chez les humains et les animaux en surpoids ou obèse. L’ensemble de ces résultats suggère que l’effet des probiotiques con tenant des

Lactobacillus sur le poids dépend à la fois de l’espèce bactérienne utilisée et de l’hôte. Enfin, l’administration de vancomycine a été associée à une prise de poids chez les animaux et les humains mais les modifications du microbiote digestif responsable de cette prise de poids n’ont pas été élucidées. Dans un travail préliminaire, nous avons retrouvé que l’administration de vancomycine était associé à l’augmentation des Lactobacillus chez l’homme. Lactobacillus reuteri, Lactobacillus fermentum et Lactobacillus sakei naturellement résistant à la vancomycine et identifié comme étant associés à la prise de poids dans d’autres études, pourrait être des candidats supportant la prise de poids sous vancomycine et des vecteurs potentiels de l’obésité.

Mots-clé : Obésité, Microbiote digestif, Lactobacillus, Bifidobacterium , Méta-analyse,

MALDI-TOF

SUMMARY

The revolution of large scale molecular sequencing methods allowed the identification of specific alterations in the gut microbiota associated with obesity such as a decreased

Bacteroidetes / Firmicutes ratio in obese individuals. Since then, many studies have described different alterations associated with obesity, including an increase in members of the

Lactobacillus genus, but results are often controversial. To clarify whether the genus

Lactobacillus was associated with obesity, we conducted two case-control studies (the second being the follow-up of the first study with a total of 263 individuals) allowing us to understand that gut microbiota alterations are more reproducible at the species level. We found a greater concentration of Lactobacillus reuteri in the gut microbiota of obese while concentrations of Bifidobacterium animalis , Methanobrevibacter smithii and Escherichia coli were reduced. We were able to establish a dose-dependent relationship between the concentration of Lactobacillus reuteri and body mass index. In addition, we performed a meta-analysis on the results of published studies and we found an association between the

Bifidobacterium (6 studies, 348 individuals) and Methanobrevibacter (3 studies, 195 individuals) with absence of obesity.

The manipulation of the gut microbiota by antibiotics and probiotics, mainly

Lactobacillus , has been used for over 50 years in agriculture for its growth promoting effect.

To clarify the effect of probiotics containing Lactobacillus on weight, we performed a meta- analysis including 17 randomized trials in humans, 51 farm animal studies and 14 studies on experimental models. Lactobacillus acidophilus , Lactobacillus fermentum and Lactobacillus ingluviei were associated with significant weight gain in animals. Lactobacillus plantarum was associated with weight loss in obese animals and Lactobacillus gasseri was associated with weight loss in both overweight or obese humans and animals. Taken together, these results suggest that the effect of probiotics containing Lactobacillus on weight depends both

4 on the bacterial species and the host. Finally, administration of vancomycin has been associated with weight gain in animals and humans but changes in digestive microbiota responsible for this weight gain have not been elucidated. In a preliminary study, we found that administration of vancomycin was associated with an increase in Lactobacillus in humans. Lactobacillus reuteri, Lactobacillus fermentum and Lactobacillus sakei naturally resistant to vancomycin and identified as being associated with weight gain in other studies, may be candidates for weight gain under vancomycin and are potential vectors of obesity.

Keywords : Obesity, Gut microbiota, Lactobacillus, Bifidobacterium , Meta-analysis,

MALDI-TOF

INTRODUCTION

L'obésité est définie par un indice de masse corporelle (IMC) > 30 kg/m2 et une augmentation de la masse grasse et est associée à une augmentation significative de la morbidité et de la mortalité incluant notamment les maladies cardio-vasculaires, l’arthrose mais aussi certains cancers. La fréquence de l'obésité est en augmentation chez les enfants, les adolescents et les adultes, et a doublé depuis 1980. Selon l'OMS, 65% de la population mondiale vit dans des pays où l'excès de poids et l'obésité tue plus de gens que l'insuffisance pondérale, y compris tous les pays à revenu élevé et la plupart des pays à revenu intermédiaire

(www.who.int).

Le microbiome humain est l’ensemble des communautés microbiennes associées au corps humain dont le nombre d’individus dépasse le nombre des cellules humaines d’au moins un ordre de grandeur. C’est un écosystème complexe qui se compose de virus, bactéries, archées, champignons et parasites. Le plus grand nombre de ces micro-organismes

(1010 à 10 14 bactéries) réside dans le tube digestif distal où ils synthétisent des acides aminés essentiels et des vitamines et assurent le métabolisme de nutriments autrement non digestibles de notre alimentation comme les polysaccharides végétaux. Son rôle dépasse le métabolisme

énergétique et la régulation du stockage des graisses et intervient notamment dans le développement de l’immunité .

Des entérotypes spécifiques ont été identifiés indépendamment de l’ origine ethnique ou géographique. Ils ont été liés à l'alimentation, et leur modulation induite par les antibiotiques peut influer le profil métabolique de l'hôte. Parce que l'intestin est un «point chaud» pour le transfert horizontal de gènes entre un nombre astronomique de bactéries (> 10 9

/ g), d’ archées et de virus, l'analyse au niveau du gène a été jugée la meilleure façon de caractériser les altérations du microbiote intestinal et leurs corrélations avec l'obésité. A

6 l'inverse, d'autres travaux rapportent que l'analyse sur une base taxonomique reste tout à fait pertinente.

Une perturbation spécifique au niveau du phylum avec un ratio Bacteroidetes /

Firmicutes diminué a d'abord été montré comme étant associé à l'obésité, mais la distinction

(c’est -à-dire la classification ou le clustering) entre le microbiote intestinal d’individus maigre et obèses est améliorée lorsque la profondeur de l'analyse taxonomique est augmentée suggérant qu’une analyse au niveau du genre, de l’espèce voire des souches bactériennes serait plus pertinente. A ce titre, un travail antérieur à notre thèse réalisé dans notre laboratoire avait montré une augmentation des représentants du genre Lactobacillus chez les individus obèses.

Enfin, la manipulation du microbiote digestif est possible par l’alimentation, les probiotiques et les antibiotiques. De façon parallèle, les antibiotiques puis les probiotiques, incluant le plus souvent des Lactobacillus , ont été utilisés comme facteur de croissance dans l’agriculture depuis plus de 50 ans. Il est donc tout à fait plausible que cet effet promoteur de croissance dépende de la modification du microbiote digestif. A partir de cette observation, il a été su ggéré que les antibiotiques ou les probiotiques pouvaient être associés à l’obésité.

Dans notre laboratoire, l’administration d’une souche de Lactobacillus ingluviei provenant d’une autruche à des poulets a, de façon tout à fait inattendue, provoqué une pr ise de poids massive et cela a été reproductible sur d’autres modèles animaux comme des souris.

Cependant le lien entre l’administration de probiotiques et la modification du poids chez les animaux et les hommes nécessitait d’être clarifié. Par ailleurs, u ne étude a montré que des patients sous vancomycine, antibiotique efficace sur les Firmicutes mais inefficace sur plusieurs espèces de Lactobacillus , voyaient leur poids augmenter et c’est pourquoi un rôle des Lactobacillus dans cette prise de poids avait alors été suspecté.

Après avoir effectué un travail préliminaire sur la méthode de « microbial culturomics » (Annexes, Article XIII à XV), l ’objectif de notre travail a été d’utiliser des techniques de culture, de spectrométrie de masse et de biologie moléculaire innovantes pour : i) Identifier les altérations du microbiote digestif associées à l’obésité (articles I à IV) et ii)

Clarifier le rôle de la manipulation du microbiote digestif sur le poids (Articles V à XII).

Partie I :

Identification des altérations du microbiote digestif

associées à l’obésité

Avant-propos

Le premier travail rapportant une altération du microbiote digestif associé à l’obésité chez l’homme a été publié en 2006 par Ley et al. 1 et a montré que les individus obèses avaient une diminution du pourcentage de séquences correspondant au phylum des

Bacteroidetes par rapport aux sujets de poids normal et que les régimes pauvres en carbohydrate ou pauvre en matière grasse entrainaient une augmentation du pourcentage de

Bacteroidetes et cette augmentation était d’autant plus importante que l’ individu perdait du poids.

Partant de cette étude, nous avons réalisé une méta-analyse sur les données publiées

(Article II) afin de clarifier quelles étaient les altérations reproductibles au niveau du phylum, du genre ou de l’espèce. Cette diminution de la proportion des Bacteroidetes n’a pas été constatée en méta-analyse alors même que nous avons retrouvé la même tendance dans deux

études observationnelles que nous avons réalisé (Article III et IV). Par contre, les altérations reproductibles et associées à des résultats concordants et significatifs en méta-analyse étaient une diminution des Bifidobacterium dans 6 études différentes réalisées dans 4 pays différents

à savoir l’Allemagne, la Finlande, l’Espagne et la Chine et une diminution des

Methanobrevibacter sp. à partir de 3 études réalisées dans 2 pays différents à savoir la France et l’Allemagne (Article II).

Par ailleurs, une précédente étude du laboratoire a rapporté une augmentation de la concentration des Lactobacillus chez les obèses 2, c’est pourquoi nous avons étudié par culture les représentants de ce genre dans le microbiote digestif de patients obèses et d’individus contrôles. Cependant, nos résultats ont rapidement montré que les Lactobacillus n’étaient ni plus prévalent ni plus abondant chez les individus obèses (culture sur milieu

LAMVAB). Afin de comprendre cette discordance, nous avons pu montrer que le système

10 d’amorces et de sonde utilisé dans l’étude précédente du laboratoire était significativement plus sensible pour certaines espèces de Lactobacillus . A partir de là, nous avons conçu plusieurs systèmes de real-time PCR spécifiques de 8 espèces de Lactobacillus identifiée en culture ou dans la littérature comme associés à l’obésité ou à un poids normal plus un système spécifique pour Lactococcus lactis et Bifidobacterium animalis . Lactobacillus reuteri (PCR) a

été associé à l’obésité alors que Lactobacillus plantarum (culture) ou Lactobacillus paracasei

(culture) ont été associés à un poids normal (Article III). De plus nous avons retrouvé de façon tout à fait inattendue une association extrêmement significative entre Bifidobacterium animalis et l’absence d’obésité. Nous avons ensuite trouvé que cela était cohérent avec notre travail de méta-analyse qui retrouvait une association entre le genre Bifidobacterium et l'absence d’obésité. Nous avons aussi confirmé que Methanobrevibacter smithii , préal ablement associé au microbiote d’individus anorexiques dans une étude de notre laboratoire 2, était associé au microbiote d’individu de poids normal par rapport à des individus obèses (Article III).

Enfin, nous avons prolongé cette étude en doublant la ta ille de l’échantillon et en

incluant des individus en surpoids et des individus anorexiques (Article IV). Dans cette

nouvelle étude, nous avons inclus Escherichia coli , associée par d’autres équipes à l’obésité.

Cela nous a permis d’identifier un effet dos e-dépendant avec un coefficient de régression

linéaire positif pour Lactobacillus reuteri (présent en concentration d’autant plus abondante

que l’indice de masse corporelle était élevé), alors que ce coefficient était négatif pour

Bifidobacterium animalis , Methanobrevibacter smithii et Escherichia coli (présents en

concentration d’autant plus faible que l’indice de masse corporelle était élevé).

Article I : REVIEW

Gut bacterial microbiota and obesity

Matthieu Million, Jean-Christophe Lagier, Dafna Yahav, Mical Paul

Published in Clinical Microbiology and Infection. Article first published online: 2 MAR 2013. DOI: 10.1111/1469-0691.12172. (IF 4.54)

REVIEW 10.1111/1469-0691.12172

Gut bacterial microbiota and obesity

M. Million 1, J.-C. Lagier 1, D. Yahav 2 and M. Paul 2 1) Unite de Recherche sur les Maladies Infectieuses et Tropicales Emergentes, Facult e de M edecine, CNRS UMR 7278, IRD 198, Aix-Marseille Universite, Marseille, France and 2) Unit of Infectious Diseases, Rabin Medical Centre, Beilinson Hospital and Sackler Faculty of Medicine, Tel-Aviv University, Tel-Aviv, Israel

Abstract

Although probiotics and antibiotics have been used for decades as growth promoters in animals, attention has only recently been drawn to the association between the gut microbiota composition, its manipulation, and obesity. Studies in mice have associated the phylum Firmicutes with obesity and the phylum Bacteroidetes with weight loss. Proposed mechanisms linking the microbiota to fat content and weight include differential effects of bacteria on the efﬁciency of energy extraction from the diet, and changes in host metabolism of absorbed calories. The independent effect of the microbiota on fat accumulation has been demonstrated in mice, where transplantation of microbiota from obese mice or mice fed western diets to lean or germ-free mice produced fat accumulation among recipients. The microbiota can be manipulated by prebiotics, probiotics, and antibiotics. Probiotics affect the microbiota directly by modulating its bacterial content, and indirectly through bacteriocins produced by the probiotic bacteria. Interestingly, certain probiotics are associated with weight gain both in animals and in humans. The effects are dependent on the probiotic strain, the host, and speciﬁc host characteristics, such as age and baseline nutritional status. Attention has recently been drawn to the association between antibiotic use and weight gain in children and adults. We herein review the studies describing the associations between the microbiota composition, its manipulation, and obesity.

Keywords: Fat, growth promoters, microbiota, obesity, probiotics

Clin Microbiol Infect

Corresponding author: M. Paul, Unit of Infectious Diseases, Rambam Healthcare Campus, Haifa 31096, Israel E-mail: [email protected]

Introduction Microbial changes in the human gut were proposed as a possible cause of obesity [5,9,10]. Certain phyla and classes of bacteria are associated with improved transfer of calories from Ten trillion to 100 trillion (10 14 ) microorganisms populate the the diet to the host, and with changes in the host metabolism of adult intestines [1,2]. The vast majority reside in the colon, absorbed calories [11]. Gut microorganisms ferment dietary where densities approach 10 11 –10 12 cells/mL. Almost all of polysaccharides into monosaccharides and short-chain fatty acids, these organisms are bacteria, and a minority are archaeons, and thus allow the extraction of calories from indigestible dietary eukaryotes, and viruses [3,4]. Bacteria are classiﬁed from the polysaccharides. One of the ways in which they affect host phylum to species level (Table 1). The two most abundant metabolism is by suppressing fasting-induced adipocyte factor, bacterial phyla in humans and in mice are the Firmicutes (60– which is a lipoprotein lipase inhibitor, and the suppression of 80%) and the Bacteroidetes (20–40%) [1,3,5]. Most of the which contributes to the deposition of triglycerides in adipocytes. representatives of these two phyla do not grow outside of their host [1]. Babies acquire their initial microbiota from the The Association between Microbiota surrounding ecosystems, especially the maternal vaginal and Composition and Obesity faecal microﬂora [2,6], and the human gut microbiome is shared among family members [7,8]. The gut microbiota composition depends on age, sex, geography, ethnicity, family, and diet, and Studies in mice have found a higher abundance of Firmicutes in can be modulated by prebiotics, probiotics, and antibiotics. obese mice and those fed on western diets, concomitant with

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases 2 Clinical Microbiology and Infection CMI

TABLE 1. Examples of the classiﬁcation of several common higher in obese than in non-obese children [23]. The gut bacteria proportions of the Bacteroides–Prevotella group were shown

Domain Bacteria Bacteria Bacteria Bacteria to increase after weight loss in obese adolescents [24]. The latter study also revealed a correlation between reductions in Phylum Firmicutes Firmicutes Bacteroidetes Actinobacteria Class Clostridia Bacilli Bacteroidetes Actinobacteria Clostridium histolyticum and Eubacterium rectale–Clostridium coc- Order Clostridiales Lactobacillales Bacteroidiales Bifidobacteriales Family Clostridiaceae Lactobacillaceae Bacteroidiaceae Bifidobacteriaceae coides (Firmicutes) proportions and weight loss. Genus Clostridium Lactobacillus Bacteroides Bifidobacterium Turnbaugh et al. [25] demonstrated more environmental gene tags of archaeons in the caecal microbiome of obese mice than in that of lean mice. Archaeons are methanogenic a decrease in the abundance of Bacteroidetes [1,11]. Within the organisms that increase the efficiency of bacterial fermenta- phylum Firmicutes, the class Mollicutes was the most common in tion. The principal methanogenic archaeon in the human gut is obese mice [11]. Bacteroidetes possess fewer genes for Methanobrevibacter smithii. Studies in mice colonized with this enzymes involved in lipid and carbohydrate metabolism than organism and/or B. thetaiotaomicron revealed that co-coloniza- Firmicutes [12]. However, within the phylum Bacteroidetes, tion increases the efficiency of polysaccharide fermentation, Bacteroides thetaiotaomicron was found to improve host nutri- leading to an increase in adiposity as compared with mice ent absorption and processing [13]. colonized with either organism alone [25,26]. Zhang et al. [27] Studies in humans found various Firmicutes/Bacteroidetes found that Methanobacteriales were present only in obese ratios in obese individuals. Some supported the finding of a individuals, after studying three obese individuals and three high Firmicutes/Bacteroidetes ratio [5,14–16], some did not find human controls. Several other studies in humans have a correlation between body mass index and the Firmicutes/ demonstrated lower levels of Methanobrevibacter in overweight Bacteroidetes ratio [3,17], and still others found an opposite and obese human volunteers [14,15,18]. ratio [18,19]. Turnbaugh et al. [7] described a lower proportion of Bacteroidetes and a higher proportion of Actinobacteria in Microbiota Transplantation Studies obese than in lean individuals, with no significant difference in the proportion of Firmicutes. A significantly higher level of Lactobacillus species (from the phylum Firmicutes) was found in The independent contribution of the microbiota to fat obese patients than in lean controls [14]. Specifically, a higher accumulation has been demonstrated in a series of elegant level of Lactobacillus reuteri and lower levels of Lactobacillus in vivo studies in mice. Germ-free mice, lacking a microbiota, casei/paracasei and Lactobacillus plantarum were associated with have significantly less body fat than normal mice, despite eating obesity [15]. Reduced proportions of butyrate-producing more [10]. Transfer of the microbiota from normal to germ- Firmicutes were described in obese subjects on weight loss free mice caused a significant increase in body fat content. diets [20] and their presence was lower in obese subjects as Transplanting germ-free mice with the microbiota from obese compared with their blood-related lean family members [8]. mice led to a significantly increased fat content as compared Another bacterial genus that has been implicated in obesity with transplantation of the microbiota from lean mice, and this is Bifidobacterium (belonging to the phylum Actinobacteria). was associated with a greater relative abundance of Firmicutes Several studies in humans found an association between lower in the guts of both the obese donors and their recipients [25]. levels of bifidobacteria and obesity [15,16,18,19,21]. Bifido- This was observed in controlled conditions, where both bacteria were found at higher levels in the intestinal microbiota groups had the same baseline weight and received the same of breast-fed infants than in that of formula-fed infants [12]. amount of feeding. Transplanting germ-free mice with the The association between bifidobacteria and obesity is probably microbiota from mice raised on a western diet led to also species-specific [22]. significantly increased body fat as compared with mice At the species level, several studies have investigated the transplanted with the microbiota from donors who had been association between specific bacterial species and obesity in fed a lean low-fat diet, rich in structurally complex plant humans. An association between Staphylococcus aureus and an polysaccharides [11]. Both carbohydrate restriction and fat overweight state was demonstrated in children and pregnant restriction from the western diet prevented the increased women [19,21]. Reduced numbers of Bacteroides and increased accumulation of fat in recipients. This was accompanied by a numbers of Staphylococcus, Enterobacteriaceae and Escherichia decrease in the presence of the Mollicutes lineage (phylum coli have been described in overweight as compared with Firmicutes) and an increase in the relative abundance of normal-weight pregnant women [16]. Levels of Faecalibacteri- Bacteroidetes. Turnbaugh et al. [28] successfully colonized mice um prausnitzii (of the phylum Firmicutes) were significantly with human faeces, and fed them with the western diet vs. the

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI CMI Million et al. Microbiota and obesity 3

low-fat/plant polysaccharide-rich diet for 2 weeks, and then be responsible for Proteobacteria abundance (order Desulfovib- transplanted their microbota into germ-free mice. Germ-free rio), but authors comparing the gut flora of malnourished mice receiving the microbiota from the obese western-diet-fed children with that of well-nourished children in Bangladesh humanized mice gained significantly more adiposity than the found a Bacteroidetes decrease and a Proteobacteria increase, mice receiving the microbiota from the low-fat-fed humanized notably for Escherichia coli and Klebsiella spp. [35]. mice. Again, this was achieved despite matching of the two groups of recipients by age, weight, and body fat, similar The Influence of Prebiotics on the Microbiota feeding of recipients with low-fat/plant polysaccharide-rich diets, and similar consumption of food. Different strains of Bifidobacterium (phylum Actinobacteria) from human volunteers’ Prebiotics are defined as food ingredients that stimulate the fresh faeces given to mice produced different effects on body growth of a limited number of microbial genus/species in the weight [22]. In all studies, the time to fat changes following gut microbiota that are hypothesized to confer health benefits microbiota manipulation was up to 2 weeks. to the host. The administration of oligofructose to high-fat-fed These studies prove that the microbiota by itself can cause mice increased the abundance of Bifidobacterium and normal- weight gain. The microbiota derived from genetically obese mice ized endotoxaemia and the inflammatory tone associated with or mice rendered obese by diet can cause fat accumulation, and the high-fat diet [36]. The administration of oligofructose to this is not mediated by increased food consumption. The studies genetically obese mice induced increases in the levels of also showed that this is mediated both by improved efficiency of Lactobacillus, Bifidobacterium, and C. coccoides –E. rectale , which transfer of calories from the diet to the host and through effects led to a reduction in intestinal permeability and an improve- on host metabolism of the absorbed calories [11]. ment in tight junction integrity and inflammatory markers, such as lipopolysaccharides and cytokines [37]. Association between Diet and the Microbiota The Influence of Probiotics on the Microbiota and Obesity Dietary habits constitute a major factor influencing the diversity of the human gut microbiota [29]. A vegetarian diet is known to affect the intestinal microbiota by decreasing the Probiotics are live bacteria that are thought to be beneficial to the amount and modifying the diversity of Clostridium cluster IV host. The first food containing probiotics ingested by humans is and Clostridium rRNA clusters XIVa and XVIII [30,31]. Walker breast milk. Two studies showed that Bifidobacterium , Lactobacil- et al. [32] successively tested obese individuals with a control lus and Enterococcus strains showed identical random amplifica- diet, a diet high in resistant starch or non-starch polysaccha- tion of polymorphic DNA profiles in breast milk samples and rides, and a reduced-carbohydrate weight loss diet. There was faeces of newborns at different sampling times, suggesting vertical no significant effect of diet on the proportions of the four main transfer of these bacteria from the mother’s milk to the infant phyla represented in the gut microbiota. E. rectale and [38,39]. The influence of probiotics on the intestinal flora is highly Ruminococcus bromii showed dramatically increased propor- dependent on their adhesion to colonocytes, resistance to acidic tions in individuals receiving the resistant starch diet, whereas pH, and bile salt tolerance [40]. In an interventional study, the proportion of Collinsella aerofaciens-related sequences was L. reuteri DSM 12246 was associated with excellent adhesion. In decreased significantly in those receiving the weight loss diet. contrast, Lactobacillus delbrueckii ssp. lactis CHCC2329 did not Gut analysis of African children from Burkina Faso showed survive at pH 2.5, and was found in only a few of those to whom it specific abundance of Prevotella, Xylanibacter and Treponema was administered. It is plausible that probiotics modulate and containing bacterial genes for cellulose and xylan hydrolysis, shape the digestive microbiota according to the antibiotic which are completely absent in European children, and are spectrum of their bacteriocins, which are antibiotic-like sub- probably linked to high intake of fibre, allowing increased stances produced by bacteria. extraction of metabolic energy from the polysaccharides of To date, the metagenomic data available from human ingested plants [33]. Wu et al. [34] found that enterotypes intervention studies with probiotics are very limited [41]. were strongly associated with long-term diets. After short- Recently, gut analysis was performed by flow cytometry and term diet modification, the gut microbiota alteration occurred fluorescence in situ hybridization in newborns’ faecal samples rapidly and was quickly reversible [32,34]. Conversely, persis- after administration of probiotics vs. placebo to mothers for tent modifications of individual enterotypes occurred during 2 months before delivery up to 2 months after delivery long-term dietary interventions [34]. Finally, a high-fat diet can (during breast-feeding) in a Finish cohort, or to newborns

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI 4 Clinical Microbiology and Infection CMI

receiving formula feeding from 1 month of age up to 4 months ATCC 4962 and ATCC 4963 (formerly named L. acidophilus ) of age in a German cohort of newborns [42]. The probiotics with weight gain in bottle-fed infants, in contrast to the results used included Lactobacillus rhamnosus LPR, Lactobacillus paraca- observed in overweight/obese people. In a randomized con- sei ST11 (only in the Finish cohort), and Bifidobacterium longum trolled trial, individuals randomized to receive fermented milk BL999. The combination of L. rhamnosus LPR and Bifidobacte- containing L. gasseri showed reductions in abdominal adiposity, rium longum BL999 had the effect of raising the level of body weight, and other measures [45]. Whether the same Lactobacillus–Enterococcus and lowering the level of Bifidobacte- Lactobacillus strains could have a growth-promoting effect in rium in the gut microbiota of the Finnish cohort (whose undernourished individuals and an anti-obesity effect in obese mothers were treated), whereas there was no such effect in individuals needs to be clarified. Overall, both the probiotic the German cohort (where infants were given probiotics). The bacterial strain and the host are important determinants of the authors concluded that probiotic treatment had different effects of probiotics on obesity, and it is possible that certain impacts on the gut microbiota composition in Finnish and marketed probiotics favour obesity. German infants, owing to differences in mode of feeding and the early commensal microbiota. The specificity of the The Influence of Antibiotics on the Microbiota Lactobacillus species or strain for its effect on the gut microbiota was demonstrated in this study. Probiotics, which had also been used for decades in Oral and intravenous antibiotics have been reported to agriculture for their growth-promoting effects, have undergone decrease the bacterial load in the digestive tract [46,47], a revival since the ban on antibiotics as growth promoters in although other studies have found that only the microbiota Europe from 1 January 2006 [43]. Probiotics are also frequently composition is changed [48]. For example, metronidazole, used by people for their proposed health benefits. We cefoperazone, or vancomycin, in contrast to amoxycillin, led to performed a meta-analysis on the effects of probiotics on alterations in community structure without a significant weight in humans and animals [44]. The bacteria most decrease in the overall bacterial biomass [49]. Studies commonly used belonged to the genera Lactobacillus, Bifidobac- summarizing the effects of antibiotics on the gut microbiota terium, Enterococcus, and Streptococcus. Lactobacillus acidophilus, in animals are summarized in Table 2. Antibiotic effects can be Lactobacillus fermentum and Lactobacillus ingluviei were associ- species-specific. Lactobacillus appears to be particularly ated with a weight gain effect in lean individuals, whereas impacted by growth-promoting antibiotics in animal studies L. plantarum and Lactobacillus gasseri strains had an anti-obesity [50]. In contrast, the levels of some bacterial genera seem to effect in overweight/obese people. It is most likely that this be reduced by the growth-promoting antibiotics, particularly effect was dependent on the strain used and the metabolic Proteobacteria (e.g. Salmonella) [51]. Finally, whereas rapid phenotype of the host, as we found one study linking L. gasseri recovery has been described after short-term antibiotic

TABLE 2. In vivo studies examining the effects of antibiotics on microbiota

Reference Host Antibiotics Microbiota changes with antibiotics Other effects

Cho, 2012 [50] Mice Subtherapeutic antibiotic Proportion of Firmicutes higher vs. controls Subtherapeutic antibiotic treatment treatment with vancomycin, Lachnospiraceae family increased altered the gene counts of genes penicillin, and chlortetracyclines involvedin the metabolism of carbohydrates to short-chain fatty acids. Increases in caecal acetate, butyrate and propionate have been observed with STAT Robinson, 2010 [49] Mice Vancomycin Increases in the phyla Proteobacteria and Tenericutes and the family Lactobacillaceae Decrease in the family Lachnospiraceae Membrez, 2008 [68] Mice (obese) Norﬂoxacin and ampicillin Decrease in aerobic and anaerobic bacteria Improved glucose tolerance Looft, 2010 [69] Pigs Chlortetracycline–sulphamethazine Increase in Proteobacteria, Escherichia coli and penicillin Kim, 2010 [70] Pigs Tylosin Lactobacillus and Sporacetigenum increased Collier, 2003 [71] Pigs Tylosin Lactobacillus increased Rettedal, 2009 [72] Pigs Chlortetracycline Lactobacillus amylovorus increased Lactobacillus johnsonii decreased Torok, 2011 [73] Chicken Avilamycin Lactobacillus crispatus, Lactobacillus reuteri, Subdoligranulum and Enterobacteriaceae increased Lachnospiraceae, Ruminococcaceae, Oxalobacteraceae and L. johnsonii decreased Torok, 2011 [74] Chicken Avilamycin L. crispatus, Lactobacillus salivarius , Improved feed conversion ratio as Lactobacillus aviarus, Escherichia coli, measured by weight gain/amount of feed Bacteroides vulgatus or consumed Faecalibacterium prausnitzii increased Dumonceaux, 2006 [51] Chicken Virginiamycin Aerobic bacteria and Lactobacillus, especially L. crispatus , increased Guban, 2006 [75] Chicken Bacitracin L. salivarius decreased

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI TABLE 3. Studies assessing associations between antibiotics and weight change in humans a CMI

Antibiotics, duration, oral or IV, dosage, Population (no., age, Design, year of antibiotic Reference frequency special population) Indication administration, country Effect on weight

Trasande, 2012 [65] Unspecified 11 532 children, <2 years (Avon Infections Longitudinal birth cohort study, WG significant at 10, 20 and Longitudinal Study of Parents and in early life 1991 –1992, UK 38 months when antibiotics were Children 1991 –1992) administered before 6 months of age Thuny, 2010 [76] Penicillin, vancomycin, and other 96 (48 treated), 45 –75 years, Endocarditis Retrospective analysis in WG significant for all treated antibiotics, IV, unspecified, patients with heart valve disease consecutive adults, 2002 –2007, patients and for subgroups treated unspecified France with vancomycin WG not significant for patients treated with amoxycillin Pirzada, 2003 [77] Azithromycin, oral, 250 mg daily for 40 (20 treated), 18 years, CFPA CF with Comparative open-label trial, 1997 – WG significant for patients treated 21 months progressive 1999, UK with azithromycin (mean duration pulmonary of 0.9 months) disease Saiman, 2003 [78] Azithromycin, 168 days, oral, 185 (87 treated), >6 years, CFPA CF Multicentre randomized double- WG significant for patients treated 250 mg if <40 kg, 500 mg if with FEV 1 > 30% blind placebo-controlled trial, with azithromycin >40 kg, 3 days a week 2000 –2002, USA Saiman, 2010 [79] Azithromycin, 168 days, oral, 260 (131 treated), 6 –18 years, CF CF Multicentre randomized double- WG significant for patients treated <

lnclMcoilg n Infection and Microbiology Clinical 250 mg if 36 kg, 500 mg if uninfected with Pseudomonas blind placebo-controlled trial, with azithromycin >36 kg, 3 days a week aeruginosa 2007 –2009, USA –Canada Saiman, 2012 [80] Azithromycin, day 168 to day 336 146 (77 treated), 6 –18 years, CF CF Multicentre open-label follow-on In the placebo –azithromycin group, (follow-on study of Saiman [79]), uninfected with P. aeruginosa study, 2007 –2009, USA –Canada the rate of WG trended towards 250 mg if <36 kg, 500 mg if improvement during the open-label >36 kg, 3 days a week study as compared with the rate observed during the placebo- controlled trial. In the azithromycin –azithromycin group, the rate of WG remained constant during the open-label study as compared with the placebo- controlled trial Clement, 2006 [81] Azithromycin, 12 months, oral, 82 (40 treated), 11 Æ 3 years, CF CF Multicentre randomized double- WG non-significant (BMI z-score ª 250 mg if <40 kg, 500 mg if infected or not with P. aeruginosa blind placebo-controlled trial, treatment effect: 0.15 03Erpa oit fCiia irbooyadIfciu Diseases, Infectious and Microbiology Clinical of Society European 2013 40 kg 2001 –2003, France (95% CI À0.03 to 0.34) Mansi, 2012 [82] Erythromycin, 10 days or until full 60 (30 treated), preterm infants Feeding Open prospective randomized WG significant in infants <32 weeks (abstract only) enteral feeding, oral, 50 mg/kg daily predominantly fed with milk intolerance controlled trial, unknown, Egypt No effect after 32 weeks formula Ng, 2012 [83] Erythromycin, 14 days, oral, 5 mg/ 45 (19 treated), very low Feeding Open randomized trial, 2007 –2009, WG significant

kg every 6 h birthweight infants ( <32 weeks, intolerance Taiwan Million <1500 g) Lane, 2011 [84] Clarithromycin, 2 weeks, oral, 1558 (787 treated), 20 –59 years, H. pylori Open randomized placebo- WG significant 500 mg twice daily (with ranitidine Helicobacter pylori -infected eradication controlled trial, 1996 –1999, UK bismuth citrate 400 mg twice daily) individuals unselected for dyspepsia Kamada, 2005 [85] Clarithromycin 400 mg twice daily, 150 (50 treated), 23 –72 years, H. pylori Open case –control study matched WG significant al. et amoxycillin 750 mg twice daily, H. pylori -positive population, only eradication in for age and sex (1 : 2 ratio), 2000, omeprazole 20 mg twice daily patients with peptic ulcers were patients with Japan Oral, 7 days treated peptic ulcer Patterson, 1977 [86] Minocyclin, period of 3 months with 100 patients, unknown, CF of the CF Unknown WG only when minocycline was the obesity and Microbiota (abstract only) a broad-spectrum antibiotic pancreas drug used, weight loss when it was rotation lasting for 2 years not the drug used. No significance results available Robinson, 1952 [62] Chlortetracycline, unknown, 50 mg/ 28 (13 treated), preterm infants (all Neonatology Antibiotic administered to the WG significant kg daily, 12 –32 days <2500 g) weaker one of twins or the weakest of triplets, unknown, Israel

ª Haight, 1955 [59] Chlortetracycline 250 mg daily and 310 (102 treated with Effect of Double-blind placebo-controlled WG signiﬁcant (chlortetracycline 03TeAuthors The 2013 procain penicillin, 100 000 units chlortetracycline, 105 treated with antibiotic trial, unknown, USA group vs. placebo group and daily, oral, 7 weeks penicillin, 103 treated with prophylaxis penicillin group vs. placebo group) placebo)18 years, healthy US navy on immune recruits response Guzman, 1958 [87] Growth CMI 5 6 Clinical Microbiology and Infection CMI

therapy [49], persistent effects have been described with some antibiotics, such as quinolones and cefoperazone [49,52], and recovery may sometimes be incomplete [53].

The inﬂuence of antibiotics on obesity

Moore et al. [54] first discovered serendipitously in 1946 that WG significant for chlortetracycline Weight loss for penicillin WG significant WG non-significant No effect WG non-significant sulphonamide administration was associated with a two-fold increase in weight in chicks fed adequate amounts of folic acid, and he noted that the total gut bacterial count increased, with

– coliform counts decreasing and lactobacilli increasing. Stokstad et al. found that both Streptomyces aureofaciens and its bacteriocin, aureomycin, were associated weight gain, leading to the 1953 – proof-of-concept of the growth-promoting effect of both probiotics and antibiotics [55,56]. Antibiotics, including mainly 1955, Guatemala

– tetracycline, glycopeptide, macrolides, and penicillin, have been controlled trial in two1953 villages, double-blind placebo-controlled trial, 1998, Guinea-Bissau Vietnam trial, Italy, 1951 1952, USA used for over 60 years to promote weight gain in animals [54], Design, year of antibiotic administration, country Effect on weight Community-based randomized Prospective cohort study, 1993, Open randomized placebo- with an optimal efficiency in pigs [57], and continue to be widely used in the USA [58]. From the beginning of their use in agriculture in the 1950s, a similar growth-promoting effect was reported in humans [59 –62], but this effect seems to have been overlooked until recently [63–65]. after measles typhoid fever Prophylaxsis Malnutrition Unknown, Africa, unknown WG significant Malnutrition Community-based randomized trial WG significant Prematurity Open alternate admission, 1951 Suspected Trasande et al. [65] found that exposure to antibiotics during the first 6 months of life is associated with consistent – , forced expiratory volume in 1 s; IV, intravenous; WG, weight gain. increases in body mass. Exposures later in infancy (6 1 14 months and 15 –23 months) was not consistently associ- ; FEV – ated with increased body mass. The authors concluded that, 10 years Poor diet Double-blind placebo-controlled

12 years, rural < – although the effects of early exposure ( 6 months) are modest – at the individual level, they could have substantial conse- 25 years 14 years (173 treated – – quences for population health. Many antibiotics have been

Pseudomonas aeruginosa associated with weight gain in children and adults (Table 3). 2500 g measles, 0 treated with chlortetracycline, 1 04 not treated), 6 Guatemalan schoolchildren undernourished children 36 months < with ciprofloxacin, 153 treatedwith ofloxacin, 223 healthy untreated age-matched controls) Shortly after the first animal studies, aureomycin (a tetracy- 84 (46 treated), patients with 260 (64 treated with penicillin, 92 Population (no., age, special population) Indication 81, malnourished children, 3 113 (57 treated), premature infants Children 1 338 (181 treated), 6 cline) was shown to induce weight gain in preterm infants after 10 days [61]. Similar effects of tetracycline were reported in premature infants, undernourished or rural children, young recruits of the US navy, and patients with cystic fibrosis and 5 days, oral

– pancreatic disease. Macrolides, which are widely used in the

5 years or 18 kg animal industry, have been linked with human weight gain, < especially azithromycin in children with cystic fibrosis in double-blind randomized placebo-controlled studies (Table 2). A recent meta-analysis confirmed a significant weight gain received paediatric tablets, oral, 7 days penicillin 50 mg daily, oral, 2 years unknown, 5 days unknown or IV unknown ofloxacin, 50 mg/kg, 3 7 months Chlortetracycline 50 mg daily, Antibiotics, duration, oral or IV, dosage, frequency Metronidazole, 20 mg/kg daily, Chlortetracycline, 37 days 72 (38 treated), 2 years, African effect of macrolides in patients with cystic fibrosis [66]. Clarithromycin was associated with weight gain when used for

) the eradication of Helicobacter pylori, with a link to acquired obesity. Sulphonamides and co-trimoxazole have been linked to weight gain when used as prophylaxis to prevent pneumonia Continued ( and other complications after measles in a community-based randomized double-blind placebo-controlled trial in Guinea- (abstract only) (abstract only) Studies were searched through PubMed and Google Scholar, unrestricted by language or date, with the following keywords: antibiotics, humans, weight, weight gain, weight loss. Exclusion criteria were comparison of antibiotics witho ut a Reference Garly, 2006 [67] Co-trimoxazole, control group, animal studies, and studies concerning tuberculosis, because it is known to be associated with important weight loss. a Heikens, 1993 [88] Coodin, 1953 [92] Terramycin, oral, 25 mg daily, Bethell, 1996 [89] Ciprofloxacin, 70 mg/kg, 7 days or McDougall, 1957 [91] Corbo, 1955 [90] Chlortetracycline, oral, 20 mg daily, BMI, body mass index; CF, cystic fibrosis; CFPA, cystic fibrosis patients colonized by

Table 3 Bissau [67]. It is difﬁcult to conclude from these studies

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI CMI Million et al. Microbiota and obesity 7

2. Ley RE, Peterson DA, Gordon JI. Ecological and evolutionary forces shaping microbial diversity in the human intestine. Cell 2006; 124: 837– 848. 3. Arumugam M, Raes J, Pelletier E et al. Enterotypes of the human gut microbiome. Nature 2011; 473: 174–180. 4. Qin J, Li R, Raes J et al. A human gut microbial gene catalogue established by metagenomic sequencing. Nature 2010; 464: 59 –65. 5. Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: human gut microbes associated with obesity. Nature 2006; 444: 1022–1023. 6. Reinhardt C, Reigstad CS, Backhed F. Intestinal microbiota during infancy and its implications for obesity. J Pediatr Gastroenterol Nutr 2009; 48: 249–256. 7. Turnbaugh PJ, Hamady M, Yatsunenko T et al. A core gut microbiome in obese and lean twins. Nature 2009; 457: 480–484. 8. Elli M, Colombo O, Tagliabue A. A common core microbiota between obese individuals and their lean relatives? Evaluation of the predispo- sition to obesity on the basis of the fecal microflora profile. Med Hypotheses 2010; 75: 350–352. 9. Angelakis E, Armougom F, Million M, Raoult D. The relationship between gut microbiota and weight gain in humans. Future Microbiol 2012; 7: 91 –109. 10. Backhed F, Ding H, Wang T et al. The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci USA 2004; 101: 15718–15723. 11. Turnbaugh PJ, Backhed F, Fulton L, Gordon JI. Diet-induced obesity is linked to marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 2008; 3: 213 –223. 12. Kallus SJ, Brandt LJ. The intestinal microbiota and obesity. J Clin Gastroenterol 2012; 46: 16 –24. 13. Hooper LV, Wong MH, Thelin A, Hansson L, Falk PG, Gordon JI. FIG. 1. Factors affecting gut microbiota and obesity. Molecular analysis of commensal host –microbial relationships in the intestine. Science 2001; 291: 881–884. whether antibiotics are associated with weight gain through 14. Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. Monitoring their beneficial effects in preventing or treating bacterial bacterial community of human gut microbiota reveals an increase in infections, or through their effects on the microbiota. It is Lactobacillus in obese patients and methanogens in anorexic patients. PLoS ONE 2009; 4: e7125. plausible that a mixture of these two mechanisms is present in 15. Million M, Maraninchi M, Henry M et al. Obesity-associated gut micro- different scenarios. biota is enriched in Lactobacillus reuteri and depleted in Bifidobacterium – In summary, intriguing data link the microbiota composition animalis and Methanobrevibacter smithii . Int J Obes 2012; 36: 817 825. 16. Santacruz A, Collado MC, Garcia-Valdes L et al. Gut microbiota to metabolism, fat accumulation and obesity in animals and composition is associated with body weight, weight gain and biochem- people. Strangely, this was exploited long before the recog- ical parameters in pregnant women. Br J Nutr 2010; 104: 83 –92. nition of the mechanism, through the use of probiotics and 17. Mai V, McCrary QM, Sinha R, Glei M. Associations between dietary habits and body mass index with gut microbiota composition and fecal antibiotics as growth promoters in animals. Figure 1 summa- water genotoxicity: an observational study in African American and rizes the known data and mechanisms. More precise delinea- Caucasian American volunteers. Nutr J 2009; 8: 49. tion of the mechanism might lead to tailored interventions or 18. Schwiertz A, Taras D, Schafer K et al. Microbiota and SCFA in lean and – preventive measures to combat one of the worst enemies of overweight healthy subjects. Obesity 2010; 18: 190 195. 19. Collado MC, Isolauri E, Laitinen K, Salminen S. Distinct composition of humanity in the current millennium, obesity. gut microbiota during pregnancy in overweight and normal-weight women. Am J Clin Nutr 2008; 88: 894–899. 20. Duncan SH, Lobley GE, Holtrop G et al. Human colonic microbiota Transparency Declaration associated with diet, obesity and weight loss. Int J Obes 2008; 32: 1720– 1724. 21. Kalliomaki M, Collado MC, Salminen S, Isolauri E. Early differences in All authors declare no conflicts of interest. fecal microbiota composition in children may predict overweight. Am J Clin Nutr 2008; 87: 534–538. 22. Yin YN, Yu QF, Fu N, Liu XW, Lu FG. Effects of four bifidobacteria on References obesity in high-fat diet induced rats. World J Gastroenterol 2010; 16: 3394–3401. 23. Balamurugan R, George G, Kabeerdoss J, Hepsiba J, Chandragunasek- aran AM, Ramakrishna BS. Quantitative differences in intestinal 1. Ley RE, Backhed F, Turnbaugh P, Lozupone CA, Knight RD, Gordon JI. Faecalibacterium prausnitzii in obese Indian children. Br J Nutr 2010; Obesity alters gut microbial ecology. Proc Natl Acad Sci USA 2005; 102: 103: 335–338. 11070–11075.

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI 8 Clinical Microbiology and Infection CMI

24. Nadal I, Santacruz A, Marcos A et al. Shifts in clostridia, bacteroides 46. Bartosch S, Fite A, Macfarlane GT, McMurdo ME. Characterization of and immunoglobulin-coating fecal bacteria associated with weight loss bacterial communities in feces from healthy elderly volunteers and in obese adolescents. Int J Obes 2009; 33: 758–767. hospitalized elderly patients by using real-time PCR and effects of 25. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. antibiotic treatment on the fecal microbiota. Appl Environ Microbiol An obesity-associated gut microbiome with increased capacity for 2004; 70: 3575–3581. energy harvest. Nature 2006; 444: 1027–1031. 47. Palmer C, Bik EM, DiGiulio DB, Relman DA, Brown PO. Development 26. Samuel BS, Gordon JI. A humanized gnotobiotic mouse model of host – of the human infant intestinal microbiota. PLoS Biol 2007; 5: e177. archaeal–bacterial mutualism. Proc Natl Acad Sci USA 2006; 103: 10011– 48. Sekirov I, Tam NM, Jogova M et al. Antibiotic-induced perturbations of 10016. the intestinal microbiota alter host susceptibility to enteric infection. 27. Zhang H, DiBaise JK, Zuccolo A et al. Human gut microbiota in obesity Infect Immun 2008; 76: 4726–4736. and after gastric bypass. Proc Natl Acad Sci USA 2009; 106: 2365–2370. 49. Robinson CJ, Young VB. Antibiotic administration alters the commu- 28. Turnbaugh PJ, Ridaura VK, Faith JJ, Rey FE, Knight R, Gordon JI. The nity structure of the gastrointestinal microbiota. Gut Microbes 2010; 1: effect of diet on the human gut microbiome: a metagenomic analysis in 279–284. humanized gnotobiotic mice. Sci Transl Med 2009; 1: 6 –14. 50. Cho I, Yamanishi S, Cox L et al. Antibiotics in early life alter the murine 29. Backhed F, Ley RE, Sonnenburg JL, Peterson DA, Gordon JI. Host – colonic microbiome and adiposity. Nature 2012; 488: 621–626. bacterial mutualism in the human intestine. Science 2005; 307: 1915 –1920. 51. Dumonceaux TJ, Hill JE, Hemmingsen SM, Van Kessel AG. Character- 30. Hayashi H, Sakamoto M, Benno Y. Fecal microbial diversity in a strict ization of intestinal microbiota and response to dietary virginiamycin vegetarian as determined by molecular analysis and cultivation. supplementation in the broiler chicken. Appl Environ Microbiol 2006; 72: Microbiol Immunol 2002; 46: 819–831. 2815–2823. 31. Liszt K, Zwielehner J, Handschur M, Hippe B, Thaler R, Haslberger AG. 52. Dethlefsen L, Huse S, Sogin ML, Relman DA. The pervasive effects of Characterization of bacteria, clostridia and Bacteroides in faeces of an antibiotic on the human gut microbiota, as revealed by deep 16S vegetarians using qPCR and PCR-DGGE fingerprinting. Ann Nutr Metab rRNA sequencing. PLoS Biol 2008; 6: e280. 2009; 54: 253–257. 53. Dethlefsen L, Relman DA. Incomplete recovery and individualized 32. Walker AW, Ince J, Duncan SH et al. Dominant and diet-responsive responses of the human distal gut microbiota to repeated antibiotic groups of bacteria within the human colonic microbiota. ISME J 2011; 5: perturbation. Proc Natl Acad Sci USA 2011; 108(suppl 1): 4554–4561. 220–230. 54. Moore PR, Evenson A et al. Use of sulfasuxidine, streptothricin, and 33. De Filippo C, Cavalieri D, Di Paola M et al. Impact of diet in shaping gut streptomycin in nutritional studies with the chick. J Biol Chem 1946; microbiota revealed by a comparative study in children from Europe 165: 437–441. and rural Africa. Proc Natl Acad Sci USA 2010; 107: 14691–14696. 55. Stokstad EL, Jukes TH et al. The multiple nature of the animal protein 34. Wu GD, Chen J, Hoffmann C et al. Linking long-term dietary patterns factor. J Biol Chem 1949; 180: 647–654. with gut microbial enterotypes. Science 2011; 334: 105–108. 56. Duggar BM. Aureomycin; a product of the continuing search for new 35. Monira S, Nakamura S, Gotoh K et al. Gut microbiota of healthy and antibiotics. Ann N Y Acad Sci 1948; 51: 177–181. malnourished children in Bangladesh. Front Microbiol 2011; 2: 228. 57. Cromwell GL. Why and how antibiotics are used in swine production. 36. Martin FP, Sprenger N, Yap IK et al. Panorganismal gut microbiome– Anim Biotechnol 2002; 13: 7–27. host metabolic crosstalk. J Proteome Res 2009; 8: 2090 –2105. 58. Food and Drug Administration. Withdrawal of Notices of Opportunity 37. Cani PD, Neyrinck AM, Fava F et al. Selective increases of bifidobac- for a Hearing; Penicillin and Tetracycline Used in Animal Feed. Federal teria in gut microflora improve high-fat-diet-induced diabetes in mice Register [Internet]. 2011; 76(246). Available from: http://www.gpo.gov/ through a mechanism associated with endotoxaemia. Diabetologia 2007; fdsys/pkg/FR-2011-12-22/html/2011-32775.htm. 50: 2374–2383. 59. Haight TH, Pierce WE. Effect of prolonged antibiotic administration of 38. Martin R, Langa S, Reviriego C et al. Human milk is a source of lactic the weight of healthy young males. J Nutr 1955; 56: 151–161. acid bacteria for the infant gut. J Pediatr 2003; 143: 754–758. 60. Ozawa E. Studies on growth promotion by antibiotics. II. Results of 39. Solis G, de Los Reyes-Gavilan CG, Fernandez N, Margolles A, aurofac administration to infants. J Antibiot (Tokyo) 1955; 8: 212 –214. Gueimonde M. Establishment and development of lactic acid bacteria 61. Perrini F. Aureomycin as a growth factor in premature infants. Boll Soc and bifidobacteria microbiota in breast-milk and the infant gut. Ital Biol Sper 1951; 27: 1151–1152. Anaerobe 2010; 16: 307–310. 62. Robinson P. Controlled trial of aureomycin in premature twins and 40. Jacobsen CN, Rosenfeldt Nielsen V, Hayford AE et al. Screening of triplets. Lancet 1952; 259: 52. probiotic activities of forty-seven strains of Lactobacillus spp. by in vitro 63. Raoult D. Human microbiome: take-home lesson on growth promot- techniques and evaluation of the colonization ability of five selected ers? Nature 2008; 454: 690–691. strains in humans. Appl Environ Microbiol 1999; 65: 4949–4956. 64. Ternak G. Antibiotics may act as growth/obesity promoters in humans 41. Gueimonde M, Collado MC. Metagenomics and probiotics. Clin as an inadvertent result of antibiotic pollution? Med Hypotheses 2005; Microbiol Infect 2012; 18(suppl 4): 32 –34. 64: 14 –16. 42. Grzeskowiak L, Gronlund MM, Beckmann C, Salminen S, von Berg A, 65. Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant Isolauri E. The impact of perinatal probiotic intervention on gut antibiotic exposures and early-life body mass. Int J Obes (Lond) 2013; 37: microbiota: double-blind placebo-controlled trials in Finland and 16 –23. Germany. Anaerobe 2012; 18: 7–13. 66. Southern KW, Barker PM, Solis-Moya A, Patel L. Macrolide 43. EUROPA. Ban on antibiotics as growth promoters in animal feed antibiotics for cystic fibrosis. Cochrane Database Syst Rev 2011; enters into effect. 2005. Available from: http://europa.eu/rapid/press- CD002203. doi: 10.1002/14651858.CD002203.pub4. release_IP-05-1687_en.htm. 67. Garly ML, Bale C, Martins CL et al. Prophylactic antibiotics to prevent 44. Million M, Angelakis E, Paul M, Armougom F, Leibovici L, Raoult D. pneumonia and other complications after measles: community based Comparative meta-analysis of the effect of Lactobacillus species on randomised double blind placebo controlled trial in Guinea-Bissau. BMJ weight gain in humans and animals. Microb Pathog 2012; 53: 100–108. 2006; 333: 1245–1247. 45. Kondo S, Xiao JZ, Satoh T et al. Antiobesity effects of Bifidobacterium 68. Membrez M, Blancher F, Jaquet M et al. Gut microbiota modulation breve strain B-3 supplementation in a mouse model with high-fat diet- with norfloxacin and ampicillin enhances glucose tolerance in mice. induced obesity. Biosci Biotechnol Biochem 2010; 74: 1656–1661. FASEB J 2008; 22: 2416–2426.

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI CMI Million et al. Microbiota and obesity 9

69. Looft T, Johnson TA, Allen HK et al. In-feed antibiotic effects on the 80. Saiman L, Mayer-Hamblett N, Anstead M et al. Open-label, follow-on swine intestinal microbiome. Proc Natl Acad Sci USA 2012; 109: 1691– study of azithromycin in pediatric patients with CF uninfected with 1696. Pseudomonas aeruginosa. Pediatr Pulmonol 2012; 47: 641–648. 70. Kim HB, Borewicz K, White BA et al. Microbial shifts in the swine distal 81. Clement A, Tamalet A, Leroux E, Ravilly S, Fauroux B, Jais JP. Long gut in response to the treatment with antimicrobial growth promoter, term effects of azithromycin in patients with cystic fibrosis: a double tylosin. Proc Natl Acad Sci USA 2012; 109: 15485–15490. blind, placebo controlled trial. Thorax 2006; 61: 895–902. 71. Collier CT, Smiricky-Tjardes MR, Albin DM et al. Molecular ecological 82. Mansi Y, Abdelaziz N, Ezzeldin Z, Ibrahim R. Randomized controlled analysis of porcine ileal microbiota responses to antimicrobial growth trial of a high dose of oral erythromycin for the treatment of feeding promoters. J Anim Sci 2003; 81: 3035–3045. intolerance in preterm infants. Neonatology 2011; 100: 290–294. 72. Rettedal E, Vilain S, Lindblom S et al. Alteration of the ileal microbiota 83. Ng YY, Su PH, Chen JY et al. Efficacy of intermediate-dose oral of weanling piglets by the growth-promoting antibiotic chlortetracy- erythromycin on very low birth weight infants with feeding intolerance. cline. Appl Environ Microbiol 2009; 75: 5489–5495. Pediatr Neonatol 2012; 53: 34 –40. 73. Torok VA, Allison GE, Percy NJ, Ophel-Keller K, Hughes RJ. Influence 84. Lane JA, Murray LJ, Harvey IM, Donovan JL, Nair P, Harvey RF. of antimicrobial feed additives on broiler commensal posthatch gut Randomised clinical trial: Helicobacter pylori eradication is associated microbiota development and performance. Appl Environ Microbiol 2011; with a significantly increased body mass index in a placebo-controlled 77: 3380–3390. study. Aliment Pharmacol Ther 2011; 33: 922–929. 74. Torok VA, Hughes RJ, Mikkelsen LL et al. Identification and charac- 85. Kamada T, Hata J, Kusunoki H et al. Eradication of Helicobacter pylori terization of potential performance-related gut microbiotas in broiler increases the incidence of hyperlipidaemia and obesity in peptic ulcer chickens across various feeding trials. Appl Environ Microbiol 2011; 77: patients. Dig Liver Dis 2005; 37: 39 –43. 5868–5878. 86. Patterson PR. Minocycline in the antibiotic regimen of cystic fibrosis 75. Guban J, Korver DR, Allison GE, Tannock GW. Relationship of dietary patients: weight gain and clinical improvement. Clin Pediatr (Phila) 1977; antimicrobial drug administration with broiler performance, decreased 16: 60 –63. population levels of Lactobacillus salivarius, and reduced bile salt 87. Guzman MA, Scrimshaw NS, Monroe RJ. Growth and development of deconjugation in the ileum of broiler chickens. Poult Sci 2006; 85: Central American children. I. Growth responses of rural Guatemalan 2186–2194. school children to daily administration of penicillin and aureomycin. Am 76. Thuny F, Richet H, Casalta JP, Angelakis E, Habib G, Raoult D. J Clin Nutr 1958; 6: 430 –438. Vancomycin treatment of infective endocarditis is linked with recently 88. Heikens GT, Schofield WN, Dawson S. The Kingston Project. II. The acquired obesity. PLoS ONE 2010; 5: e9074. effects of high energy supplement and metronidazole on malnourished 77. Pirzada OM, McGaw J, Taylor CJ, Everard ML. Improved lung function children rehabilitated in the community: anthropometry. Eur J Clin Nutr and body mass index associated with long-term use of macrolide 1993; 47: 160–173. antibiotics. J Cyst Fibros 2003; 2: 69 –71. 89. Bethell DB, Hien TT, Phi LT et al. Effects on growth of single short 78. Saiman L, Marshall BC, Mayer-Hamblett N et al. Azithromycin in courses of fluoroquinolones. Arch Dis Child 1996; 74: 44 –46. patients with cystic fibrosis chronically infected with Pseudomonas 90. Corbo S, Frontali G, Jolliffe N, Lanciano O, Maggioni G. Effects of aeruginosa: a randomized controlled trial. JAMA 2003; 290: 1749– chlortetracycline on weight gain of Italian children ages 6 to 10 on diets 1756. relatively low in animal protein. Antibiot Annu 1955; 3: 19 –26. 79. Saiman L, Anstead M, Mayer-Hamblett N et al. Effect of azithromycin 91. Macdougall LG. The effect of aureomycin on undernourished African on pulmonary function in patients with cystic fibrosis uninfected with children. J Trop Pediatr 1957; 3: 74 –81. Pseudomonas aeruginosa: a randomized controlled trial. JAMA 2010; 303: 92. Coodin FJ. Studies of tetramycin in premature infants. Pediatrics 1953; 1707–1715. 12: 652–656.

ª2013 The Authors Clinical Microbiology and Infection ª2013 European Society of Clinical Microbiology and Infectious Diseases, CMI

Article II : REVIEW

The relationship between gut microbiota and weight

gain in humans

Emmanouil Angelakis, Fabrice Armougom, Matthieu Million,

Didier Raoult

Published in Future Microbiol. 2012 Jan;7(1):91-109. (IF 3.82)

For reprint orders, please contact: [email protected] Review The relationship between gut Future Microbiology microbiota and weight gain in humans

Emmanouil Angelakis*, Fabrice Armougom, Matthieu Million & Didier Raoult Unité des Rickettsies, URMITE -CNRS UMR 6236 IRD 198, IFR 48, Faculté de Médecine, Université de la Méditerranée, 27 Bd Jean Moulin, 13385 Marseille Cedex 05, France *Author for correspondence: Tel.: + 33 491 38 55 17 Fax: + 33 491 83 03 90 [email protected]

The human gut microbiota is a metabolic organ that is determined by a dynamic process of selection and competition. Age, dietary habits and geographical origin of people have an important impact on the intestinal microbiota. The role of the microbiota is still largely unknown, but the bacteria of the gut flora do contribute enzymes that are absent in humans and play an essential role in the catabolism of dietary fibers. Germ-free mice provide a complementary approach for characterizing the properties of the human gut microbiota. Recently, microbial changes in the human gut were proposed to be one of the possible causes of obesity. This review summarizes the latest research on the association between microbial ecology and host weight.

Obesity is a major, public health concern that gastrointestinal microbiota [8] . In addition, affects at least 400 million individuals and is evidence implicating the role of microbiota in associated with severe disorders including dia- inflammatory bowel disease was supported by a betes and cancers [1] . The causes that drive certain degree of effectiveness of antibiotics in obesity appear to be complex, and a consensus the prevention and treatment of colonic inflam- hypothesis is emerging that proposes that obe- mation in both human patients and animal sity is influenced by a mixture of environmen- models, as well as by the presence of microbes tal, genetic, neural and endocrine factors [1] . and microbial components in inflammation- Infectious agents have also been proposed to be induced colonic lesions [9] . The association of the causes of obesity, and in human obesity, have gut microbiota with cancer is most commonly been associated with small EDRK-rich factor 1A observed with gastrointestinal tumors, although (SMAM-1), an avian adenovirus and adenovi- there are examples of these microbiota modify- rus 36 [2] . Human genetics is believed to play a ing the cancer risk to other systems, such as in part in determining body weight [3] . In total, breast tumors [7] . Moreover, the notion that gut 32 genes were linked to BMI, but their total vari- microbiota is important in the onset and devel- ance contribution to BMI in the population was opment of diabetes, cardiovascular dyslipidemia less than 2% [4] . It is believed that other factors and metabolic endotoxemia is becoming more also play a role in obesity, such as the availabil- widely accepted as the evidence base grows [7,10] , ity of inexpensive, calorically dense foods or the and the beneficial effect of bariatric surgery in reduction in physical activity in our daily lives. decreasing cardiovascular risk and cancer was Recently, microbial changes in the human gut associated with the increase of Enterobacter was proposed to be another possible cause of hormaechei in the gut microbiota [11] . obesity [5] and it was found that the gut microbes The role of the digestive microbiota in the from fecal samples contained 3.3 million nonre- human body is still largely unknown, but the dundant microbial genes [6] . However, it is still bacteria of the gut flora do contribute enzymes poorly understood how the dynamics and com- that are absent in humans for food digestion position of the intestinal microbiota are affected [12] . Moreover, the link between obesity and the by diet or other lifestyle factors. Moreover it has microbiota is likely to be more sophisticated than been difficult to characterize the composition of the simple phylum-level Bacteroidetes :Firmicutes the human gut microbiota due to large variations ratio that was initially identified [13] , and it is Keywords between individuals. likely to involve a microbiota–diet interaction gut flora microbiota The human gut microbiota has been also [14] . Phages have also been proposed to play a obesity associated with a number of disease states that possible role in driving the biodiversity of the gut include allergy, inflammatory bowel disease, flora by their influence on their bacterial hosts cancer and diabetes [7] . Allergy, for example, [15] and, recently, a novel pathway that involves has been associated with perturbations in the dietary lipid phosphatidylcholine and choline part of

10.2217/FMB.11.142 © Raoult D et al. Future Microbiol. (2012) 7(1), 91–109 ISSN 1746-0913 91 Review Angelakis, Armougom, Million & Raoult

metabolism, an obligate role for the intestinal [18] , and in a study of 650 individuals, the prev- microbial community, and regulation of sur- alence of M. smithii was 95.5%, whereas the face expression levels of macrophage scavenger prevalence of Methanosphaera stadtmanae was receptors that were known to participate in the 29.4% in the human gut [20] . Moreover, molecu- atherosclerotic process was proposed [16] . More lar analyses provided various degrees of evidence subtle alterations in the levels of other bacteria for the presence of groups of archaea, including in the gut may also impact human health. In the Methanosarcina , Thermoplasma , Crenarchaeota last few years, new technologies have been devel- and halophilic archaea in the human gastro- oped that have allowed researchers to attempt intestinal tract, but isolates have not been more systematic studies on intestinal bacterial obtained [21] . flora and have given more realistic information about its composition (by way of detecting non- Age & gut flora modification cultivable species). As a result, an increasing During the first days to months of life, the micro- number of studies have related imbalances in biota of the infant gut and the temporal pattern the composition of the gut microbiota to obesity in which it evolves is remarkably variable from and its associated diseases. The approaches used individual to individual [22] . At birth, humans to characterize the human gut flora vary widely, are essentially free of bacteria and over time, in a and this might explain, in part, why specific process of colonization that begins shortly after alterations in the microbiota that are associated delivery and continues through to adulthood, with excess body fat or weight loss, can also vary the body becomes a host to complex microbial between studies. This review summarizes the communities. The initial infant gut microbiota latest research on the association between the is usually dominated by Bifidobacteria , and microbial ecology and host weight. through a series of successions and replace- ments, it migrates to a more complex, adult pat- Human gut microbiota tern [22] . Vael et al. found that the population of The gut microbiota harbors large bacterial popu- Bacteroides fragilis in the microbiota increased in lations in the intestine and colon, approximately infants from the age of 3 weeks until the age of 10 11–12 microorganisms per gram of content, and 1 year, whereas the populations of Staphylococcus , are comprised of mainly anaerobes (95% of the Lactobacillus , Bifidobacterium , Clostridium and total organisms). The initial overview of the com- total anaerobes decreased starting at the age of position of the gut microbiota was culture based, 3 weeks and remained stable until 52 weeks [23] . and the predominant cultivable species that were Traditionally, it has been thought that identified included Bacteroides sp., Eubacterium between 1 and 2 years of age, the human gut sp., Bifidobacterium sp., Peptostreptoccocus sp., microbiota start to resemble that of an adult Fusobacterium sp., Ruminococcus sp., Clostridium [22] . Young children between 1 and 7 years of sp. and Lactobacillus spp. [17] . The first, large- age presented higher numbers of enterobacteria scale, 16S rDNA sequencing analysis of the gut than adults [24] . Moreover, a large-scale study microbiota by Eckburg et al. [18] revealed a high by Enck et al. found significant shifts in relative inter-individual variability at the species taxo- genus abundances during the first 2 years of life nomic level that was not recovered at the phylum and no noticeable changes in children between level, as only nine phyla out of 70 were repre- 2 and 18 years of age, including stable levels of sented [1] . The overall and individual microbiota Bifidobacterium and Lactobacillus [25] . In a recent structures were dominated by the Bacteroidetes study, the comparison of intestinal microbiota and Firmicutes phyla [18] . Finally, three gut micro- composition between adolescents and adults biota studies [19] assigned 98% of 16S rRNA revealed a statistically significantly higher abun- sequences to only four bacterial phyla: Firmicutes dance of genera Bifidobacterium and Clostridium (64%), Bacteroidetes (23%), Proteobacteria (8%) among adolescent samples [26] . and Actinobacteria (3%). Verrucomicrobia , The adult intestinal microbiota has been Fusobacteria and the TM7 p hylum together shown to be relatively stable over time [27] and accounted for the remaining 2%. is sufficiently similar between individuals. This The earliest large-scale, 16S rRNA or metage- observation allowed for identification of a core nomic studies identified Methanobrevibacter microbiome that was comprised of 66 dominant, smithii as the dominant, methanogenic archaeon operational, taxonomic units that corresponded species in the human gut microbiota [18] . to 38% of the sequence reads from 17 individu- M. smithii in three healthy individuals com- als [28] . Turroni et al. found that Bifidobacterium prised up to 11.5% of the gut microorganisms pseudolongum and Bifidobacterium bifidum , are

92 Future Microbiol. (2012) 7(1) future science group Gut flora & weight gain Review exclusively dominant in the adult bifidobacte- the cohort. Of these, 99.1% of the genes had rial population, whereas Bifidobacterium longum , bacterial origin, and the remainder was mostly Bifidobacterium breve , Bifidobacterium pseudoca- archaeal, with only 0.1% of eukaryotic or viral tenulatum and Bifidobacterium adolescentis , were origins [6] . Therefore, it seems that important found to be widely distributed, irrespective of variations in the gut flora between close coun- host age [29] . tries do not exist. As a result, Dicksved et al. In the elderly, both Bacteroides numbers and did not observe differences between fecal sam- species diversity is declined [30,31] . The analyses ples collected from children from Germany, of fecal samples collected from subjects from Switzerland and Sweden by the use of termi- four European study groups indicated higher nal restriction fragment length polymorphism proportions of enterobacteria in all elderly vol- [36] . Lay et al., when testing the composition unteers [32] . Zwielehner et al. , found that the of the fecal microbiota assessed by FISH com- proportion of Bacteroidetes in the fecal micro- bined with flow cytometry, also did not find a biota of 17 institutionalized, elderly subjects significant correlation between the microbial was significantly higher than in younger adults, compositions, with regard to age, geographical but these patients had lower proportions of origin, or gender, among subjects from France, Bifidobacterium and Clostridium cluster IV [33] . Denmark, Germany, the Netherlands and the Analysis of the core microbiota in the elderly UK [37] . However, 16S rDNA pyrosequencing showed a clear shift to a more Clostridium ana lysis revealed that geographical origin has cluster IV-dominated community [34] . an important impact on the intestinal micro- Several host factors have been correlated with biota. As a result, differences in the gut micro- methanogenic archaea carriage, and it has been biota have been observed between people living proposed that the acquisition of methanogenic in northern and southern European countries. archaea occurs by environmental contamination. For instance, Fallani et al. observed that human Additionally, it has been hypothesized that once infants from northern European countries were methanogenic archaea find favorable physico- associated with higher Bifidobacteria in their chemical conditions and available substrates in gut microbiota, whereas infants with higher the gut, stable colonization is established [21] . Bacteroides and lactobacilli were characteristic archaea were not detected in children who were of southern countries [38] . Mueller et al. found younger than 27 months, but it has been shown that the proportion of Bifidobacteria was two- that carriage increases with age, up to 60% in to three-fold higher in Italians than in the 5-year-old children. Moreover, it is possible that French, Germans or Swedes [32] . A bigger dif- an adult diet may create an intestinal microbiota ference has been observed between European that is favorable for the implantation of metha- and Africans, and De Filippo et al. found that nogenic archaea [35] . A possible direct, mother- children from a rural African village presented to-child route of transmission has also been pro- more Actinobacteria and Bacteroidetes but posed because archaea have been detected in the less Firmicutes and Proteobacteria in their gut vaginal flora of pregnant women [21] . flora than European children [39] . Moreover, African children presented significantly more Gut flora variations among different short-chain fatty acids in their gut flora than populations European children [39] . Li et al. found that there It is not yet completely understood how the were distinct microbiota profiles at the species different environments and wide range of diets level between a Chinese family and American that modern humans around the world experi- volunteers. Moreover, they identified a higher ence has affected the microbial ecology of the proportion of Bacteroidetes thetaiotaomicron in human gut. Certain lifestyles of a person may males than in females [40] . Finally, Arumugam have an impact on the composition of his/her et al. , by combining 22 sequenced, fecal metage- gut microbiota (F IGURE 1) , but these impacts are nomes of individuals from four countries, iden- currently poorly understood. Qin et al. , in the tified three enterotype clusters that were not largest study to date, found that only one-third nation- or continent-specific [41] . Enterotype 1 of the bacterial gene clusters that were conserved was enriched in Bacteroides and seemed to derive across individuals of all 124 European (Nordic energy primarily from carbohydrates and pro- and Mediterranean) origins could be associated teins through fermentation. Enterotype 2 was with a broad functional assignment [6] . Nearly enriched in Prevotella and Desulfovibrio , which 40% of the genes from each individual were can act in synergy to degrade mucin glycopro- shared with at least half of the individuals of teins that are present in the mucosal layer of the

future science group www.futuremedicine.com 93 Review Angelakis, Armougom, Million & Raoult

Dietary habits Age

First days of life Vegetarian diet Mostly Bifidobacteria 1. Increase Bacteroidetes Human microbiota 2. Decrease Clostridia Stable: Firmicutes (64%) Adults Bacteroidetes (23%) Proteobacteria (8%) Actinobacteria (3%)

1. Increase Bacteroidetes Elderly 2. Decrease Bifidobacteria

Different species level Increase Bifidobacteria 1. Increase Actinobacteria and Bacteroidetes Southern vs northern 2. Decrease Firmicutes Europeans and Proteobacteria

Origin Chinese vs Americans Europeans vs Africans

Figure 1. Impact factors for the composition of the human gut microbiota.

gut. Enterotype 3 was the most frequent and carboxymethylcellulase and endoglucanase [39] . was enriched in Ruminococcus and Akkermansia, Moreover, Bacteroides and Faecalibacterium spe- which degrade mucins [41] . Moreover, entero- cies and particularly Faecalibacterium prausnit- types 1 and 2 were capable of biosynthesis of zii , which were found in both children popula- different vitamins. The authors proposed that tions, could generally indicate the importance these three enterotypes used different routes of maintaining a microflora with potential anti- to generate energy from fermentable substrates inflammatory capability [39,44] . Liszt et al. found that were available in the colon, reminiscent of that a vegetarian diet affected the intestinal a potential specialization in ecological niches or microbiota, especially by decreasing the amount guilds [41] . and changing the diversity of Clostridium cluster IV [45] . Similar results found by Hayashi Effect of the alimentation on human et al. , who based their studies on RFLP analysis, gut flora revealed that the major composition of the veg- Dietary habits are considered to be one of the etarian gut microbiota consisted of Clostridium main factors that contribute to the diversity of rRNA subcluster XIVa and Clostridium rRNA the human gut microbiota [42] , and the pattern cluster XVIII [46] . Recently, Walker et al. tested of variation in copy number of the human sali- overweight men with a control diet, diet high vary amylase gene is consistent with a history of in resistant starch or nonstarch polysaccharides diet-related selection pressures, demonstrating and a reduced carbohydrate weight loss diet, over the importance of starchy foods in human evolu- 10 weeks and they found no significant effect tion [43] . Prevotella , Xylanibacter and Treponema of diet upon the proportions of Bacteroidetes , were present in the gut flora of children from Firmicutes , Actinobacteria or Proteobacteria a rural African village but not from Europe, within the fecal microbiota [47] . However, two and the authors of this study hypothesized that individual phylotypes, Eubacterium rectale and the presence of these three genera could be a Ruminococcus bromii , showed increased propor- consequence of high fiber intake, maximiz- tions on the resistant starch diet while Collinsella ing metabolic energy extraction from ingested aerofaciens showed decreased proportions on the plant polysaccharides [39] . These bacteria could weight loss diet [47] . Finally, Wu et al. analyzed ferment both xylan and cellulose through car- the fecal samples from 98 individuals and found bohydrate-active enzymes, such as xylanase, that fecal communities clustered into enterotypes

94 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review [5] [51] [53] [59] [52] [57] [58] [54] Ref. predict Ob/Ov in the Ob population levels in Ob subjects Actinobacteria levels sp. in Obsp. subjects family of in Ov pregnant women Bacteroides Staphylococcus aureus and rectale/Roseburia Bacteroides

and weight loss in Ob subjects Prevotellaceae in Ob subjects in Ob subjects Methanobrevibacter Coriobacteriaceae levels between Ob and N subjects Eubacterium and Staphylococcus aureus in Ob subjects Bacteroidetes and levels, independent diet, of in Ob versus N subjects Clostridium perfringens Bacteroidetes Bacteroidetes Bifidobacteria Bacteroidetes Methanobacteriales in Ob subjects (not significant) Bacteroides Bacteroidetes Firmicutes Bacteroidetes n gut microbiota. Decrease in Significant decrease in Correlation between an increase in Lower number bifidobacteria of and greater number of Significantly reduced levels of More Significantly reduced level of phenotype No significant difference in Correlation between excessive weight gain and high Higher numbers of Significant diet-dependent reduction in No difference in Ob microbiota somewhat enriched in the Ob microbiota were significantly enriched in Significant increase in † † yrosequencing; qPCR: Quantitative real-time PCR. † † † Bacteroidetes Bifidobacteria Bacteroidetes Bifidobacteria Firmicutes Staphylococcus aureus Firmicutes Firmicutes Lactobacilli Clostridia Staphylococcus aureus Bacteroidetes Bifidobacteria Bacteroidetes Bacteroidetes Firmicutes Eubacterium rectale/ Clostridium coccoides Actinobacteria Bacteroidetes Proteobacteria Fusobacteria Verrucomicrobia FISH sequencing qPCR Method Community measured Major finding Ov N N NOb qPCR Firmicutes Bacteroidetes Significant increase in Ob Culture Ob/Ov children Ob Pyro 16S N children Ob FISH N pregnant N Ob clonal 16S NOb qPCR FISH category N Ob pregnant FCM-FISH and

et al. et

et al. et et al. et et al. et et al. et Indicates that the difference is signiﬁcant. Schwiertz et al. et Kalliomäki al. et Zuo Zhang Mai Duncan al. et Table 1. Weight gain-associated 1. Table bacterial population shifts in huma StudyLey Sample Collado Collado † FCM: Flow cytometry; N: Normal weight; Ob: Obese; Ov: Overweight; Pyro: P

future science group www.futuremedicine.com 95 Review Angelakis, Armougom, Million & Raoult

[13] [49] [60] [55] [56] Ref. in Ob subjects

) Clostridium histolyticum Firmicutes iated with: Firmicutes and and levels between Ob and N subjects groups Bacteroidetes levels (belonging to Actinobacteria Bifidobacterium in Ob versus N subjects in Ob versus N subjects and Lactobacillus levels in Ob versus N subjects and Bacteroidetes Bacteroidetes Lactobacillus Bacteroidetes Eubacterium rectale, Clostridium coccoides C. coccoides C. Bacteroides/Prevotella Actinobacteria Faecalibacterium prausntzi Bacteroides fragilis n gut microbiota (cont.). Significant reduction in Correlation with weight Significant increase in Increase the in Significantly reduced levels of Present after an Ob group submitted a weight to program lost >4 kg No significant difference in Nearly half the of lean-enriched genes were from Significant reduction of Significant increase in Significant increase of Ob microbiome enriched genes in that belong to Significantly higher levels of No difference in dietary intake Greater weight loss after a multidisciplinary treatment program assoc † † yrosequencing; qPCR: Quantitative real-time PCR. † † Lactobacillus/Enteroccocus Enteric group Lactobacillus Clostridium leptum Bifidobacterium Escherichia coli bacteria Total Eubacterium rectale/ Clostridium coccoides Clostridium coccoides Clostridium hystolyticum Firmicutes Bacteroides fragilis Bifidobacterium Proteobacteria Bifidobacterium Bacteroidetes Bacteroidetes Actinobacteria Lactobacillus acidophilus Eubacterium rectale Faecalibacterium prausnitzii Bacteroidetes Lactobacillus Methanobrevibacter smithii Bacteroidetes/Prevotella Method Pyro16S V2 of Community measuredclonal Sanger sequencing Major finding Pyro16S V6 of qPCR category Ob, N twins and mother Ov adolescents N Ob qPCR Ob qPCR Firmicutes Significantly reduced levels of N Anorexic Ob FISH

et al. et Indicates that the difference is signiﬁcant. Santacruz al. et † FCM: Flow cytometry; N: Normal weight; Ob: Obese; Ov: Overweight; Pyro: P Table 1. Weight gain-associated 1. Table bacterial population shifts in huma StudyTurnbaugh Sample al. et Balamurugan Balamurugan al. et Armougom al. et Nadal

96 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

distinguished primarily by levels of Bacteroides [62]

Ref. and Prevotella [48] . They also found that long- term diet, particularly protein and animal fat versus carbohydrate diet were strongly associated with enterotype partitioning. Moreover, in a controlled-feeding study authors found that and the microbiome composition changed detect- ably within 24 h of initiating a high-fat/low-fiber or low-fat/high-fiber diet, but that enterotype identity remained stable [48] . Staphylococcus Bacteria species & obesity The Bacteroidetes phylum Armougom et al. found a significant reduction of Bacteroidetes proportions in obese, compared and increased with lean and anorexic, individuals [49] and reported lower Bacteroidetes concentrations in obese subjects (T ABLE 1) [50] . Moreover, the ana lysis

Bacteroides of 16S rDNA sequences from 154 individuals indicated that the microbiota of obese subjects and was associated with a decrease in the diversity level and was composed of significantly fewer Bacteroidetes [13] . On the other hand, Schwiertz et al. quantified bacterial communities in overweight, obese and lean individuals and found Bifidobacterium a significant increase in the proportions of Bacteroidetes in obese and overweight groups [51] . levels in Ov pregnant women Likewise, before pregnancy, overweight women have a higher number of Bacteroidetes than women of normal weight, and excessive weight n gut microbiota (cont.). gain during pregnancy is associated with an Significantly reduced Escherichia coli increase in Bacteroidetes numbers [52] . Assuming that Type 2 diabetes and reduced glucose tolerance is linked to obesity, Larsen and colleagues also found higher levels of Bacteroidetes in dia-

yrosequencing; qPCR: Quantitative real-time PCR. betic patients than in control patients [10] . Using † † †

group 16S rDNA pyrosequencing, Zhang et al. studied

† the composition of the gut microbiota in morbidly obese, normal-weight and post-gastric- bypass subjects [53] . Their results indicated that the obese microbiota is significantly enriched in Bifidobacterium Escherichia coli Staphylococcus bacteria Total Lactobacillus Clostridium coccoides Clostridium leptum Bacteroides Prevotellaceae, a subgroup of Bacteroidetes [53] . Zuo et al. , using culture methods for organisms found in the feces of obese and normal weight participants, found that obese people had fewer cultivable Bacteroides than control individuals Method Community measured Major finding [54] . Moreover, they found that obese individuals with a Pro/Ala genotype of the nuclear hormone receptor peroxisome proliferator-activated receptor J2, which modulates cellular differentiation and lipid accumulation during adipogenesis, had category Ob pregnant qPCR Ov pregnant lower levels of Bacteroides than obese participants with a Pro/Pro genotype [54] . Interestingly, the monitoring of the proportions of two major bacterial communities in obese participants dur- Indicates that the difference is significant. Table 1. Weight gain-associated 1. Table bacterial population shifts in huma StudySantacruz Sample al. et † FCM: Flow cytometry; N: Normal weight; Ob: Obese; Ov: Overweight; Pyro: P ing a weight loss program resulted in linking an future science group www.futuremedicine.com 97 Review Angelakis, Armougom, Million & Raoult

increase in levels of Bacteroidetes to weight loss, Others studies have not found any correla- independent of energy intake [5] . The impact tion between the proportions of Bacteroidetes of an obesity treatment program, including and obesity or type of diet. Both qPCR and a calorie-restricted diet and increase of physi- FISH methods have been applied to subsets of cal activity on gut microbiota composition in lean and obese subjects, and both have failed overweight and obese adolescents was reported to associate a reduced level of Bacteroidetes to [55,56] . The FISH method indicated that a sig- obesity [57] . In an attempt to study whether the nificant increase in the ratio of Bacteroides and composition of early gut microbiota can affect Prevotella correlated to weight loss in the ado- weight development throughout early child- lescent group that exhibited the highest weight hood, Kalliomäki et al. monitored weight, loss [55] . Using the same population, the results height and bacterial community abundances in obtained by FISH [55] were verified by a quanti- children of 6 months, 12 months and 7 years of tative PCR (qPCR) method, which detected a age. Children who became overweight or obese notable increase in Bacteroides fragilis after the at 7 years did not present any significant reduc- weight loss program [56] . Lastly, Vael et al. found tion in the proportion of Bacteroides-Prevotella , that high intestinal Bacteroides fragilis concen- compared with those maintaining a normal trations and low Staphylococcus concentrations in weight [58] . The relationships between weight infants between the ages of 3 weeks and 1 year loss and Bacteroidetes abundance were examined were associated with a higher BMI in preschool in adults, but no difference between obese and children [23] . nonobese subjects was observed [59] .

Group by Study (year) Subgroup within study Sample size SDM and 95% CI phyla Ow/obese Control Ley et al . (2006) 16S clonal sequencing 12 2 Turnbaugh et al . (2009) V2 pyrosequencing African, ancestry 62 8 Turnbaugh et al . (2009) V2 pyrosequencing European, ancestry 42 26 Zhang et al . (2009) Pyrosequencing 3 3 Bacteroidetes relative count (% of total sequences) Collado et al . (2008) FCM-FISH 18 36 Armougom et al . (2009) qPCR 20 20 Schwiertz et al . (2010) qPCR 33 30 Million et al . (2011) qPCR 53 39 Bacteroidetes absolute count (log cells or copies of DNA) Ley et al. (2006) 16S clonal sequencing 12 2 Turnbaugh et al . (2009) V2 pyrosequencing, African ancestry 62 8 Turnbaugh et al . (2009) V2 pyrosequencing, European ancestry 42 26 Firmicutes relative count (% of total sequences) Armougom et al . (2009) qPCR 20 20 Schwiertz et al . (2010) qPCR 33 30 Million et al . (2011) qPCR 53 39 Firmicutes absolute count (log copies DNA)

-2.00 -1.00 0.00 1.00 2.00

Lean status Ow /obese

Figure 2. Meta-ana lysis of the obesity-associated gut microbiota alterations at the phylum level ( Bacteroidetes and Firmicutes ) comparing the absolute (abs) or relative (percentage of total sequences) number of sequences (generated by quantitative PCR or cloning/sequencing or pyrosequencing) or cells (flow cytometry-FISH). Meta-analysis was performed with the comprehensive meta-analysis software version 2 [93,94] . Each line represents a comparison between an obese group (right) and a control group (left). The first reported alteration [5] was a decrease in the relative proportion of Bacteroidetes (percentage decrease) represented by a deviation of the square (standardized difference in the means) to the left. The size of the square represents the relative weight of each comparison (random model). The length of the horizontal line represents the 95% CI and the diamond represents the summarized effect. The presence of a square to the right and left of the midline means studies with conflicting results corresponding to a substantial heterogeneity (I 2 >50%). Here, the only reproducible and significant alteration at the phylum level is the decrease in the absolute number of sequences of Firmicutes in obese subjects. Relative count of Bacteroidetes (n = 4; SDM = -0.51; 95% CI = -1.7–0.67; p = 0.40 [I 2 = 81%]); absolute count of Bacteroidetes (n = 4; SDM = -0.07; 95% CI = -0.78–0.65; p = 0.86 [I 2 = 85]); relative count of Firmicutes (n = 3; SDM = 0.88; 95% CI = -0.21–1.97; p = 0.11 [I 2 = 79%]); absolute count of Firmicutes (n = 3; SDM = -0.43; 95% CI = -0.72 to -0.15; p = 0.003 [I 2 = 0%]). FCM: Flow cytometry; Ow: Overweight; qPCR: Quantitative PCR; SDM: Standardized difference in the means.

98 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

Group by Study (year) Sample size SDM and 95% CI genus Obese Control

Collado et al. (2008) 18 36 Kalliomäki et al. (2008) 25 24 Schwiertz et al. (2009) 33 30 Balamurugan et al. (2010) 15 13 Santacruz et al. (2010) 16 34 Zuo et al. (2011) 52 52 Bifidobacterium (log copies DNA/ml)

Armougom et al. (2009) 20 20 Zuo et al. (2011)52 52 Million et al. (2011) 53 39 Lactobacillus (log copies DNA/ml)

-2.00-1.99 0.00 1.00 2.00 Lean status Ow/obese

Figure 3. Meta-ana lysis of the obesity-associated gut microbiota alterations at the genus level for Bifidobacteria and Lactobacilli comparing the absolute number of sequences generated by genus-specific quantitative PCR. For Bifidobacteria , a consistent difference was found by our meta-ana lysis between 159 obese subjects and 189 controls from six published studies showing that the digestive microbiota of the obese group was significantly depleted in Bifidobacteria . Low heterogeneity (I 2 = 17%) shows that this result is very robust. Additional tests have shown that there was no small studies bias (Egger’s regression intercept test, p = 0.92; no change after Duval and Tweedie’s trim and fill). For Lactobacilli , no consistent and significant summary effect was found comparing 127 obese subjects and 110 controls from three studies. Bifidobacterium sp. (n = 6; SDM = -0.45; 95% CI = -0.69 to -0.20; p < 0.001 [I 2 = 17%]); Lactobacillus spp. (n = 3; SDM = 0.29; 95% CI = -0.31–0.90; p = 0.34 [I 2 = 80%]). Ow: Overweight; SDM: Standardized difference in the means.

Meta-analysis of the obesity-associated Indian children presented significantly higher gut microbiota alteration at the phylum level levels of Faecalibacterium prauznitzii but no (Bacteroidetes ) comparing the absolute (abs) or difference between the levels of Bacteroides and relative (percentage of total sequences) number that of Prevotella , Bifidobacterium species, the of sequences (generated by qPCR or cloning/ Lactobacillus acidophilus group or Eubacterium sequencing or pyrosequencing) or cells (flow rectal , compared with lean children [60] . Duncan cytometry [FCM]-FISH) was performed for the et al. identified a significant, diet-dependent seven studies [5,13,49–53] . These studies revealed reduction in levels of Roseburia-E. rectale , a no difference in the Bacteroidetes concentrations group of butyrate-producing Firmicutes , for obese between obese people and people of normal patients that were on a weight-loss diet [59] . Zuo weight (F IGURE 2) . et al. found a lower amount of C. perfringens and a higher proportion of Enterococci in obese sub- The Firmicutes phylum jects when compared with normal-weight indi- Ley et al. reported that the reduced level of viduals [54] . Finally, Schwiertz et al. found that Bacteroidetes found in obese humans was counter- overweight and obese volunteers exhibited lower balanced by a proportional increase in Firmicutes cell numbers of the Ruminococcus flavefaciens [5] . The greater Firmicutes proportion tended subgroup [51] . to decrease when patients were submitted to a Meta-analysis of the obesity associated gut weight-loss program [5] . These results were in microbiota alteration at the phylum level agreement with other works, which found that (Firmicutes ) comparing the absolute (abs) or significantly reduced levels of Clostridium hys- relative (percentage of total sequences) number toliticum , Eubacterium rectale and Clostridium of sequences (generated by qPCR or cloning/ coccoides correlated to weight loss in an obese, sequencing or pyrosequencing) or cells (FCM- adolescent population [55,56] . Moreover, obese, FISH) was performed for the five studies

future science group www.futuremedicine.com 99 Review Angelakis, Armougom, Million & Raoult

Study (year) Subgroup within study Sample size SDM and 95% CI

Ow/obese Control

Armougorn et al. (2009) M. smithii specific qPCR 20 20 Schwiertz et al. (2010) Methanobrevibacter sp. qPCR 33 30 Million et al. (2011) M. smithii specific qPCR 53 39

-2.00 -1.00 0.00 1.00 2.00 Lean status Ow/obese

Figure 4. Meta-ana lysis of the obesity-associated gut microbiota alterations for archaea representatives comparing the absolute number of archaeal sequences generated by quantitative PCR. One study, focused on the Methanobacteriales order level, comparing only three obese subjects and three controls, found an increase of this bacterial group in the obese group [53] (square deviated to the right) instead of the three other studies. Our meta-analysis showed, by observing the funnel plot, that this study was an outlier that was subsequently excluded. The comparison of 106 obese subjects and 89 controls including ana lysis at the Methanobrevibacter genus level by Schwiertz et al. [51] and at the Methanobrevibacter smithii species level [49,50] is justified because it shows a consistent and reproducible effect with a significant reduction of Methanobrevibacter sp. in obese subjects (Egger’s regression intercept test, p value = 0.39; and Duval’s and Tweedie’s trim and fill did not change these results). Methanobrevibacter sp. (n = 3; SDM = -0.51; 95% CI: -0.79 to -0.22; p = 0.001 [I 2 = 0%]). Ow: Overweight; SDM: Standardized difference in the means.

[5,13,49–51] . The only reproducible and significant The Actinobacteria phylum alteration at the phylum level is the decrease in Recent gut microbiota studies that have been the absolute number of sequences of Firmicutes associated with obesity have focused on shifts in obese (n = 3; standardized difference in the in Firmicutes and Bacteroidetes populations. means [SDM] = -0.43; 95% CI = -0.72 to However, the Actinobacteria phylum, which is 2 -0.15; p = 0.003 [I = 0%]) (F IGURE 2) . comprised of the Bifidobacterium genus as well Recent studies suggest a role for Lactobacillus as other genera, has also been linked to weight spp. in weight changes, and the quantification gain. Indeed, in an investigation of gut micro- of Lactobacillus species in lean, anorexic and bial communities of 18 lean or obese twins and obese subjects revealed significantly higher their mothers, the obese subjects showed higher Lactobacillus concentrations in nearly half of levels of Actinobacteria [13] . Interestingly, most the obese population [49] . Obese Type 2 dia- of the obesity related genes were found to be betic patients displayed significantly higher from Actinobacteria (75%), and many of the levels of Bacilli and Lactobacillus spp. in their obesity associated genes that were identified gut microbiota [10] . However, an increase in were involved in carbohydrate, lipid and amino Lactobacillus number in an obese, adolescent acid processing [13] . In addition, the sequencing group after a weight-loss program was also ana lysis by Zhang and colleagues revealed that reported [56] . Thuny et al. reported significant the Coriobacteriaceae family of Actinobacteria weight gain in patients with infected endo- was enriched in the obese microbiota [53] . carditis after treatment with high doses of The fecal concentration of the Bifidobacterium vancomycin and proposed that Lactobacillus genus was reported to be significantly lower in spp. that were resistant to vancomycin were obese subjects when compared with lean sub- responsible for this weight gain [61] . Similarly, jects [51,52,58,62] . Moreover, Santacruz et al. found Million et al. found that L. reuteri was asso- significantly lower Bifidobacteria counts in obese ciated with obesity [50] . Meta-ana lysis of the subjects after they had been subjected to a dietary obesity associated gut microbiota alteration at program [56] . Furthermore, Zuo et al . found a the genus level for lactobacilli comparing the nonsignificant decrease in the concentration of absolute number of sequences generated by bifidobacteria between obese and normal weight genus-specific qPCR revealed a nonsignificant humans [54] . Meta-analysis of the obesity-asso- summary effect in Lactobacillus spp. levels in ciated gut microbiota alteration at the genus obese subjects (F IGURE 3) . level for bifidobacteria comparing the absolute

100 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

number of sequences generated by genus specific fermentation efficiency [21] . The importance of qPCR revealed that the obese group was consis- methanogenic Archaea to humans lies in their tently and significantly depleted in Bifidobacteria ability to improve fermentation efficiency by [21] (n = 6; SDM = -0.45; 95% CI = -0.69 to -0.20; removing H 2 from the gut . It has been 2 p < 0.001 (I = 17%) (F IGURE 3) . This is extremely speculated that the coexistence of Prevotellaceae important because bifidobacteria depletion with methanogenic Archaea species in the obese seems to be the more reproducible alteration in gut allows for greater efficiency of dietary poly- obese gut microbiota and the best candidate to saccharide fermentation and therefore increases have an antiobesity effect. their conversion into short-chain fatty acids, resulting in their excessive storage [53] . Archaea & obesity Meta-analysis of the obesity-associated gut Using the data of Armougom et al ., but cal- microbiota alteration at the genus level for culating means of log 10 copies DNA/ml of Methanobrevibacter spp., main representative M. smithii , we found, contrary to Armougom of Archaea known in the digestive microbiota, et al ., that there was a decrease in the M. smithii comparing the absolute number of sequences load in the obese group, compared with the generated by qPCR revealed that obese sub- normal group [49] . Correspondingly, Zhang jects presented less Methanobrevibacter than et al. found more M. smithii in obese individu- nonobese subjects (F IGURE 4) . However, the rea- als than in lean controls [53] , and Schwiertz sons linking methanogens to weight gain still et al. identified lower levels of M. smithii in remain unclear. To date, Methanobrevibacter is obese subjects compared with lean subjects [51] . the main representative of archaea in the gut However, Million et al. recently found higher microbiota but archaea could not be extrapo- concentrations of M. smithii in nonobese sub- lated from Methanobrevibacter assessment. This jects [50] . Overall, methanogenic archaea could is extremely important since domain-level and indirectly promote caloric intake by the colon genus-level could lead to very different results. and further fat accumulation-related obesity in individuals who were on a high-fiber diet Ability to process polysaccharides [21] . During the fermentation process, the The gut microbiome is also involved in the accumulation of excess H 2 reduces the yield of complex carbohydrate metabolism of food ATP, which leads to a gradual decrease in the owing to its ability to process indigestible

Major carbon sources of the gut microbiota Nondigestible food components Substrates Starch, pectin, cellulose, mucilage xylan, inulin, fructans

Processes Anaerobic fermentation

– Intermediate products Succinate, formate… Lactate

Major end Butyrate Acetate Propionate H excess CO 2 products 2 SO 2- + Short-chain fatty acids 4

Acetate Methane SH 2 Colonocyste Lipid metabolism H2 removal mechanisms Bacteroides as major propionate producers Bacteroides and Firmicutes as major acetate producers Firmicutes as major butyrate producers Lactobacillus , Bifidobacterium and Streptococcus as major lactate producers Desulfovibrio as major sulfato-reducers Methanobrevibacter smithii archaeon species as major methane producers

Figure 5. Outline of carbohydrate fermentation by gut microbiota.

future science group www.futuremedicine.com 101 Review Angelakis, Armougom, Million & Raoult

components of diets, such as plant polysaccha- of CAZymes in gut microbes is highly diverse, rides [6,13,63] . The human gut microbiome plays exemplified by Bacteroides thetaiotaomicron , an essential role in the catabolism of dietary which contains 261 glycoside hydrolases and fibers, the part of plant material in the human polysaccharide lyases, as well as 208 homologs diet that is not metabolized by the upper of susC and susD genes, which code for two digestive tract, because the human genome outer membrane proteins that are involved in does not encode for an adequate carbohydrate starch utilization [65,66] . The CAZymes repre- active enzyme (CAZymes) (F IGURE 5) . Dietary sent, on average, 2.6% of the sequenced genes fibers are the components of vegetables, cere- in each microbiome [13] . As the human genome als, leguminous seeds, and fruits that are not encodes, at best, 20–25 digestive enzymes from digested in the stomach or in the small intes- CAZyme families (i.e., GH1 [lactase], GH13 tine. Instead, they are fermented in the colon [D-amylase] and GH31 [maltase, isomaltase by the gut microbiome and/or excreted in the and sucrase]), the ability to digest dietary plant feces. Additionally, dietary fibers have been carbohydrates resides entirely in gut microbi- identified as strong, positive dietary factors in omes [67] . The CAZymes represented in dif- the prevention of obesity [64] . The human gut ferent human populations that consume dif- bacteria produce a huge panel of CAZymes, ferent diets may be influenced by their varied with widely different substrate specificities, to cultural traditions. Hehemann et al. found that degrade these compounds into metabolizable porphyranase and agarase genes are specifically monosaccharides and disaccharides. The array encountered in Japanese gut bacteria and are

Table 2. Major bacteria and archaea in the human gut microbiota and their possible association with obesity. Representative phyla Class Genera Proven association with obesity Ref.

Bacteria Firmicutes Clostridia Clostridium Yes [54,55,56] Eubacterium Yes [55,59] Faecalibacterium Yes [60] Peptostreptococcus Ruminococcus Roseburia Yes [59]

Bacilli Lactobacillus Yes [10,49]

Enterococcus Yes [54] Staphylococcus Yes [52,58,62]

Bacteroidetes Bacteroidia Bacteroides Yes [52,54–56,62] Prevotella Xylanibacter Proteobacteria Deltaproteobacteria Desulfovibrio Gammaproteobacteria Escherichia Yes [62] Epsilonproteobacteria Helicobacter

Actinobacteria Actinobacteria Biﬁdobacterium Yes [51,52,58,62] Fusobacteria Fusobacteria Fusobacterium Synergistetes Synergistia Synergistes Spirochaetes Spirochaetes Treponema Verrucomicrobia Cyanobacteria Archaea

Euryarchaeota Methanobacteria Methanobrevibacter Yes [49–51,53] Methanobacteria Methanosphaera

102 Future Microbiol. (2012) 7(1) future science group Gut flora & weight gain Review probably absent in the microbiome of western Coriobacteriaceae individuals [68] . The authors proposed that consumption of sushi that contains algae from the Lactobacillus genus Porphyramay , which is associated with Enterococcus the marine bacteria Zobellia galactanivorans and Bacteroides plebeius , has been the route Faecalibacterium prausnitzii Obese individuals through which these CAZymes were acquired Prevotella in human gut bacteria [68,69] . Recently, Benjdia et al. hypothesized that Clostridium sulfatases are critical, evolved fitness factors Eubacterium [70] . To be active, sulfatases must undergo a critical post-translational modification that is Roseburia catalyzed in anaerobic bacteria by the radical Staphylococcus AdoMet enzyme, anaerobic sulfatase-maturat- Escherichia coli ing enzyme (anSME). They found that human gut Bacteroidetes possessed an anSME gene, and Methanobrevibacter several genes that encoded sulfatases were pres- Treponema ent within many species, including B. fragilis , Xylanibacter Lean individuals Bacteroides dorei or Parabacteroides distasonis Bacteroides [70] . On the other hand, Firmicutes did not possess genes encoding predicted sulfatases, and Bifidobacterium it was proposed that this demonstrated that sulfatases were an important and evolution- Figure 6. Population of bacteria found to increase in obese and lean ary conserved feature among Bacteroidetes that individuals. inhabited the human digestive tract [70,71] . higher gonadal fat content than germ-free mice, Gut flora of twins even though they consumed less food than their Turnbaugh et al. compared the fecal microbial germ-free counterparts [73] . When the microbi- communities of young, adult female monozy- ota of normal mice were transplanted into gno- gotic and dizygotic twin pairs, who were either tobiotic mice, there was a 60% increase in body lean or obese, along with those of their moth- fat within 2 weeks without any increase in food ers, to assess the gut microbiota relationship to consumption or obvious differences in energy host weight. Comparisons between all partici- expenditure [73] . Moreover, in a separate study pants showed that obesity was associated with using genetically modified (fasting-induced reduced bacterial diversity and a reduced rep- adipocyte factor [Fiaf]) knockout mice, the resentation of the Bacteroidetes [13] . In a more same authors showed that gut microbes sup- recent study, they found that the majority of press intestinal Fiaf. Fiaf suppression resulted species-level phylotypes were shared between in increased lipoprotein lipase activity in adi- deeply sampled monozygotic twins, despite pocytes and promoted storage of calories as fat. large variations in the abundance of each phy- These findings suggested that the gut micro- lotype [72] . From the gene clusters present in biota could affect both sides of the energy bal- their microbiome bins, only 17% were shared ance equation, influencing energy harvest from between the two co-twins. Bins exhibited dietary substances (Fiaf) and affecting genes differences in their degree of sequence varia- that regulate how energy is expended and stored tion, gene content, including the repertoire of [74] . Turnbaugh et al. were the first to determine carbohydrate active enzymes present within, that differences in the microbial community and between twins (e.g., predicted cellulases, could be a factor for obesity [75] . They found dockerins) and transcriptional activities [72] . that transfer of the gut microbiota from obese (ob/ob) mice to germ-free, wild-type recipients Gnotobiotic mice for the ana lysis of led to an increase in fat mass in the recipients. human gut microbes This led to speculation that the gut microbiota Germ-free mice provide a complementary promoted obesity by increasing the capacity approach for characterizing the properties of of the host to extract energy (calories) from the human gut microbiome. Backhed et al. ingested food [75] . Controlled diet manipula- found that young, conventionally reared mice tion in gnotobiotic mice, which were colonized have a 40% higher body fat content and 47% with a complete human gut (fecal) microbiota,

future science group www.futuremedicine.com 103 Review Angelakis, Armougom, Million & Raoult

revealed that the composition of their human a fecal community [80] . Colonization of germ- gut microbial communities changed dramati- free mice that consumed a plant polysaccharide- cally within a single day after the animals were rich or a simple sugar diet with wild-type or switched from a plant polysaccharide-rich anSME-deficient strains revealed that active chow to a high-fat, high-sugar ‘‘western’’ diet sulfatase production by B. thetaiotaomicron [14] . Goodman et al. developed an approach was essential for competitive colonization of the called insertion-sequencing (INSeq), which gut, especially when the organism was forced is based on a mutagenic transposon that cap- to adaptively forage on host mucosal glycans tures adjacent chromosomal DNA to define its because complex dietary polysaccharides were genomic location [76] . In this approach, complex not available [70] . The authors proposed that populations of tens of thousands of transposon anSME activity and the subsequent activation mutants are simultaneously introduced into of sulfatases represented an important pathway wild-type or genetically manipulated, germ- that allowed this Bacteroidetes species to adapt free mice in the presence or absence of other to life in the gut [70] . Fleissner et al. showed microbes. Using this assay, they discovered that that changes in energy expenditure rather than B. thetaiotaomicron employed the products of “energy harvest” were responsible for changes five adjacent genes ( BT1957–49 ) in response to in fat deposition and weight gain in mice as variations in vitamin B12 levels [76] . Moreover, they found no difference in body weight gain mice colonized with complete or cultured fecal between germ-free and conventional mice fed communities from two human donors displayed a semi-synthetic low-fat diet [81] . By contrast, significantly greater fat pad to body weight germ-free mice gained more body weight and ratios than germ-free controls [77] . Notably, body fat than conventional mice on a high- 18 species-level phylotypes were significantly fat diet. Moreover they found that the pro- affected when these gnotobiotic mice received portion of Firmicutes increased in both mice a western diet for 2 weeks. Specifically, the rela- high-fat and on a western diet. This increase tive proportion of representatives of one class of was mainly due to the proliferation of the Firmicutes (the Erysipilotrichi ) was increased, Erysipelotrichaceae [81] . Murphy et al. treated and the relative proportion of the Bacteroidia ob/ob mice with a low-fat diet and wild-type class was decreased [77] . Hildebrandt et al. found mice with either a low-fat diet or a high-fat diet that both wild-type and RELM E knockout and found that the proportions of Firmicutes , mice were lean on a standard chow diet, but Bacteroidetes and Actinobacteria did not corre- upon switching to a high-fat diet, the wild-type late with energy harvesting markers [82] . Higher mice became obese, whereas RELM E knockout concentrations of taurine-conjugated bile acids mice remained comparatively lean [78] . After the were identified in the livers and intestines of switch to the high-fat diet, the proportions of germ-free mice [83] and in those colonized by Proteobacteria, Firmicutes and Actinobacteria human baby microbiota [84] compared with con- increased, whereas the levels of Bacteriodetes ventional animals. Historically, bile acids have decreased [78] . When adult, germ-free, male been primarily viewed as detergent molecules mice were colonized with Marvinbryantella for- important for the absorption of dietary fats and matexigens and B. thetaiotaomicron , it was found lipidsoluble vitamins in the small intestine and that B. hydrogenotrophica targeted aliphatic and the m aintenance of c holesterol homeostasis in aromatic amino acids and increased the effi- the liver [83] . ciency of fermentation by consuming reducing equivalents, thereby maintaining a high NAD +/ Conclusion NADH ratio and boosting acetate production Obese and lean subjects presented increased [79] . By contrast, M. formatexigens consumed oli- levels of different bacterial populations (T ABLE 2 gosaccharides, did not impact the redox state & F IGURE 6) . In addition, a caloric diet restriction of the gut and boosted the yield of succinate impacted the composition of the gut microbiota [79] . Normalized RNA-Seq counts, generated in obese/overweight individuals and weight loss from the cecal contents and fecal samples of [5,55,56] . Interestingly, the initial microbiota of the mice revealed that prophages in M. for- overweight adolescents, before any treatment, matexigens were completely activated and that drove the efficiency of weight loss [56] , and two gene pairs were constitutively expressed differences in the gut composition at infancy in all fecal and cecal samples [80] . The authors could lead to weight gain [23,58] . Studies using proposed that a prophage might be liberated gnotobiotic mice have shown that the gut from its host cell when that cell is present in microbiota was critical for normal digestion of

104 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

nutrients [74] . It was proposed that the meta- per gram of feces [49] . Indeed, the characteriza- bolic activities of the gut microbiota facilitated tion of the 10 11 bacterial copies per gram of feces the extraction of calories from ingested dietary that was used in these studies remains superfi- substances, helped to store these calories in host cial. The use of FISH and qPCR methods were adipose tissue for later use and provided energy dependent on both sensitivity and specificity and nutrients for microbial growth and prolifer- of the targeted bacterial group. Additionally, ation [85] . A more recent hypothesis is based on the Bac303 probe, which was used in most of data from vegetarian human populations who the FISH- and qPCR-based studies [55,57–59] , presented bacteria that were commonly found underestimated the Bacteroidetes proportions in plants, like B. thetaiotaomicron , which pro- because the probe targeted only the Bacteroides- duced CAZymes and metabolized monosaccha- Prevotella groups, and it was inadequately sen- rides and disaccharides [6,13,62] . Moreover, it was sitive to the Prevotella group [92] . Ley et al. sug- predicted that other unknown factors in the gested that it will be interesting to study and microbiota and, recently, the manipulation of compare the effects of these molecular methods gut microbial with probiotics, prebiotics, anti- using the same sample stool [12] . An integration biotics or other interventions, were factors for of mechanistically based investigations and weight gain and obesity [1,86,87] , which should microbial ecology studies using high-through- be investigated more [88,89] . These results sug- put sequencing will provide insights into how gest that manipulating the composition of the to best reshape host–microbial interactions to gut microbiota may prevent weight gain or promote weight loss. facilitate weight loss in humans. Food is a source of bacteria and viruses, and changes in patterns of food consumption Future perspective results in differences in human gut flora among During the last few years, an increasing num- different groups of people. A question being ber of studies have related imbalances in the investigated is whether it is important to iden- composition of the gut microbiota to obesity. tify the source of the gut microorganisms as Many studies have reported shifts in the relative the most are ingested with food, drinks, and in abundances of bacterial communities in the gut the course of physical contact and interhuman microbiota of obese relative to normal-weight relationships. Data from agriculture, laboratory individuals, and each study has attempted to animals and humans show that manipulating link obesity with a species- or genus-specific gut microbiota results in weight modifications composition profile of the gut microbiota. and, recently, it was proposed that is neces- However, it is possible that the design and/or sary to further investigate the effects of rou- interpretation of the results has been affected tinely adding high amounts of bacteria to food by a conflict of interest of each team. It has [1,86,87] . In the last few years, the number of recently been shown that published papers in published descriptions of the organisms and nutrition and obesity research in which the genes that comprise and manipulate the gut authors were funded by industry were more microbiota is increasing dramatically, but these likely than other papers to contain results or studies have so far been limited to fairly small interpretations that favored the industry or populations. Moreover, little effort has been company that was producing the product or made to standardize the microbiota ana lysis service that was being studied [90] . Moreover, methodology and different sample collection, the heterogeneous methods that were utilized storage and ana lysis methods have only been in individual microbiota studies to estimate superficially investigated in human studies. bacterial proportions prevented rational com- This makes it almost impossible to directly parisons of results [12] . Notably, 16S rRNA compare findings from different groups, lim- sequencing-based methods are biased by the iting our ability to generalize findings. Further heterogeneity of the copy number of the 16S well-designed studies should be conducted rRNA gene that is present in an individual bac- into how gut microbial communities normally terial genome [91] and can lead to an overesti- operate, how they shape host physiology, and mation of bacterial proportions. However, it is how they may be altered by probiotic, prebi- noteworthy that the current 16S rDNA pyrose- otic, antibiotic or other interventions. For that quencing [53] , as well as clonal, Sanger sequenc- reason, massive parallel sequencing technolo- ing, studies [5] of gut microbiota within obese gies and the necessary bioinformatics tools to populations were not able to detect bacterial handle the resulting large datasets should be concentrations that were below 10 7 organisms adapted for human microbiota ana lysis.

future science group www.futuremedicine.com 105 Review Angelakis, Armougom, Million & Raoult

Financial & competing interests disclosure This includes employment, consultancies, honoraria, The authors have no relevant affiliations or financial stock ownership or options, expert testimony, grants or involvement with any organization or entity with a patents received or pending, or royalties. financial interest in or financial conflict with the sub- No writing assistance was utilized in the production ject matter or materials discussed in the manuscript. of this manuscript.

Executive summary Human gut microbiota The gut microbiota harbors approximately 10 11–12 microorganisms per gram of content. At birth, humans are essentially free of bacteria and over time, in a process of colonization that begins shortly after delivery, the body becomes a host to complex microbial communities. 16S rDNA pyrosequencing ana lysis revealed that geographical origin has an important impact on the intestinal microbiota. Dietary habits are considered to be one of the main factors that contribute to the diversity of the human gut microbiota. Bacteria species & obesity Meta-analyses revealed no difference in the Bacteroidetes concentrations between obese and humans of normal weight. Meta-analyses revealed that obese subjects present less Firmicutes than nonobese subjects in their gut flora. Meta-analyses revealed that obese subjects presented less Bifidobacteria than nonobese subjects. Meta-analyses revealed that obese subjects presented less Methanobrevibacter spp. than nonobese subjects. Ability to process polysaccharides The gut microbiota plays an essential role in the catabolism of dietary fibers into metabolizable monosaccharides and disaccharides by adequate carbohydrate active enzymes. Dietary fibers have been identified as strong, positive dietary factors in the prevention of obesity. The human gut bacteria produce a huge panel of carbohydrate active enzymes to degrade dietary fibers into metabolizable monosaccharides and disaccharides. Gnotobiotic mice for the ana lysis of human gut microbes Germ-free mice provide a complementary approach for characterizing the properties of the human gut microbiota. It was first demonstrated in experimental mice models that that differences in the gut microbiota could be a factor for obesity. Conclusion Microbial changes in the human gut are one of the possible causes of obesity.

References metagenomic sequencing. Nature 464(7285), 13. Turnbaugh PJ, Hamady M, Yatsunenko T Papers of special note have been highlighted as: 59–65 (2010). et al. A core gut microbiome in obese and of interest 7. Holmes E, Li JV, Athanasiou T, Ashrafian H, lean twins. Nature 457(7228), 480–484 of considerable interest Nicholson JK. Understanding the role of gut (2009). 1. Raoult D. Obesity pandemics and the microbiome-host metabolic signal disruption 14. Turnbaugh PJ, Ridaura VK, Faith JJ, Rey FE, modification of digestive bacterial flora. Eur. in health and disease. Trends Microbiol. Knight R, Gordon JI. The effect of diet on J. Clin. Microbiol. Infect. Dis. 27(8), 631–634 19(7), 349–359 (2011). the human gut microbiome: a metagenomic (2008). 8. Sekirov I, Russell SL, Antunes LC, Finlay ana lysis in humanized gnotobiotic mice. Sci. Transl. Med. 1(6), 6ra14 (2009). 2. Vasilakopoulou A, le Roux CW. Could a virus BB. Gut microbiota in health and disease. contribute to weight gain? Int. J. Obes. Physiol. Rev. 90(3), 859–904 (2010). One of the studies showing that gut (Lond.). 31(9), 1350–1356 (2007). 9. Barnich N, Darfeuille-Michaud A. Role of transplantation can lead to increased 3. Farooqi S, O’Rahilly S. Genetics of obesity in bacteria in the etiopathogenesis of adiposity, establishing the causal link humans. Endocr. Rev. 27(7), 710–718 (2006). inflammatory bowel disease. World between gut microbiota and obesity. J. Gastroenterol. 13(42), 5571–5576 (2007). 4. Speliotes EK, Willer CJ, Berndt SI et al. 15. Ventura M, Sozzi T, Turroni F, Matteuzzi D, Association analyses of 249,796 individuals 10. Larsen N, Vogensen FK, van den Berg FW van SD. The impact of bacteriophages on reveal 18 new loci associated with body mass et al. Gut Microbiota in human adults with probiotic bacteria and gut microbiota index. Nat. Genet. 42(11), 937–948 (2010). Type 2 diabetes differs from non-diabetic diversity. Genes Nutr. 6(3), 205–207 (2010). adults. PLoS ONE 5(2), e9085 (2010). 5. Ley RE, Turnbaugh PJ, Klein S, Gordon JI. 16. Wang Z, Klipfell E, Bennett BJ et al. Gut Microbial ecology: human gut microbes 11. Li JV, Ashrafian H, Bueter M et al. Metabolic flora metabolism of phosphatidylcholine associated with obesity. Nature 444(7122), surgery profoundly influences gut microbial- promotes cardiovascular disease. Nature 1022–1023 (2006). host metabolic cross-talk. Gut 60(9), 472(7341), 57–63 (2011). 1214–1223 (2011). Pioneering study linking obesity and gut 17. Moore WE , Holdeman LV. Human fecal microbiota. 12. Ley RE. Obesity and the human microbiome. flora: the normal flora of 20 Japanese- Curr. Opin. Gastroenterol. 26(1), 5–11 Hawaiians. Appl. Microbiol. 27(5), 961–979 6. Qin J, Li R, Raes J et al. A human gut (2010). (1974). microbial gene catalogue established by

106 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

18. Eckburg PB, Bik EM, Bernstein CN et al. 31. Woodmansey EJ, McMurdo ME, Macfarlane 43. Perry GH, Dominy NJ, Claw KG et al. Diet Diversity of the human intestinal microbial GT, Macfarlane S. Comparison of and the evolution of human amylase gene flora. Science 308(5728), 1635–1638 (2005). compositions and metabolic activities of fecal copy number variation. Nat. Genet. 39(10), 19. Frank DN, St Amand AL, Feldman RA, microbiotas in young adults and in antibiotic- 1256–1260 (2007). Boedeker EC, Harpaz N, Pace NR. treated and non-antibiotic-treated elderly 44. Sokol H, Pigneur B, Watterlot L et al. Molecular-phylogenetic characterization of subjects. Appl. Environ. Microbiol. 70(10), Faecalibacterium prausnitzii is an antimicrobial community imbalances in human 6113–6122 (2004). inflammatory commensal bacterium inflammatory bowel diseases. Proc. Natl Acad. 32. Mueller S, Saunier K, Hanisch C et al. identified by gut microbiota ana lysis of Sci. USA 104(34), 13780–13785 (2007). Differences in fecal microbiota in different Crohn disease patients. Proc. Natl Acad. Sci. 20. Dridi B, Henry M, El Khechine A, Raoult D, European study populations in relation to age, USA 105(43), 16731–16736 (2008). Drancourt M. High prevalence of gender, and country: a cross-sectional study. 45. Liszt K, Zwielehner J, Handschur M, Hippe Methanobrevibacter smithii and Appl. Environ. Microbiol. 72(2), 1027–1033 B, Thaler R, Haslberger AG. Methanosphaera stadtmanae detected in the (2006). Characterization of bacteria, clostridia and human gut using an improved DNA detection 33. Zwielehner J, Liszt K, Handschur M, Lassl C, Bacteroides in faeces of vegetarians using protocol. PLoS ONE 4(9), e7063 (2009). Lapin A, Haslberger AG. Combined qPCR and PCR-DGGE fingerprinting. Ann. 21. Dridi B, Raoult D, Drancourt M. Archaea as PCR-DGGE fingerprinting and quantitative- Nutr. Metab. 54(4), 253–257 (2009). emerging organisms in complex human PCR indicates shifts in fecal population sizes 46. Hayashi H, Sakamoto M, Benno Y. Fecal microbiomes. Anaerobe 17(2), 56-63 (2011). and diversity of Bacteroides , bifidobacteria and microbial diversity in a strict vegetarian as Clostridium cluster IV in institutionalized 22. Palmer C, Bik EM, Digiulio DB, Relman DA, determined by molecular ana lysis and elderly. Exp. Gerontol. 44(6–7), 440–446 Brown PO. Development of the human infant cultivation. Microbiol. Immunol. 46(12), (2009). intestinal microbiota. PLoS Biol. 5(7), e177 819–831 (2002). (2007). 34. Claesson MJ, Cusack S, O’Sullivan O et al. 47. Walker AW, Ince J, Duncan SH et al. Composition, variability, and temporal 23. Vael C, Verhulst SL, Nelen V, Goossens H, Dominant and diet-responsive groups of stability of the intestinal microbiota of the Desager KN. Intestinal microflora and body bacteria within the human colonic elderly. Proc. Natl Acad. Sci. USA mass index during the first three years of life: microbiota. ISME J. 5(2), 220–230 (2011). 108(Suppl. 1), S4586–S4591 (2011). an observational study. Gut Pathog. 3(1), 8 48. Wu GD, Chen J, Hoffmann C et al. Linking (2011). 35. Rutili A, Canzi E, Brusa T, Ferrari A. long-term dietary patterns with gut Intestinal methanogenic bacteria in children of 24. Hopkins MJ, Sharp R, Macfarlane GT. Age microbial enterotypes. Science 334(6052), different ages. New Microbiol. 19(3), 227–243 and disease related changes in intestinal 105–108 (2011). (1996). bacterial populations assessed by cell culture, 49. Armougom F, Henry M, Vialettes B, Raccah 16S rRNA abundance, and community 36. Dicksved J, Floistrup H, Bergstrom A et al. D, Raoult D. Monitoring bacterial cellular fatty acid profiles. Gut 48(2), Molecular fingerprinting of the fecal community of human gut microbiota reveals 198–205 (2001). microbiota of children raised according to an increase in Lactobacillus in obese patients different lifestyles. Appl. Environ. Microbiol. 25. Enck P, Zimmermann K, Rusch K, Schwiertz and Methanogens in anorexic patients. PLoS 73(7), 2284–2289 (2007). A, Klosterhalfen S, Frick JS. The effects of ONE 4(9), e7125 (2009). maturation on the colonic microflora in 37. Lay C, Rigottier-Gois L, Holmstrom K et al. 50. Million M, Maraninchi M, Henry M, infancy and childhood. Gastroenterol. Res. Colonic microbiota signatures across five Armougom F, Raoult D. Obesity-associated Pract. 752401 (2009). northern European countries. Appl. Environ. Gut microbiota is enriched in Lactobacillus Microbiol. 71(7), 4153–4155 (2005). 26. Agans R, Rigsbee L, Kenche H, Michail S, reuteri and depleted in Bifidobacterium Khamis HJ, Paliy O. Distal gut microbiota of 38. Fallani M, Amarri S, Uusijarvi A et al. animalis and Methanobrevibacter smithii . Int. adolescent children is different from that of Determinants of the human infant intestinal J. Obesity doi:10.1038/ijo.2011.153 (2011) adults. FEMS Microbiol. Ecol. 77(2), 404–412 microbiota after introduction of first (Epub ahead of print). (2011). complementary foods in five European centres. 51. Schwiertz A, Taras D, Schafer K et al. Microbiology 157(Pt 5),1385–1392 (2011). 27. Zoetendal EG, Akkermans AD, de Vos WM. Microbiota and SCFA in lean and Temperature gradient gel electrophoresis 39. De Filippo C, Cavalieri D, Di PM et al. overweight healthy subjects. Obesity (Silver ana lysis of 16S rRNA from human fecal Impact of diet in shaping gut microbiota Spring) 18(1), 190–195 (2010). samples reveals stable and host-specific revealed by a comparative study in children 52. Collado MC, Isolauri E, Laitinen K, communities of active bacteria. Appl. Environ. from Europe and rural Africa. Proc. Natl Acad. Salminen S. Distinct composition of gut Microbiol. 64(10), 3854–3859 (1998). Sci.USA 107(33), 14691–14696 (2010). microbiota during pregnancy in overweight 28. Tap J, Mondot S, Levenez F et al. Towards the 40. Li M, Wang B, Zhang M et al. Symbiotic gut and normal-weight women. Am. J. Clin. human intestinal microbiota phylogenetic microbes modulate human metabolic Nutr. 88(4), 894–899 (2008). core. Environ. Microbiol. 11(10), 2574–2584 phenotypes. Proc. Natl Acad. Sci.USA 105(6), 53. Zhang H, DiBaise JK, Zuccolo A et al. (2009). 2117–2122 (2008). Human gut microbiota in obesity and after 29. Turroni F, Foroni E, Pizzetti P et al. Exploring 41. Arumugam M, Raes J, Pelletier E et al. gastric bypass. Proc. Natl Acad. Sci. USA the diversity of the bifidobacterial population Enterotypes of the human gut microbiome. 106(7), 2365–2370 (2009). in the human intestinal tract. Appl. Environ. Nature 473(7346), 174–180 (2011). 54. Zuo HJ, Xie ZM, Zhang WW et al. Microbiol. 75(6), 1534–1545 (2009). 42. Backhed F, Ley RE, Sonnenburg JL, Peterson Gut bacteria alteration in obese people and 30. Woodmansey EJ. Intestinal bacteria and DA, Gordon JI. Host-bacterial mutualism in its relationship with gene polymorphism. ageing. J. Appl. Microbiol. 102(5), 1178–1186 the human intestine. Science 307(5717), World J. Gastroenterol. 17(8), 1076–1081 (2007). 1915–1920 (2005). (2011).

future science group www.futuremedicine.com 107 Review Angelakis, Armougom, Million & Raoult

55. Nadal I, Santacruz A, Marcos A et al. Shifts 65. Cantarel BL, Coutinho PM, Rancurel C, that is one of the putative mechanisms for in clostridia , bacteroides and Bernard T, Lombard V, Henrissat B. The gut microbiota-associated obesity. immunoglobulin-coating fecal bacteria carbohydrate-active enzymes database 76. Goodman AL, McNulty NP, Zhao Y et al. associated with weight loss in obese (CAZy): an expert resource for Identifying genetic determinants needed to adolescents. Int. J. Obes. (Lond.) 33(7), glycogenomics. Nucleic Acids Res. establish a human gut symbiont in its habitat. 758–767 (2009). 37(Database issue), D233–D238 (2009). Cell Host Microbe 6(3), 279–289 (2009). 56. 66. Santacruz A, Marcos A, Warnberg J et al. Martens EC, Koropatkin NM, Smith TJ, 77. Goodman AL, Kallstrom G, Faith JJ et al. Interplay between weight loss and gut Gordon JI. Complex glycan catabolism by Extensive personal human gut microbiota microbiota composition in overweight the human gut microbiota: the Bacteroidetes culture collections characterized and adolescents. Obesity (Silver Spring) 17(10), Sus-like paradigm. J. Biol. Chem. 284(37), manipulated in gnotobiotic mice. Proc. Natl 1906–1915 (2009). 24673–24677 (2009). Acad. Sci. USA 108(15), 6252–6257 (2011). 57. 67. Mai V, McCrary QM, Sinha R, Glei M. Turnbaugh PJ, Henrissat B, Gordon JI. 78. Hildebrandt MA, Hoffmann C, Sherrill-Mix Associations between dietary habits and Viewing the human microbiome through SA et al. High-fat diet determines the body mass index with gut microbiota three-dimensional glasses: integrating composition of the murine gut microbiome composition and fecal water genotoxicity: an structural and functional studies to better independently of obesity. Gastroenterology observational study in African American and define the properties of myriad carbohydrate- 137(5), 1716–1724 (2009). Caucasian American volunteers. Nutr. J. 8, active enzymes. Acta Crystallogr. Sect. F. 49 (2009). Struct. Biol. Cryst. Commun. 66(Pt 10), Proved that diet modifies gut microbiota independently of obesity. 58. Kalliomäki M, Collado MC, Salminen S, 1261–1264 (2010). Isolauri E. Early differences in fecal 68. Hehemann JH, Correc G, Barbeyron T, 79. Rey FE, Faith JJ, Bain J et al. Dissecting the microbiota composition in children may Helbert W, Czjzek M, Michel G. Transfer of in vivo metabolic potential of two human gut predict overweight. Am. J. Clin. Nutr. 87(3), carbohydrate-active enzymes from marine acetogens. J. Biol. Chem. 285(29), 534–538 (2008). bacteria to Japanese gut microbiota. Nature 22082–22090 (2010). 464(7290), 908–912 (2010). 80. Reyes A, Haynes M, Hanson N et al. Viruses Bifidobacteria protect children from becoming overweight. This is confirmed 69. Kurokawa K, Itoh T, Kuwahara T et al. in the faecal microbiota of monozygotic twins and their mothers. Nature 466(7304), by the fact that our meta-analysis found Comparative metagenomics revealed 334–338 (2010). that the obesity-associated gut microbiota commonly enriched gene sets in human gut are depleted in Bifidobacteria . microbiomes. DNA Res. 14(4), 169–181 81. Fleissner CK, Huebel N, Abd El-Bary MM, Bifidobacteria and Lactobacillus strains (2007). Loh G, Klaus S, Blaut M. Absence of are currently main candidates for 70. Benjdia A, Martens EC, Gordon JI, Berteau intestinal microbiota does not protect mice from diet-induced obesity. Br. J. Nutr. 104(6), antiobesity probiotics. O. Sulfatases and a radical AdoMet enzyme are key for mucosal glycan foraging and 919–929 (2010). 59. Duncan SH, Lobley GE, Holtrop G et al. fitness of a prominent human gut. Human colonic microbiota associated with Showed that modification of gut Bacteroides . J. Biol. Chem. 286(29), diet, obesity and weight loss. Int. J. Obes. microbiota is not the only way for diet to 25973–25982 (2011). (Lond.) 32(11), 1720–1724 (2008). induce obesity since germ-free mice are not 71. Berteau O, Guillot A, Benjdia A, Rabot S. A protected against diet-induced obesity. 60. Balamurugan R, George G, Kabeerdoss J, new type of bacterial sulfatase reveals a novel Diet, gut microbiota and obesity are Hepsiba J, Chandragunasekaran AM, maturation pathway in prokaryotes. J. Biol. Ramakrishna BS. Quantitative differences in associated by a triangular causal link. Chem. 281(32), 22464–22470 (2006). intestinal Faecalibacterium prausnitzii in 82. Murphy EF, Cotter PD, Healy S et al. obese Indian children. Br. J. Nutr. 103(3), 72. Turnbaugh PJ, Quince C, Faith JJ et al. Composition and energy harvesting capacity 335–338 (2010). Organismal, genetic, and transcriptional of the gut microbiota: relationship to diet, variation in the deeply sequenced gut obesity and time in mouse models. Gut 61. Thuny F, Richet H, Casalta JP, Angelakis E, microbiomes of identical twins. Proc. Natl Habib G, Raoult D. Vancomycin treatment 59(12), 1635–1642 (2010). Acad. Sci. USA 107(16), 7503–7508 (2010). of infective endocarditis is linked with 83. Swann JR, Tuohy KM, Lindfors P et al. recently acquired obesity. PLoS ONE 5(2), 73. Backhed F, Ding H, Wang T et al. The gut Variation in antibiotic-induced microbial e9074 (2010). microbiota as an environmental factor that recolonization impacts on the host metabolic regulates fat storage. Proc. Natl Acad. Sci. phenotypes of rats. J. Proteome Res. 10(8), 62. Santacruz A, Collado MC, Garcia-Valdez L USA 101(44), 15718–15723 (2004). et al. Gut microbiota composition is 3590–3603 (2011). associated with body weight, weight gain and 74. Backhed F, Manchester JK, Semenkovich CF, 84. Martin FP, Dumas ME, Wang Y et al. A biochemical parameters in pregnant women. Gordon JI. Mechanisms underlying the top-down systems biology view of Br. J. Nutr. 104, 83–92 (2010). resistance to diet-induced obesity in microbiome-mammalian metabolic germ-free mice. Proc. Natl Acad. Sci. USA interactions in a mouse model. Mol. Syst. Biol. 63. Gloux K, Berteau O, El OH, Beguet F, 104(3), 979–984 (2007). Leclerc M, Dore J. A metagenomic 3, 112 (2007). E-glucuronidase uncovers a core adaptive 75. Turnbaugh PJ, Ley RE, Mahowald MA, 85. DiBaise JK, Zhang H, Crowell MD, function of the human intestinal Magrini V, Mardis ER, Gordon JI. An Krajmalnik-Brown R, Decker GA, Rittmann microbiome. Proc. Natl Acad. Sci. USA obesity-associated gut microbiome with BE. Gut microbiota and its possible 108(Suppl. 1), S4539–S4546 (2011). increased capacity for energy harvest. Nature relationship with obesity. Mayo Clin. Proc. 444(7122), 1027–1031 (2006). 83(4), 460–469 (2008). 64. Grabitske HA , Slavin JL. Low-digestible carbohydrates in practice. J. Am. Diet. Assoc. Pioneering study linking gut microbiota 86. Raoult D. Probiotics and obesity: a link? Nat. 108(10), 1677–1681 (2008). and increased capacity for energy harvest Rev. Microbiol. 7, 619 (2009).

108 Future Microbiol. (2012) 7(1) future science group Gut ﬂora & weight gain Review

87. Raoult D. Human microbiome: take-home 90. Thomas O, Thabane L, Douketis J, Chu R, 92. Hoyles L, McCartney AL. What do we mean lesson on growth promoters? Nature Westfall AO, Allison DB. Industry funding when we refer to Bacteroidetes populations in 454(7205), 690–691 (2008). and the reporting quality of large long-term the human gastrointestinal microbiota? FEMS 88. Gordon JI , Klaenhammer TR. A rendezvous weight loss trials. Int. J. Obes. (Lond.) Microbiol. Lett. 299(2), 175–183 (2009). with our microbes. Proc. Natl Acad. Sci. USA 32(10), 1531–1536 (2008). 93. Borenstein M, Hedges L, Higgins J, Rothstein 108(Suppl. 1), S4513–S4515 (2011). 91. Hattori M , Taylor TD. The human H. Comprehensive Meta Analysis Version 2 . 89. Khoruts A , Sadowsky MJ. Therapeutic intestinal microbiome: a new frontier of Biostat, Englewood, NJ, USA (2005). transplantation of the distal gut microbiota. human biology. DNA Res. 16(1), 1–12 94. Borenstein M, Hedges L, Higgins J, Mucosal. Immunol. 4(1), 4–7 (2011). (2009). Rothstein H. Introduction to Meta-Analysis . Wiley, Hoboken, NJ, USA (2009).

future science group www.futuremedicine.com 109

Article III :

Obesity-associated gut microbiota is enriched in

Lactobacillus reuteri and depleted in Bifidobacterium

animalis and Methanobrevibacter smithii

Matthieu Million, Marie Maraninchi, Mireille Henry, Fabrice Armougom, Hervé Richet, Patrizia Carrieri, René Valero, Denis Raccah, Bernard Vialettes, Didier Raoult

Published in . Int J Obes (Lond). 2012 Jun;36(6):817-25. (IF 4.69)

International Journal of Obesity (2012) 36, 817–825 & 2012 Macmillan Publishers Limited All rights reserved 0307-0565/12 www.nature.com/ijo ORIGINAL ARTICLE Obesity-associated gut microbiota is enriched in Lactobacillus reuteri and depleted in Bifidobacterium animalis and Methanobrevibacter smithii

M Million 1, M Maraninchi 2, M Henry 1, F Armougom 1, H Richet 1, P Carrieri 3,4,5 , R Valero 2, D Raccah 6, B Vialettes 2 and D Raoult 1

1URMITE -CNRS UMR 6236 IRD 198, IFR 48, Faculte´ de Me ´decine, Universite´ de la Me ´diterrane´e, Marseille, France; 2Service de Nutrition, Maladies Me ´taboliques et Endocrinologie, UMR-INRA U1260, CHU de la Timone, Marseille, France; 3INSERM, U912(SE4S), Marseille, France; 4Universite´ Aix Marseille, IRD, UMR-S912, Marseille, France; 5ORS PACA, Observatoire Re ´gional de la Sante ´ Provence Alpes Co ˆte d’Azur, Marseille, France and 6Service de Nutrition et Diabe ´tologie, CHU Sainte Marguerite, Marseille, France

Background: Obesity is associated with increased health risk and has been associated with alterations in bacterial gut microbiota, with mainly a reduction in Bacteroidetes, but few data exist at the genus and species level. It has been reported that the Lactobacillus and Bifidobacterium genus representatives may have a critical role in weight regulation as an anti-obesity effect in experimental models and humans, or as a growth-promoter effect in agriculture depending on the strains. Objectives and methods: To confirm reported gut alterations and test whether Lactobacillus or Bifidobacterium species found in the human gut are associated with obesity or lean status, we analyzed the stools of 68 obese and 47 controls targeting Firmicutes, Bacteroidetes, Methanobrevibacter smithii, Lactococcus lactis, Bifidobacterium animalis and seven species of Lactobacillus by quantitative PCR (qPCR) and culture on a Lactobacillus-selective medium. Findings: In qPCR, B. animalis (odds ratio (OR) ¼ 0.63; 95% confidence interval (CI) 0.39–1.01; P ¼ 0.056) and M. smithii (OR ¼ 0.76; 95% CI 0.59–0.97; P ¼ 0.03) were associated with normal weight whereas Lactobacillus reuteri (OR ¼ 1.79; 95% CI 1.03–3.10; P ¼ 0.04) was associated with obesity. Conclusion: The gut microbiota associated with human obesity is depleted in M. smithii . Some Bifidobacterium or Lactobacillus species were associated with normal weight (B. animalis ) while others ( L. reuteri ) were associated with obesity. Therefore, gut microbiota composition at the species level is related to body weight and obesity, which might be of relevance for further studies and the management of obesity. These results must be considered cautiously because it is the first study to date that links specific species of Lactobacillus with obesity in humans. International Journal of Obesity (2012) 36, 817–825; doi:10.1038/ijo.2011.153; published online 9 August 2011

Keywords: gut microbiota; Methanobrevibacter smithii; Lactobacillus reuteri; Bifidobacterium animalis

Introduction predisposes children to adulthood obesity. 4 Its prevalence is increasing steadily among adults, adolescents and children, Obesity, defined as a body mass index (BMI) over 30 kg m À2 (ref. 1) and has doubled since 1960; and obesity is now considered and a massive expansion of fat, is related to a significantly a worldwide epidemic as, for example, over 30% of the increased mortality and is a risk factor for many diseases, population of North America is obese. The WHO data including diabetes mellitus, hypertension, respiratory disorders, indicate that obesity currently affects at least 400 million ischemic heart disease, stroke and cancer.2,3 Obesity can be people worldwide and 1.6 billion are overweight. The WHO considered as a transmissible disease because maternal obesity further projects that by 2015, B2.3 billion adults will be overweight and more than 700 million will be obese. 5 Correspondence: Professor D Raoult, Unite´ des Rickettsies, URMITE -CNRS The causes behind the obesity epidemic appear to be UMR 6236 IRD 198, IFR 48, Faculte´ de Me´decine, Universite´ de la complex and involve environmental, genetic, neural and Me´diterrane´e, 27 Bd Jean Moulin, Marseille, 13005 France. endocrine origins.6 E-mail: [email protected] Received 24 November 2010; revised 27 June 2011; accepted 2 July 2011; More recently, obesity has been associated with a specific published online 9 August 2011 profile of the bacterial gut microbiota, including a decrease Gut microbiota and obesity M Million et al 818 in the Bacteroidetes/Firmicutes ratio7–10 and a decrease in Caucasian and were approached in different geographical Methanobrevibacter smithii, the leading representative of the locations using a snowball approach. This approach was gut microbiota archaea.11 Since these pioneering studies, helpful in making the period of recruitment of cases and significant associations were found between the increase of controls comparable. The exclusion criteria included the some bacterial groups and obesity (Lactobacillus,12 Staphylo- following: non-assessable BMI value, BMIo19 kg m À2, coccus aureus,13–15 Escherichia coli,15 Faecalibacterium praus- BMI425 kg m À2 and o30 kg m À2, gastric bypass, history of nitzii16 ). Conversely, other groups have been associated with colon cancer, bowel inflammatory diseases, acute or chronic lean status, mainly belonging to the Bifidobacterium diarrhea in the previous 4 weeks and antibiotic administra- genus.11,13–16 To date, controversial studies make it clear tion o1 month before stool collection. Clinical data (gender, that the connection between the microbiome and excess date of birth, clinical history, weight, height and antibiotic weight is complex. 17 use) were recorded using a standardized questionnaire. The As many probiotic strains of Lactobacillus and Bifidobacterium samples, collected using sterile plastic containers, were are marketed in products for human consumption, altering transported as soon as possible to the laboratory and frozen the intestinal flora18 and stimulating indigenous lactobacilli immediately at À80 1C for later analysis. For Firmicutes, and bifidobacteria strains,19 we hypothesized that wide- Bacteroidetes, M. smithii and Lactobacillus species, analyses spread ingestion of probiotics may promote obesity by were first performed on the whole population and then after altering the intestinal flora.20–22 However, this remains exclusion of common subjects with our previous study. 12 controversial.23,24 In a first step to elucidate the interactions between probiotics for human consumption and obesity, only a few studies have compared the obese and lean subjects Analysis of gut microbiota by focusing on the Lactobacillus and Bifidobacterium genera Culture on specific Lactobacillus medium (LAMVAB at the species level 13,16 and they have not been able to medium). After thawing at room temperature, 100 mg of demonstrate significant differences probably because of a stool was suspended in 900 ml of cysteine–peptone–water too small sample size. As a result, by increasing the sample solution26 and homogenized. A serial dilution was under- size, we analyze the composition of the digestive microbiota taken in phosphate buffered saline. Samples diluted to 1/10 for Firmicutes, Bacteroidetes, the archaea M. smithii , and 1/1000 were inoculated using a 10 ml inoculation loop Lactobacillus genus, L. lactis , and explore the relationships on LAMVAB medium. 27 After a 72-hour incubation in jars between seven selected species of Lactobacillus and one (AnaeroPack, Mitsubishi Gas Chemical America, Inc., New species of Bifidobacterium, used elsewhere in marketed York, NY, USA) in an anaerobic atmosphere (GasPak EZ probiotics for human consumption and obesity. Anaerobe, Becton Dickinson, Heidelberg, Germany) at 37 1C, the number of morphotypes were identified and 1–4 colonies per morphotype were placed on four spots of an MTP 384 Target plate made of polished steel (Bruker Daltonics GmbH, Materials and methods Bremen, Germany) and stored in trypticase cases in soy culture medium (AES, Bruz, France). Ethics, participants and samples All aspects of the study were approved by the local ethics Lactobacillus strain collection and MALDI-TOF spectra committee ‘Comite´ d’e´thique de l’IFR 48, Service de database. The Lactobacillus strain collection of our labora- Me´decine Le´gale’ (Faculte´ de Me´decine, Marseille, France) tory has been completed by the strains from the Pasteur and under the accession number 10–002, 2010. Only verbal DSMZ collections, and reference spectra have been created consent was necessary from patients for this study. This is from those missing in the Bruker database. Bacterial identifica- according to the French bioethics decree Number 2007– tion was undertaken with an Autoflex II mass spectrometer 1220, published in the official journal of the French (Bruker Daltonik GmbH). Data were automatically acquired Republic. Obese patients, as defined by a BMI 430 kg m À2 using Flex control 3.0 and Maldi Biotyper Automation (BMI: weight over height squared (kg mÀ2)), were selected Control 2.0. (Bruker Daltonics GmbH). Raw spectra, ob- from two endocrinology units (Hopital La Timone and tained for each isolate, were analyzed by standard pattern Hopital Sainte Marguerite, Marseilles, France) from a group matching (with default parameter settings) against the of patients attending the clinic for excessive body weight. spectra of species used as a reference database. An isolate BMI provides the most useful population-level measure of was regarded as correctly identified at the species level when overweight and obese, as it is the same for both sexes and for at least one spectrum had a score X1.9, and one spectrum all ages of adults. 5 However, it may not correspond to the had a score X1.7.28 The reproducibility of the method was same degree of fatness in different individuals (The Y-Y evaluated by the duplicate analysis of 10 samples. paradox).25 Control subjects were healthy volunteers over 18 years of age with BMIs between 19 and 25 kg m À2. Only a few Quantitative real-time PCR for M. smithii , Bacteroidetes, patients had participated in the previous study conducted by Firmicutes and Lactobacillus genus. DNA was isolated from our laboratory. 12 The control subjects were predominantly stools as described in Dridi et al .29 The purified DNA samples

International Journal of Obesity Gut microbiota and obesity M Million et al 819 were eluted to a final volume of 100 ml and stored at À80 1C account possible confounders like age or gender, a logistic until analysis. Real-time PCR was performed on a Stratagene regression model was used. Variables with a liberal Po0.20 in MX3000 system (Agilent, Santa Clara, CA, USA) using the univariate logistic regression analysis were considered QuantiTect PCR mix (Qiagen, Courtaboeuf, France) as eligible for the multiple logistic regression analyses. 30 A described previously.12 secondary analysis based on logistic regression analysis was used to identify which culture variables ( Lactobacillus species concentration) where associated with obesity. Data analyses Quantitative real-time PCR specific for Lactococcus lactis, were conducted using SPSS v.9.0 (SPSS Inc., Chicago, IL, USA). Bifidobacterium animalis and seven Lactobacillus species. The primer and probe sequences were located on the Tuf (elongation factor Tu) gene. The Tuf gene from the Lactobacillus strains, reported in Supplementary Table 1, were Results sequenced and compared, where possible, to the sequence Patients reported in Genbank as described in Supplementary Text 1. In total, 115 subjects (68 obese patients and 47 controls) All of these sequences were compared by ClustalX (1.8; were included. Thirteen obese subjects and nine controls http://www.clustal.org) using global-multiple sequence were part of the previous study conducted in our labora- alignment by the progressive method. A distance is calcu- tory. 12 The two populations were homogeneous in sex and lated between every pair of sequences and these are used to height, but not in age (Table 1). construct the phylogenetic tree, which guides the final multiple alignment. The scores are calculated from separate pairwise alignments using the dynamic programming Culture method. A consensus sequence was obtained and compared In total, 68 obese and 44 controls samples were analyzed. with the Tuf sequences of Lactobacillus acidophilus, Lactoba- The number of positive samples was greater among the cillus casei-paracasei, Lactobacillus plantarum, Lactobacillus controls vs obese (32/44 vs 30/68, Fisher’s exact test, reuteri, Lactobacillus gasseri, Lactobacillus fermentum and P ¼ 0.002). For positive samples, the concentration was not Lactobacillus rhamnosus, Bifidobacterium animalis and Lacto- significantly different between obese subjects and controls, coccus lactis, and sequences of primers and probes of highly respectively (median 4.15 (interquartile range 4–6) vs 5.2 specific real-time PCR were established. The primer and (4–6) log10 CFU ml À1, Mann–Whitney test, P ¼ 0.93). The probe sequences are reported in Supplementary Table 2. The proportion (Table 2) and non-parametric quantitative com- Lactobacillus strain-specific detection proceeded in duplex parison of the concentration of Lactobacillus species between real-time PCR: L. acidophilus (FAM) and L. casei/paracasei obese subjects and controls has been achieved for the species (VIC), L. plantarum (FAM) and L. reuteri (VIC), L. gasseri (FAM) present in at least six individuals. L. paracasei was found and fermentum (VIC), and L. rhamnosus (FAM). B. animalis more frequently in controls (17/44 vs 10/68, Fisher’s exact (VIC) and Lactococcus lactis (FAM) detection utilized simplex test, P ¼ 0.004). L. reuteri was found more frequently in obese real-time PCR. The duplex real-time PCR was executed as patients (6/68 vs 1/44, Fisher’s exact test, P ¼ 0.15), although described above and in Armougom et al .12 The specificity was this was not significant. L. plantarum was found only in tested on the DNA of the reference strains reported in Supplementary table 1. The stool-purified DNA was analyzed in samples that were pure, diluted at 1/10, and diluted at 1/ 100 to confirm the absence of inhibitors. Negative controls Table 1 Baseline characteristics were included on each plate. The different lactobacilli, B. Obese (n ¼ 68) Controls ( n ¼ 47) P (obese vs controls) animalis and Lactococcus lactis were quantified using a ± ± a plasmid standard curve from 10 7–10 copies per assay. Age 50.5 14.4 42.6 17.5 0.01 Malesex 31(45.6%) 21(51.2%) 0.35 b Body mass index 43.6 ±7.8 22.1 ±1.8 o0.0001 a

aMann–Whitney test. bFisher’s exact test Statistical Analysis First, the results of Lactobacillus-specific culture and quantitative PCR (qPCR) were compared in the two groups (obese Table 2 Results of Lactobacillus-specific culture and control group) using the Fisher’s exact test when Obese ( n ¼ 68) Controls( n ¼ 44) P-valuea comparing proportions, and the Mann–Whitney test when L. paracasei 10(14.7%) 17(38.6%) 0.004 comparing bacterial concentrations. A difference was con- L. plantarum 0 (0%) 8 (18.2%) 0.0004 o sidered statistically significant when P 0.05. In order to L. reuteri 6 (8.8%) 1 (2.3%) 0.16 identify which qPCR bacterial groups ( Bacteroidetes, B. animalis, L. rhamnosus 3 (4.4%) 4 (9.1%) 0.27 Lactococcus lactis, L. acidophilus, L. casei/paracasei, L. fermentum, L. ruminis 3 (4.4%) 4 (9.1%) 0.27 L. salivarius 5 (7.4%) 2 (4.5%) 0.43 L. gasseri, L. plantarum, L. reuteri, L. rhamnosus) was most associated with the likelihood of being obese while taking into aSpecies present in at least six individuals. Fisher’s exact test.

International Journal of Obesity Gut microbiota and obesity M Million et al 820

Figure 1 Quantification of L. paracasei, L. plantarum and L. reuteri in culture (LAMVAB medium) Àlog (colony forming units per ml of feces) FMann–Whitney test.

Table 3 Results of Bacteroidetes, Firmicutes, Methanobrevibacter smithii and found after the exclusion of the common subjects from Lactobacillus genus quantitative PCR our previous study (Mann–Whitney test; higher level of Obese (n ¼ 67) Controls ( n ¼ 45) P Lactobacillus genus in obese people, P ¼ 0.026; lower level of M. smithii , P ¼ 0.008 and lower level of Bacteroidetes, a Presence of phyla, genus or species P ¼ 0.09). Bacteroidetes 41(61.2%) 27(60%) 0.52 Firmicutes 67(100%) 45(100%) F Lactobacillus 23(34.3%) 8(17.8%) 0.04 Methanobrevibacter smithii 50(74.6%) 40(88.9%) 0.05 Bifidobacterium–Lactococcus–Lactobacillus species-specific

À1 b qPCR Quantitative comparison (log copies DNAml ) Bacteroidetes 4.26(0–5.82) 5.65(0–6.37) 0.25 The different Bifidobacterium–Lactococcus–Lactobacillus spe- Firmicutes 6.43 (5.32–7.29) 6.62 (5.86–7.21) 0.30 cies-specific real-time PCRs were tested for their specificity Lactobacillus 0 (0–3.31) 0 (0–0) 0.039 against purified DNA of the strains reported in Supplemen- Methanobrevibacter smithii 2.31 (0–3.51) 3.78 (1.71–5.30) 0.002 tary Table 1. The different real-time PCR systems were tested aValues noted as number (percentage), Fisher’s exact test. bValues noted as for their sensitivity and we obtained a cycle threshold of À1 log copies DNAml , median (interquartile range), Mann–Whitney test. about 35 for 10 copies of DNA per 5 ml of sample. All of these real-time PCRs have good sensitivity and specificity (Supple- mentary Table 3). In total, 64 obese samples and 43 control controls (8/44 vs 0/68, Fisher’s exact test, P ¼ 0.0004). samples were analyzed. The presence of B. animalis was Quantitative comparison found higher levels of L. paracasei associated with normal weight (Table 4, Fisher’s exact test, and L. plantarum in controls (Mann–Whitney test, P ¼ 0.005 P ¼ 0.007), and L. reuteri was associated with obesity (Fisher’s and P ¼ 0.0004, respectively), while L. reuteri was higher in exact test, P ¼ 0.03). Comparison using non-parametric the obese subjects; however, this was not significant (Mann– statistics found that levels of B. animalis were lower Whitney test, P ¼ 0.14) (Figure 1). Variables eligible for the (Mann–Whitney test, P ¼ 0.004) and that of L. reuteri were final logistic regression model were L. paracasei , L. reuteri, higher in obese people (Mann–Whitney test, P ¼ 0.02) L. plantarum, L. brevis, L. fermentum and age. The final (Figure 3). By comparing the culture and the Lactobacillus multiple logistic regression model showed that after adjust- species-specific PCR, the sensitivity was higher for all seven ment for age, only L. paracasei was significantly associated tested species by PCR vs culture except for L. acidophilus, with lean status (odds ratio ¼ 0.79; 95% confidence interval which was not found by culture or species-specific PCR. 0.64–0.97; P ¼ 0.03). Overall, results of culture and PCR were consistent for the presence of L. casei/paracasei (Fisher’s exact test, P ¼ 0,017), L. plantarum (Fisher’s exact test, P ¼ 0,05) and L. reuteri Firmicutes, Bacteroidetes, M. smithii and Lactobacillus (Fisher’s exact test, P ¼ 0,00001). species-specific qPCR M. smithii was found more frequently in controls (40/ 45(89%) vs 50/67(75%), Fisher’s exact test, P ¼ 0.05). The Logistic regression analysis analysis did find a lower concentration of M. smithii in The results of the logistic regression analysis on the qPCR obese subjects (Mann–Whitney test, P ¼ 0.002; Table 3) and results are presented in Table 5. Variables eligible for the final a higher concentration of Lactobacillus (Mann–Whitney model were L. casei/paracasei , L. reuteri, L. gasseri, B. animalis, test, P ¼ 0.04). Bacteroidetes was found in lower concen- M. smithii and age. The final multiple logistic regression tration in obese, but this result was not significant (Mann– model showed that after adjustment for age, L. reuteri, Whitney test, P ¼ 0.25) (Figure 2). The same results were B. animalis and M. smithii were significantly associated with

International Journal of Obesity Gut microbiota and obesity M Million et al 821

Figure 2 Quantification of Bacteroidetes, Firmicutes, M. smithii and Lactobacillus genus by qPCR FMann–Whitney test.

Table 4 Results of Bifidobacterium animalis, Lactococcus lactis and seven Discussion Lactobacillus species-specific quantitative PCR

Obese ( n ¼ 64) Controls ( n ¼ 43) P To our knowledge, we report the largest case–control study comparing human obese gut microbiota to controls focusing a Presence of targeted taxa on Archaea, Bacteroidetes, Firmicutes, Lactobacillus genus, L. acidophilus 0 (0%) 0 (0%) F Lactococcus lactis and B. animalis and, for the first time, we L. casei/paracasei 24(37.5%) 24(55.8%) 0.047 L. fermentum 11(17.2%) 9(20.9%) 0.40 used a culture-dependent and culture-independent method L. gasseri 21(32.8%) 9(20.9%) 0.13 to compare the Lactobacillus population at the species level L. plantarum 14(21.9%) 12(27.9%) 0.31 between obese and normal-weighted humans. Our results L. reuteri 16(25.0%) 4(9.3%) 0.03 confirm global alteration in obese gut microbiota with a L. rhamnosus 11(17.2%) 9(20.9%) 0.40 Lactococcus lactis 55(85.9%) 34(79.1%) 0.25 lower level of M. smithii as already reported in the Bifidobacterium animalis 1 (1.6%) 7 (16.3%) 0.007 literature,11 and newly report lower levels of B. animalis, L. paracasei, L. plantarum and higher levels of L. reuteri in aValues expressed as number (percentage). Fisher’s exact test. obese gut microbiota. The qPCR system used in this study to detect and quantify obesity. L. reuteri was the only one which showed higher Bacteroidetes, Firmicutes, Lactobacillus genus and M. smithii in levels in obese individuals while B. animalis and M. smithii human feces has already been evaluated and validated.12,29 were found at greater levels in non-obese subjects. LAMVAB-selective media has also been used successfully to

International Journal of Obesity Gut microbiota and obesity M Million et al 822

Figure 3 Quantification of B. animalis , L. casei/paracasei, L. plantarum and L. reuteri by qPCR FMann–Whitney test.

Table 5 Factors associated with obesity based on multiple logistic regression targeting species associated with obesity or normal weight in (qPCR results, logistic regression analysis, n ¼ 107) our preliminary culture study, and targeting other species OR(95%CI) P -value present in marketed probiotics products as Lactococcus lactis and B. animalis . Species-specific Lactobacillus PCR based Lactobacillus reuteri 1.79(1.03–3.10) 0.04 on the Tuf gene and designed for this new study showed Bifidobacterium animalis 0.63(0.39–1.01) 0.056 good reproducibility, sensitivity and specificity. However, Methanobrevibacter smithii 0.76(0.59–0.97) 0.03 Age 1.05(1.01–1.08) 0.006 we found significant discrepancies between culture and Lactobacillus species-specific PCR species. First, L. gasseri Abbreviations: CI, confidence interval; OR, odds ratio; qPCR, quantitative PCR. and L. acidophilus could not be identified in culture due to the presence of vancomycin in the LAMVAB medium. identify and enumerate lactobacilli from human feces.27 As Conversely, although qPCR was much more sensitive than in our previous study, 12 we found an increase in Lactobacillus culture to detect selected species of Lactobacillus, we showed in obese patients using the same Lactobacillus genus-specific that the two methods were consistent for L. casei/paracasei , PCR system. However, we found that its sensitivity profile L. plantarum and L. reuteri . For these three Lactobacillus was heterogeneous among the Lactobacillus species found in species, both techniques resulted in the same effect direction human feces by culture (data not shown). We subsequently with human obesity gut microbiota enriched in L. reuteri, developed a novel Lactobacillus species-specific qPCR system and depleted in L. casei/paracasei and L. plantarum.

International Journal of Obesity Gut microbiota and obesity M Million et al 823 The decrease of Bacteroidetes was historically the first L. plantarum or L. rhamnosus .37,40 In vivo and in vitro analyses alteration significantly associated with obesity as reported of physiological modifications imparted by CLA on protein by Ley and Turnbaugh, 8 in mice and in North American and gene expression suggest that CLA exerts its delipidating individuals,7,9 and by Santacruz et al ., 15 who observed effects by modulating energy expenditure, apoptosis, fatty overweight pregnant women in Spain. We found the same acid oxidation, lipolysis, stromal vascular cell differentia- correlation in our previous study, 12 and the same effect tion and lipogenesis. 37 Authors who have investigated the direction in the present study with the same PCR system on mechanisms linking conjugated linoleic acid and anti- the whole population and after the exclusion of common obesity effects have reported the upregulated expression of subjects. Schwiertz et al .11 reported opposite results, but the genes encoding uncoupling proteins (UCP-2), which could methodology was objectionable because the Bacteroidetes be a primary mechanism through which CLA increases proportion was obtained by summing Bacteroides and energy expenditure and produces an anti-obesity effect. 40 Prevotella genera. Other studies found no interaction L. reuteri has been associated here with obesity. L. reuteri between the relative or absolute abundance of Bacteroidetes has been one of the most studied probiotic species especially and obesity.31–33 for its ability to inhibit the growth of other potentially In our previous study, 12 abundance of M. smithii was pathogenic microorganisms by secreting antibiotic sub- significantly higher in patients with anorexia but not in stances such as reuterin. 41 When introduced in pigs, turkeys lean controls. In this new study, we found that M. smithii and rats, L. reuteri led to a significant weight gain and was was less frequent and significantly less abundant in obese isolated in higher concentrations from feces after probiotic patients on the whole population and after the exclusion of administration.42–44 The mechanism by which L. reuteri is common subjects. Schwiertz et al .11 using a specific qPCR for able to support the healthy growth of these animals is not Methanobrevibacter species, found similar results in a German entirely understood. It is possible that L. reuteri simply serves population. These results are in contradiction to those of to protect livestock against illness caused by Salmonella Zhang et al .33 who found that Methanobacteriales was present typhimurium and other pathogens. However, other studies only in obese individuals using a qPCR but only three obese have revealed that L. reuteri can also help when the growth vs three controls were compared. depression is caused entirely by a lack of dietary protein and In this study, we report an association between lower levels not by contagious disease. 45 This raises the possibility that of B. animalis and obesity for the first time. Five studies L. reuteri somehow improves the intestines’ ability to absorb reported a decreased number of Bifidobacterium representatives and process nutrients, and increase food conversion.46 in the feces of obese subjects at the genus level. 11,13–16 At the As a theoretical basis for the causal link between the species level, Kalliomaki et al .13 using a Bifidobacterium species- gut microbiota alterations and obesity, several mechanisms specific PCR, found that Bifidobacterium longum and Bifidobac- have been suggested. First, the gut microbiota could inter- terium breve were higher in normal weight controls, but this act with weight regulation by hydrolysis of indigestible result was not significant probably because of a small sample polysaccharides to monosaccharides easily absorbable acti- size. Experimental data report that administration of a B. breve vating lipoprotein lipase. Consequently, glucose is rapidly strain to mice with high-fat diet-induced obesity led to a absorbed producing substantial elevations in serum glucose significant weight decrease. 34 Administering four different and insulin, both factors that trigger lipogenesis and fatty Bifidobacterium strains to high-fat diet induced obese rats, Yin acids excessively stored with de novo synthesis of triglycerides et al .35 reported that one strain increased body weight gain, derived from liver, these two phenomena causing weight another induced a decrease and the two other strains lead to no gain.47 Second, the composition of gut microbiota has been significant change in body weight but species were not shown to selectively suppress the angiopoietin-like protein mentioned in this study. In this way, Cani et al .36 reported 4/fasting-induced adipose factor in the intestinal epithelium, that high-fat feeding was associated with higher endotoxaemia known as a circulating lipoprotein lipase inhibitor and and lower Bifidobacterium species cecal content in mice. The regulator of peripheral lipid and glucose metabolism. 48 selective increase of bifidobacteria by oligofructose, improving Third, it has been suggested that bacterial isolates of gut mucosal barrier function, significantly and positively correlated microbiota may have pro- or anti-inflammatory properties, with improved glucose tolerance, glucose-induced insulin impacting weight as obesity, having been associated with a secretion and decreased endotoxaemia. low-grade systemic inflammation corresponding to higher L. plantarum and L. paracasei were associated with normal plasma endotoxin lipopolysaccharide concentrations defined weight in culture, consistent with experimental models in as metabolic endotoxaemia. 49–52 Fourth, extracting crude fat the literature reporting an anti-obesity effect of L. plantarum in feed and excreta, Nahashon et al .53 reported that feeding in mice. 37 Other Lactobacillus strains have shown an anti- laying Leghorn with Lactobacillus improved significantly obesity effect in animals and humans similar to the L. gasseri retention of fat with increased cellularity of the Peyer’s SBT2055 (LG2055) strain in lean Zucker rats 38 and in patches of the ileum, which indicated ileal immune response. humans.39 This anti-obesity effect may be linked to the Conversely, Bifidobacterium and Lactobacillus species have production of specific molecules that can interfere with been cited to deconjugate bile acids, which may decrease host metabolism, such as conjugated linoleic acid (CLA) for fat absorption.54

International Journal of Obesity Gut microbiota and obesity M Million et al 824 Finally, specific strains of Lactobacillus and Bifidobacterium References fed to farm animals have been shown to increase daily weight gain,55 and this fact has been used for decades in 1 Appropriate body-mass index for Asian populations and its agriculture to increase feed conversion. In this context, one implications for policy and intervention strategies. Lancet 2004; 363 : 157–163. cannot exclude that the ‘growth promoter’ effect in animals 2 Whitlock G, Lewington S, Sherliker P, Clarke R, Emberson J, associated with oral administration of specific probiotics Halsey J et al. Body-mass index and cause-specific mortality in strains is similar to the mechanisms involved in human 900 000 adults: collaborative analyses of 57 prospective studies. 373 obesity. For instance, Abdulrahim et al .56 reported that Lancet 2009; : 1083–1096. 3 Yanovski SZ, Yanovski JA. Obesity. N Engl J Med 2002; 346 : L. acidophilus significantly increased abdominal fat deposi- 591–602. tion in female chickens when administered alone and up to 4 Lawlor DA, Smith GD, O’Callaghan M, Alati R, Mamun AA, 31% when it was associated with zinc bacitracin. Further Williams GM et al. Epidemiologic evidence for the fetal over- studies are therefore mandatory in exploring the interactions nutrition hypothesis: findings from the mater-university study of pregnancy and its outcomes. Am J Epidemiol 2007; 165 : 418–424. between probiotics and weight regulation. 5 World health organization. Obesity and overweight. Fact sheet N1311. 2011. 6 Tilg H, Moschen AR, Kaser A. Obesity and the microbiota. 136 Conclusion Gastroenterology 2009; : 1476–1483. 7 Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, Ley RE et al. A core gut microbiome in obese and lean twins. In conclusion, reduced levels of M. smithii has been Nature 2009; 457 : 480–484. confirmed as being associated with obesity. In addition, 8 Ley RE, Backhed F, Turnbaugh P, Lozupone CA, Knight RD, Gordon JI. Obesity alters gut microbial ecology. Proc Natl Acad Sci higher levels of B. animalis, L. paracasei or L. plantarum were USA 2005; 102 : 11070–11075. associated with a normal weight whereas higher levels of 9 Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: L. reuteri were associated with obesity, suggesting a human gut microbes associated with obesity. Nature 2006; 444 : possible interrelationship between certain probiotic species, 1022–1023. 10 Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, marketed elsewhere for human consumption, and obesity. Gordon JI. An obesity-associated gut microbiome with increased These results must be considered cautiously because it is the capacity for energy harvest. Nature 2006; 444 : 1027–1031. first study to date that links specific species of Lactobacillus 11 Schwiertz A, Taras D, Schafer K, Beijer S, Bos NA, Donus C et al. with obesity in humans. This issue will be of critical Microbiota and SCFA in lean and overweight healthy subjects. 18 importance in the management of the twenty-first century Obesity (Silver Spring) 2010; : 190–195. 12 Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. worldwide epidemic that is obesity and especially consider- Monitoring bacterial community of human gut microbiota ing the booming market of probiotics. reveals an increase in Lactobacillus in obese patients and Methanogens in anorexic patients. PLoS One 2009; 4: e7125. 13 Kalliomaki M, Collado MC, Salminen S, Isolauri E. Early differences in fecal microbiota composition in children may Conflict of interest predict overweight. Am J Clin Nutr 2008; 87 : 534–538. 14 Collado MC, Isolauri E, Laitinen K, Salminen S. Distinct The authors declare no conflict of interest. composition of gut microbiota during pregnancy in overweight and normal-weight women. Am J Clin Nutr 2008; 88 : 894–899. 15 Santacruz A, Collado MC, Garcia-Valdes L, Segura MT, Martin- Lagos JA, Anjos T et al. Gut microbiota composition is associated Acknowledgements with body weight, weight gain and biochemical parameters in pregnant women. Br J Nutr 2010; 104 : 83–92. 16 Balamurugan R, George G, Kabeerdoss J, Hepsiba J, Chandragu- We thank all the volunteers without whom this study would nasekaran AM, Ramakrishna BS. Quantitative differences in not have been possible. intestinal Faecalibacterium prausnitzii in obese Indian children. Br J Nutr 2010; 103 : 335–338. 17 Pennisi E. Microbiology. girth and the gut (bacteria). Science 2011; 332 : 32–33. Author contributions 18 Fujimoto J, Matsuki T, Sasamoto M, Tomii Y, Watanabe K. Identification and quantification of Lactobacillus casei strain Shirota in human feces with strain-specific primers derived from Conceived and designed the experiments: DR. Performed the randomly amplified polymorphic DNA. Int J Food Microbiol 2008; clinical study: MM, MM, RV, BV and DR. Performed the 126 : 210–215. experiments: MM and MH. Analyzed the data: FA, HR and 19 Ohashi Y, Inoue R, Tanaka K, Matsuki T, Umesaki Y, Ushida K. PC. Wrote the paper: MM, MH, HR and DR. Lactobacillus casei strain Shirota-fermented milk stimulates indigenous lactobacilli in the pig intestine. J Nutr Sci Vitaminol (Tokyo) 2001; 47 : 172–176. 20 Raoult D. Obesity pandemics and the modification of digestive Disclaimer bacterial flora. Eur J Clin Microbiol Infect Dis 2008; 27 : 631–634. 21 Raoult D. Human microbiome: take-home lesson on growth promoters? Nature 2008; 454 : 690–691. The funders had no role in study design, data collection and 22 Raoult D. Probiotics and obesity: a link? Nat Rev Microbiol 2009; analysis, decision to publish or preparation of the manuscript 7: 616.

International Journal of Obesity Gut microbiota and obesity M Million et al 825 23 Delzenne N, Reid G. No causal link between obesity and 41 Talarico TL, Casas IA, Chung TC, Dobrogosz WJ. Production and probiotics. Nat Rev Microbiol 2009; 7: 901. isolation of reuterin, a growth inhibitor produced by Lactoba- 24 Ehrlich SD. Probiotics Flittle evidence for a link to obesity. Nat cillus reuteri. Antimicrob Agents Chemother 1988; 32 : 1854–1858. Rev Microbiol 2009; 7: 901. 42 Chang YH, Kim JK, Kim HJ, Kim WY, Kim YB, Park YH. Selection 25 Yajnik CS, Yudkin JS. The Y-Y paradox. Lancet 2004; 363 : 163. of a potential probiotic Lactobacillus strain and subsequent in 26 Jackson MS, Bird AR, McOrist AL. Comparison of two selective vivo studies. Antonie Van Leeuwenhoek 2001; 80 : 193–199. media for the detection and enumeration of lactobacilli in 43 Lu YC, Yin LT, Chang WT, Huang JS. Effect of Lactobacillus human faeces. J Microbiol Methods 2002; 51 : 313–321. reuteri GMNL-263 treatment on renal fibrosis in diabetic rats. 27 Hartemink R, Domenech VR, Rombouts FM. LAMVAB-A new J Biosci Bioeng 2010; 110 : 709–715. selective medium for the isolation of lactobacilli from faeces. 44 England JA, Watkins SE, Saleh E, Waldroup PW. Effects of J Microbiol Methods 1997; 29 : 77–84. Lactobacillus reuteri on live performance and intestinal develop- 28 Seng P, Drancourt M, Gouriet F, La Scola B, Fournier PE, Rolain JM ment of male turkeys. J Appl Poultry Sci 1996; 5: 311–324. et al. Ongoing revolution in bacteriology: routine identification 45 Dunham HJ, Casas IA, Edens FW, Parkhurst CR, Garlich JD, of bacteria by matrix-assisted laser desorption ionization time-of- Dobrogosz WJ. Avian growth depression in chickens induced by flight mass spectrometry. Clin Infect Dis 2009; 49 : 543–551. environmental, microbiological, or nutritional stress is moder- 29 Dridi B, Henry M, El Khechine A, Raoult D, Drancourt M. High ated by probiotic administrations of Lactobacillus reuteri. Biosc prevalence of Methanobrevibacter smithii and Methanosphaera Microflor 1998; 17 : 133–139. stadtmanae detected in the human gut using an improved DNA 46 Casas IA, Dobrogosz WJ. Validation of the probiotic concept: detection protocol. PLoS One 2009; 4: e7063. Lactobacillus reuteri confers broad-spectrum protection against 30 Hosmer DW, Lemeshow S. Applied Logistic Regression 2nd edn. disease in humans and animals. Microb Ecol Health Dis 2000; 12 : Wiley: New York, 2000. 247–285. 31 Mai V, McCrary QM, Sinha R, Glei M. Associations between 47 Backhed F, Manchester JK, Semenkovich CF, Gordon JI. Mechanisms dietary habits and body mass index with gut microbiota underlying the resistance to diet-induced obesity in germ-free mice. composition and fecal water genotoxicity: an observational study Proc Natl Acad Sci USA 2007; 104 : 979–984. in African American and Caucasian American volunteers. Nutr J 48 Backhed F, Ding H, Wang T, Hooper LV, Koh GY, Nagy A et al. 2009; 8: 49. The gut microbiota as an environmental factor that regulates 32 Duncan SH, Lobley GE, Holtrop G, Ince J, Johnstone AM, Louis P fat storage. Proc Natl Acad Sci USA 2004; 101 : 15718–15723. et al. Human colonic microbiota associated with diet, obesity and 49 Bastard JP, Maachi M, Lagathu C, Kim MJ, Caron M, Vidal H weight loss. Int J Obes (Lond) 2008; 32 : 1720–1724. et al. Recent advances in the relationship between obesity, 33 Zhang H, DiBaise JK, Zuccolo A, Kudrna D, Braidotti M, Yu Y et al. inflammation, and insulin resistance. Eur Cytokine Netw 2006; Human gut microbiota in obesity and after gastric bypass. Proc 17 : 4–12. Natl Acad Sci USA 2009; 106 : 2365–2370. 50 Hotamisligil GS. Inflammation and metabolic disorders. Nature 34 Kondo S, Xiao JZ, Satoh T, Odamaki T, Takahashi S, Sugahara H 2006; 444 : 860–867. et al. Antiobesity effects of Bifidobacterium breve strain B-3 51 Sbarbati A, Osculati F, Silvagni D, Benati D, Galie M, Camoglio FS supplementation in a mouse model with high-fat diet-induced et al. Obesity and inflammation: evidence for an elementary obesity. Biosci Biotechnol Biochem 2010; 74 : 1656–1661. lesion. Pediatrics 2006; 117 : 220–223. 35 Yin YN, Yu QF, Fu N, Liu XW, Lu FG. Effects of four bifidobacteria 52 Fogarty AW, Glancy C, Jones S, Lewis SA, McKeever TM, on obesity in high-fat diet induced rats. World J Gastroenterol Britton JR. A prospective study of weight change and systemic 2010; 16 : 3394–3401. inflammation over 9 y. Am J Clin Nutr 2008; 87 : 30–35. 36 Cani PD, Neyrinck AM, Fava F, Knauf C, Burcelin RG, Tuohy KM 53 Nahashon SN, Nakaue HS, Snyder SP, Mirosh LW. Performance of et al. Selective increases of bifidobacteria in gut microflora single comb White Leghorn layers fed corn-soybean meal and improve high-fat-diet-induced diabetes in mice through a barley-corn-soybean meal diets supplemented with a direct-fed mechanism associated with endotoxaemia. Diabetologia 2007; microbial. Poult Sci 1994; 73 : 1712–1723. 50 : 2374–2383. 54 Shimada K, Bricknell KS, Finegold SM. Deconjugation of bile 37 Lee K, Paek K, Lee HY, Park JH, Lee Y. Antiobesity effect of trans- acids by intestinal bacteria: review of literature and additional 10, cis-12-conjugated linoleic acid-producing Lactobacillus plan- studies. J Infect Dis 1969; 119 : 73–81. tarum PL62 on diet-induced obese mice. J Appl Microbiol 2007; 55 Fuller R. Probiotics in man and animals. J Appl Bacteriol 1989; 66 : 103 : 1140–1146. 365–378. 38 Hamad EM, Sato M, Uzu K, Yoshida T, Higashi S, Kawakami H 56 Abdulrahim SM, Haddadin MS, Odetallah NH, Robinson RK. et al. Milk fermented by Lactobacillus gasseri SBT2055 influences Effect of Lactobacillus acidophilus and zinc bacitracin as adipocyte size via inhibition of dietary fat absorption in Zucker dietary additives for broiler chickens. Br Poult Sci 1999; 40 : rats. Br J Nutr 2009; 101 : 1–9. 91–94. 39 Kadooka Y, Sato M, Imaizumi K, Ogawa A, Ikuyama K, Akai Y et al. Regulation of abdominal adiposity by probiotics (Lactobacillus gasseri SBT2055) in adults with obese tendencies in a randomized This work is licensed under the Creative controlled trial. Eur J Clin Nutr 2010; 64 : 636–643. Commons Attribution-NonCommercial-No 40 Lee HY, Park JH, Seok SH, Baek MW, Kim DJ, Lee KE et al. Human Derivative Works 3.0 Unported License. To view a copy originated bacteria, Lactobacillus rhamnosus PL60, produce conjugated linoleic acid and show anti-obesity effects in diet- of this license, visit http://creativecommons.org/ induced obese mice. Biochim Biophys Acta 2006; 1761 : 736–744. licenses/by-nc-nd/3.0/

Supplementary Information accompanies the paper on International Journal of Obesity website (http://www.nature.com/ijo)

International Journal of Obesity

Article IV :

Correlation between body mass index and gut

concentrations of Lactobacillus reuteri ,

Bifidobacterium animalis , Methanobrevibacter smithii

and Escherichia coli

Matthieu Million, Emmanouil Angelakis, Marie Maraninchi, Mireille Henry, Roch Giorgi, René Valero, Bernard Vialettes, Didier Raoult

Published in Int J Obes (Lond). 2013 Mar 5. doi: 10.1038/ijo.2013.20. [Epub ahead of print] (IF 4.69)

ORIGINAL ARTICLE Correlation between body mass index and gut concentrations of Lactobacillus reuteri , Biﬁdobacterium animalis , Methanobrevibacter smithii and Escherichia coli

M Million 1,2,7 , E Angelakis 1,7 , M Maraninchi 3, M Henry 1, R Giorgi 4,5 , R Valero 3,6 , B Vialettes 6 and D Raoult 1,2

BACKGROUND: Genus and species level analysis is the best way to characterize alterations in the human gut microbiota that are associated with obesity, because the clustering of obese and lean microbiotas increases with the taxonomic depth of the analysis. Bifidobacterium genus members have been associated with a lean status, whereas different Lactobacillus species are associated both with a lean and an obese status. OBJECTIVES AND METHODS: We analyzed the fecal concentrations of Bacteroidetes, Firmicutes , Methanobrevibacter smithii , the genus Lactobacillus, five other Lactobacillus species previously linked with lean or obese populations, Escherichia coli and Bifidobacterium animalis in 263 individuals, including 134 obese, 38 overweight, 76 lean and 15 anorexic subjects to test for the correlation between bacterial concentration and body mass index (BMI). Of these subjects, 137 were used in our previous study. FINDINGS: Firmicutes were found in 498.5%, Bacteroidetes in 67%, M. smithii in 64%, E. coli in 51%, Lactobacillus species between 17 and 25% and B. animalis in 11% of individuals. The fecal concentration of Lactobacillus reuteri was positively correlated with BMI (coefficient ¼ 0.85; 95% confidence interval (CI) 0.12–0.58; P ¼ 0.02) in agreement with what was reported for Lactobacillus sakei. As reported, B. animalis (coefficient ¼ À 0.84; 95% CI À 1.61 to À 0.07; P ¼ 0.03) and M. smithii (coefficient ¼ À 0.43, 95% CI À 0.90 to 0.05; P ¼ 0.08) were negatively associated with the BMI. Unexpectedly, E. coli was found here for the first time to negatively correlate with the BMI (coefficient ¼ À 1.05; 95% CI À 1.60 to À 0.50; Po0.001). CONCLUSION: Our findings confirm the specificity of the obese microbiota and emphasize the correlation between the concentration of certain Lactobacillus species and obesity.

International Journal of Obesity advance online publication, 5 March 2013; doi:10.1038/ijo.2013.20 Keywords: body mass index; Lactobacillus; Biﬁdobacterium; probiotics; Methanobrevibacter smithii; Escherichia coli

INTRODUCTION that analysis on a taxonomic basis remains fully relevant, 9 Obesity is defined by a body mass index (BMI) 430 kg m À 2 (ref. 1) specifically at the species level. and a massive expansion of fat and is associated with a significant A decreased Bacteroidetes/Firmicutes ratio was initially shown 10 increase in morbidity and mortality. 2,3 The frequency of obesity is to be associated with obesity, but the discrimination between rising among children, adolescents and adults and has doubled lean and obese gut microbiota is improved when the taxonomic 11 since 1980. According to the WHO, 65% of the world’s population depth of the analysis is increased. For instance, the lives in countries where excess weight and obesity kills more Bifidobacterium genus was associated with lean humans in a people than underweight conditions, including all high-income meta-analysis, including studies from Finland, Germany, Spain and and most middle-income countries (www.who.int). China.12 Conversely, we showed that among Lactobacillus The digestive microbiota is a complex ecosystem that consists species previously associated with obesity13,14 and diabetes, 15 of viruses, bacteria, archaea, fungi and parasites. Specific some have also been associated with weight gain9,16 while others enterotypes have been identified regardless of ethnic or have more of a protective effect. 16–18 Other bacterial species, such geographical origins.4 They have been linked to diet, 5 and their as Tropheryma whipplei, have been associated with acquired antibiotic-mediated modulation can impact the metabolic profile obesity.19 Finally, Karlsson et al. 20,21 linked the Enterobacteriaceae of the host. 6 Because the gut is a ‘hot spot’ for horizontal gene and specifically Escherichia coli to overweight and obesity. transfer between an astronomical number of bacteria ( 410 9 g À 1), Here, we looked at the inter-relationships among E. coli , archaea and viruses,7 analysis at the gene level was found to be one of the main representatives of the Enterobacteriaceae, the best way to characterize gut microbiota alteration and its Methanobrevibacter smithii, a leading representative of the gut correlation with obesity.8 Conversely, we and others have found archaea,22 Bifidobacterium animalis and 5 Lactobacillus species

1URMITE, UM63, CNRS 7278, IRD 198, Inserm 1095, Aix Marseille Universite´, Marseille, France; 2APHM, CHU Timone, Poˆle Infectieux, Marseille, France; 3INSERM UMR1062, INRA UMR1260, Faculte´ de Me´decine, Aix-Marseille Universite´, Marseille, France; 4INSERM, IRD, SESSTIM UMR S 912, Aix-Marseille Universite´, Marseille, France; 5Service de Sante´ Publique et d’Information Me´dicale, CHU de la Timone, APHM, Marseille, France and 6Service de Nutrition, Maladies Me´taboliques et Endocrinologie, CHU de la Timone, APHM, Marseille, France. Correspondence: Professor D Raoult, URMITE–CNRS UMR 7278, INSERM U1095, IRD 198, Faculte´ de Me´decine, Aix-Marseille Universite´, CNRS, 27 Bd Jean Moulin, Marseille 13385, France. E-mail: [email protected] 7These authors contributed equally to this work. Received 24 October 2012; revised 14 January 2013; accepted 28 January 2013 Gut microbiota is linked to the body mass index M Million et al 2 (Lactobacillus reuteri, Lactobacillus plantarum, Lactobacillus Initially, we tested whether the bacterial prevalence was different rhamnosus, Lactobacillus fermentum, Lactobacillus acidophilus). All between each BMI group using the bilateral Pearson Chi-square test. A 24 of the above species have been associated with weight in bilateral Barnard exact test was used when the Pearson Chi-square test previous studies.9,16 Based on our previous case-control study, 9 was not applicable. Because it is unknown whether overweight individuals we have more than doubled the sample size, having included should be considered as individuals with a disease or controls, all the groups were compared either with group I (obese subjects who were both anorexic and overweight patients in this study, and finally considered as cases) or with group III (lean subjects who were used as we have analyzed the correlations between the BMI and the controls). A logistic regression using the ascendant maximum likelihood considered taxa. model was used to identify bacteria whose presence was associated with the BMI groups in a multivariate analysis. Three models were used as follows: considering age, sex, Bacteroidetes, Firmicutes and M. smithii METHODS (phylum level); considering age, sex and Lactobacillus (genus level); or Patients considering age, sex, M. smithii , E. coli , B. animalis , L. reuteri , L. plantarum , This study was approved by the local ethics committee (accession number L. fermentum and L. rhamnosus (species level). 10-002, 2010). Fecal samples were obtained from hospitalized patients and As a second step, we tested whether the bacterial concentrations were outpatients at the Nutrition Unit (Hopital La Timone, Marseille, France) who different according to the BMI groups. Because of a generally non-Gaussian were overweight, obese or anorexic. The controls were healthy individuals distribution, comparisons were performed using the Kruskal–Wallis test. recruited based on a snowball approach and included subjects of our Following that step, we tested for the correlation between each bacterial previous study and outpatients who were not treated with antibiotics at concentration and BMI. As most of the bacterial clades were present the infectious disease unit (Hopital La Timone, Marseille, France). Anorexic in a minority of individuals, a dose-dependent relationship (BMI vs subjects met the DSM-IV criteria (Diagnostic and Statistical Manual of bacterial load) was explored graphically, and the correlation was tested Mental Disorders, Fourth Edition) for anorexia nervosa. The inclusion using the Spearman method only on patients harboring each of the criteria were adults for whom the BMI value and a fecal sample were clades considered (carriers). Linear regression was used to identify readily available. The exclusion criteria were patients o18 years of age, a bacteria whose concentrations were correlated with BMI on the whole history of colon cancer, the presence of an inflammatory bowel disease, an population. Three models were used as follows: considering age, sex, acute or a chronic diarrhea in the previous 4 weeks and an antibiotic Bacteroidetes, Firmicutes and M. smithii (phylum level); considering age, administration o6 months before the fecal sampling. Clinical data (gender, sex and Lactobacillus (genus level); or considering age, sex, M. smithii , date of birth, clinical history, weight, height and antibiotic use) were E. coli , B. animalis , L. reuteri , L. plantarum , L. fermentum and L. rhamnosus recorded using a standardized questionnaire. Other factors, such as yogurt (species level). M. smithii is the major representative of the gut archaeal phylum (pro- and prebiotics) intake, vegetarian habits, ethnicity or familial obesity, 21 were not taken into consideration in the analysis of the data. Four groups Euryarchaeota and has been included in the analyses both at the phylum 4 À 2 and at the species level. All the tests were bilateral and considered were identified as follows: group I: obese subjects (BMI 30 kg m ), o group II: overweight subjects (BMI425 and o30 kg m À 2), group III: lean significant when P 0.05. The analyses were performed using the SPSS subjects (BMI419 and o25 kg m À 2) and group IV: anorexic subjects v20.0 (IBM, Paris, France), R version 2.14.0 (R-foundation, Vienna, Austria) (BMIo19 kg m À 2). A total of 137 patients from our previous study 9 and and XLSTAT v12 (Addinsoft, Paris, France) software. 126 new subjects were included, of whom 15 were anorexic, 30 were lean controls, 21 were overweight and 60 were obese. Data from our previous study were also included, and most samples from that study RESULTS were analyzed further for the presence of E. coli . All new samples were Of the 263 patients enrolled in this study, there were 134 obese, 38 also analyzed for the presence of Bacteroidetes, Firmicutes, genus overweight, 76 lean and 15 anorexic subjects (Table 1). The average Lactobacillus, E. coli , M. smithii , L. reuteri , L. plantarum , L. rhamnosus , age was 50 ±s.d. 17 years, and 138 (52.5%) of them were males. L. fermentum and L. acidophilus . As was expected, anorexic patients were more frequently found to be younger women. PCR detection and quantification was PCR performed on 262 individuals to study the levels of Bacteroidetes, PCR analysis was performed as previously described 9 except for E. coli , Firmicutes, M. smithii and the Lactobacillus genus; on 219 for which the protocol was the same, but the primers and probes were individuals to study each Lactobacillus species and B. animalis ; 0 0 0 the following: Forward, 5 -GCTGCGCGTGCAAATGCG-3 ; Reverse, 5 -CATGGT and on 165 individuals to investigate the levels of E. coli . CATCGCTTCGGTCT-30; and probe, 5 0-CATCAGAAACTGAACACCAC-30. The 9 The prevalences of each bacterial clade were heterogeneous. primers for L. reuteri were evaluated in our previous study and have a Firmicutes was found in all the individuals ( 498.5%), whereas very high specificity at the species level with a low cross-reactivity (cycle threshold 435 for DNA extracted from pure culture) with Bacteroidetes was detected in only 67% (Supplementary Table S1). Lactobacillus oris and Lactobacillus pontis. However, it cannot be At the species level, B. animalis was found to be the rarest species excluded that the detection of L. oris , exceptionally present in the (11%), whereas M. smithii (64%) was shown to be more prevalent human gut,23 could have yielded false-positive results. Conversely, L. pontis than E. coli (51%). Lactobacillus genus was found in only one-third has never been reported in the human gut. The results in this study are of the subjects (28%), with different species ranging from 17 to depicted as log10 DNA copies ml À 1. 25% in frequency. In agreement with our previous study, 8 L. acidophilus was not detected by our system in any sample, Statistical analysis despite the positive amplification of the type strain L. acidophilus As an exploratory step, a principal component analysis was performed, CIP7613 in our in silico study. 9 À 1 including BMI and the concentrations of all taxa present at the phylum and When present, the Firmicutes (10 DNA copies ml ) was the species levels. most abundant clade before the Bacteroidetes (108). At the

Table 1. Population characteristics

Anorexic subjects (n ¼ 15) Lean subjects (n ¼ 76) Overweight subjects (n ¼ 38) Obese subjects (n ¼ 134) P-valuea

Age (mean ±s.d.) 27.3 ±10.8 49.5±18.6b 54.1±17.8 51.8±14.7 o0.0001 Male sex (n (%)) 1 (7%) 40 (57%) 32 (84%) 65 (49%) o0.0001 BMI (median, IQR) 13.5 (11.7–14.6) 22.4 (20.7–23.7) 27.1 (25.9–28.6) 40.0 (36.4–46.8) o0.0001 Abbreviations: BMI, body mass index; IQR, interquartile range. aMann–Whitney U test for age and BMI, Pearson chi-square for sex. bData unavailable for seven patients.

International Journal of Obesity (2013) 1 – 7 & 2013 Macmillan Publishers Limited Gut microbiota is linked to the body mass index M Million et al 3 species level, when found, E. coli (107) was 10 times more Bacterial clades associated with lean status 6 abundant than M. smithii (10 ), whereas Lactobacillus species were Bacteroidetes. The difference in the occurrence of Bacteroidetes 4 5 À 1 present at much lower concentrations (10 –10 DNA copies ml ; between the obese and the lean groups was not significant (60 vs Po0.0001 when comparing Lactobacillus with E. coli ). 70%, respectively; P ¼ 0.18); however, we found a decreased Preliminary analysis by principal component analysis and frequency of Bacteroidetes in obese compared with non-obese density plots suggested that some bacterial species or phyla individuals (60 vs 74%; P ¼ 0.02). Moreover, prevalence was were differentially distributed according to the BMI (Figure 1 and increased in overweight compared with obese subjects (84 vs Supplementary Figure S1) and this was confirmed by further 60%; P ¼ 0.008, Supplementary Figure S2). In a logistic regression, analyses (Figures 2 and 3). the presence of Bacteroidetes was associated with the absence of obesity (OR ¼ 0.51; 95% CI 0.30–0.87; P ¼ 0.01) or overweight individuals when compared with obese population (OR ¼ 0.28; Bacterial clades associated with obesity 0.11–0.74; P ¼ 0.01, Supplementary Table S2). Genus Lactobacillus. There was a trend towards a higher Finally, we found a trend towards decreased concentrations of prevalence of Lactobacillus in obese compared with lean patients Bacteroidetes in obese patients compared with lean controls (32 vs 20%; P ¼ 0.06) and a higher frequency of Lactobacillus in (P ¼ 0.054), and this decrease in Bacteroidetes concentration was patients with BMIs425 vs BMIs o25 kg m À 2 (32 vs 20.8%; significant when comparing obese with non-obese ( P ¼ 0.01) or P ¼ 0.06, Supplementary Table S1 and Supplementary Figure S2). with overweight individuals (P ¼ 0.017, Figure 2). No correlation In a logistic regression, the presence of Lactobacillus was not was found between the Bacteroidetes concentration and BMI in associated with any BMI group (Supplementary Table S2). the carrier subgroup. In a linear regression, Bacteroidetes The Lactobacillus concentration was higher in obese patients concentration was not correlated with BMI. compared with lean patients (Po0.05) and in individuals with À 2 À 2 BMIs425 kg m vs individuals with BMIs o25 kg m (Po0.05, M. smithii . There was a trend towards an increased prevalence of Figure 2). We also found a positive correlation between the M. smithii in lean compared with obese individuals (72 vs 60%; concentration of Lactobacillus and BMI in the carriers (patients P ¼ 0.07), and this frequency difference was significant when positive for the genus Lactobacillus, correlation coefficient 0.25; individuals with BMIso25 kg m À 2 were compared with indivi- P ¼ 0.03). No significant result was found in a linear regression. duals with BMIs425 kg m À 2 (72 vs 60%; P ¼ 0.04, Supplementary Figure S2). In a logistic regression, the presence of M. smithii was L. reuteri . There was a threefold increase in the L. reuteri not associated with the absence of obesity but was associated occurrence in obese patients compared with lean subjects with lean compared with overweight individuals (OR ¼ 0.001; 95% (22 vs 8%; P ¼ 0.01), a fourfold increase between overweight CI 0–0.98; P ¼ 0.049, Supplementary Table S2). patients and lean subjects (34 vs 8%; P ¼ 0.001) and a threefold The M. smithii concentration was lower in obese compared increase between individuals with BMIs425 kg m À 2 compared with either lean (P ¼ 0.008) or non-obese individuals ( P ¼ 0.01). with individuals with BMIso25 kg m À 2 (20 vs 7%; P ¼ 0.001, Moreover, the M. smithii concentration was higher in patients o À 2 Supplementary Figure S2). In a logistic regression, the presence having BMIs 25 kg m compared with patients having 4 À 2 of L. reuteri was associated with obesity (odds ratio (OR) ¼ 5.31; BMIs 25 kg m (P ¼ 0.005, Figure 2). We also found a negative 95% confidence interval (CI) 1.04–27.1; P ¼ 0.04), overweight correlation between the BMI values and M. smithii concentration in (OR ¼ 2.8 Â 10 7; 95% CI 6.9–10 14 ; P ¼ 0.03) or BMI 425 kg m À 2 patients harboring M. smithii (correlation coefficient À 0.20; (OR ¼ 8.07; 95% CI 2.06–31.5; P ¼ 0.003). P ¼ 0.01, Figure 4). In a linear regression, M. smithii was not The L. reuteri concentration was greater in obese vs lean associated with BMI when analyzed at the phylum level as individuals (Po0.05), in overweight vs lean individuals ( Po0.005) the M. smithii phylum is the leading representative of the and in individuals with BMIs 425 kg m À 2 compared with indivi- Euryarchaeota in the gut microbiota. Conversely, there was a duals with BMIso25 kg m À 2 (Po0.005, Figure 3). Furthermore, trend towards a correlation between a higher BMI and a lower we found a positive correlation between the concentration of M. smithii concentration at the species level ( P ¼ 0.08, Table 2). L. reuteri and BMI (patients positive for L. reuteri , coefficient correlation 0.44; P ¼ 0.004, Figure 4). In a linear regression, a B. animalis . The prevalence of B. animalis was very low in our higher concentration of L. reuteri was associated with a higher population, between 6 and 15%, but there was a trend towards a BMI (Table 2). significantly lower occurrence in obese compared with lean

Body mass index Euryarchaeota

L. reuteri L. rhamnosu s Firmicutes L. plantarum 2nd Component (25%)

2nd component (16%) B. animalis L. fermentum M. smithii Bacteroidetes Body mass index

E. coli

1st Component (33%) 1st Component (20%) Figure 1. Primary component analysis associating the gut microbial phylum and species to the BMI. Principal component analysis, including (a) BMI and phylum or ( b) species found in the gut microbiota ( Lactobacillus acidophilus was not included because it was not found by our quantitative PCR system). The preliminary analyses shown in this ﬁgure were performed on the whole population.

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 1 – 7 Gut microbiota is linked to the body mass index M Million et al 4

Figure 2. Scatter plots at the phylum and genus levels. Methanobrevibacter smithii is considered to be the leading representative of the Euryarchaeota phylum. *Po0.05, ** Po0.005. The medians and the interquartile ranges are shown.

Figure 3. Scatter plots at the species level. * Po0.05, ** Po0.005. The medians and the interquartile ranges are shown.

International Journal of Obesity (2013) 1 – 7 & 2013 Macmillan Publishers Limited Gut microbiota is linked to the body mass index M Million et al 5 Methanobrevibacter smithii Escherichia coli 10 9

8 8

7 6 6

log10 copies DNA/ml 4 5

2 20 40 60 80 10 20 30 40 50 60 70

kg/m2 kg/m2

Bifidobacterium animalis Lactobacillus reuteri 9 9 8 8 7 7 6 6 5 5 4

log10 copies DNA/ml copies log10 4 log10 copies DNA/ml log10 copies DNA/ml 3 3

20 30 40 50 10 20 30 40 50 kg/m2 kg/m2 Figure 4. Correlation between the BMI and speciﬁc bacterial clades. Plots represent analyses performed only on the carriers for each bacterial clade studied. Spearman correlation test: Methanobrevibacter smithii r ¼ À 0.20, P ¼ 0.01. Lactobacillus reuteri r ¼ 0.44, P ¼ 0.004. No correlation was found in the patients positive for E. coli (P ¼ 0.80) or Biﬁdobacterium animalis (P ¼ 0.99).

Table 2. BMI linear regression according to each bacterial clade a higher concentration of B. animalis was associated with a lower BMI (P ¼ 0.03, Table 2). Speciesa Coefficient (95% CI) P-value E. coli . The prevalence of E. coli was lower in obese compared Methanobrevibacter smithii À 0.43 ( À 0.90 to 0.05) 0.08 with lean (36 vs 60%; P ¼ 0.006), overweight (36 vs 75%; P ¼ 0.004) b À À À o Escherichia coli 1.05 ( 1.60 to 0.50) 0.001 and non-obese individuals (36 vs 47%; Po0.001, Supplementary Bifidobacterium animalis À 0.84 ( À 1.61 to À 0.07) 0.03 Lactobacillus reuteri 0.85 (0.12 to 1.58) 0.02 Figure S2). The prevalence was also significantly lower in individuals with BMIs 425 kg m À 2 compared with those with Abbreviations: BMI, body mass index; CI, confidence interval. aLinear BMIso25 kg m À 2 (31 vs 51%; P ¼ 0.004). In a logistic regression, regression, adjusted by age and sex, was performed on 218 patients for the presence of E. coli was associated with the absence of whom data for M. smithii , B. animalis , L. reuteri , L. plantarum , L. fermentum o b obesity (OR ¼ 0.25; 95% CI 0.1–0.5; P 0.001, Supplementary and L. rhamnosus were available. E. coli concentration was available only Table S2), with lean when compared with obese individuals for 133 of these patients and was replaced by the mean for the 85 lacking ¼ ¼ data. (OR 0.3; 95% CI 0.1–0.8; P 0.01), with overweight when compared with obese individuals (OR ¼ 0.15; 95% CI 0.03–0.9; P ¼ 0.01) and with individuals with BMIs o25 kg m À 2 vs individuals with BMIs 425 kg m À 2 (OR ¼ 0.3; 95% CI 0.1–0.6; P ¼ 0.002). ¼ individuals (6 vs 15%; P 0.052) and a significant decrease in the A lower concentration of E. coli was found in obese vs anorexic incidence of B. animalis in obese compared with non-obese (P ¼ 0.001), lean ( P ¼ 0.02), overweight individuals ( P ¼ 0.012) and ¼ À individuals (6 vs 15%; P 0.04, Supplementary Figure S2). Using a in individuals with BMIs 425 vs o25 kg m 2 (P ¼ 0.02). Moreover, a logistic regression, there was a trend towards an association lower concentration of E. coli was found when comparing obese between the presence of B. animalis and lean compared with with non-obese individuals ( P ¼ 0.001, Figure 3). No correlation was ¼ ¼ obese individuals (OR 0.22; 95% CI 0.05–1.03; P 0.054). found in the subgroup of individuals positive for E. coli (correlation Furthermore, the presence of B. animalis was associated with lean coefficient 0.03, P ¼ 0.8, Figure 4). In a linear regression, a higher ¼ individuals when compared with overweight subjects (OR 0; concentration of E. coli was associated with a lower BMI (Table 2). 95% CI 0–0.76; P ¼ 0.045). The concentration of B. animalis was significantly lower in obese population compared with lean (P ¼ 0.045) and non-obese populations (P ¼ 0.03, Figure 3) but no correlation was found DISCUSSION between the B. animalis concentration and BMI when we In this study, we found a relatively low prevalence of Lactobacillus performed a univariate analysis (Figure 4). In a linear regression, species because it was detected in only 30% of the individuals, but

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 1 – 7 Gut microbiota is linked to the body mass index M Million et al 6 L. reuteri was detected in 20% of the study population with REFERENCES occurrence increasing along with BMI values (7, 8, 34 and 22% for 1 WHO Expert Consultation. Appropriate body-mass index for Asian populations anorexic, lean, overweight and obese individuals, respectively). and its implications for policy and intervention strategies. Lancet 2004; 363 : Lactobacillus species, and specifically L. reuteri , have been 157–163. previously associated with obesity as it has been reported in our 2 Whitlock G, Lewington S, Sherliker P, Clarke R, Emberson J, Halsey J et al. previous case-control studies.9,13 However, this is the first time Body-mass index and cause-specific mortality in 900 000 adults: collaborative that a correlation between the bacterial loads of this species and analyses of 57 prospective studies. Lancet 2009; 373 : 1083–1096. BMI is reported. To our knowledge, only one other previous study 3 Yanovski SZ, Yanovski JA. Obesity. N Engl J Med 2002; 346 : 591–602. 4 Arumugam M, Raes J, Pelletier E, Le Paslier D, Yamada T, Mende DR et al. identified a correlation between the Lactobacillus species, and 14 Enterotypes of the human gut microbiome. Nature 2011; 473 : 174–180. specifically Lactobacillus sakei, and BMI. 5 Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: human gut microbes Other prokaryotes have been associated with a lower BMI, as associated with obesity. Nature 2006; 444 : 1022–1023. has been previously reported in other publications, such as 6 Cho I, Yamanishi S, Cox L, Methe BA, Zavadil J, Li K et al. Antibiotics in Bacteroidetes,5,10,13,25 B. animalis 9,12,26,27 and the archeal species early life alter the murine colonic microbiome and adiposity. Nature 2012; 488 : M. smithii .28 In contrast to previous studies, 29 we found a lower 621–626. frequency and lower bacterial loads for E. coli in obese individuals 7 Kurokawa K, Itoh T, Kuwahara T, Oshima K, Toh H, Toyoda A et al. Comparative with a strong statistical significance. This finding demands a word metagenomics revealed commonly enriched gene sets in human gut of caution and requires further confirmation. Moreover, our results microbiomes. DNA Res 2007; 14 : 169–181. 8 Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, Gordon JI. An obesity- suggest a ‘dose-dependent’ relationship between certain species associated gut microbiome with increased capacity for energy harvest. Nature of bacteria and archaea in the human gut and BMI. 2006; 444 : 1027–1031. A limitation of our study, as it is for most studies, is that the 9 Million M, Maraninchi M, Henry M, Armougom F, Richet H, Carrieri P et al. Obesity- analysis of the digestive microbiota associated with obesity was associated gut microbiota is enriched in Lactobacillus reuteri and depleted in performed by analyzing stool samples. 5,25 However, as 95% of fat Bifidobacterium animalis and Methanobrevibacter smithii. Int J Obes (Lond) 2012; is absorbed before the cecum, 30 the proximal gut microbiota may 36 : 817–825. be critical for the analysis of factors associated with obesity and 10 Ley RE, Backhed F, Turnbaugh P, Lozupone CA, Knight RD, Gordon JI. diabetes.31–33 The analysis of the fecal microbiota reflects only Obesity alters gut microbial ecology. Proc Natl Acad Sci USA 2005; 102 : indirectly the upper intestinal flora. Indeed, several studies have 11070–11075. 11 Murphy EF, Cotter PD, Hogan A, O’Sullivan O, Joyce A, Fouhy F et al. Divergent shown a significant difference in the gut microbiota composition 34 35 metabolic outcomes arising from targeted manipulation of the gut microbiota in according to the gut section in animals and humans with a diet-induced obesity. Gut 2012; 62 : 220–226. proximal (small bowel) enrichment in aerobic Firmicutes 12 Angelakis E, Armougom F, Million M, Raoult D. The relationship between gut 34,35 (Streptococcaceae and Lactobacillaceae) and Actinobacteria. microbiota and weight gain in humans. Future Microbiol 2012; 7: 91–109. Finally, obesity is a multifactorial disease. The causes that drive 13 Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. Monitoring bacterial obesity appear to be influenced by a mixture of environmental, community of human gut microbiota reveals an increase in Lactobacillus in obese genetic, neural and endocrine factors along with microbes that are patients and Methanogens in anorexic patients. PLoS One 2009; 4: e7125. also thought to have a role in weight gain. 12,36 Accumulating data 14 Stsepetova J, Sepp E, Kolk H, Loivukene K, Songisepp E, Mikelsaar M. Diversity and has shown that the gut microbiota is associated with both obesity metabolic impact of intestinal Lactobacillus species in healthy adults and the and diet, and there is evidence that modulation of the gut flora by elderly. Br J Nutr 2011; 105 : 1235–1244. 6,37 38 16,36 15 Larsen N, Vogensen FK, van den Berg FW, Nielsen DS, Andreasen AS, Pedersen BK antibiotics, during pregnancy or by probiotics causes et al. Gut microbiota in human adults with type 2 diabetes differs from non- weight gain. The repertoire of bacteria, and especially Lactobacillus diabetic adults. PLoS One 2010; 5: e9085. species, that protect or result in weight gain should be 16 Million M, Angelakis E, Paul M, Armougom F, Leibovici L, Raoult D. Comparative determined at the strain level as the genomic variations within a meta-analysis of the effect of Lactobacillus species on weight gain in humans and single species of Lactobacillus may be dramatic (only 64% of animals. Microb Pathog 2012; 53 : 100–108. protein genes are common between Lactobacillus johnsonii FI9785 17 Luoto R, Kalliomaki M, Laitinen K, Isolauri E. The impact of perinatal probiotic and L. johnsonii NCC 53339 ). intervention on the development of overweight and obesity: follow-up study from birth to 10 years. Int J Obes (Lond) 2010; 34 : 1531–1537. 18 Kadooka Y, Sato M, Imaizumi K, Ogawa A, Ikuyama K, Akai Y et al. Regulation of abdominal adiposity by probiotics ( Lactobacillus gasseri SBT2055) in adults CONCLUSION with obese tendencies in a randomized controlled trial. Eur J Clin Nutr 2010; 64 : This work confirms the link between the microbiota and obesity. 636–643. This link appears to be the result of both diet 5 and the cause of 19 Fenollar F, Nicoli F, Paquet C, Lepidi H, Cozzone P, Antoine JC et al. Progressive dementia associated with ataxia or obesity in patients with Tropheryma whipplei the weight gain as demonstrated by microbiota transplantation 8,38 encephalitis. BMC Infect Dis 2011; 11 : 171. from obese individuals or pregnant women to axenic animals. 20 Karlsson CL, Molin G, Fak F, Johansson Hagslatt ML, Jakesevic M, Hakansson A et al. Effects on weight gain and gut microbiota in rats given bacterial supplements and a high-energy-dense diet from fetal life through to 6 months of age. CONFLICT OF INTEREST Br J Nutr 2011; 106 : 887–895. 21 Karlsson CL, Onnerfa¨lt J, Xu J, Molin G, Ahrne´ S, Thorngren-Jerneck K. The The authors declare no conflict of interest. microbiota of the gut in preschool children with normal and excessive body weight. Obesity (Silver Spring) 2012; 20 : 2257–2261. 22 Dridi B, Henry M, El Khechine A, Raoult D, Drancourt M. High prevalence of ACKNOWLEDGEMENTS Methanobrevibacter smithii and Methanosphaera stadtmanae detected in the human gut using an improved DNA detection protocol. PLoS One 2009; 4: e7063. We thank all the volunteers because without them, this study would not have been 23 Larsen N, Vogensen FK, Gobel R, Michaelsen KF, Abu Al-Soud W, Sorensen SJ et al. possible. The sponsors took no part in the design of the study, data collection and Predominant genera of fecal microbiota in children with atopic dermatitis are not analysis, decision to publish or preparation of the manuscript. altered by intake of probiotic bacteria Lactobacillus acidophilus NCFM and Bifidobacterium animalis subsp. lactis Bi-07. FEMS Microbiol Ecol 2011; 75 : 482–496. 24 Barnard GAA. New Test for 2 Â 2 Tables. Nature 1945; 156 : 177. AUTHOR CONTRIBUTIONS 25 Turnbaugh PJ, Hamady M, Yatsunenko T, Cantarel BL, Duncan A, Ley RE et al. A core gut microbiome in obese and lean twins. Nature 2009; 457 : 480–484. DR conceived and designed the experiments. MM, EA, MM, RV, BV and DR 26 Waldram A, Holmes E, Wang Y, Rantalainen M, Wilson ID, Tuohy KM et al. performed the clinical study. MM, EA and MH performed the experiments. Top-down systems biology modeling of host metabotype-microbiome associa- MM and RG analyzed the data. MM, EA and DR wrote the manuscript. tions in obese rodents. J Proteome Res 2009; 8: 2361–2375.

International Journal of Obesity (2013) 1 – 7 & 2013 Macmillan Publishers Limited Gut microbiota is linked to the body mass index M Million et al 7

27 Kalliomaki M, Collado MC, Salminen S, Isolauri E. Early differences in fecal 34 Torok VA, Allison GE, Percy NJ, Ophel-Keller K, Hughes RJ. Influence of microbiota composition in children may predict overweight. Am J Clin Nutr 2008; antimicrobial feed additives on broiler commensal posthatch gut microbiota 87 : 534–538. development and performance. Appl Environ Microbiol 2011; 77 : 3380–3390. 28 Schwiertz A, Taras D, Schafer K, Beijer S, Bos NA, Donus C et al. Microbiota and 35 Frank St DN, Amand AL, Feldman RA, Boedeker EC, Harpaz N, Pace NR. Molecular- SCFA in lean and overweight healthy subjects. Obesity (Silver Spring) 2010; 18 : phylogenetic characterization of microbial community imbalances in human 190–195. inflammatory bowel diseases. Proc Natl Acad Sci USA 2007; 104 : 13780–13785. 29 Santacruz A, Collado MC, Garcia-Valdes L, Segura MT, Martin-Lagos JA, Anjos T 36 Million M, Raoult D. The role of the manipulation of the gut microbiota in obesity. et al. Gut microbiota composition is associated with body weight, weight gain Curr Infect Dis Rep 2012; 15 : 25–30. and biochemical parameters in pregnant women. Br J Nutr 2010; 104 : 83–92. 37 Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant antibiotic 30 Carriere F, Renou C, Ransac S, Lopez V, De Caro J, Ferrato F et al. Inhibition of exposures and early-life body mass. Int J Obes (Lond) 2012; 37 : 16–23. gastrointestinal lipolysis by Orlistat during digestion of test meals in healthy 38 Koren O, Goodrich JK, Cullender TC, Spor A, Laitinen K, Backhed HK et al. Host volunteers. Am J Physiol Gastrointest Liver Physiol 2001; 281 : G16–G28. remodeling of the gut microbiome and metabolic changes during pregnancy. Cell 31 Zhang H, DiBaise JK, Zuccolo A, Kudrna D, Braidotti M, Yu Y et al. Human gut 2012; 150 : 470–480. microbiota in obesity and after gastric bypass. Proc Natl Acad Sci USA 2009; 106 : 39 Lukjancenko O, Ussery DW, Wassenaar TM. Comparative genomics of 2365–2370. Bifidobacterium, Lactobacillus and related probiotic genera. Microb Ecol 2012; 63 : 32 Rubino F, Forgione A, Cummings DE, Vix M, Gnuli D, Mingrone G et al. 651–673. The mechanism of diabetes control after gastrointestinal bypass surgery reveals a role of the proximal small intestine in the pathophysiology of type 2 diabetes. Ann Surg 2006; 244 : 741–749. This work is licensed under a Creative Commons Attribution- 33 Kremen AJ, Linner JH, Nelson CH. An experimental evaluation of the nutritional NonCommercial-NoDerivs 3.0 Unported License. To view a copy of importance of proximal and distal small intestine. Ann Surg 1954; 140 : 439–448. this license, visit http://creativecommons.org/licenses/by-nc-nd/3.0/

Supplementary Information accompanies this paper on International Journal of Obesity website (http://www.nature.com/ijo)

& 2013 Macmillan Publishers Limited International Journal of Obesity (2013) 1 – 7

Partie II :

Le rôle de la manipulation du microbiote digestif

dans l’obésité

Avant- propos

Elie Metchnikoff est considéré comme le pionnier des probiotiques modernes

évoquant un lien entre la consommation régulière de bactéries lactiques et la longévité des populations paysannes bulgares ("La prolongation de la vie ", publié en 1907). En 1935,

Minoru Shirota a découvert une souche de Lactobacillus casei baptisée Shirota à l’origine d’une boisson probiotique commercialisée, Yakult®, associée à un succès commercial encore actuel. Depuis, le marché des probiotiques est en expansion exponentielle. Bien que la définition actuellement acceptée des probiotiques est la suivante : « micro-organismes vivants qui, lorsqu'ils sont administrés en quantités adéquates, sont associés à un bénéfice pour la santé de l'hôte », les probiotiques sont souvent associés à des propriétés non démontrées.

Les premiers paramètres reconnus pour influencer le microbiote digestif ont été l’alimentation, les antibiotiques et les probiotiques. Le rôle de ces derniers sur le poids a été démontré dès les années 1950, quand Jukes et Stokstad 3 ont associé l’administration d’une souche bactérienne, Streptomyces aureofaciens , a un effet promoteur de croissance chez l’animal. Ils ont par la suite démontré que cet effet passait par la chlortetracycline, bactériocine naturellement secrétée par cette bactérie. Leurs travaux pionniers ont été à l’origine de l’utilisation massive des antibiotiques et des probiotiques dans l’agriculture.

Partant de là, un lien entre la consommation des probiotiques et l’obésité a été proposé 4.

Afin de clarifier l’effet des probiotiques contenant des Lactobacillus sur le poids, nous avons effectué une méta-analyse incluant 17 essais randomisées chez l’homme, 51 études chez l’animal et 14 études sur des modèles experimentaux (Article VI). Lactobacillus acidophilus , Lactobacillus ingluviei et Lactobacillus fermentum étaient associés à une prise de poids significative chez les animaux. Lactobacillus plantarum était associé à une perte de poids chez des animaux obèses et Lactobacillus gasseri était associé avec une perte de poids à

la fois chez les humains et les animaux en surpoids ou obèse. L’ensemble de ces résultats suggère que l’effet des probiotiques contenant des Lactobacillus sur le poids dépend à la fois de l’espèce bactérienne utilisée et de l’hôte. Cela a été confirmé par une autre équipe qui a retrouvé une relation linéaire entre la concentration de Lactobacillus sakei et l’indice de masse corporelle 5.

Notre travail initial incluait une étude publiée en 1952 par Robinson et al. 6 qui avait montré un effet significatif de l’administration d’une souche identif iée comme Lactobacillus acidophilus sur la prise de poids chez les nouveaux-nés surtout s’ils étaient nourris au biberon et non pas au sein. Nous avons reçu un commentaire mentionnant que cette souche avait été reclassée comme Lactobacillus gasseri 7. A la suite de cela, nous avons revérifié l’identification de l’intégralité des souches des études inclues dans notre méta -analyse et, excluant cette étude, nous avons confirmé que l’effet des probiotiques contenant des

Lactobacillus sur le poids dépend à la fo is de l’espèce bactérienne utilisée et de l’hôte (Article

VII). Quoiqu’il en soit, cette étude 6 est la seule, à notre connaissance, a clairement lier l’administration d’une souche de Lactobacillus à la prise de poids chez des nouveaux nés humains et c’est pourquoi nous pensons que la manipulation du microbiote juste après la naissance est la plus à même d’être responsable d’obésité acquise. De façon similaire, il a été montré que l’administration d’antibiotiques pendant cette période était associée à une o bésité acquise 8.

Dans un deuxième travail (Article VIII), nous avons examiné le biais de publication des études sur l'administration de probiotiques contenant des Lactobacillus chez les animaux de ferme à partir de notre premier travail sur ce sujet (Article VI). Après élimination des valeurs aberrantes correspondant à 3 publications, nous avons constaté une persistance du biais de publication significative en observant le funnel plot et par le test de régression d'Egger (intercept 1,05, p-value < 10 -6). C’est -à-dire qu’il est très probable que des études

ayant trouvé une absence de prise de poids des animaux de ferme à qui il avait été administré des probiotiques contenant des Lactobacillus n’aient pas été publiées. Et ce biais de publication était co nstaté dans deux autres études de la littérature sur d’autres possibles effets bénéfiques des probiotiques (Article VIII). L’effet à long terme des probiotiques chez l'homme devrait être analysé par une recherche indépendante afin d'éviter le même désagrément déjà signalé pour l'usage délétère des probiotiques chez les personnes souffrant de pancréatite 9.

A ce titre, nous avons contribué à la publication d’une série de cas rapportant des bactériémies à Lactobacillus rhamnosus (Article IX), confirmant que ces prétendues bactéries bénéfiques pouvaient être délétères.

Jukes et Stokstad 3 avait bien montré dès les années 1950 que l’effet promoteur de croissance de Streptomyces aureofaciens était lié à la production d’une bactériocine, l’auréomycine plu s connue sous le nom de chlortetracycline. Ils ont donc fait le lien entre probiotiques, antibiotiques et prise de poids. Plus récemment, Murphy et al. 10 ont montré qu’une bactériocine produite par un probiotique pouvait altérer de manière significative l a composition du microbiote digestif in vivo alors que la même souche bactérienne sans la bactériocine n’entrainait aucune altération. Dans ce sens, nous avons lu avec intérêt un article publié par Archambaud et al. 11 qui rapportait que l’antagonisme entr e deux souches de

Lactobacillus, Lactobacillus casei BL23 et Lactobacillus paracasei CNCM I-3689 et Listeria monocytogenes passait par une modulation du transcriptome de l’hôte et de L. monocytogenes . En utilisant les bases de données de bactériocines Bagel et Bactibase pour analyser les souches utilisées dans cette étude, nous avons pu trouver une prébacteriocine

(GenBank ID: YP_001988475) dans L. casei BL23, et une Lactocin-705 (GenBank:

LC70_LACPA) parmi les génomes disponibles de L. paracasei (Article X). Prises ensemble, ces données suggèrent que l'impact d'une souche de Lactobacillus sur le microbiote digestif

est principalement déterminé par ses activités directes antibiotiques, y compris les bactériocines.

Enfin, de nombreux antibiotiques continuent à être utilisés comme facteur de croissance dans l’agriculture notamment aux Etats -unis alors que leur utilisation a été interdite en France depuis quelques années. Dans un modèle animal, l'administration d’antibiotiques à doses infrathérapeutiques dès l e sevrage augmente l’adiposité et les hormones liées au métabolisme 12 . Chez ces animaux ont été observé des changements taxonomiques importants dans le microbiome, des changements dans le nombre de copies de gènes clés impliqués dans le métabolisme des glucides en acide gras à chaîne courte (SCFA pour short chain fatty acid principalement représentés par le propionate, butyrate et acétate), l'augmentation des acides gars à chaines courtes dans le colon, et des changements dans la régulation du métabolisme hépatique des lipides et du cholestérol.

Dans cette dernière partie, nous avons résumé dans la tableau 3 de l’article I les études ayant évalué la modification du poids sous antibiotiques chez les humains. Plusieurs études ont mis en relation la prise d’antibiotiques chez l’homme avec une prise de poids voire une augmentation du risque d’obésité quand les antibiotiques sont administrés dans les premiers mois de vie 8. De plus, nous avons résumé dans le tableau 1 de l’article XI, les modifications du microb iote digestif associées à chaque classe d’antibiotiques.

Enfin, une étude antérieure du laboratoire a associé une prise de poids et une obésité acquise à l’administration de vancomycine intraveineuse pendant 4 à 6 semaines chez des patients traités pour une endocardite 13 . Afin de prolonger cette étude et de clarifier les changements du microbiote pouvant être responsable de cette prise de poids, nous avons effectué une étude observationnelle sur le poids de 98 patients dont 41 était sous vancomycine et 57 étaient sous amoxicilline (Article XII), Nous avons retrouvé, comme Thuny et al. 13 dans l’étude antérieure du laboratoire, une augmentation de la fréquence des patients ayant

une augmentation de l’indice de masse corporelle de plus de 10% à un an et de la fréquence d’obésité acquise chez les patients sous vancomycine. Analysant 192 échantillons de selles dont 83 avaient été prélevés sous amoxicilline, 67 sous vancomycine et 42 avaient été prélevés chez des contrôles, nous avons retrouvé une augmentation dans la concentration en

Firmicutes et Lactobacillus et une diminutions de la concentration en Methanobrevibacter smithii dans les échantillons prélevés chez des patients sous vancomycine. Bien que l’augmentation des Lactobacillus puissant être une consequence des antibiotiques sans lien avec la prise de poids importante chez certains patients, l’existence de nombreuse donnés de la littérature et certaines de nos études (Article III, IV et VI) sont en faveur du rôle clé de certaines espèces de Lactobacillus sur la prise de poids, comme Lactobacillus reuteri ,

Lactobacillus fermentum et Lactobacillus sakei , espèces naturellement résistantes à la vancomycine, retrouvés dans le microbiote digestif humain et associées à la prise de poids ou

à l’obésité.

Article V : REVIEW

The role of the manipulation of the gut microbiota in

obesity

Matthieu Million, Didier Raoult

Published in Curr Infect Dis Rep. 2013 Feb;15(1):25-30. (IF ND)

Article VI :

Comparative meta-analysis of the effect of

Lactobacillus species on weight gain in humans and

animals.

Matthieu Million, Emmanouil Angelakis, Mical Paul, Fabrice

Armougom, Leonard Leibovici, Didier Raoult

Published in Microb Pathog. 2012 Aug;53(2):100-8. (IF 1.94)

Microbial Pathogenesis 53 (2012) 100 e108

Contents lists available at SciVerse ScienceDirect

Microbial Pathogenesis

journal homepage: www.elsevier.com/locate/micpath

Comparative meta-analysis of the effect of Lactobacillus species on weight gain in humans and animals

Matthieu Million a,1, Emmanouil Angelakis a,1, Mical Paul b, Fabrice Armougom a, Leonard Leibovici c, Didier Raoult a,* a URMITE-CNRS UMR 7278 IRD 198, IFR 48, Faculté de Médecine, Université de la Méditerranée, 27 bd jean moulin, Marseille, France b Sackler Faculty of Medicine, Tel-Aviv University ’ Tel-Aviv, Israel c Rabin Medical Center, Beilinson Hospital, Petah-Tiqva, Israel article info abstract

Article history: Background: Obesity is associated with alteration of the gut microbiota. In order to clarify the effect of Received 27 April 2012 Lactobacillus-containing probiotics (LCP) on weight we performed a meta-analysis of clinical studies and Received in revised form experimental models. We intended to assess effects by Lactobacillus species. 11 May 2012 Methods: A broad search with no date or language restriction was performed. We included randomized Accepted 16 May 2012 controlled trials (RCTs) and comparative clinical studies in humans and animals or experimental models Available online 24 May 2012 assessing the effect of Lactobacillus -containing probiotics on weight. We primarily attempted to extract and use change from baseline values. Data were extracted independently by two authors. Results were Keywords: Probiotics pooled by host and by Lactobacillus species and are summarized in a meta-analysis of standardized Lactobacillus difference in means (SMDs). Weight Results: We identi fied and included 17 RCTs in humans, 51 studies on farm animals and 14 experimental Obesity models. Lactobacillus acidophilus administration resulted in signi ficant weight gain in humans and in Meta-analysis animals (SMD 0.15; 95% con fidence intervals 0.05 e0.25). Results were consistent in humans and animals. Lactobacillus fermentum and Lactobacillus ingluviei were associated with weight gain in animals. Lactobacillus plantarum was associated with weight loss in animals and Lactobacillus gasseri was associated with weight loss both in obese humans and in animals. Conclusions: Different Lactobacillus species are associated different effects on weight change that are host-speci fic. Further studies are needed to clarify the role of Lactobacillus species in the human energy harvest and weight regulation. Attention should be drawn to the potential effects of commonly marketed lactobacillus-containing probiotics on weight gain. Ó 2012 Elsevier Ltd. All rights reserved.

1. Introduction some bacterial groups and human obesity ( Lactobacillus [7], Staphylococcus aureus [8 e10] , Escherichia coli [10] and Faecali- The prevalence of obesity is increasing steadily among adults, bacterium prausnitzii [11] ). Conversely, other bacterial groups have adolescents and children and is now considered a worldwide been associated with lean status, mainly belonging to the Bi fido- epidemic [1]. The causes driving the obesity appear to be complex bacterium genus [6,8e11] . We found recently that different Lacto- and include environmental, genetic, neural and endocrine factors bacillus species may have a paradoxical effect with higher levels of [2] but infectious agents have also been proposed [3]. More recently Lactobacillus reuteri and lower levels of Lactobacillus plantarum and obesity was associated with a speci fic pro file of the bacterial gut paracasei in obese gut microbiota [12] . In contrast, symbiotics (the microbiota [4] and was shown to be a transmissible phenotype by combination of prebiotics and probiotics) have been proposed in microbiota transplantation [5]. First studies on obesity reported the management of malnutrition with promising results on a decrease in the Bacteroidetes/Firmicutes ratio [4] and a decrease in mortality [13] . the archae Methanobrevibacter smithii [6]. Since these pioneering As many probiotic strains of Lactobacillus and Bi fidobacterium studies, significant associations were found between the increase of are marketed in products for human consumption, altering the intestinal flora [14] , we hypothesized that widespread ingestion of probiotics may promote obesity by altering the intestinal flora [15] . * Corresponding author. Tel.: þ33 491 38 55 17; fax: þ33 491 83 03 90. E-mail address: [email protected] (D. Raoult). However, this remains controversial [16] . On the other hand, 1 These authors contributed equally to this work. manipulation of the gut microbiota by probiotics has been used for

0882-4010/$ e see front matter Ó 2012 Elsevier Ltd. All rights reserved. doi:10.1016/j.micpath.2012.05.007 M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108 101 growth promotion in farm animals for at least 30 years [17] . 2.4. Statistical analysis and heterogeneity investigation Indeed, Lactobacillus acidophilus, L. plantarum , Lactobacillus casei, Lactobacillus fermentum, L. reuteri are the most commonly used We used RevMan v5.1 [22] to carry out meta-analysis of the Lactobacillus spp. in agriculture [18] . All these data strongly suggest standardized difference in means (SMD) with 95% con fidence that Lactobacillus containing probiotics (LCP) may impact the interval for weight change after probiotics administration. When weight regulation in humans and animals. means were available and the p-value was described as > 0.05 (or Many studies have reported the effects of Lactobacillus con- “not significant”), a two-tailed p-value of 0.9 was attributed in taining probiotics (LCP) on weight but according to recent data [12] , order to increase the sensitivity of this pioneering work in this area. this effect is at least species dependent. To our knowledge, no meta- Each trial could contribute more than one comparison but analysis has been performed to con firm this difference among comparisons were pooled for each study only if experimental Lactobacillus containing probiotics. For this purpose, we pooled conditions were similar. Heterogeneity was assessed by the data from animal and human studies to obtain suf ficient power to I-squared value, 50% being considered as substantial. Summary detect a signi ficant effect at the species level. measures were determined by a random effect model assuming significant clinical heterogeneity regardless of the I-squared value. 2. Methods We primarily investigated heterogeneity by stratifying results by Lactobacillus species including comparisons with only one Lacto- 2.1. Data sources bacillus species in the probiotic product. In addition, subgroup analyses were planned a priori to discern weight changes by host According to PRISMA 2009 guidelines [19] (Table A.1 ), PubMed, category; overweight/obese animals or humans; and very low Medline, ISI Web of knowledge, Google scholar, Google, Cochrane weight birth (VLWB) newborns. The effect of studies ’ risk of bias Central Register of Controlled Trials ( www.cochrane.org), meta- was assessed through sensitivity analysis. Funnel plot was used to Register of Controlled Trials ( www.controlledtrials.com/mrct ), identify outliers subsequently excluded and to assess small studies clinicaltrials.gov and a recent evidence report/technology assess- and publication bias. Classic fail-safe N, Egger ’s test for asymmetry ment [20] were searched for articles, unrestricted by language, were also used to assess small studies bias and Duval and Tweedie ’s from 1950 to August 2011. Search terms included: probiotics, Trim and Fill adjustment with random effects model was used to Lactobacillus, weight, weight gain, weight loss, weight change, provide an estimate of the unbiased effect size. A standardized growth, performance, randomized controlled trials, placebo- difference in means > 0.10 was considered clinically relevant as it controlled and associated author names. correspond to a 1 kg weight difference for a 70 kg man based on data from a sample of 5000 healthy human individuals [23]. 2.2. Study selection and data extraction 3. Results We retrieved the full text of studies including Lactobacillus -containing probiotics and looked for weight assessment as primary or The search yielded 200 studies of which 118 were excluded secondary outcome. Inclusion was limited to experimental studies because of probiotic or host group de finitions, study design or and randomized controlled trials in farm animals, experimental missing outcome data (Fig. 1). 82 studies, involving 153 compari- models and healthy humans. Authors were contacted when pub- sons, were included in the quantitative synthesis. Included human lished data were incomplete. Exclusion criteria included hosts with studies involved 15 double-blind randomized controlled trials and underlying diseases (except for obesity) or pregnant women, pro- two open-labelled randomized trials ( Table 1 ). Included animal biotics given only to the mother, symbiotics (probiotics associated studies involved 14 studies on experimental models and 51 on farm with prebiotics), other nutrients given exclusively to the intervention animals. After exclusion of two studies with high risk of bias group, non-direct fed microbials (probiotics in silage), non viable (weight signi ficantly different at baseline, open label design), LCPs probiotic administration, recombinant probiotics, hosts challenged were not associated with signi ficant weight change in human prior to probiotic administration by viruses or bacteria, hosts with adults (SMD ¼ 0.18; 95%CI ( À0.43e0.79)), infants (SMD ¼ 0.004; diarrhoea or colitis, before-after intervention studies, inappropriate 95%CI (À0.20e0.21)), or preterm newborn infants (SMD ¼ À 0.10; control group (prebiotics, probiotics or antibiotics administration e 95%CI (À0.32e0.12)). Meta-analysis of all comparisons in healthy traditional yogurt including Lactobacillus delbrueckii subsp. bulgar- humans and animals (134 comparisons, overweight and VLWB icus and Streptococcus thermophilus was accepted as control inter- newborns excluded) resulted in weight gain but signi ficant vention), unavailable statistical data and double publications. Data heterogeneity (SMD ¼ 0.15; 95%CI (0.12 e0.18); p < 0.001; I2 ¼ 85%) were extracted independently by two authors (MM, EA). and thus we proceeded directly to the subgroup analyses, primarily assessing effects by species. 2.3. Risk of bias assessment and outcome measures 3.1. Lactobacillus acidophilus The Jadad score [21] was used for the assessment of bias in eval- uating human trials to determine studies to exclude and allowing The meta-analysis of 13 studies and 18 comparisons including sensitivity analysis based on this quality score. In animals, all studies 3307 subjects (879 humans) on L. acidophilus administration were included except those with major methodological concerns, showed a signi ficant weight gain effect (SMD ¼ 0.15; 95%CI and a score was calculated with one point for each of these terms: (0.05e0.25); p ¼ 0.005; I2 ¼ 42%) (Fig. 2). Using classic fail-safe N, dropouts mentioned, dropouts < 10%, outcome expressed as aweight 34 unpublished studies would have been necessary to bring difference (and not weight at the end) and absence of other risk of p-value > 0.05, Duval and Tweedies ’s trim and fill did not change bi as. Studies with a score > 2 were considered at low risk for bias. this result, and Egger ’s asymmetry test was not signi ficant (two- The primary outcome was the effect on weight. Weight change tailed p-value ¼ 0.66) making this summary effect robust and from baseline, weight at the end of the study, daily weight publication bias unlikely. Direction of effect favouring weight gain change, weight/age ratio, delta-BMI (Body Mass Index) and weight was consistent in humans and animals. A sensitivity analysis percentile were considered as outcome measures. We primarily including only studies with a quality score > 2 reduced heteroge- attempted to extract and use change from baseline values. neity and found a consistent and signi ficant result ( I2 ¼ 28%, 102 M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108

p ¼ 0.03; I2 ¼ 74%) ( Fig. 3b). No human studies included L. plantarum . Small studies bias was unlikely because 19 unpublished studies would have been necessary to bring the p-value to > 0,05, Duval and Tweedies ’s trim and ﬁll did not change this result and Egger ’s asymmetry test was not signi ﬁcant (two-tailed p-value ¼ 0.52). All these comparisons in experimental animals had a medium to low risk of bias.

3.5. Lactobacillus gasseri

L. gasseri was associated with a trend for weight loss in lean animals in three studies and four comparisons including 48 pigs and rats ( p ¼ 0.09) (Fig. 3a). In obese animals and humans ( Fig. 3b), three studies and three comparisons including 87 humans and 36 rats found an anti-obesity effect (SMD ¼ À 0.67; 95%CI ( À1.17 to À0.16); p ¼ 0.009; I2 ¼ 29%). Using classic fail-safe N, six unpublished studies would have been necessary to bring the p-value to > 0,05, Duval and Tweedies ’s trim and fill did not change this result and Egger’s asymmetry test was not signi ficant (two- tailed p-value ¼ 0.88) making this summary effect robust and small studies bias unlikely. This effect was consistent between humans and animals. Two L. gasseri strains (SBT2055 [25e27] and BNR17 [28]) have a signi ficant anti-obesity effects in individual studies. All these studies had a medium to low risk of bias. In obese individuals, the difference (SMD ¼ À 0.57) was clinically relevant since it correspond to a weight loss of 6 kg in humans. Fig. 1. Studies flow through meta-analysis according to PRISMA guidelines [19] .

3.6. Other species p ¼ 0.005). The difference (SMD ¼ 0.15) was clinically relevant as it corresponds to a weight gain of 1.5 kg for a 70 kg man. Other species (L. reuteri , L. casei , Lactobacillus rhamnosus and Lactobacillus sporogenes) were not associated with signi ficant and 3.2. Lactobacillus fermentum consistent effects. Only L. delbrueckii was signi ficantly associated with weight gain ( five comparisons; I2 ¼ 0%; SMD ¼ 0.39; 95%CI The meta-analysis of 3 studies and 12 comparisons including (0.06e0.71); p ¼ 0.02) but this effect was summarized from only 598 chicks, pigs and ducks but no humans on L. fermentum found two different studies in chicks and rats. a signi ficant weight increase (SMD ¼ 0.81; 95%CI (0.12 e1.50); p ¼ 0.02; I2 ¼ 90%). After exclusion of one outlier, 34 unpublished 4. Discussion studies would have been necessary to bring p-value > 0,05 using classic fail-safe N, Duval and Tweedies ’s trim and fill one study but 4.1. Signi ficant impact of Lactobacillus-containing probiotics on found a consistent result (SMD ¼ 0.53; 95% CI (0.18 e0.87)) and weight Egger’s asymmetry test was not signi ficant (two-tailed p- value ¼ 0.28) making this summary effect robust. All these studies In this meta-analysis, we showed that some Lactobacillus species have a quality score > 2. were signi ficantly associated with weight modi fications in human and animals: L. acidophilus , L. ingluviei , L. fermentum were linked to 3.3. Lactobacillus ingluviei weight gain whereas L. gasseri and L. plantarum were linked to weight loss or an anti-obesity effect. The latter effect seemed The meta-analysis of three studies and 11 comparisons particularly evident in overweight or obese individuals. Wide including 198 chicks, ducks and mice on one L. ingluviei strain variation in response was explained by probiotic species and host. isolated from an ostrich found a signi ficant weight increase effect Strati fication only by probiotic species revealed signi ficant and (SMD ¼ 0.97; 95%CI (0.49 e1.45); p < 0.001; I2 ¼ 59%). After consistent results. In a second step, we showed that the host was exclusion of one outlier [24], 27 unpublished studies would have a covariate explaining part of the heterogeneity found for a speci fic been necessary to bring a p-value > 0.05 using classic fail-safe N, probiotic species ( Table A.2 ). The differences found were clinically Egger’s asymmetry test was not signi ficant (p ¼ 0.59) and Duval relevant as they correspond to a weight change that ranged from and Tweedies’s trim and fill give a similar signi ficant result 1.5 kg gain in leans for L. acidophilus or 6 kg loss in overweight for (SMD ¼ 0.76; 95% CI (0.48 e1.03)). All these studies had a quality L. gasseri based on the statistics of a population of >600 healthy score > 2. No human trials included L. ingluviei . men with an average weight of 70 kg (standard deviation of 9.8 kg) [23]. 3.4. Lactobacillus plantarum 4.2. Lactobacillus species associated with weight gain Pooled analysis of three studies, three comparisons including 335 lean chicks, rats and mice on L.plantarum showed a weight loss With the results of our meta-analysis, bacteria candidates for effect direction but result was not signi ficant (Fig. 3a). However, L. increasing energy ef ficiency in humans are L. acidophilus and plantarum was associated with signi ficant weight loss effect in L. fermentum . To our knowledge, L. ingluviei was not identi fied in the overweight/obese animals in four studies and five comparisons human digestive microbiota but only in the intestinal tract of including 64 mice and rats (SMD ¼ À 1.33; 95%CI ( À2.50 to À0.16); pigeons, chickens and ostrich and is not contained in probiotics Table 1 Characteristics of included human studies.

prospective, double-blind, weight 2500 g) enrolled at congenital malformation, gain (g/d) on 120 d e

formula supplemented or not 108 randomized trial; low (4) 14 d of age, solely formula fed; formula intolerance, weight at with8 L. rhamnosus strain GG M/F ¼ 94/94; (188, 55, 133) e cfu/g of formula powder); 14 d of age 98% of birth (10 one group using different weight, large for gestational age 120 days casein formula was excluded; born from a mother diabetic at treated/control: 63/70 childbirth, immunode ﬁciency, fever, antibiotic within 7 d, systemic steroid since birth, LGG-suppl diet since birth, diarrhoea within 24 h ﬂ Vendt et al., 2006 Finland; 2002; multicentric; Healthy infants from 0 to 2 Not mentioned (reasons for L. rhamnosus strain GG Growth b and fecal ora on 6 7 prospective, double-blind, months on formula at least half discontinuation: colic pain, ATCC53103 (1 Â 10 cfu); till months randomized trial; medium (3) of their daily feedings; M/ cow ’s milk protein intolerance, the age of 6 months F ¼ 60/60; (120, 15, 105); constipation, diarrhoea, treated/control: 51/54 excessive breastfeeding) Weizman et al., 2006 Israel; 2006; monocentric; Full term healthy infants aged <36 wks gestation, birth L. reuteri ATCC55730 (BioGAIA Growth parameters, daily 8 prospective, double-blind, 3e65 days solely formula fed; weight < 2500 g, congenital AB, Sweden) (1 Â 10 cfu); 4 characteristics of feeding, randomized trial; low (4) M/F ¼ 26/13; (39; 7;32); anomalies, chronic disease, weeks stooling and behaviour and side treated/control: 16/17. failure to thrive (weight loss effects. of > 2 percentiles), allergy or atopic disease and recent (less (continued on next page ) 103 Table 1 (continued ) 104

Study source Location; period of inclusion; Subjects included; age and sex; Exclusion criteria Probiotic (dose); duration of Outcomes (primary in first); mono or multicentric; study sample size (subjects enrolled, treatment weight change assessment design; risk of bias (Jadad score) dropped out, used); no of (unit) treated /control subjects than four weeks) exposure to probiotics or antibiotics. Lactobacillus probiotics in lean adults 9 to Lipid pro file and body weight De Roos et al., 1999 The Netherlands; period of Healthy adults between 18 and Heart disease, diabetes, liver or L.3 Â acidophilus10 10 cfu d)L-1 (5 Â 10 change; weight change inclusion not mentioned; 65 years; BMI 24 Æ 3, at least kidney disease, medications difference (kg) monocentric; prospective, 50% of the enrolled volunteers known to affect blood lipid double-blind, randomized trial; had serum cholesterol levels metabolism, serum total medium (3) over 5 mmol/L, M/F ¼ 22/56; cholesterol concentration (85,7,78); treated/control: higher than 8 mmol/L or 39/39 a triacylglycerol concentration higher than 4 mmol/L Ò Fabian et al., 2007 Austrich; period of inclusion Healthy adults women (BMI: Smoking, Actimel (L. paracasei subsp. Antioxidants and oxidant 2 not mentioned; monocentric; 21 Æ 3 kg/m ) e 22 e29 years hypercholesterolemia, Paracasei (L. casei DN-114 001 ) parameters in plasma; weight 8 prospective, randomized trial (33, 1, 32); treated/control : pregnancy, overweight or (3.6 Â 10 cfu/g)); 4 weeks change (kg) b (blinding not mentioned); high 16/16 metabolic disease, allergies or 100 (2012) 53 Pathogenesis Microbial / al. et Million M. (2) intolerance, regular medications except oral contraceptive fi Sadrzadeh-Yeganeh Iran; period of inclusion not Healthy adults women Smoking, kidney, liver or L. acidophilus La1 and Lipid pro le; weight change fl et al., 2010 mentioned; prospective, (cholesterol < 6.2 mmol/l, in ammatory intestinal Bi fidobacterium. lactis Bb12 (kg) 7 double-blind, randomized trial; TAG < 2.3 mmol/l, BMI < 30 kg/ disease, thyroid disorders, (4 Â 10 cfu); 6 weeks 2 medium (3) m ) (90,1,89) e one group diabetes, immunode ficiency, excluded (no yoghurt) lactose intolerance; taking Treated/control: 30/29 supplements or medication, probiotic consumption in the last 2 months, elite athletes, pregnant or breastfeeding women Lactobacillus probiotics in overweight/obese adults Agerholm-Larsen Denmark; period of inclusion Healthy weight-stable Diabetes, kidney or liver 2 strains of L. acidophilus Lipid pro file and body weight; 7 et al., 2000 not mentioned; monocentric; overweight and obese disease, high blood pressure, (2 Â 10 /ml) and 1 strain of weight change (kg) 2 7 prospective, double-blind, (25 < BMI <37.5 kg/m ); mean pregnancy, breastfeeding, elite S. thermophilus (10 Â 10 /ml); 2 ¼ randomized trial; low (4) 38 years, M/F 4/12, 5/9 and 4/ athlete, chronic ethylism strains of S. thermophilus e 10; (73,3,70); treated (8 Â 10 8/ml) and 1 strain of 108 L. acidophilus 16/ L. rhamnosus L. rhamnosus (2 Â 10 8/ml) 14/control 14 10 Kadooka et al., 2010 Japan; 2008; multicentric Healthy adults with body mass Serious disorders, including L. gasseri SBT2055 (5 Â 10 cfu/ Abdominal adiposity and body (n ¼ 10); prospective, double- index (BMI) between 24.2 and internal organ diseases, 100 g) e200 g/day; 12 weeks weight; weight change (kg) 2 blind, randomized trial; 30.7 kg/m , abdominal visceral diabetes and hypersensitivity to medium (3) fat area between 81.2 and dairy products. 178.5 cm 2 aged 33 e63 years, M/F ¼ 59/28; (87,0,87); treated/ control: 43/44 Ò Woodard et al., 2009 USA; 2006 e2007; monocentric; Morbidly obese patients No exclusion criteria Puritan ’s Pride (2.4 Â 10 9 cfu/ Bacterial overgrowth, weight prospective, double-blind, undergoing Roux-en-Y gastric mentioned pill) one pill a day e no loss, quality of life; percent randomized trial; low (5) bypass (RNYGB) (BMI w 45 kg/ characterization of the excess weight loss (%) 2 m ); Age 40 e50 yrs, M/F ¼ 5/ Lactobacillus strains; 6 months 36; (44, 8, 35); treated/control:15/20 a g calculated from ounce, Â28.35. b Data given by the authors under request e in this study, weight of treated group was signi ficantly lower at baseline. M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108 105

Fig. 2. Forest plot of three Lactobacillus species associated with weight gain in humans and animals (a) L. acidophilus , (b) L. fermentum , (c) L. ingluviei (strain isolated from an ostrich [24]). A weight gain effect is represented by a deviation of the square (standardized difference in the means) to the right. The length of the horizontal line represents the 95% conﬁdence interval and the diamond represents the summarized effect. Substantial heterogeneity was cancelled after sensitivity analysis for L. acidophilus . L. acidophilus and L. fermentum , when administered in overweight/obese humans or animals didn ’t have a signi ﬁcant anti-obesity effect (data not shown).

marketed for humans. The L. ingluviei comparisons included in this effect was observed with two species having a signi ficant anti- analysis involved only one strain, isolated from an ostrich gut, obesity effect; L. gasseri and L. plantarum (Fig. 3b). This anti- showing an astonishing weight gain effect both in farm animals and obesity effect was consistent with absence of signi ficant weight- experimental models [24,29] . One candidate for the transmission of gain effect in lean individuals ( Fig. 3a). A recent study con firmed the obese phenotype, L. acidophilus , is widely present in many prod- our results showing a 6-months weight-loss effect of L. plantarum ucts for human consumption as the “acidophilus milk ”, traditionally DSM 15313 in high-energy-dense diet rats when administered to consumed in the United States or in other formulations such as freeze- mother and offspring [31] . This anti-obesity effect could be an dried products sold without any regulation on the internet. The important adjunct in the treatment of obesity, since, apart from consumption of this species is particularly prevalent in the United surgery, no medical treatment can support ef ficiently the fight States [30] where the prevalence of obesity is particularly important. against obesity to date.

4.3. Lactobacillus species associated with anti-obesity effect 4.4. Limitations

On the other hand, many bacteria appear to be protective against However, the paucity of data in individual hosts impelling us to obesity. In our study, a strong species dependent anti-obesity the pool animal and human data limits the generalization of these 106 M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108

Fig. 3. Forest plot of three Lactobacillus species associated with an anti-obesity effect in humans and animals. Lactobacillus plantarum and L. gasseri and weight changes in (a) lean (b) and overweight/obese humans and animals. Weight loss effect is represented by a deviation of the square (standardized difference in the means) to the left. The length of the horizontal line represents the 95% confidence interval and the diamond represents the summarized effect. Anti-obesity effect was consistent for these two species. data to humans. Moreover, effect size and standard deviation are papers in nutrition and obesity research in which the authors were probably very different in experimental models and in the general funded by industry were more likely than other papers to contain human population. Only few clinical studies have been conducted results or an interpretation that favored the industry or company to test a weight gain effect assessing only one Lactobacillus species that was producing the product or service that was being studied because, unlike animal studies, this effect was generally not sought [33] . Furthermore, while a comprehensive search was performed in humans. L. acidophilus increased weight gain both in bottle-fed using several strategies, we cannot be sure that all studies exam- and breast-fed newborns but this effect was stronger in bottle- ining the effects of LCPs on weight have been identi fied. fed newborns [32], L. gasseri SBT2055 resulted in signi ficant weight loss in human adults with obese tendencies [27] . 4.6. Perspectives

4.5. Conflict of interest in nutrition and obesity research In the next future, new double-blind randomized human trials should assess long-term growth in newborn infants receiving Finally, it is possible that the design and/or interpretation of the Lactobacillus-containing probiotics. A critical point is to stratify results of each individual study had been affected by a con flict of according to the initial weight [34]. For species associated here with interest of each team. It has recently been shown that published a signi ficant weight change and used for human consumption as M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108 107

L. acidophilus , L. fermentum , L. plantarum and L. gasseri , trials eval- Funding uating weight gain as a primary outcome measure will be needed. The long-term evaluation to at least 3 e5 years of age will be critical URMITE -CNRS UMR 6236 IRD 198. to identify a difference that could have been undetected by shorter studies [35]. According to the register Clinicaltrials.gov , at least one Ethical approval current study is testing the average weight gain as a primary endpoint among newborns receiving a probiotic containing Not required. L. fermentum (trial number NCT01346644). This bacterium has been associated with obesity in our study. In addition, case econtrol Acknowledgements studies comparing obese and lean children according to their consumption of Lactobacillus-containing probiotics in the first We thanks all authors that kindly answer and send us requested weeks of life should be carried out. data and especially Asal Ataie, Wageha Awad, Jonna Aaltonen, Eli- sabeth Fabian, Kirsi Laitinen, Kostas Mountzouris, Denis Roy, Nancy Szabo, Neve Vendt and Christina West. 5. Conclusion Appendix A. Supplementary material Food is a source of bacteria and viruses and changes in patterns of food consumption result in differences in human gut flora among Supplementary material associated with this article can be different groups of people [36,37] . As a result, it is necessary to found, in the online version, at doi:10.1016/j.micpath.2012.05.007 . further investigate the effects of routinely adding high amounts of bacteria in food [38]. Our systematic analysis found that the References manipulation of the gut microbiota by L. acidophilus , L. ingluviei or e L. fermentum results in weight gain whereas speci fic strains of [1] Yanovski SZ, Yanovski JA. Obesity. N Engl J Med 2002;346:591 602. [2] Tilg H, Moschen AR, Kaser A. Obesity and the microbiota. Gastroenterology L. gasseri and L. plantarum used as food supplements presented an 2009;136:1476e83. anti-obesity effect. Only two studies including these species were [3] Dhurandhar NV. A framework for identi fication of infections that contribute to available in humans, one showing a signi ficant weight gain effect of human obesity. Lancet Infect Dis 2011;11:963e9. [4] Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: human gut L. acidophilus in newborns whereas L. gasseri was found to have microbes associated with obesity. Nature 2006;444:1022e3. a signi ficant anti-obesity effect in the first and only well-designed [5] Turnbaugh PJ, Backhed F, Fulton L, Gordon JI. Diet-induced obesity is linked to study to date assessing the impact of Lactobacillus-containing marked but reversible alterations in the mouse distal gut microbiome. Cell Host Microbe 2008;3:213e23. probiotics on overweight humans. L. acidophilus and L. gasseri were [6] Schwiertz A, Taras D, Schafer K, Beijer S, Bos NA, Donus C, et al. Microbiota and associated with the same effect direction both in animals and SCFA in lean and overweight healthy subjects. Obesity (Silver Spring) 2010; humans. L. fermentum and L. ingluviei were associated with an 18:190e5. [7] Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. Monitoring bacterial astonishing weight gain effect in ducks, chicks and mice but have community of human gut microbiota reveals an increase in Lactobacillus in never been studied in humans. Next-generation human probiotic obese patients and Methanogens in anorexic patients. PLoS One 2009;4:e7125. species should contain Lactobacillus spp. that are not associated [8] Kalliomaki M, Collado MC, Salminen S, Isolauri E. Early differences in fecal with weight gain in humans. Of note, on 24 August 2007, the FDA microbiota composition in children may predict overweight. Am J Clin Nutr 2008;87:534e8. issued regulations that require current good manufacturing prac- [9] Collado MC, Isolauri E, Laitinen K, Salminen S. Distinct composition of gut tice for dietary supplements to be phased in over the next few years microbiota during pregnancy in overweight and normal-weight women. Am J e [39]. These regulations should focus on weight assessment Clin Nutr 2008;88:894 9. [10] Santacruz A, Collado MC, Garcia-Valdes L, Segura MT, Martín-Lagos JA, Anjos T, outcome according to probiotic species and strains. Finally, selec- et al. Gut microbiota composition is associated with body weight, weight gain tion of speci fic Lactobacillus containing probiotics could take part in and biochemical parameters in pregnant women. Br J Nutr 2010;104:83 e92. the future management of the two major health problems in the [11] Balamurugan R, George G, Kabeerdoss J, Hepsiba J, Chandragunasekaran AM, Ramakrishna BS. Quantitative differences in intestinal Faecalibacterium 21st century, malnutrition and obesity. prausnitzii in obese Indian children. Br J Nutr 2010;103:335 e8. [12] Million M, Maraninchi M, Henry M, Armougom F, Richet H, Carrieri P, et al. Obesity-associated gut microbiota is enriched in Lactobacillus reuteri and Competing interest statement depleted in Bi fidobacterium animalis and Methanobrevibacter smithii. Int J Obes (Lond); 2011. doi:10.1038/ijo.2011.153 [Epub ahead of print]. [13] Kerac M, Bunn J, Seal A, Thindwa M, Tomkins A, Sadler K, et al. Probiotics and All authors have completed the Uni fied Competing Interest form prebiotics for severe acute malnutrition (PRONUT study): a double-blind at www.icmje.org/coi_disclosure.pdf (available on request from the ef ficacy randomised controlled trial in Malawi. Lancet 2009;374:136 e44. fi corresponding author) and declare that (1) no authors have support [14] Fujimoto J, Matsuki T, Sasamoto M, Tomii Y, Watanabe K. Identi cation and quantification of Lactobacillus casei strain Shirota in human feces with strain- from probiotics companies for the submitted work; (2) no authors specific primers derived from randomly ampli fied polymorphic DNA. Int J have relationships with probiotics companies that might have an Food Microbiol 2008;126:210e5. interest in the submitted work in the previous 3 years; (3) their [15] Raoult D. Human microbiome: take-home lesson on growth promoters? e fi Nature 2008;454:690 1. spouses, partners, or children have not nancial relationships that [16] Delzenne N, Reid G. No causal link between obesity and probiotics. Nat Rev may be relevant to the submitted work; and (4) no authors have Microbiol 2009;7:901. non-financial interests that may be relevant to the submitted work. [17] Fuller R. Probiotics in man and animals. J Appl Bacteriol 1989;66:365 e78. [18] Anadon A, Martinez-Larranaga MR, Aranzazu MM. Probiotics for animal nutrition in the European Union. Regulation and safety assessment. Regul Toxicol Pharmacol 2006;45:91e5. Contributors [19] Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gøtzsche PC, Ioannidis JP, et al. The PRISMA statement for reporting systematic reviews and meta-analyses of DR conceived and designed the study, MM & EA extracted the studies that evaluate health care interventions: explanation and elaboration. PLoS Med 2009;6:e1000100. data, MM, FA, MP, LL analysed the data. MM, EA and DR wrote the [20] Hempel S, Newberry S, Ruelaz A, Wang Z, Miles JNV, Suttorp MJ, et al. Safety of manuscript. MP & LL revised the paper. MM had full access to all of probiotics to reduce risk and prevent or treat disease. Evidence report/tech- the data in the study and takes responsibility for the integrity of the nology assessment no. 200. Agency for Healthcare Research and Quality [21] Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, Gavaghan DJ, et al. data and the accuracy of the data analysis and is guarantor. MM and Assessing the quality of reports of randomized clinical trials: is blinding neces- EA contributed equally to this work. sary? Control Clin Trials 1996;17:1 e12. 108 M. Million et al. / Microbial Pathogenesis 53 (2012) 100 e108

[22] The Cochrane Collaboration. Review Manager (RevMan) [computer program]. [34] Vendt N, Grunberg H, Tuure T, Malminiemi O, Wuolijoki E, Tillmann V, et al. Version 5.1. Copenhagen: The Nordic Cochrane Centre; 2011. Growth during the first 6 months of life in infants using formula enriched [23] Khosla T, Lowe C. Indices of obesity derived from body weight and height. Br J with Lactobacillus rhamnosus GG: double-blind, randomized trial. J Hum Nutr Prev Soc Med 1967;21:122e8. Diet 2006;19:51e8. [24] Khan M, Raoult D, Richet H, Lepidi H, La Scola B. Growth-promoting effects of [35] Chouraqui JP, Grathwohl D, Labaune JM, Hascoet JM, de Montgol fier I, single-dose intragastrically administered probiotics in chickens. Br Poult Sci Leclaire M, et al. Assessment of the safety, tolerance, and protective effect 2007;48:732e5. against diarrhea of infant formulas containing mixtures of probiotics or pro- [25] Hamad EM, Sato M, Uzu K, Yoshida T, Higashi S, Kawakami H, et al. biotics and prebiotics in a randomized controlled trial. Am J Clin Nutr 2008; Milk fermented by Lactobacillus gasseri SBT2055 in fluences adipocyte 87:1365e73. size via inhibition of dietary fat absorption in Zucker rats. Br J Nutr 2009;101: [36] Raoult D. The globalization of intestinal microbiota. Eur J Clin Microbiol Infect 1e9. Dis 2010;29:1049e50. [26] Sato M, Uzu K, Yoshida T, Hamad EM, Kawakami H, Matsuyama H, et al. Effects [37] Gordon JI, Klaenhammer TR. A rendezvous with our microbes. Proc Natl Acad of milk fermented by Lactobacillus gasseri SBT2055 on adipocyte size in rats. Sci U S A 2011;108(Suppl. 1):4513 e5. Br J Nutr 2008;99:1013 e7. [38] Raoult D. Obesity pandemics and the modi fication of digestive bacterial flora. [27] Kadooka Y, Sato M, Imaizumi K, Ogawa A, Ikuyama K, Akai Y, et al. Regulation Eur J Clin Microbiol Infect Dis 2008;27:631 e4. of abdominal adiposity by probiotics ( Lactobacillus gasseri SBT2055) in adults [39] Sanders ME. Probiotics: de finition, sources, selection, and uses. Clin Infect Dis with obese tendencies in a randomized controlled trial. Eur J Clin Nutr 2010; 2008;46(Suppl. 2):S58e61. 64:636e43. [28] Kang JH, Yun SI, Park HO. Effects of Lactobacillus gasseri BNR17 on body weight Glossary and adipose tissue mass in diet-induced overweight rats. J Microbiol 2010;48: 712e4. Probiotics: Probiotics are de fined as “Live microorganisms which when adminis- [29] Angelakis E, Raoult D. The increase of Lactobacillus species in the gut flora of tered in adequate amounts confer a health bene fit on the host. ” According to newborn broiler chicks and ducks is associated with weight gain. PLoS ONE FAO/WHO. 2010;5:e10463. Obesity: According to the WHO, Obesity is de fined by a BMI > 30 kg/m 2 and [30] Chandan RC. History and consumption trends. In: Chandan RC, White CH, a massive expansion of fat, and is associated with a signi ficant increase in Kilara A, Hui YH, editors. Manufacturing Yogurt and Fermented Milks. Ames: morbidity and mortality. Blackwell Publishing; 2006. p. 3 e15. Lactobacillus: Lactobacillus is a genus of Gram-positive facultative anaerobic or [31] Karlsson C, Molin G, Fak F, Johansson Hagslätt ML, Jakesevic M, Håkansson Å, microaerophilic rod-shaped bacteria. They are a major part of the lactic acid et al. Effects on weight gain and gut microbiota in rats given bacterial bacteria group, named as such because most of its members convert lactose and supplements and a high-energy-dense diet from fetal life through to 6 months other sugars to lactic acid. In humans they are present in the vagina and the of age. Br J Nutr 2011;106:887 e95. gastrointestinal tract. They are largely present in food products for human and [32] Robinson EL, Thompson WL. Effect on weight gain of the addition of Lacto- animal consumption as probiotics. bacillus acidophilus to the formula of newborn infants. J Pediatr 1952;41: Meta-analysis with random effect model : A meta-analysis combines the results of 395e8. several studies by identi fication of a common measure of effect size, of which [33] Thomas O, Thabane L, Douketis J, Chu R, Westfall AO, Allison DB. Industry a weighted average might be calculated. A random effect model assumes that funding and the reporting quality of large long-term weight loss trials. Int J heterogeneity is due at least in part to the different experimental conditions Obes (Lond) 2008;32:1531e6. between individual studies.

Article VII :

Species and strain specificity of Lactobacillus

probiotics effect on weight regulation

Matthieu Million, Didier Raoult

Published in Microb Pathog. 2013 Feb;55:52-4. (IF 1.94)

Microbial Pathogenesis 55 (2013) 52 –54

Contents lists available at SciVerse ScienceDirect

Microbial Pathogenesis

journal homepage: www.elsevier.com/locate/micpath

Letter to the Editor Species and strain speci ﬁcity of Lactobacillus probiotics effect on weight regulation

abstract

Keywords: Certain strains of Lactobacillus appear to have a reproducible effect on weight as a weight-gain effect in Obesity lean humans and animals or a weight-loss effect in overweight/obese humans and animals. These results Probiotics are completely sufﬁcient to capture the attention of the scienti ﬁc community to assess the effect on the Meta-analysis weight of Lactobacillus-containing probiotics sold for human consumption. Lactobacillus Ó 2012 Elsevier Ltd. All rights reserved. Weight

Dear Editor, described in three separate studies on chicks (broiler chicks Ross308, Hubbard JV, Cobb500) and ducks (Hybrid PKB) [12 –14] We read with interest the comments of Dr. Morelli regarding our and has been confirmed by another study not included in our article published in Microbial pathogenesis entitled “Comparative meta-analysis with a signi ficant 5% weight-gain in chickens [15] , meta-analysis of the effect of Lactobacillus species on weight-gain whereas this effect was less intense (2%) and was not signi ficant in humans and animals. ” for Enterococcus faecium M74. The L. fermentum CCM7158 strain is Dr. Morelli pointed out the fact that the Lactobacillus acidophilus widely marketed under the trade name PROPOUL Ô in order to group includes several Lactobacillus species usually susceptible to fatten chickens. Another L. fermentum strain, not deposited but vancomycin and whose taxonomy has been dif ficult to clarify typed through standard morphological, biochemical, physiological because even until very recently strains identi fied as L. acidophilus tests and 16S rRNA gene by sequence analysis by the China General are being reclassi fied in other species [1]. Taxonomy of this group Microbiology Culture Collection Center [16] was associated with identified 3 clusters by the DNA homology with L. acidophilus , a weight-gain of 9 –20% among weaned pigs and these results Lactobacillus gasseri and Lactobacillus johnsonii [2]. Indeed, accord- were signi ficant. The latter study was conducted at the National ing to data from the American Type Culture Collection, the strains Key Lab of Animal Nutrition, Beijing China so that consistent results (ATCC 4962 and 4963) used by Robinson et al. [3] in 1952 identi fied were obtained from 3 different countries (Slovaky, France and initially as L. acidophilus corresponds in realty to L. gasseri . In addi- China) using three different strains of L. fermentum . tion, the NP51 strain identi fied as L. acidophilus used in the article of For L. gasseri , studies showing a protective effect against obesity in Elam et al. [4] in 2003 has been reclassi fied as Lactobacillus animalis humans and animals correspond to two strains SBT2055 [17 –19] [5]. However in this study we couldn ’t verify if the other strain used and BNR17 [20] , which correspond to recent articles and in combination (LA45 deposited in ATCC as PTA-6749) corresponds therefore are most probably correctly identi fied at the species level. to the species L. acidophilus or not. These discrepancies pointed out For L. plantarum , Karlsson [21] , who published in British Journal of the importance of characterizing each strain using the most recent Nutrition, used the DSM15313 strain with a 11% decrease in weight taxonomic means with the deposition of discriminating DNA at 6 months compared with control rats fed a high-energy-dense sequences in international database, and we highlight this critical diet. Lee et al. [22] used a strain isolated from human feces by the point in a recent article on probiotics for human consumption [6]. authors identi fied as L. plantarum PL62 sequencing the 16S gene but Nevertheless, taking into account the remarks of Dr. Morelli, we authors did not gave the obtained sequence. Takemura [2 3] used repeated the meta-analysis on L. acidophilus excluding studies cited a L. plantarum strain (strain 14) for which we could not obtain the by Dr. Morelli namely Robinson, 1952 [3] – Elam, 2003 [4] – Bra- identi fication techniques. shears, 2003 [7] – Peterson, 2007 [8]. Even after exclusion of these Overall, even taking into account the taxonomic corrections studies, the weight-gain effect of L. acidophilus remains signi ficant mentioned by Dr. Morelli, the main message of our work is essen- (Fig. 1 (random model, I2 ¼ 46%, p-value for overall effect ¼ 0.01)). tially the same: that certain strains of Lactobacillus appear to have In addition, Dr. Morelli seems to completely neglect the consis- a reproducible effect on weight as a weight-gain effect in lean tent results obtained for Lactobacillus fermentum and Lactobacillus humans and animals or a weight-loss effect in overweight/obese ingluviei in lean animals with a weight-gain effect and consistent humans and animals. These results are completely suf ficient to results for an anti-obesity effect of L. gasseri and Lactobacillus plan- capture the attention of the scienti fic community to assess the tarum found specifically in overweight/obese animals. effect on the weight of Lactobacillus-containing probiotics sold for In our laboratory, three studies have found a similar weight-gain human consumption. effect with the same strain isolated from an ostrich [9 –11] . Outside The fact that two strains of the same species can have contradic- our laboratory, the weight-gain effect of L. fermentum CCM7158 was tory effects is not impossible as it has been shown that different

step alerting the scienti fic community to clarify the effect Lactoba- cillus-containing probiotics on weight. This effect is not limited to the inhibition of pathogenic bacteria since this weight-gain effect after Lactobacillus probiotic administration is present when the loss depends only on a protein de ficit [32], improving the intestines’ ability to absorb and process nutrients [33] and significantly increasing retention of fat [34]. Finally, it may be noticed that our works enrage mainly people having direct links with the food industry such as Ehrlich [35].

References

[1] Voronina OL, Kunda MS, Bondarenko VM, Shabanova NA, Lunin VG. Re fine- ment of taxonomic position of Lactobacillus genus probiotic strains by 16S rDNA and rpoA gene sequencing. Zh Mikrobiol Epidemiol Immunobiol 2012;3:18–24. [2] Johnson JL, Phelps CF, Cummins CS, London J, Gasser F. Taxonomy of the Lacto- bacillus acidophilus group. Int J Syst Bacteriol 1980;30:53 –68. [3] Robinson EL, Thompson WL. Effect on weight gain of the addition of Lactoba- cillus acidophilus to the formula of newborn infants. J Pediatr 1952;41:395 –8. [4] Elam NA, Gleghorn JF, Rivera JD, Galyean ML, Defoor PJ, Brashears MM, et al. Effects of live cultures of Lactobacillus acidophilus (strains NP45 and NP51) and Propionibacterium freudenreichii on performance, carcass, and intestinal characteristics, and Escherichia coli strain O157 shedding of finishing beef steers. J Anim Sci 2003;81:2686–98. [5] Randhawa S, Brashears MM, McMahon KW, Fokar M, Karunasena E. Compar- ison of phenotypic and genotypic methods used for the species identi fication of Lactobacillus NP51 and development of a strain-speci fic PCR assay. Probiot- Fig. 1. Forest plot of three Lactobacillus species associated with weight gain in animals ics Antimicrob Prot 2010;2:274–83. (one study by de Roos et al. focused on Lactobacillus acidophilus included humans with [6] Angelakis E, Million M, Henry M, Raoult D. Rapid and accurate bacterial iden- fi a non signi cant weight gain effect). A weight gain effect is represented by a deviation ti fication in probiotics and yoghurts by MALDI-TOF mass spectrometry. J Food of the square (standardized difference in the mean) to the right. The length of the hori- Sci 2010;76:M568–72. zontal line represents the 95% confidence interval and the diamond represents the [7] Brashears MM, Galyean ML, Loneragan GH, Mann JE, Killinger-Mann K. Prev- summarized effect. alence of Escherichia coli O157:H7 and performance by beef feedlot cattle given Lactobacillus direct-fed microbials. J Food Prot 2003;66:748 –54. [8] Peterson RE, Klopfenstein TJ, Erickson GE, Folmer J, Hinkley S, Moxley RA, et al. Effect of Lactobacillus acidophilus strain NP51 on Escherichia coli O157:H7 fecal shedding and finishing performance in beef feedlot cattle. J Food Prot 2007; Bi fidobacterium strains might improve (strain M13-4) or decrease 70:287–91. body weight (strain L66-5) of high-fat diet obese rats [24], in addi- [9] Khan M, Raoult D, Richet H, Lepidi H, La Scola B. Growth-promoting effects of single-dose intragastrically administered probiotics in chickens. Br Poult Sci tion the whole genome content of two strains of the same species 2007;48:732–5. (L. johnsonii NCC 533 and FI9785) share only 64% of the proteins [10] Angelakis E, Bastelica D, Ben AA, El Filali A, Dutour A, Mege JL, et al. An eval- showing high intra-species variability [25]. uation of the effects of Lactobacillus ingluviei on body weight, the intestinal microbiome and metabolism in mice. Microb Pathog 2012;52:61 –8. As the global probiotic market generated US $15.9 billion in [11] Angelakis E, Raoult D. The increase of Lactobacillus species in the gut flora of 2008 ( http://www.marketsandmarkets.com) and the global probi- newborn broiler chicks and ducks is associated with weight gain. PLoS One otic market is expected to exceed US $28.8 billion by 2015, accord- 2010;5:e10463. ing to a report by Global Industry Analysts, Inc. ( http://www.prweb. [12] Weis J, Baranska B, Pal G, Hrncar C. Performance of the broiler duck males after application of two different probiotic preparations. Anim Sci Biotechnol- com), the risk of con flict of interest seems extremely important ogies 2010;43:300–2. [26]. Indeed, we recently published a work identifying signi ficant [13] Weis J, Hrncar C, Mindek S. Effect of probiotic preparates with different publication biases in the literature about probiotics and bene ficial strain on meat production of broiler ducks. Zootehnie si Biotehnologii 2008;41:717 –20 . results in human health and probiotics and growth-promoting [14] Weis J, Hrncar C, Pal G, Baranska B, Bujko J, Policka M, et al. Effect of probiotic results in farm industry [36]. This is why we caution researchers strain Lactobacillus fermentum CCM7158 supplement on performance and not to eliminate a priori the hypothesis that probiotics may be asso- carcass characteristics of broiler chickens. Acta fytotechnica et zootechnica 2010;4:96–8. ciated with weight-gain in humans. It is interesting to note that the [15] Capcarova M, Weiss J, Hrncar C, Kolesarova A, Pal G. Effect of Lactobacillus fer- first article published by Dr. Morelli available on PubMed [27] was mentum and Enterococcus faecium strains on internal milieu, antioxidant funded by Bracco Spa. which owns patents on Lactobacillus-con- status and body weight of broiler chickens. J Anim Physiol Anim Nutr (Berl) 2010;94:e215–24. taining probiotics as Lactobacillus paracasei CNCM 1-1390 and [16] Yu HF, Wang AN, Li XJ, Qiao SY. Effect of viable Lactobacillus fermentum on the that Dr. Morelli participated in a workshop of a food industry growth performance, nutrient digestibility and immunity of weaned pigs. J marketing several probiotics products [28]. Anim Feed Sci 2008;17:61–9. [17] Sato M, Uzu K, Yoshida T, Hamad EM, Kawakami H, Matsuyama H, et al. Effects The hypothesis that probiotics may be linked to human obesity of milk fermented by Lactobacillus gasseri SBT2055 on adipocyte size in rats. must be tested scienti fically with a maximum accuracy on the strain Br J Nutr 2008;99:1013 –7. identification and the maximum power in order not to neglect such [18] Hamad EM, Sato M, Uzu K, Yoshida T, Higashi S, Kawakami H, et al. Milk fer- fl an effect. For instance, after one of us (DR) pointed out that both pro- mented by Lactobacillus gasseri SBT2055 in uences adipocyte size via inhibition of dietary fat absorption in Zucker rats. Br J Nutr 2009;101:716 –24. biotics and antibiotics could be linked with a weight-gain in humans [19] Kadooka Y, Sato M, Imaizumi K, Ogawa A, Ikuyama K, Akai Y, et al. Regulation of [29], the effect of antibiotics used as growth-promoter in agriculture abdominal adiposity by probiotics ( Lactobacillus gasseri SBT2055) in adults with – for 60 years has been recognized very recently in humans when they obese tendencies in a randomized controlled trial. Eur J Clin Nutr 2010;64:636 43. fi [20] Kang JH, Yun SI, Park HO. Effects of Lactobacillus gasseri BNR17 on body are administered in rst months of life [30] and in an animal model weight and adipose tissue mass in diet-induced overweight rats. J Microbiol as promoting adiposity [31] . 2010;48:712 –4. The same paradigm has to be clari fied for probiotics and we [21] Karlsson CL, Molin G, Fak F, Johansson Hagslatt ML, Jakesevic M, Hakansson A, et al. Effects on weight gain and gut microbiota in rats given bacterial supple- think our work, even if comporting some limitations inevitably ments and a high-energy-dense diet from fetal life through to 6 months of linked with the difficult taxonomy of Lactobacillus genus, is a first age. Br J Nutr 2011;106:887 –95. 54 Letter to the Editor / Microbial Pathogenesis 55 (2013) 52 –54

[22] Lee K, Paek K, Lee HY, Park JH, Lee Y. Antiobesity effect of trans-10, cis-12- nutritional stress is moderated by probiotic administrations of Lactobacillus conjugated linoleic acid-producing Lactobacillus plantarum PL62 on diet- reuteri. Biosc Micro flor 1998;17:133 –9. induced obese mice. J Appl Microbiol 2007;103:1140 –6. [33] Casas IA, Dobrogosz WJ. Validation of the probiotic concept: Lactobacillus reu- [23] Takemura N, Okubo T, Sonoyama K. Lactobacillus plantarum strain no. 14 teri confers broad-spectrum protection against disease in humans and reduces adipocyte size in mice fed high-fat diet. Exp Biol Med (Maywood ) animals. Microb Ecol Health Dis 2000;12:247–85. 2010;235:849–56. [34] Nahashon SN, Nakaue HS, Snyder SP, Mirosh LW. Performance of single [24] Yin YN, Yu QF, Fu N, Liu XW, Lu FG. Effects of four Bi fidobacteria on obesity in comb white leghorn layers fed corn-soybean meal and barley-corn- high-fat diet induced rats. World J Gastroenterol 2010;16:3394 –401. soybean meal diets supplemented with a direct-fed microbial. Poult Sci [25] Lukjancenko O, Ussery DW, Wassenaar TM. Comparative genomics of Bi fidobac- 1994;73:1712 –23. terium, Lactobacillus and related probiotic genera. Microb Ecol 2012;63:651–73. [35] Ehrlich SD. Probiotics – little evidence for a link to obesity. Nat Rev Microbiol [26] Thomas O, Thabane L, Douketis J, Chu R, Westfall AO, Allison DB. Industry 2009 Dec;7:901. funding and the reporting quality of large long-term weight loss trials. Int J [36] Million M, Raoult D. Publication Biases in Probiotics. Eur J Epidemiol, in press . Obes (Lond) 2008;32:1531–6. [27] Morelli L, Zonenschain D, Callegari ML, Grossi E, Maisano F, Fusillo M. Assess- ment of a new synbiotic preparation in healthy volunteers: survival, persistence Matthieu Million, Didier Raoult* of probiotic strains and its effect on the indigenous flora. Nutr J 2003;2:11. Unité de Recherche sur les Maladies Infectieuses et Tropicales [28] Morelli L. The microbiological risk. In: Nestle Nutr workshop ser pediatr Emergentes, Faculté de Médecine, CNRS UMR 7278, IRD 198, program, vol. 60; 2007. p. 79 –90. [29] Raoult D. Human microbiome: take-home lesson on growth promoters? Aix-Marseille Université, Marseille, France Nature 2008;454:690–1. [30] Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant antibiotic * Corresponding author. Tel.: þ33 491 38 55 17; exposures and early-life body mass. Int J Obes (Lond) 2012 Aug 21. http:// dx.doi.org/10.1038/ijo.2012.132 [Epub ahead of print]. fax: þ33 491 83 03 90. [31] Cho I, Yamanishi S, Cox L, Methe BA, Zavadil J, Li K, et al. Antibiotics in early life E-mail address: [email protected] (D. Raoult) alter the murine colonic microbiome and adiposity. Nature 2012;488:621–6. [32] Dunham HJ, Casas IA, Edens FW, Parkhurst CR, Garlich JD, Dobrogosz WJ. Avian growth depression in chickens induced by environmental, microbiological, or Available online 26 October 2012

Article VIII :

Publication biases in probiotics

Matthieu Million, Didier Raoult

Published in Eur J Epidemiol. 2012 Nov;27(11):885-6. (IF 4.71)

L. casei BL23, and one Lactocin-705 (GenBank: LC70_LACPA) LETTER among available L. paracasei genomes. Taken together, these data suggest that the impact of a Occam ’s razor and probiotics activity Lactobacillus strain on the microbiota flora is mainly determined on Listeria monocytogenes by its direct antibiotic activities, including bacteriocins, as recently rediscovered (5). Matthieu Million 1, Emmanouil Angelakis 1, Fatima Drissi, and We read with interest the article by Archambaud et al. (1) on 2 Lactobacillus casei BL23 and Lactobacillus paracasei CNCM Didier Raoult Unité de Recherche sur les Maladies Infectieuses et Tropicales I-3689, which were able to limit the Listeria monocytogenes Emergentes, Unité Mixte de Recherche (UMR) Centre National de dissemination in a gnotobiotic humanized mouse model. The la Recherche Scienti fique (CNRS) 7278, Institut de Recherche pour authors suggested that changes in the expression of IFN-stimu- le Développement (IRD) 198, Institut National de la Santé et de la lated genes and of mi-RNA, together with the L. monocytogenes Recherche Médicale (INSERM) 1095, Faculté de Médecine, Aix metabolism redirection by Lactobacillus strains, may explain the Marseille Université, 13005 Marseille, France modulation of the infection. However, according to the Occam ’s razor principle postulating that a simpler explanation is more 1. Archambaud C, et al. (2012) Impact of lactobacilli on orally acquired listeriosis. Proc Natl Acad Sci USA 109(41):16684 –16689. likely to be true, we believe that the role of bacteriocins is critical 2. Jacobsen CN, et al. (1999) Screening of probiotic activities of forty-seven strains of in this instance. Lactobacillus spp. by in vitro techniques and evaluation of the colonization ability of five selected strains in humans. Appl Environ Microbiol 65(11):4949 –4956. Bacteriocins have been used as bio-preservatives, especially 3. Vignolo G, Fadda S, de Kairuz MN, de Ruiz Holgado AA, Oliver G (1996) Control of against L. monocytogenes contamination in vegetable food Listeria monocytogenes in ground beef by ‘Lactocin 705 ’, a bacteriocin produced by Lactobacillus casei CRL 705). Int J Food Microbiol 29(2-3):397 –402. matrices, for ∼20 y. The ability of Lactobacillus to inhibit 4. Bendali F, Gaillard-Martinie B, Hebraud M, Sadoun D (2008) Kinetic of production and pathogens in vitro is well documented. In one study, all of the mode of action of the Lactobacillus paracasei subsp. paracasei anti-listerial bacteriocin, an Algerian isolate. LWT –Food Science and Technology 41(10):1784 –1792. L. casei and L. paracasei strains inhibited L. monocytogenes 5. Kim HB, et al. (2012) Microbial shifts in the swine distal gut in response to the treatment growth (2). The Lactocin 705 produced by L. casei CRL705 is with antimicrobial growth promoter, tylosin. Proc Natl Acad Sci USA 109(38):15485 –15490. bacteriostatic on L. monocytogenes (3), whereas a strain of L. paracasei subsp. paracasei has been shown to produce another Author contributions: D.R. designed research; M.M., E.A., and F.D. performed research; substance inhibiting L. monocytogenes growth and leading to and M.M., E.A., and D.R. wrote the paper. cellular lysis (4). Using the Bagel and Bactibase bacteriocin The authors declare no con flict of interest. databases to analyze the strains used in this study, we were able 1M.M. and E.A. contributed equally to this work. to find one prebacteriocin (GenBank ID: YP_001988475) in 2To whom correspondence should be addressed. E-mail: [email protected].

www.pnas.org/cgi/doi/10.1073/pnas.1218418110 PNAS | January 2, 2013 | vol. 110 | no. 1 | E1

Article IX :

Lactobacillus rhamnosus bacteremia: an emerging

clinical entity

Frédérique Gouriet, Matthieu Million, Mireille Henri, Pierre-Edouard

Fournier, Didier Raoult

Published in Eur J Clin Microbiol Infect Dis. 2012 Apr 28 (IF 2.86)

Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480 DOI 10.1007/s10096-012-1599-5

ARTICLE

Lactobacillus rhamnosus bacteremia: an emerging clinical entity

F. Gouriet & M. Million & M. Henri & P.-E. Fournier & D. Raoult

Received: 20 November 2011 /Accepted: 29 February 2012 /Published online: 28 April 2012 # Springer-Verlag 2012

Abstract Lactobacillus spp. are ubiquitous commensals of have enhanced the recognition of new bacterial species the normal human flora that are only occasionally found in [1]. These methods have also improved the discrimi- clinical infections. Their role in human disease is established nation between bacterial species, allowing studies of for infectious endocarditis but is controversial for other the emerging clinical importance of microorganisms infections. We sought to characterize clinically associated that have been previously misidentified or unidenti- Lactobacillus spp. We conducted a retrospective study, fied. More recently, matrix-assisted laser desorption/ which consisted of the screening of Lactobacillus isolates ionization time-of-flight mass spectrometry (MALDI- obtained in our laboratory from January 2004 to December TOF MS) has been added as a rapid identification tool 2009. The polymerase chain reaction (PCR) assay was [2, 3]. selected as the gold standard method. The isolates were first Lactobacillus spp. have occasionally been associated identified using API Coryne strips, matrix-assisted laser with serious infections, especially among immunocom- desorption/ionization time-of-flight mass spectrometry promised patients and those suffering from endocarditis (MALDI-TOF MS), and 16S rRNA gene sequencing. [4]. The risk of infection due to lactobacilli and bifi- Lactobacillus tuf gene-based identification was used when dobacteria is extremely rare and is estimated to repre- the 16S rRNA results were inconclusive. Among the 60 sent 0.05 –0.4 % of cases of infective endocarditis and strains of Lactobacillus spp. obtained in our laboratory, L. bacteremia [ 5]. Meanwhile, historically, Lactobacillus rhamnosus was the most commonly isolated species and spp. found in food have been considered to be insig- was found in blood cultures from 16 patients. Combined nificant [ 6], with little clinical significance, and they with 45 patients reported in the literature, we found that are often regarded as contaminants when isolated from patients presenting with L. rhamnosus bacteremia experi- patient samples. Their role as commensals in the mam- enced nosocomial infections associated with both immuno- malian flora and their established safety in various suppression (66 %) and catheters (83 %). foods and probiotics support this conclusion. However, hazardous adverse effects of probiotics have been reported [ 7], such as acute pancreatitis, in which pro- Introduction biotic prophylaxis was associated with an increased risk of mortality [ 8]. Microbial identification methods using 16S rRNA gene In this work, we analyzed Lactobacillus spp. isolat- sequencing, which have increased by 456 % in 15 years, ed under pathological conditions over a five-year period. L. rhamnosus was the most commonly isolated F. Gouriet : M. Million : M. Henri : P.-E. Fournier : D. Raoult ( *) species. Our aim was to characterize the clinical entity Unité des Rickettsies, CNRS UMR 6236, IRD 198, Faculté de associated with this particular Lactobacillus species. Médecine, Université de la Méditerranée, We also reviewed the literature in order to identify 27 Bd. Jean Moulin, 13385 Marseille Cedex 05, France specific clinical conditions associated with L. rhamno- e-mail: [email protected] sus bacteremia. 2470 Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480

Materials and methods score ≥1.9 or 4/4 had a score ≥1.2. For each product, the colony-forming unit (CFU) and MALDI-TOF MS analyses Bacterial strain isolation were conducted three times independently. For all Lactobacillus spp. obtained from the clinical specimens, Our laboratory serves a university-affiliated tertiary care 16S rRNA gene-based identifications were performed as institution (3,820 beds) and routinely processes human sam- previously described [9]. The sequences determined were ples for the culture-based diagnosis of bacterial infectious compared with those available in the GenBank database diseases, including those due to anaerobes, aerobes, myco- using BlastN software ( http://www.ncbi.nlm.nih.gov/ bacteria, and spirochetes. The blood-culture vials used for BLAST/ ). Only one isolate per patient was further identified aerobic and anaerobic cultures (BACTEC; Becton by 16S rRNA gene sequence analysis. Dickinson, Sparks, MD) were incubated for five days. After the incubation period, Gram staining was performed, PCR and sequencing and the samples were cultured on 5 % sheep blood and chocolate agar at 37 °C under aerobic and anaerobic atmo- When the 16S rRNA gene sequence analysis was not conclu- spheric conditions for all positive blood cultures. Lung sive, we used Lactobacillus tuf gene-based identification. abscess biopsy specimens were cultured on chocolate agar Primer sequences Lac_tuf for: AYGGATGGTGC and 5 % sheep blood with or without nalidixic acid and KATCTTART and Lac_tuf rev: TCAGTGGTGTGG colistin at 37 °C under aerobic and anaerobic atmospheric AAGTAGAA were used. A negative control was introduced conditions for 10 days. Urine specimens were cultured on for each assay. The polymerase chain reaction (PCR) program 5 % sheep blood agar for 24 h, and pleural effusions were was: 15 min at 95 °C, followed by 40 cycles of 95 °C for 30 s, cultured on chocolate agar at 37 °C under the appropriate 45 s at 55 °C, 72 °C for 1 min using HotStar polymerase atmospheric conditions for 10 days. (Qiagen). After analysis on agarose gel electrophoresis, the PCR products were purified and sequenced by using the Patients BigDye Terminator 1.1 Cycle Sequencing kit (Applied Biosystems, Courtaboeuf, France) and the 3130 Genetic From January 2004 to December 2009, a retrospective study Analyzer (Applied Biosystems). The sequences were analyzed including the screening of Lactobacillus strains was con- by the SeqScape program (Applied Biosystems) and their ducted. We collected information for each patient on age, similarity with previously published sequences was determined sex, admitting hospital department, type of Lactobacillus using the online BLAST program at the NCBI and analyzed in infection, and source of Lactobacillus isolate. We focused a phylogenetic tree on patients with bacteremia due to L. rhamnosus , which was the most commonly identified species, and reviewed the Statistical analyses clinical data, including underlying disease, possible predisposing factors, and antibiotic treatment. For data comparison, we used EpiInfo version 6.0 software (Centers for Disease Control and Prevention [CDC], Atlanta, Lactobacillus identification GA). A p-value <0.05 was considered to be significant.

From January 2004 to September 2008, isolates were iden- Literature review tified using API Coryne strips (bioMérieux, Marcy l’Etoile, France). The identification was considered to be satisfactory Cases reports related to L. rhamnosus or Lactobacillus GG when the score was >80 %. From September 2008 to were identified through a MEDLINE ( http://www.ncbi.nlm. December 2009, MALDI-TOF MS was also used for the nih.gov/sites/entrez ) search for these terms, which was limited identification [2]. Measurements were performed using an to the English language. Additional cases were identified from Autoflex II mass spectrometer (Bruker Daltonics, Bremen, the references cited in the case reports. An attempt was made Germany) equipped with a 337-nm nitrogen laser. For each to obtain the original publication in each case. In situations spectrum, a maximum of 100 peaks was considered, and where the original publication was either unavailable via these peaks were compared with peaks in the database. The interlibrary loan or written in a non-English language, infor- 15 bacterial species exhibiting the most similar protein mation involving the case was based on information in the patterns compared to each isolate were ranked by their article(s) referenced by the case report. identification scores. For the MALDI-TOF MS analysis, The following data elements were extracted from each we adapted the score values proposed by Seng et al. [ 2, 3]. case: patient age, patient gender, patient comorbidity, type More specifically, an isolate was considered to be correctly of Lactobacillus infection, source of Lactobacillus isolate, identified by MALDI-TOF MS when ≥2/4 spectra had a species of Lactobacillus recovered, treatment regimen, and Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480 2471 overall mortality. The case was included in our study if it Table 1 Patient characteristics and species isolated overall and in contained three or more of these elements. Microbiological bacteremia methods used to identify the Lactobacillus organism in each Characteristic All cases (n060) Bacteremia (n028) case were not evaluated. Additionally, the distinction between community-acquired infections and nosocomial Age (years) infections was not examined. Median 55.87 57 Range 8–92 8 –89 Unknown 4 1 Results Sex Male 29 16 Identification methods Female 31 10 Species PCR methods were chosen as the gold standard. A total of Lactobacillus rhamnosus 25 16 60 Lactobacillus strains were isolated from 60 patients in Lactobacillus gasseri 12 1 our laboratory between January 2004 and December 2009. Lactobacillus casei 7 4 From 2004 to 2008, 48 Lactobacillus spp. were isolated and Lactobacillus delbrueckii 3 1 50 % (24) of the isolates were identified at the genus level Lactobacillus fermentum 3 1 using the phenotypic API Coryne strip system. In 2009, 12 Lactobacillus murinus 3 0 Lactobacillus spp. were isolated and 41 % (5) of the isolates Lactobacillus minutus 1 1 were identified at the species level using MALDI-TOF MS. Lactobacillus plantarum 1 1 From this five-year period, all 60 isolates were also identi- Lactobacillus zeae 2 1 fied prospectively using 16S rRNA sequencing: 75 % (44/ Lactobacillus reuteri 1 1 60) were identified at the species level. For the five isolates Lactobacillus spp. 2 1 identified by MALDI-TOF MS, the 16S rRNA results were consistent in only two of the cases. In 16 cases, the 16S rRNA sequence analyses could not discriminate between L. 0.0004). During this period, only one death associated with rhamnosus and L. casei . We retrospectively identified these Lactobacillus bacteremia was noted in our study. The most strains using Lactobacillus tuf gene-based identification, but commonly isolated species were as follows: L. rhamnosus (16 only 14 (13 L. rhamnosus and one L. casei ) were available. isolates); L. casei (four isolates); L. gasseri , L. delbrueckii , L. fermentum , L. minutus , L. zeae , L. murinus , L. plantarum , and Patient demographics L. reuteri (one isolate each). Most L. rhamnosus isolates, 16/25 (64 %), were obtained All patient characteristics are listed in Tables 1 and 2. The from the blood (Table 3). These patients were hospitalized sex ratio was equivalent (M/F029/31) and the patient age in the oncology department (seven patients), the critical care range was 8–92 years. Lactobacillus spp. were isolated from unit (five patients), and the emergency department (one various sites, but most of the strains (28 strains) were patient). The most common predisposing factor was immu- isolated from blood. The hospitalization departments of the nosuppressive therapy in patients with cancer ( n06), hepatic patients included the critical care unit (14 cases), oncology or lung transplantation ( n02), and skin allograft ( n01). (nine cases), and internal medicine (eight cases). Only one patient was not undergoing immunosuppressive Lactobacillus spp. were associated with other microorgan- therapy. In 91 % of cases, the patients had a central venous isms in only three cases (5 %) as follows: one ascites fluid catheter (11/12 patients; for four patients, the information sample, one pleural effusion sample, and one pacemaker was not available), and the bacteremia was nosocomial in device sample. those cases (Table 3). In nine cases, L. rhamnosus was isolated from various sites, including the lungs (two cases), Bacteremia pleural effusion (two cases), as well as sputum, mediastinal abscess, pharynx, intracardiac device, and bile (one case In 2004 and 2005, the blood cultures positive with each). None of the patients were undergoing probiotic Lactobacillus represent 0.06 % (3/3,146) and 0.11 % (4/ therapy. 3,267) in 2006, 0.14 % (4/3,589) and 0.15 % (5/3,195) in 2007, 0.08 % (3/3,864) in 2008, and 0.24 % in 2009 (9/ Literature review 3,632). The number of Lactobacillus spp. obtained from blood cultures in 2009 (9/3,632) increased significantly We found 45 reported cases of L. rhamnosus infection in the compared to the previous five years (15/17,061; p< literature [ 10 –45 ]. Most occurred in patients with the 2472 Table 2 Patient characteristics and Lactobacillus species isolated overall and by type of infection in the present series

Patient Sex Age (years) Sample Department Phenotypic identification 16S rRNA-based identification Similarity (%) tuf gene-based identification

1 M 38 Blood culture Oncology Lactobacillus spp. L. casei 99 2 M 64 Kidney Critical care unit Lactobacillus spp. L. fermentum . 99 3 F 64 Mediastinal abscess Thoracic surgery Lactobacillus spp . L. rhamnosus /casei 99 L. rhamnosus 4 M 36 Blood culture Internal medicine Lactobacillus spp. L. rhamnosus /casei 99 L. casei 5 M 67 Pleural effusion Critical care unit Lactobacillus spp. L. rhamnosus /casei 99 L. rhamnosus 6 F 92 Lung abscess Internal medicine Lactobacillus spp. L. gasseri 99 7 M 83 Ascites Critical care unit Lactobacillus spp., L. rhamnosus /casei 99 NA Escherichia coli 8 M 53 Bile Oncology Lactobacillus spp L. murinus 99 9 M 88 Blood culture Urology Lactobacillus spp. L. casei 99 10 F 33 Peritoneal abscess Gastroenterology Lactobacillus spp . L. gasseri 99 11 M 35 Blood culture Emergency Lactobacillus spp. L. rhamnosus /casei 99 L. rhamnosus 12 M 61 Blood culture 2/2 Oncology Lactobacillus spp. L. casei 99 13 M 57 Blood culture Oncology Lactobacillus spp. L. rhamnosus /casei 99 L. rhamnosus 14 M 60 Pleural effusion Thoracic surgery Lactobacillus spp., Lactobacillus spp. 99 L. rhamnosus Enterococcus faecalis , Candida glabrata 15 M Eye Ophthalmology BGP vancomycin R L. gasseri 99 16 F 64 Cervical abscess ORL L. gasseri L. gasseri 17 M 84 Blood culture Emergency L. fermentum L. fermentum 99 18 M 57 Bone abscess Orthopedic BGP vancomycin R L. casei 99 u lnMcoilIfc i 21)31:2469 (2012) Dis Infect Microbiol Clin J Eur 19 M 43 Blood culture 1/3 Infectious disease BGP vancomycin R L. reuteri 98 20 M 31 Renal lithiasis Urology BGP vancomycin R L. gasseri 99 21 M 44 Lung abscess Critical care unit BGP vancomycin R Lactobacillus spp. 99 L. rhamnosus 22 F 16 Blood culture 1/1 Critical care unit Lactobacillus spp. Lactobacillus spp. 99 L. rhamnosus 23 M 45 Defibrillator Cardiology Lactobacillus spp., L. casei /L. rhamnosus 99 L. rhamnosus Propionibacterium acnes , Staphylococcus epidermidis 24 M 8 Blood culture 2/2 Oncology BGP vancomycin R Lactobacillus spp. 99 L. rhamnosus 25 F 89 Blood culture 1/3 Nephrology BGP vancomycin R L. casei /L. rhamnosus 99 NA 26 F 65 Bone Orthopedic surgery BGP vancomycin R L. murinus 99 27 F Urine Infectious disease BGP vancomycin R L. delbrueckii 99 28 F Urine Internal medicine BGP vancomycin R L. gasseri 99 29 F 78 Urine Emergency BGP vancomycin R L. delbrueckii

30 M Pharynx Oncology BGP vancomycin R L. casei 99 – 2480 31 F 63 Left kidney Urology BGP vancomycin R L. gasseri 99 u lnMcoilIfc i 21)31:2469 (2012) Dis Infect Microbiol Clin J Eur Table 2 (continued)

Patient Sex Age (years) Sample Department Phenotypic identification 16S rRNA-based identification Similarity (%) tuf gene-based identification

32 M 62 Blood culture 1/3 Nephrology BGP vancomycin R L. zeae 99 33 F 83 Urine Emergency Lactobacillus spp. L. gasseri 99 34 F 43 Blood culture Critical care unit Lactobacillus spp. L. rhamnosus /casei 100 L. rhamnosus 35 F 84 Blood culture 2 Critical care unit Lactobacillus spp. L. rhamnosus /casei 99 L. rhamnosus 36 M NA Blood culture Gastroenterology Lactobacillus spp. L. rhamnosus 99 37 M 65 Arm abscess Emergency Lactobacillus spp. L. gasseri 99 38 F 76 Urine Emergency Lactobacillus spp. L. gasseri 100 39 F 90 Urine Emergency Lactobacillus spp. L. gasseri 99

40 F 56 ORL ORL Lactobacillus spp. L. rhamnosus 100 – 2480 41 M 66 Blood culture Critical care unit Lactobacillus spp. L. casei /rhamnosus 99 L. rhamnosus 42 F 56 Bone biopsy Orthopedic surgery Lactobacillus spp. L. murinus 99 43 F 60 Blood culture Respiratory care unit Lactobacillus spp. Lactobacillus spp. ( rhamnosus 99 L. rhamnosus casei zeae ) 44 F 47 Sputum Critical care unit Lactobacillus spp. L. fermentum 99 45 F 8 Sputum Oncology Lactobacillus spp. L. rhamnosus 99 46 F 11 Kidney abscess Critical care unit Lactobacillus spp. L. casei 99 47 M 14 Abdominal abscess Gastroenterology L. casei L. zeae 99.9 48 F 49 Blood culture Critical care unit L. casei L. rhamnosus 100 49 F 49 Blood culture Critical care unit L. casei L. rhamnosus 99 50 M 83 Blood culture Oncology No identification L. rhamnosus 100 51 F 71 Blood culture 3/3 Urology No identification L. delbrueckii 99.7 52 M 58 Blood culture 2/6 Respiratory care unit No identification L. rhamnosus 99.9 53 F 22 Blood culture Critical care unit No identification L. rhamnosus 99.9 54 F 54 Blood culture 1/1 Oncology No identification L. rhamnosus 99.6 55 F 82 Blood culture 2/2 Internal medicine No identification L. minutus 99.2 56 M 54 Blood culture 1/2 Gastroenterology No identification L. gasseri 99.9 57 M 83 Bile Gastroenterology No identification L. rhamnosus 100 58 M 50 Blood culture 1/2 Oncology No identification L. rhamnosus 99.9 59 F 70 Lung Respiratory care unit No identification L. rhamnosus 99.9 60 M 69 Blood culture 1/1 Internal medicine No identification L. plantarum 99.9

NA: not available; R resistant 2473 2474

Table 3 Data of patients with Lactobacillus rhamnosus bacteremia in the present series

Patient Sex Age Sample Department Nosocomial Immunosuppression Predisposing factors Intravenous Antibiotics Death (years) catheter

1 M 35 Blood culture Emergency No Na Na Na Na No 2 M 57 Blood culture Oncology No Radiotherapy and Lingual epidermoid Central Amoxicillin acid, clavulanic No chemotherapy carcinoma acid, ciprofloxacin, fluconazole, and vancomycin 3 F 16 Blood culture 1/1 Critical care unit Yes No Fracture and luxation of Central Na No dorsal rachis 4 M 8 Blood culture 2/2 Oncology No Chemotherapy Medulloblastoma Central Vancomycin and ceftazidime No 5 F 43 Blood culture Critical care unit Yes Na Na Na Na No 6 F 84 Blood culture 2/2 Critical care unit Yes No Skin allograft, burn Central Na No 7 M Blood culture Gastroenterology Yes Yes Gastric carcinoma Central Na No 8 M Blood culture Critical care unit Yes Yes Pharynx carcinoma Central Na No 9 F 60 Blood culture Respiratory care unit Yes No Dilatation of bronchi, No Na No pulmonary fibrosis 10 F 49 Blood culture Critical care unit Yes Na Na Na Na No 11 F 49 Blood culture Critical care unit Yes Immunosuppressor Hepatic transplantation Na NA No amylose 12 M 83 Blood culture Oncology No No Gastric adenocarcinoma Central Ceftriaxone and ciprofloxacin No 31:2469 (2012) Dis Infect Microbiol Clin J Eur 13 M 58 Blood culture 2/6 Respiratory care unit No Chemotherapy Pleural mesothelioma Central Na No 14 F 22 Blood culture Critical care unit Yes Immunosuppressor Pulmonary transplantation, Central Na No cystic fibrosis 15 F 54 Blood culture 1/1 Oncology No Chemotherapy Jugular epidermoid Central Na No carcinoma 16 M 50 Blood culture 1/2 Oncology Yes Radiotherapy, Laryngeal epidermoid Central Amoxicillin acid, clavulanic Yes chemotherapy carcinoma acid, metronidazole, and piperacillin + tazobactam

NA: not available – 2480 Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480 2475 following clinical characteristics: immunosuppressive ther- were L. rhamnosus (41 %), L. gasseri (19 %), and L. casei apy ( n02), dialysis ( n02), transplantation ( n04), cancer (11 %). Cannon et al. [ 4] reported 200 cases of (n08), diabetes mellitus (n03), intravenous catheters (n0 Lactobacillus spp.-associated infections from 1950 to 12), prior surgical interventions (n02), and prior digestive 2003. Similar to our study and other works, L. casei disorders (n08) (Table 4). Seventeen patients without any (35.7 %) and L. rhamnosus (22.9 %) were the most com- predisposing factors other than valvulopathy had L. rham- monly isolated species from blood cultures. Additionally, nosus infectious endocarditis. Six patients were undergoing the antimicrobial pressure link to the increased use of van- probiotic therapy. Antibiotics treatment was available for 21 comycin to which lactobacilli are resistant may also have patients. The antibiotic treatment was appropriate in 17 contributed to the increasing isolation of colonizing lacto- cases including aminopenicillin or carbapenem. bacilli. Several reports of L. rhamnosus -related infectious endocarditis [52 , 53 ] and bacteremia have been published [5, 54 –56 ]. In one review [ 4], 73 cases of infectious endo- Discussion carditis were reported. L. rhamnosus and L. casei were the most commonly isolated species in these cases. We received Lactobacillus spp. are commensals found in the gastrointes- 267 endocarditis cases during this five-year period; despite tinal tract, oral cavity, and female urogenital tract [ 46 ]. L. the fact that we are a referral center for the diagnosis of rhamnosus specifically belongs to the normal flora of infectious endocarditis, bacteremia was more common than healthy human rectal, oral [ 46 ], and vaginal mucosa [ 47 ]. infectious endocarditis (16 vs. 1) in the current study. We Because most (47 %) Lactobacillus spp. were isolated from found 45 cases of L. rhamnosus infections reported in the blood cultures, our findings confirm prior observations that literature (Table 4). Upon combining published data with Lactobacillus spp. can exhibit virulence. These species are those from our study, we identified 47 patients with L. used worldwide as starter cultures for the production of rhamnosus bacteremia without infectious endocarditis. fermented dairy products, and they are among the most Thirty-one (66 %) presented with cancer leukemia or had commonly used microorganisms in probiotics for human received a transplant, and 33/40 (82 %) may have had a consumption. For example, the L. rhamnosus strain, GG, possible catheter-related infection. Indeed, the 18 patients has been used in the treatment of infantile diarrhea [ 48 ], with L. rhamnosus infectious endocarditis had significantly antibiotic-associated diarrhea [ 49 ], and candidal vaginitis fewer catheters (2/17) and did not undergo immunosuppres- [50 ]. Lactobacillus spp. are occasionally found in human sion (0/16; p<0.004). In contrast, underlying cardiac lesions clinical infections and are often considered as contaminants (12/17, 68 %) were more common in infectious endocarditis or opportunistic pathogens. Because of their low signifi- compared to isolated bacteremia (1/50, 2 %; p<0.0004). In cance and special growth requirements, they are often over- patients with L. rhamnosus -associated bacteremia, the looked or incorrectly identified. Furthermore, identification reported mortality rate ranges from 12 to 48 %, depending at the species level is often difficult. Over the past decade, on the treatment [ 54 , 55 ]. Documented cases of systemic the taxonomy has changed, and accurate Lactobacillus spp. infections associated with the consumption of lactic acid identification has, indeed, improved with molecular analysis bacteria are extremely rare, with only nine published cases. [51 ]. However, in most microbiology laboratories, commer- However, adverse events are not systematically assessed or cial systems routinely employed for Gram-positive rod iden- reported in clinical trials [ 57 ]. Probiotic supplementation tification are inefficient, and only 30 –50 % of isolates are must be used with caution in individuals who may be at correctly identified [ 5]. In the current study, we identified risk for the dissemination of these live microbes to sterile only 41 % of the isolates using this routine method. In sites [58 ]. contrast, all of the isolates in the present study were identi- In conclusion, the frequency of L. rhamnosus isolation fied at the species level using molecular techniques: 76 % of from blood cultures has increased. Because few cases of L. the isolates were identified using 16S rRNA gene sequenc- rhamnosus bacteremia are reported in the literature com- ing and 100 % by tuf gene-based identification. pared to infectious endocarditis, we suspect that bacteremia At our institution, Lactobacillus spp. have been uncom- cases are underestimated compared to L. rhamnosus -associ- mon clinical isolates, identified in only 0.05 % of positive ated infectious endocarditis. We believe that, when >50 blood cultures; over a five-year period, the identification of cases of infection are reportedly caused by a specific spe- the most common species, L. rhamnosus , increased from 11/ cies, it is useful to describe its clinical entity. The increase in 17,068 cases to 5/3,632 cases, which represents an increase reported Lactobacillus spp.-associated infections raises to 0.13 % of the positive blood cultures. The increase of questions. It is important for clinicians to be aware of the Lactobacillus spp. isolates over this five-year period may be potential risks of microbial therapy. Our analysis found that due to the recent incorporation of a more sensitive molecular L. rhamnosus bacteremia occurs in 66 % of immunosup- technique. The most common species recovered in our study pressed patients and 82.5 % of catheterized patients. 2476 Table 4 Review of reported clinical cases involving Lactobacillus rhamnosus infection and data from patients with L. rhamnosus infection in the present series

Case Age Sex Sample Predisposing factor Neoplasia Intravenous Diagnosis Treatment Outcome Reference report (years*) catheter

1 79 M Pleural effusion COPD, diabetes mellitus No No Lung abscess pleuritis Na Cured [ 8] 2 74 M Blood culture, pus, Tonsillar carcinoma, Mirizzi Yes No Liver abscess Na Cured [ 9] and gallbladder syndrome, diabetes mellitus 3 74 F Pus Drinks containing L. No No Liver abscesses Na Cured [ 10 ] rhamnosus GG 4 5 M Pericardial effusion Bone marrow transplant Yes Yes Pericardial infection Na Cured [ 11 ] and blood culture for aplastic anemia 5 57 M Peritoneal dialysate Peritoneal dialysis No Yes Peritonitis Na Cured [ 12 ] culture 6 35 F Pus Na No No Angiocholitis and pancreatitis Ciprofloxacin + imipenem, Cured [ 15 ] amoxicillin + rifampicin, surgery 7 70 F Angiocholitis pus, Hypothyroid No No Pancreatitis with necrosis and Amoxicillin + clavulanic Cured [ 15 ] blood culture bacteremia acid, ofloxacin, amoxicillin + rifampicin, surgery 8 59 M Peritoneal fluid Continuous ambulatory No Yes Peritonitis Vancomycin, gentamicin, Cured [ 16 ] peritoneal dialysis erythromycin 9 32 M Lymph nodes Erysipeloid No Adenitis Na Died [ 17 ] 10 61 F Sputum Myeloid leukemia Yes Na Chest infection Na Na [ 18 ] 11 73 M Sputum Emphysema No No Pneumonia/lung abscess Na Na [ 19 ] 12 56 M Pleural fluid Double lung transplantation No Yes Pleuritis Ampicillin sulbactam Cured [ 44 ] HIV 13 73 F Purulent material Chronic cholecystis Na Na Empyema gallbladder Na Na [ 42 ] 14 Na Na Blood culture Gore-Tex patch in the inferior Na Bacteremia Na Na [ 20 ] vena cava 15 Na Na Blood culture, Acute leukemia Yes Yes Bacteremia, pneumonia Imipenem + erythromycin [ 21 ]

bronchoalveolar 31:2469 (2012) Dis Infect Microbiol Clin J Eur lavage 16 Na Na Blood culture Acute leukemia, concomitant Yes Yes Bacteremia Imipenem + erythromycin Na [ 21 ] popular skin rash 17 Na Na Blood culture Acute leukemia, concomitant Yes Yes Bacteremia Imipenem + erythromycin Na [ 21 ] popular skin rash 18 Na Na Blood culture Lung transplantation No Yes Catheter-related bacteremia Na Na [ 22 ] 19 43 M Blood culture Ulcerative colitis No Na Bacteremia Na Na [ 23 ] 20 73 F Blood culture Diabetes mellitus No Na Liver abscess and bacteremia Ampicillin + gentamicin Cured [ 24 ] 21 42 F Blood culture, Sjogren ’s syndrome, No Yes Na Tazocin + gentamicin + Died [ 25 ] central line immunosuppressive therapy, vancomycin, rifampicin diarrhea with vancomycin 22 6 F Blood culture Cerebral palsy, microcephaly, No Yes Bacteremia Vancomycin + ceftazidime, Cured [ 26 ] mental retardation, amoxicillin gastrojejunostomy feeding, probiotic therapy for diarrhea 23 Na Na Blood culture Depressed immune status Na Na Bacteremia Na Na [ 27 ] –

24 11 months Na Blood culture Short gut syndrome, probiotic No Yes Catheter-related bacteremia Ampicillin + gentamicin Na [ 28 ] 2480 therapy u lnMcoilIfc i 21)31:2469 (2012) Dis Infect Microbiol Clin J Eur Table 4 (continued)

Case Age Sex Sample Predisposing factor Neoplasia Intravenous Diagnosis Treatment Outcome Reference report (years*) catheter

25 36 weeks ’ M Blood culture Parenteral nutrition short residual Na Na Bacteremia Ceftriaxone ampicillin Na [ 29 ] gestation intestine, probiotic therapy 26 34 week ’ M Blood culture Gastroschisis, gastrostomy, Na Na Bacteremia Ceftriaxone ampicillin Na [ 29 ] gestation and jejunostomy for bowel infarction 27 Na Na Blood culture, Allogeneic hematopoietic stem Yes Yes Meningitis Na Na [ 30 ] cerebrospinal cell transplantation for acute fluid leukemia 28 14 Na Blood cultures Acute myeloid leukemia Yes Bacteremia Na Cured [ 31 ] 29 35 M Blood culture Na No Na Bacteremia Na Cured Present report 30 57 M Blood culture Lingual epidermoid carcinoma, Yes Yes Bacteremia Amoxicillin, clavulanic acid, Cured Present report radiotherapy and chemotherapy ciprofloxacin, fluconazole, –

and vancomycin 2480 31 16 F Blood culture 1/1 Fracture and luxation of dorsal No Yes Bacteremia Na Cured Present report rachis 32 8 M Blood culture 2/2 Medulloblastoma, chemotherapy Yes Yes Bacteremia Vancomycin and ceftazidime Cured Present report 33 43 F Blood culture Na Na Na Bacteremia Na Cured Present report 34 84 F Blood culture 2 Skin allograft, burn Yes Yes Bacteremia Na Cured Present report 35 49 F Blood culture Na Na Na Bacteremia Na Cured Present report 36 49 F Blood culture Hepatic transplantation, No Na Bacteremia NA Cured Present report amylase, immunosuppressor 37 83 M Blood culture Gastric adenocarcinoma Yes Yes Bacteremia Ceftriaxone and ciprofloxacin Cured Present report 38 58 M Blood culture Pleural mesothelioma, Yes Yes Bacteremia Na Cured Present report chemotherapy 39 22 F Blood culture Pulmonary transplantation, No Yes Bacteremia Na Cured Present report cystic fibrosis, immunosuppressor 40 54 F Blood culture Jugular epidermoid carcinoma, Yes Yes Bacteremia Na Cured Present report chemotherapy 41 50 M Blood culture Laryngeal epidermoid Yes Yes Bacteremia Amoxicillin, clavulanic acid, Cured Present report carcinoma, radiotherapy, metronidazole, and chemotherapy piperacillin + tazobactam 42 66 M Blood culture Gastric neoplasia Yes Yes Bacteremia Na Cured Present report 43 Na M Blood culture Pharynx carcinoma Yes Yes Bacteremia Na Cured Present report 44 60 F Blood culture Dilatation of bronchi No Na Bacteremia Na Cured Present report 45 66 M Blood culture Mitral No Na Endocarditis Na Na [ 32 ] 46 29 M Blood culture Prolapse of mitral valve No No Endocarditis Penicillin + gentamicin, surgery Cured [ 33 ] 47 73 M Blood culture Prosthetic aortic valve No No Endocarditis Vancomycin + gentamicin, Cured [ 34 ] amoxicillin + rifampicin, surgery 48 65 M Blood culture Dairy product consumption, No Na Endocarditis Penicillin + gentamicin and Cured [ 35 ] colonoscopy ceftriaxone + clindamycin + ciprofloxacin 49 6 weeks old M Blood culture Double outlet right ventricle No Yes Endocarditis Penicillin G + gentamicin Cured [ 26 ] 2477 and pulmonary stenosis 2478 Table 4 (continued)

Case Age Sex Sample Predisposing factor Neoplasia Intravenous Diagnosis Treatment Outcome Reference report (years*) catheter

surgery, probiotic therapy for diarrhea 50 Na Na Blood culture Na Na Na Endocarditis Na Na [ 36 ] 51 31 M Blood culture Rheumatic valvular disease No Na Endocarditis Na Cured [ 17 ] 52 17 F Blood culture Marfan syndrome No Na Endocarditis Na Cured [ 17 ] 53 36 F Blood culture Coarctation of the aorta + No Na Endocarditis Na Cured [ 17 ] aneurysm aorta ascending 54 6 F Blood culture Tooth extraction No No Endocarditis Na Na [ 37 ] 55 7 F Blood culture Tricuspid atresia carious teeth No No Endocarditis Na Na [ 38 ] 56 68 M Pus covering Severe atrial disease No Na Endocarditis Na Na [ 39 ] aortic graft 57 71 F Blood culture Heart disease aortic valve No Na Endocarditis Na Na [ 39 ] replaced 58 45 M Blood culture Dental surgery No Na Endocarditis Na Na [ 40 ] 59 Na M Blood culture Probiotic yogurt consumption No Na Endocarditis Na Na [ 41 ] and septic arthritis 60 67 M Blood culture Probiotic tabs consumption, No No Endocarditis Ampicillin + gentamicin Cured [ 43 ] mitral regurgitation 61 24 F Blood culture Aortic mechanical valve No Yes Endocarditis Na Cured [ 45 ] previous1 endocarditis, probiotic tabs with L. rhamnosus

*Unless otherwise stated Na: not available u lnMcoilIfc i 21)31:2469 (2012) Dis Infect Microbiol Clin J Eur – 2480 Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480 2479

References 22. Carretto E, Barbarini D, Marzani FC et al (2001) Catheter-related bacteremia due to Lactobacillus rhamnosus in a single-lung transplant recipient. Scand J Infect Dis 33:780 –782 1. Janda JM, Abbott SL (2007) 16S rRNA gene sequencing for 23. Farina C, Arosio M, Mangia M et al (2001) Lactobacillus casei bacterial identification in the diagnostic laboratory: pluses, perils, subsp. rhamnosus sepsis in a patient with ulcerative colitis. J Clin and pitfalls. J Clin Microbiol 45:2761 –2764 Gastroenterol 33:251–252 2. Seng P, Drancourt M, Gouriet F et al (2009) Ongoing revolution in 24. Notario R, Leardini N, Borda N et al (2003) Hepatic abscess and bacteriology: routine identification of bacteria by matrix-assisted bacteremia due to Lactobacillus rhamnosus. Rev Argent Microbiol laser desorption ionization time-of-flight mass spectrometry. Clin 35:100–101 Infect Dis 49:543–551 25. MacGregor G, Smith AJ, Thakker B et al (2002) Yoghurt biother- 3. Seng P, Rolain JM, Fournier PE et al (2010) MALDI-TOF-mass apy: contraindicated in immunosuppressed patients? Postgrad Med spectrometry applications in clinical microbiology. Future Micro- J 78:366 –367 biol 5:1733–1754 26. Land MH, Rouster-Stevens K, Woods CR et al (2005) Lactobacillus 4. Cannon JP, Lee TA, Bolanos JT et al (2005) Pathogenic relevance sepsis associated with probiotic therapy. Pediatrics 115:178 –181 of Lactobacillus: a retrospective review of over 200 cases. Eur J 27. Arpi M, Vancanneyt M, Swings J et al (2003) Six cases of Lacto- Clin Microbiol Infect Dis 24:31–40 bacillus bacteraemia: identification of organisms and antibiotic 5. Salminen MK, Tynkkynen S, Rautelin H et al (2002) Lactobacillus susceptibility and therapy. Scand J Infect Dis 35:404 –408 bacteremia during a rapid increase in probiotic use of Lactobacil- 28. De Groote MA, Frank DN, Dowell E et al (2005) Lactobacillus lus rhamnosus GG in Finland. Clin Infect Dis 35:1155 –1160 rhamnosus GG bacteremia associated with probiotic use in a child 6. Adams MR, Marteau P (1995) On the safety of lactic acid bacteria with short gut syndrome. Pediatr Infect Dis J 24:278 –280 from food. Int J Food Microbiol 27:263 –264 29. Kunz AN, Noel JM, Fairchok MP (2004) Two cases of Lactoba- 7. Bernardeau M, Vernoux JP, Henri-Dubernet S et al (2008) Safety cillus bacteremia during probiotic treatment of short gut syndrome. assessment of dairy microorganisms: the Lactobacillus genus. Int J J Pediatr Gastroenterol Nutr 38:457 –458 Food Microbiol 126:278–285 30. Robin F, Paillard C, Marchandin H et al (2010) Lactobacillus 8. Besselink MG, van Santvoort HC, Buskens E et al (2008) Pro- rhamnosus meningitis following recurrent episodes of bacteremia biotic prophylaxis in predicted severe acute pancreatitis: a rando- in a child undergoing allogeneic hematopoietic stem cell transplan- mised, double-blind, placebo-controlled trial. Lancet 371:651–659 tation. J Clin Microbiol 48:4317 –4319 9. Drancourt M, Bollet C, Carlioz A et al (2000) 16S ribosomal DNA 31. Majcher-Peszynska J, Heine W, Richter I et al (1999) Persistent sequence analysis of a large collection of environmental and clin- Lactobacillus casei subspecies rhamnosus bacteremia in a 14 year ical unidentifiable bacterial isolates. J Clin Microbiol 38:3623 – old girl with acute myeloid leukemia. A case report. Klin Padiatr 3630 211:53 –56 10. Shoji H, Yoshida K, Niki Y (2010) Lung abscess and pleuritis 32. Naudé WD, Swanepoel A, Böhmer RH et al (1988) Endocarditis caused by Lactobacillus rhamnosus in an immunocompetent pa- caused by Lactobacillus casei subspecies rhamnosus. A case re- tient. J Infect Chemother 16:45 –48 port. S Afr Med J 73:612 –614 11. Chan JF, Lau SK, Woo PC et al (2010) Lactobacillus rhamnosus 33. Monterisi A, Dain AA, Suárez de Basnec MC et al (1996) Native- hepatic abscess associated with Mirizzi syndrome: a case report valve endocarditis produced by Lactobacillus casei sub. rhamno- and review of the literature. Diagn Microbiol Infect Dis 66:94 –97 sus refractory to antimicrobial therapy. Medicina (B Aires) 12. Rautio M, Jousimies-Somer H, Kauma H et al (1999) Liver ab- 56:284–286 scess due to a Lactobacillus rhamnosus strain indistinguishable 34. Wallet F, Dessein R, Armand S et al (2002) Molecular diagnosis of from L. rhamnosus strain GG. Clin Infect Dis 28:1159–1160 endocarditis due to Lactobacillus casei subsp. rhamnosus. Clin 13. Kalima P, Masterton RG, Roddie PH et al (1996) Lactobacillus Infect Dis 35:e117 –e119 rhamnosus infection in a child following bone marrow transplant. J 35. Avlami A, Kordossis T, Vrizidis N et al (2001) Lactobacillus Infect 32:165–167 rhamnosus endocarditis complicating colonoscopy. J Infect 14. Klein G, Zill E, Schindler R et al (1998) Peritonitis associated with 42:283–285 vancomycin-resistant Lactobacillus rhamnosus in a continuous 36. Golledge C (1988) Vancomycin resistant lactobacilli. J Hosp Infect ambulatory peritoneal dialysis patient: organism identification, 11:292 antibiotic therapy, and case report. J Clin Microbiol 36:1781 –1783 37. Fritsche D, Mennicken U, Vielhaber K (1973) Endocarditis caused 15. Brahimi M, Mathern P, Fascia P et al (2008) Two cases of Lacto- by diphtheroids and lactobacilli (author ’s transl). Dtsch Med bacillus rhamnosus infection and pancreatitis. Med Mal Infect Wochenschr 98:2239 –2242 38:29–31 38. Tornos MP, Perez-Soler R, Fernandez-Perez R (1980) Lactobacil- 16. Sanyal D, Bhandari S (1992) CAPD peritonitis caused by Lacto- lus casei endocarditis in tricuspid atresia. Chest 77:713 bacillus rhamnosus. J Hosp Infect 22:325 –327 39. Holliman RE, Bone GP (1988) Vancomycin resistance of clinical 17. Sharpe ME, Hill LR, Lapage SP (1973) Pathogenic lactobacilli. J isolates of lactobacilli. J Infect 16:279 –283 Med Microbiol 6:281–286 40. Griffiths JK, Daly JS, Dodge RA (1992) Two cases of endocarditis 18. Rahman M (1982) Chest infection caused by Lactobacillus casei due to Lactobacillus species: antimicrobial susceptibility, review, ss rhamnosus . Br Med J (Clin Res Ed) 284:471 –472 and discussion of therapy. Clin Infect Dis 15:250 –255 19. Namnyak SS, Blair AL, Hughes DF et al (1992) Fatal lung abscess 41. Presterl E, Kneifel W, Mayer HK et al (2001) Endocarditis by due to Lactobacillus casei ss rhamnosus. Thorax 47:666 –667 Lactobacillus rhamnosus due to yogurt ingestion? Scand J Infect 20. Jureen R, Søndenaa K, Høiby EA et al (2002) Lactobacillus Dis 33:710–714 rhamnosus septicaemia in a patient with a graft in the inferior vena 42. Allison D, Galloway A (1988) Empyema of the gall-bladder due to cava. Scand J Infect Dis 34:135 –136 Lactobacillus casei. J Infect 17:191 21. Chomarat M, Espinouse D (1991) Lactobacillus rhamnosus septi- 43. Mackay AD, Taylor MB, Kibbler CC et al (1999) Lactobacillus cemia in patients with prolonged aplasia receiving ceftazidime – endocarditis caused by a probiotic organism. Clin Microbiol Infect vancomycin. Eur J Clin Microbiol Infect Dis 10:44 5:290–292 2480 Eur J Clin Microbiol Infect Dis (2012) 31:2469 –2480

44. Luong ML, Sareyyupoglu B, Nguyen MH et al (2010) Lactoba- Lactobacillus acidophilus , Lactobacillus casei group, Lactobacillus cillus probiotic use in cardiothoracic transplant recipients: a link to delbrueckii , and Bifidobacterium longum in commercial dairy prod- invasive Lactobacillus infection? Transpl Infect Dis 12:561 –564 ucts. J Food Prot 72:93 –100 45. Kochan P, Chmielarczyk A, Szymaniak L et al (2011) Lactobacil- 52. Salvana EM, Frank M (2006) Lactobacillus endocarditis: case lus rhamnosus administration causes sepsis in a cardiosurgical report and review of cases reported since 1992. J Infect 53:e5 –e10 patient—is the time right to revise probiotic safety guidelines? 53. Yagi S, Akaike M, Fujimura M et al (2008) Infective endocarditis Clin Microbiol Infect 17:1589–1592 caused by Lactobacillus. Intern Med 47:1113 –1116 46. Ahrné S, Nobaek S, Jeppsson B et al (1998) The normal Lactoba- 54. Salminen MK, Rautelin H, Tynkkynen S et al (2004) Lactobacillus cillus flora of healthy human rectal and oral mucosa. J Appl bacteremia, clinical significance, and patient outcome, with special Microbiol 85:88–94 focus on probiotic L. rhamnosus GG. Clin Infect Dis 38:62–69 47. Kiss H, Kögler B, Petricevic L et al (2007) Vaginal Lactobacillus 55. Salminen MK, Rautelin H, Tynkkynen S et al (2006) Lactobacillus microbiota of healthy women in the late first trimester of pregnan- bacteremia, species identification, and antimicrobial susceptibility cy. BJOG 114:1402 –1407 of 85 blood isolates. Clin Infect Dis 42:e35 –e44 48. Szajewska H, Kotowska M, Mrukowicz JZ et al (2001) Efficacy of 56. Ouwehand AC, Saxelin M, Salminen S (2004) Phenotypic differ- Lactobacillus GG in prevention of nosocomial diarrhea in infants. ences between commercial Lactobacillus rhamnosus GG and L. J Pediatr 138:361 –365 rhamnosus strains recovered from blood. Clin Infect Dis 39:1858– 49. Biller JA, Katz AJ, Flores AF et al (1995) Treatment of recurrent v 1860 colitis with Lactobacillus GG. J Pediatr Gastroenterol Nutr 57. Hempel S, Newberry S, Ruelaz A, Wang Z, Miles JNV, Suttorp 21:224–226 MJ, Johnsen B, Shanman R, Slusser W, Fu N, Smith A, Roth B, 50. Falagas ME, Betsi GI, Athanasiou S (2007) Probiotics for the Polak J, Motala A, Perry T, Shekelle PG (2011) Safety of pro- treatment of women with bacterial vaginosis. Clin Microbiol Infect biotics to reduce risk and prevent or treat disease. Agency for 13:657–664 Healthcare Research and Quality (US), Rockville 51. Sheu SJ, Hwang WZ, Chen HC et al (2009) Development and use of 58. Sanders ME, Akkermans LM, Haller D et al (2010) Safety assess- tuf gene-based primers for the multiplex PCR detection of ment of probiotics for human use. Gut Microbes 1:164 –185

Article X :

Occam's razor and probiotics activity on Listeria

monocytogenes

Matthieu Million, Emmanouil Angelakis, Fatima Drissi, Didier Raoult

Published in Proc Natl Acad Sci U S A. 2013 Jan 2;110(1):E1. (IF 9.77)

102

www.pnas.org/cgi/doi/10.1073/pnas.1218418110 PNAS | January 2, 2013 | vol. 110 | no. 1 | E1

Article XI : REVIEW

Human gut microbiota: repertoire and variations

Jean-Christophe Lagier, Matthieu Million, Perrine Hugon, Fabrice

Armougom, Didier Raoult

Published in Front Cell Infect Microbiol. 2012;2:136 (IF: ND)

104

REVIEW ARTICLE published: 02 November 2012 CELLULARANDINFECTIONMICROBIOLOGY doi: 10.3389/fcimb.2012.00136 Human gut microbiota: repertoire and variations

Jean-Christophe Lagier, Matthieu Million, Perrine Hugon, Fabrice Armougom and Didier Raoult*

URMITE, UM63, CNRS 7278, L’Institut de Recherche pour le Développement 198, INSERM 1095, Aix-Marseille Université, Marseille, France

Edited by: The composition of human gut microbiota and their relationship with the host and, conse- Lorenza Putignani, Children’s Hospital quently, with human health and disease, presents several challenges to microbiologists. and Research Institute Bambino Gesù, Italy Originally dominated by culture-dependent methods for exploring this ecosystem, the Reviewed by: advent of molecular tools has revolutionized our ability to investigate these relationships. Nikhil Thomas, Dalhousie University, However, many biases that have led to contradictory results have been identified. Microbial Canada culturomics, a recent concept based on a use of several culture conditions with identifica- Jun Lin, The University of Tennessee, tion by MALDI-TOF followed by the genome sequencing of the new species cultured had USA allowed a complementarity with metagenomics. Culturomics allowed to isolate 31 new *Correspondence: Didier Raoult, URMITE, UMR CNRS bacterial species, the largest human virus, the largest bacteria, and the largest Archaea 7278, L’Institut de Recherche pour le from human. Moreover, some members of this ecosystem, such as Eukaryotes, giant Développement 198, INSERM U1095, viruses, Archaea, and Planctomycetes, have been neglected by the majority of studies. Faculté de Médecine, Aix-Marseille In addition, numerous factors, such as age, geographic provenance, dietary habits, antibi- Université, 27 Boulevard Jean Moulin, 13385 Marseille Cedex 5, France. otics, or probiotics, can influence the composition of the microbiota. Finally, in addition e-mail: [email protected] to the countless biases associated with the study techniques, a considerable limitation to the interpretation of studies of human gut microbiota is associated with funding sources and transparency disclosures. In the future, studies independent of food industry funding and using complementary methods from a broad range of both culture-based and molecular tools will increase our knowledge of the repertoire of this complex ecosystem and host-microbiota mutualism.

Keywords: gut microbiota, culturomics, metagenomics, archaea, transparency disclosures, antibiotics

INTRODUCTION 2008), Crohn’s disease ( De Hertogh et al., 2006 ; Manichanh et al., The exhaustive description of human microbiota and their rela- 2006; Scanlan et al., 2006 ), and metabolic diseases such as type II tionship with health and disease are major challenges in the diabetes ( Larsen et al., 2010 ) and obesity ( Ley et al., 2006b ; Turn- twenty-first century ( Turnbaugh et al., 2007 ). To assess the impor- baugh et al., 2006, 2009 ; Armougom et al., 2009 ; Santacruz et al., tance of this challenge, we used the ISI Web of Knowledge to 2009). demonstrate the dramatically renewed interest of scientists in this Based on these early data and to complete the description of the subject. To extend the chart presented by Sekirov et al. (2010) ; human gut composition, considerable funds have been granted. Marchesi (2011) , which lists the number of publications per year Among the projects pursuing this line of research, the human involving human gut microbiota, we found that in 2011, there microbiome project is an international consortium with the aim of were more than 4 times as many citations referencing human sequencing 1,000 bacterial genomes and multiplication by metage- gut microbiota than in 2005 ( Figure 1A ), when Eckburg et al. nomic analysis to characterize the complexity of microbial com- (2005) published the seminal large-scale gut metagenomics study. munities at several body sites, including the human gut, to deter- In addition, in 2011, there were approximately as many published mine whether there is a core microbiome ( Turnbaugh et al., 2007 ). items investigating human gut microbiota than during the 10 years Despite these advances in knowledge of gut microbiota compo- between 1993 and 2002 ( Figure 1B ). sition, the relationships of the microbiota with their host and, The human gut microbiota is composed of approximately consequently, with health and disease are still largely unknown, 10 11–12 microorganisms per gram of content, including diverse as reflected in several contradictory results ( Sekirov et al., 2010 ). populations of bacteria, mainly anaerobes (95% of the total), Moreover, molecular tools and by extension, experimental mod- which is 10 times higher than the total number of human cells ( Ley els, often reflect a reductionist approach as opposed to a holistic et al., 2006a ). In the study of human gut microbiota, two major approach ( Fang and Casadevall, 2011 ). Nevertheless, an appealing technological periods can be distinguished: schematic microscopic approach that was recently applied to the study of oral microbiota observation and culture-based methods before 1995 followed by will allow us to detect the minor bacterial populations, which are the advent of culture-independent methods. This technology- usually neglected, using dilution to obtain a threshold below 10 6 driven progress led to suggest relationships between gut micro- bacteria per ml or DNA >1 pg per µl ( Biesbroek et al., 2012 ). biota composition and diverse diseases, such as irritable bowel We propose here an inventory of current knowledge regarding syndrome ( Kassinen et al., 2007 ), polyposis or colorectal can- gut microbiota composition, the techniques used for this study cer ( Scanlan et al., 2008 ), necrotizing enterocolitis ( Siggers et al., and the relationships with the host. Finally, further research on

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 1 Lagier et al. Human gut microbiota

Nevertheless, these studies attempted especially to culture anaer- A Citaons in each year obic bacterial species whereas some gut bacteria preferentially 20000 grown in microaerophilic conditions. 15000 Among other unique problems associated with bacterial culture, Moore have also observed a major discordance between the 10000 culture counts and the microscopic counts of species ( Moore and 5000 Holdeman, 1974b); these discrepancies have been named by Staley 0 and Konopka (1985) as the “great plate count anomaly”.Indeed, it is generally accepted that only 1% of bacteria can be easily grown 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 in vitro (Vartoukian et al., 2010). Consequently, the major popula- Citaons in each year tion easily isolated from stools is composed of bacteria that grow B Pu blish ed item in eac h year quickly in classical high-nutrient growth media,with the usual carbon or electron sources at mesophilic temperatures ( Hugenholtz, 500 2002),and this constitutes the most studied bacteria. It is estimated 400 that approximately 75% of published studies by microbiologists 300 before the advent of molecular tools focused on only nine bacterial 200 genera among four phyla ( Actinobacteria , Proteobacteria , Firmi- 100 cutes , Bacteroidetes ; Galvez et al., 1998), whereas we know now 0 that more than 30 different phyla compose the gut microbiota 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 (Figure 2 ; Rajilic-Stojanovic et al., 2007). Nevertheless, studies of Published item in each year these fast-growing and easily cultured bacteria neglect the minority bacterial populations,including potentially pathogenic bacteria FIGURE 1 | Using the key words “human gut microbiota” or “human such as Salmonella typhi . fecal flora” and using the ISI Web of Knowledge database, (A) shows Finally, considering the main first culture-based studies the citations in each year regarding this subject, and (B) shows the number of bacterial species was estimated at approximately 400– number of published items each year, both between 1993 and 2011. 500 (Mata et al., 1969; Moore and Holdeman, 1974a; Finegold et al., 1977). In addition to the necessary use of stringent anaerobic conditions to culture bacteria from human stools, the usual phe- human gut microbiota is the subject of considerable funding by notypic identification methods are time consuming and expensive the food industry. Consequently, to perform an efficient analysis of (Seng et al., 2009). Indeed, the exponential technological advances this subject, the design and/or interpretation of the results of each in molecular tools led microbiologists to progressively abandon study can be associated with a conflict of interest. For example, it the culture-based approach for studies of the gut microbiota has recently been shown that published papers in obesity research ecosystem. in which the authors were funded by the food industry were more likely than other papers to contain results or an interpretation that METAGENOMICS AND PYROSEQUENCING favored the industry or company that was producing the product As often occurs during scientific progress, technological advances or service that was being studied ( Thomas et al., 2008). in microbiology allowed scientists to revisit the knowledge base (Rajilic-Stojanovic et al., 2007). Since 2000, large-scale 16S rRNA REPERTOIRE or metagenomic studies have allowed scientists to dramatically CULTURE expand the known diversity of the human gut microbiome, illu- Culturing has been the first method used to characterize a bac- minating new ways ( Eckburg et al., 2005; Andersson et al., 2008). terial ecosystem ( Finegold et al., 1974, 1977; Moore and Holde- It is now commonly accepted that approximately 80% of the bac- man, 1974a). Gut composition was first studied by microscopic teria found by molecular tools in the human gut are uncultured, observation and axenic culture. Gram staining has been widely and hence can be characterized only by metagenomic studies ( Eck- used by microbiologists to describe stool composition. Using this burg et al.,2005). Whereas the number of species was limited in the technique, gram-positive bacteria accounted for only 2–45% of seminal studies using culture-based methods ( Finegold et al.,1974; the cells observed ( Gossling and Slack, 1974). However, a dis- Moore and Holdeman, 1974b), Turnbaugh et al. (2010) estimated crepancy arises because culture counts reveal a predominance of 473 phylotypes using V2 pyrosequencing. There is a significant gram-positive bacteria in human feces. Indeed, one of the first discrepancy between bacterial observations with a microscope and culture studies of human stools showed that anaerobes always most of the molecular studies, which observe a striking dominance constitute the major component of the culturable flora of chil- of gram-positive bacteria ( Eckburg et al., 2005; Andersson et al., dren and adults ( Mata et al., 1969), with a predominance of 2008; Turnbaugh et al., 2010; Figure 2 ). gram-positive cells. Moore and Holdeman (1974a), in a study Indeed, these recent methods generate contradictory results of 20 individuals, revealed 113 different bacteria, including more reflecting the biases in every step of the Polymerase Chain Reac- gram-positive bacteria ( Bifidobacterium , Eubacterium , Peptostrep- tion procedure. A dramatic divergence in the proportion of the tococcus , Ruminococcus , Lactobacillus , and Clostridium genera) different phyla was observed depending of the type of extraction than gram-negative bacteria ( Bacteroides , Fusobacteria genera;). kit used, notably for the Fusobacteria (2–40%) and Bacteroidetes

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 2 Lagier et al. Human gut microbiota

100%

80%

60%

G ram - 40% G ram + 20%

0% Gram ME Pyroséquençage

Phylum Reads %

Firmicutes 44769 71.12 Actinobacteria 5797 9.21 Other 6627 10.53 Bacteroidetes 3983 6.33

Proteobacteria 1747 2.78 Cyanobacteria 21 0.03 Verrucomicrobia 4 0.01 Gram staining ( ×100 oil immersion) Electron microscopy ( ×7100) 53% bacteria Gram -positive 60% bacteria Gram -positive Tota l 62948 100 47% bacteria Gram-negative 40% bacteria Gram-negative Pyrosequencing 80% bacteria Gram-positive 9% bacteria Gram-negative 11% not available

FIGURE 2 | A comparison of Gram staining, electron microscopy, and pyrosequencing to determine the proportion of Gram-positive/Gram-negative bacteria in the same stool sample (personal data).

(40–60%) phyla ( Wu et al., 2010). In addition, the relative abun- to more quickly analyze longer reads sequenced to study larger dance of a phylum depends significantly on the 16S hypervariable cohort samples in low taxonomic level. region, independent of pyrosequencing chemistry. For example, Finally,molecular methods detected bacteria present at concen- the 454 titanium and Illumina next-generation sequencing (NGS) trations greater than approximately 10 6 and neglected minority methods reveal a dominance of the Bacteroidetes phylum using 16S populations. Among these neglected populations are potentially rDNA v4v5 region primers, whereas Firmicutes was predominant pathogenic bacteria such as S. typhi , Yersinia enterocolitica , and using v3v4 primers on the same gut microbiota ( Claesson et al., Tropheryma whipplei , which may be present in human stools at 2010). Using 454 titanium, Ralstonia genera have been detected concentrations below 10 5 cfu per ml ( Raoult et al., 2010), the cur- only by V4/V5 primers, whereas Bifidobacteria have been detected rent threshold of the latest NGS method ( Turnbaugh et al., 2010; only by V3/V4 primers ( Turnbaugh et al., 2010). In parallel, Hong Lagier et al., 2012a; Figure 3 ). The depth is directly correlated with et al. (2009) have described that the rRNA approach misses half of the number of generated sequences, and no plateau was obtained the bacteria in environmental microbiology. in the number of phylotypes observed, although close to 1,000,000 Although controversial, the higher taxonomic level analyses (as 16S rRNA gene amplicons have been sequenced by Turnbaugh phylum level) have suggested an association between obesity and et al. (2010). Firmicutes/Bacteroidetes proportion ( Ley et al., 2005). The genus- level analysis has allowed to hypothesize specific enterotypes com- VIRUSES positions despite controversies ( Arumugam et al., 2011). In addi- Research in the human gut has been focused on bacterial com- tion, Murphy et al. (2012) has recently observed in a study from the position ( Walker, 2010). Early studies suggested that most DNA manipulation of the mice gut microbiota in diet-induced obesity viruses found in the intestine were phages and that most RNA that a better separation of lean and diet-induced obese mice was viruses were plant viruses ( Breitbart et al., 2008). Nevertheless, observed at the family and genus-level than at the phylum level. a recent metagenomic study carried out over 1 year, with three However, the large inter-individual variability leads the analysis of stools analyzed from each monozygotic adult twin and their lower taxonomic-level to complex results because of small number mother, revolutionized virome knowledge ( Reyes et al., 2010). of samples. Finally, the optimization of primers able to detect gen- The authors carried out shotgun pyrosequencing to generate over era often misdetected by pyrosequencing, as Bifidobacteria (Sim 280 Mb of sequence and, at the same time, a pyrosequencing of et al., 2012), and technology progress in pyrosequencing, will allow 16S rRNA genes to identify the bacterial species. Approximately

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 3 Lagier et al. Human gut microbiota

Microorganisms per gram of feces

12 10 Bacteroides spp. Eubacterium spp . Ley et al. 10 11 Clostridium spp. 2006 Andersson SANGER et al. 2008 Turnbaugh Peptostreptococcus spp. Qin et al. 10 et al. 10 GS 20 2010 Bifidobacterium spp. 201 0 Vibrio cholerae SOLEXA GS FLX 10 9 E. coli O157 Shigella dysenteriae Enterotoxigenic E. coli 10 8 Streptococcus spp. 10 7 Aeromonas hydrophila Clostridium difficile Campylobacter jejuni 10 6 Lactobacillus spppp .

10 5 Salmonella Typhimurium Tropheryma whipplei 10 4 Yersinia enterocolitica 10 3 Salmonella Typhi

10 2

10 1

FIGURE 3 |The statistical detection thresholds of metagenomic methods. The statistical detection thresholds of metagenomic methods are correlated with the number of bacteria in the ecosystem studied by the number of sequences generated.

80% of sequencing reads did not match any known viruses in for creating arrayed species collections that do not detect minority the database corresponding to prophages or temperate phages. populations. These populations were persistent in each individual, with no In addition to the stringent culture conditions, some of the dif- significant clustering between co-twins or between twins and ficulties linked to culture include the cost and the amount of time their mothers, contrasting with the bacterial similarity between required for bacterial identification ( Seng et al., 2009). These diffi- twins ( Turnbaugh et al., 2009). In addition, Minot et al. (2011) culties have recently been overcome by mass spectrometry, which observed that a change of diet is associated with a change in virome enables quick and effective identifications in routine bacteriology composition. (Seng et al., 2009, 2010) and allow the researcher to quickly check the major population and to concentrate interest on the minor- CULTUROMICS ity population. We have recently reported a breakthrough in this There has been a renewed interest in culture methods for these field of research with the microbial culturomics concept ( Lagier “non-cultivable” species ( Vartoukian et al., 2010). Initially, envi- et al., 2012a). We applied 212 different culture conditions in two ronmental microbiologists were confronted with the fact that the African stools and a French obese stool samples, including enrich- majority of bacteria do not grow in classical Petri dishes. These ment techniques, Escherichia coli phage cleaning, and innovative first studies used prolonged incubation and stringent anaerobic conditions (using rumen fluid, sterile human stools). We analyzed conditions, notably, diffusion chambers ( Kaeberlein et al., 2002; 32,500 colonies by MALDI-TOF, allowed us to culture 340 dif- Bollmann et al., 2007), with the aim of simulating the natural ferent bacterial species among seven phyla and 117 genera. This environment of these “uncultivable” microorganisms ( Kaeberlein included 174 species never described in the human gut. More- et al., 2002). This technique enlarged the diversity of the envi- over 31 new species were found, including five new genera, as well ronmental microorganisms that were isolated ( Epstein, 2009). In two species from rare phyla ( Deinococcus-Thermus and Synergis- parallel, a recently published study proposed an anaerobic culture tetes ). Genome sequencing and description of each new species of a single stool sample to complement 16S rRNA sequencing, is in progress ( Kokcha et al., 2012; Lagier et al., 2012b,c; Mishra using rumen fluid or an extract of fresh stools to mimic the nat- et al., 2012a,b,c,d). By comparison, pyrosequencing of 16S rDNA ural environment of the gut bacteria. Goodman et al. (2011) have amplicons from the three stools noted a dramatic discrepancy with recovered 36 cultured species: four uncultured described species culturomics as only 51 species identified by 16S rDNA amplifi- and 53 unknown isolates with different v2 sequences. However, cation and sequencing were also found among the 340 cultured these authors used the most probable number (MPC) technique species highlighting the renewed interest for the culture in the gut

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 4 Lagier et al. Human gut microbiota

Gram+ Gram- No staining LPS absent Pepdoglycan absent Phylum present in human Verrucomicrobia SR1 Bacteroidetes Sphaerobacter thermophilus Chrysiogenetes Thermobaculum terrenum Spirochaetes Thermotogae Chloroﬂexis, Nitrospira Chloroﬂexi Herpesphon Dehalococcoides Cyanobacteria Acidobacteria

Lensphaerae Unculvated phyla Deferribacteres -OP1, OP3, OP8, OP9,OP11 Chlorob i -BRC1 Elusimicrobia (Termite -OD1 Group 1) -ABY1 SR1 Dictyoglomi -TM6 Fusobacteria -TM7 Ca ldiser i ca -WS2, WS3, WS5, WS 6 Deinococcus-Thermus (OP5) -… Fibrobacteres Proteobacteria

Synergistetes Armamonadetes (OP10)

Gemmamonadetes Aquiﬁcae Planctomycetes Acidaminococcus Firmicutes Eubacterium Chlamydiae Oscillibacter Tenericutes Acnobacteria Mycobacterium (triple layered structure)

FIGURE 4 | A non-exhaustive representation of different bacterial Mycobacterium ) or without a cell wall ( Tenericutes ) have abnormal phyla found in culture (outer star in blue) or phyla with no Gram staining and are shown in pink. The purple triangle represents the representative in culture (inner star in gray). Gram-positive bacteria absence of lipopolysaccharide in the outer membrane of Gram-negative are colored in green, and Gram-negative bacteria are colored in white. bacteria. The red square symbolizes phyla that do not have a Bacteria with an atypical cell wall (triple-layered structure of peptidoglycan structure. microbiota study. Culturomics allowed us break several “records” rDNA amplicons from the three stools identified 698 phylotypes with the largest number of bacteria cultured from a single stool including 282 phylotypes of known bacterial species and 416 phy- (219 species), the first bacteria from Deinococcus-Thermus phylotypes of uncultured bacteria. We noted a dramatic discrepancy lum isolated from human, the largest human virus and the largest with culturomics as only 51 species identified by 16S rDNA ampli- bacteria from human ( Lagier et al., 2012a). fication and sequencing were also found among the 340 cultured species. Consequently,microbial culturomics increased by 30% the COMPARISON OF THE TECHNIQUES microbial repertoire of the human gut studied by pyrosequencing There are currently no rational explanations for the typical (Lagier et al., 2012a). observed proportions of gram-positive/negative bacteria, which are highly divergent microscopically ( Turnbaugh et al., 2007) with GAPS IN KNOWLEDGE culture, ( Gossling and Slack, 1974) and the proportions obtained In addition to the bias previously described, some components of by sequence detection ( Eckburg et al., 2005; Figures 2 and 4). In human gut microbiota have been partially neglected by the current 2002, Hayashi compared the digestive microbiota of three indi- tools ( Figure 5 ). viduals by cloning/sequencing and anaerobic culture using the “plate-in-bottle” method. These researchers isolated between 48 EUKARYOTES and 65 phylotypes in the cloning of individuals and 48 species, Eukaryotes are an important part of the human gut microbiome of which three individuals were potentially three new species and play different beneficial or harmful roles. Some species may be (Hayashi et al., 2002b). In light of the phylogenetic tree described commensal or mutualistic, whereas others may be opportunistic in this publication, these authors found significant discrepancies or parasitic ( Parfrey et al., 2011). The eukaryotic component of the between these two techniques, which were somewhat surprising human gut microbiome remains unexplored because these organ- given the low number of species and phylotypes identified. Sev- isms are of limited interest ( Marchesi, 2010). Culture-dependent eral species in culture had no equivalent in cloning. A previous techniques and microscopy-based approaches have been mainly study compared these same techniques, but the number of species used to explore eukaryotes in the human gut,and identification has and phylotypes was even lower ( Wilson and Blitchington, 1996). frequently been based on morphological and physiological tech- In this study, of 48 species, 25 were detected only by cloning, nine niques with numerous biases. Moreover,this approach detects only were common to both techniques, and 14 were identified only by a small fraction of microorganisms, including Candida and Sac- culture. In addition, in our microbial culturomics study, by com- charomyces spp., but the growth requirements for many eukaryotic parison with the 340 bacteria cultured, pyrosequencing of 16S species remain unknown.

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 5 Lagier et al. Human gut microbiota

FIGURE 5 | A non exhaustive overview of human gut microorganisms among bacterial, Archaea, viral, and Eukaryota domains.

Using culture-independent methods, Scupham et al. (2006) giant virus that is frequently missed by large-scale virome metage- have identified a large number of fungi, including Ascomycota , nomics studies that use only 0.22 µm filters, making giant virus Basidiomycota , Chytridiomycota , and Zygomycota phyla, in stud- detection unlikely ( Willner et al., 2009; Reyes et al., 2010). ies of mouse feces. Furthermore, Scanlan and Marchesi (2008), In our laboratory, in an effort to obtain fastidious bacteria from studying the human distal gut, have shown that the diversity and an African stool sample by amoeba ( Acanthamoeba polyphaga ) co- abundance of eukaryotes is low relative to those of bacteria. Only culture, we obtained a new giant virus strain named Senegal virus members of the genera Gloeotinia /Paecilomyces and Galactomyces (Lagier et al., 2012a), which we sequenced (Genbank JF909596– have been identified as the most abundant. Nevertheless, we have JF909602). These findings indicate that giant viruses may be a shown that due to a large variety of primers used, the human gut part of the gut microbiota and that virome metagenomic studies contains a broader eukaryotic diversity than predicted ( Hamad should use different filter sizes. Because the potential pathology et al., 2012). In parallel, applying traditional and modern labo- of the giant viruses is currently unknown, it is unreasonable to ratory techniques (using intergenic spacers for 18S rDNA), the neglect them ( Boyer et al., 2009). repertoire of intensive care unit pneumonial microbiota has been considerably extended, notably regarding fungal microbiota and Archaea plants (Bousbia et al., 2012). Nottingham and Hungate (1968) isolated a previously unidentified methanogenic Archaea from human feces using a non- Giant viruses selective medium and a stringent anaerobic atmosphere composed Giant viruses growing in amoebae have previously been isolated of 80% H 2 and 20% CO 2. Miller et al. (1982) isolated Methanobre- in the environment, e.g., in the water of cooling towers, in rivers vibacter smithii from human stool specimens from four healthy and lakes, in seawater, in decorative fountains, and in soil ( Pag- adults using anaerobic cultures enriched with the same H 2–CO2 nier et al., 2008). Mimivirus DNA has been obtained from the anaerobic atmosphere pressurized to two bars. Illustrating the bronchoalveolar lavage of patients ( Raoult et al., 2007; Lysholm technical limitations of the fastidious Archaea culture, in our lab- et al., 2012), and a laboratory infection was documented by serol- oratory, we have recently achieved the isolation of the fourth ogy ( Raoult et al., 2006). In addition, Lysholm et al. (2012), in a methanogenic Archaea species in humans and the first cultured viral microbiome metagenomic study performed in 210 children representative of a new order of Archaea ( Methanomassiliicoccus and adults with lower respiratory infections, recently identified luminyensis ) after a 16-month tentative culturing procedure. We Mimivirus. Because the authors used two pools and filtered with obtained this strain after subtle modifications in the composition 0.22 and 0.45 µm pore-size disk filters, they were able to isolate a of the culture medium (enzyme co-factors) and adaptation of the

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 6 Lagier et al. Human gut microbiota

atmospheric pressure (the culture medium is patented; Dridi et al., COMPOSITION 2012a). In addition, the genome sequencing of this new species The composition of the human gut ecosystem is influenced by represents the largest genome of a methanogenic euryarchaeota multiple and diverse factors, some physiological (age, origin, envi- isolated from humans ( Gorlas et al., 2012). ronment) and others linked to external factors, such as dietary In addition, recent molecular studies indicated that human habits, antibiotics, and probiotics ( Figure 6 ). Archaea constitute an expanding world ( Dridi et al., 2011). Using 16S rDNA sequencing, many studies confirmed the presence of M. AGE smithii and M. stadtmanae in the human gut, with variable and In a pioneering study using microarrays to detect small rRNAs, low prevalence ( Dridi et al., 2009). Nevertheless, in our study, our Palmer followed a cohort of newborns, including a pair of twins, new Archaea was detected in stools in 4% of individuals, and its during the first year of life. It was shown that despite considerable prevalence increases with age, although its role in human health is temporal variations and environmental influences, the compo- unknown ( Dridi et al., 2012b). Regarding the influence of Archaea sition of the intestinal ecosystem tended to be characteristic of on human health, a recent meta-analysis compared the number of adulthood at the end of this period ( Palmer et al., 2007). The pro- sequences of Methanobrevibacter spp. in stools. Obese individuals portion of Bacteroides fragilis increased from 1 month to 1 year had fewer Methanobrevibacter genera by quantitative polymerase (Vael et al., 2011). In a 2.5-year case study, Koenig analyzed chain reaction (qPCR) than non-obese subjects ( Angelakis et al., >300,000 16S rRNAs from 60 fecal samples from healthy children 2012a). Previous studies had reported discordant results concern- and showed that infant gut variation is associated with life events. ing the levels of detection of M. smithii in the obese gut ( Zhang The phylogenetic diversity of the microbiome increased gradually et al., 2009; Schwiertz et al., 2010; Million et al., 2011). In addition, over time with progressive temporal changes but, inversely, the the detection of Archaea in the vaginal flora of pregnant women major phyla, genera, and species composition showed rough shifts allowed us to hypothesize a possible mother-to-child transmission in abundance corresponding to modifications in diet or health (Dridi et al., 2011). (Koenig et al., 2011). Nevertheless, using microbiota array to analyze gut microbiota Planctomycetes composition in adolescent subjects, Agans et al. (2011) found The phylum Planctomycetes, phylogenetically closely related to a statistically significantly higher abundance of Bifidobacterium Verrucomicrobia and Chlamydiae, is composed of environmen- and Clostridium genera, contrary to current knowledge, suggest- tal microorganisms characterized by a peptidoglycan-free cell wall ing that the gut microbiome of adolescents is different from that and cell compartmentalization ( Fuerst and Sagulenko, 2011). The of adults. At the other extreme of life, using pyrosequencing of 16S culture is fastidious and requires the addition of appropriate rRNA gene V4 region amplicons, the gut microbiota composition antibiotics (peptidoglycan synthesis inhibitors) and amphotericin of elderly subjects was distinct from that of younger adults, with B to prevent contamination of the culture medium. Undetected by a greater temporal stability over a limited time, particularly in the conventional 16S rRNA PCR or standard culture techniques, this proportion of Bacteroides spp . (Claesson et al., 2011). phylum has been reported in black-and-white colobus monkey stools ( Yildirim et al., 2010) and, in one instance, in the human GEOGRAPHICAL PROVENANCE AND ENVIRONMENT gut microbiota, using metagenomics ( De Filippo et al., 2010). In Discordant results have also been published regarding a geographic our laboratory, preliminary results (unpublished data) confirmed signature of the gut microbiota depending on the technique used. the presence of specific Planctomycetes DNA in human stools. To investigate the hypothetical association between gut composi- Several species of Planctomycetes and, more generally, of species tion and cancer, early culture-dependent studies compared pop- including the superphyla Verrucomicrobia, Planctomycetes, and ulations at high-risk (western countries) and at low risk (Japan, Chlamydiae, are undergoing genome sequencing. It is expected Uganda, India) and reported different compositions of microbiota that this sequencing will increase our knowledge of this specific (Hill et al., 1971; Drasar et al., 1973; Finegold et al., 1974). The branch of the tree of bacterial life ( Wagner and Horn, 2006). high-risk population had a microbiota composed primarily of Bacteroidetes , and there were specific differences, including pat- The variability depending of the gut samples terns of food consumption, between western countries and Asian “The gut microbiota is non-homogenous with a progressive or African populations,although multiculturalism and population increase of bacterial concentration from the stomach (approxi- exchanges have reduced these differences. Only a few large-scale mately 10 3 bacteria per gram to the colon (approximately 10 11 bac- molecular studies have used stool samples collected from Asia or teria per gram; O’Hara and Shanahan, 2006). Nevertheless, most Africa ( De Filippo et al., 2010; Lee et al., 2011), where approx- of studies explored stools samples reflecting mainly the colonic imately 75% of the population of the world lives; nevertheless, composition. However, differences in compositions have been the findings have suggested a possible signature of biogeography reported between small intestine biopsies (most of Streptococ- (Lee et al., 2011). Indeed, most of the large-scale metagenomic or caceae belonging to Firmicutes phylum and Actinomycinaeae and pyrosequencing studies used stools collected from American or Corynebacteriaceae belonging to Actinobacteria phylum) whereas European individuals ( Ley et al., 2006b; Turnbaugh et al., 2006; colonic biopsies were enriched by Bacteroidetes phylum and Lach- Claesson et al., 2011). nospiraceae among the Firmicutes phylum ( Frank et al., 2007). Lay, characterizing 91 European gut microbiota using FISH Intestinal analysis of tiered samples will allow to exhaustively combined with flow cytometry, did not observe a significant describe the gut composition.” grouping with regard to country of origin (France, Netherlands,

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 7 Lagier et al. Human gut microbiota

Asian vs American European vs African Geographic Southern vs Northern Europeans provenance Age Diﬀerences in species level but Dietary Habits Contradictory results First month of life : Mostly Biﬁdobacteria

1 month to 1 year of life : Increase Bacteroides fragilis Vegetarians : Human gut Increase Bacteroidetes contradictory results Adolescent : composion Decrease Clostridia

Adult : Stable: Increase Lactobacillus spp. ? Firmicutes > Bacteroidetes > Proteobacteria> Acnobacteria Probiocs

Elderly : Increase Bacteroidetes Decrease Biﬁdobacteria Malnurished children : Decrease bacterial load ? Increase Proteobacteria Alteraon of com posion ? Decrease Bacteroidetes

Malnutrion Anbiocs

FIGURE 6 |The inﬂuence of external factors determining the composition of the human gut microbiota.

Denmark, UK, and Germany; Lay et al., 2005). With the same tech- The authors hypothesized that the abundance of these genera nology, Mueller et al. (2006) found differences in Bifidobacteria could be a consequence of the high intake of fiber, similar to species between European individuals. Grzeskowiak et al. (2012), the diet of early human settlements at the time of the birth of using flow cytometry-FISH and qPCR, have shown that sev- agriculture, maximizing the extraction of metabolic energy from eral species ( Bifidobacterium adolescentis , Staphylococcus aureus , the polysaccharides of ingested plants ( De Filippo et al., 2010). and Clostridium perfringens ) were absent in Malawian children A vegetarian diet affects the intestinal microbiota, specifically by but present in 6-month-old Finnish infants. Fallani comparing decreasing the amount and modifying the diversity of Clostrid- infants living in northern or southern European countries by 16S ium cluster IV ( Liszt et al., 2009). Based their studies on RFLP rDNA pyrosequencing, have found that geographical provenance analysis, Hayashi et al. (2002a) Hayashi found that the digestive is important, with a higher proportion of Bifidobacteria in north- microbiota of vegetarians harbored Clostridium rRNA clusters ern infants and more Bacteroidetes and Lactobacilli in southern XIVa and XVIII. Recently, Walker et al. (2011) tested overweight European countries. people successively with a control diet, a diet high in resistant Finally, Arumugam studied 22 fecal metagenomes of individ- starch (RS) or non-starch polysaccharides (NSP) and a reduced uals from four different countries and identified three different carbohydrate weight loss (WL) diet for 10 weeks by two different enterotype clusters, which were independent of geographic prove- methods: large-scale sequencing and quantitative PCR. No signifi- nance. The three different enterotypes were, respectively, richer in cant effect was observed at the phylum level, but at finer taxonomic Bacteroides , in Prevotella , and in Ruminococcus for the Enterotype level, Eubacterium rectale and Ruminococcus bromii showed signif- 3. Arumugam suggested that each enterotype used a different route icant and dramatic (fourfold) increased proportions in the RS diet, to generate energy ( Arumugam et al., 2011). whereas the proportion of Collinsella aerofaciens -related sequences was decreased significantly on the WL diet ( Walker et al., 2011). DIETARY HABITS In this study, reproducible changes were found only at the phy- Dietary habits are thought to be a major factor contributing to lotype level, whereas no differences were significant at a broader the diversity of the human digestive microbiota ( Backhed et al., taxonomic level (the phylum or family Ruminococcaceae level), 2005). Part of the geographic diversity of the gut microbiota and the analysis suggested that the amplified 16S rRNA sequence seems to be explained by differences in diet. For example, African clustered more strongly by individuals than by diet. These changes children from a rural area in Burkina Faso showed a specific are entirely in agreement with studies using RNA-based stable iso- abundance of Prevotella and Xylanibacter , known to contain a tope probing, which showed that R. bromii was the first starch set of bacterial genes for cellulose and xylan hydrolysis, com- degrader in the human gut ( Kovatcheva-Datchary et al., 2009). pletely lacking in European children ( De Filippo et al., 2010). Wu analyzed stool samples from 98 individuals and found that

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 8 Lagier et al. Human gut microbiota

enterotypes were strongly associated with long-term diets, espe- Consequently, glucose is rapidly absorbed, producing substantial cially for animal fat and protein ( Bacteroides ) vs. carbohydrate elevations in serum glucose and insulin, both factors that trigger (Prevotella ). Changes in gut microbiota related to short-term diet lipogenesis. In addition, fatty acids are stored excessively, with de modifications occurred rapidly, were detectable within 3–4 days, novo synthesis of triglycerides derived from the liver. Together, and were rapidly reversed ( Walker et al., 2011; Wu et al., 2011). these phenomena cause weight gain ( Backhed et al., 2007). Using Conversely, Wu et al. (2011) suggested that long-term dietary Fasting-Induced Adipocyte Factor (Fiaf) knockout mice, Backhed interventions might allow the pervasive modulation of an individ- et al. (2007) showed that gut microbiota suppressed intestinal Fiaf, ual’s enterotype to improve health. In animal models of obesity consequently increasing the storage of calories. induced by diet (DIO), Turnbaugh et al. (2008) showed that a Second, the composition of gut microbiota has been shown high-fat diet could significantly alter the intestinal flora of exper- to selectively suppress angiopoietin-like protein 4/fasting-induced imental models with a bloom in a single uncultured clade within adipose factor in the intestinal epithelium. This molecule is a cir- the Mollicutes class of the Firmicutes . Hildebrandt et al. (2009) culating lipoprotein lipase inhibitor and a regulator of peripheral suggested that a high-fat diet altered the intestinal flora regardless lipid and glucose metabolism ( Backhed et al., 2004). Backhed et al. of weight change. These authors observed a bloom of Clostridia (2004) showed that when the microbiota of normal mice were and Proteobacteria associated with the high-fat diet. The major transplanted into germ-free mice, after 2 weeks, body fat increased group of Proteobacteria that increased in abundance was the Delta- by 60% without increased food consumption, modifications of Proteobacteria phylum, order Desulfovibrio . Finally, Monira et al. energy expenditure, or relative insulin resistance, and there was a (2011) have recently published a study comparing the gut flora 2.3-fold higher production of triglycerides in the liver, suggesting of malnourished children with that of well-nourished children in that the gut operates in host energy homeostasis and adiposity. Bangladesh and found a decrease in Bacteroidetes and an increase Third, it has been suggested that bacterial isolates of gut micro- in Proteobacteria phyla, including E. coli and Klebsiella spp. biota may have pro- or anti-inflammatory properties, impacting weight. Obesity has been associated with a low-grade sys- OBESITY AND GNOTOBIOTIC MICE temic inflammation corresponding to higher plasma endotoxin Beginning in 2005, obesity has been associated with a specific lipopolysaccharide (LPS) concentrations, defined as metabolic profile of bacterial gut microbiota, including a decrease in the endotoxemia ( Bastard et al., 2006; Hotamisligil, 2006; Sbarbati Bacteroidetes/Firmicutes ratio ( Ley et al., 2005, 2006b; Turnbaugh et al., 2006; Fogarty et al., 2008). Cani et al. (2008) showed that et al., 2006, 2009) and decreased bacterial diversity ( Turnbaugh antibiotics can lower LPS levels in mice fed a high-fat diet and et al., 2009). Since these pioneering studies, significant associa- in ob/ob mice and, consequently, can reduce glucose intolerance, tions have been found between obesity and an increase in some body weight gain, and fat mass. Conversely, some Bifidobacterium bacterial groups, including Lactobacillus , S. aureus, E. coli , and Fae- and Lactobacillus species have been reported to deconjugate bile calibacterium prausnitzii (Collado et al., 2008; Kalliomaki et al., acids, which may decrease fat absorption ( Shimada et al., 1969). 2008; Armougom et al., 2009; Santacruz et al., 2009; Balamuru- gan et al., 2010). In a recent review, we found no reproducible and ANTIBIOTICS AND PROBIOTICS significant alteration linking obesity and gut microbiota at the phy- Antibiotics and total bacterial count lum level ( Angelakis et al., 2012a). Conversely, meta-analysis at the According to the literature, oral or intravenous antibiotics tend genus-level found decreased levels of bifidobacteria ( Collado et al., to decrease the bacterial load in the digestive tracts of infants 2008; Kalliomaki et al., 2008; Santacruz et al., 2009; Balamurugan (Palmer et al., 2007) and elderly patients ( Bartosch et al., 2004). et al., 2010; Schwiertz et al., 2010) and Methanobrevibacter spp. However, other studies reported that only the microbiota compo- (Armougom et al.,2009; Schwiertz et al.,2010; Million et al.,2011), sition is altered,and the total biomass is not modified by antibiotics the leading known representative of Archaea in the human gut, in (Sekirov et al., 2008). In contrast to amoxicillin and metronidazole overweight/obese people. To date, controversial studies show that or cefoperazone, Robinson noted that the alterations in commu- the connection between the microbiome and excess weight is com- nity structure associated with vancomycin specifically occurred plex ( Pennisi,2011). Wefound a difference at the species level,with without a significant decrease in the overall bacterial biomass L. reuteri enriched in obese gut microbiota, whereas L. plantarum (Robinson and Young, 2010). was increased in lean individuals ( Million et al., 2011). At the gene level, obesity has been associated with an altered representa- Structural disruption tion of bacterial genes and metabolic pathways. Turnbaugh et al. Antibiotic administration has a reproducible effect on the com- (2009) showed that diversity of organismal assemblages yields a munity structure of the indigenous gastrointestinal microbiota core microbiome at a functional level and that deviations from this in mice ( Robinson and Young, 2010). A very recent study found core are associated with different physiological states (obese com- that the administration of a commercial growth-promoting antibi- pared with lean), with obese gut microbiota having an increased otic combination (ASP250: chlortetracycline-sulfamethazine and capacity for energy harvest. penicillin) entailed a reproducible bloom in proteobacteria (1– As a theoretical basis for the causal link between alterations in 11%) in swine gut microbiota ( Looft et al., 2012). This shift was the gut microbiota and obesity, several mechanisms have been driven by an increase in E. coli populations. In humans, analy- suggested. First, the gut microbiota may interact with weight sis of the fecal microbial populations of infants after antibiotic regulation by hydrolyzing indigestible polysaccharides to easily therapy showed a major alteration as measured by SSU rDNA absorbable monosaccharides and by activating lipoprotein lipase. microarray analysis ( Palmer et al., 2007) or culture-based methods

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 9 Lagier et al. Human gut microbiota

(Savino et al., 2011). In adults, the same dramatic shift has been The reversibility of structural gut microbiota modification reported, depending on the antibiotic. Clindamycin ( Donskey The recovery of the gut community toward baseline after short- et al., 2003; Jernberg et al., 2005a) has the strongest effect com- term antibiotic therapy has been reported in animal models pared to oral cephalosporin, which is responsible for minor or no (Robinson and Young, 2010), but pervasive disturbance to the changes ( Swedish Study Group, 1991a,b). Of note, the extremely community has been observed several weeks after withdrawal of moderate effect of cephalosporin on gut microbiota ( Donskey certain antibiotics, including cefoperazone ( Robinson and Young, et al., 2003) has been linked with the low activity of this mol- 2010) and quinolones ( Dethlefsen et al., 2008). Changing the ecule on intestinal anaerobes. Moreover, the fecal elimination of intestinal microbiota of termites with antibiotics offers a privileged carbapenems is very limited, explaining why changes in the intesti- experimental model and has shown that prolonged antibiotic nal microflora are only moderate, whereas these agents have the treatment with rifampicin has an irreversible effect not only broadest spectra of the beta-lactam antibacterial agents ( Sullivan on microbial diversity but also on longevity, fecundity and the et al., 2001). The characterization of gut microbiota alteration weight (weight gain compared to controls) of two termite species, by metagenomic analysis of the v3–v6 region has been stud- Zootermopsis angusticollis and Reticulitermes flavipes. ied in three patients on ciprofloxacin ( Dethlefsen et al., 2008). Ciprofloxacin decreased to one-third the abundance of taxa [num- Probiotics ber of ref Operational taxonomic units (OTU)], their diversity Probiotics were initially used in agriculture to prevent diarrhea in and distribution. However, comparing gut microbiota alterations poultry because they reduce intestinal colonization by Salmonella by DGGE analysis, the rate of similarity with the pre-treatment spp. and C. perfringens (Angelakis and Raoult, 2010), but the use profile was 73% with ciprofloxacin but only 11–18% with clin- of probiotics such as Lactobacillus spp. can led to a rapid weight damycin ( Donskey et al.,2003). In addition,ciprofloxacin has been increase in chickens ( Angelakis and Raoult, 2010). L. acidophilus , reported to have little or no impact on anaerobic intestinal micro- L. plantarum , L. casei , L. fermentum , and L. reuteri are the most biota ( Nord, 1995; Edlund and Nord, 1999b). Glycopeptides, used commonly used Lactobacillus species in agriculture ( Anadon et al., widely in agriculture as growth promoters, are associated with 2006). The inoculation of L. ingluviei in mice is responsible for natural resistance of most of the lactobacilli and have no effect gut flora alterations associated with an increase in weight gain and on gram-negative bacteria, including Enterobacteria (Barna and liver enlargement ( Angelakis et al., 2012b). In parallel, probiotics Williams, 1984). Analyzing vancomycin-associated gut microbiota are increasingly used in human foods, notably in the milk indus- alterations in mice by cloning sequencing, Robinson found that try ( Raoult, 2008). Although the mechanisms are not yet known, vancomycin increased members of the Proteobacteria and Tener- many studies suggest that probiotics function through direct or icutes phyla and the Lactobacillaceae family, whereas members indirect impacts on colonizing microbiota of the gut ( Sanders, of the Lachnospiraceae family decreased ( Robinson and Young, 2011). Million et al. (2011) recently found that different Lacto- 2010). Using a continuous-culture colonic model system, Macca- bacillus species may have a paradoxical effect, with higher levels ferri et al. (2010) demonstrated that rifaximin, reported to induce of L. reuteri and lower levels of L. plantarum and L. paracasei in clinical remission of active Crohn’s disease while not altering obese gut microbiota. A recent systematic meta-analysis reported the overall structure of the human colonic microbiota, increased that the administration of L. acidophilus is responsible of weight Bifidobacterium , Atopobium , and F. prausnitzii and led to a vari- gain in human and animals and that the use of L. fermentum and L. ation of metabolic profiles associated with potential beneficial ingluviei resulted of weight gain in animals ( Million et al., 2012). effects on the host. The effects of tetracycline on gut micro- Thuny et al. (2010) observed a weight gain in patients treated biota in humans are of particular interest because this antibiotic with vancomycin and hypothesized that the gain was induced by is commonly used in poultry production as a growth promoter, the growth-promoting effect of Lactobacillus spp., as these species suggesting dramatic changes in intestinal microbial populations. are resistant to glycopeptides. In contrast, symbiotics (the combi- One notable effect of tetracycline is a decrease in bifidobac- nation of prebiotics and probiotics) have been proposed for the teria ( Nord et al., 2006; Saarela et al., 2007). Overall, specific management of malnutrition, with promising results on mortality gut microbiota changes are associated with specific antibiotics (Kerac et al., 2009). After gavage of gnotobiotic mice with a combi- (Table 1 ). nation of bacteria, including B. animalis subsp. lactis , L. delbrueckii The effects of three growth-promoting antibiotics (avilamycin, subsp. bulgaricus ,Lactococcus lactis subsp. cremoris ,and Streptococ- zinc bacitracin, and flavomycin) on broiler gut microbial com- cus thermophilus , only anecdotal changes were noted in microbiota munity colonization and bird performance were investigated composition, whereas significant changes were observed in the (Torok et al., 2011). OTU linked to changes in gut microbiota in expression of microbiome-encoded enzymes involved in meta- birds on antimicrobial-supplemented diets were characterized and bolic pathways, notably, carbohydrate metabolism ( McNulty et al., identified. Lachnospiraceae, L. johnsonii , Ruminococcaceae, and 2011). However, these suggestions of a relationship between probi- Oxalobacteraceae genera were less prevalent in the guts of chicks otics and obesity remain controversial ( Delzenne and Reid, 2009). fed antimicrobial-supplemented diets. L. crispatus , L. reuteri , Sub- In addition, the reports of the anti-diabetic and anti-inflammatory doligranulum , and Enterobacteriaceae were more prevalent in the effects of Lactobacilli should be considered cautiously because the guts of chicks raised on the antimicrobial diet ( Torok et al., 2011). translation of findings based on animal models to humans is haz- These results suggest that antibiotic effects on gut microbiota ardous ( Kootte et al., 2012). Finally, all these results should be may be relevant at the species level because different Lactobacillus interpreted with caution in view of the substantial funding of species-related OTUs showed paradoxical changes. obesity research by the food industry, creating conflicts of interest.

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 10 Lagier et al. Human gut microbiota

Table 1 | Modiﬁcations of gut ﬂora linked to antibiotics.

Antibiotic Method References

PENICILLINS Ampicillin Decrease in enterococci Cultivation Black et al. (1991) Decrease in streptococci Cultivation Black et al. (1991) Decrease in E. coli strains Cultivation Black et al. (1991) Slight decrease in anaerobic Gram-positive bacteria Cultivation Black et al. (1991) Amoxicillin Increase in aerobic Gram-negative rods, such as Cultivation Brismar et al. (1993), Floor et al. (1994), Stark et al. (1996) enterobacteria, other than E. coli (Klebsiella, Enterobacter) Increase in anaerobic Gram-positive rods Cultivation Swedish Study Group (1991b) Increase in Bacteroides Cultivation Swedish Study Group (1991b) Decrease in streptococci and Staphylococci Cultivation Brismar et al. (1993) Decrease in anaerobic Gram-positive cocci, such as Cultivation Brismar et al. (1993), Stark et al. (1996) eubacteria Amoxicillin/clavulanic acid Increase in enterococci and E. coli Cultivation Lode et al. (2001) Decrease in lactobacilli, clostridia, bifidobacteria Cultivation Lode et al. (2001) Disappearance of Clostridium cluster XIVa Cloning/sequencing Young and Schmidt (2004) (cloning/sequencing) Decrease in Faecalibacterium spp. Cloning/sequencing Young and Schmidt (2004) Piperacillin/tazobactam* Decrease in enterobacteria Cultivation Nord et al. (1993) Decrease in bifidobacteria, eubacteria, lactobacilli Cultivation Nord et al. (1993) Decrease in anaerobic Gram-positive cocci like clostridia Cultivation Nord et al. (1993) CEPHALOSPORINS Cefepime Decrease in E. coli and bifidobacteria Cultivation Bacher et al. (1992) Increase in clostridia and Bacteroides Cultivation Bacher et al. (1992) Ceftriaxone Decrease in the total numbers of anaerobes Cultivation Welling et al. (1991) Dramatic decrease in clostridia, lactobacilli, bifidobacteria Cultivation Vogel et al. (2001) Dramatic decrease in Gram-negative rods (enterobacteria) Cultivation Cavallaro et al. (1992), Vogel et al. (2001), Welling et al. (1991) Increase in enterococci Cultivation Vogel et al. (2001), Welling et al. (1991) Carbapenems Meropenem Decrease in enterobacteria and streptococci Cultivation Bergan et al. (1991) Increase in enterococci Cultivation Bergan et al. (1991) Decrease in clostridia, Gram-negative cocci, and bacteroides Cultivation Bergan et al. (1991) FLUOROQUINOLONES Ciprofloxacin Dramatic decrease in enterobacteria Cultivation Bergan et al. (1986), Borzio et al. (1997), Brismar et al. (1990), Brumfitt et al. (1984), Enzensberger et al. (1985), Esposito et al. (1987), Holt et al. (1986), Krueger et al. (1997), Ljungberg et al. (1990), Rozenberg-Arska et al. (1985), Van Saene et al. (1986), Wistrom et al. (1992) Decrease in aerobic Gram-positive cocci Cultivation Bergan et al. (1986), Brismar et al. (1990), Brumfitt et al. (1984), Ljungberg et al. (1990), Van Saene et al. (1986) Decrease in streptococci Cultivation Brismar et al. (1990), Brumfitt et al. (1984), Ljungberg et al. (1990) Decrease in enterococci Cultivation Bergan et al. (1986), Brismar et al. (1990), Ljungberg et al. (1990), Van Saene et al. (1986)

(Continued)

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 11 Lagier et al. Human gut microbiota

Table 1 | Continued

Antibiotic Method References

Increase in enterococci Cultivation Borzio et al. (1997) Decrease in anaerobic bacteria Cultivation Bergan et al. (1986), Brismar et al. (1990), Rozenberg-Arska et al. (1985) Suppression of Bacteroides putredinis , Ruminococcus DGGE Donskey et al. (2003) torques Norfloxacin Dramatic decrease in enterobacteria Cultivation de Vries-Hospers et al. (1985), Edlund et al. (1987), Leigh et al. (1985), Pecquet et al. (1986) Decrease in aerobic Gram-positive cocci Cultivation de Vries-Hospers et al. (1985), Pecquet et al. (1986) Decrease in streptococci Cultivation Pecquet et al. (1986) Decrease in enterococci Cultivation de Vries-Hospers et al. (1985) Ofloxacin Dramatic decrease in enterobacteria Cultivation Edlund et al. (1988), Edlund et al. (1997b), Pecquet et al. (1987) Decrease in aerobic Gram-positive cocci Cultivation Edlund et al. (1988), Edlund et al. (1997b), Pecquet et al. (1987) Decrease in enterococci Cultivation Edlund et al. (1988), Pecquet et al. (1987) Decrease in lactobacilli, bifidobacteria, eubacteria Cultivation Edlund et al. (1988) Decrease in anaerobic bacteria Cultivation Edlund et al. (1988) Decrease in Veillonella and Bacteroides spp. Cultivation Levofloxacin, Gatifloxacin,Trovafloxacin, Moxifloxacin Dramatic decrease in enterobacteria Cultivation Edlund et al. (1997b), Edlund and Nord (1999a), van Nispen et al. (1998) Strong decrease in aerobic Gram-positive cocci Cultivation Edlund et al. (1997b), Edlund and Nord (1999a), van Nispen et al. (1998) Levofloxacin, gatifloxacin: decrease in clostridia Cultivation Edlund et al. (1997b), Edlund and Nord (1999a) Gatifloxacin: decrease in fusobacteria Cultivation Edlund and Nord (1999a) GLYCOPEPTIDS Oral vancomycin Decrease in enterococci Cultivation Edlund et al. (1997a), Lund et al. (2000) Decrease in staphylococci Cultivation Van der Auwera et al. (1996) Overgrowth of lactobacilli (and pediococci) Cultivation Edlund et al. (1997a), Lund et al. (2000), Van der Auwera et al. (1996) Strong suppression or elimination of bacteroides Cultivation Edlund et al. (1997a), Lund et al. (2000) Decrease in clostridia and bifidobacteria Cultivation Lund et al. (2000) Oral teicoplanin Increase in Gram-negative aerobic rods and total numbers Cultivation Van der Auwera et al. (1996) of aerobes Increase in lactobacilli and pediococci Cultivation Van der Auwera et al. (1996) LINEZOLID Reduction of enterococci Cultivation Lode et al. (2001) Reduction of bifidobacteria, lactobacilli, clostridia, and Cultivation Lode et al. (2001) bacteroides Increase in Klebsiella Cultivation Lode et al. (2001) TETRACYCLINES Doxycycline Decrease in bifidobacteria Cultivation Saarela et al. (2007) Tigecycline Decrease in enterococci Cultivation Nord et al. (2006) Decrease in E. coli Cultivation Nord et al. (2006) Increase of other enterobacteria ( Klebsiella and Cultivation Nord et al. (2006) Enterobacter spp.)

(Continued)

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 12 Lagier et al. Human gut microbiota

Table 1 | Continued

Antibiotic Method References

Marked reduction of lactobacilli and bifidobacteria Cultivation Nord et al. (2006) Increase in yeasts Cultivation Nord et al. (2006) MACROLIDES, LINCOSAMIDES, SYNERGISTINS Erythromycin Dramatic decrease in streptococci and enterobacteria Cultivation Brismar et al. (1991) Decrease in clostridia, lactobacilli, bifidobacteria, and Cultivation Brismar et al. (1991) bacteroides Clarithromycin Reduction of enterobacteria, E. coli , and streptococci Cultivation Brismar et al. (1991), Edlund et al. (2000b) Dramatic decrease in clostridia, and bacteroides Cultivation Brismar et al. (1991) Reduction of lactobacilli and bifidobacteria Cultivation Brismar et al. (1991), Edlund et al. (2000b) Telithromycin Decrease in E. coli but overgrowth of non- E. coli Cultivation Edlund et al. (2000a) enterobacteria Reduction of lactobacilli and bifidobacteria Cultivation Edlund et al. (2000a) Clindamycin Increase in enterobacteria T-RFLP Jernberg et al. (2005b) Decrease in total anaerobic bacteria Cultivation Nord et al. (1997) Decrease in lactobacilli and Bacteroides Cultivation Nord et al. (1997), Sullivan et al. (2003) Decrease in clostridia Cultivation Nord et al. (1997) Disappearance of bifidobacteria Cultivation Jernberg et al. (2005b), Nord et al. (1997) Dramatic decrease in Bifidobacterium, Clostridium T-RFLP Jernberg et al. (2005b) (particularly C. coccoides subgroup as Eubacterium) and Bacteroides Suppression of B. vulgatus , B. acidofasciens , F.prausnitzii , DGGE Donskey et al. (2003) C. indolis , and C. leptum cluster No change in B. thetaiotaomicron and B. uniformis DGGE Donskey et al. (2003) Streptogramins: Quinupristin/dalfopristin Decrease in anaerobic Gram-negative bacteria Cultivation Scanvic-Hameg et al. (2002) Increase in enterococci and enterobacteria Cultivation Scanvic-Hameg et al. (2002) OTHERS Cotrimoxazole Suppression of Enterobacteriaceae Cultivation Mavromanolakis et al. (1997) Metronidazole No significant change but not enough data available Cultivation Sullivan et al. (2001) Nitrofurantoin No impact on intestinal microflora Cultivation Mavromanolakis et al. (1997)

REDUCTIONIST APPROACH AND BIASES genetics, and environmental conditions. The first germ-free model In their attempts to reduce ignorance and in contrast to the was used by Pasteur in 1885. Since that time, various models have holistic approach based on the combination of conventional tech- been used to study gut microbiota, including germ-free neona- niques and technology-driven methods, which enable researchers tal pigs ( Meurens et al., 2007), zebrafish ( Rawls et al., 2004), and to study and make sense of a complex ecosystem, diverse stud- gnotobiotic mice, which is the most effective tool ( Backhed et al., ies based on experimental models have induced reductionism 2007; Goodman et al., 2011). in the understanding of human microbiota and have generated Conversely, based on observations and not supported by any contradictory results ( Raoult, 2010; Fang and Casadevall, 2011). preconceived hypothesis, several findings by different research teams have shown a significant reduction in Bacteroidete s propor- HYPOTHESIS-DRIVEN RESEARCH VERSUS HOLISTIC-DRIVEN tions in obese patients ( Armougom et al., 2009; Turnbaugh et al., RESEARCH 2009; Million et al., 2011). Comparing the composition of the Because it has been suggested that gut microbiota play a role in gut microbiota between young adult female monozygotic or dizy- health and disease, it has been attractive to find a stable model gotic twins who are obese or lean and their mothers, Turnbaugh to help scientists to understand host-gut microbiota mutualism, found that obesity was associated with reduced bacterial diversity but this relationship is very complex and involves control diets, and, notably, a reduced proportion of Bacteroidetes . Moreover, the

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 13 Lagier et al. Human gut microbiota

genes over-represented within obese individuals exclusively belong 67% were supported by the food industry. Moreover, the results to the Firmicutes phylum ( Turnbaugh et al., 2009). Nevertheless, of industry-supported trials were significantly associated with a previous investigations based on genomic data and experimental higher quality of reporting score associated with long-term WL. models have shown that co-colonization of B. thetaiotaomicron Moreover, compounding this problem, some scientists do not and M. smithii in the digestive tract of xenobiotic mice was declare their conflicts of interest. Based on these data, and the responsible for significant weight gain, with discordance between considerable financial involvement associated with human gut sequence analysis results and the initial hypothesis ( Xu et al., 2003; microbiota research, notably in obesity, we regret that there is not Samuel and Gordon, 2006). more public funding ( Smith, 2005), and that conflict of interest with food industry are not actively required as for pharmaceutical CONFLICTS OF INTEREST industry. Finally, to exhaustively review the topic of human gut microbiota composition and mutualism with the host, it would be morally CONCLUDING REMARKS objectionable not to address the central influence of funding Factors affecting the composition of the gut microbiota and the sources and transparency disclosures ( Million and Raoult, 2012). relationship with the hosts are of considerable complexity. Both A transparency declaration of conflict of interest is important physiological and external factors are often unstable over time, for publication in the medical literature. Lundh et al. (2010), based influencing the gut microbiota. Despite the contribution of recent on the articles published in six of the most prestigious medical technologies, the repertoire of this ecosystem remains incomplete. journals, showed that the publication of studies financed by indus- As a striking example, despite the dramatic increase in the num- tries was associated with an increase in the impact factor of the ber of publications regarding gut microbiota, simplistic anomalies journal. Regarding the economic aspect and the payment of physi- persist, such as the discordance among microscopic observation, cians by five manufacturers of hip and knee prostheses, a recent pyrosequencing, and culture results. We regret that fewer studies study confirmed that only approximately 80% of direct payments are based on observation and description in opposition to studies and 50% of indirect payments to physicians have been disclosed performed to confirm a hypothesis. Indeed, it is paradoxical to (Okike et al., 2009). Some authors even consider that publications design experiments and models to confirm a hypothesis because in medical journals are a marketing tool for the pharmaceutical the ecosystem is only partially described. Finally, the central prob- industry ( Smith, 2005). In the beverage and food industry, Levine lem of funding sources and transparency declarations lead us to studied the financial relationships between industry and authors hope that public funding will develop food-industry-independent who have published research on alimentary substitutes. Classifying research to increase confidence in the results. these publications as neutral, critical, or supportive toward the ali- In the future, we think that culturomics followed by the high- mentary substitutes,the authors suggested a significant association throughput genome sequencing and its applications as the explo- between the authors who support the efficiency of the substitute ration of host-pathogen interactions will allows to capture the rela- and the authors with financial relationships with the industrial tionships in the gut microbiota. In addition, technology advances company ( Levine et al., 2003). in pyrosequencing with higher reads fragment analysis, may facil- Finally, Thomas et al. (2008) has recently shown that of itate the analysis to low taxonomic level (genera, species) reducing 63 randomized trials published regarding nutrition and obesity, consequently the depth bias.

REFERENCES Angelakis, E., Bastelica, D., Ben, A. A., El human gut microbiome. Nature 473, in germ-free mice. Proc. Natl. Acad. Agans, R., Rigsbee, L., Kenche, H., Filali,A., Dutour,A., Mege, J. L., et al. 174–180. Sci. U.S.A. 104, 979–984. Michail, S., Khamis, H. J., and Paliy, (2012b). An evaluation of the effects Bacher, K., Schaeffer, M., Lode, H., Balamurugan, R., George, G., Kabeer- O. (2011). Distal gut microbiota of of Lactobacillus ingluviei on body Nord, C. E., Borner, K., and doss, J., Hepsiba, J., Chandragu- adolescent children is different from weight, the intestinal microbiome Koeppe, P. (1992). Multiple dose nasekaran, A. M., and Ramakr- that of adults. FEMS Microbiol. Ecol. and metabolism in mice. Microb. pharmacokinetics, safety, and ishna, B. S. (2010). Quantita- 77, 404–412. Pathog. 52, 61–68. effects on faecal microﬂora, of tive differences in intestinal Fae- Anadon, A., Martinez-Larranaga, M. Angelakis, E., and Raoult, D. (2010). cefepime in healthy volunteers. calibacterium prausnitzii in obese R., and Aranzazu, M. M. (2006). The increase of Lactobacillus species J. Antimicrob. Chemother. 30, Indian children. Br. J. Nutr. 103, Probiotics for animal nutrition in in the gut ﬂora of newborn broiler 365–375. 335–338. the European Union. Regulation chicks and ducks is associated with Backhed, F.,Ding, H., Wang, T.,Hooper, Barna, J. C., and Williams, D. H. (1984). and safety assessment. Regul. Toxicol. weight gain. PLoS ONE 5, e10463. L. V., Koh, G. Y., Nagy, A., et al. The structure and mode of action of Pharmacol. 45, 91–95. doi:10.1371/journal.pone.0010463 (2004). The gut microbiota as an glycopeptide antibiotics of the van- Andersson, A. F., Lindberg, M., Jakob- Armougom, F., Henry, M., Vialettes, environmental factor that regulates comycin group. Annu. Rev. Micro- sson, H., Backhed, F., Nyren, P., B., Raccah, D., and Raoult, D. fat storage. Proc. Natl. Acad. Sci. biol. 38, 339–357. and Engstrand, L. (2008). Com- (2009). Monitoring bacterial com- U.S.A. 101, 15718–15723. Bartosch, S., Fite, A., Macfarlane, G. parative analysis of human gut munity of human gut microbiota Backhed, F., Ley, R. E., Sonnenburg, J. T., and McMurdo, M. E. (2004). microbiota by barcoded pyrose- reveals an increase in Lactobacillus L., Peterson, D. A., and Gordon, J. Characterization of bacterial com- quencing. PLoS ONE 3, e2836. in obese patients and Methanogens I. (2005). Host-bacterial mutualism munities in feces from healthy doi:10.1371/journal.pone.0002836 in anorexic patients. PLoS ONE in the human intestine. Science 307, elderly volunteers and hospital- Angelakis, E., Armougom, F., Million, 4, e7125. doi:10.1371/journal.pone. 1915–1920. ized elderly patients by using real- M., and Raoult, D. (2012a). The 0007125 Backhed, F., Manchester, J. K., time PCR and effects of antibi- relationship between gut microbiota Arumugam, M., Raes, J., Pelletier, E., Semenkovich, C. F.,and Gordon, J. I. otic treatment on the fecal micro- and weight gain in humans. Future Le, P. D., Yamada, T., Mende, D. (2007). Mechanisms underlying the biota. Appl. Environ. Microbiol. 70, Microbiol. 7, 91–109. R., et al. (2011). Enterotypes of the resistance to diet-induced obesity 3575–3581.

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 14 Lagier et al. Human gut microbiota

Bastard, J. P., Maachi, M., Lagathu, C., in patients undergoing colorec- Eyken, P., et al. (2006). Validation of (HMR 3647) and clarithromycin Kim, M. J., Caron, M.,Vidal, H., et al. tal surgery. Antimicrob. Agents 16S rDNA sequencing in microdis- on the oropharyngeal and intesti- (2006). Recent advances in the rela- Chemother. 34, 481–483. sected bowel biopsies from Crohn’s nal microflora. J. Antimicrob. tionship between obesity, inflam- Brismar, B., Edlund, C., and Nord, C. E. disease patients to assess bacter- Chemother. 46, 741–749. mation, and insulin resistance. Eur. (1991). Comparative effects of clar- ial flora diversity. J. Pathol. 209, Edlund, C., Beyer, G., Hiemer-Bau, M., Cytokine Netw. 17, 4–12. ithromycin and erythromycin on the 532–539. Ziege, S., Lode, H., and Nord, C. Bergan, T., Delin, C., Johansen, S., normal intestinal microflora. Scand. de Vries-Hospers, H. G., Welling, G. W., E. (2000b). Comparative effects of Kolstad, I. M., Nord, C. E., J. Infect. Dis. 23, 635–642. and van der Waaij, D. (1985). Nor- moxifloxacin and clarithromycin on and Thorsteinsson, S. B. (1986). Brismar, B., Edlund, C., and Nord, C. floxacin for selective decontamina- the normal intestinal microflora. Pharmacokinetics of ciprofloxacin E. (1993). Impact of cefpodoxime tion: a study in human volunteers. Scand. J. Infect. Dis. 32, 81–85. and effect of repeated dosage proxetil and amoxicillin on the nor- Prog. Clin. Biol. Res. 181, 259–262. Edlund, C., Barkholt, L., Olsson- on salivary and fecal microflora. mal oral and intestinal microflora. Delzenne, N., and Reid, G. (2009). No Liljequist, B., and Nord, C. E. Antimicrob. Agents Chemother. 29, Eur. J. Clin. Microbiol. Infect. Dis. 12, causal link between obesity and pro- (1997a). Effect of vancomycin on 298–302. 714–719. biotics. Nat. Rev. Microbiol. 7, 901. intestinal flora of patients who pre- Bergan, T., Nord, C. E., and Thorsteins- Brumfitt, W., Franklin, I., Grady, D., Dethlefsen, L., Huse, S., Sogin, M. viously received antimicrobial ther- son, S. B. (1991). Effect of Hamilton-Miller, J. M., and Iliffe, A. L., and Relman, D. A. (2008). apy. Clin. Infect. Dis. 25, 729–732. meropenem on the intestinal (1984). Changes in the pharmaco- The pervasive effects of an antibi- Edlund, C., Sjostedt, S., and Nord, microflora. Eur. J. Clin. Microbiol. kinetics of ciprofloxacin and fecal otic on the human gut micro- C. E. (1997b). Comparative effects Infect. Dis. 10, 524–527. flora during administration of a 7- biota, as revealed by deep 16S of levofloxacin and ofloxacin on Biesbroek, G., Sanders, E. A., Roesel- day course to human volunteers. rRNA sequencing. PLoS Biol. 6,e280. the normal oral and intestinal ers, G., Wang, X., Caspers, M. P., Antimicrob. Agents Chemother. 26, doi:10.1371/journal.pbio.0060280 microflora. Scand. J. Infect. Dis. 29, Trzcinski, K., et al. (2012). Deep 757–761. Donskey, C. J., Hujer, A. M., Das, S. M., 383–386. sequencing analyses of low density Cani, P. D., Bibiloni, R., Knauf, Pultz, N. J., Bonomo, R. A., and Rice, Edlund, C., Bergan, T., Josefsson, microbial communities: working at C., Waget, A., Neyrinck, A. M., L. B. (2003). Use of denaturing gra- K., Solberg, R., and Nord, C. E. the boundary of accurate microbiota Delzenne, N. M., et al. (2008). dient gel electrophoresis for analysis (1987). Effect of norfloxacin on detection. PLoS ONE 7, e32942. Changes in gut microbiota control of the stool microbiota of hospital- human oropharyngeal and colonic doi:10.1371/journal.pone.0032942 metabolic endotoxemia-induced ized patients. J. Microbiol. Methods microflora and multiple-dose phar- Black, F., Einarsson, K., Lidbeck, A., inflammation in high-fat diet- 54, 249–256. macokinetics. Scand. J. Infect. Dis. Orrhage, K., and Nord, C. E. induced obesity and diabetes in Drasar, B. S., Crowther, J. S., Goddard, 19, 113–121. (1991). Effect of lactic acid produc- mice. Diabetes 57, 1470–1481. P.,Hawksworth,G.,Hill,M. J.,Peach, Edlund, C., Kager, L., Malmborg, A. ing bacteria on the human intesti- Cavallaro, V., Catania, V., Bonaccorso, S.,et al. (1973). The relation between S., Sjostedt, S., and Nord, C. E. nal microflora during ampicillin R., Mazzone, S., Speciale, A., Di diet and the gut microflora in man. (1988). Effect of ofloxacin on oral treatment. Scand. J. Infect. Dis. 23, Marco, R., et al. (1992). Effect of Proc. Nutr. Soc. 32, 49–52. and gastrointestinal microflora in 247–254. a broad-spectrum cephalosporin on Dridi, B., Fardeau, M. L., Ollivier, patients undergoing gastric surgery. Bollmann, A., Lewis, K., and Epstein, the oral and intestinal microflora B., Raoult, D., and Drancourt, Eur. J. Clin. Microbiol. Infect. Dis. 7, S. S. (2007). Incubation of environ- in patients undergoing colorectal M. (2012a). Methanomassiliicoccus 135–143. mental samples in a diffusion cham- surgery. J. Chemother. 4, 82–87. luminyensis , gen. nov., sp. nov., a Edlund, C., and Nord, C. E. (1999a). ber increases the diversity of recov- Claesson, M. J., Cusack, S., O’Sullivan, novel methanogenic Archaea iso- Ecological effect of gatifloxacin ered isolates. Appl. Environ. Micro- O., Greene-Diniz, R., de Weerd, H., lated from human feces. Int. J. Syst. on the normal human intestinal biol. 73, 6386–6390. and Flannery, E. (2011). Composi- Evol. Microbiol. 62, 1902–1907. microflora. J. Chemother. 11, 50–53. Borzio, M., Salerno, F., Saudelli, M., tion, variability, and temporal stabil- Dridi, B., Henry, M., Richet, H., Edlund, C., and Nord, C. E. (1999b). Galvagno, D., Piantoni, L., and Fra- ity of the intestinal microbiota of the Raoult, D., and Drancourt, M. Effect of quinolones on intestinal giacomo, L. (1997). Efficacy of oral elderly. Proc. Natl. Acad. Sci. U.S.A. (2012b). Age-related prevalence of ecology. Drugs 58(Suppl. 2), 65–70. ciprofloxacin as selective intestinal 108(Suppl. 1), 4586–4591. Methanomassiliicoccus luminyensis Enzensberger, R., Shah, P. M., and decontaminant in cirrhosis. Ital. J. Claesson,M. J.,Wang,Q.,O’Sullivan,O., in the human gut microbiome. Knothe, H. (1985). Impact of oral Gastroenterol. Hepatol. 29, 262–266. Greene-Diniz, R., Cole, J. R., Ross, R. APMIS 120, 773–777. ciprofloxacin on the faecal flora Bousbia, S., Papazian, L., Saux, P., P., et al. (2010). Comparison of two Dridi, B., Henry, M., El Khéchine, of healthy volunteers. Infection 13, Forel, J. M., Auffray, J. P., Mar- next-generation sequencing tech- A., Raoult, D., and Drancourt, 273–275. tin, C., et al. (2012). Repertoire nologies for resolving highly com- M. (2009). High prevalence of Epstein, S. S. (2009). Microbial awaken- of intensive care unit pneumonia plex microbiota composition using Methanobrevibacter smithii and ings. Nature 457, 1083. microbiota. PLoS ONE 7, e32486. tandem variable 16S rRNA gene Methanosphaera stadtmanae Esposito, S., Barba, D., Galante, doi:10.1371/journal.pone.0032486 regions. Nucleic Acids Res. 38, e200. detected in the human gut using D., Gaeta, G. B., and Laghezza, Boyer, M.,Yutin, N., Pagnier, I., Barrassi, Collado, M. C., Isolauri, E., Laitinen, an improved DNA detection O. (1987). Intestinal microflora L., Fournous, G., Espinosa, L., et al. K., and Salminen, S. (2008). Dis- protocol. PLoS ONE 4, e7063. changes induced by ciprofloxacin (2009). Giant Marseillevirus high- tinct composition of gut microbiota doi:10.1371/journal.pone.0007063 and treatment of portal-systemic lights the role of amoebae as a melt- during pregnancy in overweight and Dridi, B., Raoult, D., and Drancourt, M. encephalopathy (PSE). Drugs Exp. ing pot in emergence of chimeric normal-weight women. Am. J. Clin. (2011). Archaea as emerging organ- Clin. Res. 13, 641–646. microorganisms. Proc. Natl. Acad. Nutr. 88, 894–899. isms in complex human microbio- Fang, F. C., and Casadevall, A. (2011). Sci. U.S.A. 106, 21848–21853. De Filippo, C., Cavalieri, D., Di Paola, mes. Anaerobe 17, 56–63. Reductionistic and holistic science. Breitbart, M., Haynes, M., Kelley, S., M., Ramazzotti, M., Poullet, J. B., Eckburg, P. B., Bik, E. M., Bernstein, C. Infect. Immun. 79, 1401–1404. Angly, F., Edwards, R. A., Felts, B., Massart, S., et al. (2010). Impact N., Purdom, E., Dethlefsen, L., Sar- Finegold, S. M., Attebery, H. R., and et al. (2008). Viral diversity and of diet in shaping gut microbiota gent, M., et al. (2005). Diversity of Sutter, V. L. (1974). Effect of diet dynamics in an infant gut. Res. revealed by a comparative study the human intestinal microbial flora. on human fecal flora: comparison of Microbiol. 159, 367–373. in children from Europe and rural Science 308, 1635–1638. Japanese and American diets. Am. J. Brismar, B., Edlund, C., Malmborg, Africa. Proc. Natl. Acad. Sci. U.S.A. Edlund, C., Alvan, G., Barkholt, L., Clin. Nutr. 27, 1456–1469. A. S., and Nord, C. E. (1990). 107, 14691–14696. Vacheron, F., and Nord, C. E. Finegold, S. M., Sutter,V.L., Sugihara, P. Ciprofloxacin concentrations and De Hertogh, G., Aerssens, J., De Hoogt, (2000a). Pharmacokinetics and T., Elder, H. A., Lehmann, S. M., and impact of the colon microflora R., Peeters, P., Verhasselt, P., Van comparative effects of telithromycin Phillips,R. L. (1977). Fecal microbial

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 15 Lagier et al. Human gut microbiota

flora in seventh day adventist popu- Hayashi, H., Sakamoto, M., and Benno, overweight. Am. J. Clin. Nutr. 87, sequence and description of Anae- lations and control subjects. Am. J. Y. (2002a). Fecal microbial diver- 534–538. rococcus senegalensis sp. nov. Stand. Clin. Nutr. 30, 1781–1792. sity in a strict vegetarian as deter- Kassinen, A., Krogius-Kurikka, L., Genomic Sci. 6, 116–125. Floor, M., van Akkeren, F., Rozenberg- mined by molecular analysis and Makivuokko, H., Rinttila, T., Paulin, Larsen, N., Vogensen, F. K., van den Arska, M., Visser, M., Kolsters, A., cultivation. Microbiol. Immunol. 46, L., and Corander, J. (2007). The fecal Berg,F.W.,Nielsen,D. S.,Andreasen, Beumer, H., et al. (1994). Effect 819–831. microbiota of irritable bowel syn- A. S., and Pedersen, B. K. (2010). Gut of loracarbef and amoxicillin on Hayashi, H., Sakamoto, M., and Benno, drome patients differs significantly microbiota in human adults with the oropharyngeal and intestinal Y. (2002b). Phylogenetic analy- from that of healthy subjects. Gas- type 2 diabetes differs from non- microflora of patients with bron- sis of the human gut microbiota troenterology 133, 24–33. diabetic adults. PLoS ONE 5, e9085. chitis. Scand. J. Infect. Dis. 26, using 16S rDNA clone libraries Kerac, M., Bunn, J., Seal, A., Thindwa, doi:10.1371/journal.pone.0009085 191–197. and strictly anaerobic culture-based M., Tomkins, A., and Sadler, K. Lay, C., Rigottier-Gois, L., Holmstrom, Fogarty, A. W., Glancy, C., Jones, S., methods. Microbiol. Immunol. 46, (2009). Probiotics and prebiotics K., Rajilic, M.,Vaughan, E. E., and de Lewis, S. A., McKeever, T. M., and 535–548. for severe acute malnutrition Vos, W. M. (2005). Colonic micro- Britton, J. R. (2008). A prospective Hildebrandt, M. A., Hoffmann, C., (PRONUT study): a double-blind biota signatures across five northern study of weight change and systemic Sherrill-Mix, S. A., Keilbaugh, S. A., efficacy randomised controlled trial European countries. Appl. Environ. inflammation over 9 y. Am. J. Clin. Hamady, M., and Chen, Y. Y. (2009). in Malawi. Lancet 374, 136–144. Microbiol. 71, 4153–4155. Nutr. 87, 30–35. High-fat diet determines the compo- Koenig, J. E., Spor, A., Scalfone, N., Lee,S.,Sung,J.,Lee,J.,and Ko,G. (2011). Frank, D. N., St Amand, A. L., sition of the murine gut microbiome Fricker, A. D., Stombaugh, J., and Comparison of the gut microbiotas Feldman, R. A., Boedeker, E. C., independently of obesity. Gastroen- Knight, R. (2011). Succession of of healthy adult twins living in South Harpaz, N., and Pace, N. R. (2007). terology 137, 1716–1724. microbial consortia in the devel- Korea and the United States. Appl. Molecular-phylogenetic characteri- Hill, M. J., Goddard, P., and Williams, oping infant gut microbiome. Proc. Environ. Microbiol. 77, 7433–7437. zation of microbial community R. E. (1971). Gut bacteria and aeti- Natl. Acad. Sci. U.S.A. 108(Suppl. 1), Leigh, D. A., Emmanuel, F. X. S., and imbalances in human inflammatory ology of cancer of the breast. Lancet 4578–4585. Tighe, C. (1985). “Pharmacokinetic bowel diseases. Proc. Natl. Acad. Sci. 2, 472–473. Kokcha, S., Mishra, A. K., Lagier, J. C., studies of norfloxacin in healthy vol- U.S.A. 104, 13780–13785. Holt, H. A., Lewis, D. A., White, L. O., Million, M., Leroy, Q., Raoult, D., et unteers, and effect on the fecal flora,” Fuerst, J. A., and Sagulenko, E. (2011). Bastable, S. Y., and Reeves, D. S. al. (2012). Non contiguous finished in Proceedings of the 14th Inter- Beyond the bacterium: plancto- (1986). Effect of oral ciprofloxacin genome sequence and description of national Congress of Chemotherapy, mycetes challenge our concepts of on the faecal flora of healthy Bacillus timonensis sp. nov. Stand. 1835–1836, Kyoto. microbial structure and function. volunteers. Eur. J. Clin. Microbiol. 5, Genomic Sci. 6, 346–355. Levine, J., Gussow, J. D., Hastings, Nat. Rev. Microbiol. 9, 403–413. 201–205. Kootte, R. S., Vrieze, A., Holleman, F., D., and Eccher, A. (2003) Authors’ Galvez, A., Maqueda, M., Martinez- Hong, S., Bunge, J., Leslin, C., Jeon, S., Dallinga-Thie, G. M., Zoetendal, E. financial relationships with the food Bueno, M., and Valdivia, E. (1998). and Epstein, S. S. (2009). Polymerase G., and de Vos, W. M. (2012). The and beverage industry and their Publication rates reveal trends in chain reaction primers miss half of therapeutic potential of manipulat- published positions on the fat sub- microbiological research. ASM News rRNA microbial diversity. ISME J. 3, ing gut microbiota in obesity and stitute olestra. Am. J. Public Health 64, 269–275. 1365–1373. type 2 diabetes mellitus. Diabetes 93, 664–669. Goodman, A. L., Kallstrom, G., Faith, Hotamisligil, G. S. (2006). Inflamma- Obes. Metab. 14, 112–120. Ley, R. E., Backhed, F., Turnbaugh, P., J. J., Reyes, A., Moore, A., Dan- tion and metabolic disorders. Nature Kovatcheva-Datchary, P., Zoetendal, E. Lozupone, C. A., Knight, R. D., and tas, G., et al. (2011). Extensive per- 444, 860–867. G., Venema, K., de Vos, W. M., and Gordon, J. I. (2005). Obesity alters sonal human gut microbiota cul- Hugenholtz, P. (2002). Explor- Smidt, H. (2009). Tools for the tract: gut microbial ecology. Proc. Natl. ture collections characterized and ing prokaryotic diversity in the understanding the functionality of Acad. Sci. U.S.A. 102, 11070–11075. manipulated in gnotobiotic mice. genomic era. Genome Biol. 3, the gastrointestinal tract. Therap. Ley,R. E.,Peterson,D. A.,and Gordon,J. Proc. Natl. Acad. Sci. U.S.A. 108, reviews0003.1–reviews0003.8. Adv. Gastroenterol. 2, 9–22. I. (2006a). Ecological and evolution- 6252–6257. Jernberg, C., Sullivan, A., Edlund, C., Krueger, W. A., Ruckdeschel, G., and ary forces shaping microbial diver- Gorlas, A., Robert, C., Gimenez, and Jansson, J. K. (2005a). Mon- Unertl, K. (1997). Influence of intra- sity in the human intestine. Cell 124, G., Drancourt, M., and Raoult, itoring of antibiotic-induced alter- venously administered ciprofloxacin 837–848. D. (2012). Complete genome ations in the human intestinal on aerobic intestinal microflora and Ley, R. E., Turnbaugh, P. J., Klein, S., sequence of Methanomassiliicoccus microflora and detection of probi- fecal drug levels when adminis- and Gordon, J. I. (2006b). Micro- luminyensis, the largest genome of a otic strains by use of terminal restric- tered simultaneously with sucralfate. bial ecology: human gut microbes human-associated Archaea species. tion fragment length polymor- Antimicrob. Agents Chemother. 41, associated with obesity. Nature 444, J. Bacteriol. 194, 4745. phism. Appl. Environ. Microbiol. 71, 1725–1730. 1022–1023. Gossling, J., and Slack, J. M. (1974). 501–506. Lagier, J. C., Armougom, F., Million, Liszt, K., Zwielehner, J., Handschur, Predominant gram-positive bacteria Jernberg, C., Sullivan, A., Edlund, C., M., Hugon, P., Pagnier, I., Robert, M., Hippe, B., Thaler, R., and in human feces: numbers, variety, and Jansson, J. K. (2005b). Monitor- C., et al. (2012a). Microbial cul- Haslberger, A. G. (2009). Charac- and persistence. Infect. Immun. 9, ing of antibiotic-induced alterations turomics: a paradigm shift in the terization of bacteria, clostridia and 719–729. in the human intestinal microflora human gut microbiome study. Clin. Bacteroides in faeces of vegetarians Grzeskowiak, L., Collado, M. C., Man- and detection of probiotic strains Microbiol. Infect. PMID: 23033984. using qPCR and PCR-DGGE fin- gani, C., Maleta, K., Laitinen, K., by use of terminal restriction frag- [Epub ahead of print]. gerprinting. Ann. Nutr. Metab. 54, and Ashorn, P. (2012). Distinct gut ment length polymorphism. Appl. Lagier, J. C., Armougom, F., Mishra, A. 253–257. microbiota in south eastern African Environ. Microbiol. 71, 501–506. K., Nguyen, T. T., Raoult, D., and Ljungberg, B., Nilsson-Ehle, I., Edlund, and northern European infants. Kaeberlein, T., Lewis, K., and Epstein, Fournier, P. E. (2012b). Non con- C., and Nord, C. E. (1990). Influ- J. Pediatr. Gastroenterol. Nutr. 54, S. S. (2002). Isolating “uncultivable” tiguous finished genome sequence ence of ciprofloxacin on the colonic 812–816. microorganisms in pure culture in and description of Alistipes timonen- microflora in young and elderly vol- Hamad, I., Sokhna, C., Raoult, D., a simulated natural environment. sis sp. nov. Stand. Genomic Sci. 6, unteers: no impact of the altered and Bittar, F. (2012). Molecu- Science 296, 1127–1129. 315–324. drug absorption. Scand. J. Infect. Dis. lar detection of eukaryotes in a Kalliomaki, M., Collado, M. C., Salmi- Lagier, J. C., El Karkouri, K., Nguyen, 22, 205–208. single human stool sample from nen, S., and Isolauri, E. (2008). Early T. T., Armougom, F., Raoult, Lode, H., Von der, H. N., Ziege, S., Senegal. PLos ONE 7, e40888. differences in fecal microbiota com- D., and Fournier, P. E. (2012c). Borner, K., and Nord, C. E. (2001). doi:10.1371/journal.pone.0040888 position in children may predict Non-contiguous finished genome Ecological effects of linezolid versus

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 16 Lagier et al. Human gut microbiota

amoxicillin/clavulanic acid on the of a consortium of fermented milk of Paenibacillus senegalensis sp. nov. physicians. N. Engl. J. Med. 361, normal intestinal microflora. Scand. strains on the gut microbiome of Stand. Genomic Sci. 7. [Epub ahead 1466–1474. J. Infect. Dis. 33, 899–903. gnotobiotic mice and monozygotic of print] Pagnier, I., Raoult, D., and La Scola, Looft, T., Johnson, T. A., Allen, H. K., twins. Sci. Transl. Med. 3, 106ra106. Monira, S., Nakamura, S., Gotoh, B. (2008). Isolation and identifi- Bayles, D. O., Alt, D. P., Stedtfeld, R. Meurens, F., Berri, M., Siggers, K., Izutsu, K., Watanabe, H., cation of amoeba-resisting bacte- D. (2012). In-feed antibiotic effects R. H., Willing, B. P., Salmon, and Alam, N. H. (2011). ria from water in human environ- on the swine intestinal microbiome. H., Van Kessel, A. G., et al. Gut microbiota of healthy ment by using an Acanthamoeba Proc. Natl. Acad. Sci. U.S.A. 109, (2007). Commensal bacteria and and malnourished children in polyphaga co-culture procedure. 1691–1696. expression of two major intesti- Bangladesh. Front. Microbiol. 2:228. Environ. Microbiol. 10, 1135–1144. Lund, B., Edlund, C., Barkholt, L., Nord, nal chemokines, TECK/CCL25 doi:10.3389/fmicb.2011.00228 Palmer, C., Bik, E. M., Di Giulio, D. C. E., Tvede, M., and Poulsen, R. L. and MEC/CCL28, and their Moore, W. E., and Holdeman, L. V. B., Relman, D. A., and Brown, (2000). Impact on human intesti- receptors. PLoS ONE 2, e677. (1974a). Human fecal flora: the nor- P. O. (2007). Development of nal microflora of an Enterococcus doi:10.1371/journal.pone.0000677 mal flora of 20 Japanese-Hawaiians. the human infant intestinal faecium probiotic and vancomycin. Miller, T. L., Wolin, M. J., Conway Appl. Microbiol. 27, 961–979. microbiota. PLoS Biol. 5, e177. Scand. J. Infect. Dis. 32, 627–632. de Macario, E., and Macario, A. J. Moore, W. E., and Holdeman, L. V. doi:10.1371/journal.pbio.0050177 Lundh, A., Barbateskovic, M., Hrob- (1982). Isolation of Methanobre- (1974b). Special problems associ- Parfrey, L. W., Walters, W. A., and jartsson, A., and Gotzsche, P. vibacter smithii from human ated with the isolation and identi- Knight, R. (2011). Microbial eukary- C. (2010). Conflicts of interest feces. Appl. Environ. Microbiol. 43, fication of intestinal bacteria in fecal otes in the human microbiome: at medical journals: the influ- 227–232. flora studies. Am. J. Clin. Nutr. 27, ecology, evolution, and future ence of industry-supported ran- Million, M., Angelakis, E., Paul, M., 1450–1455. directions. Front. Microbiol. 2:153. domised trials on journal impact Armougom, F., Leibovici, L., and Mueller, S., Saunier, K., Hanisch, C., doi:10.3389/fmicb.2011.00153 factors and revenue – cohort Raoult, D. (2012). Comparative Norin, E., Alm, L., and Midtvedt, T. Pecquet, S., Andremont, A., and Tan- study. PLoS Med. 7, e1000354. meta-analysis of the effect of Lac- (2006). Differences in fecal micro- crede, C. (1986). Selective antimi- doi:10.1371/journal.pmed.1000354 tobacillus species on weight gain biota in different European study crobial modulation of the intesti- Lysholm, F., Wetterbom, A., Lin- in humans and animals. Microb. populations in relation to age, gen- nal tract by norfloxacin in human dau, C., Darban, H., Bjerkner, Pathog. 53, 100–108. der, and country: a cross-sectional volunteers and in gnotobiotic mice A., and Fahlander, K. (2012). Million, M., Maraninchi, M., Henry, M., study. Appl. Environ. Microbiol. 72, associated with a human fecal flora. Characterization of the viral Armougom, F., Richet, H., and Car- 1027–1033. Antimicrob. Agents Chemother. 29, microbiome in patients with rieri, P. (2011). Obesity-associated Murphy, E. F., Cotter, P. D., Hogan, 1047–1052. severe lower respiratory tract gut microbiota is enriched in A., O’Sullivan, O., Joyce, A., Fouhy, Pecquet, S., Andremont, A., and infections, using metagenomic Lactobacillus reuteri and depleted F., et al. (2012). Divergent meta- Tancrede, C. (1987). Effect of sequencing. PLoS ONE 7: e30875. in Bifidobacterium animalis and bolic outcomes arising from targeted oral ofloxacin on fecal bacteria doi:10.1371/journal.pone.0030875 Methanobrevibacter smithii. Int. J. manipulation of the gut micro- in human volunteers. Antimicrob. Maccaferri, S., Vitali, B., Klinder, A., Obes. (Lond.) 36, 817–825. biota in diet-induced obesity. Gut . Agents Chemother. 31, 124–125. Kolida, S., Ndagijimana, M., and Million, M., and Raoult, D. (2012). Pub- PMID:22345653. [Epub ahead of Pennisi, E. (2011). Microbiology. Girth Laghi, L. (2010). Rifaximin mod- lication biases in probiotics. Eur. J. print]. and the gut (bacteria). Science 332, ulates the colonic microbiota of Epidemiol. PMID: 23086285. [Epub Nord, C. E. (1995). Effect of quinolones 32–33. patients with Crohn’s disease: an ahead of print]. on the human intestinal microflora. Rajilic-Stojanovic, M., Smidt, H., and de in vitro approach using a contin- Minot, S., Sinha, R., Chen, J., Li, H., Keil- Drugs 49(Suppl. 2), 81–85. Vos, W. M. (2007). Diversity of the uous culture colonic model sys- baugh, S. A., and Wu, G. D. (2011). Nord, C. E., Brismar, B., Kasholm- human gastrointestinal tract micro- tem. J. Antimicrob. Chemother. 65, The human gut virome: inter- Tengve, B., and Tunevall, G. (1993). biota revisited. Environ. Microbiol. 9, 2556–2565. individual variation and dynamic Effect of piperacillin/tazobactam 2125–2136. Manichanh, C., Rigottier-Gois, L., Bon- response to diet. Genome Res. 21, treatment on human bowel Raoult, D. (2008). Obesity pandemics naud, E., Gloux, K., Pelletier, E., and 1616–1625. microflora. J. Antimicrob. and the modification of digestive Frangeul, L. (2006). Reduced diver- Mishra, A. K., Gimenez, G., Lagier, J. C., Chemother. 31(Suppl. A), 61–65. bacterial flora. Eur. J. Clin. Microbiol. sity of faecal microbiota in Crohn’s Robert, C., Raoult, D., and Fournier, Nord, C. E., Lidbeck, A., Orrhage, Infect. Dis. 27, 631–634. disease revealed by a metagenomic P. E. (2012a). Non contiguous fin- K., and Sjostedt, S. (1997). Oral Raoult, D. (2010). Technology-driven approach. Gut 55, 205–211. ished genome sequence and descrip- supplementation with lactic acid- research will dominate hypothesis- Marchesi, J. R. (2010). Prokaryotic and tion of Alistipes senegalensis sp. nov. producing bacteria during intake of driven research: the future of eukaryotic diversity of the human Stand. Genomic Sci. 6, 304–314. clindamycin. Clin. Microbiol. Infect. microbiology. Future Microbiol. 5, gut. Adv. Appl. Microbiol. 72, 43–62. Mishra, A. K., Lagier, J. C., Robert, 3, 124–132. 135–137. Marchesi, J. R. (2011). Human distal gut C., Raoult, D., and Fournier, P. Nord, C. E., Sillerstrom, E., and Raoult, D., Fenollar, F., Rolain, J. M., microbiome. Environ. Microbiol. 13, E. (2012b). Non contiguous fin- Wahlund, E. (2006). Effect of Minodier, P., Bosdure, E., and Li, W. 3088–3102. ished genome sequence and descrip- tigecycline on normal oropha- (2010). Tropheryma whipplei in chil- Mata, L. J., Carrillo, C., and Villatoro, tion of Clostridium senegalense ryngeal and intestinal microflora. dren with gastroenteritis. Emerging E. (1969). Fecal microflora in health sp. nov. Stand. Genomic Sci. 6, Antimicrob. Agents Chemother. 50, Infect. Dis. 16, 776–782. persons in a preindustrial region. 386–395. 3375–3380. Raoult, D., La Scola, B., and Birtles, R. Appl. Microbiol. 17, 596–602. Mishra, A. K., Lagier, J. C., Robert, Nottingham, P. M., and Hungate, R. E. (2007). The discovery and charac- Mavromanolakis, E., Maraki, S., Samo- C., Raoult, D., and Fournier, P. E. (1968). Isolation of methanogenic terization of Mimivirus, the largest nis, G., Tselentis, Y., and Cranidis, (2012c). Non contiguous finished bacteria from feces of man. J. Bac- known virus and putative pneu- A. (1997). Effect of norfloxacin, genome sequence and description teriol. 96, 2178–2179. monia agent. Clin. Infect. Dis. 45, trimethoprim-sulfamethoxazole of Peptoniphilus timonensis sp. nov. O’Hara, A. M., and Shanahan, F. (2006). 95–102. and nitrofurantoin on fecal flora of Stand. Genomic Sci. 7. [Epub ahead The gut flora as a forgotten organ. Raoult, D., Renesto, P., and Brouqui, women with recurrent urinary tract of print] EMBO Rep. 7, 688–693. P. (2006). Laboratory infection of infections. J. Chemother. 9, 203–207. Mishra, A. K., Lagier, J. C., Rivet, Okike, K., Kocher, M. S., Wei, E. a technician by mimivirus. Ann. McNulty, N. P., Yatsunenko, T., Hsiao, R., Raoult, D., and Fournier, P. E. X., Mehlman, C. T., and Bhandari, Intern. Med. 144, 702–703. A., Faith, J. J., Muegge, B. D., and (2012d). Non contiguous finished M. (2009). Accuracy of conflict- Rawls, J. F., Samuel, B. S., and Goodman, A. L. (2011). The impact genome sequence and description of-interest disclosures reported by Gordon, J. I. (2004). Gnotobiotic

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 17 Lagier et al. Human gut microbiota

zebrafish reveal evolutionarily con- microbiota in colorectal cancer and Sim, K., Cox, M. J., Wopereis, H., Identification and characterization served responses to the gut micro- polyposis. Environ. Microbiol. 10, Martin, R., Knol, J., Li, M. S., of potential performance-related biota. Proc. Natl. Acad. Sci. U.S.A. 789–798. et al. (2012). Improved detec- gut microbiotas in broiler chick- 101, 4596–4601. Scanlan, P. D., Shanahan, F., O’Mahony, tion of bifidobacteria with opti- ens across various feeding tri- Reyes, A., Haynes, M., Hanson, N., C., and Marchesi, J. R. (2006). mised 16S rRNA-gene based pyrose- als. Appl. Environ. Microbiol. 77, Angly, F. E., Heath, A. C., Rohwer, Culture-independent analyses of quencing. PLos ONE 7, e32543. 5868–5878. F., et al. (2010). Viruses in the temporal variation of the dominant doi:10.1371/journal.pone.0032543 Turnbaugh, P. J., Backhed, F., Ful- faecal microbiota of monozygotic fecal microbiota and targeted bacte- Smith, R. (2005). Medical journals ton, L., and Gordon, J. I. (2008). twins and their mothers. Nature 466, rial subgroups in Crohn’s disease. J. are an extension of the mar- Diet-induced obesity is linked to 334–338. Clin. Microbiol. 44, 3980–3988. keting arm of pharmaceutical marked but reversible alterations in Robinson, C. J., and Young, V. B. Scanvic-Hameg,A., Chachaty, E., Rey, J., companies. PLoS Med. 2, e138. the mouse distal gut microbiome. (2010). Antibiotic administration Pousson, C., Ozoux, M. L., Brunel, doi:10.1371/journal.pmed.0020138 Cell Host Microbe 3, 213–223. alters the community structure of E., et al. (2002). Impact of quin- Staley, J. T., and Konopka, A. (1985). Turnbaugh, P. J., Hamady, M., Yat- the gastrointestinal micobiota. Gut upristin/dalfopristin (RP59500) on Measurement of in situ activi- sunenko, T., Cantarel, B. L., Duncan, Microbes 1, 279–284. the faecal microflora in healthy vol- ties of nonphotosynthetic microor- A., and Ley, R. E. (2009). A core gut Rozenberg-Arska, M., Dekker, A. W., unteers. J. Antimicrob. Chemother. ganisms in aquatic and terrestrial microbiome in obese and lean twins. and Verhoef, J. (1985). Ciprofloxacin 49, 135–139. habitats. Annu. Rev. Microbiol. 39, Nature 457, 480–484. for selective decontamination of Schwiertz, A., Taras, D., Schafer, K., Bei- 321–346. Turnbaugh, P. J., Ley, R. E., Hamady, the alimentary tract in patients jer, S., Bos, N. A., Donus, C., et Stark, C. A., Adamsson, I., Edlund, C., M., Fraser-Liggett, C. M., Knight, with acute leukemia during remis- al. (2010). Microbiota and SCFA in Sjosted, S., Seensalu, R., Wikstrom, R., and Gordon, J. I. (2007). The sion induction treatment: the effect lean and overweight healthy sub- B., et al. (1996). Effects of omepra- human microbiome project. Nature on fecal flora. J. Infect. Dis. 152, jects. Obesity (Silver Spring) 18, zole and amoxycillin on the human 449, 804–810. 104–107. 190–195. oral and gastrointestinal microflora Turnbaugh, P. J., Ley, R. E., Mahowald, Saarela, M., Maukonen, J., von Wright, Scupham, A. J., Presley, L. L., Wei, B., in patients with Helicobacter pylori M. A., Magrini, V., Mardis, E. R., A.,Vilpponen-Salmela, T.,Patterson, Bent, E., Griffith, N., and McPher- infection. J. Antimicrob. Chemother. and Gordon, J. I. (2006). An obesity- A. J., and Scott, K. P. (2007). Tetra- son, M. (2006). Abundant and 38, 927–939. associated gut microbiome with cycline susceptibility of the ingested diverse fungal microbiota in the Sullivan, A., Barkholt, L., and Nord, increased capacity for energy har- Lactobacillus acidophilus LaCH-5 murine intestine. Appl. Environ. C. E. (2003). Lactobacillus aci- vest. Nature 444, 1027–1031. and Bifidobacterium animalis subsp. Microbiol. 72, 793–801. dophilus, Bifidobacterium lactis and Turnbaugh, P. J., Quince, C., Faith, J. J., lactis Bb-12 strains during antibi- Sekirov, I., Russell, S. L., Antunes, L. C., Lactobacillus F19 prevent antibiotic- McHardy, A. C., Yatsunenko, T., and otic/probiotic intervention. Int. J. and Finlay, B. B. (2010). Gut micro- associated ecological disturbances of Niazi,F.(2010). Organismal,genetic, Antimicrob. Agents 29, 271–280. biota in health and disease. Physiol. Bacteroides fragilis in the intestine. J. and transcriptional variation in the Samuel, B. S., and Gordon, J. I. (2006). Rev. 90, 859–904. Antimicrob. Chemother. 52, 308–311. deeply sequenced gut microbiomes A humanized gnotobiotic mouse Sekirov, I., Tam, N. M., Jogova, M., Sullivan, A., Edlund, C., and Nord, of identical twins. Proc. Natl. Acad. model of host-archaeal-bacterial Robertson, M. L., Li, Y., Lupp, C., et C. E. (2001). Effect of antimicro- Sci. U.S.A. 107, 7503–7508. mutualism. Proc. Natl. Acad. Sci. al. (2008). Antibiotic-induced per- bial agents on the ecological balance Vael, C., Verhulst, S. L., Nelen, V., U.S.A. 103, 10011–10016. turbations of the intestinal micro- of human microflora. Lancet Infect. Goossens, H., and Desager, K. N. Sanders, M. E. (2011). Impact of pro- biota alter host susceptibility to Dis. 1, 101–114. (2011). Intestinal microflora and biotics on colonizing microbiota enteric infection. Infect. Immun. 76, Swedish Study Group. (1991a). A body mass index during the first of the gut. J. Clin. Gastroenterol. 4726–4736. randomized multicenter trial to three years of life: an observational 45(Suppl.), S115–S119. Seng, P., Drancourt, M., Gouriet, F., La compare the influence of cefaclor study. Gut Pathog. 3, 8. Santacruz, A., Marcos, A., Warnberg, Scola,B.,Fournier,P.E.,Rolain,J. M., and amoxycillin on the colonization Van der Auwera, P., Pensart, N., J., Marti, A., Martin-Matillas, M., et al. (2009). Ongoing revolution in resistance of the digestive tract in Korten, V., Murray, B. E., and and Campoy, C. (2009). Interplay bacteriology: routine identification patients with lower respiratory tract Leclercq, R. (1996). Influence of between weight loss and gut micro- of bacteria by matrix-assisted laser infection. Infection 19, 208–215. oral glycopeptides on the fecal biota composition in overweight desorption ionization time-of-flight Swedish Study Group. (1991b). A flora of human volunteers: selec- adolescents. Obesity (Silver Spring) mass spectrometry. Clin. Infect. Dis. randomized multicenter trial to tion of highly glycopeptide-resistant 17, 1906–1915. 49, 543–551. compare the influence of cefaclor enterococci. J. Infect. Dis. 173, Savino, F., Roana, J., Mandras, N., Seng, P., Rolain, J. M., Fournier, P. E., and amoxycillin on the colonization 1129–1136. Tarasco, V., Locatelli, E., and Tul- La Scola, B., Drancourt, M., and resistance of the digestive tract in van Nispen, C. H., Hoepelman, A. I., lio, V. (2011). Faecal microbiota Raoult, D. (2010). MALDI-TOF- patients with lower respiratory tract Rozenberg-Arska, M., Verhoef, J., in breast-fed infants after antibiotic mass spectrometry applications in infection. Infection 19, 208–215. Purkins, L., and Willavize, S. A. therapy. Acta Paediatr. 100, 75–78. clinical microbiology. Future Micro- Thomas, O., Thabane, L., Douketis, J., (1998). A double-blind, placebo- Sbarbati, A., Osculati, F., Silvagni, D., biol. 5, 1733–1754. Chu, R., Westfall, A. O., and Allison, controlled, parallel group study Benati, D., Galie, M., and Camoglio, Shimada, K., Bricknell, K. S., and Fine- D. B. (2008). Industry funding and of oral trovafloxacin on bowel F. S. (2006). Obesity and inflam- gold, S. M. (1969). Deconjugation the reporting quality of large long- microflora in healthy male volun- mation: evidence for an elementary of bile acids by intestinal bacte- term weight loss trials. Int. J. Obes. teers. Am. J. Surg. 176, 27S–31S. lesion. Pediatrics 117, 220–223. ria: review of literature and addi- (Lond.) 32, 1531–1536. Van Saene, J. J., Van Saene, H. K., Scanlan, P. D., and Marchesi, J. R. tional studies. J. Infect. Dis. 119, Thuny, F., Richet, H., Casalta, J. Geitz, J. N., Tarko-Smit, N. J., and (2008). Micro-eukaryotic diversity 273–281. P., Angelakis, E., Habib, G., and Lerk, C. F. (1986). Quinolones and of the human distal gut microbiota: Siggers, R. H., Siggers, J., Boye, M., Thy- Raoult, D. (2010). Vancomycin colonization resistance in human qualitative assessment using culture- mann, T., Molbak, L., and Leser, T. treatment of infective endocardi- volunteers. Pharm. Weekbl. Sci. 8, dependent and -independent analy- (2008). Early administration of pro- tis is linked with recently acquired 67–71. sis of faeces. ISME J. 2, 1183–1193. biotics alters bacterial colonization obesity. PLoS ONE 5, e9074. Vartoukian, S. R., Palmer, R. M., and Scanlan, P. D., Shanahan, F., Clune, and limits diet-induced gut dysfunc- doi:10.1371/journal.pone.0009074 Wade, W. G. (2010). Strategies Y., Collins, J. K., O’Sullivan, G. C., tion and severity of necrotizing ente- Torok, V. A., Hughes, R. J., Mikkelsen, for culture of “unculturable” bac- and O’Riordan, M. (2008). Culture- rocolitis in preterm pigs. J. Nutr. 138, L. L., Perez-Maldonado, R., Bald- teria. FEMS Microbiol. Lett. 309, independent analysis of the gut 1437–1444. ing, K., and MacAlpine, R. (2011). 1–7.

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 18 Lagier et al. Human gut microbiota

Vogel,F.,Ochs,H. R.,Wettich,K.,Kalich, the bacterial enzymatic activity in W. (2010). Effects of polymerase, in obesity and after gastric bypass. S., Nilsson-Ehle, I., Odenholt, I., et the intestinal tract. Infection 19, template dilution and cycle number Proc. Natl. Acad. Sci. U.S.A. 106, al. (2001). Effect of step-down ther- 313–316. on PCR based 16 S rRNA diversity 2365–2370. apy of ceftriaxone plus loracarbef Willner, D., Furlan, M., Haynes, analysis using the deep sequencing versus parenteral therapy of ceftri- M., Schmieder, R., Angly, F. E., method. BMC Microbiol. 10, 255. Conflict of Interest Statement: The axone on the intestinal microflora in and Silva, J. (2009). Metage- doi:10.1186/1471-2180-10-255 authors declare that the research was patients with community-acquired nomic analysis of respiratory tract Xu, J., Bjursell, M. K., Himrod, J., Deng, conducted in the absence of any com- pneumonia. Clin. Microbiol. Infect. 7, DNA viral communities in cystic S., Carmichael, L. K., and Chiang, mercial or financial relationships that 376–379. fibrosis and non-cystic fibrosis H. C. (2003). A genomic view could be construed as a potential con- Wagner, M., and Horn, M. (2006). individuals. PLoS ONE 4, e7370. of the human-Bacteroides thetaio- flict of interest. The planctomycetes, verrucomi- doi:10.1371/journal.pone.0007370 taomicron symbiosis. Science 299, crobia, chlamydiae and sister phyla Wilson, K. H., and Blitchington, R. B. 2074–2076. Received: 30 August 2012; paper pending comprise a superphylum with (1996). Human colonic biota stud- Yildirim, S., Yeoman, C. J., Sipos, published: 22 September 2012; accepted: biotechnological and medical rel- ied by ribosomal DNA sequence M., Torralba, M., Wilson, B. A., 16 October 2012; published online: 02 evance. Curr. Opin. Biotechnol. 17, analysis. Appl. Environ. Microbiol. and Goldberg, T. L. (2010). Char- November 2012. 241–249. 62, 2273–2278. acterization of the fecal micro- Citation: Lagier J-C, Million M, Hugon Walker, A. (2010). Gut metagenomics Wistrom, J., Gentry, L. O., Palmgren, A. biome from non-human wild pri- P, Armougom F and Raoult D (2012) goes viral. Nat. Rev. Microbiol. 8, C., Price, M., Nord, C. E., Ljungh, mates reveals species specific micro- Human gut microbiota: repertoire and 841. A., et al. (1992). Ecological effects of bial communities. PLoS ONE 5, variations. Front. Cell. Inf. Microbio. Walker, A. W., Ince, J., Duncan, S. short-term ciprofloxacin treatment e13963. doi:10.1371/journal.pone. 2:136. doi: 10.3389/fcimb.2012.00136 H., Webster, L. M., Holtrop, G., of travellers’ diarrhoea. J. Antimi- 0013963 Copyright © 2012 Lagier, Million, and Ze, X. (2011). Dominant and crob. Chemother. 30, 693–706. Young, V. B., and Schmidt, T. M. Hugon, Armougom and Raoult. This is diet-responsive groups of bacteria Wu, G. D., Chen, J., Hoffmann, C., (2004). Antibiotic-associated diar- an open-access article distributed under within the human colonic micro- Bittinger, K., Chen, Y. Y., and Keil- rhea accompanied by large-scale the terms of the Creative Commons Attri- biota. ISME J. 5, 220–230. baugh, S. A. (2011). Linking long- alterations in the composition of the bution License, which permits use, distri- Welling, G. W., Meijer-Severs, G. J., term dietary patterns with gut fecal microbiota. J. Clin. Microbiol. bution and reproduction in other forums, Helmus, G., van Santen, E., Tonk, microbial enterotypes. Science 334, 42, 1203–1206. provided the original authors and source R. H., de Vries-Hospers, H. G., et 105–108. Zhang, H., Dibaise, J. K., Zuccolo, A., are credited and subject to any copy- al. (1991). The effect of ceftriaxone Wu, J. Y., Jiang, X. T., Jiang, Y. X., Kudrna, D., Braidotti, M., and Yu, right notices concerning any third-party on the anaerobic bacterial flora and Lu, S. Y., Zou, F., and Zhou, H. Y. (2009). Human gut microbiota graphics etc.

Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2012 | Volume 2 | Article 136 | 19

Article XII :

Vancomycin-associated Gut Microbiota Alteration

and Weight Gain in Human Adults

Matthieu Million, Franck Thuny, Manolis Angelakis, Jean-Paul

Casalta, Gilbert Habib, Didier Raoult

Work in progress

124

1 Vancomycin-associated Gut Microbiota Alteration and Weight Gain in Human Adults

2 Matthieu MILLION 1, Franck THUNY 1, 2 , Manolis ANGELAKIS 1, Jean-Paul CASALTA 3,

3 Gilbert HABIB 2, Didier RAOULT 1,3 *

4 1. Unité de Recherche sur les Maladies Infectieuses et Tropicales Emergentes, Faculté de

5 Médecine, CNRS UMR 7278, IRD 198, Aix-Marseille Université, 27 Bd Jean Moulin, 13005

6 Marseille, France.

7 2. Service de Cardiologie, Hôpital de la Timone, Marseille, France,

8 3. Pôle de Maladies Infectieuses, Hôpital de la Timone, Marseille, France

10 *Corresponding author.

11 Phone: (33) 491 38 55 17

12 Fax: (33) 491 83 03 90

13 Email: [email protected]

14 Abstract: XXX

15 Text word count: 2,921

25 Keywords: Antibiotics, weight gain, obesity, gut microbiota

26 Introduction

27 Antibiotics, including mainly glycopeptide, tetracycline, macrolides and penicillin,

28 have been used for over 60 years to promote weight gain in animals (1) and continue to be

29 widely used in the USA (2). A report (3) estimated that 1600 tons of antibiotic were used for

30 their growth-promoter effect in European agriculture in 1997 before their ban (4) due to

31 bacterial resistance issues that represent a threat for human health. From the beginning of their

32 use in agriculture in the 50s, a similar effect on growth was reported in humans (5-7) but

33 seems to have been neglected until very recently. We and others stress the need to assess

34 carefully the impact of prolonged antibiotics on weight in humans (8-10) while antibiotics in

35 early infancy have been linked very recently to excess risk of being overweight later in

36 childhood (10).

37 Among the many antibiotics used for their growth factor effect, it was observed that

38 the glycopeptides showed the same effect in animals and humans. Indeed, avoparcin,

39 originally isolated from Streptomyces candidus (11), was associated with weight gain in farm

40 animals (12) while vancomycin isolated from Amycolatopsis orientalis (formerly named

41 Streptomyces orientalis) (13) was associated with significant weight gain and acquired obesity

42 in humans (14). Both these glycopeptides are known to be active only on Gram positive

43 bacteria with the notable exception of most of Pediococcous and Lactobacillus. As an

44 exception, members of the Lactobacillus acidophilus group ( L. acidophilus, L. gasseri, L.

45 crispatus, L. johnsonii and L. delbruecki) have been found to be susceptible to glycopeptides

46 (15).

47 It has been shown using axenic animals that the weight gain of antibiotic growth

48 promoter (AGP) was linked to changes in the intestinal flora (16;17). Many studies have

49 reported the modification of digestive microbiota after administration of AGP in animals (18-

50 21) and some studies from the 1950s’ reported an E. coli decrease by culture in bottle-fed

51 infant gaining weight after chlortetracycline administration (22). To our knowledge, no recent

52 study has linked the modification of the intestinal flora and weight gain in humans by

53 molecular methods.

54 In a former work, we found that antibiotics (ampicillin, vancomycin and gentamycin)

55 were linked with increased body weight but gut flora was not explored (14). In this work, we

56 have tested the weight gain effect associated with vancomycin in a new series in humans and

57 compared the change in digestive microbiota in patients receiving vancomycin or amoxicillin

58 by molecular techniques.

60 Methods

61 Ethics Committee

62 Written consent was obtained from each participant and approved by the ethics committee of

63 the Faculty of Medicine La Timone, Marseille, France under number 08 –002.

64 Patients

65 We retrospectively included all IE patients treated in the departments of Cardiology, Hospital

66 de la Timone, Marseille, France from January 2008 until January 2011. Endocarditis was

67 defined by the modified Duke criteria (23). Patients were treated with amoxicillin or

68 vancomycin for at least 4 weeks initially associated or not to gentamycin. For each case of

69 endocarditis, the decision for surgery was made by a multidisciplinary discussion following

70 our protocol (24). The use of another antibiotic than amoxicillin, vancomycin or gentamycin

71 for more than 7 days was an exclusion criterion. A control group included 42 controls for

72 whom the stools were analysed in a previous study (25).

73 Analysis of weight change

74 A standardized questionnaire was used to collect demographic, clinical and therapeutic data in

75 all patients treated by antibiotics. The baseline weight (one month before the onset of the

76 disease), weight at one year and height were collected prospectively or retrospectively for

77 each patient based on clinical records, systematic follow-up consultations or phone call.

78 Analysis of intestinal flora

79 From April 2009, all patients with a diagnosis of endocarditis had a stools sample analyzed at

80 diagnosis, and repeated as possible once a week during 4-6 weeks of treatment. DNA was

81 isolated from the stool as described in Dridi et al. (26). The purified DNA was eluted from

82 samples to a final volume of 100 µl and stored at -80 ° C until analysis. The real-time PCR

83 was performed on a Stratagene MX3000 (Agilent, Santa Clara, CA, United States) using the

84 Quantitect PCR mix (Qiagen, Courtaboeuf, France) as described previously (27) targeting the

85 Bacteroidetes, Firmicutes, Lactobacillus, Methanobrevibacter smithii. Quantification of all

86 bacteria was performed as previously described (26). A second real-time PCR was performed

87 on BioRad FLX96 targeting Lactobacillus reuteri, Lactobacillus plantarum, Lactobacillus

88 rhamnosus , Bifidobacterium animalis as previously reported (25). Escherichia coli was

89 quantified using an in-house PCR system using the following primers ( ECOmpGMGBAluId

90 GCTGCGCGTGCAAATGCG, ECOmpGMGBAluIr CATGGTCATCGCTTCGGTCT) and a

91 MGB probe (ECOmpGMGB 6FAM-CATCAGAAACTGAACACCAC ) yielding a 100 bp

92 amplicons with the previously reported PCR protocol (25).

94 Statistical Analysis

95 The baseline BMI was that obtained one month before the first symptoms. First, the changes

96 of BMI at one year were compared between the two groups. Proportions were compared using

97 the Fisher bilateral exact test. For the continuous variables, the student t test or Mann-

98 Whitney test were used for comparison between the two groups according to the distribution

99 assessed by Kolmogorov-Smirnov test. Logistic regression was used to determine the

100 pr edictors of an increase of BMI ≥10% at one year. The following variables were tested as

101 potential predictors: age, sex, cardiac surgery within the year after admission, and the type of

102 antibiotic used in association with gentamycin (i.e., vancomycin or amoxicillin). In a second

103 analysis, concentration of Bacteroidetes , Firmicutes , Methanobrevibacter smithii and

104 Lactobacillus were compared between intervention groups (vancomycin, amoxicillin,

105 controls). In a third analysis, specific bacterial species present in the first stool sample

106 collected before 7 days of treatment were tested as potential predictors of weight change at

107 one year. P<0.05 was considered significant and a Bonferroni's correction was used in case of

108 multiple comparisons. All statistical analyses were performed using EpiInfo software version

109 3.4.1 (Centers for Disease Control and Prevention, Atlanta, GA, USA). To confirm data on a

110 broader group we later combine this study with our former work (14).

111

112 Results

113 Characteristics of patients

114 98 patients were included in the weight study, 57 patients were included in the amoxicillin

115 group and 41 in the vancomycin group (Table 1). Sex, Age and BMI were not significantly

116 different at baseline. Surgery was more frequent in the amoxicillin group; 34/54(62%), than in

117 the vancomycin group; 18/38(47%) but this was not significant (p=0.13). In the second part of

118 the study, comparing bacterial concentration between intervention groups, 192 stool samples

119 were analyzed for Bacteroidetes, Firmicutes, Lactobacillus or M. smithii (67 during

120 vancomycin treatment, 83 during amoxicillin treatment, 42 without antibiotics) corresponding

121 to 80 patients including 35 patients on vancomycin and 45 patients on amoxicillin and 42

122 controls. For the third part of our study, 53 samples were available for the analysis of the

123 initial gut microbiota targeting different Lactobacillus species, Bifidobacterium animalis and

124 E. coli corresponding to 53 patients.

125

126 Weight change

127 Globally, there was no significant difference between vancomycin and amoxicillin groups in

128 BMI change over a year and after stratification by age, sex, initial weight or surgery (Table 2).

129 However, the proportion of patients with an increase in BMI greater than 10% was higher in

130 the vancomycin group (5 / 41 (12.2%) compared to 1 / 57 (1.7%) for the amoxicillin group,

131 Barnard bilateral exact test, p = 0.038). It was not significant for an increase in BMI greater

132 than 5% (9/41 in the vancomycin group vs 7/57 in the amoxicillin group (p=0.18). Again in

133 logistic regression model including age, sex, surgery, initial weight category (lean,

134 overweight, obese), vancomycin was associated with an increase in BMI greater than 10%

135 (adjusted OR = 0.54 (95%CI 0.003 – 1.074); p = 0.049). Women of less than 60 years seemed

136 to be more at risk, indeed 2 out of 4 women of less than 60 years treated by vancomycin

137 showed an increase in BMI greater than 10% with acquired obesity in one case (Initial BMI at

138 25.7, BMI at one year 30.6). This initially lean woman gain 11 kg in one year. When we

139 combined this study with our first study, we did not observe any significant change in weight

140 for all individuals nor for an increase in BMI of 5%. Conversely, there was a significant

141 increase in the number of patients with increased BMI of over 10% (12/52 vs 8/84, p = 0.034)

142 and the number of patients with acquired obesity (7/52 vs. 2/84, p = 0.011). However, even

143 after stratification for age, sex, surgery or initial BMI, it was not possible to identify a

144 subgroup with a significant weight gain.

145

146 Gut microbiota alteration following antibiotic treatment

147 192 stool samples were analyzed including 83 during amoxicillin treatment, 67 during

148 vancomycin treatment and 42 in the control group. The number of sample per patient was not

149 significantly different between the 2 antibiotic groups (p=0.47) so samples were pooled by

150 antibiotic. The amount of Firmicutes were significantly increased in the vancomycin group

151 compared to the amoxicillin group (p = 0.027) and controls (p = 0.005) while the amoxicillin

152 group showed no difference for Firmicutes compared to the control group (p = 0.66) (figure

153 1). The amount of Bacteroidetes was also increased in the vancomycin groups but only

154 compared to controls (p <0.0001). Similarly the amount of Bacteroidetes was increased in the

155 amoxicillin group (p = 0.002). Conversely, the amount of Methanobrevibacter smithii was

156 significantly decreased in both groups of patients on antibiotics compared to controls (p =

157 0.013 and p = 0.011 respectively for vancomycin and amoxicillin). Conversely, there was no

158 significant difference for M. smithii between patients receiving vancomycin and amoxicillin

159 (p = 0.86). Lactobacillus were significantly increased in the vancomycin group both

160 compared to the amoxicillin group and controls (p = 0.0009 and p = 0.037 respectively).

161 There was no significant difference for Lactobacillus bacterial count between the amoxicillin

162 group and the control group. Finally, the total number of bacteria was increased in both

163 groups of patients receiving antibiotics compared to controls (p <0.0001 and p = 0.002 for

164 vancomycin and amoxicillin, respectively). Similarly, the total number of bacteria was higher

165 in the vancomycin group compared to the amoxicillin group (p = 0.04).

166

167 Testing bacterial predictors for weight changes after antibiotic treatment

168 Analysing the presence of specific species previously related to obese or lean human status

169 (B. animalis, L. plantarum, L. reuteri, L. rhamnosus, E. coli ), we found no significant results

170 for an association with BMI change. Considering presence or absence of each analysed

171 species by individuals and by weight change category (>1%, >5%, >10% BMI increase),

172 consistent effect direction but not significant was noted with Bifidobacterium animalis,

173 Lactobacillus plantarum and Escherichia coli associated with a protective effect against

174 weight gain (Odds ratio < 1) and Lactobacillus reuteri favoring weight gain (Odds ratio > 1)

175 (Figure 2). E. coli , reported elsewhere as associated with obesity (27), unexpectedly show

176 here a protective effect (OR < 1 for each category of weight gain) even if result was not

177 significant. The woman with acquired obesity who gained 11kg after vancomycin treatment

178 harboured Lactobacillus reuteri in her gut microbiota, identified previously having a pro-

179 obesity link (2).

180 Finally, we didn’t found any statistical significant antagonism between any Lactobacillus

181 species or Bifidobacterium animalis and E. coli (data not shown).

182

183 Discussion

184 We found in this work an increased frequency of patients with acquired obesity and

185 BMI increase over 10% after prolonged treatment with vancomycin and gentamycin for

186 endocarditis compared with amoxicillin and gentamycin treatment, taken here as a control.

187 We found that vancomycin and amoxicillin were associated with an increase in Bacteroidetes

188 and total bacterial concentration and a decrease of Methanobrevibacter smithii . Furthermore,

189 we found a significant increase in Firmicutes , Lactobacillus and global bacterial

190 concentration in vancomycin-treated patients gut microbiota compared with patients treated

191 by amoxicillin. Finally, we found, even if no significant results were obtain, that pre-existing

192 profile of the gut microbiota could predict weight gain after prolonged antibiotic treatment.

193 Weight gain associated with antibiotics have been reported since the 1940 s’ in animals

194 (28) and since the 1950s in humans (5-7)(Table 3). Avoparcin, a glycopeptides bacteriocin

195 first isolated from a Streptomyces sp. (11), have been used for decades as a growth-promoter

196 in the farm industry (12). In humans, Thuny et al. (14) first found that another glycopeptide,

197 vancomycin, was associated with significant weight gain, increased frequency of acquired

198 obesity and of BMI increase over 10% in patients treated for infective endocarditis.

199 Amoxicillin was also associated with weight gain over 10% compared to controls without

200 antibiotics. In this study, we confirm an association between vancomycin and an increased

201 frequency of acquired obesity and BMI increase over 10%. Amoxicillin has not been retested

202 but experimental studies found a similar effect with a growth-promoting effect of penicillin

203 (29).

204 Independently of weight gain, we recently reviewed the reported effects of antibiotics

205 on gut microbiota (30). Here, we found significant alterations in vancomycin treated humans

206 gut microbiota with an increase in Bacteroidetes , Firmicutes and Lactobacillus and a decrease

207 in M. smithii . In the literature (Table 4), the effects of vancomycin and amoxicillin on the

208 digestive flora are completely different. The most striking differences is an increase in

209 Lactobacillacaeae under vancomycin while they are reduced as amoxicillin. Searching for

210 association between gut bacteria, antibiotics and weight change, penicillin have been shown

211 to increased weight gain in conventional animals whereas germ free animals did not respond

212 to the antibiotic (17). Clostridium perfringens when implanted in the gut of germ-free chicks,

213 caused growth depression reversible by penicillin (31). Enterococcus faecalis or Clostridium

214 perfringens have been linked with decreased fat absorption in gnotobiotic animals, whereas

215 Lactobacillus didn’t impact the fat absorption (32). Torok et al. found certain Lactobacillus

216 species linked to weight gain in the proximal gut of chicks whereas others were specifically

217 associated with non responding birds (20). In one of the most recent studies on antibiotics

218 generating adiposity including vancomycin and penicillin, Cho et al. (29) reported an increase

219 in the number of sequences of the Lachnospiraceae but also Lactobacillaceae. The authors

220 found a significant change in the proportion of sequences at the phylum level only for

221 vancomycin with a decrease in Bacteroidetes and a Firmicutes increased.

222 In our study, we found that the total bacterial amount is surprisingly increased under

223 antibiotic and is significantly increased in the vancomycin group as compared with penicillin.

224 Jukes noted in 1955 that when the usual “feed” antibiotics are given to animals, there is

225 typically an increase in the number of intestinal bacteria that can be plated out, and the new

226 population is predominantly antibiotic-tolerant or resistant (33). Cho et al. (29) reported

227 similar results. Indeed, in their study, vancomycin and vancomycin associated with penicillin

228 were associated with a non-significant increase in the total number of such sequences

229 suggesting an increase of total gut bacteria with antibiotics. Conversely, Vijay-Kumar et al.

230 (34) reported that a very broad spectrum antibiotics regimen including ampicillin and

231 neomycin reduces the gut total bacterial load by 90% and corrects the metabolic syndrome in

232 mice experimentally induced TLR5-KO. These results suggest that the metabolic syndrome is

233 linked with gut bacteria. The gut microbiota seems to be only structurally altered but not

234 quantitatively reduced by antibiotics with limited anti-bacterial spectra and this alteration

235 could lead to increased adiposity and weight gain.

236 Coates et al. (17) reported a decreased weight gain in germ-free animals fed a standard

237 diet whereas Backed et al. (35) found that germ-free mice were resistant to diet-induced

238 obesity. Whatever, as a putative explanation, it is plausible that changes in the microbiota

239 following antibiotic treatment could reproduce the alterations observed in the digestive

240 microbiota of obese namely a decrease in Bifidobacterium in animals treated by antibiotics

241 (29) or in obese (36), an increase in Lactobacillus under antibiotics (present study, (37)) and

242 obese patients (25;27;38), an increase in Staphylococcaceae under antibiotics (29) and in

243 obese patients (36;39), an increase of Lachnospiraceae on antibiotics (29) and obese-induced

244 animals (40). Finally, antibiotics were linked here with a Methanobrevibacter smithii

245 decrease, previously found in obese gut microbiota (25;36). Conversely, Membrez et al . (41)

246 reported that administration of ampicillin and norfloxacin in ob/ob and diet-induced obese and

247 insulin-resistant mice were linked with a non-significant weight loss, a significant total fat pad

248 weight decrease and a significant improvement in fasting glycemia and oral glucose tolerance.

249 More generally, to our knowledge, fluoroquinolones had never been linked to human or

250 animal weight gain in the literature (42) but a weight gain suppression (43). These results

251 suggest that antibiotics, and especially vancomycin, can enhance weight gain through gut

252 microbiota manipulation decreasing bacteria associated with lower energy harvest and fat

253 absorption ( Clostridium perfringens, Enterococcus sp. usually susceptible to vancomycin)

254 while favoring bacteria promoting the absorption and accumulation of energy (carbohydrate

255 and lipid) by the host (Lactobacillus ).

256 Finally, our work suggests that the structure of the microbiota before antibiotic

257 administration predict which individuals are more likely to gain weight and to present an

258 acquired obesity as a result of treatment with a specific antibiotic. In this sense, even if not

259 significant, Lactobacillus reuteri , linked with obesity elsewhere (25) has the same effect and

260 predict a BMI increase > 10%, while Bifidobacterium animalis and Lactobacillus plantarum

261 linked elsewhere with lean status (25) have the opposite effect and could prevent weight gain

262 after antibiotic administration.

263

264 Conclusion

265 Our work confirms that vancomycin is associated with weight gain in humans in relation to a

266 specific gut alteration including a Firmicutes and Lactobacillus increase and

267 Methanobrevibacter smithii decrease. It appears that antibiotics modulate the digestive

268 microbiota depending on the pre-existent gut microbiota and the spectrum of the antibiotic

269 administered. This work extends the previous findings showing that vancomycin led to

270 dramatic weight gain in certain individuals. More generally, in view of our results, it seems

271 necessary to inform the patient of the risk of weight gain and obesity acquired during long-

272 term antibiotic treatment especially vancomycin. Further studies are needed to identify which

273 individuals are more susceptible to have this weight gain side effect characterizing their

274 digestive microbiota prior to antibiotics.

275

276 Acknowledgement

277 We thank the cardiology department for their helpful help in the recruitment of the IE

278 patients.

279

280 Funding source

281 No funding source

282

283

284

285

286

287

288

289

290

291

292

293

294

295

296 Reference List

297

298 (1) OZAWA E. Studies on growth promotion by antibiotics. I. Effects of chlortetracycline on 299 growth. J Antibiot (Tokyo) 1955 Dec;8(6):205-11.

300 (2) Food and drug administration. Withdrawal of Notices of Opportunity for a Hearing; 301 Penicillin and Tetracycline Used in Animal Feed. 79697-79701. 22-12-2011. 302

303 (3) Harvey J, Mason L. The Use and Misuse of Antibiotics in UK agriculture. 1998. 304

305 (4) Ban on antibiotics as growth promoters in animal feed enters into effect. 22-12-2005. 306

307 (5) HAIGHT TH, PIERCE WE. Effect of prolonged antibiotic administration of the weight of 308 healthy young males. J Nutr 1955 May 10;56(1):151-61.

309 (6) OZAWA E. Studies on growth promotion by antibiotics. II. Results of aurofac administration 310 to infants. J Antibiot (Tokyo) 1955 Dec;8(6):212-4.

311 (7) Robinson P. Controlled trial of aureomycin in premature twins and triplets. Lancet 1952 Jan 312 5;259(6697):52.

313 (8) Ternak G. Antibiotics may act as growth/obesity promoters in humans as an inadvertent 314 result of antibiotic pollution? Med Hypotheses 2005;64(1):14-6.

315 (9) Raoult D. Human microbiome: take-home lesson on growth promoters? Nature 2008 Aug 316 7;454(7205):690-1.

317 (10) Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant antibiotic exposures and 318 early-life body mass. Int J Obes (Lond) 2012 Aug 21.

319 (11) Michel KH, Shah RM, Hamill RL. A35512, a complex of new antibacterial antibiotics 320 produced by Streptomyces candidus. I. Isolation and characterization. J Antibiot (Tokyo) 321 1980 Dec;33(12):1397-406.

322 (12) Feighner SD, Dashkevicz MP. Subtherapeutic levels of antibiotics in poultry feeds and their 323 effects on weight gain, feed efficiency, and bacterial cholyltaurine hydrolase activity. Appl 324 Environ Microbiol 1987 Feb;53(2):331-6.

325 (13) MCCORMICK MH, MCGUIRE JM, PITTENGER GE, PITTENGER RC, STARK WM. Vancomycin, a 326 new antibiotic. I. Chemical and biologic properties. Antibiot Annu 1955;3:606-11.

327 (14) Thuny F, Richet H, Casalta JP, Angelakis E, Habib G, Raoult D. Vancomycin treatment of 328 infective endocarditis is linked with recently acquired obesity. PLoS One 2010;5(2):e9074.

329 (15) Klare I, Konstabel C, Werner G, Huys G, Vankerckhoven V, Kahlmeter G, et al. Antimicrobial 330 susceptibilities of Lactobacillus, Pediococcus and Lactococcus human isolates and cultures 331 intended for probiotic or nutritional use. J Antimicrob Chemother 2007 May;59(5):900-12.

332 (16) Gaskins HR, Collier CT, Anderson DB. Antibiotics as growth promotants: mode of action. 333 Anim Biotechnol 2002 May;13(1):29-42.

334 (17) COATES ME, Fuller R, HARRISON GF, LEV M, SUFFOLK SF. A comparison of the growth of 335 chicks in the Gustafsson germ-free apparatus and in a conventional environment, with and 336 without dietary supplements of penicillin. Br J Nutr 1963;17:141-50.

337 (18) Looft T, Johnson TA, Allen HK, Bayles DO, Alt DP, Stedtfeld RD, et al. In-feed antibiotic 338 effects on the swine intestinal microbiome. Proc Natl Acad Sci U S A 2012 Jan 339 31;109(5):1691-6.

340 (19) Kim HB, Borewicz K, White BA, Singer RS, Sreevatsan S, Tu ZJ, et al. Microbial shifts in the 341 swine distal gut in response to the treatment with antimicrobial growth promoter, tylosin. 342 Proc Natl Acad Sci U S A 2012 Sep 18;109(38):15485-90.

343 (20) Torok VA, Hughes RJ, Mikkelsen LL, Perez-Maldonado R, Balding K, MacAlpine R, et al. 344 Identification and characterization of potential performance-related gut microbiotas in 345 broiler chickens across various feeding trials. Appl Environ Microbiol 2011 Sep;77(17):5868- 346 78.

347 (21) Torok VA, Allison GE, Percy NJ, Ophel-Keller K, Hughes RJ. Influence of antimicrobial feed 348 additives on broiler commensal posthatch gut microbiota development and performance. 349 Appl Environ Microbiol 2011 May;77(10):3380-90.

350 (22) OZAWA E. Studies on growth promotion by antibiotics. II. Results of aurofac administration 351 to infants. J Antibiot (Tokyo) 1955 Dec;8(6):212-4.

352 (23) Li JS, Sexton DJ, Mick N, Nettles R, Fowler VG, Jr., Ryan T, et al. Proposed modifications to 353 the Duke criteria for the diagnosis of infective endocarditis. Clin Infect Dis 2000 354 Apr;30(4):633-8.

355 (24) Botelho-Nevers E, Thuny F, Casalta JP, Richet H, Gouriet F, Collart F, et al. Dramatic 356 reduction in infective endocarditis-related mortality with a management-based approach. 357 Arch Intern Med 2009 Jul 27;169(14):1290-8.

358 (25) Million M, Maraninchi M, Henry M, Armougom F, Richet H, Carrieri P, et al. Obesity- 359 associated gut microbiota is enriched in Lactobacillus reuteri and depleted in 360 Bifidobacterium animalis and Methanobrevibacter smithii. Int J Obes (Lond) 2012 361 Jun;36(6):817-25.

362 (26) Dridi B, Henry M, El KA, Raoult D, Drancourt M. High prevalence of Methanobrevibacter 363 smithii and Methanosphaera stadtmanae detected in the human gut using an improved 364 DNA detection protocol. PLoS One 2009;4(9):e7063.

365 (27) Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. Monitoring bacterial community of 366 human gut microbiota reveals an increase in Lactobacillus in obese patients and 367 Methanogens in anorexic patients. PLoS One 2009;4(9):e7125.

368 (28) Moore PR, Evenson A, Luckey TD, McCoy E, Elvehjem CA, Hart EB. Use of sulfasuxidine, 369 streptothricin, and streptomycin in the nutritional studies with the chick. J Biol Chem 370 1946;165:437-41.

371 (29) Cho I, Yamanishi S, Cox L, Methe BA, Zavadil J, Li K, et al. Antibiotics in early life alter the 372 murine colonic microbiome and adiposity. Nature 2012 Aug 30;488(7413):621-6.

373 (30) Lagier J-C, Million M, Hugon P, Armougom F, Raoult D. Human gut microbiota: repertoire 374 and variations. Front Cell Inf Microbio. In press 2012.

375 (31) LEV M, FORBES M. Growth response to dietary penicillin of germ-free chicks and of chicks 376 with a defined intestinal flora. Br J Nutr 1959;13(1):78-84.

377 (32) Cole JR, Jr., Boyd FM. Fat absorption from the small intestine of gnotobiotic chicks. Appl 378 Microbiol 1967 Sep;15(5):1229-34.

379 (33) Jukes TH. Antibiotics in Animal Feeds and Animal Production. BioScience 1972;22(9):526- 380 34.

381 (34) Vijay-Kumar M, Aitken JD, Carvalho FA, Cullender TC, Mwangi S, Srinivasan S, et al. 382 Metabolic syndrome and altered gut microbiota in mice lacking Toll-like receptor 5. Science 383 2010 Apr 9;328(5975):228-31.

384 (35) Backhed F, Manchester JK, Semenkovich CF, Gordon JI. Mechanisms underlying the 385 resistance to diet-induced obesity in germ-free mice. Proc Natl Acad Sci U S A 2007 Jan 386 16;104(3):979-84.

387 (36) Angelakis E, Armougom F, Million M, Raoult D. The relationship between gut microbiota 388 and weight gain in humans. Future Microbiol 2012 Jan;7(1):91-109.

389 (37) Robinson CJ, Young VB. Antibiotic administration alters the community structure of the 390 gastrointestinal micobiota. Gut Microbes 2010 Jul;1(4):279-84.

391 (38) Stsepetova J, Sepp E, Kolk H, Loivukene K, Songisepp E, Mikelsaar M. Diversity and 392 metabolic impact of intestinal Lactobacillus species in healthy adults and the elderly. Br J 393 Nutr 2011 Apr;105(8):1235-44.

394 (39) Santacruz A, Collado MC, Garcia-Valdes L, Segura MT, Martin-Lagos JA, Anjos T, et al. Gut 395 microbiota composition is associated with body weight, weight gain and biochemical 396 parameters in pregnant women. Br J Nutr 2010 Jul;104(1):83-92.

397 (40) Hildebrandt MA, Hoffmann C, Sherrill-Mix SA, Keilbaugh SA, Hamady M, Chen YY, et al. 398 High-fat diet determines the composition of the murine gut microbiome independently of 399 obesity. Gastroenterology 2009 Nov;137(5):1716-24.

400 (41) Membrez M, Blancher F, Jaquet M, Bibiloni R, Cani PD, Burcelin RG, et al. Gut microbiota 401 modulation with norfloxacin and ampicillin enhances glucose tolerance in mice. FASEB J 402 2008 Jul;22(7):2416-26.

403 (42) Bethell DB, Hien TT, Phi LT, Day NP, Vinh H, Duong NM, et al. Effects on growth of single 404 short courses of fluoroquinolones. Arch Dis Child 1996 Jan;74(1):44-6.

405 (43) Takizawa T, Hasimoto K, Itoh N, Yamashita S, Owen K. A comparative study of the repeat 406 dose toxicity of grepafloxacin and a number of other fluoroquinolones in rats. Hum Exp 407 Toxicol 1999 Jan;18(1):38-45.

408 (44) Dubray C, Ibrahim SA, Abdelmutalib M, Guerin PJ, Dantoine F, Belanger F, et al. Treatment 409 of severe malnutrition with 2-day intramuscular ceftriaxone vs 5-day amoxicillin. Ann Trop 410 Paediatr 2008 Mar;28(1):13-22.

411 (45) Pirzada OM, McGaw J, Taylor CJ, Everard ML. Improved lung function and body mass index 412 associated with long-term use of Macrolide antibiotics. J Cyst Fibros 2003 Jun;2(2):69-71.

413 (46) Saiman L, Marshall BC, Mayer-Hamblett N, Burns JL, Quittner AL, Cibene DA, et al. 414 Azithromycin in patients with cystic fibrosis chronically infected with Pseudomonas 415 aeruginosa: a randomized controlled trial. JAMA 2003 Oct 1;290(13):1749-56.

416 (47) Saiman L, Mayer-Hamblett N, Anstead M, Lands LC, Kloster M, Goss CH, et al. Open-label, 417 follow-on study of azithromycin in pediatric patients with CF uninfected with Pseudomonas 418 aeruginosa. Pediatr Pulmonol 2012 Jul;47(7):641-8.

419 (48) Saiman L, Anstead M, Mayer-Hamblett N, Lands LC, Kloster M, Hocevar-Trnka J, et al. Effect 420 of azithromycin on pulmonary function in patients with cystic fibrosis uninfected with 421 Pseudomonas aeruginosa: a randomized controlled trial. JAMA 2010 May 5;303(17):1707- 422 15.

423 (49) Clement A, Tamalet A, Leroux E, Ravilly S, Fauroux B, Jais JP. Long term effects of 424 azithromycin in patients with cystic fibrosis: A double blind, placebo controlled trial. Thorax 425 2006 Oct;61(10):895-902.

426 (50) Southern KW, Barker PM, Solis-Moya A, Patel L. Macrolide antibiotics for cystic fibrosis. 427 Cochrane Database Syst Rev 2011;(12):CD002203.

428 (51) Mansi Y, Abdelaziz N, Ezzeldin Z, Ibrahim R. Randomized controlled trial of a high dose of 429 oral erythromycin for the treatment of feeding intolerance in preterm infants. Neonatology 430 2011;100(3):290-4.

431 (52) Lane JA, Murray LJ, Harvey IM, Donovan JL, Nair P, Harvey RF. Randomised clinical trial: 432 Helicobacter pylori eradication is associated with a significantly increased body mass index 433 in a placebo-controlled study. Aliment Pharmacol Ther 2011 Apr;33(8):922-9.

434 (53) Kamada T, Hata J, Kusunoki H, Ito M, Tanaka S, Kawamura Y, et al. Eradication of 435 Helicobacter pylori increases the incidence of hyperlipidaemia and obesity in peptic ulcer 436 patients. Dig Liver Dis 2005 Jan;37(1):39-43.

437 (54) Azuma T, Suto H, Ito Y, Muramatsu A, Ohtani M, Dojo M, et al. Eradication of Helicobacter 438 pylori infection induces an increase in body mass index. Aliment Pharmacol Ther 2002 439 Apr;16 Suppl 2:240-4.

440 (55) Patterson PR. Minocycline in the antibiotic regimen of cystic fibrosis patients: weight gain 441 and clinical improvement. Clin Pediatr (Phila) 1977 Jan;16(1):60-3.

442 (56) Edlund C, Barkholt L, Olsson-Liljequist B, Nord CE. Effect of vancomycin on intestinal flora 443 of patients who previously received antimicrobial therapy. Clin Infect Dis 1997 444 Sep;25(3):729-32.

445 (57) Lund B, Edlund C, Barkholt L, Nord CE, Tvede M, Poulsen RL. Impact on human intestinal 446 microflora of an Enterococcus faecium probiotic and vancomycin. Scand J Infect Dis 447 2000;32(6):627-32.

448 (58) Van der Auwera P, Pensart N, Korten V, Murray BE, Leclercq R. Influence of oral 449 glycopeptides on the fecal flora of human volunteers: selection of highly glycopeptide- 450 resistant enterococci. J Infect Dis 1996 May;173(5):1129-36.

451 (59) Yap IK, Li JV, Saric J, Martin FP, Davies H, Wang Y, et al. Metabonomic and microbiological 452 analysis of the dynamic effect of vancomycin-induced gut microbiota modification in the 453 mouse. J Proteome Res 2008 Sep;7(9):3718-28.

454 (60) Swedish Study Group. A randomized multicenter trial to compare the influence of cefaclor 455 and amoxycillin on the colonization resistance of the digestive tract in patients with lower 456 respiratory tract infection. Infection 1991 Jul;19(4):208-15.

457 (61) Brismar B, Edlund C, Nord CE. Impact of cefpodoxime proxetil and amoxicillin on the 458 normal oral and intestinal microflora. Eur J Clin Microbiol Infect Dis 1993 Sep;12(9):714-9.

459 (62) Floor M, van AF, Rozenberg-Arska M, Visser M, Kolsters A, Beumer H, et al. Effect of 460 loracarbef and amoxicillin on the oropharyngeal and intestinal microflora of patients with 461 bronchitis. Scand J Infect Dis 1994;26(2):191-7.

462 (63) Stark CA, Adamsson I, Edlund C, Sjosted S, Seensalu R, Wikstrom B, et al. Effects of 463 omeprazole and amoxycillin on the human oral and gastrointestinal microflora in patients 464 with Helicobacter pylori infection. J Antimicrob Chemother 1996 Dec;38(6):927-39.

465 (64) Schumann A, Nutten S, Donnicola D, Comelli EM, Mansourian R, Cherbut C, et al. Neonatal 466 antibiotic treatment alters gastrointestinal tract developmental gene expression and 467 intestinal barrier transcriptome. Physiol Genomics 2005 Oct 17;23(2):235-45.

468 (65) Million M, Angelakis E, Paul M, Armougom F, Leibovici L, Raoult D. Comparative meta- 469 analysis of the effect of Lactobacillus species on weight gain in humans and animals. 470 Microb Pathog 2012 Aug;53(2):100-8. 471 472

473

474 Table 1. Baseline characteristics of patients and results of the weight change study

Amoxicillin (n=57) Vancomycin (n=41) P-value (test)

Sexe (Male) 46/57 28/41 0.24 (Fisher)

Age 61.9 ± 12.1 65.3±12.8 0.24 (Student)

Surgery* 34/54 18/38 0.20 (Fisher)

Baseline BMI 26.8 ± 5.1 26.2 ± 4.8 0.53 (Student)

475 * Data unavailable for 6 patients

476

477

478

479

480

481 Table 2. Weight change and acquired obesity according to antibiotics

Vanco Amoxicillin P-value (a) OR

Former study (n=39) (20)

%deltaBMI(b) 15.5 (0 to 17.7) 1.31 (-1.33 to 8) 0.15

BMI10% 7/11 7/28 0.03 5.25 (1.17 - 23.46)

Acquired obesity 3/11 2/28 0.12 4.87 (0.68 - 34.50)

Our study (n=98)

%deltaBMI 0 (-5.15 to 3.17) -0.48 (-4.82 to -0.48) 0.35

BMI10% 5/41 1/57 0.047 7.63 (0.85 - 68.10)

Acquired obesity 4/41 0/57 0.029 undefined

All (n = 137)

%deltaBMI 0.52 (-5.15 to 7.5) 0.56 (-4.49 to 2.63) 0.28

BMI10% 12/52 8/85 0.03 2.85 (1.07 - 7.54)

Acquired obesity 7/52 2/85 0.0159 6.37 (1.27 - 32.00)

(a) Mann-Withney test for continuous data, Fisher bilateral exact test for

proportions (b) median (IQR) - calculated with epi-info v7

482 Table 3. Studies reporting a significant weight gain associated with antibiotic administration in humans

Antibiotics Indication Reference

All Infections in the early life Trasande, Int J Obesity, 2012 (10)

Glycopeptides

Vancomycin Endocarditis Thuny, PlosOne, 2010 (14)

Betalactamines

Amoxicillin Malnutrition Dubray, Ann Trop Paediatr, 2008 (44)

Ceftriaxone Malnutrition Dubray, Ann Trop Paediatr, 2008 (44)

Macrolides

Azithromycin Cystic fibrosis Pirzada, J Cystic fibrosis, 2003 (45)

Saiman, JAMA, 2003 – Saiman, JAMA, 2010 -

Saiman, Pediatr Pulmonol, 2012 (46-48)

Clement, Thorax, 2006 (49)

Southern, Cochrane, 2011 (50)

Erythromycin Neonatology Mansi, Neonatology, 2012 (51)

Clarithromycin Eradication of Helicobacter pylori Lane, Aliment Pharmacol Ther, 2011 (52)

Kamada, Dig Liver Dis, 2005 (53)

Azuma, Aliment Pharmacol Therap, 2002 (with

amoxicillin) (54)

Tetracyclines

Minocyclin Cystic fibrosis Patterson, Clin Pediatr (Phila), 1977 (55)

Chlortetracycline Neonatology Robinson, Lancet, 1952 (7)

Chlortetracycline Antibiotic prophylaxis on Haight, J Nutr, 1955 (5)

immune response

483

484

485 Table 4. Effect of vancomycin and amoxicillin on gut microbiota in the literature

Host Method Ref.

Vancomycin

Overgrowth of lactobacilli (a) (and pediococci) Humans Cultivation (56-58)

Increase of Lactobacillaceae (a) Humans Cloning (37)

Increase of Proteobacteria (a) Humans Cloning (37)

Decrease of enterococci Humans Cultivation (56;57)

Decrease of staphylococci Humans Cultivation (58)

Strong suppression or elimination of Bacteroides (b) Humans Cultivation (56;57)

Decrease of clostridia and bifidobacteria (b) Humans Cultivation (57)

Decrease of Firmicutes , especially C. leptum, C. coccoides, C. symbosium Mouse PCR-DGGE (59)

Amoxicillin

Increase of aerobic Gram-negative rods like enterobacteria other than E. coli Humans Cultivation (60-63) (Klebsiella , Enterobacter ) Increase of anaerobic Gram-positive rods Humans Cultivation (60)

Increase of Bacteroides (b) Humans Cultivation (60)

Decrease of streptococci and staphylococci Humans Cultivation (61)

Almost suppression of lactobacillus (a) Rat pups Cultivation (64)

Depletion of enterococci Rat pups Cultivation (64)

Depletion of enterobacteriacae (b) Rat pups Cultivation (64)

486 (a) Specific lactobacilli (members of the Lactobacillaceae family) have been associated with obesity (25), correlated with BMI (38) and

487 associated with weight gain (65). (b) Specific Proteobacteria , enterobacteriaceae and Escherichia coli have been linked with weight gain

488 and obesity whereas Bacteroidetes , Bacteroides and bifidobacteria have been linked to a statistical anti-obesity effect (36).

489

490 Figure legends

491 Figure 1. Global modification of the gut microbiota durting protracted amoxicillin or

492 vancomycin treatment

493 * p < 0.05, **p < 0.005, ***p < 0.0005 compared to controls

494 Figure 2. Principal component analysis identifying potential bacterial predictors of weight

495 gain after amoxicillin or vancomycin treatment

496 Analysis was done with XLSTAT 2012 software (Addinsoft, paris, France)

497

498 Figure 1.

499

500

501 Figure 2.

502

503 -

Conclusion Générale et Perspectives

Notre travail a permis, pour la première fois à notre connaissance, d’identifier quelles

étaient les altérations du microbiote digestif associées à l’obésité de façon reproductible indépendamment des études et des pays par la méthode de méta-analyse. De plus, nous avons confirmé que l’analyse au niveau de l’espèce est primordial dans l’identification de altérations associées à l’obésité et avons retrouvé une espèce de Lactobacillus associée de manière dose- dépendante à l’indice de masse corporelle. Cela a d éjà été confirmé par une autre équipe pour une autre espèce de Lactobacillus . Cependant, il n’est pas possible d’éliminer à ce jour un facteur confondant comme le régime qui pourrait être responsable à la fois de la prise de poids et de la modulation du mi crobiote digestif incluant l’augmentation de la concentration de certains Lactobacillus . Dans le futur, d’autres études observationnelles associant des approches de culture et de biologie moléculaire ciblant des genres d’intérêt mais discriminatif au nivea u de l’espèce voire de la souche permettront de confirmer la reproductibilité de nos résultats et d’identifier de nouveaux candidats microbiens pour l’obésité.

Enfin, notre travail a permis d’éclaircir le rôle de la manipulation du microbiote digestif su r l’obésité. Nous avons montré que l’effet des probiotiques contenant des

Lactobacillus sur le poids dépendait de l’espèce et de l’hôte. Basé sur la littérature, nous

émettons l’hypothèse que l’administration des probiotiques a un impact majeur sur le poid s à l’âge adulte quand ils sont administrés dans les premiers mois de vie pendant la formation du microbiote digestif, et cela pourrait être associée à une obésité acquise comme cela a été montré pour les antibiotiques. Cependant, afin de confirmer un rôle causal chez l’homme, suggéré par les expériences de transplantation de microbiote chez l’animal, de nouvelles

études sont nécessaires. Enfin, devant toutes ces données évoquant un rôle causal de certaines souches de probiotiques sur l’obésité, nous suggér ons que le poids soit systématiquement

151

évalué dans les études à venir sur les probiotiques afin de ne pas négliger une augmentation de l’obésité associée à la commercialisation massive de ces produits. En conséquence, les

études sur les probiotiques sont s ujettes à des biais de publication importants, c’est pourquoi une politique active sur les conflits d'intérêts des études sur les probiotiques devrait être encouragée.

152

References

1. Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: human gut microbes associated with obesity. Nature 2006;444(7122):1022-1023.

2. Armougom F, Henry M, Vialettes B, Raccah D, Raoult D. Monitoring bacterial community of human gut microbiota reveals an increase in Lactobacillus in obese patients and Methanogens in anorexic patients. PLoS One 2009;4(9):e7125.

3. Stokstad EL, Jukes TH, . The multiple nature of the animal protein factor. J Biol Chem 1949;180(2):647-654.

4. Raoult D. Human microbiome: take-home lesson on growth promoters? Nature 2008;454(7205):690-691.

5. Stsepetova J, Sepp E, Kolk H, Loivukene K, Songisepp E, Mikelsaar M. Diversity and metabolic impact of intestinal Lactobacillus species in healthy adults and the elderly. Br J Nutr 2011;105(8):1235-1244.

6. Robinson EL, Thompson WL. Effect on weight gain of the addition of Lactobacillus acidophilus to the formula of newborn infants. J Pediatr 1952;41(4):395-398.

7. Morelli L. Million et al. "Comparative meta-analysis of the effect of Lactobacillus species on weight gain in humans and animals." Letter to editors. Microb Pathog 2013;55:51.

8. Trasande L, Blustein J, Liu M, Corwin E, Cox LM, Blaser MJ. Infant antibiotic exposures and early-life body mass. Int J Obes (Lond) 2013;37(1):16-23.

9. Besselink MG, van Santvoort HC, Buskens E et al. Probiotic prophylaxis in predicted severe acute pancreatitis: a randomised, double-blind, placebo-controlled trial. Lancet 2008;371(9613):651-659.

10. Murphy EF, Cotter PD, Hogan A et al. Divergent metabolic outcomes arising from targeted manipulation of the gut microbiota in diet-induced obesity. Gut 2013;62(2):220-226.

11. Archambaud C, Nahori MA, Soubigou G et al. Impact of lactobacilli on orally acquired listeriosis. Proc Natl Acad Sci U S A 2012;109(41):16684-16689.

12. Cho I, Yamanishi S, Cox L et al. Antibiotics in early life alter the murine colonic microbiome and adiposity. Nature 2012;488(7413):621-626.

13. Thuny F, Richet H, Casalta JP, Angelakis E, Habib G, Raoult D. Vancomycin treatment of infective endocarditis is linked with recently acquired obesity. PLoS One 2010;5(2):e9074.

153

ANNEXES

154

Article XIII :

Microbial Culturomics: Paradigm shift in the human

gut microbiome study

Jean-Christophe Lagier, Fabrice Armougom, Matthieu Million, Perrine Hugon, Isabelle Pagnier, Catherine Robert, Fadi Bittar, Ghislain Fournous, Gregory Gimenez, Marie Maraninchi, Jean- François Trape, EugeneV. Koonin, Bernard La Scola, Didier Raoult.

Published in Clin Microbiol Infect 2012 Dec;18(12):1185-93. (IF 4.54)

155

ORIGINAL ARTICLE BACTERIOLOGY

Microbial culturomics: paradigm shift in the human gut microbiome study

J.-C. Lagier1, *, F. Armougom 1, *, M. Million 1, P. Hugon 1, I. Pagnier 1, C. Robert 1, F. Bittar 1, G. Fournous 1, G. Gimenez 1, M. Maraninchi 2, J.-F. Trape 3, E. V. Koonin 4, B. La Scola 1 and D. Raoult 1 1) Aix Marseille Universite´, URMITE, UM63, CNRS 7278, IRD 198, INSERM 1095, 2) Service de Nutrition, Maladies Me´taboliques et Endocrinologie, UMR-INRA U1260, CHU de la Timone, Marseille, France, 3) IRD, UMR CNRS 7278-IRD 198, Route des Pe`res Maristes, Dakar, Se´ne´gal and 4) National Centre for Biotechnology Information, National Library of Medicine, National Institutes of Health, Bethesda, MD, USA

Abstract

Comprehensive determination of the microbial composition of the gut microbiota and the relationships with health and disease are major challenges in the 21st century. Metagenomic analysis of the human gut microbiota detects mostly uncultured bacteria. We studied stools from two lean Africans and one obese European, using 212 different culture conditions (microbial culturomics), and tested the colonies by using mass spectrometry and 16S rRNA amplification and sequencing. In parallel, we analysed the same three samples by pyrosequencing 16S rRNA amplicons targeting the V6 region. The 32 500 colonies obtained by culturomics have yielded 340 species of bacteria from seven phyla and 117 genera, including two species from rare phyla (Deinococcus-Thermus and Synergistetes, five fungi, and a giant virus (Senegalvirus). The microbiome identified by culturomics included 174 species never described previously in the human gut, including 31 new species and genera for which the genomes were sequenced, generating c. 10 000 new unknown genes (ORFans), which will help in future molecular studies. Among these, the new species Microvirga massiliensis has the largest bacterial genome so far obtained from a human, and Senegalvirus is the largest virus reported in the human gut. Concurrent metagenomic analysis of the same samples produced 698 phylotypes, including 282 known species, 51 of which overlapped with the microbiome identified by culturomics. Thus, culturomics complements metagenomics by overcoming the depth bias inherent in metagenomic approaches.

Keywords: Culturomics, gut microbiota, MALDI-TOF MS, metagenomic analysis, uncultured bacteria Original Submission: 20 August 2012; Revised Submission: 28 August 2012; Accepted: 29 August 2012 Editor: G. Greub Article published online: 4 September 2012 Clin Microbiol Infect 2012; 18: 1185–1193 10.1111/1469-0691.12023 metagenomic analysis have dramatically expanded the known Corresponding author: D. Raoult, Aix-Marseille Universite´, UR- diversity of the human gut microbiome [4,5,7,8]. It is commonly MITE, UMR CNRS 7278, IRD 198, INSERM U1095, Faculte´ de Me´decine, 27 Bd Jean Moulin, Cedex 5, 13385 Marseille, France accepted that c. 80% of the bacterial species found by molecular E-mail: [email protected] tools in the human gut are uncultured or even unculturable [1]. *These authors contributed equally to this work. However, several drawbacks of the current metagenomic approaches, including major discrepancies among different Introduction studies, apparently reflect biases of the employed techniques. In particular, sequence-based techniques miss clinically relevant minority populations, including potentially pathogenic bacteria, The composition of the human gut microbiome, the determina- such as Salmonella Typhi, Tropheryma whipplei, and Yersinia tion of which represents a major challenge in the 21st century enterocolitica, that may be present at concentrations lower than [1], has been studied with different tools, leading to increasingly 10 5/mL; this major problem is known as the depth bias. complex results [2–6]. The first approach used to study the gut Recently, there has been a renewed interest in culture microbiota employed microbial culture [2]. Subsequent studies methods for ‘non-cultivable’ species [3,9]. One of the grid- that involved amplification and sequencing of 16S rRNA and later locks of the traditional bacteriological culture methods has

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases 1186 Clinical Microbiology and Infection, Volume 18 Number 12, December 2012 CMI

been recently overcome by advances in mass spectrometry and sterile stool extract to mimic the natural environment [9]. (MS) techniques, which can accurately and rapidly identify Moreover, with the aim of selecting a minority population, we microorganisms with matrix-assisted laser desorption ioniza- used antibiotics, both active and passive filtration, and bacterio- tion time-of-flight (MALDI-TOF), allowing rapid screening of phages. We used MALDI-TOF MS to quickly identify a maximum large numbers of colonies [10]. In this study, we combined of colonies. When the strains remained unrecognized, the 16S the MS approach with extensive sequencing and exploration rRNA gene was sequenced. As previously described, a threshold of the potential of numerous known and new culture meth- similarity of >98.7% was chosen to define a new bacterial species ods to introduce culturomics as a major complement to me- [11]. The same three stool samples were tested by pyrosequenc- tagenomics in the study of the human gut microbiome. ing of a 16S rRNA amplicon targeting the V6 region, the most variable region, as previously described [7] NCBI accession number=SRA049748. The new bacterial genera and species were Materials and Methods sequenced with a paired-end strategy for high-throughput pyrosequencing on the 454-Titanium instrument. Senegalvirus All of the data are detailed in the Supporting information. We was sequenced with the Roche 454 FLX-Titanium platform. used two African stools, both from healthy young males living in rural Senegal, and a stool from a French obese individual with a Results body mass index of 48.2 kg/m 2. Each patient’s consent was obtained, and the study was approved by the local ethics committee of IFR48 (agreement number 09-022; Marseille, France). We Proof-of-concept designed 212 culture conditions, using variable physicochemical We studied stool samples from two young lean Africans conditions, pre-incubation in blood culture bottles, rumen fluid from a rural environment in Senegal ( Fig. 1) and one obese

(d) (a)

(b)

(c)

FIG. 1. The source of material for culturomics and the record-breaking virus and bacterium from the human gut. (a) The geographical locations of the Dielmo and N’diop villages (Sources: Wikitravel.org and Google Earth) from which the two African stool samples analysed in this work were obtained. (b) Electronmicrograph of the giant Senegalvirus, which was isolated from a stool sample of an individual from N’Diop. (c) Comparison of the Senegalvirus genome with the genomes of related giant viruses, Marseillevirus and Lausannevirus. (d) Electronmicrograph of Microvirga massiliensis (the bacterium with the largest genome ever isolated from humans), which was isolated from the Dielmo stool sample.

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases, CMI, 18 , 1185–1193 CMI Lagier et al. Culturomics revolutionizes gut 1187

French individual, using 212 different conditions, including With the two other stool samples, only those culture amoebal co-culture. For the first African sample, 56 different conditions that proved to be efficient with the first sample culture methods were applied, including different physico- were used again, and many additional culture conditions chemical conditions and the addition of specific nutrients or were applied to maximize the chance of isolation of new inhibitors (Table S1). Using these approaches, we isolated species (Table S1). This optimized approach yielded 191 dis- 3000 colonies, which were subjected to MALDI-TOF MS tinct bacterial isolates, including two new genera and six new analysis for rapid identification of microbial species [10]. This species from the obese individual’s stool sample; the largest analysis resulted in the identification of 99 bacterial species, number of bacteria ever identified in a single stool (219 bac- 42 of which had never been found in the human gut teria and five fungi, including three new genera and 18 new (Fig. S1), and two of which had never been described bacterial species) were isolated from the second African (Table 1 Fig. 2). sample (Tables 1 and 2; Fig. 2; Table S2) [12–16].

TABLE 1. Characteristics of the 23 new bacterial species and genera cultured from the Senegalese stools [12–16]

Diameter Genome size ORFan Estimated GC Genbank Phylum Initial culture conditions (lm) (EM) estimate (Mb) (%) content (%) no.

N’Diop stool sample New species Oceanobacillus massiliensis Firmicutes Filtration brain–heart infusion 5% sheep 0.70 3.6 5.6 41 HQ586877 blood 0.45-lm aerobe, 37 °C Bacillus timonensis Firmicutes Brain–heart infusion + sheep blood 5%, 0.66 4.7 6.8 38.3 JF824810 aerobe, 37 °C Dielmo stool sample New species Kurthia massiliensis Firmicutes CNA aerobe 2.5% CO 2, 37 °C 1.08 3.3 11.9 39.7 JF824795 Kurthia senegalensis Firmicutes Filtration 5% sheep blood agar 1.2- lm 1.03 2.9 11.3 39.6 JF824796 aerobe, 37 °C Kurthia timonensis Firmicutes HTM, aerobe, 2.5% CO 2, 37 °C 0.94 4.1 16.2 39 JF824797 Anaerococcus senegalensis Firmicutes Brucella anaerobe, 37 °C 0.68 1.8 3 28.5 JF824805 Paenibacillus senegalensis Firmicutes Schaedler kanamycin vancomycin, aerobe, 0.66 5.7 10.7 48.3 JF824808 37 °C Bacillus massiliosenegalensis Firmicutes 5% sheep blood agar, aerobe, 28 °C 0.64 4.9 7.7 37.7 JF824800 Clostridium senegalense Firmicutes Inoculation in blood culture bottle for 1.05 3.9 11.5 29.3 JF824801 5 days with 5 mL of sheep blood, 5% sheep blood agar, anaerobe, 37 °C Peptoniphilus senegalensis Firmicutes Inoculation in blood culture bottle for 0.64 1.8 3.9 32.5 JF824803 10 days with 5 mL of sheep blood, 5% sheep blood agar, anaerobe, 37 °C Peptoniphilus timonensis Firmicutes Inoculation in blood culture bottle anaerobe 0.91 1.7 9.3 31 JN657222 for 14 days with 8 mL of rumen fluid, 5% sheep blood agar, anaerobe, 37 °C Ruminococcus massiliensisa Firmicutes Inoculation in blood culture bottle anaerobe 0.96 5.1 25 57 JN657221 for 14 days with 8 mL of rumen fluid 5% sheep blood agar, anaerobe, 37 °C Alistipes senegalensis Bacteroidetes Schaedler kanamycin vancomycin, anaerobe, 0.53 4 3.8 58.3 JF824804 37 °C Alistipes timonensis Bacteroidetes Inoculation in blood culture bottle anaerobe 0.62 3.5 2.9 58.8 JF824799 for 5 days, Schaedler kanamycin vancomycin, anaerobe 37 °C Cellulomonas massiliensis Actinobacteria Passive filtration with Leptospira broth, 5% 0.48 3.4 7.9 73.9 JN657218 sheep blood agar, aerobic atmosphere, 37 °C Aeromicrobium massiliense Actinobacteria 5% sheep blood agar, aerobe, 37 °C 1.04 3.3 10.5 72.6 JF824798 Brevibacterium senegalense Actinobacteria Brucella, aerobe, 37 °C 0.68 3.4 9.6 69.9 JF824806 Enterobacter massiliensis Proteobacteria Phage T1 + T4, then 5% sheep blood agar, 1.02 4.9 3 55.4 JN657217 aerobe, 37 °C Herbaspirillum massiliense Proteobacteria Passive filtration with Leptospira broth, 5% 0.44 4.2 8.1 59.7 JN657219 sheep blood agar, aerobic atmosphere, 37 °C Microvirga massiliensis Proteobacteria MOD 2, aerobe, 37 °C 2.28 9.35 24.1 59.2 JF824802 New genera Dielma fastidiosa Firmicutes Inoculation in blood culture bottle anaerobe 0.59 3.6 10.5 40 JF824807 for 10 days, brain–heart infusion, anaerobe, 37 °C Senegalemassilia anaerobia Actinobacteria Inoculation in blood culture bottle anaerobe 0.70 2.3 6.3 61.8 JF824809 for 5 days. 5% sheep blood aga,r anaerobe, 37 °C Timonella senegalensis Actinobacteria Inoculation in blood culture bottle anaerobe 0.59 3 11.9 61.3 JN657220 for 14 days with 8 mL of rumen fluid, 5% sheep blood agar, anaerobe, 37 °C

EM : Electron Microscopy. aThe characterization of this bacterial species was not performed, because the impossibility of subculture.

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases, CMI, 18 , 1185–1193 1188 Clinical Microbiology and Infection, Volume 18 Number 12, December 2012 CMI

FIG. 2. Phylogenetic tree representing the new bacterial species and genera obtained by culturomics. Red labels indicate the new species found in the Senegalese patients and obese patient. Dark labels indicate the closest neighbour species deﬁned as Isolates and Type in the RDP-II database. Tree branches in red, dark green, purple and blue represent the phyla Bacteroidetes, Proteobacteria, Actinobacteria , and Firmicutes, respectively. Green squares denote new species found in the obese patient. Dark circles indicate that the genome sequence is available for the closest neighbour species.

TABLE 2. Characteristics of the eight new bacterial species and genera cultured from the stools of the obese individual

Diameter Genome size ORFan Estimated GC Genbank Phylum Initial culture conditions (lm) (EM) estimate (Mb) (%) content (%) no.

New species Anaerococcus obesiensis Firmicutes Inoculation in blood culture bottle with 0.71 2.05 3.7 30.1 JN837490 thioglycolate for 4 days, 5% sheep blood agar, anaerobe, 37 °C Brevibacillus massiliensis Firmicutes M17, aerobe, 37 °C 0.73 5.1 7.2 53 JN837488 Peptoniphilus grossensis Firmicutes Inoculation in blood culture bottle for 0.77 2.1 5.5 34.5 JN837491 26 days with rumen and sheep blood, 5% sheep blood agar, anaerobe, 37 °C Peptoniphilus obesiensis Firmicutes Inoculation in blood culture bottle for 0.85 1.77 4.7 30.4 JN837495 26 days with rumen and sheep blood, 5% sheep blood agar, anaerobe, 37 °C Alistipes obesiensis Bacteroidetes Inoculation in blood culture bottle for 0.61 3.1 7.3 58.5 JN837494 11 days with rumen, 5% sheep blood agar, anaerobe, 37 °C Actinomyces grossensis Actinobacteria Inoculation in blood culture bottle with 0.49 1.87 5.2 56 JN837492 thioglycolate for 4 days, 5% sheep blood agar, anaerobe, 37 °C New genera Enorma massiliensis Actinobacteria Inoculation in blood culture bottle with 0.57 2.3 7.9 61.8 JN837493 thioglycolate for 4 days, 5% sheep blood agar, anaerobe, 37 °C Kallipyga massiliensis Firmicutes Inoculation in blood culture bottle for 0.67 1.77 6.3 51.4 JN837487 26 days with rumen and sheep blood, 5% sheep blood agar, anaerobe, 37 °C

EM : Electron Microscopy.

Optimization of rapid screening by MALDI-TOF MS ticular infections. Therefore, for the second and the third With a low level of operator training, MALDI-TOF MS allows samples, we developed an enrichment culture technique rapid discrimination between identified bacteria (present in the involving several days of direct pre-incubation of stools in an current data bank) and unidentified bacteria. Indeed, subsequent aerobic or anaerobic blood culture bottle, allowing the to the analysis of the first sample, with the two other tested growth of 29 bacterial species that were not detected by samples we were able to identify a significantly greater number standard axenic culture (including 24 anaerobic species). This of colonies: without automatic colony picking, we tested a total approach yielded three new genera and three new species. of 29 500 different colonies from these two samples, as com- Addition of sheep blood to the blood culture bottle allowed pared with the 3000 colonies tested from the first sample. In us to identify three additional species. addition to the better training of the operators, as and when necessary, each previously unknown bacterial spectrum was New culture conditions added to our data bank, facilitating the screening for further To increase the growth of bacteria under culture conditions studies. MS, which obviates the need for both the time-consum- that mimic their natural environment, and drawing from pre- ing Gram-staining procedure and the usual biochemical tests, vious studies on environmental bacteria [9], we used sterile seems to be the current method of choice for the identification rumen fluid [3] (Fig. S2) and sterile fresh human stools with of microorganisms, and has the potential to supplant traditional or without pre-incubation in blood culture bottles. This microbiological methods [10]. Indeed, with the third stool sam- approach allowed us to isolate 17 strains that were not ple, we tested, under eight different culture conditions, 50–100 recovered in classic axenic conditions, including two new colonies that were indistinguishable in appearance. This experi- genera, three new species, and one species of the Deinococ- ment allowed us to identify several species, notably those from cus-Thermus phylum that has not been previously cultured the genus Enterococcus , for which identification in routine bacte- from human clinical samples [18]. riology is mainly based on colony morphology. Finally, in an effort to obtain fastidious bacteria by amoebal co-culture with Acanthamoeba polyphaga [19], we identi- Eliminating the predominant population fied from the three stool samples four additional bacterial The high concentration of bacteria in the human gut (10 12 to species that have not been detected by axenic culture. Ser- 10 11 bacteria per gram of stools) [1] hampers non-selective endipitously, we also isolated, from the first African stool, a culture analysis. Therefore, we used antibiotics in culture new giant virus strain, which we named Senegalvirus and that media to eliminate sensitive organisms and thus facilitate the has the largest genome among the viruses isolated from identification of resistant ones. We developed different strate- humans (Figs 1 and S3), with the exception of two reports gies to extend the use of ‘classic selective media’. First, to of mimivirus detection [19]. The isolation of a non-filterable identify new proteobacteria, we had to develop alternative giant virus from a human stool indicates that giant viruses strategies, because Escherichia coli is the overwhelmingly domi- could constitute a component of the gut microbiome that is nant bacterial species in the human gut under aerobic condi- missed by metagenomic studies with 0.22- lm filters [6,20]. tions. We used a cocktail of E. coli lytic bacteriophages [17] that allowed us to clear the culture of E. coli and to identify Microbial culturomics: a general perspective an unknown enterobacterial species ( Enterobacter massiliensis ) Only 45 bacterial species obtained in culture in the present that was not detected by classic axenic culture. work were common to all three analysed stool samples and Otherwise, an effective method to remove the major bac- could be named ‘the culturomics core microbiome’. The terial population was active filtration with successive mem- majority of the isolated species (63.6%) were cultured from branes (from 5 to 0.2 lm); this procedure allowed us to only one stool sample, indicative of large inter-individual identify eight new bacterial species. Finally, using the physical diversity of the human gut microbiome (Fig. S4). Thus, characteristics of certain bacteria, we applied passive filtration, microbial culturomics allowed the detection of numerous which resulted in the identification, in the second African new bacteria from each tested sample (Tables 1 and 2). stool sample, of three motile bacteria that have not been pre- Although we used a total of 212 different culture condi- viously detected in the human gut, including two new species tions for the three stool samples, 100% of the species (Table 1 and Table S1). grew under only 70 culture conditions, and 73% of the species were identified with only 20 conditions (Fig. S5; Enrichment of samples in blood culture bottles Table S3). These results provide guidance for future cul- Incubation of clinical samples in blood culture bottles is turomics studies, which will benefit from using the set of known to promote the growth of Kingella kingae in osteoar- conditions shown to be efficient in this study before

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases, CMI, 18 , 1185–1193 1190 Clinical Microbiology and Infection, Volume 18 Number 12, December 2012 CMI

developing new culture approaches. Although a limited ria ranged from 1.7 to 9.35 Mb. Among the predicted gene number of culture conditions allowed us to grow the products of the new genomes, 2.9–24.1% had no readily majority of the common bacterial species, a more thor- detectable homologues, i.e. they represented ORFans. Alto- ough approach that used many ‘exotic’ culture conditions gether, the present study yielded c. 10 000 previously substantially expanded the repertoire by allowing the isola- unknown genes. tion of less abundant bacteria. Comparison with 16S rRNA sequencing ‘Giant’ bacteria and giant virus The pyrosequencing that was performed as part of this The typical diameter of the isolated bacteria ranged from 0.5 study identified 126, 138 and 157 phylotypes correspond- to 1.5 lm [21]. However, the largest isolated bacterium, Mi- ing to known species from the three stools, respectively crovirga massiliensis, reached 2.28 lm according to transmis- (Tables S4, S5 and S6). For the three stools taken sion electron microscopy, and was also shown to possess together, the microbial culturomics approach yielded 340 the largest genome (9.35 Mb) of any bacterium previously bacterial species from seven phyla and 117 genera, obtained from a human sample (Tables 1 and 2; Fig. 1). The whereas pyrosequencing identified 282 species from six giant (194 nm in diameter) Senegalvirus isolated by amoebal phyla and 91 genera. However, a dramatic difference was co-culture is the first giant virus ever isolated from the observed with culturomics: only 51 phylotypes were com- human gut. The genome of Senegalvirus (Genbank JF909596– mon between the two approaches (15% of the culturomics JF909602) is closely related to those of Marseillevirus (96% set) (Fig. 3a). Among the ‘culturomics core microbiome’ of identity) and Lausannevirus [22] (Fig. S3), suggesting that this 45 cultured species, only 12 (26%) were detected by py- is a new strain of Marseillevirus. Preliminary work indicates rosequencing. Similarly, only 44 genera (38% of the cul- the presence of antibodies against this virus in the serum turomics set) were shared between the two approaches and stool of the subject. (Fig. 3b). Altogether, 416 phylotypes of previously uncultured bacteria were identified. Notably, the sequence from Genome sequencing of new bacteria a new genus cultured here ( Senegalemassilia anaerobia) had The genomes of all 31 new bacterial species and genera been previously identified as an uncultured bacterium by (Tables 1 and 2; Fig. 2) isolated from the two African stool metagenomics, demonstrating the capacity of culturomics samples and the French stool sample were sequenced, gener- to grow such supposedly ‘unculturable’ microorganisms ating a total of 110.4 Mb of unique sequence, and are freely [23–27]. Finally, the molecular techniques did not identify available in the EMBL database (http://www.ebi.ac.uk/embl/ pathogenic bacteria, such as Salmonella, that were detected Submission/index.html). The genome sizes of the new bacte- by culturomics ( Fig. 4).

(a) (b)

FIG. 3. Identification of bacteria in the human gut by culturomics and metagenomics. (a) The two ‘icebergs’ represent the 340 cultured bacterial species and the 698 phylotypes identified by pyrosequencing. The overlap between the two sets of species, i.e. the 51 species detected by both approaches, is shown in purple. Below the ‘sea level’ is the projected unknown part of the human gut microbiome. (b) Taxonomic distribution of organisms identified by culturomics and pyrosequencing. The ovoid shape denoted ‘Culture’ indicates all of the bacterial and fungal genera identified by culturomics from the three stool samples. The ovoid shape denoted ‘Pyro’ indicates genera identified by 16S amplicon pyrosequencing of the three stool samples. The dashed coloured lines show the phylum membership of the respective genus node. The two shapes with dashed lines at left and right represent the bacterial and fungal genera identified by only one technique (pyrosequencing or culturomics), whereas the shape in the middle represents the genera identified by both culturomics and pyrosequencing.

S. Aureus S G G S G G G Bacteroides, S Eubacterium, A S S O S S P C N L Clostridium F G F E F F C E 2 L X L L L U Peptostreptococcus X Bifidobacterium R 0 X A X X R L Vibrio cholerae E. coli O157 T Shigella dysenteriae U Enterotoxigenic E. coli Streptococcus R Aeromonas hydrophila Clostridium difficile 35% E Campylobacter jejuni

Salmonella Typhimurium 65% Tropheryma whipplei Yersinia enterocolitica Salmonella Typhi

FIG. 4. The detection thresholds of metagenomic and culturomic approaches. The detection threshold of metagenomic methods correlates with the concentration of bacteria in the investigated sample divided by the number of generated sequences. The blue pointed shapes show the detection depth of different published metagenomic analyses of the human gut microbiome. The upper dotted red line shows the detection threshold of the most powerful available metagenomic methods, the middle line shows the detection threshold of PCR, and the lower line shows the detection threshold of culturomics. The latter two thresholds were determined by detection of Staphylococcus aureus that was added to the samples in varying concentrations (indicated by green pointed shapes). Among the 340 cultivated bacterial species, 29 were identiﬁed only after several days of incubation in an anaerobic blood culture bottle, so their concentrations in the original samples could not be estimated. Among the remaining 311 bacteria, 203 (65%) were found at concentrations of <10 6 CFU/g of stool, i.e. below the detection threshold of metagenomic methods.

metagenomic approaches (Fig. 4). In support of this conclusion, Discussion culture methods allowed the detection of Staphylococcus aureus that was added to the stools at a concentration that was Metagenomics is currently thought of as the mainstream of 100 times lower than the concentration detectable by molec- microbiome studies, in particular as applied to the human ular tools (Fig. 4). gut. Unexpectedly, however, in a direct comparison, we The paradigm shift in microbiome study that seems to be described more known bacterial species by systematically brought about by culturomics became possible thanks to the applying a large sample of culture conditions (the approach breakthrough in the application of MALDI-TOF MS [10]. In we denoted culturomics) than by pyrosequencing (Fig. 3a). comparison with the most rapid conventional phenotypic Moreover, we found a dramatic divergence between the sets identification method for bacteria (Vitek System; Biomerieux, of bacteria identified by the two approaches at the level of Marcy I’Etoile, France), MALDI-TOF MS reduces by c. 55-fold both species and genera. The detection by culturomics of the time to bacterial identification and reduces the costs by numerous bacteria that go undetected in genomic and meat least a factor of 5 [10]. Therefore, MALDI-TOF MS is cur- tagenomic studies is far from being trivial, even if most of rently the most time-effective and cost-effective identification these microorganisms are of low abundance. Undoubtedly, a method available for culture-based microbiota studies. minority population, as in the famous short story [28], can In the present culturomics study, the actual analysis of have a substantial effect on the ecology of the gut microbiota microbial cultures involved only three students (JCL,MM,PH) and on human health. Indeed, c. 65% of bacterial species that performed the experience on the three stools during from the three samples were detected at concentrations 2 years in a single laboratory. Nevertheless, this limited effort between 10 3 and 10 6 CFU/mL, which are below the yielded 174 bacterial species that have not been previously detection thresholds of large-scale molecular studies, demon- reported from the human gut microbiota. Genome sequencing strating the major ‘depth bias’ that is characteristic of the of these bacteria would increase by c. 18% the number of

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases, CMI, 18 , 1185–1193 1192 Clinical Microbiology and Infection, Volume 18 Number 12, December 2012 CMI

sequenced bacterial genomes from the human gut (975) that De´veloppement), and Aix-Marseille Universite´ (cre´dits have been identified by several laboratories within the human recurrents). The authors have declared that no competing microbiome project [29]. interests exist. The present limited culturomics study shows that microbial biodiversity in the human gut is substantially broader than pre- Supporting Information dicted on the basis of genomic and metagenomic analyses [27,30]. Interestingly, culturomics also ‘broke the records’ for the largest bacterium and virus isolated from humans so far. Additional Supporting Information may be found in the By using different atmospheres, temperatures, pH, nutrients, online version of this article: minerals, antibiotics or phages, ‘microbial culturomics’ pro- Figure S1. A comparison of the species identified from vides comprehensive culture conditions simulating, reproduc- the cultures of the different samples. ing or mimicking the entirety of selective constraints that have Figure S2. The protocol used for the rumen fluid prepa- shaped the gut microbiota for millions of years. In fact, each ration. isolated microorganism is one among the possible viable solu- Figure S3. A genomic comparison between Marseillevi- tions to the evolutionary equation whose constants are the rus, Senegalvirus and Lausannevirus. selective constraints of the environment, corresponding here Figure S4. A Venn diagram representing the number of to the human gut. This is why microbial culturomics is the best species cultivated from each of the stool samples. way to capture the functional and viable gut microbiota biodi- Figure S5. The number of species that were cultivated in versity of each human individual through large-scale isolation, 10–70 conditions out of the 212 tested culture conditions. and to capture the deepest informational genetic gut biodiver- Figure S6. The percentage of new species and of the sity by sequencing the complete genomes of the previously iso- total number of species cultured that grew in only one cul- lated microorganisms. In the future, the use of the most ture condition or in multiple culture conditions. effective conditions and automatic colony picking will further Table S1. Culture conditions for microbial culturomics deepen this field of research. characterization from the stool samples of the N’Diop and Dielmo individuals and the obese French individual. Table S2. The 345 bacterial and fungal species cultured Acknowledgements from the N’Diop, Dielmo and obese patient stool samples. Table S3. The 20 best culture conditions, which facili- The authors wish to thank B. Davoust for the sheep rumen tated the identification of 73% of the bacterial species. collection, R. Rivet for his technical assistance, and I. Combe Table S4. Cultivated species identified in a Senegalese for her administrative assistance. stool sample from Dielmo village. Table S5. Cultivated phylotypes identified in the stool sample from an obese patient. Author Contributions Table S6. Cultivated phylotypes identified in a Senegalese stool sample from N’Diop village. Conception and and design of the experiments: D. Raoult. Per- Please note: Wiley-Blackwell are not responsible for the formance of the experiments: J. C. Lagier, M. Million, P. Hugon, content or functionality of any supporting materials supplied I. Pagnier, C. Robert, and B. La Scola. Analysis of the data: J. C. by the authors. Any queries (other than missing material) Lagier, F. Armougom, P. Hugon, I. Pagnier, F. Bittar, G. Four- should be directed to the corresponding author for the article. nous, G. Gimenez, E. V. Koonin, B. La Scola, and D. Raoult. Contribution of reagents/material/analysis tools: J. C. Lagier, F. References Armougom, P. Hugon, I. Pagnier, F. Bittar, G. Fournous, G. Gimenez, M. Million, B. La Scola, and D. Raoult. Writing of the 1. Turnbaugh PJ, Ley RE, Hamady M, Fraser-Liggett CM, Knight R, paper: J. C. Lagier, F. Armougom, E. V. Koonin, and D. Raoult. Gordon JI. The human microbiome project. Nature 2007; 449: 804–810. 2. Finegold SM, Attebery HR, Sutter VL. Effect of diet on human fecal Transparency Declaration flora: comparison of Japanese and American diets. Am J Clin Nutr 1974; 27: 1456–1469. 3. Goodman AL, Kallstrom G, Faith JJ et al. Extensive personal human This work was funded by the CNRS, Centre National de la gut microbiota culture collections characterized and manipulated in Recherche Scientifique, the IRD (Institut de Recherche et gnotobiotic mice. Proc Natl Acad Sci USA 2011; 108: 6252–6257.

4. Andersson AF, Lindberg M, Jakobsson H, Backhed F, Nyren P, Eng- 16. Mishra AK, Lagier JC, Robert C, Raoult D, Fournier PE. Non-contigu- strand L. Comparative analysis of human gut microbiota by barcoded ous finished genome sequence and description of Clostridium senega- pyrosequencing. PLoS ONE 2008; 3: e2836. lense sp. nov. Stand Genomic Sci 2012; 6: 386–395. 5. Turnbaugh PJ, Quince C, Faith JJ et al. Organismal, genetic, and tran- 17. Sillankorva S, Oliveira D, Moura A et al. Efficacy of a broad host scriptional variation in the deeply sequenced gut microbiomes of range lytic bacteriophage against E. coli adhered to urothelium. Curr identical twins. Proc Natl Acad Sci USA 2010; 107: 7503–7508. Microbiol 2010; 68: 1128–1132. 6. Reyes A, Haynes M, Hanson N et al. Viruses in the faecal microbiota 18. Tian B, Hua Y. Carotenoid biosynthesis in extremophilic Deinococcus- of monozygotic twins and their mothers. Nature 2010; 466: 334–338. Thermus bacteria. Trends Microbiol 2010; 18: 512–520. 7. Claesson MJ, Wang Q, O’Sullivan O et al. Comparison of two next- 19. Pagnier I, Raoult D, La Scola B. Isolation and identification of generation sequencing technologies for resolving highly complex mic- amoeba-resisting bacteria from water in human environment by using robiota composition using tandem variable 16S rRNA gene regions. an Acanthamoeba polyphaga co-culture procedure. Environ Microbiol Nucleic Acids Res 2010; 38: e200. 2008; 10: 1135–1144. 8. Wu GD, Lewis JD, Hoffmann C et al. Sampling and pyrosequencing 20. Willner D, Furlan M, Haynes M et al. Metagenomic analysis of respi- methods for characterizing bacterial communities in the human gut ratory tract DNA viral communities in cystic fibrosis and non-cystic using 16S sequence tags. BMC Microbiol 2010; 10: 206. fibrosis individuals. PLoS ONE 2009; 4: e7370. 9. Kaeberlein T, Lewis K, Epstein SS. Isolating ‘uncultivable’ microorgan- 21. Dworkin M, Falkow S. The prokariotes, 3rd edn. NY: springer, 2006. isms in pure culture in a simulated natural environment. Science 2002; 22. Thomas V, Bertelli C, Collyn F et al. Lausannevirus, a giant amoebal 296: 1127–1129. virus encoding histone doublets. Environ Microbiol 2011; 13: 1454– 10. Seng P, Drancourt M, Gouriet F et al. Ongoing revolution in bacteri- 1466. ology: routine identification of bacteria by matrix-assisted laser 23. Qin J, Li R, Raes J et al. A human gut microbial gene catalogue estab- desorption ionization time-of-flight mass spectrometry. Clin Infect Dis lished by metagenomic sequencing. Nature 2010; 464: 59–65. 2009; 49: 543–551. 24. De Filippo C, Cavalieri D, Di Paola M et al. Impact of diet in shaping 11. Stackebrandt E, Ebers J. Taxonomic parameters revisited: tarnished gut microbiota revealed by a comparative study in children from Eur- gold standards. Microbiol Today 2006; 33: 152–155. ope and rural Africa. Proc Natl Acad Sci USA 2010; 107: 14691–14696. 12. Lagier JC, El Karkouri K, Nguyen TT, Armougom F, Raoult D, Fournier 25. Ley RE, Turnbaugh PJ, Klein S, Gordon JI. Microbial ecology: human PE. Non-contiguous finished genome sequence and description of An- gut microbes associated with obesity. Nature 2006; 444: 1022–1023. aerococcus senegalensis sp. nov. Stand Genomic Sci 2012; 6: 116–125. 26. Turnbaugh PJ, Hamady M, Yatsunenko T et al. A core gut microbiom- 13. Lagier JC, Armougom F, Mishra AK, Nguyen TT, Raoult D, Fournier e in obese and lean twins. Nature 2009; 457: 480–484. PE. Non-contiguous finished genome sequence and description of Al- 27. Arumugam M, Raes J, Pelletier E et al. Enterotypes of the human gut istipes timonensis sp. nov. Stand Genomic Sci 2012; 6: 315–324. microbiome. Nature 2011; 473: 174–180. 14. Mishra AK, Gimenez G, Lagier JC, Robert C, Raoult D, Fournier PE. 28. Dick PK. The minority report . NY: Pantheon, 1956. Non-contiguous finished genome sequence and description of Alistipes 29. Human microbiome project catalogue. Available at: http://www.hmp- senegalensis sp. nov. Stand Genomic Sci 2012; 6: 304–314. dacc-resources.org/cgi-bin/hmp_catalog/main.cgi?section=HmpSum- 15. Kokcha S, Mishra AK, Lagier JC et al. Non-contiguous finished gen- mary&page=showSummary (last accessed 1 November 2011). ome sequence and description of Bacillus timonensis sp. nov. Stand 30. Fournier PE, Drancourt M, Raoult D. Bacterial genome sequencing Genomic Sci 2012; 6: 346–355. and its use in infectious diseases. Lancet Infect Dis 2007; 7: 711–723.

ª2012 The Authors Clinical Microbiology and Infection ª2012 European Society of Clinical Microbiology and Infectious Diseases, CMI, 18 , 1185–1193

Article XIV :

Non contiguous finished genome sequence and

description of Bacillus timonensis sp. nov.

Sahare Kokcha, Jean-Christophe Lagier, Matthieu Million, Rivet R,

Grégory Gimenez, Didier Raoult, Pierre-Edouard Fournier

Published in Stand. Genomic Sci. 2012;6:346-355. (IF 1.62)

165

Standards in Genomic Sciences (2012) 6:346-355 DOI:10.4056/sigs.2776064

Non contiguous-finished genome sequence and description of Bacillus timonensis sp. nov.

Sahare Kokcha 1, Ajay Kumar Mishra 1, Jean-Christophe Lagier 1, Matthieu Million 1, Quentin Leroy 1, Didier Raoult 1 and Pierre-Edouard Fournier 1*

1 Unité de Recherche sur les Maladies Infectieuses et Tropicales Emergentes, UMR CNRS 6236 – IRD 198, Faculté de médecine, Aix-Marseille Université

* Corresponding author: Pierre-Edouard Fourner ([email protected])

Key words : Bacillus timonensis , genome

Bacillus timonensis strain MM10403188 T sp. nov. is the type strain of a proposed new species within the genus Bacillus . This strain, whose genome is described here, was isolated from the fecal flora of a healthy patient. B. timonensis is an aerobic Gram-negative rod shaped bacterium. Here we describe the features of this organism, together with the complete genome sequence and annotation. The 4,632,049 bp long genome (1 chromosome but no plasmid) contains 4,610 protein-coding and 74 RNA genes, including 5 rRNA genes.

Introduction Bacillus timonensis strain MM10403188 T (= CSUR Gram-positive, motile, and spore-forming bacteria, P162 = DSM 25372) is designated as the type is made of 256 species and 7 subspecies with val- strain of B. timonensis , a new Gram-negative aero- idly published names [7]. Members of the genus bic, indole-positive bacillus that was isolated from Bacillus are ubiquitous bacteria isolated from var- the stool of a healthy Senegalese patient as part of ious environments including soil, fresh and sea a “culturomics” study aiming at cultivating indi- water, food, and occasionally from humans in vidually all species within human feces. whom they are either pathogens, such as B. To date, DNA-DNA hybridization and G+C content anthracis and B. cereus , or opportunists in determination [1] remain the gold standard meth- immunocompromised patients [7]. Apart from ods for the definition of bacterial species, despite anthrax, caused by B. anthracis [8], and toxi- the development of 16S rRNA PCR and sequenc- infections caused by B. cereus , Bacillus species ing which have deeply changed bacterial taxono- may be involved in a variety of aspecific human my [2]. Over recent years, high throughput ge- infections, including cutaneous, ocular, central nome sequencing provided a wealth of genetic nervous system or bone infections, pneumonia, information [3]. In an effort to include genomic endocarditis and bacteremia [9]. data in bacterial taxonomy we recently used a polyphasic approach [4] that includes genomic Classification and features data, MALDI-TOF spectrum and main phenotypic A stool sample was collected from a healthy 16- characteristics to describe new bacterial species year-old male Senegalese volunteer patient living [5,6] . in Dielmo (a rural village in the Guinean-Sudanian Here we present a summary classification and a zone in Senegal), who was included in a research set of features for B. timonensis sp. nov. strain protocol. The patient gave an informed and signed MM10403188 T together with the description of consent, and the agreement of the National Ethics the complete genomic sequencing and annotation. Committee of Senegal and the local ethics commit- These characteristics support the circumscription tee of the IFR48 (Marseille, France) was obtained of the species B. timonensis . under agreements 09-022 and 11-017). The fecal specimen was preserved at -80°C after collection The genus Bacillus (Cohn 1872) was created in and sent to Marseille. Strain MM10403188 (Table 1872 [6]. To date, this genus, mostly comprised of 1) was isolated in June 2011 by cultivation on 5% The Genomic Standards Consortium

Kokcha et al. sheep blood-enriched Brain Heart Infusion agar value was lower than the 98.7% 16S rRNA gene with (Becton Dickinson, Heidelberg, Germany). sequence threshold recommended by This strain exhibited a 98.2% nucleotide sequence Stackebrandt and Ebers to delineate a new species similarity with Bacillus humi , the phylogenetically without carrying out DNA-DNA hybridization [2]. closest validated Bacillus species (Figure 1). This

Table 1. Classification and general features of Bacillus timonensis strain MM10403188 T MIGS ID Property Term Evidence code a Current classification Domain Bacteria TAS [10] Phylum Firmicutes TAS [11-13] Class Bacilli TAS [14,15] Order Bacillales TAS [16,17] Family Bacillaceae TAS [16,18] Genus Bacillus TAS [16,19,20] Species Bacillus timonensis IDA Type strain MM10403188 T IDA Gram stain negative IDA Cell shape rod IDA Motility motile IDA Sporulation sporulating IDA Temperature range mesophile IDA Optimum temperature 37°C IDA MIGS-6.3 Salinity growth in BHI medium + 5% NaCl IDA MIGS-22 Oxygen requirement aerobic IDA Carbon source unknown NAS Energy source unknown NAS MIGS-6 Habitat human gut IDA MIGS-15 Biotic relationship Free living IDA MIGS-14 Pathogenicity Unknown NAS Biosafety level 2 Isolation human feces MIGS-4 Geographic location Senegal IDA MIGS-5 Sample collection time September 2010 IDA MIGS-4.1 Latitude 13.7167 IDA MIGS-4.1 Longitude -16.4167 IDA MIGS-4.3 Depth Surface IDA MIGS-4.4 Altitude 51 m above sea level IDA Evidence codes - IDA: Inferred from Direct Assay; TAS: Traceable Author Statement (i.e., a direct report exists in the literature); NAS: Non-traceable Author Statement (i.e., not directly observed for the living, isolated sample, but based on a generally accepted property for the species, or anecdotal evidence). These evidence codes are from the Gene Ontology project [21]. If the evidence is IDA, then the property was directly observed for a live isolate by one of the authors or an expert mentioned in the acknowledgements. http://standardsingenomics.org 347 Bacillus timonensis sp. nov.

Figure 1. Phylogenetic tree highlighting the position of Bacillus timonensis strain MM10403188 T relative to other type strains within the Bacillus genus. GenBank accession numbers are indicated in parentheses. Sequences were aligned using CLUSTALW, and phylogenetic inferences obtained using the maximum-likelihood method within the MEGA software. Numbers at the nodes are percentages of bootstrap values obtained by repeating the analysis 500 times to generate a majority consensus tree. Clostridium botulinum was used as an outgroup. The scale bar represents a 2% nucleotide sequence divergence.

Different growth temperatures (25, 30, 37, 45°C) Strain MM10403188 T exhibited oxidase activity but were tested. Growth occurred at all tested tempera- not catalase activity, and was positive for indole. tures, but optimal growth occurred between 30 and Using API 50CH, a positive reaction was obtained for 37°C. Colonies were 3 mm in diameter on blood- L-arabinose, D-lactose, D-melibiose, D-trehalose, D- enriched BHI agar. Growth of the strain was tested saccharose, and D-turanose fermentation. A weak under anaerobic and microaerophilic conditions reaction was obtained for aesculin. Other tests were using GENbag anaer and GENbag microaer systems, negative. Using API-ZYM, positive reactions were respectively (BioMérieux), and in aerobic conditions, obtained for ester - - with or without 5% CO 2. Growth was achieved in - -glucosinidase. B. aerobic (with and without CO 2) and microaerophilic timonensis was susceptibleǡ to Ƚ penicillin ǡ G, amoxicil- Ⱦ conditions. No growth was observed in anaerobic lin, ǡ vancomycin, Ƚ gentamicin, Ⱦ erythromycin, conditions. Gram staining showed Gram negative doxycyclin, rifampicin, and ciprofloxacin but re- bacilli (Figure 2). A motility test was positive. Cells sistant to trimethoprim/sulfamethoxazole. grown on agar are sporulated and have a mean diameter of 0.66 µm (Figure 3). 348 Standards in Genomic Sciences Kokcha et al.

Figure 2. Gram staining of B. timonensis strain MM10403188 T

Figure 3. Transmission electron microscopy of B. timonensis strain MM10403188 T, using a Morgani 268D (Philips) at an operating voltage of 60kV. The scale bar represents 900 nm. http://standardsingenomics.org 349 Bacillus timonensis sp. nov. By comparison with B. humi , B. timonensis differed shots at a variable laser power. The time of acquisi- in Gram staining, in culture atmosphere, as B. humi tion was between 30 seconds and 1 minute per was able to grow anaerobically, in catalase activity, spot. The four MM10403188 spectra were import- in spore forming capacity, in indole production, and ed into the MALDI BioTyper software (version 2.0, in carbohydrate metabolism, notably for arbutin, Bruker) and analyzed by standard pattern match- salicin, L-arabinose, melibiose, turanose, and treha- ing (with default parameter settings) against the lose [22]. main spectra of 3,769 bacteria including 129 spec- Matrix-assisted laser-desorption/ionization time- tra from 98 Bacillus species, notably B. humi , used of-flight (MALDI-TOF) MS protein analysis was as reference data, in the BioTyper database. The carried out as previously described [23]. Briefly, a method of identification included the m/z from pipette tip was used to pick one isolated bacterial 3,000 to 15,000 Da. For every spectrum, 100 peaks colony from a culture agar plate, and to spread it as at most were taken into account and compared a thin film on a MTP 384 MALDI-TOF target plate with spectra in the database. A score enabled the (Bruker Daltonics, Leipzig, Germany). Four distinct presumptive identification and discrimination of deposits were done for strain MM10403188 from the tested species from those in the database: a four isolated colonies. Each smear was overlaid score > 2 with a validated species enabled the iden- with 2µL of matrix solution (saturated solution of tification at the species level, a score > 1.7 but < 2 alpha-cyano-4-hydroxycinnamic acid) in 50% ace- enabled the identification at the genus level; and a tonitrile, 2.5% tri-fluoracetic-acid, and allowed to score < 1.7 did not enable any identification. For dry for five minutes. Measurements were per- strain MM10403188 T, the obtained score was 1.2, formed with a Microflex spectrometer (Bruker). thus suggesting that our isolate was not a member Spectra were recorded in the positive linear mode of a known species. We incremented our database for the mass range of 2,000 to 20,000 Da (parame- with the spectrum from strain MM10403188 (Fig- ter settings: ion source 1 (IS1), 20 kV; IS2, 18.5 kV; ure 4). The spectrum was made available online in lens, 7 kV). A spectrum was obtained after 675 our free-access URMS database [24].

8000 Intens. [a.u.] Intens.

6000

4000

2000

2000 4000 6000 8000 10000 12000 14000 16000 18000 m/z

Figure 4. Reference mass spectrum from B. timonensis strain MM10403188 T. Spectra from 12 individual colonies were compared and a reference spectrum was generated.

350 Standards in Genomic Sciences Kokcha et al. Genome sequencing information Genome project history The organism was selected for sequencing on the on a DNA labchip 7500 with an optimal size of basis of its phylogenetic position and 16S rRNA simi- 3.345kb. The library was constructed according to the larity to other members of the genus Bacillus , and is 454 GS FLX Titanium paired-end protocol. Circulari- part of a “culturomics” study of the human digestive zation and nebulization were performed and generat- flora aiming at isolating all bacterial species within ed a pattern with an optimum at 492 bp. After PCR human feces. It was the 60 th genome of a Bacillus amplification through 15 cycles followed by double species and the first genome of Bacillus timonensis size selection, the single stranded paired end library sp. nov. A summary of the project information is was then quantified on the Quant-it Ribogreen kit shown in Table 2. The Genbank accession number is (Invitrogen) on the Genios Tecan fluorometer at 339 CAET00000000 and consists of 146 contigs. pg/µL. The library concentration equivalence was calculated as 12,6E+08 molecules/µL. The library was Growth conditions and DNA isolation stored at -20°C until further use. T B. timonensis sp. nov. strain MM10403188 , CSUR The shotgun library was clonally amplified with 3cpb P162, DSM 25372, was grown aerobically on 5% and the paired-end library was amplified with lower sheep blood-enriched BHI agar at 37°. Four petri cpb (1 cpb) in 4 emPCR reactions with the GS Titani- dishes were spread and growth from the plates um SV emPCR Kit (Lib-L) v2 (Roche). The yields of the was resuspended in 3x500µl of TE buffer and emPCR was 5.97% for the shotgun and 15.92% for stored at 80°C. Then, 500µl of this suspension the paired end as expected by the range of 5 to 20% were thawed, centrifuged 3 minutes at 10,000 from the Roche procedure. rpm and resuspended in 3x100µL of G2 buffer (EZ1 DNA Tissue kit, Qiagen). A first mechanical Approximately 790,000 beads for a 1/4 region and lysis was performed by glass powder on the 340,000 beads for a 1/8 region were loaded on the GS Fastprep-24 device (Sample Preparation system, Titanium PicoTiterPlate PTP Kit 70×75 and se- MP Biomedicals, USA) using 2x20 seconds cycles. quenced with the GS FLX Titanium Sequencing Kit DNA was then treated with 2.5µg/µL lysozyme (30 XLR70 (Roche). The run was performed overnight minutes at 37°C) and extracted using the BioRobot and then analyzed on the cluster through the EZ1 Advanced XL (Qiagen). The DNA was then gsRunBrowser and Newbler assembler (Roche). For concentrated and purified using the Qiamp kit the shotgun sequencing, 112,962 passed filter wells (Qiagen). The yield and the concentration was were obtained and generated 34.48Mb with a length measured by the Quant-it Picogreen kit (Invitro- average of 322 bp. For the shotgun sequencing, gen) on the Genios Tecan fluorometer at 50ng/µl. 213,882 passed filter wells were obtained and generated 50.6 Mb with a length average of 236 bp. The Genome sequencing and assembly passed filter sequences were assembled Using DNA (5 µg) was mechanically fragmented on a Newbler with 90% identity and 40bp as overlap. The Hydroshear device (Digilab, Holliston, MA,USA) with final assembly identified 11 scaffolds and 89 contigs an enrichment size at 3-4kb. The DNA fragmentation (>1500bp) generating a genome size of 4.6 Mb. was visualized through the Agilent 2100 BioAnalyzer

Table 2. Project information MIGS ID Property Term MIGS-31 Finishing quality High-quality draft MIGS-28 Libraries used 454 GS shotgun and paired-end 3- kb libraries MIGS-29 Sequencing platform 454 GS FLX Titanium MIGS-31.2 Sequencing coverage 19× MIGS-30 Assemblers Newbler version 2.5.3 MIGS-32 Gene calling method PRODIGAL INSDC ID 112529 Genbank Date of Release February 28 th , 2012 Gold ID Gi13534 NCBI project ID CAET00000000 MIGS-13 Project relevance Study of the human gut microbiome http://standardsingenomics.org 351 Bacillus timonensis sp. nov. Genome annotation Open Reading Frames (ORFs) were predicted us- genes than B. licheniformis (4,684 and 4,356, re- ing Prodigal [25] with default parameters but the spectively), and more genes assigned to COGs predicted ORFs were excluded if they were span- (3,399 and 3,130, respectively). However, the dis- ning a sequencing gap region. The predicted bac- tribution of genes into COG categories (Table 4) terial protein sequences were searched against was highly similar in both genomes. In addition, B. the GenBank database [26] and the Clusters of timonensis shared a mean 86.10% (range 76.4- Orthologous Groups (COG) databases using 93%) sequence similarity with B. licheniformis at BLASTP. The tRNAScanSE tool [27] was used to the genome level. find tRNA genes, whereas ribosomal RNAs were found by using RNAmmer [28] and BLASTn Although the degree of 16S rRNA similarity was against the GenBank database. ORFans were iden- elevated (98.2%) between strain MM10403188 tified if their BLASTP E-value was lower than 1e- and B. humi strain DSM 16318, both strains exhib- 03 for alignment length greater than 80 amino ited several phenotypic and genomic differences, acids. If alignment lengths were smaller than 80 and we formally propose the creation of Bacillus amino acids, we used an E-value of 1e-05. Such timonensis sp. nov. that contains the strain parameter thresholds have already been used in MM10403188 T. This strain has been found in Sen- previous works to define ORFans. egal. To estimate the mean level of nucleotide sequence Description of Bacillus timonensis sp. nov. similarity at the genome level between Bacillus Bacillus timonensis (tim.on.en´sis. L. gen. masc. n. species, we compared the ORFs only using timonensis , of Timone, the name of the hospital BLASTN and the following parameters: a query where strain MM10403188 T was cultivated.) Iso- coverage of t 70% and a minimum nucleotide lated from stool from an asymptomatic Senegalese length of 100 bp. patient. B. timonensis is an aerobic Gram-negative bacterium. Grows on axenic medium at 37°C in an Genome properties aerobic atmosphere. Colonies were 3 mm in diam- The genome is 4,632,049 bp long (1 chromosome, eter on blood-enriched BHI agar. Cells grown on but no plasmid) with a 37.30% GC content (Figure agar are sporulated and have a mean diameter of 5 and Table 3). Of the 4,684 predicted genes, 4,610 0.66 µm. A positive reaction was obtained for L- were protein-coding genes and 74 were RNAs. A arabinose, D-lactose, D-melibiose, D-trehalose, D- total of 3,399 genes (75.56%) were assigned a saccharose, and D-turanose fermentation. Positive putative function. Three hundred forty genes were reactions were obtained for oxidase, ester - identified as ORFans (7.4%). The remaining genes -glucorinidase, and - - were annotated as hypothetical proteins. The glucosinidase activity. No catalase activityǡ was Ƚ properties and the statistics of the genome are ǡexhibited. Positive Ⱦ for indole. By comparisonȽ with Ⱦ summarized in Tables 3. The distribution of genes B. humi , B. timonensis differs in Gram staining, in into COGs functional categories is presented in culture atmosphere, as B. humi grows anaerobical- Table 4. ly, in catalase activity, in spore forming capacity, in indole production, and in carbohydrate metabo- Comparison with the genomes from lism, notably for arbutin, salicin, L-arabinose, Bacillus melibiose, turanose, and trehalose. B. timonensis is other species susceptible to penicillin G, amoxicillin, vanco- Genome sequences are currently available for mycin, gentamicin, erythromycin, doxycyclin, ri- more than 25 validly named Bacillus species. Here fampicin, and ciprofloxacin but resistant to trime- we compared the genome sequence of B. thoprim/sulfamethoxazole. Motile. The G+C con- timonensis strain MM10403188 T with that of B. tent of the genome is 37.30%. The 16S rRNA and licheniformis strain ATCC 14580, the most closely genome sequences are deposited in GenBank un- related phylogenetic neighbor for which the ge- der accession numbers JF824810 and nome sequence is available. The draft genome CAET00000000, respectively. The type strain sequence of B. timonensis is larger than B. MM10403188 T (= CSUR P162 = DSM 253720) was licheniformis (4.6 Mb and 4.2 Mb, respectively) but isolated from the fecal flora of a healthy patient its G+C content is lower (37.30 and 46.19%, re- from Senegal. spectively). B. timonensis has more predicted 352 Standards in Genomic Sciences Kokcha et al.

Figure 5 . Graphical circular map of the chromosome. From outside to the center: Genes on the forward strand (colored by COG categories), genes on the reverse strand (colored by COG categories), RNA genes (tRNAs green, rRNAs red), GC content, and GC skew.

Table 3 . Nucleotide content and gene count levels of the genome Attribute Value % of total a Genome size (bp) 4,632,049 DNA Coding region (bp) 3,959,694 85.48 DNA G+C content (bp) 1,727,754 37.3 Total genes 4,684 100 RNA genes 74 1.58 Protein-coding genes 4,610 98.42 Genes with function prediction 3,643 77.75 Genes assigned to COGs 3,399 75.56 Genes with peptide signals 189 4.03 Genes with transmembrane helices 1,261 26.92 aThe total is based on either the size of the genome in base pairs or the total number of protein coding genes in the annotated genome http://standardsingenomics.org 353 Bacillus timonensis sp. nov. Table 4. Number of genes associated with the 25 general COG functional categories Code Value % age a Description J 181 3.93 Translation, ribosomal structure and biogenesis A 0 0 RNA processing and modification K 310 6.72 Transcription L 169 3.67 Replication, recombination and repair B 1 0.02 Chromatin structure and dynamics D 39 0.85 Cell cycle control, mitosis and meiosis Y 0 0 Nuclear structure V 71 1.54 Defense mechanisms T 193 4.19 Signal transduction mechanisms M 197 4.27 Cell wall/membrane biogenesis N 67 1.45 Cell motility Z 0 0 Cytoskeleton W 0 0 Extracellular structures U 49 1.06 Intracellular trafficking and secretion O 114 2.47 Posttranslational modification, protein turnover, chaperones C 184 3.99 Energy production and conversion G 349 7.57 Carbohydrate transport and metabolism E 412 8.94 Amino acid transport and metabolism F 97 2.10 Nucleotide transport and metabolism H 121 2.62 Coenzyme transport and metabolism I 150 3.25 Lipid transport and metabolism P 245 5.31 Inorganic ion transport and metabolism Q 100 2.17 Secondary metabolites biosynthesis, transport and catabolism R 594 12.89 General function prediction only S 361 7.83 Function unknown - 606 13.15 Not in COGs aThe total is based on the total number of protein coding genes in the annotated genome

References 1. Rossello-Mora R. DNA-DNA Reassociation Meth- Microbiol 2010; 60 :249-266. PubMed ods Applied to Microbial Taxonomy and Their Criti- http://dx.doi.org/10.1099/ijs.0.016949-0 cal Evaluation. In : Stackebrandt E (ed), Molecular 5. Lagier JC, El Karkouri K, Nguyen TT, Armougom F, Identification, Systematics, and population Structure Raoult D, Fournier PE. Non-contiguous finished ge- of Prokaryotes. Springer, Berlin, 2006, p. 23-50. nome sequence and description of Anaerococcus 2. Stackebrandt E, Ebers J. Taxonomic parameters re- senegalensis sp. nov. Stand Genomic Sci 2012; visited: tarnished gold standards. Microbiol Today 6:116-125. PubMed 2006; 33 :152-155. http://dx.doi.org/10.4056/sigs.2415480 3. Welker M, Moore ER. Applications of whole-cell 6. Mishra AK, Lagier JC, Robert C, Raoult D, Fournier matrix-assisted laser-desorption/ionization time-of- PE. 2012. Non-contiguous finished genome se- flight mass spectrometry in systematic microbiology. quence and description of Clostridium senegalense Syst Appl Microbiol 2011; 34 :2-11. PubMed sp. nov. Stand.Genomic.Sci. In press5. Cohn http://dx.doi.org/10.1016/j.syapm.2010.11.013 F. Untersuchungen über Bakterien. Beitrage zur Biologie der Pflanzen Heft 1872; 1:127-224. 4. Tindall BJ, Rosselló-Móra R, Busse HJ, Ludwig W, Kämpfer P. Notes on the characterization of prokar- 7. Jernigan JA, Stephens DS, Ashford DA, Omenaca C, yote strains for taxonomic purposes. Int J Syst Evol Topiel MS, Galbraith M, Tapper M, Fisk TL, Zaki S, 354 Standards in Genomic Sciences Kokcha et al. Popovic T, et al . Bioterrorism-related inhalational Manual of Determinative Bacteriology , Eighth Edi- anthrax: the first 10 cases reported in the United tion, The Williams and Wilkins Co., Baltimore, States. Emerg Infect Dis 2001; 7:933-944. PubMed 1974, p. 529-550. http://dx.doi.org/10.3201/eid0706.010604 19. Mathews WC, Caperna J, Toerner JG, Barber RE, 8. Bottone EJ. Bacillus cereus , a volatile human patho- Morgenstern H. Neutropenia is a risk factor for gen. Clin Microbiol Rev 2010; 23 :382-398. PubMed Gram-negative Bacillus bacteremia in human im- http://dx.doi.org/10.1128/CMR.00073-09 munodeficiency virus-infected patients: results of a nested case-control study. Am J Epidemiol 1998; 9. Woese CR, Kandler O, Wheelis ML. Towards a 148 :1175-1183. PubMed natural system of organisms: proposal for the do- http://dx.doi.org/10.1093/oxfordjournals.aje.a00960 mains Archae, Bacteria , and Eukarya. Proc Natl 6 Acad Sci USA 1990; 87 :4576-4579. PubMed http://dx.doi.org/10.1073/pnas.87.12.4576 20. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, Davis AP, Dolinski K, Dwight SS, 10. Gibbons NE, Murray RGE. Proposals Concerning Eppig JT, et al . Gene ontology: tool for the unifica- the Higher Taxa of Bacteria . Int J Syst Bacteriol tion of biology. The Gene Ontology Consortium. 1978; 28 :1-6. http://dx.doi.org/10.1099/00207713- Nat Genet 2000; 25 :25-29. PubMed 28-1-1 http://dx.doi.org/10.1038/75556 11. Garrity GM, Holt JG. The Road Map to the Manual. 21. Stackebrandt E, Ebers J. Taxonomic parameters re- In: Garrity GM, Boone DR, Castenholz RW (eds), visited: tarnished gold standards. Microbiol Today Bergey's Manual of Systematic Bacteriolog y, Second 2006; 33 :152-155. Edition, Volume 1, Springer, New York, 2001, p. 119-169. 22. Heyrman J, Rodriguez-Diaz M, Devos J, Felske A, Logan NA, De Vos P. Bacillus arenosi sp. nov., Ba- 12. Murray RGE. The Higher Taxa, or, a Place for Every- cillus arvi sp. nov., Bacillus humi sp. nov., isolated thing...? In: Holt JG (ed), Bergey's Manual of Sys- from soil. Int J Syst Evol Microbiol 2005; 55 :111- tematic Bacteriology , First Edition, Volume 1, The 117. PubMed http://dx.doi.org/10.1099/ijs.0.63240- Williams and Wilkins Co., Baltimore, 1984, p. 31- 0 34. 23. Seng P, Drancourt M, Gouriet F, La SB, Fournier PE, 13. List Editor. List of new names and new combinations Rolain JM, Raoult D. Ongoing revolution in bacteri- previously effectively, but not validly, published. List ology: routine identification of bacteria by matrix- no. 132. Int J Syst Evol Microbiol 2010; 60 :469-472. assisted laser desorption ionization time-of-flight http://dx.doi.org/10.1099/ijs.0.022855-0 mass spectrometry. Clin Infect Dis 2009; 49 :543- 14. Ludwig W, Schleifer KH, Whitman WB. Class I. 551. PubMed http://dx.doi.org/10.1086/600885 Bacilli class nov. In: De Vos P, Garrity G, Jones D, 24. URMS database. http://ifr48.timone.univ- Krieg NR, Ludwig W, Rainey FA, Schleifer KH, mrs.fr/portail2/index.php?option=com_content&task Whitman WB (eds), Bergey's Manual of Systematic =view Bacteriology , Second Edition, Volume 3, Springer- Verlag, New York, 2009, p. 19-20. 25. Prodigal. http://prodigal.ornl.gov 15. Skerman VBD, Sneath PHA. Approved list of bacte- 26. Benson DA, Karsch-Mizrachi I, Clark K, Lipman DJ, rial names. Int J Syst Bact 1980; 30 :225-420. Ostell J, Sayers EW. GenBank. Nucleic Acids Res http://dx.doi.org/10.1099/00207713-30-1-225 2012; 40 :D48-D53. PubMed http://dx.doi.org/10.1093/nar/gkr1202 16. Prevot AR. Dictionnaire des bactéries pathogens. In : Hauduroy P, Ehringer G, Guillot G, Magrou J, 27. Lowe TM, Eddy SR. t-RNAscan-SE: a program for Prevot AR, Rosset, Urbain A ( eds ). Paris, Masson, imroved detection of transfer RNA gene in genomic 1953, p.1-692. sequence. Nucleic Acids Res 1997; 25 :955-964. PubMed 17. Fischer A. Untersuchungen über bakterien. Jahrbücher für Wissenschaftliche Botanik 1895; 28. Lagesen K, Hallin P, Rodland EA, Staerfeldt HH, 27 :1-163. Rognes T, Ussery DW. RNAmmer: consistent and rapid annotation of ribosomal RNA genes. Nucleic 18. Gibson T, Gordon RE. Genus I. Bacillus Cohn 1872, Acids Res 2007; 35 :3100-3108. PubMed 174; Nom. gen. cons. Nomencl. Comm. Intern. Soc. http://dx.doi.org/10.1093/nar/gkm160 Microbiol. 1937, 28; Opin. A. Jud. Comm. 1955, 39. In: Buchanan RE, Gibbons NE (eds), Bergey's http://standardsingenomics.org 355

Article XV :

Rapid and accurate bacterial identification in probiotics and yoghurts by MALDI-TOF mass

spectrometry

Emmanouil Angelakis, Matthieu Million, Mireille Henry,

Didier Raoult

Published in J Food Sci. 2011 Oct;76(8):M568-72. (IF 1.66) + photo de couverture

176

R ° in ies otic species spp. thelium cid bac- for LAB Lactobacillus Lactobacillus MALDI- , Bifidus , instead of and Streptococcus L. lactis and instead of Lactobacillus reuteri 2011 Institute of Food Technologists Further reproduction without permission is prohibited C ° doi: 10.1111/j.1750-3841.2011.02369.x spp., spp. for bacteria that do not produce Lactobacillus bulgaricus, , Streptococcus thermophilus. S. thermophilus spp. DNA/g. All products contained and Bacillus , and Pediococcus, Enterococcus, Bifidobacterium Lactobacillus casei (Saulnier and others 2009). Other probiotic species in- Lactobacillus , imultaneously with starter cultures in fermentation studies showed different features at the strain level for adhe- Propionibacteria L. delbrueckii specific agar and tested by quantitative real-time PCR Food is a major source of bacteria and viruses, and it modifies colony forming units/g. These bacteria were identified as lactic acid (Huys and othersshown 2006). Probiotic to properties be have been species specific and some authors reported that vitro sion, autoaggregation, and immunomodulatory effect (ReidKotzamanidis 1999; and others 2010). Identificationand of strains bacterial from commercializedmostly probiotics using has molecular beenSchillinger conducted methods and others (Holzapfelothers 2003; and 2007; Huys Marcobal others and andOur others 2001; objective others was 2008; 2006; to Sheu assess Pereaassisted and the laser and others accuracy desorption/ionisation and 2009). feasibility time-of-flightetry of mass (MALDI-TOF matrix- spectrom- MS) for bacteriallevel in identification various at fermented the dairyto spec species foods announced and on to each. compare our results and clude the genera acidophilus rhamnosus bile acid challenge, and (v) adherence to the intestinal epi of the host (Perea and others 2007). the microbial balance in the intestineteria (Raoult (LAB) 2010). and in Probi ucts the United contain States, probiotic-containing prod- bacterial strains used in dairy products are mostly lactic a copies of 9 sp. Bacterial identification was performed by MALDI-TOF 7 ics are species and strain dependent, we found a discrepancy ified. t cies and strain dependent, we found a discrepancy between the os- to 10 mote 6 to 10 and 16S rDNA genes. We tested 13 probiotic food and yoghurts Lactobacillus ly with 6 (GRAS; nimalis, Lactobacillus delbrueckii, tuf ´ edecine, Univ. Lactobacillus Authors are with . ´ e de M species, MALDI-TOF MS, probiotics, yoghurts com). Vol. 76, Nr. 8, 2011 r Lactobacillus MALDI-TOF MS is rapid and specific for the identification of bacteria in probiotic food and Probiotic food is manufactured by adding probiotic strains s functional food, ´ ee, 27 Bd Jean Moulin, 13385 Marseille Cedex 05, France. Direc spp. which was mentioned on the label and for another evidences of these health benefits are unclear except for

Journal of Food Science ´ editerran

in vivo ´ e des Rickettsies, CNRS UMR 6020, IFR 48, Facult MALDI-TOF MS allows a rapidAlthough and the accurate safety bacterial and identification functionality at of probiotics the are species spe level in probiotic food and yoghurts. bacterial strain announced on the label and the strain ident Bifidus TOF MS presented 92% specificity compared to the molecular assays. In one product we found Lactobacillus casei, Lactococcus lactis, Bifidobacterium a tanks. Here, we investigate thespectrometry accuracy (MALDI-TOF and MS) for feasibility bacterial offood identification matrix-assisted at and laser the yoghurts species desorption/ionisation level were time-of-flight in cultured mass probiotic food in and Columbia yoghurts. Probiotic and and we identified by qPCR that they presented 10 yoghurts. Although the safetybetween and the functionality bacterial of strain probiot announced on the label and the strain identified. analysis and by amplification and sequencing of (qPCR) for the detection and quantification of Practical Application: Keywords: Abstract: very large numbers of living bacteria varying from 10 The international endorsed definition for probiotics is live mi- generally recognized as safe),storage, (iii) (ii) antagonistic viability effects against during pathogens; (iv) processing tolerance to and reducing infectious diarrhea (Reid 1999).by Probiotics the are Food regulated and Drugand Administration by (FDA) the in the European Commission United2006). States in The Europe essential characteristics (Anadon for and bacteriaotics others to during be manufacturing used include as probi- (i) recognition as safe M568 MS 20110320 Submitted 3/14/2011, Accepted 7/18/2011 starter cultures inparticularly fermentation children, have been tanks consumingfermented (Saxelin probiotic especially dairy 2008). in products Humans, (Raoult 2008). Manufacturers pro croorganisms that, when administered infer adequate a amounts, con- healthare benefit manufactured on by adding the probiotic host strains (Sanders simultaneous 2008). Probiotic food Introduction fermented food with probiotics underitive the effects pretense on of immunological, digestive, having and p but respiratory functions Emmanouil Angelakis, Matthieu Million, Mireille Henry, and Didier Raoult Spectrometry in Probiotics and Yoghurts by MALDI-TOF Mass Rapid and Accurate Bacterial Identification Unit de la M inquiries to author Raoult (E-mail: didier.raoult@gmail.

M: Food Microbiology & Safety n hs ek eecmae ihpasi h aaae The database. the 219 in including peaks organisms 3290 with contains compared database account, were into peaks taken were these peaks 100 and of maximum a spectrum, each odn oD a,Rgs,adSap Mrk Darmstadt, conditions cultivation 37 (Merck, 3 at in Sharpe anaerobic 1960) (1) and others and Rogosa, Man (De Man, Germany) De to cording ehdo dnicto nlddthe included The identification Control Germany). of Automation Bremen, method GmbH, Biotyper Daltonics Maldi (Bruker acqui software and automatically 2.0. 3.0 were control Data equipped Flex laser. U.S.A.) using nitrogen Mass., 337-nm a Billerica, with Daltonics, (Bruker trometer dilut we (CFU), units 100 forming colony of number the enumerate eildltoswr ltdo ouba5 he lo agar or blood sheep France) 5% l’Etoile, Columbia Marcy on plated (BioMerieux, were dilutions Serial fmnindo h product. in the noted on sold were mentioned lo- species are if from bacterial that bought Labeled were Marseille. brands products in supermarkets All different study. cal this 5 in from used were yoghurts France 5 and ghurts) lt eecutd n h atrafo needn colonie spectrometry. independent mass 5 by from identified bacteria were the and counted, were plate t30at iue1Figure spectrometryMass Culture Materials Lactobacillus esrmnswr efre iha uoe Ims spec- mass II Autoflex an with performed were Measurements to and bacteria live contained products whether determine To ih ucinlfos( rboi rnsad4poitcyo- probiotic 4 and drinks probiotic (4 foods functional Eight µ ◦ fec etdsml n1m fseiewtr(iue1). (Figure water sterile of mL 1 in sample tested each of L .Atr7- nclto ooisotie nec agar each on obtained colonies inoculation 72-h After C. − iuinpoeuefrteietfiaino iebcei c bacteria live of identification the for procedure Dilution p.i od . . food. in spp. ◦ ,()5 CO 5% (2) C, 2 t37at m/z rm3t 5ka For kDa. 15 to 3 from ◦ Lactobacillus ,ad()5 CO 5% (3) and C, Lactobacillus grac- agar nandi ir products. diary in ontained spec- red ed 2 s : NucleoSpin ina h pce ee Splmnaytbe1.Frec prod- each independently. times For repeated were 3 1). analyses table MALDI-TOF identifica- and (Supplementary CFU-enumeration good level uct, showed species the Database at Bruker tion the in present already sso pcr rm19 from spectra of ysis oeua assaysMolecular h cr auspooe ySn n tes(20) htis, MALDI-TOF That by 2009). identified ( correctly others considered when and was Seng isolate by an proposed adapted values we score analysis, MALDI-TOF the For by ranked score. exh were isolate identification species the with their bacterial tho pattern peptide 15 from similar The most the created database. ing Bruker been the have in DSMZ spectra missing and Pasteur reference the and from strains collections by completed been has ratory r nteBue aaaefo a 00 n 5 atrain bacteria 850 and 2010, May The from database. database local the Bruker the in tra xrcinngtv oto ih100 with control extraction-negative yMnr n tes(2008). others and Menard by ecie yCroioadohr 06.PRamplification PCR 2006). ( others previ as and constructed Carcopino was by that described plasmid quantification a using C qC)ratoswr efre na MX3000 quantitative an in Real-time performed extractions. were reactions DNA (qPCR) of PCR series each in duced in eluted was DNA procedure. 100 manufacturer’s the to according n unicto of detec quantification The and Netherlands). the Amsterdam, Europe, (Stratagene N a xrce rm100 from extracted was DNA µ ≥ feuinbfe n trdat stored and buffer elution of L / pcr a score a had spectra 2/4 ° R iseMn i Mcee ae,Hed,France) Hoerdt, Nagel, (Macherey Kit Mini Tissue o.7,N.8 2011 8, Nr. 76, Vol. Lactobacillus Lactobacillus Lactobacillus ≥ Lactobacillus r ora fFo Science Food of Journal p a efre sreported as performed was sp. ..Uigteeciei,anal- criteria, these Using 1.9. tancleto forlabo- our of collection strain µ µ tan rmorlaboratory our from strains fseiewtrwsintro- was water sterile of L fec rdc ya by product each of g − 20 p.wr quantified were spp. ◦ ni sd An used. until C TM system M569 ously ibit- tion se

M: Food Microbiology & Safety rs s- ∗ ∗ io e- nd nd ort ation versy nt agars and 16S Labeled Not labeled Not labeled Not labeled Not labeled Not labeled Not labeled Not labeled Not labeled Bifidus sp. . This dis- tuf CFU/g found 5 L. casei may lead to differ- CFU/g (Figure 2). L. casei, L. paracasei, 9 instead of to 10 L. delbrueckii S. thermophilus 6 S. thermophilus, L. delbrueckii, L. paracasei specific agar) under various temper- L. paracasei gene 16S rDNA gene strain tuf ) which cannot be distinguished by conven- PCR amplification and sequencing L. delbrueckii S. thermophilus Lactobacillus S. thermophilus, L. delbrueckii, strains (Schillinger and others 2003). In addition group includes a number of species ( Lactobacillus ia in functional foods and yogurts. Functional food and yoghurts contained very large numbers Mass spectrometry has been used to control quality of dairy pro- to phenotypic identificationabout limitations, to taxonomic reject contro and (Dicks others and 2002) the others species 1996) name or to retain (Dellagl ent denomination for the2006). same probiotic strain (Huys and others of living bacteria varying from 10 L. rhamnosus, L. zeae tional phenotypic propertiesand (Klein others and 2001). others Theseof 1998; difficulties probiotic Holzapfel lactobacilli in have theof led correct to identification the frequent misclassification rDNA genes was alsothe species validated level for (Chavagnat the and2003). others identification 2002; of Ventura LAB and others at biotics determining a changeothers in 1999a; milk protein 1999b) profile2010). or (Fedele chemical In fingerprinting a the (Liufully and last employed to othe decade, achieve the mass characterization(Claydon of spectrometry and different bacteria others has 1996) beenof and bacterial succes allow strains qualitative to characteriz beand used others in 1999b). the However, to production of ourof yoghurt knowledge, bacterial there (Fedele is identification no from rep commercialized probioticby products this method. MALDI-TOFbacteria MS species presented level highone in specificity product the we at identified fermentedcordant dairy in food, results can andL. probably only casei be in explained by the fact that the mass spectrometry. The accuracy ofqPCR standard quantification assay of our wasamplification assessed of on samples the withproducibility basis known of concentrations the of quantification and the inothers the each linearity 2008). r PCR run of PCR (Menard amplification DNA a and sequencing of ature conditions was theMoreover, in reason 2 why webacteria products found we we higher found identified CFU/g. the a label. and discrepancy Numerous between the surveys the the have microorganisms revealed stated deficiencies in on the These concentrations were higher than the 10 (Columbia and by Schillinger when(Schillinger 1999). tested Probably the probiotics fact that yoghurts we used by 2 differe MRS agar L. delbrueckii S. thermophilus 6 S. thermophilus, L.delbrueckiiL.delbrueckiiL.delbrueckii L.delbrueckii L.delbrueckii L.delbrueckii L.delbrueckii L.delbrueckii L.delbrueckii L.paracaseiL.caseiL. lactisB.animalisL.delbrueckii L.caseiL.delbrueckiiL.delbrueckiiL.delbrueckii L.casei B.animalis L.L. L.delbrueckii lactis delbrueckii, L.delbrueckii L.delbrueckii L.casei L.delbrueckii L.delbrueckii B.animalis L.casei L.delbrueckii L. lactis L.delbrueckii L.delbrueckii L.casei Bifidussp. L.casei Bifidus sp. ol th- 10 nce and spp. spp.. 7 8 8 8 7 6 6 7 7 7 7 6 × Ventura 9 10 10 10 10 10 10 10 10 10 10 10 10 0.05 was 10 × × × × × × × × × × × × Bifidus in probiotic 2 8 9 1 3 4 8 9 4 2 4 3 < Bifidus spp. DNA/g Lactobacillus L. delbrueckii 7 7 7 7 7 7 7 7 9 7 7 8 8 value CFU/mL spp. DNA copies/g 0.44). Live bacteria 10 10 10 10 10 10 10 10 10 10 10 10 10 P = × × × × × × × × × × × × × , instead of instead of agar. In addition, we 1 1 1 1 3 5 6 1 2 1 9 3 1 P Lactobacillus spp. DNA copies/g ob- . In 2 products, we identi- agar cultures were positive sp. Columbia L. lactis Lactobacillus Vol. 76, Nr. 8, 2011 6 6 6 6 6 7 6 7 7 6 7 7 CFU/g, probiotic yoghurts con- r L. casei 6 10 10 10 10 10 10 10 10 10 10 10 10 8 copies of spp. DNA copies. As a result, all Lactobacillus × × × × × × × × × × × × 7 10 S. thermophilus Lactobacillus 0.80) (Table 1). As a result, probiotic × instead Lactobacillus in yoghurts (Table 1). The 92% of prod- Lactobacillus = to 10 CFU/g, and yoghurts contained 9 and 6 P 7 to 3 in probiotic yoghurts and 7 10 Lactobacillus × agar ( spp. in food. . . L. paracasei to 8 CFU/g. The number of L. delbrueckii Molecular and MALDI-TOF identification of lactic acid bacter 7 9 L. delbrueckii − Journal of Food Science genes were used for the identification of LAB at the species 10 × tuf Lactobacillus L. casei, Lactococcus lactis, and Bifidobacterium animalis For this study, our results were validated by independent me Overall, qPCR was positive in all the samples and based on the We compared the CFU/g obtained by the culture of prod- Both Columbia and Discordance between the labeled and identified bacteria. ods based on molecular assays, on agar dilution assays, and on Yoghurt 4Yoghurt 5 4 5 1 2 P. yoghurt 2P. yoghurt 3P. yoghurt 4Yoghurt 2 1 2 2 1 2 1 1 2 Yoghurt 3 3 1 and Prevention, Atlanta,considered Ga., significant. U.S.A.). A tained by qPCRused EpiInfo and version the 6.0 software CFU/g. (Centers For for data Disease Contr comparison, we which was mentioned onidentified the label. Similarly for yoghurt 1, we fied a discrepancy between thelabel and bacterial the strain strain announced identified.by For on the both the probiotic techniques drink the 3, presence we of found Product Trademark DNA copies/g agar agar MALDI-TOF MS P. drink 2P. drink 3P. yoghurt 1 2 1 1 3 4 10 compared the number of plasmid quantification we determinedlarge that all numbers products of contained Results and Discussion for all products.CFU/g No obtained significant when difference productson were was cultured observed on in Columbia or the ∗ M570 P: Probiotic. Yoghurt 2 1 5 P. drink 4 2 1 P. drink 1 1 1 Table 1 Statistical analysis Lactobacillus and sequencing targeting theand 16S ribosomal RNA (16S rRNA) ucts on Columbia and on drinks contained 10 level, as previously described (Chavagnat andand others 2002; others 2003). products contained 10 tained 10 ucts were correctly identifiedthe by MALDI-TOF CFU-enumeration MS and compared qPCR to we results identified as only in one product Streptococcus thermophilus of food was comparable to the CFU/g ( (Table 1). to 2 drinks, were identified by mass spectrometry.of We identified the prese

M: Food Microbiology & Safety oeta elhrs rnt(adrofadYug20) The 2008). Young and (Vanderhoof not or risk prese the health supplement However, on potential the 2006). whether benefit determine others health manufacturers and only a (Anadon 2001) confer (FAO/WHO label. number host” the adequate ad- when on in which stated microorganisms ministered “live microorganisms bac- as the promoted the are between 2011), and Probiotics discrepancy others identified a we and found we teria (Scharl products, some bacteria in of and concentrations or species as otie h irognsssae ntelbl(huisnand some (Theunissen Moreover, label 2005). the others on ot stated yoghurts microorganisms and probiotic the the (Marcobal contained of 54.5% label only product that found the Theunissen 2008). on listed species unad-missing contain products probiotic additional product Many vertized the 2007). in present others microorganism and probiotic (Perea purchased scien- the probiotics full of the the indicate name oth- not tific of did and often information and Masco vague, label rather 2005; was products the and others and Temmerman and 2005) 2003; Drisko ers others 2003b; and Fasoli 2003a, product others 2002; probiotic others of and labeling ung and quality microbiological 2Figure Lactobacillus Conclusion level. species of or LAB level cases of 46 genus identified the cultures and at use 213 misidentification nutritional collected or probiotic (2006) EU-fund for the others intended In and 2005). others Huys and project, (Theunissen label be- the without on detected ing were microorganisms various and identified ucinlfo otishg ocnrtoso atrasuch bacteria of concentrations high contains food Functional .casei L. − rmsano yoghurt. a of stain Gram and .animalisB. p.i od . . food. in spp. lactobacilli ucinlfoshv ostandardized no have foods Functional . and Bifidobacterium bifidobacteria p.wr incorrectly were spp. hra tesare others whereas , (Ye-s t ants hers ed ohrsMLITFM ol eaueu olfrbacteria for tool useful and a food be functional could identification. in MS that proved MALDI-TOF We yoghurts products. pre- their bacteria the in of clearlysented concentrations should the manufactures and species and the 2006), depen-characterize strain others and and species (Anadon are dent probiotics of functionality and safety eeeL ealaR atsot ,PnliC rliP 199 P. Traldi C, Pinelli B, Battistotti R, Seraglia 199 L, P. Traldi Fedele C, Pinelli B, Battistotti R, Seraglia L, Fedele aoiS aztoM iztiL os ,DlaloF Torri F, Dellaglio F, Rossi L, Rizzotti 2 M, R. Marzotto S, McCallum RV, Fasoli Rao M, Adelson C, Giles B, Bischoff J, Drisko ik M uPessE,DlaloF ae .19.Rcasfiainof Reclassification 1996. E. Lauer F, Dellaglio EM, Plessis Du LM, Dicks elgi ,FlsG,Trin .20.Tesau ftespecies the of status The 2002. S. Torriani GE, Felis F, Dellaglio hvga ,HutrM ieoJ ae G 02 Comparison 2002. MG. Casey J, Jimeno M, Haueter F, Chavagnat H, Richet AS, Fallabregues D, Benmoura M, Henry X, Probio Carcopino 2006. MM. Aranzazu MR, Martinez-Larranaga A, Anadon References eMnJ,Rgs ,Sap E 90 eimfrtecultivati the for medium A 1960. ME. Sharpe M, Rogosa JC, Man De r The 1996. DB. Gordon V, Edwards-Jones SN, Davey MA, Claydon fbceilsrisepoe ntepouto fygr ymti-sitdlsrdesorp- laser matrix-assisted by yogurt of 34:1385–8. production Spectrom Mass the J spectrometry. in mass employed tion/ionization strains bacterial of pro- yogurt in digestion 34:1338–45. protein Spectrom Mass bacterial J monitoring duction. for spectrometry mass tion/ionization Microbiol Food J Int analysis. PCR-DGGE by evaluated 82:59–70. as products probiotic commercial of paracasei Lactobacillus rboi rdcsfrlblcam yDAetato n pol 50:1113–7. and Sci extraction Dis DNA Dig by claims label for products probiotic einto fAC 3 stenoyeof neotype the as 334 ATCC of designation casei pno.ItJSs vlMcoil52:285–7. Microbiol Evol Syst J Int opinion. 96 asnadLse 91and 1971 Lessel and Hansen 1916) 23:130–5. Bacteriol eemnto fHVtp 6ad1 ia odi evclser fwmnrfre to Lett Microbiol referred FEMS women lactobacilli. of of smears identification the cervical for in load viral 18 78:1131–40. Virol and Med 16 J colposcopy. type HPV of 45:91–5. Determination Pharmacol Toxicol Regul assessment. safety and Regulation Union. European the irognssuigms pcrmty a itcnl14:1584–86. Biotechnol Nat spectrometry. mass using microorganisms TC33and 393 ATCC o.7,N.8 2011 8, Nr. 76, Vol. n ytBceil46:337–40. Bacteriol Syst J Int . atbclu rhamnosus Lactobacillus atbclu paracasei Lactobacillus TC180a atbclu eenm rev., nom. zeae Lactobacillus as 15820 ATCC r .casei L. ora fFo Science Food of Journal subsp. oln ta.18.Rqetfran for Request 1989. al. et Collins n .20.Bceilcomposition Bacterial 2003. S. ani 217:177–83. a arxasse ae desorp- laser Matrix-assisted 9a. b ulttv characterization Qualitative 9b. casei mrs hi ecinanalysis. reaction chain ymerase atbclu casei Lactobacillus fpriltfgn sequences gene tuf partial of n eeto ftename the of rejection and , obiL aae .2006. C. Tamalet L, Boubli isfraia urto in nutrition animal for tics pdietfiaino intact of identification apid 0.Eauto ffive of Evaluation 005. nof on atbclu casei Lactobacillus lactobacilli (Orla-Jensen M571 Appl J . subsp.

M: Food Microbiology & Safety rticle. Lactobacillus , ent and use of tuf -independent analysis of -specific identification of Raoult D. 2009. Ongoing . Analysis, characterization, p-specific and RAPD-PCR in commercial dairy products. s. Clin Infect Dis 46 (Suppl 2): –22. species and their direct application Lactobacillus acidophilus obiol 81:1–10. eria by matrix-assisted laser desorption strains from probiotic yogurts. Curr Microbiol Bifidobacterium Bifidobacterium longum and Lactobacillus , and Lactobacillus Lactobacillus delbrueckii group, S67–72. and loci of the tuf genes in revolution in bacteriology: routine identification of bact ionization time-of-flight mass spectrometry. Clin Infect Dis 49:543–51. gene-based primers for the multiplex PCR detection of casei probiotic products by69:220–6. denaturing gradient gel electrophoresis.ganisms Appl in South Environ African98:11–21. Microbiol products using PCR-based DGGE analysis. Int J Food Microbiol J Food Prot 72:93–100. bacterial isolates from probiotic products. Int J Food Micr for species identification. Appl Environ Microbiol 69, 6908 commercial probiotic strains. J Dairy Sci 85:1039–51. 47:453–6. analyses for rapid differentiation of Ventura M, Canchaya C, Meylan V, Klaenhammer TR, Zink R. 2003 Sheu SJ, Hwang WZ, Chen HC, Chiang YC, Tsen HY. 2009. Developm Temmerman R, Scheirlinck I, Huys G, Swings J. 2003b. Culture Supporting Information The following supporting information is available forSupplementary this Table a 1. Please note: Wiley-Blackwell is notfunctionality responsible of for the any content supporting or materialsAny supplied queries by (other the than authors. the corresponding missing author material) for should the article. be directed to Theunissen J, Britz TJ, Torriani S, Witthuhn RC. 2005. IdentificationVanderhoof of JA, probiotic Young R. microor- 2008. Probiotics in the United State Temmerman R, Pot B, Huys G, Swings J. 2003a. Identification and antibiotic susceptibility of Yeung PS, Sanders ME, Kitts CL, Cano R, Tong PS. 2002. Species Seng P, Drancourt M, Gouriet F, La SB, Fournier PE, Rolain JM, Schillinger U, Yousif NM, Sesar L, Franz CM. 2003. Use of grou h promoters? Nature ependent and culture- lecular quantification of siology of probiotic lactic . Appl Environ Microbiol ition. Am J Clin Nutr 73: n Infect Dis 46 (Suppl 2): U. 2001. Taxonomy and im- on of the bacterial composition technol 20:135–41. he current probiotics market, urt: the number of living bacteria anisms of probiosis and prebiosis: odulatory activity of presumptive takis N, Yiangou M. 2010. Evalu- acilli from novel-type probiotic and . Eur J Clin Microbiol Infect Dis Lactobacillus Vol. 76, Nr. 8, 2011 r loads to predict bacterial vaginosis. Clin Infect Dis Atopobium vaginae strains. Int J Food Microbiol 140:154–63. and spp. in food. . . Lactobacillus Journal of Food Science 65:3763–6. (Suppl 2):S58-61. considerations for enhanced functional foods. Curr Opin Bio and changes inS76-9. the marketplace: a European perspective.in probiotic Cli yoghurt decreases under exposure to room temperature.mild Digestion yoghurts 83:13–7. and their stability during refrigerated storage. Int J Food Microbiol 47:79–87. 47:33–43. 2007. Identification andColumbian characterization dairy of products. J starter Appl lactic Microbiol 103:666–74. acid bacteria and probiotics from 454:690–1. Gardnerella vaginalis of commercial probioticanalysis. products J Pediatr by Gastroenterol Nutr terminal 46:608–11. restriction fragmentindependent length qualitative analysis polymorphism of probioticFood products Microbiol claimed 102:221–30. to contain bifidobacteria. Int J ionization mass spectrometry. Rapid Commun Mass Spectrom 24:1365–70. portant features of365S-73S. probiotic microorganisms in food and nutr of species identity ofRes commercial Microbiol bacterial 157:803–10. cultures intended for probiotic or nutritionalacid use. bacteria. Int J Food Microbiol 41:103–25. ation of adhesion capacity, cellprobiotic surface traits and immunom Sanders ME. 2008. Probiotics: definition,Saulnier sources, DM, Spinler selection, JK, Gibson GR, and Versalovic J. 2009. uses. Mech Clin Infect Dis 46 Scharl M, Geisel S, Vavricka SR, Rogler G. 2011. Dying in yogh Schillinger U. 1999. Isolation and identification of lactob Saxelin M. 2008. Probiotic formulations and applications, t Perea VM, Hermans K, Verhoeven TL, Lebeer SE, VanderleydenRaoult J, D. De 2008. Keersmaecker SC. Human microbiome: take-home lesson on growt Raoult D. 2010. The globalizationReid of G. intestinal 1999. microbiota The scientific basis for probiotic strains of Masco L, Huys G, De BE, Temmerman R, Swings J.Menard 2005. Culture-d JP, Fenollar F, Henry M, Bretelle F, Raoult D. 2008. Mo Liu J, Qiu B, Luo H. 2010.Marcobal Fingerprinting A, Underwood of MA, Mills yogurt DA. 2008. products Rapid determinati by laser desorption spray post- M572 Lactobacillus Holzapfel WH, Haberer P, Geisen R, Bjorkroth J, Schillinger Huys G, Vancanneyt M, D’Haene K, Vankerckhoven V, Goossens H, Swings J. 2006. Accuracy Klein G, Pack A, Bonaparte C, Reuter G.Kotzamanidis 1998. C, Taxonomy Kourelis and A, phy Litopoulou-Tzanetaki E, Tzane

M: Food Microbiology & Safety

REMERCIEMENTS

A ma femme, Estelle, qui est ma bonne étoile,

A Eva Catalina et Satine Alizée, mes amours qui ont changé ma vie,

A mes parents, Claire et Claude, qui m’ont transmis le goût de la Science.

Au Professeur Didier Raoult,

Qui m’a fait confiance

Qui m’a réellement appris l’esprit de la recherche scientifique

Qui m’a donné les moyens d’aller jusqu’au bout de ces travaux

Et que je remercie profondément pour m’avoir accueilli dans son équipe.

Je lui exprime ici mon plus grand respect

et com bien j’apprécie ses idées et ses méthodes souvent originales, parfois

paradoxales et toujours si efficaces.

J’exprime également mes remerciements au Professeur Jean -Louis Mège qui a

accepté de présider le Jury de cette thèse.

Ma gratitude s’adresse aussi à mes rapporteurs qui ont bien voulu juger ce travail.

Je remercie l’équipe de recherche URMITE et l’équipe du service de Nutrition,

Maladies Métaboliques et Endocrinologie du CHU de la Timone et notamment le

Professeur Vialettes, le Professeur Valero et Marie Maraninchi.

Un grand merci à Sophie, Jean-Christophe, Fadi, Isabelle, Manolis, Fabrice,

Caroline, Stéphanie, Aurélie, Annick, Alpha, Véronique, Mical, Roch, Hervé, Patrizia,

Mireille, Juline, Franck, Jean-Marie sans qui cela n’aurait pas été possibl e. 182