Research Article

Predominant Role of Host Genetics in Controlling the Composition of Gut Microbiota

  • Zaruhi A. Khachatryan,

    Affiliation: Laboratory of Molecular Genetics, Institute of Molecular Biology of Armenian National Academy of Sciences, Yerevan, Armenia

  • Zhanna A. Ktsoyan,

    Affiliation: Laboratory of Molecular Genetics, Institute of Molecular Biology of Armenian National Academy of Sciences, Yerevan, Armenia

  • Gayane P. Manukyan,

    Affiliation: Group of Molecular and Cellular Immunology, Institute of Molecular Biology of Armenian National Academy of Sciences, Yerevan, Armenia

  • Denise Kelly,

    Affiliation: Rowett Institute of Nutrition and Health, University of Aberdeen, Aberdeen, United Kingdom

  • Karine A. Ghazaryan,

    Affiliation: Group of Molecular and Cellular Immunology, Institute of Molecular Biology of Armenian National Academy of Sciences, Yerevan, Armenia

  • Rustam I. Aminov mail

    Affiliation: Rowett Institute of Nutrition and Health, University of Aberdeen, Aberdeen, United Kingdom

  • Published: August 26, 2008
  • DOI: 10.1371/journal.pone.0003064



The human gastrointestinal tract is inhabited by a very diverse symbiotic microbiota, the composition of which depends on host genetics and the environment. Several studies suggested that the host genetics may influence the composition of gut microbiota but no genes involved in host control were proposed. We investigated the effects of the wild type and mutated alleles of the gene, which encodes the protein called pyrin, one of the regulators of innate immunity, on the composition of gut commensal bacteria. Mutations in MEFV lead to the autoinflammatory disorder, familial Mediterranean fever (FMF, MIM249100), which is characterized by recurrent self-resolving attacks of fever and polyserositis, with no clinical signs of disease in remission.

Methodology/Principal Findings

A total of 19 FMF patients and eight healthy individuals were genotyped for mutations in the MEFV gene and gut bacterial diversity was assessed by sequencing 16S rRNA gene libraries and FISH analysis. These analyses demonstrated significant changes in bacterial community structure in FMF characterized by depletion of total numbers of bacteria, loss of diversity, and major shifts in bacterial populations within the Bacteroidetes, Firmicutes and Proteobacteria phyla in attack. In remission with no clinical signs of disease, bacterial diversity values were comparable with control but still, the bacterial composition was substantially deviant from the norm. Discriminant function analyses of gut bacterial diversity revealed highly specific, well-separated and distinct grouping, which depended on the allele carrier status of the host.


This is the first report that clearly establishes the link between the host genotype and the corresponding shifts in the gut microbiota (the latter confirmed by two independent techniques). It suggests that the host genetics is a key factor in host-microbe interaction determining a specific profile of commensal microbiota in the human gut.


The human gut microbiota has been shaped by the long co-evolutionary history of symbiotic host-microbe interaction [1] and has evolved to play an important role in maintaining human health by preventing colonization by pathogens, degrading dietary and in situ-produced compounds, producing nutrients, and shaping and maintaining the normal mucosal immunity [2][4]. Recently, other important functions of commensal microbiota became apparent including an anti-inflammatory effect on gut epithelium [5], [6], the influence on lipid metabolism of the host [7] and association with obesity [8], [9] as well as the involvement in intestinal homeostasis, repair and angiogenesis [10], [11].

Recently, the intricate molecular mechanisms behind this host-microbe cross-talk began to emerge. The innate immune system of the mammalian host senses bacteria through an impressive array of receptors, called Toll-like (TLR) and Nod-like (NLR) (also called CATERPILLER, NOD-LRR, and NACHT-LRR) receptors. The first mammalian TLR was discovered a decade ago [12] and the functionality of the first NLRs (then called NODs) as sensors of bacterial ligands was established in 2001 [13]. TLRs are predominantly but not exclusively located in the transmembrane region of host cells, while NLR proteins are mostly located in the cytoplasm [14]. Both groups of receptors contain leucine-rich repeats (LRR), which are involved in detection of danger signals and, in particular, of whole bacteria and bacterial products through the recognition of pathogen associated molecular patterns (PAMP). PAMP signalling through the innate immunity receptors appeared to be important not only for the recognition of, and launching defences against, the invading pathogens but also for the maintenance of intestinal epithelial homeostasis and for protection against gut injury and associated mortality [11]. Bacterial products are also crucial for the normal immune system development; for example, a single bacterial polysaccharide (PSA) from a commensal bacterium Bacteroides fragilis was shown is capable of directing the complete cellular and physical maturation of the developing immune system [15]. In murine experimental colitis models, signalling by probiotic bacterial DNA ligands through TLR9 exerted a strong anti-inflammatory effect [16] while signalling through TLR2 seems important for regulation of mucosal inflammation and maintenance of intestinal epithelial barrier integrity [17], [18].

Another type of regulation of the host metabolism by microbes can be through the low molecular weight bacterial metabolites such as short chain fatty acids (SCFAs) and, in particular, butyrate. Butyrate is the preferred energy source for the colonic epithelium and may play an important role in colonic health and prevention of colorectal cancer and colitis [19][21].

Because of limitations imposed by the extreme diversity of microbiota in the conventional gut, which does not allow monitoring the behaviour of a single bacterium, much less is known of how the bacterial counterpart responds to the host-generated molecular signals. Initially, the bacterial response was considered to be limited to various strategies used by gut bacteria to forage the endogenously produced substrates of the host [22], [23]. Interestingly, however, one of the aspects of this interaction, namely the specific fucosylation, seems important for both sides involved in the host-bacterial dialogue. On one side, the intestinal bacteria induce mucin fucosylation in the gut epithelium thus creating a specific nutritional econiche [22], [24]. On the other side, fucosylation of the cell surface in a mammalian-like fashion appeared to be essential for the gut bacteria to establish the successful mutualistic relationship with the host. The lack of this property may result in the failure of bacteria to colonize the mammalian intestine [25]. Yet another “language” of host-bacterial communication could be through the quorum sensing (QS), which utilizes hormone-like compounds and, in E. coli, has been implicated in a potential cross-communication between the luxS/AI-3 bacterial QS system and the epinephrine host signalling system [26]. Bacteria may also sense the immune status of the host and respond by the expression of a quorum-sensing dependent virulence determinant [27]. These two examples of quorum-sensing-mediated signalling between the host and bacteria are taken from the pathogenic bacteria but it cannot be excluded that this type of communication is common among gut symbiotic bacteria as well.

What happens if this fine-tuned molecular cross-talk is compromised and the host fails to recognize bacterial components as, for example, in the case of inflammatory bowel disease (IBD) such as Crohn's disease (CD)? In a subset of CD patients, mutations in the NOD2/CARD15 protein have been associated with susceptibility to the disease [28], [29]. Bacterial ligand for this receptor has been recently identified as bacterial muramyl dipeptide [30], [31] and failure to recognise this ligand in CD patients leads to the breach of tolerance and to the launch of an aggressive Th1-skewed response against the antigens of normal diet and commensal bacteria as well as to production of autoantibodies [32][34]. The microbiota of these patients demonstrates diminished diversity, in particular, due to the depletion of bacteria belonging to the Firmicutes and Bacteroidetes phyla [35], [36]. Interestingly, the loss of the transcription factor T-bet, which regulates the innate immune system, influences bacterial populations in the murine gut in a way that they become colitogenic and this colitis is communicable to genetically intact hosts [37]. Presently, it is not clear how the composition of the normal microbiota is restructured to become a “pathobiota” but, nevertheless, this is the first experimental proof of the view that the community as a whole may demonstrate pathogenic properties, not a single pathogenic bacterium.

In our study, we investigated the effect of mutations in a single gene of the host on commensal gut microbiota. The gene under investigation, MEFV (for MEditerranean FeVer), encodes a protein called pyrin/marenostrin, which is involved in regulation of innate immunity [38]. Structurally, this 781-residue protein consists of a 92-amino acid N-terminal PYRIN (PYD) domain, a B-box zinc finger, a coiled-coil region and a ~200-amino acid C-terminal B30.2/rfp/SPRY domain [39]. Mutations in this gene lead to an autoinflammatory disorder, familial Mediterranean fever (FMF, MIM249100), which is characterised by short recurrent self-resolving attacks of fever and polyserositis, with no clinical signs of disease in remission periods. The carrier rate and occurrence of disease is high in populations originating from the Mediterranean basin, including Sephardic Jews, Armenians, Arabs, and Turks. Articles describing positional cloning of the gene by two independent consortia appeared in 1997 [40], [41] and to date 166 MEFV mutations and single nucleotide polymorphisms (SNPs) are detected in the gene ( The hotspot of MEFV mutations is localized in exon 10, which encodes the B30.2/rfp/SPRY domain, and this single locus inheritance greatly facilitated our analysis of host genetics in comparison with other disorders with more complex genetics. Genome-wide scans for IBD susceptibility, for example, have resulted in identification of seven loci that are confirmed and replicated in several studies, but with many more that need independent confirmation and verification [42].

The present work is the first attempt to analyze simultaneously the genetics of both sides involved in the host-gut microbiota relationship. All study participants were analyzed for MEFV mutations and the composition of gut microbiomes was assessed by two culture-independent techniques, fluorescent in situ hybridization (FISH) and 16S rDNA gene library analyses. For the first time, our findings conclusively establish the key role of the host genotype in host-microbe interaction, which determines a highly specific composition of commensal microbiota in the human gut.


MEFV Genotyping

Nineteen FMF patients with clinically confirmed disease and eight healthy individuals were subjected to genotyping (Table 1). We performed genotyping of the exon regions of the MEFV gene, which is thought to be a single genetic determinant of FMF, in particular exons 2 and 10, where the majority of disease-associated mutations were identified ( FMF-associated mutations were found in all FMF patients displaying the clinical signs of the disease, as well as in one healthy Armenian control subject (in a simple heterozygous state) without disease-related symptoms (Table 1). All mutations were located in exon 10, only synonymous substitutions were detected in exon 2, irrespectively of the health status. The majority of mutations in FMF patients were M694V, which were present in the compounded form with other disease alleles in eight patients and in a simple heterozygous form in six patients. In the latter group, no other disease-related mutations in exon 2, except neutral SNPs, were detected. Homozygotes were represented exclusively by the M680I mutation (Table 1). Interestingly, the SNPs in the control group were all in the heterozygous state, while the homozygous forms were encountered only among FMF patients.


Table 1. MEFV genotypes of FMF patients and control subjects and gut microbiota analysis methods applied.


Phylogenetic and comparative analysis of 16S libraries

Three 16S rDNA libraries from the fecal samples of healthy controls as well as of FMF patients in remission and disease attack were generated (designated as S, SR and SA in Table 1). After the quality control a total of 1328 validated sequences (572 for healthy controls, 629 for FMF remission and 127 for FMF attack) were analyzed and phylogenetic analysis was performed to establish taxonomic positioning of sequences obtained (Table 2). Among 1328 clones analyzed, there were 268 distinct OTUs and the vast majority (95.95%) fell into the two major phyla, Bacteroidetes with 99 OTUs (38.15%) and 714 sequences (53.76%) and Firmicutes with 158 OTUs (57.8%) and 580 sequences (43.67%). Proteobacteria were represented by 9 OTUs (3.47%) and 30 sequences (2.26%). Only four Actinobacteria sequences (2 OTUs) were detected, which were all derived from one FMF patient in remission.


Table 2. Phylogenetic distribution of 16S rRNA gene sequences and phylotypes retrieved from healthy individuals and FMF patients in remission and attack.


In all three libraries, bacteria belonging to the Bacteroidetes phylum were the most abundant, containing more than the half of the sequences in each library (from 50.72% to 56.64%) but a relatively smaller proportion of OTUs (from 35.19% to 40.91%), whereas the Firmicutes phylum had the larger proportion of phylotypes (from 55.30% to 60.49%) with the less copious sequence numbers (from 41.43% to 45.95%) (Table 2). At the phylum level, the relative proportions of Bacteroidetes and Firmicutes were not significantly different among the three libraries, however, pairwise comparisons of each 16S rDNA library to every other library by using the online library compare tool at RDP-II (Naive Bayesian rRNA Classifier) revealed significant differences within the lower taxa levels (Table 3). In particular, in asymptomatic FMF patients in remission, the proportion of Enterobacteriaceae, Acidaminococcaceae, Ruminococcus and Megasphaera was significantly increased in comparison with control subjects, while Roseburia was significantly reduced (Table 3). In the active disease, the proportion of Prevotellaceae, Dialister and Prevotella was significantly lower than in healthy controls while the Porphyromonadaceae, Phascolarctobacterium, Faecalibacterium, and Parabacteroides were significantly increased (Table 3). Interestingly, there were also differences due to the disease state (remission or attack) and the affected bacterial groups were Acidaminococcaceae, Porphyromonadaceae, Megasphaera, Dialister, Faecalibacterium, and Parabacteroides (Table 3).


Table 3. Comparison of 16S rRNA gene libraries derived from healthy controls and FMF patients in remission and attack.


Community analysis of 16S rDNA libraries

The estimates of diversity, richness, evenness and library coverage for the three 16S rDNA clone libraries studied are shown in Table 4. At the phylotype level, the most diverse gut bacterial community was observed in FMF patients in remission. The slightly lower indices were found for the healthy microbiota, while the bacterial diversity during disease attacks was consistently lower. The similar trend was observed for another community parameter, species richness, which was the highest in the gut of FMF patients in remission and in healthy volunteers but much lower during disease attacks. On the contrary, evenness index, which describes how evenly the individual sequences are represented in phylotypes, was the highest in the disease attack community (Table 4). Good's coverage was 89.35–90.73% for the sequence sets from FMF patients in remission and healthy controls but only 77.2% for sequences from FMF patients in attack reflecting the lower number of subjects and sequences collected.


Table 4. Indices of diversity, richness and evenness and library coverage for 16S rRNA gene libraries.


Rarefaction analysis of clone libraries essentially confirmed these findings and suggested that FMF patients in remission possess the most diverse gut bacterial community, which is even higher than in healthy subjects (Fig. 1). Because of limited number of sequences in the clone library from the acute disease, we extrapolated the curve using an empiric regression function (for details, see the footnote to Fig. 1). The extrapolated curve confirmed the initial findings suggesting diminished diversity of gut bacteria in FMF attack in comparison with disease remission and control.


Figure 1. Rarefaction curves obtained for clone libraries from healthy controls and FMF patients.

The rarefaction curves were generated by plotting the number of phylotypes (defined at 99% level) against the number of clones sequenced. The shape of the curves of observed phylotype richness indicates a trend of diminishing chance of finding new phylotypes as the sampling continues. Data for library in attack were extrapolated by empiric regression function a×x2b×exp(−c×x). Coefficient of regression determination (R2) was 99.9%. Regression statistics: adjusted R2 – 0.9991; standard error – 0.1720; a = 1.8690; b = 0.7104; c = 8.0466E-04. Healthy controls are shown in dark blue; FMF patients in remission – in red and attack – in green, the extrapolated curve is shown in light blue.


Collector's curves of the observed and estimated phylotype richness are shown in Fig. 2. In healthy subjects, both estimates, Chao1 and abundance-based coverage estimator (ACE), were highly similar and, up to the point of collection of ca. 320 clones, were in a steady and congruent increase together with the observed phylotype number (Fig. 2A). In this sampling area, almost any new clone sequence added represented a novel phylotype. Adding new sequences after this point, however, led to the gap expansion between the estimated and observed phylotype numbers suggesting repeated sampling of the phylotypes already present in the sample. In FMF remission, the gap between the observed and estimated phylotype richness was constant after the sampling point of 290 clones (Fig. 2B). Confirmatory to the previous analyses, the phylotype number estimates at the sampling point with 500 clones gave the higher numbers of phylotypes in FMF remission than in the norm (220 vs. 160). The difference between the estimates and observed phylotype richness is highest in acute FMF possibly reflecting the insufficient coverage in this group (77.2%).


Figure 2. Collector's curves of the observed and estimated phylotype richness of 16S libraries.

Collector's curves of the observed (shown in dark blue) and estimated (Chao1 (light blue) and ACE (red)) phylotype richness calculated for healthy subjects (A), FMF remission (B), and FMF attack (C) with the phylogenetic depth defined at 99% level.


FISH analysis

Twelve Cy3-labelled oligonucleotide probes that target the total and specific predominant groups of human gut bacteria were used in this study to enumerate bacteria in the feces of FMF patients and healthy controls (for details, see Materials and Methods). Hybridization signals were detected for all bacterial groups, indicating their presence as dominant microbiota in all fecal samples analyzed.

FISH analysis revealed that the composition of fecal bacteria in FMF patients in both inactive and active phases of the disease is different from that of the healthy controls. The results of enumeration of bacterial cells in each group and their relative proportions versus the universal bacterial probe Eub338 are summarized in Table 5. All fecal samples were processed exactly in the same way and the count of cells hybridized with Eub338 can be calculated to give the absolute numbers of bacteria per gram of feces. No significant difference in concentration of total fecal bacteria was found between the healthy controls and FMF patients in remission, the mean values 5.47×1010/g and 5.12×1010/g, respectively (Table 5). In the acute stage of the disease, however, the concentration of bacteria in fecal material was substantially lower, at 3.56×1010/g.


Table 5. Mean numbers of predominant bacteria in fecal samples of FMF patients in both stages of the disease and healthy controls detected by FISH.


Among the taxonomic groups tested, the highest proportion was detected with the Bac303 probe, which targets the Bacteroides group. It gave the concentration of these bacteria at 1.51×1010/g (27.3% of total bacteria) in healthy controls, with elevation in FMF patients to 1.86×1010/g (39.1%) in remission and to 1.35×1010/g (38.6%) in the attack period (Table 5). Cluster XIVa bacteria (Clostridium coccoides group) was the second most abundant group enumerated with the Erec482 probe comprising 1.09×1010/g (21.1%) in the healthy controls and approximately 1.01×1010/g in both, FMF remission and attack phases (18% and 28.9% of total bacteria, respectively). The E. rectale-Roseburia probe Rrec584, which is a nested probe within the C. coccoides group (enumerated with Erec482), detected 6.5%, 8.3% and 18.0% within the total bacteria probe, Eub338, and 41.5%, 43.8% and 59.8% within the Erec482 probe in healthy controls, FMF remission and FMF attacks, respectively. There were significant populations of Fecalibacterium prausnitzii related bacteria (Fprau645), the second important group of butyrate producers within clostridial cluster IV, in all studied groups, with the tendency to increase in acute FMF. These bacteria comprised 0.54×1010/g (11.3%) in healthy controls, 0.5×1010/g (10.4%) in FMF remissions, and 0.53×1010/g (13.8%) in acute stage of FMF. Another group within clostridial cluster IV, ruminococci, which was detected with the combination of probes Rfla729 and Rbro730, demonstrated lower levels in FMF patients, especially in the acute phase of the disease, compared to healthy controls. In particular, the members of the R. flavefaciens/R. bromii group accounted for 0.59×1010/g (9.9%) in healthy controls, 0.29×1010/g (5.3%) in FMF remissions, and 0.09×1010/g (2.6%) in acute FMF. Similarly, clostridial cluster IX representatives, detected with the probe Prop853, were estimated to be at 0.64×1010/g (11%) in healthy group, whereas in FMF remission these bacteria were reduced to 0.44×1010/g (7.1%), with even more significant reduction in the acute stage of FMF, accounting for 0.13×1010/g (3.5%). Atopobium group also tended to be less represented in gut microflora of FMF patients compared with healthy subjects, steadily decreasing from 0.42×1010/g (7.8%) in healthy controls to 0.29×1010/g (6.3%) in remission and 0.17×1010/g (4.5%) in attack. Members of the Bifidobacterium, Enterobacteriaceae and Lactobacillus-Enterococcus groups, which were detected by the probes Bif164, EnterobacD and Lab158, respectively, did not reveal significant differences between the studied groups because they were found in low numbers and demonstrated high inter-individual variability (Table 5).

We also attempted to evaluate the difference in gut microbiota composition in remission and attack periods of disease in the “pure” form, when the genetic, phenotypic and environmental variables are removed from the analysis. For this, we monitored the intra-individual shifts in total and main gut bacterial populations during the remission and attack phases of the disease in two patients and these observations confirmed and reinforced the tendencies found between the larger combined cohorts of FMF sufferers in the remission and acute stages (Table 6). First, the total numbers of bacteria in the gut enumerated with the Eub338 probe decreased about two-fold. Second, in a number of bacterial groups, the tendencies detected earlier became more profound and statistically significant. In particular, these changes included a dramatic increase in concentration of cluster XIVa bacteria and particularly of E. rectale-Roseburia cluster during the acute phase of the disease, with the reduction in concentration of cluster IX bacteria and ruminococci. The fall of proportion of populations of lactic acid bacteria enumerated with the Bif164 and Lab158 probes during the disease attack was significant in one patient, FMF 13-123, while in the other, FMF 87, the differences in this group were not significant. Thus, the monitoring of intra-individual bacterial shifts during the remission and attack phases validated and confirmed the findings involving the larger cohorts (Tables 5 and 6).


Table 6. Distribution of main bacterial groups in two FMF patients in remission and attack.


The part of FMF patients was undergoing colchicine therapy and we compared the composition of gut bacteria in respect to this treatment (Table 7). There was no difference in acute disease but in remission a certain tendency to normalization was observed in the Bacteroides group and in clostridial cluster IX bacteria. Nevertheless, these differences were not statistically significant and the data sets of colchicine-treated and untreated patients were combined together for the microbiota analyses.

Multiple comparative analyses of relative proportions of bacterial populations in the gut of the three cohorts were performed using the Kruskal-Wallis test and the results are presented in Fig. 3. The proportion of Bacteroides was significantly higher in both diseased cohorts in comparison with the healthy controls, while the proportion of cluster XIVa bacteria was only increased in the acute phase of the disease in comparison with the control group and remission (Fig. 3A). The fraction of E. rectale-Roseburia cluster among total bacteria was substantially increased, especially in active disease (Fig. 3B), while bacteria of cluster IX and Atopobium decreased in the attack phase of the disease in comparison with controls and remission (Fig. 3C). The proportion of ruminococci was significantly lower in FMF patients (Fig. 3D), while no significant differences were detected among F. prausnitzii, bifidobacteria, lactic acid bacteria, and enterobacteria (Fig. 3B, 3D and 3E).


Figure 3. Comparison of the main bacterial groups identified by FISH between healthy and diseased groups.

Comparative analysis of relative proportions of bacterial groups detected by different oligonucleotide probes in fecal samples between FMF patients in remission and attack periods and healthy control group (by Kruskal-Wallis test for multiple comparisons). On box-and-whisker plots the points within boxes indicate the mean values of bacterial proportions, the boxes represent standard errors (SE) and the vertical bars represent standard deviations (SD).


Table 7. Distribution of main bacterial groups in respect to colchicine treatment.


Discriminant function analysis (DA) is a powerful tool to extract specific patterns in multiple datasets and we used this approach to analyze the specific gut microbiota distribution patterns in our three cohorts. The summary of discriminant function analysis with 20 subjects and 10 variables is shown in Fig. 4. It revealed three clusters in bacterial distribution that are highly specific, well separated and distinct for healthy controls, FMF remission, and FMF attack. As shown in Table 8, the total percentage of correct classification was 95%, thus 95% of total subjects were classified into the correct groups (100% in healthy controls, 83.3% in remission and 100% in attack). The validity of the variables in differentiating the groups was supported by the Wilks's lambda coefficient, which may vary from 0 to 1 (the smaller the lambda, the more the variable differentiates the groups), and in this case it had a value of 0.045. The F-test of Wilks's lambda, which is used to test if the discriminant model as a whole is significant, produced (20,16) = 2.95; p<0.016. There was only one case of misclassification from a total of 20 when the remission stage was assessed as acute. This patient refused the colchicine therapy and was suffering from very frequent FMF attacks and there is a possibility that at the time of sampling the microbiota was not restructured back to the composition specific for the remission stage. Thus the model displayed good predictive accuracy and suggested the specific patterns of gut microbiota distribution in the healthy and diseased states.


Figure 4. Discriminant function analysis of FISH data.

Number of variables in model: 10; grouping - 3 groups. Wilks's lambda coefficient – 0.045; F-test - (20,16) = 2.95, p<0.016. Root 1, 2 – discriminant functions 1 and 2 (1st and 2nd canonical roots). Regions of group distribution are outlined by ellipses; healthy is shown in blue triangles, remission – in red circles, and attack – in green squares.


Table 8. Classification table of correct estimates based on FISH variables.



This is the first work reporting the specific changes in the human gut microbiota due to the mutations in a single gene of the host. Suggestions that the composition of gut microbiota is host genotype-dependant have been proposed earlier [43], [44], however, these works were based on twin studies and the host genotype was not assessed. Gut inflammatory diseases such as IBD have been long known as having a substantial genetic predisposition component and a recent review of genome-wide scans reported on the seven confirmed and replicated cases of genetic susceptibility loci, with many more, however, that need independent confirmation and verification [42]. Although the gut microbiota of IBD patients was subjected to several diversity analyses [35], [36], [45], the host genetics was not assessed, most probably because of complicated genetics of IBD. Thus, uncovering the genetic mechanisms that are involved in regulation of composition of human gut microbiota remains a challenging task. Evidence obtained in a study of bacterium-derived cellular fatty acids in the stool samples of six mouse strains congenic for the major histocompatibility complex (MHC) suggested the importance of host immune system in this process [46]. The obvious advantage of our model is that the genetics of FMF is well defined and confined to a single gene thus facilitating the host genotype analysis. Thus, for the first time, we had the opportunity for direct comparison of the host and gut community genotypes.

Initially, we genotyped the FMF patients and healthy controls for the presence of disease-specific mutations in exons 2 and 10 of the MEFV gene. In all clinically confirmed cases we were able to detect the presence of known mutations, all of them in exon 10, which encodes the SPRY/B30.2 domain. Among our very limited cohort of seven healthy controls of Armenian origin, one heterozygote was detected, suggesting that the mutation carrier rate in this population is at least 14%, which is within the range of carrier rates found in populations of other ethnic origin [47][50]. The high carrier frequency in affected populations suggests the selective advantage conferred by the heterozygous state but no perceptible biological advantage of the carrier state is presently identified [49]. Positive selection to maintain the high frequency of the heterozygotes should also overcome the negative selection imposed by the increased morbidity and mortality rates among the homozygotes and compounded heterozygotes. Clinical manifestations during the disease attack are consistent with a strong inflammatory response, with fever, massive influx of polymorphonuclear leukocytes into the affected tissues, neutrophilia, and acute-phase response. In some patients this leads to systemic amyloidosis resulting in renal failure and death.

The penetrance of mutations that are all missense mutations, varied widely. Apart from the combined heterozygotes, which support the recessive nature of mutations, there were also simple heterozygotes but with different phenotypes. The simple heterozygosity in clinically confirmed disease was exclusively exemplified by a M694V mutation and, similarly to another study [51], suggested the autosomal dominant inheritance probably due to the importance of this residue in the interaction of pyrin with caspase-1 [52]. On the other end of the mutation spectrum is the simple M680I heterozygosity found in one of the healthy volunteers, C6b, which did not result in any clinical or subclinical (cytokine profile) FMF manifestation. Thus, depending on the locus involved, the outcome could be either an asymptomatic carrier state or a full-blown disease. In the case of M694V mutation, a 50% complementation by the normal allele may be not sufficient to provide the protection against clinical manifestations of the disease. This is the most abundant mutation in our FMF cohort and is present in 70% of the patients. Taking into the consideration the high penetrance and associated fitness cost of carrying this allele even in the heterozygous form it seems counterintuitive to suggest that it may confer any selective advantage in the heterozygous carrier state under the present day living conditions. Phenotypically, the asymptomatic heterozygous carriers of FMF mutations are characterized by the elevated levels of markers of inflammation such as CRP, SAA, and INF-γ [53], [54] and by specific glycosylation of AGP [55]. Whether this subclinical inflammation in asymptomatic patients may be protective against specific pathogens, probably endemic for the Mediterranean basin, remains to be seen since the previous studies failed to identify such agents [49].

In a parallel study, we measured the levels of systemic markers of inflammation in acute phase and remission of FMF [56]. The study confirmed earlier observations that even in remission, where the clinical signs of disease are absent, the level of certain markers of inflammation remains elevated suggesting chronic subclinical inflammation in FMF. How this inflammation may affect the gut microbiota? To answer this question, we used two independent techniques to assess gut molecular bacterial diversity in health and disease, the sequence analysis of 16S rRNA gene libraries and FISH. Both these techniques have their own advantages and disadvantages but in combination they are complementary and allow obtaining a more balanced overview of gut microbiota. In particular, 16S libraries may suffer from PCR biases introduced by elevated number of cycles [57] or poor amplification of certain templates [58] while FISH analysis is less sensitive since it targets higher taxonomic entities and no community analysis involving individuals and species/phylotypes, unless they very are numerous, is possible. On the other hand, FISH technique consistently detects higher levels of Bifidobacteria than clone library sequencing, it is less expensive and can be easily automated in conjunction with flow cytometry or computer-aided automatic count under the microscope for analysis of large cohorts. In order to minimize inter-individual variations, 16S rRNA gene libraries were constructed from fecal samples of 12 FMF patients in remission, three patients in the acute stage of FMF, six healthy Armenians, and one healthy subject of non-Armenian origin. The number of amplification cycles was maintained low (10 cycles) to prevent a potential PCR bias.

Both analyses of gut microbiota produced the similar results in terms of general distribution of major phyla representatives in the human gut. Similarly to the previous studies [8], [36], [59] the vast majority of bacterial sequences and cells hybridized to specific probes in the gut fell into the Bacteroidetes and Firmicutes phyla. The Bacteroidetes contained the largest number of sequences but fewer phylotypes while the diverse Firmicutes were dominant in terms of the numbers and the relative proportion of phylotypes detected. The Bacteroidetes was also the largest group detected by the Bac303 probe accounting for 27–39% of the total bacterial count, which is lower than the clone library estimates. On the contrary, the combination of various FISH probes targeting the Firmicutes produced a higher estimate of bacteria belonging to this phylum than the clone libraries. The most profound difference between the two techniques in detection of bacteria in the human gut, however, concerned the enumeration of high G+C Gram-positive bacteria such as Atopobium spp. and Bifidobacterium spp.: FISH technique consistently detected a sizable population of these genera in the gut while they were hardly represented in PCR-generated libraries. Taking into the consideration an important role played by, for example, Bifidobacteria in maintaining the gastrointestinal health, consistent underestimates of these bacteria in clone library analyses may obscure the effect of this group in various gut diseases such as IBD as well as in the healthy gut.

Examination of gut bacterial diversity in healthy and diseased cohorts by the two methods arrived essentially at the same conclusion suggesting the specific shifts in composition of bacteria imposed by mutations in the MEFV gene of the host. The gut microbiota in active disease was characterized by diminished bacterial diversity and major population restructuring within the phyla. In particular, the proportion of phylotypes belonging to of Porphyromonadaceae, Phascolarctobacterium, Faecalibacterium, and Parabacteroides were significantly increased while Prevotellaceae, Dialister and Prevotella were significantly lower in acute disease in comparison with control. Both approaches detected the fall in proportion of clostridial cluster IX in FMF attack: FISH demonstrated it for the whole group, while the sequence libraries elaborated this general trend as involving Acidaminococcaceae, Megasphaera, and Dialister. Evenness values for microbiota in active disease were higher reflecting the lower number of dominant phylotypes, although this could be a consequence of the smaller number of sequences collected. Also, in a parallel FISH approach, the fall in absolute numbers of bacteria during the inflammation attacks was detected. In most cases, this analysis is less sensitive than sequencing because it targets the larger taxonomic entities such as, for example, the whole Bacteroides group and cannot detect the differences at the lower taxa levels, e.g., in the Prevotellaceae and Porphyromonadaceae families and the Prevotella and Parabacteroides genera while sequence analysis operating at the individual sequence level is capable to do. In some cases, however, the situation may be opposite: Rfla729/Rbro730 FISH probe, for example, detects a subset of ruminococci, which is significantly reduced in attack (Table 5), while at the genus level the differences are not significant (Table 3). Thus, FISH served as a complementary tool and allowed analyzing bacterial populations that are routinely underestimated in sequence analyses. In particular, FISH detected the fall in absolute numbers of high G+C bacteria such as Bifidobacterium and Atopobium, which are not represented well in our and others SSU rRNA clone libraries [59]. Thus, in disease attack, the drop of total bacterial count and restructuring of bacterial composition in the gut may be the consequence of acute inflammation, which is accompanied by fever and influx of polymorphonuclear neutrophils (PMNs) into the affected areas including peritoneum. Elevated body temperatures (up to 40°C) and the localized respiratory burst of PMNs producing reactive oxygen species could be responsible for gut microbiota selection.

As mentioned earlier, despite the absence of any clinical signs of disease in remission, certain systemic markers of inflammation are still elevated suggesting subclinical chronic inflammation [56]. Interestingly, the diversity values in remission were higher than in the healthy gut suggesting that the low-level subclinical inflammation in FMF may favour a greater diversity of intestinal bacteria. DA analysis of FISH data also confirmed the specific composition of bacterial populations in remission, which is different from control and forms a separate cluster. An additional internal control subject of different ethnic background was included in our study and the classification of the subject was consistently in the healthy group. The specific shifts of bacterial composition were also confirmed in two patients that were investigated during the remission and attack periods to exclude the influence of other genetic or environmental factors and these analyses confirmed the specific bacterial composition shifts due to disease state such as the increase of butyrate producers and the fall of clostridial cluster IX bacteria during disease attacks. While these shifts in attack may be explained by factors such as acute inflammation, with fever and PMN infiltration, the factors that govern the specific bacterial shifts in remission, which is free of disease symptoms, are harder to suggest. Nevertheless, our DA model consistently classified the patients in remission as a separate group, which suggests the specific patterns of gut microbiota in this cohort.

Thus the question is: what are the molecular mechanisms responsible for these microbiota shifts imposed by the mutated version of pyrin? In a broad sense, the functional role of this protein is the regulation of innate immunity [38]. The protein is expressed in granulocytes, activated monocytes, and serosal and synovial fibroblasts [60][62]. Recently it becomes evident that the regulatory functions of pyrin are carried out through the interaction of two domains of the protein, PYRIN and B30.2/rfp/SPRY, with the proinflammatory caspase-1 activating complex, called the inflammasome. The N-terminal PYRIN domain of the protein binds and competes for ASC (apoptosis-associated speck-like protein containing a caspase-recruitment domain), an inflammasome component, thereby decreasing its availability for the cryopyrin/NALP3/CIAS1 inflammasome. This results in reduction of caspase-1 activation and pro-interleukin-1beta processing and secretion thus halting the inflammatory response [63]. The mutational hotspot in FMF, however, is located in the C-terminal B30.2/rfp/SPRY domain, which modulates the inflammatory response through the interaction with several components of the inflammasome. In addition to interaction with cryopyrin/NALP3/CIAS1, it interacts directly with caspase-1 and its substrate pro-IL-1β [52], [64]. The net effect of these interactions is the suppression of IL-1β activation and block of inflammation. It was suggested that the heightened IL-1β responsiveness could be a factor selecting for mutations in the MEFV gene [52]. Taking into consideration the overall role of pyrin as a modulator/suppressor of the inflammatory response, another factor contributing to the autoinflammatory nature of the disease could be the reduced MEFV messenger RNA expression in patients with familial Mediterranean fever [65]. First, it may contribute to the lower titre of pyrin and its PYRIN domain molecules in the cell thus making more ASC molecules available to initiate caspase-1 activation. Second, the reduced concentration of pyrin and therefore of its B30.2/rfp/SPRY domain, which, in addition, is mutated to the loss of caspase-1 suppressor function in most of FMF cases, may provoke easier triggering the inflammation cascade through caspase-1 activation. Thus, both consequences of MEFV mutations may lead to the heightened responsiveness of cryopyrin/NALP3/CIAS1, which can be oligomerized and activated in response to a very diverse range of ligands such as bacterial muramyl dipeptide, ATP, toxins, bacterial and viral RNA, small antiviral compounds, Staphylococcus, Listeria, and uric acid crystals as well as by low intracellualr potassium concentration [66][71]. While these exo- and endogenous stimuli are normally not perceived as danger signals or are efficiently suppressed, the facilitated oligomerization of the cryopyrin/NALP3/CIAS1, which is not adequately suppressed by the mutated pyrin in FMF, may be the underlying cause in this disease, resulting in excessive pro-caspase-1 and pro-IL1β processing. Indeed, monocytes from FMF patients in remission fail to induce LPS homologous tolerance and exhibit heightened sensitivity to bacterial endotoxin [72], one of the important activators of the cryopyrin/NALP3/CIAS1 inflammasome [73].

In conclusion, we established in this work that mutations in a single host gene lead to specific restructuring of commensal gut microbiota. Although the exact molecular mechanisms behind this process are not fully understood, at our current genetic stage of analysis the work points to the host genotype as a key factor in symbiotic host-microbe interaction. Further developments in this area may focus on details of this interaction at the biochemical and physiological levels. Other models of gut inflammation such as IBD may help to establish the role of other host genes affecting the composition of commensal gut microbiota. Comparative community analyses from genetically different hosts are particularly interesting in identifying gut microbiota alterations specific for a particular disease genotype. At the same time, it cannot be excluded that the changes in the community structure may share some common features reflecting the similarities in disease phenotypes such as inflammation. And finally, how the gut microbiota is transformed in disease to become a “pathobiota” [37] remains an interesting question to answer.

Materials and Methods

Subjects and sampling

A total of 19 Armenian FMF patients with clinically confirmed disease, seven healthy individuals of the same ethnic background, and one healthy subject of non-Armenian origin participated in this study (Table 1). All FMF cases were diagnosed based on Tel-Hashomer criteria [74]. Blood and fecal samples were collected from FMF patients (15 males, four females; aged from 15 to 64 years, mean age–27 years) and control subjects (one male, seven females; aged from 32 to 67 years, mean age–43.8 years) for genotyping and gut bacterial diversity analyses (see below). None of the FMF patients and healthy individuals had used antibiotics within the three months prior to sampling. In the beginning of this study, ten FMF patients were not receiving any drugs and nine were undergoing regular colchicine therapy. During the study, one patient from the colchicine-free group was prescribed this therapy. All subjects were informed about the aim of this study and gave their consent to participate in it. The study was approved by the local ethical committee at IMB.

Genetic diagnostics

Venous blood samples were obtained from the clinically confirmed 19 FMF patients and eight healthy controls to reveal the MEFV mutation carrier status. Genetic analysis was based on screening of two mutational hot spots in the MEFV gene (exons 2 and 10). For this, genomic DNA was isolated from the anticoagulated venous blood samples using the Wizard Genomics DNA Purification kit (Promega, UK), according to the manufacturer's instructions. Exon 2 of the MEFV gene was PCR amplified from genomic DNA using a newly designed forward and reverse primer set: Exon2F - 5′-ATTCTCTCTCCTCTGCCCTG-3′, Exon2R - 5′-CCATTCTTTCTCTGCAGCCG-3′, yielding a 839 bp amplicon. Exon 10 was amplified using the primers Exon10F 5′-CCAGAAGAACTACCCTGTCCC-3′ and Exon10R 5′-TCCTCCTCTGAAATCCATGG-3′, yielding a product of 887 bp [39]. For exon 10, amplification was performed in a final volume of 50 µl, in a mix containing 10 mM Tris-HCl, 1.5 mM MgCl2, 50 mM KCl, 200 µM of each dNTP, 10.0 pmol of each primer, approximately 20 ng of genomic DNA, and 1 U of Red Taq polymerase (Bioline, UK). The cycling conditions included the initial denaturation step for 5 min at 94°C, followed by 35 cycles consisting of denaturation at 94°C for 30 sec, annealing at 61°C for 60 sec, and extension at 72°C for 60 sec, with a final extension at 72°C for 10 min. For exon 2, PCR amplification was performed similarly, except that the PCR mix contained 1 M betaine (Sigma-Aldrich, UK). Amplicons of the expected size were excised from 1.5% agarose gel and purified using a Wizard SV Gel and PCR Clean-up system (Promega, UK), according to the manufacturer's instructions. Purified PCR products were sequenced on both directions, using the same primer set as for generation of templates and the sequences were read on an automated 8-channel capillary sequencer (Beckman, USA) The wild-type sequences of exons 2 and 10 of the MEFV gene were retrieved from GenBank (accession number AF111163).

Generation of 16S rDNA clone libraries

In order to derive a detailed phylogenetic inventory of the gut microflora in FMF and in healthy controls, we constructed bacterial 16S rDNA gene libraries from fecal samples. For this, freshly voided feces were collected into sterile containers from 12 FMF patients in remission, three patients in the acute stage of FMF, six healthy Armenians, and one healthy subject of non-Armenian origin (Table 1). One patient, FMF 13-123, provided the samples both in remission and attack. Total DNA was extracted from fecal samples using QIAamp DNA Stool Mini Kit (Qiagen, UK), according to the manufacturer's instructions. DNA was additionally purified by phenol-chloroform-isoamyl alcohol (25:24:1) extraction, ethanol precipitated, and dissolved in TE buffer. Amplification of the 16S rRNA genes was carried out with the universal primer set–27F (5′-AGAGTTTGATCMTGGCTCAG-3′; positions 8 to 27 in the Escherichia coli 16S rRNA gene) and 1492R (5′-ACGGCTACCTTGTTACGACTT-3′; positions 1510 to 1492 in the E. coli 16S rRNA gene) [75]. The PCR reaction mixture contained 10 mM Tris-HCl, 2 mM MgCl2, 50 mM KCl, 200 µM of each dNTP, 10.0 pmol of each primer, 1U of Bio-X-Act Short DNA polymerase (Bioline, UK), and 100 ng of template DNA in a final volume of 50 µl. To reduce PCR bias caused by elevated number of PCR cycles [57], amplifications were performed using the following conditions: initial denaturation of template DNA at 94°C for 5 min, followed by 10 cycles of denaturation at 94°C for 30 sec, annealing at 57°C for 30 sec, and extension at 72°C for 2 min, with a final extension at 72°C for 10 min. PCR products were visualized on 1% agarose gel in TBE buffer stained with GelStar (Cambrex, UK). 16S PCR products were purified using Wizard SV Gel and PCR Clean-up system (Promega, UK), according to the manufacturer's instructions. This was followed by ethanol precipitation and the final pellets were suspended in 5 µl of TE buffer (pH 8.0). The purified PCR products were directly cloned into pCR-4 cloning vector and then transformed into E. coli TOP10 chemically competent cells using the TOPO TA Cloning Kit (Invitrogen, UK), according to the manufacturer's protocols. Colonies were blue/white screened on Luria-Bertani (LB) agar plates supplemented with ampicillin (50 µg/ml) and X-Gal (80 µg/ml). No IPTG was necessary in this system because the strain and plasmid do not carry the lac repressor. White recombinant colonies were randomly picked up into 200 µl of LB medium with ampicillin (50 µg/ml) in a 96-well format, sealed with a gas-permeable membrane, and incubated overnight at 37°C in a horizontal shaker. Two insert-surrounding primers, T3 (5′-ATTAACCCTCACTAAAGGGA-3′) and T7 (5′-TAATACGACTCACTATAGGG-3′) were used to amplify inserts using a 35-cycle colony PCR. Products were visualized after electrophoresis on 1% agarose gel in TAE buffer and ethidium bromide staining.

Sequencing and phylogenetic analysis

Successfully amplified 16S rDNA inserts were purified and sequenced on the automated 8-channel capillary sequencers (Beckman, USA). The sequencing reactions were carried out with 926R primer (5′-CCGTCAATTCCTTTGAGTTT-3′; positions 926 to 907 in the E. coli 16S rRNA gene) [76]. Nucleotide sequences were aligned using the multiple sequence alignment program CLUSTALX v1.83 [77]. Each sequence was manually edited in conjunction with its chromatogram. Sequences were examined for the possible chimeras using CHIMERA_CHECK online analysis at RDP-II website (​uSSU). Similarity search of the 16S rDNA sequences against database entries was performed using online Basic Local Alignment Search Tool (BLAST) program at the NCBI website ( Distance and similarity matrices were constructed with the DNADIST program v3.6 from PHYLIP (​/phylip/phylip.html) according to the Jukes-Cantor model. Phylogenetic analysis was performed using the neighbour-joining method [78]. Statistical significance of branching was verified by bootstrap analysis involving the construction and analysis of 1000 trees from bootstrapped data sets. Operational taxonomic units (OTUs), or phylotypes, were defined at 99% level as recommended for gut microbiota analyses [59]. To assign sequences quickly and accurately to OTUs, additionally, Jukes-Cantor corrected distance matrices were analyzed with DOTUR (Distance-Based OTU and Richness) by using the furthest-neighbor method [79].

About the half of OTUs (165 out of 346) were sequenced to completion. For this, a set of additional three universal bacterial primers, 27F (5′-AGAGTTTGATCMTGGCTCAG-3′; positions 8 to 27 in the E. coli 16S rRNA gene), 519R (5′-GTATTACCGCGGCTGCTG-3′; positions 536 to 519 in the E. coli 16S rRNA gene) [76] and RP2 (5′- ACGGCTACCTTGTTACGACTT -3′; positions 1510 to 1492 in the E. coli 16S rRNA gene) was used. These partial rRNA gene sequences (the average single sequence read length was ~700 bp), including the sequence previously generated by the 926R primer, were assembled with ChromasPro v1.33 program to produce a consensus sequence for each of the 165 clones. Initially, all sequence data were assembled according to their overlapping sequence fragments into a contiguous consensus sequence using default program settings or, in some cases, adjusting the program settings, if necessary, followed by verification by eye. Afterwards, all chromatogram contigs obtained were manually edited removing ambiguous chromatogram areas, which resulted in the nearly full-length 16S bacterial rDNA sequences of about 1500 bp.

All full-length sequences were checked for possible chimeras using Chimera Check program, v2.7 (online analysis at Ribosomal Database Project II website,​uSSU). Chimeras and sequences of poor quality were omitted from further phylogenetic analyses. Phylogenetic analysis was performed in two stages. First, the almost full-length sequences from our libraries were combined with the closest relatives, cultivated and uncultivated, from databases, aligned, and subjected to phylogenetic analysis using the neighbor-joining method [78]. In the second stage, the partial sequences were aligned to the full-length sequence profile using the guide tree generated during the first stage. A total of 1328 validated 16S rDNA sequences participated in the final tree computing. Validation of the tree branching was done with bootstrap analysis [80] involving 1,000 reiterations with re-sampled data sets.

Community structure analysis

Rarefaction and collector's curves of observed phylotypes, richness estimates and diversity indices were determined with DOTUR program using Jukes-Cantor corrected distance matrix. The bias-corrected Chao 1 richness estimator was calculated after 1000 randomizations of sampling without replacement. Collector's curves of observed and estimated (Chao 1 and the abundance-based coverage estimator, ACE) richness were constructed. Diversity was estimated by the Shannon and Simpson indices, the Simpson reciprocal index was calculated as 1/D, and another version of the Simpson diversity index–as 1-D. The evenness was calculated as E = eH/N, where H is the Shannon diversity index, N is the number of phylotypes. The Good's coverage percentage was calculated with the formula [1−(n/N)]×100, where n is the number of phylotypes in a sample represented by one clone (singletons) and N is the total number of sequences in that sample [81].

Fluorescent in situ hybridization (FISH)

FISH analysis was applied to quantify the predominant groups of bacteria in fecal samples using the universal bacterial and group-specific 16S rRNA-based Cy3-labelled oligonucleotide probes (Table 9). The probes hybridized with the next bacterial groups: Eub338 with the total bacteria [82]; Bac303 with the Bacteroides group [83]; Erec482 with the members of the Clostridium coccoides group (cluster XIVa) [84]; Rrec584 with the Eubacterium rectale-Roseburia cluster, which is a component of clostridial cluster XIVa [85]; Fprau645 for Faecalibacterium prausnitzii-related bacteria–a component of clostridial cluster IV [86]; the Rfla729 and the Rbro730 probes are specific for ruminococci in clostridial cluster IV [87]; Prop853 for the members of clostridial cluster IX [88]; Bif164 for the genus Bifidobacterium [89]; Ato291 for the Atopobium group [90]; EnterobactD for the genus Enterobacteriaceae [91]; and Lab158 for lactobacilli and enterococci [92] (Table 9). Together, the group-specific probes detected the vast majority of the total bacterial cells in the normal human gut [84]. The Rrec584 probe was excluded from this summation, since it detects the E. rectale-Roseburia cluster within the group already covered by the probe Erec482 [85]. The probe Rfla729 was used in conjunction with the probe Rbro730 as the target bacterial groups are overlapping [87].


Table 9. Oligonucleotide probes and hybridization conditions used in this study.


Freshly voided feces from six FMF patients in remission, seven in the acute phase, six healthy Armenians and one healthy subject of non-Armenian descent were collected (Table 1). Two patients, FMF 13-123 and FMF 87, provided fecal samples at both disease stages. The samples in sterile plastic bags were kept at 4°C for no longer than 12 h prior processing. The specimens were kneaded mechanically for 5 min at 4°C to distribute the sample evenly. 0.5 g of each sample was suspended in 4.5 ml of filtered (0.2 µm pore-size filter) ice-cold phosphate-buffer-saline (PBS) and vortexed with a dozen of glass beads for at least 3 min to homogenize the sample and dislodge the bacteria from feed particles. The suspension was centrifuged at 700×g for 1 min to remove debris. One ml of supernatant was added to three ml of 4% paraformaldehyde (PFA) in PBS, fixed at 4°C for 16 hours, washed with ice-cold PBS, and stored in 50% (vol/vol) ethanol-PBS at −20°C until analysed.

Depending on the expected number of target cells, the samples were diluted 40- to 1600-fold, and 10 µl of a diluted bacterial suspension was applied to gelatin-coated slides, air-dried and fixed in 100% ethanol. 10 µl of 50 ng/µl solution of oligonucleotide probes in 100 µl of hybridization buffer was added to the slides and the samples were hybridized overnight (except for the Bac303 probe, which was incubated for 2 h) in a precision incubator at different temperatures, depending on the probe (Table 9). If more stringent conditions were required, formamide was added to the hybridization buffer. To improve bacterial cell permeability samples were incubated before hybridization in 10 µl of lysozyme (final concentration 1 mg/ml) in 100 mM Tris-HCl (pH 8.5) at 37°C. The full panel of FISH oligonucleotide probes with their target bacterial groups and hybridization conditions used in this study is shown in Table 9.

After hybridization, the slides were soaked in a washing buffer for 20 min at 50°C, rinsed with Milli-Q water, and rapidly dried with compressed air. Slides then were covered with 50 µl of Vectashield (Vector Laboratories, Burlington, California) to prevent the fading of fluorescence and a coverslip. The enumeration was performed when the hybridization signals were clear and morphologically distinguishable. Bacterial cells were counted automatically using image analysis software Quantimet HR600 with a Leica DMRXA epifluorescence microscope (Wetzlar, Germany). Depending on the number of labelled cells, from 20 to 50 fields were counted to bring the coefficient of variance value below 10.

Statistical analyses

Statistical analyses were performed using the Statsoft Statistica package ( Comparisons were made among groups of subjects using the Mann-Whitney U test and Student's t-test. Multiple comparisons were carried out using Kruskal-Wallis test. In order to classify cases data were subjected to discriminant function analysis (DA). The P-values below 0.05 were considered statistically significant.

Statistical comparison of 16S rDNA clone libraries was performed using online library compare tool at RDP-II website (​sp). This tool uses the RDP Naive Bayesian classifier to provide rapid classification of library sequences into the bacterial taxonomy proposed by the Bergey's Trust. The P values <0.01 were considered as having significant differences.

Data deposition: The sequences reported in this paper have been deposited in GenBank with accession numbers EU531867-EU532010 and EU532765-EU533948.


We thank Sylvia Duncan, Alan Walker and Harry Flint for help with automated FISH.

Author Contributions

Conceived and designed the experiments: ZK RA. Performed the experiments: ZK GM KG. Analyzed the data: ZK GM KG RA. Contributed reagents/materials/analysis tools: DK RA. Wrote the paper: ZK RA.


  1. 1. Ley RE, Peterson DA, Gordon JI (2006) Ecological and evolutionary forces shaping microbial diversity in the human intestine. Cell 124: 837–848.
  2. 2. Finegold SM, Sutter VL, Mathisen GE (1983) Human Intestinal Microflora in Health and Disease. In: Hentges DJ, editor. New York: Academic Press. pp. 3–31.
  3. 3. Falk PG, Hooper LV, Midtvedt T, Gordon JI (1998) Creating and maintaining the gastrointestinal ecosystem: what we know and need to know from gnotobiology. Microbiol Mol Biol Rev 62: 1157–1170.
  4. 4. Salminen S, Bouley C, Boutron-Ruault MC, Cummings JH, Franck A, et al. (1998) Functional food science and gastrointestinal physiology and function. Br J Nutr 80: S147–S171.
  5. 5. Neish AS, Gewirtz AT, Zeng H, Young AN, Hobert ME, et al. (2000) Prokaryotic regulation of epithelial responses by inhibition of IkappaB-alpha ubiquitination. Science 289: 1560–1563.
  6. 6. Kelly D, Campbell JI, King TP, Grant G, Jansson EA, et al. (2004) Commensal anaerobic gut bacteria attenuate inflammation by regulating nuclear-cytoplasmic shuttling of PPAR-gamma and RelA. Nature Immunol 5: 104–112.
  7. 7. Backhed F, Ding H, Wang T, Hooper LV, Koh GY, et al. (2004) The gut microbiota as an environmental factor that regulates fat storage. Proc Natl Acad Sci U S A 101: 15718–15723.
  8. 8. Ley RE, Turnbaugh PJ, Klein S, Gordon JI (2006) Microbial ecology: human gut microbes associated with obesity. Nature 444: 1022–1023.
  9. 9. Turnbaugh PJ, Ley RE, Mahowald MA, Magrini V, Mardis ER, et al. (2006) An obesity-associated gut microbiome with increased capacity for energy harvest. Nature 444: 1027–1031.
  10. 10. Stappenbeck TS, Hooper LV, Gordon JI (2002) Developmental regulation of intestinal angiogenesis by indigenous microbes via Paneth cells. Proc Natl Acad Sci U S A 99: 15451–15455.
  11. 11. Rakoff-Nahoum S, Paglino J, Eslami-Varzaneh F, Edberg S, Medzhitov R (2004) Recognition of commensal microflora by toll-like receptors is required for intestinal homeostasis. Cell 118: 229–241.
  12. 12. Medzhitov R, Preston-Hurlburt P, Janeway CA Jr (1997) A human homologue of the Drosophila Toll protein signals activation of adaptive immunity. Nature 388: 394–397.
  13. 13. Inohara N, Ogura Y, Chen FF, Muto A, Nunez G (2001) Human Nod1 confers responsiveness to bacterial lipopolysaccharides. J Biol Chem 276: 2551–2554.
  14. 14. Kaparakis M, Philpott DJ, Ferrero RL (2007) Mammalian NLR proteins; discriminating foe from friend. Immunol Cell Biol 85: 495–502.
  15. 15. Mazmanian SK, Liu CH, Tzianabos AO, Kasper DL (2005) An immunomodulatory molecule of symbiotic bacteria directs maturation of the host immune system. Cell 122: 107–118.
  16. 16. Rachmilewitz D, Katakura K, Karmeli F, Hayashi T, Reinus C, et al. (2004) Toll-like receptor 9 signaling mediates the anti-inflammatory effects of probiotics in murine experimental colitis. Gastroenterology 126: 520–528.
  17. 17. Cario E, Gerken G, Podolsky DK (2004) Toll-like receptor 2 enhances ZO-1-associated intestinal epithelial barrier integrity via protein kinase C. Gastroenterology 127: 224–238.
  18. 18. Cario E, Gerken G, Podolsky DK (2007) Toll-like receptor 2 controls mucosal inflammation by regulating epithelial barrier function. Gastroenterology 132: 1359–1374.
  19. 19. Hague A, Paraskeva C (1995) The short-chain fatty acid butyrate induces apoptosis in colorectal tumour cell lines. Eur J Cancer Prev 4: 359–364.
  20. 20. Csordas A (1996) Butyrate, aspirin and colorectal cancer. Eur J Cancer Prev 4: 221–231.
  21. 21. Ahmad MS, Krishnan S, Ramakrishna BS, Mathan M, Pulimood AB, et al. (2000) Butyrate and glucose metabolism by colonocytes in experimental colitis in mice. Gut 46: 493–499.
  22. 22. Hooper LV, Xu J, Falk PG, Midtvedt T, Gordon JI (1999) A molecular sensor that allows a gut commensal to control its nutrient foundation in a competitive ecosystem. Proc Natl Acad Sci U S A 96: 9833–9838.
  23. 23. Sonnenburg JL, Xu J, Leip DD, Chen CH, Westover BP, et al. (2005) Glycan foraging in vivo by an intestine-adapted bacterial symbiont. Science 307: 1955–1959.
  24. 24. Meng D, Newburg DS, Young C, Baker A, Tonkonogy SL, et al. (2007) Bacterial symbionts induce a FUT2-dependent fucosylated niche on colonic epithelium via ERK and JNK signaling. Am J Physiol Gastrointest Liver Physiol 293: G780–787.
  25. 25. Coyne MJ, Reinap B, Lee MM, Comstock LE (2005) Human symbionts use a host-like pathway for surface fucosylation. Science 307: 1778–1781.
  26. 26. Sperandio V, Torres AG, Jarvis B, Nataro JP, Kaper JB (2003) Bacteria-host communication: the language of hormones. Proc Natl Acad Sci U S A 100: 8951–8956.
  27. 27. Wu L, Estrada O, Zaborina O, Bains M, Shen L, et al. (2005) Recognition of host immune activation by Pseudomonas aeruginosa. Science 309: 774–777.
  28. 28. Hugot JP, Chamaillard M, Zouali H, Lesage S, Cezard JP, et al. (2001) Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn's disease. Nature 411: 599–603.
  29. 29. Ogura Y, Bonen DK, Inohara N, Nicolae DL, Chen FF, et al. (2001) A frameshift mutation in NOD2 associated with susceptibility to Crohn's disease. Nature 411: 603–606.
  30. 30. Inohara N, Ogura Y, Fontalba A, Gutierrez O, Pons F, et al. (2003) Host recognition of bacterial muramyl dipeptide mediated through NOD2. Implications for Crohn's disease. J Biol Chem 278: 5509–5512.
  31. 31. Girardin SE, Boneca IG, Viala J, Chamaillard M, Labigne A, et al. (2003) Nod2 is a general sensor of peptidoglycan through muramyl dipeptide (MDP) detection. J Biol Chem 278: 8869–8872.
  32. 32. Landers CJ, Cohavy O, Misra R, Yang H, Lin YC, et al. (2002) Selected loss of tolerance evidenced by Crohn's disease-associated immune responses to auto- and microbial antigens. Gastroenterology 123: 689–699.
  33. 33. Lodes MJ, Cong Y, Elson CO, Mohamath R, Landers CJ, et al. (2004) Bacterial flagellin is a dominant antigen in Crohn disease. J Clin Invest 113: 1296–1306.
  34. 34. Adams RJ, Heazlewood SP, Gilshenan KS, O'brien M, McGuckin MA, et al. (2007) IgG antibodies against common gut bacteria are more diagnostic for Crohn's disease than IgG against mannan or flagellin. Am J Gastroenterol. Oct 6 [Epub ahead of print].
  35. 35. Manichanh C, Rigottier-Gois L, Bonnaud E, Gloux K, Pelletier E, et al. (2006) Reduced diversity of faecal microbiota in Crohn's disease revealed by a metagenomic approach. Gut 55: 205–211.
  36. 36. Frank DN, St Amand AL, Feldman RA, Boedeker EC, Harpaz N, et al. (2007) Molecular-phylogenetic characterization of microbial community imbalances in human inflammatory bowel diseases. Proc Natl Acad Sci U S A 104: 13780–13785.
  37. 37. Garrett WS, Lord GM, Punit S, Lugo-Villarino G, Mazmanian SK, et al. (2007) Communicable Ulcerative Colitis Induced by T-bet Deficiency in the Innate Immune System. Cell 131: 33–45.
  38. 38. Ting JP, Kastner DL, Hoffman HM (2006) CATERPILLERs, pyrin and hereditary immunological disorders. Nat Rev Immunology 3: 183–195.
  39. 39. Brydges S, Kastner DL (2006) The systemic autoinflammatory diseases: inborn errors of the innate immune system. Curr Top Microbiol Immunol 305: 127–160.
  40. 40. The International FMF Consortium (1997) Ancient missense mutations in a new member of the RoRet gene family are likely to cause familial Mediterranean fever. Cell 90: 797–807.
  41. 41. The French FMF Consortium (1997) A candidate gene for familial Mediterranean fever. Nat Genet 17: 25–31.
  42. 42. Van Limbergen J, Russell RK, Nimmo ER, Satsangi J (2007) The Genetics of Inflammatory Bowel Disease. Am J Gastroenterol 102: 2820–31.
  43. 43. Zoetendal EG, Akkermans ADL, Akkermans van-Vliet WM, de Visser JA, de Vos WM (2001) The host genotype affects the bacterial community in the human gastrointestinal tract. Microbial Ecol Health Dis 13: 129–134.
  44. 44. Stewart JA, Chadwick VS, Murray A (2005) Investigations into the influence of host genetics on the predominant eubacteria in the faecal microflora of children. J Med Microbiol 54: 1239–1242.
  45. 45. Seksik P, Rigottier-Gois L, Gramet G, Sutren M, Pochart P, et al. (2003) Alterations of the dominant faecal bacterial groups in patients with Crohn's disease of the colon. Gut 52: 237–242.
  46. 46. Toivanen P, Vaahtovuo J, Eerola E (2001) Influence of major histocompatibility complex on bacterial composition of fecal flora. Infect Immun 69: 2372–2377.
  47. 47. Stoffman N, Magal N, Shohat T, Lotan R, Koman S, et al. (2000) Higher than expected carrier rates for familial Mediterranean fever in various Jewish ethnic groups. Eur J Hum Gen 8: 307–310.
  48. 48. Gershoni-Baruch R, Shinawi M, Leah K, Badarnah K, Brik R (2001) Familial Mediterranean fever: prevalence, penetrance and genetic drift. Eur J Hum Genet 9: 634–637.
  49. 49. Kogan A, Shinar Y, Lidar M, Revivo A, Langevitz P, et al. (2001) Common MEFV mutations among Jewish ethnic groups in Israel: high frequency of carrier and phenotype III states and absence of a perceptible biological advantage for the carrier state. Am J Med Genet 102: 272–276.
  50. 50. Yilmaz E, Ozen S, Balci B, Duzova A, Topaloglu R, et al. (2001) Mutation frequency of Familial Mediterranean Fever and evidence for a high carrier rate in the Turkish population. Eur J Hum Genet 9: 553–555.
  51. 51. Booth DR, Gillmore JD, Lachmann HJ, Booth SE, Bybee A, et al. (2000) The genetic basis of autosomal dominant familial Mediterranean fever. QJM 93: 217–221.
  52. 52. Chae JJ, Wood G, Masters SL, Richard K, Park G, et al. (2006) The B30.2 domain of pyrin, the familial Mediterranean fever protein, interacts directly with caspase-1 to modulate IL-1beta production. Proc Natl Acad Sci U S A 103: 9982–9987.
  53. 53. Tunca M, Kirkali G, Soyturk M, Akar S, Pepys MB, et al. (1999) Acute phase response and evolution of familial Mediterranean fever. Lancet 353: 1415.
  54. 54. Aypar E, Ozen S, Okur H, Kutluk T, Besbas N, et al. (2003) Th1 polarization in familial Mediterranean fever. J Rheumatol 30: 2011–2013.
  55. 55. Poland D, Drenth J, Rabinovitz E, Livneh A, Bijzet J, et al. (2001) Specific glycosylation of α1-acid glycoprotein characterises patients with familial Mediterranean fever and obligatory carriers of MEFV. Ann Rheum Dis 60: 777–780.
  56. 56. Manukyan GP, Ghazaryan KA, Ktsoyan ZhA, Tatyan MV, Khachatryan ZA, et al. (2008) Cytokine profile of Armenian patients with Familial Mediterranean fever. Clin Biochem 41: 920–922.
  57. 57. Bonnet R, Suau A, Dore J, Gibson GR, Collins MD (2002) Differences in rDNA libraries of faecal bacteria derived from 10- and 25-cycle PCRs. Int J Syst Evol Microbiol 52: 757–763.
  58. 58. Tajima K, Aminov RI, Nagamine T, Matsui H, Nakamura M, et al. (2001) Diet-dependent shifts in the bacterial population of the rumen revealed with real-time PCR. Appl Environ Microbiol 67: 2766–2774.
  59. 59. Eckburg PB, Bik EM, Bernstein CN, Purdom E, Dethlefsen L, et al. (2005) Diversity of the human intestinal microbial flora. Science 308: 1635–1638.
  60. 60. Centola M, Wood G, Frucht DM, Galon J, Aringer M, et al. (2000) The gene for familial Mediterranean fever, MEFV, is expressed in early leukocyte development and is regulated in response to inflammatory mediators. Blood 95: 3223–3231.
  61. 61. Matzner Y, Abedat S, Shapiro E, Eisenberg S, Bar-Gil-Shitrit A, et al. (2000) Expression of the familial Mediterranean fever gene and activity of the C5a inhibitor in human primary fibroblast cultures. Blood 96: 727–731.
  62. 62. Diaz A, Hu C, Kastner DL, Schaner P, Reginato AM, et al. (2004) Lipopolysaccharide-induced expression of multiple alternatively spliced MEFV transcripts in human synovial fibroblasts: a prominent splice isoform lacks the C-terminal domain that is highly mutated in familial Mediterranean fever. Arthritis Rheum 50: 3679–3689.
  63. 63. Chae JJ, Komarow H, Cheng J, Wood G, Raben N, et al. (2003) Targeted disruption of pyrin, the FMF protein, causes heightened sensitivity to endotoxin and a defect in macrophage apoptosis. Mol Cell 11: 591–604.
  64. 64. Papin S, Cuenin S, Agostini L, Martinon F, Werner S, et al. (2007) The SPRY domain of Pyrin, mutated in familial Mediterranean fever patients, interacts with inflammasome components and inhibits proIL-1beta processing. Cell Death Differ 14: 1457–1466.
  65. 65. Notarnicola C, Didelot MN, Kone-Paut I, Seguret F, Demaille J, et al. (2002) Reduced MEFV messenger RNA expression in patients with familial Mediterranean fever. Arthritis Rheum 46: 2785–2793.
  66. 66. Martinon F, Agostini L, Meylan E, Tschopp J (2004) Identification of bacterial muramyl dipeptide as activator of the NALP3/cryopyrin inflammasome. Curr Biol 14: 1929–1934.
  67. 67. Mariathasan S, Weiss DS, Newton K, McBride J, O'Rourke K, et al. (2006) Cryopyrin activates the inflammasome in response to toxins and ATP. Nature 440: 228–232.
  68. 68. Kanneganti TD, Ozoren N, Body-Malapel M, Amer A, Park JH, et al. (2006) Bacterial RNA and small antiviral compounds activate caspase-1 through cryopyrin/Nalp3. Nature 440: 233–236.
  69. 69. Kanneganti TD, Body-Malapel M, Amer A, Park JH, Whitfield J, et al. (2006) Critical role for Cryopyrin/Nalp3 in activation of caspase-1 in response to viral infection and double-stranded RNA. J Biol Chem 281: 36560–36568.
  70. 70. Martinon F, Petrilli V, Mayor A, Tardivel A, Tschopp J (2006) Gout-associated uric acid crystals activate the NALP3 inflammasome. Nature 440: 237–241.
  71. 71. Petrilli V, Papin S, Dostert C, Mayor A, Martinon F, et al. (2007) Activation of the NALP3 inflammasome is triggered by low intracellular potassium concentration. Cell Death Differ 14: 1583–1589.
  72. 72. Davtyan TK, Hakopyan GS, Avetisyan SA, Mkrtchyan NR (2006) Impaired endotoxin tolerance induction in patients with familial Mediterranean fever. Pathobiology 73: 26–39.
  73. 73. Mariathasan S, Monack DM (2007) Inflammasome adaptors and sensors: intracellular regulators of infection and inflammation. Nat Rev Immunol 7: 31–40.
  74. 74. Livneh A, Langevitz P, Zemer D, Zaks N, Kees S, et al. (1997) Criteria for the diagnosis of familial Mediterranean fever. Arthritis Rheum 40: 1879–1885.
  75. 75. Weisburg WG, Barns SM, Pelletier DA, Lane DJ (1991) 16S ribosomal DNA amplification for phylogenetic study. Journal of Bacteriology 173: 697–703.
  76. 76. Lane DJ, Pace B, Olsen GJ, Stahl DA, Sogin ML, et al. (1985) Rapid Determination of 16S Ribosomal RNA Sequences for Phylogenetic Analyses. Proc Natl Acad Sci U S A 82(20): 6955–6959.
  77. 77. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG (1997) The ClustalX windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Research 24: 4876–4882.
  78. 78. Saitou N, Nei M (1987) The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol Biol Evol 4(4): 406–425.
  79. 79. Schloss PD, Handelsman J (2005) Introducing DOTUR, a computer program for defining operational taxonomic units and estimating species richness. Appl Environ Microbiol 71(3): 1501–1506.
  80. 80. Felsenstein J (1985) Confidence intervals on phylogenies: an approach using the bootstrap. Evolution 39: 783–791.
  81. 81. Good IJ (1953) The population frequencies of species and the estimation of population parameters. Biometrika 40: 237–264.
  82. 82. Amann RI, Binder BJ, Olson RJ, Chisholm SW, Devereux R, et al. (1990) Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Appl Environ Microbiol 56: 1919–1925.
  83. 83. Manz W, Amann R, Ludwig W, Vanvanneyt M, Schleifer KH (1996) Application of a suite of 16S rRNA-specific oligonucleotide probes designed to investigate bacteria of the phylum Cytophaga-Flavobacter-Bacteroides in the natural environment. Microbiology 142: 1097–1106.
  84. 84. Franks AH, Harmsen HJ, Raangs GC, Jansen GJ, Schut F, et al. (1998) Variations of bacterial populations in human feces measured by fluorescent in situ hybridization with group-specific 16S rRNA-targeted oligonucleotide probes. Appl Environ Microbiol 64: 3336–3345.
  85. 85. Aminov RI, Walker AW, Duncan SH, Harmsen HJ, Welling GW, et al. (2006) Molecular diversity, cultivation, and improved detection by fluorescent in situ hybridization of a dominant group of human gut bacteria related to Roseburia spp. or Eubacterium rectale. Appl Environ Microbiol 72: 6371–6376.
  86. 86. Suau A, Rochet V, Sghir A, Grammet S, Breways S, et al. (2001) Fusobacterium prausnitzii and related species represent a dominant group within the human faecal flora. Syst Appl Microbiol 24: 139–145.
  87. 87. Harmsen HJ, Raangs GC, He T, Degener JE, Welling GW (2002) Extensive set of 16S rRNA-based probes for detection of bacteria in human feces. Appl Environ Microbiol 68: 2982–2990.
  88. 88. Walker AW, Duncan SH, McWilliam Leitch EC, Child MW, Flint HJ (2005) pH and peptide supply can radically alter bacterial populations and short-chain fatty acid ratios within microbial communities from the human colon. Appl Environ Microbiol 71: 3692–3700.
  89. 89. Langendijk PS, Schut F, Jansen GJ, Raangs GC, Kamphuis GR, et al. (1995) Quantitative fluorescence in situ hybridization of Bifidobacterium spp. with genus-specific 16S rRNA-targeted probes and its application in fecal samples. Appl Environ Microbiol 61: 3069–3075.
  90. 90. Harmsen HJ, Wildeboer-Veloo AC, Grijpstra J, Knol J, Degener JE, et al. (2000) Development of 16S rRNA-based probes for the Coriobacterium group and the Atopobium cluster and their application for enumeration of Coriobacteriaceae in human feces from volunteers of different age groups. Appl Environ Microbiol 66: 4523–4527.
  91. 91. Ootsubo M, Shimizu T, Tanaka R, Sawabe T, Tajima K, et al. (2002) Oligonucleotide probe for detecting Enterobacteriaceae by in situ hybridization. J Appl Microbiol 93: 60–68.
  92. 92. Harmsen HJ, Elfferich P, Schut F, Welling GW (1999) A 16S rRNA-targeted probe for detection of lactobacilli and enterococci in fecal samples by fluorescent in situ hybridization. Microb Ecol Health Dis 11: 3–12.