Segmented filamentous bacteria in human ileostomy samples after high-fiber intake


Correspondence: Hans Jonsson, Swedish University of Agricultural Sciences, Department of Microbiology, Box 7025, SE-750 02 Uppsala, Sweden. Tel.: +46 (0)18-673327; fax: +46 (0)18-673327; e-mail:


Segmented filamentous bacteria (SFB) are inhabitants of the small intestine of various animals, where they can be detected microscopically due to their specific morphology and intimate association with the intestinal epithelium. SFB colonize the distal part of the small intestine in a host-specific manner and affects important functions of the immune system, such as the induction of secretory IgA production and regulation of T-cell maturation. Considering the influences SFB have on immune functions, they could be regarded as a key species in host–microbial interactions of the gastrointestinal tract. Although these influences might be executed by other microorganisms, a human-adapted variant of SFB is not unlikely. In this study, ileostomy samples from 10 human subjects were screened with PCR, using primers derived from sequences of SFB from rat and mouse. PCR products were obtained from samples taken from one individual at two time points. Sequencing revealed the presence of a 16S rRNA gene with high similarity (98%) to the corresponding genes from SFB of mouse and rat origin, thus indicating the presence of a human variant of SFB. The findings presented in this study will hopefully encourage research to elucidate whether this intriguing organism is a persistent member of the normal human microbiota.


The human intestinal microbiota is nowadays recognized as being of crucial importance for the development and maintenance of fundamental functions in their host (Tremaroli & Bäckhed, 2012). Substantial efforts have thus been made to elucidate the underlying mechanisms of this intricate host–bacterial relationship. One strategy has been to, with the aid of new powerful high-throughput technology, define the components of the system and correlate the overall microbial composition with health and disease (Claesson et al., 2012). Another approach deals with the identification of microorganisms that could be of specific importance by mediating core functions in host–bacterial crosstalk established through coevolutionary forces (Ivanov & Honda, 2012).

Segmented filamentous bacteria (SFB) exhibit a rare if not unique ability to intimately associate with the intestinal epithelium where they are tightly anchored by a specialized cell at one end of the filament (Davis & Savage, 1974). A variety of animals are colonized by these bacteria in a spatially and temporally regulated manner, so that large populations of SFB can be found around the time of weaning and are mainly located at the distal ileum where they associate with areas near Peyer's patches (Klaasen et al., 1992a). Furthermore, colonization is host specific where variants of SFB colonize different hosts (Tannock et al., 1984). SFB have gained increased attention in the last few years due to the strengthened recognition of their potential to affect important functions of the immune system. Thus, they can induce production of secretory IgA and orchestrate coordinated maturation of T-cell responses (Talham et al., 1999; Gaboriau-Routhiau et al., 2009). Several well-designed studies have shown that SFB could be regarded as a key species in the intricate host–microbiota driven maturation and maintenance of immune functions (Ivanov et al., 2009; Salzman et al., 2010).

Although SFB are not cultivable in the laboratory, mono-colonization of germ-free animals has allowed isolation and taxonomic characterization (Klaasen et al., 1991; Snel et al., 1994, 1995), and in recent years, several genome sequences of SFB isolated from rat and mouse have been published (Prakash et al., 2011; Sczesnak et al., 2011). Phylogenetically, SFB are now identified as a distinct clade within the Clostridiaceae. Based on predicted number of genes and GC content, they may be regarded as being positioned between obligate and facultative symbionts, and thus are highly dependent on host functions for nutrition (Pamp et al., 2012).

The seemingly outstanding features of SFB regarding their immune modulatory functions would make any intestinal microbiologist eager to reveal whether SFB can be found also in humans. Contradictory data exist on this matter. While old light microscope findings indicate the presence of a filamentous organism, tentatively being SFB, on the ileal mucosa of a human subject (Klaasen et al., 1993), new large-scale sequence searches in metagenomic datasets conclude that SFB are not present in humans (Sczesnak et al., 2011). Very recently, Yin et al. (2013) reported the presence of SFB in human feces from a large number of individuals. In contrast to what has been seen so far regarding host specificity of SFB, the 16S rRNA gene sequences ascribed to human SFB in their work were not unique but instead clustered together with mouse SFB sequences.

One obstacle to find human SFB would be their assumed spatial and temporal abundance, as mentioned above. The human distal ileum is not easily accessed for sampling and even less so when it comes to weaning infants, which could represent the most likely reservoir considering data regarding SFB colonization in animals. Thus, most analyses of the human microbiota have been performed on fecal material although the composition of the microbiota in feces will not fully reflect the composition in other part of the gastrointestinal tract (Hayashi et al., 2005). The current study employed ileostomy effluents from adult human subjects, which were collected in an earlier work aimed at recording physiological parameters in response to different intake of dietary fiber (Lundin et al., 2004). The study had a crossover design were all 10 participants started out on a low-fiber diet for 2 weeks, followed by a 1-week wash-out period and a 2-week period with high-fiber intake. Ileal effluents were collected and pooled so that four samples representing the four dietary periods were available for each person in the study (Fig. 1). The aim of the present study was to investigate whether SFB could be detected in these ileostomy samples since they are more representative than fecal samples as to the location previously described for SFB.

Figure 1.

Schematic outline of the dietary study. Ten human ileostomy patients participated in the study where all started with a low-fiber diet for 2 weeks (blue box) followed by 1 week of ‘wash-out’(white box), and 2 weeks of high-fiber diet (red box). During both fiber treatments, the group was divided so that five individuals followed a high meal frequency regime for 1 week and then switched to a low meal frequency for 1 week (red line). The other group followed a low meal frequency regime for 1 week and switched to a high meal frequency for 1 week (blue line). The study was divided into four dietary periods (gray boxes) representing the 5 weekdays. Sampling was performed at day 3–5 within each dietary period and pooled, which resulted in four samples per subject.

Material and methods

Samples and DNA preparation

Ileostomy effluents were collected in an earlier study, performed with the aim to investigate physiological response to the intake of diets with different fiber content (Lundin et al., 2004). DNA was prepared from the lyophilized ileostomy samples using the QIAamp DNA Stool Mini Kit (QIAGEN) with an added bead beating treatment as the first step. Bead beating was performed with 0.1 mm zirconium/silica beads (Biospec), 2 times 45 s with setting 5 using the MP FastPrep-24 (MP-Biomedicals).


Primers were designed by alignment of published 16S rRNA sequences from SFB from mouse (GenBank: X77814), rat (GenBank: X87244), and chicken (GenBank: X80834). Six different primers were manually designed to fit to conserved regions in the aligned sequences while being as discriminative to other sequences as possible. The primers were as follows: SFBf1 5′-GGAGTCTGCGGCACATTAG-3′, SFBf2 5′-GCGACGATGTGTAGCCGGT-3′, SFBf3 5′-CGCAAACGCAATAAGTACCC-3′, SFBr1 5′-CACCTTAGACTGCTGCCTC-3′, SFBr2 5′-CCTAGTTAACCTAGGCTGTC-3′, SFBr3 5′-GTACTTATTGCGTTTGCGACG-3′. Additional primers were designed from the sequence generated by the first set of primers; SFBf2a 5′- GCGACGATGTGTAGCTGGT-3′, SFBf3a 5′-AATAGCAATCTGTGCCGTCG-3′, SFBr2a 5′- CTCAGCTTCACCTGCTAGC-3′. The specificity of the primers was evaluated in silico with blast at NCBI and Probe Match at Ribosomal Database Project. All primers were from Life Technologies. The PCR program was as follows: (95 °C 30 s, 65 °C 30 s, 72 °C 2 min) 35×, 72 °C 10 min. PuReTaq™Ready-To-Go™ PCR beads (GE Healthcare) and FideliTaq™ (Usb) were used for the PCR reactions.

DNA sequence determination

The PCR products generated with the primer combinations SFBf1 + SFBr3, SFB f1 +  SFBr2, SFBf2a + SFBr1, and SFBf3 + SFBr1 were purified with QIAquick PCR Purification Kit (QIAGEN) and sequenced. The sequences were manually edited and used for searches in the GenBank database using standard nucleotide blast at NCBI. Ribosomal Database Project 10 Sequence match was also used for homology searches. DECIPHER at University of Wisconsin was used for chimera-check of the sequence. The sequence was deposited in GenBank at NCBI under accession number: KC135882.

Phylogenetic analysis

The 1222-bp sequence generated in this work (GenBank: KC 135882), and sequences from SFB-mouse (GenBank: X77814), SFB-rat (GenBank: X87244, GenBank: AP012210 region 46043.47549), SFB-chicken (GenBank: X80834), and SFB-trout (AY007720) were aligned using Muscle 3.8.31, and phylogeny was estimated with Phyml 3.0, both at the Mobyle Portal, Mega 5.1 (Tamura et al., 2011) was used to construct a tree based on the alignment. The closest cultured recognized species Clostridium cadaveris (AB542932) was used as outgroup. All sequences were adjusted to the same length using the new sequence as template. The tree (Fig. 3) also included the sequences presented in the recent work by Yin et al. (2013); human predominant sequence (HP1) GenBank: JQ361233, mouse predominant sequences 1-2 (MP1) GenBank: JQ361402, (MP2) GenBank: JQ361372, chicken predominant sequences 1-3 (CP1) GenBank: JQ361161, (CP2) GenBank: JQ361171, (CP3) GenBank: JQ361180.

Results and discussion

In silico evaluation of the primers using blast and Probe Match at RDP revealed that none of the primer combination would be expected to amplify a product from human intestinal material. This was supported by the fact that most samples did not give rise to a visible PCR product and also by sequencing of positive samples, which only produced one sequence.

The PCR analysis included forty ileostomy samples, representing four samples, separated in time, taken from each of the 10 different individuals. Using six primers derived from published SFB sequences, it was possible to generate PCR products from two samples coming from the same individual. The primer combinations SFBf1-SFBr2, SFBf1-SFBr3, SFBf3-SFBr1, and SFBf3-SFBr2 generated products in the initial experiments. Sequencing revealed a mismatch in SFBf2 explaining the lack of PCR product in PCR runs with this primer. The primer SFBf2a was thus designed to replace SFBf2 in the following experiments. Also, primers SFBf3 and SFBr2 were replaced with SFBf3a and SFBr2a, respectively, as the latter performed more robustly. A summary of the PCR strategy is depicted in Fig. 2. Sequencing of 1222 bp of the PCR products revealed the presence of a 16S rRNA gene with high similarity (98%) to the corresponding genes from SFB of mouse and rat origin. All mismatches in the alignment with SFB-mouse and SFB-rat were manually scrutinized using PCR products from several DNA preparations. The phylogenetic analysis based on the sequence described in this work and already known SFB sequences, places the new human-derived sequence at close phylogenetic distance to the other SFB sequences, the distance being shortest to that from mouse SFB (Fig. 3). This is further supported by the result from the sequence match analysis at the Ribosomal Database Project. Altogether, the results strongly suggest that the sequence presented in this work originates from a hitherto unknown SFB variant, possibly unique to humans.

Figure 2.

Schematic representation of the PCR strategy. Conserved regions, specific for SFB, but discriminatory against related species, were identified by alignment of the 16S rRNA gene sequences of SFB from rat, mouse, and chicken. Six primers were designed based on this alignment, and eight primer combinations were employed to screen for SFB in the ileostomy samples. Additional primers were designed based on the obtained sequence. The primer combinations SFBf1-SFBr2a, SFBf1-SFBr3, SFBf3a-SFBr1, and SFBf2a-SFBr1 were used to generate PCR products for sequencing. The used primers and their positions are in bold. In total, the sequences enclose a 1222-bp region covering position 218-1410 in mouse SFB (X77814).

Figure 3.

Phylogenetic tree of SFB. The alignment of SFB sequences from mouse, rat, chicken, and trout with the new sequence of tentative human-adapted SFB (in bold) was used to construct a phylogenetic tree with maximum likelihood using Phyml. The closest cultured recognized species, Clostridium cadaveris, was used as outgroup. HP1, MP1-2, and CP1-3 refer to the predominant SFB sequences found in humans, mice, and chickens, respectively, as recently reported by Yin et al. (2013). The accession number for the new sequence reported in this work is GenBank: KC135882. Bootstrap values are supported by 500 replicates and are denoted as percentage.

In a recent article by Yin et al. (2013), SFB were reported to be present in a large number of human individuals, and colonization was seen primarily in young individuals. One conclusion from their work is that host specificity is not reflected in a unique 16S rRNA gene sequence when it comes to human SFB, because this sequence was identical to sequences they could also retrieve from mice. The sequence generated from the ileostomy effluents is only 97% homologous to the predominant human-derived sequence presented in the work by Yin et al. It is noteworthy that the primers used in their work (779f and 1380r) both contains one mismatch to the sequence presented here and therefore could fail to generate a product corresponding to this variant of SFB if present in the fecal samples they examined.

The sequences described in the work by Yin et al.(2013) were included in the phylogenetic analysis, and it could be clearly seen that the sequence they generate from human fecal samples is distinct from the sequence reported here. They reported one predominant sequence from human samples (HP1), two predominant sequences from mouse (MP1, MP2), and three predominant sequences from chicken (CP1-3). As seen in the phylogenic tree (Fig. 3), the human sequence and one of the mouse sequences are identical and cluster closely with the three sequences they found in chickens and also with the earlier reported chicken sequence. Besides the predominating sequences, they also reported some polymorphism in the 16S rRNA gene sequence. All polymorphic variants of the human-derived SFB sequences are 97–98% homologous to the sequence reported in this work. At this stage, it is thus reasonable to conclude that more than one type of SFB could be present in humans, the type reported in the current study possibly being unique.

The 10 individuals that participated in the dietary study were all protocolectomized for ulcerative colitis, meaning that they had all or part of the colon removed. Although a dramatic influence on intestinal architecture, it has been concluded that ileostomy effluents exhibit microbial compositions similar to what can be found in healthy subjects (Zoetendal et al., 2012). The fact that SFB sequences could be detected in only one of the 10 subjects in the study indicate that this is a relatively uncommon situation. As all subjects were adults, this is in correspondence with reports of SFB colonizing preferably in young animals (Jiang et al., 2001). The time between sampling points, corresponding to the two positive samples, was approximately 5 days. This increases the likelihood that the organism indeed colonized this particular individual as transit time from mouth to cecum is on the hour scale (Gilmore, 1990). However remote, the possibility that an external source, like fecally contaminated foodstuff, has contributed to the results cannot be definitely ruled out.

Interestingly, the two positive samples came from the same individual during intake of a high-fiber diet (dietary period 3 and 4, Fig. 1). Repeated attempts to obtain PCR products from the two samples this individual had contributed during low-fiber intake were unsuccessful. The intestinal microbiota in humans is known to be affected by diet and certain groups of bacteria are responding more strongly to changes in carbohydrate availability. Furthermore, the response is likely to be different on an individual basis, reflecting the overall individual variability in intestinal microbiota composition (Flint et al., 2012). Most studies regarding the effect of diet on the microbiota have been performed with focus mainly on the colonic microbiota as manifested in fecal material. However, also bacteria in the small intestine are likely to be affected by such changes in nutrient availability. It is tempting to speculate that the fact that SFB increase during weaning in many animals, could be a result of the introduction of dietary fiber in the suckling animal when mother's milk is gradually substituted with solid food. Interestingly, two early studies can be notified in this context; in the first study, the addition of milk proteins in the form of whey during a stress experiment with mice resulted in a decrease of SFB (Koopman et al., 1989). The second study where ordinary beans, which are rich in dietary fiber, were added to a natural diet, reported an increase of SFB colonization in mice (Klaasen et al., 1992b). The milk proteins and the dietary fiber might thus have opposite effects on the growth of SFB.

The large and diverse set of microorganisms inhabiting the gastrointestinal tract affects important functions of the host. Although the underlying mechanisms for these influences and the contribution of the individual microbial components are largely unknown, a number of ‘autobionts’ have been suggested which might be regarded as key players in view of their immunomodulatory activity (Ivanov & Honda, 2012). The study by Yin et al. (2013) indicates that one of the most spectacular members of this group, SFB, is commonly part of the human microbiota. The data presented in the current study, although limited, open up the possibility that another variant of SFB, with a unique 16S rRNA gene sequence, is also present in humans. It will be of great interest to see the role of this group of microorganisms in the human host.


I thank Göran Hallmans for supplying the ileostomy samples. I thank Johan Dicksved and Stefan Roos for comments on the manuscript.