Introduction

Irritable bowel syndrome (IBS) is a multifactorial gastrointestinal disorder that affects approximately one in ten people worldwide, although prevalence estimates vary widely [1]. Characteristic symptoms include abdominal pain, bloating, and changes in stool form and frequency that begin at least 6 months prior to diagnosis [2]. The three main subtypes are diarrhea-predominant IBS (IBS-D), constipation-predominant IBS (IBS-C), and mixed IBS with constipation and diarrhea (IBS-M) [2].

There is strong evidence that the gut microbiome plays an important role in the pathophysiology of IBS [3]. A meta-analysis of 23 case–control studies identified some consistent findings when stool samples from IBS subjects were compared to those from healthy controls, including a decrease in fecal levels of the genera Lactobacillus and Bifidobacterium in IBS subjects, as well as increases in fecal levels of family Enterobacteriaceae and species Escherichia coli [4]. A smaller earlier meta-analysis also identified lower Lactobacillus and Bifidobacterium levels in IBS subjects, as well as lower Faecalibacterium prausnitzii, when compared to healthy controls [5]. However, there is increasing evidence that IBS subtypes should be studied separately, as IBS-C and IBS-D appear to be associated with different gut microbial compositions (microtypes) [6]. Specifically, subjects with IBS-C have higher levels of methane (CH4)-producing archaea in their stool when compared to healthy controls [7], including higher levels of Methanobrevibacter smithii [8], which is the predominant methanogen in the human gut. Moreover, levels of this methanogen correlate with CH4 levels on a lactulose breath test [8,9,10]. In a recent publication, we identified distinct gut microbial compositions and associated breath gas profiles in subjects with IBS-C vs. IBS-D, with increased breath CH4 in IBS-C subjects that correlated with increased levels of methanogens, including M. smithii, in stool samples, whereas IBS-D subjects had increased breath levels of both hydrogen (H2) and hydrogen sulfide (H2S) [6]. A recent study by Algera et al., found that increased breath CH4 levels were associated with decreased oroanal transit times in IBS subjects, and while that study did not measure H2S levels, they did find associations between increased breath H2 levels and increased oroanal transit times in IBS [11]. The authors proposed that these findings resulted from differences in gut microbial compositions in different groups of IBS subjects [11], which is consistent with our identification of distinct gut microtypes and breath gas profiles in IBS-C vs. IBS-D subjects [6]. Further, we found that increased breath H2S levels in IBS-D subjects correlated with increased abundances of H2S-producing bacterial species, including species from the genera Fusobacterium and Desulfovibrio, in stool samples [6], and microbial metabolic pathway analysis suggested enrichment of KEGG modules associated with methane production in CH4-positive IBS-C subjects, and enrichment of modules associated with H2S production in IBS-D subjects [6]. While these findings suggest that increased M. smithii and H2S producers may contribute to stool phenotypes in IBS-C and IBS-D subjects, respectively, these are associations and do not prove cause-and-effect. In this study, we used rat models to explore the effects of increased levels of methanogens or H2S producers on stool phenotypes, in order to identify the potential microbial drivers of phenotypes seen in the different IBS subtypes.

Methods

Adult male Sprague Dawley rats (Envigo, Madison, WI) were used to test the effects of (1) the methanogen M. smithii, and (2) the H2S producers Desulfovibrio piger and Fusobacterium varium, on stool phenotypes. Rats were housed under regular light–dark cycles of 12/12 h and had free access to food and water at all times. Both studies were approved by the Cedars-Sinai Institutional Animal Care and Use Committee (IACUC).

Effects of Methanobrevibacter smithii on Stool Consistency

Previous work showed that Sprague–Dawley rats are endogenously colonized with M. smithii and that M. smithii gavage only transiently augmented its absolute abundance [12]. However, when rats were placed on high-fat diet (HFD), M. smithii levels increased significantly and remained stable for the duration of HFD feeding [12]. Therefore, HFD was used to assess the effects of persistent M. smithii elevation on stool consistency (% wet weights). Sprague–Dawley rats were placed on a HFD (60% energy from fat [D12492; Research Diets, New Brunswick, NJ]). Three-day stool collections (for determination of % wet weights [13]) were performed at baseline and on days 12, 59, and 71, and M. smithii levels were determined at baseline, on days 3 and 52, and at euthanasia (Fig. 1). To confirm that any effects on stool consistency were associated with absolute levels of M. smithii, and not secondary HFD effects, rats were subsequently divided into 3 groups and gavaged for 10 days with either 1.5 mg/ml lovastatin lactone, 1.5 mg/ml lovastatin hydroxyacid, or water (as controls), after which additional stool collections were performed (Fig. 1). Lovastatin is a natural inhibitor of methanogens, with the lactone and hydroxyacid forms having different mechanisms of action. Specifically, lovastatin hydroxyacid can inhibit archaeal cell wall biosynthesis by inhibiting the hydroxy‐methylglutaryl coenzyme A (HMG‐CoA) reductase enzyme (HMGR), and lovastatin lactone can independently inhibit the activity of methylenetetrahydromethanopterin dehydrogenase (mtd), a key methanogenesis enzyme, by competing with coenzyme F420, a central catabolic cofactor required by mtd [14, 15]. Stool and small bowel (ileal) contents were collected at euthanasia and stored at − 80 °C prior to analysis.

Fig. 1
figure 1

Animal study design. A Investigation of the effects of diet-induced increases in the methanogen Methanobrevibacter smithii, and of methanogenesis inhibitors, on stool consistency. B Investigation of the effects of gavage with H2S producers Desulfovibrio piger and Fusobacterium varium on stool consistency

Full size image

DNA Extraction

DNAs were extracted from stool and small bowel contents using the MagAttract PowerSoil DNA KF Kit (Qiagen) with some modifications [16]. Extracted DNAs were purified using a KingFisher Duo automated system (ThermoFisher Scientific, Waltham, MA), and DNA purity and concentration were determined using a NanoDrop One spectrophotometer (ThermoFisher Scientific).

Analysis of M. smithii Levels by Quantitative Polymerase Chain Reaction (qPCR)

Levels of the methanogen M. smithii in stool and small bowel contents were determined by qPCR using primers and probes targeting the beta subunit of the RNA polymerase (rpoB) gene [17]. Assays were optimized by Applied Biosystems (Custom Taqman Gene Expression Assays). Real-time qPCR was performed on a QuantStudio 6 Flex System (ThermoFisher Scientific) as follows: 1 µL of 20 × Custom TaqMan Gene Expression assay solution (ThermoFisher Scientific), 10  µL of TaqMan Fast Advanced Master Mix (ThermoFisher Scientific), 7 µL of PCR grade water and 2 µL of template DNA (25 ng/µl), at 50 °C for 2 min, 95 °C for 2 min, 40 cycles of 95 °C for 1 s, and 60 °C for 20 s. DNA from an M. smithii stock culture was extracted using the same protocol, and standard curves with tenfold serial dilutions were prepared for use as qPCR standards.

Effects of H2S Producers on Stool Consistency

The effects of H2S producers on stool consistency were determined using D. piger and F. varium. D. piger ATCC29098 (ATCC, Virginia) was grown anaerobically in sterile 1249 modified Baar’s medium for sulfate reducers (ATCC) and plated on trypticase soy agar (TSA) with 5% sheep blood (BD, New Jersey). Plates were incubated anaerobically at 37◦C for 48–96 h to obtain single colonies. H2S production was confirmed by growing isolated D. piger colonies in 1249 modified Baar’s medium for sulfate reducers with 5% of ferrous ammonium sulfate.

F. varium clinical isolates were obtained from the Cedars-Sinai Microbiology Department and grown anaerobically in sterile peptone yeast extract broth (Anaerobe Systems, Morgan Hill, CA) and plated on TSA with 5% sheep blood (BD). Isolated single F. varium colonies were cultured in SIM (Sulfide, Indole, Motility) medium (Hardy Diagnostics, CA) to confirm H2S production.

Gavage with H2S Producers

Rats were divided into three groups and gavaged with (1) sterile 1 × PBS (controls), (2) 1 × 108 CFU/mL D. piger, or (3) 1 × 108 CFU/mL F. varium. Liquid cultures were grown from single colonies as described above until 1 × 108 CFU/mL was achieved. Cultures were centrifuged for 10 min at 3000×g at 4 °C, and bacterial pellets were washed twice and resuspended in 1 × PBS to achieve 1 × 108 CFU/mL. Rats were gavaged on days 0, 2, and 4 (Fig. 1). These rats were fed a standard chow diet (13% energy from fat [PicoLab Rodent Diet 20; LabDiet, St. Louis, MO]).

Stool Collection and H2S Measurements

Stool samples were collected by anal stimulation [18] at baseline (3-day collection) and on days 5, 7, 10, 12, and 20 (single collections, see Fig. 1) for determining stool consistency (% wet weights). H2S production in stool from all rats was measured on day 20. Samples were immediately homogenized with sterile 1 × PBS, placed in sterile Erlenmeyer flasks sealed with rubber stoppers connected to a stopcock and incubated at 37 °C for 2 h. Gas samples were withdrawn into gas-impermeable bags and sent for measurement of gases by gas chromatography (Gemelli Biotech, Raleigh, NC).

16S rRNA Sequencing and Analysis

Stool samples collected at euthanasia from animals gavaged with H2S producers were used for DNA isolation and preparation of 16S rDNA hypervariable V3 and V4 region libraries following an Illumina protocol (https://support.illumina.com/documents/documentation/chemistry_documentation/16s/16smetagenomic-library-prep-guide-15044223-b.pdf) (Illumina, San Diego, CA), using primers S-D-Bact0341-b-S-17 and SD-Bact-0785-a-A-21 [19] as described previously [16]. Final libraries were analyzed using Agilent High Sensitivity DNA chips (Agilent) on an Agilent 2100 Bioanalyzer System and quantified using Qubit 1X dsDNA High Sensitivity Assay kits (Invitrogen by Thermo Fisher Scientific, Waltham, MA, USA) on a Qubit 4 Fluorometer (Invitrogen). Pooled libraries were paired-sequenced (2 × 301) on a MiSeq instrument (Illumina). Reads were trimmed and merged using CLC Genomics Workbench software version 22.0.2 (Qiagen). Operational Taxonomic Unit (OTU) clustering and taxonomic analyses were carried out using CLC Microbial Genomics Module 22.1.2 (Qiagen) against the SILVA Small Subunit (SSU) rRNA Database v. 138.1 (2020), with 97% of similarity and using default parameters. Creation of new OTUs was not allowed. Alpha diversity (Shannon index) and beta diversity (Unweighted UniFrac) were calculated using the CLC Microbial Genomics Module (Qiagen). The PERMANOVA test was used to calculate differences in beta diversity. Differences in microbial relative abundances between groups were calculated using the CLC Microbial Genomics Module (Qiagen). Significance was determined by Wald test and Wilcoxon test and, due to the small number of samples, a False Discovery Rate (FDR) < 0.1 was considered significant. Microbial metabolic pathways were analyzed using MetaCyc Pathway Database (2022–05) and CLC Microbial Genomics Module (Qiagen).

Statistical Analysis

Descriptive analyses are presented as mean ± standard deviation. Continuous variables were compared by t-test or Mann–Whitney U-test for comparisons between two groups. Comparisons between three or more groups were analyzed by one-way ANOVA or Kruskal–Wallis. Correlations between variables were calculated by Spearman’s test (two-tailed). ROC curves were used to analyze thresholds for H2S levels. Statistical analysis was performed using SPSS 24.0 (SPSS® Inc., Chicago, IL), and GraphPad Prism® 9.5.1 (GraphPad Software, La Jolla, CA). Graph construction was performed using GraphPad Prism® 9.5.1 (GraphPad Software). Significance was set at P < 0.05.

Results

Methanobrevibacter smithii Is Associated with a Constipation-Like Phenotype in Rats

To determine the effects of increased M. smithii levels on stool phenotypes, adult male Sprague Dawley rats (N = 30) were placed on a HFD. Absolute levels of stool M. smithii were increased significantly after 3 days on HFD when compared to baseline (1.69 × 105 ± 1.57 × 105 copies/gram vs. 0.91 × 105 ± 0.71 × 105 copies/gram, P = 0.0127) and were further increased after 52 days on HFD (3.06 × 105 ± 2.89 × 105 copies/gram, P < 0.001, Fig. 2A). Stool % wet weights were significantly decreased after 12 days on HFD compared to baseline (P < 0.0001) and remained significantly decreased after 59 days on HFD (P < 0.0001, Fig. 2B). Further, the decrease in % wet weight after 59 days on HFD was associated with the increase in M. smithii levels (R = − 0.38, P = 0.037), indicating that higher M. smithii levels were associated with drier stool.

Fig. 2
figure 2

Changes in absolute levels of M. smithii in stool (A) and stool % wet weights (B) of rats after 59 days on HFD. Changes in absolute levels of M. smithii in stool (C) and stool wet weights (D) after 10 days of treatment with water (controls), lovastatin hydroxyacid, or lovastatin lactone. E Changes in absolute levels of M. smithii in the small bowel (ileum) after 10 days of treatment with water (controls), lovastatin hydroxyacid, or lovastatin lactone. Horizontal bars denote mean ± SD

Full size image

To confirm that these effects were due to increased M. smithii, rather than the effects of the HFD alone, rats were randomized into 3 groups and treated for 10 days with lovastatin lactone (N = 10), lovastatin hydroxyacid (N = 10), or water (controls, N = 10) while being maintained on the HFD. Levels of M. smithii in stool samples were not significantly different between controls and rats treated with either form of lovastatin (Fig. 2C). However, rats treated with lovastatin hydroxyacid exhibited a significant increase in % wet weight when compared to controls (P = 0.0246, Fig. 2D), indicating a partial reversal of the constipation-like phenotype.

In contrast to stool, rats treated with lovastatin lactone exhibited significant reductions in absolute M. smithii levels in the distal small bowel (ileum) (29.24 × 105 ± 19.91 × 105 copies/gram) when compared both to controls (144.80 × 105 ± 82.15 × 105 copies/gram, P < 0.0006) and to rats treated with lovastatin hydroxyacid (P = 0.0003, Fig. 2E). Levels of M. smithii in the small bowel of rats treated with lovastatin hydroxyacid were not significantly different from those in controls (Fig. 2E).

H2S-Producing Bacteria Induce H2S Production and a Diarrhea-Like Phenotype in Rats

To determine the effects of increased levels of H2S producers on stool phenotypes, adult male Sprague Dawley rats were gavaged with D. piger (N = 16), F. varium (N = 16) ,or PBS (controls, N = 8). Rats gavaged with D. piger had increased % wet weights on day 10 compared to controls (P < 0.0001, Fig. 3A). Rats gavaged with F. varium had significantly increased % wet weights at all time points when compared to controls [day 5 (P = 0.019), day 7 (P = 0.005), day 10 (P < 0.0001), day 12 (P = 0.027), and day 20 (P = 0.032)], with the peak occurring on day 10 (Fig. 3B). In addition, H2S production in stool samples from rats gavaged with D. piger (P = 0.0005) or F. varium (P = 0.006) was significantly greater than H2S production in stool samples from control rats (Fig. 3C). Using a cutoff of 0.48 ppm of H2S (95% Cl 0.65–1.0), stool H2S levels in rats gavaged with F. varium were positively associated with stool % wet weight on day 20 (r = 0.447, P = 0.042).

Fig. 3
figure 3

Changes in stool wet weight in rats gavaged with A D. piger and B F. varium. P-values denote comparisons between SRB-gavaged rats and controls. C H2S production in stool collected from D. piger- or F. varium-gavaged rats and controls. Horizontal bars denote mean ± SD. Statistical analyses by Mann–Whitney U-test

Full size image

H2S Producers Induce Changes in Stool Microbial Profiles

16S rRNA gene sequencing was performed using stool samples obtained at euthanasia, as rats gavaged with H2S producers exhibited increased stool H2S levels when compared to controls at this timepoint. Alpha diversity analysis (Shannon index) showed no differences between rats gavaged with H2S producers and controls, D. piger vs. controls, P = 0.5; F. varium vs. controls, P = 0.2). Although D. piger and F. varium were not detected in stool samples at euthanasia (17 days after the last gavage) and alpha diversity was not different between groups, there were significant differences in the stool microbial profiles of rats gavaged with H2S producers when compared to controls. Beta diversity (PERMANOVA) was different between control rats and rats gavaged with D. piger (P = 0.0001) and between control rats and rats gavaged with F. varium (P = 0.001) (Fig. 4). Moreover, beta diversity was positively associated with stool H2S levels (PCo3: r = − 0.401, P = 0.014).

Fig. 4
figure 4

Principal coordinates analysis (PCoA) plot of the unweighted UniFrac distances between control rats (green), rats gavaged with D. piger (purple), and rats gavaged with F. varium (red)

Full size image

Taxonomic differences were identified even at the phylum level between rats gavaged with H2S producers and controls, including increases in relative abundance (RA) of Deferribacterota (D. piger vs. controls, fold change (FC) = 8.93, P = 0.0006, Adj. P-value = 0.006; F. varium vs. controls, FC = 7.70, P = 0.002, Adj. P-value = 0.007), Bacteroidota (D. piger vs. controls, FC = 1.58, P = 0.002, Adj. P-value = 0.01; F. varium vs. controls, FC = 1.72, P = 0.0003, Adj. P-value = 0.004), Desulfobacterota (D. piger vs. controls, FC = 2.13, P = 0.017, Adj. P-value = 0.06; F. varium vs. controls, FC = 2.61, P = 0.003, Adj. P-value = 0.008). F. varium-gavaged rats also exhibited increased RA of phylum Firmicutes compared to controls (FC = 1.78, P = 0.002, Adj. P-value = 0.007).

At the family level, rats gavaged with D. piger exhibited increased RA of Bacteroidaceae (FC = 1.83, P < 0.0001, Adj. P-value = 0.002), Deferribacteraceae (FC = 10.15, P = 0.0005, Adj. P-value = 0.015), Lachnospiraceae (FC = 2.21, P = 0.0007, Adj. P-value = 0.015), Enterobacteriaceae (FC = 7.12, P = 0.001, Adj. P-value = 0.016), Staphylococcaceae (FC = 4.54, P = 0.003, Adj. P-value = 0.032), and Rikenellaceae (FC = 1.60, P = 0.01, Adj. P-value = 0.074), and decreased RA of Eggerthellaceae (FC = − 1.52, P = 0.006, Adj. P-value = 0.055), when compared to controls. Rats gavaged with F. varium exhibited increased RA of family Butyricicoccaceae (FC = 2.24, P = 0.0004, Adj. P-value = 0.023) when compared to controls. However, at the genus level, the RA of several genera were found to be different in F. varium-gavaged rats compared to controls, including increased RA of an unknown genus from family Lachnospiraceae (FC = 19.94, P = 0.0008, Adj. P-value = 0.039) and of Bilophila (FC = 4.77, P = 0.001, Adj. P-value = 0.048), and decreased RA of Akkermansia (FC = − 19.03, P = 0.004, Adj. P-value = 0.076). D. piger-gavaged rats also exhibited increased RA of the unknown genus from family Lachnospiraceae (FC = 18.96, P = 0.0009, Adj. P-value = 0.052), as well as Frisingicoccus (FC = 106.70, P = 0.0005, Adj. P-value = 0.052) and Mucispirillum (FC = 6.86, P-value = 0.002, Adj. P-value = 0.069), and a trend toward increased RA of Bilophila (FC = 3.66, P = 0.007, Adj. P-value = 0.116) and Sutterella (FC = 31.47, P = 0.006, Adj. P-value = FDR P = 0.116) (Supplemental Table 1).

A Spearman correlation was performed, including all animals, between stool H2S levels and the RA of known H2S-producing bacteria, as well as the most abundant bacteria found in the stool of these animals. There were positive associations between stool H2S levels and the RA of genera Bilophila (r = 0.282, P = 0.091), Escherichia-Shigella (r = 0.315, P = 0.058), the unknown genus from family Lachnospiraceae (r = 0.28, P = 0.093), and Sutterella (r = 0.331, P = 0.046). The positive association between H2S levels and the unknown genus from family Lachnospiraceae (r = 0.49, P = 0.024) was stronger in rats gavaged with F. varium, whereas the association with Sutterella was stronger in rats gavaged with D. piger (r = 0.592, P-value = 0.003).

Predicted microbial metabolic pathways were also analyzed and compared to those in controls, and several pathways were found to be different between groups. Importantly, the L-homocysteine biosynthesis pathway, which is used by specific bacteria to consume and produce H2S, was positively associated with stool % wet weight on day 20 (r = 0.30, P = 0.06). This pathway was increased in rats gavaged with F. varium compared to controls (P < 0.001) and was positively associated with H2S levels in these animals (r = 0.5, P = 0.021), but this was not observed when D. piger-gavaged animals were analyzed (r = 0.24, P = 0.264).

Discussion

In this study, we demonstrate that induced increases in levels of the methanogen M. smithii result in a constipation-like phenotype (decreased stool % wet weights) in a rat model, whereas increases in either of the H2S-producing species D. piger or F. varium result in a diarrhea-like phenotype (increased stool % wet weights) and increased stool H2S production. These findings build on our recent human studies linking increased levels of stool methanogens, including M. smithii, to increased breath methane CH4 in IBS-C subjects [6], and linking increased abundances of H2S-producing taxa, including the genera Fusobacterium and Desulfovibrio, to increased breath H2S in IBS-D subjects [6]. Taken together, these results suggest that methanogens such as M. smithii, and H2S producers such as D. piger and F. varium, may contribute to the constipation-predominant and diarrhea-predominant subtypes in IBS subjects, respectively.

M. smithii and other hydrogenotrophic methanogens use H2 produced by syntrophic bacterial species for the generation of CH4 [20]. CH4 has been directly linked to slowing of intestinal transit in an animal model and to an increased motility index in CH4-producing IBS subjects [21]. Several studies have linked breath CH4 to constipation [22, 23] and to IBS-C [3, 24,25,26], and M. smithii has been shown to be the predominant methanogen in CH4-postive IBS-C subjects [8]. We recently found that increased CH4 on breath test correlated with increased abundance of methanogens, including M. smithii, in stool samples from IBS-C subjects, and predicted microbial metabolic pathway analysis indicated enrichment of pathways associated with methanogenesis in these subjects [6]. These findings are also consistent with an independent study which recently found that increased CH4 on breath test was associated with slower oroanal transit times in IBS subjects, whereas increased H2 on breath test was associated with more rapid oroanal transit times [11]. In the present study, increased stool M. smithii levels correlated with a constipation-like phenotype (decreased stool % wet weights) in a rat model. These rats were also treated with different forms of lovastatin, which have anti-methanogenesis effects based on their ability to compete with the key cofactor F420 for its binding site on the mtd enzyme (lactone form), as well as their ability to inhibit archaeal cell wall biosynthesis by inhibiting the HMGR enzyme (hydroxyacid form) [14, 15]. Treatment with lovastatin lactone resulted in a significant reduction in M. smithii levels in the distal small bowel (ileum), and treatment with lovastatin hydroxyacid resulted in looser stools, when compared to rats maintained on the HFD, indicating that inhibiting methanogens resulted in normalization of % wet weights away from the constipation-like phenotype. Of note, while methanogens are typically most abundant in the colon in humans, we have previously demonstrated high abundance of methanogens in the small intestine, both in rats [12] and in some human subjects [27]. These findings suggest that increased levels of methanogens, particularly M. smithii, directly contribute to a constipation phenotype in subjects with IBS-C.

H2S-producing bacteria use different substrates in the production of H2S—F. varium primarily uses the amino acid cysteine, whereas D. piger uses sulfate—and compete with methanogens for H2 [28]. In contrast to CH4, H2S is also produced by mammalian cells, albeit in lower amounts [29]. Although H2S is often considered to be a toxic gas (at least when present in excess [30]), in recent years it has been shown to be a signaling molecule involved in many physiological functions, with anti-inflammatory and anti-oxidative effects in lung injury [31] as well as beneficial effects in the cardiovascular system [32]. With regard to the gastrointestinal tract, we previously linked higher breath H2S levels to diarrhea [33], which is consistent with animal studies demonstrating that H2S acts as a smooth muscle relaxant, through the inhibition of L-type calcium channels [34]. In addition, H2S has been linked to the diarrheal disorder ulcerative colitis [35], and F. varium specifically has been identified in Japanese patients with ulcerative colitis [36, 37]. We recently identified higher relative abundances of genus Fusobacterium and Desulfovibrio species in IBS-D subjects, with strong correlations between Fusobacterium and H2S on breath test [6]. In the present study, we show for the first time that increased levels of F. varium or D. piger result in a diarrhea-like phenotype in a rat model, and an increase in stool production of H2S. These findings suggest that increased levels of H2S-producing bacteria may contribute to a diarrheal phenotype in subjects with IBS-D.

Interestingly, the 16S rRNA sequencing results did not identify the gavaged H2S producers in stool at the time of euthanasia. It is possible that colonization occurred in a more proximal part of the gut, or that colonization with the gavaged bacteria was more transient. However, gavage with H2S producers appears to result in increased RA of other H2S-producing bacteria in these animals. The RA of Bilophila, a well-known H2S producer [38], was increased in both rat models but particularly in rats gavaged with F. varium, whereas the RA of Sutterella, a sulfur metabolizing bacterium [39], was more significantly increased in rats gavaged with D. piger. Furthermore, Sutterella RA was also found to be associated with stool H2S levels. Rats gavaged with D. piger also exhibited an increase in family Lachnospiraceae. This bacterial family was recently shown to increase the production of reactive sulfur species, such as H2S, in the host [40]. We hypothesize that these bacteria could have taken a niche created by the H2S producers with which these rats were initially gavaged, although further work would be needed to prove this. Lastly, the L-homocysteine biosynthesis pathway, which is used to both consume and produce H2S [41], was positively associated with stool % wet weight on day 20 and was predicted to be increased in rats gavaged with F. varium compared to controls. Using a cutoff of 0.48 ppm H2S, we also identified a positive association between this pathway and stool H2S levels in rats gavaged with F. varium, which could explain the diarrhea-like phenotype in these animals.

This study has some limitations. M. smithii levels were induced indirectly through feeding a high-fat diet, rather than by a more direct method such as gavage. This approach was chosen as we have previously shown that gavage does not result in persistent increases in the abundance of M. smithii, which is strictly anaerobic, whereas significant and stable increases in M. smithii levels can be induced via a HFD [12]. The decreases in stool wet weights correlated with increases in M. smithii levels, and were partly reversible using methanogenesis inhibitors, suggesting that these effects were due to M. smithii rather than non-specific effects of the HFD. More studies are needed to study the mechanistic effects of these organisms and the gases that they produce in the gastrointestinal tract.

In conclusion, this study further substantiates the presence of distinct microbiome-based subtypes (microtypes) that drive the predominant bowel phenotypes in IBS, consistent with the recent suggestion by Algera et al. that IBS subjects with more rapid vs. slow oroanal times have different underlying gut microbial compositions and associated breath gas profiles [11]. Specifically, we show that increased levels of methanogenic archaea such as M. smithii appear to drive a constipation phenotype, and increases in H2S-producing bacteria such as F. varium or D. piger appear to drive a diarrheal phenotype. These findings, and an increased understanding of the underlying mechanisms, may facilitate the development of targeted microtype-based therapies in IBS.