Genomic and functional analysis of Romboutsia ilealis CRIB^T reveals adaptation to the small intestine

Global genome features

R. ilealis CRIB^T contains a single, circular chromosome of 2,581,778 bp and a plasmid of 6,145 bp (Table 1 and Fig. 1). The chromosome contains 2,351 predicted protein CDS, of which 321 were annotated as hypothetical and for 91, only a domain of unknown function could be assigned. The plasmid carries eight predicted protein CDS, of which none was recognized for having a metabolic or replicative function. Furthermore, it appears to be a non-mobilizable plasmid, given that it lacks any known mobilization-associated genes. The overall G + C content of the genome is 27.9%, which is in good agreement with a G + C content of 28.1 mol% previously determined for R. ilealis CRIB^T by HPLC methods (Gerritsen et al., 2014).

With a total of 14 copies of the 16S ribosomal RNA (rRNA) gene, R. ilealis CRIB^T is among the species with the highest number of 16S rRNA gene copies reported up to this date (Lee, Bussema & Schmidt, 2009). High numbers of rRNA operons have been proposed to be indicative for fast growth and to allow microbes to respond quickly to changes in available resources (Klappenbach, Dunbar & Schmidt, 2000). In addition, a high copy number of the rRNA operon has been suggested to be essential for successful sporulation and germination (Yano et al., 2013). This is also reflected in the observation that in general the species that contain the highest number of reported rRNA operons, including R. ilealis CRIB^T, belong to the spore-forming bacterial orders Bacillales and Clostridiales. Not all of the 16S rRNA gene copies in R. ilealis CRIB^T are embedded in the conserved 16S-23S-5S rRNA operon structure. Of the fifteen locations containing rRNA genes, ten are in the classical order 16S-23S-5S. The other five operons are characterized by duplicated or missing rRNA genes, or a different order of the genes. It should be noted that the current assembly contains three gaps, all of which are located within rRNA operons. Diverging rRNA operon structures have been reported for other genomes containing multiple rRNA operons, as a result of duplications (Bensaadi-Merchermek et al., 1995; Schwartz, Gazumyan & Schwartz, 1992).

A cluster of orthologous genes (COG) category (Tatusov, Koonin & Lipman, 1997) could be assigned to 1,647 of the predicted proteins (70%) including 372 proteins (16%) assigned to the categories R (general function prediction only) and S (function unknown) (Fig. S1). With InterProScan a predicted function could be assigned to 82% of the predicted proteins. Based on the InterPro and PRIAM classifications (Claudel-Renard et al., 2003), an enzymatic function could be predicted for more than 500 proteins.

General metabolic pathways

Analysis of the CDS predicted from the R. ilealis CRIB^T genome revealed the presence of a complete set of enzymes for the glycolytic pathway. In line with the anaerobic lifestyle of the organism, enzymes for the oxidative phase of the pentose phosphate pathway could not be detected. Additionally, the genes that encode enzymes involved in the tricarboxylic acid cycle were lacking. Subsequently a metabolic model was constructed with Pathway tools v18.0. A flux balance analysis with the model was performed, suggesting that R. ilealis CRIB^T is a mixed acid fermenter as previously reported (Gerritsen et al., 2014). Predicted end products of fermentation are a mixture of acetate, formate, lactate and ethanol, with the possibility of gas formation (CO₂ and H₂). In addition to ethanol, which can be produced during mixed acid fermentation, 1,2-propanediol was predicted to be formed via the L-fucose degradation pathway. The fermentation end products formate, acetate and lactate are predicted to be produced from pyruvate. No other solvents were predicted to be produced by the metabolic model. The only metabolite produced by R. ilealis CRIB^T that was not accounted for by the metabolic model, was propionate. None of the three established pathways for propionate production in the intestinal tract, i.e., the succinate, acrylate or the propanediol pathway (Reichardt et al., 2014), could be identified at the genetic level in the genome of R. ilealis CRIB^T. Although propionate is only produced in low amounts (max. 3 mM in 24 h) it is noteworthy because propionate production was observed repeatedly during in vitro growth in this study (see Table 2) and as previously reported (Gerritsen et al., 2014).

The analysis of the genome and the prediction by the model indicated that fermentation is probably the main process for energy conservation in R. ilealis. However, the presence of a sulfite reductase gene cluster (CRIB_1284-CRIB_1286) of the dissimilatory asrC-type (Dhillon et al., 2005) points at possible anaerobic respiration. Similar siroheme-dependent sulfite reductases are found in many close-relatives of R. ilealis such as I. bartlettii, Clostridium sordellii and C. difficile (Czyzewski & Wang, 2012). Sulfite reduction by R. ilealis CRIB^T, and close relatives, has been previously demonstrated in vitro (Gerritsen et al., 2014), and increased growth yield and metabolite production was observed in the presence of sulfite for R. ilealis CRIB^T (Table S7). In the intestinal tract, sulfite is derived from food sources that contain sulfite as a preservative, and it has been shown that neutrophils release sulfite as a part of the host defence against microbes (Mitsuhashi et al., 1998).

Metabolism of growth factors and cofactors

Complete pathways are present for the biosynthesis of the amino acids aspartate, asparagine, glutamate, glutamine and cysteine, using carbon skeletons available from central metabolites or via conversion of other amino acids. However, many genes encoding enzymes required for biosynthesis of other amino acids appeared to be absent in R. ilealis CRIB^T. As most missing genes are part of well-studied pathways, it is unlikely these functionalities are encoded by unknown genes and likely represent true auxotrophies. The absence of genes to produce branched-chain amino acids (leucine, isoleucine and valine) was also reflected in the absence of branched chain fatty acids in the cell membrane of R. ilealis, which is characteristic for the genus Romboutsia (Gerritsen et al., 2014). From these observations it can be concluded that R. ilealis depends on a number of exogenous amino acids, peptides and/or proteins to fuel protein synthesis. The dependency on an exogenous source of amino acids is reflected by the identification of multiple amino acid transporters, including an arginine/ornithine antiporter, multiple serine/threonine exchangers, a transporter for branched amino acids, and several amino acid symporters and permeases without a predicted specificity. Furthermore, numerous genes were annotated as protease or peptidase, including several with a signal peptide.

R. ilealis CRIB^T appears to contain all genes for de novo purine and pyrimidine synthesis, as well as for the production of the coenzymes NAD and FAD via salvage pathways from niacin and riboflavin, respectively. While some organic cofactors can be produced by R. ilealis CRIB^T, it mainly relies on salvage pathways (e.g., for lipoic acid) or exogenous sources for the supply of precursors, mainly in the form of vitamins (e.g., thiamin, riboflavin, niacin, pantothenate, pyridoxine, biotin, vitamin B12).

Carbohydrate transport and metabolism

As previously reported, R. ilealis CRIB^T is able to utilize a wide variety of carbohydrates (Gerritsen et al., 2014). Previously, good growth of R. ilealis on L-fucose, glucose, raffinose and sucrose was described, in addition to moderate growth on D-arabinose and D-galactose and weak growth on D-fructose, inulin, lactose, maltose and melibiose. Growth on L-fucose, fructose, galactose, glucose, lactose, maltose, melibiose, raffinose and sucrose was predicted from the genome-scale metabolic model as well. For these different carbohydrates, the genes encoding the specific carbohydrate degradation enzymes were found distributed throughout the genome in gene clusters together with their respective transporters and transcriptional regulator. The only carbohydrate utilized by R. ilealis CRIB^T that was not predicted based on the metabolic model, was D-arabinose. Although a separate arabinose transporter, similar to the maltose and sucrose transporters, could be identified in the genome R. ilealis CRIB^T, no separate pathway for the use of D-arabinose could be predicted, However, it is likely that the L-fucose degradation pathway (encoded by genes CRIB_1294-CRIB_1298) is also used for D-arabinose utilization as is also observed in other intestinal species (LeBlanc & Mortlock, 1971). In addition to the carbohydrates for which growth was studied, a gene cluster involved in the degradation of the host-derived carbohydrate sialic acid could be predicted (CRIB_613-CRIB_619) (Almagro-Moreno & Boyd, 2009). The structure of this gene cluster is similar to the one identified in C. difficile (Ng et al., 2013). The ability to degrade the predominantly host-derived carbohydrates, L-fucose and sialic acid, suggest a role in the utilization of mucin, an abundant host-derived glycoprotein in the intestinal tract (Derrien et al., 2010; Ouwerkerk, De Vos & Belzer, 2013). However, no growth on mucin was observed (Table S7), which is in line with the lack of a predicted extracellular fucosidase and/or sialidase.

Other genes encoding niche-specific functionalities

A gene cluster encoding a urease, consisting of three subunits (ureABC), and a number of urease accessory genes was identified (CRIB_1381-CRIB_1388). The gene cluster identified in R. ilealis CRIB^T is very similar to the urease gene cluster in the genome of C. sordellii (Fig. S3), a species in which the urease activity is used to phenotypically distinguish C. sordellii strains from C. bifermentans strains (Roggentin et al., 1985). Furthermore, a possible ammonium transporter (CRIB_1389) was identified in the genome of R. ilealis CRIB^T next to the urease gene cluster. Ureases are nickel-containing metalloenzymes that catalyse the hydrolysis of urea to ammonia and carbon dioxide, and thereby these enzymes allow microbes to use urea as nitrogen source by assimilation via glutamate. They are ubiquitous proteins occurring in diverse organisms (Mobley, Island & Hausinger, 1995). In the intestinal environment, where urea is abundantly present (Fuller & Reeds, 1998), some bacteria use ureases to survive the acidic conditions in the upper part of the intestinal tract as urea hydrolysis leads to a local increase in pH (Rutherford, 2014).

Another gene encoding a niche-specific functionality is the predicted choloylglycine hydrolase. Proteins within the choloylglycine hydrolase family are bile salt hydrolases (BSHs), also known as conjugated bile acid hydrolases (CBAHs), that are widespread among intestinal microbes (Ridlon, Kang & Hylemon, 2006). They are involved in the hydrolysis of the amide linkage in conjugated bile salts, releasing primary bile acids. There is a large heterogeneity among BSHs, for example with respect to their substrate specificity. The BSH of R. ilealis CRIB^T was found to be the most similar to the one found in Clostridium butyricum. Although the physiological advantages of BSHs for the microbes are not completely understood, it has been hypothesized that they constitute a mechanism to detoxify bile salts and thereby enhance bacterial colonization (Czyzewski & Wang, 2012).

Metabolite and transcriptome analysis of R. ilealis CRIB^T during growth on different carbohydrates

To study key pathways predicted to be involved in carbohydrate utilization and their regulation in more detail, a genome-wide transcriptome analysis was performed, focussing on four experimental conditions. Firstly, growth on glucose, a preferred substrate for many microbes present in the intestinal tract, was studied. Secondly the growth on fructans, oligo-and polysaccharides present in many food items was examined. Previously weak growth on inulin, a polysaccharide consisting of long chains of β1 → 2 linked fructose units, was observed (Gerritsen et al., 2014). For this study a shorter fructan (FOS P06, DP2-4) was chosen, because growth on shorter fructans is likely more relevant for microbes living in the small intestine (Zoetendal et al., 2012). Thirdly, growth on L-fucose was examined, as growth on this substrate was found to be unique for R. ilealis CRIB^T compared to other related microbes. Finally, R. ilealis CRIB^T was also grown in the basal medium in the absence of an additional carbon source for comparison (control condition).

Based on measurements of optical density and pH during growth (growth characteristics of individual cultures can be found in Table S1), samples were drawn in the mid-exponential phase (∼8–10 h incubation; used for transcriptome analysis) and in stationary phase (24 h incubation), and sugar utilization and fermentation products were measured with HPLC (Table 2). In neither of the experimental conditions the supplied carbohydrates were depleted, and metabolites were still produced at the time of sampling at ∼8–10 h and 24 h, which further confirmed that samples obtained for transcriptome analysis at ∼8–10 h were taken during exponential growth. In the FOS cultures, an accumulation of extra-cellular fructose was observed. As predicted from the metabolic model, growth on glucose resulted in the production of formate, acetate and lactate (Table 2).

Growth on FOS was marginally lower than that on glucose, however, after 24 h of growth, the same fermentation products were observed in similar amounts (Table 2). Growth on L-fucose showed production of 1,2-propanediol instead of lactate. The fact that 1,2-propanediol was observed in one of the control cultures could be explained by the fact that an L-fucose grown culture was used as inoculum for this culture, leading to carry-over of minor amounts of metabolites.

For the genome-wide transcriptome analysis of triplicate cultures grown in the four different conditions (i.e., a total of 12 cultures), a total of 159,250,634 150 bp-reads were generated by RNA-seq (overview in Table S2). Principal component analysis of the transcriptomes of the individual cultures showed that the cultures clustered by condition (Fig. 2).

Differential expression of genes involved in carbohydrate degradation and fermentation in R. ilealis CRIB^T

To identify differentially regulated genes, pairwise comparisons were done with cuffdiff (Trapnell et al., 2013) using a cut off of ≥1.5 log2 (fold-change) and q-value ≤ 0.05. Figure 3 shows a heat map of all differentially regulated genes, and exact numbers can be found in Tables S3–S6.

The gene cluster involved in glycolysis (CRIB_186-CRIB_191) was most abundantly expressed in the conditions that support the highest growth rates determined by the highest cell density reached in the time period that was measured (glucose, followed by FOS; Fig. 3). This was also reflected in the fact that expression of genes encoding proteins involved in replication such as ribosomal proteins, proteins involved in cell wall biosynthesis and general cell division processes were most strongly expressed during growth in the presence of glucose and to a lesser extent FOS. Other genes involved in the central sugar metabolic pathways (e.g., CRIB_1849, CRIB_140, CRIB_2223, and CRIB_105) were upregulated in these conditions, albeit not significantly differentially regulated. This suggests that these genes are less tightly regulated at the transcriptional level, probably because they are also involved in other processes than sugar degradation (Commichau et al., 2009). The metabolic model suggests that this is indeed the case, as some of the enzymes produce intermediates which can be consumed by fatty acid biosynthesis and amino acid biosynthesis processes.

Altogether, the transcriptome of R. ilealis CRIB^T grown on FOS was very similar to its transcriptome when grown on glucose (Fig. 2), with only 18 genes significantly upregulated during growth in the presence of FOS compared to glucose (Table S4). Apparent was the upregulation of the gene clusters that code for proteins involved in the transport and degradation of the respective sugars or their derivatives (Fig. 3). In the presence of glucose, the glucose-specific PTS system (CIRB_2017-CRIB_2018) was significantly upregulated, together with its associated transcriptional regulator (CRIB_2019). In turn, in the presence of FOS, two clusters predicted to be involved in sucrose degradation (CRIB_148-CRIB_152 and CRIB_1458-1461) were significantly upregulated. The third gene cluster predicted to be involved in sucrose degradation (CRIB_1399-1400) was not significantly regulated during growth on FOS. However, it should be noted that these genes are located in a cluster functionally annotated to melibiose metabolism and are most likely regulated by the transcriptional regulator in this cluster. In addition to the two sucrose degradation clusters, a transport cluster of unknown function (CRIB_1506-CRIB_1509) was upregulated during growth on FOS, albeit only significantly when compared to growth on glucose. During growth in the presence of L-fucose, the gene cluster predicted to be involved in L-fucose degradation (CRIB_1294-CRIB_1298) was significantly upregulated, including the gene encoding the corresponding transcriptional regulator (CRIB_1299). An overview of the main carbohydrate degradation pathways regulated in the different conditions is given in Fig. 4.

During growth on glucose, L-lactate dehydrogenase (CRIB_684) was significantly upregulated, albeit not significantly compared to growth on FOS. This enzyme catalyses the reduction of pyruvate resulting in the production of L-lactate and the reoxidation of the NADH formed during glycolysis. Only at the time point of 24 h, lactate was observed (Table 2). This suggests that at time point ∼8–10 h the cells were starting to regenerate NAD by upregulating this gene. In the presence of L-fucose, NAD⁺ regeneration is achieved via the reduction of lactaldehyde to 1,2-propanediol by lactaldehyde reductase (CRIB_1300), which was upregulated in the presence of L-fucose together with the L-fucose degradation gene cluster. In the spent medium of L-fucose grown cells, 1,2-propanediol was already seen at time point ∼8–10 h, whereas no lactate production was observed. Another way to regenerate NAD⁺ is to reduce pyruvate to ethanol (Fig. 4). In the presence of both glucose and FOS, an upregulation was seen for the gene encoding the bifunctional aldehyde/alcohol dehydrogenase (CRIB_2231), which converts acetyl-CoA to ethanol. However, there were no samples in which ethanol was measured by HPLC analysis.

During growth on FOS, a small gene cluster (CRIB_601-CRIB_603) that includes a gene encoding an alternative sigma factor was significantly upregulated. This was also apparent in the control culture that was inoculated with FOS-preconditioned cells. This suggests that in the presence of FOS (or its derivatives sucrose or fructose) transcription is also regulated by RNA polymerase promoter recognition.

Expression and regulation of other environmentally relevant functions in R. ilealis CRIB^T

Noteworthy was the significant upregulation of a gene cluster related to iron transport (CRIB_892-CRIB-898) during growth on glucose and FOS compared to growth on L-fucose. The significance of this gene cluster for carbohydrate utilization is not known, however, several enzymes could be identified in the genome of R. ilealis CRIB^T that use different forms of iron as cofactor, for example the hydrogenases involved in hydrogen metabolism (Calusinska et al., 2010), several ferredoxins, and the L-threonine dehydratase (CRIB-426) that was significantly upregulated during growth on L-fucose. As multiple transporters involved in the transport of iron compounds were predicted, it is possible that the uptake of iron provides a competitive advantage to other microbes that are dependent on iron for respiration and other metabolic processes (Kortman et al., 2014).