In vitro bile acid transformation and impact of 13C-CA
Two human isolates C. scindens ATCC 35704 and C. scindens VPI 12708 and one murine isolate E. muris DSM 28561 (SJ24) were tested for their ability to 7-DH-ate human and mouse primary BAs in vitro. The human isolates were chosen due to their role as representative species and the murine isolate was selected due to E. muris SJ24 being the only 7-DH-ing murine strain isolated that grows rapidly (growth within 24 hours).
CA is known to induce the expression of bai genes (hence the name bile acid inducible genes) but it is unknown whether other BAs also promote gene expression. Thus, the goal of these experiments was to confirm CA 7-DH-ion and to investigate whether the 7-DH-ion of CDCA, UDCA and MCA (a and b) can occur in the absence of upregulation by CA.
As expected, all three strains 7-DH-ed CA but to varying extents (Figure 3). C. scindens ATCC 35704 and C. scindens VPI 12708 showed strong 7-DH-ing activity with 97% and 80% CA conversion to 7-DH-ed BAs after 48 hours, respectively. E. muris SJ24, on the other hand, only converted 9% of the CA provided into 7-DH-ed forms (Figure 3). C. scindens ATCC 35704 produced up to 52.26 µM DCA after 32 hours with some of the DCA subsequently oxidized to 12-oxolithocholic acid (12-oxoLCA) (Figure 3A). In contrast, C. scindens VPI 12708 produced the highest amount of DCA after 48 hours (71.42 µM), with little to no oxidized DCA forms (0.19 µM of 3-oxoDCA at 48 hours) (Figure 3B). Moreover, parallel experiments were performed by amending the cultures with 13C-CA in addition to other individual BAs to test whether the 7-DH-ion of CDCA, aMCA, bMCA and UDCA could be induced by CA (Supplementary Table 1). The lack of 12-oxo forms from the C. scindens VPI 12708 strain was expected since the 12α-hydroxysteroid dehydrogenase (12α-HSDH) required for this process was not detected by PCR in this strain (the full genome is currently unavailable) (data not shown). Finally, E. muris SJ24 only produced 8.1 µM of DCA after 48 hours with very low amounts of oxidized forms of DCA (0.21 µM of 12-oxoLCA) (Figure 3C). It is important to highlight that E. muris does not possess a 3a-HSDH encoded by baiA2 which was recently identified as an important component of the CA 7-DH-ion pathway [24]. BaiA1/3 has lower affinity to CA than BaiA2 [35] and baiA1/3 is present outside the bai operon [36]. Finally, all three strains also showed a modicum of 7-oxidation activity (Figure 3), resulting in the formation of 7-oxoDCA, which cannot be 7-DH-ed.
CDCA 7a-dehydroxylation is very limited for all three strains. Indeed, C. scindens ATCC 35704 only produced 1.59 µM LCA and C. scindens VPI 12708 only 1.55 µM LCA (Figures 4A and 4B), whereas for E. muris, no LCA was detected. The latter is in line with previous reports [22]. The amendment of 13C-CA significantly increased the transformation of CDCA for both C. scindens strains (p-value < 0.001 two-way ANOVA) but had no impact on E. muris SJ24 (Figure 4C). Indeed, the LCA yield increased to 9.77 µM for strain ATCC 35704 and to 40.4 µM for strain VPI 12708 (Figure 4A and 4B). No change was observed for E. muris SJ24.
For UDCA, 7a-dehydroxylation to LCA was observed only with amendment of 13C-CA (Figure 5). None of the strains exhibited any detectable level of activity from cultures that included only UDCA. 14.92 µM of LCA as well as extremely low amounts of 3-oxoLCA (with a maximum of 0.12 µM at 32 hours) were detected with C. scindens ATCC 35704 when 13C-CA was included. An unknown oxidized form labelled X-oxoUDCA was detected with a maximum concentration of 2.41 µM after 32 hours (Figure 5A). It is likely that this BA corresponds to 3-oxoUDCA (3-oxo-7β-hydroxy-5β-cholan-24-oic acid) as we can exclude 7-oxoLCA (the other product of oxidation of UDCA) (Figure 1). Another BA with the same ionized mass as UDCA was detected at a maximum concentration of 4.24 µM after 24 hours. We propose that this could be an isoform of UDCA with the hydroxyl group of the C3 carbon in the β conformation (3β,7β-dihydroxy-5β-cholan-24-oic acid). However, the identity of these compounds remains unconfirmed due to the lack of standards. In the presence of 13C-CA, the 7-DH-ing activity of C. scindens VPI 12708 was comparable to that of the ATCC strain in the presence of 13C-CA, with 13.61 µM of LCA and 0.68 µM of 3-oxoLCA after 48 hours (Figure 5B). X-oxoUDCA was also detected at very small concentrations around 0.5 µM from 12 hours until the end of the experiment. The potential isoform of UDCA was detected at up to 9.92 µM at the 24-hour time point. Following the same trend observed with the other primary BAs, E. muris SJ24 did not show any detectable activity with UDCA with or without 13C-CA. The chromatograms for the unknown bile acids and the standards used can be found in Supplementary Figure 2 and Supplementary Table 4 respectively.
As expected, neither C. scindens strain nor E. muris were capable of aMCA 7α-dehydroxylation in the absence of 13C-CA (Figure 6). C. scindens ATCC 35704 only produced minute amounts of an unknown oxo form of αMCA (labelled Y-oxoaMCA) (0.85 µM at 32 hours). In the C. scindens ATCC 35704 culture amended with 13C-CA, 6-oxoMDCA was detected at 2.7 µM after 48 hours (Figure 6A). This secondary bile acid has been 7a-DH-ed but also the hydroxyl at C6 oxidized. Moreover, several intermediates for which standards are unavailable were also detected after 48 hours. These were unknown oxidized forms of αMCA (labelled X- and Y- oxoαMCA) at concentrations not exceeding 5 µM each. A third unknown BA was detected (albeit at very low concentrations, 0.41 µM at 32 hours) with the same mass as 6-oxoMDCA, suggesting that it is an MCA species with one oxidation and one dehydroxylation. This would indicate the production of another 7a-dehydroxylated form of αMCA in vitro (Figure 1, Figure 6A). As for the ATCC 35704 strain, C. scindens VPI 12708 exhibited an increase in the quantity of products from αMCA transformation in the presence of 13C-CA relative to its absence (Figure 6B). This includes the 7-DH-ed BA 6-oxoMDCA that reached a concentration of 8.18 µM after 48 hours and the X- and Y- αMCA forms that were detected at maximum concentrations of 4.22 µM (4 hours) and 1.69 µM (32 hours), respectively. The aforementioned αMCA-derived bile acid with one ketone group and one dehydroxylation was also detected at a maximum concentration of 3.49 µM after 48 hours (Figure 6B). Surprisingly, E. muris SJ24 did not exhibit any observable 7-DH-ing activity with or without 13C-CA. Nevertheless, a small amount of X-oxoαMCA was detected at all time points, with a stable concentration at around 2.4 µM without and 1.9 µM with 13C-CA (Figure 6C). The results for βMCA were very similar to those for aMCA and are discussed in further detail in the supplementary information.
The concentration of 13C-CA was also measured over time to ascertain that CA was being metabolized. It was observed to decrease until it disappeared after 48 hours in the C. scindens strains except in the presence of CDCA, for which the concentration decreased slowly over time. We attribute this observation to the toxicity of CDCA at that concentration [20]. On the other hand, the concentration of 13C-CA in E. muris remained stable over time and in all conditions, confirming lack of transformation (Supplementary Figure 3).
bai gene expression in the presence of bile acids
To assess whether the amendment of 13C-CA to the culture resulted in bai operon expression as hypothesized, the relative expression of baiCD and of baiE were measured. In addition, the expression of baiJ (an accessory gene) was also monitored. Gene expression was normalized using at least three reference genes and was calculated relative to the expression levels in a control group without BAs. The E. muris strains (SJ24, DSM 28561) has a truncated baiJ gene [22] but we deemed it worth of investigation here. Additionally, baiO was also analysed for E. muris SJ24 as an alternative accessory bai gene [37].
Results show that the expression of bai operon genes in both C. scindens strains was highly upregulated in response to exposure to CA or to CDCA but not to the other BAs (Figure 7).
For C. scindens ATCC 35704, the three genes tested were highly upregulated when 13C-CA was present along with another BA (CDCA, UDCA, aMCA, or bMCA) (Figure 7A). In the CDCA dataset, statistically significant differences relative to the single BA condition were observed as all genes were slightly more upregulated in the presence of 13C-CA (p-value < 0.001 linear model), but baiCD was more so than the other genes (Figure 7A). Most interestingly, UDCA, aMCA, or bMCA did not activate the expression of bai genes on their own, consistent with the lack of 7-DH-ing activity with these BA substrates alone (Figure 7A).
A similar pattern was observed for C. scindens VPI 12708 but with the significant difference that baiJ was not upregulated under any conditions (Figure 7B) (p-value < 0.001 linear model). In the CDCA dataset, the addition of 13C-CA had an upregulatory effect if assessed with a paired Wilcoxon test (Supplementary Figure 4) but this effect was not found to be significant when using the linear statistical model displayed in Figure 7. Similar to the case of C. scindens ATCC 35704, 13C-CA amendment had a dramatic effect on the expression levels of baiCD and baiE in the presence of UDCA, αMCA, or βMCA, with upregulation reaching the expression levels observed with CA or CDCA alone (Figure 7B).
E. muris SJ24 showed a slight upregulation of baiCD and baiE in the CA dataset but it was not significant and did not occur in any of the other conditions (Figure 7C), consistent with its very poor 7-DH-ing activity in vitro (Figure 3). In fact, the increased baiCD and baiE gene expression ratio observed in the CA group was probably caused by a single biological replicate that had a higher expression level than the others.
Thus, CA had a large effect on bai expression, but in a strain-specific manner. Genes of the bai operon in the two C. scindens strains (ATCC 35704 and VPI 12708) exhibited a similar response to CA amendment but the accessory baiJ differed in its response. It was upregulated in strain ATCC 35704 but not in strain VPI 12708. In contrast, CA had no significant effect on the expression of any of the bai genes considered in E. muris SJ24.
The rhaS_1 gene (HDCHBGLK_01429) is immediately upstream of the bai operon promoter on the opposite strand and has been proposed as bile acid-regulatory A (barA) due to its potential implication in bai regulation [7]. The expression of rhaS1 and rhaS2 (a copy of rhaS1 elsewhere in the genome) was shown to have background levels in the presence of all BAs (Supplementary Figure 5). This was tested in C. scindens ATCC 35704 without the amendment of 13C-CA. Results indicate that rhaS is not upregulated by the presence of the BAs tested.
Thus, the question remains about the conditions propitious for bai gene expression and robust 7-DH-ion in E. muris SJ24. We hypothesized that other mouse-specific BAs may be key regulators.
bai gene regulation by other BAs in E. muris SJ24
Because the presence of 13C-CA did not affect bai gene expression in E. muris SJ24, we tested four BA cocktails to probe whether other BAs commonly found in the BA pool could promote bai expression. The BA pool was divided into four cocktails: tauro-conjugated BAs, oxidized BAs, sulfonated BAs, or ωMCA. The addition of these BAs mixtures to E. muris SJ24 did not yield the production of any detectable secondary BAs (Figure 8). A small CA concentration (<2 µM) was detected with the tauro-BA cocktail (Figure 8A) but this was likely the result of the presence of CA as an impurity in the TCA standard, as it was also detected at time 0. In the oxidized BA cocktail, 12-oxoCDCA was almost fully reduced to CA after 16 hours (Figure 8B). Small quantities of CDCA and βMCA were detected, while both are likely to be impurities from the standards used (detected at time 0), it is worth highlighting that the concentration of CDCA increased from an average of 2.74 µM (time 0) to 5.35 µM (time 24), meanwhile, the concentration of βMCA remained stable around 2 µM. No reduction of 3-oxo forms was detected, likely due to the absence of baiA2 [24]. Finally, neither sulfonated BAs nor ωMCA were transformed by E. muris SJ24 in any way (Figure 8C-D). Therefore, we considered unlikely that these other tested BAs could upregulate bai expression without being substrates for 7-DH-ion.
The expression of baiCD, baiE, the pseudogene baiJ and baiO was nonetheless measured in the BA cocktail experiments and compared with a CA-only reference group. Given the lack of 7-DH-ion of the BAs within the cocktails, it is not surprising that no significant upregulation was observed in any of the BA cocktail groups when compared to the CA control. (Supplementary Figure 6A).
Bile acid 7-DH-ion by E. muris SJ24 in the presence of mouse cecal content
CA 7-DH-ion by E. muris SJ24 was investigated in the presence of cecal content from either germ-free mice or a stable gnotobiotic murine model, Oligo-Mouse-Microbiota (Oligo-MM12) [38] in order to further investigate potential non-BA triggers for 7-DH-ion. A significant fraction of CA was conjugated with Co-enzyme A (CoA) and therefore could not be rigorously identified or quantified, as there are no standards for CoA- forms. In the controls (no cecal content), the DCA concentration averaged 4.26 µM after 48 hours which corresponded to the transformation of approximately 7% of the initial CA (Figure 9A). The amendment of cecal content from germ-free mice increased the DCA produced to 7.6 µM which corresponded to 12% of the initial CA (Figure 9B). Finally, the addition of cecal content from Oligo-MM12 mice produced only 0.67 µM of DCA but 18.6 µM of 7-oxoDCA (Figure 9C). In all conditions, DCA was detected after 12 hours of incubation and gradually increased. 12-oxoCDCA was detected in all conditions, while 3-oxoCA was only found in the no cecal content and germ-free groups (Figure 9). The control groups of cecal content without E. muris SJ24 showed no change in CA concentration other than the potential conjugation with CoA by the Oligo-MM12 mouse case (Supplementary Figure 7).
Despite the measurable impact on 7-DH-ion by the addition of germ-free mouse cecal content of CA 7-DH-ion, it was not sufficient to significantly upregulate bai expression when compared to the CA-only reference group (Supplementary Figure 6B). In both assays, the gene expression ratio of bai genes was never above 3.
E. muris SJ24 in vivo 7-DH-ion and bai gene expression
The ability of E. muris strain DSM 28560 (JM40) to 7-DH-ate in vivo has been previously documented [22]. Here, colonisation of Oligo-MM12 gnotobiotic mice was performed with the DSM 28561 strain (E. muris SJ24) to confirm 7-DH-ion in vivo and assess bai gene expression. The bile acid composition confirms active 7-DH-ion in vivo in Oligo-MM12 mice. Indeed, DCA, LCA and MDCA, were exclusively identified in the sDMDMm2 + E. muris SJ24 group (Supplementary Figure 8).
As above, expression of baiCD, baiE, the pseudogene baiJ and baiO was assessed, but no relative quantification was performed due to the absence of bai genes in the Oligo-MM12 control mice (not colonized with E. muris SJ24). Transcripts of all bai genes were detected in the Oligo-MM12 mice colonized with E. muris SJ24. We observed non-specific amplification of bai genes in the Oligo-MM12 control mice, but at a very low level compared to the E. muris-colonized Oligo-MM12 mice (Supplementary Table 2). Considering the evidence of secondary BAs produced in the colonized mice (Supplementary Figure 8), we conclude that E. muris expresses bai genes in vivo.