Experimental design for acetate-dependent transcriptome analysis
Acetate consumption is the main driver of butyrate production by genus Faecalibacterium, as shown in Fig. 1A. To gain insight into the effect of acetate on the transcriptome of F. duncaniae A2-165, two sets of cultures were established in media that contained low (3 mM) or high (23 mM) concentrations of acetate. The resulting growth kinetics are shown in Fig. 1B. In both acetate conditions, there was no difference in growth kinetics over the first 7 hours, and the stationary growth phase was reached after 9 hours of culture in both treatments. In the early stationary phase, the biomass in high-acetate conditions (Sa) was about 1.5-fold higher than in low-acetate conditions (S) (2.30 ± 0.13 OD600 vs. 1.50 ± 0.18 OD600, N = 4, p < 0.05). Moreover, the growth rate was 1.25-fold higher in high-acetate conditions (Fig. 1C), suggesting that the general metabolism of F. duncaniae A2-165 was highly active in these cultures. Indeed, the generation time was reduced by about 25 minutes in this group compared to that of low-acetate cultures (1.67 h vs. 2.08 h). In high- and low-acetate conditions, we quantified butyrate production and acetate consumption in both the late exponential growth phase (Ea and E, respectively) and the early stationary growth phase (Sa and S, respectively) (Fig. 1D). As expected, in both acetate conditions, butyrate production was significantly higher in the early stationary phase than in the late exponential phase. Moreover, in the late exponential phase (7 hours of growth) there was no difference between acetate conditions in either butyrate production or acetate consumption. In contrast, in the early stationary phase (10 hours of growth), butyrate production and acetate consumption were significantly different between acetate conditions (butyrate: 11.94 mM ± 0.89 for Sa, 6.37 mM ± 1.07 for S, p < 0.01). Moreover, after 10 hours of growth, almost all acetate had been consumed in the low-acetate cultures (-2.16 mM ± 0.38), whereas only one-third of the acetate had been consumed in the high-acetate cultures (-7.66 mM ± 0.58). The acetate limitation experienced in the low-acetate conditions strongly affected the growth of F. duncaniae A2-165 (Fig. 1BC). We thus investigated the acetate-growth effect at the transcriptional level in both acetate-limited and acetate-saturated conditions using RNA-Seq. For this, gene expression profiles were analyzed at both E/Ea and S/Sa time points (Fig. 1B).
Overview of transcriptomes in low- and high-acetate conditions
The RNA-Seq datasets contained a total of 2,882 genes out of the 3,013 predicted in the RefSeq database (i.e., over 95%). Of these, 1,161 genes were found to be differentially expressed (DE) (log2 fold change (FC) ≥ |2| and FDR-adjusted p-value ≤ 0.01) between treatment conditions (Table S1 in Additional file 2). All DE genes are presented in Additional file 2 (Tables S2–S5). Eight of these genes were differentially expressed in early exponential cultures, compared to 542 in the late stationary phase (340 and 202 transcripts up- and down-regulated, respectively, Fig. 2A), which is consistent with the different patterns of growth kinetics. This result clearly demonstrates that, under our experimental conditions, acetate concentration (3 vs. 23 mM) had little impact on transcriptomic profiles in F. duncaniae A2-165 for the first 7 hours of growth (i.e., up to the late exponential growth phase), but had a significant impact when the cultures entered the stationary phase of growth. This prompted us to perform a detailed analysis of the adaptive changes that occur in the transcriptome in response to acetate availability between 7 and 10 hours of growth. As shown in Fig. 2B, there was a large difference in the adaptive response between high-acetate (Sa/Ea, 118 DE genes) and low-acetate (S/E, 492 DE genes) conditions in this three-hour period. Specifically, we found that, under high-acetate conditions, only 67 and 51 genes were up- and downregulated, respectively, whereas under low-acetate conditions, 216 and 276 genes were up- and downregulated, respectively. When comparing the DE transcript lists in low- and high-acetate conditions (Sa/Ea compared to S/E, Fig. 2C), it clearly appeared that the acetate-limiting condition triggered a larger and more-specific adaptive transcriptional response at the onset of the stationary growth phase compared to the acetate-saturated conditions.
Using the COG and PATRIC databases, we classified DE genes into 12 functional categories (Table S6, Additional file 2). As shown in Fig. 3A, for all transcriptomes (high-acetate, low-acetate, and early stationary), the largest category (27.2–58.8% of genes) was the superclass “Poorly characterized protein-genes”, which included the categories “Hypothetical proteins” and “General function prediction”. It is likely that these unknown DE genes contribute to the adaptive responses of F. duncaniae A2-165 as cells enter the stationary phase. In addition, clear differences between low- and high-acetate transcriptomes were observed with respect to the categories “Protein synthesis”, “Energy metabolism”, “Import system”, “Defense system”, “Stress response”, and “Transcription and Post-transcriptional regulation” (Fig. 3A, Table S6). Interestingly, in acetate-limiting conditions (S vs. E), the highest proportions of up- and downregulated genes were related to “Stress response” (9.3%, i.e., 20 DE genes) and “Protein synthesis” (27.9%, i.e., 77 DE genes) respectively, whereas in acetate-saturated conditions (Sa vs. Ea), the highest proportions of affected genes were related to “Import system” (13.4%, i.e., 9 upregulated genes, and 23.5%, i.e., 12 downregulated genes).
Overall, our transcriptome analysis strongly suggests that F. duncaniae A2-165 is able to tightly regulate the expression of metabolic and stress-response genes according to acetate level. Remarkably, in all comparative transcriptional analyses, the CG447_03795 gene encoding the MAM protein did not appear among the DE genes; the transcription of this gene was unchanged between acetate conditions as well as between the growth phases examined here (Fig. S2 in Additional file 3).
To further characterize the acetate responses we observed, we investigated in more detail these functional categories: “Protein synthesis”, “Stress response”, “Transcription and Post-transcriptional regulation” and “Import system” (Fig. 3B).
A general stress response under acetate-limiting condition
As depicted in Fig. 3B, we clearly observed significant differences in the transcriptional responses of F. duncaniae A2-165 under acetate-limiting condition compared with acetate-saturated condition. Interestingly, in low-acetate condition, we found a high number of downregulated transcripts in the “Protein synthesis” category (77 DE genes, log2FC range: 2–5, Table S4) that were not detected in high-acetate condition. These were mainly transcripts involved in translation machinery, such as translation initiation/elongation factors and aminoacyl-tRNA synthetases/transferases, but also included genes involved in ribosome biogenesis and/or stability. These findings likely indicated a major slowdown of translational processes under low-acetate growth conditions as F. duncaniae cells entered the stationary phase.
We also noted a drastic shift in the “Stress response” category (20 DE genes, Table S4) which again was not observed in high-acetate condition. Seven of these genes (log2FC range: 2.2–2.9) encoded chaperone proteins involved in protein remodeling, in the repair of proteins following stress damage, or even in the ubiquitin machinery and proteasome. The remaining 13 (log2FC range: 2.1–6.2) genes were associated with type II toxin/antitoxin (TA) systems [28]. More specifically, we observed the upregulation of four putative operons for RelE TA systems, one putative operon for a TA module in the Doc family, two genes encoding antitoxin proteins, and one gene encoding a toxin protein. Of these, the DinJ/YafQ system (CG447_14090/14095) was the most activated TA system (log2FC: 6.1–6.2). Overall, our data suggest that cells of F. duncaniae A2-165 in low-acetate condition experienced severe general stress upon entry to the early stationary growth phase.
Bacteria employ different transcriptional regulators and sigma transcription factors to respond to changing environments. In the low-acetate transcriptome examined here, we found a higher number of upregulated genes encoding transcriptional regulators and sigma transcription factors (18 DE genes, log2FC range: 2.1–5.5, Table S4) compared to high-acetate condition (4 DE genes, log2FC range: 3.0–5.0, Table S5). Moreover, in all bacteria studied thus far, the global regulator that mediates the general stress response is a specialized sigma factor [29]. In the low-acetate transcriptome, we specifically found five upregulated genes encoding sigma factors (CG447_02225/05485/06440/06875/08160, Table S4), while in high-acetate conditions there was only one, which was also upregulated in low-acetate conditions (CG447_02215, Tables S4-S5). It is thus possible that these five sigma factors could be involved in the general stress response described above when cells enter the stationary phase.
Deciphering import systems under high- and low-acetate conditions
Transcription of transporter genes is usually regulated in response to substrate availability [30]. As shown in Fig. 3B, we clearly observed two distinct import system transcriptomes of F. duncaniae A2-165, under acetate-limiting and saturated conditions. These findings likely indicated two major adaptation responses to substrate availability, according to acetate conditions. A large number of genes belonging to the “Import system” category were found to be downregulated under low-acetate conditions (46 DE genes, Table S4); this number was four-fold higher than under high-acetate conditions. These were mainly transporter genes in the ABC family (25 DE genes) and metal ion transporter genes (9 DE genes) (Fig. 3B, Table S4). Within this group, we noticed a considerable reduction in gene expression related to ferrous iron uptake (feoAABC putative operon, CG447_12740-55, log2 FC range: -4.9–5.9, Fig. 3B), and ferric iron uptake (fhu putative operon, CG447_03300-10, log2FC range: -6.1–8.0). This suggested that the expression of major systems of iron transport was severely impaired in low-acetate conditions. Conversely, we found that expression of the feoAABC operon was upregulated under high-acetate condition (log2FC: 1.5–2.1, Table S5, Fig. 3B). This was particularly marked in early stationary transcriptome, in which we observed an even higher fold-change in expression (log2FC: 4.5–5.9, Table S3, Fig. 3B). Many studies have explored the role of the FeoB transporters in ferrous iron uptake in iron-poor environments [31, 32]. Our result suggests that in the early stationary phase in an acetate-saturated environment, ferrous iron is no longer abundant, and activation of the feoAABC system may allow F. duncaniae A2-165 cells to maintain iron homeostasis in these conditions.
Negative regulation of feoB, but not feoAB in the presence of ferrous sulfate in high-acetate conditions
In order to confirm our hypothesis of limited iron availability in the high-acetate culture conditions, we analyzed feoB expression in the presence of excess ferrous sulfate (Fig. 4). Figure 4A illustrates ferrous iron uptake and iron homeostasis in F. duncaniae. As a control, we also analyzed gene expression in low-acetate conditions, as well as the expression of feoAB (CG447_08795, log2FC: -4.3, Table S5, Fig. 3B) and butCoA, which encodes the terminal enzyme required for butyrate production (Fig. 4A). A new set of F. duncaniae A2-165 cultures was prepared with high or low levels of acetate and with or without ferrous sulfate, and sampling was performed as previously described (i.e., in the late exponential phase, indicated by E, and the early stationary phase, indicated by S, Fig. 4B). Figure 4B presents the results regarding the expression ratios of feoB, feoAB, and butCoA obtained under the different conditions. These clearly demonstrated that the feoB expression was strongly activated only in high-acetate condition. Moreover, its regulation was dependent on the availability of an iron source in the medium, with strong activation and repression in the absence and presence of ferrous sulfate, respectively. For the feoAB gene, the expression ratios were similar in the absence or presence of an iron source in both acetate growth conditions, indicating that the regulation of feoAB is independent of iron availability in our growth condition. For the butCoA gene, we observed a positive growth phase effect in both acetate and iron conditions with stronger activation in the high-acetate conditions (Fig. 4B). Overall, these results demonstrate that regulation of the expression of feoAABC and feoAB differs significantly and suggest that the FeoB and FeoAB systems of F. duncaniae A2-165 play nonredundant roles in ferrous iron acquisition.
FeoB peptides from F. duncaniae A2-165 detected in healthy human fecal samples
As our culture data suggested that feoB gene expression was tightly regulated, we were interested in studying its expression in vivo, in the human gut microbiome. For this, we analyzed a metaproteomic dataset obtained from eight healthy individuals, which included 123 425 peptides from the envelope fraction of the gut microbiota (Fig. 5) [33]. The metaproteomic data are presented in Additional file 4. Among these peptides, 236 matched with 10 of the 42 transporter genes that we had found to be upregulated in high-acetate conditions (Table S7 in additional file 2). Remarkably, the second highest degree of protein coverage (43.6%, 51 peptides) was found for the FeoB transporter. Of those 51 peptides, 9 were specific to a single protein in the metaproteomic dataset (‘a5.b59.a1’, sheet ‘CG447_12750 FeoB’ in Additional file 4), which was identified as FeoB of the A2-165 strain (K04759 in the KEGG orthology database). This demonstrated that the FeoB protein from A2-165 is expressed (i.e., transcribed and translated) in the human gut under healthy conditions. Furthermore, the fact that it was detected in four of the subjects at different levels, but not in the remaining four, suggests that feoB is also tightly regulated in vivo in the human gut.