Ethanolamine Utilization and Bacterial Microcompartment Formation Are Subject to Carbon Catabolite Repression

Ethanolamine (EA) is a compound commonly found in the gastrointestinal (GI) tract that can affect the behavior of human pathogens that can sense and utilize it, such as Enterococcus faecalis and Salmonella. Therefore, it is important to understand how the genes that govern EA utilization are regulated. In this work, we investigated two regulatory factors that control this process. One factor, a small RNA (sRNA), is shown to be important for generating the right levels of gene expression for maximum efficiency. The second factor, a transcriptional repressor, is important for preventing expression when other preferred sources of energy are available. Furthermore, a global bioinformatics analysis revealed that this second mechanism of transcriptional regulation likely operates on similar genes in related bacteria.

ABSTRACT

Ethanolamine (EA) is a compound prevalent in the gastrointestinal (GI) tract that can be used as a carbon, nitrogen, and/or energy source. Enterococcus faecalis, a GI commensal and opportunistic pathogen, contains approximately 20 ethanolamine utilization (eut) genes encoding the necessary regulatory, enzymatic, and structural proteins for this process. Here, using a chemically defined medium, two regulatory factors that affect EA utilization were examined. First, the functional consequences of loss of the small RNA (sRNA) EutX on the efficacy of EA utilization were investigated. One effect observed, as loss of this negative regulator causes an increase in eut gene expression, was a concomitant increase in the number of catabolic bacterial microcompartments (BMCs) formed. However, despite this increase, the growth of the strain was repressed, suggesting that the overall efficacy of EA utilization was negatively affected. Second, utilizing a deletion mutant and a complement, carbon catabolite control protein A (CcpA) was shown to be responsible for the repression of EA utilization in the presence of glucose. A predicted cre site in one of the three EA-inducible promoters, PeutS, was identified as the target of CcpA. However, CcpA was shown to affect the activation of all the promoters indirectly through the two-component system EutV and EutW, whose genes are under the control of the PeutS promoter. Moreover, a bioinformatics analysis of bacteria predicted to contain CcpA and cre sites revealed that a preponderance of BMC-containing operons are likely regulated by carbon catabolite repression (CCR).

IMPORTANCE Ethanolamine (EA) is a compound commonly found in the gastrointestinal (GI) tract that can affect the behavior of human pathogens that can sense and utilize it, such as Enterococcus faecalis and Salmonella. Therefore, it is important to understand how the genes that govern EA utilization are regulated. In this work, we investigated two regulatory factors that control this process. One factor, a small RNA (sRNA), is shown to be important for generating the right levels of gene expression for maximum efficiency. The second factor, a transcriptional repressor, is important for preventing expression when other preferred sources of energy are available. Furthermore, a global bioinformatics analysis revealed that this second mechanism of transcriptional regulation likely operates on similar genes in related bacteria.

INTRODUCTION

Organisms that inhabit the gastrointestinal (GI) tract exhibit metabolic diversity and flexibility, including the ability of some to catabolize ethanolamine (EA) (1, 2). EA is a highly accessible source of nitrogen, carbon and/or energy, as it is a breakdown product of phosphatidylethanolamine, an abundant phospholipid in both mammalian and bacterial cell membranes (3, 4). The host diet and the bacterial and epithelial cells in the intestine are rich sources of EA, resulting in gut luminal concentrations of up to 2 mM (5, 6). A number of bacteria that inhabit the GI tract and other environments, including but not limited to species of Salmonella, Enterococcus, Klebsiella, Mycobacterium, Clostridium, Escherichia, and Pseudomonas, can utilize EA (1, 2).

The eut locus of Enterococcus faecalis encodes approximately twenty enzymatic, structural, and regulatory components required for EA utilization (7, 8). For example, eutB and eutC encode the two subunits of the adenosylcobalamin (AdoCbl)-requiring ethanolamine ammonia lyase, EutBC. EutBC catalyzes the first step in EA utilization, yielding ammonia, which can serve as a cellular supply of reduced nitrogen, and acetaldehyde, which can be converted into the metabolically useful compound acetyl coenzyme A (acetyl-CoA). However, to prevent loss and damage from gaseous and toxic acetaldehyde, this reaction, as well as those downstream, takes place inside a proteinaceous organelle-like structure called a bacterial microcompartment (BMC) composed of the structural proteins encoded by eutM, eutS, eutK, eutL, and eutN (reviewed in references 2 and 9). Considering how metabolically expensive it would be to make all these components if EA were not available, it is perhaps not surprising that the regulation of eut gene expression is dependent on the detection of EA, which is accomplished by an EA-responsive two-component system (TCS) encoded by eutW and eutV. Upon EA detection, the sensor histidine kinase, EutW, autophosphorylates and phosphotransfers to the response regulator, EutV (8). A model depicting the current knowledge of how the eut genes are regulated and what will be contributed in this new work is shown in Fig. 1.

FIG 1.

Model depicting the regulation of eut gene expression. As previously established, the TCS, EutV and EutW, is activated by EA and positively regulates eut gene expression by preventing termination at the nascent transcripts generated by the eutP, eutG, and eutS promoters (shown in gray). However, the sRNA, EutX, also contains a EutV binding site (depicted as two gray loops) and will act as a “sponge,” preventing positive regulation by EutV. AdoCbl disrupts this negative regulation by binding to the riboswitch (red loop) when EutX is being transcribed, causing an early termination event resulting in a shortened form of EutX that can no longer bind EutV. By this manner, AdoCbl positively regulates eut gene expression. As shown in this current work and circled in red, glucose negatively regulates eut gene expression by activating CcpA, which binds the cre site within the eutS promoter, directly blocking transcription and preventing expression of the TCS. Expression at the eutP and eutG promoters is thereby indirectly blocked by CCR due to the lack of the TCS. +, positive effect on eut gene expression; −, negative effect.

Interestingly, the regulatory mechanisms thus far characterized for E. faecalis eut gene expression all occur at the postinitiation level. Phosphorylated and active EutV dimerizes and binds to nascent transcripts produced by constitutive promoters within the eut locus. The binding neutralizes terminators located 5′ to open reading frames (ORFs) that otherwise prevent gene expression (10). However, EA alone does not activate eut gene expression; AdoCbl is additionally required (7, 10). AdoCbl is sensed by a riboswitch aptamer located in a constitutively expressed small RNA (sRNA), EutX, which also contains a EutV binding site that sequesters the activated response regulator. To prevent EutV sequestration, AdoCbl binds the riboswitch aptamer, causing an early termination event that results in a shortened form of EutX that lacks the EutV binding site (Fig. 1) (11–15). We previously demonstrated that deletion of eutX results in eut gene expression dependent only on EA rather than EA and AdoCbl. Amazingly, normal-appearing BMCs were still formed, though EA could not be catabolized because the key enzyme EutBC requires AdoCbl as a cofactor (13). One goal of the work herein was to further examine the functional consequences of the loss of eutX on EA utilization and BMC formation. We hypothesized that loss of this negative regulator would increase the number of BMCs formed and thereby affect the efficiency of catabolism.

Ethanolamine utilization is repressed by glucose in E. faecalis as well as in Salmonella and Escherichia coli (7, 16–19). Recall that glucose commonly represses the use of alternative sources of carbon in Gram-positive bacteria by activating the catabolite control protein A, CcpA. CcpA binds catabolite responsive elements (cre) in the promoters of genes encoding alternative carbon catabolism pathways, blocking transcription initiation (20). Therefore, another goal of this study was to ascertain if CcpA was responsible for the observed repression by glucose and to identify the cre site(s) responsible. Finally, to elucidate how widespread carbon catabolite repression (CCR) is in regulating BMC-containing operons, a global bioinformatic search for cre sites in these loci was undertaken in the Firmicutes.

RESULTS

E. faecalis can utilize ethanolamine in chemically defined medium under aerobic conditions.

To study the factors that affect EA utilization, we first needed to establish rigorous medium conditions. Previous studies examining the role of EA usage on the growth of E. faecalis utilized a modified M9 minimal medium buffered with HEPES and supplemented with yeast extract (7). To ensure control over the content of our medium, particularly the source(s) of carbon, a strictly defined chemical medium (CDM) was designed and purchased (see Table S1 in the supplemental material) based on defined medium that was used previously for growth studies of other Gram-positive cocci (21–24). Because ethanolamine utilization requires the presence of the cofactor AdoCbl (7, 10), 40 μg/ml AdoCbl was added to the base CDM. Under aerobic conditions, achieved by growing the cultures in air with shaking, the reference strain E. faecalis OG1RF failed to grow when inoculated in CDM supplemented with 40 μg/ml AdoCbl (Fig. 2A, left graph). However, robust growth was achieved when 0.2% ribose, which is a moderate non-phosphotransferase system (non-PTS) sugar, was added (Fig. 2A, left graph). Interestingly, the reference strain grew to a significantly greater density when 33 mM EA was added to CDM in addition to 0.2% ribose (Fig. 2A, left graph). It was reasoned that the growth advantage was a result of successful utilization of EA as an energy source, because no significant growth was observed when 33 mM EA was added as the sole carbon source. While EA can be processed to generate ATP and acetyl-CoA, we speculated that bacteria that lack the tricarboxylic acid (TCA) cycle, such as E. faecalis, require a sugar like ribose that can enter the pentose phosphate pathway to generate biosynthetic intermediates (see Discussion) (25). Nearly identical growth curves were observed under microaerobic conditions (sealed plate, no air, and no shaking) compared to those under aerobic conditions (Fig. 2A, right graph), suggesting that oxygen conditions do not affect the utilization of EA in E. faecalis, unlike what was previously reported (7).

FIG 2.

E. faecalis utilizes EA and forms BMCs in CDM under aerobic conditions. (A) Graphical representation of the growth of E. faecalis OG1RF (WT) strain when grown in CDM. Growth curves were initiated at an OD₆₀₀ of ∼0.1, and cells were allowed to grow for 12 h in air with shaking (aerobic; left graph) (A) or covered with no air (microaerobic; right graph) in CDM containing either no carbon source, 33 mM EA, 0.2% ribose, or both 33 mM EA and 0.2% ribose. All cultures were supplemented with 40 μg/ml AdoCbl. (B) Representative transmission electron micrographs showing the formation of BMCs (arrows) in E. faecalis OG1RF (WT) strain. BMC formation was induced by addition of 33 mM EA (bottom panels) compared to the control condition where no EA was added (top panels). Left panels, aerobic conditions; right panels, microaerobic conditions.

In addition to measuring cell growth, cells from the reference strain under the aforementioned growth conditions were examined for BMC formation. Recall that it is within these proteinaceous structures that EA is catabolized (2, 26). To test for BMC formation, logarithmically growing cells, were collected and prepped for transmission electron microscopy (TEM). Under inducing conditions (CDM plus 0.2% ribose plus 40 μg/ml AdoCbl plus 33 mM EA), the formation of white opaque structures indicative of BMCs was observed within the cytoplasm. These structures were nonexistent in the absence of EA (Fig. 2B). Therefore, the TEM micrographs and the growth curves collectively indicate that E. faecalis is capable of forming ethanolamine utilization (EUT) BMCs and utilizing EA for metabolic activities when grown under these defined medium conditions.

Increased number of BMCs in an E. faecalis ΔeutX mutant negatively affects growth.

Once the defined culture conditions were established, we asked how certain regulatory and metabolic factors affected the efficiency of EA utilization in E. faecalis. As mentioned above, full-length EutX contains a EutV binding site that sequesters EutV-P generated by the EA-sensing kinase EutW. While the presence of AdoCbl causes a shorter form of EutX to be generated by a riboswitch-mediated transcriptional termination event, the termination is not 100% efficient (13). Since any remaining full-length EutX would dampen EutV activity, it was predicted and observed that a complete deletion of eutX results in overexpression of the eut genes (13).

To better understand the effects of the loss of EutX on the efficiency of EA catabolism, we examined both BMC formation and growth. TEM micrographs indicated the formation of structurally normal BMCs within the cytoplasm of the ΔeutX mutant cells grown in CDM under aerobic conditions (Fig. 3A). However, quantification of the BMCs showed that while there were 1.73 ± 0.172 BMCs/cell in the reference strain, the ΔeutX mutant cells displayed BMC counts of 5.93 ± 0.462 per cell (Fig. 3B). To test if this increase in the number of BMCs resulted in more efficient use of EA as an energy source, the reference strain and ΔeutX and ΔeutV mutants were grown with or without EA, with the ΔeutV mutant serving as a control for EA utilization. As expected, the reference strain showed successful utilization of EA, resulting in a greater cell culture density when supplemented with EA than without EA, while the ΔeutV mutant showed no differences in growth characteristics with or without EA (Fig. 3C). The ΔeutX mutant also grew to greater cell culture density upon addition of EA versus without EA, but somewhat surprisingly, the level of growth was lower than that of the reference (Fig. 3C). The growth defect was complemented by genetic insertion of eutX back into the deletion strains (see Fig. S1). Therefore, our results point to an inverse correlation between the number of BMCs per cell and the efficacy of EA utilization for growth.

FIG 3.

Bacterial microcompartment numbers and their effect on the growth of E. faecalis mutants. (A) Representative transmission electron micrographs displaying the formation of BMCs (arrows) in E. faecalis strains OG1RF (WT) and the ΔeutX mutant when induced with 33 mM EA (bottom) versus no EA control condition (top). (B) Graphical representation of the number of BMCs per cell calculated for E. faecalis OG1RF (WT) and ΔeutX mutant grown under inducing conditions. BMCs were counted in 30 cells per strain. An unpaired t test with Welch’s correction for the number of BMCs per cell observed in the ΔeutX mutant versus that in the WT was used to calculate the P values. (C) Graphical plot of the growth of E. faecalis strains OG1RF (WT) and ΔeutV and ΔeutX mutants with or without 33 mM EA. All the cultures were supplemented with 0.2% ribose and 40 μg/ml AdoCbl and were grown under aerobic conditions for 12 h.

EA utilization is regulated by CcpA acting through a cre site in the promoter of eutS.

Previous studies showed that eut gene expression induced by EA and AdoCbl is repressed in the presence of glucose (7, 19). We predicted that glucose would therefore also repress the use of EA as an energy source, which was tested by adding 0.2% glucose to CDM plus AdoCbl, with or without EA. The OG1RF reference strain grew more robustly in the presence of glucose than when the same concentration of ribose was used, with the culture entering into the logarithmic phase earlier and achieving higher cell densities. However, no growth enhancement was observed when 33 mM EA was added, unlike the addition of EA to ribose-containing CDM (Fig. 4). To examine the effects of glucose on eut gene expression, we employed quantitative reverse transcription-PCR (qRT-PCR) to measure gene expression of eutP, eutG, and eutS. Each of these genes is preceded by a promoter subject to EutV regulation and requires EA to be induced (7, 8, 10, 19) (Fig. 1). In the reference strain background, indicated as wild type, we found that the expression of these genes, normally induced by EA in ribose-containing medium, was not induced in glucose-containing medium (Fig. 4). These observations were consistent with the previously reported effects of glucose on eut gene expression and are consistent with CCR.

FIG 4.

Glucose-associated carbon catabolite repression of ethanolamine utilization in E. faecalis. Graphical representation of E. faecalis OG1RF (WT) strain growth characteristics when grown in CDM under aerobic conditions for 12 h. Cultures utilizing 0.2% ribose as the carbon source with or without the addition of 33 mM EA were compared to cultures containing 0.2% glucose as an alternative carbon source, similarly supplemented with 33 mM EA; 40 μg/ml AdoCbl was added to all the cultures.

Like many other Gram-positive bacteria, E. faecalis harbors CcpA, which has been shown to affect the expression of certain genes in this bacterium, including ace (adhesin of collagen from E. faecalis) (27, 28). Using a ΔccpA deletion mutant (27), we examined the effects of glucose on eut gene expression. Unlike the reference strain background, the eut genes were expressed in medium containing glucose and EA in a ΔccpA background as measured by qRT-PCR of eutP, eutG, and eutS (Fig. 5). In fact, the levels of gene expression in the ΔccpA background were higher than in the induced wild-type strain, suggesting that CcpA may still exert some inhibitory effect in the EA-plus-ribose condition. A newly generated complement strain in which ccpA is expressed from a heterologous locus, ΔccpA::ccpA, mostly repressed the eut gene induction in the glucose-plus-EA condition. We attributed the lack of complete complementation to less expression of ccpA in the heterologous locus, which was confirmed by measuring ccpA expression by qRT-PCR (see Fig. S2). These results indicate that CcpA mediates CCR repression of the eut locus.

FIG 5.

CcpA negatively regulates expression of the eut locus in E. faecalis. qRT-PCR analysis of transcript levels of eutP, eutG, and eutS relative to that of the control, gyrA, for the E. faecalis strains OG1RF (WT), the ΔccpA mutant, and the ΔccpA::ccpA complement. Cultures for each strain grown using 0.2% ribose, with or without the addition of 33 mM EA, were compared to those cultured with 0.2% glucose with or without 33 mM EA.

Next, we wanted to understand where in the eut locus CcpA was binding to regulate gene expression. Possible cre sites had previously been compiled for E. faecalis by searching the genome using the consensus sequence developed for Bacillus subtilis (29). One such site was found in the promoter of eutS (GATGAAATCGATAACAT) (Fig. 6A). The site sits between the −35 and −10 sequences, and CcpA binding in this region is predicted to prevent RNA polymerase binding. Using site-directed mutagenesis, we altered this site in order to add a restriction site (to aid in screening for the change) and disrupt the predicted cre without altering the −35, the −10, or the spacing of these sequences to generate a PeutS::cre* mutation (Fig. 6A). In this background, adding glucose to the medium containing EA failed to repress the eut genes in contrast to that in the reference strain (Fig. 6B). As modeled in Fig. 1, these data support the assertion that CcpA regulates eut gene expression via this cre site in the eutS promoter. It is noteworthy that the eutS promoter controls an operon that includes eutV and eutW, which encode the EA-sensing TCS necessary to activate gene expression. By directly controlling this promoter, CcpA is able to indirectly affect the expression of all the EutV-regulated promoters in the eut locus (Fig. 1).

FIG 6.

The cre site substitution and its influence on eut gene expression in E. faecalis. (A) Illustration showing the intergenic region between eutG and eutS, bearing the modified cre site (AATCGA was substituted for CCATGG to bring in an NcoI restriction site). The EutV regulatory region corresponds to the terminator loops in the 5′ untranslated region (UTR) region of the nascent eutS transcripts to which EutV binds to facilitates antitermination. SD, ribosomal binding site upstream of eutS open reading frame. (B) Transcript levels of eutP, eutG, and eutS normalized to the control, gyrA, for the E. faecalis strains OG1RF (WT) and the P_eutS::cre* mutant. Each strain was grown aerobically in 0.2% ribose with or without 33 mM EA, and its transcript levels were compared to those grown in 0.2% glucose with or without 33 mM EA.

cre site regulation of BMC-encoding loci are common in the firmicutes.

Finally, we extended the analysis of CCR by CcpA to additional microcompartment loci and bacterial species. In addition to cre site sequences from B. subtilis, we utilized cre site motifs compiled from Streptococcus suis and Peptoclostridium difficile (30, 31). We searched the genomes of a broad selection of previously identified bacterial species containing BMC loci (32) to identify nonoverlapping cre site motifs found in the intergenic regions of each BMC locus. Firmicutes contained 1.57 putative cre sites per 1 kb of intergenic sequence in BMC loci. Non CcpA-containing phyla with many BMC loci, Cyanobacteria and Proteobacteria, contain substantially fewer cre motif sequences, 0.9 and 0.56 per 1 kb of intergenic sequence, respectively, and non-Firmicutes BMC loci contain 0.73 per 1 kb (Fig. 7A). Comparing by type of BMC, all, except for those that were incomplete or encode ethanol utilization, contain a preponderance of cre sites (Fig. 7B).

FIG 7.

Identification of cre motifs in bacterial microcompartment loci. (A) Frequencies of cre motifs found in intergenic regions of BMC loci in Firmicutes (CcpA containing) and non-Firmicutes. (B) Frequencies of cre motifs found in intergenic regions of different microcompartment subtype loci in Firmicutes genomes. Unknown, metabolosome of unknown function; PDU, 1,2-propanediol utilization; PVM, Planctomycetes and Verrucomicrobia type; BUF, BMC of unknown function; GRM, glycyl radical enzyme containing; EUT, ethanolamine utilization; MIC, incomplete core locus; ETU, ethanol utilization. Schematics of eut or pdu-eut loci from four bacterial genomes as identified in reference 32. (C) Enterococcus faecalis OG1RF. (D) Listeria monocytogenes EGD-e. (E) Streptococcus sanguinis SK36. (F) Peptoclostridium difficile 630. Genes are colored by association with different components of the BMC loci. Green, eut-associated genes; blue, pdu-associated genes; brown, cobalamin biosynthesis or transport genes; orange, regulatory genes; gray, other/unrelated genes. Regulatory RNAs, eutX and aspocR, are show in light green and light orange, respectively. Only putative promoter sequences associated with cre motif sequences are identified by directional arrows.

We selected four representative Firmicutes genomes containing eut or pdu-eut gene clusters (1,2-propanediol and ethanolamine utilization hybrid clusters) to manually identify potential CcpA-regulated promoters by identifying vegetative promoter sequences overlapping or adjacent to putative CcpA binding sites (Fig. 7C to F). These representative genomes include E. faecalis OG1RF (eut) (Fig. 7C), Listeria monocytogenes EGD-e (pdu-eut) (Fig. 7D), Streptococcus sanguinis SK36 (pdu-eut) (Fig. 7D), and Peptoclostridium difficile 630 (eut) (Fig. 7F). A putative CcpA-regulated promoter was identified upstream potential transcripts containing eutVW in all four eut loci. In pdu-eut hybrid loci, potential CcpA-regulated promoters were also found controlling either a regulator of pdu, pocR (Fig. 7D), or the primary pdu transcript (Fig. 7E). In several cases, other nearby transcripts containing cobalamin biosynthesis genes are found downstream of cre site-associated promoters. Thus, CCR of BMC-containing loci appears to commonly occur. Moreover, it often manifests via CcpA/cre site transcriptional control of promoters proceeding relevant regulator-encoding genes.

DISCUSSION

While a previous study suggested that E. faecalis can only utilize EA under anaerobic conditions (7), the current work shows that this catabolism also occurs under aerobic conditions. The data conform with the current paradigm that EA catabolism is not an inherently anaerobic process and are consistent with previous findings that other facultative anaerobes, such as E. coli and Salmonella enterica serovar Typhimurium, catabolize EA under aerobic conditions (reviewed in reference 33). The chemically defined minimal medium utilized contains purines, pyrimidines, essential vitamins, and amino acids (34, 35). While previous work defining the minimal conditions necessary for E. faecalis growth showed that only a subset of amino acids are absolutely required, better growth is achieved when additional amino acids are provided (35). Because of this requirement for amino acids, it is not possible to test E. faecalis’s ability to use EA as a sole source of nitrogen. Though previous work claimed that E. faecalis can grow on EA as a sole source of carbon, the medium used contained a small amount of yeast extract, providing a potential carbon source (7). We found that E. faecalis did not grow on EA alone in CDM (Fig. 2). Because E. faecalis lacks the TCA cycle, biosynthetic intermediates cannot be generated from a two-carbon compound such as EA. Ribose, however, is a five-carbon compound capable of entering the pentose phosphate pathway and generating intermediates (25). The addition of EA to CDM containing ribose provided a significant boost to growth that we surmise is from additional ATP being generated by EA catabolism (Fig. 2). The carbon-containing products predicted to form from this catabolism include ethanol and/or acetyl-CoA (2). Without a TCA cycle to convert it into pyruvate, acetyl-CoA is predicted to be fermented into acetate, generating ATP in the process (25, 33). However, confirmation of the catabolism end products of EA utilization in E. faecalis will require carbon labeling or metabolomic experiments.

In previous work, it was shown that the EutX sRNA negatively regulates EA utilization in E. faecalis by sequestering the activated antiterminator protein EutV in a process relieved by the presence of AdoCbl. When eutX is deleted, eut gene expression is no longer dependent on AdoCbl, and levels are significantly increased due to loss of this negative regulator (15). In this work, we investigated the physiological implications and discovered that there was increased BMC formation in a ΔeutX mutant (Fig. 3B). However, having more BMCs to catabolize more EA did not correlate with better growth (Fig. 3C). We speculate that the negative effects of misregulation of BMC development on cell growth could be caused by the following processes. Because expression and translation of close to twenty genes to form BMCs are energy intensive processes, there could be a tipping point at which generating more BMCs requires more energy than is obtained from the additional capacity to catabolize EA. Alternatively, or in addition, the increased numbers of BMCs could result in more production of toxic products such as acetaldehyde and/or acetate that have negative effects on the cells.

In addition to EutX, we studied another negative regulator of eut gene expression, CcpA. While EutX functions to ensure that the eut genes are not induced when AdoCbl is absent, CcpA prevents expression when PTS sugars such as glucose are present. Generally, in Gram-positive bacteria, CcpA functions to curtail the use of nonpreferred energy and carbon sources when more energy-rich alternatives are present. High-energy conditions are sensed by the presence of fructose-1,6-phosphate, a product of glycolysis, activating HprK which phosphorylates Hpr on a serine residue. Hpr(Ser-P) then binds CcpA, licensing the protein complex to bind cre sites and activate or repress transcription (Fig. 1). Whether the CcpA complex is activating or repressing often depends on where the cre site is located within the promoter (36). The eutS cre site, located in the middle of the −35 and −10 sites, would be predicted to be repressing, as was found in this work (Fig. 6A). Elimination of the cre site or CcpA derepressed eut gene expression under glucose-replete conditions (Fig. 5 and 6B). The fact that the cre site lies within the eutS promoter upstream of the genes encoding the EA-sensing TCS enables both direct and indirect control as modeled in Fig. 1. Due to the cre site, there is direct control at the level of transcription initiation at the eutS promoter that affects the expression of eutVW encoding the TCS. The intracellular concentration of the TCS provides a mechanism of indirect control at the postinitiation level by affecting the frequency of antitermination events occurring in the nascent transcripts downstream of the eutP, eutG, and eutS promoters, all of which contain terminators. By this mechanism, it is proposed that the single cre site in the eutS promoter can control expression of all the eut genes in this complex multipromoter locus.

In addition to the CcpA-regulated cre site found to control eutVW expression in E. faecalis, sequence analysis of several other eut microcompartment loci in L. monocytogenes, S. sanguinis, and P. difficile revealed potential cre sites associated with promoter sequences controlling analogous transcripts in each of the eut clusters. Although the exact gene compositions of the putative transcripts differ, in all cases, at least eutVW are likely to be repressed by CcpA binding, allowing similar control of the full eut gene cluster by catabolite repression. Two of these gene clusters include not just a eut pathway but a set of 1,2-propanediol utilization genes as well. In both cases, putative CcpA-controlled promoters were identified either directly controlling pdu genes or controlling a regulator of pdu gene expression (pocR), indicating that catabolite repression of metabolome gene clusters may be a general feature. In a catalog of BMC loci, cre motif sequences were identified at a higher frequency in CcpA-containing Firmicutes, and most types of microcompartment loci were associated with a high level of cre motif occurrences in intergenic spaces, again suggesting widespread catabolite repression of these metabolic pathways in these species.

In conclusion, this study establishes that CcpA regulates eut gene expression in E. faecalis by binding a specific cre site within the promoter that generates the transcripts for the EA-sensing TCS. Moreover, bioinformatics analysis strongly suggests that regulation by CcpA of BMC-containing loci is widespread in the Firmicutes. Likely, CCR of BMC loci is a general occurrence in bacteria, but since not all bacteria use CcpA to carry out CCR, establishment of this fact would require study of species that utilize different mechanisms of CCR.

MATERIALS AND METHODS

Bacterial strains and media.

All the bacterial strains used in this study are listed in Table S2 in the supplemental material. Sigma and Thermo Fisher Scientific were used for the procurement of chemicals and reagents. All growth media, with the exception of the CDM, were sourced from either BD Difco or BD Bacto. The CDM was a custom order from Alpha Biosciences, bearing the catalog number C03-201 and lacking cyanocobalamin. E. coli strains were cultured in Luria-Bertani (LB) broth at 37°C with shaking on a Thermo Scientific culture rotator at a 45° angle. Erythromycin (300 μg/ml), spectinomycin (100 μg/ml), and chloramphenicol (15 μg/ml) were added when necessary. E. faecalis strains were cultured similarly to that described above in CDM. Chloramphenicol (15 μg/ml), rifampin (100 μg/ml), and erythromycin (50 μg/ml) were used for selection.

Primers, sequencing, and DNA manipulations.

All primers used in this study are listed in Table S2. Primers were procured from Integrated DNA Technologies (Coralville, IA) and Sigma-Aldrich (St. Louis, MO). Phusion High-Fidelity DNA polymerase (NEB, Ipswich, MA) was used to perform all PCRs as per the manufacturer’s instructions. Restriction enzymes for restriction-based cloning were purchased from NEB, while T4 DNA ligase was purchased from Promega (Madison, WI). All restriction-free clonings (Gibson assemblies) were performed using NEBuilder HiFi DNA Assembly master mix (NEB). Plasmid extractions and PCR purifications were carried out utilizing kits from Qiagen (Hilden, Germany), as per the recommendations of the manufacturer. The sequencing of all constructs was sourced to Genewiz (South Plainfield, NJ).

Plasmid construction and allelic exchange.

All plasmids used in this study are listed in Table S2. To construct plasmid pKK6, primers KK12/KK13 and primers KK16/KK17 were employed to amplify genes ef2238 and ef2239, respectively, from E. faecalis OG1RF genomic DNA. The multiple-cloning site (MCS) from pCJK47 was amplified using primers KK14/KK15. The three fragments were spliced together by overlap extension PCR, with the MCS lying in between the ef2238 and ef2239 fragments, using primers KK18/KK19. pCR2.1 was used as the vector into which the spliced fragment was inserted using TA cloning and then transformed into E. coli TOP10 to give the plasmid, pKK5. Restriction enzymes XbaI and NcoI were used to subclone the fragment into similarly digested pCJK47. The ligated product was then electroporated into E. coli EC1000 resulting in plasmid pKK6. To construct plasmid pKK31, ccpA, including its native promoter, was amplified from E. faecalis OG1RF genomic DNA using primers KK116/KK117 and then inserted into PCR linearized pKK6 (primers KK114/KK115) using the NEBuilder HiFi DNA Assembly kit as per the manufacture’s instructions. The assembled plasmid was then transformed E. coli EC1000 using electroporation. Plasmid pKK36 utilizes a clustered regularly interspaced short palindromic repeat (CRISPR)-Cas gene-editing system and was constructed by first generating four separate DNA fragments. The first fragment, which was amplified from E. faecalis OG1RF genomic DNA using primers KK202/KK217, brought in the modified cre site, including a 800-bp stretch directly upstream of it. The second fragment brought in the same modified cre site along with an 800-bp section directly downstream of it using primers KK216/KK205. The third fragment, constructed with primers KK201/KK206 using pGR-ermB as a template, brought in the cre site spacer sequence. The fourth fragment was constructed in a manner similar to the third fragment, however, utilizing primers KK200/KK207 instead. The four DNA fragments were then spliced together using Gibson assembly and transformed into chemically competent E. coli TOP10. To construct plasmid pKK40, eutX as well as its upstream and downstream flanking regions were amplified as two separate fragments using the primer sets KK244/KK245 and KK246/KK247, respectively. Primers KK245 and KK246 brought in a T-to-C substitution upstream of the region coding for the adenosylcobalamin riboswitch of EutX, to help distinguish the ΔeutX complementation strain from the wild type. The upstream and downstream fragments were spliced with the PCR-linearized pCJK47 plasmid (KK242/KK243) using Gibson assembly and cloned into E. coli as described above.

Plasmids pKK31, pKK36, and pKK40 were then transformed into the conjugative E. faecalis strain CK111 by electroporation, resulting in strains EFKK27, EFKK33, and EFKK35, respectively. Allelic exchange was achieved using one of two approaches. The first approach was performed as described by Kristich et al. (37) after conjugation of strain EFKK27 with the ΔccpA mutant and EFKK35 with SD289, resulting in strains EFKK29 and EFKK36, respectively. After conjugation of strain EFKK33 with OG117, the second approach, which is based on a CRISPR-Cas system, was performed as described by Hullahalli et al. (38), resulting in strain EFKK34.

Growth curves.

E. faecalis strains were initially cultured overnight in 5 ml of brain heart infusion (BHI) broth. The bacterial cells, however, were prewashed the following day in CDM lacking EA to remove any contaminating carbon and nitrogen sources carried over from the overnight cultures before inoculating the CDM. The cultures were started in 96-well plates from Falcon (Corning, NY) at an optical density at 600 nm (OD₆₀₀) of 0.05, with incubation at 37°C and shaking. The plates were sealed using the Breathe-Easy sealing membrane (Diversified Biotech, Dedham, MA), achieving aerobic or anaerobic growth by either removing or leaving on the top carrier sheet of the sealing membrane sandwich, respectively. Growth was monitored using the Cytation 5 plate reader system (BioTek Instruments, Inc., VT), with readings taken every 10 min until the point where the bacterial cells entered the stationary phase.

Transmission electron microscopy.

E. faecalis strains OG1RF (WT) and SD289 (ΔeutX) were cultured either in the presence or absence of AdoCbl and EA to induce or not the expression of the eut genes. The cultures were then subsequently stained and thin sectioned for transmission electron microscopy as described in DebRoy et al. (13). Image acquisition was performed using the JEOL 1200 TEM system, equipped with the Gatan 2k by 2k charge-coupled device (CCD) camera. For quantification, the numbers of BMCs in 30 imaged cells from each strain were counted and averaged. An unpaired t test with Welch’s correction for the number of BMCs/cell observed was used to calculate the P values.

Preparation of RNA and quantitative reverse-transcription PCR.

E. faecalis strains OG1RF (wild type [WT]), the ΔccpA mutant, EFKK29 (ΔccpA::ccpA), and EFKK34 (P_eutS::cre*) were cultured aerobically as mentioned previously, with either 0.2% ribose or 0.2% glucose, with or without the addition of 33 mM EA in CDM. Samples were collected for RNA isolation at the mid-log phase of growth. The RiboPure kit (Invitrogen, Carlsbad, CA) was used to isolate RNA as per the manufacturer’s instructions. Any contaminating DNA was removed by DNase I treatment of the RNA samples using the Turbo DNA-free kit (Invitrogen, Carlsbad, CA). qRT-PCR was performed on a CFX96 Touch real-time PCR detection system (Bio-Rad, Hercules, CA) using the Power SYBR green RNA-to-C_T 1-step kit (Applied Biosystems, Foster City, CA), following quantification of the RNA. The fold changes of transcript levels relative to that of gyrA were determined using the comparative threshold cycle (C_T) method. The primers used for amplification of eutP (MRC94/MRC95), eutG (MRC88/MRC89), eutS (MRC90/MRC91) and gyrA (MRC86/MRC87) are listed in Table S2.

Identification of CcpA binding sequences (cre sites) in bacterial microcompartment loci.

We generated motifs for MEME 5.0.2 (39) using the B. subtilis high- and low-affinity cre sequences (40), the pseudopalindromic cre sequences from Streptococcus suis (31), and cre sequences identified in Peptoclostridium difficile (30). For each genome containing a bacterial microcompartment locus (32), we identified potential CcpA binding sites using FIMO (41). We merged overlapping motif sequences from multiple cre variant motifs into a single putative CcpA binding site. Frequencies of cre motif appearance in microcompartment loci as previously identified in reference 32 were generated by identifying motif sequences found entirely in intergenic spaces in the BMC locus and normalizing to the total amount of intergenic sequence found in microcompartment loci for each subdivision. Candidate CcpA-regulated promoters in Enterococcus faecalis OG1RF, Listeria monocytogenes EGD-e, Streptococcus sanguinis SK36, and Peptoclostridium difficile 630 were identified by analyzing 200 bp of sequence surrounding each candidate intergenic cre motif using BPROM to identify potential vegetative promoters (42).

Statistical analysis.

Statistical analyses were conducted using Prism, v7.0a (GraphPad Software, Inc., San Diego, CA). P values were calculated using an unpaired t test with Welch’s correction for the number of BMCs per cell observed in the ΔeutX mutant versus that in the WT.