Carbohydrate Utilization in Bacteria: Making the Most Out of Sugars with the Help of Small Regulatory RNAs

ABSTRACT

Survival of bacteria in ever-changing habitats with fluctuating nutrient supplies requires rapid adaptation of their metabolic capabilities. To this end, carbohydrate metabolism is governed by complex regulatory networks including posttranscriptional mechanisms that involve small regulatory RNAs (sRNAs) and RNA-binding proteins. sRNAs limit the response to substrate availability and set the threshold or time required for induction and repression of carbohydrate utilization systems. Carbon catabolite repression (CCR) also involves sRNAs. In Enterobacteriaceae, sRNA Spot 42 cooperates with the transcriptional regulator cyclic AMP (cAMP)-receptor protein (CRP) to repress secondary carbohydrate utilization genes when a preferred sugar is consumed. In pseudomonads, CCR operates entirely at the posttranscriptional level, involving RNA-binding protein Hfq and decoy sRNA CrcZ. Moreover, sRNAs coordinate fluxes through central carbohydrate metabolic pathways with carbohydrate availability. In Gram-negative bacteria, the interplay between RNA-binding protein CsrA and its cognate sRNAs regulates glycolysis and gluconeogenesis in response to signals derived from metabolism. Spot 42 and cAMP-CRP jointly downregulate tricarboxylic acid cycle activity when glycolytic carbon sources are ample. In addition, bacteria use sRNAs to reprogram carbohydrate metabolism in response to anaerobiosis and iron limitation. Finally, sRNAs also provide homeostasis of essential anabolic pathways, as exemplified by the hexosamine pathway providing cell envelope precursors. In this review, we discuss the manifold roles of bacterial sRNAs in regulation of carbon source uptake and utilization, substrate prioritization, and metabolism.

*These authors contributed equally to the manuscript.

INTRODUCTION

Carbohydrates are degraded in central metabolic pathways, namely, glycolysis, the pentose phosphate pathway, and the tricarboxylic acid (TCA) cycle, to fuel cells with energy and building blocks to synthesize all biomolecules. A functional carbohydrate metabolism requires sufficient supply with carbon sources but also coordination with the availability of other nutrients and cellular activities. Hence, bacterial carbohydrate metabolism is controlled at all levels by large and densely interconnected regulatory networks (1). In recent years, posttranscriptional mechanisms involving small regulatory RNAs (sRNAs) have emerged as an additional layer in these networks. Extensive cross talk of sRNAs with transcriptional regulators ensures a fine-tuned and coordinated metabolism.

Bacterial sRNAs come in two flavors. (i) cis-Encoded sRNAs are transcribed from the opposite strand of their target genes. Due to their perfect complementarity, they form extensive RNA duplexes with their target transcripts, influencing transcription, translation, or degradation of the target (2). (ii) trans-Encoded sRNAs regulate distantly encoded targets that can be either RNA or protein. They regulate translation or RNA stability, either negatively or positively, through imperfect base-pairing (3). In addition, modulation of transcription termination by sRNAs has also been observed (4). In Gram-negative bacteria, trans-encoded sRNAs often require protein Hfq for protection from degradation and RNA duplex formation (5, 6). The activities of sRNAs are tightly controlled, at the level of either biogenesis or their decay (7–9). A recently emerging mechanism is decoy and sponge RNAs that are capable of sequestering sRNAs by base-pairing (10).

Here, we review the manifold roles of sRNAs in regulation of carbohydrate metabolism. The outline of the review is illustrated in Fig. 1. First, specialized sRNAs that regulate consumption of particular carbohydrates are described (Fig. 1A). Bacteria residing in mixed environments often select the energetically most favorable carbon source. The regulatory contribution of sRNAs to substrate prioritization will be discussed subsequently (Fig. 1B). Next, we will dissect how transcriptional and posttranscriptional mechanisms cooperate to coordinate metabolic activities with carbohydrate availability and other cues such as iron and oxygen availability (Fig. 1C). Finally, we will discuss the amino sugar pathway generating precursors for cell envelope synthesis as an exemplary anabolic pathway regulated by sRNAs (Fig. 1D). To date, sRNAs are most thoroughly investigated in the Gram-negative model bacteria Escherichia coli and Salmonella, but knowledge from unrelated species has increased and is incorporated.

FIGURE 1.

Manifold roles of sRNAs in regulation of carbohydrate metabolism in bacteria. The figure summarizes the major roles of sRNAs (depicted in red) in regulation of carbohydrate metabolism in bacteria. (A) Regulation of uptake and utilization of particular carbohydrates by sRNAs in various species. In Enterobacteriaceae, the trans-encoded sRNA SgrS counteracts phosphosugar stress through repression of glucose transporters and activation of the sugar phosphatase YigL. sRNA ChiX downregulates the chitosugar-specific porin ChiP, setting the threshold concentration for induction of degrading enzymes. Further examples include regulation of host glycan and mannitol uptake by cis-encoded sRNAs in Bacteroides and Vibrio species, respectively. (B) Role of sRNAs in CCR. In Enterobacteriaceae and Vibrionaceae, the sRNA Spot 42 represses genes for utilization of secondary carbon sources. Spot 42 is repressed by cAMP-CRP and therefore only active in the presence of preferred sugars generating low cAMP levels. In Pseudomonas, translation of mRNAs for utilization of secondary carbon sources is repressed by Hfq. In the absence of preferred substrates, the CbrA/CbrB TCS activates expression of the decoy sRNA CrcZ, titrating Hfq from target transcripts. (C) sRNAs coordinate carbohydrate metabolism with carbohydrate, oxygen, and iron availability. The RNA-binding protein CsrA activates glycolysis and represses gluconeogenesis by binding to corresponding RNAs. CsrA activity is counteracted through sequestration by sRNAs CsrB/CsrC, whose levels are regulated by signals from metabolism. In the absence of oxygen, sRNAs such as FnrS in E. coli and RoxS in B. subtilis redirect metabolism from oxidative phosphorylation to anaerobic respiration or fermentation. Upon iron starvation, sRNA RyhB represses TCA cycle enzymes to save iron for essential processes. (D) Example of an anabolic pathway regulated by sRNAs. In Enterobacteriaceae, two homologous sRNAs regulate the key enzyme GlmS to achieve homeostasis of glucosamine-6-phosphate (glucosamine-6-P), an essential precursor for cell envelope synthesis.

sRNAs REGULATING UTILIZATION OF PARTICULAR CARBON SOURCES

Heterotrophic bacteria such as E. coli and Salmonella can grow on a plethora of compounds as sole sources of carbon and energy (11). To preserve resources, genes required for uptake and utilization of a particular carbon source are tightly regulated by substrate availability. Traditionally this task is thought to be achieved by dedicated transcription factors, for which the lactose repressor provides the classical paradigm (12). In fact, Jacob and Monod speculated that the Lac repressor could be an RNA acting at the posttranscriptional level (13), but this idea was largely forgotten until the first report of gene regulation by a small antisense RNA in bacteria (14). Meanwhile, several bacterial carbohydrate utilization systems are known to be fine-tuned by sRNAs. The sRNAs in these circuits may limit the response to substrate availability, set the threshold concentration, or modulate the delay time required for activation and shutdown of the system.

Posttranscriptional Regulation of Glucose Uptake

Many bacteria, including E. coli, preferentially utilize glucose when growing in a mixture of carbon sources (15, 16), which also holds true for many enterobacterial pathogens when residing in mammalian host cells (17). In E. coli, glucose is internalized by the glucose transporter PtsG and to a minor degree by mannose transporter ManXYZ (18). Both transporters belong to the phosphotransferase system (PTS). PTS transporters generate phosphosugars during transport. The phosphoryl groups derived from phosphoenolpyruvate (PEP) are transferred via phosphotransferases enzyme I and HPr to the transporters, including the EIIA^Glc protein, which phosphorylates PtsG.

While phosphosugars are a primary energy source, high intracellular concentrations are toxic (19, 20). Such conditions cause rapid degradation of ptsG mRNA, limiting further glucose uptake, which relieves stress (21, 22). The dedicated transcriptional regulator SgrR senses phosphosugar stress and induces expression of sRNA SgrS (23, 24). Hfq-assisted base-pairing of SgrS with ptsG inhibits translation and recruits endoribonuclease RNase E to degrade ptsG mRNA at the cytoplasmic membrane (23, 25–27). SgrT, a short peptide encoded by SgrS, contributes to stress relief by blocking glucose transport via direct inhibition of PtsG independently of SgrS base-pairing (28–30). In E. coli, SgrS regulates at least eight mRNAs by direct base-pairing (31, 32). Downregulation of the manXYZ mRNA through a dual base-pairing mechanism prevents leaky glucose uptake (33, 34). Stabilization of the yigL mRNA, encoding a sugar phosphatase, by masking an RNase E cleavage site upon base-pairing allows export of sugars following their dephosphorylation (35–37).

While regulation of manXYZ and yigL clearly contributes to phosphosugar stress relief (36), the roles of the remaining targets adiY, asd, folE, ptsI, and purR are less obvious (31, 32). The ptsI gene encodes enzyme I, which delivers phosphoryl groups to all 21 PTS transporters in E. coli (38, 39). Some of these transporters internalize sugars that generate glucose-6-phosphate or other phosphosugars upon catabolism. Thus, global deceleration of PTS activity may contribute to the phosphosugar stress response. Interestingly, phosphosugar stress elicited by a glucose analog or a block in glycolysis, e.g., by pgi mutation, can be rescued by addition of glycolytic intermediates downstream of the block (21, 40). This suggests that toxicity results from the depletion of a downstream metabolite, most likely PEP, rather than from accumulation of glucose-6-phosphate itself. This could explain the physiological roles of SgrS targets such as asd, which encodes an enzyme that converts aspartate to other amino acids. Downregulation of asd may preserve aspartate to replenish PEP and relieve stress (31).

The SgrS-mediated phosphosugar stress response seems conserved in Enterobacteriaceae and Aeromonas species (41, 42). However, PTS-type glucose transporters are much more widespread (18, 43). Do these bacteria also encounter phosphosugar stress, and how do they cope? Downregulation of the ptsG transcript by glucose in a pgi mutant has also been observed for the Gram-positive Corynebacterium glutamicum (44). C. glutamicum lacks Hfq, and therefore the underlying mechanism must differ from E. coli. A phosphosugar stress response has not been reported for any of the Gram-positive Firmicutes species. However, as demonstrated for Bacillus subtilis, these bacteria may activate a glycolytic bypass, the methylglyoxal pathway, to prevent deleterious accumulation of phosphosugars (45).

Regulation of Chitin and Chitosugar Utilization by sRNAs

Chitin is one of the most abundant polysaccharides on earth and particularly ample in aquatic environments, representing an important carbon source for aquatic bacteria such as Vibrionaceae. In Vibrio cholerae, an important facultative human pathogen, chitin even serves as signal for natural competence. Chitin is sensed by the orphan sensor kinase ChiS, which activates expression of chitin utilization genes by a still-unknown mechanism (46, 47). ChiS further activates transcription factor TfoS, which is necessary for expression of the Hfq-dependent sRNA TfoR. This sRNA stimulates translation of TfoX, a regulator required for induction of competence (48, 49). In addition, TfoX induces expression of type VI secretion systems for killing of nonimmune cells and subsequent acquisition of the released DNA (50). This mechanism provides a mechanistic basis for the high degree of genomic diversity observed in V. cholerae.

For E. coli and Salmonella, chitin-derived carbohydrates represent a secondary carbon source as they become sporadically available as part of the hosts’ diet. These species rely on excreted chitinases of other bacteria to convert chitin to chitosugars. Multiple transcriptional regulators and the sRNA ChiX are employed to restrict expression of chitosugar utilization genes to conditions of substrate sufficiency (51–53). Chitoporin ChiP, required for uptake of chitosugars across the outer membrane, is encoded in the chiPQ operon. ChiX inhibits chiP translation initiation by base-pairing with its 5′ untranslated region (5′ UTR) (54, 55) but also represses the distal cistron chiQ by facilitating Rho-dependent transcriptional termination (56). Interestingly, ChiX is not codegraded with its target chiP but with a decoy RNA derived from the chb operon (54, 57). The chb operon encodes a PTS transporter and enzymes for chitosugar uptake and degradation. Expression of the chb operon is activated by the operon-specific transcription regulator ChbR in response to chitosugar availability. When chb transcription rates are sufficiently high, base-pairing with the chb RNA trap sequesters ChiX and relieves chiPQ repression, boosting synthesis of chitoporin ChiP. Thereby, ChiX likely sets the delay time and threshold concentration for chitosugar utilization. ChiX only affects the chb transcript under noninducing conditions, leading to efficient silencing of the mRNA (53).

In the chitinolytic bacterium Serratia marcescens, ChiX coordinates synthesis of ChiP and chitin-degrading chitinases (58). Whereas chiP/ChiX base-pairing is conserved, the ChiX target site within the chb mRNA is lacking. In contrast, ChiX represses chiR, encoding a transcriptional activator of chitinase genes. Upon induction of chiP expression, ChiX is sequestered by base-pairing and repression of chiR is relieved (58). Thereby, ChiX couples induction of degrading enzymes to the expression of the specific transporters, coordinating extracellular breakdown of chitin with uptake of the products.

Regulation of Mannitol Uptake by a cis-Encoded sRNA in V. cholerae

In addition to chitin, mannitol represents an important carbon source for V. cholerae, as it is produced in large quantities by marine algae. In V. cholerae, synthesis of the mannitol PTS transporter MtlA is controlled by the cis-encoded sRNA MtlS through an Hfq-independent mechanism (59). MtlS is transcribed antisense to the mannitol mtlADR operon and shares 71 nucleotides of perfect complementarity with the 5′ UTR of mtlA. MtlS and mtlA form a stable duplex inhibiting mtlA translation without impairing transcript stability. How this affects the cotranscribed mtlDR genes is unknown. Close proximity of the mtlA and mtlS loci is required to efficiently repress mtlA, presumably by enabling rapid formation of the RNA duplex (60). Mannitol represses mtlS transcription, but the responsible regulator has not been identified. The MtlR repressor protein, which is encoded in the mtlADR operon itself, and MtlS appear to operate independently from each other (59, 61). MtlA was shown to activate biofilm formation, suggesting that mannitol serves as extracellular signal for V. cholerae to colonize beneficial habitats (62). Mannitol may also act as compatible solute, helping V. cholerae to withstand the high osmolarity in the human intestine (59). How these additional roles are integrated into the mannitol operon remains to be addressed.

Regulation of Polysaccharide Utilization Genes by cis-Encoded sRNAs in Bacteroides

Gram-negative Bacteroidetes is a dominating phylum of the microbiota in the human colon (63) and specializes in utilizing a wide variety of dietary polysaccharides and glycans derived from the mucosa of the gut (64). To this end, Bacteroides spp. carry a large number of polysaccharide utilization loci (PULs), each one dedicated to the uptake and utilization of a specific glycan or polysaccharide. Each PUL encodes its own protein regulators for substrate-dependent induction of the locus. Transcriptome sequencing analysis of Bacteroides fragilis revealed that many of these PULs transcribe sRNAs from the opposite strand (65). The antisense RNAs seem to be conserved, as they are also observed in other Bacteroides. Overexpression of such an sRNA, DonS in B. fragilis, triggers loss of the corresponding pul transcript, causing disability to utilize corresponding host glycans. DonS may target the cognate pul mRNA towards degradation or act through transcriptional interference by RNA polymerase collision, as observed for other antisense RNAs (2). Regulation by DonS might become relevant when the concentration of the inducing substrate declines, leading to an excess of constitutively produced DonS over the pul transcript. Interestingly, the PULs shown to include antisense sRNAs are all involved in the utilization of host-derived glycans (65). Species like Bacteroides thetaiotaomicron preferentially utilize dietary polysaccharides if available and consequently repress the PULs for glycan utilization (66). Hence, it is possible that the DonS-like sRNAs mediate substrate prioritization in Bacteroides.

CARBON CATABOLITE REPRESSION AT THE POSTTRANSCRIPTIONAL LEVEL

In mixed environments, bacteria often selectively utilize the carbon source favoring fastest growth (15, 67). In E. coli, uptake of the preferred substrate glucose triggers dephosphorylation of the PTS, which activates mechanisms that prevent uptake and utilization of less-preferred carbon sources—collectively known as carbon catabolite repression (CCR) (43, 68). Accumulation of nonphosphorylated EIIA^Glc inhibits uptake of less-preferred carbon sources by inducer exclusion. It also impedes production of cyclic AMP (cAMP), thereby preventing activation of carbohydrate utilization genes by the global transcription regulator cAMP receptor protein (CRP). While these mechanisms are well studied, the involvement of posttranscriptional mechanisms in CCR emerged only recently. A global omics study in E. coli found >90 genes to be posttranscriptionally regulated by CCR (69). In the evolutionary distant pseudomonads, CCR even appears to operate solely at the posttranscriptional level (70).

Spot 42: the Third Pillar of CCR in Enterobacteriaceae

The sRNA Spot 42 cooperates with cAMP-CRP in coherent feedforward loops to regulate multiple carbohydrate metabolic genes (Fig. 2A) (71). It might therefore be considered the third pillar of CCR, working in addition to the well-established mechanisms involving inducer exclusion and cAMP (43). Spot 42, encoded by spf, is one of few genes that are repressed by cAMP-CRP in E. coli (72, 73). The first Spot 42 target discovered was galK, encoding galactokinase for galactose utilization (74). In the presence of glucose, Spot 42 accumulates and selectively downregulates GalK without affecting the other proteins encoded in the galETKM operon, which have additional functions for synthesis of UDP-sugars (Fig. 3) (74). Subsequent work revealed that Spot 42 has a global role in carbohydrate metabolism (75, 76). To date, its validated regulon contains 29 genes (Fig. 2B), but is expected to increase further as many additional potential Spot 42 targets were identified by RNA interaction by ligation and sequencing (RIL-seq) (77). This novel methodology identifies sRNA/mRNA pairs by Hfq pulldown and subsequent ligation of bound RNAs. Impressively, RIL-seq recovered 11 previously validated Spot 42 targets, emphasizing the reliability of the method (Fig. 2B). Again, many of the newly identified candidate targets have roles in carbohydrate metabolism (77).

FIGURE 2.

The transcriptional regulator cAMP-CRP and sRNA Spot 42 cooperate to trigger CCR in Enterobacteriaceae. (A) CRP and Spot 42 participate in coherent feedforward loops to prevent utilization of the indicated secondary carbon sources when the preferred carbon source glucose is present. In addition to cAMP-CRP, Spot 42 is regulated by base-pairing with the sponge RNA PspH. (B) The validated Spot 42 regulon to date. Target genes that are also positively controlled by cAMP-CRP at the level of transcription are boxed. Microarray analysis of Spot 42 pulse expression (75) and improved software prediction algorithms (32, 76) fostered the identification of most targets. Additional targets were identified by human inference or by a CLIP-seq approach mapping Hfq binding sites on a global scale (74, 82, 173, 181). Several of these targets were recovered by RIL-seq (77).

FIGURE 3.

Posttranscriptional regulation of central carbon metabolic pathways in E. coli. Effects of sRNAs (depicted in red) and of the RNA-binding protein CsrA (blue) on synthesis of enzymes involved in glycolysis, gluconeogenesis, and the TCA cycle. A green asterisk and bold letters indicate direct regulation by CsrA (106). Anabolic pathways directing synthesis of glycogen, UDP-sugars, and the biofilm compound PGA are also shown.

Most targets are repressed by Spot 42, and where known, their transcription is activated by cAMP-CRP, fostering the hypothesis that Spot 42 cooperates with cAMP-CRP in coherent feedforward loops to regulate carbohydrate utilization genes (Fig. 2) (75). In the presence of glucose, Spot 42 prevents leaky expression of these genes by targeting the few mRNAs produced despite inactivity of CRP. Furthermore, Spot 42 also shapes the dynamics of gene expression when cells shuttle between CCR and CCR-free conditions. Upon a shift to glucose-rich growth conditions, Spot 42 accelerates repression of the secondary carbohydrate utilization genes, which may facilitate adaptation to the more favorable growth condition. Vice versa, upon activation of CRP by cAMP, Spot 42 delays target activation, perhaps to prevent their premature activation in case glucose reappears (71, 75). Interestingly, many secondary carbon sources whose utilization is repressed by Spot 42 are available in the mucosa of mammalian guts (e.g., arabinose, N-acetylneuraminic acid, and l-fucose; Fig. 2A), E. coli’s natural habitat (78), where Spot 42 may be particularly important for carbon source selection (71).

Until recently, cAMP-CRP was the only known regulator of Spot 42. However, RIL-seq identified a sponge sRNA, PspH, whose overexpression reduces Spot 42 levels and thus derepresses its targets (77). The role of this interaction is unknown. Another study reported induction of spf expression by pyruvate independent of cAMP-CRP (79), suggesting that the spf promoter is controlled by additional transcription factor(s) that remain to be identified.

Spot 42 base-pairing sites overlap or are close to the ribosomal binding site (RBS) (75, 76), and work on galK demonstrated that Spot 42 inhibits translation in an Hfq-dependent manner (74, 80). However, Spot 42 also alters target mRNA levels (75). In-depth study of the galETKM operon revealed that Spot 42 stimulates Rho-dependent transcription termination at the galT-galK junction, recapitulating observations for ChiX (56, 81). A noncanonical mechanism of Spot 42 action was observed for the sdhCDAB mRNA, encoding succinate dehydrogenase (82). Spot 42 pairs far upstream of the sdhC RBS and merely recruits Hfq, which inhibits translation. A direct role for Hfq as a translation repressor has also been reported for other mRNAs (83, 84), even in species beyond E. coli (70). Spot 42 contains three unstructured regions, each of them involved in target regulation, explaining the high conservation of the entire Spot 42 sequence (75). In some cases, Spot 42 employs multiple base-pairing sites to regulate a single target, which might improve regulatory strength (76). Perhaps multisite pairing provides the flexibility required to regulate multiple targets, as observed for other sRNAs controlling exceptionally large regulons (77, 85, 86).

Outside of the Enterobacteriaceae, the spf gene is found in four additional orders of Gammaproteobacteria, including Vibrionaceae (87). In the latter order, the role of Spot 42 was also studied in the fish pathogen Aliivibrio salmonicida and in Vibrio parahaemolyticus, which causes diarrhea and gastroenteritis in humans through consumption of contaminated seafood (88–90). In A. salmonicida, spf expression is negatively regulated by cAMP-CRP, like in E. coli (88). Microarray analysis of an spf deletion mutant revealed upregulation of genes involved in sugar catabolism, motility, and chemotaxis (88). Thus, A. salmonicida Spot 42 may also impact on carbohydrate metabolism, but by targeting different genes than in E. coli, and carrying out additional roles. In V. parahaemolyticus, Spot 42 is strongly upregulated during infection and impacts on the activities of two type III secretion systems (89, 90). It is tempting to speculate that Spot 42 coordinates the activities of the type III secretion systems with carbohydrate availability in the host.

CCR in Pseudomonas by Hfq-Mediated Translational Repression

In pseudomonads, CCR is regulated exclusively at the posttranscriptional level. Contrary to E. coli, pseudomonads do not prefer glucose. Rather, succinate elicits the highest degree of CCR in Pseudomonas aeruginosa but only weakly affects CCR in Pseudomonas putida, which prefers alkanes and branched-chain amino acids (91). CCR requires sRNA antagonists, the catabolite repression control protein Crc, and Hfq as master regulator. Hfq binds the 5′ UTRs of mRNAs encoding transporters and catabolic enzymes for less-preferred carbon sources and directly blocks translation initiation (70). Protein Crc is required for efficient CCR and forms stable ternary complexes with Hfq and RNAs containing A-rich motifs (92, 93). Crc contributes to the stability of these complexes through interaction with both Hfq and RNA (93).

Availability of the preferred carbon source coincides with low levels of sRNA antagonists for Hfq. P. aeruginosa possesses one such sRNA, CrcZ. P. putida encodes two, CrcY and CrcZ, and other species possess similar sRNAs (94–96). Expression of Crc sRNAs is induced by the two-component system (TCS) CbrA/CbrB in the absence of preferred carbon sources (92, 94–97). Kinase CbrA presumably senses internal stimuli reflecting the energetic state of the cell, such as the α-ketoglutarate/glutamine ratio (91, 97). CrcZ of P. aeruginosa and CrcY/CrcZ of Pseudomonas fluorescens bind Hfq with an ∼5-to-20-fold-higher affinity as compared to the mRNAs targeted by Hfq (70, 95, 98). Consequently, increased levels of the Crc sRNAs effectively sequester Hfq from its target transcripts, relieving repression (92, 94). Through competition for Hfq, CrcZ may also interfere with riboregulation exerted by other sRNAs and thus indirectly impact their regulatory potential (98).

POSTTRANSCRIPTIONAL CONTROL OF CENTRAL CARBOHYDRATE METABOLISM

Posttranscriptional Mechanisms Coordinating Central Metabolism with Carbohydrate Availability

The activity of central carbohydrate metabolism is tightly coordinated with carbon supply by adjusting the amounts of corresponding enzymes in response to key metabolites, namely, fructose-1,6-bisphosphate (FBP) and the PEP/pyruvate ratio (1, 99). In E. coli, FBP is sensed by the transcriptional regulator Cra, which represses glycolytic genes and activates genes involved in gluconeogenesis (100, 101). The PEP/pyruvate ratio determines the phosphorylation state of EIIA^Glc, which regulates adenylate cyclase and thus cAMP synthesis (102, 103). Importantly, FBP decreases the PEP/pyruvate ratio through feedforward activation of pyruvate kinase and PEP carboxylase (1). A high FBP level activates glycolysis through inactivation of Cra and decreases activity of cAMP-CRP by inhibiting phosphorylation of EIIA^Glc. Of note, cAMP-CRP activates expression of TCA cycle enzymes (104), whose transcripts are repressed by Spot 42 (Fig. 3). Therefore, cAMP-CRP and Spot 42 also cooperate to redirect metabolism from oxidative phosphorylation to fermentation when glycolytic carbon sources are available. Similar to CRP, Cra possesses a counterpart at the posttranscriptional level, which is the carbon storage regulatory (Csr) system.

Regulation of Glycolysis and Gluconeogenesis by the Csr System

Protein CsrA represents a global posttranscriptional regulator of diverse activities across bacterial species. In E. coli and other Gram-negative bacteria, CsrA controls carbohydrate metabolic pathways, carbon source and nutrient acquisition, biofilm formation, motility, stress responses, and virulence (105, 106). CsrA binds mRNA substrates at GGA motifs and mostly represses translation, but examples of positive regulation also exist (107). For instance, CsrA activates glycolysis by positively regulating mRNAs of several glycolytic enzymes while repressing synthesis of enzymes for gluconeogenesis and the TCA cycle (Fig. 3) (106, 108, 109). In fact, CsrA is essential for growth on glycolytic substrates, reflecting its crucial role in an undisturbed carbohydrate metabolism (110). Flux analysis showed that stabilization of the pfkA mRNA, encoding phosphofructokinase, is crucial for regulation of glycolytic activity by CsrA (106, 109). CsrA also inhibits accumulation of the carbon storage compound glycogen and synthesis of the exopolysaccharide poly-β-1,6-N-acetyl-d-glucosamine (PGA), a major component of biofilm matrices (Fig. 3). Of note, CsrA was recently shown to bind Spot 42 and to activate target genes that are repressed by this sRNA (106). It remains to be shown whether CsrA binding inhibits Spot 42 base-pairing, thereby also influencing CCR. Further, binding to cra mRNA was also demonstrated, but the physiological consequences are so far unclear. These sophisticated interconnections may expand the already complex regulatory network governing carbohydrate metabolism.

CsrA activity is antagonized by the decoy sRNAs CsrB and CsrC, which sequester CsrA by presenting multiple binding sites (111, 112). Transcription and decay of these sRNAs are controlled by signals derived from carbohydrate metabolism. The BarA/UvrY TCS activates transcription of both sRNAs in response to short-chained carboxylic acids, e.g., acetate and formate, which accumulate when cells have expended glycolytic carbon sources and transition into stationary phase (Fig. 4) (113, 114). Degradation of the sRNAs by RNase E requires the protein CsrD (115, 116). CsrD is activated by interaction with nonphosphorylated EIIA^Glc in the presence of glycolytic substrates (102, 117). Together, activation of csrB/csrC transcription and slowdown of CsrB/CsrC decay increases abundance of these sRNAs when preferred carbon sources have been consumed (Fig. 4). The resulting shutdown of CsrA activity promotes the shift to stationary-phase metabolism by repression of glycolytic genes and derepression of gluconeogenetic and glycogen biosynthetic mRNAs (117, 118).

FIGURE 4.

Model of the interconnection of the CsrA system with central carbon metabolism. Decoy sRNAs CsrB and CsrC regulate CsrA activity by sequestering the protein from its target mRNAs. CsrA indirectly activates csrB/csrC transcription, creating a negative feedback loop. In fast-growing cells, when CsrB/C levels are low, CsrA activates glycolytic genes and represses the TCA cycle, gluconeogenesis, and glycogen synthesis. Metabolism of glycolytic carbon sources causes accumulation of FBP, which activates pyruvate kinase, thereby reducing the PEP/pyruvate ratio. Intake of PTS substrates and a low PEP/pyruvate ratio trigger dephosphorylation of EIIA^Glc, leading to activation of CsrD, which triggers degradation of CsrB/CsrC by RNase E. Upon accumulation of short carboxylic acids (R-COOH) as metabolic end products, expression of csrB/csrC is induced by the BarA/UvrY TCS. Deceleration of glycolytic activity elevates the PEP/pyruvate level and increases EIIA^Glc phosphorylation, leading to stabilization of CsrB/CsrC and titration of CsrA. EIIA^Glc∼P stimulates adenylate cyclase CyaA, which converts ATP to cAMP. The cAMP-CRP complex inhibits transcription of csrB/csrC. Involvement of EIIA^Glc in regulation of CsrB/CsrC synthesis as well as decay allows integration of further cues.

EIIA^Glc has an additional role in the activity of the Csr system in E. coli, as it also controls transcription of CsrB/CsrC through cAMP-CRP (119). cAMP-CRP represses csrB indirectly and csrC directly by blocking access of response regulator UvrY to the csrC promoter. Thus, nonphosphorylated EIIA^Glc has opposing effects, as it activates the turnover but also transcription of these sRNAs (Fig. 4), creating an incoherent feedforward loop with the potential to integrate further cues (119). For instance, the activity of adenylate cyclase is also inhibited by α-ketoglutarate, signaling nitrogen limitation (Fig. 4) (120), which may affect CsrB/CsrC expression, but not degradation.

The Csr system is conserved in Proteobacteria, albeit the number of CsrA paralogs and Csr sRNAs may vary (105). The regulatory links between Csr and EIIA^Glc/CRP may likewise differ, as CsrD is absent in most Proteobacteria beyond the families Enterobacteriaceae, Shewanellaceae, and Vibrionaceae (116). Similarly, control of csrB/csrC transcription may be different; e.g., in Yersinia pseudotuberculosis, a close relative of E. coli, expression of csrC is activated by PhoP/PhoQ rather than the BarA/UvrY TCS (121). In sum, the Csr system provides a further tier of controlling fluxes through central carbohydrate metabolic pathways in response to carbohydrate availability. In addition, CsrA may cross-talk to CCR and integrate information on the metabolic status into other intricately regulated processes, namely biofilm formation, motility, and pathogenicity. For more information on this topic, see reference 182.

Regulation of TCA Cycle Activity by sRNAs

ATP can be produced either by substrate-level phosphorylation or by oxidative phosphorylation. Respiration yields more ATP but is also more costly, as it requires more proteins. Bacteria sense the availability of carbon, oxygen, and energy to efficiently regulate the TCA cycle and respiration. As already discussed, in E. coli the information on carbohydrate availability is integrated into the TCA cycle by CsrA, cAMP-CRP, and sRNA Spot 42. sRNAs with comparable functions may also exist in unrelated bacteria. For instance, pathogenic Neisseria species employ two homologous sRNAs to repress transcripts of TCA cycle enzymes (122, 123). Overexpression of these sRNAs impairs growth of Neisseria meningitidis in cerebrospinal fluid but not in blood, suggesting that they integrate information about the metabolic status into the decision to colonize different niches in the host (122). In addition, activity of the TCA cycle is strongly shaped by availability of iron and oxygen. Again, sRNAs play prominent roles in these adaptations.

Downregulation of TCA Cycle Activity by sRNAs in Response to Iron Limitation

Iron is indispensable for activity of numerous enzymes operating within major metabolic pathways. Upon limitation, bacteria redirect iron from nonessential to essential processes with the aid of transcription factor Fur and sRNA RyhB (124). Fur represses transcription of ryhB under iron sufficiency (125). However, upon iron starvation, RyhB is relieved from repression and downregulates nonessential iron-containing proteins, including TCA cycle enzymes (Fig. 3) (126), prompting cells to resort to fermentation (127). This trade-off enables essential pathways involving iron-dependent enzymes to remain functional when iron is scarce. Iron limitation in particular is encountered by pathogenic bacteria within the host (128). Staphylococcus aureus was shown to switch to fermentation inside the host, thereby producing lactate, which lowers the surrounding pH. This increases iron availability through release from host iron storage proteins (129). Switch to fermentation is restricted to bacteria that can grow anaerobically. In the obligate aerobe Azotobacter vinelandii, the functional analog of RyhB named ArrF does not affect TCA cycle-related enzymes, but rather represses genes involved in nitrogen fixation, a nonessential process (130).

Coordination of Carbon Metabolism with Oxygen Availability by sRNA FnrS in Enterobacteriaceae

Enterobacteriaceae are facultative anaerobes. In the absence of oxygen, E. coli uses alternative electron acceptors to procure anaerobic respiration. If oxygen is not available, NAD⁺ is regenerated by fermenting carbon sources to mixed acids and ethanol (131). Two global transcription factors, ArcA and Fnr, reprogram metabolism in response to anaerobiosis (132, 133). Fnr senses oxygen directly, whereas response regulator ArcA is activated by its cognate kinase ArcB when the redox state of the quinone pool changes. Upon anaerobiosis, Fnr and ArcA collectively activate genes of alternative electron transport chains and repress functions of aerobic metabolism, including the TCA cycle, the glyoxylate shunt, and respiratory NADH dehydrogenases (134–137).

Notably, Fnr and ArcA also employ sRNAs in their regulons. One of them, sRNA FnrS, is conserved among Enterobacteriaceae. FnrS is only detectable in the absence of oxygen, as its transcription strictly depends on Fnr and to a minor extent on ArcA (86, 138). Globally, FnrS appears to extend the regulons of Fnr and ArcA by acting as a noncoding regulator to repress functions that are not required in absence of oxygen, including enzymes of aerobic carbohydrate metabolism (Fig. 3) (86, 138). For other targets, e.g., mqo (Fig. 3), FnrS cooperates in coherent feedforward loops, as these genes are also directly repressed by Fnr or ArcA (86, 139). This also applies to acnA and fumC, but here FnrS acts indirectly through repression of MarA, which is a transcriptional activator of these TCA cycle genes—a regulatory scenario known as a multistep coherent feedforward loop (Fig. 3) (32, 138). RIL-seq revealed many additional metabolism-related transcripts putatively base-pairing with FnrS, including the fnr mRNA itself (77), hinting at a feedback loop balancing Fnr and FnrS levels. FnrS is Hfq dependent and appears to act primarily by inhibition of translation initiation (138). Interestingly, FnrS uses distinct sequences to base-pair with subsets of its targets. Transcripts linked to oxidative stress and folate metabolism appear to base-pair with the 5′ end of FnrS, whereas mRNAs of central metabolic enzymes are regulated by a single-stranded region in the sRNA body (86). This functional specialization may reflect evolution of FnrS by fusion of two originally distinct sRNAs (86).

The E. coli and Salmonella ArcA regulon contains an additional sRNA, ArcZ, which is encoded downstream of the arcB gene and is only expressed under aerobic conditions. ArcZ limits accumulation of active ArcA through destabilization of the arcB mRNA (140) and targets further diverse functions, but is apparently not involved in regulation of carbohydrate metabolism (140–143). Recently, the sRNA EsrE was shown to activate synthesis of subunit SdhD of succinate dehydrogenase in E. coli (Fig. 3) (144). EsrE appears somewhat as an aerobic opponent of FnrS, as it is essential for aerobic growth on TCA cycle substrates, but the signal to which it responds remains unknown.

An sRNA activated in response to anaerobiosis was also identified in pathogenic Neisseria species (145, 146). These bacteria, which likely face oxygen limitation during host colonization, are capable of anaerobic respiration (147–149). The anaerobically induced sRNA was named AniS in N. meningitidis and FnrS in Neisseria gonorrhoeae, and both clearly belong to the Fnr regulon, albeit sequence homology to enterobacterial FnrS is lacking (145, 146). So far, only a few targets for these sRNAs are known, and they do not contribute to a common metabolic process (146, 150), leaving it open whether these sRNAs are indeed functional equivalents of enterobacterial FnrS.

RsaE: a Functional Equivalent of FnrS in Gram-Positive Bacteria?

sRNAs also play a role in regulation of central carbohydrate metabolism in Gram-positive Firmicutes. RsaE—later renamed RoxS in B. subtilis (151)—is besides the ubiquitous 6S RNA the sole trans-acting sRNA known to be conserved between staphylococci and Bacillaceae (152). Two independent studies linked RsaE of S. aureus to regulation of carbohydrate metabolism, amino acid transport, and the folate pathway for one-carbon metabolism (152, 153). In particular, RsaE represses pyruvate dehydrogenase and several TCA cycle enzymes (152, 153). Consistently, downregulation of TCA cycle enzymes was also observed for RoxS in B. subtilis (151, 154).

Expression of RsaE/RoxS is induced by the response regulator ResD of the ResD/ResE TCS (151). S. aureus and B. subtilis are facultative anaerobes and can switch to fermentation or nitrate respiration in absence of oxygen. The ResD/ResE TCS (named SrrA/SrrB in S. aureus) responds to oxygen limitation or increased nitric oxide (NO) levels and activates genes required for anaerobic metabolism and NO detoxification (155). Nitrate respiration produces NO as a by-product, which is likely sensed as an indicator of nitrate availability and leads to induction of RsaE/RoxS expression through ResD (SrrA) (151). Therefore, RoxS (RsaE) may extend the regulon of the ResD/ResE TCS, contributing to adaptation to anoxia. B. subtilis RoxS is additionally controlled by transcription factor Rex, which represses genes for fermentation under oxic conditions when the NADH/NAD⁺ ratio is low (154). RoxS is transiently released from Rex repression when malate is utilized, which generates NADH in the early steps of catabolism (154). By stimulating synthesis of the malate transporter YflS, RoxS ensures continuous uptake of malate (154).

Detailed analysis of yflS regulation by RoxS revealed a novel mechanism for how RNA degradation may be counteracted by sRNAs in Gram-positive bacteria. RoxS base-pairs with the 5′ end of the yflS mRNA, thereby protecting it from RNase J1, which degrades RNA in 5′-to-3′ direction—an activity absent in Enterobacteriaceae (154). Among the negatively regulated RoxS targets, the ppnKB mRNA was studied in detail (151). Base-pairing inhibits translation but also creates an RNase III cleavage site destabilizing the mRNA. RoxS uses a C-rich motif for base-pairing—a feature shared by many Gram-positive sRNAs to prevent ribosome recruitment (156). RoxS is cleaved by endoribonuclease RNase Y. Intriguingly, processed RoxS and full-length RoxS exhibit distinct regulatory potentials, albeit the physiological meaning of this functional specialization remains unclear (151).

Posttranscriptional Regulation of Anabolic Carbohydrate Pathways

A number of anabolic pathways using carbohydrates as substrates are regulated at the posttranscriptional level. One example is provided by CsrA, which regulates gluconeogenesis. Another important example is provided by the posttranscriptional control of biosynthesis of cell wall precursors, which must be safeguarded in growing cells, regardless of the nature of the carbon source and the catabolic pathway. In Enterobacteriaceae, this task is achieved by two hierarchically acting sRNAs, GlmY and GlmZ.

Regulation of the Hexosamine Pathway by sRNAs GlmY and GlmZ

Glucosamine-6-phosphate (GlcN6P) synthase (GlmS) catalyzes the first and rate-limiting step in the hexosamine biosynthesis pathway by converting fructose-6-phosphate to GlcN6P (Fig. 3), an essential precursor for cell wall and outer membrane biogenesis (157). Intracellular GlcN6P levels dictate the need for GlmS, whose amount is fine-tuned by posttranscriptional regulatory mechanisms. In Gram-positive bacteria, GlcN6P serves as cofactor for a ribozyme present in the 5′ UTR of the glmS mRNA (158, 159). Following self-cleavage, the glmS mRNA is rapidly degraded by RNase J1 (160). In Enterobacteriaceae, GlmS levels are feedback-regulated by two homologous sRNAs: GlmY and GlmZ (Fig. 5) (161, 162). Only GlmZ is a direct activator of glmS translation (163–165). When GlcN6P is plentiful, GlmZ is inactivated by RNase E cleavage, which requires the dedicated adaptor protein RapZ (166). However, under GlcN6P depletion, cleavage of GlmZ is counteracted to elevate GlmS amounts and replenish the GlcN6P pool. This is achieved through sequestration of RapZ by the decoy sRNA GlmY, whose levels increase when amounts of the metabolite decline (Fig. 5) (164, 166). Consequently, full-length GlmZ base-pairs with the glmS leader and activates expression by disrupting an inhibitory stem-loop structure, thereby exposing the RBS (163, 165). GlmZ is an Hfq-dependent sRNA and a substrate of RNase E, whereas GlmY is not recognized by either of the two proteins (166, 167).

FIGURE 5.

Role of RNase E adaptor protein RapZ in feedback regulation of GlmS synthesis in E. coli. When GlcN6P is plentiful in the cell, RapZ prevents glmS upregulation by targeting its activating sRNA GlmZ to cleavage by RNase E. Within the tripartite complex formed, the sRNA is envisioned to be sandwiched between the tetrameric RapZ protein and the N-terminal domain (NTD) of RNase E, which also forms a tetramer (168). Processing results in functional inactivation of GlmZ and subsequent decline in GlmS levels. Conversely, under GlcN6P depletion, RapZ is predominantly sequestered in complexes with the homologous sRNA GlmY, whose levels increase under this condition. Consequently, GlmZ remains in its active full-length form and stimulates glmS expression. Higher levels of GlmS replenish GlcN6P levels in the cell. Whether RapZ has an active role in sensing GlcN6P via direct binding of the metabolite within its C-terminal domain (CTD) is currently under investigation. The unusual tetrameric structure of RapZ is schematically depicted in the box in the upper left. Each monomer is represented by one color and consists of two globular domains, NTD and CTD, connected via flexible linkers. Three distinct surfaces involved in self-interaction can be discerned: CTD-CTD, NTD-NTD, as well as CTD-NTD (168).

RapZ represents a highly specialized RNA-binding protein, as it exclusively binds GlmY and GlmZ (Fig. 5) (166). Upon binding, no major structural rearrangements are observable in the sRNAs, suggesting that RapZ stimulates cleavage of GlmZ by RNase E through protein-protein interaction (for a detailed discussion of RNase E, see reference 183). Recently, the crystal structure of RapZ revealed an unusual quaternary structure comprising a domain-swapped dimer of dimers—an arrangement that is a prerequisite for RapZ activity in vivo (Fig. 5) (168). The RNA-binding function is located in the C terminus, which bears homology to a subdomain of 6-phosphofructokinase, implying that RapZ may have evolved through repurposing of enzyme components from central metabolism. Putative RNA-binding residues are surface exposed and form basic patches around an extended loop. Intriguingly, a binding pocket for a nonprotein ligand is observed in close vicinity to the presumptive RNA-binding domain (168). It remains to be seen whether this site binds GlcN6P, potentially interfering with sRNA binding. Identification of the GlcN6P binding site may foster the rational design of artificial ligands that can be used for antimicrobial chemotherapy (162).

CONCLUSION AND PERSPECTIVES

A decade ago, when a first review on the current topic was published, only a single target had been identified for SgrS and Spot 42, and together with GlmZ these were the only base-pairing sRNAs known to regulate carbohydrate metabolic genes (169). Meanwhile, such sRNAs have become common, and further examples are expected to follow. For instance, CCR in Gram-positive bacteria may also include posttranscriptional mechanisms, as two sRNAs are controlled by the CCR master regulator CcpA in Streptococcus mutans (170). Even though the contribution of posttranscriptional mechanisms to regulation of metabolism is evident, they are usually neglected in studies assessing metabolic flux control and CCR (171, 172). In addition, the regulatory mechanisms employed by sRNAs are much more diverse than previously envisioned. Novel principles include modulation of target accessibility to degrading RNases, regulation of Rho-dependent transcription termination, recruitment of Hfq as translational repressor, and employment of decoy RNAs sequestering sRNAs or their interacting proteins.

RNA-seq has pushed the development of sophisticated omics approaches facilitating assessment of posttranscriptional regulators on a global scale. RNA immunoprecipitation sequencing (RIP-seq) and UV cross-linking and immunoprecipitation sequencing (CLIP-seq) provide snapshots of substrates bound to RNA-binding proteins at a given time and also identify RNA-binding sites (see, e.g., reference 173). CLASH (cross-linking, ligation, and sequencing of hybrids) and RIL-seq enable the recovery of sRNA-mRNA duplexes, revealing whole RNA networks in a single experiment (77, 174). Most recently, CLIP-seq, ribosome profiling, transcriptomics, and proteomics are combined in “multi-omics” approaches exploring several layers of regulation in parallel and genome-wide. Application of multi-omics to E. coli CsrA revealed novel targets and physiological roles but also confirmed the global character of this posttranscriptional regulator in coordinating bacterial lifestyles with metabolic cues (106). The additional integration of metabolic flux analyses could reveal the specific contribution of sRNAs such as Spot 42 to reprogramming of metabolism.

There is an intimate connection between metabolism and virulence for which carbohydrate-related sRNAs and their protein interaction partners play an important role (175, 176). In fact, bacterial pathogenesis can be regarded as a developmental program granting access to nutrients in a hostile host environment. For instance, mutants lacking CsrA are severely compromised in establishing an infection, which not only is a consequence of dysregulated metabolism but may also result from discoordinated expression of virulence factors (177, 178). A recurrent theme observed in pathogenic bacteria is that sRNAs from the core genome are recruited to regulate horizontally acquired virulence functions, which also applies to SgrS, Spot 42, and GlmY/GlmZ (89, 90, 179, 180). In line with these observations, SgrS and GlmY are strongly upregulated in Y. pseudotuberculosis during infection (177). For more information on this interesting topic, see reference 184.