Regulation of Aerobic Succinate Transporter dctA of E. coli by cAMP-CRP, DcuS-DcuR, and EIIA^Glc: Succinate as a Carbon Substrate and Signaling Molecule

Recently, the significance of C4-dicarboxylates (C4-DC) has been recognized in pathogenic enteric bacteria, serving either as an electron acceptor [1, 2] or electron donor [3] for growth or as a signaling molecule during host infection [4]. Primarily, attention has been directed toward the anaerobic aspects of C4-DC metabolism when C4-DCs such as fumarate, l-malate, or l-aspartate serve as electron acceptors, given that Enterobacteriaceae in a host environment commonly encounter anaerobic to microaerobic conditions. Nevertheless, the aerobic type of C4-DC metabolism is intriguing as it provides insights into the hierarchical carbon utilization in Escherichia coli when C4-DCs compete with monosaccharides as a carbon source. Carbon catabolite repression (CCR) largely relies on cAMP-CRP signaling and fructose 1,6-bisphosphate formation [5]. Furthermore, the phosphoenolpyruvate (PEP)-to-pyruvate ratio serves as a crucial indicator of nutrient quality and availability, with elevated PEP/pyruvate ratios observed in cells experiencing nutrient deprivation, such as during starvation [6, 7].

Under aerobic conditions, DctA functions as the primary C4-DC transporter, utilizing the proton potential for driving the reaction. Under more acidic conditions, DauA supports DctA in C4-DC uptake [8]. DctA is responsible for the uptake of fumarate, succinate, l-malate, and d-malate [9‒12]. In addition, a small part of the DctA population is part of a complex with the DcuS sensor kinase to form the functional DctA × DcuS sensor complex under aerobic conditions [12‒14]. The transcription of dctA is controlled by the C4-DC two-component system DcuS-DcuR [15, 16]. Moreover, the transcription of dctA is under the regulatory control of CCR mediated by the cyclic AMP receptor protein (CRP) and is influenced by aerobic conditions through the ArcBA two-component system [16‒18]. The PEP:glucose-specific phosphotransferase system (PTS) plays a pivotal role in utilizing a diverse range of carbohydrates. This system serves as a key mediator in regulating carbon metabolism in glucophilic E. coli, where glucose stands as the preferred carbon source [19]. The PTS mediates the transfer of the phosphoryl group from the donor PEP through the EI, HPr, and EIIA^Glc proteins to the glucose transporter EIIBC (shown in Fig. 1).

The phosphorylation status of EIIA^Glc is contingent upon the availability of glucose and PEP. In the presence of glucose, phosphorylated EIIA^Glc transfers the phosphoryl group to EIIB, catalyzing the phosphorylation of glucose to glucose-6-phosphate during its uptake through group translocation (shown in Fig. 1). In the presence of glucose, EIIA^Glc is primarily in its dephosphorylated state, and it can hinder the activity of transporters responsible for alternative substrates, such as the lactose permease LacY – a phenomenon known as inducer exclusion [20, 21]. Conversely, in the absence of glucose, EIIA^Glc is primarily phosphorylated. This phosphorylated form stimulates cAMP production by activating adenylate cyclase CyaA [22, 23]. The resulting cAMP-CRP complex induces the expression of glucose-repressed genes [24]. Carbohydrates and related compounds, including glucose, mannitol, and mannose, which are translocated by PTS-linked transporters, are categorized as PTS substrates. On the other hand, non-PTS substrates are transported by primary or secondary transporters. In glycolytic bacteria like E. coli, where the PTS forms the basis for CCR, the PEP/pyruvate ratio serves as a crucial indicator of carbon metabolism [6, 7, 19]. However, non-PTS substrates like glycerol, maltose, and melibiose, which generate PEP during degradation, impact the phosphorylation status of EIIA^Glc and consequently influence cAMP-CRP signaling [7, 25, 26]. CRP regulates the expression of more than 180 genes in E. coli [27]. Most of these genes are involved in catabolism of secondary (or non-PTS) carbon sources, such as lactose, glycerol, and maltose. Additionally, CRP is involved in a multitude of other processes, e.g., nitrogen assimilation [28], osmoregulation, and virulence [29]. A recent proteomic analysis identified metabolic and transcriptional regulation that is influenced by a network between C4-DC-specific (DcuS-DcuR) and CCR (cAMP-CRP) [30]. CRP is activated by the secondary messenger cAMP and is subsequently able to bind to DNA [19, 31]. This study explored the transcriptional control of dctA expression in the presence of diverse carbohydrate sources. Similarly, to gain an initial understanding of posttranslational regulation, the interaction between DctA and EIIA^Glc from the glucose-specific PTS was investigated using a bacterial two-hybrid system (BACTH).

The E. coli K12 strains are listed in Table 1 [32, 33, 34, 35, 36, 37]. All molecular methods, including phage P1 transduction, were performed according to standard procedures [38‒41]. Phage P1 transduction to obtain P_dctA-lacZ mutants was performed as described previously [39, 42]. The recipient strain was either IMW385 (P_dctA-lacZ) or IMW386 (P_dctA-lacZ, dctA::Spc^R), and the donor strains were from the Keio collection [32]. Confirmation of successful recombination was achieved through sequencing the region of interest or by observing growth on selective agar plates. The region of interest was amplified using Phusion™ High-Fidelity DNA Polymerase (Thermo Fisher Scientific), and the PCR products were purified with the QIAquick PCR purification kit (QIAGEN) for sequencing with the corresponding forward primer (shown in Table 2). Bacterial cultures were grown aerobically at 37°C in lysogeny broth (LB) as specified. In general, effectors for the different experiments were supplied in 20 mm concentration. Antibiotics were applied at 100 μg/mL for ampicillin and 50 μg/mL for kanamycin. When two antibiotics were concurrently used, the concentrations were reduced by half. All experiments requiring bacterial growth were conducted in 48-well cell culture plates (Sarstedt) incubated in a plate incubator (Heidolph) set at 37°C with a rotation speed of 900 rpm.

Protein-protein interaction analyses were conducted employing a BACTH, as described by Karimova et al. [35] [1998; 2001]. Fusion constructs were generated by combining the T25 and T18 domains with the target genes, including dctA, and crr (EIIA^Glc). DctA fused N-terminally to either T18 or T25 was already constructed and published [40]. The fusion of crr (EIIA^Glc) to both T18 and T25 was accomplished using the crr_for (BACTH) and crr_rev (BACTH) primers for amplification with chromosomal DNA from BW25113. These primers incorporated BamHI and KpnI restriction sites into the crr PCR product, facilitating the cloning of crr into pKT25 and pKNT25 (shown in Tables 1, 2). The cloning process yielded pMW2787 (_T25EIIA^Glc) and pMW2788 (_T25EIIA^Glc), and both constructs were confirmed through sequencing. Additionally, T25-zip and T18-zip, integral components of the BACTH system, were utilized as positive controls for the experiments. Zip refers to the leucine zipper region within the yeast protein GCN4, known for its robust dimerization capability. This methodology allowed for the exploration of potential interactions among the specified proteins. BTH101 cells were co-transformed with T18/T25 crr and dctA variants and cultured in selective LB medium containing ampicillin and kanamycin. On the day of the experiment, a subculture was prepared, inducing the expression of protein fusions with IPTG (250 μm) after 2 h. After 4 h, the cell cultures were used in the β-galactosidase assay.

Bacteria (shown in Table 1) were grown aerobically in LB medium with effectors as stated in the experiment. Interactions in the BACTH system were measured in terms of β-galactosidase activity [39]. BACTH experiments were conducted as described previously [41] with slight modifications [43]. The reported activities represent the mean values from at least two independent experiments, each with four replicates. In both experiments, measuring reporter gene activity of the dctA promoter (P_dctA) fused to lacZ, and the BACTH experiment involving EIIA^Glc and DctA, cultures were grown in 48-well plates (Sarstedt) with a volume of 250 μL per well. After reaching the desired OD₅₇₀, either exponential phase (OD₅₇₀ = 0.5) or stationary phase (OD₅₇₀ > 1.2), cells were permeabilized by adding 200 μL of culture to 800 μL β-galactosidase buffer composed of KPi buffer (potassium phosphate [100 mm, pH 7.0], potassium chloride [10 mm], magnesium chloride [1 mm], cetyltrimethylammonium bromide [0.005% w/v], sodium deoxycholate [0.0025% w/v] with dithiothreitol [8 mm]). For the β-galactosidase assay, 150 μL of permeabilized cells were transferred to 96-well plates and incubated at 30°C. To initiate the reaction, 30 μL of ortho-Nitrophenyl-β-d-galactoside (4 mg/mL) was added. The reaction was halted after 20 min by adding 70 μL Na₂CO₃ (1 m). Each strain was measured in two to four biological repeats, with four independent samples each.

For testing the response of the dctA promoter (P_dctA) to different carbon sources, a genetic fusion was constructed between P_dctA and the lacZ reporter gene, encoding β-galactosidase [40]. The expression of P_dctA-lacZ was assessed in the presence of various pentoses, hexoses, disaccharides, sugar alcohols, pyruvate, lactate, and C4-DCs (shown in Fig. 2a). Furthermore, an analysis of P_dctA-lacZ expression was carried out in a reporter strain lacking dctA, resulting in a constitutively active DcuS (shown in Fig. 2b). This setup renders the expression of P_dctA-lacZ independent of C4-DC-mediated induction, as DcuS-DcuR is already fully activated in this genetic context [12, 16, 40, 44, 45]. The bacteria were grown aerobically in LB medium in presence of different carbon sources. As expected and previously shown [12, 16], the activity of P_dctA-lacZ was approximately fivefold higher in the dctA-deficient reporter strain compared to the wild type (shown in Fig. 2a, b). Notably, only the reporter strain with a wild-type dctA background exhibited an increase of P_dctA-lacZ expression in the presence of the C4-DCs fumarate, l-malate, and l-aspartate, compared to the reference activity in LB medium without the addition of a carbon source (shown in Fig. 2a). The dctA-deficient reporter strain that exhibits higher expression of the reporter, generally showed a higher repression of P_dctA-lacZ by the substrates compared to the control. Moreover, even C4-DCs repress P_dctA-lacZ compared to the LB reference (shown in Fig. 2b). Notably, the order in which carbon sources repress P_dctA-lacZ expression is quite similar in both strains. As expected, glucose showed the highest repression of P_dctA-lacZ. Maltose, glycerol, l-arabinose, and d-xylose stood out as non-PTS sugars by their high repression. These substrates are followed in their repression by pyruvate and rhamnose, and eventually C4-DCs, and the disaccharide lactose. The C4-DCs and lactose showed the weakest repression of P_dctA-lacZ (shown in Fig. 2b). In conclusion, the behavior of P_dctA-lacZ underscores E. coli’s capacity to adjust gene expression to carbon sources, but the response depends largely on the specific carbon source.

The pentose d-xylose and the sugar alcohol glycerol displayed pronounced repression of P_dctA (shown in Fig. 2), despite neither of these carbohydrates being transported via the PTS. The extent of P_dctA repression by the various carbon sources (shown in Fig. 3a) allows no conclusion whether CCR or other specific substrate-related regulation play the major role in repression. Glycerol and d-xylose repress P_dctA with high efficiency without being PTS substrates. The following experiments further delve into the role of d-xylose and glycerol on dctA expression to evaluate whether these substrates act by specific transcriptional regulation or by CCR and cAMP-CRP. To investigate the involvement of intermediates from the corresponding pathways in the repression of P_dctA-lacZ, the respective kinases for glycerol (GlpK, or glpK gene) and for d-xylose (XylB, or xylB gene) and transcriptional regulators GlpR (glpR gene) and XylR (xylR gene) were genetically inactivated in both P_dctA-lacZ reporter strains. The bacteria were cultured aerobically in LB medium, with either glycerol or d-xylose serving as an effector.

The repression of dctA by glycerol was abolished in the glpK-deficient strain. GlpK catalyzes the phosphorylation of glycerol to glycerol-3-phosphate (G3P). The regulator GlpR associated with glycerol metabolism exhibited no effect on P_dctA-lacZ expression (shown in Fig. 3b). These findings corroborate similar observations for P_malT-lacZ expression, where the repression of malT by glycerol was demonstrated to rely on cAMP-CRP [25]. A similar observation was made for d-xylose, where genetic inactivation of the xylulokinase XylB resolved dctA repression. Interestingly, the XylR-deficient mutant showed the same effect (shown in Fig. 3c). This observation can be explained by a previous work that showed how xylR-deficiency abolished the expression of the xylAB and xylFGHR operons [46], suggesting that this phenotype is due to the inability of the xylR-deficient strain to express XylB. This implies xylulose 5-phosphate-specific repression, akin to the observation with G3P, suggesting that the degradation of the respective sugar is crucial for repressing P_dctA-lacZ expression through cAMP-CRP regulation in both cases.

cAMP-CRP is essential for the activation of catabolite-controlled genes associated to the degradation of carbohydrates other than glucose [47‒49]. The PEP/pyruvate ratio influences cAMP production through the PTS, where PEP serves as the phosphoryl group donor for EIIA^Glc. The phosphorylation state of EIIA^Glc directly influences cAMP production through the stimulation of adenylate cyclase CyaA [7]. Accordingly, cAMP levels fluctuate depending on transport (PTS vs. non-PTS substrate), the metabolic link of the substrate to PEP production, and additionally by individual regulation. To disrupt activation by cAMP-CRP, the adenylate cyclase CyaA was genetically inactivated in the P_dctA-lacZ reporter strains. In contrast, glucose was employed as an effector, resulting in the lowest cAMP levels for comparison. The expression of P_dctA-lacZ shows a 6.3- and 4.6-fold decrease in the presence of glucose or in the cyaA-deficient strain, respectively (shown in Fig. 4a). As expected, in the dctA-deficient P_dctA-lacZ reporter strain, activity showed a more substantial reduction of 13.4- and 13.8-fold since the basal activity of P_dctA-lacZ is higher (shown in Fig. 4b). In both instances, the absence of cyaA and the presence of glucose demonstrate comparable repression of P_dctA-lacZ. These results suggest that the C4-DC-dependent induction mediated by DcuS-DcuR is subordinate to the activation driven by cAMP-CRP. This induction likely occurs only when there are ample amounts of cAMP-CRP complexes present. To summarize, the transcription of dctA is primarily contingent on cAMP-CRP activation, serving as the pivotal determinant for the transcription of P_dctA. In contrast, C4-DC-specific stimulation by DcuS-DcuR is less potent and requires the presence of cAMP-CRP.

EIIA^Glc of the glucose-specific PTS is known for inhibiting the transport of metabolites with lower preference, as observed in the case of the lactose permease LacY [50]. To investigate the potential interaction between DctA and EIIA^Glc, which could form the foundation for such an inhibitory interaction, a BACTH was employed. BACTH relies on the reconstitution of adenylate cyclase activity through distinct T18 and T25 adenylate cyclase domains of Bordetella pertussis [36]. The N- and C-termini of DctA have cytoplasmic location [37], which is important since cAMP needs to be produced cytoplasmically to induce the reporter lacZ. Fusions of DctA to the T18 and T25 fragments adenylate cyclase were produced in all combinations. Similarly, the N- and C-termini of EIIA^Glc were genetically fused to the T18 and T25 fragments. The co-expression of DctA and EIIA^Glc, fused to T18 and T25, respectively, in the same strain resulted in a significantly elevated β-galactosidase activity, comparable to or even surpassing the positive control (_T18Zip × _T25Zip) and clearly exceeded the background activity observed in _T18Zip × DctA_T25 combination (shown in Fig. 5 [35, 36]). These findings strongly indicate a distinct interaction between DctA and EIIA^Glc.

E. coli can grow on a mixture of sugars and other carbon sources but consumes the available substrates in sequential order through a process known as CCR [51]. The order depends, among others, on the growth rate or the metabolic suitability of the substrate for E. coli [52, 53]. This ensures that bacterial cells adapt their metabolic resources to the preferred carbon source, which in turn optimizes the growth rate. CCR is a central regulatory mechanism that is involved in the regulation of 5–10% of all genes [51, 52, 54, 55]. The classic example of CR is the biphasic growth (diauxie) of E. coli on a glucose and lactose-containing medium, with the bacteria initially using glucose and then lactose [56, 57]. In the absence of PTS substrates, EIIA^Glc is phosphorylated and stimulates cAMP production by CyaA. The increase in intracellular cAMP leads to the formation of cAMP-CRP complexes and subsequent induction of genes requiring transcriptional activation by cAMP-CRP. However, also non-PTS substrates, such as glycerol, maltose, and melibiose, affect the phosphorylation state of EIIA^Glc via PEP formation and thus cAMP production [7, 25, 26]. These regulatory features highlight the ability of cAMP-CRP to indirectly sense the quality of carbon sources and thus modulate transcriptional regulation.

Class I CRP-dependent promoters necessitate cAMP-CRP for transcriptional activation. The P_dctA reporter gene belongs to this category, relying on cAMP-CRP for activation (Fig. 4). In addition, P_dctA features a CRP-binding site at a position characteristic of class I promoters. In class I promoters, the CRP dimer engages with the α-C-terminal domain of the α-subunit of the RNA polymerase (RNAP) [58]. This interaction enhances the affinity of RNAP for the promoter DNA [59‒61], ensuring proper positioning for open complex formation [62]. cAMP-CRP induces a nucleic acid bend of approximately 87° at the CRP consensus sequence [63, 64]. However, the extent of this bend is contingent on the nucleotide sequence of the CRP-binding site. In the case of P_dctA, the CRP and DcuR-binding sites are situated at positions −81.5 and −105.5 upstream of the transcription start, respectively [16, 65] (shown in Fig. 6a [16, 53, 63, 65, 66]). Both expression and positioning data indicate that the induction by DcuR relies on cAMP-CRP-induced DNA bending, which strategically places DcuR in proximity to RNAP (shown in Fig. 6d). Proteomic data for E. coli cultured under various carbon sources in minimal M9 medium revealed a relatively constant level of CRP at 2,943 (±535 ≜ 18.17%) copies per cell [67]. The concentration of cAMP exhibited significant variability in the presence of glucose (35 μm), glycerol (83 μm ≜ 137% increase), and acetate (146 μm ≜ 317% increase) [68], affirming that intracellular cAMP concentrations essentially govern cAMP-CRP activation. Furthermore, the quality and quantity of the CRP-binding site are crucial factors. Quality pertains to the consensus sequence of the CRP-binding site, while quantity reflects the number of CRP-binding sites preceding a transcription start.

The consensus sequence of CRP-binding sites (5′-AAA​TGT​GAT​CTA​GAT​CAC​ATT​T- 3′) is palindromic, featuring a consensus half site of 5′-A₁A₂A₃T₄G₅T₆G₇A₈T₉C₁₀T₁₁-3′ [53, 69] (shown in Fig. 6c). The first half of the binding site is dominant in CRP binding and therefore conserved. CRP directly interacts with G₅, G₇, and A₈ in the core motif of the first half site [53, 70, 71]. Mutations in the CRP site toward the consensus lead to an enhanced ability of cAMP-CRP to activate the promoter and thereby reprogram the hierarchy in sequential carbon utilization [55, 72, 73]. Interestingly, in the promoter region of dctA, G₅ of the first half site of the CRP-binding site is replaced by T₅, potentially weakening the interaction with cAMP-CRP (shown in Fig. 6c). Variations in the second half of the CRP-binding site are less important for cAMP-CRP interaction [53, 70, 71].

In addition to the quality of the CRP-binding site, the total number of CRP-binding sites represents a crucial characteristic in gene regulation [53]. The mannitol-specific PTS enzyme II MtlA facilitates the transport and phosphorylation of mannitol. Mannitol-1-phosphate 5-dehydrogenase MtlD then converts mannitol-1-phosphate to fructose-6-phosphate, which subsequently enters glycolysis [74]. The mtlAD operon comprises five CRP-binding sites [75], exhibiting a conservation level ranging from 58% to 75% compared to the CRP consensus [63]. On the other hand, a significant portion of CRP-controlled genes possesses only one CRP-binding site, akin to the situation observed in P_dctA [66]. In summary, P_dctA represents a particular example of cAMP-CRP activation for the regulatory fine-tuning in response to both global and specific transcriptional regulators, including cAMP-CRP and DcuS-DcuR.

Feed-forward loops (FFLs) represent crucial regulatory networks in E. coli, involving two transcription factors (TFs) and one or more target genes. In FFLs, TF1 directly regulates the expression of the target gene and of TF2, while TF2 also influences expression of the target gene, resulting in the indirect regulation of the target gene by TF1 via TF2. If the direct effect (positive or negative) of TF1 on the target gene is the same as the indirect effect of TF2 on the target gene, the FFL is termed coherent [76‒78]. In the dctA FFL (shown in Fig. 6b), CRP (TF1) directly activates dctA (target) and DcuR (TF2) transcription [16, 79], and DcuR positively regulates dctA [65]. Thus, this network motif is characterized as a coherent CRP-FFL (shown in Fig. 6b) [78]. This CRP-FFL suggests that sufficient cAMP-CRP must be present for specific transcriptional regulation to occur. This introduces a second layer of regulation to dctA. In the presence of glucose, low cAMP levels hinder the transcriptional activation of dcuR and dctA. Conversely, when alternative carbon sources like C4-DCs are present, elevated cAMP levels activate the transcription of dcuR and dctA via cAMP-CRP, enabling specific transcriptional activation by DcuS-DcuR in response to C4-DCs. In summary, the regulation of dctA involves two essential layers: (I) the interaction of cAMP-CRP with DcuR at P_dctA for the DcuR-mediated induction in the first place, and (II) the cAMP-CRP activation for dcuR expression to obtain sufficient DcuR levels. This establishes a two-step transcriptional activation process required for aerobic C4-DC uptake to take place (shown in Fig. 6d).

The expression of dctA is tightly controlled by cAMP-CRP in response to carbon availability as described above. In addition, the PTS is known for its ability to post-translationally inhibit transporters of secondary metabolic pathways through EIIA^Glc. The BACTH system demonstrated a robust interaction between DctA and EIIA^Glc, reminiscent of the interaction observed in LacY-EIIA^Glc. This interaction results in the inhibition of LacY, a phenomenon known as inducer exclusion, which restricts the uptake of lactose [50]. Although the interaction between DctA and EIIA^Glc is evident, its role and physiological consequences remain unclear. The presumed outcome is to exclude C4-DCs as alternative substrates from the cell when a preferred carbon source like glucose is available. Importantly, this inhibition of C4-DC uptake does not result in classical inducer exclusion, as C4-DCs are sensed in the periplasm by the two-component system DcuS-DcuR [80, 81]. The proposed model introduces the concept of “substrate exclusion,” which differs from classical inducer exclusion known for the inhibition of lactose permease LacY or the maltose ABC transporter [50, 82, 83]. Classical inducer exclusion prevents the uptake of substrates that also act as inducers in the cytoplasm, thereby inhibiting the induction of alternative metabolic pathways. The interaction between DctA and EIIA^Glc suggests an additional layer of regulation on DctA, complementing the transcriptional regulation of dctA by cAMP-CRP. C4-DCs play a dual role, serving as both a growth substrate [1, 84] and as signaling molecules for host-bacteria interaction [4]. The exclusion of these substrates makes sense, as bacteria can so differentiate between degradation and preserving the signaling function of these metabolites. Therefore, the “substrate exclusion” might support the recently identified signaling role of succinate in host-bacteria interaction [4]. In summary, the control of dctA underscores the intricate interplay between global and specific transcriptional regulation. This coordination fine-tunes gene expression in response to both general environmental cues, such as carbohydrates, and specific stimuli, like C4-DCs. Furthermore, it may be imperative to implement regulations at multiple levels to effectively coordinate and integrate the dual purpose of carbon substrates for growth and signaling in bacterial host interactions for this metabolite group.

We thank Dr. Alexandra Kleefeld for the construction of the parent reporter strains IMW385 and IMW386. We also thank Florence Best for experimental help. We are also grateful to the National BioResource Project (NIG, Japan) for providing E. coli strains and to Dr. D. Ladant (Paris) for supplying the strains and plasmids for the BACTH system.

Ethic approval is not required for this study (no experiments with humans or animals). A preprint version of this article is available on bioRxiv [85].

The authors have no conflicts of interest to declare.

We are grateful to Deutsche Forschungsgemeinschaft for funding G.U. (UN 49/19-1 and UN 49/21-1) and C.S. (SCHU 3606/1-1). Thanks to Prof. Dr. Wolf-Dietrich Hardt for financial support to C.S. and NCCR Microbiomes grant (51NF40_180575) to W.-D.H.

C.S. contributed to the design of the study, the acquisition and interpretation of the data, and the writing, and G.U. contributed to the design of the study, interpretation of the data, and writing.

All the data produced in the course of this investigation are incorporated within this article. For additional inquiries, please contact the corresponding author.