In an acidic (pH 5.8) and lysine-rich environment Escherichia coli induces expression of the cadBA operon which encodes CadA, catalysing the decarboxylation of lysine to cadaverine, and CadB, the lysine/cadaverine antiporter. cadBA expression is dependent on CadC, a membrane-integrated transcriptional activator which belongs to the ToxR-like protein family and directly binds to the DNA in the cadBA promoter region. Here we describe that CadC senses the extracellular lysine not directly but indirectly requiring the interplay with the lysine permease LysP. Biochemical analyses of isolated CadC or the periplasmic domain of CadC (CadC188–512) revealed an unexpectedly low affinity for lysine, making it unlikely that CadC is a direct sensor for this substrate. Moreover, CadC hybrid proteins, in which the transmembrane domain or single amino acids were replaced, supported lysine-independent cadBA expression but retained a pH-dependent regulation. These CadC mutants were resistant to the effect of an overproduction of LysP, which represses cadBA expression in wild-type cells. Our results suggest a model according to which CadC is inactivated by an interaction with LysP at a low external lysine concentration. When lysine is abundantly available, the interaction between LysP and CadC is released, and CadC becomes susceptible to activation by low pH.
Bacteria respond to changes in the external pH by altering various physiological processes and their pattern of gene expression. In Escherichia coli several genes are induced upon growth at acidic external pH to support survival under this unfavourable condition. Among these genes are those for the degradative amino acid decarboxylase systems adi, gad and cad (Gale and Epps, 1942).
The Cad system of E. coli consists of the cytoplasmic protein CadA and the transmembrane proteins CadB and CadC. CadA is a lysine decarboxylase that catalyses decarboxylation of lysine to cadaverine resulting in the consumption of a cytoplasmic proton. The alkaline cadaverine is concomitantly excreted by the lysine/cadaverine antiporter CadB (Auger et al., 1989; Soksawatmaekhin et al., 2004). CadB is probably anchored in the cytoplasmic membrane with 12 transmembrane helices, displaying similarity to the ornithine/putrescine antiporter PotE. cadA and cadB are organized in one operon, localized at 93.7 min on the E. coli chromosome (Meng and Bennett, 1992; Soksawatmaekhin et al., 2004). The cadBA operon is under the control of the PCad promoter. Expression of the cadBA operon is induced by external acidification, and the simultaneous presence of lysine in the medium. The end-products of lysine decarboxylation, cadaverine (Neely et al., 1994) and CO2 (Takayama et al., 1994) inhibit cadBA expression. Under anaerobic conditions the expression level is almost 10 times higher than under aerobic conditions (Sabo et al., 1974). CadC is the positive regulator of cadBA expression (Watson et al., 1992) and acts on the cadBA promoter via direct binding to its binding sequences CAD1 and CAD2 (Küper and Jung, 2005).
The cadC gene lies upstream of the cadBA operon and encodes a 58 kDa inner membrane protein. CadC, a member of the ToxR-like transcriptional activators (Miller et al., 1987), consists of a cytoplasmic N-terminal domain (amino acids 1–158), a single transmembrane domain (amino acids 159–187) and a periplasmic C-terminal domain (amino acids 188–512) (Watson et al., 1992; Dell et al., 1994; Fig. 1). The cytoplasmic domain shows sequence similarity to the ROII-subgroup of DNA-binding domains of response regulators like PhoP from Bacillus subtilis, VirG from Agrobacterium tumefaciens or OmpR from E. coli (Watson et al., 1992). In contrast to OmpR-like transcriptional regulators, signal transduction in CadC functions without any chemical modification, e.g. phosphorylation. Thus, CadC and all other ToxR-like proteins represent the simplest form of a stimulus-response mechanism in bacteria. Based on CadC derivatives with altered sensing properties owing to single amino acid replacements within the periplasmic domain, it was suggested that this domain is the signal input domain and that pH and lysine are sensed independently from each other (Dell et al., 1994).
Two proteins, LysP and H-NS, are known to be involved in the negative regulation of cadBA expression under non-inducing conditions (Shi et al., 1993; Neely et al., 1994). The global regulator H-NS participates in the formation of a repression complex at PCad under non-inducing conditions (Küper and Jung, 2005). For LysP, the lysine permease of E. coli, a role in the lysine-dependent regulation of the Cad system was postulated because a loss of functional LysP resulted in lysine-independent cadBA expression (Neely et al., 1994).
Until now it was not yet clear how the function of LysP is linked to the ability of CadC to induce cadBA expression. If CadC acts as a lysine sensor and depends on lysine for activity, it is possible that LysP regulates CadC indirectly by diminishing the external lysine concentration. Alternatively, LysP might regulate CadC by direct interaction, and lysine availability would be sensed solely by LysP.
Our analysis revealed that CadC or CadC188–512 are hardly able to bind lysine and thus CadC is most likely not a lysine sensor. Instead, perception of the lysine signal was found to be strongly dependent on the lysine permease LysP. Amino acid replacements in CadC revealed that the activity of this protein is modulated by LysP via the transmembrane domain of CadC.
CadC has an extremely low affinity for lysine
In order to address the question if CadC is a lysine sensor, a non-radioactive binding assay was developed. The assay utilized the effect of ligands on tryptophane fluorescence of a truncated, soluble CadC derivative representing the periplasmic domain of the protein (CadC188–512; Fig. 1). CadC188–512 contains four tryptophane residues (W211, W318, W450, W488) that can contribute to the intrinsic fluorescence detected in this assay. Measuring the intrinsic tryptophane fluorescence is a tool for detecting conformational alterations in proteins such as after binding of ligands. For example, this method was applied to determine the effect of sugar binding on melibiose permease (Mus-Veteau et al., 1995), the affinity of CaiT for l-carnitine (Jung et al., 2002), and the affinity of OpuAC for certain osmoprotectants in B. subtilis (Horn et al. 2005).
Based on earlier results it was proposed that lysine and cadaverine influence the CadC activity directly (Dell et al., 1994). Therefore, both ligands were tested independently in this assay. In the absence of any ligand, the fluorescence spectrum was characterized by a maximum emission wavelength of 337 nm (Fig. S1). Addition of 1 mM cadaverine caused a quenching of the intrinsic tryptophane fluorescence of CadC188–512 and thereby a significant decrease in the fluorescence measured at 337 nm while no shift in the emission maximum was detectable. To exclude unspecific effects, the same experiment was carried out with spermidine as a negative control, a polyamine that was without effect on cadBA expression in vivo (Neely et al., 1994). Changes in tryptophane fluorescence of CadC188–512 recorded for cadaverine were corrected with the values obtained for spermidine. Titration experiments with varying concentrations of cadaverine resulted in fluorescence alterations (ΔF/F) of CadC188–512 and plotting ΔF/F against the substrate concentration yielded a typical saturation curve (Fig. 2A). After plotting of the data according to Scatchard (1949), the apparent affinity of CadC188–512 for cadaverine was determined as 96 ± 18 μM (Fig. 2B). The same experiment was performed for lysine. However, lysine did reduce tryptophane fluorescence only marginally, and plotting the data did not allow a calculation of the KD value (Fig. 2B). Even addition of high amounts of lysine (up to 30 mM) and performing the experiment at pH 5.8, gave only a slight shift in the signal intensity, indicating an extremely low affinity of CadC188–512 for lysine (data not shown). Alternatively, the affinity of full-length CadC for lysine was determined by isothermal titration calorimetry (ITC), and a KD value of 22 mM was calculated (Fig. S2).
Taking this low affinity of CadC into account, we determined whether cadBA expression would increase with higher external lysine concentrations. Starting at a lysine concentration of 10 mM, the system was highly induced as determined by the amount of CadA produced. An increase of the exogenous lysine concentration up to 250 mM did not lead to a further increase in the amount of CadA (data not shown). These results indicate that the activation of CadC becomes independent of the exogenous lysine concentration once a threshold concentration is exceeded. Remarkably, the threshold concentration of external lysine (around 5 mM, Neely et al., 1994) was found to be below the KD value determined for CadC.
The role of the transmembrane domain of CadC for lysine-dependent cadBA expression
As CadC is the essential transcriptional activator of cadBA expression (Küper and Jung, 2005), but hardly a lysine sensor, and as its activity is completely inhibited by overproduction of LysP (Neely et al., 1994), it was reasonable to assume that sensing of the lysine signal by CadC is solely mediated by an interplay or direct interaction with the lysine permease LysP. LysP is a permease with 12 transmembrane helices with relatively short interconnecting loops (the largest loop contains 43 amino acids; Ellis et al., 1995). Thus both the periplasmic domain and the transmembrane domain of CadC might provide potential interaction surfaces.
Earlier, the region around amino acid 265 within the periplasmic domain of CadC was described as important for sensing the lysine signal (Dell et al., 1994). However, it is important to note, that all of the mutants isolated after a random mutagenesis carried amino acid replacements which alter the side-chain dramatically, e.g. R265C, N263K, Q266P and G284D (Dell et al., 1994), and so it might be that not the amino acid replacements per se disturb sensing, but rather the introduced amino acids cause long-range structural and conformational changes. To investigate this hypothesis, we replaced arginine at position 265 with lysine, serine or glutamine and compared the effect of these mutations with the original replacement (cysteine). Moreover, the whole periplasmic domain of CadC was replaced with the periplasmic domain of ToxR from Vibrio harveyi for which to our knowledge no lysine sensing is known. Plasmids encoding the corresponding CadC derivatives were introduced into reporter strain E. coli EP314, which lacks a functional chromosomal cadC gene and carries a cadA–lacZ fusion gene (Neely et al., 1994). All mutants of CadC were able to induce cadBA expression (Fig. 3A) and were inserted into the membrane properly (Fig. 3B). Moreover, mutant CadC-R265C was the only one which supportedlysine-independent cadBA expression. Even the mutant CadC_ToxR which carries a replacement of the whole periplasmic domain mediated cadBA expression in a pH- and lysine-dependent manner (14-fold induction in the absence of lysine and 65-fold induction in the presence of lysine). These results reveal that the periplasmic domain is hardly the sensor domain for the lysine signal.
Dell et al. (1994) also isolated one lysine-independent mutant which carries an amino acid replacement within the transmembrane domain of CadC (CadC-G170D). In order to investigate the importance of the transmembrane domain of CadC for the lysine-dependent regulation, we replaced the transmembrane domain of CadC completely with a transmembrane domain (TM) of an unrelated membrane protein (TM4 or TM6 of PutP; Jung et al., 1998), resulting in CadC_PutP_TM4 and CadC_PutP_TM6. Mutants producing these hybrid proteins supported a pH-dependent but lysine-independent induction of cadBA expression indicating that sensing of the pH by the mutants remained intact whereas sensing of lysine did not (Fig. 4). It should be noted, that CadC_PutP_TM4 induced cadBA expression to almost the same level as did the wild type. The maximal induction level was generally lower in cells producing CadC_PutP_TM6 (Fig. 4). In a separate experiment it was demonstrated that both CadC derivatives were inserted into the membrane to the same extent as the wild-type protein (Fig. 5B). Thus differences in the expression level cannot be attributed to different amounts of CadC molecules inserted into the membrane. It should be noted, that chromosomally encoded CadC is hardly detectable in a Western blot (about 20 copies). Therefore, visualization of CadC requires overproduction of the corresponding proteins, whereas complementation of a ΔcadC mutant requires a low number of molecules to ensure a regulation as observed in the wild type (Küper and Jung, 2005). Remarkably, both mutants with an exchanged transmembrane domain were still able to respond to cadaverine, which downregulates cadBA expression. These results indicate that the transmembrane domain rather than the periplasmic domain is important for the lysine-dependent cadBA expression. In accord with the in vitro results, the inhibitor cadaverine binds to the periplasmic domain, and downregulates activity of both CadC derivatives with a replaced transmembrane domain.
The importance of a cluster of aromatic amino acids within the transmembrane domain of CadC for lysine-dependent cadBA expression
The transmembrane domain of CadC contains a cluster of aromatic amino acids (Fig. 1) which is very unusual for transmembrane helices. In order to investigate if this cluster plays a role in lysine-dependent cadBA expression, CadC variants with alterations within this cluster were tested. When the whole cluster of aromatic amino acids was deleted or replaced with nonaromatic amino acids (CadC_ΔF159-165 and CadC_F159L-F165A), cadBA expression of the corresponding mutants became lysine-independent to approximately the same degree as the CadC derivatives in which the whole transmembrane helix was replaced (Fig. 5A). The values presented indicate the degree to which cadBA expression became lysine-independent after measurement of cadBA expression at pH 5.8 in the absence and presence of 10 mM lysine and calculating the ratio thereof.
A detailed view on the cluster of aromatic amino acids allows the recognition of two aromatic patterns, F159,W160 and W162,F163,F164,F165, which are separated by Val161 (Fig. 1). Subsequently, both patterns were investigated in more detail by replacement of the corresponding amino acids. This resulted in mutants CadC_F159L,W160V (first pattern) and CadC_W162A-F165A (second pattern). Whereas the CadC mutant with a replacement of the first aromatic pattern regulated cadBA expression in a lysine-dependent manner, a replacement of the second pattern caused a lysine-independent phenotype (Fig. 5A).
Finally, CadC mutants with single amino acid replacements within the second aromatic pattern were investigated. cadBA expression in cells producing CadC_F165A was found to be lysine-independent to the same degree as in cells producing CadC derivatives in which all amino acids of the cluster were replaced. cadBA expression of cells producing CadC_W162A was lysine-independent as well, but to a minor degree (Fig. 5A). The determined β-galactosidase activities of all mutants under inducing conditions were in the range of those determined for cells producing wild-type CadC. Moreover, all CadC derivatives were found to be inserted into the membrane (Fig. 5B). These results reveal that a replacement of amino acids within the second pattern of aromatic amino acids, specifically of F165, is sufficient to convert CadC into a pure pH sensor.
The effect of LysP on CadC and its derivatives
It is known that LysP overproduction prevents cadBA expression and thus inhibits CadC activation under inducing conditions (Neely et al., 1994). Consequently, we tested whether LysP is still able to exert an effect on CadC derivatives with replacements in the transmembrane domain. These CadC mutants were analysed for their ability to activate PCad under simultaneous overproduction of LysP. Overproduction of LysP was achieved by using the pBAD expression system (Guzman et al., 1995), and Western blot analysis revealed equal amounts of LysP in all mutants (data not shown). cadBA expression in wild-type cells was inhibited by simultaneous LysP overproduction to 63% (Table 1). In cells producing CadC derivatives with modifications in the transmembrane domain, inhibition of cadBA expression by LysP was overridden to various degrees. Replacement of the whole transmembrane domain in CadC (CadC_PutP_TM4 and CadC_PutP_TM6) completely desensitized the corresponding mutants against the effect of LysP, as indicated by an almost unchanged level of cadBA expression (Table 1). The same was true for a CadC derivative with a deletion of a cluster of aromatic amino acids (CadC_ΔF159-165). LysP overproduction was also less effective in mutants producing CadC derivatives with a replacement of the cluster of aromatic amino acids (CadC_F159L-F165A) or replacements involving the second aromatic pattern (CadC_W162A-F165A and CadC_F165A). In general, it can be stated that the greater the lysine-independence of the mutants (Fig. 5A), the lower was the effect of LysP overproduction (Table 1).
Table 1. Influence of LysP overproduction on the ability of CadC and CadC derivatives to activate cadBA expression in the presence or absence of lysine.
Degree of inhibition (%)
Without LysP overproduction
With LysP overproduction
Reporter gene assays were performed with E. coli EP314 (ΔcadC; cadA::lacZ fusion) which was transformed with plasmid-encoded cadC or the indicated cadC derivative and co-transformed with a second plasmid carrying the lysP gene. Cells were cultivated under microaerophilic conditions in minimal medium at pH 5.8 in the presence or absence of 10 mM lysine. Overproduction of LysP was initiated by addition of 0.2% (w/v) arabinose. The activity of the reporter enzyme β-galactosidase was determined according to Miller (1992). It is given in Miller units together with the standard deviation and serves as a measurement for cadBA expression. Data were obtained from at least three independent experiments with two replicates in each case.
Cells were grown at pH 5.8 in the presence of lysine
180 ± 48.0
67.0 ± 22.5
243 ± 68.8
211 ± 72.7
86.5 ± 21.2
77.2 ± 21.4
244 ± 20.5
236 ± 42.2
253 ± 37.5
203 ± 49.9
330 ± 16.6
220 ± 42.6
303 ± 66.0
180 ± 48.5
242 ± 66.8
183 ± 63.0
230 ± 26.9
91.1 ± 35.8
Cells were grown at pH 5.8 in the absence of lysine
2.0 ± 0
2.4 ± 0
257 ± 67.6
250 ± 73.2
82.7 ± 19.4
86.7 ± 20.6
234 ± 30.7
205 ± 55.3
168 ± 45.2
64.5 ± 18.2
185 ± 22.0
47.5 ± 11.4
36.2 ± 11.1
7.9 ± 3.0
192 ± 36.2
45.6 ± 15.8
357 ± 79.8
117 ± 60.5
The influence of LysP overproduction on the CadC mutants was also investigated when external lysine was absent. This condition represents the uninduced state for which a strong inhibitory effect of LysP on CadC activity is predicted. Consequently, cadBA expression was repressed to nearly 100% in wild-type cells even without LysP overproduction (Table 1), and thus only those CadC mutants could be tested that supported a lysine-independent cadBA expression. Mutants CadC_PutP_TM4, CadC_PutP_TM6 and CadC_ΔF159-165 exhibited no decrease in cadBA expression by LysP overproduction, suggesting that LysP had completely lost its regulatory influence on these CadC derivatives (Table 1). In the other CadC mutants cadBA expression was downregulated by LysP overproduction up to 78% (CadC_W162A) but never completely as observed in the wild type.
These data reveal an interaction between CadC and LysP via the transmembrane domains which is stronger in the absence of external lysine. To strengthen this hypothesis, a CadC derivative with an intact transmembrane domain and extremely shortened periplasmic domain was studied (CadC 1–215). cadBA expression of the corresponding mutant was inhibited by LysP overproduction to the same degree of about 60% as the wild type in the presence of lysine (Table 1). It should be noted that CadC1–215 is only able to activate cadBA expression when overproduced, probably owing to a diminished stability of the protein without the intact periplasmic domain. The effect of a reduced stability by mutations in the periplasmic domain was also observed for the closely related ToxR from Vibrio cholerae (Pfau and Taylor, 1988). However, overproduced CadC (data not shown) or CadC1–215 support stimulus-independent cadBA expression which explains induction of cadBA by CadC1–215 in the absence of lysine (Table 1). Moreover, simultaneous overproduction of LysP and CadC1–215 was regulated by the same expression system. Therefore, equal production of both proteins was confirmed by Western blot analysis (data not shown). Importantly, regardless of the presence or absence of external lysine, mutant CadC1–215 was significantly influenced by LysP overproduction. In addition to that, CadC_ToxR, the CadC derivative with the exchanged periplasmic domain, was found to be extremely sensitive towards LysP. An elevated copy number due to the plasmid-encoded LysP was sufficient to completely inhibit CadC_ToxR dependent cadBA expression (data not shown). In contrast, a plasmid-encoded defective LysP did not result in a loss of activity of CadC_ToxR. The results for the two CadC derivatives in which the periplasmic domain was altered but the transmembrane domain remained intact (CadC1–215 and CadC_ToxR) underline the necessity of an intact transmembrane domain of CadC for the interaction with LysP.
Cross-linking of purified CadC and/or LysP in proteoliposomes
In order to analyse a potential physical interaction between CadC and LysP, cross-linking studies were performed with proteoliposomes containing either CadC or LysP alone or co-reconstituted CadC and LysP using the amino-specific cross-linker disuccinimidyl suberate (DSS). As a negative control proteoliposomes co-reconstituted with CadC and the proline transporter PutP were used. Following the cross-linking procedure, formation of high molecular weight complexes was observed in proteoliposomes co-reconstituted with CadC and LysP; however, these complexes hardly entered the sodium dodecyl sulphate (SDS) gel (Fig. 6). Concurrently, the protein bands corresponding to the monomers declined. Remarkably, CadC and LysP were cross-linked to a higher degree in comparison to the degree of cross-link found for each molecule alone. In contrast, no cross-linking was observed with proteoliposomes co-reconstituted with CadC and PutP. These results provide first evidence for a direct interaction between CadC and LysP.
CadC combines two functions in one polypeptide: it is a sensor for environmental stimuli and a transcriptional activator. Earlier, the sensory function was ascribed to the periplasmic domain of CadC, because a screen for mutants which induced cadBA expression independently of the presence of lysine and/or low pH unraveled CadC derivatives with single amino acid replacements in the periplasmic domain (Dell et al., 1994). However, it was still unclear how the external stimuli are sensed by CadC and how the signals are transduced over the membrane. For perception of the exogenous lysine signal an additional role for the lysine permease LysP was proposed, because insertional mutations within lysP conferred a lysine-independent cadBA expression (Neely et al., 1994), and LysP overproduction reduced wild-type CadC-mediated cadBA expression under otherwise inducing conditions (Neely et al., 1994). The presented work focuses solely on the question how CadC senses the external lysine concentration.
Based on published data, two hypotheses of how CadC could sense the lysine signal were conceivable: (i) CadC senses lysine directly, and LysP with its high affinity for lysine (KM = 10 μM, Rosen, 1971) acts as a scavenger for this substrate. The competition of both proteins for the same substrate leads to a scenario according to which CadC can only be active when it also has a high affinity for lysine or at sufficient high external lysine concentrations; (ii) CadC is not a sensor for lysine, and the lysine signal is perceived solely by LysP and transduced to CadC via alterations in its interaction with LysP. Furthermore, based on the repressing effect of an overproduction of LysP on cadBA expression, it could be proposed that under lysine-limiting conditions there is a strong interaction between CadC and LysP, which is released when sufficient lysine is available. It was the aim of this study to distinguish between these two models to get a better understanding about the molecular mechanism of lysine sensing by the transcriptional activator CadC.
The essential prerequisite for a lysine sensor is its ability to bind lysine. Therefore, we used purified full-length CadC or the purified periplasmic domain of CadC (CadC188–512) to determine its affinity for lysine. Application of ITC for measuring the affinity of full-length CadC turned out to be extremely difficult because of a very low signal intensity which was at the detection limit. The measurements with CadC revealed a KD value of about 22 mM lysine. Alternatively, we utilized the effect of ligands on tryptophane fluorescence of CadC188–512 to determine its apparent affinity. Besides lysine we tested cadaverine, the product of lysine decarboxylation which exerts a feedback inhibition on the Cad system (Neely et al., 1994). The concentration-dependent alterations of tryptophane fluorescence indicated an affinity of the periplasmic domain for cadaverine of about 100 μM which is in the range of the affinity determined e.g. for binding of trimethylamine N-oxide by the periplasmic binding protein TorT in E. coli (KD = 150 μM; Baraquet et al., 2006). In contrast, lysine did induce only a minor conformational change in CadC even at a low pH which is in agreement with the weak ITC signal. The unexpectedly low affinity of CadC for lysine suggested that CadC can only be activated at high external lysine concentrations. Therefore, we tested whether cadBA expression is proportional to the external lysine concentration. We found that cadBA expression reached a maximal level when a certain threshold concentration was attained (5 mM), indicating that CadC activity is not proportional to the available lysine concentration. Taken together, these results led to the conclusion that CadC is not a direct lysine sensor. Consequently, the lysine-dependent regulation of the transcriptional activator is mediated by the lysine permease LysP.
Investigation of various replacements of Arg265 in CadC as well as studies with a CadC mutant in which the whole periplasmic domain was replaced with the corresponding domain of ToxR (CadC_ToxR) revealed that the periplasmic domain of CadC is not directly involved in the lysine-dependent activation of cadBA expression. Instead, the described effect of the substitution R265C (Neely et al., 1994) could be a result of an additional cysteine which might interfere with the interactions of the three native cysteines (C172, C208, C272) of CadC and thereby might alter the conformation of CadC. An oxidation-induced aggregation of CadC-R265C was not observed. It cannot be excluded that parts of the periplasmic domain interact with the periplasmic loops of LysP and thereby influence the regulatory process to a minor extent because cadBA expression of a mutant producing CadC_ToxR was not completely downregulated in the absence of lysine. It should be mentioned that nothing is known about the stimuli perceived by ToxR of V. harveyi but a lysine-dependent regulation seems rather unlikely considering the results for ToxR of V. cholerae (Skorupski and Taylor, 1997).
As LysP is a membrane-integrated protein with 12 transmembrane domains (Ellis et al., 1995), it was tempting to speculate that it exerts its effect on the transmembrane domain of CadC. In order to test this prediction, two CadC derivatives, CadC_PutP_TM4 and CadC_PutP_TM6, were constructed in which the transmembrane domain was substituted with transmembrane helices of PutP, the Na+/proline permease of E. coli (Jung et al., 1998). These CadC derivatives supported a pH-dependent but completely lysine-independent cadBA expression. It should be noted, that the β-galactosidase activity of cells producing CadC_PutP_TM4 were in the range of wild-type CadC, but the activity was significantly reduced in case of CadC_PutP_TM6. The reduced activity of this derivative might be explained with a restricted conformational flexibility of TM6 of PutP which contains a proline residue within the helix (Jung, 1998). Remarkably, both mutants with an exchanged transmembrane domain induced cadBA expression regardless of the availability of lysine, but cadaverine still had a downregulating effect. These data are consistent with the in vitro results according to which only for cadaverine an affinity in the micromolar range was detectable whereas the affinity for lysine was very low.
Taken together, these results allowed the conclusion that the transmembrane domain of CadC is involved in the lysine-dependent regulation of CadC, and consequently, the transmembrane domain was examined in more detail. Interestingly, this domain contains six clustered aromatic amino acids (Fig. 1). A similar motif was not found in any other membrane protein as revealed by a Blast search (Altschul et al., 1990). A detailed analysis of this motif seemed promising as aromatic residues have already been shown to mediate the self-assembly of soluble proteins through π–π interactions of the planar aromatic rings and also seem to promote the dimerization of transmembrane proteins (Sal-Man et al., 2007). The detailed analysis of CadC mutants in which this cluster of aromatic amino acids was replaced or deleted revealed its importance for lysine-dependent cadBA expression. Moreover, within the cluster two patterns of aromatic amino acids separated by V161 were found. Individual replacements of the two patterns and single amino acid replacements narrowed the residues responsible for the lysine-dependent phenotype clearly to the second aromatic pattern, specifically phenylalanine at position 165. This result proposes a role of a single aromatic amino acid for the mediation of protein–protein interactions between transmembrane domains as it is also discussed by Sal-Man et al. (2007). They demonstrated the importance of aromatic residues for homodimerization of biological transmembrane domains. In case of CadC aromatic residues might be involved in the mediation of heterooligomerization of transmembrane domains of different proteins. In support of this proposal, each transmembrane helix of LysP contains at least one aromatic residue that might serve as a partner for F165 of CadC in the heterodimerization.
To strengthen the role of the transmembrane domain of CadC as a communication tool with LysP, we tested the capability of CadC derivatives with altered transmembrane domains to induce cadBA expression under conditions of concurrent LysP overproduction. It is known that LysP overproduction represses cadBA expression in the wild type under otherwise inducing conditions (Neely et al., 1994). As expected, LysP overproduction was almost without effect in cells producing CadC derivatives with a replaced transmembrane domain (CadC_PutP_TM4, CadC_PutP_TM6), or with the deletion of the cluster of aromatic amino acids (CadC_Δ159-165). In these cells cadBA was only minimally repressed. Moreover, for all other CadC derivatives we found a reverse relation between lysine-independent cadBA expression and the effect of LysP overproduction. Specifically, the more these derivatives supported lysine-independent cadBA expression, the less was the effect of LysP. Our results for the truncated CadC derivative CadC1–215 that lacks most of the periplasmic domain and for CadC_ToxR in which the periplasmic domain was replaced by the corresponding domain of ToxR are in support of this hypothesis. cadBA expression was inhibited by LysP overproduction in cells producing these derivatives, indicating that not the periplasmic domain but rather the transmembrane domain is responsible for this regulatory effect.
Based on these results we propose an interaction between CadC and LysP via the membrane domains to sense the external lysine concentration (Fig. 7). Our model suggests that in the absence of lysine LysP inhibits CadC by an interaction with its transmembrane domain. In the presence of lysine LysP acts as a lysine permease and somehow loses its ability to interact and thereby inhibit CadC. The copy number of CadC is below 20 copies (our own observation). The number of LysP copies is unknown thus far. As overproduction of CadC results in constitutive cadBA expression and overproduction of LysP prevents cadBA induction, the stoichiometry of CadC and LysP seems to be very sensitive and important for proper regulation. Cross-linking studies with the amino-specific cross-linker DSS provided first evidence for a physical interaction between CadC and LysP (Fig. 6).
As it is still open which transmembrane helix of LysP might interact with CadC, it is interesting to note that LysP belongs to the sort of amino acid symporters that contain 12 transmembrane helices whereof only the first 10 have an internal symmetry. The function of the two additional transmembrane helices (TM11 and TM12) is thus far unknown (Jung et al., 2007). These two additional transmembrane helices might be responsible for the interaction with CadC, a hypothesis which we will address in future studies. Finally, it cannot be excluded that additional factors are involved in this regulatory process that have not been elucidated to date. However, the existence of additional factors does not seem necessary to explain the mechanism of the lysine-dependent regulation of the Cad system, taking into account that the phenotype of a LysP-deficient strain (Neely et al., 1994) and our results obtained with different CadC derivatives and LysP overproduction are sufficient to explain sensing of the external lysine concentration.
In summary, the presented data clearly show that CadC is not a direct sensor for lysine. The lysine-dependent regulation is mediated by an interplay of CadC and LysP presumably via the transmembrane domains. These results fit well in the increasing line of evidences about the coupling between transcriptional regulation and transport processes via protein–protein interactions. Known examples are the interplay between the virulence regulator PfrA and the phosphotransferase system (PTS) of Listeria monocytogenes (Marr et al. 2006), as well as between the global transcriptional regulator Mlc and the glucose transporting subunit of the PTS protein PtsG (Lee et al., 2000).
Bacterial strains and growth conditions
Strains and plasmids are listed in Tables 2 and 3. E. coli JM109 was used as carrier for the plasmids described. E. coli BL21(DE3) pLysS was used for expression of cadC and cadC variants from the T7-promoter and for the expression of lysP from the lac promoter, and E. coli Origami B (DE3) pLysS was used for the expression of trx-cadC188–512. E. coli EP314 carries a cadA–lacZ fusion gene and a deletion of cadC. This strain was complemented with plasmids (pET16b) encoding cadC and its derivatives, and used for cadBA transcriptional analysis. Overproduction of LysP was performed in E. coli EP314 transformed with plasmid pBAD33-lysP by inducing the arabinose promoter with 0.2% (w/v) arabinose. E. coli strains were grown on Luria–Bertani (LB) medium (Maniatis et al., 1982) for strain maintenance and protein overproduction. To probe signal transduction in vivo, cells of E. coli EP314 transformed with the indicated plasmids were grown in minimal medium (Epstein and Kim, 1971); the phosphate buffer of the medium was adjusted to either pH 5.8 or pH 7.6. Lysine was added at a concentration of 10 mM, and cadaverine at a concentration of 1.3 mM. For selection of plasmid containing cells appropriate antibiotics were added at concentrations of 100 μg ml−1 (ampicillin), 50 μg ml−1 (kanamycin) and 34 μg ml−1 (chloramphenicol).
Single amino acid exchanges are described by giving the native amino acid following the one letter code in combination with the position in the primary protein sequence and the amino acid after exchange.
All cadC variants were constructed by polymerase chain reaction (PCR) with mismatch primers either by single-step or by two-step PCR (Ho et al., 1989). To facilitate construction, two unique restriction sites [SpeI: G→A (524), BssHII: T→C (803) and T→G (806), numbers indicate the nucleotide positions in the cadC sequence] were introduced into plasmid pET16b-CadC by silent mutation resulting in plasmids pET16b-cadC1 (BssHII) and pET16b-cadC2 (BssHII, SpeI). Subsequently, cadC variants were constructed. In order to obtain CadC with a replaced transmembrane domain, the part of putP encoding the respective transmembrane domain (domains 4 and 6) was amplified by PCR. The PCR products were spliced to the fragments encoding the cytoplasmic and periplasmic domain of CadC by splicing-by-overlapPCR (SOE-PCR, Ho et al., 1989) and cloned into pET16b-cadC1 resulting in pET16b-cadC_PutP_TM4 and pET16b-CadC_PutP_TM6. In order to obtain CadC with a replaced periplasmic domain, the part of toxR encoding the periplasmic domain was amplified by PCR from genomic DNA of Vibrio harveyi BB120. The PCR products were spliced to the fragment encoding cytoplasmic and transmembrane domain of CadC by SOE-PCR (Ho et al., 1989).
CadC_ΔF159-165 carries a deletion of amino acids 159–165, and in CadC_F159L-F165A the aromatic amino acids of this cluster were exchanged as follows: F159L, W160V, W162A, F163L, F164V, F165A. CadC_F159L,W160V carries the replacements F159L and W160V, and CadC_W162A-F165A carries the exchanges W162A, F163L, F164V and F165A. In addition to this, the single amino acid replacements W162A, F163L, F164V and F165A were introduced. All site-specific mutations were directed by synthetic oligonucleotide primers containing the required nucleotide exchanges. PCR fragments were cloned into the expression vector pET16b with the restriction enzymes NdeI and BamHI so that all constructs carry an N-terminal His-Tag consisting of 10 histidine residues. The truncated derivative CadC1–215 was obtained by inserting three stop codons after triplet 215 in cadC. This construct also encodes an N-terminal His-Tag and is expressed under the control of the arabinose promoter (plasmid pBAD24). The corresponding nucleotides encoding the soluble periplasmic domain, CadC188–512, were amplified by PCR and cloned into vector pET32a (LaVallie et al., 1993) with the restriction enzymes NcoI and BamHI resulting in a hybrid protein consisting of thioredoxin, an N-terminal His-Tag and CadC188–512.
Detection of CadC derivatives in the membrane fraction
Cells of E. coli BL21(DE3) pLysS transformed with the indicated plasmids were grown to an optical density at 600 nm (OD600) of 0.5 in LB medium (Maniatis et al., 1982). Subsequently, gene expression was induced by addition of 0.5 mM IPTG. After growth for 3 h, cells were harvested. Membrane vesicles were prepared essentially as described (Jung et al., 1997). Protein was assayed by the method described by Peterson (1977) using bovine serum albumin as standard. Proteins were separated by SDS-PAGE (Laemmli, 1970) using 12.5% acrylamide gels. For Western blot analysis, protein was blotted onto a nitrocellulose membrane, and probed with monoclonal anti-His antibodies (Qiagen). Immunodetection was colorimetrically performed using a goat anti-(mouse IgG)-alkaline phosphatase antibody.
Measurement of CadC signal transduction activity in vivo
Signal transduction activity of different CadC derivatives in vivo was probed with a β-galactosidase based reporter gene assay. In order to achieve a regulation as in the wild type, it was important to use an expression system for the complementation studies with the CadC derivatives that does not result in an overproduction but allows for a production similar to that in the wild type. Using a pET-based vector in combination with the reporter strain E. coli EP314 that does not posses a T7 polymerase resulted in a low expression that was nevertheless sufficient to allow the complementation (Küper and Jung, 2005).
Cells carrying the appropriate plasmids of an overnight culture (grown in minimal medium under non-inducing conditions) were inoculated into fresh minimal medium (pH 5.8 or pH 7.6, supplemented with lysine or cadaverine as necessary), and cell density resulting in an OD600 of 0.05 was adjusted. Cells were grown under microaerophilic conditions at 37°C to mid-logarithmic growth phase, and harvested by centrifugation. To test β-galactosidase activity, cells were re-suspended in sodium phosphate buffer (100 mM sodium phosphate buffer, pH 7.0, 10 mM KCl, 1 mM MgSO4, 50 mM β-mercaptoethanol). An aliquot of this suspension was used to determine the OD600, and another aliquot was used for determination of β-galactosidase activity after cells were permeabilized with chloroform and SDS. β-Galactosidase activity was determined from at least three different experiments, and is given in Miller units calculated as described by Miller (1992).
The specific activity of the lysine decarboxylase CadA as a measurement for cadBA expression was determined according to Küper and Jung (2005), and is given as μmol min−1 (mg protein)−1 .
Production and purification of CadC188–512
The periplasmic domain of CadC was overproduced in E. coli Origami B (DE3) pLysS pET32a-CadC188–512 as a thioredoxin hybrid protein containing an N-terminal His-Tag to facilitate folding, increase solubility and allow easy purification of the protein. Cells were cultivated in LB medium (Maniatis et al., 1982) at 30°C and gene expression was induced by addition of 0.1 mM IPTG. After 3 h cells were harvested by centrifugation, and cells were disrupted by passage through a high pressure cell disrupter (Constant Systems) in 50 mM sodium phosphate buffer, pH 7; 1 mM EDTA; 1 mM PMSF and 30 μg ml−1 DNase. After subsequent ultracentrifugation (1 h, 244 000 g, 4°C), the cytosolic fraction (supernatant) was obtained. CadC188–512 was purified by using Ni2+-NTA affinity chromatography in batch. The Ni2+-NTA resin was pre-equilibrated with buffer A (50 mM sodium-phosphate, pH 7; 250 mM NaCl, 30 mM imidazole). The cytosolic fraction was supplemented with 30 mM imidazole and 250 mM NaCl, and incubated with the Ni2+-NTA resin (1.5 ml resin per 50 ml cytosol) at 4°C for 45 min. The resin–protein complex was washed several times with buffer A containing 50 mM imidazole. CadC188–512 was eluted with buffer A containing 250 mM imidazole. The thioredoxin-CadC188–512 hybrid protein was cleaved by treatment with biotinylated thrombin using the Thrombin Cleavage Capture Kit according to the instructions of the supplier (Novagen).
Cross-linking studies of CadC and LysP
Overproduction of His10-CadC and LysP-His6 was performed with E. coli BL21(DE3) pLysS transformed with the plasmids pET16b-cadC and pT-lysP. Overexpression and production of membrane vesicles were performed as described above. CadC and LysP were purified essentially as described in Küper and Jung (2005) except that LysP was extracted by solubilization of the inverted membrane vesicels at a concentration of 5 mg protein ml−1 and that instead of lauryldimethylamine-oxide n-dodecyl-β-d-maltoside in a final concentration of 1% (w/v) was used as detergent. Buffers used for affinity chromatography were changed to 2 mM β-mercaptoethanol; 300 mM NaCl, 0.04% (w/v) n-dodecyl-β-d-maltoside and 30 mM imidazole (CadC) or 15 mM imidazole (LysP) respectively. The solubilized proteins were supplemented with NaCl (final concentration: 300 mM) and imidazole (final concentration: 30 mM for CadC and 15 mM for LysP). PutP was purified as described in Jung et al. (1998). Purified CadC, LysP and PutP were reconstituted either separately or together in equimolar amounts into E. coli phospholipids in a ratio of 1:25 essentially as described in Küper and Jung (2005). The pellet was re-suspended in 50 mM Tris/HCl, pH 7.5, 10% (v/v) glycerol. For cross-linking experiments the buffer was exchanged to 20 mM sodium phosphate, pH 7.2; 10% (v/v) glycerol by washing the proteoliposomes several times. Proteoliposomes were either used instantly or stored in liquid nitrogen. Cross-linking studies with DSS were performed as follows: Proteoliposomes containing either LysP or CadC alone or CadC + LysP or CadC + PutP (each at a molar concentration of 5 μM) were incubated with a 10-fold excess of DSS at room temperature for 15 min. The reaction was stopped by adding 50 mM Tris/HCl, pH 7.5 and subsequent incubation for 15 min. Proteins were separated by SDS-PAGE (Laemmli, 1970) using 7.5% acrylamide gels and subsequently stained with Coomassie blue (Weber and Osborn, 1969).
Fluorescence was measured using a FluoroMax-3 spectrofluorometer (Horiba Jobin Yvon GmbH) at 25°C. The excitation wavelength was set to 287 nm, and emission was recorded between 300 and 500 nm with a slit width of 2 nm for excitation and 4 nm for emission. Measurements were carried out with 0.1 mg ml−1 CadC188–512 in a total volume of 250 μl in 150 mM sodium phosphate buffer (pH 7 or pH 5.8) and 150 mM NaCl. The samples were dialysed against the reaction buffer before the experiment. Ligands (solved in dialysis buffer) were added at the indicated concentrations, and the mixture was incubated for 5 min at 25°C prior to measurement. All measurements were performed a minimum of three times. To correct unspecific effects the same experiments were performed with spermidine. Changes in fluorescence are expressed as ΔF/F, where ΔF is the change in fluorescence intensity at 337 nm upon addition of ligand at the indicated concentration and F is the initial fluorescence intensity measured without ligand. ΔF/F-values obtained for spermidine were substracted from the corresponding values obtained for cadaverine. The affinity constant KD was calculated from the slope of the plot of ΔF/F (= bound ligand) against (ΔF/F)/(L) (= quotient of bound and free ligand) according to Scatchard (1949).
Isothermal titration calorimetry
For ITC measurements, full-length CadC was produced, solubilized and purified as described (Küper and Jung, 2005). Prior to measurements, solubilized CadC was dialysed against 50 mM sodium phosphate buffer, pH 5.8, 10% (v/v) glycerol, 10 mM β-mercaptoethanol, 600 mM NaCl and 0.2% (v/v) lauryldimethylamine-oxide. The ITC measurements were performed with a VP-ITC-Microcalorimeter (MicroCal). CadC at a concentration of 16 μM (total volume of 1.4 ml) was stepwise titrated with lysine (10 μl per titration, 29 times), which was dissolved in dialysis buffer at a concentration of 10 mM. The measurements were performed at 20°C, while stirring with 310 r.p.m. The affinity constant KD was obtained by analysing the raw ITC data with the software Origin supplied with the instrument.
This work was supported by the Deutsche Forschungsgemeinschaft (JU270/5–1 and Exc114/1). We are thankful to L. Moroder and his group, MPI for Biochemistry (Munich), for providing access to the ITC instruments, and D. Hilger, LMU Munich, for preparing purified PutP.