Molecular Design of a Signaling System Influences Noise in Protein Abundance under Acid Stress in Different Gammaproteobacteria

ABSTRACT

Bacteria have evolved different signaling systems to sense and adapt to acid stress. One of these systems, the CadABC system, responds to a combination of low pH and lysine availability. In Escherichia coli, the two signals are sensed by the pH sensor and transcription activator CadC and the cosensor LysP, a lysine-specific transporter. Activated CadC promotes the transcription of the cadBA operon, which codes for the lysine decarboxylase CadA and the lysine/cadaverine antiporter CadB. The copy number of CadC is controlled translationally. Using a bioinformatics approach, we identified the presence of CadC with ribosomal stalling motifs together with LysP in species of the Enterobacteriaceae family. In contrast, we identified CadC without stalling motifs in species of the Vibrionaceae family, and the LysP cosensor is missing. Therefore, we compared the outputs of the Cad system in single cells of the distantly related organisms E. coli and Vibrio campbellii using fluorescently tagged CadB as the reporter. We observed a heterogeneous output in E. coli, and all the V. campbellii cells produced CadB. The copy number of the pH sensor CadC in E. coli was extremely low (≤4 molecules per cell), but it was 10-fold higher in V. campbellii. An increase in the CadC copy number in E. coli correlated with a decrease in heterogeneous behavior. This study demonstrated how small changes in the design of a signaling system allow a homogeneous output and, thus, adaptation of Vibrio species that rely on the CadABC system as the only acid resistance system.

IMPORTANCE Acid resistance is an important property for bacteria, such as Escherichia coli, to survive acidic environments like the human gastrointestinal tract. E. coli possesses both passive and inducible acid resistance systems to counteract acidic environments. Thus, E. coli evolved sophisticated signaling systems to sense and appropriately respond to environmental acidic stress by regulating the activity of its three inducible acid resistance systems. One of these systems is the Cad system, which is induced only under moderate acidic stress in a lysine-rich environment by the pH-responsive transcriptional regulator CadC. The significance of our research lies in identifying the molecular design of the Cad systems in different proteobacteria and their target expression noise at the single-cell level during acid stress conditions.

INTRODUCTION

Bacteria are exposed to changing environments, which include exposure to life-threatening compounds, such as antibiotics, or acidic conditions (1). Therefore, bacteria have evolved sophisticated signaling systems to sense and respond to environmental alterations and to change their behavior and physiological state to protect themselves against stressful conditions. In addition to a uniform response to environmental stress conditions, different phenotypes can emerge within genetically identical bacterial populations, such as formation of persisters (2), protection against antibiotics (3), or response to nutrient limitations (4); this phenomenon is called phenotypic heterogeneity (5, 6).

Acid resistance (AR) is an important property of Escherichia coli and other enterobacteria that enables the organism to survive acidic environments, such as the human gastrointestinal tract or acidic soils (7, 8). E. coli possesses both passive and inducible AR systems to counteract acidic environments (7, 9). The three inducible AR systems of E. coli are the Cad system (AR4), the glutamate decarboxylase (Gad) system (AR2), and the arginine decarboxylase (Adi) system (AR3) (10–12). All three inducible AR systems rely on the activities of a proton-consuming reaction catalyzed by specific amino acid decarboxylases and the corresponding antiporters (9, 11, 13, 14).

The Cad system protects members of the Enterobacteriaceae against inorganic and organic acids in the intestinal tract (8) and against fermentation acids under phosphate-limiting conditions (8). This AR system is induced only under moderate acidic stress in a lysine-rich environment. The three principal components of the Cad system are the lysine decarboxylase CadA, the antiporter CadB, and the membrane-integrated pH sensor CadC (15). The CadC pH sensor, similar to other members of the ToxR family, combines sensory and output functions within one polypeptide, and signal transduction is mediated without chemical modification (16). These receptors are characterized by a modular structure as follows: a periplasmic sensory domain followed by a single transmembrane helix is connected via a flexible linker to a cytoplasmic DNA-binding domain (17). A decrease in external pH induces dimerization of the periplasmic sensory domain of CadC, thus enabling cadBA expression (18). There are two CadC-binding sites (CadC1 and CadC2) within the cadBA promoter, and each binding site is occupied by one CadC homodimer (18–20). CadA converts lysine under the consumption of a proton into the more alkaline cadaverine and carbon dioxide, thus increasing the intracellular pH. The CadB antiporter transports lysine into the cells and exports cadaverine, thereby increasing the extracellular pH. Two additional regulatory elements are responsible for fine-tuning the functionality of the Cad system in E. coli by controlling the activity and the copy number of CadC. First, LysP inhibits the activity of CadC under noninducing conditions, and its inhibitory effect is alleviated in the presence of external lysine (21, 22). Furthermore, CadC activity is inhibited by external cadaverine through additional negative feedback (18, 23). Second, production of CadC is dependent on the translation elongation factor P (EF-P) due to the presence of a polyproline (PPP) motif in its linker that causes a stalling arrest during translation. In wild-type cells, only an average of 1 to 3 CadC molecules per cell are present (24). Additionally, CadC itself is distributed heterogeneously (25).

In this study, we elucidated the influence of the molecular design of the Cad signaling systems on the noise of the target protein(s) at the single-cell level in two distantly related bacteria, namely, E. coli and Vibrio campbellii. Using a bioinformatics approach, we identified CadCs with ribosomal stalling motifs together with the presence of LysP in species of the Enterobacteriaceae family. In contrast, we identified CadC without stalling motifs in species of the Vibrionaceae family, and the LysP cosensor is missing. By fluorescently tagging the CadB antiporter of both organisms, we observed a heterogeneous output in E. coli, whereas all the cells of V. campbellii produced CadB. Overall, by reducing the complexity of the Cad system, a homogeneous output is possible, ensuring the acid resistance of all the cells in bacteria that have only one protection system, such as V. campbellii.

RESULTS

Phylogenetic distribution of CadC within gammaproteobacteria.

We used a bioinformatics approach to investigate the presence and design elements of CadC within the bacterial kingdom. We also focused on a potential coevolution of CadC and LysP within gammaproteobacteria and the presence of ribosome-stalling motifs (XPP or PPX) within the CadC sequences (Fig. 1a). According to our study, there are 3,223 homologs of E. coli CadC that are concentrated mainly in the Enterobacteriaceae family (72.9%) and distributed in the genera Salmonella (38.3%), Escherichia (20.3%), Klebsiella (11.6%), and Serratia (3.2%), with a sequence identity of >30%. CadC homologs are also present at a lower abundance in members of the Vibrionaceae family (17.1%), such as Vibrio (15.8%) and Photobacterium (1.3%), the Aeromonadaceae (4.9%), the Yersiniaceae (3.3%), and others (1.8%), with a sequence identity of <22% (Fig. 1a).

FIG 1.

Phylogenetic tree of CadC and cooccurrence with LysP and/or stalling XPP or PPX motifs in gammaproteobacteria. (a) The protein sequences of 3,223 CadC homologs were aligned, and a phylogenetic tree was generated and displayed as a circular cladogram. The branches of the tree are colored according to the family of the organisms containing a CadC homolog. The presence of an XPP or PPX motif is displayed in the outer blue/dark gray ring, and the presence/absence of LysP is displayed in the inner salmon/light gray ring. The CadC homolog sequences were examined for XPP or PPX motifs, which were grouped according to their stalling strength as follows: strongest PPP, dark blue; other strong XPP or PPX, blue; moderate/weak XPP or PPX, light blue; or no stalling motif, dark gray. (b) Percentage of CadC sequences containing strong stalling XPP or PPX, moderate/weak XPP or PPX, or no XPP or PPX stalling motif. Sequences and XPP or PPX motifs are summarized in Data Set S1 in the supplemental material. Arrows indicate the occurrence of CadC in the E. coli and V. campbellii clades.

Of note, 77.2% of the species that have a CadC homolog also contain a LysP homolog (Fig. 1a, inner ring, salmon), which is characteristic for the Enterobacteriaceae family, as the majority of these species of this family have CadC and LysP homologs (99.6%) (Fig. 1a).

CadC of E. coli contains a polyproline (PPP) motif, which causes strong ribosome stalling (24, 26). There is a hierarchy of arrest strength depending on the composition of the XPP or PPX motifs. Triplets, such as PPP, DPP, PPD, PPW, APP, GPP, PPG, PPE, PPS, and PPN, cause strong ribosome stalling, whereas L/PP/L, CPP, or HPP results in rather weak translational pauses (27). Our bioinformatics study revealed that 71.0% of the 3,223 CadC sequences have a strong stalling XPP or PPX motif, where the strong stalling PPP triplet is present in 64.2% of the sequences (Fig. 1b). The following strong stalling motifs are less frequent: APP (3.2%), PPS (1.5%), PPE (1.3%), GPP (0.6%), PPD (0.1%), and DPP (0.1%) (Fig. 1b). The strong stalling XPP or PPX motifs within CadC are mainly found within genera of the Enterobacteriaceae family (73.0%), such as Escherichia (97.9%), Salmonella (97.2%), and Klebsiella (99.6%), and these motifs are less frequent among the genera Vibrio (1.4%) and Aeromonas (3.7%). Moderately or weakly stalling XPP or PPX motifs are present in 16.4% of the CadC sequences, belonging mainly to Yersiniaceae (5.2%), Aeromonadaceae (4.4%), Enterobacteriaceae (Klebsiella spp. and Enterobacter spp.) (4.1%), and Vibrionaceae (2.0%). Furthermore, 12.6% of the CadC sequences lack a polyproline (polyPro) motif, including sequences in Vibrionaceae (10.8%), Enterobacteriaceae (0.6%), and Hafniaceae (0.4%) (Fig. 1a, outer ring, gray).

E. coli and Salmonella enterica CadC are examples that show a high conservation between strong polyPro triplets in CadC and the presence of LysP. CadC of E. coli strains contain PPP (97.6%), DPP (0.4%), or a weak triplet (2.0%), and CadC of S. enterica strains contain PPP (95.8%), DPP (0.3%), APP (1.1%), or a weak triplet (2.8%) (see Data Set S1 in the supplemental material). Of the strains possessing LysP, 87.5% contain a strong stalling polyPro motif within the CadC sequence. Examples of species that do not have a strong stalling motif in CadC but possess LysP are mainly Serratia, Edwardsiella, Klebsiella, Yersinia, and Enterobacter species and Escherichia albertii (Fig. 1a, outer ring, light blue and gray).

Overall, the prototypical design of the CadC signaling system, including its activity control by the LysP cosensor and translational control by a stalling motif, is a predominant feature of members of the Enterobacteriaceae family. In contrast, members of the Vibrionaceae family possess an apparently simpler CadC signaling system without the additional regulatory inputs by LysP and the polyPro motif.

Role of the Cad system in E. coli and V. campbellii.

Based on our phylogenetic analysis, we selected two distantly related organisms, E. coli and V. campbellii, as representatives for the comparison of two differently designed CadC signaling systems.

E. coli has, in addition to the Cad system, two other inducible AR systems, the glutamate decarboxylase (Gad) system and the arginine decarboxylase (Adi) system, which counteract acidic stress (10–12). However, the Cad system in V. campbellii is the only inducible AR mechanism (28), and V. campbellii (Fig. 2a) and Vibrio cholerae (28) generally have higher susceptibility to acid stress than E. coli or S. enterica serovar Typhimurium (29). The Cad system in V. campbellii is activated when this bacterium faces low pH in an amino acid-rich environment. The activation is lysine independent due to the lack of the lysine cosensor LysP. Nevertheless, CadA of V. campbellii functions as a lysine decarboxylase, similar to its function in enterobacteria (Fig. 2b). The Cad system of V. campbellii is important to counteract moderate acid stress, as the deletion of cadC, cadA, or cadB prevents V. campbellii from increasing the external pH (see Fig. S1a in the supplemental material).

FIG 2.

Acid resistance of V. campbellii and CadA decarboxylase activity. (a) Survival of E. coli and V. campbellii at pH 3.0 in complex medium (LB or LM, respectively). At the indicated times, the samples were collected, and the number of CFU was analyzed. (b) Lysine decarboxylase activity of purified CadA of V. campbellii (1 μg/ml) was tested in the presence of increasing lysine concentrations and fitted to the Michaelis-Menten equation (gray line) by nonlinear regression calculations (K_m = 4.7 ± 2.7 mM).

The Cad system is heterogeneously activated in E. coli but homogenously activated in V. campbellii.

We next studied the spatiotemporal output of the cad operon in single cells of both organisms using fluorescently tagged CadB as a reporter. The respective fusions of cadB and the enhanced green fluorescent protein (eGFP) gene (egfp) were chromosomally integrated and expressed by the native promoters. The hybrid proteins were inserted into the cytoplasmic membrane (Fig. S1b) and were fully functional in E. coli and V. campbellii, respectively. Furthermore, the acid stress adaptation of E. coli or V. campbellii cells expressing cadB-egfp was similar to that of the respective wild-type cells (Fig. S1a and c).

E. colicadB-egfp cells were exposed to acid stress (pH 5.8 in the presence of lysine), and the presence of CadB-eGFP was analyzed in single cells over time (Fig. 3). CadB-eGFP became visible within 1 h after exposure of cells to acid stress, and its production was heterogeneously distributed among single cells of the E. coli population. The highest production was observed at approximately 2.5 h, reaching a plateau, which is in perfect agreement with previous bulk experiments (15). At this time point, CadB-eGFP production displayed a high noise value of 0.304 (defined as the standard deviation divided by the mean of log-transformed intensity values [30–32]). Furthermore, the percentage of cells in the CadB-eGFP on state increased over the time course from 34.6% up to 70% (Fig. 3, inset, dark green bars; see Table S1 in the supplemental material). The distribution of CadB-eGFP among single E. coli cells under acid stress was not a symmetric Gaussian-like distribution but rather an asymmetric right-skewed distribution (see Fig. S2a in the supplemental material). This distribution suggested that only some cells produce CadB-eGFP at a varying high level and approximately one-third of the cells are in the off state (Fig. 3, inset). As a control, unstressed cells that do not express cadBA were examined. These cells exhibited low fluorescence and noise values (Fig. 3, no stress at 2.5 h), and these results were comparable to those obtained with the nontagged E. coli MG1655 cells (Table S1).

FIG 3.

Heterogenous CadB production in E. coli in response to acid stress. At time zero, E. coli was shifted from pH 7.6 to pH 5.8 in the presence of lysine. CadB-eGFP intensity was quantified in single E. coli cells over time (upper graph; the gray line indicates background fluorescence). Representative fluorescence and phase-contrast images of E. coli cells containing CadB-eGFP are shown. As a control, E. coli cells containing CadB-eGFP were analyzed under nonstress conditions (pH 7.6, 2.5 h, first column). The noise and mean relative fluorescence intensity (RF) were calculated for 1,000 cells per condition and time point. Single-cell fluorescence intensity was quantified by microscopy and the use of the ImageJ software. The inset graph illustrates the distribution of CadB-eGFP (percentage of fluorescence) among single cells (gray, cells in the off state; dark green, cells with a high CadB level [on state]) (see Table S1 in the supplemental material). Noise, standard deviation/mean of log-transformed values; PH, phase contrast; GFP, green fluorescent channel. Scale bar, 5 μm. Arrows mark exemplary cells in the off state.

Second, we determined the spatiotemporal output of the Cad system in single V. campbellii cells. As described above, V. campbellii contains a simpler design of the Cad system, as LysP is absent and CadC has only a weakly stalling motif (PPI). Similarly to V. cholerae (30), V. campbellii activates the Cad system after exposure to pH 5.8 in an amino acid-rich environment (Fig. 4). The distribution of CadB-eGFP was analyzed in single V. campbellii cells over time. CadB-eGFP fluorescence became detectable within 1 h after exposure to acid stress, and the intensity constantly increased over time and reached a maximum after 2.5 h (Fig. 4). CadB-eGFP was homogeneously distributed among V. campbellii single cells under acid stress conditions with a symmetric Gaussian distribution (Fig. S2b). The percentage of cells with CadB-eGFP in the on state increased already after 1 h to 76% and reached a plateau of 95% over the time course (Fig. 4, inset, dark green bars; see Table S2 in the supplemental material). The expression of CadB-eGFP in acid-stressed cells was characterized by low noise values of 0.1 to 0.14, which were in the same range as the noise values of unstressed cells (Fig. 4 and Table S2). The noise levels were slightly increased (0.19 compared to 0.1) at times when the mean CadB-eGFP production was very low. It was observed previously that noise becomes higher with lower gene expression (31).

FIG 4.

Homogeneous CadB production in V. campbellii in response to acid stress. At time zero, V. campbellii was shifted from pH 7.6 to pH 5.8 in tryptone-containing (LM) medium. The CadB-eGFP intensity was quantified in single V. campbellii cells over time (upper graph; the gray line indicates background fluorescence). Representative fluorescence and phase-contrast images of V. campbellii cells containing CadB-eGFP are shown. As a control, V. campbellii cells containing CadB-eGFP were analyzed under nonstress conditions (pH 7.6, 2.5 h, first column). The noise and mean relative fluorescence intensity (RF) were calculated for 1,000 cells per condition and time point. Single-cell fluorescence intensity was quantified by microscopy and the use of the ImageJ software. The inset graph illustrates the distribution of CadB-eGFP (percentage of fluorescence) among single cells (gray, cells in the off state; dark green, cells with a high CadB level [on state]) (see Table S2 in the supplemental material). Noise, standard deviation/mean of log-transformed values; PH, phase contrast; GFP, green fluorescent channel. Scale bar, 5 μm.

CadC copy number influences the degree of heterogenous CadB-eGFP distribution in E. coli.

As the output of the Cad system is heterogeneous in E. coli but homogeneous in V. campbellii, we next investigated how the heterogeneity is established in E. coli. Therefore, we focused on the following two regulatory mechanisms: (i) the translational control of CadC by the polyproline motif and (ii) the control of CadC activity by LysP.

We previously found that the copy numbers of LysP and CadC are crucial for an appropriate dual-stimulus response of the Cad system (24). Furthermore, the natural copy number of CadC in E. coli is extremely low (1 to 3 molecules per cell) (24). The inactivation or removal of the strong PPP-stalling motif (CadC-PAP or CadC-AAA) increases the number of CadC molecules per cell by three to six times and interferes with the tight regulation of cadBA expression (24). Due to the low copy number, CadC is distributed heterogeneously in E. coli (25). Thus, we tested whether an increase in the CadC molecule number per cell affects the heterogeneous output of the Cad system. We found an inverse relation between the number of CadC molecules and the heterogenous distribution of CadB-eGFP in E. coli (Fig. 5a). A higher CadC level decreased the noise of CadB-eGFP production, and the overall amount of CadB-eGFP increased, which was reflected by an increase in the mean fluorescence (Fig. 5a). Moreover, a population with a high CadC number per cell (∼100 molecules per cell) was characterized by a homogenous, symmetric Gaussian distribution of CadB-eGFP (Fig. 5a and S2a). E. coli cells which produce such a high CadC number grow somewhat slower than wild-type cells under acid stress and stress-free conditions (see Fig. S3 in the supplemental material).

FIG 5.

CadC copy number influences the degree of heterogenous production of CadB within single E. coli cells. (a) Quantification of the CadB-eGFP intensity of single E. coli cells at 2.5 h after a shift to acid stress conditions (pH 5.8 plus lysine) and various levels of CadC protein copy number due to plasmid-carried cadC (upper graph; the gray line indicates background fluorescence). Representative fluorescence and phase-contrast images of quantified E. coli cells containing CadB-eGFP are shown. The noise and mean relative fluorescence intensity (RF) were calculated for 1,000 cells per condition and time point. Single-cell fluorescence intensity was quantified by microscopy and the use of the ImageJ software. The inset graph illustrates the distribution CadB-eGFP (percentage of fluorescence) among single cells (gray, cells in the off state; dark green, cells with a high CadB level [on state]) (see Table S1 in the supplemental material). The arrows indicate the cells with low CadB content. (b) Quantification of mCherry-LysP intensity of single E. coli cells at 2.5 h after a shift to acid stress conditions (pH 5.8 plus lysine) and nonstress conditions (pH 7.6) (upper graph; the gray line indicates background fluorescence). The noise and mean relative fluorescence intensity (RF) were calculated for 1,000 cells per condition and time point. Noise, standard deviation/mean of log-transformed values; PH, phase contrast; GFP, green fluorescent channel; mCH, red fluorescent channel (mCherry). Scale bar, 5 μm.

We next investigated the spatial distribution of LysP during acid stress to elucidate its role in establishing a heterogeneous output of the Cad system in E. coli. We analyzed the distribution of the cosensor and CadC inhibitor LysP in single E. coli cells using a chromosomal encoded mCherry-lysP fusion, whose product is inserted in the membrane (Fig. S1d). Interestingly, LysP was homogeneously distributed independently of the external pH (Fig. 5b). Therefore, these findings indicated that the homogeneous distribution of LysP has no influence on the heterogenous output of the Cad system, which is established via the low copy number of CadC and leads to an uneven distribution in the population of single E. coli cells.

Finally, to strengthen our hypothesis that the naturally occurring low copy number of CadC is sufficient to generate a heterogeneous output of the Cad system in E. coli, we quantified the CadC copy numbers in the two organisms. Under acid stress, the copy number of CadC was determined to be 55 ± 14 copies/cell in V. campbellii, which was 13 times higher than that in E. coli (Fig. 6). Moreover, the CadC copy number increased by a factor of five under acid stress in V. campbellii. This increase in the CadC copy was expected, as the promoter of cadC is under the positive control of the LysR-type transcriptional regulator AphB in Vibrio species, including V. vulnificus and V. cholerae (32–34). In E. coli, the copy number of CadC per cell under acid stress is slightly increased to 4 ± 1 (Fig. 6). Whether this effect is related to the repressor CsiR needs to be investigated (14).

FIG 6.

CadC and LysP copy numbers in E. coli and V. campbellii. Dot blot analysis was used to determine the CadC copy number in the indicated strains of E. coli and V. campbellii under nonstress (pH 7.6, absence of lysine) conditions and acid stress conditions (pH 5.8 plus 10 mM lysine or LM medium at pH 5.8). The copy number of fluorescently tagged LysP was quantified in E. coli under nonstress (pH 7.6) conditions and acid stress conditions (pH 5.8 plus 10 mM lysine). The numbers below the dot blot images indicate the quantified protein copy numbers per strain and condition. The E. coli ΔcadC and V. campbellii ΔcadC mutants were used as background controls. The E. coli ΔcadC strain carrying the pET-mCherry-cadC plasmid (24, 25) was used as the reference for quantification using ImageJ (47).

We also determined the copy number of LysP, which decreased due to the presence of 10 mM lysine under acid stress (Fig. 5b and 6). This result was in agreement with the regulation of lysP expression depending on the external availability of lysine by the transcriptional regulator ArgP and the global regulator leucine-responsive protein Lrp (35).

Overall, CadB-eGFP is heterogeneously distributed in single E. coli cells after exposure to acid stress, which is achieved by the few and heterogeneously distributed copies of the pH sensor CadC.

DISCUSSION

Acid stress sensing and adaptation are important for bacteria to survive in acidic environments, such as the human gastrointestinal tract or acidic soils (7, 8). Therefore, many bacteria possess inducible acid resistance (AR) systems that rely on H⁺-consuming amino acid decarboxylases and their corresponding antiporters. The latter function as importers for the corresponding amino acids and as exporters for the decarboxylated products. The number and complexity of the three inducible AR systems, namely, the lysine decarboxylase (Cad) system, the glutamate decarboxylase (Gad) system, and the arginine decarboxylase (Adi) system, vary among bacteria and reflect an adaptation to the needs of their individual natural habitats (28). Enteropathogenic bacteria, such as Escherichia, Salmonella, or Yersinia spp., generally possess several AR systems to survive the extremely low pH in the stomach (pH 2.5), but they are also equipped to live in the moderately acidic colon (pH 5.8). E. coli, for example, contains all three inducible AR systems. Salmonella has only the Adi and Cad systems, and Shigella has only the Gad system (28). Vibrio species are generally more sensitive to acid (Fig. 2a) and contain only the Cad system. However, not only does the number of AR systems possessed by bacteria differ, but also the complexity of the individual AR systems varies. For example, the number of different glutamate decarboxylases may be up to three, and the number of antiporters may be up to two in some bacteria (36).

In this study, we focused on the influence of molecular design on the noise of the target protein(s) of the Cad system in two distantly related bacteria, namely, E. coli and V. campbellii (Fig. 7). The Cad system is best studied in E. coli and responds to low pH in a lysine-rich environment due to the activity of the membrane-integrated transcriptional activator CadC, which leads to the expression of cadBA (15) (Fig. 7, left panels).

FIG 7.

The molecular design of a signaling system influences gene expression noise. E. coli (left panels) contains the three inducible AR systems, namely, Gad, Adi, and Cad, whereas V. campbellii (right panels) contains only the Cad system. The main components of the Cad system, including (i) the membrane-integrated pH-responsive regulator CadC, (ii) the lysine decarboxylase CadA, and (iii) the lysine/cadaverine antiporter CadB, are similar in E. coli and V. campbellii. However, E. coli has additional regulatory elements. CadC activity is regulated by the cosensor LysP, which is a lysine transporter, and CadC has a low copy number due to the presence of the strong stalling PPP motif. Due to the differences in the molecular design of the Cad systems, the output reflected by the distribution of CadB-eGFP can be either heterogeneous, such as in E. coli, or homogeneous, such as in V. campbellii. PP, periplasm; CM, cytoplasmic membrane; CP, cytoplasm.

Our phylogenetic analysis revealed a higher complexity and more regulatory elements for the Cad system in E. coli than for that in V. campbellii (Fig. 1 and 7). E. coli has the main components of the Cad system, including (i) the membrane-integrated pH-responsive regulator CadC, (ii) the lysine decarboxylase CadA, and (iii) the lysine/cadaverine antiporter CadB, but it also has additional regulatory elements. Thus, CadC activity is regulated by the lysine permease LysP, and the number of CadC copies per cell is rather low due to the presence of the strong ribosome-stalling PPP motif (Fig. 7, left panels). In contrast, the Cad system of V. campbellii consists of only the main components, namely, CadC, CadA, and CadB (Fig. 7, right panels). Using fluorescently tagged CadB, we studied the output of the Cad system in E. coli and V. campbellii at the single-cell level, and we found a heterogeneous behavior in E. coli (Fig. 3) but not in V. campbellii (Fig. 4). In addition, we found that the degree of heterogeneity of CadB-eGFP production correlates with the copy number of the transcriptional activator CadC in E. coli (Fig. 5). CadC itself is heterogeneously distributed between the cells of the E. coli population (25), probably as a result of dilution of the low number of CadC molecules (≤4 CadC per cell) during cell division. Heterogeneous activation of the cadBA operon leads to a division of labor, and the investment to produce the lysine decarboxylase CadA and the lysine/cadaverine antiporter CadB is reduced to a subpopulation of E. coli cells. However, the whole population will benefit from the increase in the external pH due to the production and secretion of the alkaline cadaverine (see Fig. S1c in the supplemental material). If the heterogenous activation of the cadBA operon is disturbed by a high CadC number, all cells activate the cadBA operon (Fig. 5a) stress independently, resulting in high CadA and CadB production in all cells (37) of the population and even causing slower growth (see Fig. S3 in the supplemental material).

In V. campbellii, we observed a homogeneous production of CadB-eGFP (Fig. 4), and this observation may be explained by the naturally higher number of CadC copies, which further increase during acid stress (Fig. 6). The transcription of cadC is under positive transcriptional control of AphB in Vibrio species, such as V. vulnificus and V. cholerae (32–34), and its translation does not lead to ribosome stalling, as strong or intermediate polyproline motifs are missing (Fig. 1 and 7). In S. enterica, cadC expression is also induced by low pH and the presence of lysine (38). In E. coli, cadC is constitutively expressed (39). However, the copy number of CadC is slightly increased during acid stress, but signaling is still dependent on the two stimuli, namely, low pH and lysine availability. A 5-fold increase in the number of CadC copies not only reduces the sensitivity of CadC to tightly respond to the two stimuli (24) but also lowers the degree of heterogeneous behavior of the population in E. coli (Fig. 5a).

These results broaden our understanding of acid stress adaptation in gammaproteobacteria and reveal a new role of polyproline motifs in general. Polyproline motifs induce ribosome stalling during translation and require the elongation factor EF-P to alleviate this translational arrest. Due to this translational burden, there is an evolutionary selection against polyproline motifs (40). Nevertheless, many signaling proteins contain XPP or PPX motifs. In the case of CadC, the presence or absence of such motifs correlates with a lower or higher copy number, with consequences for the tight regulation of cadBA expression (24) and the degree of noise. In the case of the osmolarity- and pH-sensing histidine kinase EnvZ, the two polyproline motifs are mainly involved in protein-protein interactions (41).

CadC homologs with a strong polyPro stalling motif are especially found in the Enterobacteriaceae family (73.0%) (Fig. 1). Moreover, these homologs most frequently cooccur with LysP homologs. Although the function of LysP as a cosensor is characterized only in E. coli thus far, this cooccurrence of CadC and LysP within the Enterobacteriaceae family emphasizes a conserved selection of the CadC/LysP interaction and a more complex design of the Cad system in these organisms. We and others have previously shown that the copy number of both CadC and LysP needs to be balanced (24, 39). In response to the two stimuli, namely, low pH and lysine, CadC becomes active, and the inhibitory effect of LysP is alleviated by decreasing the strength of the interaction between CadC and LysP (22) and decreasing the copy number of LysP (Fig. 6).

The CadC signaling system of the Vibrionaceae family is simpler, as CadC lacks the regulatory inputs by LysP and the polyPro motif. Without this tight regulation, all the cells of the V. campbellii population activate the Cad system, which is important, as the Cad system is the only AR system in this bacterium. In contrast, E. coli has three decarboxylase-based AR systems, which might overlap. Heterogeneous activation of the gadB promoter has recently been shown (42). It is suggested that individual E. coli cells activate one or the other system to save energy, but the whole population is equipped with several systems to withstand a wide range of pH variations and even strong acid stress (Fig. 2).

Overall, this work presents another example of how phenotypic heterogeneity correlates with the molecular design of signaling systems (43).

MATERIALS AND METHODS

Bacteria and growth conditions.

Bacterial strains and plasmids used in this study are listed in Table 1, and oligonucleotides used in this study are listed in Table S3 in the supplemental material. E. coli strains were cultivated in LB medium (10 g/liter NaCl, 10 g/liter tryptone, 5 g/liter yeast extract) or in Kim-Epstein (KE) medium (44) adjusted to pH 5.8 or pH 7.6, using the corresponding phosphate buffer. E. coli strains were incubated aerobically in a rotary shaker at 37°C. KE medium was always supplemented with 0.2% (wt/vol) glucose. Generally, lysine was added to a final concentration of 10 mM unless otherwise stated.

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant genotype or description	Reference
Strains
E. coli
MG1655	K-12 F– λ– ilvG rfb-50 rph-1	55
MG1655 cadB-egfp	Chromosomally integrated C-terminal cadB-egfp fusion in E. coli MG1655	This work
MG1655 mCherry-LysP	Chromosomally integrated N-terminal mCherry-lysP fusion in E. coli MG1655	This work
V. campbellii
ATCC BAA-1116	Wild type	45
ATCC BAA-1116 cadB-egfp	Chromosomally integrated C-terminal cadB-egfp fusion	This work
ATCC BAA-1116 ΔcadB	Clean deletion of cadBVC	This work
ATCC BAA-1116 ΔcadC	Clean deletion of cadCVC	This work
ATCC BAA-1116 ΔcadA	Clean deletion of cadAVC	This work
ATCC BAA-1116 mCherry-cadC	Chromosomally integrated N-terminal cadC-mCherry fusion	This work
E. coli
DH5αλpir	endA1 hsdR17 glnV44 (= supE44) thi-1 recA1 gyrA96 relA1 ϕ80′lacΔ(lacZ)M15 Δ(lacZYA-argF)U169 zdg-232::Tn10 uidA::pir+	56
WM3064	thrB1004 pro thi rpsL hsdS lacZΔM15 RP4-1360 Δ(araBAD)567 ΔdapA1341::[erm pir]	W. Metcalf, University of Illinois, Urbana
BL21(DE3)	F− ompT gal dcm lon hsdSB(rB− mB−) λ(DE3)	57
MG1655 ΔcadC	Clean deletion of cadCEC in MG1655	19
MG1655 ΔcadB	cadB::Km in MG1655, Kmr	This work
Plasmids
pET-cadC	cadCEC under control of IPTG-inducible T7 polymerase-dependent promoter in pET16b, Ampr	19
pET-mCherry-cadC	N-terminal fusion of cadC-mCherry in pET16b, Ampr	25
pBAD-cadC	cadCEC under control of arabinose-inducible promoter in pBAD24, Ampr	24
pBAD24	Arabinose-inducible PBAD promoter, pBR322 ori, Ampr	58
pNTPS138-R6KT	mobRP4+ori-R6K sacB; suicide plasmid for in-frame deletions, Kmr	46
pNPTS138-R6KT-cadB-egfp-EC	pNPTS-138-R6KT-derived suicide plasmid for in-frame insertion of cadB-egfp in E. coli MG1655, Kmr	This work
pNPTS138-R6KT-cadB-EC	pNPTS-138-R6KT-derived suicide plasmid for clean deletion of cadBEC in E. coli MG1655, Kmr	This work
pNPTS138-R6KT-mCherry-lysP-EC	pNPTS-138-R6KT-derived suicide plasmid for in-frame insertion of lysP-mCherry in E. coli MG1655, Kmr	This work
pNPTS138-R6KT-cadB-egfp-VC	pNPTS-138-R6KT-derived suicide plasmid for in-frame insertion of cadB-egfp in V. campbellii BAA-1116, Kmr	This work
pNPTS138-R6KT-cadB-VC	pNPTS-138-R6KT-derived suicide plasmid for clean deletion of cadBVC in V. campbellii, Kmr	This work
pNPTS138-R6KT-cadC-VC	pNPTS-138-R6KT-derived suicide plasmid for clean deletion of cadCVC in V. campbellii, Kmr	This work
pNPTS138-R6KT-cadA-VC	pNPTS-138-R6KT-derived suicide plasmid for clean deletion of cadAVC in V. campbellii, Kmr	This work
pNPTS138-R6KT-mCherry-cadC-VC	pNPTS-138-R6KT-derived suicide plasmid for in-frame insertion cadC-mCherry in V. campbellii, Kmr	This work
pNPTS138-R6KT-egfp	pNPTS-138-R6KT-derived suicide plasmid containing egfp, Kmr	This work
pET28a	His-tagging expression vector, Kmr	Novagen
pET28a-cadA-VC	N-terminal His6-tagged cadAVC, Kmr	This work

V. campbellii ATCC BAA-1116 (formerly known as Vibrio harveyi ATCC BAA-1116 [45]) and derivatives were routinely cultivated in Luria marine (LM) medium (20 g/liter NaCl, 10 g/liter tryptone, 5 g/liter yeast extract) and incubated aerobically on a rotary shaker at 30°C.

If necessary, media were supplemented with 100 μg/ml ampicillin or 50 μg/ml kanamycin sulfate. To allow the growth of the conjugation strain E. coli WM3064, we added meso-diamino-pimelic acid (DAP) to a final concentration of 300 μM.

Construction of chromosomally integrated fluorescent fusions and deletion strains.

To fluorescently label CadB of E. coli, an E. coli MG1655 strain harboring chromosomally egfp-tagged cadB under the control of its native promoter was constructed. In-frame insertion of egfp was achieved in E. coli MG1655 using the suicide plasmid pNPTS138-R6KT-cadB-egfp-EC. Briefly, 500 bp upstream and downstream of cadB were amplified by PCR using MG1655 genomic DNA as the template. In order to amplify egfp, the plasmid pNPTS138-R6KT-egfp was used as the template (720 bp). Using overlap extension PCR, the three fragments were assembled via their homologous regions. The overlap PCR fragment was cloned into plasmid pNPTS138-R6KT (46) using BamHI and PspOMI restriction sites, resulting in the pNPTS138-R6KT-cadB-egfp-EC plasmid. The resulting plasmid was introduced into E. coli MG1655 by conjugative mating using E. coli WM3064 as a donor in LB containing DAP. Single-crossover integration mutants were selected on LB plates containing kanamycin but lacking DAP. Single colonies were grown over a day without antibiotics and plated onto LB plates containing 10% (wt/vol) sucrose but lacking NaCl to select for plasmid excision. Kanamycin-sensitive colonies were checked for targeted deletion by colony PCR using primers bracketing the site of the deletion. Insertion of egfp was verified by colony PCR and sequencing.

The plasmid pNPTS138-R6KT-egfp was generated by cloning an egfp PCR fragment into pNPTS138-R6KT using the restriction enzymes SpeI and PspOMI. The egfp PCR fragment was generated by amplification of egfp from pEGFP-C1 (Clontech).

The E. coli MG1655 strain harboring chromosomally mCherry-tagged lysP under the control of its native promoter was constructed as described above, with the only difference being that LysP is tagged N terminally and 500 bp upstream and downstream of lysP were amplified, resulting in the plasmid pNPTS138-R6KT-mCherry-lysP-EC.

Deletion of cadB in MG1655 was done by Red/ET recombination using the E. coli Quick and Easy gene deletion kit (Gene Bridges, Heidelberg, Germany). Briefly, primers (cadB_KO_sense and cadB_KO_anti) were designed according to the manual. These primers target the area surrounding of cadB in order to amplify the FRT-PGK-gb2-neo-FRT template. A double-stranded PCR fragment was introduced via electroporation into E. coli MG1655 according to the manual. Deletion of cadB was verified by colony PCR and sequencing.

To fluorescently label CadB of V. campbellii, a V. campbellii strain harboring chromosomally egfp-tagged cadB under the control of its native promoter was constructed as described above. However, instead of 500-bp-long flanking regions, 600-bp-long flanking regions were amplified by PCR using V. campbellii genomic DNA as the template, resulting in the plasmid pNPTS138-R6KT-cadB-egfp-VC. Moreover, plasmid excision was performed on LB agar plates containing 10% (wt/vol) sucrose for V. campbellii cells.

To fluorescently label CadC of V. campbellii, a V. campbellii strain harboring chromosomally mCherry-tagged cadC under the control of its native promoter was constructed as described above, using 600-bp-long flanking regions and amplifying the flanking regions of V. campbellii cadC, resulting in the plasmid pNPTS138-R6KT-mCherry-cadC-VC. mCherry and CadC are connected via a 22-amino-acid (AMGHHHHHHHHHHSSGHIEGRH) linker.

Construction of the ΔcadB, ΔcadA, and ΔcadC markerless in-frame deletion strains in V. campbellii was achieved using the suicide plasmids pNPTS138-R6KT-cadB-VC, pNPTS138-R6KT-cadA-VC, and pNPTS138-R6KT-cadC-VC, respectively, and V. campbellii genomic DNA as the template as described above.

Construction of N-terminally His₆-tagged CadA of V. campbellii.

The V. campbellii cadA gene was cloned into the pET-28a vector using BamHI and XhoI as restriction sites, resulting in an added N-terminal His₆ tag sequence.

In vivo fluorescence microscopy and data analysis.

To analyze the spatiotemporal localization of CadB-GFP of E. coli and V. campbellii and mCherry-LysP of E. coli, overnight cultures were prepared in KE medium (pH 7.6) (E. coli) or LM medium (pH 7.6) (V. campbellii) and aerobically cultivated at 37°C or 30°C, respectively. The overnight cultures were used to inoculate 1-day cultures (optical density at 600 nm [OD₆₀₀] of 0.1) in fresh medium at pH 7.6. At an OD₆₀₀ of 0.5, cells were gently centrifuged and resuspended, thereby exposing them to different conditions, e.g., different low pHs or different lysine concentrations. The cultures were then cultivated aerobically at 37°C or 30°C. Every 30 min after the shift to different conditions, 2 μl of the culture was spotted on 1% (wt/vol) agarose pads (prepared with the different media), placed on microscope slides, and covered with a coverslip. In case of V. campbellii, cultures grown in the complex LM medium were spotted on a 1% (wt/vol) agarose pad prepared with KE medium (pH 5.8) supplemented with 2% NaCl in order to reduce the fluorescence of LM medium. Subsequently, images were taken on a Leica DMi8 inverted microscope equipped with a Leica DFC365 FX camera (Wetzlar, Germany). An excitation wavelength of 546 nm and a 605-nm emission filter with a 75-nm bandwidth were used for mCherry fluorescence with an exposure of 500 ms, gain of 5, and 100% intensity for the E. coli mCherry-lysP strain. An excitation wavelength of 485 nm and a 510-nm emission filter with a 75-nm bandwidth were used for eGFP fluorescence with an exposure of 500 ms, gain of 5, and 100% intensity for the E. coli and V. campbellii cadB-egfp strains.

To analyze the influence of different CadC copy numbers, the E. coli cadB-egfp strain was transformed with plasmid pET-cadC or pBAD-cadC by electroporation. E. coli strains carrying the plasmids were cultivated and analyzed as described above.

To quantify relative fluorescent intensities (RF) representing CadB-eGFP or mCherry-LysP of single cells, phase-contrast and fluorescent images were analyzed using the ImageJ (47) plug-in MicrobeJ (48). The default settings of MicrobeJ were used for cell segmentation (fit shape, rod-shaped bacteria) except for the following settings: area, 0.1 to max μm²; length, 1.2 to 5 μm; width, 0.1 to 1 μm; curvature, 0 to 0.15; and angularity, 0 to 0.25 for E. coli cells. For V. campbellii cells, the parameter width was altered to 0.6 to 1.5 μm in MicrobeJ. In total, 1,000 cells were quantified per strain, condition, and time point. The background reading for the agarose pad was subtracted from that for each cell per field of view. The mean and standard deviation (SD) of the RF were quantified using MicrobeJ, and background was subtracted. Noise was defined as SD/mean of log-transformed values per sample. Statistical analysis and presentation were performed using GraphPad Prism 5.03.

Detection of fluorescently tagged CadB and LysP via Western blotting.

To verify the location of fluorescently tagged CadB or LysP in different compartments of the cell, the E. coli cadB-egfp and mCherry-lysP strains and the V. campbellii cadB-egfp strain were grown to an OD₆₀₀ of 0.5 in KE or LM medium, respectively. Cells were then shifted to inducing conditions (pH 5.8, KE medium plus lysine or LM medium), harvested after 2.5 h, and then adjusted to an OD₆₀₀ of 1. Cells were disrupted by passage through a high-pressure cell disrupter (Constant Systems, Northants, United Kingdom) in ice-cold disruption buffer (50 mM Tris-HCl [pH 7.5], 10% [vol/vol] glycerol, 10 mM MgCl₂, 100 mM NaCl, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride [PMSF], and 0.03 mg ml⁻¹ DNase). After removal of intact cells and cell debris (pellet) via centrifugation (5,000 × g, 30 min, 4°C), membrane vesicles were collected by ultracentrifugation (45,000 × g, 60 min, 4°C), where the pellet contained the membrane fraction and the supernatant the cytoplasm. These fractions were separated by SDS-PAGE (49) on 12.5% acrylamide gels and transferred to a nitrocellulose membrane. Tagged proteins were labeled with primary polyclonal anti-mCherry antibody (Invitrogen) or anti-GFP antibody (Roche), and anti-rabbit or anti-mouse alkaline phosphatase-conjugated antibody (Rockland Immunochemicals) was used as the secondary antibody according to the manufacturer’s recommendations. Localization of the secondary antibody was visualized using colorimetric detection of alkaline phosphatase activity with 5-bromo-4-chloro-3-indolylphosphate and nitroblue tetrazolium chloride. As a ladder, the PageRuler prestained protein ladder (10 to 180 kDa; Thermo Fisher) was used.

Relative protein quantification using the dot blot method.

To determine the relative amount of CadC per cell of V. campbellii in comparison to the copy number of CadC of E. coli, we remeasured the abundance of CadC of E. coli using the same strains and antibodies and calculated the absolute numbers based on the former study (24).

mCherry-tagged CadC strains (E. coli MG1655 ΔcadC pET-mCherry-cadC and V. campbellii mCherry-CadC) and control strains (E. coli MG1655 ΔcadC and V. campbellii ΔcadC) were cultivated in LB or LM medium (pH 7.6), respectively, to an OD₆₀₀ of 0.3 and shifted to inducing conditions at pH 5.8 for 2.5 h or to noninducing conditions (pH 7.6). Subsequently, cells having an OD₆₀₀ of 0.4 to 0.6 were adjusted to an OD₆₀₀ of 25 in Tris-buffered saline (50 mM Tris, 150 mM NaCl, pH adjusted with HCl to pH 7.6.) and lysed by adding 0.5% (wt/vol) SDS and boiling for 5 min. Two microliters of cell extracts were then spotted on Amersham Protran 0.45 NC nitrocellulose transfer membranes (GE Healthcare), air dried, and further handled as described in the previous section using the anti-mCherry antibody as the primary antibody. As a reference, E. coli MG1655 was transformed with pET-mCherry-cadC, resulting in 3 to 5 CadC copies per cell (24). Background subtraction was performed from V. campbellii ΔcadC and E. coli MG1655 ΔcadC cells. All quantification was performed with the ImageJ software (47).

The relative amount of mCherry-LysP of E. coli was also determined in comparison to E. coli MG1655 pET-mCherry-cadC as described above based on the numbers from the former study (24).

AR assay.

Acid resistance (AR) was determined essentially as described previously (10, 29) with the following modifications. E. coli and V. campbellii wild-type strains were cultivated in pH 7.6 LB or LM medium, respectively, to an OD₆₀₀ of 0.6 to 0.8, and then cultures were adjusted to an OD₆₀₀ of 0.5 and resuspended in LB or LM medium with a pH of 3.0. The low-pH challenge was conducted at 37°C or 30°C, and samples were collected immediately after resuspension (t = 0), every 15 min for the first hour, and then hourly for 4 h. Samples were serially diluted and plated onto LB or LM agar plates to assess the number of colonies surviving the acid challenge.

Growth experiment.

In order to compare the influence of the CadC copy number on the growth of E. coli under acid stress and no stress, strains E. coli MG1655 ΔcadC pBAD-cadC and E. coli MG1655 pBAD24 were cultivated in KE5.8 plus lysine and KE7.6 plus lysine supplemented with 0.01% (wt/vol) arabinose. Due to the plasmid pBAD-cadC in the strain MG1655 ΔcadC, the copy number of CadC is elevated to ∼100 CadC molecules per cell, whereas the E. coli MG1655 pBAD24 strain contains ≤4 CadC molecules per cell. These two cultures were adjusted to an OD₆₀₀ of 0.1 in KE5.8 plus lysine or KE7.6 plus lysine and then aerobically cultivated in 96-well plates at 37°C. Growth was determined every 15 min in a microtiter plate with a Tecan Infinite F500 system (Tecan, Crailsheim, Germany). Data are reported as OD₆₀₀ versus time.

Functionality of CadB-eGFP fusion.

In order to test the functionality of the CadB fusion with eGFP, a liquid-based colorimetric assay using a pH indicator was used. The assay is based on the color change of the pH indicator bromocresol purple and detects an increase of external pH by the production and secretion of cadaverine (39, 50, 51). Briefly, cells are cultivated in lysine-decarboxylase differential medium (0.5% [wt/vol] peptone, 0.3% [wt/vol] yeast extract, 0.1% [wt/vol] glucose, 1% [wt/vol] l-lysine, 0.002% [wt/vol] bromocresol purple, pH 5.0 [200 mM NaCl for Vibrio strains]) at a starting OD of 0.5 overnight.

Alignment and construction of phylogenetic tree.

To identify nonredundant CadC orthologs, a Protein BLAST search of the NCBI RefSeq protein database (52) using the full-length CadC from E. coli MG1655 as the query sequence (expected value of <1 × 10⁻²⁰) was carried out. A tolerance of 10% of the amino acid length was set as the default parameter. A pairwise alignment of 3,223 sequences was done with a progressive algorithm from the software CLC Workbench 8.0 (CLC Bio Qiagen, Hilden, Germany) using the following parameters: gap open cost, 10; gap extension cost, 1; and high accuracy (53). The results served as the basis for construction of a phylogenetic tree by the software’s high-accuracy, distance-based neighbor-joining algorithm (100 bootstrap replicates and the Jukes-Cantor distance correction as default parameters). The branch lengths therefore represent the degree of evolutionary divergence between any two nodes in the tree. We screened the organisms containing a CadC ortholog for the presence of LysP by searching for orthologs of E. coli MG1655 LysP with NCBI Protein BLAST (blastp algorithm; expected value, 10).

In addition, the presence of polyproline motifs within the CadC sequences was analyzed, which are classified into strong stalling XPP or PPX motifs (PPP, APP, DPP, GPP, PPN, PPD, PPG, PPW, PPE, and PPS), moderately/weakly stalling XPP or PPX motifs, or no XPP or PPX motif according to reference 40 (see Data Set S1 in the supplemental material).

CadA purification and lysine decarboxylase assay.

In order to purify CadA of V. campbellii and test its functionality as a lysine decarboxylase, E. coli BL21(DE3) carrying plasmid pET28a-cadA-VC was cultivated in LB supplemented with kanamycin at 37°C. At an OD₆₀₀ of 0.5, 0.5 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was added to induce cadA expression, and cells were further cultivated at 18°C for 4 h. Cells were then harvested, resuspended, and disrupted by passage through a high-pressure cell disrupter (Constant Systems, Northants, United Kingdom) in ice-cold disruption buffer (25 mM Tris-HCl [pH 7.5], 200 mM NaCl, 5% [vol/vol] glycerol, 1 mM dithiothreitol [DTT], 3 mg DNase, and 0.5 mM phenazine methosulfate in double-distilled water [ddH₂O]). After removal of intact cells and cell debris via centrifugation (5,000 × g, 30 min, 4°C), membrane vesicles were separated by ultracentrifugation (45,000 × g, 60 min, 4°C), and the supernatant was loaded on a Ni-nitrilotriacetic acid (NTA) column and eluted with 250 mM imidazole. Fractions containing V. campbellii CadA were pooled and dialyzed against 50 mM morpholineethanesulfonic acid (MES) buffer–100 mM NaCl, pH 6.0. Afterwards, the specific activity of purified V. campbellii CadA was measured as described before (54); however, only 1 μg/ml CadA was used per sample.