c-di-AMP assists osmoadaptation by regulating the Listeria monocytogenes potassium transporters KimA and KtrCD

Many bacteria and some archaea produce the second messenger cyclic diadenosine monophosphate (c-di-AMP). c-di-AMP controls the uptake of osmolytes in Firmicutes, including the human pathogen Listeria monocytogenes, making it essential for growth. c-di-AMP is known to directly regulate several potassium channels involved in osmolyte transport in species such as Bacillus subtilis and Streptococcus pneumoniae, but whether this same mechanism is involved in L. monocytogenes, or even whether similar ion channels were present, was not known. Here, we have identified and characterized the putative L. monocytogenes' potassium transporters KimA, KtrCD, and KdpABC. We demonstrate that Escherichia coli expressing KimA and KtrCD, but not KdpABC, transport potassium into the cell, and both KimA and KtrCD are inhibited by c-di-AMP in vivo. For KimA, c-di-AMP–dependent regulation requires the C-terminal domain. In vitro assays demonstrated that the dinucleotide binds to the cytoplasmic regulatory subunit KtrC and to the KdpD sensor kinase of the KdpDE two-component system, which in Staphylococcus aureus regulates the corresponding KdpABC transporter. Finally, we also show that S. aureus contains a homolog of KimA, which mediates potassium transport. Thus, the c-di-AMP–dependent control of systems involved in potassium homeostasis seems to be conserved in phylogenetically related bacteria. Surprisingly, the growth of an L. monocytogenes mutant lacking the c-di-AMP–synthesizing enzyme cdaA is only weakly inhibited by potassium. Thus, the physiological impact of the c-di-AMP–dependent control of potassium uptake seems to be less pronounced in L. monocytogenes than in other Firmicutes.

Both B. subtilis and S. aureus contain well-described potassium uptake systems. B. subtilis uses the high-affinity transporters KtrAB and KimA and the low-affinity transporter KtrCD (5,43). By contrast, S. aureus relies on the high-affinity transporter KdpFABC, whose synthesis and activity is controlled by the two-component system KdpDE and c-di-AMP, respectively (41,44,45). S. aureus also contains the low-affinity potassium transport systems KtrCB and KtrCD sharing the accessory protein KtrC (44). A BLASTp sequence analysis revealed that the L. monocytogenes genome codes for the KdpABCDE (Lmo2682-Lmo2678) and KtrCD (Lmo1023, Lmo0993) proteins, which show about 31-56% and 51-64% overall amino acid identity with the homologs from S. aureus and B. subtilis, respectively. The kdpF gene that has been shown to be important for proper function of the Kdp system in E. coli does not exist in the L. monocytogenes genome (46). A homolog of the high-affinity potassium transporter KimA from B. subtilis is also present in L. monocytogenes and S. aureus (5). The KimA homologs from B. subtilis, L. monocytogenes, and S. aureus are from now on designated as KimA Bsu , KimA Lmo (Lmo2130), and KimA Sau (Sacol2443), respectively. KimA Lmo and KimA Sau show about 59 and 57% overall amino acid identity, respectively, with the B. subtilis homolog (Fig. 1A). The membrane topology was illustrated using the web-based tool Protter (47). Like KimA Bsu , KimA Lmo also contains an N-terminal extracellular domain, 11 transmembrane helices, and a C-terminal intracellular domain, which might be important for activity control of the transporter (Fig. 1B). To conclude, although B. subtilis, L. monocytogenes, and S. aureus are phylogenetically related, each species uses a different set of transporters for potassium uptake.

In vivo activities of the L. monocytogenes potassium transporters
To assess whether KdpABC, KimA, and KtrCD from L. monocytogenes are active in potassium transport, we cloned the kdpABC, kimA, and ktrCD genes using the plasmid pWH844, which allows IPTG-dependent expression of heterologous genes in E. coli (48). We also cloned a truncated kimA Lmo gene encoding the ⌬C-KimA Lmo variant lacking 152 amino acids of the C-terminal cytosolic domain. Furthermore, we cloned the kimA Sau gene from S. aureus, to evaluate whether KimA homologs from other Firmicutes are involved in potassium uptake. The resulting plasmids were used to transform the E. coli strain LB650 that is unable to take up potassium via the native uptake systems Kup, KdpABC, TrkG, and TrkH (49). The strain is suitable to study potassium transporters because it is only viable in minimal medium supplemented with potassium concentrations above 15 mM KCl (see Fig. 4). The empty plasmid and a plasmid encoding the B. subtilis ktrAB genes, which were previously shown to mediate potassium transport in E. coli LB650 (5), served as negative and positive controls, respectively. The cells were grown during the day in M9 medium with 50 mM KCl and without IPTG induction, col-

Listerial potassium transporters
lected by centrifugation, washed in potassium-free medium, and propagated on M9 minimal medium plates without and with 10 mM IPTG. As shown in Fig. 2, with the exception of the strain harboring the plasmid for kimA Sau expression, the bacteria could not grow in the absence of IPTG. The weak growth of the cells containing the kimA Sau gene could be due to a leaky promoter and due to the high affinity of the encoded KimA Sau transporter for potassium (see below). By contrast, the strains carrying the ktrAB Bsu , kimA Lmo , and ktrCD Lmo , could grow with low amounts of K ϩ when these genes were induced with IPTG  (63). Amino acids in black, gray, and white have an amino acid similarity of Ͼ80, 60 -80, or Ͻ60%, respectively. B, predicted membrane topology of KimA Lmo overlaid with a MUSCLE alignment between KimA Bsu and KimA Lmo . The dashed line indicates the position at which the KimA Lmo protein was truncated. Amino acids in black are identical; amino acids in gray are similar, and amino acids in white are nonsimilar. (Fig. 2). Moreover, the kimA Lmo variant lacking the C-terminal domain (kimA Lmo ⌬C) also supported growth of the strains, albeit less well than the full-length protein. These results indicate that the N-terminal extracellular domain and the 11 transmembrane helices of KimA Lmo from L. monocytogenes are sufficient for mediating potassium import in E. coli (Fig. 2). Expression of ktrAB Bsu from B. subtilis restores growth on low potassium concentrations, agreeing with previous reports that KtrAB Bsu is a high-affinity potassium transporter (5,43). Expression of KimA Sau from S. aureus in the E. coli strain LB650 resulted in much better growth than those strains expressing KimA Lmo and KtrCD, indicating that KimA Sau is likely a high-affinity potassium transporter (Fig. 2). Thus, KimA Lmo and KtrCD Lmo from L. monocytogenes as well as KimA Sau from S. aureus are indeed potassium transporters. The putative potassium transporter KdpABC Lmo did not support growth of the E. coli strain LB650 irrespective of whether the kdpABC Lmo genes were expressed from the IPTG-and arabinose-dependent plasmids pWH844 and pBAD24 (data not shown; see "Experimental procedures"). Therefore, the KdpABC Lmo system was not further analyzed in regard to its affinity to potassium ions and in vivo inhibition by c-di-AMP.

Apparent affinities of KimA and KtrCD for potassium
To determine the apparent affinities of KimA Lmo , the KimA Lmo ⌬C terminus variant (⌬C-KimA Lmo ) and KtrCD from L. monocytogenes and KimA Sau from S. aureus, we determined the growth rates of the E. coli strain LB650 synthesizing the potassium transporters in M9 minimal medium supplemented with different amounts of potassium. The strains carrying the empty plasmid and expressing the B. subtilis ktrAB genes served as negative and positive controls, respectively. The growth rates were plotted against the potassium concentrations and fitted to the Michaelis-Menten equation (5). The V max values and the apparent affinities are summarized in Table 1. As shown in Figs. 3 and 4, each E. coli strain required a different concentration of external potassium to reach half-maximal growth; the strains synthesizing the transporters KtrCD Lmo , KimA Lmo , ⌬C-KimA Lmo , KimA Sau , and KtrAB Bsu required 6.30 Ϯ 2.06, 0.35 Ϯ 0.12, 2.99 Ϯ 0.65, 0.14 Ϯ 0.02, and 0.03 Ϯ 0.01 mM, respectively. These results demonstrate that KtrCD Lmo and KimA Lmo from L. monocytogenes are transporters with low and moderately high affinities for potassium, respectively. Moreover, the C-terminal intracellular domain of KimA Lmo is important for full activity of the transporter (Figs. 1B and 4). In contrast to KtrCD Lmo and KimA Lmo , KimA Sau from S. aureus is a high-affinity potassium transporter, which is in line with the observation that the E. coli strain LB650 synthesizing KimA Sau and KtrAB Bsu grew comparatively well with low amounts of potassium (Fig. 2).

Inhibition of KimA and KtrCD potassium transport activity by c-di-AMP
Several recent studies indicate that c-di-AMP is essential for viability of Gram-positive bacteria like B. subtilis, L. lactis, L. monocytogenes, S. agalactiae, and S. aureus because the nucleotide controls influx of osmolytes like potassium whose accumulation leads to cell lysis due to water uptake (5,14,15,25,26). Thus, either synthesis of the potassium transporters or their activity or both need to be tightly regulated. As shown in Fig. 5, the IPTG-dependent overexpression of the ktrAB Bsu and kimA Lmo genes encoding high-affinity potassium transporters KtrAB Bsu and KimA Lmo , respectively, in E. coli during growth in M9 minimal medium caused a strong increase of the cellular volume. Moreover, the growth of E. coli synthesizing the higher-affinity KtrAB Bsu transporter was in addition significantly reduced, as illustrated by the decline of the optical density (Fig.  5, top right corners). By contrast, in the absence of the inducer IPTG, the growth and the volume of the cells containing the ktrAB Bsu and kimA Lmo genes were indistinguishable from that of the cells carrying the empty vector. Thus, once sufficient potassium has been taken up by the bacteria to cope with the osmolarity of the environment, the activities of osmolyte transporters have to be reduced to prevent further ion uptake and cell lysis. It has indeed been demonstrated that the activity of the cytoplasmic gating component of the transporters KtrCB and KtrCD from S. aureus as well as the KimA Bsu transporter from B. subtilis are inhibited by c-di-AMP (7,39). Like KimA Bsu from B. subtilis, the KimA Lmo homolog from L. monocytogenes belongs to a novel class of high-affinity potassium transporters (see above) (5). However, whether c-di-AMP directly binds to KimA Lmo and KtrCD Lmo to inhibit the transport activity of the proteins has not been tested so far.
To assess whether c-di-AMP affects the activity of KimA Lmo and KtrCD Lmo , we established a co-expression system using the E. coli strain LB2003, which carried unmarked mutations in the kdp, kup, and trk genes, and enable the use of multiple plasmids encoding chloramphenicol and ampicillin resistance genes (49). Like the E. coli strain LB650, LB2003 is deficient in the Kdp, Kup, and Trk potassium uptake systems and is therefore only able to grow at low potassium concentrations when synthesizing a potassium transporter. Moreover, E. coli lacks c-di-AMP-producing and c-di-AMP-degrading enzymes, which is a prerequisite to assess the phenotypic effect of c-di-AMP on

Listerial potassium transporters
the activity of KimA Lmo and KtrCD Lmo . The plasmids pBP384 (kimA Lmo ), pBP396 (⌬C-kimA Lmo ), and pBP371 (ktrCD Lmo ) were used for the IPTG-dependent expression of ⌬C-Ki-mA Lmo , KimA Lmo , and KtrCD Lmo , respectively. The empty plasmid pWH844 served as a negative control. The L. monocytogenes DAC CdaA and the inactive CdaA* variant D171N (50) are encoded by the arabinose-inducible plasmids pBP370 and pBP373, respectively. The strains carrying pWH844, pBP384, pBP396, and pBP371 as well as either of the two DAC-encoding plasmids were grown in M9 minimal medium supplemented with 30, 0.35, 3, and 7 mM KCl, respectively, conditions that allow half-maximal growth of the bacteria. As shown in Fig. 6, growth of the strains carrying the empty plasmid pWH844, and synthesizing the active and the catalytically inactive CdaA and CdaA* variants, respectively, was not reduced. Thus, neither the DAC proteins nor c-di-AMP affect growth of the E. coli strain. By contrast, growth of the bacteria synthesizing KimA Lmo and KtrCD Lmo was reduced when the active DAC CdaA was co-produced, indicating that c-di-AMP inhibits the transporter with a moderately high affinity for potassium and to a lesser extent also the low-affinity transporter (Fig. 6). Growth was not affected in the absence of a functional DAC and, thus, of c-di-AMP production. Moreover, c-di-AMP did not affect the activity of the C-terminally truncated ⌬C-KimA Lmo variant, indicating that the C-terminal intracellular domain of the transporter contributes to c-di-AMP-dependent regulation (see "Discussion"). Surprisingly, c-di-AMP did not inhibit the activity of KimA Sau from S. aureus (data not shown). To conclude, the potassium transporters KimA Lmo and KtrCD Lmo from L. monocytogenes are both inhibited by c-di-AMP.

Effect of potassium on growth of a c-di-AMP-free L. monocytogenes strain
Previously, it has been shown that c-di-AMP is essential in B. subtilis to control the uptake of potassium to toxic levels (5). To investigate whether c-di-AMP is also essential for the control of potassium uptake in L. monocytogenes, we constructed a markerless deletion of the cdaA gene, encoding the sole c-di-AMP-synthesizing enzyme. As described previously, we confirmed that the cdaA mutant is not viable on complex, but on chemically defined growth medium (10,11). We prepared Listeria synthetic medium (LSM) (11) without potassium (LSM-K ϩ ) and observed that potassium concentrations below 1 mM impair the growth of the L. monocytogenes WT strain (Fig. 7). The growth behavior of the WT strain was not affected at potassium chloride concentrations higher than 1 mM. The L. monocytogenes cdaA mutant shows a slightly slower growth than the WT at potassium concentrations above 1 mM. However, high potassium concentrations did not fully inhibit growth of the L. monocytogenes cdaA mutant strain as it has been shown for a c-di-AMP-free strain of B. subtilis (see Fig. 7) (5).

c-di-AMP interaction with the KimA homologs and KtrCD Lmo
To assess the interaction between c-di-AMP and the potassium transporters or their regulators, we performed a differential radial capillary action of ligand assay (DRaCALA) with the proteins KimA Lmo , ⌬C-KimA Lmo , KimA Sau , KtrC Lmo , KdpAB-C Lmo , and KdpD Lmo (see "Experimental procedures"). We also tested the interaction between c-di-AMP and the 156-and 158amino-acid-long C-terminal cytosolic domains of KimA Lmo and KimA Sau , respectively. This domain could be involved in the c-di-AMP-dependent control of KimA potassium transport activity. The lysate of the E. coli strain DH5␣ containing the empty plasmid pWH844 or pGP172 served as a negative controls. Whereas the majority of the proteins showed no specific interaction with c-di-AMP in the DRaCALA assay, KtrC Lmo , the cytosolic protein of the KtrCD potassium transporter, and KdpD Lmo , the sensor kinase of the KdpDE twocomponent system, gave positive results (Fig. 8). To conclude, the potassium transport activity of KtrCD Lmo from L. monocytogenes is inhibited by c-di-AMP in vivo, and the nucleotide binds to the KtrC subunit of the KtrCD Lmo transporter and to KtrD of the KdpDE Lmo two-component system in vitro. Due to toxicity, we were unable to purify the full-length KimA Lmo protein. The failure of purification of the C-terminal part of KimA also precludes further in vitro characterization.

Discussion
Here, we have identified and characterized the potassium transporters KtrCD Lmo and KimA Lmo from L. monocytogenes. As stated above, the KtrCD homologs from L. monocytogenes and B. subtilis show 64% overall sequence identity and have similar affinities for potassium (see Table 1) (43). We also demonstrate that S. aureus possesses a homolog of KimA (Sacol2443). The KimA homologs from S. aureus and L. monocytogenes belong to a novel class of high-affinity potassium transporters that are active at low external potassium concentrations (5). Moreover, the potassium transport activity of KtrCD Lmo and KimA Lmo from L. monocytogenes is inhibited by c-di-AMP (Fig. 6). Furthermore, we show that the C-terminal cytosolic domain is important for the c-di-AMP-mediated regulation of KimA Lmo in vivo because the C-terminally truncated variant lacking 156 amino acids did not respond to the nucleotide. Recently, it has been shown that c-di-AMP binds to the KimA homolog from B. subtilis and controls the uptake of potassium by the transporter in vivo (39). Unfortunately, we could not show the binding of c-di-AMP to the full-length Table 1 Michaelis-Menten constants of the potassium transporters Mean values of the K m and S.E. are shown (n ϭ 4). p values were always Ͻ0.0001 (****) compared with empty plasmid alone (F (8,18) ϭ 54.69, ANOVA with Dunnett's post hoc test). Mean values of the V max and S.E. are shown (n ϭ 4). p Ͻ 0.01 (**), p Ͻ 0.001 (***), p Ͻ 0.0001 (****) compared with empty plasmid alone (F (5,18)

Listerial potassium transporters
KimA Lmo protein and to the C-terminal domain of KimA Lmo . However, we speculate that binding of c-di-AMP to the cytosolic domain is required of for regulation of KimA Lmo in vivo (Fig.  6). Therefore, it might be worthwhile to study the role of the C-terminal domain in controlling the activity of the high-affinity potassium transporter KimA. Surprisingly, the KimA Lmo transporter from L. monocytogenes has a much lower affinity for potassium than the homolog from B. subtilis (39,43). As the external concentrations of potassium are rather low, it is tempting to speculate that L. monocytogenes possesses an additional high-affinity potassium transporter to be able to compete with other bacteria when the extracellular potassium is scarce. The phylogenetically related bacteria B. subtilis and S. aureus contain two high-affinity potassium transport systems that are active during growth at low potassium concentrations. B. subtilis employs the high-affinity potassium transporters KtrAB and KimA under potassium-limiting growth conditions (5,39,43). Previously, it has been shown that S. aureus relies on the

Listerial potassium transporters
high-affinity transporter KdpFABC, whose synthesis and activity is regulated by the two-component system KdpDE (38,44,45). KimA Sau could also be important for growth of S. aureus when the extracellular potassium concentrations are low. In contrast to S. aureus, the KdpABC homolog of L. monocytogenes does not seem to contribute to potassium uptake. It has been shown previously that the small membrane protein KdpF is required for proper function of the E. coli Kdp potassium transport system (46). As described above, no KdpF homolog is present in L. monocytogenes. Therefore, the lack of KdpF in L. monocytogenes could be the reason why the KdpABC system is not active in potassium transport. Interestingly, we found that c-di-AMP binds to the sensor kinases KdpD of the KdpDE two-component system that might be involved in controlling the expression of the kdpABC genes in L. monocytogenes (Fig.  8). However, our comparative RNA-Seq experiments using the WT strain and a c-di-AMP-free cdaA mutant strain in chemically defined medium revealed that the lack of c-di-AMP does not alter the expression of genes involved in potassium uptake (data not shown). In B. subtilis, it has been shown that the 5Ј-UTRs of the kimA and ktrAB genes contain ydaO riboswitches preventing synthesis of the transporters in the presence of c-di-AMP (5, 51). However, c-di-AMP-dependent riboswitches that could be involved in controlling the expres-sion of potassium transporter genes are absent in L. monocytogenes. Therefore, it remains to be elucidated under which conditions the potassium transporter genes are transcribed. Moreover, it has to be investigated whether c-di-AMP controls the expression of the kimA, ktrC, ktrD, and kdpABC genes in L. monocytogenes at all. As stated above, during growth under hyperosmotic conditions many bacteria take up potassium ions to prevent water efflux from the cytosol and to increase the cellular turgor (1)(2)(3)(4). Once the cellular turgor has been adjusted to the environmental osmolarity, the transport of potassium ions across the cell membrane has to be reduced to prevent osmotic swelling and cell lysis (1)(2)(3)(4). A reduction of the ion uptake might be achieved either by proteolytic degradation or by controlling the activity of the transporters through binding of low-molecular weight ligands. It has indeed been shown that transport systems are rapidly degraded when the respective substrates are not available (52). However, the cellular turgor is a physical variable that changes rapidly and needs to be tightly adjusted (1)(2)(3)(4). Thus, it is obvious that the proteolytic degradation of transport systems would be too slow to allow the bacteria to prevent potassium uptake to toxic levels. However, the tight control of the cellular turgor requires the existence of low-molecular weight ligands, which specifically modulate the activity of potassium transport-

Listerial potassium transporters
ers and other osmolyte uptake systems. In S. aureus, it has been shown that the low-affinity potassium transporters KtrCB and KtrCD are inhibited by the second messenger c-di-AMP that binds to the RCK_C (regulator of conductance of K ϩ ) domain of the KtrC gating component (7). Moreover, c-di-AMP binds to the CabP protein and prevents potassium uptake by the CabP-TrkH protein complex in S. pneumoniae (8). The cytoplasmatic regulatory subunit KtrC of the KtrCD potassium transporter is also bound by c-di-AMP in Mycoplasma pneumoniae (53). Recently, it has been demonstrated that the potassium importers KupA and KupB of L. lactis are inhibited by c-di-AMP (38). Here, we show that the potassium transporters KtrCD Lmo and KimA Lmo from L. monocytogenes are inhibited by the second messenger c-di-AMP. This study also revealed that the uncontrolled influx of potassium ions via the KtrAB Bsu and KimA Lmo results in osmotic swelling of E. coli (Fig. 5). Recently, it has been shown that c-di-AMP inhibits the potassium transport activity of the KimA homolog from B. subtilis (39). In this organism, c-di-AMP is required to reduce potassium uptake to toxic levels. As described above, c-di-AMP also controls the uptake of potassium at the level of transcription. For instance, c-di-AMP inhibits the sensor kinase KdpD of the KdpDE two-component system and thus reduces the expression of the kdpFABC operon encoding the high-affinity Kdp-FABC potassium transport system from S. aureus (41,45). Moreover, c-di-AMP prevents the expression of the ktrAB and kimA mRNAs in B. subtilis, thereby reducing expression of the high-affinity potassium transporters KtrAB and KimA, respectively (5,51). It should be noted that c-di-AMP also inhibits the uptake of other osmolytes, such as glycine betaine and carnitine (11-13, 15, 26). Thus, c-di-AMP plays a central role in controlling the activities of potassium transporters and other osmolyte uptake systems, and the c-di-AMP-dependent regulation can occur at two different levels in a variety of bacteria.
Recently, it has been demonstrated that the control of potassium uptake is an essential function of c-di-AMP in B. subtilis (5). A B. subtilis strain lacking all c-di-AMP-producing enzymes was only viable in medium containing low potassium concentrations. c-di-AMP is also essential in bacteria like L. monocytogenes, S. agalactiae, and S. aureus to prevent uptake of osmolytes to toxic levels (10,15,26). However, in these bacteria, the control of glycine betaine and amino acid uptake seems to be the essential function of c-di-AMP. This could explain why an increase of external osmolarity, either by sodium or potassium chloride, rescues the growth defect of a cdaA mutant strain in complex media, irrespective of the ion (11). We furthermore show that high amounts of potassium only slightly inhibit the growth of the cdaA mutant in defined medium (Fig. 7). Thus, the physiological impact of the c-di-AMP-dependent control of the potassium transporters seems to be less pronounced in L. monocytogenes than in bacteria like B. subtilis (5). In fact, phylogenetically related bacteria have evolved species-specific mechanisms to regulate the cellular turgor using different osmolytes, but they all use c-di-AMP in

Listerial potassium transporters
this essential process (6). It remains to be elucidated how c-di-AMP controls potassium homeostasis in L. monocytogenes. Moreover, it will be crucial to identify the osmo-signal-sensing mechanism of the c-di-AMP system, which could be conserved among different bacteria (6,15).

Bacterial strains and growth conditions
The bacterial strains are listed in Table 2. The E. coli strains XL1-Blue (Stratagene), Rosetta (DE3) (Novagen), and T7 Express I q (New England Biolabs) were used for cloning and protein overproduction. E. coli was grown in LB medium, and transformants were selected on LB plates (15 g/liter Bacto agar (Difco)) containing kanamycin (50 g ml Ϫ1 ), ampicillin, carbenicillin (100 g ml Ϫ1 ), or chloramphenicol (30 g ml Ϫ1 ). The L. monocytogenes WT strain EGD-e (laboratory strain collection) was cultivated in brain heart infusion medium (Sigma-Aldrich, Darmstadt, Germany). For the deletion of the cdaA gene, the LSM was used as described previously (11), with the following minor changes (equimolar substitutions): riboflavin-5Ј-monophosphate instead of riboflavin; L-isoleucine, L-methionine, and L-valine instead of the DL-enantiomers; and L-cysteine⅐HCl⅐H 2 O instead of L-cysteine⅐2HCl. For experiments with defined potassium concentrations, the KH 2 PO 4 component of the "phosphate stock solution" was replaced by equimolar concentrated NaH 2 PO 4 , and for the other 43 components, no chemicals containing potassium salts were used. The LSM without potassium was subsequently adjusted to the depicted concentrations, by the addition of KCl. Erythromycin and 5-bromo-4-chloro-3-indolyl-␤-D-galactopyranoside (X-Gal) were used in the deletion process at concentrations of 5 or 100 g ml Ϫ1 , respectively. For pouring minimal medium agar plates, 2-fold concentrated medium was prewarmed to 37°C and mixed with 70°C prewarmed 2-fold Bacto agar, directly before pouring the plates. The B. subtilis WT strain 168 (laboratory strain collection) was cultivated in LB medium. Potassium transporter-deficient E. coli strains LB650 and LB2003 were cultivated in LB-K medium (NaCl substituted by 1% KCl (w/v)) (49). M9 medium was used for E. coli growth experiments with the following composition: 37.85 mM Na 2 HPO 4 , 22.05 mM KH 2 PO 4 , 18.75 mM NH 4 Cl, 1 mM MgSO 4 , 0.1 mM CaCl 2 , 0.5 M FeCl 3 , 28 mM D-glucose or glycerol as sources of carbon. For the E. coli strain LB650, the M9 medium was supplemented with amino acids L-valine, L-isoleucine, L-methionine, L-proline, and L-serine (each 0.02% (w/v)) and 3 M thiamine. For the E. coli strain LB2003, the M9 medium was supplemented with 0.0066% (w/v) casein hydrolysate (acid) (Oxoid), 0.004% (w/v) L-proline, and 3 M thiamine. For experiments with defined potassium concentrations, the KH 2 PO 4 salt was replaced by NaH 2 PO 4 , and KCl was added as indicated.
If not otherwise specified, IPTG was used at a concentration of 50 M, and L-arabinose was used at 0.005% (w/v).

DNA manipulation
Transformation of E. coli was performed using standard procedures (54). Plasmid DNA was extracted using the NucleoSpin Plasmid Kit (Macherey and Nagel). Restriction enzymes, T4 DNA ligase, and DNA polymerases were used as recommended by the manufacturers. DNA fragments were purified using the PCR purification kit (Qiagen). DNA sequences were determined by the Microsynth sequencing laboratories (Göttingen, Germany). Chromosomal DNA of L. monocytogenes or B. subtilis was isolated using the NucleoSpin Microbial DNA Kit (Macherey and Nagel). Chromosomal DNA of S. aureus COL was kindly provided by Dr. Jan Pané-Farré (University of Greifswald, Germany). Oligonucleotides were purchased from Sigma-Aldrich (Darmstadt, Germany).

Plasmid construction
The genes encoding putative potassium transporters were introduced into the vector pWH844, allowing IPTG-dependent expression in E. coli (48). The kimA Lmo and kimA Sau genes were amplified using the oligonucleotide pairs JH95/JH96 and JH97/ JH98, respectively ( Table 3). The PCR products were EcoRI/ BamHI-digested and ligated to pWH844. The resulting plasmids were designated as pBP384 and pBP385 (Table 4). To study the role of the C-terminal domain of KimA Lmo , the plasmid pBP396 was constructed. The truncated kimA Lmo gene was amplified with the oligonucleotide pair JH95/JH120, digested with EcoRI/BamHI, and ligated to pWH844. The plasmid pBP371 for the expression of the L. monocytogenes ktrCD genes was constructed as follows. The ktrC and ktrD genes were amplified using the oligonucleotide pairs JH59/JH60 and JH61/

Listerial potassium transporters
JH62, respectively, and fused by splicing by overhang extension (SOE) PCR using primer pair JH59/JH62 (55). The resulting PCR product was digested with EcoRI and BamHI and ligated to pWH844. The plasmids pBP559 and pBP563 were constructed for the expression of the L. monocytogenes kdpABC genes. The kdpABC genes were amplified using the oligonucleotide pairs MI1/MI2 and MI21/MI2, respectively. The PCR products were digested with BamHI/PstI and KpnI/PstI and ligated to the plasmids pWH844 and pBAD24, respectively. The plasmid pBP560 served for the expression of the L. monocytogenes kdpD gene. The kdpD gene was amplified by PCR with the oligonucleotide pair MI11/MI13. The BamHI/PstI-digested PCR product was ligated to the plasmid pWH844. The plasmids pBP370 and pBP373 were constructed for producing the WT CdaA enzyme and the inactive D171N variant (50). The cdaA gene was amplified using the oligonucleotide pair JH51/JH52 and introduced into the XbaI/PstI sites of pBAD33 (56). For the construction of plasmid pBP373, we used the oligonucleotide pair JH51/JH52 together with the 5Ј-phosphorylated oligonucleotide JR18 to introduce the D171N mutation via the combined chain reaction (57). The pBAD33 and pWH844 expression vectors have compatible selection markers and origin of replications, allowing the co-expression of potassium transporter (from pWH844) and cdaA genes (from pBAD33). The plasmids pBP345, pBP346, and pBP347 were constructed to study the binding of c-di-AMP to KtrC and the cytosolic domains of KimA Lmo (aa 452-607) and KimA Sau (aa 452-609). The respective genes were amplified using the oligonucleotide pairs GH5/GH6, GH7/GH8, and GH9/GH10, digested with BamHI/SalI, and ligated to pWH844 cut with the same enzymes. The genes encoding the full-length KimA Lmo and KimA Sau proteins as well as the C-terminally truncated KimA Lmo variant (aa 1-455) were amplified using oligonucleotide pairs JH142/JH96, JH143/JH98, and JH142/JH120, respectively. The PCR products were digested with SacI/ BamHI and ligated to pGP172 (58). The resulting plasmids were  For the chromosomal deletion of the cdaA gene, pBP352 was constructed ( Table 4). The up-and downstream regions of cdaA, while leaving the cdaA ORF out, were amplified using oligonucleotide pairs JH05/JH06 and JH07/JH08, respectively ( Table 3). The resulting PCR products were fused by SOE PCR using oligonucleotides JH05 and JH08, digested with EcoRI and BamHI, and ligated to pMAD (55,59), which was digested using the same enzymes.

Deletion of the cdaA gene
The chromosomal deletion of the cdaA gene in strain BPL77 was performed as follows. The plasmid pBP352 (pMAD-⌬cdaA) was introduced into the WT strain EGD-e by electroporation, and the cells were plated on LSM with erythromycin and X-Gal at 30°C for up to 72 h. Single blue colonies were streaked on the same medium and incubated for up to 72 h at 42°C to facilitate the selection for integrants. Blue colonies were used to inoculate 5 ml of LSM without antibiotics at 30°C for 4 h, and the temperature was shifted to 42°C for 6 h, after which serial dilutions were plated on LSM with X-Gal and incubated at 37°C for up to 72 h. Erythromycin-sensitive, X-Galnegative bacteria that did grow on LSM but not on brain heart infusion were subjected to colony PCR as described previously (60). The cdaA deletion and the absence of ectopic suppressor mutations was confirmed by whole-genome sequencing and Sanger sequencing, and the strain was designated BPL77 ( Table 2).

Growth of L. monocytogenes in LSM
Single colonies of the L. monocytogenes WT and the cdaA mutant strains were grown overnight in LSM-K ϩ with 1 mM KCl. Overnight cultures were harvested by centrifugation at 4000 ϫ g for 5 min at room temperature and resuspended in LSM-K ϩ . These cell suspensions were used to inoculate 10 ml of LSM-K ϩ to an OD 600 of 0.1 and grown for about 4 h. Cells were washed again as described in LSM-K ϩ , the OD 600 was adjusted to 0.2, and 100 l were used to inoculate wells of a 96-well plate (Microtest Plate 96 Well, F, Sarstedt), containing 100 l of LSM-K ϩ with a 2-fold concentration of the indicated potassium concentrations. The 96-well plate was incubated at 37°C with medium orbital shaking at 237 cpm (4 mm) in an Epoch 2 microplate spectrophotometer (BioTek Instruments), and growth was measured at an optical density (OD 600 ) in 15-min intervals.

Drop dilution assay
Single colonies of the E. coli strain LB650 harboring the plasmid pWH844, pBP371, pBP372, pBP384, pBP385, or pBP396 were taken from LB-K plates and used to inoculate 4 ml of LB-K medium supplemented with kanamycin, ampicillin, and chloramphenicol. The cultures were incubated at 37°C and 220 rpm. The precultures were used to inoculate 4 ml of M9 medium supplemented with glucose, antibiotics, and 50 mM KCl to an OD 600 of 0.001. The cultures were incubated for about 16 h at 37°C. The next day, the cultures were used to inoculate 10 ml of the same medium to an OD 600 of 0.1. At an OD 600 between 0.3 and 0.5, the cells were harvested by centrifugation at 4000 ϫ g for 10 min at room temperature. The cell pellets were washed twice in 10 ml of M9 medium lacking KCl. The cell suspension was adjusted to an OD 600 of 0.1, and 5 l of the diluted cells were spotted onto M9 minimal medium plates, which were incubated for 24 h at 37°C. M9 plates were prepared by mixing 2ϫ M9 medium (prewarmed to 37°C) and 2ϫ Bacto agar (prewarmed to 70°C before mixing). The finial medium contained glucose as a carbon source, 10 mM KCl, and 50 M IPTG when required.

Determination of kinetic parameters of the potassium transporters
To determine the growth characteristics of the E. coli strain LB650 synthesizing potassium transporters from L. monocytogenes and S. aureus, the bacteria were grown until the early exponential phase and harvested by centrifugation at 4000 ϫ g for 10 min. The pellet was resuspended in 10 ml of M9 medium with glucose, ampicillin, and 50 M IPTG without KCl. The cells were incubated for 1 h at 37°C, harvested by centrifugation, and washed twice. The cultures were adjusted to an OD 600 of 0.2, and 50 l were used to inoculate a 96-well plate (Microtest Plate 96 Well, F, Sarstedt) containing 50 l of M9 medium with glucose, ampicillin, 50 M IPTG, and KCl concentrations ranging from 0 to 100 mM. The 96-well plate was incubated at 37°C with medium orbital shaking at 237 cpm (4 mm) in an Epoch 2 microplate spectrophotometer (BioTek Instruments). The growth rates were calculated ( ϭ (2.303⅐(log(OD 2 ) Ϫ log(OD 1 )))/(t 2 Ϫ t 1 )), plotted against the KCl concentrations, and fitted to the Michaelis-Menten equation using the solver tool of Excel 2013 (Microsoft) to calculate V max ( (h Ϫ1 )) and the apparent K m (mM KCl).

c-di-AMP in vivo inhibition assay
The potassium transporter-deficient E. coli strain LB2003 was co-transformed with the plasmid pWH844 or derivatives Listerial potassium transporters (pBP371, pBP384, or pBP396) and the pBAD33 derivatives (pBP370 or pBP373) on LB-K plates containing 0.5% (w/v) glucose, ampicillin, and chloramphenicol. Single colonies were used to inoculate 4 ml of LB-K medium containing 0.2% (w/v) glucose, ampicillin, and chloramphenicol, and the exponentially growing cultures were used to inoculate M9 medium containing 0.2% (w/v) glycerol and 0.02% (w/v) glucose to an OD 600 of 0.001. The cultures were incubated overnight at 37°C and used to re-inoculate the same medium (without glucose) to an OD 600 of 0.1. After reaching early exponential phase (OD 600 ϭ 0.3-0.5), the cells were washed, and 50 l of the suspensions were used to inoculate a 96-well plate. The M9 medium was supplemented with glycerol, 50 M IPTG, ampicillin, chloramphenicol, and KCl with or without L-arabinose. Final concentrations of KCl were equal to the determined K m values (see Table 2), and either no or 0.005% (w/v) L-arabinose was present, as indicated. Growth was monitored in an Epoch 2 microplate spectrophotometer (BioTek Instruments).

Protein expression and DRaCALA
The binding of c-di-AMP to the potassium transporters was analyzed using the E. coli strain Rosetta (DE3) for pGP172 and derivatives or strain NEB T7 Express I q for pWH844 and derivatives. Single colonies were used to inoculate 10 ml of LB-K medium containing carbenicillin and chloramphenicol. After incubation overnight at 30°C, the precultures were used to inoculate 1.5 ml of LB-K medium to an OD 600 of 0.1. 1 mM IPTG was added at an OD 600 of 1.0 -1.5 to induce gene expression. After incubation for 4 h, the cultures were harvested by centrifugation (4000 ϫ g, 10 min, 4°C), the cell pellets were resuspended in 150 l of Tris-NaCl buffer (10 mM Tris, pH 8.0, 100 mM NaCl). Cells were lysed by three freeze/thaw cycles of Ϫ80°C and room temperature. DRaCALA was performed by mixing 1 l of [ 32 P]c-di-AMP with 20 l of cell lysates. After a 1-min incubation, 2 l of the mixture was spotted on dry nitrocellulose, dried, exposed to a PhosphorImager screen, and imaged using an FLA-7000 PhosphorImager. The fraction bound was calculated using the inner and total areas and intensities, as described previously (61).

Microscopic analysis
Derivatives of the LB650 strain harboring the plasmids pWH844 (empty plasmid), pBP372, or pBP384 were in 4 ml of LB-K medium containing ampicillin, kanamycin, and chloramphenicol at 37°C. The next day, the cultures were washed twice and used to inoculate 10 ml of M9 medium (containing 22.05 mM KH 2 PO 4 ) with or without 1 mM IPTG to an OD 600 of 0.1. Cells were transferred to standard microscope slides (Carl Roth) and examined using an Axioskop 40 FL fluorescence microscope, equipped with an Axio-Cam MRm digital camera, objectives of the Neofluar series at 1000-fold primary magnification, and the AxioVision Rel 4.8.2 software (Carl Zeiss). Images were later equally processed using ImageJ 1.48 software (62).

Statistical analysis
All data are presented as means with n representing the number of independent experiments. Data were statistically evalu-ated by analysis of variance (ANOVA) tests with post hoc Dunnett's or Tukey tests using the GraphPad Prism version 8.2.1 software (GraphPad Software, La Jolla, CA).