By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
Quorum sensing, the population density-dependent regulation mediated by N-acylhomoserine lactones (AHSL), is essential for the control of virulence in the plant pathogen Erwinia carotovora ssp. carotovora (Ecc). In Erwinia carotovora ssp. the AHSL signal with an acyl chain of either 6 or 8 carbons is generated by an AHSL synthase, the expI gene product. This work demonstrates that the AHSL receptor, ExpR1, of Ecc strain SCC3193 has strict specificity for the cognate AHSL 3-oxo-C8-HSL. We have also identified a second AHSL receptor (ExpR2) and demonstrate a novel quorum sensing mechanism, where ExpR2 acts synergistically with the previously described ExpR1 to repress virulence gene expression in Ecc. We show that this repression is released by addition of AHSLs and appears to be largely mediated via the negative regulator RsmA. Additionally we show that ExpR2 has the novel property to sense AHSLs with different acyl chain lengths. The expI expR1 double mutant is able to act in response to a number of different AHSLs, while the expI expR2 double mutant can only respond to the cognate signal of Ecc strain SCC3193. These results suggest that Ecc is able to react both to the cognate AHSL signal and the signals produced by other bacterial species.
Quorum sensing (QS) is a central cell-to-cell communication system that bacteria employ to monitor their population density and coordinate functions for which high population density is profitable (Fuqua et al., 2001; Waters and Bassler, 2005). QS is achieved by production and sensing of diffusible chemical signals that in gram-negative bacteria are usually N-acylhomoserine lactones (AHSLs). AHSLs control a number of diverse functions in bacteria, such as bioluminescence (Eberhard et al., 1981; Engebrecht and Silverman, 1984), conjugal transfer (Zhang et al., 1993), production of antibiotics as well as secondary metabolites and virulence in both plant and animal pathogens (Whitehead et al., 2001). QS is required for the production of virulence determinants and biofilm formation in the opportunistic human pathogen Pseudomonas aeruginosa (Winson et al., 1995), exopolysaccharide production in the plant pathogen Pantoea stewartii (Beck von Bodman and Farrand, 1995) and production of plant cell wall degrading enzymes (PCWDEs) and antibiotics in another plant pathogen Erwinia carotovora (Jones et al., 1993; Pirhonen et al., 1993).
The LuxR/LuxI system controlling bioluminescence in Vibrio fischerii was the first to be characterized and has become the paradigm of QS. The QS system is minimally executed by an AHSL synthase, a LuxI-type protein, and a QS regulator, a LuxR-type protein, controlling transcription of downstream genes (Fuqua et al., 2001; Whitehead et al., 2001). There is high degree of specificity in QS determined by substrate specificity of the AHSL synthase and specific recognition of the cognate AHSL by the LuxR-type protein, modulating the expression of QS-regulated target genes. Different bacteria produce AHSLs with diverse acyl side chain lengths, ranging from 4 to 16 carbons, and with alterations in the oxidative status of carbon 3 (Fuqua and Greenberg, 2002; Pappas et al., 2004). The LuxR-type proteins act as QS regulators, and they distinguish between different AHSLs by showing binding of cognate, but not non-cognate AHSLs suggesting that they are mainly involved in intraspecies signaling (Lazdunski et al., 2004). The LuxR-type proteins share a similar structure with a ligand (AHSL) recognizing domain at the amino-terminus (N-terminus) and usually a very conserved DNA-binding domain at the carboxy-terminus (C-terminus). Although LuxR-type proteins have similar structures, their operative mechanisms can be different. Many LuxR-type activators, including CarR, LuxR and TraR, the only crystallized LuxR-type protein so far (Qin et al., 2000; Vannini et al., 2002; Zhang et al., 2002), form dimers or multimers upon binding to AHSLs (Whitehead et al., 2001; Pappaset al., 2004). This stable complex then binds to a target gene promoter, in many cases to a 20 bp palindromic lux box, in order to activate target gene transcription (Fuqua et al., 2001; Lazdunski et al., 2004). In contrast, EsaR, a LuxR-type protein of the plant-pathogen Pantoea stewartii (Beck von Bodman and Farrand 1995) binds the target DNA in the absence of AHSL and represses transcription, but after addition of the cognate AHSL EsaR is thought to be released from DNA and the target gene is derepressed (Minogue et al., 2002; 2005).
QS is central to regulation of virulence of the gram-negative, broad host range plant pathogen Erwinia carotovora ssp. carotovora (Ecc), and it also controls production of carbapenem antibiotics in some strains of Ecc (Pirhonen et al., 1991; 1993; Jones et al., 1993). We have shown that AHSL synthesis is required for the production of PCWDEs, such as cellulases, polygalacturonases and pectinases, the main virulence determinants of this pathogen and that QS is responsible for density-dependent and coordinated production of these enzymes to establish a successful infection (Pirhonen et al., 1991; 1993). Mutants deficient in the AHSL synthase (the expI gene product) are impaired in the production of PCWDEs and are thus avirulent (Pirhonen et al., 1991; 1993; Jones et al., 1993). Depending on the strain, the main AHSLs produced and recognized by Ecc are 3-oxo-hexanoylhomoserine lactone (3-oxo-C6-HSL) or 3-oxo-octanoylhomoserine lactone (3-oxo-C8-HSL) with 3-oxo-C8-HSL being the cognate AHSL of the Ecc strain SCC3193 (Brader et al., 2005). The so far best characterized LuxR-type protein of Ecc is CarR, which positively regulates the production of carbapenem antibiotics, in response to its cognate autoinducer 3-oxo-C6-HSL in the Ecc strain GS101 (Welch et al., 2000).
The QS regulators controlling PCWDE production have so far remained more elusive. We originally identified ExpR of the Ecc strain SCC3193 as a potential QS regulator (Andersson et al., 2000). Although mutations in expR did not show a clear phenotype, overexpression studies suggested that ExpR might act as repressor of PCWDE synthesis (Andersson et al., 2000). Interestingly, von Bodman et al. (2003) have demonstrated that ExpRSCC3193 can bind to DNA in the absence of AHSL, but that this binding is inhibited by AHSL addition. A recent study by Cui et al. (2005) showed that a related LuxR-type protein ExpREcc71 from another Ecc strain Ecc71 binds to the promoter of a target gene rsmA, activating its transcription in an AHSL free state. Addition of the cognate AHSL released the ExpREcc71 from this promoter leading to repression of the target gene that encodes the global negative regulator RsmA (Chatterjee et al., 2005; Cui et al., 2005). RsmA is an RNA-binding protein that represses the production of PCWDEs (Chatterjee et al., 1995). These results supported a close relation between RsmA and the QS system already indicated in previous studies (Chatterjee et al., 1995; Kõiv and Mae, 2001).
In this study we identify a novel QS regulation where two LuxR-type proteins, ExpR1 and ExpR2, act synergistically as negative regulators of PCWDE production in the Ecc strain SCC3193. This negative regulation released by accumulation of AHSLs appears to be largely mediated by the global negative regulator RsmA. Intriguingly, we demonstrate that the two ExpR proteins have distinct AHSL specificities: while ExpR1 is specific to the cognate AHSL, the newly identified ExpR2 protein shows broad signal sensing capacity and responds also to non-cognate AHSL, allowing both intra- and interspecies communication.
Inactivation of expR alters AHSL sensing specificity
We have demonstrated that the AHSL synthase encoded by expI of Ecc strain SCC3193 (Pirhonen et al., 1993) produces mainly 3-oxo-C8-HSL (Brader et al., 2005). An expI mutant (SCC3065) of SCC3193 is not able to produce the PCWDEs (Pirhonen et al., 1993), but can be specifically rescued by addition of the cognate 3-oxo-C8-HSL (Brader et al., 2005). A much higher concentration (200-fold) of the non-cognate AHSL 3-oxo-C6-HSL is required to rescue the PCWDE negative phenotype of this mutant (Brader et al., 2005). To address the role of the ExpR protein of SCC3193 (Andersson et al., 2000) in this recognition, we characterized the ability of the added non-cognate 3-oxo-C6-HSL or cognate 3-oxo-C8-HSL to restore the cellulase (Cel) activity in the expI mutant (SCC3065) and the expI expR double mutant (SCC6005) (Fig. 1) using a Carboxymethylcellulose (CMC) plate assay (Pirhonen et al., 1993). Interestingly, the presence or absence of the expR gene had a profound effect on AHSL sensing specificity; while the Cel activity of the expI mutant was restored only by the cognate 3-oxo-C8-HSL, in the expI expR double mutant the Cel production was restored both by the non-cognate 3-oxo-C6-HSL and the cognate 3-oxo-C8-HSL. To confirm that this altered specificity was due to loss of the ExpR function the expRSCC3193 gene was expressed in trans in the expI expR double mutant (SCC6005) background. This resulted in restoration of the requirement for the cognate 3-oxo-C8-HSL in Cel production (Fig. 1). These results demonstrate that the expR mutant has broader substrate specificity and suggest that ExpR is involved in sensing of 3-oxo-C8-HSL.
Identification of a second expR homologue in Ecc strain SCC3193
The altered AHSL sensing specificity (Fig. 1) and the lack of clear PCWDE phenotype (Andersson et al., 2000) of the expR mutant could be explained by the existence of an additional ExpR homologue or possibly another partly redundant QS regulator responsible for sensing the non-cognate 3-oxo-C6-HSL. Interestingly, two expR homologues have been reported in the newly sequenced genome of Erwinia carotovora ssp. atroseptica strain SCRI1043 (Eca); one (expR) is linked to the expI gene similar to the organization of expI and expR in SCC3193 (Andersson et al., 2000) and the other one (ECA1561) exists separately (Bell et al., 2004). To explore the possibility of a second expR homologue in Ecc strain SCC3193 we used the sequence data of Eca strain SCRI1043 to design primers for PCR identification of a potential second expR homologue. Subsequently the presence of a second expR gene in SCC3193 was demonstrated (designated expR2, with the previously identified/characterized expR renamed as expR1). The genomic organization of sequences flanking expR2SCC3193 is rather similar between the two subspecies of Erwinia carotovora. Downstream of the expR2SCC3193 gene is an open reading frame (orf) of 252 aa that is 83% identical to a CDP-diacylglycerol pyrophosphatase of Eca strain SCRI1043, while in the upstream region an approximately five kb fragment, present in the Eca genome, is lacking between expR2 and the next orf (partially sequenced) that is 88% identical to a chemotaxis signal transduction protein (ECA1568) of the Eca strain SCRI1043 (Fig. 2A).
The expR2 orf consists of 735 bp encoding a putative ExpR2 protein of 245 aa. Structural comparison of ExpR2 with other LuxR-type proteins suggested that ExpR2 is likely a QS regulator (Fig. 2B). The ExpR2 protein shows 94% aa identity (97% similarity) to a potential QS regulator (ECA1561) of Eca SCRI1043 and 62% aa identity (81% similarity) to ExpR of Eca SCRI1043, 58% aa identity (81% similarity) to ExpR1 of Ecc SCC3193, 62% aa identity (80% similarity) to a putative ExpR of Ecc SCC1, 54% aa identity to EsaR of Pantoea stewartii ssp. stewartii and 22% aa identity (46% similarity) to TraR of Agrobacterium tumefaciens (Fig. 2B). The DNA binding domain is highly conserved between the Erwinia carotovora and the Pantoea strains, while the AHSL binding domains show more sequence variety. Taken together, these results indicate that Ecc SCC3193 has indeed two LuxR-type proteins and suggest that this redundancy could explain the lack of a clear PCWDE phenotype in expR1 mutants.
Inactivation of both expR1 and expR2 suppresses the cellulase-negative phenotype of an AHSL deficient strain
To explore the functional roles of the two ExpR proteins in the QS system of E. carotovora we generated both single and double expR1 and expR2 insertion mutants in the wild-type and expI genetic backgrounds of SCC3193. Interestingly, neither the expR1 or expR2 single mutants nor the expR1 expR2 double mutant showed any clear impairment of PCWDE production or virulence in the wild-type SCC3193 background suggesting that the corresponding proteins do not act as positive regulators of virulence. As there was the distinct possibility that one or both of the ExpR proteins of SCC3193 could function as negative regulators of virulence and enzyme production as suggested previously for ExpR1, EsaR and for the ExpREcc71 and VirR of other E. carotovora strains (von Bodman et al., 1998; Andersson et al., 2000; Cui et al., 2005; Burr et al., 2006) we assessed the phenotypic effects of expR1 and expR2 mutations in the expI mutant background. The results firstly demonstrate that the presence of either ExpR1 or ExpR2 is sufficient for the PCWDE-negative phenotype of expI mutants (Fig. 3A). In contrast, inactivation of both expR genes in the expI mutant background restores PCWDE production to wild-type levels (Fig. 3A). This was initially shown for Cel production (Fig. 3A), but a similar restoration of other enzyme activities including polygalacturonase (Peh) and protease (Prt) was also evident (data not shown). The effect was independent of growth conditions as the PCWDE phenotypes were identical in both rich and minimal medium. These data strongly indicate that both ExpR1 and ExpR2 act as negative regulators of PCWDE production and hence of virulence in Ecc and suggest that the function of AHSLs could be in relieving this repression at high population density.
AHSL specificity of ExpR1 and ExpR2
The generation of expI expR1 and expI expR2 double mutants allowed us to explore the AHSL specificity of each ExpR protein. Mutant strains were grown in the presence of either the cognate autoinducer 3-oxo-C8-HSL or the non-cognate 3-oxo-C6-HSL and Cel activity was characterized (Fig. 3A). The analysis showed that the expI expR2 mutant (SCC908) responded specifically to 3-oxo-C8-HSL, in contrast to the broader specificity shown by the expI expR1 double mutant (SCC6005). According to these results ExpR1 is a specific LuxR-type protein, activating Cel production only after addition of 3-oxo-C8-HSL. At physiological AHSL concentrations ExpR1 was indeed only able to respond to the cognate signal 3-oxo-C8-HSL, while the addition of 3-oxo-C6-HSL did not suppress the Cel negative phenotype of the expI expR2 double mutant. Addition of 3-oxo-C6-HSL to a concentration of 10 μM or more was required to restore Cel activity to this mutant (data not shown).
To determine the specificity of ExpR1 and ExpR2 as AHSL receptors, expI expR1 and expI expR2 double mutants were exposed to a series of AHSLs: C4-HSL, C6-HSL, 3-oxo-C6-HSL, C7-HSL, C8-HSL, 3-oxo-C8-HSL, 3-oxo-C10-HSL, C12-HSL or 3-oxo-C14-HSL, and assayed for Cel activity (Fig. 3B). As suggested by previous results (Fig. 3A) ExpR1 appeared to have a narrow AHSL specificity responding best to the cognate 3-oxo-C8-HSL, although it also could respond to some extent to 3-oxo-C10-HSL and to C8-HSL. On the other hand ExpR2 appeared to be promiscuous and respond to all the AHSLs tested except for the shortest (C4-HSL) and the longest ones (C12-HSL and 3-oxo-C14-HSL). Thus ExpR2 had a much broader AHSL recognition capacity compared to ExpR1, which clearly is the more specific LuxR-type protein mainly responding to the cognate 3-oxo-C8-HSL.
Having established the specificity of AHSL sensing for each ExpR protein, we wanted to explore the joint effect of the two ExpR proteins in PCWDE regulation. As a tool we employed the triple mutant (SCC906) lacking the AHSL synthase as well as the two ExpR proteins. As shown this triple mutant exhibited wild-type levels of Cel activity without the addition of any AHSL (Fig. 3A and C). Complementation studies were used to demonstrate the specific roles for each ExpR protein. Plasmids carrying either expR1 or expR2 were introduced to the expI expR1 expR2 triple mutant and Cel activities were assessed in the presence and absence of 3-oxo-C6-HSL or 3-oxo-C8-HSL (Fig. 3C). These results support the hypothesis that both ExpR proteins have distinct roles in AHSL recognition. Introduction of expR1 into the triple mutant made the strain dependent of the cognate 3-oxo-C8-HSL in Cel production whereas introduction of expR2 generated a dependency of a variety of AHSLs. These data demonstrate that the ExpR1 and ExpR2 have individual roles in AHSL recognition and indicate that either ExpR is sufficient to repress PCWDE production.
ExpR1 and ExpR2 mediated repression of PCWDE activity is population density-dependent
Since the expI expR1 expR2 triple mutant was constitutively Cel positive without the addition of AHSL, we wanted to elucidate the role of the QS system in population density-dependent regulation of PCWDEs. Nothern analysis was used to explore the role of ExpR1 and ExpR2 in regulation of PCWDE genes in the presence of endogenous AHSLs. We compared PCWDE gene expression in the wild-type (SCC3193), the expR1 expR2 mutant (SCC907) and the expI expR1 expR2 mutant (SCC906) strains during the logarithmic growth using PCWDE gene specific probes (Fig. 4). Both mutant strains exhibited substantially up-regulated PCWDE transcript accumulation compared to the wild-type strain. Similar expression levels were only reached by the wild-type at the stationary growth phase. In accordance with this also the production of one of the enzymes, Cel was clearly up-regulated in the expR1 expR2 double mutant at the early phases of growth (Fig. 4). This analysis demonstrates that ExpR1 and ExpR2 are essential for the growth phase dependent control of PCWDE production.
ExpR1 and ExpR2 control expression of RsmA, a negative regulator of PCWDE
Our results (Figs 3 and 4) demonstrate that ExpR1 and ExpR2 are negative regulators of PCWDE gene expression and PCWDE production. However, they do not indicate weather this repression is direct or mediated through some other regulator. An involvement of the global negative regulator RsmA was recently suggested by Cui et al. (2005) showing that in Ecc strain Ecc71 ExpR71 binds to the rsmA promoter and activates its expression in the absence of AHSL. The activation of rsmA transcription was prevented by the addition of 3-oxo-C6-HSL (Chatterjee et al., 2005; Cui et al., 2005). We explored the possibility of an RsmA-mediated mechanism of ExpR1 and ExpR2 control in Ecc SCC3193 using an rsmA-gusA promoter fusion. The rsmA promoter from Ecc strain SCC3193 used in this construct contained a putative ExpR-box, similar to the ExpR-binding site defined in Ecc71 (Chatterjee et al., 2005). Our results from β-glucuronidase (GUS) assays in the following mutant backgrounds expI, expI expR1, expI expR2 and expI expR1 expR2 demonstrate that in the absence of AHSL ExpR1 and ExpR2 were both, either together or separately, able to activate the expression of rsmA (Fig. 5A). In contrast, in the expI expR1 expR2 triple mutant the rsmA expression was significantly down-regulated with or without the addition of AHSLs suggesting that at least one of the ExpRs is needed for the transcriptional activation of rsmA.
To further elucidate the role of ExpR1 and ExpR2 in the transcriptional activation of rsmA the effect of 3-oxo-C6-HSL and 3-oxo-C8-HSL was characterized (Fig. 5A). Addition of the cognate AHSL of SCC3193 3-oxo-C8-HSL, resulted in a substantially decreased GUS activity in both expR1 and expR2 mutants. This argues that both ExpR1 and ExpR2 are able to bind the cognate autoinducer, which will subsequently prevent the activation of rsmA. The residual GUS activity was at the level of that found in the expI expR1 expR2 triple mutant. In contrast addition of the non-cognate AHSL 3-oxo-C6-HSL led to substantially decreased GUS activity only in the expI expR1 mutant, likely due to release/inactivation of ExpR2. ExpR2 was not able to activate the transcription of rsmA in the presence of 3-oxo-C6-HSL, while under the same growth conditions ExpR1 activates the transcription of rsmA in the expI expR2 mutant. These data demonstrate that in the absence of AHSLs either of the ExpRs is sufficient for transcriptional activation of rsmA, which in turn down-regulates expression of PCWDE genes. These data also suggest that accumulation of the cognate AHSL (3-oxo-C8-HSL) will release both ExpRs from rsmA leading to coordinate activation of PCWDE genes. Interestingly, also the non-cognate AHSL (3-oxo-C6-HSL) appears to release/inactivate one of the ExpRs, ExpR2, suggesting that signals from other bacterial species may modulate this response.
The binding of AHSL by ExpR1 and ExpR2 correlates with rsmA transcriptional activity
To provide additional evidence for the mode of action of the ExpR1 and ExpR2 proteins and to elucidate whether the transcription of rsmA is indeed controlled by AHSL binding to ExpRs, we measured AHSL binding using the same strains that were characterized for rsmA transcriptional activation (Fig. 5B). As expected the triple mutant, lacking both ExpR proteins, was not able to bind either of the AHSLs used, indicating that the presence of ExpR proteins is indeed required for AHSL binding. In contrast, the expI mutant with both ExpR proteins present binds effectively both the cognate 3-oxo-C8-HSL and the non-cognate 3-oxo-C6-HSL. Binding of the cognate AHSL in expI mutant background was clearly more effective than in strains lacking either one of the ExpR proteins indicating that ExpR1 and ExpR2 cooperate in this binding. The AHSL binding studies also support the observed specificity of the AHSL action. The expI expR1 double mutant with only ExpR2 present was able to bind 3-oxo-C6-HSL equally well as the expI mutant indicating that 3-oxo-C6-HSL binding can be explained by the binding capacity of ExpR2. On the other hand the expI expR2 mutant with the specific ExpR1 protein present was as expected able to bind only the cognate 3-oxo-C8-HSL. These AHSL binding results correlate well with the ExpR mediated rsmA regulation shown with the rsmA promoter fusion assay. Taken together these results strongly indicate that ExpR1 and ExpR2 act synergistically in binding of AHSLs and suggest that this binding leads to down-regulation of rsmA transcription.
Role of RsmA in QS regulation of PCWDEs
Our data show that QS controls rsmA expression and suggest that QS regulation of PCWDEs could be mediated by RsmA. However, our data does not exclude the presence of other RsmA-independent mechanisms by which ExpR proteins could regulate the PCWDE production. To explore this possibility we employed an expI rsmA double mutant. The effect of rsmA on PCWDE production was assessed by comparing Cel activities of an expI mutant and an expI rsmA double mutant in the presence and absence of AHSLs (Fig. 6; Table 1). In accordance with earlier studies, using another Ecc strain (Chatterjee et al., 1995), the expI rsmA double mutant of SCC3193 showed a Cel positive phenotype even in the absence of AHSL supporting the role of RsmA as a major QS regulator of PCWDE production in Ecc SCC3193. Interestingly, addition of the cognate, but not the non-cognate AHSL to the expI rsmA double mutant resulted in somewhat higher Cel activity. To quantify this effect we assayed Cel activity and could show a 30% increase in Cel activity with the cognate AHSL added (Fig. 6). These results suggest that QS control of PCWDE production is largely mediated by RsmA but that in response to the cognate 3-oxo-C8-HSL there is an additional RsmA-independent pathway to fine tune PCWDEs production.
Table 1. The Cel activity was additionally measured with a quantitative Cel assay, measuring the amount of nmol reduced sugar/h/OD600 released.
Parallel samples to the experiment in Fig. 6 were used.
20 ± 8
17 ± 13
60 ± 4
96 ± 9
97 ± 9
130 ± 1
The AHSL specificity of Ecc strain SCC3193 can be altered with an ExpR protein from Ecc strain SCC1
To further elucidate the specificity of the QS system among Erwinia strains, we tested whether the system is compatible with the LuxR-type protein expRSCC1 from SCC1, a 3-oxo-C6-HSL producing strain of Ecc (Brader et al., 2005). The expRSCC1 gene was expressed in trans in different expR1 and expR2 mutant backgrounds of the expI mutant of SCC3193, and the specificity for 3-oxo-C6-HSL and 3-oxo-C8-HSL was characterized by assessing the Cel phenotype (Fig. 7). Introduction of the expRSCC1 gene into the triple mutant (expI expR1 expR2) background repressed the Cel-positive phenotype of this mutant demonstrating that ExpRSCC1 is functional in this heterologous background. The repression could be released by the cognate AHSL of the SCC1 strain (Brader et al., 2005) 3-oxo-C6-HSL, but not by the cognate AHSL (3-oxo-C8-HSL) of SCC3193 strain (Fig. 7). Thus, introduction of expRSCC1 into the triple mutant (SCC906) changed the strain's AHSL sensing specificity from 3-oxo-C8-HSL to 3-oxo-C6-HSL. The specificity of the ExpRSCC1 protein was also evident when expRSCC1 was expressed in expI expR1 mutant background (SCC6005). The expI expR1 mutant strain with the broader AHSL ligand specificity was altered to the specific 3-oxo-C6-HSL sensing when expressing expRSCC1 in trans. Expression of expRSCC1 in trans in the presence of ExpR1 as in the expI mutant (SCC3065) or the expI expR2 double mutant strain (SCC908) resulted in a Cel-negative phenotype independent of the addition of AHSL. This might be explained by the simultaneous presence of two AHSL receptors ExpR1SCC3193 and ExpRSCC1 with distinct AHSL specificities. These data demonstrate that the AHSL sensing specificity of the QS system in SCC3193 is determined by the ExpR1 protein of either Ecc strain SCC3193 or SCC1.
Lack of a functional quorum sensing system enhances plant maceration
The QS system is one of the most important virulence regulators in Ecc and controls PCWDE production required for plant tissue maceration during infection. To elucidate the role of the ExpR proteins on maceration capacity, we inoculated potato tubers (Van Gogh cultivar) and Arabidopsis wild-type (Col-0) plants with wild-type strain SCC3193, and the different mutants and the extent of maceration was determined (Fig. 8). As expected the avirulent expI mutant showed almost no maceration, while the expR1 expR2 double mutant and the expI expR1 expR2 triple mutant strains showed even slightly enhanced maceration compared to the wild-type (Fig. 8A and B). Wild-type Arabidopsis plants were inoculated with the same Ecc strains as used for potato above. Similar results as in the potato assay were obtained with Arabidopsis: while no maceration was observed in plants inoculated with the expI mutant, clear maceration was evident in plants inoculated with wild-type, the expR1 expR2 double mutant or the expI expR1 expR2 triple mutant (Fig. 8A). These results are in agreement with our results from both the PCWDE assays and Northern analysis, showing enhanced production of PCWDEs, the main virulence factors of Ecc in strains lacking both ExpR proteins.
QS is central in the control of virulence and the production of main virulence determinants like PCWDEs in the plant pathogen Erwinia carotovora ssp. carotovora (Ecc). Previous studies have established that the QS system of Ecc strain SCC3193 consists of AHSL synthase, encoded by the expI gene, producing 3-oxo-C8-HSL as cognate AHSL, and a convergently transcribed expR gene, encoding a QS regulator ExpR (renamed ExpR1 in this article), which we proposed to be a negative regulator of PCWDEs in the absence of AHSL (Pirhonen et al., 1993; Andersson et al., 2000; Brader et al., 2005). In this study we present several major results: First we demonstrate that two ExpR proteins, ExpR1 and ExpR2, exist in Ecc SCC3193 and that ExpR1 and ExpR2 act synergistically to repress the production of PCWDEs and thus also virulence in the absence of AHSL. Second, we show that the two ExpR proteins produced by the Ecc strain SCC3193, have distinct AHSL recognition and binding specificities with ExpR1 responsible for recognition of cognate AHSL and ExpR2 responsible for recognition of both cognate and non-cognate AHSLs. Third, we provide data suggesting that the ExpR control of PCWDE gene expression is to a large extent, but not solely, mediated by the negative regulator RsmA (Fig. 9).
The lack of a clear PCWDE phenotype in the expR1 mutant (Andersson et al., 2000) suggested a redundancy in the AHSL recognition system of Ecc SCC3193. That this was indeed the case was demonstrated in this study by identifying a second LuxR-type protein ExpR2 in Ecc SCC3193. Our studies clearly demonstrated that (i) both ExpR1 and ExpR2 function as AHSL receptors and indeed bind AHSLs and (ii) they act as repressors of PCWDE production in the absence of AHSLs. By inactivating the QS system of SCC3193, including the expI, expR1 and expR2 genes, we could show that the triple mutant strain was able to produce wild-type levels of PCWDEs without the addition of any AHSLs and macerated plant tissue as well as or even better than the wild-type. The presence of either ExpR1 or ExpR2 in the expI mutant background essentially abolished the production of PCWDEs in the absence of AHSLs. This repression was relieved by addition of the cognate 3-oxo-C8-HSL to both expI expR1 and expI expR2 double mutants.
These results suggest a novel QS mechanism where two partially redundant QS regulators act in concert to control a single characteristic, i.e. PCWDE production. The repression by single ExpRs was not always absolute: an expI expR1 double mutant showed partial relief from the PCWDE repression, observed as haloes around bacterial colonies grown on CMC indicator plates (data not shown). This was not seen from the culture supernatants of liquid cultures probably due to still too low level of PCWDE production. This result is in agreement with recent findings that show partial restoration of PCWDE production in an expI expR double mutant of another Ecc strain (Cui et al., 2005) and the slight increase in PCWDEs in the expR1 single mutant strain (Andersson et al., 2000). Also the recently sequenced Eca strain SCR1043 (Bell et al., 2004) harbors two expR genes in its genome that are also organized in a similar fashion as in Ecc SCC3193 suggesting a similar type of QS regulation as observed here. During preparation of this manuscript Burr et al. (2006) showed mutant analysis of the expR2 homologue of Eca (virR) suggesting that this gene codes for a repressor of PCWDE production. Interestingly, the virR mutant of Eca showed partial restoration of PCWDE production suggesting that the Eca ExpR (Ecc ExpR1 homologue) might not be as strong repressor of PCWDEs as ExpR1 in Ecc.
How is the control of PCWDE production by AHSLs and the two ExpR proteins executed? We showed that the transcript levels of the PCWDE genes are affected. Northern analysis was used to this aim and demonstrated that at early growth phase the transcript levels examined were indeed constitutive in the expR1 expR2 double mutant. Here we present for the first time evidence showing that the ExpR proteins are truly responsible for the delay in PCWDE gene expression and demonstrate that QS is essential for population density-dependent regulation of these genes. These data do not, however, demonstrate weather the ExpR proteins directly control PCWDE genes or weather this regulation is executed at some other level. Recent data from Cui et al. (2005) indicated that the global negative regulator RsmA is controlled by ExpR in another Ecc strain. To explore this possibility in Ecc SCC3193 we employed an rsmA-gusA promoter fusion and demonstrated that ExpR-dependent repression of PCWDE genes was apparently due to transcriptional control of rsmA. In the absence of AHSL both ExpR1 and ExpR2 were able to activate rsmA transcription, leading to down-regulation of PCWDE genes (Figs 5A and 9) while the addition of the cognate 3-oxo-C8-HSL abolished this activation leading to up-regulation of PCWDE genes. To further characterize the mechanism of this regulation we examined the AHSL binding capacities of the different expR mutant strains. These results clearly show that ExpR1 and ExpR2 are able to bind either together or separately the cognate 3-oxo-C8-HSL. These results strongly support the results gained with the rsmA transcription activation assay. This model is in agreement with the data by Cui et al. (2005) showing that in the Ecc strain Ecc71 the ExpREcc71 protein binds to the promoter region of rsmA and activates its transcription in the absence of AHSLs and that the DNA binding of ExpREcc71 and hence activation of rsmA transcription was inhibited by the presence of the cognate AHSL (Chatterjee et al., 2005; Cui et al., 2005).
Although rsmA-mediated repression appears to be the main control of PCWDE production we show that this is not the only pathway by which the AHSLs exert their control. Interestingly, an expI rsmA double mutant was still able to respond to the cognate 3-oxo-C8-HSL and increase its Cel activity accordingly. This result suggests a further fine-tuning in the system. This could be accomplished by a dual function of ExpR proteins (Fig. 9) whereby the AHSL-bound forms of ExpR could act as direct or indirect regulators of PCWDE production. Alternatively ExpR could control expression of another yet unknown negative regulator of PCWDEs similarly to its control of rsmA.
In most known AHSL receptors the N-terminal ligand binding domain has been shown to be very specific distinguishing between cognate and non-cognate ligands (Luo et al., 2003; Chai and Winans, 2004). We propose that (Fig. 9) in Ecc the two ExpR proteins have synergistic, but individual roles, with ExpR1 acting as a cognate ligand specific regulator, while ExpR2, interestingly, responds to both the cognate and the non-cognate signals. We demonstrate that ExpR2 has a broad AHSL sensing capacity, in contrast to the other LuxR-type proteins of Erwinia strains known today (Welch et al., 2000; Chatterjee et al., 2005). This conclusion is based on four key observations: First, recently we showed that the PCWDE activity of an expI mutant of Ecc strain SCC3193 could only be restored with physiological levels of the cognate AHSL 3-oxo-C8-HSL (Brader et al., 2005). Here we show that the expI expR1 double mutant of SCC3193 had lost its specificity for the cognate AHSL and reacted also to non-cognate AHSLs, such as 3-oxo-C6-HSL indicating the presence of another AHSL receptor. We subsequently identified a second ExpR protein (ExpR2) that was shown to be responsible for the sensing of the non-cognate AHSLs. Second, by monitoring Cel activity we show that ExpR2 senses AHSLs with acyl chains ranging between C6 and C10. The inability to sense the AHSLs with even longer acyl chains (C12 and C14), might depend on the lack of a reasonable transport system for these long chain AHSLs into the cell (Fuqua et al., 2001). Third, in the expI expR1 double mutant the transcription of rsmA can be prevented with either 3-oxo-C6-HSL or 3-oxo-C8-HSL, and leading to subsequent production of PCWDEs. In contrast expression of rsmA is only abolished by the addition of 3-oxo-C8-HSL in, the expI expR2 mutant (Fig. 9). Fourth, we show that the expI mutant bind both 3-oxo-C6-HSL and 3-oxo-C8-HSL, while the the expI expR2 mutant is only able to bind 3-oxo-C8-HSL. This is a clear evidence that ExpR2 is responsible of the binding of 3-oxo-C6-HSL.
The special feature of ExpR2 to respond to several different AHSLs raises questions of its role. Why does this Ecc strain have both a specific and an unspecific QS regulator? The existence of expR2 could be a consequence of horizontal gene transfer from other bacteria (Pappas et al., 2004). This is supported by the fact that expR2 is located separately from the expI-expR1 cassette. A biological advantage of having a receptor for various non-cognate AHSLs could be the ability to identify neighboring bacteria by sensing the accumulation of different kinds of AHSLs. Thus eavesdropping on possible competitors (Lazdunski et al., 2004; Waters and Bassler, 2005) or establishing cooperation with other bacteria to overwhelm the plant host could be beneficial for the success of Ecc as a pathogen. The special feature of ExpR2 to recognize AHSLs produced by its own species in addition to AHSLs produced by neighboring bacteria could further enhance survival in a crowded niche. It is also possible that ExpR2 in addition possess unique target sites not recognized by ExpR1 and involved in interactions with other bacterial species.
We present in this article a novel mechanism for QS (Fig. 9): the simultaneous need for two individual, but cooperatively acting QS regulators in controlling expression of the same target gene. In this study we have shown that ExpR1 and ExpR2 can function as activators either alone or synergistically when both proteins are present. Our results suggest that the amount of bound AHSLs correlate with the amount of ExpR proteins present and that the AHSL binding of the ExpR proteins correspond to the transcriptional activity of rsmA. We propose that ExpR1 and ExpR2 act in synergy integrating cognate and non-cognate AHSL signals to control expression of a central regulatory gene rsmA. However, we show that they can also act independently and therefore are not necessarily physically interacting. LuxR-type proteins use different modes of action and operate in various ways depending on bacterial species and tasks to be accomplished. A hierarchical organization model where one LuxR-type protein regulates the transcription of another LuxR-type protein is described in, e.g. Yersinia pseudotuberculosis (Atkinson et al., 1999) and Pseudomonas aeruginosa (Pesci et al., 1997). The possibility that ExpR1 would control expression of expR2 or vice versa, was tested using expR-gusA promoter fusions, but no transcriptional regulation between ExpR1 and ExpR2 was observed (data not shown). Thus we prefer a model where the two ExpR proteins directly control expression of the rsmA gene and possible other target genes. LuxR-type proteins, such as CarR, LuxR and TraR dimerize/multimerize after binding AHSL leading to subsequent stabilization of the protein and activation of target gene expression (Qin et al., 2000; Welch et al., 2000; Urbanowski et al., 2004). In contrast, the ExpR proteins appear to bind target DNA and act as activators in the absence of AHSL and then be inactivated/released by AHSLs. Whether these states involve dimerization remains to be demonstrated. Alternatively it could be possible that after AHSL binding, the two ExpR proteins stay at the target gene promoter just shifting in their conformation to block the function of the RNA polymerase. E.g. RhlR, the AHSL receptor of Pseudomonas aeruginosa has such a dual role always binding to DNA, but acting as an activator or a repressor depending on the presence of AHSL (Medina et al., 2003). An open question is also whether the ExpR1 and ExpR2 bind the same promoter site or are there possibly several ExpR binding sites in the rsmA promoter? In the rsmA promoter of Ecc71 one ExpR binding site was recently identified (Chatterjee et al., 2005). We could identify a similar DNA element (ExpR box) in the rsmA promoter of Ecc strain SCC3193 and in the sequenced Eca strain SCRI1043. Furthermore, we cannot rule out the possibility that the ExpR proteins of Ecc SCC3193 have a dual role, working both in an AHSL bound state and in a ligand-free state. This hypothesis is partly supported by the results with the expI rsmA mutant strain, showing enhanced Cel activity only with the addition of the cognate 3-oxo-C8-HSL. Such a model has also been proposed for EsaR of Pantoea stewartii that binds to its target promoter in the absence of AHSL and is released by the addition of AHSL (Minogue et al., 2005).
The ability of Ecc strains lacking the whole QS system to grow and macerate plant tissues as well as the wild-type under laboratory conditions indicate that the biological relevance of the QS system is mainly in the natural habitat, where the densities and the composition of bacterial populations fluctuate in response to environmental cues. An ecological study would be essential to elucidate the significance of QS in controlling the success of Ecc in the environment (Manefield and Turner, 2002; Redfield, 2002; Toth and Birch, 2005; Keller and Surette, 2006).
Bacterial strains and media
Bacterial strains and plasmids used in this study are listed in Table 2. Escherichia coli strains were cultured in L medium (Miller, 1972) at 37°C and Erwinia carotovora ssp. carotovora strains at 28°C. Ampicillin (Amp) was added to media at 150 μg ml−1, chloramphenicol (Cm) at 50 μg ml−1 and kanamycin (Km) at 50 μg ml−1 when required and if not otherwise mentioned. AHSLs were used at a concentration of 1 μM if not otherwise mentioned.
Table 2. Bacterial strains, plasmids used in this study.
expR1SCC3193 cloned into pQE30 EcoRI and BamHI sites
expR2SCC3193 cloned into pQE30 EcoRI and BamHI sites
expRSCC1 cloned into pQE30 EcoRI and BamHI sites
rsmASCC3193 promoter (189 nt) and partial CDS (67 nt) cloned into pGUS102 SalI and HindIII sites
889 bp DNA fragment containing upstream region of expR2
cat gene and km gene cloned into pSMS100
expR2SCC3193::Km, expR2SCC3193::Cm cloned into pBluescript ApaI and SpeI sites
expR2SCC3193::Km, expR2SCC3193::Cm cloned into pGP704 ApaI and SpeI sites
Library screening and construction of mutant strains
Recombinant DNA techniques were used according to standard procedures (Sambrook and Russell, 2001). PCR amplifications were performed with proofreading Pfu polymerase (Stratagene) and Dynazyme II (Finnzymes). Based on the sequence of Eca strain SCRI1043 an expR2 specific probe was PCR amplified from wild-type SCC3193 genomic DNA using primers ProbR2F (5′-TCTGTATTTTGCTCTGATAA-3′) and ProbR2R (5′-CAGATCGCCATACTGTTTTA-3′). The expR2 probe was used to screen Lambda DASH library (Stratagene) containing 17–22 kb BamHI fragments of SCC3193. An expR2 positive clone was amplified for lambda DNA isolation and the purified DNA containing an approximately 20 kb insert was cut with BamHI and EcoRI and verified with Southern blot analysis to be expR2 positive. Primers cdhF (5′-TGATTGCTATAGGTCCTCAG-3′) and cheR (5′-GGTGAGGTTTGTTCTCTCATC-3′) were used to PCR amplify a 3 kb DNA fragment, for subsequent sequencing and cloning procedures. To obtain a deletion mutant of expR2, the upstream region of the expR2 gene was amplified by PCR from SCC3193 with primers expRBapaR (5′-CCGGGCCCCCTGCGGCTATTGTGATAACG-3′) and expRBhindF (5′-CCAAGCTTTCTGGCTGCGTTATCGATTATG-3′). The resulting 1461 bp PCR product was digested with ApaI and HindIII resulting in a 889 bp DNA fragment (the reduction in DNA length was due to an unobserved HindIII restriction site) and inserted into these sites in pBluescript (pSMS100). The cat gene and the kanamycin resistance (km) gene was PCR amplified with following primers: CatSmaF (5′-CCCCCGGGTTCGACCGAATAAATACCTGT-3′) and CatHindR (5′-CCAAGCTTCTATCGTCAATTATTACCTCCA-3′); KmSmaF (5′-CCCCCGGGCAGCTACTGGGCTATCTGGA-3′) and KmHindR (5′-CCAAGCTTGCGTCAATACGGGATAATAGTG-5′). Each fragment containing an antibiotic resistance marker gene was digested with SmaI and HindIII and inserted to pSMS100 to confer one chloramphenicol and one kanamycin resistant plasmid pSMS103(Km) and pSMS103(Cm). The down-stream region of expR2 gene was PCR amplified with primers expRBspeF (5′-GACTAGTGTGTAGCGTAGTCAGGCAAC-3′) and expRBsmaR (5′-CCCCCGGGCTACTGTTACCCCATGATATCAC-3′). The product was digested with SpeI and SmaI and inserted into plasmids pSMS103(Km) and pSMS103(Cm) digested with same enzymes, resulting in the constructs pSMS104(Km) and pSMS104(Cm). The DNA fragment (expR2::Km; expR2::Cm) of pSMS104 (Km/Cm) was digested with ApaI and SpeI and inserted into suicide vector pGP704 digested with same enzymes. The resulting plasmids pSMS105/Km and pSMS105/Cm were transformed into E. coli S17-1 λ pir and further transformed by conjugation into Ecc strain SCC3193, SCC3065 (expI), SCC5003 (expR1) and SCC6005 (expI expR1) (de Lorenzo and Timmis, 1994). Transconjugants were plated on either M9 minimal medium supplemented with 0.2% sucrose and chloramphenicol 10 μg ml−1 or kanamycin 10 μg ml−1 or on L medium plates supplemented with chloramphenicol 50 μg ml−1 and kanamycin 50 μg ml−1. The resulting strains SCC905 (Cm) (expR2::Cm), SCC905 (Km) (expR2::Km), SCC906 (expIexpR1::Cm expR2::Km), SCC907 (expR1::Cm expR2::Km) and SCC908 (expI::Km expR2::Cm) were verified to be CmR AmpS, KmR AmpS or CmR KmR AmpS and confirmed genotypically.
Construction of plasmids
The expR1 and expR2 genes were amplified by PCR from wild-type SCC3193 genomic DNA using the primers: ExpR(3193)eco2 (5′-CGGAATTCGAGATGTCGCAGTTATTCTACA-3′) and ExpR(3193)bam (5′-CGGGATCCGCCTATGACTGAACCGGTCG-3′); ExpR2 SMS17 Rev (5′-CGGGATCCCTATAGTGGTTCTGGCTTGATG-3′) and ExpRB pOKF (5′-CGGAATTCATGTCTGTATTTTGCTCTGATAATG-3′). The 737 bp expR1 PCR product and the 738 bp expR2 PCR product was digested with BamHI and EcoRI and ligated into pQE30, digested with corresponding enzymes, resulting in pSMS20 and pSMS21, respectively. The expRSCC1 was amplified by PCR from wild-type SCC1 genomic DNA using primers: expR(1)eco (5′-CGGAATTCGAGATGTCGCCATTATTCACTG-3′) and expR(1)bam (5′-CGGGATCCTACCTGCCGCTATTGCACAGG-3′). The 729 bp expRSCC1 PCR product was digested with BamHI and EcoRI and ligated into pQE30, digested with corresponding enzymes, resulting in pSMS22. For promoter fusion studies a plasmid (pSMS18) was constructed containing the rsmA promoter region and partial coding sequence amplified by PCR using primers: RsmAF prom (5′-GCGTCGACCTGTTGTTGTGATAACAAAAG-5′) and RsmAR prom (5′-CCAAGCTTACCGTTACCTCATCGCCGA-3′). The 189 bp PCR product was digested with SalI and HindIII and ligated into pGUS102 digested with corresponding enzymes.
RNA isolation and Northern blot analysis
Erwinia cells from overnight cultures were diluted 1/100 in L medium and grown at 28°C. Samples for RNA isolation were taken at indicated time points and the growth was monitored by measuring the OD600. Total RNA was isolated as described by Sambrook et al. (1989). Northern analysis was performed with 10 μg of total RNA separated in 1.5% formaldehyde gel. Filters were probed with specific digoxigenin labeled DNA fragments for celV1, pehA, pelB and 16S rRNA (Hyytiäinen et al., 2003). The blotting, hybridization and digoxigenin detection was performed according to the instructions of the manufacturer (Roche).
Cellulase (Cel) activities were analysed from 10 μl of supernatant of overnight grown liquid cultures on CMC indicator plates using Congo Red to determine the cellulose digestion (Pirhonen et al., 1993). Quantitative Cel assay was performed as described previously (Pirhonen et al., 1991). For β-Glucuronidase (GUS) activity assay Erwinia cells from overnight cultures were diluted 1/100 in L medium and grown at 28°C. Samples for GUS activity assay were taken at indicated time points and the growth was monitored by measuring the OD600. GUS activity was measured by using p-nitrophenyl β-D-glucuronide as substrate (Novel et al., 1974; Marits et al., 2002). ρ-nitrophenol (ρ-NP), was detected at an absorbance of 405 nm and the specific activity of GUS was expressed as nmol ρ-NP liberated min−1.
Assay for AHSL Binding
The samples for determining the AHSL binding capacity were prepared as follows: 15 ml of bacteria was grown in L media complemented with no AHSL, 1 μM 3-oxo-C6-HSL or 1 μM 3-oxo-C8-HSL. The overnight grown bacterial cells were collected and washed twice with 0.9% NaCl to remove AHSL from the supernatant. Washed cells were then resuspended in 1.5 ml of lysis buffer (50 mM NaH2PO4 pH 8.0, 300 mM NaCl, 10 mM imidazole) and lysozyme was added. After 30 min incubation on ice, cells were sonicated and the cell debris was removed. The AHSLs were extracted twice with equal amounts of ethylacetate and the extracts dried in a Speed-Vac, with subsequent resuspension into 30 μl acetonitrile:0.1% formic acid (1:1 v/v) for LC-MS analysis as described (Brader et al., 2005) or 50 μl l-medium for bioluminescence assays. Here, overnight grown E. coli carrying pSB402 (Guard-Pette, 1998) was diluted 1:100 and grown for 5 h. After this 50 μl of E. coli and the 50 μl AHSL extract was mixed and incubated for 2 h. The bioluminescence was measured with the 1420 multilabel counter VICTOR2.
Synthesis and analysis of AHSLs
AHSL standards have been purchased from Sigma-Aldrich (C7-, C8-, 3-oxo-C6-HSL) or synthesized (C4-, C6-, C12-HSL, 3-oxo-C8-HSL, 3-oxo-C10-HSL, 3-oxo-C14-HSL) as described (Zhang et al., 1993). AHSL standards and profiles of culture supernatants have been analyzed by LC-MS as described earlier (Brader et al., 2005).
Assay of maceration capacity
Erwinia strains were grown overnight, diluted into 0.9% NaCl and samples containing 105 bacterial cells ml−1 were used for inoculation of potato tubers (Solanum tuberosum cv. Van Gogh). The inoculation site was bored with a sterile toothpick. Infected potatoes were incubated at 28°C for 72 h under humid conditions with wet tissue paper in the incubation box. The amount of soft rot was measured by cutting the potato tubers in half and scraping and subsequently weighing the rotted tissue. Arabidopsis thaliana Col-0 was infiltrated with a syringe without needle with 105 bacterial cells ml−1 prepared as described above. The development of disease symptoms was documented after 48 h, kept in 22°C, 16 h light and high humidity.
Nucleotide sequence accession number
The DNA sequence data determined in this study has been submitted to the DDB/EMBL/GenBank databases under accession numbers DQ333187 and DQ333188.
We thank Leila Miettinen for excellent technical assistance. We thank Hannu Saarilahti (University of Helsinki, Finland) for kindly providing us the SCC3193 Lambda Dash II library. We thank Robert Andersson for kindly providing us the expI rsmA mutant. This study was supported by the Helsinki Graduate School in Biotechnology, Molecular Biology and Academy of Finland (projects 388033, 44252 and 44883; Finnish Centre of Excellence Programme 2000-05), Biocentrum Helsinki and a grant from Leonardo da Vinci II Programme (to G. K.).