Cyclic nucleotides represent second messenger molecules in all kingdoms of life. In bacteria, mass sequencing of genomes detected the highly abundant protein domains GGDEF and EAL. We show here that the GGDEF and EAL domains are involved in the turnover of cyclic-di-GMP (c-di-GMP) in vivo whereby the GGDEF domain stimulates c-di-GMP production and the EAL domain c-di-GMP degradation. Thus, most probably, GGDEF domains function as c-di-GMP cyclase and EAL domains as phosphdiesterase. We further show that, in the pathogenic organism Salmonella enterica serovar Typhimurium, the nosocomial pathogen Pseudomonas aeruginosa and the commensal species Escherichia coli, GGDEF and EAL domains mediate similar phenotypic changes related to the transition between sessility and motility. Thus, the data suggest that c-di-GMP is a novel global second messenger in bacteria the metabolism of which is controlled by GGDEF and EAL domain proteins.
In the prokaryotic kingdom, signalling by cyclic nucleotides seems pretty simple. Only cAMP has been identified as a bona fide general second messenger that plays a central role in transcriptional regulation (Kolb et al., 1993). In Escherichia coli and Salmonella enterica serovar Typhimurium (S. Typhimurium), the cAMP circuit consists of one adenylate cyclase, a cAMP receptor protein and a phosphodiesterase respectively (Imamura et al., 1996). Uniquely, cellulose biosynthesis in Gluconacetobacter xylinus has been demonstrated to require the allosteric activator cyclic-di(3′→5′)-guanylic acid (c-di-GMP) (Ross et al., 1987).
The mass sequencing of bacterial genomes detected an abundance of genes encoding proteins containing domains of unknown function (Galperin et al., 2001), whereby the GGDEF [also called DUF (Domain of unknown function) 1] and the EAL (DUF2) domains are especially abundant. For example, in 93 sequenced prokaryotic genomes 691 and 503 copies of GGDEF and EAL domain have been detected, whereby S. Typhimurium harbours 12 proteins with GGDEF and 14 proteins with EAL domain. However, functional characterization laggs behind, up to now for only a handful of those genes phenotypes have been demonstrated related to bacterial morphogenesis and/or motility (Hecht and Newton, 1995; Ausmees et al., 1999; Jones et al., 1999; Ko and Park, 2000; Römling et al., 2000; Boles and McCarter, 2002; D’Argenio et al., 2002). For example, a GGDEF containing protein stimulates autoaggregation in Pseudomonas aeruginosa, while overexpression of an EAL domain overcomes hns-mediated non-motility (Ko and Park, 2000; D’Argenio et al., 2002).
Recently, a link between c-di-GMP metabolism and the function of GGDEF and EAL domains could be established, because the three diguanylate cyclases and phosphodiesterases involved in turnover of c-di-GMP in G. xylinus contain GGDEF and EAL domains (Tal et al., 1998). Nevertheless, although molecular modelling has predicted the GGDEF domain to encode a nucleotide cyclase (Pei and Grishin, 2001), functional analysis is not trivial, because each diguanylate cyclase and phosphodiesterase contains a GGDEF as well as an EAL domain (Tal et al., 1998).
We have recently established that S. Typhimurium, E. coli and other Enterobacteriaceae are capable of producing cellulose whereby the structural genes are encoded by the bacterial cellulose synthesis (bcsABZC) operon (Zogaj et al., 2001; 2003). Cellulose biosynthesis in S. Typhimurium is conveniently detected by the characteristic pink, dry and rough (pdar) morphology of bacterial colonies on Congo Red (CR) agar plates (Römling et al., 2000). Activation of cellulose biosynthesis requires solely AdrA, but not the response regulator CsgD that transcriptionally activates adrA (Römling et al., 2000; Zogaj et al., 2001). AdrA is a 371 amino acid (aa) long protein with an N-terminal domain, which contains four transmembrane spanning regions, and a C-terminal GGDEF domain of 180 amino acids (Römling et al., 2000). Because transcription of the bcsABZC operon is not influenced by AdrA, activation of cellulose biosynthesis by AdrA takes place on a post-transcriptional level (Zogaj et al., 2001).
We have used the system of regulation of cellulose biosynthesis in S. Typhimurium to dissect the function of GGDEF and EAL domains in proteins. We could show that solely GGDEF domains elevated c-di-GMP concentration in vivo, while EAL domains diminished c-di-GMP concentration in the cell, most probably by functioning as a cyclase and phosphodiesterase respectively. Expression of GGDEF and EAL domain proteins and consequently c-di-GMP levels regulated bacterial behaviour in two major pathogens, S. Typhimurium and P. aeruginosa, and a commensal E. coli strain. These findings strongly suggest that c-di-GMP acts as a universal second messenger that directs the transition from sessility to motility.
The GGDEF domain of AdrA is required for the production of c-di-GMP in vivo
Activation of cellulose biosynthesis in S. Typhimurium by AdrA (Römling et al., 2000; Zogaj et al., 2001) was used as the model system to study the function of the GGDEF domain. AdrA expression from the tightly regulated pBAD30 plasmid in the adrA knockout strain S. Typhimurium MAE103 lead to the activation of cellulose biosynthesis as monitored by an intensive pdar morphotype on Congo Red agar plates and fluorescent colonies on Calcofluor plates, while cells remained white and non-fluorescent with the vector control (Zogaj et al., 2001; Fig. 1A; data not shown). To further confirm that the morphotype change caused by AdrA overexpression is related to elevated cellulose biosynthesis, the AdrA expression plasmid was introduced into MAE190, a knockout strain of the catalytic subunit of the cellulose synthase bcsA. A morphotype switch to pdar was not observed in MAE190 when AdrA was overexpressed confirming the specificity of the morphotype and colour change (data not shown). The GGDEF domain is, in combination with the EAL domain, involved in the turnover of c-di-GMP in G. xylinus (Tal et al., 1998). To elucidate the role of AdrA and the GGDEF domain in c-di-GMP metabolism, we established a protocol for nucleotide extraction from plate-grown bacteria (Amikam et al., 1995). To detect c-di-GMP production, extracts were analysed by reversed-phase high-pressure liquid chromatography (HPLC) coupled with MALDI-TOF analysis using conditions optimized for the separation and detection of c-di-GMP. A strong signal at the retention time of synthetic c-di-GMP was detected in extracts from cells where AdrA had been overexpressed, whereas no signal was present in extracts from the vector control (Fig. 2A). In order to further verify the identity of the compound, fractions were subjected to MALDI-TOF analysis. A prominent [M-H]– ion at mass/charge (m/z) ratio of 689 was detected in fractions from extracts of cells, which overexpressed AdrA (Fig. 2B). The MS/MS spectrum of the [M + H]+ ion m/z 691 yielded identical fragmentation products for the isolated compound and synthetic c-di-GMP (Fig. 2C and D). Consequently, those experiments suggested that AdrA is involved, directly or indirectly, in c-di-GMP production in S. Typhimurium.
Next, we investigated whether expression levels of the AdrA protein correlated with c-di-GMP levels and the amount of cellulose biosynthesis. Therefore, we modulated AdrA concentrations by variation of the concentration of l-arabinose that activates transcription from the araC promoter on the pBAD30 plasmid. The expression level of AdrA was checked by Western blotting (Fig. 3A). Cell growth was not affected even at the highest expression of AdrA (data not shown). Increased expression of AdrA lead to a more pronounced pdar morphotype, which indicated higher cellulose expression, but a slight morphotype change was already observed without the inducer l-arabinose (Fig. 3B). C-di-GMP levels were investigated by HPLC and subsequent MALDI-TOF analysis (Fig. 3C). As it can be seen in Fig. 3, c-di-GMP levels correlated with increasing amounts of AdrA protein. Therefore, a connection between the expression of AdrA, c-di-GMP levels and cellulose biosynthesis could be established.
The c-di-GMP concentration at highest AdrA expression reached 0.64 nmol mg−1 cells (Fig. 3). C-di-GMP is made from two molecules of GTP. Theoretically, the artificially high c-di-GMP concentration caused by AdrA overexpression could deplete the GTP pool, which is required for RNA synthesis and translation. The GTP concentration was determined by HPLC after nucleotide extraction. When AdrA was not induced in strain MAE103, the GTP concentration was 100-fold higher than the c-di-GMP concentration. At the highest AdrA expression however, the c-di-GMP concentration exceeded the GTP concentration 10-fold. Nevertheless, the GTP concentration went only 30% down as compared to the situation where AdrA was not induced (data not shown). The 30% reduction in GTP concentration did not seem to affect the physiological role of GTP, because cell growth was not reduced by AdrA overexpression.
Theoretically, either the N-terminal transmembrane domain, the C-terminal GGDEF domain or both domains can be involved in c-di-GMP production. To show that the GGDEF domain of AdrA was required for c-di-GMP production, an AdrA protein where the two glycines in the GGDEF motif at position 288 and 289 were exchanged to alanine (AdrA G288A G289A) was constructed. The mutant protein was expressed at lower amounts than wild-type AdrA at the same arabinose concentration in the adrA knockout strain S. Typhimurium MAE103. However, when the phenotypes of cells with comparable concentrations of wild-type and mutant protein were investigated, the mutant protein did not activate cellulose biosynthesis and c-di-GMP production in contrast to the wild type (Fig. 4 and data not shown).
The non-functionality of AdrA G288A G289A showed that the GGDEF domain is involved in c-di-GMP production, but does not completely exclude that both domains of AdrA are required for c-di-GMP production. To gain indirect evidence for the sole role of the GGDEF domain in c-di-GMP synthesis, the E. coli encoded GGDEF containing protein YhcK (Ausmees et al., 2001) was expressed in the adrA knockout strain S. Typhimurium MAE103. YhcK, which contains an N-terminal domain non-homologous to the one of AdrA, mediated cellulose biosynthesis and c-di-GMP synthesis (Table 1 and data not shown).
Table 1. Colony morphology and c-di-GMP concentration in S. Typhimurium, E. coli TOB1 and Pseudomonas aeruginosa PAO containing plasmid-encoded GGDEF and EAL domain genes.
c-di-GMP concentration (pmol/mg cells)
nd, c-di-GMP was not detected as a peak in HPLC analysis.
S. Typhimurium MAE103 (pBAD30)
S. Typhimurium MAE103 (pWJB30); AdrA+
S. Typhimurium MAE103 (pWJB30, pRGS3); AdrA+, YhjH+
S. Typhimurium MAE103 (pYhcK); YhcK+
S. Typhimurium MAE103 (pYhcK, pRGS2); YhcK+, STM1827+
E. coli TOB1 (pBAD30)
E. coli TOB1 (pWJB30); AdrA+
P. aeruginosa PAO (pLAFR3)
P. aeruginosa PAO (pWJB9); AdrA+
The EAL domain is required for breakdown of cyclic-di-GMP in vivo
Because of many conserved acidic residues that could participate in metal binding and might form the phosphodiesterase active site, the EAL domain is predicted to encode a phosphodiesterase (pfam00563). The S. Typhimurium genome harbours 14 genes encoding for EAL domain proteins, whereby two are pretty simple in that they solely consist of an EAL domain. We cloned one of those genes [STM3611 (yhjH)] and assessed its function. When yhjH was expressed in the cellulose-proficient strain S. Typhimurium MAE97, it lead to the disappearance of the pdar morphotype on agar plates (Fig. 1B). Therefore YhjH, directly or indirectly, counteracted cellulose biosynthesis. To establish a connection between the effect of YhjH on cellulose biosynthesis and c-di-GMP levels in vivo, AdrA and YhjH were co-expressed in the adrA knockout strain S. Typhimurium MAE103. AdrA expression was modulated by l-arabinose as previously (Fig. 3), while the level of YhjH was kept constantly high. All AdrA/YhjH expression ratios led to a weaker expression of the pdar morphotype on plates as well as to significantly lower c-di-GMP concentrations (Fig. 5), when compared with the same strain that only expressed AdrA (Figs 3 and 5). However, only at the lowest AdrA/YhjH ratio, cellulose biosynthesis was completely abolished. This finding indicates that YhjH cannot induce degradation of c-di-GMP as efficiently as it is synthesized via AdrA. Alternatively, not all c-di-GMP produced via AdrA is accessible for degradation by YhjH.
Multiple sequence alignment classified the EAL domains encoded by the S. Typhimurium genome into three subclasses (data not shown). However, to convert the specificity of, for example, nucleotide cyclases, only few amino acids need to be exchanged (Hurley, 1998). To show that an EAL domain from another subclass elucidates the same function as YhjH STM1827 was cloned and expressed. STM1827 was, as YhjH, able to counteract cellulose biosynthesis and to mediate degradation of c-di-GMP (Table 1 and data not shown).
To investigate which amino acids of YhjH are involved in the activity to degrade c-di-GMP, we constructed a mutant of YhjH by exchanging the highly conserved amino acid glutamic acid at position 136 to alanine. When mutant YhjH E136A was introduced into cellulose producing S. Typhimurium MAE97, cellulose expression did not decrease, although YhjH E136A was expressed at the same level as wild-type YhjH (data not shown). YhjH E136A did also not alter the c-di-GMP levels when coexpressed together with AdrA in S. Typhimurium MAE 103 (Fig. 6).
AdrA mediates c-di-GMP production in various bacterial species
C-di-GMP production by AdrA or YhcK could be dependent on a specific pathway present in S. Typhimurium. To show that c-di-GMP production by AdrA is an intrinsic feature of AdrA independent of the genetic background, AdrA was expressed in the commensal strain E. coli TOB1. AdrA mediated high level of c-di-GMP concentration in TOB1 (Table 1). But E. coli is very closely related to S. Typhimurium and contains an AdrA homologue (Römling et al., 2000). Therefore, we also expressed AdrA in P. aeruginosa PAO, which does not contain an AdrA homologue, in contrast to its close relative P. fluorescens Pf0-1 (Nikolskaya et al., 2003). The latter fact made us positive that AdrA could anyway function in P. aeruginosa PAO. And indeed, we could observe enhanced c-di-GMP production coupled to a colony morphology change when AdrA was expressed in P. aeruginosa PAO (Table 1 and data not shown).
Effect of GGDEF and EAL domain on multicellular behaviour in S. Typhimurium UMR1
The results have shown that the GGDEF domain is required for c-di-GMP production, while the EAL domain is required for c-di-GMP degradation. In S. Typhimurium and E. coli, the change in c-di-GMP level is correlated with cellulose biosynthesis (this work and data not shown), which is consistent with the situation in G. xylinus, where c-di-GMP was identified as an allosteric activator of cellulose biosynthesis (Ross et al., 1991).
But we suspected that the c-di-GMP level could influence other phenotypes of S. Typhimurium besides cellulose biosynthesis. To that end we introduced AdrA and YhjH into S. Typhimurium UMR1, a typical wild-type S. Typhimurium strain, which shows a highly regulated multicellular behaviour called the rdar (red, dry and rough) morphotype, which is characterized by the expression of cellulose and curli fimbriae (Römling et al., 1998a; 2003). As a consequence of the strict regulation, multicellular behaviour of the rdar morphotype can be up- as well as downregulated in S. Typhimurium UMR1.
AdrA and YhjH had opposite effects on several modes of multicellular behaviour of S. Typhimurium UMR1 (Fig. 6 and data not shown). The expression of the rdar morphotype on agar plates (Römling et al., 1998a) was enhanced by AdrA and abolished by YhjH (data not shown). Using various model systems and different media, we found that other features of the rdar morphotype such as adherence to abiotic surfaces, pellicle formation and clumping were enhanced by AdrA and diminished by YhjH (Fig. 7A–D and data not shown). The magnitude of the observed effect depended on the experimental conditions, therefore, we hypothesized that other GGDEF and EAL domain proteins intrinsically expressed under specific environmental conditions influence the outcome of the experiments.
GGDEF and EAL domain proteins had an opposite effect on motility function. Swarming of S. Typhimurium UMR1, a form of multicellular behaviour where the bacteria move over an agar surface with the help of flagella, was significantly stimulated by YhjH and inhibited by AdrA (Fig. 7E). Motility, swimming in low-concentration agar plates with the help of flagella, was slightly enhanced by YhjH and dramatically decreased, although not abolished, by AdrA (Fig. 7F or data not shown). Inhibition of swimming motility occurred gradually as AdrA expression increased (Fig.S1). In conclusion, those experiments showed that AdrA and YhjH inversely regulate sessility versus motility. However, the effects of AdrA and YhjH were not absolute, suggesting that c-di-GMP regulates transition of sessility versus motility downstream of major regulatory pathways. Alternatively, AdrA and YhjH do not perform complete c-di-GMP turnover.
Effect of GGDEF and EAL domain on multicellular behaviour in P. aeruginosa PAO and E. coli TOB1
To generalize the effect of the GGDEF and EAL domain on bacterial behaviour, we investigated the effect of AdrA and YhjH in P. aeruginosa PAO, a major nosocomial pathogen and in the commensal isolate E. coli TOB1. When AdrA was expressed in strain PAO, we could observe a change in colony morphology resembling the rdar morphotype of S. Typhimurium (data not shown). All forms of motility such as swimming, swarming and twitching (movement on surfaces with the help of type IV pili) were repressed by AdrA and enhanced by YhjH (Fig. 8A). On the other hand, biofilm formation, pellicle formation and clumping were enhanced by the expression of AdrA and repressed by YhjH (Fig. 8B). Also in E. coli TOB1, AdrA and YhjH expression controls biofilm formation and non-motility versus motility (Fig. 8C and D and data not shown).
We have dissected the function of the GGDEF and EAL domains, namely c-di-GMP turnover, whereby the GGDEF domain represents the dinucleotide cyclase, while EAL, most probably, represents the cyclic dinucleotide phosphodiesterase. During the review of this manuscript, the group of Urs Jenal unambiguously demonstrated in vitro dinucleotide cyclase activity of the GGDEF domain of PleD from Caulobacter crescentus (Paul et al., 2004). Although a remote possibility, we cannot completely exclude that the EAL domain does not function as a phosphodiesterase but inhibits GGDEF domain activity. In any case, our findings strongly suggest that c-di-GMP acts as a general second messenger in bacteria, because GGDEF and EAL domain superfamilies are highly abundant in prokaryotes, especially in free-living Gram-negative bacteria (Galperin et al., 2001). In the pioneering work by Benziman (Ross, 1987), c-di-GMP has been identified as an allosteric activator of cellulose synthase in G. xylinus, an apathogenic soil bacterium, and was subsequently found in Agrobacterium tumefaciens, another soil-born species that produces cellulose (Amikam and Benziman, 1989).
But c-di-GMP has much broader impact on bacterial cell physiology. In this work, we show that c-di-GMP is produced by pathogenic as well as commensal species, whereby c-di-GMP levels influence several cellular functions. From our experiments we could conclude that c-di-GMP levels do not simply mediate the transition from a multicellular biofilm mode of growth to a planktonic (single) and motile cell status, but rather promote the transition between sessile and moving cells (Figs 7, 8 and 9A). This regulatory concept by c-di-GMP makes sense, because biofilm formation involves a motility component and motility can be a multicellular phenomenon. Highly regulated motility by flagella or type IV pili on the level of individual bacterial cells is actually required for the development of a sophisticated biofilm architecture (Klausen et al., 2003). On the other hand, swarming motility is a multicellular behaviour dependent on cell–cell interactions and ‘slime’ production (Harshey, 1994). S. Typhimurium has in total 19 proteins with GGDEF and/or EAL domains, whereby P. aeruginosa has 44. As far as we can extrapolate from our data and recent findings by others (Paul et al., 2004), all those proteins might be involved in c-di-GMP turnover. Why is such a complex and tight regulation of a single molecule in a bacterial cell required? Or in other words, why is the regulatory network for movement and sessility so sophisticated? Maybe, up to now, we have simply underestimated the requirement for a coordinated behaviour of bacterial cells by the assumption that directed behavioural processes, if at all, do only rudimentary exist in prokaryotes.
In eukaryotes, cyclic nucleotide concentrations are tightly temporally and spatially regulated by cyclase and phosphodiesterase superfamily members, whereby their individual members show distinct domain architecture and tissue-specific expression pattern (Houslay and Milligan, 1997). Supported by circumstantial observations such a scenario can also be envisaged for the regulation of c-di-GMP in bacteria. In S. Typhimurium the overall c-di-GMP concentration is very low (data not shown). In C. crescentus, PleD, a GGDEF domain protein has been reported to be localized at one cell pole (Paul et al., 2004). Messenger RNAs for GGDEF and EAL domain proteins have been found to be highly unstable (our unpublished data; Merkel et al., 2003). In addition, a substantial fraction of proteins encode for an EAL domain C-terminal to a GGDEF domain. This close proximity of counteracting domains may ensure a highly localized c-di-GMP response whereas when proteins that contain only the GGDEF domain such as AdrA produce c-di-GMP the second messenger diffuses into the cytoplasm (Fig. 9B).
Not much is known about the regulatory network directing expression or activation of GGDEF/EAL domain proteins. In S. Typhimurium, the response regulator CsgD mediates a specific form of multicellular behaviour characterized by the expression of cellulose and curli fimbriae, the rdar morphotype (Römling et al., 1998b; 2000). When CsgD is expressed, it transcriptionally activates expression of AdrA that subsequently triggers cellulose biosynthesis (Römling et al., 2000; Zogaj et al., 2001; Fig. 9B). In Bordetella pertussis, the EAL domain protein bvgR is transcriptionally regulated by the two component system BvgS/BvgA (Merkel et al., 2003), whereas in C. crescentus the two histidine sensor kinases DivJ and PleC phosphorylate an N-terminal CheY-like receiver domain of PleD, whereby the C-terminal GGDEF domain is activated (Paul et al., 2004). Therefore, GGDEF/EAL domain proteins are an integral part or immediately downstream of the signalling cascade of two-component systems. Often GGDEF/EAL domain proteins contain at least one N-terminal sensor domain such as GAF or PAS (Galperin et al., 2001), which regulate their activity by a stimulus such as oxygen (Chang et al., 2001). This stimulus might be different from the one that activates the corresponding two component system (concept of double stimulus, Fig. 9B), thus GGDEF/EAL domain proteins provide a second level of signal integration and decision making with which they contribute to the overall information processing in a bacterial cell.
But how does c-di-GMP work? In G. xylinum, it is suggested that c-di-GMP binds to BcsB whereby cellulose biosynthesis is activated (Kimura et al., 2001). There exists also a membrane protein, yet unidentified, which has high affinity to c-di-GMP and thereby modulates intracellular c-di-GMP concentrations (Weinhouse et al., 1997). However, up to now, a c-di-GMP binding site has not been identified. Cyclic nucleotides in pro- and eukaryotes excert their effects by binding to different classes of proteins such as to transcriptional regulators, protein kinases, ion channels and even regulate their own synthesis by binding to phosphodiesterases and nucleotide cyclases (Kolb et al., 1993; Soderling and Beavo, 2000; Kopperud et al., 2003). Where c-di-GMP binds and what effects will be exerted will be a subject of subsequent studies.
Bacterial strains and growth conditions
The features of bacterial strains are shown in Table 2. Cellulose is co-expressed with curli fimbriae in S. Typhimurium ATCC14028 (Römling et al., 2000). Thus, to unambiguously detect cellulose biosynthesis, strains with knockouts of csgBA, the structural genes for curli fimbriae were used. E. coli DH5α was grown in Luria–Bertani (LB) medium at 37°C supplemented with ampicillin (100 µg ml−1) or tetracycline (20 µg ml−1), when required. Unless otherwise stated, S. Typhimurium strains and E. coli TOB1 were grown on LB agar plates without NaCl supplemented with antibiotics and 0.1%l-arabinose, when required. P. aeruginosa PAO was grown on LB agar plates supplemented with tetracycline (70 µg ml−1) and 0.1%l-arabinose, when required.
Table 2. Bacterial strains and primers used in this study.
Newly introduced restriction site are underlined, nucleotides encoding for a His-tag are in italics and mutated nucleotides are shown in bold.
AdrA, YhjH and STM1827 were cloned in pBAD30 with a C-terminal His-tag resulting in plasmids pWJB30 (primers, PBAD30.YAIC.A, PBAD30.YAIC.E shown in Table 2), pRGS1 (primers, YhjH-start, YhjH-stop) and pRGS2 (primers, STM1827-Start, STM-1827-stop) respectively. YhjH cloned in pBAD30 was subcloned into pLAFR3 (pRGS3). AdrA cloned in pBAD30 was amplified together with the araC regulator region by primers ARA-C and BLA-30 (Table 1) and subsequently cloned into pLAFR3 (plasmid pWJB9). Mutant adrA and yhjH alleles were constructed by polymerase chain reaction (PCR) based mutagenesis using overlapping primers (Table 2) and subsequently cloned into pBAD30. Mutations introduced into adrA led to the replacement of the two glycines at positions 288 and 289 in the GGDEF motif by alanine (AdrA G288A, G289A) and plasmid pRGS6. The mutation introduced into yhjH led to the replacement of the glutamic acid at position 136 by alanine (YhjH E136A), plasmid pRGS4; and subcloning of YhjH E136A into pLAFR3 to plasmid pRGS5. pYhcK (received from Martin Lindberg and Nora Ausmees) has been described (Ausmees et al., 2001). Recombinant DNA techniques were performed according to Sambrook et al. (1989). PCR amplifications were performed using genecraft (Biostar). Plasmids were electroporated into S. Typhimurium via strain LB5010 as described before (Römling et al., 1998a). pLAFR3 derivatives were mobilized from E. coli to other bacteria by triparental mating using pRK2013 helper plasmid.
Bacteria were harvested from LB agar plates without NaCl after growth for 20 h at 37°C. Whole cell extracts were processed as it will be described elsewhere (manuscript in preparation). Approximately 10 µg protein was separated on a 12% SDS–polyacrylamide gels and afterwards blotted onto polyvinylidene difluoride (PVDF) membranes (Millipore) in transfer buffer (15 mM TRIS, 120 mM glycine, 10% methanol). Membranes were incubated in blocking buffer [TBS (20 mM Tris-HCl pH 7.6, 130 mM NaCl) containing 3% bovine serum albumine (BSA) and 0.1% Tween 20]. Anti-His-tag antibody (Quiagen) was used as the primary antibody at concentration of 0.4 µg ml−1, while the secondary antibody horseradish peroxidase-conjugated goat anti-mouse IgG (Jackson Immunolaboratories) was used at a dilution of 1:2000. Reactive bands were visualized using chemoluminescence (Roche), which was detected with a charge-coupled device camera (Geldoc 2000, FujiFilm).
To detect cellulose biosynthesis, S. Typhimurium was grown on Congo Red (CR) or Calcofluor agar plates for 20 h at 37°C or for 48 h at 28°C as described (Römling, 2001). The appearance of a pdar colony morphology is indicative for cellulose biosynthesis (Römling et al., 2000). Swimming motility was observed using 0.3% LB agar plates inoculated with a constant cell number from overnight cultures. Swarming motility was performed on 0.5% LB agar plates supplemented with 0.5% glucose. Twitching motility of P. aeruginosa was assayed by the subsurface agar stab method (Huang et al., 2003). Biofilm formation was observed in a steady state assay, using polystyrene wells without shaking or glass tubes (shaking 150 or 210 r.p.m.). Overnight cultures of S. Typhimurium were inoculated in LB without NaCl or M9 minimal medium and for P. aeruginosa LB medium was used. After staining with crystal violett, the biofilms was dissolved in dimethylsulfoxide and the absorbance at 540 nm was measured. Cell clumping and motiliy was observed by phase contrast microscopy.
Chemical synthesis of c-di-GMP
C-di-GMP was synthesized as described (Hsu and Dennis, 1982; de Vroom et al., 1987). C-di-GMP was purified by reversed-phase HPLC. The run was carried out in 0.1 M triethyl-ammonium-bicarbonate (TEAB/4% acetonitrile) at 1 ml min−1 in a linear gradient of acetonitrile. The identity of the substance was verified by thin layer chromatography, 1H and 31P nuclear magnetic resonance (1H- and 31P-NMR) spectroscopy and electrospray-mass spectrometry (ESI-MS) in the positive mode. The purity of the substance was >99%.
Isolation and detection of c-di-GMP from S. typhimurium
Cells (300 µg) were harvested from LB agar plates without NaCl after growth for 20 h at 37°C. Nucleotides were extracted by heat or acid (Amikam et al., 1995; Weinhouse et al., 1997). For heat extraction, cells were resuspended in water, heated at 100°C for 5 min and nucleotides were extracted twice with 0.5 ml of 65% ice-cold ethanol (Amikam et al., 1995). For acid extraction, cells were extracted with 0.6 M HClO4, denatured protein was removed by centrifugation and the supernatant was neutralized with 5 M K2CO3 (Weinhouse et al., 1997). In both methods, the extract was lyophilized, resuspended in 500 ml of water certified for HPLC analysis and filtered (0.2 µm pore size). Extracts equivalent to 150 mg cells (wet weight) were adjusted to 500 µl in 0.1 M tri-ethyl-ammonium acetate (TEAA) buffer pH 4.5 and subjected to HPLC separation.
HPLC was performed on a 250 × 4.6 mm reverse phase column (Hypersil® ODS 5 µ; Hypersil-Keystone) at room temperature, detection at 260 and 280 nm, on an Äktabasic apparatus (Amersham-Pharmacia). Running conditions were optimized using synthetic c-di-GMP and other relevant nucleotides as references. Runs were carried out in 0.15 M TEAA buffer pH 4.5 at 1 ml min−1, using a multistep gradient of acetonitrile. Relevant fractions of 1 ml were collected, lyophilized and resuspended in 10 µl H2O.
For quantification, a standard curve was established whereby synthetic c-di-GMP was added to relevant cell extracts. The area of the c-di-GMP peak was used to estimate the amount of c-di-GMP in a sample referred to wet cell weight.
Mass spectrometric analysis of c-di-GMP
One microlitre of fractions collected by HPLC or synthetic c-di-GMP at various concentrations was applied on a stainless-steel target by the fast evaporation method (matrix: α-cyano 4-hydroxycinnamic acid) and allowed to dry. MALDI-TOF analysis was performed on a Bruker Reflex II (Bruker-Franzen-Analytik) mass spectrometer using the negative ion mode. The detection limit of c-di-GMP by MALDI-TOF was 1 fMol. To confirm the identity of the substance, relevant peaks were fragmented by ESI-MS using the positive ion mode.
We cordially thank Mats Andersson for generous access to the HPLC apparatus and for technical advice. W. Pfleiderer and C. Ramamurthy (University of Konstanz, Germany) kindly provided precursors required for the synthesis of c-di-GMP. Werner Bokranz constructed pWJB9 and Nora Ausmees provided plasmid pYhcK. This work was supported by the Karolinska Institutet (Elitforskartjänst to U.R.), Mukoviszidose e.V and the Swedish Natural Science Research Council.
Fig. S1. Expression of AdrA correlates with decreased motility, while expression of YhjH stimulates motility.