SEARCH

SEARCH BY CITATION

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Aeromonas hydrophila uses the type II secretion system (T2SS) to transport protein toxins across the outer membrane. The inner membrane complex ExeAB is required for assembly of the ExeD secretion channel multimer, called the secretin, into the outer membrane. A putative peptidoglycan-binding domain (Pfam number PF01471) conserved in many peptidoglycan-related proteins is present in the periplasmic region of ExeA (P-ExeA). In this study, co-sedimentation analysis revealed that P-ExeA was able to bind to highly pure peptidoglycan. The protein assembled into large multimers in the presence of peptidoglycan fragments, as shown in native PAGE, gel filtration and cross-linking experiments. The requirement of peptidoglycan for multimerization was abrogated when the protein was incubated at 30°C and above. These results provide evidence that the putative peptidoglycan-binding domain of ExeA is involved in physical contact with peptidoglycan. The interactions facilitate the multimerization of ExeA, favouring a model in which the protein forms a multimeric structure on the peptidoglycan during the ExeAB-dependent assembly of the secretin multimer in the outer membrane.


Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

The type II secretion system (T2SS) or the main terminal branch (MTB) of the general secretion pathway (GSP) is used by a wide variety of Gram-negative bacteria for secretion of protein toxins and degradative enzymes. The proteins are translocated into the periplasmic space via the Sec or Tat systems, and through the T2SS cross the outer membrane to the extracellular environment (for reviews, see Filloux, 2004;Johnson et al., 2006). The apparatus is generally comprised of 12–16 proteins, including GspC–O, GspAB and/or GspS, and spans both the inner and outer membranes. It features an inner membrane GspE-F-L-M platform, an outer membrane 12- to 14-member multimer of GspD, called the secretin, and a GspG-H-I-J-K pseudopilus that may extend from the platform to the secretin (Bitter et al., 1998; Sauvonnet et al., 2000; Py et al., 2001; Korotkov and Hol, 2008). It is possible that the secreted proteins are pushed through the secretin, which has a central pore of 50–100 Å in diameter (Bitter et al., 1998; Nouwen et al., 1999; 2000; Chami et al., 2005), by a retractable pseudopilus, powered by the GspE ATPase that forms a hexametric ring in the inner membrane platform (Robien et al., 2003; Camberg and Sandkvist, 2005).

The T2SS is used by Aeromonas hydrophila for transport of a variety of extracellular proteins, including the pore-forming toxin aerolysin. The gsp genes encoding the T2SS in A. hydrophila are clustered in two operons, termed exeAB and exeC–N (Jiang and Howard, 1991; Jahagirdar and Howard, 1994). In addition, TapD, a type IV prepilin peptidase, is required for processing the T2SS pseudopilins (Pepe et al., 1996). Although ExeAB homologues GspAB are not found in all T2SSs, the complex is essential for normal secretion in A. hydrophila (Jahagirdar and Howard, 1994). The 60 kDa ExeA is comprised of a 31 kDa N-terminal/cytoplasmic domain, a short central hydrophobic inner membrane domain and a 28 kDa C-terminal/periplasmic domain. The 25 kDa ExeB resides mainly in the periplasm with only a small N-terminal domain exposed on the cytoplasmic side (Howard et al., 1996). The two proteins form a complex in the inner membrane (Schoenhofen et al., 1998). The cytoplasmic domain of ExeA contains an ATP binding site which is required for normal secretion and was shown to have ATPase activity in vitro (Howard et al., 1996; Schoenhofen et al., 2005). Further studies showed that ExeAB is required for assembly of the ExeD secretin into the outer membrane (Ast et al., 2002). In the absence of ExeAB, ExeD remains in the inner membrane as a monomer. In silico analysis revealed a putative peptidoglycan-binding domain in the periplasmic region of ExeA. Although the purified ExeA periplasmic domain shows no peptidoglycan hydrolase activity, mutagenesis studies showed that the putative peptidoglycan-binding domain is essential for ExeD secretin assembly and the cross-linking of ExeA to peptidoglycan in vivo (Howard et al., 2006). These findings suggest that interactions with peptidoglycan are involved in the function of ExeAB, which may be to generate gaps in peptidoglycan, perhaps by interference with normal peptidoglycan biogenesis, to allow ExeD to traverse the peptidoglycan barrier and assemble in the outer membrane (Dijkstra and Keck, 1996; Koraimann, 2003; Howard et al., 2006).

The putative peptidoglycan-binding domain (Pfam number PF01471) is found in numerous proteins, including a variety of enzymes related to bacterial cell wall degradation, for example, the spore cortex-lytic enzyme SleB in Bacillus cereus and the hydrolase metallo (Zn) DD-peptidase in Streptomyces albus G (Pfam database; Dideberg et al., 1982; Moriyama et al., 1996). The crystal structure of the latter protein has been solved (PDB entry 1LBU; Dideberg et al., 1982). It is comprised of an N-terminal putative peptidoglycan-binding domain (a.a. 1–90) and a C-terminal Zn-containing catalytic domain (a.a. 91–213) that cleaves the D-Ala-D-Ala amino bond of muropeptide. The putative peptidoglycan-binding domain features three alpha helices. The domain is proposed to have a general peptidoglycan binding function; however, no direct evidence of physical interactions has been reported, nor has the peptidoglycan binding site or the actual ligand been identified, in spite of the fact that peptidoglycan has been extensively analysed by HPLC and Mass Spectrometry (MS) in the past 20 years and many muropeptide species have been separated and identified (Glauner, 1988; Bacher et al., 2001; Antignac et al., 2003).

In this study, interactions between peptidoglycan and the periplasmic domain of ExeA were demonstrated by co-sedimentation, native PAGE, gel filtration and cross-linking in vitro. We also show that the interactions caused the periplasmic ExeA domain to form large multimers. These findings confirm the function of the putative peptidoglycan-binding domain and provide further understanding of the role of ExeA in secretin assembly.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

P-ExeA interacts with the peptidoglycan backbone

Since ExeA is an insoluble membrane protein, the C-terminal/periplasmic region of ExeA (P-ExeA) from a.a. 296 to a.a. 547 was cloned with a His-tag at the N-terminal end (N-His P-ExeA) or at the C-terminal end (C-His P-ExeA). We also constructed N-His F487S, N-His L493S and N-His G500D with substitution mutations of highly conserved amino acid residues in the putative peptidoglycan-binding domain. These three mutations greatly disrupted the function of full-length ExeA for normal secretion, secretin assembly and cross-linking to peptidoglycan in vivo although formation of the ExeAB complex was not impaired (Howard et al., 2006). Each of these proteins was purified by His-tag affinity chromatography and ion exchange chromatography. We purified non-denatured peptidoglycan by washing A. hydrophila cell envelopes several times with cold 2% Triton X-100 solutions to remove membrane lipids (Schoenhofen et al., 1998), and also prepared denatured peptidoglycan with boiling SDS and pronase hydrolysis treatment (Glauner, 1988). The effects of these treatments were confirmed by SDS-PAGE, which showed that many protein contaminants were present in the non-denatured sample; however, no detectible protein was present in the denatured sample (data not shown). The peptidoglycan samples were quantified by a colorimetric method that measures the muramic acid content (Hoijer et al., 1995).

The wild-type (N-His P-ExeA) and mutant (N-His G500D) proteins were incubated with peptidoglycan, followed by centrifugation to sediment the peptidoglycan, and the pellet samples were examined by immunoblot with ExeA antiserum. Wild-type P-ExeA, but not the G500D mutant, co-sedimented with the Triton X-100-treated peptidoglycan, suggesting specific co-sedimentation and the involvement of the putative peptidoglycan-binding domain in these interactions (Fig. 1A). A similar amount of the wild-type protein was also found in the sample of SDS/pronase-treated peptidoglycan and again much less of the mutant protein was bound (Fig. 1B). This indicates that P-ExeA interacted with the peptidoglycan backbone rather than proteins or lipids associated with the peptidoglycan. It was noticed, however, that more of the mutant P-ExeA co-sedimented with the pure peptidoglycan, possibly because of elimination of impurities that may block the binding sites on peptidoglycan.

image

Figure 1. Co-sedimentation of P-ExeAs with different peptidoglycan preparations. A. Co-sedimentation of P-ExeAs with non-denatured A. hydrophila peptidoglycan. N-His P-ExeA (wild type) and N-His G500D were used in the assay. The samples were applied to SDS-PAGE and immunoblotted with ExeA antiserum. The pellet samples were 30 times concentrated during resuspension. B. Co-sedimentation of P-ExeAs with different peptidoglycan preparations. Peptidoglycan samples were added at 100 µM muramic acid units for A. hydrophila and E. coli, and 600 µM for B. subtilis. The pellet samples were analysed by SDS-PAGE and ExeA immunoblot.

Download figure to PowerPoint

Peptidoglycan samples extracted from E. coli and the Gram-positive Bacillus subtilis were also used in the assays. Figure 1B shows that similar results for wild-type and mutant P-ExeAs were obtained with E. coli and B. subtilis peptidoglycan preparations. These results suggest that P-ExeA could interact with a peptidoglycan structure conserved in different bacteria, consistent with the widespread occurrence of the putative peptidoglycan-binding domain in a variety of species.

We also examined the C-His-tagged P-ExeA and the F487S- and L493S-binding domain mutants for interactions with peptidoglycan. Surprisingly, the addition of the tag to the C-terminal side of the protein almost completely abrogated co-sedimentation, so that it bound much less than even the mutant N-His G500D (Fig. 2). This may be because the putative peptidoglycan-binding domain (a.a. 441–507) is located near the C-terminal end of P-ExeA. The C-His tag might hinder peptidoglycan from accessing its binding site on P-ExeA, and/or affect the multimerization of the protein, which was investigated in later experiments (see below). The F487S and L493S variants almost completely abolished the peptidoglycan binding (data not shown). However, gel exclusion analysis revealed that the purified F487S and L493S proteins eluted as large aggregates (data not shown), suggesting that they did not fold properly and making it difficult to evaluate the roles of the two residues in peptidoglycan binding. In contrast, N-His P-ExeA, N-His G500D and C-His P-ExeA eluted as apparent dimers (data not shown). As a result, only the latter three proteins were further investigated.

image

Figure 2. Co-sedimentation analysis of P-ExeA variants. Different concentrations of N-His P-ExeA, N-His G500D and C-His P-ExeA were incubated with SDS/pronase-treated A. hydrophila peptidoglycan in the co-sedimentation assay. The supernatant (S) and pellet (P) samples were analysed by SDS-PAGE and immunoblotted with ExeA antiserum.

Download figure to PowerPoint

Characterization of P-ExeA–peptidoglycan binding

Co-sedimentation of P-ExeA and peptidoglycan was examined in buffers of different pHs and salt concentrations. Figure 3A shows that the co-sedimentation was pH-dependent. At pH 7.4 or above, very little P-ExeA co-sedimented with peptidoglycan, whereas at pH 6.0, the co-sedimentation was highest. These results were not due to precipitation of the protein at lower pH, as indicated by non-peptidoglycan controls (data not shown). The co-sedimentation was also affected by ionic strength. The protein showed decreased co-sedimentation at 50 mM NaCl and above in a buffer containing 40 mM sodium phosphate (pH 6.5) (Fig. 3B). Trials for affinity analysis did not yield a satisfactory saturable binding curve (data not shown). This may have been caused by the complex nature of the P-ExeA–peptidoglycan interactions, which we found to be accompanied by multimerization of the protein (see later).

image

Figure 3. Co-sedimentation of P-ExeA with peptidoglycan at various pHs and salt concentrations. The co-sedimentation was performed in 40 mM sodium phosphate buffers of various pHs (A) and pH 6.5 with various NaCl concentrations (B). The samples were analysed by SDS-PAGE and anti-ExeA immunoblot. The pellet samples in (B) were five times concentrated during resuspension.

Download figure to PowerPoint

Self-assembly of P-ExeA into large multimers at elevated temperature

Self-multimerization of P-ExeA was first observed by a temperature scan of dynamic light scattering (DLS), in which the protein started to multimerize at temperatures above 25°C (data not shown). The dynamic process of multimerization of N-His P-ExeA at 30°C is shown in Fig. 4A. After purification, the protein was stored at 4°C. Under this condition, the 29 kDa protein showed a hydrodynamic radius (RH) of 3.2 nm with a polydispersity of 31.1%. If a monomodal algorithm and a globular protein model were used, a 55 kDa species of apparent dimers was dominant in the sample; however, other species, possibly monomers and higher oligomers, might exist. This is consistent with an apparent dimer peak in gel filtration analysis (data not shown). The RH was increased to 6.3 nm with a polydispersity of 64.6% after incubation at 30°C, suggesting large multimers were formed. The RH was decreased when the sample was returned to 4°C; however it stabilized around 5.5 nm, suggesting the multimerization was only partially reversible. The particle size distribution was also analysed as shown in Fig. 4B. After incubation at 30°C, the scattered light intensity peak was significantly widened and extended to over 10 nm. Interestingly, a peak at ∼10 nm and a peak at ∼3 nm (corresponding to globular molecular masses of 700 kDa and 44 kDa respectively) were evident after the sample was returned to 4°C, suggesting that the protein was able to form stable multimers of RH 10 nm while the intermediate multimers were less stable. The presence of the two peaks was confirmed by gel filtration analysis, which indicated apparent molecular masses of ∼550 kDa and ∼50 kDa (data not shown). As also shown in Fig. 4, the G500D mutant showed a similar multimerization pattern as the wild-type protein; however, the C-His-tagged protein exhibited little if any multimerization.

image

Figure 4. Dynamic light scattering (DLS) analysis of P-ExeA multimerization at elevated temperature. Approximately 1 mg ml−1 N-His P-ExeA (▪), N-His G500D (▴) and C-His P-ExeA (◆) samples were analysed by DLS during incubation at 30°C for 1 h followed by 4°C for 11 h. A. Hydrodynamic radius (RH) curves of the P-ExeA variants. The RH values were calculated with a monomodal algorithm. B. Regularization histogram analysis of particle size distribution in the P-ExeA samples. The percentages of scattered light intensity are plotted against RH.

Download figure to PowerPoint

We also used a three-stage native gradient PAGE method modified from blue native PAGE (Schagger, 2001) to examine the multimerization. The purified N-His P-ExeA migrated predominantly as a single band (Fig. 5). However, after incubation at 37°C, a portion of the protein ran as multiple bands with high molecular mass. The distinctive multiple bands suggests that the components of the multimers had defined structures, and thus that the multimers were unlikely to be caused by denaturation and aggregation. To examine the stability of the P-ExeA multimers via this method, one sample was returned to 4°C overnight before being loaded on the gel. It showed that a decreased but still significant amount of multimer was present in the sample, confirming the DLS results.

image

Figure 5. Native PAGE of temperature-induced P-ExeA multimers. Approximately 2 mg ml−1 N-His P-ExeA samples were incubated at 4°C or 37°C for 1 h, followed by incubation at 4°C for 1 h or overnight* before electrophoresis at 4°C in a three-stage gradient native PAGE gel. The gel was stained with Coomassie Brilliant Blue (CBB) R-250.

Download figure to PowerPoint

These data suggest that P-ExeA has the potential to self-multimerize, although it should be noted that the temperature-induced self-assembly of P-ExeA in the above experiments was observed at high protein concentration (1–2 mg ml−1).

Separation of peptidoglycan fragments interacting with P-ExeA

Since peptidoglycan is a net-like structure with very large mass, cleavage by lytic enzymes such as lysozyme was required in order to solubilize the P-ExeA-binding components so that the interactions could be studied in more detail. The boiling SDS/pronase-purified A. hydrophila peptidoglycan was hydrolysed by lysozyme, and the fragments were applied to a Superose 6 gel filtration column. The resulting fractions were assayed for muramic acid concentration (Fig. 6A). A single major muramic acid peak eluted after the 12.4 kDa protein marker, suggesting that the peptidoglycan was almost completely hydrolysed.

image

Figure 6. Gel filtration and native PAGE analysis of lysozyme-hydrolysed peptidoglycan fragments. A. Muramic acid concentration distribution of peptidoglycan gel filtration fractions. SDS/pronase-purified A. hydrophila peptidoglycan was hydrolysed with lysozyme and applied to a Superose 6 column. The fractions were assayed for muramic acid concentration. The elution positions of protein markers are indicated at the top of the figure. B. Native PAGE analysis of interactions between P-ExeA and the peptidoglycan fractions. N-His P-ExeA was incubated with the peptidoglycan fractions at 4°C before native PAGE and anti-ExeA immunoblot. A sample induced to form multimers at 37°C in the absence of peptidoglycan fragments was also applied for comparison. The monomer and multimer bands of P-ExeA are indicated by the arrow and arrowhead respectively.

Download figure to PowerPoint

The peptidoglycan gel filtration fractions were examined for their ability to interact with P-ExeA by native PAGE. N-His P-ExeA was incubated with the peptidoglycan fragments at 4°C before being analysed in a native PAGE gel, which in this case had to be transferred and immunoblotted with ExeA antisera due to the small amounts of N-His P-ExeA in the assay. Figure 6B shows that the muramic acid peak fractions caused a portion of the protein to migrate at a higher molecular mass. An N-His P-ExeA sample induced to multimerize at 37°C in the absence of peptidoglycan fragments was also applied to the gel for comparison. Although the ladder of individual multimer bands that could be resolved in the Coomassie Brilliant Blue (CBB)-stained native PAGE gels (Fig. 5) could not be observed in these immunoblots, the results suggest that interactions between P-ExeA and the peptidoglycan fragments induced the protein to form multimers. We cannot rule out the possibility that the shift in migration is due to peptidoglycan binding rather than multimerization; however, this is unlikely given the small size of the fragments in the muramic acid peak used in the experiment.

Gel filtration analysis of P-ExeA multimerization in the presence of peptidoglycan fragments

We also used gel filtration to examine the P-ExeA–peptidoglycan interactions. Lysozyme-digested A. hydrophila peptidoglycan gel filtration fractions, which contained the muramic acid peak (Fig. 6A), were incubated with N-His P-ExeA at 4°C, followed by gel filtration on a Superose 6 column. The resulting fractions were analysed by anti-ExeA immunoblot (Fig. 7). In the absence of peptidoglycan fragments, the 29 kDa protein eluted with an apparent molecular mass of ∼18 kDa, possibly because of weak interactions with the column matrix in the low-salt buffer which was used to prevent interference with the peptidoglycan binding in these experiments. Upon incubation with the peptidoglycan fractions, however, the protein eluted in much higher molecular mass fractions, peaking with an apparent molecular mass of ∼400 kDa. These results confirmed the interactions between P-ExeA and the peptidoglycan fragments observed in the native PAGE assay (Fig. 6B). More interestingly, the P-ExeA-containing complex apparent molecular mass of 400 kDa was much higher than the apparent molecular mass combination of P-ExeA (18 kDa) and the peptidoglycan fragments (less than 12 kDa), suggesting that the protein formed large multimers.

image

Figure 7. Gel filtration analysis of P-ExeA multimers induced by peptidoglycan fragments. N-His P-ExeA, N-His G500D and C-His P-ExeA were incubated with peptidoglycan gel filtration fragments at 4°C before being applied to a Superose 6 column. P-ExeAs without incubation with peptidoglycan fragments were also applied as controls. The fractions were analysed by anti-ExeA immunoblot to determine the protein distribution. The positions of protein markers are indicated at the top of the figure.

Download figure to PowerPoint

Since the multimerization of P-ExeA required the presence of peptidoglycan fragments, it is apparent that the interactions with peptidoglycan induced the protein to multimerize. The two P-ExeA variants, N-His G500D and C-His P-ExeA, which showed decreased interactions with peptidoglycan (Figs 1 and 2), were also examined in the assay. Figure 7 shows that the G500D mutant shifted only slightly in apparent molecular mass. It thus either formed much smaller multimers or increased in size only due to peptidoglycan fragment binding. C-His P-ExeA, which showed little binding in the co-sedimentation assay, had almost no ability to form multimers.

Cross-linking analysis of P-ExeA multimerization in the presence of peptidoglycan fragments

We also used in vitro cross-linking to examine the interactions between P-ExeA and peptidoglycan fragments. If P-ExeA interacts with peptidoglycan fragments and consequently multimerizes, the protein should be cross-linked in different patterns compared with the non-peptidoglycan control. After incubation with peptidoglycan fragments at 4°C, N-His P-ExeA was cross-linked with 0–5 mM 3′-Dithiobis[sulphosuccinimidylpropionate] (DTSSP) on ice. A sample with no peptidoglycan and a sample induced to multimerize at 37°C in the absence of peptidoglycan were also cross-linked as controls. Figure 8 shows that the protein was cross-linked into closely spaced double bands as well as larger bands in the presence of the peptidoglycan fragments, as was observed for the P-ExeA multimers induced at 37°C, whereas the non-peptidoglycan control showed very little cross-linking. The cross-linked multimer bands corresponded to dimers and higher multimers of P-ExeA by comparison of their apparent molecular weights with protein markers. The cross-linked bands disappeared after treatment with β-mercaptoethanol (data not shown). This result indicates that the interactions between P-ExeA and the peptidoglycan fragments cause the protein to multimerize in a similar arrangement as that of the temperature-induced multimers. In contrast, the cross-linking of the G500D mutant in the presence of peptidoglycan showed little increase relative to the control, confirming the role of peptidoglycan binding in the multimerization (Fig. 8).

image

Figure 8. Cross-linking analysis of P-ExeA multimers. N-His P-ExeA and N-His G500D were incubated with peptidoglycan fragments at 4°C and cross-linked with 0–5 mM DTSSP on ice. Control samples with no peptidoglycan added and multimer samples induced at 37°C were also cross-linked. After cross-linking, the samples were applied to SDS-PAGE and immunoblotted with ExeA antiserum.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

In this study, interactions between the C-terminal/periplasmic region of ExeA (P-ExeA) and peptidoglycan were demonstrated by co-sedimentation, native PAGE, gel filtration and cross-linking studies in vitro. After treatment with boiling SDS and pronase, the peptidoglycan samples were still able to interact with P-ExeA, indicating that the protein interacts directly with peptidoglycan, rather than with proteins or other impurities that may have been included in the samples (Fig. 1). Strong evidence in support of specific interactions was provided by the G500D mutant which contained a substitution mutation in the putative peptidoglycan-binding domain. This mutant consistently showed a twofold decrease in co-sedimentation assays (Figs 1 and 2). Thus it was still able to significantly bind pure peptidoglycan in vitro, although it was barely cross-linked to peptidoglycan in vivo (Howard et al., 2006). The C-His P-ExeA construct, in which the His-tag was added close to the putative peptidoglycan-binding domain of wild-type P-ExeA, almost totally lost the ability to interact with peptidoglycan and therefore provided a good control for the maximum level of non-specific binding (Fig. 2). In addition, 40 mM sodium phosphate, 0.05% Tween 20 and 10 µg ml−1 bovine serum albumin (BSA) were used in the binding buffer to overcome weak and non-specific interactions. The P-ExeA–peptidoglycan interactions were characterized in different buffer conditions. The co-sedimentation was highest at pH 6.0 and almost abrogated at pH 7.4 and above (Fig. 3A). The periplasm is slightly acidic (pH 6.5) (Mitchell, 1966; Dhungana et al., 2003). The observed pH requirement for the co-sedimentation thus fits with the in vivo conditions.

The multimerization of P-ExeA at elevated temperature is intriguing. The multimers were not likely caused by denaturation and aggregation at high temperature, since they ran as distinctive bands in native PAGE and thus had defined structures (Fig. 5). The physiological significance of the P-ExeA multimers was suggested by the native PAGE and gel filtration experiments, in which the multimerization was induced by peptidoglycan at 4°C (Figs 6 and 7). Cross-linking analysis suggested that the peptidoglycan-induced P-ExeA multimers may have a similar arrangement as that of the temperature-induced multimers (Fig. 8). The G500D mutant, which showed decreased peptidoglycan co-sedimentation, had greatly decreased peptidoglycan-induced multimerization (Figs 7 and 8). Since the mutant showed similar multimerization as the wild type at elevated temperature (Fig. 4), it is more likely defective in peptidoglycan binding rather than multimerization. The C-His tagged variant, in contrast, was apparently unable to multimerize (Figs 4 and 7). The interactions with peptidoglycan perhaps introduce conformational changes in ExeA or lower the energy barrier to facilitate multimerization, which can alternatively be achieved by elevated temperature and high protein concentration. We cannot exclude the possibility however that the two forms of multimers have different ultrastructures, for example, circular versus linear filaments. It would be interesting to compare the ultrastructures of the multimers induced by peptidoglycan at different temperatures to facilitate the determination of the physiological form.

In vivo evidence of ExeA multimers was provided by our previous studies, in which the protein was cross-linked into an apparent dimer and higher multimers in addition to an ExeAB complex (Schoenhofen et al., 1998). Gel filtration analysis of detergent-solubilized A. hydrophila membranes also showed that the 85 kDa ExeAB complex eluted with apparent molecular masses of approximately 185 kDa and 500 kDa, suggesting that ExeAB was part of higher-order multimers (Schoenhofen et al., 2005).

The finding that P-ExeA forms multimers on peptidoglycan is novel but not unanticipated. Two components of the T2SS apparatus, GspE and GspD, have been shown to form a hexameric ring in the inner membrane platform and a 12- to 14-member multimeric ring in the outer membrane respectively (Nouwen et al., 1999; 2000; Robien et al., 2003; Chami et al., 2005). In fact ring-like structures are prevalent in many trans-envelope machineries, for example, the type IV pilus, flagella and type III secretion systems (Cornelis, 2006;Craig and Li, 2008). The predominant peptidoglycan-induced P-ExeA multimers were ∼400 kDa (Fig. 7), calculated to contain approximately 12 subunits of the 29 kDa domain, consistent with a secretin multimer containing 12–14 GspD subunits (Chami et al., 2005). These findings suggest that ExeAB facilitates T2SS assembly by forming a multimer, possibly a ring-like structure, on peptidoglycan. An ExeA ring (presumably also containing ExeB) could act as a scaffold to direct the assembly of the secretin through ExeA–D and/or ExeB–D interactions.

During this research, a report of physical interactions between peptidoglycan and a similar domain was published (Briers et al., 2007). In those studies, the putative peptidoglycan-binding domains of bacteriophage peptidoglycan-lytic enzymes KZ181 and EL183 were fused to green fluorescent protein (GFP). Both of the fusion proteins were able to stain Pseudomonas aeruginosa peptidoglycan sacculi prepared with SDS treatment, in buffers that did not contain detergents. The binding peaked at pH 8.0 and 8.5, higher than the optimum pH for the lytic activities of the two enzymes (6.2 and 7.3). In addition, the pH of the highest binding observed in our studies was 6.0 and the in vivo pH of the periplasmic space is ∼6.5. Although the discrepancy could be caused by different characteristics of KZ181/EL183 and ExeA, other explanations are possible. In our studies, we found that peptidoglycan samples could cause non-specific co-sedimentation of BSA and both wild-type and mutant P-ExeAs, at levels ∼1000 times greater than when 0.05% Tween 20 was added (data not shown).

The identity of the peptidoglycan fragment interacting with ExeA is not clear. It must be a common component conserved in different bacteria, because P-ExeA was able to interact with peptidoglycan samples from A. hydrophila, E. coli and B. subtilis (Fig. 1). This is also suggested by the wide distribution of the peptidoglycan-binding domain among both Gram-positive and Gram-negative bacteria (Pfam database for PF01471). However, a commercial muropeptide, GluNAc-MurNAc-dipeptide, failed to interact with P-ExeA (data not shown). It is possible that full-length muropeptides are required for the interactions. It is also possible that ExeA binds to special peptidoglycan structures that may include chemical modifications, for example, N-deacetylation, O-acetylation and N-glycolylation, or unusual arrangements, for example, tri-muropeptides and tetra-muropeptides (Vollmer, 2008). The peptidoglycan-binding domain may therefore not have a general peptidoglycan binding function, but a role in binding a special peptidoglycan structure during peptidoglycan biogenesis or at certain cell locations. Isolation and analysis of individual peptidoglycan components by HPLC and MS are required to further identify the peptidoglycan ligand of the peptidoglycan-binding domain of ExeA, as are ultrastructural studies of the ExeA multimers.

Experimental procedures

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

Strains and culture conditions

For production of the C-terminal/periplasmic domains of ExeA (P-ExeAs), E. coli BL21 (DE3) containing P-ExeA-encoding plasmids were grown at 35°C in 2×YT containing 50 µg ml−1 Kanamycin (Kn) to an OD600 of 0.6–0.8 and induced for 2 h by adding 0.4 mM IPTG. For purification of peptidoglycan, A. hydrophila C5.84, an exeA::Tn5–751 insertion mutant that is unable to produce either ExeA or ExeB (Ast et al., 2002), E. coli BL21 (DE3) or B. subtilis ATCC 6633 were grown in 2×YT to an OD600 of 1.5–1.8 at 30°C (A. hydrophila) or 37°C.

Construction and purification of P-ExeAs

DNA fragments encoding N-His- or C-His-tagged P-ExeAs were amplified from plasmid pRJ31.1 (wild type), pCG1 (F487S), pCG2 (L493S) or pCG3 (G500D) (Howard et al., 2006) by PCR with Pfu DNA polymerase. For N-His tag, 5′-CATATGCACCATCATCATCATCATCAGTTCTTCGGCTTCTTCCCC and 5′-CTCGAGTCAGGAAGCCTCCTCCGACAATGTG were used as primers. For C-His tag, 5′-CATATGCAGTTCTTCGGCTTCTTCCCCGAAC and 5′-CTCGAGGGAAGCCTCCTCCGACAATGTGAC were used. The fragments were inserted into the NdeI/XhoI restriction sites of the pET30a vector to construct p-exeA plasmids. All plasmids were sequenced to rule out the possibility of unwanted mutations before being electroporated into E. coli BL21 (DE3) for protein expression. For purification of P-ExeAs, cell pellets were resuspended in His-tag buffer A (20 mM Tris, pH 8.0, 250 mM NaCl, 10 mM imidazole, 10% glycerol). After addition of 100 µg ml−1 RNase, 1 µg ml−1 DNase and protease inhibitor cocktail tablet (Roche), the cells were disrupted by two passes through a French press at 12 000 lb in−2, followed by centrifugation to separate the P-ExeA-containing supernatant from cell debris. The samples were applied to HisTrap HP 5 ml columns (GE Healthcare) and P-ExeAs were eluted with a gradient of His-tag buffer B (20 mM Tris, pH 8.0, 250 mM NaCl, 500 mM imidazole, 10% glycerol). After desalting into anion exchange buffer A (20 mM Tris, pH 8.0, 25 mM NaCl, 0.2 mM PMSF, 1 mM EDTA, 10% glycerol) on a HiPrep 26/10 column (GE Healthcare), the samples were applied to a Resource Q 6 ml column (GE Healthcare) and eluted with a gradient of anion exchange buffer B (20 mM Tris, pH 8.0, 500 mM NaCl, 0.2 mM PMSF, 1 mM EDTA and 10% glycerol). The chromatography was performed on a FPLC system (GE Healthcare) at 4°C. The protein preparations were examined by SDS-PAGE for purity. Protein concentrations were determined by measuring absorbance at UV280 with theoretical extinction coefficients.

Purification of peptidoglycan and muramic acid assay

To purify non-denatured peptidoglycan, A. hydrophila C5.84 cells were disrupted by one pass through a French press at 12 000 lb in−2, followed by a low-speed centrifugation at 5000 g for 15 min to remove unbroken cells and a high-speed centrifugation at 40 000 g for 1 h to pellet cell envelopes (Sprott et al., 1994). The cell envelopes were washed with 2% Triton X-100 containing 20 mM MgCl2 and 5 mM EDTA, separately, to remove membrane lipids (Schoenhofen et al., 1998). After each wash, the peptidoglycan was collected by ultracentrifugation at 130 000 g for 1 h. After four washes with water, the peptidoglycan was resuspended in water. All steps were performed at 4°C to reduce peptidoglycan autolysis. Boiling SDS and pronase-treated A. hydrophila and E. coli peptidoglycan samples were prepared as described previously (Glauner, 1988). B. subtilis peptidoglycan was prepared as described previously (Bacher et al., 2001), except for an additional treatment with DNase and RNase to reduce nucleic acid contamination (Antignac et al., 2003). The peptidoglycan samples were quantified by a colorimetric method that measures the muramic acids of peptidoglycan (Hoijer et al., 1995). Muramic acid (0–1 mM) (Sigma) solutions were used as standards. Purified peptidoglycan samples were stored at 4°C.

Co-sedimentation

Peptidoglycan preparations (100 µM muramic acid units, 600 µM for B. subtilis) were incubated with 0.5–10 nM purified P-ExeAs in 40 mM sodium phosphate (pH 6.5), 0.05% Tween 20 and 10 µg ml−1 BSA at 4°C for 1 h. The mixtures were centrifuged at 21 000 g at 4°C for 1 h to pellet the peptidoglycan. The mixture, supernatant and pellet samples were applied to SDS-PAGE gels and immunoblotted with ExeA antiserum.

Dynamic light scattering

Dynamic light scattering was carried out on a DynaPro (Protein Solutions) instrument. Purified proteins (1 mg ml−1) in 40 mM sodium phosphate, pH 6.5 and 150 mM NaCl were analysed during incubation at 30°C for 1 h followed by 4°C for 11 h. At least 15 valid readings were recorded for each measurement.

Native gradient PAGE

Native gradient PAGE was performed as described previously (Schagger, 2001) with some modifications. The electrophoresis was run in a Mini-PROTEAN II system (Bio-Rad). Three-stage gradient acylamide gels were made by adding the following solutions in sequence: 2.5 ml of 18% T/3% C (total/cross-linker), 0.5 ml of 10.5% T/3% C and 1.5 ml of 3% T/3% C. N-His P-ExeA samples were mixed with 2× sample buffer (100 mM imidazole/HCl, pH 7.0, 20% glycerol) before loading. No CBB G-250 was added in the cathode buffer. The electrophoresis was performed at 4°C at 100 V for 1.5 h. The proteins were visualized by CBB R-250 staining as for SDS-PAGE gels or transferred to PVDF membranes and immunoblotted with ExeA antiserum.

Gel filtration chromatography

Gel filtration experiments were performed at 4°C with a flow rate of 0.5 ml min−1 on a FPLC system (GE Healthcare). For gel filtration of peptidoglycan fragments, 1 ml of 3.5 mM muramic acid units of A. hydrophila peptidoglycan were hydrolysed with 20 µg of lysozyme (Sigma) in 40 mM sodium phosphate (pH 6.5) at 37°C for 6 h, followed by incubation at 95°C for 20 min to inactivate the enzyme. Insoluble material was removed by centrifugation at 21 000 g at 4°C for 1 h. Approximately 0.5 ml of the supernatant was applied to a Superose 6 10/300 GL column (GE Healthcare) in the same buffer and fractionated into 1 ml fractions. For multimerization of P-ExeA in the presence of peptidoglycan fragments, 100 nM protein concentrations were incubated with 300 µM muramic acid units of peptidoglycan gel filtration fractions in 40 mM sodium phosphate (pH 6.5) with 0.05% Tween 20 at 4°C for 2 h. Approximately 0.5 ml of the mixtures were applied to a Superose 6 10/300 GL column (GE Healthcare) in the same buffer and collected in 1 ml fractions. The fractions were immunoblotted to determine the P-ExeA distribution. The columns were calibrated with a molecular weight marker kit (Sigma).

Chemical cross-linking

N-His P-ExeA (80 nM) was incubated with 200 µM muramic acid units of peptidoglycan fragments in 40 mM sodium phosphate (pH 6.5) with 0.05% Tween 20 at 4°C for 2 h. After an equal volume of 50 mM sodium phosphate (pH 8.5) containing 0–5 mM DTSSP was added, the mixtures were incubated on ice for 30 min and mixed with 2× SDS-PAGE sample buffer (no β-mercaptoethanol added) to stop the cross-linking. The samples were applied to 10% SDS-PAGE gels, transferred to PVDF membranes and immunoblotted with ExeA antiserum.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References

This research was supported by grants from the Natural Sciences and Engineering Research Council of Canada and from the Canadian Institutes of Health Research. DLS experiments at the Saskatchewan Structural Sciences Centre were supported by a group grant to the Molecular Design Research Group from the Saskatchewan Health Research Foundation. We thank Harold Bull for helpful discussions.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Results
  5. Discussion
  6. Experimental procedures
  7. Acknowledgements
  8. References