|Home | About | Journals | Submit | Contact Us | Français|
In Gram-negative bacteria, the peptidoglycan (PG) cell wall is a significant structural barrier for outer membrane protein assembly. In Aeromonas hydrophila, outer membrane multimerization of the type II secretion system (T2SS) secretin ExeD requires the function of the inner membrane assembly factor complex ExeAB. The putative mechanism of the complex involves the reorganization of PG and localization of ExeD, whereby ExeA functions by interacting with PG to form a site for secretin assembly and ExeB forms an interaction with ExeD. This mechanism led us to hypothesize that increasing the pore size of PG would circumvent the requirement for ExeA in the assembly of the ExeD secretin. Growth of A. hydrophila in 270 mM Gly reduced PG cross-links by approximately 30% and led to the suppression of secretin assembly defects in exeA strains and in those expressing ExeA mutants by enabling localization of the secretin in the outer membrane. We also established a heterologous ExeD assembly system in Escherichia coli and showed that ExeAB and ExeC are the only A. hydrophila proteins required for the assembly of the ExeD secretin in E. coli and that ExeAB-independent assembly of ExeD can occur upon overexpression of the d,d-carboxypeptidase PBP 5. These results support an assembly model in which, upon binding to PG, ExeA induces multimerization and pore formation in the sacculus, which enables ExeD monomers to interact with ExeB and assemble into a secretin that both is inserted in the outer membrane and crosses the PG layer to interact with the inner membrane platform of the T2SS.
IMPORTANCE The PG layer imposes a strict structural impediment for the assembly of macromolecular structures that span the cell envelope and serve as virulence factors in Gram-negative species. This work revealed that by decreasing PG cross-linking by growth in Gly, the absolute requirement for the PG-binding activity of ExeA in the assembly of the ExeD secretin was alleviated in A. hydrophila. In a heterologous assembly model in E. coli, expression of the carboxypeptidase PBP 5 could relieve the requirement for ExeAB in the assembly of the ExeD secretin. These results provide some mechanistic details of the ExeAB assembly complex function, in which the PG-binding and oligomerization functions of ExeAB are used to create a pore in the PG that is required for secretin assembly.
The secretion of protein toxins and enzymes via the type II secretion system (T2SS) is an important virulence mechanism as well as a major route of scavenging enzyme secretion and surface protein assembly for many Gram-negative bacteria (1,–4). The T2SS is a large complex composed of 12 to 15 highly conserved proteins designated GspC through GspO together with a complement of assembly proteins, including GspAB and GspS, which are variably found among different bacterial species. The basic structure of the T2SS includes an inner membrane platform (GspC, GspE, GspF, GspL, and GspM), a periplasmic pseudopilus (GspG through GspK), and an outer membrane channel formed by the secretin protein GspD. Substrates of the T2SS are translocated from the cytosol to the periplasmic space by the Sec or Tat pathways, and folded proteins are then exported across the outer membrane through the pore formed by GspD (5,–9). The T2SS secretin is homologous in sequence and structure to secretins from other cell envelope-spanning complexes, including the type IV pilus system and the type III secretion system (T3SS) (10, 11).
GspD is an 80-kDa protein that assembles into a dodecameric secretin in the outer membrane (5, 12, 13). The secretin has a C-terminal transmembrane domain and a large N-terminal periplasmic domain comprised of four subdomains designated N0 to N3 (14, 15). In Aeromonas hydrophila, assembly of the ExeD secretin requires a complex of two inner membrane assembly factors, the 60-kDa AAA ATPase protein ExeA and the 25-kDa basic protein ExeB (16,–18). Mutation of ExeAB results in accumulation of ExeD monomers in the inner membrane, suggesting that ExeAB may be required for the transport of ExeD to the outer membrane as well as for assembly of the secretin (18). ExeB is required to maintain the stability of ExeA (17); in fact, in Vibrio vulnificus, the GspA and GspB proteins are fused to create one protein that has the periplasmic domain of GspB fused to the C-terminal end of GspA (19). The periplasmic domain of ExeA has a peptidoglycan (PG)-binding domain that has been shown to bind PG from A. hydrophila, Escherichia coli, and Bacillus subtilis (20). Interaction with PG results in multimerization of ExeA in vitro, and in vivo ExeAB assembles into large complexes of up to 12 monomers of each protein (20, 21). Point mutations in conserved residues within the PG-binding domain of ExeA demonstrated that PG binding is required for ExeD assembly (20). Recent results have also indicated that ExeB directly interacts with the amino terminal domain of ExeD (22).
Structural analyses of the T2SS and T3SS secretins suggest that they completely span the periplasm and would thus be built directly through the PG net (10). Despite several decades of research, current models for the assembly of cell envelope-spanning complexes, such as the T2SS, have not adequately addressed the structural impediment of the PG sacculus. The covalently closed structure and relatively small (approximately 20-Å) pore size of the PG are significant structural barriers that must be overcome to allow large proteins to access the outer membrane (23). The pore size of the PG is predicated by the degree of peptide cross-linking (24). It has been determined that the size limit for diffusion of proteins through isolated sacculi is approximately 50 kDa; therefore, modification of cross-linking density (at least locally) would be required for both the transport of the 80-kDa ExeD monomer into the outer membrane and assembly of the trans-PG secretin multimer (25,–27). Several transenvelope complexes, including the type IV pili, flagella, and type III and type IV secretion systems, have dedicated PG-degrading enzymes that may be involved in localized hydrolysis of peptide cross-links (23, 28); however, there are no known PG-degrading enzymes associated with the T2SS.
Our current working model for ExeAB-mediated assembly of ExeD is that the ExeAB complex in the inner membrane binds to PG through interactions involving ExeA and recruits ExeD for assembly via interactions involving ExeB (18, 20, 22). Given the structural barrier imposed by the PG sacculus, we further hypothesize that the function of the ExeA PG-binding domain is to modify the pore size of the PG, possibly by locally perturbing the growth of the sacculus, to allow the transport and assembly of ExeD in the outer membrane. As one test of this hypothesis, in this study, we used chemical and genetic methods to reduce the amount of cross-linking in the PG and analyzed secretin assembly and secretion in a number of strains expressing ExeA mutant proteins. Growth in the presence of high concentrations of Gly and d-amino acids has been shown to reduce cross-linking in B. subtilis and E. coli, respectively (29, 30). ExeA mutants of A. hydrophila that were grown in medium containing 270 mM Gly or 38 mM d-Met assembled the ExeD multimer by an ExeAB-independent mechanism. In addition, we have developed a heterologous assembly system for ExeD in E. coli, in which ExeD secretin assembly is dependent upon the coexpression of ExeAB. In this system, the overexpression of the carboxypeptidase PBP 5 resulted in the assembly of the secretin even in the absence of ExeAB expression. Collectively, these results demonstrate that alterations in PG cross-linking enable ExeA-independent assembly of ExeD and suggest that the function of the PG-binding domain of ExeA is to alter the PG structure to allow transport of ExeD across the cell wall and assembly of the secretin within the cell envelope.
Growth in high concentrations of Gly has been shown to alter the stem peptide structure of PG and prevent cross-linking in several Gram-positive bacterial species, including B. subtilis, which has the same stem peptide sequence as A. hydrophila (29). A. hydrophila strain Ah65 was grown in buffered Luria-Bertani (LB) medium with various concentrations of Gly to determine the effect on growth and outer membrane integrity. Growth in up to 270 mM Gly did not adversely affect the growth rate of A. hydrophila (Fig. 1A) after 4 h (mid-log phase). Cell lysis can lead to spontaneous secretin assembly; therefore, assays of β-lactamase activity in supernatants from Ah65 grown in 270 mM Gly were also conducted (Fig. 1B). No leakage of periplasmic contents was detected, thereby indicating that growth in 270 mM Gly does not lead to significant cell lysis. As a result, the 270 mM concentration of Gly was used for further experimentation.
Growth in 270 mM Gly significantly reduced the intrinsic resistance of A. hydrophila to carbenicillin and ampicillin. The MICs for carbenicillin and ampicillin were reduced 5- to 10-fold from 1 mg/ml to 0.2 or 0.1 mg/ml for all of the strains tested. This positive synergistic effect suggests that Gly alters A. hydrophila PG structure through a mechanism similar to that demonstrated for Gram-positive bacteria (29). Growth in buffered LB medium with 270 mM Gly also induced filamentation in some cells and lysis after 12 h of growth (data not shown). These cell morphology changes are similar to those observed for β-lactam antibiotics, providing further support to the hypothesis that growth in Gly also induces alterations in the PG structure of A. hydrophila.
Growth in Gly had not previously been shown to alter PG structure in any Gram-negative bacteria, including A. hydrophila. Therefore, we isolated PG from Ah65 grown with or without 270 mM Gly and analyzed the structure by ultraperformance liquid chromatography (UPLC) (Fig. 2). The major peaks observed in the absence of Gly corresponded to the tetra monomer (M4) and tetra-tetra dimer (D44) muropeptides, which is consistent with the previous results observed in E. coli (31). The PG from cells grown with 270 mM Gly had additional major peaks that indicated the incorporation of Gly at position 4 of the stem peptide (M4G and D44G). The cross-linking of the PG was also reduced by approximately 30% in the Gly samples, from 27.7% ± 1.56% (LB medium) to 17.8% ± 0.820% (LB medium, 270 mM Gly). This reduction is significantly higher than the approximate 18% decrease observed in B. subtilis (29).
To determine whether chemical modification of the PG structure by growth in Gly or d-Met can compensate for the lack of the PG-binding activity of ExeA mutants, we grew wild-type A. hydrophila and exeA strains expressing ExeA mutants in medium with or without 270 mM Gly and analyzed secretin assembly by immunoblotting (Fig. 3A). The strains that were tested included the wild type, a strain with a Tn5 insertion in exeA (C5.84), and strains expressing four different point mutations within the region encoding the PG-binding domain in exeA together with exeB in the exeA mutant background. Three of the exeA point mutations (Q488A, T503A, and L507A) have been shown to interfere with the PG binding of ExeA and prevent ExeD secretin assembly, whereas the substitution mutation G497A does not (32). Consistent with previous reports, we found that only the wild type and the exeA strain expressing the G497A mutant of ExeA (together with ExeB) were able to assemble an appreciable amount of ExeD secretin when cells were grown without Gly. However, when the strains were grown with 270 mM Gly, we observed secretin assembly in all of the mutant strains tested.
In addition to secretin assembly, chemical modification of PG by growth in Gly also restored secretion of the T2SS substrate lipase in all of the mutants tested (Fig. 3B). Growth in Gly led to a less than 2-fold increase in extracellular lipase in the exeD deletion strain AhD14, whereas supernatants from the exeA mutants tested had statistically significant increases in lipase activity of approximately 4- to 10-fold when grown in Gly compared to that when grown without (P < 0.05; Student's t test). Since Gly does not cause significant cell lysis or leakage of periplasmic constituents in the growth conditions used in this study, these lipase assay results suggested that the ExeD secretin assembly observed upon growth in Gly reflects functional assembly of the secretion system.
In E. coli, growth in a medium containing high concentrations of d-Met was shown to severely reduce the amount of cross-linking in PG due to the incorporation of d-Met into the stem peptide (30). The synthetic enzymes and structure of A. hydrophila PG are very similar to those of E. coli; therefore, it is reasonable to expect that d-Met would have a similar effect on the A. hydrophila PG. To determine whether growth in d-Met would also enable ExeD secretin assembly in ExeA mutants, the strains described above were grown in buffered LB medium with and without 38 mM d-Met, and secretin assembly was analyzed. In a manner similar to growth in Gly, growth in d-Met did not affect secretin assembly in the wild type or the exeA(pMMB207)::exeA(G497A)B strain. However, addition of d-Met to the growth medium enabled partial restoration of secretin assembly in the exeA strain C5.84 and in the exeA mutants expressing the three PG-binding site mutations in ExeA (Q488A, T503A, and L507A) (Fig. 4). Note that the MIC of d-Met was determined to be >50 mM for A. hydrophila.
In addition to the PG-binding function of ExeA, the assembly factor complex ExeAB also contains a cytoplasmic ATPase domain associated with ExeA, whose function is essential for secretin assembly (17), and an ExeD-binding function associated with ExeB (18, 22). Growth in Gly restored ExeD secretin assembly in the C5.84 exeA mutant; however, the amount of assembled secretin observed was significantly less than that in the wild type and in several strains expressing ExeA mutants with point mutations in the PG-binding domain (Fig. 3A). These results suggested the possibility that Gly-mediated secretin assembly is partially dependent on the other functions of ExeAB described above. Therefore, we performed an induction experiment with the exeA C5.84 mutant that had ExeA(G497A)B expressed from the inducible plasmid pMMB207. The ExeA(G497A)B derivative was used in lieu of wild-type ExeA because the plasmid containing the wild-type gene displays basal expression levels sufficient to partially complement the C5.84 mutant (18). The wild type and the C5.84 strain with pMMB207::exeA(Q488A)B (incapable of binding PG) were used as positive and negative controls, respectively. In the absence of Gly, the amount of assembled ExeD in the C5.84(pMMB207)::ExeA(G497A)B strain was dependent on the amount of ExeAB expressed (Fig. 5), whereas the C5.84(pMMB207)::ExeA(Q488A)B strain did not assemble ExeD regardless of the amount of ExeA expressed. In contrast, growth with 270 mM Gly enabled assembly of ExeD under all conditions for all strains tested, even when no detectable ExeA was present. These results suggest that Gly-mediated assembly of ExeD occurs by an ExeAB-independent mechanism.
Sucrose gradient fractionation was used to separate cell envelope preparations into inner and outer membrane fractions (Fig. 6). Fractions containing inner membrane vesicles were identified by assaying for activity of the inner membrane protein NADH oxidase (Fig. 6A). SDS-PAGE and Coomassie blue staining of fractions were used to detect the outer membrane-containing fractions by identifying the outer membrane porins OmpC and OmpF (Fig. 6B). Anti-ExeD immunoblots of the fractions from Ah65 samples grown with or without Gly showed that the majority of the secretin is associated with the outer membrane fractions. In the exeA mutant, the secretin is associated with the inner membrane fractions, a finding that is consistent with previously reported results. However, upon chemical modification of PG by growth in 270 mM Gly, a significant amount of secretin was localized within the outer membrane fractions in this strain. These results confirmed that the decreased PG cross-linking that occurs due to chemical modification of the PG enables localization and assembly of the ExeD secretin in the outer membrane.
Our current working model for the assembly of the ExeD secretin in A. hydrophila is that the ExeAB complex interacts with PG through the PG-binding domain of ExeA (20, 32) and recruits ExeD for assembly through interaction with ExeB (22). Since interactions between PG, ExeAB, and ExeD are the main components of this model, we sought to establish a heterologous secretin assembly system in E. coli by coexpressing only ExeAB and ExeD. In addition to providing further support for our proposed model, such an assembly system would have the advantage of being amenable to genetic analysis using the tools and mutant collections available for E. coli. To construct strains for assembly assays, plasmids pVACD, pRJ100.1, and pBAD(AB), which express proteins ExeCD, ExeD, and ExeAB, respectively, were transformed individually or together into an E. coli W3110 strain, which contains deletions in both the ara gene to allow induction of pBAD(AB) with arabinose and degP to prevent degradation of the ExeD secretin (data not shown).
In W3110 strains that expressed ExeCD without ExeAB, ExeD was stably expressed primarily in the monomeric form. However, in cells coexpressing ExeCD and ExeAB, the amount of multimer detected corresponded with the concentration used to induce ExeAB expression (Fig. 7). In the absence of ExeC, the relative amount of ExeD present in both the monomeric and multimeric forms is negligible in comparison to that in conditions in which ExeC is coexpressed with ExeD. This result would suggest that the expression of ExeC is beneficial to maintaining the stability of ExeD in the E. coli model. Interestingly, ExeAB-independent assembly was observed when the E. coli strains were grown in concentrations greater than 130 mM Gly (data not shown), a result that is consistent with the results described above for A. hydrophila.
In addition to chemical mechanisms, it is possible to alter PG structure genetically by altering the expression levels of the penicillin-binding proteins (PBPs) that synthesize and modify the sacculus. To take advantage of previously constructed PBP-expressing plasmids in E. coli, we used the E. coli assembly system described above. There are 12 known PBPs in E. coli that have been well-characterized (33); however, mutagenesis of PBPs is a difficult approach for altering PG structure because of the redundancy in the PBP network. For example, the two main PG synthetic enzymes, PBP 1a and PBP 1b, have redundant activities, but deletion of both enzymes is lethal (34).
To genetically alter the PG structure of E. coli, we overexpressed the low-molecular-weight carboxypeptidase PBP 5 (encoded by dacA), which is the most abundant PBP in E. coli (35). Carboxypeptidases cleave the terminal d-alanine from the stem peptide and are important for regulating the number and type of peptide cross-links (36, 37). Overexpression of PBP 5 leads to aberrant cell morphology after prolonged growth, and it is proposed to inhibit cell wall elongation by competing with the class A and B PBPs that catalyze transpeptidation reactions (35, 37, 38).
To determine the effect of genetically altering the PG structure in E. coli, W3110 strain TS40 was cotransformed with pVACD and plasmid pLP4 or pLP4::dacA (37) and assembly assays were performed (Fig. 8). The TS40 strain had both the wild-type dacA and the plasmid-carried dacA that was expressed from an isopropyl-β-d-thiogalactopyranoside (IPTG)-inducible promoter. In the control TS40 strain with the pLP4 empty vector, ExeD was observed in monomeric and multimeric forms. Overexpression of PBP 5 from pLP515, however, resulted in more multimer production than that in the absence of PBP 5 overexpression. To confirm that PBP 5 was overexpressed upon induction of pLP4::dacA, penicillin-binding proteins were detected by Bocillin FL labeling (39) in cell envelope preparations separated by SDS-PAGE. As shown in Fig. 8B, according to the relative fluorescence units (RFU) of the PBP 5 band, the amount of PBP 5 observed increased 15% upon induction with 0.01 mM IPTG, 22% upon induction with 0.05 mM IPTG, and 32% upon induction with 0.1 mM IPTG above the wild-type level of PBP 5.
Elucidating the mechanism of T2SS secretin assembly in the outer membrane has been a major focus of research; however, current models do not account for the structural barrier imposed by the PG sacculus in the cell envelope or for the size of the secretin in relation to the position of the PG in the periplasm (23, 40). In A. hydrophila, assembly of the ExeD secretin requires the assembly factors ExeA and ExeB, components of the inner membrane ExeAB complex with a PG-binding domain in ExeA and an ExeD-binding domain in ExeB. In this study, we have used chemical and genetic approaches to demonstrate that increasing the PG pore size by reducing the amount of peptide cross-linking circumvents the requirement for ExeA in assembly of the ExeD secretin. These results provide evidence that the main function of ExeA is to modify the pore size of PG to allow transport of ExeD to the outer membrane.
Previously, we provided evidence that suggests that ExeB acts as a scaffold by interacting directly with both ExeA and ExeD (22). In addition, PG binding by ExeA induces multimerization of ExeA (20). Based on these results, we have proposed the following model to explain the mechanism in which ExeAB supports assembly of the ExeD secretin: PG binding by ExeA induces multimerization and the formation of a pore in the sacculus that enables ExeD monomers, through interaction with ExeB, to assemble into a secretin that crosses the PG layer and is inserted in the outer membrane. To provide evidence that ExeA creates a pore in the PG, we initially asked whether chemical agents that restructure the PG can replace the function of ExeA. The most well-known PG-modifying chemicals are the β-lactam antibiotics that function by resembling the stem peptide and binding irreversibly to PBPs, thereby inhibiting the transpeptidation enzymes that form the PG cross-links; however, A. hydrophila is resistant to high concentrations of β-lactam antibiotics due to the presence of a number of chromosomally encoded β-lactamases (41,–43). Alteration of the PG structure by chemical means has been shown in B. subtilis and E. coli, whereby growth in elevated concentrations of Gly and d-Met, respectively, results in the replacement of d-Ala in position 4 or 5 of the stem peptide and decreases in cross-linking of 18% (Gly) and 25% (d-Met) (29, 30). We found that growth in 270 mM Gly also resulted in significant incorporation of Gly into position 4 of the stem peptide of the A. hydrophila PG (Fig. 2). In addition, cross-linking in the PG was reduced from 27.7% ± 1.56% to 17.8% ± 0.82%. Incorporation of Gly or d-Met in place of d-Ala in A. hydrophila PG may prevent cross-linking by several putative mechanisms, such as by inhibiting high-molecular-weight PBPs or because the modified stem peptides may be poor substrates for transpeptidation (29, 30).
We used several methods to ensure that the secretin assembly observed in strains expressing ExeA mutants was not an artifact of cell lysis induced by alterations in PG cross-linking. For instance, the growth conditions that we used did not alter the growth rate of the cultures, and samples were taken during log phase (Fig. 1A). Lipase activity assays showed significant increases in extracellular lipase in supernatant from strains expressing PG-binding site mutants of ExeA grown in 270 mM Gly but not in an exeD deletion strain, indicating that growth in Gly enables specific secretion of the lipase, as opposed to nonspecific leakage from the periplasm (Fig. 3B). Furthermore, β-lactamase activity assays of cell extracts and supernatants showed that the PG-altering conditions used in the study did not result in nonspecific leakage of periplasmic contents (Fig. 1B).
Once it was established that the chemical agent Gly did modify PG by decreasing the amount of PG cross-linking and that growth in this agent did not cause a pleiotropic growth defect that leads to leakage of periplasmic constituents, we were able to determine the effect on secretin assembly in A. hydrophila. In wild-type A. hydrophila, the ExeD secretin is localized in the outer membrane, whereas in an exeAB mutant, ExeD is localized to the inner membrane. Addition of Gly to the growth medium did not affect the localization of the secretin in the wild type but enabled the ExeD secretin to assemble in the outer membrane of the exeAB mutant (Fig. 6C). Likewise, growth in Gly supported the assembly of the ExeD secretin (Fig. 3A) and lipase secretion (Fig. 3B) in strains of A. hydrophila that expressed PG-binding site mutations in ExeA. These results confirmed that chemical modification of PG by Gly and the resultant decrease in PG cross-linking can complement the lack of the PG-binding protein ExeA for assembly of the ExeD secretin multimer in the outer membrane. Similar increases in secretin assembly in the absence of functional ExeAB were observed when the cells were grown in the presence of d-Met (Fig. 4), which has also been shown to reduce PG cross-linking (30).
Without chemical modification of PG, strains that expressed the ExeA PG-binding site mutation Q488A, T503A, or L507A exhibited the same low level of assembled ExeD secretin observed in the exeAB strain C5.84. However, upon chemical modification of PG, expression of the ExeA PG-binding site mutants increased the amount of ExeD secretin observed in comparison to that in the exeAB strain (Fig. 3A and and4).4). Likewise, when the Q488A PG mutant was induced in a gradient manner in the exeAB strain under PG modification conditions, a larger amount of the ExeD multimer was observed at the highest induction level (Fig. 5). These results suggest that while Gly-mediated ExeD assembly can occur independently of all of the functions of ExeAB, the ATPase function of ExeA and the ExeD-binding function of ExeB are still able to increase the efficiency of secretin assembly. These results also provide some mechanistic information regarding the function of the ExeAB assembly complex. The finding that the expression of ExeAB containing mutations in the ExeA PG-binding site supports additional secretin assembly only upon chemical modification of PG suggests that in wild-type cells the PG-modifying function of the ExeA PG-binding domain is required before the ExeD-binding function of ExeB can take place. These results are preliminary, however, and will require more investigation.
Chemical modification of PG that results in less cross-linking of the sacculus may perform a function analogous to the PG-binding domain of ExeA since our results show that the assembly of the ExeD secretin and function of the T2SS can occur in the absence of ExeAB under conditions that lower PG cross-linking. However, the amount of secretin that is assembled in the absence of ExeAB under conditions of PG chemical modification is much smaller than that observed in the wild type without chemical modification. Despite this, wild-type levels of lipase were secreted by the exeA mutant under these conditions (Fig. 3), which may reflect an excess secretin level in wild-type cells as was previously observed in Vibrio cholerae (44). The reduced level of multimer formation observed in these cells is not surprising given that the assembly factors ExeAB direct efficient assembly of the T2SS, whereas assembly in the mutants grown in Gly is likely proceeding via diffusion or some other nonspecific mechanism. This hypothesis is further supported by the results of the sucrose gradient separation of inner and outer membranes, which showed that ExeD is localized to the inner and outer membranes in the ExeA mutant grown in Gly whereas all of the secretin is localized to the outer membrane in the wild-type strain.
We sought to establish a heterologous ExeD secretin assembly system in E. coli and found that ExeAB are required to observe the assembly of ExeD secretin. In addition, coexpression of ExeC significantly facilitated the assembly and/or stability of the ExeD secretin in E. coli (Fig. 7). This result was perhaps not surprising given the numerous studies that have demonstrated several different points of interaction between GspC and GspD (45,–48); as a result, coexpression of ExeC with ExeD would more closely resemble a wild-type situation than expression of ExeD alone. The necessity for additional, highly conserved assembly factors, such as the Bam complex, has not been determined in A. hydrophila, although Bam-independent assembly of the T2SS secretin has been shown for other bacteria (49, 50).
Overexpression of the carboxypeptidase PBP 5 has been shown to prevent PG cross-linking by cleaving the terminal d-Ala residue from the stem peptide (35). In E. coli expressing plasmid-encoded ExeD, the protein was observed in monomeric and multimeric forms, whereas coexpression of PBP 5 resulted in a significant increase in the secretin multimer. These results thus provided evidence that alteration of the PG structure by genetic means can also support the assembly of the ExeD secretin without the function of ExeAB.
The PG-binding domain of ExeA has no known enzymatic activity; therefore, the mechanism by which it can modify PG cross-linking is not immediately apparent. It has previously been shown that the periplasmic domain of ExeA will interact with the PG from E. coli and B. subtilis, both of which have the same stem peptide sequence as A. hydrophila (20, 29). With the results described here, we envision a model in which the ExeA component of the ExeAB complex facilitates secretin assembly by establishing sites in the PG network through which ExeD can pass and assemble in the outer membrane to form the secretin. ExeA may bind to the stem peptide and block PG cross-linking by preventing PBPs from accessing their substrates. Multimerization of ExeA may then allow the formation of a pore stabilized by the ExeAB complex that recruits ExeD through interaction with ExeB. Further studies are required to determine the nature of the interactions between ExeA and PG and/or the PG synthetic machinery.
The strains and plasmids used are listed in Table 1. A. hydrophila strains were grown at 30°C in buffered Luria-Bertani (LB) medium, pH 7.5 (18). E. coli strains were cultured in LB medium at 37°C. Antibiotics were used at the following final concentrations (μg · ml−1) when necessary: ampicillin (Ap), 100; chloramphenicol (Cm), 2.5; kanamycin (Km), 50; tetracycline (Tc), 10; and gentamicin (Gm), 50. To determine the effect of 270 mM Gly and 38 mM d-methionine on growth, strains were subcultured 1:100 into 10 ml of buffered LB medium with 1.25 μg · ml−1 Cm, 0.04 mM IPTG, and Gly or d-methionine as indicated and grown at 30°C to an optical density at 600 nm (OD600) of 2.0. Secretin assembly and lipase secretion were analyzed as described below. MIC assays for carbenicillin and ampicillin were performed by subculturing strains 1:100 in 1 ml of LB medium with 0.04 mM IPTG and serial dilutions of carbenicillin or ampicillin (0 to 10 mg · ml−1). Results were determined after overnight incubation at 30°C with shaking. At least three independent replicates were performed for each strain.
Ah65 was subcultured 1:100 into 400 ml of buffered LB medium or buffered LB medium with 270 mM Gly. Cultures were harvested at an OD600 of approximately 2.0. The PG was isolated using the boiling SDS method (31). Muropeptides for UPLC analysis were prepared by digesting PG samples with 20 μg lysozyme at 37°C overnight. Digested PG samples were freeze-dried prior to further analysis. Supernatants were reduced by adding 150 μl of 0.5 M sodium borate (pH 9.5) and sodium borohydride to a final concentration of 10 mg/ml and incubating at room temperature (RT) for 30 min. Finally, samples were adjusted to pH 3.5 with phosphoric acid.
UPLC analyses of muropeptides were performed on an Acquity UPLC BEH C18 column (130 Å, 1.7 μm, and 2.1 mm by 150 mm; Waters) and detected at 204 nm. Muropeptides were separated using a linear gradient from buffer A (phosphate buffer, 50 mM, pH 4.35) to buffer B (phosphate buffer, 50 mM, pH 4.95; methanol, 15% [vol/vol]) over a 20-min run.
The identity of the peaks was determined by comparison of the retention times and profiles to other chromatograms and confirmed by mass spectrometry. The percentage of cross-linkage was calculated as described previously (31). The values are the means from three independent experiments.
SDS-PAGE gels (8% to 10%) were routinely used to analyze protein samples. For analysis of ExeD secretin, 3% to 8% Criterion precast polyacrylamide Tris-acetate gradient gels (Bio-Rad) were used. Gradient gel samples were standardized to 0.01 OD600 per lane. For immunoblotting, the proteins were transferred to polyvinylidene difluoride (PVDF) membranes (GE Healthcare Life Sciences) using a Bio-Rad Trans-blot Turbo apparatus at 1 A and 25 V for 45 min, which resulted in the best transfer of the secretin but often resulted in poor visualization of the ExeD monomer, presumably because it was driven through the membrane. Visualization of ExeD was achieved by incubation with the appropriate rabbit antiserum followed by incubation with peroxidase-conjugated mouse anti-rabbit IgG (Sigma). The signal was developed with a chemiluminescent substrate kit (GE Healthcare Life Sciences).
Lipase activity was assayed by measuring the increase in absorbance at 405 nm from the release of p-nitrophenol from p-nitrophenol caprylate (pNPC) as described previously (51). The culture supernatants (200 μl) were added to 800 μl of substrate buffer containing 1 mM p-nitrophenol caprylate, 100 mM Tris (pH 8.0), and 0.2% Triton X-100, and the reaction mixture was incubated for 30 min at RT, during which the OD was measured at 5-min intervals. One unit of lipase activity equals 1 nmol pNPC hydrolyzed per min. The lipase activity in units per milliliter per OD600 was compared to that of the wild-type strain to calculate the percentage of lipase secretion.
For β-lactamase assays, cells were harvested at an OD600 of approximately 2.0. Supernatants were recovered by centrifugation followed by filtration through a 0.45-μm filter. Cell pellets were lysed with a French press, and cell debris was removed by centrifugation at 15,000 × g for 15 min at 4°C. A 50-μl aliquot of supernatant or cell extract was used for the assay. The β-lactamase assay buffer contained 50 mM Tris-HCl (pH 8.0) and 0.1 mg/ml nitrocefin, and the change in absorbance at 510 nm was monitored.
To remove the gIII signal sequence from pBAD/gIII, a 120-bp fragment containing an NdeI site was amplified from pSK-T using primers US45 (5′-AGATTAGCGGATCCTACCTG-3′) and US46 (5′-CGAGCTCCATGGTCATATGTAATTCCTCCT-3′) and cloned into the BamHI and NcoI sites of pBAD/gIII-C. Restriction sites in primers are underlined. A 2.4-kb fragment containing exeAB was amplified with primers US47 (5′-TGGATCCATATGTACACACAG-3′) and US48 (5′-CTCAGGCTCCCTCTAGAAATGGACG-3′) and cloned into the NdeI and XbaI sites of the modified vector. Successful cloning was verified by restriction digestion and sequencing.
To construct strains for assembly assays, pVACD-P and pBAD(AB), which express ExeCD and ExeAB, respectively, were transformed into TS43. To determine the effect of PBP 5 expression on assembly, pVACD-P was cotransformed with pLP4 or pLP515 into TS40. Assembly assays were performed by subculturing strains 1:100 into LB medium with arabinose and/or IPTG, as indicated, and growing at 30°C to an OD600 of approximately 2.0. The amount of assembled ExeD was determined by gradient gel electrophoresis and immunoblotting of whole-cell samples as described above.
An unpaired two-sided Student t test was used for all statistical analyses. Values were considered significantly different at a P value of less than 0.05.
We thank Kevin Young and Gang Li for the gift of the pLP4 and pLP515 plasmids.