|Home | About | Journals | Submit | Contact Us | Français|
The Pla surface protease of Yersinia pestis activates human plasminogen and is a central virulence factor in bubonic and pneumonic plague. Pla is a transmembrane β-barrel protein and member of the omptin family of outer membrane proteases which require bound lipopolysaccharide (LPS) to be proteolytically active. Plasminogen activation and autoprocessing of Pla were dramatically higher in Y. pestis cells grown at 37°C than in cells grown at 20°C; the difference in enzymatic activity by far exceeded the increase in the cellular content of the Pla protein. Y. pestis modifies its LPS structure in response to growth temperature. We purified His6-Pla under denaturing conditions and compared various LPS types for their capacity to enhance plasmin formation by His6-Pla solubilized in detergent. Reactivation of His6-Pla was higher with Y. pestis LPSs isolated from bacteria grown at 37°C than with LPSs from cells grown at 25°C. Lack of O antigens and the presence of the outer core region as well as a lowered level of acylation in LPS were found to enhance the Pla-LPS interaction. Genetic substitution of arginine 138, which is part of a three-dimensional protein motif for binding to lipid A phosphates, decreased both the enzymatic activity of His6-Pla and the amount of Pla in Y. pestis cells, suggesting the importance of the Pla-lipid A phosphate interaction. The temperature-induced changes in LPS are known to help Y. pestis to avoid innate immune responses, and our results strongly suggest that they also potentiate Pla-mediated proteolysis.
Plague is a zoonosis that primarily affects rodents but also has caused three waves of plague pandemics in the human population, leading to tens of millions of deaths (51). The etiological agent of plague, Yersinia pestis, is transmitted to humans primarily by the bite of infected fleas. The common form of the disease is bubonic plague, where the bacteria spread from the intradermal fleabite site into the lymphatic system, multiply in lymph nodes, and cause formation of bubos (51, 71). Bubonic plague may progress to systemic infection and reach the lungs, causing pneumonic plague. A high level of bacteremia is essential for the blood feeding by fleas and the efficient transmission of Y. pestis from host to host (51).
The Pla surface protease of Y. pestis is encoded by the Y. pestis-specific plasmid pPCP1 (3, 64, 65). Loss of pPCP1 or specific deletion of the pla gene causes a million-fold attenuation of Y. pestis in subcutaneously infected mice (8, 63). It was recently shown that development of bubonic plague depends on Pla (59), and in keeping with its central role, pla was identified as one of the most highly expressed genes in Y. pestis isolated from bubos of infected rats (60). Pla is also central for progression of pneumonic plague and multiplication of Y. pestis in the lungs (45). In contrast, Pla is not critical in colonization of the flea by Y. pestis (25) or in establishment of primary septicemic plague (48, 59), where bacteria are injected directly into dermal blood vessels. Some Pla-negative Y. pestis strains that are virulent even in the case of subcutaneous infection have been isolated from natural plague foci as well as being generated by elimination of pPCP1 (2).
Pla is an efficient plasminogen (Plg) activator and also destroys the main circulating control system for plasmin, the antiprotease α2-antiplasmin (41, 64), which leads to uncontrolled plasmin activity. Plasmin is a powerful serine protease with several important physiological functions, such as degradation of fibrin clots (fibrinolysis), proteolytic activation of procollagenases and progelatinases, and degradation of laminin, a major glycoprotein of basement membranes (44, 46, 50). In accordance with the important role of the Plg system in plague pathogenesis in mammals, bubonic and pneumonic plague are associated with enhanced fibrinolysis (14, 45), and Plg-deficient mice show a 100-fold increase in resistance to Y. pestis infection compared to normal mice (21).
Pla is a β-barrel transmembrane protein and belongs to the omptin family of surface-expressed aspartic proteases of Gram-negative bacteria (36, 40, 68). The omptin barrel is formed by 10 antiparallel β-strands and five surface-exposed loops (68), and the omptins share the same predicted protein fold, the predicted mature molecular size of approximately 292 amino acids, a conserved catalytic site formed by a His-Asp dyad and an Asp-Asp couple, and 50 to 70% overall sequence identity of the mature polypeptides (36, 41, 43, 68). A further conserved feature is the presence of a three-dimensional protein motif for binding to lipid A phosphates (16, 68), which is spatially positioned on the outer surface of the omptin β-barrel immediately above the outer leaflet of the outer membrane. The motif is made up by basic amino acids and shared by a number of prokaryotic and eukaryotic proteins which bind to lipopolysaccharide (LPS) (16); these include the Escherichia coli porin FhuA (17) and the human MD-2 protein that presents LPS to Toll-like receptor 4 (TLR4) (70). Indeed, the presence of LPS is needed to reactivate enzymatic function of the purified omptins OmpT of E. coli (6, 37, 38) and Pla of Y. pestis (39) reconstituted in detergent micelles, strongly suggesting that interaction with LPS is important for omptin functions in general. Further, the modification of LPS structure inside the mouse macrophage was recently found important for the high degree of in vivo activity of the PgtE omptin of Salmonella enterica (42).
Y. pestis has rough LPS and lacks genes encoding synthesis of the O antigen present in the smooth LPS form (31, 32, 54, 55, 62, 69), whereas the other human pathogenic yersiniae, Yersinia pseudotuberculosis and Yersinia enterocolitica, express O antigens that are important for bacterial colonization in infected mice (61). Pla-mediated proteolysis is sterically inhibited by the O antigen, and one of the selective advantages of the loss of the O antigen in Y. pestis is the ability to express high Pla-mediated proteolytic activity (39).
The transmission of Y. pestis from the flea to the mammal host involves a change of growth temperature from approximately 20 to 25°C to 37°C. The adaptation to higher growth temperature is associated with modification of LPS into a tetra-acylated form which has low immunostimulatory activity (15, 29, 30, 55). Biological activities of LPSs are determined by the shape of their hydrophobic lipid A portions, which, on the other hand, are strongly dependent on the acylation and on the charge of endotoxin (57, 58). A Y. pestis strain genetically modified to express hexa-acylated lipid A only was avirulent in a mouse bubonic plague model (48), indicating that evasion of the LPS-induced inflammatory response is critical for virulence of Y. pestis. Temperature also affects other structural properties of LPS, i.e., the level of substitution in lipid A phosphates and the fine structure of the core oligosaccharide (15, 30, 31, 32). Cavanaugh (11) and McDonough and Falkow (47) reported that the fibrinolytic activity of Y. pestis is dependent on bacterial growth temperature and is enhanced in cells grown at 37°C. Their studies did not address the amount of the Pla protein in Y. pestis cells grown at different temperatures, whereas recent proteomic studies have indicated approximately a 2-fold increase in the Pla protein content in cells grown at 37°C (12, 52). The reason for the growth temperature-dependent fibrinolytic activity of Y. pestis has remained open; conformational changes in the Pla protein have been suggested but not documented (47). As omptin functions are critically dependent on LPS (7, 37, 38, 39), we hypothesized and confirmed here that the temperature-induced changes in Y. pestis LPS modulate activity of Pla and thus indirectly affect proteolysis by Y. pestis.
The bacterial strains and plasmids used in this study are listed in Table Table1.1. For His6-Pla purification, the strain E. coli BL21(pREP4) His6-Pla (41) and its derivatives were first cultivated for 18 h at 37°C under shaking in Luria broth supplemented with ampicillin (100 μg/ml) and kanamycin (50 μg/ml). To induce expression of the His6-Pla proteins, the culture was inoculated 1/10 into prewarmed (37°C) Luria broth supplemented with ampicillin and kanamycin, and the suspension was incubated for 1 h at 37°C. Isopropyl thiogalactoside (IPTG; Promega) was then added at a 1 mM concentration, and the culture was incubated for 5 h at 37°C. For extraction of smooth LPS from the Y. pseudotuberculosis strain PB1 (53), the bacteria were grown at 20°C; the medium was supplemented with kanamycin in the case of the rough mutant Y. pseudotuberculosis PB1 Δwb strain (62). After cultivation, the Y. pseudotuberculosis cells were washed and lyophilized for LPS isolation. For the assays Y. pestis strains were cultivated overnight at 20°C on brain heart infusion (BHI) agar plates supplemented with hemin (40 μg/ml). Ten milliliters of BHI-hemin broth was inoculated with the bacteria, and cultivation was for 17 h at 20°C or for 40 h at 37°C. To allow the Y. pestis cells to adapt to the temperatures, the cultivation at 20°C or at 37°C was done twice. Recombinant E. coli were cultivated for 17 h in Luria broth supplemented with glucose (0.2%, wt/vol), ampicillin (100 μg/ml), and tetracycline (12.5 μg/ml). Recombinant Y. pestis strains were cultivated for 40 h at 37°C in 10 ml of BHI-hemin broth supplemented with glucose (0.2%, wt/vol) and ampicillin (100 μg/ml). To obtain high levels of expression of the recombinant proteins, the bacteria were then collected and suspended in 150 μl of phosphate-buffered saline (PBS), pH 7.1, and 100 μl of the suspension was plated on Luria or BHI-hemin agar plates containing 5 μM IPTG and antibiotics. For the assay bacteria were collected in PBS to a cell density 109 or 2 × 109 cells/ml (41).
The structures of the LPSs isolated from Y. pestis strains KM218, EV11M, and 1146 have been determined previously (15, 31, 69). The Y. pestis LPSs had been extracted from dried cells with phenol-chloroform-light petroleum ether (19), enzymatically digested first with nucleases and then with proteases, and further purified by ultracentrifugation. The E. coli LPSs used in this study, i.e., K-12 LPS (26, 49), as well as Re LPS and lipid A, were available from previous studies (7, 58). The hexa-acyl and penta-acyl Re LPS and lipid A from E. coli have been described previously (57), and the smooth O8 LPS was isolated from E. coli O8:K27 (27). The LPS from Y. pseudotuberculosis O1b strain PB1 and its isogenic rough Δwb derivative were purified in a manner similar to that used for the Y. pestis LPSs (19) and visualized by SDS-PAGE, and their 3-deoxy-d-manno-oct-2-ulosonic acid (Kdo) contents were chemically determined (28).
The His6-Pla proteins were purified from E. coli BL21(pREP4) carrying the pla derivatives, using denaturing conditions according to the manufacturer's protocol (Qiagen, Valencia, CA). The reconstitution of His6-Pla proteins was based on the previously described procedures for OmpT and His6-Pla purification and reactivation (38, 41). In brief, the His6-Pla protein in 8 M urea buffer was extensively dialyzed against decreasing concentrations of urea (starting from 4 M) in 20 mM HEPES buffer, pH 7.0, in the presence of 1 mM detergent n-dodecyl-N,N-dimethyl-3-ammonio-1-propanesulphonate (DodMe2NprSO3; Sigma). In the final step, the protein was dialyzed against 20 mM HEPES alone. The His6-Pla proteins were stored at 4°C and used within 30 days. In activation assays, 200 pmol of His6-Pla was incubated with an equimolar amount of purified LPS for 1 h at 37°C in a rotary shaker (1,000 rpm; Thermomixer Confort, Eppendorf, Germany) and used immediately for the assays.
Plg activation by 8 × 107 bacteria was performed as detailed earlier (39, 41). In reactivation of His6-Pla proteins, 200 pmol of the His6-Pla protein and 4 μg of human Glu-Plg were mixed in a final volume of 170 μl of HEPES buffer; 30 μl of the chromogenic plasmin substrate S-2251 was added, and degradation of the substrate was measured at 405 nm in a microtiter plate reader (Labsystem Multiskan, Helsinki, Finland). Each assay was performed in triplicates, and in each case at least three recombinant protein preparations were tested.
Substitutions in the LPS-binding motif of Pla were constructed by a recombinant PCR amplification procedure (24) as previously described for substitutions in the pla gene (41). The internal primer pairs contained the desired mutation sequence, and the external primer pair was used for amplification of the entire pla gene. Single- and two-residue substitutions to encode glutamic acid were made in the DNA encoding arginines R138 and R171 (17, 39, 66). The internal primers were designed on the basis of the pla sequence (accession numbers X15136, AF053945, and AL109969) and were CTGAACTCTGTTTCCTGATATCC (RevR138E), ATCAGGAAACAGAGTTCAGTTGG (ForR138E), CATAGAAAACTCCTGGTTATAACC (RevR171E), and GTTATAACCAGGAGTTTTCTATGC (ForR171E). The pla derivatives were expressed in the inducible pSE380 vector plasmid in either E. coli XL1 or Y. pestis KIM D34 (39, 41). Their correct nucleotide sequences were ascertained by nucleotide sequencing. The relative amounts of different Pla forms (α, β, and γ) produced by the His6-Pla substitution constructs or by wild-type Pla encoded on plasmid pPCP1 were assessed by Western blotting as described previously using whole cells of Y. pestis (41).
Semiquantitative analyses have earlier shown that fibrinolytic activity of Y. pestis is increased in cells grown at temperatures above 30°C (11, 47). On the other hand, proteomic studies have reported an approximately 2-fold increase in the amount of Pla isoforms when the Y. pestis cells are shifted from 26°C to 37°C (12, 52), but the two assay lines have not been correlated. We therefore compared plasminogen activation as well as Pla expression by Y. pestis KIM D27 grown at 20°C or at 37°C (Fig. (Fig.1).1). The plasminogen activation by the two cultures was first evaluated on the basis of equal bacterial mass and then, after cellular Pla content in the cells was assessed, on the basis of equal amounts of the Pla protein in the assay. The plasmin formation by KIM D27 was dramatically higher in cells from 37°C cultures. The amount of Pla protein in cells from 37°C was 1.7 times higher than that in cells from 20°C cultures (Fig. (Fig.1,1, inset), and the difference was more pronounced in the amount of the β-Pla isoform, which is autocleaved at residue K262 (41). The amount of α-Pla, which is the mature secreted form of Pla (65), showed a smaller increase in cells grown at higher temperatures (Fig. (Fig.1).1). When the number of KIM D27 cells from 20°C was corrected for the amount of the Pla protein, only a marginal increase in the plasminogen activation was detected (Fig. (Fig.1).1). We therefore concluded that the higher Pla enzymatic activity in cells from 37°C cultures could not be explained by their higher content of the Pla protein.
Purified, denatured omptins solubilized in detergent can be reactivated by exogenous LPS (37, 39). Major temperature-dependent structural changes in the LPS of Y. pestis are seen in the level of acylation as well as in the 4-amino-4-deoxy-l-arabinose (Ara4N) content in lipid A, and minor changes are seen in the core carbohydrates (15, 29, 31, 32) (Table (Table2).2). We compared reactivation of His6-Pla by several LPS species which differ in these aspects. The structures of the E. coli and Y. pestis LPSs used here have been determined previously (15, 26, 31, 49, 69) (Table (Table2).2). Natural LPS molecules are intrinsically heterogeneous in structure and molecular mass due to, e.g., different oligosaccharide chain lengths and the acylation and substitution levels. The molar amounts of LPSs in this study were calculated using the median mass of major components in the preparations (15, 26, 31, 49, 69); also, we measured their Kdo contents to be certain about the correct molar ratios.
To directly analyze the effect of LPS modification on Pla proteolysis, we first reactivated purified, denatured His6-Pla contained in detergent micelles with six LPS preparations isolated from three Y. pestis strains grown at 25°C or at 37°C. These LPSs have been characterized by chemical methods, high-resolution mass spectrometry, and nuclear magnetic resonance (NMR) spectroscopy (15, 31, 69) (Table (Table2).2). We used a molar ratio of 1:1 (200 pmol of His6-Pla and 200 pmol of LPS) after initially testing various LPS/His6-Pla ratios. The 1:1 ratio gave a marked effect on plasmin formation, and the amount of added LPS did not inhibit plasmin activity (data not shown). A low level of plasmin was formed by the His6-Pla protein without exogenous LPS, whereas addition of Y. pestis LPSs increased proteolysis (Fig. (Fig.2).2). In each case, the effect was higher with the LPS isolated from Y. pestis cells grown at 37°C; the individual Y. pestis LPSs, however, showed a strain-dependent effect in the reactivation. Using Western blotting of His6-Pla-treated plasminogen, we confirmed that no cleavage of the plasminogen molecule took place in the absence of exogenous LPS (data not shown). Also, no plasminogen activation was detected in the presence of LPS alone or in assays with the catalytic-site mutant His6-Pla(D206A) (His6-Pla with the mutation D206A) (41) (data not shown). We concluded that reactivation of His6-Pla by purified Y. pestis LPSs showed a similar temperature dependence as did the Pla enzyme activity in Y. pestis cells.
We have earlier reported that (i) Pla is nearly inactive when expressed in the smooth strain of Y. pseudotuberculosis O1b, from which serotype Y. pestis probably has evolved (62), and (ii) Pla is active when expressed in the Δwb rough derivative of Y. pseudotuberculosis O1b (39), which has an LPS with a full core region similar to that of Y. pestis (62). A similar dependence on LPS structure was observed in reactivation of His6-Pla by the LPSs from the two Y. pseudotuberculosis strains: the protein was successfully reactivated by the O1b Δwb LPS but remained poorly active with the O1b smooth LPS (Fig. (Fig.3).3). A low level of reactivation was also observed with the Y. pestis EV11M LPS (Fig. (Fig.3),3), which is a deep rough LPS mutant with one Kdo and one Ko (d-glycero-d-talo-oct-2-ulosonic acid) residue attached to lipid A (Table (Table22).
We also tested rough and smooth LPSs from E. coli in enhancement of His6-Pla-mediated plasmin formation. The rough (Ra) E. coli K-12 LPS was efficient in enhancement of Plg activation, whereas the smooth O8 LPS was only poorly active. However, a similarly low level of plasmin formation was seen with His6-Pla reconstituted with the deep rough Re LPS, and even lower activity was seen with lipid A (Fig. (Fig.3).3). These results confirmed the inhibitory effect of LPS O chains on Pla functions and suggested that a complete rough core is optimal for Pla activity.
An immunologically important temperature-induced change in the LPS of Y. pestis lowers the acylation level in lipid A from a mixture of tetra-, penta-, and hexa-acylated LPSs at 25°C into a mixture of tetra- and penta-acylated forms in cells grown at 37°C (15, 29, 31) (Table (Table2).2). We compared hexa-acylated and penta-acylated Re and lipid A preparations from E. coli in His6-Pla reactivation. The penta-acylated Re LPS gave a higher plasmin formation by His6-Pla than the hexa-acylated form, whereas no difference was observed with the two lipid A preparations (Fig. (Fig.4).4). The result showed that lower acylation in rough LPS is associated with higher proteolytic activity of His6-Pla.
E. coli lipid A molecules lacking either the 4′ phosphate or the glycosidic phosphate were available for testing and did not stimulate plasminogen activation by His6-Pla (data not shown). However, the enhancement of His6-Pla enzymatic activity by the intact lipid A preparation was also very low, and the results remained inconclusive. To gain more insight into the lipid A-Pla interaction, we disrupted the lipid A-binding motif in Pla. Arginines 138 and 171 are conserved and oriented outward on the surface of the omptin β-barrel (68) and form a part of the three-dimensional motif for protein binding to lipid A phosphates (16). We replaced R138 and R171 in His6-Pla with glutamates individually as well as in combination. The substitution R171E caused a reduction in the reactivation of His6-Pla by Y. pestis KM218 LPS, and the substitutions R138E and R138E R171E nearly completely abolished the enhancement of plasmin formation by His6-Pla (Fig. (Fig.5A5A).
We next expressed wild-type and mutated pla genes in the vector plasmid pSE380 using recombinant Y. pestis KIM D34 (Fig. (Fig.5B)5B) and E. coli XL1 (Fig. (Fig.5C)5C) as host bacteria. In both bacterial species, Pla with the mutation R138E [Pla(R138E)] and Pla with the substitutions R138E R171E [Pla(R138E/R171E)] were diminished in plasmin formation, whereas no effect by the R171E substitution was seen in Y. pestis KIM D34, and in E. coli this substitution caused a partial reduction.
We also assessed by Western blotting the amount of the Pla proteins in Y. pestis KIM D34 cells. The substitutions R138E and R138E R171E caused a reduction in the amount of Pla isoforms in KIM D34 cells (Fig. (Fig.5D).5D). The reduction mainly involved the amount of the unprocessed pre-Pla, and the amount of β-Pla was reduced by 50%. The γ-Pla isoform was not detected in the R138E mutant bacteria; it represents mature Pla that is folded differentially from α-Pla (38, 41). These results showed that disruption of the lipid A-binding motif affected both the reactivation of His6-Pla as well as the amount of Pla present in Y. pestis cells.
Numerous studies have reported that the temperature-induced modification of LPS in Y. pestis renders the bacteria less immunostimulatory and makes it a weak inducer of TLR4-mediated innate immunity responses (15, 29, 30, 48, 55). We report here that these LPS modifications also modulate activity of the Pla protease. We confirmed the early findings by McDonough and Falkow (47) and by Cavanaugh (11) that plasmin formation by Y. pestis is considerably higher in cells from cultures grown at 37°C than in cells grown at 20 to 25°C. Another indication of the higher proteolytic activity of Pla was the increased autoprocessing, i.e., formation of β-Pla through self-cleavage at K262 (41), which was observed in cells grown at 37°C. We also observed that the cellular content of the Pla protein isoforms was only marginally increased at 37°C, suggesting that the increase in plasminogen activation by Pla involves other factors. Involvement of LPS alterations is suggested by our findings that reconstitution of the proteolytic activity of His6-Pla was more efficient with Y. pestis KM218, KM260, and 1146 LPSs isolated from cells cultured at 37°C than with LPSs of the same strains cultured at 25°C. The correlation between the modification of LPS structure and the cellular content and activity of Pla, as well as reactivation of His6-Pla by different LPSs, indicates that Y. pestis tunes its proteolytic activity via LPS modification.
The amount of Pla encoded on plasmid pPCP1 in Y. pestis KIM D27 was 1.7 times higher in cells grown at 37°C than in cells grown at 20°C. This difference is very close to the approximately 2- to 4-fold increases in the Pla protein amount reported by recent analyses of the Y. pestis proteome from different growth temperatures (12, 52). Also, high temperature causes a 2-fold increase in transcription of the pla gene (23). We found that the increase in the proteolytic activity of Pla at 37°C by far exceeded the increase in Pla expression, and therefore the increase in Pla protease activity at the higher temperature involves additional factors. Our results indicate that LPS alterations are a factor in cell surface-associated plasminogen activation but do not rule out a role for other surface changes. For example, expression of the F1 capsular antigen was recently found to decrease Pla-mediated cleavage of the antimicrobial peptide LL-37 but not to interfere with the activity of Pla on Plg (20). The conclusion was that the inhibitory effect of F1 on Pla is substrate specific.
The main temperature-induced changes in Y. pestis LPS are the lowering of the acylation and the Ara4N substitution levels in lipid A (15, 29, 30, 31, 55). We present here evidence that these LPS features affect the Plg activation by Pla. The lack of hexa-acylated LPS in Y. pestis cells from 37°C cultures results from downregulation of the acyltransferase genes that add C12 and C16:1 groups to the lipid A precursor (55). The LPS preparations from the three Y. pestis strains showed better reactivation of His6-Pla with LPSs from 37°C cultures, and also the penta-acylated Re LPS was more potent than the hexa-acylated Re LPS in reactivation of purified His6-Pla.
The penta- and tetra-acylated rough LPS molecules adopt a lamellar (cylindrical) structure and are associated with weak immunostimulatory activity, whereas the hexa-acylated rough LPS has a conical (concave) structure associated with strong endotoxic activity (13, 57). Structure and charge of LPS molecules also affect their aggregation and intercalation into phospholipid bilayers (57, 58). We found an increase in the amount of processed, i.e., secreted, Pla α- and β-isoforms in Y. pestis cells from 37°C cultures. This may result from an easier incorporation of Pla into the outer membrane that is more fluid and more spacious due to reduced LPS-LPS interactions (4). The increased fluidity of the outer membrane lipid bilayer at 37°C is known to increase outer membrane-associated functions in yersiniae (4, 5). This might also explain the increase in the amount of the β-Pla, which is produced by autocleavage. OmpT was crystallized as a monomer (68), and a possible polymerization of omptins has not been documented. It seems likely that the enhanced formation of β-Pla at 37°C results from increased Pla-Pla contacts, which are due to higher incorporation and mobility of Pla isoforms in the more fluid outer membrane of Y. pestis. We observed formation of γ-Pla in Y. pestis cells that were overexpressing the pla gene, whereas the γ isoform was not detectable when pla was expressed under its natural promoter in the pPCP1 plasmid. The γ-Pla is an isoform differentially folded from α-Pla (22, 38), and the biological relevance of its formation has not been studied.
Montminy and coworkers (48) expressed the lpxL acyltransferase gene from E. coli in Y. pestis and found that the resulting recombinant strain makes hexa-acylated and more immunostimulatory lipid A. In their study the change in lipid A acylation did not change the activity of Pla-mediated Plg activation by Y. pestis cells. Here, while we found that hexa- and penta-acylated lipid A molecules did not differ in promoting His6-Pla activity, we detected a considerable difference between penta- and hexa-acylated Re LPSs. This result indicates that the effect of acylation change in lipid A is surpassed by the lack of a core oligosaccharide chain that is present in the Y. pestis, Y. pseudotuberculosis O1b Δwb, and the E. coli K-12 (Ra) LPSs, which all were highly efficient in reactivation of His6-Pla. In contrast, the deep rough LPSs of Y. pestis EV11M as well as the Re LPS and lipid A of E. coli were poor in reactivation of His6-Pla. This suggests that the presence of a complete core but not its precise carbohydrate structure is critical for Pla activity. The importance of the full LPS core is further supported by our recent analysis of Pla activity in an isogenic set of Y. pestis LPS mutants (1).
The LPS of the Y. pestis strain 1146 that is naturally deficient in pPCP1, as also are other representatives of Y. pestis subsp. caucasica (2), was a poor activator of His6-Pla. This LPS differed from KM218 and KM260 LPSs in that its Ara4N substitution at lipid A phosphate groups remains high irrespective of the growth temperature, and, conversely, the amount of unsubstituted phosphate groups remains low (15, 31) (Table (Table2).2). The poor activity of the 1146 LPSs suggests that unsubstituted lipid A phosphates have a role in the Pla-LPS interaction. We approached this question by disrupting the lipid A-binding motif in Pla; this motif had been identified in omptins by sequence homology to the lipid A-binding site in the FhuA outer membrane protein (16, 17, 68) and occurs in eukaryotic LPS-binding proteins as well (16, 17). The mutations were constructed in the arginine residues at positions 138 and 171, which are located in β-strands 5 and 6 and are spatially close to each other at the correct height of the omptin barrel to bind lipid A phosphates (68). The mutated proteins Pla(R138E) and Pla(R138E/R171E) were almost completely inactive in reactivating His6-Pla, whereas the mutation R171E had only a minor effect. This indicates that the Arg residue at 138 is involved in the Pla-LPS interaction; this is slightly different from the MD-2 protein-LPS interaction, where both Arg residues in the motif influence LPS binding and TLR4 activation (70). When the Pla proteins were overexpressed in bacteria, the Plg activation was lower with the Pla(R138E) and Pla(R138E/R171E) proteins; the effect was more pronounced in the E. coli background than in Y. pestis. Also, compared to Pla, the cellular contents of Pla(R138E) and Pla(R138E/R171E) in Y. pestis were reduced. The reduction in the Pla protein mostly concerned the amount of the nonprocessed, intracellular pre-Pla, which seems to be destabilized by the R138E substitution. The smaller amounts of the cell wall-associated α-, β-, and γ-isoforms of Pla(R138E) and Pla(R138E/R171E) may result from instability of the preform and are in line with the observation that LPS assists the folding and membrane insertion of the outer membrane β-barrel protein OmpA (9). OmpA contains 8 while the omptins contain 10 β-strands in their transmembrane β-barrels. While it is becoming evident that the folding and insertion into the outer membrane are varied with different outer membrane proteins, a recent study identified similar behaviors of OmpT and OmpA in folding to bilayers of both natural and synthetic lipids (10).
Exactly how LPS binding enhances activity and expression of Pla and other omptins remains to be defined. Kramer and coworkers (37) did not detect by spectroscopy any gross conformational changes in OmpT of E. coli upon LPS binding, and overall, the β-barrel is a sturdy structure that tolerates large deletions and substitutions at loop structures (33, 34). The binding of LPS to omptins in the outer membrane probably has a minor conformational effect that stabilizes and orients the surface loops, in particular, the large L3 between β-strands 5 and 6, for correct recognition of polypeptide substrates (22, 37). The temperature-induced LPS modifications enhance virulence of Y. pestis by dampening the TLR4-mediated innate immune response to the bacteria (48). We show here that another virulence-associated consequence of LPS modification by Y. pestis at the mammalian temperature is to enhance Plg activation by the Pla protease.
We have been supported by the Academy of Finland (The Microbe and Man Research Programme and grant numbers 105824, 116507, and 211300) and the European Union Network of Excellence EuroPathogenomics program. A.P.A. and Y.A.K. were supported by the Russian Foundation for Basic Research (grant 06-04-49280).
Editor: J. B. Bliska
Published ahead of print on 5 April 2010.