INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Mesophilic aeromonads are increasingly being reported as important pathogens of humans and lower vertebrates, including amphibians, reptiles, and fish (11). Infections produced by mesophilic aeromonads in humans can be classified into two major groups, i.e., noninvasive disease (such as gastroenteritis) and systemic illnesses (10). Aeromonas strains have been serogrouped on the basis of the O-antigen lipopolysaccharide (LPS) (31), particularly the polysaccharide chains in the smooth LPS, also known as the somatic antigen. Recently, a group of virulent Aeromonas hydrophila and Aeromonas veronii strains isolated from humans and fish have been described (12, 16), which are related serologically by their O-antigen lipopolysaccharide (serogroup O:11) with a known chemical structure (30) and which have a surface array protein with a molecular weight of ca. 52,000 (termed S-layer) (25, 29). The strains from this serogroup are the most common isolates from septicemia caused by mesophilic Aeromonas species (12). Serogroup O:34 strains of mesophilic Aeromonas species have been recovered from moribund fish or from clinical specimens (21, 24). O:34 is the single most common Aeromonas serogroup, accounting for 26.4% of all infections. Previous investigations have documented O:34 strains as an important cause of infections in humans (21, 24).

The complement system plays a key role in humoral defense against microbial pathogens and has extensively been reviewed (34). Its importance is clearly seen in individuals with complement deficiencies because they have a higher risk of developing severe and recurrent microbial infections (7). Therefore, resistance to complement action is a requisite for pathogenic microorganisms, which have developed a variety of mechanisms to ensure survival in nonimmune serum (7). Gram-negative bacteria activate complement via the classical or alternative pathway (CPC and APC, respectively), and more frequently, both pathways are required for the effective elimination of serum-sensitive strains (39). Activation of the CPC usually requires the presence of antibodies bound to bacterial antigens, whereas the APC is activated by certain bacterial polysaccharides by an antibody-independent mechanism (15).

In the present study, we focused on defining the mechanisms of complement sensitivity in this bacterium. Only the CPC is effective in the elimination of Aeromonas serum-sensitive strains in nonimmune serum as we previously reported (20, 22). Activation of the CPC by these strains was studied in more detail, and we have identified a bacterial outer membrane protein (OMP), presumably porin II (14), that binds C1q and activates this pathway in nonimmune serum and in agammaglobulinemic serum in an antibody-independent manner.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacteria. A. hydrophila strains from serogroups O:11 and O:34, as well as their derivative mutants, were previously described (20, 22). Mesophilic Aeromonas strains from different O serogroups were kindly provided by T. Shimada (31). Tryptic soy broth or agar was used as the regular medium for bacterial growth.

Human sera. A pool of nonimmune human sera (NHS) was obtained from healthy volunteers. NHS diluted 1/50 did not react with OMPs from Aeromonas strains of serogroups O:11 and O:34 in Western blot (immunoblot) experiments. NHS was made deficient in C1q as described previously (17). The C1q titer in C1q-deficient serum was 5.9 × 10⁵ hemolytically active molecules per µl; in NHS, it was 1.9 × 10⁹ hemolytically active molecules per µl. CPC activity was less than 1% in C1q-deficient serum, as measured by hemolytic assay detailed in previous work (1). Serum depleted of C1q and then reconstituted with purified C1q was obtained as previously described by us (3). Agammaglobulinemic serum was obtained as previously described by us (1) and also depleted of C1q and reconstituted with purified C1q as described above for NHS.

C1q purification and labeling. C1q was purified from NHS and tested for purity by polyacrylamide gel electrophoresis (PAGE) as previously described (1). Iodination of purified C1q was carried out with lactoperoxidase-glucose oxidase as described previously (35) and biotinylated with sulfosuccinimidyl 2-(biotinamido)ethyl-1,3-dithiopropionate (Pierce) at a molar ratio of 1:25 according to the manufacturer's procedure. Purified and labeled C1q (both iodinated and biotinylated) were hemolytically active and able to interact with immune complexes (5). The globular regions and the collagen-like fragment of C1q were isolated by the methods of Paques et al. (27) and Reid (30) respectively, as previously described by us (3).

Bacterial cell surface isolation and analysis. Bacterial cell envelopes, containing cytoplasmic and outer membranes, were obtained by French press cell lysis and centrifugation. OMPs were isolated as sodium lauryl sarcosinate-insoluble material (8). C1q-binding OMPs were isolated and identified by electrophoresis and Western blot analysis as previously described, using either purified biotinylated C1q or NHS with rabbit anti-C1q (1).

Binding of C1q to bacterial cells. Mid-logarithmic-phase bacterial cells were recovered by centrifugation, washed with phosphate-buffered saline (PBS), and incubated sequentially with biotin-labeled C1q (0.2 mg in PBS, 1 h) and colloidal gold-labeled avidin (Sigma; diluted 1 to 50 in PBS, 1 h). Washing steps with PBS were included after each incubation period. Controls (whole cells) to show the specificity of the C1q binding to the bacterial cells were treated only with the gold-labeled reagent. Cells were observed with a Hitachi H600 electron microscope at 75 kV. Binding was also studied with radiolabeled C1q as described previously (1).

Antisera. Antiserum against C1q was obtained as previously described (1). Antiserum against the purified C1q-binding OMP was obtained in New Zealand White rabbits by using purified protein as antigen and by the immunization procedure previously described (3).

Bacterial survival in human serum. Bacterial cells (10⁸ CFU) of the serum-sensitive strains in the logarithmic phase were suspended in PBS with 10% serum and incubated at 37°C. Viable bacterial counts were made at different times by dilution and plating.

Complement assays. Human sera were incubated for 45 min at 37°C with purified C1q-binding OMP or purified LPS from serum-sensitive strains. After incubation, 50% hemolytic complement (CH₅₀) (29) and C4 consumption (4) were measured.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Complement sensitivity of mesophilic Aeromonas. Killing of A. hydrophila serum-sensitive strains AH-53 and AH-26 was mediated by the CPC. An example of this type of experiment, performed with strain AH-53 (20), is shown in Fig. 1. Figure 1 shows that after incubation in NHS for 1 h, there is a decrease of 4 orders of magnitude in bacterial viability. Control incubations with heat-inactivated serum showed that complement was responsible for the loss of viability observed. To further assess the role of the CPC and APC, we incubated the bacteria in C1q- or factor B-deficient serum. After incubation in C1q-deficient serum, no loss of viability was observed, while a loss of viability similar to the one observed for complete NHS was observed in factor B-deficient serum. For a control, we also incubated bacteria in C1q-deficient serum reconstituted with purified C1q, and a decrease of 3 to 4 orders of magnitude in bacterial viability was observed after 1 h of incubation in this serum.

View larger version (16K): [in this window] [in a new window]   FIG. 1.   Survival of A. hydrophila AH-53 (serum sensitive [20]) in nonimmune serum (), heat-inactivated (30 min, 56°C) nonimmune serum (), C1q-deficient serum (), factor B-deficient serum (), and C1q-deficient serum reconstituted with purified C1q (). The results are the averages of at least three independent experiments (values are means ± standard deviations).

Binding of C1q to bacterial cells. Binding of C1q was quantitated for different strains by using ¹²⁵I-labeled C1q. As shown in Fig. 2, cells of the serum-sensitive strains bind C1q more efficiently than cells from the serum-resistant strains (wild types). No C1q binding was observed for wild-type strain TF7 (has O:11 and S-layer). Accessibility and binding of C1q to bacterial cells were studied by electron microscopy with biotin-labeled purified C1q and gold-labeled avidin. A representative example of this type of experiment is depicted in the insert of Fig. 2. Cells of the serum-sensitive strain AH-26 (22) treated with biotinylated C1q and colloidal gold-labeled avidin bound C1q, and C1q was visualized as gold spheres associated with the cells (panel A in the insert of Fig. 2). Control experiments with cells incubated with the gold-labeled reagent in the absence of biotinylated C1q (panel B in the insert of Fig. 2) or with cells from the serum-resistant strain TF7 (panel C in the insert of Fig. 2) treated as described for strain AH-26 in the legend to panel A of the Fig. 2 insert showed no gold spheres bound to these cells, i.e., biotinylated C1q was not bound by these cells.

View larger version (32K): [in this window] [in a new window]   FIG. 2.   Binding of 125I-labeled C1q to A. hydrophila whole cells as described in Materials and Methods of wild-type strains AH-3 () and TF7 () (serum resistant [20, 22]) and their LPS mutants AH-53 () and AH-26 () (serum sensitive [20, 22]), respectively. The results are the averages of at least three independent experiments (values are means ± standard deviations). (Insert) Binding of biotinylated C1q to A. hydrophila whole cells as described in Materials and Methods. Cells were incubated with biotinylated C1q and avidin-colloidal gold spheres (A and C) or with avidin-colloidal gold spheres alone (B). Cells in panels A and B correspond to the A. hydrophila LPS mutant AH-26 (serum sensitive [22]), and that in panel C corresponds to wild-type strain TF7 (serum resistant [22]). Bars, 0.4 µm.

Identification and isolation of bacterial surface molecules involved in C1q binding. Since strains AH-53 and AH-26 are devoid of LPS O antigen (O:34 and O:11, respectively) and AH-26 also has no S-layer, candidate surface molecules for C1q binding were OMPs and rough LPS. Figure 3A, lanes 1 and 2, shows the OMP profiles of strains AH-3 (wild type) and AH-53 after SDS-PAGE and Coomassie blue staining, and only a few proteins were visualized, as is typical in gram-negative OM. The binding of C1q to OMPs and identification of the protein(s) responsible for this binding were studied by Western blotting using biotinylated C1q as indicated in Materials and Methods. A single band with a molecular mass of approximately 39 kDa was strongly stained (Fig. 3B). Lane 4 of Fig. 3B, containing blotted bovine serum albumin (BSA), was treated in the manner described in the legend for lanes 1 and 2 to demonstrate the specificity of the analysis of C1q binding. The C1q-binding OMP was purified as described in Materials and Methods and is showed on Fig. 3A boiled and not boiled in sample buffer (lanes 3 and 4, respectively) to see heat modification and analyzed by Western blotting with purified biotinylated C1q for the C1q binding (Fig. 3B, lane 3). It was concluded that C1q binding to this OMP is antibody independent. Purified LPSs from these strains are shown in Fig. 3C, and the purified LPSs of serum-sensitive strains (rough LPS) were shown to be unable to bind C1q by the Western procedure (Fig. 3B, lanes 5 and 6).

View larger version (63K): [in this window] [in a new window]   FIG. 3.   Isolation of C1q-binding OMP from A. hydrophila AH-53 (serum sensitive [20]). (A) SDS-PAGE and Coomassie blue staining. Lanes: 0, molecular mass standards (in kilodaltons); 1, OMPs of wild-type strain AH-3 (O:34); 2, OMPs of strain AH-53; 3, the purified C1q-binding OMP (probably porin II) boiled in sample buffer for 10 min; 4, as in lane 3 without boiling. (B) Western blot analysis with biotinylated C1q of OMPs and LPS as described in Materials and Methods. Lanes: 0, molecular mass standards (20, 31, 43, 62, and 97 kDa, respectively); 1, 2, and 3, as in panel A; 4, 25 µg of purified BSA; 5 and 6, purified LPS from strains AH-53 and AH-26, respectively. (C) SDS-PAGE and silver staining of purified LPS from different strains (20, 22). Lanes: 1, AH-3 (O:34); 2, AH-53; 3, TF7 (O:11); 4, AH-26.

Contribution of cell surface components to C1q binding and CPC activation. Since the cell surfaces of mesophilic Aeromonas serum-sensitive strains are formed mainly by OMP and rough LPS, it was important to study the relative contribution of these components to the C1q-binding phenomenon. We have studied this point by dot blot analysis of the purified C1q-binding OM porin II isolated as described above and purified LPS from serum-sensitive strains (rough LPS). The results of this analysis are shown in Fig. 4. Row A contained twofold dilutions of the purified OM porin II (starting at 0.5 mg/ml), and rows B and C contained purified LPS from strain AH-53 (2 mg/ml) and BSA (as a control, 1 mg/ml), respectively. After incubation with biotinylated C1q and avidin-alkaline phosphatase, only the OM porin II containing dots (row A) were visualized. We conclude that C1q binding is due to OM porin II, and although C1q binding to rough LPS cannot be ruled out because rough LPS activates CPC (20, 22), it is at least 64-fold less efficient than the C1q binding to OM porin II. It is important to note that this binding was observed in an antibody-free experiment, as was the binding observed in Fig. 2 and 3. Also, in contrast to the proteins of SDS-PAGE and Western analysis, the proteins used in this dot blot assay had not been boiled, indicating that C1q binds to porin II in its native state. Similar results were obtained for strain AH-26.

View larger version (44K): [in this window] [in a new window]   FIG. 4.   Dot blot analysis with purified C1q labeled with biotin. Purified 39-kDa OMP (probably OM porin II) (row A) and rough LPS from strain AH-53 (row B) and BSA (row C) were dot blotted starting at 0.5 µg (row A), 2 µg (row B), and 1 µg (row C) (column 1) and then with twofold dilutions (columns 2 through 7). Binding of biotinylated C1q was detected with alkaline phosphatase-labeled avidin.

View larger version (27K): [in this window] [in a new window]   FIG. 5.   Activation of the CPC by the purified 39-kDa OMP (probably porin II). (A) NHS depleted of C1q and factor D and reconstituted with purified C1q was incubated with different amounts (0 to 15 µg) of the porin. After 45 min at 37°C, the remaining C4 (white bars) and CH50 (cross-hatched bars) hemolytic activities were determined as described in Materials and Methods. (B) NHS, NHS depleted of C1q and factor D and reconstituted with purified C1q (RQD+Q), and agammaglobulinemic (Agamma) serum were incubated for 45 min at 37°C with 15 µg of the porin. After incubation, the remaining C4 (white bars) and CH50 (cross-hatched bars) hemolytic activities were assayed as described above for panel A. All the results are the averages of at least three independent experiments (values are means ± standard deviations).

View larger version (19K): [in this window] [in a new window]   FIG. 6.   Antibody-independent activation of the CPC by OM components. An agammaglobulinemic serum depleted of C1q and factor D and reconstituted with purified C1q was incubated with different amounts of the purified 39-kDa OMP (probably porin II) () or purified LPS from serum-sensitive strains AH-26 () and AH-53 (). After 45 min at 37°C, the remaining CH50 was determined as described in Materials and Methods. The results are the averages of at least three independent experiments (values are means ± standard deviations).

View larger version (107K): [in this window] [in a new window]   FIG. 7.   OMPs obtained as described in Materials and Methods from different clinical mesophilic Aeromonas strains of O serogroups 2, 3, 6, 11, 12, 14, 16, 17, 18, 23, 29, 33, 34, and 43 (the most common around the world). (A) OMP profiles obtained by SDS-PAGE and Coomassie blue staining. Lanes: 0, molecular mass standards (14, 21, 30, 41, 66, and 92 kDa); 1, serogroup O:2; 2, serogroup O:3; 3, serogroup O:6; 4, serogroup O:11; 5, serogroup O:12; 6, serogroup O:14; 7, serogroup O:16; 8, serogroup O:17; 9, serogroup O:18; 10, serogroup O:23; 11, serogroup O:29; 12, serogroup O:33; 13, serogroup O:34; 14, serogroup O:43. (B) Western blot analysis of the same OMPs incubated with biotinylated C1q as described in Materials and Methods. Lanes 1 to 14 as in panel A.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The mechanisms that pathogenic bacteria use to evade the complement action have been extensively studied and reviewed (15, 34). We also previously published the mechanisms used for different mesophilic Aeromonas strains from serogroups O:11 (22) and O:34 (20). It is also important to define the bacterial molecules acting as targets for complement activation and deposition, since this probably represents the basis of the natural immunity to bacterial infections. Mesophilic Aeromonas rarely causes infections in healthy individuals, whereas it is a pathogen for immunodepressed patients. Most people, then, are successfully dealing with these strains through the action of the complement system, a fact which stresses the need for a better definition of the molecules involved in complement activation by this species.

We previously reported (20, 22) and reconfirmed here that only the CPC is involved in the killing of A. hydrophila O:11 and O:34 serum-sensitive strains, while the APC seems not to be involved, which indicated that these polysaccharides (O:11 and O:34 antigen LPSs) are nonactivating surfaces for the APC.

Enterobacterial porin proteins from different bacterial species, such as Salmonella typhimurium (9), Salmonella minnesota (33), or Klebsiella pneumoniae (1, 3), have also been shown to activate the CPC after C1q binding. However, it is the first time to our knowledge that a nonenteric porin (probably porin II of A. hydrophila) has been described to bind C1q and activate the CPC and furthermore in a bacterium able to activate only the CPC and not the APC. Furthermore, after C1q digestion with collagenase and pepsin, the globular and collagen-like C1q fragments were prepared, respectively (3, 27, 30). The preliminary results (data not shown) demonstrate that the interaction between C1q and the purified OM porin II is by the globular region of C1q and not by its collagen-like fragment, as with the OmpK36 porin of K. pneumoniae (3).

Binding of C1q to other OM components of these Aeromonas serum-sensitive strains, in particular the rough LPS, could not be showed. It now seems clear that lipid A is the only component of the rough LPS able to interact with C1q and in an antibody-independent manner activate the CPC (6). This binding and activation phenomenon has been shown to be restricted to chemotype Re and isolated lipid A (39), but other rough LPSs may not interact with C1q and therefore would not activate the CPC. We suggested that the rough LPS of our A. hydrophila serum-sensitive strains does not belong to chemotype Re and that lipid A is not accessible to C1q and is thus unable to activate the CPC by C1q binding, as with other rough LPSs as we described in K. pneumoniae (1).

A. hydrophila serum-sensitive strains (O:11 and O:34) are killed by complement via the CPC, and because the killing takes place without an apparent or significant C1q binding to the rough LPS, it seems clear that the 39-kDa OMP (probably OM porin II) plays a major role on the CPC activation. This fact is supported by the results obtained with the purified protein, which is more effective than the purified LPS in terms of C1q binding, C4 consumption, and the reduction of the total hemolytic activity of serum. The C1q binding to the purified protein (probably OM porin II) is an antibody-independent process. This fact has been proved by the ability of the purified protein (with LPS contamination of <10³) to bind purified C1q and its ability to activate the CPC in both nonimmune and agammaglobulinemic sera. The present results coincide with those reported by other researchers and ourselves for different bacterial OMPs (mainly porins) interacting with C1q as antibody-independent phenomenon (1, 3, 23). Furthermore, this phenomenon would be more important in the case of A. hydrophila strains, because this is the first case described in which this phenomenon takes place on a bacterium able to activate only the CPC.

Finally, we observed that this 39-kDa OMP (probably porin II) or a similar one is largely conserved among different mesophilic Aeromonas strains from the more common serogroups found in clinical samples, despite the variability in the OMP profiles described for the different strains of mesophilic Aeromonas species (18, 41). Furthermore, on Western blots, this OMP or a similar one is able to bind C1q, as shown in strains of serogroups O:11 and O:34. These results seem to indicate that the general way for the CPC activation by mesophilic Aeromonas strains may be to bind C1q to one OMP (probably porin II or a similar protein) in an antibody-independent manner. It is interesting that a porin II-deficient mini-Tn5 Km1 transposon mutant from strain AH-53, with a rough LPS devoid of the O:34 antigen LPS, is practically unable to activate complement (data not shown), which indicates the major role of porin II in CPC activation.