|Home | About | Journals | Submit | Contact Us | Français|
Recently, we isolated human IgG from normal human sera (NHS) using lipooligosaccharide (LOS) from gonococcal strain JW31R as an affinity ligand. We provided evidence that the oligosaccharide (OS) moiety of LOS was immunogenic in humans and that NHS contains functional antibodies that bind to the branched OS. The present study aimed to identify bactericidal antibodies that bind to partial core OS structures or their adjacent sites expressed in the 3,4-branched and 2,3:3,4-dibranched neisserial LOSs. Using 15253 LOS from serum-resistant gonococcal strain 15253 as an affinity ligand, we isolated IgG2 and found that this preparation contained at least three different species. (i) One IgG2 species recognized a cross-reactive epitope that is expressed on 3,4-branched and 2,3:3,4-dibranched neisserial LOSs. (ii) Another IgG2 species was specific for JW31R LOS from a pyocin-resistant gonococcal strain; this IgG-defined epitope was not shared with the aforementioned branched LOSs. (iii) The third IgG2 species bound to the “Salmonella minnesota” Rb and Re mutant lipopolysaccharides (LPSs); this IgG2 recognizes a KDOα2-4KDO residue at the reducing end of the carbohydrate moiety of each LPS. The IgG2 was also found to be functional and facilitated the killing of strain 15253. The current results show that neisserial LOS contains several epitopes within its OS moiety that are recognized by human antibodies.
Lipooligosaccharide (LOS) is a major antigenic and immunogenic component of pathogenic Neisseria species such as Neisseria meningitidis and N. gonorrhoeae (34). LOS produced by these bacteria consists of an oligosaccharide (OS) moiety and lipid A, and structural variation of the OS leads to LOS heterogeneity (34). As shown in Fig. Fig.1A,1A, Neisseria bacteria biosynthesize a core pentasaccharide, N-acetylglucosamine (GlcNAc); two heptose (Hep) and two 2-keto-3-deoxymannooctulosonic acid (KDO) residues are sequentially α-glycosidically linked. GlcNAc is linked to the 2 position of Hep[II], which is linked to Hep[I] via its 3 position. The latter Hep[I] is linked to the 5 position of the terminal KDO of KDO(α2-4)KDO. The structural variation of neisserial OS stems from how glycoses are further biosynthesized from the inner core disaccharide, Hep[II](α1-3)Hep[I], of the pentasaccharide (Fig. (Fig.1A).1A). Sequential addition of glycoses to the 4 position of the 3-substituted Hep[I] produces a 3,4-branched OS structure (Fig. (Fig.1B),1B), and addition of both to the same Hep[I]-4 position and to the 3 position of the 2-substituted Hep[II] produces a 2,3:3,4-dibranched OS (Fig. (Fig.1C)1C) (4, 16, 18, 25, 41, 42). Furthermore, functional groups such as phosphate and acetyl groups are expressed in the GlcNAc(α1-2)Hep[II] portion of the core (Fig. (Fig.1A);1A); O-phosphorylation takes place at the 3 and 6 positions of Hep[II] (6, 35), and O-acetylation may occur at the 3 position of GlcNAc (17).
Recently, we reported the isolation of human IgG from normal human sera (NHS) using a 2,3:3,4-dibranched LOS (Fig. (Fig.1C)1C) from serum-sensitive gonococcal strain JW31R as an affinity ligand (43). The purified IgG did not bind to any 3,4-branched neisserial LOSs examined and bound only to the ligand JW31R LOS and the truncated form of LOS from serum-resistant strain 15253. Both LOSs contain a disaccharide, Galβ1-4Glc (lactose), on Hep[I], and the structural difference between the two LOSs is the length of the OS linked to Hep[II]; the same disaccharide on Hep[I] is present on Hep[II] of 15253 LOS (41), whereas a tetrasaccharide, an elongated form of lactose, resides on JW31R LOS (5). We have determined that the IgG was specific for the OS moieties of the JW31R and 15253 LOSs but not their lipid A portions.
The strain 15253 LOS described above is recognized by murine monoclonal antibody (MAb) 2C7, and an OS epitope recognized by MAb 2C7 is frequently expressed in vitro and in vivo on the surface of the gonococci (10). This MAb requires the strain 15253 OS structure containing lactose on both Hep[I] and Hep[II] for binding, but it does not bind to JW31R LOS (42). Although this binding difference between MAb 2C7 and the above-mentioned anti-JW31R IgG exists, recognition of strain 15253 LOS by the latter IgG showed that a 2C7-like epitope is also recognized by human IgG. This human IgG was also found to be bactericidal against strain JW31R. These previous results provided evidence that the OS moiety of LOS is immunogenic in humans and also showed that NHS contains functional antibodies specific for the branched OS.
Other investigators have also reported that NHS contains bactericidal antibodies that recognize neisserial LOS (3, 15) and that the OS moiety is immunogenic in humans. However, very little is known about the OS structures of LOS that are recognized by human antibodies. Humans are the only natural host of the pathogenic neisseriae, and the molecular basis of the recognition specificity of human antibodies to LOS is important for us to understand the immune responses of the host to these bacteria.
In the present study, we aimed to isolate and characterize human IgG by using 15253 LOS, which contains the MAb 2C7 epitope (10, 42), as an affinity ligand. We anticipated identifying bactericidal antibodies that bind to partial core OS structures or their adjacent sites expressed in the 3,4-branched and 2,3:3,4-dibranched LOSs. We found that IgG2 isolated by affinity chromatography from NHS bound to the aforementioned branched LOSs and also to Salmonella mutant Rb and Re lipopolysaccharides (LPSs). The IgG2 was also found to be functional and able to facilitate the killing of serum-resistant strain 15253, which expresses the ligand LOS. The current results demonstrated that neisserial OS contains several epitopes that are recognized by human antibodies.
Gonococcal strains (JW31R, 15253, and MS11mkA) were provided by R. E. Mandrell (USDA/ARS, Albany, CA). S. Gulati (University of Massachusetts Medical School, Worcester) provided gonococcal strains PID-2, WG, 24-1, 302, and 220. Most of the above gonococcal strains have been studied for sensitivity to the bactericidal activity of NHS (10, 11, 37); 15253, 302, and 220 were designated serum resistant, and F62, 24-1, and JW31R were designated serum sensitive. Gonococci were cultured on a GC agar base (Difco Laboratories, Detroit, MI) containing 1% defined supplement in a CO2 incubator at 37°C (39, 43). We used the following neisserial LOSs that have been immunochemically and/or structurally characterized: 2,3:3,4-dibranched LOSs (JW31R [5, 42], 15253 , 15253 lgtE mutant , and WG ) and 3,4-branched LOSs (F62 , 24-1 , MS11mkA , PID-2 , 302 , and 220 ). LOS samples from proteinase K-digested (PK) cell lysates of gonococcal strains were prepared using the method of Hitchcock and Brown (13). The molecular weights of the 3,4-linked LOSs were estimated by using the values used for the six PID-2 LOS components as described previously (28). Escherichia coli LPS and the following Salmonella rough mutant LPSs were purchased from Sigma (St. Louis, MO): Ra (Salmonella enterica serovar Typhimurium TV119), Rc (“S. minnesota” R5), Rd (S. minnesota R7), and Re (S. minnesota Re 595). The Rb mutant LPS (S. minnesota R345) was from List Biological Laboratories, Inc. (Campbell, CA). 15253 LOS, MS11mkA LOS, and S. minnesota Re 595 LPS were each dephosphorylated with 48% hydrofluoric acid as described previously (44). WG LOS was sequentially treated with β-galactosidase and β-N-acetylhexosaminidase as described previously (42).
To purify anti-LOS antibodies, we used the same pooled sera used in our previous work (43). In addition, we used commercially available NHS (pooled male sera) and human IgG (Sigma Chemical Co., St. Louis, MO). Mouse anti-human IgG MAb (γ chain specific, alkaline phosphatase [AP] conjugate) was purchased from Sigma Chemical Co. Strain 15253 LOS was coupled to CNBr-activated Sepharose 4B (GE Healthcare, Japan) as described previously (43), and the coupling of 15253 LOS was confirmed by treating the resulting gel matrix sequentially with MAb 2C7 (10, 42), the anti-mouse IgG antibody (Ab; AP conjugate; Sigma Chemical Co.), and Western Blue-stabilized substrate (Promega Co., Madison, WI).
Anti-15253 LOS antibodies were purified as described previously (42). Briefly, the pooled human sera (8 ml) diluted in phosphate-buffered saline (PBS) were loaded onto the LOS affinity column (0.7 by 6 cm), and the proteins bound to the column were eluted with 0.1 M glycine/HCl (pH 2.7). The eluted fractions were neutralized (2 M Tris/HCl, pH 9.0), pooled, dialyzed, concentrated, and then passed through a column (0.7 by 6 cm) of inactivated Sepharose. The fractions not bound to the Sepharose column were pooled, concentrated, and then applied to a HiTrap Protein G column (GE Healthcare) to purify an anti-15253 IgG. NHS (from Sigma) was pretreated with 50% ammonium sulfate and then processed as described above, and human IgG (from Sigma) was passed through the LOS affinity column and then the inactivated matrix column. The presence of IgG during the chromatographic steps was determined by enzyme-linked immunosorbent assay (ELISA) using U16 Maxisorp plates (Nunc, Roskilde, Denmark) and a Multiskan MS reader (Labsystems, Helsinki, Finland) (43). Goat anti-human IgG (AP conjugate; Sigma Chemical Co.) and p-nitrophosphate were used as the probe and substrate, respectively. The binding of IgG to LOS and LPS samples was also analyzed by ELISA; U16 Maxisorp wells coated with 200 ng each of LOS, and LPS samples were blocked with 1% bovine serum albumin (BSA) in PBS and then sequentially treated with the IgG, the anti-human IgG, and the substrate as described above. The subclass of the purified IgG was determined with a Human Ig Subclass Typing kit (The Binding Site, Birmingham, United Kingdom), and the protein concentration was measured with a BCA Protein Assay kit (Pierce, Rockford, IL) as described previously (43).
Both Protean II and Protean III cells (Bio-Rad Laboratories, Hercules, CA) were used for gel electrophoresis. LOS and LPS samples separated by polyacrylamide gel electrophoresis (PAGE) were electroblotted onto nitrocellulose or polyvinylidene difluoride (PVDF) membranes (both from Bio-Rad) as described previously (28, 40, 43). Anti-human IgG (AP conjugate; Sigma Chemical Co.) and the Western Blue substrate were used for immunostaining. For chemiluminescence detection, the PVDF membrane was incubated sequentially with the affinity-purified IgG, anti-human IgG (1:25,000, peroxidase conjugate; Sigma Chemical Co.), and SuperSignal West Pico substrate (Pierce Biotechnology, Rockford, IL). Each incubation was performed for 1 h at room temperature unless otherwise stated. A model LAS-1000 apparatus (Fujifilm, Tokyo, Japan) was used for chemiluminescence detection.
Dot blot analysis of lactosylceramide (Wako Pure Chemicals, Osaka, Japan) was completed with glass fiber sheets impregnated with silica gel (Gelman Sciences, Ann Arbor, MI) as described previously (40, 43). LOS and LPS samples were dotted onto the nitrocellulose membrane, and the membrane was treated as described above for Western blotting. Partial acid hydrolysates of LOSs (15253 LOS and MS11mkA LOS) and LPS (Re mutant LPS) samples were prepared by treating each LOS or LPS with 1% acetic acid for 1 h at 100°C. Each hydrolysate was analyzed by thin-layer chromatography (TLC) immunostaining (44); three TLC plates (Silica gel G60; Merck, Darmstadt, West Germany) were developed using a solvent system of chloroform-methanol-water-ammonium hydroxide (50:25:4:2, vol/vol/vol/vol). One plate was sprayed with methanolic H2SO4 and then heated at 120°C for 5 min in an oven. Two plates were coated with polymethacrylate and then blocked with 1% casein in PBS. One plate was sequentially treated with the affinity-purified IgG and the anti-human IgG, and the other was treated with an anti-lipid A Ab (Sanbio, Uden, Netherlands) and anti-mouse IgG (AP conjugate; Sigma Chemical Co.). Both plates were stained with the Western Blue substrate.
For Western blot analysis of IgG preabsorbed with LOS or LPS, we used the IgG prepared from NHS (from Sigma) as described earlier, and the concentration of the Ab obtained after sequential affinity chromatography was ~20 μg/ml. The above IgG was diluted to 1:20 or 1:50, and each diluted IgG was examined for binding to each (10 ng) of the following neisserial LOSs through chemiluminescence detection: PID-2 LOS, MS11mkA LOS, JW31R LOS, and 15253 LOS. To quantify the IgG and LOS or LPS, molecular masses of 150,000 and 4,000 Da were used, respectively. Because the concentration of the 1:50 dilution of IgG (400 ng/ml) was inadequate to clearly visualize its binding to each LOS, we employed the 1:20 dilution of IgG (1 μg/ml). Preliminary absorption analysis of the IgG was performed using MS11mkA LOS. U16 microtiter wells were coated with 100 ng (25 pmol, ~19 molar equivalents) or 400 ng (~100 pmol, ~75 molar equivalents) of the antigen. After blocking with 1% BSA in PBS containing 10 mM MgCl2 for 1 h at room temperature, each coated well was incubated with the IgG2 (200 μl = 200 ng [~1.33 pmol]/well) for 1 h at room temperature. The two absorbed IgG preparations were then used for blot analysis with the following antigens (10 ng of each): PID-II LOS, MS11mkA LOS, JW31R LOS, and 15253 LOS. Each PVDF membrane was incubated with each of the preabsorbed IgG samples overnight at 4°C and then treated as described earlier for chemiluminescence detection. The 1:20 dilution of IgG was also absorbed with 15253 LOS, MS11mkA LOS, and Re 595 LPS (1 μg, ~250 pmol, ~188 molar equivalents) as described above, and each preabsorbed IgG was used for blot analysis with 10 ng of PID-II LOS, MS11mkA LOS, JW31R LOS, 15253 LOS, Rb LPS (S. minnesota), and Re 595 LPS (S. minnesota) as described above.
Bactericidal activities of the IgG against the JW31R and 15253 strains were measured using the method of McQuillen et al. (24) with the following modifications. The amount of test serum, bacterial suspension, and Gey's balanced salt solution (GBSS) containing 0.5% BSA was 50 μl (37) instead of 25 μl. Each bacterial suspension was prepared by diluting bacteria at a mid-log-phase concentration (optical densities at 650 nm, 0.215 and 0.203 for strains JW31R and 15253, respectively) with GBSS/BSA. We used the same pooled human sera that had been collected in a previous study (43) and stored at −82°C. The complement source was prepared by rotating the above sera (5 ml) with either glutaraldehyde-fixed JW31R or 15253 for 1 h at 4°C. After removal of the bacteria by centrifugation, the absorbed serum was filtered through a 0.22-μm filter. Aliquots of each of the absorbed complements were stored at −82°C and thawed immediately before use in each experiment. For control experiments, the NHS and complement source were inactivated by incubation at 56°C for 30 min. The bactericidal reaction mixtures were prepared by sequentially adding 50 μl of the IgG, untreated NHS, or heat-inactivated NHS, 25 μl of each bacterial suspension, and 50 μl of either the complement source or GBSS/BSA. The mixtures were incubated at 37°C for 45 min in a shaking water bath. Samples (50 μl) were removed from each bactericidal reaction mixture and spread onto GC agar plates. Colonies produced by surviving organisms were counted after incubation of the plates in a CO2 incubator overnight at 37°C. Bactericidal activity was expressed as follows: [1 − (CFU in the test sample with intact complement/CFU in the buffer control)] × 100 CFU. The average of duplicate plates was used in the analysis.
We prepared an affinity column using 15253 LOS in a manner similar to that described previously (43) and confirmed that the LOS was coupled to the column matrix using MAb 2C7, which recognizes the ligand LOS (42). A typical purification procedure is shown in Fig. Fig.2A.2A. Three columns, 15253 LOS, HiTrap protein G, and inactivated Sepharose matrix, were used sequentially for the isolation of anti-15253 LOS IgG. The ligand 15253 OS has Galβ1-4Glc (lactose) on both Hep[I] and Hep[II], and this disaccharide is, respectively, β1-4 and α1-3 linked to Hep[I] and Hep[II] (Fig. (Fig.2B)2B) (41). As shown in lane 1 of Fig. Fig.2C,2C, the fraction bound to the LOS column contained large amounts of serum albumin and other proteins, and most of these proteins nonspecifically bound to the LOS column (Fig. (Fig.2C,2C, lane 2) were removed after affinity purification with a HiTrap Protein G column (Fig. (Fig.2C,2C, lane 3). Further purification (Fig. (Fig.2C,2C, lane 4) with the inactivated affinity matrix yielded the anti-15253 LOS IgG. The average yield from 8 ml of the sera was ~25 μg, and 37 μg was the best yield. Subclass analysis of this purified IgG showed that it was mainly composed of IgG2 and a residual amount of IgG1 (Fig. (Fig.1D).1D). As expected, ELISA analysis (Fig. (Fig.2E)2E) indicated that the IgG bound to both the affinity ligand 15253 LOS and its OS elongated form, JW31R LOS. In addition, binding of the IgG to S. minnesota Re 595 LPS was also seen by ELISA.
The binding of IgG to neisserial LOS was confirmed using PK lysates of several gonococcal clinical isolates and a pyocin-resistant mutant strain (JW31R), and the LOSs from these strains have been structurally and/or immunochemically characterized previously. Figure Figure33 A shows PAGE/Western blot assay results for three 2,3:3,4-dibranched LOSs (15253, JW31R, and WG), which confirmed the results obtained by ELISA (Fig. (Fig.2E).2E). The IgG2 bound to 15253 LOS and its Hep[II]-elongated form (Fig. (Fig.3B),3B), JW31R LOS (5). It also bound to the Hep[I]-elongated form (Fig. (Fig.3C)3C) of 15253 LOS, WG LOS, which consists of the major and minor components (Fig. (Fig.3A)3A) (42). The disaccharide at the nonreducing end of the Hep[I]-OS on the major WG LOS component is Galβ1-4GlcNAc (N-acetyllactosamine), which is linked to lactose to form a tetrasaccharide, lactoneotetraose. This OS is identical to that expressed on a human glycosphingolipid, paragloboside. The minor component is an elongated form of the major LOS; N-acetylgalactosamine (GalNAc) is linked to the Gal of lactoneotetraose, and the terminal disaccharide of the Hep[I]-OS on this minor LOS, GalNAcβ1-3Gal, is a blood group P antigen.
In addition to the three dibranched LOSs described above, the IgG2 bound to the 3,4-branched LOSs that have OS only on Hep[I] (Fig. (Fig.4A).4A). As expected from the results obtained with the above WG LOS (Fig. (Fig.3),3), the IgG2 bound to the two F62 LOS components (40), the 4.5- and 4.8-kDa F62 LOSs (Fig. 4B and C). The former and latter OSs on Hep[I] are identical to those expressed in the major and minor WG LOS components, respectively (40). Similarly, the Ab bound to the 5.1-kDa PID-2 LOS component that contains a dimer of N-acetyllactosamine, Galβ1-4GlcNAcβ1-3Galβ1-4GlcNAc. It also bound to the 4.8-kDa PID-2 LOS, whose OS on Hep[I] is different from that of the 4.8-kDa F62 LOS; the nonreducing end, the terminal carbohydrate residue, is GlcNAc, instead of GalNAc (38). Of the six components produced by PID-2, the IgG2 bound most strongly to the 3.6-kDa LOS, which contains lactose on Hep[I]. The immunostaining of this LOS component was too intense to confirm whether the IgG also bound to the lowest-molecular-weight PID-2 LOS, which expresses only Glc on Hep[I].
The IgG2 did not bind to any of the 24-1 LOS components (27) (Fig. (Fig.4B,4B, silver staining and immunostaining) or to any of the silver-stained LOS components of 220 and 302. It only bound to each of their ~3.6-kDa minor components. The 220 LOS was recognized by MAb 2-1-L8 (23), which is specific for the 3.6-kDa LOS containing lactose on Hep[I] and phosphoethanolamine in the Hep[II] 3 position (18, 30). Therefore, the OS structure on Hep[I] of the ~3.6-kDa 220 component may be the same as that of the corresponding PID-2 LOS. The IgG binding to the ~3.6-kDa 302 LOS showed that the IgG epitope does not reside on the major LOS containing a trisaccharide composed of hexose (Hex) on Hep[I] (21).
In addition to the PK lysates described above, we examined the 3,4-branched and 2,3:3,4-dibranched LOSs that have been purified from the corresponding strains. We found that the patterns of binding to these purified LOSs were almost identical to those of LOS preparations of the cell lysates (data not shown). These results showed that the IgG2 bound to the 3,4-branched and 2,3:3,4-dibranched LOSs in spite of the presence of human carbohydrate epitopes on their OS.
The recognition of the 3,4-branched LOSs by the IgG2 and the residual binding to Re 595 LPS observed in the ELISA experiment (Fig. (Fig.2E)2E) prompted us to examine whether the IgG2 truly binds to the mutant LPS. Salmonella mutant LPSs (Ra, Rb, Rc, Rd, and Re) were analyzed by Western blotting. As shown in Fig. Fig.5B,5B, the same disaccharide, KDOα2-4KDO, is expressed at the reducing end, the innermost core of the OS moieties among these mutant LPSs and neisserial 3,4-branched and 2,3:3,4-dibranched LOSs. The inner core trisaccharide linked to the KDO-KDO is Hep[III]α1-2Hep[II]α1-3Hep[I], instead of GlcNAcα1-2Hep[II]α1-3Hep[I], which is expressed in neisserial LOS. In addition to this structural difference, the OS elongation pattern is different; no OS resides on Hep[I], and the OS is elongated from Hep[II]. Of these mutant LPSs, the IgG bound to the Rb and Re LPSs, but not to the Ra, Rc, and Rd LPSs. The binding of IgG to the neisserial LOSs, as well as to the two truncated LPSs, demonstrated that the isolated IgG2 was different from the anti-JW31R LOS Ab (43), which did not bind to either the 3,4-branched LOSs or the two mutant LPSs used in this study. This finding also suggested that the 2:3,3,4-dibranched OS structure of 15253 LOS contains several epitopes that are recognized by human antibodies.
Before exploring whether the Ab recognizes either the common KDOα2-4KDO epitope or different carbohydrate epitopes expressed on the neisserial LOSs and the two truncated LPSs, we examined whether the Ab is specific for the OS moiety. Because neisserial bacteria often express phosphoethanolamine in the core OS (6, 35), we first examined whether IgG2 binding requires a phosphorylated core and found that the Ab bound dephosphorylated 15253 LOS, MS11mkA LOS, and Re 595 LPS (data not shown). This result indicated that the major binding site(s) of the Ab is neither the phosphoethanolamine nor the phosphorylated Hep residue(s). This finding was also supported by the binding of IgG to PID-2 LOS components that do not contain any phosphoethanolamine residues within their OSs (38). We then confirmed lack of Ab binding to the lipid A components by TLC immunostaining (44). Figure Figure66 shows the results obtained using the IgG2 and a partial hydrolysate obtained after treating 15253 LOS with 0.5% acetyl hydroxide for 1 h at 100°C. As shown in Fig. Fig.6B,6B, an anti-lipid A Ab bound to the lipid A components released by this hydrolysis, as well as to the LOS, which did not migrate along the TLC plate. However, the IgG2 did not bind to the lipid A components but only bound to the LOS (Fig. (Fig.6C).6C). Similar results were also obtained using the partial hydrolysates of both MS11mkA LOS and Re 595 LPS (data not shown). Analysis of the dephosphorylated LOS and LPS samples and the TLC immunostaining results revealed that (i) the phosphoethanolamine residues or phosphorylated Hep or KDO are not the major epitopes recognized by the IgG2 and (ii) the Ab is specific for the OS moiety but not lipid A.
After confirming that IgG2 is specific for the OS moiety, we examined whether the LOS and LPS samples were also recognized by commercially available NHS and human IgG. As expected, preliminary ELISA and dot blot analyses confirmed that both the NHS and the immunoglobulin contained IgG that binds to 15253 LOS and JW31R LOS (data not shown). Next, anti-15253 LOS IgG was purified from both the NHS and human IgG preparations by affinity chromatography in a manner similar to that described earlier. Western blot analysis showed that the two IgG preparations bound to neisserial LOSs (15253, JW31R, MS11mkA, PID-2, and F62) and the two truncated LPSs (Rb and Re). The partial blot results obtained with the purified IgG from the commercially available NHS are shown below. This analysis indicated that the binding patterns and titers of the two preparations were similar to those of the IgG2 from the in-house NHS. These results showed that the binding of IgG to neisserial LOS and truncated LPS occurs with different sources of adult NHS.
We examined whether binding of the IgG2 to the neisserial LOSs and the two truncated LPSs is due to the KDOα2-4KDO commonly expressed on those LOSs and LPSs. First, we examined the expression of the IgG-defined epitope on the 2,3:3,4-dibranched neisserial LOSs using enzyme-treated WG LOS and lgtE mutant LOS. As described earlier, the major and minor WG LOS components have N-acetyllactosamine (Galβ1-4GlcNAc) and the P antigen (GalNAcβ1-3Gal) at the nonreducing ends of the Hep[I]-OS, respectively (Fig. (Fig.7C).7C). Because these disaccharides are expressed on human cells, we speculated that these self antigens were unlikely to be the IgG2-defined epitopes on the two WG LOS components. To confirm that these epitopes do not reside at the nonreducing ends but exist within the 15253 OS structure of the two components, WG LOS was sequentially treated with two exoglycosidases, β-galactosidase and β-N-acetylhexosaminidase, and these enzyme-treated LOS samples were analyzed by Western blotting. For the detection of IgG binding to those samples, we used chemiluminescence because the binding of the IgG to WG LOS was much weaker than to 15253 LOS (Fig. (Fig.7A,7A, lanes 1 and 2, immunostaining) and a large amount of the resulting LOS sample was lost during dialysis after the second enzymatic treatment (Fig. (Fig.7A,7A, lane 4, silver staining). The IgG bound to the Gal-depleted LOS component (shown as −Gal in Fig. Fig.7A,7A, lanes 3, and C) after the β-galactosidase treatment. Exposure of the 15253 LOS structure (shown as −Gal-GlcNAc in Fig. Fig.7A,7A, lanes 4, and C) after the removal of GlcNAc of the Gal-depleted component by hexosaminidase treatment greatly enhanced IgG binding, although loss of the hexosaminidase-treated LOS sample occurred, as described above. The results obtained with the enzyme-treated WG LOS indicated that the IgG-defined epitope on the two WG components is shared by the 15253 OS moiety, and in addition, the affinity of the IgG for this epitope probably becomes much lower when the nonreducing end, the terminal Gal on Hep[I] of 15253 LOS, is further glycosylated (Fig. (Fig.3C3C).
In contrast, truncation of the 15253 OS structure resulted in a loss of IgG binding. As shown in Fig. Fig.7C,7C, lgtE mutant LOS lacks the two nonreducing Gal residues of 15253 LOS (1), and only a Glc residue is linked to both Hep[I] and Hep[II]. The IgG2 did not bind to this mutant LOS, even though 100 ng of the LOS was loaded (Fig. (Fig.7B,7B, lane 2, immunostaining) and chemiluminescence detection was used. If the major binding detected between IgG2 and 15253 LOS was due to the presence of the KDO dimer at the reducing end, the IgG2 would also bind to the lgtE mutant LOS. This lack of binding to the lgtE LOS suggested that the binding of the IgG to the Re LPS is due to a different population of IgG that is specific for KDOα2-4KDO.
The lack of binding to this mutant LOS also showed that the IgG2 requires either or both of the nonreducing ends for the recognition of 15253 LOS. However, a simple lactose is not the IgG2-defined epitope on the above dibranched LOS because dot blot analysis illustrated that the IgG2 showed no binding to lactosylceramide, even at a level of 1 μg (data not shown). Therefore, these results indicated that (i) the binding of most of the IgG to 15253 LOS and possibly WG LOS is not due to KDOα2-4KDO and that (ii) either or both of the nonreducing ends of the 15253 OS are necessary for the epitope expression of the major IgG species that bound to 15253 LOS.
The noninvolvement of the KDO disaccharide in the major binding site(s) of the IgG2 on the 3,4-branched LOS was also supported by Western blot analysis using IgG absorbed with neisserial LOS or Re LPS. For the following experiments, we used the IgG purified from commercially available NHS. Because the amount of IgG available after the time-consuming sequential affinity chromatography protocol was limited, the IgG was diluted to 1:20 and 1:50 and examined by Western blot analysis. For this preliminary analysis, we used several neisserial LOSs that were bound strongly by the IgG. The 1:50 dilution of IgG (400 ng/ml) was inadequate to clearly visualize its binding to each LOS (data not shown), and therefore, the 1:20 dilution of IgG (1 μg/ml) was chosen. Figure Figure8A8A shows the results of the binding of the 1:20 dilution of IgG to the 3,4-branched (PID-2 and MS11mkA) and 2,3:3,4-dibranched (15253 and JW31R) LOSs. Of the above two 3,4-branched LOSs, the OS of the latter MS11mkA LOS (18) contains Galβ1-4Glc on its Hep[I] and is identical to that of the 3.6-kDa LOS (Fig. (Fig.4B)4B) expressed on PID-2 LOS, except that phosphoethanolamine is present in the 3 position of its Hep[II]. This MS11mkA LOS (Fig. (Fig.8A,8A, lane 2) was found to be more strongly bound by the above 1:20 dilution of IgG than 15253 LOS under the conditions used. Thus, we chose MS11mkA LOS to examine the amounts of LOS or LPS needed to absorb specific species of IgG2. Figure 8B and C show the results obtained with the IgG2 absorbed with ~19 and ~75 molar equivalents of the antigen, respectively. As expected, the binding of the IgG2 to MS11mkA and PID-2 LOSs was abolished after incubation with ~75 molar equivalents of MS11mkA LOS (Fig. (Fig.8C,8C, lanes 1 and 2), although very weak residual binding was observed when the IgG was absorbed with ~19 molar equivalents (Fig. (Fig.8B).8B). However, binding to the 2,3:3,4-dibranched 15253 LOS was also eliminated after treatment of the IgG2 with the greater amount of the 3,4-branched MS11mkA LOS (Fig. (Fig.8C,8C, lane 4). This lack of binding to the 3,4-branched MS11mkA and PID-2 LOSs, as well as the 2,3:3,4-dibranched 15253 LOS, indicated that a cross-reacting epitope is present among the three LOSs.
In contrast to 15253 LOS, the absorbed IgG2 maintained its ability to bind JW31R LOS (Fig. (Fig.8C,8C, lane 3). We expected that the IgG would not bind JW31R LOS because this LOS is the Hep[II]-elongated form (Fig. (Fig.3B)3B) of 15253 LOS. This unexpected binding to JW31R LOS suggested the presence of another minor population of IgG2 that was specific for this LOS. Expression of the IgG2-defined epitopes was further investigated using IgG that was preabsorbed with 3,4-branched MS11mkA LOS, 2,3:3,4-dibranched 15253 LOS, or the truncated Re LPS.
Figure Figure99 shows partial results of the Western blot analysis with each of the preabsorbed IgGs using the following LOSs and LPSs: 2,3:3,4-dibranched LOSs (15253 and JW31R), 3,4-branched LOSs (MS11mkA and PID-II), and mutant LPSs (Rb and Re). To decrease the possibility of leaving behind major IgGs specific for each LOS or LPS after absorption, the 1:20 dilution of IgG2 was incubated with ~188 molar equivalents of each of the three antigens. Figure 9A and B show the silver staining of the LOSs and IgG binding to the LOSs before absorption, respectively. As shown in Fig. Fig.9C,9C, the IgG preabsorbed with 2,3:3,4-dibranched 15253 LOS showed the same binding patterns as the IgG absorbed with 3,4-branched MS11mkA LOS (Fig. (Fig.8);8); it lost its ability to bind to the homologous LOS (lane 1) and the two 3,4-branched LOSs (MS11mkA and PID-2, lanes 3 and 4, respectively) but maintained its ability to bind to JW31R LOS (lane 2). In addition to JW31R LOS, the preabsorbed IgG bound to the two truncated Rb and Re LPSs (lanes 5 and 6). The results obtained with the IgG preabsorbed with MS11mkA LOS (data not shown) were almost identical to those obtained with the IgG preabsorbed with 15253 LOS. Thus, the binding of the two preabsorbed IgG preparations to the Rb and Re LPSs indicated the presence of a third IgG2 population that was specific for the KDO dimer. This finding was also confirmed by the experiment using the IgG absorbed with Re LPS.
As expected, the IgG after absorption with the Re LPS lost its binding to the Rb LPS and the homologous LPS (Fig. (Fig.9D,9D, lanes 5 and 6). However, this preabsorbed IgG bound to the 3,4-branched and 2,3:3,4-dibranched LOSs, including JW31R LOS. Because the molecular weight of Re LPS [(KDO)2-lipid A] is close to half of the designated molecular mass (4,000 Da) used for quantifying the LOS and LPS samples, the actual Re LPS used in the absorption is ~350 molar equivalents. However, no decrease in the binding of this preabsorbed IgG to the neisserial LOSs was observed.
The IgG2 binding results obtained after absorption with each of the three antigens revealed the following. (i) The binding to the Rb and Re LPSs is due to a minor but separate IgG2 fraction that recognizes KDOα2-4KDO. (ii) The IgG2 that binds to the neisserial LOSs, except for JW31R LOS, is due to a cross-reactive epitope among the 3,4-branched and 2,3:3,4-dibranched LOSs. (iii) Another IgG2 specific for JW31R LOS exists in this preparation.
We examined the bactericidal activity of the Ab against two N. gonorrhoeae strains, serum-sensitive JW31R and serum-resistant 15253. The initial bacterial concentration of each of the two strains was adjusted to 108 CFU/ml and diluted with GBSS/BSA until each strain was killed by NHS. As shown in Fig. Fig.10,10, NHS was able to kill strain JW31R (Fig. 10A, lane 5) at a concentration of 107 CFU/ml. However, it was necessary to dilute strain 15253 (Fig. 10B, lane 5) to 104 CFU/ml in order for it to be killed by NHS. In both cases, we found that the affinity-purified IgG increased the killing of the two strains. In the presence of the complement source, both the Ab and heat-inactivated NHS killed strain JW31R (Fig. 10A), and the killing activity of the IgG2 was much higher than that of GBSS.
As shown in Fig. 10B, the IgG2 and complement killed the 15253 organisms more efficiently than NHS alone, inactivated NHS and complement, or GBSS and complement, which indicated that the higher killing activity is due to the addition of the purified IgG2. The moderate killing activity of the serum-resistant strain by NHS or GBSS and complement may be explained by the amount of the bactericidal IgG used in this study, and the residual killing of GBSS and complement also indicates that antibodies specific for the organism may not be fully absorbed out of the source complement. A slightly higher killing activity observed in the inactivated NHS and NHS than the complement source alone may result from other blocking antibodies present in NHS (31, 36). These antibodies would lower the killing capability of anti-LOS bactericidal antibodies. As described earlier and as shown in Fig. Fig.88 and and9,9, anti-15253 LOS IgG was also present in commercially available NHS, and this IgG is presumably common to adult human sera.
Because purification of the large amount of IgG from the pooled NHS used for this experiment was not possible and the lytic activities of the IgG and the source complement vary from batch to batch, we did not perform the bactericidal experiment using higher concentrations of the bacteria and did not compare the bactericidal activities of the “in-house” NHS with those of Caucasian or other NHSs. Therefore, it is not clear whether the Japanese population possesses higher concentrations of anti-15253 LOS IgG. Nevertheless, the current results indicate that the anti-LOS IgG2 used in this study was functional and able to facilitate the killing of both of serum-sensitive and serum-resistant strains under the conditions studied.
Using 15253 LOS as an affinity ligand, we isolated IgG2 that recognized several 3,4-branched and 2,3:3,4-dibranched LOSs and Rb and Re mutant LPSs. The results obtained indicated the presence of three different IgG2 species. (i) The first IgG species recognizes a cross-reacting epitope expressed among the 3,4-branched and 2,3:3,4-dibranched LOSs. (ii) The second IgG species is specific for JW31R LOS, which is produced by the pyocin-resistant mutant JW31R (11), and this IgG-defined epitope is not shared with the two aforementioned branched LOSs. (iii) The third species recognizes the KDO dimer epitope that is also present in the Rb and Re LPSs.
Because the IgG was obtained by using 15253 LOS as an affinity ligand, the IgG preabsorbed with the ligand would, in theory, not bind to the epitopes expressed on the LOS. However, this expected inability to bind to those epitopes may occur if the expression level of each of several epitopes residing on 15253 LOS is similar and if each epitope-specific IgG has a similar affinity or avidity for each of the antigens used in this study. The fact that the IgG preabsorbed with 15253 LOS was able to bind to JW31R LOS and the two truncated LPSs (Fig. (Fig.9C)9C) indicated that several epitopes residing on 15253 LOS are not equally expressed. In addition, the binding affinity of each specific IgG for each LOS is probably different, as exemplified by the greatly enhanced IgG binding to the exposed 15253 LOS from the major WG LOS after removal of the nonreducing end, Galβ1-4GlcNAc, of Hep[I] (Fig. (Fig.7A,7A, lane 4). The KDO dimer epitope within the 3,4-branched and 2,3:3,4-dibranched LOSs is probably sterically hindered due to the presence of Hep[I]-elongated OS, and therefore, the binding affinity of the KDO dimer-specific IgG for those neisserial LOSs may become extremely low compared with that for the Re LPS. Therefore, the minor IgG species specific for the KDO dimer may not be completely removed from the 15253 IgG preparation, even with an excess amount of 15253 LOS. The residual KDOα2-4KDO-specific IgG that remained in the mixture could bind to the Re LPS. The binding of IgG to JW31R LOS after absorption with 15253 LOS may also be explained in a manner similar to that described above.
In regard to the binding of IgG2 to Rb LPS, we speculate that this event occurs at the same site as binding to Re LPS. We have observed similar binding patterns with peptides screened from a phage-displayed library using the same Re LPS as a ligand; the peptides selected from the LPS bound to Rb but not to Ra, Rc, or Rd (29). It has also been reported by other investigators that one of the murine MAbs raised against Re mutant LPS binds to Rb2 LPS (26). These results suggest that the same KDO epitope is expressed on both Rb and Re LPSs. They also suggest that the stereochemical or structural environments of the dimeric KDO region of the Ra, Rc, and Rd LPSs are different from those expressed on the Rb and Re LPSs. To our knowledge, this is the first report to provide conclusive evidence that human IgG in NHS binds to these truncated LPSs.
With respect to the cross-reactive epitope of the 3,4-branched and 2,3:3,4-dibranched LOSs, the structural difference between the MS11mkA and 15253 LOSs provided an insight into the expression of this epitope. Because these LOSs differ in the presence of Galβ1-4Glc on Hep[II] and because the IgG recognized the MS11mkA LOS, which lacks this disaccharide residue, the Hep[I]-elongated OS may be essential for the expression of the cross-reactive epitope. Furthermore, the IgG binding to the 2,3:3,4-dibranched 15253 and WG LOSs (Fig. (Fig.3A)3A) and the lack of binding to the lgtE LOS containing only Glc on both Hep[I] and Hep[II] (Fig. 7B and C) indicated that the IgG epitope residing on the 15253 and WG LOSs may be similar to that of MAb 2C7. However, MAb 2C7 does not recognize 3,4-branched LOSs (10, 42). Therefore, the cross-reacting epitope identified in this study is distinct from the above 2C7 epitope.
Structural determination of the 3,4-dibranched 220 and 302 LOS components that were not recognized by the IgG2 (Fig. (Fig.4A)4A) may provide further insight into this epitope. We also observed that the IgG2 showed no binding to the two major 24-1 LOS components (Fig. (Fig.4A),4A), whose OS structures have been suggested to be the same as those of the 4.5- and 4.8-kDa F62 LOS components (13). However, the mobilities of the two major 24-1 LOS components in PAGE (Fig. (Fig.4B)4B) were not the same as those of F62 LOS, which was also observed using purified LOS samples (data not shown). The partial structures of the two major 24-1 OSs may be the same as those of F62, but the entire OS structures may not be identical, which could result in the lack of binding observed.
Because of the difficulties arising from the oligoclonal nature of the IgG2 and the amount available for each sequential affinity chromatography step, we concentrated our efforts on characterizing whether the binding of the Ab to the neisserial LOSs and the mutant LPSs was due to a single epitope that was shared among the LOS and LPS samples. Our results indicate that the affinity-purified IgG2 contains at least three different IgG2 species, which showed that the dibranched neisserial OS used in this study contains several epitopes that are recognized by human antibodies. As mentioned earlier, other investigators have also reported that NHS contains antineisserial LOS antibodies, and their partial results are summarized together with our current work in Table 1 . Estabrook et al. (3) have isolated bactericidal human IgG using a 3,4-branched LOS expressing both Galβ1-4Glc and Galα1-3Galβ1-4Glc on Hep[I] as an affinity ligand. These LOSs containing the former and latter OSs correspond to meningococcal serotypes L8 and L1, respectively, and these OS structures are also expressed on gonococci (34). Their flow cytometric analysis showed that the IgG bound to meningococcal strains expressing LOS components (serotypes L3,7,9 and L2) whose OS on Hep[I] is identical to that of the 4.5-kDa component of the F62 and PID-2 LOSs (Table (Table1).1). Also, Jäkel et al. (15) have affinity purified IgG by using a deacylated, 3,4-branched LOS conjugate which contained only Glc on Hep[I]. In an LPS ELISA analysis, they have also shown that the isolated IgG bound not only to the ligand LOS but also to the 3,4-branched LOS, whose OS on Hep[I] is either GlcNAcβ1-3Galβ1-4Glc or Galβ1-4GlcNAcβ1-3Galβ1-4Glc (Table (Table1).1). Their IgG was found to be bactericidal only with the mutant strain that biosynthesized the OS used for the affinity ligand. The subclasses of the human IgGs described in the two studies mentioned above were not characterized. In addition, it is not clear whether the IgG isolated by the former group truly binds to specific LOS serotype structures, because LOS on the meningococcal surface is phase variable. Nevertheless, their results also support the notion that human anti-LOS antibodies recognize several epitopes within the branched neisserial OS.
The importance of antibacterial antibodies in immunity to and protection against pathogenic neisseriae, especially N. meningitidis, is well established and has been described in a review article by Pollard and Frasch (33). Serum bactericidal activity in the newborn is due to transplacental transfer of immunoglobulin, and 50% of newborns have antimeningococcal bactericidal activity. Serum bactericidal activity steadily increases through childhood (50 to 80% bactericidal activity by 12 years of age), so that by adulthood 65 to 85% of individuals have bactericidal activity against N. meningitidis (7, 8). Of the several immunogenic meningococcal surface components, LOS is also immunogenic and anti-LOS antibodies are detected in sera following infection (2, 9, 22). It was shown that convalescent-phase sera from children infected with N. menigitidis contained bactericidal anti-LOS antibodies (9), and a specific LOS component was suggested to be a possible target for the bactericidal antibodies (2).
Because meningococci and gonococci have cross-reactive surface antigens, antigonococcal LOS antibodies in the host have also been investigated; the presence of IgG that binds to the outer membrane proteins, as well as LOS, in the serum of men (92%) with no history of prior sexually transmitted diseases had also been shown in an earlier work by Hicks et al. (12). They also reported that urethral infection by gonococci induced anti-LOS IgG in six of nine patients who had no detectable Ab in their preinfection sera. The importance of serum antibodies in the course of gonococcal infection has been reported by Plummer et al., who have shown that antibodies against selected gonococcal surface antigens can either increase or decrease the susceptibility to disease (31, 32). Furthermore, the specificity of serum and local antibodies of female patients have been analyzed (14, 20), and IgG and IgA against pili, LOS, porin protein I, and opacity proteins have been detected in both serum and vaginal fluids. These studies on the above Ab-antigen specificities indicate that serum antibodies against specific gonococcal antigens may be relevant to local immune responses.
The presence of anti-LOS antibodies in NHS is probably due to the exposure to N. meningitidis, N. lactamica, and/or other bacteria that have cross-reacting epitopes with neisserial bacteria. Most meningococcal infections are asymptomatic, and adults are likely to carry meningococci in the nasopharynx at some time during their lives. Also, it is well known that nonpathogenic N. lactamica has many antigenic determinants in common with pathogenic N. meningitidis and N. gonorrhoeae (12, 19, 22). An earlier immunochemical characterization utilizing murine MAbs has shown that host carbohydrate epitopes such as Galβ1-4GlcNAc and Galβ1-4Glc expressed on Hep[I] of meningococcal and gonococcal LOS (19) are also present on N. lactamica LOS. Although a structural analysis of the core OS of N. lactamica LOS has not yet been done, a recent genetic analysis of the genes involved in the biosynthesis of neisserial OS (45) suggests that N. lactamica may be able to biosynthesize the same core OS expressed on meningococcal and gonococcal LOSs. Although the development and biological functions of anti-core OS antibodies remain to solved, these antibodies probably function in protection against pathogenic neisserial bacteria, as well as other Gram-negative bacteria that contain epitopes that are cross-reactive with those of nonpathogenic and pathogenic neisserial bacteria.
In summary, we isolated IgG2 from NHS by using 15253 LOS as an affinity ligand and found that the IgG2 bound to 2,3:3,4-dibranched LOSs, 3,4-branched LOSs, and truncated LPSs. The results obtained in this study showed that the IgG2 was composed of at least three different species: an IgG2 that was specific for the cross-reactive epitope expressed on the two different branched LOSs, another IgG2 that was specific for LOS from pyocin mutant strain JW31R, and a third IgG2 that recognized the KDOα2-4KDO epitope that was common to the Rb and Re LPSs. The IgG2 was also found to be functional and was able to facilitate the killing of serum-resistant strain 15253 under the conditions used in this study. Our results show that neisserial OS contains several epitopes that are recognized by human antibodies. They are directed not toward human carbohydrate-mimicking epitopes such as Galβ1-4GlcNAc and GalNAcβ1-3Gal but against epitopes that reside within the branched OS. Differentiation of such epitopes would lead to the development not only of effective vaccines against pathogenic Neisseria species but also of potential analytic and diagnostic tools for the identification of specific microorganisms and the infections they cause.
This work was supported in part by grants-in-aid from the Ministry of Education, Culture, Sports, Science, and Technology of Japan (10306022 and 14360207).
We thank Sanjay Ram (University of Massachusetts Medical School, Worcester) for reading parts of the Results and Discussion and providing useful comments.
Editor: J. N. Weiser
Published ahead of print on 17 May 2010.