Previous studies of Ehrlichia chaffeensis infection in the mouse have demonstrated that passive transfer of polyclonal Abs from resistant immunocompetent mice to susceptible SCID mice ameliorated infection and disease, even when Abs were administered during established infection. To identify particular Abs that could mediate bacterial clearance in vivo, E. chaffeensis-specific mAbs were generated and administered to infected SCID mice. Bacterial infection in the livers was significantly lowered after administration of either of two Abs of different isotypes (IgG2a and IgG3). Moreover, repeated administration of one Ab (Ec56.5; IgG2a) rescued mice from an otherwise lethal infection for at least 5 wk. Both protective Abs recognized the E. chaffeensis major outer membrane protein (OMP)-1g. Further studies revealed that both Abs recognized closely related epitopes within the amino terminus of the first hypervariable region of OMP-1g. Analyses of human sera showed that E. chaffeensis-infected patients also generated serological responses to OMP-1g hypervariable region 1, indicating that humans and mice recognize identical or closely related epitopes. These studies demonstrate that OMP-specific mAbs can mediate bacterial elimination in SCID mice, and indicate that Abs, in the absence of cell-mediated immunity, can play a significant role in host defense during infection by this obligate intracellular bacterium.
It is commonly believed that Abs play little or no role in immunity to intracellular bacteria, and that cell-mediated immunity is the principal mechanism of host defense (1). However, studies of several intracellular pathogens have indicated that under some conditions Abs can provide effective immunity (Refs. 2, 3, 4, 5 ; reviewed in Ref. 6). Similarly, studies of ehrlichiae immunity have also demonstrated that passive Ab administration could protect immunocompetent (7, 8) and immunodeficient SCID mice (9) from infection. Nevertheless, a role for Abs in host defense against intracellular bacteria is novel, and the exact mechanism(s) involved have only begun to be explored in depth.
To assess the role of Abs during intracellular bacterial infection in more detail, this study has used Ehrlichia chaffeensis, an obligate intracellular pathogen that infects cells of the monocyte/macrophage lineage (10). The bacterium, which is tick transmitted, is the agent of human monocytic ehrlichiosis. The factors influencing disease susceptibility in humans have not been defined, although immunocompromised individuals appear to be particularly susceptible to serious disease (11). In previous studies, which used a mouse model for E. chaffeensis infection, it was demonstrated that susceptibility was correlated with immunodeficiency (12, 13). Infected SCID mice developed severe and fatal disease that bore resemblance to human monocytic ehrlichiosis (12). Studies of the immunological basis of disease resistance revealed that passive transfer of polyclonal Abs provided significant protection to the susceptible SCID mice (9). Most remarkable was the ability of Abs to mediate bacterial clearance from the livers of SCID mice, even when the Abs were administered well after infection had been established in this tissue (9). To begin to understand the mechanism(s) whereby Abs could mediate intracellular bacterial clearance, two mAbs that protect SCID mice from infection and disease have been generated. Both Abs recognized an outer membrane protein (OMP)3 that is immunodominant in both humans and mice, indicating that OMPs are likely to be major targets of the protective immune response. Moreover, the ability of Abs to mediate clearance of these intracellular bacteria in the absence of T or B lymphocytes indicates that Abs alone can play a significant role in host defense during intracellular bacterial infection.
Materials and Methods
Animals and infections
C57BL/6, C57BL/6-scid, and BALB/c-scid mice were obtained from The Jackson Laboratory (Bar Harbor, ME), or were bred in the Wadsworth Center Animal Care Facility (Albany, NY) under institutional guidelines for animal care and use. The mice were routinely infected with 1–2 × 106 E. chaffeensis-infected DH82 cells via the peritoneum, as described previously (12). Institutional animal care and use guidelines did not permit the use of death as an experimental endpoint in these studies. In some experiments body weight was used to monitor animal health and Ab efficacy.
Bacteria and cell lines and PCR analyses
E. chaffeensis was cultured, as described previously (12), in the canine monocytic cell line DH82. The Arkansas isolate of E. chaffeensis was used for all mouse infections in this study, and was cultured from an early passage obtained from the Centers for Disease Control (Atlanta, GA). The bacteria were quantitated by semiquantitative PCR assay, as described previously (12, 9), using E. chaffeensis-specific 16S rDNA oligonucleotide primers (14). PCR amplification of the mouse glucose-6 phosphate dehydrogenase gene was performed using the oligonucleotide primers 5′-GACCTGCAGAGCTCCAATCAAC-3′ (sense) and 5′-CACGACCCTCAGTACCAAAGGG-3′ (antisense). Genomic DNA (200 ng) was amplified for 40 cycles (94°/30s; 52°/45s; 72°/60s) using 0.5 U AmpliTaq polymerase (Perkin-Elmer, Wellesley, MA) in reaction buffer containing 1.5 mM MgCl2.
The PCR products were electrophoresed in 1.5% agarose gels and visualized with ethidium bromide. The limit of bacteria detection by PCR has been estimated to be ∼1 × 106 organisms per gram of liver tissue (12). The JAX and St. Vincent isolates of E. chaffeensis were generously provided by Dr. Christopher Paddock (Centers for Disease Control), and were also cultured in the DH82 cells.
C57BL/6 mice were infected i.p. with E. chaffeensis-infected DH82 cells, in the absence of adjuvant, two to four times at 2- to 4-wk intervals. Splenocytes were harvested and fused to the myeloma cell line SP2/0 using standard protocols. The hybridoma supernatants were screened for reactivity to E. chaffeensis by immunofluorescence assay, as described previously (9), and cells that produced specific Abs were expanded and subcloned by limiting dilution. Heavy and light chain Ab isotypes were determined by ELISA using isotype-specific polyclonal reagents (Southern Biotechnology Associates, Birmingham, AL).
Ab purification and administration
Abs were purified by fast performance liquid chromatography from hybridoma culture supernatants using protein A- or protein G-Sepharose (for IgGs; Amersham Pharmacia Biotech, Piscataway, NJ) or IgM purification columns (Amersham Pharmacia Biotech), following the instructions of the manufacturer. Abs were administered to infected mice via the peritoneum (100–200 μg/mouse). Ab concentrations were determined by measurement of absorbance at 280 nm. Irrelevant isotype control Abs used were KJ1-26 (IgG2a; anti-TCR; Ref. 15 ; purified as described above), and a mouse myeloma IgG3 (Sigma, St. Louis, MO).
Production and purification of rOMPs
Production and purification of the full-length rOMP-1g and OMP-1d have been described previously (16). To generate truncated forms of OMP-1g, the OMP-1g gene was also cloned into the expression vector pET32-LIC (Novagen, Madison, WI). The full-length OMP-1g was first isolated by PCR from E. chaffeensis DNA obtained from infected DH82 cells using the oligonucleotides 5′-GACGACGACAAGATGGACCCAGCAGGTAGT-3′ (sense) and 5′-GAGGAGAAGCCCGGTTTAGAAAGCAAACCTTCC-3′ (antisense), and the PCR product was cloned using a ligation independent cloning strategy, as described by the manufacturer (Novagen). The three-step cycling PCR conditions were as follows: an initial 3-min denaturation at 94°C; 30 cycles of a 1-min denaturation at 94°C, 1-min annealing at 55°C, and 1-min polymerization at 72°C; followed by a 10-min extension at 72°C. The truncated forms of OMP-1g were subsequently generated by PCR using the cloned OMP-1g plasmid as a template, by pairing the OMP-1g sense oligonucleotide with the following antisense oligonucleotides: 5′-GAGGAGAAGCCCGGTTTATATGTCAACTAATCC-3′ (to generate OMPΔ3), 5′-GAGGAGAAGCCCGGTTTAAAACGGGTTGTTTTC-3′ (OMPΔ2/3), and 5′-GAGGAGAAGCCCGGTTTACTTCAGTCCAAAC-3′ (OMPΔ1/2/3). The positions of the resulting truncations are shown in Fig. 4. OMPΔ1 was generated using the sense oligonucleotide 5′-GACGACGACAAGATGTTAGGTTTTGCAGGA-3′, paired with the OMP-1g antisense oligonucleotide. This construct introduced an additional methionine at the immediate amino terminus of the OMP-1g Δ1 coding sequence, which began at amino acid residue 101 of the wild-type protein. The OMPs were expressed in pET32-LIC as thioredoxin fusion proteins, and contained, in addition, a 6X histidine peptide tag. The nucleotide sequences of the cloned genes were verified. The predicted sizes of these fusion proteins are as follows: OMP-1g, 48.2 kDa; Δ3, 32.9 kDa; Δ2/3, 25.2 kDa; Δ1/2/3, 21.7 kDa; and Δ1, 40.2 kDa. The OMP expression plasmids were transfected into E. coli BL21(DE3) competent cells (Novagen), and protein expression was induced for 4 h at 37°C using 100 mM isopropyl-β-d-thiogalactopyranoside (Amresco, Solon, OH). Purification of the rOMPs was performed using nickel chelate chromatography. The bacteria were lysed in 8 M urea buffer containing 50 mM Tris-HCl (pH 7.5), and the rOMPs were purified from cleared lysates by fast performance liquid chromatography using a HiTrap chelating chromatography column (Amersham Pharmacia Biotech). The bound proteins were eluted from the column with an elution buffer containing 8 M urea, 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 50 mM EDTA. The purified proteins were dialyzed in 1 M urea and 50 mM Tris-HCl (pH 7.5), and stored at 4°C. In some cases, the proteins were further dialyzed in PBS.
Western analyses of lysates of bacteria-infected cells were performed as described previously (9). SDS-PAGE sample buffer containing 2-ME (2% SDS, 2% 2-ME, 10% glycerol, 50 mM Tris-HCl (pH 6.9)) was added to 0.1 μg of each rOMP, the proteins were boiled and electrophoresed in an 8–20% gradient SDS-PAGE gel, and transferred to polyvinyl difluoride blotting membranes. The membranes were blocked with 1% nonfat dry milk in PBS, and probed with C57BL/6 normal or immune serum at a 100-fold dilution, or mAbs at 10 μg/ml in blocking solution. Bound Abs were detected using HRP-conjugated anti-mouse Ig secondary reagents (at a concentration of 1:200; Southern Biotechnology Associates), and the blots were developed using a chemiluminescent substrate (ECL Plus; Amersham Pharmacia Biotech).
Purified rOMPs were adsorbed overnight to 96-well microtiter plates (Dynex Technologies, Chantilly, VA) at a concentration of 3 μg/ml in PBS. Peptides were adsorbed in sodium carbonate buffer (pH 9.6), at a concentration of 10 μg/ml. The peptides were synthesized by Genemed Synthesis (South San Francisco, CA; peptide 61–90), or by the Wadsworth Center Peptide Synthesis Core Facility. The microtiter plates were blocked with 1% nonfat dry milk in PBS. Bound Abs were detected using alkaline phosphatase-conjugated subclass-specific anti-mouse IgG secondary Abs (Southern Biotechnology Associates), or alkaline phosphatase-conjugated anti-human Ig (heavy plus light) secondary Abs (Southern Biotechnology Associates), followed by p-nitrophenyl phosphate (Sigma), a colorimetric substrate for alkaline phosphatase. The absorbance was read at 405 nm using a ThermoMax microplate reader (Molecular Devices, Sunnyvale, CA).
A single cubic polynomial “growth curve” was initially employed to model the time dependency of weight simultaneously for all four experimental groups shown in Fig. 2 c. Parameter estimates were determined by classical least squares regression and a stepwise model selection technique proceeded as follows. Groups of four parameters, one for each experimental group, were sequentially added to the initial model if they produced a significant contribution: four intercepts, four slopes, and four quadratic terms. The group of four cubic terms was not significant. A backwards elimination procedure was then employed to sequentially remove or merge the individual most insignificant parameter until only significant terms remained. The validity of using least squares was investigated with Bartlett’s test for homoscedasticity (17), the Kolomogrovo-Smirnov test for normal residuals (18), and by comparing the results to a robust regression (19). Because the cubic term was nonsignificant, the final set of significant parameters potentially included up to four quadratic equations in 12 parameters. However, three parameters not significantly different from zero and matches between six other parameters reduced this number by half. The final models contain linear growth for Ec56.5 and three similar parabolas in the remaining groups. The final model showed homoscedastic errors, p = 0.34; normal residuals, p = 0.07; and estimates closely similar to those obtained from robust regression. No outliers were present.
E. chaffeensis mAbs mediated bacterial clearance in SCID mice
Our previous studies demonstrated that polyclonal Abs mediated clearance of E. chaffeensis from the livers of SCID mice (9). To identify and characterize mAbs that could mediate ehrlichial immunity, mAbs were generated from infected immunocompetent C57BL/6 mice. E. chaffeensis-specific Abs produced by 27 hybridomas were tested in vivo for their abilities to mediate bacterial elimination from the livers of infected SCID mice. Bacteria were quantitated in liver tissue because this organ is a major site of ehrlichia infection (12). Of nine Abs that were found to mediate bacterial clearance (i.e., partial to apparently complete elimination of bacterial infection), two Abs were chosen for in depth study. In vivo administration of the Abs Ec56.5 (IgG2a) or Ec18.1 (IgG3) provided significant protection from infection when administered to C57BL/6-scid mice either before infection, or 10 days postinfection (Fig. 1,a). Bacterial infection was decreased, often to below detectable levels, within 4 days of Ab administration. Administration of Ec56.5 resulted in nearly complete bacterial elimination from infected liver tissue. Administration of Ec18.1 was also effective, but perhaps less efficient, because clearance was not always complete (Fig. 1,a). Abs were also effective in BALB/c-scid mice, and isotype matched irrelevant Abs did not mediate bacterial clearance (Fig. 1 b). The timing of Ab administration was not critical because bacterial clearance was observed when administration was performed before, or well after infection had been established. These data extend our previous studies by demonstrating that both monoclonal and polyclonal Abs could provide effective immunity during E. chaffeensis infection in SCID mice.
Our previous studies also demonstrated that repeated administration of polyclonal Abs protected susceptible SCID mice from lethal E. chaffeensis infection (9). Repeated administration of the polyclonal Abs was necessary for long-term protection apparently because the Abs were depleted in the SCID mice. To determine whether the mAbs could similarly provide long term protection if administered at regular intervals, infected C57BL/6-scid mice remained untreated, or were treated weekly with C57BL/6 immune serum, Ec56.5, or Ec18.1, beginning on day 10 postinfection. In these experiments, the untreated mice became moribund within 35–42 days, and exhibited characteristic weight loss, lack of mobility, hunched posture, and liver necroses (12). Analysis of the untreated and Ec56.5 and immune serum-treated mice 35 days postinfection revealed that the untreated mice were heavily infected, but bacterial infection was low to undetectable in the livers of the immune serum- or Ab-treated mice (Fig. 2,a). An additional group of Ec56.5-treated mice that continued to receive weekly Ab treatment showed no signs of disease 70 days postinfection. PCR analyses revealed undetectable or very low levels of bacterial infection in the livers of these mice (Fig. 2,a). Repeated administration of Ec18.1 also enhanced survival and resulted in significantly lower levels of infection when examined on day 42 postinfection (Fig. 2,b). However, long term immunity mediated by Ec18.1 was somewhat less effective because the onset of morbidity was delayed, but was not prevented, in the Ec18.1-treated mice. The reason for the delayed appearance of morbidity relative to earlier published experiments (9) is not known, but may have been attributable to attenuation of the bacteria in culture. These observations were supported by analyses of mouse body weight, which provided an independent measure of Ab efficacy (Fig. 2,c). Mice that did not receive Abs exhibited a pronounced loss of body weight beginning about 3 wk postinfection. Weight loss was delayed in the immune serum or Ec18.1-treated mice, and was completely abrogated in the Ec56.5-treated mice. However, bacterial infection was reduced in the livers of the Ec18.1-treated mice on day 59 (Fig. 2 b, panel 4), indicating that decreased bacterial infection in the liver did not always correlate with loss of morbidity. The data provided further evidence that mAb administration effectively cleared the ehrlichiae from infected livers, and identified one Ab, Ec56.5, that protected SCID mice from an otherwise lethal infection for at least 35 days.
Ab recognition of E. chaffeensis OMPs
To begin to understand the mechanism(s) whereby Abs mediate bacterial clearance, the Ag(s) recognized by the protective mAbs were identified. Both Ec56.5 and Ec18.1 detected by Western analysis an Ag of ∼28 kDa in detergent extracts of infected DH82 cells, and the Ag(s) comigrated with an immunodominant Ag identified previously using polyclonal mouse sera (Fig. 3,a; 9). The 28 kDa Ag was only found in infected cells, and was not recognized by normal sera, sera raised against uninfected DH82 cells, or irrelevant isotype matched Abs (Fig. 3,a). The molecular size of the 28-kDa Ag was characteristic of a previously described E. chaffeensis OMP (20, 21, 22). To determine whether the mAbs also recognized OMPs, Western analysis was performed using rOMP-1g (also known as open reading frame (ORF)5). OMP-1g is an expressed OMP in the E. chaffeensis Arkansas isolate used in these studies (16). Both mAbs recognized the rOMP-1g (Fig. 3 b, top), thus identifying OMP-1g as a target of the Abs, and confirming OMP-1g as an immunodominant Ag in the mouse (9).
The E. chaffeensis OMPs are encoded by a large multigene family of related proteins that exhibit overall conservation, but contain short regions of high diversity (21, 22). To determine whether the epitopes recognized by the Abs were shared or unique among related OMPs, Western blotting was performed using rOMP-1g, and OMP-1d (also known as ORF2; Ref. 16). OMP-1d, also obtained from the Arkansas isolate, is not thought to be an expressed gene (22). Neither Ec18.1 or Ec56.5 recognized OMP-1d, which indicated that the relevant epitope(s) were not shared by OMP-1g and OMP-1d (Fig. 3,b). OMP-1g and OMP-1d differ largely in three regions, known as hypervariable regions (HVRs), that are divergent among all of the E. chaffeensis OMPs (21, 22, 23). Therefore, the data suggested that the relevant mAb epitopes were located within the HVRs of OMP-1g (Fig. 4,a). The polyclonal Ab response was also largely directed at HVRs, because OMP-1d was recognized only weakly by the immune serum (Fig. 3 b). Thus, the HVRs are highly antigenic, which probably explains why these regions also exhibit the greatest degree of diversity.
E. chaffeensis OMPs exhibit allelic differences within the HVRs among clinical isolates (23), so Ab recognition of OMPs putatively expressed by two additional isolates was also examined. Both mAbs recognized putative OMPs from the Arkansas, St. Vincent, and JAX isolates, which indicated that the Ab epitope(s) were conserved among the OMPs expressed by the three isolates (Fig. 3,c). The apparently weaker recognition of the JAX OMP(s) was due to lower levels of bacterial infection in the cultures. It is not known whether OMP-1g is the only OMP expressed by these isolates because it is possible that the Abs also recognized other OMPs expressed by the isolates. However, the data from these analyses, combined with the above finding that the Ec18.1 and Ec56.5 epitope(s) were found in OMP-1g, but not OMP-1d, indicated that the epitopes were likely to be located within the HVRs, in particular, at residues in the HVRs that were conserved among the clinical isolates, and yet differed between OMP-1g and OMP-1d (Fig. 4 a).
To identify the OMP-1g HVR(s), which encoded the Ab epitopes, truncated forms of rOMP-1g were produced and purified (Fig. 4, b and c). ELISAs indicated that both Abs recognized epitopes within HVR1 (residues 70–100) of OMP-1g (Fig. 5). This was evident because deletion of HVR3 and HVR2 had no effect, but deletion of HVR1 (in both Δ1/2/3 and Δ1) abrogated nearly all mAb recognition (Fig. 5). Moreover, a significant portion of the reactivity of the immune sera could similarly be attributed to recognition of HVR1 (Fig. 5), indicating that within OMP-1g HVR1 was immunodominant.
To further characterize the epitopes recognized by the mAbs, synthetic peptides were generated and examined by ELISA (Fig. 6,a). A peptide containing residues 71–90 of OMP-1g was recognized by neither Ab, but peptides 61–90 and 65–78 were recognized by both Abs, indicating that critical portions of the epitopes were contained within or adjacent to residues 65–70 (Fig. 6,b). The weaker binding of Ec18.1 to peptide 65–78 relative to peptide 61–90 suggested that residues flanking peptide 65–78 also contributed to recognition by this Ab. Within peptide 65–70, only two residues, L68 and K69, differed between OMP-1d, which was not recognized, and OMP-1g expressed by the Arkansas, and presumably the JAX and St. Vincent isolates (Fig. 6; Ref. 23). The conservative leucine to isoleucine substitution at position 68 between OMP-1g and OMP-1d was not required for recognition. However, substitution of the OMP-1d-encoded glutamic acid for lysine at position 69 in OMP-1g (peptide K69E) abolished reactivity by Ec18.1, but not Ec56.5, indicating that K69 was critical for Ab recognition by Ec18.1. Note that although K69 was present in truncation Δ1/2/3, which was not recognized by Ec18.1, the location of this residue at the carboxyl terminus of this truncation probably abrogated Ab recognition. Ec56.5 recognition of the peptide containing the K69E substitution was only partially affected (Fig. 6), so the contribution of other nearby residues was also examined. Ec56.5, as well as Ec18.1, recognized N71D, as expected, because this residue was encoded by the JAX and St. Vincent OMP-1gs. However, neither Ab recognized the doubly substituted peptide Q70N/N71D, indicating that residue Q70 in OMP-1g was critical for recognition by both Abs (Fig. 6). Residue Q70 was essential, but not sufficient for Ec56.5 binding, because only very weak binding of Ec56.5 to OMP-1d was detected by Western analysis (Fig. 3,b). Because Ec56.5 required Q70, but not K69, and because some reactivity was detected using the truncation mutant Δ1/2/3 (Fig. 5), the data suggested that additional residues amino-terminal to the truncation breakpoint in Δ1/2/3, excluding K69, probably also contributed to Ag recognition. In contrast, Ec18.1 required as part of its epitope both K69 and Q70 (Fig. 6). Thus, although subtle differences in specificity were observed, two independently isolated protective Abs recognized nearly identical epitopes in OMP-1g HVR1.
Humans and mice recognized similar OMP epitopes
Humans also mount an immunodominant OMP Ab response (24, 25). To determine whether similar OMP epitopes might be recognized by both mice and humans, acute and convalescent sera from E. chaffeensis-infected human patients was examined by ELISA using the Arkansas OMP-1g peptide 61–90. Patient sera exhibited significantly higher levels of anti-peptide reactivity when compared with normal human sera (Fig. 7). Although the bacteria that infected the human patients had not been characterized, it is likely that OMPs containing epitopes conserved with OMP-1g were expressed in the patients. Therefore, the data indicated that similar HVR1 epitopes were recognized by both mouse and man.
mAbs mediated bacterial elimination during active infection
We previously demonstrated that passive transfer of immune serum not only protected SCID mice from E. chaffeensis infection, but also ameliorated disease, even when administered during a well-established infection (9). Here, these findings have been extended by demonstrating that similar protection could be observed using either of two mAbs. The mAbs demonstrated equal or better efficacy than that achieved using immune serum, as determined by direct assay for bacterial infection, and by analyses of body weight. Repeated administration of one Ab (Ec56.5) rescued SCID mice from an otherwise fatal infection for at least 5 wk, and may do so indefinitely. To our knowledge, this is the first report of such efficacy of a mAb in an immunocompromised host during an ehrlichia infection. The data confirm and extend our and others’ previous studies that demonstrated the efficacy of Abs during some ehrlichia infections (7, 8, 9), and support the assertion that, in contrast to prevailing dogma, Abs can play a significant role during host defense against an obligate intracellular bacterium.
The role of OMPs in ehrlichia immunity
Both protective mAbs recognized immunodominant E. chaffeensis OMPs. Moreover, of a total of 39 mAbs recovered from three independent B cell fusions, 22 Abs recognized OMP-1g, and in preliminary studies, nine of these OMP Abs were protective after administration to SCID mice (J.S.L. and G.M.W., unpublished data). The effectiveness of the OMP immune response is supported by a previous study that demonstrated that immunization of immunocompetent mice with purified OMP-1g protected them from infection (21). Thus, the OMPs are major targets of the protective immune response in the mouse, and are also likely to be important in human immunity (26).
The OMPs in E. chaffeensis and related ehrlichiae are encoded by multigene families (10, 21, 22, 27, 28, 29, 30), so it is likely that variation in the expression of different OMPs contribute to immune evasion in the natural host for E. chaffeensis, the white tailed deer (31). Indeed, cyclic rickettsemia has been reported to occur in cattle infected with the related erythrocyte tropic ehrlichial pathogen Anaplasma marginale (32, 33). However, a role for antigenic variation during opportunistic infection by E. chaffeensis has not yet been demonstrated, and antigenic variation has not been reported to occur during ehrlichial infection of either mice or humans. The potential for OMP variation suggests, nevertheless, that the bacteria have evolved mechanisms to avoid protective humoral immune responses in the natural hosts.
Epitope characterization studies indicated that the protective Abs Ec18.1 and Ec56.5 recognized nearly identical regions of HVR1, because the glutamine residue at residue 70 in OMP-1g was critical for recognition by both Abs. Of the 22 OMP-1g-specific mAbs recovered during hybridoma analyses, at least 15 recognized HVR1 epitopes (J.S.L and G.M.W., unpublished data), indicating that among the OMP-1g HVRs, HVR1 was immunodominant. Moreover, a significant component of the polyclonal Ab response was directed at HVR1 in OMP-1g. Therefore, epitope use within OMP-1g may be critical for Ab efficacy. The high antigenicity of HVR1 suggests that this region of OMP-1g is particularly exposed on the bacterium, and is likely to be a strong candidate for the development of vaccines designed to elicit protective humoral responses.
Comparative studies indicated that infected humans produced Abs that recognized epitopes in OMP-1g similar to those recognized by the mouse Abs. These data suggest that mice and humans generate similar Ab responses to these Ags. It is not yet known whether Abs will show similar efficacy in infected humans, but it is likely that similar immune mechanisms will be used by both mice and humans.
The epitope characterization studies presented here suggested that changes in OMP expression might have been sufficient to allow the bacteria to avoid elimination by the mAbs. However, upon repeated Ab treatment mice were protected for at least 70 days postinfection, which suggested that significant numbers of bacterial escape variants did not arise during this time. Low levels of infection were observed in the liver on day 70 postinfection in some of the Ec56.5-treated mice, but it is not yet known whether these possibly resistant bacteria resulted from the expression of variant OMPs. Of the five OMPs for which sequences are currently available, four likely encode the Ec56.5 epitope, but it is nevertheless possible that expression of other uncharacterized E. chaffeensis OMPs might have contributed to resistance to mAb-mediated elimination in the long term protection studies.
Mechanisms of humoral immunity
Of particular interest is the observation that humoral immunity was effective in SCID mice in the absence of an adaptive cell-mediated immune response. This finding contrasts with other examples of humoral immunity to intracellular pathogens such as Cryptococcus neoformans, where Abs were effective only in the presence of T cells (34). Therefore, the mechanism(s) of Ab-mediated protection is likely to differ during the different infections, probably because of differences in the pathogenesis and lifecycles of the different organisms.
The data also suggest that Ab isotype is critical for Ab efficacy. Although protective Abs of IgG2a and IgG3 isotypes were recovered, in similar experiments several IgMs failed to provide significant protection in SCID mice (J.S.L. and G.M.W., unpublished data). Moreover, greater efficacy was achieved with Ec56.5 (IgG2a) than Ec18.1 (IgG3) in the extended protection experiments, even though Ec56.5 and Ec18.1 are both high-affinity Abs and recognize very similar OMP-1g epitopes. IgG2a are more effective at complement fixation and FcR binding than IgG3 (35), so complement fixation and FcR-mediated IgG recognition are likely to be important for Ab-mediated bacterial clearance. Protective immunity was similarly correlated with the development of a T cell-dependent IgG2 major surface protein Ab response during infection of cattle by the related ehrlichia Anaplasma marginale (36). However, the efficacy of IgG3 suggests that alternative mechanisms may be involved, because phagocytosis mediated by IgG2a and IgG3 Abs probably occurs via different FcRs (37). Ab FcR interactions are likely to be critical because of the failure of OMP-1g-specific IgM Abs (some which recognized HVR1), which do not bind murine FcRs, to provide protection. One hypothesis is that ehrlichia-IgG immune complexes mediate an FcR-dependent oxidative burst in phagocytes that contributes to bacterial clearance. Mouse macrophages express the high-affinity type I, and the low-affinity type II and type III FcγR (38), and immune complex mediated oligomerization of some of these receptors has been demonstrated to activate intracellular signaling pathways associated with microbicidal activity (39, 40).
Why are Abs so effective against this obligate intracellular bacterium? The details of the E. chaffeensis lifecycle in the infected host have not been well characterized, and it is possible that during part of their lifecycle, the bacteria are exposed to Abs in the extracellular milieu, perhaps during intercellular transfer. Alternatively, immune complex might trigger effector functions in macrophages or other phagocytes and thereby directly or indirectly mediate clearance of resident intracellular organisms. Kupffer cells, a principal target of the Abs, may be particularly sensitive to such stimuli. The mouse model of ehrlichia infection will provide the tools for dissecting the mechanism(s) involved.
We thank Dr. Donal Murphy (Wadsworth Center) for critical reading of the manuscript, Dr. Susan Wong (Wadsworth Center) for the human patient sera, Dr. Christopher Paddock of the Centers for Disease Control (Atlanta, GA) for the St. Vincent and JAX E. chaffeensis clinical isolates, and Frank Abbruscato for excellent technical assistance. We also thank the Wadsworth Center Peptide Synthesis and the Immunology Core Facilities.
This work was supported in part by U.S. Public Health Service Grant 5R29CA69710-02.
Abbreviations used in this paper: OMP, outer membrane protein; ORF, open reading frame; HVR, hypervariable region.