INTRODUCTION
Cholera is a severe dehydrating illness of humans caused almost exclusively by
Vibrio cholerae of the O1 serogroup. Over 1 billion people remain at risk for cholera in 51 countries of endemicity, and there are an estimated 3 million cases and 95,000 deaths each year from cholera (
1). The current global pandemic began in 1961 and gives no indication of abating, as evidenced by recent large outbreaks in Haiti and Yemen (
2). This reality has led to enhanced commitments to cholera control strategies (
3). Such strategies now include vaccination against cholera, along with efforts to improve water and sanitation (
4). Currently available oral killed cholera vaccines are an important addition to these control efforts; however, these vaccines may provide limited durable protection, especially in immunologically naive individuals, including children under 5 years of age who bear a large proportion of the global cholera burden (
2). In comparison, survivors of cholera, including young children, have high-level protective immunity that persists for years (
5).
The development and optimal use of cholera vaccines has been hampered by the relatively limited understanding of the immunologic mechanisms of protection against cholera.
V. cholerae is a noninvasive luminal intestinal pathogen, and it is likely that antibodies that bind the pathogen or its products in the intestinal lumen contribute to protective immunity (
6). Cholera is a toxin-mediated disease;
V. cholerae express cholera toxin (CT), an ADP-ribosylating enzyme, at the intestinal surface, and the actions of this toxin on intestinal epithelial cells lead to the large-volume secretory diarrhea characteristic of cholera (
7). Despite this, immune responses that target CT do not provide meaningful protection against cholera (
8).
An
in vitro vibriocidal assay is currently our best predictor of protection against cholera; however, the vibriocidal response appears to be a surrogate marker of an as-yet-to-be-identified mucosal antibody response(s) (
7). We have shown that the vibriocidal response largely targets the O-specific polysaccharide (OSP) of
V. cholerae (
9). Moreover, we found that OSP-specific antibody and memory B cell responses correlate with protection against cholera in household contacts of cholera index patients in Bangladesh (
10). In North American recipients of an oral cholera vaccine, OSP-specific antibody responses correlate with protection against cholera in challenge studies (
11).
How OSP-specific antibodies protect against
V. cholerae in the intestinal lumen is currently unclear. Possible mechanisms include direct bactericidal, enchaining, or agglutinating activity (
12). However, we hypothesized that inhibition of motility could be a potential mechanism as well, a possibility supported by previous work (
13–16).
V. cholerae is a highly motile bacterium that has a single polar flagellum sheathed in lipopolysaccharide (LPS) displaying OSP, and
V. cholerae motility correlates with virulence (
17–19). Furthermore, several
in vitro studies have shown that antibodies targeting
V. cholerae LPS impede
V. cholerae motility (
13–16), and studies in suckling mice have suggested that the impedance of motility by anti-OSP antibodies provides protection in this model (
13,
14,
16,
20,
21). Here, we used antibodies recovered from humans surviving cholera in Bangladesh, including monoclonal antibodies cloned from plasmablasts homing to the intestinal mucosa (
22), to investigate the role of anti-OSP antibodies in protection against cholera. Our findings suggest that the inhibition of motility by the bivalent binding of anti-OSP antibodies contributes to immune protection from cholera.
DISCUSSION
We found that adult humans recovering from cholera develop antibody responses that inhibit
V. cholerae motility. These antibodies, which impact motility at subagglutinating concentrations of antibody, target
V. cholerae OSP and may mediate their effect via cross-linking. This work extends previous animal and
in vitro LPS-based analyses (
13–16).
V. cholerae motility is associated with pathogenesis (
23).
V. cholerae has a single sheathed polar flagellum and swims rapidly in an arcing pattern that can be visualized by dark-field microscopy of freshly passed cholera stools. This
V. cholerae distinctive swimming pattern can be used to rapidly identify
V. cholerae in the stool of potentially infected patients via dark-field microscopy and is referred to as “shooting star” motility (
24). As shown in our videos (see
Movies S1 to S3 in the supplemental material), the impact of the anti-OSP antibody on
V. cholerae shooting star motility is significant. Motility is involved not only with horizontal movement of
V. cholerae in the intestine (along the axis from stomach to colon) but also vertical localization in intestinal tissue (most importantly, along the axis created by villi and crypts in the small intestine) (
25). Once ingested, the few
V. cholerae cells that survive the acidic environment of the stomach enter the small intestine, and a subset swim into the lower villi and intestinal crypts (
25). Motility is required for this vertical localization (
25).
V. cholerae motility is then involved in the initial penetration of intestinal mucin that overlays the intestinal epithelium (
26). Following this initial penetration,
V. cholerae cells become nonmotile (with many bacteria losing their flagella), transit the mucin (in a poorly understood process), adhere at the intestinal epithelial surface, where they form microcolonies, and secrete cholera toxin (
26). This transition from a motile bacterium to an adherent microaggregated form at the intestinal surface is closely controlled by an interplay of quorum sensing by LuxO and HapR, motility, and
V. cholerae virulence networks (
26–33).
Following microcolony formation and the actions of cholera toxin on the intestinal epithelial cell leading to fluid secretion into the intestinal lumen, microcolonies break apart, and
V. cholerae cells regain their motility in a poorly understood process. This “mucosal escape response” involves RpoS-dependent downregulation of virulence genes and upregulation of motility genes (
34).
V. cholerae motility facilitates the release of organisms from microcolonies (
35). Detached organisms are then flushed/swim into the intestinal lumen and are passed into the environment in the rice water stools characteristic of cholera. These highly motile organisms have high transmission potential, a state that has been termed the “hyperinfectious state” (
36,
37).
Based on this sequence of events, inhibition of
V. cholerae mobility could impact
V. cholerae pathogenesis at a number of critical points. Inhibition of motility could alter intestinal passage, reducing the number of bacteria that move horizontally through the intestinal tract. It could also impact the ability of
V. cholerae to reach optimal ecologic niches in the lower third of intestinal villi and crypts and could affect the ability of
V. cholerae to penetrate the protective mucin layer overlying the intestinal epithelium. Here, using the suckling mouse model, we found that the largest impact of anti-OSP antibodies was in the proximal small intestine, the region with the thickest mucin layer (
25). Given the prominent interactions among the
V. cholerae quorum sensing, motility, and virulence networks, the inhibition of motility by anti-OSP antibodies could modify the pathogen’s intestinal colonization, toxin expression, and virulence. Altered
V. cholerae motility could also affect the ability of microcolonies to break apart and the ability of
V. cholerae to reenter the intestinal lumen and be flushed from the patient, thus affecting the hyperinfectious and transmission potential of passed
V. cholerae.
Our findings suggest that the effect of OSP-specific antibodies on
V. cholerae motility requires at least bivalent binding of antibody. Fab fragments of polyclonal antibodies retained their capacity to bind OSP but did not impede
V. cholerae motility. We also found that convalescent-phase IgM and IgA antibodies had a greater effect on
V. cholerae motility than IgG; this may be secondary to the pentavalent and dimeric structures of IgM and IgA, respectively, which markedly increase their valency. Other groups have assessed the impact of murine antibodies on
V. cholerae motility with variable results. Analysis of Fab fragments generated from polyclonal antibody generated in mice vaccinated against
V. cholerae suggested that elimination of multivalency eliminated the effect of the antibodies on
V. cholerae motility (
16). However, Fab fragments generated from a number of murine monoclonal anti-LPS antibodies apparently affect
V. cholerae motility (
14,
15,
38).
Our observations may help explain this discrepancy. Despite clearly demonstrating complete cleavage of antibody and removal of the Fc (fragment, crystallizable) fraction, we were unable to prevent aggregation of Fab fragments of human monoclonal antibodies under any physiologic nonreducing condition at relevant concentrations. We did not observe aggregation in Fab fragments generated from human polyclonal antibodies, which were used in our analyses. The propensity of fragmented monoclonal antibodies to form aggregates, presumably due to their inability to form proper disulfide bonds, was described previously (
39).
The requirement for multivalent binding to affect
V. cholerae motility suggests at least two possible mechanisms for the OSP-specific effect on
V. cholerae motility, depending on the concentrations of bacteria and antibody and the length of interaction time. The first requires sufficient bacteria, antibody, and time and involves interbacterial cross-linking of bacteria, i.e., agglutination. The second involves only the concentration of antibody and impact on individual bacteria. Our results strongly suggest that both conditions can impact
V. cholerae motility, and in the human intestine, both conditions could exist. Not surprisingly, we were able to show an impact on
V. cholerae motility under agglutinating conditions. More interestingly, we demonstrated the impact of human anti-OSP antibodies at antibody concentrations less than those associated with agglutination, and we were able to confirm inhibition of motility at the single bacterial level by microscopy. How anti-OSP antibodies could affect
V. cholerae motility at subagglutinating concentrations is currently unclear. One possibility is flagellar tethering or kinking due to antibody-mediated OSP cross-linking (
14). We did indeed detect flagellar tethering using electron microscopy, although we also observed occasional tethering with control antibody as well. Other mechanisms of action are also possible. For instance, the colonization defect of the flagellated but nonmotile
motB mutant strain in comparison to that of the wild-type strain was not completely reversed when both were exposed to anti-OSP antibody, suggesting that the genetically nonmotile strain may be less able to respond to intestinal environmental signals and activate colonization and virulence cascades (
40). An effect on transcriptional or protein expression in the setting of OSP-specific binding is also possible (
41).
A limitation of our study was our inability to assess IgA and IgM fractions due to the small amounts recovered from polyclonal serum.
V. cholerae is a noninvasive mucosal infection, and IgM and IgA are thought to be the predominant antibodies involved in mediating protection (
11). We were, however, able to include analysis of monoclonal antibodies cloned from mucosa-homing plasmablasts in patients with cholera (
22), and analyses comparing the impacts of IgM, IgA, and IgG isotype anti-OSP antibodies on
V. cholerae motility are planned.
Despite this limitation, our results show that humans surviving cholera develop antibodies that inhibit V. cholerae motility through binding of the OSP component of LPS, under both subagglutinating and agglutinating conditions. These data build upon a growing body of evidence that OSP-specific antibody responses mediate protection against cholera and suggest that such antibodies may act by mechanisms other than agglutination. Finally, our findings also support a novel mechanism of antibody-mediated protection through direct inhibition of bacterial motility.
MATERIALS AND METHODS
Study subject selection and sample collection.
Venous blood was collected at the acute phase (day 2) and convalescent phase of infection (day 7) from 10 adult patients (age, 18 to 55 years) presenting to the International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b) hospital with culture-confirmed V. cholerae O1 El Tor Ogawa infection. Patients were treated with intravenous fluids and/or oral rehydration solution and antimicrobials at the discretion of the attending physician. This study was approved by the Research and Ethical Review Committees of the icddr,b and the human studies committee of Massachusetts General Hospital.
Bacterial strains.
The
V. cholerae O1 strains used in this study are listed in
Table S2 in the supplemental material (
42). Bacterial strains were grown at 37°C in Luria-Bertani (LB) medium with or without antibiotics as appropriate. Antibiotics and their concentrations were streptomycin (Sm) at a concentration of 100 μg/ml and kanamycin (Km) at 45 μg/ml. 5-Bromo-4-chloro-3-indolyl-β-
d-galactopyranoside (X-Gal) (200 μg/ml) plus isopropyl β-
d-1-thiogalactopyranoside (IPTG; 0.1 mM) were used for blue/white colony screening. For experiments with the rough strain, bacteria were resuspended in 0.5% NaCl to prevent autoagglutination after growth in LB medium.
Purification of antibody isotypes from plasma.
IgG was purified from heat-inactivated human plasma using protein G HP spin trap columns (GE Healthcare) and then incubated with CaptureSelect IgA affinity matrix (Thermo Scientific) to remove contaminating IgA per the manufacturer’s instructions. IgA and IgM antibodies were then subsequently purified by sequentially running the flowthrough from the protein G column onto columns loaded with CaptureSelect IgA and IgM affinity matrix (Thermo Scientific). The quantity and purity of IgG, IgA, and IgM isotype fractions were determined by enzyme-linked immunosorbent assay (ELISA) using standard curves of ChromPure human IgG, IgA, and IgM (Jackson ImmunoResearch).
Fab fragmentation of plasma IgG.
Fab fragments of IgG were generated using the Fab preparation kit (Pierce) per the manufacturer’s instructions. Briefly, immobilized papain was used to digest the desalted and purified IgG, and the resultant Fab fragments were purified using a Protein A Plus spin kit (Pierce). The purity of the Fab fragments was assessed by Western blotting with 100 ng of sample per well and direct probing with anti-IgG F(ab)2 and anti-IgG Fc conjugated to horseradish peroxidase (Jackson ImmunoResearch).
Monoclonal antibody production.
Recombinant human monoclonal antibodies were generated from the cholera-induced day-7 plasmablast population of
V. cholerae-infected Bangladeshi patients by single-cell expression as previously described (
22). For this analysis, we used the following previously characterized human IgG monoclonal antibodies: high-affinity anti-OSP G1 (CF21.2.G01), low-affinity anti-OSP A4 (CF21.1.A04), low-affinity anti-OSP B4 (CF21.1.B04), and anti-FlaA-flagellin AT11 (AT11.1.B12) (
22).
Generation of V. cholerae O1 OSP.
OSP from
V. cholerae O1 Ogawa strain PIC158 was purified and conjugated to bovine serum albumin (BSA) as previously described (
43).
Vibriocidal assays.
Vibriocidal assays were performed on heat-inactivated plasma samples as previously described (
44), and the target strain
V. cholerae O1 Ogawa O395 was used. Heat-inactivated plasma and exogenous guinea pig complement (EMD Millipore) were incubated with the target strain. Vibriocidal titers were defined as the reciprocal of the highest serum dilution resulting in a 50% reduction in optical density at 600 nm (OD
600) compared to that of no serum controls.
OSP ELISA.
OSP responses were measured using an ELISA as previously described (
44). Microplates were coated with 100 ng/well of O1 Ogawa OSP-BSA. Samples were added to plates (plasma, 1:50; purified IgG, 10 μM; purified Fab, 1 μM), and IgG responses were detected with goat anti-human IgG conjugated to horseradish peroxidase [anti-IgG F(ab)
2 for Fab] (Jackson ImmunoResearch). IgM and IgA anti-OSP reactivity of plasma was measured using goat anti-human IgM and IgA conjugated to horseradish peroxidase (Jackson ImmunoResearch). Peroxidase activity was measured with the substrate 2,2-azinobis (ethylbenzthiazolinesulfonic acid) for plasma,
o-phenylenediamine for purified IgG/M/A, and SuperSignal West Femto maximum sensitivity substrate (Thermo Scientific) for purified Fab fragments.
Agglutination assay.
Agglutination assays were performed as previously described with modification to match conditions used in assessing motility via high-speed microscopy (
22). Briefly,
V. cholerae O1 classical Ogawa strain O395 was grown to mid-log phase in bovine heart infusion medium. Bacteria were pelleted, washed twice with phosphate-buffered saline (PBS), and diluted to an OD
600 of 0.1 with PBS. In a U-bottom microtiter plate blocked with BSA, 25 μl of bacterial samples was mixed with equal volumes of day-7 patient samples to a final starting dilution/concentration of plasma 1:8; purified IgG at 2.5, 1, or 0.5 μM depending on sample availability, purified IgG Fab at 1 μM, or monoclonal antibody starting at a concentration of 2.048 μM (0.320 mg/ml). The minimum agglutinating titer was determined by serial 2-fold dilutions of antibody sample. Plates were incubated at room temperature for 5 min and imaged using a UV imaging system (ChemiDoc; Bio-Rad) to assess agglutination. If no agglutination was evident, the value was marked as greater than the most concentrated value.
Viability assay.
V. cholerae O1 classical Ogawa strain O395 was grown to mid-log phase in LB medium. Bacterial samples, adjusted to an OD600 of 0.1, were added to equal volumes of PBS or patient’s sample (heat-inactivated convalescent plasma or purified IgG for final antibody concentration of 1:256 for plasma and 0.25 μM and 2.5 μM for purified IgG). The mixture was incubated for 5 min at 37°C. The effect of the antibody on bacterial viability was determined by serially diluting samples 10-fold and enumerating bacteria on LB plates. Experiments were performed in quadruplicates.
Motility inhibition assay.
V. cholerae motility was assessed using a modification of a previously reported approach (
13). Briefly,
V. cholerae O1 O395 were grown to mid-log-phase in LB broth, and the OD
600 was adjusted to 0.1. This OD was found to approximate 3 × 10
6 CFU/ml. Heat-inactivated convalescence antibodies were mixed with bacterial sample at a 1:1 dilution of acute or convalescence antibodies (final antibody concentrations of plasma and purified IgG were at subagglutinating levels: 1:256 (plasma), 0.25 μM [polyclonal IgG], and 0.5 μM [Fab generated from purified polyclonal IgG]), and 10 μl of the mixture was placed on a standard glass slide. Concentration dependence was also assessed for purified polyclonal IgG and monoclonal antibodies using 10-fold serial dilutions (2.5 to 0.025 μM for purified IgG; 2.5 to 0.00025 μM for monoclonal antibodies) using slides blocked with BSA. Motility inhibition of pooled samples of IgM and IgA was assessed at 0.25 μM; sample volumes were too low to assess motility individually for each patient using these antibody isotypes. PBS mixed 1:1 with bacteria was used as a negative control. Slides were incubated for 5 min at 37°C before visualization of motility by dark-field microscopy (Nikon Eclipse Ti-E inverted microscope stand, Nikon Plan-Fluor 40×/0.75 lens objective, Nikon TI-DF dark-field dry condenser [numerical aperture, 0.95 to 0.80], Hamamatsu ORCA-ER camera, and MetaMorph imaging software). Images were taken at 400-ms exposure time over 100 frames with readout time frame of 111.72 ms (movie play time of 1/30th s per frame). Frames 1, 50, and 98 were frozen as still shots, and bacteria were counted as lines (motile bacteria) or dots (nonmotile bacteria). Results were expressed as percent motile (motile/total bacteria) and averaged across the three screen shots. To confirm that the antimotility effect was secondary to OSP-specific antibodies, we performed experiments with antibody samples that were first adsorbed with OSP-BSA. Antibody samples were first incubated overnight with OSP-BSA before mixing with the bacterial sample for microscopy. Adsorption was performed using a 1:100 molar ratio (antibody/OSP) for purified antibody, and 250 μg of OSP-BSA for diluted plasma samples. PBS mixed with OSP-BSA was used as a negative control. All assays were performed in at least triplicates.
Electron microscopy.
Scanning electron microscopy (SEM) was performed at the Harvard University Center for Nanoscale Systems (CNS) using a FESEM Supra55VP microscope with an SE2 laser. V. cholerae O1 O395 was imaged in the presence or absence of anti-flagellin monoclonal antibody (AT11, 0.025 μM) at a magnification range of 45,300 to 46,700 or anti-OSP monoclonal antibody (G1, 0.005 μM) at a magnification range of 63,800 to 66,500.
Neonatal challenge.
The mouse neonatal
V. cholerae challenge assay was used to assess protection afforded by OSP-specific antibodies as previously described (
6). Briefly, 3- to 5-day-old CD-1 mice were separated from their dams for 3 h and then orally inoculated with 50 μl of a mixture containing 10
9 V. cholerae O1 Inaba C6706 and PBS or anti-OSP monoclonal antibody (G1) at a final concentration of 0.25 μM. This strain was used due to the availability of mutant strains used in the competition assays (see below). Mice were then housed away from the dams at 30°C and monitored for survival for 30 h. All mouse studies were approved by the MGH Institutional Animal Care and Use Committee.
Competitive colonization assay.
A single colony of each V. cholerae strain was inoculated into LB medium with Sm for C6706lacZ− (referred to below as wild type [WT]) and LB medium with Sm and Kan for C6706lacZ+ transposon mutants (rough [VC0244::Kanr] and nonmotile [motB::Kanr]) and incubated at 37°C overnight on a roller drum. Cultures were then regrown to mid-log phase, and the rough and comparator wild-type (WT) strains were mixed together 1:1 (105 organisms) and resuspended in 50 μl of LB medium without antibiotic mixed with PBS or 0.025 μM monoclonal antibody G1. A higher inoculum of the nonmotile strain was required to enable sufficient recovery after infection for enumeration and calculation of a competitive index (input, 106 mutant and 104 comparator). Three- to 5-day-old CD-1 infant mice were inoculated by oral gavage. After 20 h, the pups were sacrificed, and the proximal and distal thirds of the small intestines were isolated and mechanically homogenized in 5 ml of LB medium. Serial dilutions were plated on LB with Sm, X-Gal, and IPTG to enumerate bacteria and determine the input and output ratios of the comparator WT and competing rough or nonmotile strains. The competitive index was calculated as the output ratio of competing strain/WT strain divided by the input ratio of competing strain/WT. The data represent infections of multiple litters on different days that were pooled for analysis. All mouse studies were approved by the MGH Institutional Animal Care and Use Committee.
Statistical analysis.
Results were expressed as medians and compared by a Wilcoxon matched-pairs signed-rank sums test or Mann-Whitney U test for within group and between group comparisons as appropriate. Results of the survival curve analysis were compared by log rank testing. The threshold for statistical significance was a two-tailed P value of <0.05. Prism 6 was used for all statistical analyses.
ACKNOWLEDGMENTS
We thank the patients for participating in this study and the field workers and research staff at the icddr,b, Bangladesh, for their support and effort in sample collection and processing.
This work was supported in part by core grants to the icddr,b and by the Government of the People’s Republic of Bangladesh, Global Affairs Canada (GAC), Swedish International Development Cooperation Agency (SIDA), and the Department of International Development (UKAid). This work was also supported by grants from the National Institutes of Health (R01AI106878 to E.T.R., F.Q., and R.C.C.; U01AI058935 to E.T.R. and R.C.C.; R01AI103055 to J.B.H., J.W., and F.Q.; R01AI137164 to R.C.C. and J.B.H.; R01AI04247 to M.K.W.; R01AI135115 to D.T.L. and F.Q.; R01AI130378 to D.T.L. and T.R.B.; R01AI137127 to J.W.), the Fogarty International Center (FIC) and NIAID training grant in vaccine development and public health (TW005572 to A.A., M.H., K.I., R.B., M. Kamruzzaman, and T.R.B.), an FIC Global Emerging Leader Award (K43TW010362 to T.R.B.), the Robert Wood Johnson Foundation Harold Amos Medical Faculty Development Program (grant 72424 to R.C.C.), and National Institutes of Allergy and Infectious Diseases (training grant T32 AI007061 to A.S.R. and K22 AI104755 to C.S.F.). The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
E.T.R., R.C.C., J.B.H., and F.Q. designed the study. F.C., A.I.K., T.R.B., and F.Q. assisted in enrolling and collecting patient samples for this analysis. R.C.C., M. Kelly, J.M.T., M.H., K.I., A.A., L.M.M.-S., J.M.V., R.B., R.L.B., and M. Kamruzzaman performed the experiments and analysis. J.B.H., J.W., A.R., and R.C.K. generated and provided the human monoclonal antibodies. P.X. and P.K. produced V. cholerae OSP and OSP-BSA. M.K.W. produced V. cholerae C6706 strains. A.S.R., S.G., K.P.N., and C.S.F. performed electron microscopy. R.C.C., M. Kelly, J.M.T., M.H., K.I., A.A., J.M.V., D.T.L., R.C.L., A.W., F.C., A.I.K., S.B.C., T.R.B., J.B.H., F.Q., and E.T.R. discussed the results. R.C.C., M. Kelly, M.K.W., and E.T.R. contributed to the writing. All authors edited and approved the final manuscript.
We declare no competing interests.