Phase-Variable Heptose I Glycan Extensions Modulate Efficacy of 2C7 Vaccine Antibody Directed against Neisseria gonorrhoeae Lipooligosaccharide

Gonorrhea, caused by the Gram-negative diplococcus Neisseria gonorrhoeae (the gonococcus) is the most common bacterial sexually transmitted infection worldwide (second most common in the United States). Although most cases result in uncomplicated infections of the lower genital tract (urethritis in men and cervicitis in women), gonorrhea may sometimes lead to complications such as pelvic inflammatory disease and disseminated gonococcal infection. Serious sequelae of gonorrhea include infertility and ectopic pregnancy. Infected individuals who are asymptomatic or minimally symptomatic constitute an important reservoir for the transmission of infection.

Globally, ∼78 million new cases of gonorrhea occur annually (1). As a result of the emergence of antibiotic-resistant strains, including strains resistant to third-generation cephalosporins such as cefixime and ceftriaxone (2) and the lack of vaccines (3, 4) or novel anti-infective therapeutics, gonorrhea has become a major public health concern. A safe and effective vaccine would be a key step in curbing the spread of multidrug-resistant gonorrhea.

An obstacle to gonococcal vaccine development is the wide antigenic variation and/or variable expression of Ags that may elicit a protective response (e.g., pilin, opacity [Opa] proteins, porin B, lipooligosacharides [LOSs]) (3–5). Additionally, certain conserved Ags elicit nonprotective and, in some instances, subversive responses; an example of the latter is reduction modifiable protein (Rmp) (6).

Despite its phase-variable nature (7), gonococcal LOS has been considered as a potential vaccine Ag (8, 9). Men who were experimentally infected with N. gonorrhoeae were less likely to become infected upon rechallenge when they elicited an anti-LOS IgG response following the initial infection (10). Previous work by our group identified an epitope on gonococcal LOS that is recognized by a mAb called 2C7 (and therefore referred to as the 2C7 epitope) and was expressed on 94% of gonococci (64 of 68) recovered directly from human cervical secretions (11). Gonococcal infection in humans elicits an Ab response against the 2C7 epitope (11). Expression of a lactose residue from heptose (Hep) II is required for binding of mAb 2C7 (12). Addition of an α-linked Glc residue at the 3 position of HepII represents the first step in synthesis of the lactose extension from HepII and is mediated by the phase-variable LOS glycosyltransferase G (lgtG) (13).

Expression of LgtG is important for murine infection (14). Passive administration of mAb 2C7, as well as active immunization with a peptide mimic (mimotope) of the 2C7 epitope that was configured as a multiantigenic peptide on a poly-l-lysine backbone significantly shortened the duration and burden of infection in the murine vaginal colonization model of gonorrhea (14). Taken together, these data suggest that the 2C7 epitope represents a promising gonococcal vaccine candidate.

Phase variation of LOS glycan extensions is mediated by slipped-strand mispairing at homopolymeric tracts within the coding regions of the lgt genes; lgtA, lgtC, and lgtD modify glycan extensions from HepI; lgtG permits glycan extensions from HepII, as discussed above. Phase variation permits gonococci to express several distinct LOS structures that differ in their glycan composition (7, 15). Modulation of mAb 2C7 function by variations in HepI glycans has not been studied, yet it is an important consideration that may impact the efficacy of a 2C7 epitope–based vaccine and forms the basis of this study.

The neisserial strains used in this study are described in Table I. N. gonorrhoeae MS11 4/3/1 is a variant of MS11 VD300 with an isopropyl-β-d-thiogalactopyranoside (IPTG)–inducible pilE that controls pilus expression (16). UMNJ60_06UM was recovered in 2013 from a symptomatic male with urethritis in Nanjing, People’s Republic of China (17), and shows intermediate resistance to ceftriaxone (Etest minimum inhibitory concentration = 0.38 μg/ml and disc = 35 mm [sensitive ≥ 35 mm]). UMNJ60_06UM belongs to NG-MAST sequence type 3289 and MLST sequence type 1600.

Gonococcal strains were routinely cultured at 37°C in an atmosphere of 5% CO₂ on chocolate agar enriched with a chemically defined supplement (termed isovitalex) used as an additive for cultivation of nutritionally fastidious microorganisms. For growth in liquid culture, Morse A supplemented with Morse B and isovitalex were used (18). When used, antibiotics were added to GC agar plates at the following concentrations: erythromycin (Erm), 5 μg/ml; kanamycin (Kan), 100 μg/ml; and streptomycin (Sm), 10 mg/ml. To induce pilus expression and enable transformation, strain MS11 4/3/1 was cultured on GC agar plates supplemented with 0.25 mM (IPTG).

Escherichia coli Top10, XL-10 Gold, and INVαF′ (Invitrogen) were cultured on Luria–Bertani agar supplemented, as needed, with antibiotics at the following concentrations: ampicillin, 125 μg/ml; Kan, 50 μg/ml; Erm, 400 μg/ml; or chloramphenicol, 50 μg/ml. INVαF′, a naturally Sm-sensitive strain, was used for propagation of all plasmids containing the Erm^R-Sm^S cassette.

We created eight LOS mutants in MS11 4/3/1 (Table I), in which expression of the four phase-variable lgt genes (lgtG, lgtA, lgtC and lgtD [shown schematically in Fig. 1A]) was genetically fixed either on or off (or deleted).

lgtG was insertionally inactivated (G⁻ mutants; HepII unsubstituted) by amplifying lgtG::kan from FA19 lpt6A lptA lgtG (provided by Dr. William Shafer, Emory University, Atlanta, GA) using lgtG_F and lgtG_R primers (Supplemental Table I), and subsequently transforming MS11 4/3/1 with the purified PCR product. The Kan marker in lgtG in FA19 was derived from pCK49 (19). Inactivation of lgtG in Kan-resistant MS11 transformants was confirmed by PCR and DNA sequencing.

lgtG was fixed on (G⁺ mutants; HepII substituted with lactose) by first exchanging the wild-type lgtG with lgtG containing the ermC′-rpsL_F62 cassette (pRYGW2ES1; Supplemental Table II) that encodes resistance to Erm and sensitivity to Sm (20). Erm-resistant transformants were subsequently transformed with an lgtG-on construct (plgtG⁺; Supplemental Table II) in which the C₁₁ homopolymer had been changed to the non–phase-variable sequence 5′-CCCCTCCGCCA-3′. lgtG-on (G⁺) mutants were selected for resistance to Sm and screened for sensitivity to Erm (20).

HepI glycan mutants were made in both the MS11 G⁺ and G⁻ backgrounds by first exchanging each of the three phase-variable HepI lgt genes (lgtA, lgtC, lgtD) with an ermC′-rpsL_F62 cassette (plgtA-ES, plgtC-ES, plgtD-ES; Supplemental Table II), followed by transformation with the respective locked on (plgtA-on, plgtC-on, plgtD-on; Supplemental Table II), locked off (plgtC-off; Supplemental Table II), or the mutated (segment deleted) form of each gene (plgtA-del and plgtD-del; Supplemental Table II).

To insert the ermC′-rpsL_F62 cassette into each lgt, the homopolymeric phase variation sequence in each lgt was deleted and an SmaI restriction site was incorporated by overlap extension PCR (using the respective Fwd-Ext/Rev-Int and Fwd-Int/Rev-Ext primers; Supplemental Table I). Each mutated (homopolymer deleted and SmaI incorporated) lgt was amplified (Fwd-Ext and Rev-Ext; Supplemental Table I) and cloned (separately) into pCR2.1 TOPO TA (Invitrogen, Carlsbad, CA). The ermC′-rpsL_F62 cassette was extracted from pFLOB4300 (provided by Dr. Janne G. Cannon, University of North Carolina, Chapel Hill, NC) with pvuII and inserted into the SmaI site of each phase-variable HepI lgt gene (see plasmids plgtA-ES, plgtC-ES, and plgtD-ES; Supplemental Table II). Plasmids carrying ermC′-rpsL_F62 were maintained in the Sm-sensitive E. coli INVαF′ (Life Technologies, San Diego, CA).

Wild-type lgtA, lgtC, and lgtD were amplified from MS11 4/3/1 chromosomal DNA by PCR using the corresponding Fwd-Ext and Rev-Ext primers (Supplemental Table I) and the amplicons ligated with pCR2.1 TOPO TA cloning vector (Life Technologies) and transformed into chemically competent E. coli TOP10 (Life Technologies) per the manufacturer’s instructions (Supplemental Table II). Plasmids with lgtA, lgtC, and lgtD locked on and lgtC locked off were generated using a QuickChange Lightning multi site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA) with the corresponding mutagenic primers (Supplemental Table I) and transformed into chemically competent XL-10 Gold E. coli cells per the manufacturer’s recommendations (Supplemental Table II). Double digestion (with BbsI/SspI and NotI/SpeI, respectively) and subsequent ligation of plasmids with wild-type lgtA and lgtD yielded plasmids with deletion mutations in lgtA and lgtD (Supplemental Table II).

Replacement of the lgtC locked on in the mutant that expressed the 3-Hex HepI/lgtG⁺ LOS structure with lgtC locked off yielded the 2-Hex-HepI/lgtG⁺ mutant. Conversely, locking lgtC on in the 2-Hex-HepI/lgtG⁺ mutant yielded a lgtG⁺ mutant that expressed 3-Hex from HepI.

UMNJ60_06UM lgtA::kan was constructed as previously described (21). Inactivation of lgtA was confirmed by PCR and Western blot using mAb 3F11 (mAb 3F11 is described below). UMNJ60_06UM lgtA::kan 2-Hex and UMNJ60_06UM lgtA::kan 3-Hex were identified by Western blot; UMNJ60_06UM lgtA::kan 2-Hex reacted with mAb L8 but not mAb L1 (both mAbs are described below), and UMNJ60_06UM lgtA::kan 3-Hex reacted with mAb L1 (recognizes the globotriose Galα1,4-Galβ1,4-Glc structure, also called the P^k-like structure) but not mAb L8 (data not shown). All UMNJ60_06UM strains reacted with mAb 2C7 by Western blot (data not shown) and by flow cytometry (FCM; see 14Results).

Fresh chocolate agar plates were inoculated with bacteria harvested from cultures grown overnight for 15 h, and bacteria were grown for 6 h. LOS was extracted, de–O-acylated, and analyzed by mass spectroscopy as described previously (22).

Anti-LOS mAbs 2-1-L8 (henceforth referred to as mAb L8) (23), 17-1-L1 (referred to as mAb L1) (24), 3F11 (25), and 2C7 (11) have been described previously. A schematic of the epitopes recognized by these mAbs is provided in Fig. 1A. mAb 2C7 was purified from tissue culture supernatants over protein A/G (Pierce, Rockford, IL). Affinity-isolated goat anti-human factor H (FH) was prepared from anti-FH antiserum (Complement Technology, Tyler, TX) by passage over FH-Sepharose as described previously (26). Alkaline phosphatase–conjugated anti-mouse IgG and anti-mouse IgM, and FITC-conjugated anti-mouse IgG and anti-goat IgG were from Sigma-Aldrich. mAb 104 that binds to domains 1 and 2 of the α-chain of human C4b-binding protein (C4BP) (27) was provided by Dr. Anna M. Blom (Lund University, Malmö, Sweden). mAb 104 blocks C4BP function (27) and also blocks C4BP binding to gonococcal porin B (28) when preincubated with serum. However, mAb 104 does not displace C4BP already bound to the gonococcal surface and was used as the detection reagent for C4BP binding, as previously described (28). C3 deposited on gonococci was detected with FITC-conjugated anti-human C3c (AbD Serotec/Bio-Rad), which detects both C3b as well as iC3b, at a dilution of 1:100. To demonstrate that mAb 104 blocked C4BP binding to bacteria, complement was incubated with mAb 104 (9 μg mAb 104 was added to 30 μl complement) on ice for 10 min, added to bacteria. C4BP bound to bacteria was detected with anti-human C4BP mAb 67 (provided by Dr. Anna M. Blom) that recognizes domain 4 of the α-chain of C4BP, followed by anti-mouse IgG Alexa Fluor 647 (Sigma-Aldrich) both at a dilution of 1:100.

Protease K–digested bacterial lysates were separated on 12% Bis-Tris gels (Invitrogen) with MES running buffer (Invitrogen) and LOS was visualized by silver stain (Bio-Rad). LOS was transferred to polyvinylidene difluoride (Millipore, Billerica, MA) by Western blotting; membranes were blocked with PBS/1% milk for 1 h at 37°C and probed with tissue culture supernatants containing anti-LOS mAbs 2C7, 3F11, L1, and L8 (described above) for 15 h at 4°C, as described previously (29). mAb-reactive LOS bands were visualized with anti-mouse IgG alkaline phosphatase (for mAbs 2C7, L1, and L8) or anti-mouse IgM alkaline phosphatase (for mAb 3F11).

To ascertain whether a terminal hexosamine (in this instance, GalNAc) was present on the lgtD-on (D⁺) mutants, bacteria were suspended in water, frozen at −20°C, and thawed at 37°C to osmotically lyse them and treated with 10 U DNAse I in DNAse buffer (Ambion) for 60 min at 37°C. Treatment with DNAse I was carried out to reduce viscosity of the sample prior to electrophoresis. Proteins were digested with 1 mg/ml protease K (Calbiochem) in SDS (final concentration of 0.01%) for 1 h at 50°C. Protease K activity was destroyed by heating at 100°C for 20 min. Terminal N-acetyl hexosamine from LOS was released by treating the sample with 30 U β-N-acetylhexosaminidase in G2 buffer (both from New England Biolabs) for 15 h at 37°C. Samples were electrophoresed on a 16.5% Tris-Tricine gel (Bio-Rad) at 100 V at 4°C and LOS was visualized with silver staining as described above.

Blood was obtained from human volunteers (informed consent approved by the University of Massachusetts Institutional Review Board) and serum was immunodepleted of IgG and IgM by passage over protein A/G plus agarose (Pierce) and anti-human IgM agarose columns (Sigma-Aldrich) to prepare complement (30). The flow-through was spin concentrated, equilibrated with PBS/0.1 mM EDTA, and sterilized by passage through a 0.22-μm filter (Millipore). Hemolytic activity was determined using a total hemolytic complement kit (Binding Site, Birmingham, U.K.). FCM using FITC-conjugated anti-human IgG and anti-human IgM (Sigma-Aldrich) showed no detectable IgG or IgM binding in the depleted serum to strains that were used in experiments. Ab-depleted serum (henceforth referred to as complement) was aliquoted and stored at −80°C until use. In some experiments C4BP function and binding to gonococci was blocked by adding mAb 104 (28, 31) to complement (30 μg mAb 104/100 μl complement).

FCM was used to measure binding of mAb 2C7, C4BP, and deposition of complement C3 to bacteria as described previously (32–34). All Abs were diluted in HBSS containing 2 mM each of Ca²⁺ and Mg²⁺ (HBSS⁺⁺). Data were collected from a BD LSR II or FACSCalibur instrument (BD Biosciences, Franklin Lakes, NJ) and analyzed using a FlowJo analysis software program (version 7.2.5; Tree Star, Ashland, OR).

Serum bactericidal assays were performed as described previously (18, 29). Briefly, bacteria harvested from an overnight culture on chocolate agar plates were repassaged onto fresh chocolate agar and grown for 6 h at 37°C in an atmosphere of 5% CO₂. Approximately 2000 CFU gonococci in HBSS⁺⁺ were incubated with complement (concentration specified for each experiment) either in the presence or absence of mAb 2C7 (concentration specified for each experiment). In some experiments, C4BP function was blocked by preincubating complement with 30 μg/ml mAb 104 as described above. Final bactericidal reaction volumes were maintained at 150 μl. Aliquots of 10 μl were plated onto chocolate agar plates in duplicate at the beginning of the assay (t₀) and again after incubation at 37°C for 30 min (t₃₀). Survival was calculated as the number of viable colonies at t₃₀ relative to t_0.

Human neutrophils were isolated from human blood over a Percoll gradient, and opsonophagocytosis assays were performed using freshly isolated IL-8–primed adherent neutrophils as previously described (35). Briefly, bacteria were incubated with mAb 2C7 (4 μg/ml) and/or human complement (20%), or with HBSS⁺⁺ alone (controls) for 15 min at 37°C to permit IgG binding and C3 deposition. Reaction mixtures were added to IL-8–primed, adherent polymorphonuclear neutrophils (PMNs) at a multiplicity of infection of 1:1 and centrifuged at 400 × g for 4 min at 10°C to achieve synchronous infection (35). Cells were washed once with PBS/0.5% BSA, placed into RPMI 1640 with 10% heat-inactivated FBS, and warmed to 37°C. Cells were washed and lysed using 1% saponin in PBS at 0 min (taken immediately after the 10°C centrifugation step), and parallel wells were similarly treated at 60 min, serially diluted in GC broth, and plated to determine viable CFUs. Survival was expressed as the percentage of CFUs at 60 min relative to CFUs at 0 min.

Comparisons between two groups were made using the two-tailed unpaired t test. One-way ANOVA was used to compare multiple groups; pairwise comparisons were made by a Tukey post hoc test, whereas comparisons with a control group were made by a Dunnett test. Two-way ANOVA was employed to compare groups when time or concentrations were independent variables.

A schematic of potential gonococcal LOS structures, the relevant enzymes involved in biosysthesis of the outer core, and the specificity of anti-LOS mAbs used to characterize LOS glycan extensions are shown in Fig. 1A. Phase-variable expression of lgtA, C, and D leads to variation in the HepI glycan extensions: HepI 2-Hex (lgtA, C, and D all off), HepI 3-Hex (lgtA off, lgtC on, and lgtD on or off; expression of lgtD is extraneous in an lgtA off background), HepI 4-Hex (lgtA on, lgtC and D off), and HepI 5-Hex (lgtA and D on, lgtC off). Phase-variable expression of lgtG controls expression of lactose on HepII. To investigate the role of HepI glycan extensions on the function of mAb 2C7, we constructed a series of mutants in the background of MS11 4/3/1 in which the phase-variable lgt loci (lgtA, C, D, and G) were genetically fixed either on or off. Lgt loci were fixed on by mutating the repetitive homopolymeric sequence found in each gene such that the homopolymer was removed but the coding sequence was not altered, as previously described (36). The lgt loci were fixed off by deletion (lgtA, lgtD), insertional inactivation (lgtG), or by removing the homopolymeric sequence and inserting stop codons in all three reading frames (lgtC).

The LOS structures expressed by individual mutants were characterized by Western blotting using the anti-LOS mAbs described in Fig. 1A; relative masses of the LOSs were determined by SDS-PAGE (Fig. 1B). For simplicity, we refer to the mutants used in this study by their longest predicted HepI structures assuming activity of all expressed Lgt enzymes. The on and off status of lgtG is indicated as G+ and G−, respectively. For example, in lane 1 of Fig. 1B, the mutant with lgtA and lgtD on is expected to have a 5-Hex HepI structure (note the use Hex in the text includes both hexoses and N-acetyl hexosamines). lgtG in this mutant is fixed on, so the mutant is referred to as 5-Hex/G⁺. This simplified designation for each mutant is provided at the bottom of Fig. 1B and in Table I.

Note that fixing an lgt on does not ensure that all of the LOS displayed on the bacterial surface will be substituted with the glycan added by the encoded Lgt enzyme(s) because transport of incomplete LOS molecules to the outer membrane from the site of assembly on the cytoplasmic side of the inner membrane may occur prior to the addition of a glycan by all Lgts that are fixed on. The amount and efficiency of each Lgt enzyme will determine the ratio of complete to incomplete LOS expressed (37). An example of the transport of incomplete LOS shows that >50% of the LOS expressed by the two strains in which lgtD has been locked on (5-Hex/G⁺ and 5-Hex/G⁻) reacts with mAb 3F11 and represent 4-Hex structures with a terminal lactosamine (the lower, more prominent band in lanes 1 and 2 shown in the silver stain row in Fig. 1B), indicating that, despite expression of lgtD, most LOS in these mutants is exported to the surface prior to the addition of the terminal GalNAc to LOS. Another example is provided by mAb L8, which reacts specifically with LOS structures that contain a lactose on HepI and no glycans from the 3-position of HepII (i.e., lgtA off and lgtG off, respectively) (23). Thus, if all the LOSs expressed by mutants with lgtA off and lgtC and/or lgtG on were substituted with a terminal α(1,4)-linked Gal on HepI and/or a proximal Glc on HepII, these mutants should not react with mAb L8. In fact, mAb L8 reacted with all three mutants that had lgtA off and lgtC and/or lgtG on (lanes 5–7 in the L8 blot in Fig. 1B), indicating export of LOS structures to the surface in these mutants prior to addition of Glc on HepII by LgtG (lanes 5 and 7) and/or the distal α(1,4)-linked Gal on HepI (LgtC; lane 5). In contrast, fixing lgtA on (lgtA⁺) did not result in any detectable short LOS structures (no L8-reactive bands seen in lanes 2 and 4 of Fig. 1B; mutants with lgtA on and lgtG off), suggesting that LgtA efficiently added GlcNAc to the proximal lactose on HepI.

Mass spectrometric analysis confirmed loss of HepII glycan extensions in the lgtG off mutants and the presence of HepII glycans in the lgtG on mutants (Supplemental Table III). Mass spectrometry also confirmed that all mutants expressed the expected HepI glycan extensions shown in Table I, as well as incomplete structures as noted above. Further evidence that supported the presence of a terminal HexNAc residue in the 5-Hex/G⁺ and 5-Hex/G⁻ mutants was provided by β-N-acetyl hexosaminidase treatment, which resulted in almost complete disappearance of the highest molecular mass band on silver staining of their LOS (Fig. 1C).

Binding of mAb 2C7 (concentrations ranging from 0.1 to 10 μg/ml) to the LOS mutants was studied by FCM. The amount of mAb 2C7 bound to bacteria measured by FCM varied across the mutants (Fig. 2). The 2-Hex/G⁺ mutant showed maximum binding and 3-Hex/G⁺ the least; 4- and 5-Hex/G⁺ mutants bound intermediate amounts of 2C7. Binding of mAb 2C7 requires lactose extension from HepII. As expected, none of the lgtG deletion (G⁻) mutants showed binding above conjugate control levels (data not shown).

The ability of mAb 2C7 to kill each of the four G⁺ mutants was studied next. Bacteria were incubated with either 2 or 4 μg/ml mAb 2C7 and 20% human complement (normal human serum [NHS] depleted of IgG and IgM); survival at 30 min was measured by bacterial CFUs relative to CFUs at 0 min (Fig. 3). As expected, control reactions (no mAb 2C7 added) showed no killing (>100% survival). Additional controls with mAb 2C7 alone (no added complement) or heat-inactivated complement also showed no killing (data not shown). The 2-Hex/G⁺ mutant showed >90% killing in the presence of 2 μg/ml mAb 2C7; the 3-Hex/G⁺ mutant was fully resistant (>100% survival) to 4 μg/ml mAb 2C7. The 4-Hex/G⁺ and 5-Hex/G⁺ mutants showed an intermediate pattern; that is, resistance (≥50% survival) to 2 μg/ml 2C7, but sensitivity (<50% survival) to 4 μg/ml 2C7 (in this instance, >90% killing was observed). The bactericidal data followed a hierarchy similar to that seen with mAb 2C7 binding (Fig. 2).

Binding of the classical pathway inhibitor C4BP to gonococci modulates the efficacy of mAb 2C7 (29) and could have contributed to differences in susceptibility to mAb 2C7. We measured binding of C4BP to the four G⁺ mutants using heat-inactivated serum as a source of C4BP and found that all G⁺ mutants bound high and similar amounts of C4BP (Supplemental Fig. 1). These findings are consistent with prior data showing that MS11 and its LOS derivatives that expressed at least two hexoses from HepI bound C4BP well (28, 38).

The gonococcal genome contains >100 phase-variable genes. To confirm that differences in the binding of mAb 2C7 and killing between the 3-Hex/G⁺ and 2-Hex/G⁺ mutants were specifically related to LOS structure, lgtC was fixed on in the 2-Hex/G⁺ strain, permitting addition of Gal-α(1,4) to HepI (strain designated as 2→3-Hex/G⁺), and lgtC was fixed off in the 3-Hex/G⁺ strain, which blocked addition of Gal-α(1,4) to HepI (strain designated as 3→2-Hex/G⁺). The LOSs expressed by the mutants were verified by silver staining and with Western blots using mAbs L8 and L1 (Fig. 4A). The two mutants, 2→3-Hex/G⁺ and 3→2-Hex/G⁺, were next examined for their ability to bind and be killed by mAb 2C7 (Fig. 4B, 4C). The results recapitulated those seen with the 3-Hex/G⁺ and 2-Hex/G+ mutants, respectively.

We next asked whether serum resistance of the 3-Hex/G⁺ mutant could be overcome by either increasing complement concentrations or by blocking C4BP binding to bacteria. As shown in Fig. 5A, in the presence of 4 μg/ml 2C7, killing of 3-Hex/G⁺ was enhanced in a dose-dependent manner by increasing the concentration of complement. Complement alone (mAb 2C7 absent), even at the highest concentration tested (70%), did not result in killing. Similar to our prior observations with strain MS11 (28), mAb 104 blocked C4BP binding to the 3-Hex/G⁺ mutant (Fig. 5B), which resulted in enhanced killing by mAb 2C7 compared with control reactions that lacked mAb 104 (Fig. 5C). Thus, increasing complement activation on the 3-Hex/G⁺ mutant either by increasing the concentration of complement or by decreasing complement inhibition by C4BP overcame its serum-resistant phenotype.

C3 fragments, in particular iC3b, deposited on bacteria enhance opsonophagocytosis. mAb 2C7 did not promote direct killing by complement of the 3-Hex/G⁺ mutant in serum bactericidal assays that used 20% complement (Fig. 3). However, we reasoned that mAb 2C7–mediated C3 deposition on the 3-Hex/G⁺ mutant supports opsonophagocytic killing and constitutes a potential mechanism of protection by vaccine Ab.

Total C3 (C3b and iC3b) deposition on the 3-Hex/G⁺ was measured by FCM; the three other G⁺ mutants were included as comparators. Bacteria were incubated with either 2 or 4 μg/ml mAb 2C7 and 20% complement; C3 deposited at 15 and 30 min was measured. In the absence of mAb 2C7 there was minimal C3 deposition on all mutants [median fluorescence intensity (MFI) was <2-fold above baseline conjugate control levels (Fig. 6)]. As expected, the 2-Hex/G⁺ mutant that was highly susceptible to complement-dependent killing showed the most rapid accumulation and the highest levels of C3 deposition. An intermediate amount of C3 was deposited on the 4-Hex mutant. The 3- and 5-Hex/G⁺ mutants bound the least; there was a trend toward less C3 on 3-Hex/G⁺ compared with 5-Hex/G⁺, but the differences were not significant.

The Opa proteins of N. gonorrhoeae encompass a phase-variable family of proteins. Gonococci possess 11 opa genes and can express three or four Opa proteins simultaneously (39) that can engage CEACAM3 on PMNs and mediate opsonophagocytic killing independent of Ab and complement (40, 41). To address the potential role of mAb 2C7–dependent complement activation in facilitating opsonophagocytosis of the 3-Hex/G⁺ mutant, we recreated the 3-Hex/G⁺ LOS structure in the background of an Opa⁻ MS11 strain (42). The 3-Hex/G⁺ Opa⁻ MS11 mutant strain bound similar (low) levels of mAb 2C7 and was fully resistant (>100% survival) to killing by 4 μg/ml mAb 2C7 plus 20% complement (Fig. 7, bar at far left), analogous to the 3-Hex/G⁺ in MS11 with its native opa genes intact. In the presence of both mAb 2C7 and complement (bar to the extreme right), PMNs caused a 60% decrease in bacterial survival (p < 0.01 compared with the control reaction with bacteria alone plus PMNs [second bar from left]). Compared to the control with bacteria and PMNs (second bar from left), reactions that contained bacteria, PMNs, and 2C7 (third bar from the left) or bacteria, PMNs, and complement (fourth bar from the left) did not show increased killing.

To ascertain whether the decreased mAb 2C7 binding and increased resistance of the 3-Hex/G⁺ mutant was generalizable and not unique to strain MS11 alone, we identified 2-Hex and 3-Hex phase variants of an lgtA mutant (∆lgtA) of a minimally passaged clinical isolate called UMNJ60_O6UM. Two natural variants of UMNJ60_O6UM ∆lgtA were selected, one that expressed lactose on HepI (UMNJ60_O6UM 2-Hex; analogous with lgtC phase varied off and therefore did not react with mAb L1) and one that expressed 3-Hex P^k-like LOS on HepI (UMNJ60_O6UM 3-Hex; analogous with lgtC phase varied on and therefore reacted with mAb L1; see Ref. 43). Both variants had lgtG phase on and therefore expressed lactose from HepII. The two variants were examined for mAb 2C7 binding and killing in a complement-dependent bactericidal assay. UMNJ60_O6UM 2-Hex variant bound more mAb 2C7 than did the UMNJ60_O6UM 3-Hex variant (Fig. 8A). UMNJ60_O6UM 3-Hex was also more resistant to complement-dependent killing by mAb 2C7 (Fig. 8B). The 2-Hex variant was killed >99 and 100% by complement in the presence of 2 and 4 μg/ml 2C7, respectively; the 3-Hex variant survived ∼70 and ∼50% under similar conditions.

Unlike the MS11 3-Hex/G⁺ mutant that was fully resistant (>100% survival) to killing by even 4 μg/ml mAb 2C7 plus 20% complement (Fig. 3), the UMNJ60_O6UM 3-Hex variant was partly susceptible to killing under similar conditions. As noted above, C4BP binds to certain N. gonorrhoeae strains, including MS11, and promotes serum resistance (28) (Supplemental Fig. 1). To determine whether UMNJ60_O6UM bound C4BP, we compared C4BP binding to the two UMNJ60_O6UM LOS variants with the corresponding MS11 LOS mutants. Both UMNJ60_O6UM variants bound significantly lower amounts of C4BP than did the corresponding MS11 LOS mutants (Fig. 8C), thus providing a probable explanation for the greater sensitivity to complement of UMNJ60_O6UM variant 3-Hex compared with MS11 3-Hex/G⁺. Collectively, these data suggest that the 3-Hex HepI structure negatively affects mAb 2C7 binding and function.

The role of LOS glycan extensions in the pathogenesis of N. gonorrhoeae, including their role in immune evasion, is well recognized. Much attention has been directed to LOS that expresses the LNnT structure on HepI, which mimics host paraglobosides (44). Unsialylated LNnT interacts with the asialoglycoprotein receptor and facilitates adhesion of gonococci to male urethral epithelial cells (45). A lectin-like interaction between the terminal lactosamine residue of LNnT and gonococcal Opa proteins plays a role in intergonococcal adhesion and the degree of colony opacity (46). Gonococci possess a surface-exposed LOS sialyltransferase (47) that catalyzes the transfer of N-acetylneuraminic acid (Neu5Ac) from the donor molecule CMP-Neu5Ac present in host secretions (48, 49) onto the 3-position of the terminal Gal residue of LNnT. LNnT sialylation is involved in the inhibition of all three pathways of complement and enables gonococci to resist killing by natural IgG/IgM and complement in NHS, called serum resistance (50–52). Sialylation of gonococcal LNnT in vivo has been demonstrated by electron micrographs of organisms in human secretions (53). Schneider et al. (54) also demonstrated the importance of phase variation of gonococcal LOS and LNnT sialylation in humans. They inoculated male volunteers intraurethrally with a variant of strain MS11 that expressed a (nonsialylated) 2-Hex HepI structure predominantly. At the onset of symptoms, several days later, almost all men shed bacteria that expressed LOS with predominantly longer (including the 4-Hex HepI), sialylatable HepI structures (54). Recently, McLaughlin et al. (55) found that gonococci present in urethral exudates of infected men displayed an lgt genotype that predicted sialylation of terminal lactosamine; lgtA was in-frame, whereas lgtC and lgtD were out-of-frame in most cases. The importance of LOS sialylation in pathogenesis has also been demonstrated in the mouse vaginal colonization/infection model; gonococcal mutants that lack LOS sialyltransferase are outcompeted by their wild-type counterparts (32, 56). The efficacy of mAb 2C7 against phase-variant and sialylated bacteria has been demonstrated both in vitro (organisms grown in media containing 2 μg/ml CMP-Neu5Ac) (29) and in vivo in mice, where LOS sialylation occurs (56).

Diversity of surface Ags generated by phase variation confers a survival advantage to microbes and enables them to adapt to different niches in the host. Phase variation of LOS also modulates resistance of N. gonorrhoeae to killing by NHS, independent of LOS sialylation (57–59). Expression of GalNAc distal to LNnT (i.e., 5-Hex HepI; lgtA and lgtD on) permits binding of natural IgM present in NHS (60) and enhances bacterial killing by complement (57). Although the terminal GalNAc enhances killing by NHS, gonococci that possess LOS with this terminal residue interact with macrophage galactose-type lectin on dendritic cells. This may result in more pronounced Th2 and Th17 responses (61), instead of protective Th1 responses (4, 62). Expression of a truncated 3.6-kDa LOS (the 2-Hex [lactosyl] HepI), which also is a host glycan mimic, is also associated with increased resistance to NHS because most humans lack natural (IgG/IgM) Ab against this epitope (58).

Our studies used mAb 2C7 and NHS depleted of IgG/IgM in bactericidal assays. In contrast to previous studies where mutants with short HepI glycan extensions (e.g., lgtA off) were more resistant to killing by NHS (57, 59), the 2-Hex/G⁺ mutant in the present study was more susceptible to killing by mAb 2C7 and complement (i.e., NHS depleted of IgG/IgM) than were the three remaining mutants because it bound the most mAb 2C7, which resulted in overwhelming complement activation. By comparison, the 5-Hex/G⁺ strain, ordinarily more sensitive to killing by IgM in NHS (60), was relatively resistant to killing by mAb 2C7 plus complement because it bound less mAb 2C7. We replicated previous studies and showed killing (100% killing in 10% serum [IgG and IgM intact]) of the 5-Hex/G⁺ mutant; the 2- and 3-Hex/G⁺ strains were fully serum resistant (>100% survival in 10% NHS).

The 2C7 LOS epitope represents a promising vaccine candidate. Separately, we have shown that HepII glycan extension, required to generate the epitope, is also important in the mouse model of gonococcal colonization/infection, where an lgtG deletion mutant of strain FA1090 was shown to be less fit (14). A role for HepII glycans in gonococcal pathogenesis is supported by the observation that >95% (96 of 101) of minimally passaged clinical isolates reacted with mAb 2C7 (11). A contemporary analysis of >70 gonococcal isolates recovered from men with urethritis attending a clinic for sexually transmitted diseases in Nanjing, China, has substantiated these findings; all the recovered isolates bind mAb 2C7 (P.A. Rice, L.A. Lewis, and S. Ram, unpublished observations).

A potential reason for 2C7 vaccine resistance would be selection of LOS variants that show decreased binding of Ab. One explanation would be natural selection of variants expressing LOS with lgtG off. However, loss of LgtG expression may reduce fitness and therefore not be favored (14). The hypothesis addressed in our study was that other different HepI LOS structures may affect binding and function of mAb 2C7, an important consideration in predicting vaccine efficacy and coverage. Strains that expressed the P^k-like LOS structure (represented by the 3-Hex/G⁺ mutant) bound the least amount of mAb 2C7 and were relatively resistant to complement-mediated killing by mAb 2C7. With the exception of LgtC, which adds the terminal Gal on HepI of the 3-Hex/G⁺ mutant via an α-linkage, glycans added by enzymes encoded by the lgtA–E operon are β-linked (57, 63–65). The α-linked terminal Gal on HepI of the 3-Hex/G⁺ may hinder access of mAb 2C7 to its epitope.

Several microbes bind C4BP, an inhibitor of the classical pathway, to evade killing by complement (66). MS11 binds C4BP and mAb 2C7 must surmount this barrier to mediate killing through membrane attack complex insertion. The inhibiting role of C4BP was evident when C4BP binding to bacteria and C4BP function were blocked using mAb 104, resulting in increased killing of bacteria. Moreover, the 3-Hex variant of a clinical strain that bound low levels of C4BP (UMNJ60_06UM) was more susceptible to killing by mAb 2C7 than was the MS11 3-Hex/G⁺ that binds high levels of C4BP.

Despite the absence of killing through the membrane attack complex, mAb 2C7 deposited sufficient C3 on 3-Hex/G⁺ bacteria to enable human PMNs to decrease CFUs by >50%. Of note, maximal opsonophagocytic killing by PMNs required both Ab and complement; either one alone did not result in killing above baseline levels when bacteria only were incubated with PMNs. The neisserial P^k-like LOS structure can also be sialylated (43, 67). Unlike sialylation of LNnT LOS, sialylation of P^k-like LOS does not enhance FH binding to bacteria and confers resistance to only low, but not high, concentrations of NHS (43). Whereas McLaughlin et al. (55) found that lgtC was out-of-frame in all seven samples of N. gonorrhoeae obtained directly from male urethras, Mandrell (68) reported that as many as 36 of 70 (51%) of strains surveyed in vitro bound mAb 3D9, which reacts with the P^k Ag. The advantage conferred by the gonococcal P^k-like LOS in vivo remains to be elucidated.

How mAb 2C7 decreases gonococcal burden and duration of infection in vivo—that is, the specific contributions of direct complement-mediated killing, opsonophagocytosis, or other novel mechanisms of Ab-mediated clearance—remains to be elucidated. Notwithstanding the differences in direct killing of the mutants by mAb 2C7 and complement, enhanced C3 deposition occurred on all the HepI/G⁺ mutants; in particular, the 3-Hex/G⁺ mutant, which was not killed directly, was opsonophagocytosed and killed. Our findings support the inclusion of the 2C7 LOS epitope in a vaccine candidate against N. gonorrhoeae.

We thank Ashild Vik and Michael Koomey (University of Oslo, Oslo, Norway) for providing strain MS11 4/3/1, Daniel C. Stein (University of Maryland, College Park, MD) for the Opa⁻ MS11 mutant and F62 ΔlgtA lgtG⁺, William M. Shafer (Emory University) for providing FA19 lpt6A lptA lgtG, Anna M. Blom (Lund University) for mAbs 67 and 104, and Alison K. Criss (University of Virginia, Charlottesville, VA) for advice with opsonophagocytosis assays. We thank Sunita Gulati for mAb 2C7, and Nancy Nowak and Bo Zheng (all from the University of Massachusetts) for expert technical assistance. We also thank the Flow Cytometry Core Facility at the University of Massachusetts Medical School for assistance.

The authors have no financial conflicts of interest.