Underlying Glycans Determine the Ability of Sialylated Lipooligosaccharide To Protect Nontypeable Haemophilus influenzae from Serum IgM and Complement

Nontypeable Haemophilus influenzae (NTHi) efficiently colonizes the human nasopharynx asymptomatically but also causes respiratory mucosal infections, including otitis media, sinusitis, and bronchitis. The lipooligosaccharide (LOS) on the cell surface of NTHi displays complex glycans that mimic host structures, allowing it to evade immune recognition. However, LOS glycans are also targets of host adaptive and innate responses.

ABSTRACT

Nontypeable Haemophilus influenzae (NTHi) efficiently colonizes the human nasopharynx asymptomatically but also causes respiratory mucosal infections, including otitis media, sinusitis, and bronchitis. The lipooligosaccharide (LOS) on the cell surface of NTHi displays complex glycans that mimic host structures, allowing it to evade immune recognition. However, LOS glycans are also targets of host adaptive and innate responses. To aid in evasion of these responses, LOS structures exhibit interstrain heterogeneity and are also subject to phase variation, the random on/off switching of gene expression, generating intrastrain population diversity. Specific LOS modifications, including terminal sialylation of the LOS, which exploits host-derived sialic acid (Neu5Ac), can also block recognition of NTHi by bactericidal IgM and complement by mechanisms that are not fully understood. We investigated the LOS sialic acid-mediated resistance of NTHi to antibody-directed killing by serum complement. We identified specific LOS structures extending from heptose III that are targets for binding by naturally occurring bactericidal IgM in serum and are protected by sialylation of the LOS. Phase-variable galactosyltransferases encoded by lic2A and lgtC each add a galactose epitope bound by IgM that results in antibody-dependent killing via the classical pathway of complement. NTHi’s survival can be influenced by the expression of phase-variable structures on the LOS that may also depend on environmental conditions, such as the availability of free sialic acid. Identification of surface structures on NTHi representing potential targets for antibody-based therapies as alternatives to antibiotic treatment would thus be valuable for this medically important pathogen.

INTRODUCTION

Haemophilus influenzae, an obligate human pathogen, interacts with the host environment and adapts accordingly to maintain its upper respiratory tract niche. H. influenzae colonizes the nasopharynx and infects the respiratory mucosa to cause infections, including otitis media, sinusitis, pneumonia, and exacerbation of chronic obstructive pulmonary disease (COPD) (1, 2). Cell surface lipooligosaccharide (LOS), a short-chain form of lipopolysaccharide lacking the repetitive O-antigen carbohydrate extension, mediates immune evasion by a variety of Gram-negative bacterial pathogens (3). The LOS of H. influenzae is structurally diverse both within and between strains. Intra- and interstrain diversity can occur via phase variation and differential LOS gene composition, respectively. The majority of circulating H. influenzae strains are nonencapsulated, nontypeable strains (nontypeable H. influenzae [NTHi]) in which the LOS is particularly important in pathogenesis. NTHi strains are unaffected by the capsular conjugate vaccines against the type b strains, and a growing proportion of strains are ampicillin resistant via β-lactamase acquisition or through intrinsic mechanisms that increase resistance to cephalosporins and carbapenems (4).

Structural variability within the LOS is generally restricted to the outermost sugar extensions of NTHi, while inner structures are more frequently conserved, including lipid A anchored within the outer membrane and the core oligosaccharide, which contains a single 2-keto-3-deoxyoctulosonic acid (Kdo) linked to three heptose (Hep) residues that each serve as a site for extension by additional sugar moieties (5). In NTHi strain 375, chain extension from heptose III (HepIII) is initiated by the glycosyltransferase LpsA, adding a β-glucose in a 1,2 linkage, followed by lic2A-dependent 1,4 linkage of a β-galactose. Glycosyltransferases LgtC and LgtD further extend this chain by adding α1,4-galactose and β1,3-N-acetylgalactosamine (GalNAc), respectively. The LOS can be terminally capped by neuraminic acid, a nine-carbon sugar often referred to as sialic acid, which H. influenzae cannot synthesize and acquires from the host, likely from the heavily sialylated respiratory mucus (6). N-Acetylneuraminic acid (Neu5Ac) is the predominant form found in humans (7), where it is abundant on the surface of all cell types to mediate cellular interactions (8). After uptake from the extracellular milieu, Neu5Ac is either utilized as a carbon and energy source or activated by a CMP-Neu5Ac synthetase (SiaB) for incorporation by sialyltransferases into the LOS structure. Virtually all NTHi strains tested to date can include Neu5Ac in their LOS but may differ in their sialyltransferase repertoire and sialylation patterns (9). As illustrated in Fig. 1, the sialyltransferases encoded by lic3A and lic3B can link Neu5Ac to the terminal galactose added by Lic2A on HepIII. In NTHi 375, HepIII chain extension can depend on phase-variable expression of the enzymes encoded by lic2A, lgtC, lic3A, and lic3B.

FIG 1.

LOS of NTHi strain 375. The diagram is based on structural information described previously (9). Glc, glucose; Gal, galactose; Neu5Ac, N-acetylneuraminic acid; GalNAc, N-acetylgalactosamine; PC, phosphorylcholine; Hep, heptose; Kdo, 2-keto-3-deoxyoctulosonic acid. Competition between Lic3A and LgtC is indicated by a dashed line. Lic3B is a bifunctional α-2,3- and α-2,8-sialyltransferase present in some NTHi strains (indicated by dashed brackets), including strain 375. Certain conditions may result in the disialylation of LOS (49).

Sialic acid incorporation onto the bacterial surface is a form of molecular mimicry that produces glycan structures similar to those on host cells, effectively disguising the pathogen. Sialylated LOSs have been implicated in the pathogenesis of a number of bacterial species, including Neisseria gonorrhoeae, N. meningitidis, and Campylobacter jejuni, in which a common pathogenic role is to inhibit bacterial recognition and killing by the complement system (10). Earlier studies revealed that the sialylated LOS of N. gonorrhoeae, which contains some structures similar to those of the LOS of NTHi, can inhibit complement by several mechanisms, including inhibition of IgM-directed bactericidal complement activity and recruitment of factor H, a negative regulator of complement (11–15). While the pathogenic mechanisms mediated by sialic acid in NTHi are not fully understood, sialic acid has been implicated in biofilm formation and evasion of host immune defenses, including resistance to antibody-dependent complement activity (9, 16, 17). In vivo, sialylated LOS is critical for colonization of the middle ear and nasopharynx in a chinchilla model of otitis media (18) as well as in a gerbil middle ear infection model and a rat pulmonary model (17).

Several LOS modifications are known to contribute to the complement resistance of NTHi, and specific structures, including sialic acid, can inhibit the binding of complement proteins, such as C3b and C4b (19–23). Recently, Oerlemans et al. showed that NTHi can utilize sialic acid to evade IgM binding and the bactericidal effect of complement in pooled human serum (23). Natural IgM antibodies present in healthy individuals exhibit varied specificity against self and nonself glycans (24). Previous studies have indicated that the majority of naturally occurring bactericidal antibody against NTHi is specific to the LOS. While IgG antibodies against NTHi are present in most individuals, the majority of individuals lack abundant IgM against the LOS of wild-type (WT) NTHi (25). Nevertheless, bactericidal natural or naturally acquired serum IgM against NTHi mutants with truncated LOS structures has been detected (21, 26), and these antibodies may recognize NTHi expressing LOS structures modified by phase variation or in response to environmental signals encountered during infection.

Although several studies have investigated sialic acid-mediated complement resistance and antibodies targeting NTHi surface structures, the specific target structures on NTHi that are protected by sialylated LOS from binding by bactericidal IgM are not known. In this study, we delineate these LOS targets and demonstrate that protection of these epitopes from antibody binding contributes to sialic acid-mediated complement resistance. Identification of conserved targets of naturally occurring bactericidal antibodies can provide information for the development of potential therapeutic strategies to subvert immune evasion by NTHi.

RESULTS

Sialic acid enhances resistance to killing via the classical pathway of complement.

To evaluate the effect of sialic acid on complement activity, we performed serum bactericidal assays. Wild-type (WT) NTHi strains 375 and NT127, which differ in their cell surface structures, including the glycans of the LOS outer core (9, 27), were grown on chemically defined agar in the presence or absence of sialic acid (MIc^±SA) and incubated with either normal human serum (NHS) or NHS treated with MgCl₂ and EGTA (MgEGTA). The classical and lectin pathways require both magnesium and calcium, while the alternative pathway requires only magnesium. Adding MgCl₂ and EGTA to serum chelates calcium to selectively block the classical and lectin pathways while leaving the alternative pathway intact (28). The survival of NTHi 375 and NT127 under each serum condition is shown in Fig. 2. In the absence of sialic acid, the survival of NTHi 375 (Fig. 2A) was just below 10% (9.8% ± 1.5%) after exposure to NHS but was increased to ∼30% (27.5% ± 2.5%) when grown in the presence of sialic acid. However, when the classical pathway was inactivated with MgEGTA, there was nearly 100% survival and no statistically significant effect of sialic acid on serum resistance. Similar results were seen with NT127 (Fig. 2B), in which an even greater enhancement of survival by sialic acid in 2% NHS than that in strain 375 was shown but in which no differential effect by sialic acid or killing in MgEGTA-treated serum was still found. These results suggest that sialic acid enhances resistance when all pathways of the complement system are intact and, as previously shown, that the alternative pathway alone does not mediate bactericidal activity against H. influenzae (19, 29).

FIG 2.

Sialic acid enhances resistance to killing via the classical pathway. Wild-type NTHi strains 375 (A) and NT127 (B) were grown on MIc^±SA and assayed for serum sensitivity following incubation with 2% normal human serum (NHS) or 2%, 20%, or 50% NHS treated with MgEGTA (10 mM) for 30 min at 37°C. Percent survival is the ratio of the number of CFU recovered from serum-treated samples after 30 min to the number of CFU recovered from samples treated with each respective heat-inactivated (NHS^Δi) serum sample. The mean for duplicate samples is shown. Survival ratios were evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (**, P < 0.01; ***, P < 0.001; n.s., not statistically significant).

Sialic acid decreases binding by serum IgM.

The classical pathway is known to be required for efficient complement killing activity against NTHi and is strongly activated by antibodies bound to the bacterial cell surface. We tested IgM and IgG binding to the NTHi 375 WT grown on MIc^±SA. Sialic acid significantly reduced the amount of IgM bound to this strain (Fig. 3A), consistent with a previous report demonstrating that sialic acid uptake confers decreased IgM binding to NTHi (23). However, IgG binding was unaffected by the presence of sialic acid (Fig. 3B). To determine whether inhibition was specific to sialic acid addition to the LOS, we evaluated IgM binding to wild-type NTHi strains 375 and NT127 and their respective isogenic siaB mutants that lack CMP-Neu5Ac synthetase activity to sialylate their LOS. In contrast to sialic acid-mediated inhibition of IgM binding to wild-type strains, there was no difference in the binding of IgM to their respective siaB mutants when grown with or without sialic acid, and the binding levels under both conditions were similar to those for their parental strains grown without sialic acid (Fig. 4), supporting the conclusion that IgM inhibition can be attributed to the sialylation of NTHi LOS. Although the LOS structures of 375 and NT127 differ (9, 27), we still saw the equivalent inhibition of serum killing and IgM binding by sialic acid.

FIG 3.

Sialic acid decreases NTHi binding to serum IgM but not IgG. The NTHi 375 WT grown on MIc^±SA was incubated with 20% NHS^Δi for 30 min at 37°C, followed by detection via flow cytometry using anti-human IgM (A) or IgG (B) conjugated to FITC. The median fluorescence intensity (MFI) values are those obtained after the subtraction of the MFI values for samples with detection antibody only. The means for duplicate samples are shown. Statistical significance was evaluated by an unpaired, two-tailed, Student's t test (*, P < 0.05; n.s., not statistically significant).

FIG 4.

Inhibition of IgM binding by sialic acid is dependent on LOS sialylation. Wild-type NTHi strains 375 (A) and NT127 (B) and their respective isogenic siaB mutants that lack CMP-Neu5Ac synthetase activity were grown on MIc^±SA and incubated with 20% NHS^Δi for 30 min at 37°C. Binding was detected via flow cytometry using anti-human IgM conjugated to FITC (n = 3). Statistical significance was evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (**, P < 0.01; ****, P < 0.0001; n.s., not statistically significant).

The antigenic heterogeneity of the LOS of H. influenzae strains is well noted. To address the possibility that strain-specific antigens are involved in antibody binding and to determine if IgM inhibition by sialic acid is a common mechanism of protection, we measured binding to a diverse set of NTHi strains with different LOS structures. We observed statistically significant differential binding by IgM in all 4 additional strains tested, supporting a common role of sialic acid in protecting NTHi strains from classical pathway activity (Fig. 5).

FIG 5.

NTHi strains show sialic acid-dependent inhibition of serum IgM binding. Rd and a panel of the indicated wild-type NTHi strains were tested for IgM binding after growth on MIc^±SA. The statistical significance for each strain was evaluated by unpaired, two-tailed, Student’s t tests (*, P < 0.05; **, P < 0.01).

Sialic acid protects NTHi from antibody-mediated killing.

To correlate the levels of antibody binding with serum resistance, the NTHi 375 WT was assayed for survival after incubation with IgG/IgM-depleted human complement (HC) and heat-inactivated NHS (NHS^Δi) as a source of serum antibody. Our results indicate that without sialic acid, serum killing was enhanced by antibody addition compared to that achieved with HC alone, but upon sialylation, this enhancement was abrogated and the level of serum resistance was not significantly lowered by antibody addition (Fig. 6), indicating that sialic acid protects from antibody-dependent killing in serum, as expected. However, significant inhibition of complement activity by sialic acid in HC alone suggested potential inhibition of the alternative pathway independent of its effect on antibody-mediated classical pathway activation. Therefore, we measured C3 deposition on the 375 WT exposed to 20% NHS and 20% NHS treated with MgEGTA. The results in Fig. 7 confirm that sialic acid inhibits C3 binding in the presence of all three pathways and when only the alternative pathway is active.

FIG 6.

Sialic acid mediates protection from antibody-dependent killing by serum. The NTHi 375 WT was incubated with human complement (HC) serum depleted of IgG and IgM in the presence or absence of 20% NHS^Δi as a source of serum antibodies. The means for four independent samples are shown. Statistical significance was evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (**, P < 0.01; ****, P < 0.0001; n.s., not statistically significant).

FIG 7.

Sialic acid mediates protection from alternative pathway activity. The NTHi 375 WT grown on MIc^±SA was incubated with 20% NHS or 20% NHS treated with MgEGTA (10 mM) for 30 min at 37°C. The C3 deposited on the cells was detected via flow cytometry using anti-human C3c conjugated to FITC. The median fluorescence intensity (MFI) values are those obtained after subtraction of the MFI values for samples with detection antibody only. The means for three independent samples are shown. Statistical significance was evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (**, P < 0.01).

IgM targets a LOS structure protected by sialylation.

Studies on the sialic acid-mediated resistance of NTHi have not elucidated the mechanism of antibody inhibition. The LOS is a major target for complement binding that contains structures recognized by monoclonal antibodies (MAbs) and serum antibodies. We sought to identify the specific LOS epitope protected by sialic acid from binding by naturally occurring serum IgM. Based on the current structural model of the NTHi 375 LOS (Fig. 1), several candidate structures were considered. We constructed a lgtD deletion mutant that lacks the outermost N-acetylgalactosamine (GalNAc) and measured the effect of sialic acid on IgM binding (Fig. 8A). The mutant retained a significant amount of IgM binding relative to the WT in the absence of sialic acid and only a partial loss in the presence of sialic acid, suggesting that a major LOS antibody target was still present.

FIG 8.

Sialic acid-mediated inhibition of binding by serum IgM and anti-CD77 MAb to LOS epitopes. The NTHi 375 WT and the indicated isogenic mutants (the lgtD or lgtC mutant) were grown on MIc^±SA and incubated with 20% NHS^Δi, followed by detection via flow cytometry using anti-human IgM (A) or anti-CD77 MAb (B) conjugated to FITC. The lgtD mutant lacks a β-1,3-N-acetylgalactosaminyltransferase that caps HepIII with GalNAc. The lgtC mutant lacks an α-1,4-galactosyltransferase that adds a galactose to the HepIII-Glc-Gal structure. Monoclonal IgM antibody 5B5, specific for human CD77 (P^k), detects Glc-(β1,4)-Gal-(α1,4)-Gal. The MFI values for FITC antibody-only or cell-only controls were subtracted from the experimental samples. The means for duplicate samples are shown. Statistical significance was evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; n.s., not statistically significant).

The major oligosaccharide extension (HepIII-Glc-Gal-Gal) on NTHi 375 resembles the human P^k blood group antigen, also known as CD77 or globotriaosylceramide (Gb3) (9, 13). Commercially available anti-human CD77 IgM monoclonal antibody (MAb) clone 5B5 allowed us to confirm the expected display of the LOS globotriaose structure in wild-type strain 375 and our lgtD deletion mutant. We were also able to directly evaluate inhibition of this specific IgM monoclonal antibody by sialic acid (Fig. 8B). Anti-CD77 binding was equally high in the WT and the lgtD mutant in the absence of sialic acid, suggesting that the CD77 epitope is not blocked by the lgtD-dependent GalNAc structure. With sialic acid present, binding to the WT was significantly decreased to a level similar to that to the lgtC mutant grown with or without sialic acid, indicating that the majority of anti-CD77 IgM binding to the lgtC-dependent Gal-(α1,4)-Gal structure is inhibited by sialylation.

Sialic acid-mediated inhibition of IgM binding and serum killing of the lgtC mutant.

Because sialylation inhibited the binding of an IgM monoclonal antibody specific to the Gal-(α1,4)-Gal structure, we examined the binding of serum IgM to the lgtC mutant, which lacks the digalactoside extension, to determine if the outermost galactose within this extension is a possible target. Removal of this lgtC-dependent epitope significantly reduced IgM binding and reduced the protective effect of sialic acid on binding (Fig. 9A), and similar results were obtained with five single donor serum samples (see Fig. S1 in the supplemental material).

FIG 9.

Sialic acid protects the lgtC mutant from both IgM binding and complement killing. (A) The NTHi 375 WT and the lgtC deletion mutant were grown on MIc^±SA and tested for IgM binding following incubation with 20% NHS^Δi. Mean MFIs for replicate samples (n = 4) are shown. (B) Strains were assayed for survival following incubation with 1% and 2% NHS. Percent survival is the ratio of the number of CFU recovered from serum-treated samples at 30 min to the number of CFU recovered from samples treated with heat-inactivated serum (NHS^Δi). The mean ratio for triplicate samples is shown. MFI values and survival ratios were evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; n.s., not statistically significant).

Although antibody binding is a key step in classical pathway activation that can subsequently lead to killing via the lytic pathway, nonbactericidal antibodies in serum are capable of binding but not enhancing host-mediated responses to eliminate bacteria (30–33). In order to determine if increased serum resistance can be attributed to sialic acid-mediated inhibition of IgM binding to the lgtC-dependent Gal-(α1,4)-Gal structure, we tested serum bactericidal activity against wild-type strain 375 and the lgtC mutant (Fig. 9B). Interestingly, we found that in the absence of sialic acid, the survival of the lgtC mutant was less than half of that of the WT. Thus, despite higher levels of IgM binding to the WT than to the mutant, which lacks a major epitope, the mutant was more sensitive to killing by complement. More importantly, our results show that sialic acid-mediated protection from serum killing is retained in the lgtC mutant. Therefore, sialic acid may inhibit binding to additional, less strongly recognized bactericidal IgM epitopes or act on downstream complement activation steps, such as through inhibition of C3 or C4 binding to LOS targets (19, 34).

Lic2A is required for sialic acid-mediated serum resistance and inhibition of IgM binding in NTHi 375.

To address whether the binding of bactericidal IgM is determined by the underlying LOS structure added by Lic2A (Fig. 1), we tested the effect of sialic acid on the serum survival of a lic2A deletion mutant versus the survival of the WT (Fig. 10A). At the lower serum concentration (1%), the lic2A mutant exhibited increased survival relative to the WT in the absence of sialic acid, and sialylation did not further increase the survival of the lic2A mutant. At a higher serum concentration (2%), the survival of the lic2A mutant remained unaffected by growth in sialic acid. In addition, at the higher serum concentration, the survival of the lic2A mutant was greatly reduced relative to that of the WT, potentially because the Gal added by Lic2A is needed to block the underlying Glc residue, which is a previously demonstrated target of bactericidal antibody (IgM) in lic2A mutants (21), and as more serum is added, the benefit of losing the lic2A epitope is overcome. Since our results indicate that the lic2A gene is required for sialic acid-mediated resistance and adds an epitope that could be targeted by IgM, we measured IgM binding to the WT and the lic2A mutant in the presence and absence of sialic acid (Fig. 10B). As predicted, we saw a significant decrease in binding to the wild type in the presence of sialic acid, but sialic acid had no effect on binding to the lic2A mutant. Deletion of lic2A effectively reduced IgM binding to levels similar to those for the wild type with sialic acid. We conclude that Lic2A likely provides the sole sialylation site on the LOS of NTHi 375 and is required for the sialic acid-mediated serum resistance of NTHi 375.

FIG 10.

Lic2A is required for sialic acid-mediated serum resistance and inhibition of IgM binding in NTHi 375. (A) The NTHi 375 WT and lic2A deletion mutant (HepIII-Glc) grown on MIc^±SA were assayed for serum sensitivity following incubation with 1% and 2% NHS. Percent survival is the ratio of the number of CFU recovered from serum-treated samples at 30 min to the number of CFU recovered from samples treated with heat-inactivated serum (NHS^Δi). The mean for triplicate samples is shown. (B) The 375 WT and lic2A mutant grown on MIc^±SA were tested for IgM binding. The MFI values for FITC antibody-only controls were subtracted from those for the experimental samples (n = 2). Survival ratios and MFI values were evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (**, P < 0.01; ***, P < 0.001; n.s., not statistically significant).

Sialylation protects from bactericidal targeting of IgM epitopes dependent on lic2A and lgtC.

Serum absorption was performed to validate the presence of antibodies specific to the lic2A- and lgtC-dependent LOS epitopes that can be blocked from binding by sialic acid. Heat-inactivated serum was subjected to multiple rounds of absorption with either WT, lic2A mutant, or lgtC mutant strains grown on MIc agar without sialic acid (MIc^−SA). The resulting absorbed sera were then used as antibody sources to assess IgM binding and serum bactericidal activity in the presence of human complement (HC) for the 375 WT and isogenic lic2A and lgtC mutants grown in the absence or presence of sialic acid (Fig. 11). Figure 11A indicates that serum preabsorbed with the lic2A mutant, which does not deplete antibodies specific to lic2A- and lgtC-dependent structures, showed sialic acid-mediated inhibition of IgM binding to WT. IgM binding to the nonsialylated WT in serum preabsorbed with the lgtC mutant, which depletes antibody specific to the lic2A-dependent epitope, was reduced by ∼2.2-fold compared to the binding in serum absorbed with the lic2A mutant, but a significant amount of inhibition of IgM binding by sialic acid remained. As anticipated, serum preabsorbed with the WT strain, which depletes antibodies specific to both the lic2A- and lgtC-dependent structures, greatly reduced IgM binding to the WT grown on MIc^−SA (∼6.8-fold) compared to that with serum preabsorbed with the lgtC mutant, reducing binding to the level present on the WT grown in the presence of sialic acid. Together, these results indicate that the majority of protection from serum IgM binding by sialic acid occurs on epitopes generated by both LgtC and Lic2A.

FIG 11.

Sialic acid protects LOS epitopes added by Lic2A and LgtC from binding by bactericidal IgM. (A) Heat-inactivated serum was preabsorbed with antibody targeting LOS epitopes using the 375 wild-type (WT), lic2A mutant, and lgtC mutant strains grown in the absence of sialic acid (MIc^−SA). (Right) The diagram depicts differences on HepIII (Fig. 1) between absorbing strains. (Left) IgM binding to the 375 WT, lic2A mutant, and lgtC mutant grown on MIc^±SA was measured after incubation with each preabsorbed serum sample (n = 2). The MFI values for the FITC antibody-only controls were subtracted from those for the experimental samples (n = 2). (B) The 375 WT and lgtC mutant grown on MIc^±SA were assayed for survival after incubation with each indicated preabsorbed serum sample and 3% HC. Percent survival is the ratio of the number of CFU recovered from absorbed serum-treated samples at 30 min relative to the number of input CFU at time zero. The means for duplicate samples are shown. The lower limit of detection (LLD) was ∼0.20% survival. MFI values and survival ratios were evaluated by one-way ANOVA with Bonferroni’s multiple-comparison test (*, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001; n.s., not statistically significant).

To determine the relative contribution of antibody binding activities against each structure to serum bactericidal activity, the same absorbed serum was used to perform serum bactericidal assays with the WT and the lgtC mutant (Fig. 11B). Our results indicate that bactericidal antibodies specifically targeting lic2A- and lgtC-dependent structures remained in lic2A mutant-absorbed serum, resulting in the lowest survival of the WT and the lgtC mutant when grown in the absence of sialic acid. Absorption with the lgtC mutant removed antibodies targeting the lic2A-dependent epitope, while antibodies specific to the lgtC-dependent epitope remained. When the WT grown in the absence of sialic acid was incubated with lgtC-absorbed serum and lic2A-absorbed serum, survival decreased by 2.8 and 6.2 times, respectively, relative to its survival in serum absorbed with the WT, which removes antibodies against both structures. These results confirm that lic2A- and lgtC-dependent IgM epitopes on the LOS of NTHi 375 are targets of bactericidal antibodies that can be inhibited by sialic acid.

DISCUSSION

The purpose of this study was to elucidate the mechanism of classical pathway inhibition by sialic acid. Specifically, we investigated antibody-dependent complement activity and identified structures on the surface of H. influenzae that are targets for binding and subsequent killing by serum antibodies and complement. Antibody and the classical pathway of complement are required for the bactericidal effect of normal human serum on NTHi (21, 26, 27). The majority of naturally occurring antibody directing the complement-mediated killing of NTHi is specific to the LOS, and anti-LOS IgM titers correlate with increased serum bactericidal activity (25). A recent report showed that sialic acid uptake-dependent inhibition of IgM binding correlated with the reduced sensitivity of NTHi to classical pathway-mediated killing in NHS (23). Consistent with these observations, we confirmed that growth on sialic acid conferred to NTHi protection from killing by the classical pathway of complement (Fig. 2) and inhibited IgM binding in 6 of 6 NTHi strains tested (Fig. 4 and 5).

Despite the heterogeneity of the LOS composition and structure, all of these strains have lic2A and lgtC genes, which likely generate targets protected by sialic acid. However, it is important to note that the total serum resistance level of a strain does not appear to directly correlate with the amount of IgM binding that occurs, based on our data and others’ (18, 27, 29, 35). Antibodies directed against specific surface structures, including LOS extensions, either can be ineffective and nonbactericidal or can make resistant strains more sensitive to killing (30–33). This highlights the importance of identifying specific LOS targets that result in optimal killing.

Protection from IgM binding was dependent on the siaB gene, encoding CMP-Neu5Ac synthetase, demonstrating the requirement for sialylated LOS (Fig. 4). We found that serum antibodies enhanced the complement-mediated killing of NTHi lacking sialylated LOS; however, bactericidal activity against NTHi with sialylated LOS was unaffected by the addition of serum antibodies (Fig. 6). Interestingly, we also saw a significant difference in survival between sialylated and nonsialylated NTHi in the presence of human complement alone. To differentiate this effect from the potential inhibition of residual antibody in the complement source, we evaluated C3 binding in normal and MgEGTA-treated serum and found that sialic acid could inhibit the alternative pathway alone in the absence of classical pathway activation (Fig. 7), consistent with the findings of studies by Figueira et al. (19). Therefore, while sialic acid inhibits antibody binding, which is required for efficient complement-mediated killing of NTHi, it also inhibits C3 binding by an independent mechanism that may also decrease total complement activity. In this report, we focused on the classical pathway and extended these studies into the mechanism of complement resistance by identifying specific LOS targets that are protected by sialic acid from IgM recognition.

Using LOS mutant strains, we identified epitopes protected by sialic acid from binding by serum IgM. GalNAc is the outermost sugar added by LgtD to HepIII on the LOS of NTHi 375 (Fig. 1). Flow cytometric analysis revealed that deletion of lgtD reduced IgM binding compared to that for the wild type, but a significant protective effect of sialic acid was retained in the mutant (Fig. 8A). We performed additional experiments with a deletion mutant lacking LgtC, which is required for addition of an underlying galactose as part of a digalactoside extension. It was previously suggested that since the digalactoside [Gal-(α1,4)-Gal] structure added by Lic2A and LgtC is also present in humans as the P^k blood group antigen, it is not a target of human antibody (12, 20). However, Huflejt et al. conducted studies on the anticarbohydrate antibodies present in normal serum and concluded that antibodies to the P1 and P^k structures are among the most abundant natural antibodies present in the majority of individuals (24). Flow cytometry results indicated that overall IgM binding to the mutant was substantially reduced compared to that to the wild type and that there was a diminished effect of sialic acid, which led us to the conclusion that antibodies specific for the lgtC-dependent structure are abundant in serum and can be inhibited from binding NTHi by LOS sialylation. However, the serum bactericidal activity against the wild type and the lgtC mutant in the presence and absence of sialic acid did not appear to directly correlate with IgM binding. In the absence of sialic acid, the survival of the lgtC mutant was greatly reduced compared to that of the wild type, even though significantly less IgM bound to the mutant’s surface (Fig. 9). It is likely that LgtC contributes to the serum resistance of this strain, as seen in strain RdAW (Rd) and other NTHi strains (R2846, R2866), and that deletion of lgtC results in extreme serum sensitivity that overcomes any benefit to survival that reduced IgM binding might have had (20, 34–36). With sialylated LOS, however, the survival of the wild type and the lgtC mutant was nearly equivalent, demonstrating that sialic acid can mediate resistance in an lgtC-independent manner.

Because a significant inhibitory effect of sialic acid on IgM binding was retained in the lgtC mutant, we postulated that underlying epitopes are targeted by IgM and protected by sialylation. Therefore, we evaluated the galactose residue added by Lic2A (Fig. 1). After testing serum bactericidal activity and IgM binding, we were able to conclude that sialic acid-mediated serum resistance is dependent on Lic2A in NTHi 375 (Fig. 10). As the lic2A-dependent galactose structure likely provides the only sialylation site in this strain (37) (Fig. 1), serum absorption experiments were conducted to determine whether the epitopes protected by sialylated LOS are dependent on lgtC or lic2A. Killing of wild-type NTHi 375 grown without sialic acid was the greatest in the presence of serum absorbed with the lic2A mutant (lacking both lic2A- and lgtC-dependent structures) and decreased sequentially as antibodies to lic2A- and lgtC-dependent structures were absorbed out of serum (Fig. 11). In the 375 WT grown without sialic acid, absorption with the lgtC mutant depleted antibody targeting the lic2A-dependent structure, resulting in increased survival. Both lic2A- and lgtC-dependent epitopes are present on the wild type grown without sialic acid, and absorption with these cells removes the antibodies of interest, resulting in the largest amount of survival of wild-type NTHi 375, though a small amount of sialic acid-mediated protection remained with this serum, either due to incomplete absorption or due to the presence of an additional unidentified epitope. Also consistent with a bactericidal antibody targeting a lic2A-dependent epitope protected by sialylation, the lgtC mutant grown without sialic acid was killed in the presence of lic2A-absorbed serum but was resistant to this serum when grown with sialic acid. As expected, serum absorbed with either the lgtC mutant or the wild type appeared to kill the lgtC mutant grown without sialic acid less efficiently than did the lic2A-absorbed serum, though this difference was not statistically significant, as the lgtC mutant was highly sensitive to killing in the absence of sialic acid, likely because it lacks complement evasion mechanisms independently mediated by LgtC, as described previously (34). In contrast, when grown with sialic acid, there was no difference in the survival of the lgtC mutant in any of the three absorbed serum samples, consistent with sialic acid-mediated protection of a lic2A-dependent IgM target in this strain. Together, these results are consistent with sialic acid-mediated protection from IgM of two LOS epitopes produced by lgtC and lic2A, respectively. Even though a large proportion of IgM bound to the lgtC-dependent structure of NTHi 375, a considerable amount of lgtC-independent, sialic acid-mediated protection from serum killing that could be attributed to the lic2A-dependent IgM target was observed (Fig. 11).

The clinical importance of serum IgM antibodies against NTHi has been identified in individuals with primary antibody deficiencies who have low serum IgM, a risk factor for respiratory tract colonization with H. influenzae (38). Currently, monoclonal antibody (MAb) therapies for the treatment of microbial diseases are underutilized and underdeveloped, despite rising antimicrobial drug resistance. Nevertheless, recent reports support the potential efficacy of MAbs against specific LOS structures and the effects of sialylation. MAb 2C7, targeting a LOS epitope, is being developed as a gonococcal vaccine candidate (39). The binding of this MAb requires extension from both HepI and HepII, which sometimes display a P^k-like extension that is capped by Neu5Ac. Ram et al. demonstrated the in vitro and in vivo activity of this MAb against N. gonorrhoeae and found that sialylation inhibits binding and bactericidal killing in vitro; however, its antibacterial activity was retained in the presence of sialic acid in vivo (40). Apicella et al. found that a Kdo monosaccharide epitope recognized by MAb 6E4 could be detected on NTHi directly after isolation from experimentally infected middle ears in the chinchilla otitis media model and that expression of this epitope in vitro is decreased in the presence of Neu5Ac, which inhibits Kdo addition to the LOS outer core and MAb 6E4 bactericidal activity (22). Therefore, detection of the Kdo epitope in ex vivo isolates suggests that NTHi may encounter low sialic acid levels during certain stages of middle ear infection, conditions that would favor exposure of the LOS IgM epitopes identified herein. Additional studies will be required to determine whether lgtC- or lic2A-dependent structures can be developed as therapeutic targets. It is important to identify targets that elicit a response that produces bactericidal antibodies that are effective in vivo and not blocking, nonbactericidal antibodies. Strategies to overcome protective sialylation could be critical in the development of antibody-based therapies and other immunotherapeutics.

MATERIALS AND METHODS

Strains and culture conditions.

H. influenzae strain RdAW (Rd) (41) and NTHi clinical isolates NT127 (42), NTHi 375 (18), Hi486 (9), PittGG (43), and R2866 (37, 44) were grown at 35°C ± 1.5°C on agar plates containing brain heart infusion supplemented with 10 μg/ml NAD and 10 μg/ml hemin (sBHI) or a chemically defined medium (MIc) (45). MIc agar contains either no sialic acid (MIc^−SA) or sialic acid (Neu5Ac) supplemented at 25 μg/ml (MIc^+SA). The phase on/off status of relevant phase-variable genes lic1A, lic2A, lgtC, lic3A, and lic3B in the wild-type NTHi 375 strain was determined by PCR, amplifying tandem repeat regions upstream of each gene followed by Sanger sequencing (Eurofins Genomics), and all five genes were found to be phase on in our isolate. DNA was transformed into naturally competent H. influenzae as described previously (46). NT127 with a deletion of siaB and NTHi 375 with a deletion of lic2A were created as previously described (26), and we thank Derek Hood for the Hi375 siaB strain (18). The NTHi 375 lgtC mutant contains an lgtC gene disrupted by a kanamycin resistance cassette (47).

lgtD deletion mutant.

An lgtD mutant strain in the NTHi 375 background was constructed by replacement of the coding sequence of lgtD (NF38_07685) with a cassette containing a kanamycin resistance gene, aphI from Tn903 (48) driven by the Escherichia coli trc promoter. The kanamycin resistance gene driven by the trc promoter from pTrcHis2B (Thermo Fisher Scientific) was amplified with primers PTrc5CGA (5′-CGACTGCACGGTGCACCAATGCTTC-3′) and kan3’+TAA (5′-TTAGAAAAACTCATCGAGCATCAAA-3′) from mariner Himar1 minitransposon mmTrcK (48) to yield a 1,065-bp product, Trc-kan. To create a nonpolar in-frame deletion of lgtD, a 999-bp PCR product containing the 5′ flanking region upstream of the lgtD start codon was amplified from NTHi 375 with primers 5′ lgtD_1kbup (5′-CGAGGCTGAATTTGTAAAAGCG-3′) and 5′ Trckan-lgtDout (5′-CATTGGTGCACCGTGCAGTCGATAAGCTCTAATTTTACACTCTT-3′). A 974-bp PCR product containing the flanking region immediately 3′ of the termination codon of lgtD was amplified from NTHi 375 with primers 3′ kan-lgtDout (5′-GCTCGATGAGTTTTTCTAAATAGAGTGTTCAGAAAGGGGA-3′) and 3′ lgtD_1kbdown (5′-TGCATTCGTTGTTTAGCAACAG-3′). The 999-bp, 974-bp, and 1,065-bp (Trc-kan) products were stitched together using overlap extension PCR with primers 5′ lgtD_1kbup and 3′ lgtD_1kbdown to yield the resultant ∼3-kb lgtD knockout PCR product. This product was transformed into competent NTHi 375 cells, and transformants were selected on sBHI plates containing 20 μg/ml kanamycin. The transformants were screened by PCR for the presence of the kanamycin resistance cassette, and deletion of lgtD was verified by PCR amplification across the inserted recombinant region with primers specific for flanking sequences (primers 5lgtD_confirm [5′-ACAAACAGAAGCATTAGCATTCGG-3′] and 3lgtD_confirm [5′-TCATACTTGCATTAAGGTTCATCC-3′]).

Absorption of sera.

Absorption of serum antibodies was performed using whole bacterial cells. NTHi 375 wild-type, lic2A mutant, and lgtC mutant strains grown on MIc agar containing no sialic acid were harvested, and 1 × 10⁸ cells of each strain were washed with Hanks’ balanced salt solution with calcium and magnesium chloride (HBSS⁺⁺), pelleted, resuspended in 100 μl of heat-inactivated normal human serum, and incubated for 30 min at 37°C, followed by centrifugation for 5 min at room temperature. Serum supernatant was collected and added to a fresh aliquot of 10⁸ bacteria, and the absorption procedure was repeated for a total of three rounds. After the final round, each serum suspension was incubated at 4°C overnight with gentle agitation before final centrifugation to collect the absorbed sera.

Serum bactericidal assays.

NTHi strains grown overnight on defined minimal medium plates were harvested and resuspended in HBSS⁺⁺–0.1% bovine serum albumin (BSA). Bacteria were diluted to ∼2,000 CFU and incubated with the concentration of pooled normal human serum (NHS; Complement Technology, Inc.) specified above or, where specified, with NHS treated with 10 mM magnesium chloride (MgCl₂) and 10 mM EGTA to selectively block the classical and lectin pathways. At time zero and after 30 min of incubation at 37°C, the reaction mixtures were plated for counting of the bacteria. Cells treated with NHS heat inactivated (NHS^Δi) for 35 min at 56°C were used as a control. Survival was calculated as the number of colonies recovered at 30 min in active serum-treated samples relative to the number of colonies in samples treated with NHS^Δi. To test the bactericidal activity of serum antibodies, an IgG/IgM antibody-depleted human complement (HC) pooled serum (Pel-Freez Biologicals) was used as a source of active complement. Samples were incubated with 20% NHS^Δi at 37°C for 5 min, prior to incubation with 5% HC serum at 37°C for 30 min. Each reaction mixture was plated at time zero and at 30 min to determine survival. All experiments were conducted on duplicate or triplicate independent samples.

Flow cytometry.

NTHi strains grown on MIc agar plates with and without sialic acid were incubated with NHS or single-donor human complement serum (SDNHS) (Innovative Research) to measure the binding of serum antibodies. Bacteria (optical density at 600 nm, ∼0.3) were incubated with NHS^Δi (final concentration, 20%) for 30 min at 37°C. Fluorescein isothiocyanate (FITC)-conjugated goat anti-human IgG or sheep anti-human IgM (Southern Biotech) was used for detection. C3 binding was detected using FITC-conjugated sheep anti-human C3c (SouthernBiotech) after the bacteria were incubated with 20% NHS or 20% NHS treated with MgEGTA (10 mM) for 30 min at 37°C. CD77-like glycoforms were detected using mouse anti-human CD77 clone 5B5 conjugated to FITC (BioLegend). All incubations with serum and antibodies were carried out in HBSS⁺⁺–1.0% BSA. Data were acquired on an Acea Biosciences NovoCyte flow cytometer and analyzed using Acea NovoExpress software. The median fluorescence intensity (MFI) shown for all experiments was calculated by subtracting the fluorescence values for the antibody-only background control of each strain from the values for the corresponding replicate samples.

Statistical analyses.

Statistical significance was determined by a two-tailed Student's t test or one-way analysis of variance (ANOVA) with Bonferroni’s multiple-comparison test using GraphPad Prism software (San Diego, CA), as indicated in the figure legends.