ABSTRACT
Candida albicans is a yeast-like pathogen and can cause life-threatening systemic candidiasis. Its cell surface is enriched with mannan that is resistant to complement activation. Previously, we developed the recombinant human IgG1 antimannan antibody M1g1. M1g1 was found to promote complement activation and phagocytosis and protect mice from systemic candidiasis. Here, we evaluate the influence of IgG subclass on antimannan antibody-mediated protection. Three IgG subclass variants of M1g1 were constructed: M1g2, M1g3, and M1g4. The IgG subclass identity for each variant was confirmed with DNA sequence and subclass-specific antibodies. These variants contain identical M1 Fabs and exhibited similar binding affinities for C. albicans yeast and purified mannan. Yeast cells and hyphae recovered from the kidney of antibody-treated mice with systemic candidiasis showed uniform binding of each variant, indicating constitutive expression of the M1 epitope and antibody opsonization in the kidney. All variants promoted deposition of both murine and human C3 onto the yeast cell surface, with M1g4 showing delayed activation, as determined by flow cytometry and immunofluorescence microscopy. M1g4-mediated complement activation was found to be associated with its M1 Fab that activates the alternative pathway in an Fc-independent manner. Treatment with each subclass variant extended the survival of mice with systemic candidiasis (P < 0.001). However, treatment with M1g1, M1g3, or M1g4, but not with M1g2, also reduced the kidney fungal burden (P < 0.001). Thus, the role of human antimannan antibody in host resistance to systemic candidiasis is influenced by its IgG subclass.
INTRODUCTION
Candida albicans is one of the leading causes of bloodstream infections in hospitalized patients (1, 2). Dissemination of the organism from the bloodstream into deep tissue leads to hematogenously disseminated candidiasis that is life-threatening and is associated with an extremely high mortality rate (3). The medical significance of disseminated candidiasis is further underscored by the emergence of resistance to antifungal therapy despite the introduction of new antifungal agents (4). Consequently, there have been efforts to seek alternative preventative and therapeutic means.
One such effort has been focused on antibody-based prophylactics and therapeutics. For example, efungumab (Mycograb), a human recombinant antibody specific for heat shock protein 90, was found to be therapeutically effective in combination with amphotericin B in patients with candidiasis (5). In animal models of systemic candidiasis, a number of passively introduced antibodies have been found to be protective; the protective antigens include β-glucan (6), a 58-kDa mannoprotein (7), and secreted aspartic proteinase 2 (8).
We along with others have been studying the protective role of antibodies against cell surface-displayed mannan, which is composed of mannose polysaccharides and oligosaccharides covalently linked to proteins (9, 10). A protective role for antimannan antibody was initially suggested by studies with murine antimannan antibodies (9) and with vaccines that induce a murine antibody response to β-mannan epitopes (11). We subsequently generated the first human recombinant antimannan IgG1 antibody, M1g1 (10). M1g1 was found to activate both mouse and human complement, to promote phagocytosis and phagocytic killing of C. albicans yeast cells by murine macrophages, and to be protective in a murine model of hematogenously disseminated candidiasis (10, 12).
Studies with antibodies reactive with mannan and other antigens have collectively demonstrated an important role of antibody-mediated immunity in host resistance to invasive candidiasis. However, little is known about the influence of IgG subclass on antibody-mediated protection against candidiasis. It is well established that antibody-mediated clearance of infectious agents is largely dependent on Fc-mediated effector mechanisms, such as recruitment of complement and phagocytes, yet there seem to be no hard-and-fast rules for the selection of a protective isotype for a particular microbial infection. For example, although murine IgG3 monoclonal antibodies (MAbs) against the capsule of Cryptococcus neoformans are nonprotective (13), murine IgG3 MAbs are protective against the encapsulated pathogen Streptococcus pneumoniae (14) and nonencapsulated pathogen C. albicans (15). Similarly, human IgG1 (hIgG1) is protective against intranasal infection by the encapsulated pathogen S. pneumoniae (16) but not protective against hematogenous infection by the encapsulated yeast C. neoformans (17).
The purpose of this study was to determine the influence of IgG subclass on human antimannan antibody-mediated resistance to hematogenously disseminated candidiasis. We constructed M1g2, M1g3, and M1g4, which are IgG subclass variants of M1g1; they contain identical M1 Fab fragments but differ in the Fc region. Our results show that each of the four M1 IgG subclass variants is able to initiate deposition of both mouse and human C3 onto C. albicans cells, and complement activation by M1g4 is mediated by its M1 Fab through the alternative pathway in an Fc-independent manner. All four subclass antibodies enhanced resistance to hematogenously disseminated candidiasis in mice, with M1g2 conferring less protection than the others.
MATERIALS AND METHODS
Yeast and Candida mannan.C. albicans 3153A was used for this study. Yeast cells were passaged daily at 37°C three times in 3 ml of GYEP (2% glucose, 1% peptone, and 0.3% yeast extract) medium and used as an inoculum for one final culture. For infection of mice, yeast cells of the final culture were washed and resuspended in Dulbecco's phosphate-buffered saline (DPBS; Lonza, Switzerland). For complement assays, yeast cells were inactivated with 1% formaldehyde, washed, resuspended in phosphate-buffered saline (PBS), and stored at −80°C (12). For purification of mannan, yeast cells were inactivated with 1% formaldehyde, washed, and resuspended in water; water-soluble mannan was isolated after removal of lipids as described previously (10).
Construction and production of IgG subclasses of antimannan M1 antibodies.The expression vector for the human recombinant antimannan antibody M1g1 (10) was used for construction of IgG subclass variants. M1g1 is an IgG1 antibody that contains a κ Fab fragment specific for a mannan epitope defined as M1 that was identified from a phage display library, as described in detail elsewhere (10). The gene for IgG1 heavy-chain domains (CH1-hinge-CH2-CH3) contained in the M1g1 expression vector was replaced with the gene for the heavy-chain domain of human IgG2, IgG3, or IgG4 (18), a generous gift from Paul Parren (Genmab, Netherlands). Thus, the expression vectors for M1g1, M1g2, M1g3, and M1g4 contained the same genes for the M1 Fab fragment but differed in the genes for the Fc domains. As needed, one or two bases of the subclass heavy-chain genes were altered to facilitate cloning, but the predicted amino acids were retained. DNA sequence analysis confirmed the presence of the expected heavy-chain gene sequences in the expression vectors for the most prevalent human IgG subclass variants found at IgBLAST (the database of human immunoglobulins at the National Center for Biotechnology Information). Chinese hamster ovarian cells (ATCC CHO-K1) were transfected with the expression vector of M1g1, M1g2, M1g3, or M1g4 for production of soluble IgG subclass antimannan antibodies as described previously (10). M1g1, M1g2, or M1g4 was purified by affinity chromatography using protein A, and M1g3 was purified using protein G according to the manufacturer's procedures (GE Biosciences, Piscataway, NJ). Purified antibodies were desalted by gel filtration (GE Biosciences), resuspended in DPBS, filtration sterilized, and stored at −80°C. Antibody concentration was determined by absorbance at 280 nm with a molar absorption coefficient adjusted for the amino acid composition of each subclass variant (19).
Characterization of IgG subclasses of human recombinant antimannan M1 antibodies.The molecular sizes and identities of affinity-purified IgG subclass variants were established with standard SDS-PAGE and Western blot analyses using myeloma IgG subclass variants as a standard (The Binding Site, Birmingham, England) and goat antibodies specific for human IgG light or heavy chains (SouthernBiotech, Birmingham, AL) (10). Subclass identity was confirmed for each variant with enzyme-linked immunosorbent assay (ELISA), as described previously; for the assay microtiter wells were first coated with purified mannan of C. albicans 3153A, and mannan-bound antibody was detected with MAbs specific for each of the four IgG subclasses (SouthernBiotech, Birmingham, AL) (10).
The binding affinity of each subclass variant for purified mannan was measured by ammonium thiocyanate disruption (20). Briefly, microtiter wells were coated with purified mannan and then incubated in duplicate with each subclass variant at a concentration that gave 1 unit of absorbance at 450 nm in the presence of increasing amounts of NH4SCN from 0 M to 10 M in PBS containing 0.5% Tween 20 (PBS-T). Wells were washed after a 15-min incubation at room temperature, and the amount of mannan-bound antibody was quantified colorimetrically with goat anti-κ horseradish peroxidase (HRP; SouthernBiotech, Birmingham, AL) and its substrate (SuperBlue; KPL, Gaithersburg, MD). The relative binding of each variant to immobilized mannan in the presence of NH4SCN compared to the maximum binding in the absence of NH4SCN was calculated; the amounts of NH4SCN that caused 50% reduction in antibody binding were estimated and statistically compared among the subclass variants by one-way analysis of variance (ANOVA) using SigmaPlot, version 11 (Systat Software, Inc., San Jose, CA).
To compare the binding pattern of each variant on the cell surface, yeast cells were incubated for 30 min at 0°C with each variant in increasing amounts, and yeast-bound antibody was detected with fluorescein isothiocyanate (FITC)-labeled anti-human κ chain antibody and either quantified by flow cytometry or visualized by immunofluorescence microscopy as described previously (12). The half-maximal effective concentrations (EC50s) for the variants were calculated and compared statistically by one-way ANOVA using SigmaPlot, version 11.
Effect of M1 IgG subclass variants on the resistance of mice to hematogenously disseminated candidiasis.A procedure previously used for studies of the protective efficacy of M1g1 (10) and murine antimannan antibodies (21) was adopted. Female BALB/c mice were purchased from the National Cancer Institute Animal Production Program (Frederick, MD). Mice at the age of 6 weeks were first given an intraperitoneal injection of M1g1, M1g2, M1g3, or M1g4 at 1 mg/mouse in 0.5 ml of DPBS or the buffer alone and challenged intravenously 4 h later with a lethal dose of 0.85 × 106 C. albicans 3153A yeast cells in 0.2 ml of DPBS. The choice of antibody dose at 1 mg per mouse was based on a prior study that compared the protective efficacy of M1g1 in various amounts (10), and this dose was also found protective against systemic cryptococcosis in mice treated with human-mouse chimeric antibodies (17). Mice were monitored twice daily for moribundity or death following the protocol approved by the Institutional Animal Care and Use Committee at the California State University—Long Beach. A moribund state was defined as being listless, unable to eat or drink, and nonreactive to touch (22). Kaplan-Meier survival curves were constructed and compared statistically with a log rank test using SigmaPlot, version 11; pairwise comparisons were done with the Bonferroni test.
The fungal burden in the kidney was determined following the method of Hans and Cutler for studies of murine antimannan antibodies (21). Mice were treated the same way as for the survival assays described above. Kidneys were retrieved 48 h postinfection and weighed. Individual kidneys were manually macerated in 0.5 ml of DPBS in a 5-ml Dounce homogenizer. Homogenate was serially diluted and cultured overnight on GYEP agar. CFU counts were determined per unit weight of kidney per mouse and compared statistically with a Mann-Whitney test between antibody-treated mice and control mice using SigmaPlot, version 11.
Analysis of in vivo expression of M1 epitope and in vivo opsonization of C. albicans cells by M1 IgG subclass variants in the kidney.To determine whether the M1 epitope was expressed during infection, mice were intravenously infected with C. albicans, and kidneys were retrieved and homogenized at 48 h postinfection as described above for the fungal burden assay. Homogenate was subjected to three cycles of washing with distilled water, maceration with a 23-gauge needle, vortexing, and freeze-thawing to lyse mouse cells. A sample of the treated homogenate was subsequently incubated with each subclass at 5 μg/ml in 100 μl of PBS–1% bovine serum albumin (BSA) at 37°C and followed by goat anti-human IgG-FITC. Cells bound with subclass variants were visualized by differential interference contrast microscopy (DIC) and immunofluorescence microscopy (12).
To determine whether C. albicans cells were opsonized by each of the subclass variants in the mouse, an ex vivo fluorescent staining method was adapted from a procedure for detection of in vivo display of Candida glucan in mice (23). Kidney homogenates from the fungal burden study were washed three times in distilled water to lyse mouse cells and then incubated with goat anti-human IgG-FITC for 1 h on ice in PBS–1% BSA, washed, and resuspended in PBS–1% BSA. C. albicans cells in different morphologies were visualized by DIC and fluorescence microscopy (12).
Effect of M1 IgG subclass variants on activation of C3 binding to yeast.Deposition of murine or human C3 onto C. albicans yeast cells in the presence of a distinct M1 IgG subclass variant was either analyzed quantitatively by flow cytometry or visualized by immunofluorescence microscopy as previously described (12). Briefly, normal mouse serum (NMS) pooled from 10-week-old female BALB/c mice was purchased from Harlan (Madison, WI). Normal human serum (NHS) was prepared from at least 30 healthy individuals after informed consent under a protocol approved by the Institutional Review Board of the California State University—Long Beach. Serum samples were stored at −80°C. For removal of naturally occurring antimannan antibodies (24, 25), NMS was adsorbed two times with yeast cells, and NHS was adsorbed four times; each adsorption was performed with 5 × 108 C. albicans 3153A yeast cells per ml of serum for 30 min at 0°C. The adsorbed serum was filtered and used immediately or stored at −80°C. C3 binding to yeast cells was measured at 37°C in barbital-buffered saline (5 mM sodium Veronal, 142 mM NaCl, pH 7.3) that contained 0.1% gelatin (gelatin-Veronal buffer, or GVB) and the following reagents: (i) 40% NMS or NHS, (ii) either 5 mM EGTA and 5 mM MgCl2 to limit complement activation to the alternative pathway or 1.5 mM CaCl2 and 1 mM MgCl2 to permit the activity of both the classical and alternative pathways (26), and (iii) one of the four variant antibodies or M1 Fab at 10 μg/ml. Cell-bound human C3 was detected by yeast-adsorbed goat anti-human C3-FITC (Kent Laboratories, Bellingham, WA), and cell-bound murine C3 was detected by MAb anti-mouse C3-FITC (Cedarlane Laboratories, Ontario, Canada).
In some experiments, human serum depleted of C1q, C2, or factor B (Quidel, Santa Clara, CA) was used to determine the requirement of the classical or alternative pathway for M1g4-mediated C3 activation (12). Each serum was adsorbed with yeast cells as described above. Yeast cells were incubated for 12 min in GVB containing 40% of each adsorbed serum and 1 mM MgCl2 with or without M1g4, and C3 deposition was determined as described above.
RESULTS
A matched set of four IgG subclasses of human recombinant antimannan M1 antibody have expected subclass characteristics and similar binding affinities.Previously, we constructed M1g1, a human recombinant antimannan antibody, that showed a protective role in host resistance to hematogenously disseminated candidiasis in mice (10). M1g1 is an IgG1 antibody and contains the Fab fragment M1 that is reactive with a mannan epitope (27). In this study, human IgG2, IgG3, and IgG4 subclass variants of M1g1 were generated, namely, M1g2, M1g3, and M1g4, and they all contained identical M1 Fab fragments. DNA sequence analysis of the expression vectors for the variants confirmed the presence of the entire light- and heavy-chain genes in correct locations. It also revealed complete agreement of the cloned gene sequences for the subclass-specific constant domains of the most prevalent allotype with the sequences at IgBLAST (data not shown). The subclass identity was confirmed by use of monoclonal antibodies specific for each of the human IgG subclasses; M1g1, M1g2, M1g3, or M1g4 bound to immobilized mannan could be recognized only by a subclass-specific detection antibody (Fig. 1). Affinity-purified subclass variants also exhibited expected molecular sizes (28), as shown by SDS-PAGE under both reducing and nonreducing conditions and by Western blot analysis using myeloma IgG subclasses as a standard (data not shown). The binding pattern of each subclass variant on the cell surface of C. albicans yeast was uniform, as determined by immunofluorescence microscopy (data not shown).
M1 IgG variants display expected subclass identities. M1g1, M1g2, M1g3, or M1g4 was serially diluted in microtiter wells containing immobilized mannan of C. albicans 3153A. Each mannan-bound subclass variant was detected with four separate MAbs, each for one specific human IgG subclass (anti-hIgG1, anti-hIgG2, anti-hIgG3, and anti-hIgG4), to establish the IgG subclass identity. Data from one of three representative experiments are shown.
The relative binding affinities of the subclass variants for purified mannan and for whole yeast cells were evaluated. An ELISA-based thiocyanate elution method was used to determine the avidity of each variant for purified mannan (20). The mean amounts of NH4SCN required to disrupt 50% of the binding of each variant to immobilized mannan were 6.4 ± 0.1 mM for M1g1 (from two independent experiments; means ± standard errors of the means [SEM]), 6.8 ± 0.2 mM for M1g2, 6.7 ± 0.1 mM for M1g3, and 6.6 ± 0.1 mM for M1g4. These values were similar for the variants (P = 0.2). The binding of each variant to yeast cells was also compared. Yeast cells were incubated with each variant in increasing amounts, and cell-bound antibody was quantified by flow cytometry and an anti-human κ chain antibody that is specific for M1 Fab, the common component in the variants. The half-maximal effective concentrations (EC50s) for the four variants were compared and found to be similar (P = 0.09) (Fig. 2).
Comparable binding of M1 IgG subclass variants to the cell surface of C. albicans. Yeast cells were incubated with each variant in various amounts and washed. The cell-bound antibody was quantified with FITC-conjugated goat anti-human κ chain antibody and flow cytometry, and the half-maximal effective concentration (EC50) for each subclass variant was calculated. Dose-dependent binding curves are representative of four assays. EC50s as shown are means from four independent experiments ± standard errors of the means and were compared statistically by one-way ANOVA (P = 0.09).
Antimannan IgG subclass variants M1g1, M1g3, and M1g4 are more protective than M1g2 against hematogenously disseminated candidiasis.A murine model of hematogenously disseminated candidiasis was used to assess the effect of each subclass variant on survival and fungal burden (10, 21). Six-week-old female BALB/c mice were treated intraperitoneally with each of the affinity-purified variant antibodies in DPBS or with DPBS alone as a control and lethally challenged intravenously 4 h later with C. albicans 3153A yeast cells. Survival was closely monitored for a period of 45 days. The mean survival time of control mice was 12 days. In contrast, treatment of mice with any one of the subclass antibodies significantly extended the survival times (P < 0.001). However, all mice treated with M1g2 died within the experiment period, whereas approximately 40% of the mice treated with M1g1, M1g3, or M1g4 survived the lethal infection (Fig. 3).
Effect of passive immunization with M1 IgG subclass variants on the survival of mice with hematogenously disseminated candidiasis. Results as shown are the combined data from two independent experiments (16 mice per group). Mice were treated 4 h before challenge by intraperitoneal injection of either PBS as a control or 1 mg of M1g1, M1g2, M1g3, or M1g4 (P < 0.001 versus the control by the Bonferroni method; P < 0.001 overall by a log rank test).
Next, fungal burden in kidneys at 48 h postinfection was assessed (21). Mice were treated as described above for the survival assay. The fungal burden in terms of CFU counts from kidneys was compared between control mice and antibody-treated mice. There was a significant reduction in CFU amounts recovered from M1g1-, M1g3-, or M1g4-treated mice compared to the CFU amounts from control mice (P < 0.001) (Fig. 4). However, M1g2-treated mice showed a level of fungal burden in the kidney similar to that of control mice (Fig. 4). These results together show that protection from disseminated candidiasis by human antimannan IgG antibody may be influenced by the antibody's subclass.
Effect of passive immunization with M1 IgG subclass variants on the kidney fungal burden in mice with hematogenously disseminated candidiasis. Results as shown are the combined data from three independent experiments. Mice were treated 4 h before challenge by intraperitoneal injection of either PBS as a control (5 mice in experiment 1 and experiment 2 and 6 mice in experiment 3) or 1 mg of M1g1, M1g2, M1g3, or M1g4 (4 mice in experiment 1, 5 mice in experiment 2, and 6 mice in experiment 3). Horizontal bars represent median values. *, P < 0.001 (versus control values by a Mann-Whitney test).
Both C. albicans yeast and hyphae express the M1 epitope in vivo and are bound by M1 IgG subclass antibodies in the kidney.Previous studies have shown that C. albicans cells introduced into the bloodstream spread to various organs within a short time (29), but progressive fungal growth occurs mostly in the kidney (29, 30), leading to renal failure (30, 31). Furthermore, passive immunization with protective antibodies does not seem to prevent the kidney from the initial infection by C. albicans (10, 21). These studies suggest that subsequent host clearance of C. albicans cells from the kidney, disseminated from the bloodstream, appears critical to mice in resistance to disseminated candidiasis. Thus, we first determined whether the M1 epitope was expressed by C. albicans cells in the kidney. Homogenate of kidneys from mice infected with C. albicans was treated with each of the four IgG variants, and human antibodies bound to C. albicans cells were then detected with FITC-conjugated anti-human antibody and visualized by both DIC and fluorescence microscopy. We found that both yeast cells and hyphae were brightly stained by each variant and that the binding pattern was uniform (data not shown), indicating an in vivo expression of the M1 epitope on all morphologies.
Next, we assessed the binding of different M1 IgG subclass antibodies to C. albicans in the kidney. Kidneys from C. albicans-infected mice that had been treated with DPBS alone or DPBS containing each of the subclass variant antibodies were retrieved and homogenized. The homogenate was subsequently treated with an FITC-conjugated detecting antibody specific for human IgG. C. albicans cells recovered from mice that had been treated with each of the four subclass variants showed bright and uniform fluorescence not only on yeast cells but also on germ tubes and hyphae (Fig. 5). There were no visually distinguishable differences in the pattern and intensity of in vivo antibody binding to C. albicans cells by each of the M1 subclass variants. These observations together indicate that the M1 epitope is constitutively expressed on all cell types and that antibody binding to the progeny of the yeast cells initially translocated from the bloodstream continues to occur in the kidney.
In vivo opsonization of C. albicans by M1g1, M1g2, M1g3, or M1g4. Mice were treated with either PBS alone as a control or with PBS containing each IgG subclass and then infected with C. albicans yeast cells intravenously. Kidneys were retrieved at 48 h postinfection and homogenized. Homogenate was incubated with 10 μg/ml FITC-conjugated goat anti-human κ chain antibody, and cell-bound antibody was visualized by fluorescence microscopy. Candida cells recovered from control mouse homogenate were also visualized by DIC (differential interference contrast) microscopy.
All M1 IgG subclass variants activate both murine and human complement, with M1g4 being the least potent.Studies with murine antimannan antibodies have demonstrated a requirement of complement for antibody-mediated protection against hematogenously disseminated candidiasis (32). Previously, we observed that C. albicans is resistant to activation of mouse complement as no C3 was detected on the cell surface of yeast in the absence of M1g1 (10). Consequently, we compared the ability of each M1 subclass variant to activate the mouse complement cascade. Initiation of C3 deposition onto yeast cells occurred rapidly in the presence of M1g1, M1g2, or M1g3 but was delayed in the presence of M1g4 (Fig. 6, top panel). However, following a 20-min incubation, substantial C3 accumulated on the cell surface in the presence of any subclass variant, in contrast to a limited binding of C3 in the absence of antibody (Fig. 6, top panel). Furthermore, C3 binding occurred over the entire cell surface in a visually uniform pattern in the presence of M1g1, M1g2, or M1g3. In contrast, M1g4 mediated patchy depositions of C3 on the cell surface, and these patches expanded over time to cover the entire cell surface (Fig. 6, bottom panel). Thus, human IgG subclasses of antimannan antibody are activators of the murine complement system.
Activation of mouse C3 deposition onto C. albicans by M1 IgG subclass variants. Yeast cells were incubated with 40% yeast-adsorbed normal mouse serum containing one of the four variants at 10 μg/ml for various times. Cell-bound C3 was detected with an FITC-conjugated MAb specific for mouse C3. C3 deposition was either quantified by flow cytometry (top panel; values are means ± standard errors of the means from three independent experiments) or visualized by fluorescence microscopy (bottom; data from one of three representative experiments are shown). Ab, antibody.
Activation of mouse C3 by M1g4, a human IgG4, was unexpected as IgG4 is known as a nonactivator of the human complement system (33–35). We therefore repeated the same assay to determine the ability of the subclass variants to initiate deposition of human C3 on C. albicans. As in mouse serum, binding of C3 to yeast cells occurred rapidly in human serum when it was supplemented with either M1g1, M1g2, or M1g3, but it followed a lag in the presence of M1g4 (Fig. 7, top panel). The patterns of C3 deposition were also similar in both mouse and human sera, where uniform binding was observed with M1g1, M1g2, and M1g3, and expanding foci of C3 on the cell surface were seen with M1g4 (Fig. 7, bottom panel). Taken together, these observations demonstrate that all IgG subclasses of the M1 antimannan antibody can initiate deposition of both human and murine C3 onto C. albicans cells but show differences in the kinetics and patterns of C3 binding.
Activation of human C3 deposition onto C. albicans by M1 IgG subclass variants. Yeast cells were incubated with 40% yeast-adsorbed normal human serum containing one of the four variants at 10 μg/ml for various times. Cell-bound C3 was detected with FITC-conjugated goat anti-human C3 antibody. C3 deposition was either quantified by flow cytometry (top panel; means ± standard errors of the means from three independent experiments) or visualized by fluorescence microscopy (bottom; data from one of three representative experiments are shown).
M1g4-mediated complement activation is associated with M1 Fab and requires the alternative pathway.Activation of the classical pathway requires binding of C1q to the CH2 of the Fc region, but the CH2 of IgG4 does not bind C1q (36, 37). The unexpected observation that the antimannan antibody M1g4 initiates deposition of mouse C3 on C. albicans (Fig. 6) led us to an analysis of the role of M1 Fab in complement activation. Our previous studies found that M1 Fab activates the human complement system through the alternative pathway in an Fc-independent manner (12). Therefore, the ability of M1g4 to activate the murine complement system may be associated with its M1 Fab. To confirm this possibility, we determined the ability of M1 Fab to activate mouse complement through the alternative pathway. Results show that C3 deposition on yeast cells was essentially absent in adsorbed NMS, but addition of M1 Fab restored C3 activation to the adsorbed serum (Fig. 8). Furthermore, the kinetics of C3 deposition were largely indistinguishable whether both the classical and alternative pathways were active (no EGTA) or only the alternative pathway was active (with EGTA) (Fig. 8). These results confirm that M1 Fab is also an activator of murine complement C3.
Activation of mouse C3 deposition on C. albicans by human Fc-free M1 Fab through the alternative pathway. Yeast cells were incubated with 40% yeast-adsorbed normal mouse serum containing 0 or 10 μg/ml M1 Fab in the presence or absence of EGTA. Cell-bound C3 was detected with an FITC-conjugated MAb specific for mouse C3. C3 deposition was either quantified by flow cytometry (means ± standard errors of the means from three independent experiments) or visualized by fluorescence microscopy (insets showing C3 binding with M1 Fab plus EGTA; data from one of three representative experiments are shown).
The requirement of the alternative pathway for M1g4-mediated complement activation was further assessed using human serum deficient in complement components. M1g4-mediated C3 deposition was abolished in factor B-deficient serum in which the alternative pathway is absent, but it could be restored with the addition of factor B (Fig. 9A). In contrast, deficiency in C1q or C2 that is required for the classical pathway had no effect on M1g4-mediated C3 deposition (Fig. 9B and C).
Requirement of an intact alternative pathway for M1g4-mediated activation of human C3 deposition on C. albicans. Yeast cells were incubated for 12 min with 40% yeast-adsorbed human serum deficient in factor B (A), C1q (B), or C2 (C) containing 0 or 10 μg/ml M1g4. For experiments using factor B-deficient serum, controls containing exogenous factor B (fB) were also included. Cell-bound C3 was detected with an FITC-conjugated goat anti-human C3 antibody and quantified by flow cytometry. The data in panel A are means ± standard errors of the means from two independent experiments. The data in panels B and C represent results of one single experiment.
These observations together assign a novel characteristic to M1 Fab in activation of both mouse and human complement systems in an Fc-independent manner. An IgG4 can be a complement activator when it contains a Fab fragment that activates the alternative pathway.
DISCUSSION
Substantial evidence is now available for the important role of antibody-mediated immunity in host resistance to systemic candidiasis and provides rationales for potential applications of immunoglobulin-based prophylactics and therapeutics. However, little is known about how the IgG subclass of an anti-Candida antibody influences its biological functions. To approach this question, we constructed a matched set of four IgG subclass variants of the human antimannan antibody M1 (10) that share identical M1 Fabs and differ only in the subclass-specific heavy chains (Fig. 1). These antibodies showed similar affinities of binding to the cell surface of C. albicans (Fig. 2). They uniformly bound to both yeast cells and hyphae in the kidney of antibody-treated and C. albicans-infected mice (Fig. 5). They activated deposition of both mouse C3 (Fig. 6 and 8) and human C3 (Fig. 7 and 9) onto the cell surface of C. albicans, with M1g4 being the least potent. Treatment with M1g1, M1g2, M1g3, or M1g4 extended the survival of mice with hematogenously disseminated candidiasis (Fig. 3), but treatment with M1g1, M1g3, or M1g4, but not M1g2, also reduced the fungal burden in the kidney (Fig. 4). To our knowledge, this is the first report on the influence of IgG subclass on the biological activities of human antimannan antibody against C. albicans.
Antimannan IgG antibody-mediated resistance to systemic candidiasis may involve neutralization, Fcγ receptor (FcγR)-dependent phagocytosis, complement activation, and possibly other mechanisms. In this study, we used a mouse model of hematogenously disseminated candidiasis where introduced C. albicans yeast cells spread to deep tissues through the bloodstream. Initial protection provided by antimannan antibody could involve blocking the binding sites in mannan that are required for adherence of yeast cells to the endothelium for invasion as Candida mannan has a demonstrated role in adhesion (38–41). The four IgG subclass variants used in this study have the same binding specificity for mannan and therefore should have similar neutralizing effects on adhesion of C. albicans to host cells and tissues. However, they differ in their Fc regions and likely also differ in Fc-dependent clearance functions.
One possibility for the reduced protection mediated by M1g2 compared to that of M1g1, M1g3, or M1g4 (Fig. 3 and 4) may be related to the poor ability of M1g2, an IgG2, to mediate FcγR-dependent phagocytic clearance. In the human system, the four IgG subclasses in FcγR-dependent effector functions follow the ranking of IgG1/IgG3 > IgG4 > IgG2. IgG2 typically binds only to the low-affinity activating FcγRIIA, whereas IgG1, IgG3, and IgG4 bind to all activating FcγRs of both high and low affinities (42). The relationship between each of the human IgG subclasses and mouse FcγRs has not been well studied, but the available data indicate that human IgG2 may behave like murine IgG3 in the murine system as a poor mediator of FcγR-dependent effector functions. In one study, human IgG1, IgG3, and IgG4 were shown to bind to all three murine activating FcγRs (I, III, and IV), whereas human IgG2 bound to only one activating FcγR (III) (43). Furthermore, the study showed that among the four IgG subclass variants of human anti-CD20 antibodies, IgG2 was the least effective mediator of phagocytosis of lymphoblasts by bone marrow-derived murine macrophages (43). Similar results have been reported by other investigators. For example, a recombinant human IgG2 anti-pneumococcal antibody was found to be unable to interact with mouse leukocyte FcR in vitro (16), and human IgG2-mouse chimeric antibodies were found to be unable either to promote phagocytosis of C. neoformans yeast cells by isolated mouse peritoneal macrophages (17) or to induce the release of thymidine from murine macrophages as a measure of antibody-dependent cytotoxicity (44). Thus, it appears that human IgG2 antibody acts as an ineffective activator of FcγR-mediated functions in the murine system.
Inefficient FcγR-mediated phagocytic clearance, as possibly in the case of M1g2 treatment, would allow more C. albicans cells to escape from the bloodstream into deep tissue and importantly might not effectively clear invading C. albicans cells from deep-tissue colonization. Time course studies have found that Candida cells introduced into the bloodstream quickly emerge in various organs, including the kidney (29), but progressive fungal growth occurs mostly in the kidney (29, 30), leading to renal failure (30, 31). Therefore, kidney-based immunity appears to be essential to host resistance to hematogenous candidiasis (45). We have previously observed that M1g1 promotes phagocytosis and phagocytic killing of C. albicans yeast cells by mouse peritoneal macrophages and that kidneys retrieved from M1g1-treated mice showed significantly fewer and smaller infection loci than control mice, suggesting an antibody-enhanced clearance in the kidney (10). Data from this study show that opsonization of both yeast cells and hyphae by each of the four subclass antibodies occurred in the kidney (Fig. 5), suggesting a continued antibody-binding to the progeny of the yeast cells initially translocated from the bloodstream. However, M1g2 opsonization did not lead to a reduced fungal burden in the kidney (Fig. 4). These observations indicate that M1g2 is not as effective as other IgG subclasses in containment and clearance of Candida infection in the kidney. It is likely that the events that occur following IgG opsonization may alter the course of infection.
The observation that M1g2 treatment produced a modest prolongation of survival without reducing the fungal burden in the kidney may suggest that M1g2 enhances the tolerance of mice to systemic candidiasis. A prolonged survival without reduction in fungal burden has also been observed with a murine model of hematogenous candidiasis with C. albicans strains differing in chitin expression (46) or in the outer-chain elongation of N-glycans (47) and with a murine model of pulmonary cryptococcosis prophylactically treated with an anti-cryptococcal capsule MAb (48). These observations together indicate that an extension of survival is likely an overall indication of the strength of host resistance to a microbial infection and that a specific measure may not reflect the complexity of elements mobilized in the host defense.
Antimannan antibody-mediated protection from hematogenously disseminated candidiasis may involve complement. A study by Han et al. (32) reported that protection by a murine antimannan IgM and its IgG3 variant against systemic candidiasis was observed only in complement-competent mice but not in complement-deficient mice. Similarly, protection by human IgG1 and IgG2 anti-pneumococcal antibodies was abolished in mice deficient in complement factors (16). We observed that deposition of mouse C3 onto C. albicans cells occurred markedly faster in the presence of M1g1, M1g2, or M1g3 than in the presence of M1g4 (Fig. 6), but M1g4 was as protective as M1g1 and M1g3 and more protective than M1g2 (Fig. 3 and 4). This observation indicates that the capacity of complement activation alone may not be an absolute determinant of antibody-mediated protection. The role of complement in antibody-mediated resistance to microbial infection is likely complex and may be influenced by the fine balance between its disease-causing effects and its effects on clearance of microbes. Such a complexity is well illustrated by the study of Beenhouwer et al. (17). They found that complement-nonactivating human IgG2- or IgG4-mouse chimeric antibodies are protective in a murine model of cryptococcosis, whereas complement-activating human IgG1- or IgG3-mouse chimeric antibodies are nonprotective (17). Thus, it is possible that factors other than complement may also be involved in antibody-mediated immunity to fungal infection.
Mannan is abundantly displayed on the cell surface of C. albicans, and it is resistant to complement activation (25). We have previously shown that the human antibody M1g1 can activate deposition of both mouse and human C3 on the cell surface of C. albicans (10, 12). This study extends this finding to the other three IgG subclasses of antimannan antibody: M1g2, M1g3, and M1g4 (Fig. 6 and 7). The ability of human IgG1, IgG2, and IgG3 to activate human complement has been well documented, and our data (Fig. 7) are in agreement with those of previous studies (49, 50). In contrast, studies of the interaction of human antibody with the mouse complement system have been quite limited. In one study, human recombinant IgG1 and IgG2 antibodies were found to activate binding of mouse C3c to pneumococci (16). In another study, binding of C1q to antibody-antigen complexes was used to assess activation of the classical pathway, and mouse C1q was found to bind to C. neoformans yeast cells treated with human IgG1- or IgG3-mouse chimeric antibodies, but not with IgG2 or IgG4 variants (17). The difference in the ability of human IgG2 to activate mouse complement in these studies is likely due to the differences in the assay format. It has been reported that the level of C1q binding may not necessarily correlate with the level of complement activation (36).
The observation that M1g4, a human IgG4, activates complement is interesting (Fig. 6 and 7). IgG4 is a nonactivator of the human complement system, likely due to the fact that the CH2 of IgG4 is unable to bind to C1q (36, 37). It is consequently considered anti-inflammatory (51). Complement can be activated through the classical pathway, the alternative pathway, or the lectin pathway. Typically, only the initiation of the classical pathway requires antibody in an Fc-dependent manner. We have previously shown that the M1 Fab that is common to all of these subclass variants is an activator of the human complement cascade through the alternative pathway in the absence of the Fc (12). In this study, we show that M1 Fab promoted deposition of mouse C3 to the cell surface of C. albicans (Fig. 8) and that activation of complement by M1g4 required an intact alternative pathway (Fig. 9). Therefore, it is likely that the Fab M1 contained in M1g4 initiates the mouse complement cascade in an Fc-independent manner, imparting to M1g4 a novel ability for complement activation.
In summary, we found that the four IgG subclass variants of human antimannan M1 antibody differ in protection against hematogenously disseminated candidiasis in mice, with IgG1, IgG3, and IgG4 being more effective than IgG2. These IgG subclass variants activate both mouse and human complement, leading to deposition of opsonic C3 fragments on C. albicans cells. The ability of an IgG4 antibody to activate complement may be influenced by its Fab, and an IgG4 antibody can be an activator of complement. These results contribute to rational selection of protective isotypes in designs of immuno-prophylactics and therapeutics that involve antibody responses against systemic candidiasis.
ACKNOWLEDGMENTS
This work was supported by National Institutes of Health grants AI052139, GM063119, and GM094080. C.T.N. was an NIH RISE fellow.
FOOTNOTES
- Received 8 July 2015.
- Returned for modification 29 July 2015.
- Accepted 6 November 2015.
- Accepted manuscript posted online 16 November 2015.
- Copyright © 2016, American Society for Microbiology. All Rights Reserved.
