The noncollagenous (NC1) domains of α3α4α5(IV) collagen in the glomerular basement membrane (GBM) are targets of Goodpasture autoantibodies or Alport posttransplant nephritis alloantibodies mediating rapidly progressive glomerulonephritis. Because the autoepitopes but not the alloepitopes become cryptic upon assembly of α3α4α5NC1 hexamers, we investigated how the accessibility of B cell epitopes in vivo influences the development of glomerulonephritis in mice passively immunized with human anti-GBM Abs. Alport alloantibodies, which bound to native murine α3α4α5NC1 hexamers in vitro, deposited linearly along the mouse GBM in vivo, eliciting crescentic glomerulonephritis in Fcgr2b−/− mice susceptible to Ab-mediated inflammation. Goodpasture autoantibodies, which bound to murine α3NC1 monomer and dimer subunits but not to native α3α4α5NC1 hexamers in vitro, neither bound to the mouse GBM in vivo nor induced experimental glomerulonephritis. This was due to quinary NC1 crosslinks, recently identified as sulfilimine bonds, which comprehensively locked the cryptic Goodpasture autoepitopes in the mouse GBM. In contrast, non-crosslinked α3NC1 subunits were identified as a native target of Goodpasture autoantibodies in the GBM of squirrel monkeys, a species susceptible to Goodpasture autoantibody-mediated nephritis. Thus, crypticity of B cell autoepitopes in tissues uncouples potentially pathogenic autoantibodies from autoimmune disease. Crosslinking of α3α4α5NC1 hexamers represents a novel mechanism averting autoantibody binding and subsequent tissue injury by posttranslational modifications of an autoantigen.

Autoimmune diseases are initiated by an abnormal engagement of the adaptive immune system against self-Ags. Although autoimmunity is primarily prevented by central or peripheral establishment of immune self-tolerance in T cells and B cells, inadvertent autoimmune responses may also be uncoupled from disease by other mechanisms. For instance, tissue injury mediated by type II or III hypersensitivity reactions can be prevented by anatomic, cellular, and molecular barriers that avert either tissue deposition of immune complexes (1, 2) or the engagement of inflammatory effectors by tissue-bound Abs (3). Another putative barrier is cryptic B cell autoepitopes, which are sites within the structure of native autoantigen normally inaccessible for autoantibody binding. The existence of autoantibodies to hidden determinants of self-Ags suggests that pathologic unmasking of cryptotopes may contribute to breaching immune self-tolerance, yet the role of cryptic epitopes in the effector phase is unknown. A paradigm for addressing this question is provided by Goodpasture (GP) disease, the prototypical autoimmune disease characterized by autoantibodies against cryptic epitopes (4).

GP disease presents clinically as life-threatening rapidly progressive glomerulonephritis and pulmonary hemorrhage, associated with circulating and tissue-bound IgG autoantibodies deposited in a linear pattern along the glomerular and alveolar basement membranes. A clinical variant without overt lung involvement is known as autoimmune anti–glomerular basement membrane (GBM) Ab disease. GP autoantibodies target two major conformational autoepitopes within the noncollagenous (NC1) domain of α3(IV) collagen (46), a tissue-restricted autoantigen abundant in the GBM, which forms supramolecular networks composed of α3α4α5(IV) collagen molecules joined at both ends. GP autoepitopes are cryptic, requiring unmasking for maximal binding of GP autoantibodies to the autoantigen from tissues (7, 8).

Crypticity of GP epitopes emerges from interactions among NC1 domains mediating the self-assembly of collagen IV networks (911). The GP epitopes are partly buried during the assembly of α3α4α5NC1 hexamers, becoming cryptic (9, 12, 13). In vitro, GP autoantibodies natively react with isoforms of α3α4α5NC1 hexamers composed of monomer subunits, which occurs in small amounts in the human GBM. However, crosslinked isoforms of α3α4α5NC1 hexamers containing NC1 dimer subunits do not react with GP autoantibodies under native conditions, unless dissociated in vitro (14). It was therefore hypothesized that GP autoantibodies target a subset of α3α4α5(IV) collagen molecules lacking NC1 crosslinks in the human GBM.

The α3α4α5NC1 hexamers are also the target of anti-GBM alloantibodies mediating Alport posttransplant nephritis (APTN), a serious complication affecting ∼3–5% of Alport patients receiving a kidney transplant (1518). APTN is the result of an alloimmune reaction to “foreign” α3α4α5(IV) collagen present in the allograft GBM but absent from the Alport patient’s tissues. APTN is most prevalent in patients with X-linked Alport syndrome, who develop alloantibodies against several alloepitopes within the α5NC1 domain (17). Upon binding to the allograft GBM, APTN alloantibodies cause aggressive glomerulonephritis with similar clinical presentation and pathology findings as in autoimmune anti-GBM disease (19). However, the APTN alloepitopes are accessible in α3α4α5NC1 hexamers of the human GBM, unlike the GP autoepitopes (4, 17).

Whether differences in the epitope specificity between GP autoantibodies and APTN alloantibodies are pathogenically relevant is not known. We postulated that APTN alloantibodies are more nephritogenic than GP autoantibodies because they bind to all isoforms of α3α4α5NC1 hexamers from the GBM. Testing this hypothesis requires a suitable animal model. A landmark study has demonstrated that GP autoantibodies injected into squirrel monkeys bind to the GBM of the recipient host, causing severe glomerulonephritis (20). However, the nephritogenicity of APTN alloantibodies has not been evaluated by passive transfer into animal models.

The purpose of the present study was to determine whether the relative inaccessibility of B cell autoepitopes in the GBM limits the severity of autoantibody-mediated glomerulonephritis. Because rodent models are preferable to nonhuman primates on humane grounds, we developed a mouse model of anti-GBM glomerulonephritis by passive immunization with human anti-GBM Abs. We show that APTN alloantibodies but not GP autoantibodies bound to mouse GBM in vivo, causing crescentic glomerulonephritis in susceptible mouse strains. Although cross-reactive with murine α3NC1 subunits in vitro, GP autoantibodies did not bind to native α3α4α5NC1 hexamers of the mouse GBM in vivo, nor did they induce disease. Resistance against GP autoantibody-mediated nephritis was due to quinary2 (intermolecular) NC1 crosslinks locking the cryptic GP autoepitopes in the mouse GBM. In contrast, the GBM of squirrel monkeys contained non-crosslinked α3NC1 subunits reactive with GP autoantibodies under native conditions. These findings indicate that quinary crosslinks of α3α4α5NC1 hexamers, recently identified as sulfilimine bonds (22), lock the cryptic GP autoepitopes, conferring protection against nephritis mediated by GP autoantibodies, but not against APTN alloantibody-mediated disease. Thus, potentially pathogenic autoantibodies can be uncoupled from autoimmune disease by posttranslational modifications of an autoantigen that render B cell autoepitopes structurally inaccessible in tissues.

Previously characterized serum or kidney-eluted anti-GBM alloantibodies from four X-linked Alport syndrome patients who developed APTN (17), as well as sera or plasma exchange fluid from six patients with GP disease (18), were used for in vitro binding studies. Only kidney-eluted alloantibodies from one patient with APTN were available in sufficient amounts for in vivo studies. Purified GP IgG autoantibodies used for in vivo experiments were isolated from the plasma exchange fluid of two patients by affinity chromatography on immobilized human α3NC1 (4). Normal human IgG (hIgG) was purchased from Sigma-Aldrich (St. Louis, MO) or purified from healthy donor sera by affinity chromatography on immobilized protein A. Nephrotoxic nephritis serum from sheep immunized with mouse GBM was a gift from Dr. A. Richard Kitching (Monash University, Clayton, Victoria, Australia).

Mouse kidneys were purchased from Pel-Freez Biologicals (Rogers, AR). Frozen kidneys from adult squirrel monkeys (Saimiri sp.) were obtained from the Squirrel Monkey Breeding and Research Resource (University of Texas MD Anderson Cancer Center, Houston, TX). Renal basement membranes including the GBM were isolated from homogenized kidney cortex and digested with bacterial collagenase (Worthington, Lakewood, NJ). The NC1 hexamers thus solubilized were purified by passage through a DE-52 ion-exchange column and gel-filtration chromatography (23).

Indirect ELISA was performed as described (17). To assay the binding of human anti-GBM Abs to mouse Ags, Nunc MaxiSorp 96-well plates were coated overnight with NC1 hexamers from mouse GBM (300 ng/well) in PBS (pH 7.4). Prior to coating, some NC1 hexamers were treated with 6 M guanidine hydrochloride for 15 min at 60°C for dissociation into subunits. Sera diluted 1:100 or purified hIgG Abs (100 ng/ml) were used as primary Abs. The hIgG bound to the immobilized Ags was detected with alkaline phosphatase-conjugated goat anti-human IgG (Rockland Immunochemical, Gilbertsville, PA). To assay the circulating mouse anti-human IgG Abs, wells coated with normal human IgG (300 ng/well) were incubated with mouse sera diluted 1:200, followed by detection with HRP-conjugated goat Abs specific for mouse IgG1, IgG2b, or IgG2c subclasses (Bethyl Laboratories, Montgomery, TX). The levels of human IgG in mouse sera were assayed by capture ELISA in plates coated with goat anti-human IgG, Fc-specific (300 ng/well).

Western blots were performed as described (17). NC1 hexamers from the mouse or human GBM (1 μg/lane) were separated by SDS-PAGE and transferred onto Immobilon-P (Millipore, Bedford, MA). The membrane was blocked with 5% of blotting grade non-fat dry milk and cut into strips. Strips were incubated with purified hIgG Abs (1 μg/ml) or anti-α3NC1 mAb 8D1, which were detected with alkaline phosphatase–conjugated secondary Abs. For immunoprecipitation, NC1 hexamers (20 μg) isolated from mouse or squirrel monkey kidneys were incubated overnight at 4°C with purified hIgG Abs (20 μg). Ag–Ab complexes bound to protein G–Sepharose 4 Fast Flow (GE Healthcare Bio-Sciences, Piscataway, NJ) were solubilized in sample buffer, separated by SDS-PAGE, and transferred onto Immobilon-P for blotting with mAbs specific for NC1 domains (24).

Cryostat sections (5 μm) of snap-frozen mouse kidneys embedded in OCT were fixed in acetone at −20°C for 10 min. Some sections were treated for 5 min on ice with 6 M urea in 0.1 M glycine (pH 2.2) to unmask cryptic epitopes. After blocking with 3% normal goat serum and 3% bovine albumin in PBS for 45 min, appropriately diluted primary Abs were added for 1 h, and then the sections were stained with Alexa Fluor 488 goat anti-human IgG (H+L) (Invitrogen, Eugene, OR), polyclonal anti-rat Ig (BD Pharmingen, San Jose, CA), or FITC-conjugated donkey anti-sheep IgG (Jackson ImmunoResearch Laboratories, West Grove, PA). Stained sections were observed with an Axioplan 2 fluorescence microscope (Carl Zeiss MicroImaging, Thornwood, NY) and images were captured with the AxioVision 4.8 software.

C57BL/6J (B6) and B6.Col4a5−/− mice were purchased from The Jackson Laboratory (Bar Harbor, ME), and B6.FcγRIIB−/− mice were from Taconic Farms (Germantown, NY). The mice were maintained in a specific pathogen-free facility with access to water and food. All mouse studies were performed in accordance with the principles for humane treatment of laboratory animals and were approved by the Institutional Animal Care and Use Committee at Vanderbilt University. Each mouse was injected once in the tail vein with 0.1 mg of purified hIgG in 100 μl of PBS, divided in four groups: kidney-eluted APTN alloantibodies, affinity-purified GP-1 or GP-2 autoantibodies, and normal human serum IgG. Blood was collected at the indicated time points by saphenous vein puncture, along with spot urine samples. Blood urea nitrogen was measured by a kinetic urease and glutamate dehydrogenase assay, and urinary creatinine was measured by alkaline picrate (Jaffe reaction), using kits from Thermo Electron (Louisville, CO). Urinary albumin excretion was assayed using a mouse albumin ELISA quantitation kit (Bethyl Laboratory, Montgomery, TX). Portions of kidneys and lungs collected from mice sacrificed at the end of experiments were embedded in paraffin or OCT for further evaluation. For light microscopic assessment of the kidney histopathology, 5-μm paraffin sections stained with H&E or with periodic acid–Schiff reagent were observed with an Axioskop 40 microscope. Images were captured with MRGrab software (Carl Zeiss MicroImaging). For direct immunofluorescence, frozen sections of OCT-embedded kidneys were fixed in acetone and stained with FITC-conjugated goat anti-mouse Ig (BD Pharmingen), Alexa Fluor 488 goat anti-human IgG (H+L), or FITC-conjugated goat anti-mouse C3c (Nordic Immunological Laboratory, Tilburg, The Netherlands).

Data were analyzed using GraphPad Prism (GraphPad Software, San Diego, CA), by two-tailed t test for two groups or ANOVA followed by Dunnett’s post hoc test for three or more groups. Statistical significance was inferred for p values <0.05.

The lack of cross-reactivity between human autoantibodies and homologous Ags from other species is a potential roadblock preventing the transfer of autoimmune disease to animal models by passive immunization with patient autoantibodies (25). To determine whether mice are a suitable model for this study, we first verified that human anti-GBM Abs cross-react with mouse collagen IV. By Western blot, NC1 domains from mouse GBM bound hIgG from patients with GP disease or APTN, but not normal hIgG (Fig. 1A). Next, we assessed the accessibility of the epitopes in the mouse GBM. By indirect ELISA (Fig. 1B), only APTN alloantibodies exhibited significant reactivity toward native NC1 hexamers from the mouse GBM, whereas GP autoantibodies bound preferentially to dissociated hexamers, revealing the crypticity of GP autoepitopes. This finding was corroborated by indirect immunofluorescence staining (Fig. 1C). Under native conditions, only APTN alloantibodies stained the GBM and tubular basement membranes of normal mouse kidneys. GP autoantibodies produced the same staining pattern on kidney sections treated with acid urea for epitope unmasking, but did not bind to mouse basement membranes under native conditions. The same results were found using kidneys from C57BL/6, 129/Sv, DBA/1, and SJL/J mice (Supplemental Fig. 1). Additionally, the alveolar basement membranes of mouse lungs were stained by APTN alloantibodies under native conditions, but by GP autoantibodies only after acid urea treatment (Supplemental Fig. 2). Finally, we verified that both APTN alloantibodies and GP autoantibodies bind specifically to mouse α3α4α5(IV) collagen because neither stained kidney basement membranes from Col4a5-null Alport mice (Fig. 1D), in contrast to sheep anti-mouse GBM Abs used to elicit murine nephrotoxic serum nephritis, which appear to target other Ags in the mouse GBM. We have thus shown that NC1 hexamers of mouse α3α4α5(IV) collagen harbor cryptic autoepitopes recognized by GP autoantibodies and accessible alloepitopes targeted by APTN alloantibodies.

FIGURE 1.

APTN alloantibodies and GP autoantibodies cross-react with the NC1 domains of mouse α3α4α5(IV) collagen. A, Western blots show the binding of hIgG to NC1 dimer and monomer subunits of hexamers from the mouse or human GBM, after incubation with GP autoantibodies, APTN alloantibodies, or normal human serum IgG. Staining with mAb 8D1 identifies the position and relative abundance of α3NC1 monomers and dimers. B, The binding of APTN alloantibodies (n = 4), GP autoantibodies (n = 4), and normal human sera IgG (n = 3) to NC1 hexamers from mouse GBM (300 ng/well), native (●) or dissociated (○), was measured by indirect ELISA (left). After subtracting the nonspecific binding to BSA (x), the relative hIgG binding to native versus dissociated mouse hexamers was calculated for each sample (right). Unlike APTN alloantibodies, GP autoantibodies reacted very weakly with native mouse GBM hexamers (***p < 0.0001). C, Binding of APTN alloantibodies and GP autoantibodies to mouse kidney basement membranes under native conditions or after acid urea treatment was evaluated by indirect immunofluorescence staining. D, Indirect immunofluorescence staining of kidneys from B6.Col4a5+/Y (wild type) and B6.Col4a5/Y (Alport) mice with APTN alloantibodies, GP autoantibodies, a mAb specific for mouse α3α4α5(IV) collagen, and nephrotoxic serum from sheep immunized with mouse GBM. Staining with GP autoantibodies was performed on acid urea–treated kidney sections. Original magnification ×400. D, dimer; M, monomer; NHS, normal human serum; NTS, nephrotoxic serum.

FIGURE 1.

APTN alloantibodies and GP autoantibodies cross-react with the NC1 domains of mouse α3α4α5(IV) collagen. A, Western blots show the binding of hIgG to NC1 dimer and monomer subunits of hexamers from the mouse or human GBM, after incubation with GP autoantibodies, APTN alloantibodies, or normal human serum IgG. Staining with mAb 8D1 identifies the position and relative abundance of α3NC1 monomers and dimers. B, The binding of APTN alloantibodies (n = 4), GP autoantibodies (n = 4), and normal human sera IgG (n = 3) to NC1 hexamers from mouse GBM (300 ng/well), native (●) or dissociated (○), was measured by indirect ELISA (left). After subtracting the nonspecific binding to BSA (x), the relative hIgG binding to native versus dissociated mouse hexamers was calculated for each sample (right). Unlike APTN alloantibodies, GP autoantibodies reacted very weakly with native mouse GBM hexamers (***p < 0.0001). C, Binding of APTN alloantibodies and GP autoantibodies to mouse kidney basement membranes under native conditions or after acid urea treatment was evaluated by indirect immunofluorescence staining. D, Indirect immunofluorescence staining of kidneys from B6.Col4a5+/Y (wild type) and B6.Col4a5/Y (Alport) mice with APTN alloantibodies, GP autoantibodies, a mAb specific for mouse α3α4α5(IV) collagen, and nephrotoxic serum from sheep immunized with mouse GBM. Staining with GP autoantibodies was performed on acid urea–treated kidney sections. Original magnification ×400. D, dimer; M, monomer; NHS, normal human serum; NTS, nephrotoxic serum.

Close modal

In vivo binding of human anti-GBM Abs to target Ags was investigated in C57BL/6J mice injected with APTN alloantibodies, GP autoantibodies from two patients, and hIgG from normal human sera. Renal function was monitored for 8 wk postinjection to assess the development of kidney injury. At all time points, urinary albumin excretion (Fig. 2A) and blood urea nitrogen levels (Fig. 2B) were found to be normal in all mice. Light microscopic examination of the kidneys revealed no histological abnormalities in any group (Fig. 2C,a–f). Direct immunofluorescence revealed bright linear staining for hIgG along the GBM in all mice injected with APTN alloantibodies (Fig. 2C,g), but no hIgG deposition was detected in the kidneys of mice injected with either GP autoantibodies (Fig. 2C,h) or normal hIgG (Fig. 2C,i). Linear GBM deposition of mouse IgG (mIgG) (Fig. 2C,j) but not of mC3 (Fig. 2C m) was also observed in APTN-injected mice, indicating that these mice produced Abs to the foreign globulin. APTN alloantibodies stained in vitro but did not bind in vivo to other murine basement membranes containing α3α4α5(IV) collagen, such as renal tubular basement membranes or lung alveolar basement membranes (Supplemental Fig. 2). These results show that APTN alloantibodies, unlike GP autoantibodies, bind specifically to the mouse GBM in vivo, but do not induce glomerulonephritis in C57BL/6J mice.

FIGURE 2.

APTN alloantibodies but not GP autoantibodies bind in vivo to the mouse GBM. C57BL/6J mice (n = 4/group) were injected with 0.1 mg of APTN alloantibodies, GP-1 or GP-2 autoantibodies, or normal human serum IgG. A and B, Kidney function was evaluated by measuring urinary albumin-to-creatinine ratios (A) and blood urea nitrogen levels (B) at the times indicated. The graphs depict the means and SEM. Differences among groups of mice injected with different human IgG Abs were not statistically significant. C, Kidney histopathology in mice sacrificed at 8 wk after the injection of APTN alloantibodies (a, d, g, j, m), GP autoantibodies (b, e, h, k, n), or normal hIgG (c, f, i, l, o) was evaluated by light microscopy on paraffin sections stained by H&E (a–c) or periodic acid–Schiff (d–f). Deposition of hIgG (g–i), mIgG (j–l), and mC3 (m–o) in the mouse glomeruli was evaluated by direct immunofluorescence staining. Original magnification ×400. G, glomeruli; NHS, normal human serum.

FIGURE 2.

APTN alloantibodies but not GP autoantibodies bind in vivo to the mouse GBM. C57BL/6J mice (n = 4/group) were injected with 0.1 mg of APTN alloantibodies, GP-1 or GP-2 autoantibodies, or normal human serum IgG. A and B, Kidney function was evaluated by measuring urinary albumin-to-creatinine ratios (A) and blood urea nitrogen levels (B) at the times indicated. The graphs depict the means and SEM. Differences among groups of mice injected with different human IgG Abs were not statistically significant. C, Kidney histopathology in mice sacrificed at 8 wk after the injection of APTN alloantibodies (a, d, g, j, m), GP autoantibodies (b, e, h, k, n), or normal hIgG (c, f, i, l, o) was evaluated by light microscopy on paraffin sections stained by H&E (a–c) or periodic acid–Schiff (d–f). Deposition of hIgG (g–i), mIgG (j–l), and mC3 (m–o) in the mouse glomeruli was evaluated by direct immunofluorescence staining. Original magnification ×400. G, glomeruli; NHS, normal human serum.

Close modal

The development of experimental glomerulonephritis after passive immunization with human anti-GBM IgG Abs was next investigated using B6.Fcgr2b−/− mice, because the absence of the inhibitory IgG Fc receptor FcγRIIB exacerbates susceptibility to Ab-mediated inflammation and autoimmune disease (26, 27). Injection of APTN alloantibodies (but not of GP-1 and GP-2 autoantibodies, nor normal hIgG) into Fcgr2b−/− mice caused increased albuminuria after 2 wk (Fig. 3A) and abnormally high blood urea nitrogen levels after 4 wk (Fig. 3B). By 8 wk, 80% (four out of five) of mice injected with APTN alloantibodies died or were sacrificed because of end-stage renal disease, compared with zero of eight mice injected with GP-1 or GP-2 autoantibodies. Light microscopic examination of the kidneys revealed crescentic necrotizing glomerulonephritis in APTN-treated mice, but not in those receiving GP autoantibodies or normal hIgG (Fig. 3C,a–f). Direct immunofluorescence staining of kidneys from Fcgr2b−/− mice injected with APTN alloantibodies revealed linear GBM binding of hIgG, along with mIgG and mC3 deposition along the capillary loops (Fig. 3C,g, j, m). However, APTN alloantibodies did not bind in vivo to mouse renal tubular or alveolar basement membranes containing α3α4α5(IV) collagen (Supplemental Fig. 2). Weak staining for hIgG, mIgG, and mC3 in a mesangial pattern was found in Fcgr2b−/− mice injected with GP autoantibodies (Fig. 3C,h, k, n) or normal hIgG (Fig. 3C,i, l, o). This pattern of hIgG deposition, judged to be nonspecific because it did not colocalize with mouse α3α4α5(IV) collagen (Fig. 3C,p, q, r), may be caused by immune complexes between hIgG and mouse anti-human IgG Abs (MAHA). MAHA were detected in mouse sera by 2 wk after passive immunization (Fig. 4A), and rising titers of MAHA paralleled a decrease in the serum levels of hIgG (Fig. 4B). In mice sacrificed at 3 d after the injection of human Abs, prior to the formation of MAHA, nonspecific mesangial deposits of hIgG were not observed (Fig. 4C). Taken together, these results demonstrate that only APTN alloantibodies bind specifically to the GBM of B6.FcγRIIB−/− mice, causing crescentic glomerulonephritis, whereas GP autoantibodies and normal hIgG are significantly less nephritogenic because neither binds specifically to the mouse GBM in vivo.

FIGURE 3.

APTN alloantibodies but not GP autoantibodies induce crescentic glomerulonephritis in Fcgr2b−/− mice. B6.Fcgr2b−/− mice were injected with 0.1 mg of APTN alloantibodies (n = 5 mice), GP-1 or GP-2 autoantibodies (n = 4 mice each), or normal human serum IgG (n = 4 mice). A and B, Kidney function was evaluated by measuring urinary albumin-to-creatinine ratios (A) and blood urea nitrogen levels (B) at the times indicated. The graphs depict the means and SEM. Mice injected with APTN alloantibodies, but not those injected with GP-1 or GP-2 autoantibodies, developed significantly higher albuminuria and blood urea nitrogen levels than did control mice injected with normal hIgG (p < 0.05 by ANOVA followed by Dunnett’s multiple comparison test). C, Kidney histopathology in Fcgr2b−/− mice sacrificed at 8 wk after the injection of APTN alloantibodies (a, d, g, j, m, p), GP autoantibodies (b, e, h, k, n, q), or normal human IgG (c, f, i, l, o, r) was evaluated by light microscopy on paraffin sections stained by H&E (a–c) or periodic acid–Schiff (d–f). Glomerular deposition of hIgG (g–i), mIgG (j–l), and mC3 (m–o) was evaluated by direct immunofluorescence staining of frozen kidney sections. Original magnification ×400. Double staining of frozen kidney sections (p–r) assessing the colocalization of hIgG (green) with mouse α3α4α5(IV) collagen (red) is shown. NHS, normal human serum.

FIGURE 3.

APTN alloantibodies but not GP autoantibodies induce crescentic glomerulonephritis in Fcgr2b−/− mice. B6.Fcgr2b−/− mice were injected with 0.1 mg of APTN alloantibodies (n = 5 mice), GP-1 or GP-2 autoantibodies (n = 4 mice each), or normal human serum IgG (n = 4 mice). A and B, Kidney function was evaluated by measuring urinary albumin-to-creatinine ratios (A) and blood urea nitrogen levels (B) at the times indicated. The graphs depict the means and SEM. Mice injected with APTN alloantibodies, but not those injected with GP-1 or GP-2 autoantibodies, developed significantly higher albuminuria and blood urea nitrogen levels than did control mice injected with normal hIgG (p < 0.05 by ANOVA followed by Dunnett’s multiple comparison test). C, Kidney histopathology in Fcgr2b−/− mice sacrificed at 8 wk after the injection of APTN alloantibodies (a, d, g, j, m, p), GP autoantibodies (b, e, h, k, n, q), or normal human IgG (c, f, i, l, o, r) was evaluated by light microscopy on paraffin sections stained by H&E (a–c) or periodic acid–Schiff (d–f). Glomerular deposition of hIgG (g–i), mIgG (j–l), and mC3 (m–o) was evaluated by direct immunofluorescence staining of frozen kidney sections. Original magnification ×400. Double staining of frozen kidney sections (p–r) assessing the colocalization of hIgG (green) with mouse α3α4α5(IV) collagen (red) is shown. NHS, normal human serum.

Close modal
FIGURE 4.

Time course for hIgG clearance from mouse sera, MAHA production, and glomerular IgG deposition at 3 d after passive immunization. A, Serum levels of MAHA in the Fcgr2b−/− mice injected with hIgG were measured by indirect ELISA. B, The levels of hIgG in mouse sera were measured by capture ELISA. C, Indirect immunofluorescence analysis of hIgG or mIgG deposition in kidneys from Fcgr2b−/− mice at 3 d after injection of APTN alloantibodies, GP autoantibodies, or normal hIgG. At this time point, before production of MAHA, kidney histology appeared normal by light microscopy (not shown), and nonspecific glomerular deposits of hIgG were not observed. Original magnification ×400.

FIGURE 4.

Time course for hIgG clearance from mouse sera, MAHA production, and glomerular IgG deposition at 3 d after passive immunization. A, Serum levels of MAHA in the Fcgr2b−/− mice injected with hIgG were measured by indirect ELISA. B, The levels of hIgG in mouse sera were measured by capture ELISA. C, Indirect immunofluorescence analysis of hIgG or mIgG deposition in kidneys from Fcgr2b−/− mice at 3 d after injection of APTN alloantibodies, GP autoantibodies, or normal hIgG. At this time point, before production of MAHA, kidney histology appeared normal by light microscopy (not shown), and nonspecific glomerular deposits of hIgG were not observed. Original magnification ×400.

Close modal

We next inquired why APTN alloantibodies elicit severe glomerulonephritis in Fcgr2b−/− but not in wild-type mice. Because the injection of APTN alloantibodies led to C3 deposition along the capillary loops in Fcgr2b−/− but not in wild-type mice, we hypothesized that the absence of the inhibitory IgG Fc receptor may augment production of complement-fixing mIgG subclasses, linking the GBM-bound hIgG to inflammatory effectors. For all mIgG subclasses, the titers of MAHA were consistently higher in Fcgr2b−/− mice than in wild-type mice, but they were less influenced by the nature of hIgG injected into mice (Fig. 5). In wild-type mice, MAHA were almost exclusively mIgG1, whereas Fcgr2b−/− mice also produced significant amounts of mIgG2b and mIgG2c MAHA. These results suggest that mIgG2b and/or mIgG2c Abs to hIgG may exacerbate glomerulonephritis in Fcgr2b−/− mice injected with APTN alloantibodies.

FIGURE 5.

Fcgr2b−/− mice have enhanced production of mIgG2b and mIgG2c anti-hIgG Abs (MAHA). Titers of circulating MAHA of mIgG (A), mIgG1 (B), mIgG2b (C), and mIgG2c (D) subclasses were measured by indirect ELISA. Sera collected from wild-type (filled bars) and Fcgr2b−/− mice (open bars) at 4 wk after the injection of patient alloantibodies (APTN), autoantibodies (GP-1, GP-2), or normal hIgG were diluted 1/500. Binding of mIgG to plates coated with normal hIgG was assayed using subclass-specific secondary Abs. The graphs depict the means and SEM of ELISA absorbance values for four or five mice in each group. The statistical significance of differences in MAHA titers between wild-type and Fcgr2b−/− mice was analyzed by two-way ANOVA followed by Bonferroni posttests (*p < 0.05; **p < 0.01; ***p < 0.0001). NHS, normal human serum.

FIGURE 5.

Fcgr2b−/− mice have enhanced production of mIgG2b and mIgG2c anti-hIgG Abs (MAHA). Titers of circulating MAHA of mIgG (A), mIgG1 (B), mIgG2b (C), and mIgG2c (D) subclasses were measured by indirect ELISA. Sera collected from wild-type (filled bars) and Fcgr2b−/− mice (open bars) at 4 wk after the injection of patient alloantibodies (APTN), autoantibodies (GP-1, GP-2), or normal hIgG were diluted 1/500. Binding of mIgG to plates coated with normal hIgG was assayed using subclass-specific secondary Abs. The graphs depict the means and SEM of ELISA absorbance values for four or five mice in each group. The statistical significance of differences in MAHA titers between wild-type and Fcgr2b−/− mice was analyzed by two-way ANOVA followed by Bonferroni posttests (*p < 0.05; **p < 0.01; ***p < 0.0001). NHS, normal human serum.

Close modal

We finally addressed the molecular basis of the inaccessibility of GP autoepitopes in the mouse GBM. Because GP autoantibodies mediate anti-GBM disease in human patients and squirrel monkeys (20) but not in mice, we hypothesized that species-specific structural features of GBM collagen IV may affect its immunoreactivity toward GP autoantibodies, thus determining the host’s susceptibility to GP autoantibody-mediated glomerulonephritis. We therefore analyzed how human anti-GBM Abs bind to NC1 domains from the GBM of mice and squirrel monkeys using immunoprecipitation (Fig. 6A). As reference, Fig. 6B depicts the reactivity of NC1 hexamer isoforms from human GBM toward GP autoantibodies (14) and APTN alloantibodies (17). In both monkeys and mice, APTN alloantibodies coprecipitated α3NC1, α4NC1, and α5NC1 monomers and dimers (Fig. 6A, lanes c, g). This pattern of reactivity, also seen with human GBM NC1 domains (17), indicates that APTN alloantibodies bind to native α3α4α5NC1 hexamers from all species (Fig. 7A, 7C, 7D). Under native conditions, GP autoantibodies mainly targeted α3NC1 and α4NC1 monomers from the monkey GBM (Fig 6A, lane b). This pattern of reactivity, previously found for human GBM NC1 domains (14), suggests that the GBM of squirrel monkeys also contains non-crosslinked hexamers (Fig. 6B). In contrast, GP autoantibodies did not bind to native NC1 hexamers from murine GBM (Fig 6A, lane f), indicating that non-crosslinked hexamers do not occur in the mouse kidneys. Nevertheless, GP autoantibodies bound to α3NC1 monomer and α3–α5NC1 dimer subunits produced by in vitro dissociation of mouse GBM NC1 hexamers (Fig. 6A, lane i). Therefore, GP autoepitopes occur in the mouse GBM only within isoforms of autoantigen impenetrable by GP autoantibodies, namely crosslinked NC1 hexamers containing NC1 dimer subunits. Consequently, mice are resistant to GP autoantibody-mediated nephritis (Fig. 7E).

FIGURE 6.

The binding of human anti-GBM Abs to NC1 domains from squirrel monkey and mouse kidneys. A, Purified GP autoantibodies (b, f, i), APTN alloantibodies (c, g, j), and normal human serum IgG (e, h) were incubated with native NC1 hexamers from squirrel monkeys (a–c) or mouse (d–g) kidneys, or with mouse NC1 subunits produced by dissociating the hexamers with 6 M guanidinium hydrochloride (h–j). After immunoprecipitating the immune complexes with protein G beads, the NC1 domains bound to human anti-GBM Abs were resolved into dimer and monomer subunits by SDS-PAGE and identified by Western blot with mAbs specific for α3, α4, and α5NC1 domains. B, Schematic depiction of NC1 hexamer isoforms identified in the human GBM (14) and their reactivity toward APTN alloantibodies and GP autoantibodies. The isoforms differ in the number of component NC1 dimer subunits produced by quinary NC1 crosslinks (horizontal lines). All hexamer isoforms react with APTN alloantibodies targeting accessible epitopes. Non-crosslinked hexamers react with GP autoantibodies, with concomitant dissociation. Native crosslinked hexamers do not react with GP autoantibodies, unless dissociated into subunits. D, dimer; M, monomer; NHS, normal human serum.

FIGURE 6.

The binding of human anti-GBM Abs to NC1 domains from squirrel monkey and mouse kidneys. A, Purified GP autoantibodies (b, f, i), APTN alloantibodies (c, g, j), and normal human serum IgG (e, h) were incubated with native NC1 hexamers from squirrel monkeys (a–c) or mouse (d–g) kidneys, or with mouse NC1 subunits produced by dissociating the hexamers with 6 M guanidinium hydrochloride (h–j). After immunoprecipitating the immune complexes with protein G beads, the NC1 domains bound to human anti-GBM Abs were resolved into dimer and monomer subunits by SDS-PAGE and identified by Western blot with mAbs specific for α3, α4, and α5NC1 domains. B, Schematic depiction of NC1 hexamer isoforms identified in the human GBM (14) and their reactivity toward APTN alloantibodies and GP autoantibodies. The isoforms differ in the number of component NC1 dimer subunits produced by quinary NC1 crosslinks (horizontal lines). All hexamer isoforms react with APTN alloantibodies targeting accessible epitopes. Non-crosslinked hexamers react with GP autoantibodies, with concomitant dissociation. Native crosslinked hexamers do not react with GP autoantibodies, unless dissociated into subunits. D, dimer; M, monomer; NHS, normal human serum.

Close modal
FIGURE 7.

Schematic representation of the mechanisms of anti-GBM Ab-mediated glomerulonephritis in human patients and mouse models. In human patients, APTN alloantibodies (A) bind to all isoforms of α3α4α5NC1 hexamers, whereas GP autoantibodies (B) bind to α3α4α5NC1 hexamers lacking NC1 crosslinks (indicated by horizontal bars), directly engaging effectors of Ab-mediated inflammation (complement and/or activating IgG Fc receptors) and causing glomerulonephritis. APTN alloantibodies injected into mice also bind to crosslinked α3α4α5NC1 hexamers in the murine GBM (C, D), where they fix MAHA. GBM fixation of mIgG2a/c and IgG2b in Fcgr2b−/− mice (D) but not of mIgG1 in wild-type mice (C) then leads to crescentic glomerulonephritis. Passively transferred GP autoantibodies do not bind to crosslinked α3α4α5NC1 hexamers in the mouse GBM (E), and hence they do not induce glomerulonephritis, but elicit MAHA, likely forming circulating immune complexes. In murine glomerulonephritis induced by immunization with heterologous α3NC1 Ags (F), nephrotropic mIgG Abs probably target epitopes of α3α4α5NC1 hexamers accessible in the GBM. In Fcgr2b−/− mice immunized with Cellmatrix-IV (G), some mIgG autoantibodies may bind to the collagenous domain of murine α1α2(IV) collagen. In murine models of nephrotoxic nephritis (H), heterologous globulin bound to unknown GBM Ags acts as a planted Ag to fix mIgG subclasses mediating glomerulonephritis. C.I.C., circulating immune complexes; GN, glomerulonephritis.

FIGURE 7.

Schematic representation of the mechanisms of anti-GBM Ab-mediated glomerulonephritis in human patients and mouse models. In human patients, APTN alloantibodies (A) bind to all isoforms of α3α4α5NC1 hexamers, whereas GP autoantibodies (B) bind to α3α4α5NC1 hexamers lacking NC1 crosslinks (indicated by horizontal bars), directly engaging effectors of Ab-mediated inflammation (complement and/or activating IgG Fc receptors) and causing glomerulonephritis. APTN alloantibodies injected into mice also bind to crosslinked α3α4α5NC1 hexamers in the murine GBM (C, D), where they fix MAHA. GBM fixation of mIgG2a/c and IgG2b in Fcgr2b−/− mice (D) but not of mIgG1 in wild-type mice (C) then leads to crescentic glomerulonephritis. Passively transferred GP autoantibodies do not bind to crosslinked α3α4α5NC1 hexamers in the mouse GBM (E), and hence they do not induce glomerulonephritis, but elicit MAHA, likely forming circulating immune complexes. In murine glomerulonephritis induced by immunization with heterologous α3NC1 Ags (F), nephrotropic mIgG Abs probably target epitopes of α3α4α5NC1 hexamers accessible in the GBM. In Fcgr2b−/− mice immunized with Cellmatrix-IV (G), some mIgG autoantibodies may bind to the collagenous domain of murine α1α2(IV) collagen. In murine models of nephrotoxic nephritis (H), heterologous globulin bound to unknown GBM Ags acts as a planted Ag to fix mIgG subclasses mediating glomerulonephritis. C.I.C., circulating immune complexes; GN, glomerulonephritis.

Close modal

Anti-GBM Abs binding to the NC1 domains of α3α4α5(IV) collagen in the human GBM mediate the most aggressive forms of rapid progressive glomerulonephritis in patients with APTN (Fig. 7A) or GP disease (Fig. 7B). Because GP autoantibodies and APTN alloantibodies bind to cryptic and accessible GBM epitopes, respectively, we investigated the role of the accessibility of B cell epitopes in vivo in the pathogenesis of these diseases. Crypticity of GP autoepitopes led us to conjecture that APTN alloantibodies would cause more severe experimental nephritis than would GP autoantibodies in a passive transfer model of anti-GBM disease. We found that only APTN alloantibodies bound in vivo to the mouse GBM, inducing crescentic glomerulonephritis in Fcgr2b−/− mice susceptible to Ab-mediated inflammation (Fig. 7C, 7D). We have thus demonstrated the nephritogenic potential of APTN alloantibodies and established the first murine model of glomerulonephritis induced by a patient’s anti-GBM Abs that reproduces the severity of human disease. In contrast, GP autoantibodies neither bound to mouse α3α4α5(IV) collagen in vivo nor induced experimental nephritis (Fig. 7E), owing to quinary NC1 crosslinks locking the GP autoepitopes in the mouse GBM. These results show that the crypticity of B cell epitopes in tissues is sufficient to uncouple potentially pathogenic autoantibodies from autoimmune disease.

Our study illustrates how seemingly small differences in the location of epitopes targeted by anti-GBM Abs impact the development of glomerulonephritis. The different accessibility of the collagen IV autoepitopes and alloepitopes in the GBM is thought to reflect the distinct immune mechanisms triggering GP disease and APTN (2830). The crypticity of GP autoepitopes in α3α4α5NC1 hexamers implies that a perturbation of the native structure of the autoantigen may be instrumental in breaching immune self-tolerance toward α3(IV) collagen in GP disease. In contrast, APTN alloantibodies target accessible alloepitopes in the α3α4α5NC1 hexamers—antigenic determinants encountered in the normal renal allograft but not in the basement membranes of the affected patients, who presumably lack immune tolerance toward missing collagen IV chains. In this study, both GP autoantibodies and APTN alloantibodies were shown to cross-react with NC1 subunits of murine α3α4α5(IV) collagen in vitro, yet only the GP autoepitopes were completely inaccessible within the collagen IV network of mouse GBM in vivo. As a result, mice were protected against anti-GBM glomerulonephritis induced by GP autoantibodies, but not by APTN alloantibodies.

Because relatively large amounts of Ag-specific hIgG were used for passive transfer experiments, only anti-GBM Abs from a small number of patients could be tested in vivo. Nevertheless, these samples were judged to be representative based on their in vitro binding to mouse GBM Ags, which are essentially identical to additional samples of GP autoantibodies and APTN alloantibodies tested only in vitro. We also note that APTN alloantibodies injected in mice were eluted from an allograft kidney, and hence they were highly enriched in the nephrotropic hIgG species. In all in vitro immunoassays, kidney-eluted and serum APTN alloantibodies bound to mouse GBM Ags similar to each other but different from GP autoantibodies (Fig. 1 and Supplemental Fig. 1). However, kidney-eluted APTN alloantibodies have higher affinity for α3α4α5NC1 hexamers than do serum APTN alloantibodies (17), which may translate into greater nephritogenic potential. To emulate the enrichment in autoantigen-specific hIgG species upon kidney binding and the subsequent elution, GP autoantibodies used for passive immunizations were affinity-purified from plasma exchange fluid using immobilized α3NC1 monomers.

The proximal step initiating crescentic glomerulonephritis in mice injected with APTN alloantibodies is the GBM deposition of hIgG. Prerequisites for this binding are the ability of human APTN alloantibodies to cross-react with mouse α3α4α5(IV) collagen, as well as the structural accessibility of the alloepitopes within the α3α4α5NC1 hexamers (Fig. 1). However, other murine basement membranes containing α3α4α5(IV) collagen, such as those in renal tubules and lung alveoli, bind APTN alloantibodies in vitro (by indirect immunofluorescence staining) but not in vivo. APTN alloantibodies readily bind to the GBM because fenestrations in glomerular endothelial cells allow the passage of large macromolecules, which is an adaptation to the blood filtration function. In most capillary beds, endothelial cells have tight junctions impermeable to molecules of IgG size, which prevent the access of circulating Abs to the underlying basement membranes under normal physiological conditions. Thus, cellular and anatomic barriers can modulate the accessibility of (auto)antigens in vivo, influencing which tissues become targets of Ab-mediated inflammation.

In vivo binding of APTN alloantibodies to the mouse GBM was necessary but not sufficient for induction of glomerulonephritis, because only Fcgr2b−/− mice were susceptible to disease. This echoes the finding that hIgG autoantibodies from patients with rheumatoid arthritis are arthritogenic in Fcgr2b−/− but not in wild-type mice (27). Several mechanisms may contribute to the susceptibility of Fcgr2b−/− mice to Ab-mediated glomerulonephritis. Because the expression of the inhibitory receptor FcγRIIB on myeloid cells counterbalances signaling through activating IgG Fc receptors (FcγRI, -III, and -IV), neutrophils and macrophages from Fcgr2b−/− mice would have a lower threshold of activation by immune complexes (3). FcγRIIB is also a negative regulator of B cells, implicated in the feedback inhibition of Ab production or apoptosis of plasma cells (31). Consequently, Fcgr2b−/− mice produce larger amounts of Abs following antigenic stimulation, as observed for MAHA titers in our study. Finally, it has not escaped our attention that significant amounts of mIgG2b/c MAHA are elicited by injection with hIgG only in Fcgr2b−/− but not in wild-type mice. Unlike mIgG1, mIgG2b and mIgG2c, an allotype of mIgG2a in mice with the Igh-1b allele (32), are proinflammatory because they can activate complement and have higher affinity for activating IgG Fc receptors (33). Fixation of mIgG2a and mIgG2b (but not mIgG1) to the mouse GBM induces acute glomerular injury due to an FcγRIII- and FcγRIV-dependent influx of neutrophils (34). In contrast, even long-term persistence of mIgG1 in the GBM of mice injected with an anti–α3NC1 mAb does not induce any glomerular pathology (35). Autoimmune pathology in Fcgr2b−/− mice immunized with bovine collagen type IV is associated with enhanced production of mIgG2a/b autoantibodies (36), suggesting that a propensity toward production of proinflammatory mIgG subclasses may contribute to the susceptibility of Fcgr2b−/− mice to Ab-mediated glomerulonephritis.

In contrast to APTN alloantibodies, affinity-purified GP IgG autoantibodies did not induce crescentic glomerulonephritis into Fcgr2b−/− mice. Even very large amounts of affinity-purified GP autoantibodies (5 mg/mouse, 50 times more than in this study) injected in XenoMouse II only cause mild nephritis with patchy mesangial expansion and glomerular hIgG deposition in a punctate rather than a linear pattern (37). The inability of GP autoantibodies to bind in vivo to the mouse GBM in significant amounts explains the inconspicuous glomerular pathology after passive immunization. In this study, we showed that quinary NC1 crosslinks of α3α4α5(IV) collagen prevent the binding of GP autoantibodies to mouse GBM both in vitro and in vivo. Previous studies of α3α4α5NC1 hexamers from human GBM have shown that crosslinks yield α3–α5NC1 heterodimers and α4–α4NC1 homodimers (9), which reinforce the hexamers and prevent their dissociation by GP autoantibodies in vitro (14), as also demonstrated for α3α4α5NC1 hexamers isolated from bovine testes (12). In this study, GP autoantibodies likewise bound to murine α3–α5NC1 dimers and α3NC1 monomers, after the NC1 hexamers from the mouse GBM had been dissociated (Fig. 6A, lane i). Because NC1 crosslinks do not prevent the binding of GP autoantibodies to α3–α5NC1 dimer subunits, the impenetrability of native murine α3α4α5NC1 hexamers by GP autoantibodies must be the combined result of the crypticity of GP autoepitopes together with the structural stabilization of NC1 hexamers by quinary crosslinks. Collectively, these findings imply that the NC1 crosslinks “lock” the cryptic GP autoepitopes within the α3α4α5(IV) collagen suprastructure.

The chemical nature of the collagen IV bonds crosslinking NC1 dimers has been elusive. Long presumed to be crosslinked by disulfide bonds (38), NC1 dimers have been shown to contain intermolecular covalent bonds between conserved Met93 and Lys211 residues, initially assigned as thioether bonds (12, 39, 40). Most recently, this unusual posttranslational modification has been identified as a sulfilimine bond, uniquely found within collagen IV networks in tissues (22). Sulfilimine bonds have been proposed to occur in most metazoa, based on the evolutionary conservation of the Met93 and Lys211 in the NC1 domains of collagen IV from vertebrates and most invertebrate phyla, except in hydra, flatworms, sponges, and Placozoa (22). This implies that the cryptic GP autoepitopes are locked in the mouse GBM by sulfilimine bonds crosslinking the NC1 domains of adjoining α3α4α5(IV) collagen molecules. Thus, posttranslational modifications of autoantigens, often implicated in autoimmune etiology (4143), can also avert autoimmune disease, as showcased by NC1 crosslinks of α3α4α5(IV) collagen.

Unlike mice in this study or XenoMouse II (37), human patients or squirrel monkeys are susceptible to anti-GBM nephritis mediated by GP autoantibodies (20). However, biochemical analyses of NC1 domains from the GBM of mice (24), humans (9), and monkeys (44) have revealed a similar architecture, with α3α4α5(IV) collagen molecules forming supramolecular networks. We found that the reactivity toward GP autoantibodies is inversely correlated with the extent to which the NC1 hexamers of the GBM undergo posttranslational modifications forming quinary crosslinks. In the murine GBM, α3α4α5NC1 hexamers undergo comprehensive crosslinking, thereby locking all GP autoepitopes. In contrast, a small proportion of α3α4α5NC1 hexamers without NC1 crosslinks that react with GP autoantibodies under native conditions has been found to occur in the GBM of humans (14) and squirrel monkeys (this study). Why the crosslinking of GP autoantigen varies among species is unknown. As mouse kidneys contained a smaller proportion of NC1 monomers than did human or monkey kidneys (Figs. 1A, 6A), we speculate that a putative enzyme producing the NC1 crosslink may be more active and/or expressed at higher levels in rodent kidneys. It is perhaps not coincidental that NC1 hexamers from rat kidneys have the smallest proportion of NC1 monomers and the lowest reactivity with GP autoantibodies among six species of mammals (45). Because susceptibility to GP Ab-mediated nephritis is associated with GBM isoforms of α3(IV) collagen not stabilized by NC1 crosslinks, these isoforms are likely implicated in the pathogenesis of GP disease as targets for GP autoantibodies directed against cryptic epitopes.

Since the α3NC1 autoepitopes targeted by GP autoantibodies are inaccessible in the mouse GBM, what are the targets of nephritogenic mIgG Abs in mouse models of anti-GBM glomerulonephritis? In mice immunized with α3NC1 from bovine GBM (4648) or recombinant human α3NC1 (37, 49), which develop glomerulonephritis associated with serum anti-α3NC1 Abs and linear GBM deposition of mIgG, it is likely that some mIgG may be produced against non-cryptic mouse α3NC1 epitopes (Fig. 7F). Supporting this view, mIgG mAb 8D1, raised against human α3NC1 monomers, binds to native murine α3α4α5NC1 hexamers in vitro (23) and in vivo (35). More intriguing is the nature of autoantibodies mediating crescentic glomerulonephritis and hemorrhagic pneumonitis (GP syndrome) in Fcgr2b−/− mice immunized with Cellmatrix-IV (36). Cellmatrix-IV consists of collagen IV fragments solubilized from bovine lens capsule basement membranes by limited pepsin digestion, a treatment that destroys the NC1 domains (50). Sera from mice immunized with Cellmatrix-IV contain mIgG Abs binding to pepsin-extracted murine collagen IV from Engelbreth–Holm–Swarm tumors, suggesting that their epitopes reside in the collagenous domain of α1α2(IV) collagen (Fig. 7G), a rare target of human anti-GBM autoantibodies (51). Mouse GBM Ags other than α3α4α5(IV) collagen are also targeted by anti–GBM Abs from sheep nephrotoxic sera (Fig. 7H).

The new mouse model of glomerulonephritis described in this study is suitable for investigating the pathogenic role of alloantibodies in APTN and for testing new experimental therapies, such as using decoy Ag to prevent the binding of pathogenic Abs to target tissues (25). For these purposes, the widely used murine model of nephrotoxic serum nephritis induced by heterologous anti-mouse GBM Abs—although useful for studies of the effector mechanisms of Ab-mediated glomerulonephritis—is inadequate, since anti-GBM Abs from nephrotoxic sera do not reproduce the restricted Ag and epitope specificity of hIgG Abs mediating anti-GBM disease in patients. Whereas our model resembles murine nephrotoxic nephritis models with regard to the development of mIgG against the foreign globulin, which amplify glomerular injury in the autologous phase, much smaller amounts of GBM-specific patient Abs (0.1 mg of hIgG per mouse) were sufficient to induce severe experimental nephritis in susceptible mice, thus mitigating against the limited availability of human anti-GBM Abs. Although the nephritogenicity of autoantibodies from GP patients cannot currently be evaluated by passive transfer into mice, this may become possible when mice unable to crosslink α3α4α5(IV) collagen are genetically engineered. Until then, primate models remain available for this purpose (20).

We thank Dr. Larry Williams and Dr. Susan Gibson for providing squirrel monkey kidneys, Dr. A. Richard Kitching for the generous gift of sheep anti-mouse GBM antiserum, Dr. Corina M. Borza for critical reading of the manuscript, and Linna Ge for technical assistance.

Disclosures The authors have no financial conflicts of interest.

This work was supported in part by National Institutes of Health Grants R21-AI069874 and R01-DK080799 (to D.B.B.). The Squirrel Monkey Breeding and Research Resource was funded by National Institutes of Health Grant P40-RR001254 (to Dr. Christian R. Abee).

The online version of this article contains supplemental material.

2

The quinary structure was introduced as an extension of the four canonical levels of proteins organization (primary through quaternary) to designate interactions among molecules within helical arrays (21). By extending this concept to basement membrane components that assemble into higher order structures, such as collagen IV and laminin, we distinguish between the obligate intracellular assembly of heterotrimeric molecules—a quintessential quaternary structure—and supramolecular assembly into sheetlike networks in the extracellular space, which can be regarded as the quinary level of organization. Thus, the crypticity of the GP autoepitopes is the result of quaternary assembly of α3α4α5(IV) collagen heterotrimers, whereas the NC1 crosslinks yielding dimer subunits are a feature of the quinary structure.

Abbreviations used in this paper:

APTN

Alport posttransplant nephritis

C.I.C.

circulating immune complexes

D

dimer

G

glomeruli

GBM

glomerular basement membrane

GN

glomerulonephritis

GP

Goodpasture

hIgG

human IgG

M

monomer

MAHA

mouse anti-human IgG Abs

mIgG

mouse IgG

NC1

noncollagenous domain 1

NHS

normal human serum

NTS

nephrotoxic serum.

1
Matsumoto
I.
,
Maccioni
M.
,
Lee
D. M.
,
Maurice
M.
,
Simmons
B.
,
Brenner
M.
,
Mathis
D.
,
Benoist
C.
.
2002
.
How antibodies to a ubiquitous cytoplasmic enzyme may provoke joint-specific autoimmune disease.
Nat. Immunol.
3
:
360
365
.
2
Wipke
B. T.
,
Wang
Z.
,
Kim
J.
,
McCarthy
T. J.
,
Allen
P. M.
.
2002
.
Dynamic visualization of a joint-specific autoimmune response through positron emission tomography.
Nat. Immunol.
3
:
366
372
.
3
Nimmerjahn
F.
,
Ravetch
J. V.
.
2008
.
Fcγ receptors as regulators of immune responses.
Nat. Rev. Immunol.
8
:
34
47
.
4
Borza
D. B.
,
Netzer
K. O.
,
Leinonen
A.
,
Todd
P.
,
Cervera
J.
,
Saus
J.
,
Hudson
B. G.
.
2000
.
The goodpasture autoantigen: identification of multiple cryptic epitopes on the NC1 domain of the α3(IV) collagen chain.
J. Biol. Chem.
275
:
6030
6037
.
5
Netzer
K. O.
,
Leinonen
A.
,
Boutaud
A.
,
Borza
D. B.
,
Todd
P.
,
Gunwar
S.
,
Langeveld
J. P.
,
Hudson
B. G.
.
1999
.
The Goodpasture autoantigen: mapping the major conformational epitope(s) of α3(IV) collagen to residues 17–31 and 127–141 of the NC1 domain.
J. Biol. Chem.
274
:
11267
11274
.
6
Hellmark
T.
,
Segelmark
M.
,
Unger
C.
,
Burkhardt
H.
,
Saus
J.
,
Wieslander
J.
.
1999
.
Identification of a clinically relevant immunodominant region of collagen IV in Goodpasture disease.
Kidney Int.
55
:
936
944
.
7
Yoshioka
K.
,
Michael
A. F.
,
Velosa
J.
,
Fish
A. J.
.
1985
.
Detection of hidden nephritogenic antigen determinants in human renal and nonrenal basement membranes.
Am. J. Pathol.
121
:
156
165
.
8
Wieslander
J.
,
Langeveld
J.
,
Butkowski
R.
,
Jodlowski
M.
,
Noelken
M.
,
Hudson
B. G.
.
1985
.
Physical and immunochemical studies of the globular domain of type IV collagen: cryptic properties of the Goodpasture antigen.
J. Biol. Chem.
260
:
8564
8570
.
9
Borza
D. B.
,
Bondar
O.
,
Todd
P.
,
Sundaramoorthy
M.
,
Sado
Y.
,
Ninomiya
Y.
,
Hudson
B. G.
.
2002
.
Quaternary organization of the Goodpasture autoantigen, the α3(IV) collagen chain. Sequestration of two cryptic autoepitopes by intrapromoter interactions with the α4 and α5 NC1 domains.
J. Biol. Chem.
277
:
40075
40083
.
10
Boutaud
A.
,
Borza
D. B.
,
Bondar
O.
,
Gunwar
S.
,
Netzer
K. O.
,
Singh
N.
,
Ninomiya
Y.
,
Sado
Y.
,
Noelken
M. E.
,
Hudson
B. G.
.
2000
.
Type IV collagen of the glomerular basement membrane: evidence that the chain specificity of network assembly is encoded by the noncollagenous NC1 domains.
J. Biol. Chem.
275
:
30716
30724
.
11
Kang
J. S.
,
Colon
S.
,
Hellmark
T.
,
Sado
Y.
,
Hudson
B. G.
,
Borza
D. B.
.
2008
.
Identification of noncollagenous sites encoding specific interactions and quaternary assembly of α3α4α5(IV) collagen: implications for Alport gene therapy.
J. Biol. Chem.
283
:
35070
35077
.
12
Vanacore
R. M.
,
Ham
A. J.
,
Cartailler
J. P.
,
Sundaramoorthy
M.
,
Todd
P.
,
Pedchenko
V.
,
Sado
Y.
,
Borza
D. B.
,
Hudson
B. G.
.
2008
.
A role for collagen IV cross-links in conferring immune privilege to the Goodpasture autoantigen: structural basis for the crypticity of B cell epitopes.
J. Biol. Chem.
283
:
22737
22748
.
13
David
M.
,
Borza
D. B.
,
Leinonen
A.
,
Belmont
J. M.
,
Hudson
B. G.
.
2001
.
Hydrophobic amino acid residues are critical for the immunodominant epitope of the Goodpasture autoantigen: a molecular basis for the cryptic nature of the epitope.
J. Biol. Chem.
276
:
6370
6377
.
14
Borza
D. B.
,
Bondar
O.
,
Colon
S.
,
Todd
P.
,
Sado
Y.
,
Neilson
E. G.
,
Hudson
B. G.
.
2005
.
Goodpasture autoantibodies unmask cryptic epitopes by selectively dissociating autoantigen complexes lacking structural reinforcement: novel mechanisms for immune privilege and autoimmune pathogenesis.
J. Biol. Chem.
280
:
27147
27154
.
15
Brainwood
D.
,
Kashtan
C.
,
Gubler
M. C.
,
Turner
A. N.
.
1998
.
Targets of alloantibodies in Alport anti-glomerular basement membrane disease after renal transplantation.
Kidney Int.
53
:
762
766
.
16
Kalluri
R.
,
Torre
A.
,
Shield
C. F.
 III
,
Zamborsky
E. D.
,
Werner
M. C.
,
Suchin
E.
,
Wolf
G.
,
Helmchen
U. M.
,
van den Heuvel
L. P.
,
Grossman
R.
, et al
.
2000
.
Identification of α3, α4, and α5 chains of type IV collagen as alloantigens for Alport posttransplant anti-glomerular basement membrane antibodies.
Transplantation
69
:
679
683
.
17
Kang
J. S.
,
Kashtan
C. E.
,
Turner
A. N.
,
Heidet
L.
,
Hudson
B. G.
,
Borza
D. B.
.
2007
.
The alloantigenic sites of α3α4α5(IV) collagen: pathogenic X-linked alport alloantibodies target two accessible conformational epitopes in the α5NC1 domain.
J. Biol. Chem.
282
:
10670
10677
.
18
Wang
X. P.
,
Fogo
A. B.
,
Colon
S.
,
Giannico
G.
,
Abul-Ezz
S. R.
,
Miner
J. H.
,
Borza
D. B.
.
2005
.
Distinct epitopes for anti-glomerular basement membrane Alport alloantibodies and Goodpasture autoantibodies within the noncollagenous domain of α3(IV) collagen: a Janus-faced antigen.
J. Am. Soc. Nephrol.
16
:
3563
3571
.
19
Kashtan
C. E.
2006
.
Renal transplantation in patients with Alport syndrome.
Pediatr. Transplant.
10
:
651
657
.
20
Lerner
R. A.
,
Glassock
R. J.
,
Dixon
F. J.
.
1967
.
The role of anti-glomerular basement membrane antibody in the pathogenesis of human glomerulonephritis.
J. Exp. Med.
126
:
989
1004
.
21
Edelstein
S. J.
1980
.
Patterns in the quinary structures of proteins: plasticity and inequivalence of individual molecules in helical arrays of sickle cell hemoglobin and tubulin.
Biophys. J.
32
:
347
360
.
22
Vanacore
R.
,
Ham
A. J.
,
Voehler
M.
,
Sanders
C. R.
,
Conrads
T. P.
,
Veenstra
T. D.
,
Sharpless
K. B.
,
Dawson
P. E.
,
Hudson
B. G.
.
2009
.
A sulfilimine bond identified in collagen IV.
Science
325
:
1230
1234
.
23
Kang
J. S.
,
Wang
X. P.
,
Miner
J. H.
,
Morello
R.
,
Sado
Y.
,
Abrahamson
D. R.
,
Borza
D. B.
.
2006
.
Loss of α3/α4(IV) collagen from the glomerular basement membrane induces a strain-dependent isoform switch to α5α6(IV) collagen associated with longer renal survival in Col4a3−/− Alport mice.
J. Am. Soc. Nephrol.
17
:
1962
1969
.
24
Heidet
L.
,
Borza
D. B.
,
Jouin
M.
,
Sich
M.
,
Mattei
M. G.
,
Sado
Y.
,
Hudson
B. G.
,
Hastie
N.
,
Antignac
C.
,
Gubler
M. C.
.
2003
.
A human-mouse chimera of the α3α4α5(IV) collagen protomer rescues the renal phenotype in Col4a3−/− Alport mice.
Am. J. Pathol.
163
:
1633
1644
.
25
Nishie
W.
,
Sawamura
D.
,
Goto
M.
,
Ito
K.
,
Shibaki
A.
,
McMillan
J. R.
,
Sakai
K.
,
Nakamura
H.
,
Olasz
E.
,
Yancey
K. B.
, et al
.
2007
.
Humanization of autoantigen.
Nat. Med.
13
:
378
383
.
26
Takai
T.
,
Ono
M.
,
Hikida
M.
,
Ohmori
H.
,
Ravetch
J. V.
.
1996
.
Augmented humoral and anaphylactic responses in FcγRII-deficient mice.
Nature
379
:
346
349
.
27
Petkova
S. B.
,
Konstantinov
K. N.
,
Sproule
T. J.
,
Lyons
B. L.
,
Awwami
M. A.
,
Roopenian
D. C.
.
2006
.
Human antibodies induce arthritis in mice deficient in the low-affinity inhibitory IgG receptor FcγRIIB.
J. Exp. Med.
203
:
275
280
.
28
Hudson
B. G.
,
Tryggvason
K.
,
Sundaramoorthy
M.
,
Neilson
E. G.
.
2003
.
Alport’s syndrome, Goodpasture’s syndrome, and type IV collagen.
N. Engl. J. Med.
348
:
2543
2556
.
29
Borza
D. B.
,
Neilson
E. G.
,
Hudson
B. G.
.
2003
.
Pathogenesis of Goodpasture syndrome: a molecular perspective.
Semin. Nephrol.
23
:
522
531
.
30
Borza
D. B.
2007
.
Autoepitopes and alloepitopes of type IV collagen: role in the molecular pathogenesis of anti-GBM antibody glomerulonephritis.
Nephron, Exp. Nephrol.
106
:
e37
e43
.
31
Xiang
Z.
,
Cutler
A. J.
,
Brownlie
R. J.
,
Fairfax
K.
,
Lawlor
K. E.
,
Severinson
E.
,
Walker
E. U.
,
Manz
R. A.
,
Tarlinton
D. M.
,
Smith
K. G.
.
2007
.
FcγRIIb controls bone marrow plasma cell persistence and apoptosis.
Nat. Immunol.
8
:
419
429
.
32
Morgado
M. G.
,
Cam
P.
,
Gris-Liebe
C.
,
Cazenave
P. A.
,
Jouvin-Marche
E.
.
1989
.
Further evidence that BALB/c and C57BL/6 γ2a genes originate from two distinct isotypes.
EMBO J.
8
:
3245
3251
.
33
Nimmerjahn
F.
,
Ravetch
J. V.
.
2005
.
Divergent immunoglobulin G subclass activity through selective Fc receptor binding.
Science
310
:
1510
1512
.
34
Giorgini
A.
,
Brown
H. J.
,
Lock
H. R.
,
Nimmerjahn
F.
,
Ravetch
J. V.
,
Verbeek
J. S.
,
Sacks
S. H.
,
Robson
M. G.
.
2008
.
FcγRIII and FcγRIV are indispensable for acute glomerular inflammation induced by switch variant monoclonal antibodies.
J. Immunol.
181
:
8745
8752
.
35
Summers
S. A.
,
Steinmetz
O. M.
,
Li
M.
,
Kausman
J. Y.
,
Semple
T.
,
Edgtton
K. L.
,
Borza
D. B.
,
Braley
H.
,
Holdsworth
S. R.
,
Kitching
A. R.
.
2009
.
Th1 and Th17 cells induce proliferative glomerulonephritis.
J. Am. Soc. Nephrol.
20
:
2518
2524
.
36
Nakamura
A.
,
Yuasa
T.
,
Ujike
A.
,
Ono
M.
,
Nukiwa
T.
,
Ravetch
J. V.
,
Takai
T.
.
2000
.
Fcγ receptor IIB-deficient mice develop Goodpasture’s syndrome upon immunization with type IV collagen: a novel murine model for autoimmune glomerular basement membrane disease.
J. Exp. Med.
191
:
899
906
.
37
Meyers
K. E.
,
Allen
J.
,
Gehret
J.
,
Jacobovits
A.
,
Gallo
M.
,
Neilson
E. G.
,
Hopfer
H.
,
Kalluri
R.
,
Madaio
M. P.
.
2002
.
Human antiglomerular basement membrane autoantibody disease in XenoMouse II.
Kidney Int.
61
:
1666
1673
.
38
Siebold
B.
,
Deutzmann
R.
,
Kühn
K.
.
1988
.
The arrangement of intra- and intermolecular disulfide bonds in the carboxyterminal, non-collagenous aggregation and cross-linking domain of basement-membrane type IV collagen.
Eur. J. Biochem.
176
:
617
624
.
39
Than
M. E.
,
Henrich
S.
,
Huber
R.
,
Ries
A.
,
Mann
K.
,
Kühn
K.
,
Timpl
R.
,
Bourenkov
G. P.
,
Bartunik
H. D.
,
Bode
W.
.
2002
.
The 1.9-Å crystal structure of the noncollagenous (NC1) domain of human placenta collagen IV shows stabilization via a novel type of covalent Met-Lys cross-link.
Proc. Natl. Acad. Sci. USA
99
:
6607
6612
.
40
Vanacore
R. M.
,
Friedman
D. B.
,
Ham
A. J.
,
Sundaramoorthy
M.
,
Hudson
B. G.
.
2005
.
Identification of S-hydroxylysyl-methionine as the covalent cross-link of the noncollagenous (NC1) hexamer of the α1α1α2 collagen IV network: a role for the post-translational modification of lysine 211 to hydroxylysine 211 in hexamer assembly.
J. Biol. Chem.
280
:
29300
29310
.
41
Mamula
M. J.
,
Gee
R. J.
,
Elliott
J. I.
,
Sette
A.
,
Southwood
S.
,
Jones
P. J.
,
Blier
P. R.
.
1999
.
Isoaspartyl post-translational modification triggers autoimmune responses to self-proteins.
J. Biol. Chem.
274
:
22321
22327
.
42
Doyle
H. A.
,
Mamula
M. J.
.
2001
.
Post-translational protein modifications in antigen recognition and autoimmunity.
Trends Immunol.
22
:
443
449
.
43
Anderton
S. M.
2004
.
Post-translational modifications of self antigens: implications for autoimmunity.
Curr. Opin. Immunol.
16
:
753
758
.
44
Borza
C. M.
,
Borza
D. B.
,
Pedchenko
V.
,
Saleem
M. A.
,
Mathieson
P. W.
,
Sado
Y.
,
Hudson
H. M.
,
Pozzi
A.
,
Saus
J.
,
Abrahamson
D. R.
, et al
.
2008
.
Human podocytes adhere to the KRGDS motif of the α3α4α5 collagen IV network.
J. Am. Soc. Nephrol.
19
:
677
684
.
45
Ryan
J. J.
,
Katbamna
I.
,
Mason
P. J.
,
Pusey
C. D.
,
Turner
A. N.
.
1998
.
Sequence analysis of the “Goodpasture antigen” of mammals.
Nephrol. Dial. Transplant.
13
:
602
607
.
46
Kalluri
R.
,
Danoff
T. M.
,
Okada
H.
,
Neilson
E. G.
.
1997
.
Susceptibility to anti-glomerular basement membrane disease and Goodpasture syndrome is linked to MHC class II genes and the emergence of T cell-mediated immunity in mice.
J. Clin. Invest.
100
:
2263
2275
.
47
Kitching
A. R.
,
Turner
A. L.
,
Semple
T.
,
Li
M.
,
Edgtton
K. L.
,
Wilson
G. R.
,
Timoshanko
J. R.
,
Hudson
B. G.
,
Holdsworth
S. R.
.
2004
.
Experimental autoimmune anti-glomerular basement membrane glomerulonephritis: a protective role for IFN-γ.
J. Am. Soc. Nephrol.
15
:
1764
1774
.
48
Dean
E. G.
,
Wilson
G. R.
,
Li
M.
,
Edgtton
K. L.
,
O’Sullivan
K. M.
,
Hudson
B. G.
,
Holdsworth
S. R.
,
Kitching
A. R.
.
2005
.
Experimental autoimmune Goodpasture’s disease: a pathogenetic role for both effector cells and antibody in injury.
Kidney Int.
67
:
566
575
.
49
Hopfer
H.
,
Maron
R.
,
Butzmann
U.
,
Helmchen
U.
,
Weiner
H. L.
,
Kalluri
R.
.
2003
.
The importance of cell-mediated immunity in the course and severity of autoimmune anti-glomerular basement membrane disease in mice.
FASEB J.
17
:
860
868
.
50
Weber
S.
,
Engel
J.
,
Wiedemann
H.
,
Glanville
R. W.
,
Timpl
R.
.
1984
.
Subunit structure and assembly of the globular domain of basement-membrane collagen type IV.
Eur. J. Biochem.
139
:
401
410
.
51
Borza
D. B.
,
Chedid
M. F.
,
Colon
S.
,
Lager
D. J.
,
Leung
N.
,
Fervenza
F. C.
.
2005
.
Recurrent Goodpasture’s disease secondary to a monoclonal IgA1-κ antibody autoreactive with the α1/α2 chains of type IV collagen.
Am. J. Kidney Dis.
45
:
397
406
.