Anti-hinge Abs (AHAs) target neoepitopes exposed after proteolytic cleavage of IgG. In this study, we explored the diversity of protease- and IgG subclass–restricted AHAs and their potential as immunological markers in healthy donors (HDs) and patients with rheumatoid arthritis (RA) or systemic lupus erythematosus (SLE). AHA reactivity against IgG-degrading enzyme of Streptococcus pyogenes (IdeS)– or pepsin-generated F(ab′)2 fragments of all four human IgG subclasses was determined. AHA reactivity against one or more out of eight F(ab′)2 targets was found in 68% (68 of 100) of HDs, 69% (68 of 99) of SLE patients, and 81% (79 of 97) of RA patients. Specific recognition of hinge epitopes was dependent on IgG subclass and protease used to create the F(ab′)2 targets, as confirmed by inhibition experiments with F(ab′)2 fragments and hinge peptides. Reactivity against IdeS-generated F(ab′)2 targets was found most frequently, whereas reactivity against pepsin-generated F(ab′)2 targets better discriminated between RA and HDs or SLE, with significantly higher AHA levels against IgG1/3/4. In contrast, AHA levels against pepsin-cleaved IgG2 were comparable. No reactivity against IdeS-generated IgG2-F(ab′)2s was detected. The most discriminatory AHA reactivity in RA was against pepsin-cleaved IgG4, with a 35% prevalence, ≥5.8-fold higher than in HDs/SLE, and significantly higher levels (p < 0.0001). Cross-reactivity for F(ab′)2s generated from different IgG subclasses was only observed for subclasses having homologous F(ab′)2 C termini (IgG1/3/4). For IgG2, two pepsin cleavage sites were identified; anti-hinge reactivity was restricted to only one of these. In conclusion, AHAs specifically recognize IgG subclass– and protease-restricted hinge neoepitopes. Their protease-restricted specificity suggests that different AHA responses developed under distinct inflammatory or infectious conditions and may be markers of, and participants in, such processes.
Rheumatoid arthritis (RA) is an autoimmune disease characterized by the presence of different autoantibodies. Autoantibodies targeting the constant (Fc) domain of IgG, the first to be described (1), are known as rheumatoid factors. More recently, anti-citrullinated protein Abs (ACPAs) (2), as well as anti-carbamylated protein Abs (3, 4), were found to be associated with RA. These autoantibodies target neoepitopes induced in the joint during inflammation by posttranslational modification of proteins. In the inflamed RA joint, increased activity of proteases contributes to tissue damage by degrading matrix proteins (5–8). Proteases can also cleave IgG molecules at specific sites in the hinge region, generating Fab or F(ab′)2 fragments and exposing neoepitopes composed of C-terminal amino acid residues (9–11). Autoantibodies that recognize these neoepitopes, but not intact IgG, known as anti-hinge Abs (AHAs) (12), are found both in healthy individuals and in RA, with higher levels in the latter (13, 14).
It is unclear what the biological roles of AHAs are and whether they are beneficial or harmful. AHAs may be involved in very diverse mechanisms, ranging from pathological immune complex formation to physiological clearance of degraded Abs and restoring effector function to IgGs cleaved by protease-producing microbes (12).
Recently, our group reported the presence of AHAs specifically targeting the pepsin-cleaved hinge of IgG4 in a subset of RA patients (15). The frequency of these subclass-specific AHAs as well as of total AHAs was low compared with previous studies, probably because the RIA used only detects high-affinity Abs (16). Nonetheless, the data suggested that AHA reactivity can be very specific and can be associated with disease. It is unclear how specific AHAs are for protease- and IgG subclass–restricted neoepitopes and whether determining their specificity can be used to discriminate between AHAs from healthy individuals and AHAs associated with specific (autoimmune) diseases such as RA or systemic lupus erythematosus (SLE).
The objective of the present study was to investigate the presence of AHA reactivity and specificity of AHAs for protease- and IgG subclass–restricted neoepitopes in healthy donors (HDs) and patients with RA or SLE, both autoantibody-associated autoimmune diseases, using a newly developed ELISA system with greater sensitivity than the RIA.
Our ELISA setup measures reactivity against the exposed hinge of F(ab′)2 fragments generated from all four IgG subclasses with the proteases pepsin and IgG-degrading enzyme of Streptococcus pyogenes (IdeS). Pepsin was used in most earlier investigations and cleaves IgG1 at the same location in the hinge as does matrix metalloproteinase (MMP)-7 (9, 10). MMPs play an important role in physiological and pathophysiological tissue remodeling, and their increased levels in RA may contribute to joint damage (6, 8). IdeS is a bacterial protease that cleaves the IgG hinge 2 aa closer to the C terminus (11, 17) and was chosen to be able to compare reactivity against targets generated in an inflammation setting with those generated in an infectious, bacterial setting. Whereas the site where IdeS cleaves the different IgG subclasses is well defined (11, 18, 19), the pepsin cleavage site can be variable in IgG1 (20) and has not been well defined in IgG2 and IgG4. Using native mass spectrometry we now define these sites more clearly.
Our results suggest that anti-hinge reactivity against pepsin- and IdeS-generated F(ab′)2s are multiple independent AHA responses and that AHAs specifically bind to protease- and IgG subclass–restricted neoepitopes. AHA responses containing reactivity against pepsin-cleaved IgG4 were found much more frequently in RA compared with HDs, and SLE and AHA responses specifically targeting pepsin-cleaved IgG4 were found almost exclusively in RA.
Materials and Methods
Three sets of serum samples were used. The first set consisted of serum samples from 100 HDs who were frequently boosted with tetanus toxoid. The second set consisted of baseline serum samples from 97 patients with active early RA included in the “COBRA-light” trial (21) and treated at various hospitals in the Amsterdam area. RA was diagnosed according to the 1987 criteria of the American College for Rheumatism (22). These patients had not been previously treated with disease-modifying antirheumatic drugs. The third set consisted of baseline serum samples from 99 unselected patients included in the Amsterdam SLE cohort (23). Patients included in this open cohort fulfill the 1997 American College for Rheumatism classification criteria for SLE (24).
All RA and SLE patients gave informed consent for use of serum samples and clinical data. No informed consent was obtained for the samples from the healthy controls, because materials used for this study were leftovers from samples taken for routine diagnostic purposes. HD materials were used anonymously without any connection to clinical or person-specific data.
Two sources of Abs were used: recombinant chimeric Abs of the four different IgG subclasses (IgG1, IgG2, IgG3, IgG4), all specific for biotin, and the therapeutic Ab formulations adalimumab (IgG1) (Humira; AbbVie), panitumumab (IgG2) (Vectibix; Amgen), and natalizumab (IgG4) (Tysabri; Biogen Idec International). The anti-biotin Abs were produced as described before (25) by cloning synthetic constructs coding for the variable domains (26, 27) and IgG1/IgG2/IgG3/IgG4 and κ constant domains (GeneArt; Invitrogen) into a pcDNA3.1 expression vector (Invitrogen). IgGs were purified by protein G affinity chromatography: Cell culture supernatants were filtered over a syringe filter, pore size of 0.20 μm (Whatman Puradisc 30; Sigma-Aldrich), followed by loading on a protein G column (HiTrap Protein G HP; GE Healthcare) and elution of the IgG with 0.1 M citric acid–NaOH (pH 3). The eluate was immediately neutralized with 2 M Tris-HCl (pH 9) and dialyzed overnight to 12.6 mM sodium phosphate, 140 mM NaCl (pH 7.4) (B. Braun, Oss, the Netherlands). After dialysis, samples were sterile filtered over a 0.20-μm syringe filter. Concentration of the purified IgG was determined by measuring absorbance at 280 nm.
Proteolysis to create F(ab′)2 fragments
To create F(ab′)2 targets and inhibitors for the anti-hinge ELISAs, IgGs were subjected to proteolysis by pepsin or IgG-degrading enzyme of Streptococcus pyogenes (IdeS). Optimal pepsin cleaving times were determined to be different for the IgG subclasses. Proteolysis by pepsin was performed by incubating Abs overnight (IgG1, IgG4), for 2 h (IgG2), or for 1 h (IgG3) at 37°C, 1:100 (w/w), at pH 3.5 (IgG1–3) or pH 4 (IgG4). The reaction was stopped by adding 1 M Tris until pH was 7.5. Proteolysis by IdeS was performed by incubating Abs at a concentration of 1 mg/ml for 2 h at 37°C in 5 ml of PBS (pH 6.6) with IdeS (FabRICATOR; Genovis). Undigested IgG was removed with a HiTrap protein G column. Purified F(ab′)2 fragments were dialyzed against PBS and analyzed by gel electrophoresis (SDS-PAGE).
Analysis of F(ab′)2 fragments by native mass spectrometry
F(ab′)2s were analyzed by native mass spectrometry at AbLab (Utrecht, the Netherlands). Fifty micrograms of each F(ab′)2 sample was exchanged into 150 mM ammonium acetate buffer (pH 7.5), using 10-kDa molecular mass cut-off spin-filter columns (Vivaspin 500 centrifugal concentrators; Sartorius). Two microliters of a 5 μM sample was sprayed on an electrospray ionization–time-of-flight mass spectrometer (LCT Mass Spectrometer; Waters, Manchester, U.K.) using gold-coated borosilicate capillaries made in-house (using a Sutter P-97 puller [Sutter Instruments, Novato, CA] and an Edwards Scancoat Six sputter-coater [Edwards Laboratories, Milpitas, CA]). Mass calibration was performed using 25 mg/ml cesium iodide. MassLynx V4.1 (Waters) was used for experimental mass determination.
Synthetic peptides for inhibition assays
Synthetic 6- or 7-mer peptide analogs of the IgG hinge were designed based on the potential C-terminal amino acid epitopes exposed by the F(ab′)2s and produced by GenScript (Table I). For use in inhibition assays the peptide analogs were dissolved in assay buffer (PBS/0.1% Tween 20), and pH was checked and corrected to pH 7–8 by titrating in a 4 M NaOH solution. Peptides were mixed with the sera before the sera were added to the ELISA wells.
|EU Numbering .||226 .||227 .||228 .||229 .||230 .||231 .||232 .||233 .||234 .||235 .||236 .||237 .||238 .||239 .||240 .||241 .||242 .||243 .|
|EU Numbering .||226 .||227 .||228 .||229 .||230 .||231 .||232 .||233 .||234 .||235 .||236 .||237 .||238 .||239 .||240 .||241 .||242 .||243 .|
Overview of the synthetic peptide analogs of the IgG1, IgG2, and IgG4 hinge used in inhibition experiments described in the present study (in bold). Amino acids sequences of IgG1, IgG2, and IgG4 starting at the core hinge (CPPC/CPSC) are displayed for reference. EU numbering of amino acid positions follows that used in (37).
–, absence of an amino acid at that position.
Nunc MaxiSorp flat-bottom 96-well plates (Thermo Scientific) were used for all ELISAs. Plates were coated overnight at 4°C with 100 μl/well IgG-depleted biotinylated human serum albumin (HSA-biotin) diluted to 10 μg/ml in PBS, then washed five times with wash buffer (PBS/0.02% Tween 20). Anti-biotin F(ab′)2s were diluted to 0.25 μg/ml in PBS/0.1% Tween 20 and HSA-biotin–coated plates were incubated with 100 μl/well diluted anti-biotin F(ab′)2 for 2 h at room temperature. Serum samples were diluted 1:400 in PBS/0.1% Tween 20 and plates were incubated with 100 μl/well diluted samples for 60 min at room temperature. Plates were washed five times and incubated with 100 μl/well monoclonal mouse anti-human IgG Abs labeled with HRP (MH16-1 HRP; Sanquin Reagents, Amsterdam, the Netherlands) diluted to 1 μg/ml in PBS/0.1% Tween-20. After 30 min incubation at room temperature, plates were washed five times and developed with 100 μl/well tetramethylbenzidine substrate solution (Interchim, Montluçon Cedex, France) diluted 1:1 in distilled water. The reaction was stopped with 100 μl/well 2 M H2SO4. Absorbance was measured at 450 and 540 nm using a BioTek microtiter plate reader, and AHA concentrations were calculated relative to a titration curve of a reference serum tested for reactivity against pepsin-generated F(ab′)2s from IgG1 on every plate and expressed in arbitrary units (AU) per milliliter. The reference serum used was from an HD that showed the highest AHA titers in the previously published RIA (15) and was defined to contain 200 AU/ml of AHAs against pepsin-generated IgG1 F(ab′)2s. The limit for positivity was determined in an inhibition experiment using 50 μg/ml IVIG-F(ab′)2 (Fig. 1C). For most subjects tested, a serum dilution of 1:400 gave AHA levels within the dynamic range of the assays, and samples were therefore tested first at a 1:400 dilution and subsequently at additional dilutions when necessary. Most sera were negative for at least one target. However, a subset of sera showing reactivity against all eight F(ab′)2 targets (19 of 296) was tested for reactivity against HSA-biotin, and in all cases these sera were found to generate a (slightly) positive signal on the HSA-biotin coat. Their AHA levels were corrected for the background anti–HSA-biotin levels.
Surface plasmon resonance imaging
For the surface plasmon resonance imaging (SPRi) experiments the IBIS MX96 imager (IBIS Technologies, Enschede, the Netherlands) was used. F(ab′)2s were spotted in duplicate at a spotting concentration of 100 nM in 10 mM MES buffer with 0.075% Tween 80 (pH 6) onto preactivated Easy2Spot G-type sensors (Ssens) using a continuous flow microspotter (Wasatch Microfluidics, Salt Lake City). Serum samples were diluted in PBS with 0.075% Tween 80 and 10 mM EDTA. Samples were flowed over the sensor for 10 min in the association phase, followed by a 5-min dissociation phase and regeneration of the sensor with two 6-s pulses of 100 mM H3PO4.
Differences between groups were analyzed with a Kruskal–Wallis ANOVA and a Dunn posttest for multiple comparisons or a z-score test for two population proportions, and correlations were analyzed with a Spearman test using GraphPad Prism software version 6. A p value <0.05 was considered significant.
Development of AHA ELISAs
Fig. 1A schematically shows the set-up of the anti-hinge ELISAs developed in the present study. ELISA plates were coated with biotinylated HSA (HSA-biotin) and subsequently opsonized with anti-biotin F(ab′)2s serving as targets for AHA binding. The F(ab′)2 targets were generated from IgG1, IgG2, IgG3, and IgG4 anti-biotin Abs with the proteases pepsin and IdeS. The specific binding of the anti-biotin F(ab′)2s to the coated HSA-biotin should provide optimal positioning of the F(ab′)2 targets for binding by AHAs and prevent interference of any Fc fragments remaining after proteolysis. Fig. 1B depicts the amino acid sequences of the hinge regions and upper CH2 regions of the four IgG subclasses and the sites where pepsin and IdeS cleave the IgG molecules to create F(ab′)2 fragments. IdeS specifically cleaves the IgG subclasses proximal to amino acid G237 (EU numbering) (18). Putative sites where pepsin cleaves the anti-biotin IgG1 and IgG4 Abs were confirmed by analyzing the generated F(ab′)2 targets with native mass spectrometry. IgG1-F(ab′)2 was shown to terminate at L234 and IgG4-F(ab′)2 at F234 (Supplemental Fig. 1). For IgG2, two dominant cleavage products were found: F(ab′)2 with a symmetrically cleaved hinge terminating in F241, and F(ab′)2 where the hinge was asymmetrically cleaved between V240 and F241 on one side and between F241 and L242 on the other side.
Prevalence and levels of AHAs targeting IgG-subclass hinge epitopes
The AHA ELISAs described above were used to compare presence and levels of anti-hinge reactivity between cohorts of HDs, early RA patients with active disease, and established SLE patients. In all three groups the AHA ELISAs were able to quantify levels of AHAs binding to pepsin-generated F(ab′)2s [pepsin-F(ab′)2s] from IgG1–4- and IdeS-generated F(ab′)2s [IdeS-F(ab′)2s] from IgG1, IgG3, and IgG4 (Fig. 2). No AHAs binding to IdeS-IgG2-F(ab′)2s were detected. Fig. 2A shows that AHAs targeting pepsin-IgG1-, IgG3-, and IgG4-F(ab′)2s were found more frequently in RA compared with HDs, and RA patients more frequently showed reactivity against IgG1- and IgG4-F(ab′)2s than did SLE patients. The most discriminatory AHA reactivity in RA, setting it apart from HDs and SLE, was against pepsin-cleaved IgG4, with a prevalence of 35% in RA, which was at least 5.8-fold higher compared with HDs and SLE, and with significantly higher levels (p < 0.0001). Levels of AHAs against pepsin-cleaved IgG1 and IgG3 were also significantly higher in RA compared HDs, and against IgG1 compared with SLE. In contrast, AHAs targeting pepsin-IgG2-F(ab′)2s were found at comparable (not significantly different) frequencies in all three groups.
Compared to AHA reactivity against pepsin-F(ab′)2s, reactivity against IdeS-F(ab′)2s was more common in all groups, with the highest frequencies in RA (Fig. 2B). Levels of AHA reactivity against IgG1- and IgG3-F(ab′)2s was significantly higher in RA patients compared with SLE, but only for IgG3-F(ab′)2s compared with HDs. No reactivity was found against IdeS-IgG2-F(ab′)2s in all three groups. Collectively, reactivity against at least one of eight F(ab′)2 targets was found in 68% of HDs, 69% of established SLE patients, and 81% of early RA patients.
Specificity of AHA responses for protease-restricted hinge neoepitopes
To determine whether binding of AHAs is dependent on the specific site where proteases cleave the IgG hinge, we analyzed co-occurrence of AHA reactivity against F(ab′)2 targets generated from the same IgG subclass with different proteases (pepsin or IdeS). The x,y plots in Fig. 3 show that in all three cohorts single positivity for AHA reactivity against pepsin-F(ab′)2s or IdeS-F(ab′)2s generated from the same IgG subclass was found more frequently than double positivity. Whereas AHA reactivity against both pepsin-F(ab′)2s and IdeS-F(ab′)2s was found frequently in individual sera, for example for IgG1 and IgG3 targets in RA (Fig. 3D, 3E), the AHA levels appeared not to be correlated. These data suggest that AHAs specifically recognize protease-restricted neoepitopes and that anti-hinge reactivity against pepsin-F(ab′)2s and anti-hinge reactivity against IdeS-F(ab′)2s are two separate responses.
To further investigate the specificity of AHAs for protease-restricted hinge neoepitopes, inhibition experiments were performed using the ELISA system and the results were confirmed using a an SPRi array. In both set-ups anti-hinge reactivity was only clearly inhibited by adding IgG-derived F(ab′)2s generated with the same protease (Fig. 4). As evidenced by Fig. 4A–D, two independent AHA reactivities against the same IgG subclass, that is, AHAs specifically targeting pepsin-IgG1-F(ab′)2s and AHAs specifically targeting IdeS-IgG1-F(ab′)2s, can be present in the same individual.
Specificity of AHA responses for IgG subclass–specific hinge neoepitopes
Because the C-terminal amino acid residues exposed after cleavage of the IgG1 and IgG3 hinge, and to a lesser extent also the IgG4 hinge, by the same protease (pepsin or IdeS) are similar (Fig. 1B), we expected to find cross-reactivity in AHA+ sera for F(ab′)2s generated from different IgG subclasses with the same protease.
To investigate this potential cross-reactivity, we analyzed co-occurrence of anti–IgG1-F(ab′)2 reactivity with reactivity against other IgG subclasses cleaved with the same protease. Fig. 5A–C show that the vast majority of samples with anti-hinge reactivity against pepsin-cleaved IgG1 and/or IgG2 is reactive with either IgG1-F(ab′)2s or IgG2-F(ab′)2s, not with both. The C-terminal amino acid residues exposed by IgG2-F(ab′)2s after proteolysis with pepsin, determined in the present study by native mass spectrometry (Fig. 1B, Supplemental Fig. 1B), are very different from those exposed by IgG1-F(ab′)2s, and the infrequent co-occurrence of anti–pepsin-IgG1-F(ab′)2 reactivity and anti–pepsin-IgG2-F(ab′)2 reactivity suggests that in HDs as well as in RA and SLE patients these are two separate AHA responses. In contrast, IgG1- and IgG3-F(ab′)2s created by pepsin cleavage should have similar exposed C-terminal amino acids residues (Fig. 1B) with a one amino acid difference from the core hinge to the C terminus (CPPCPAPEL in IgG1 versus CPRCPAPEL in IgG3). In the RA and SLE cohorts most AHA+ sera showed AHA reactivity against both pepsin-IgG1- and pepsin-IgG3-F(ab′)2s (Fig. 5E, 5F). The quantitative correlation between AHA levels against IgG1- and IgG3-F(ab′)2s was statistically significant for the RA group, and there was a statistically significant association between anti–pepsin-IgG1-F(ab′)2 and anti–pepsin-IgG3-F(ab′)2 reactivity for all three groups (Fisher exact test two-tailed p value <0.001). Analogous to IgG1- and IgG3-F(ab′)2s, the C-terminal amino acid residues exposed by IgG1- and IgG4-F(ab′)2s after pepsin cleavage are also similar, but with two amino acid differences from core hinge to C terminus (CPPCPAPEL in IgG1 versus CPSCPAPEF in IgG4), including the C-terminal residue. As noted above, the prevalence of anti–pepsin-IgG4-F(ab′)2 reactivity was much higher in the RA group compared with the HD and SLE cohorts (35 versus 6 and 5%). In the HD cohort and the SLE cohort most sera contained only reactivity against IgG1-F(ab′)2s (Fig. 5G, 5I), whereas in the RA cohort there were as many double-positive sera as there were single-positive sera (Fig. 5H). The correlation between levels of AHA reactivity against pepsin-IgG1- and pepsin-IgG4-F(ab′)2s was not statistically significant when taking into account the many single-positive samples. Interestingly, single positivity for IgG4-F(ab′)2s was not seen in HDs, whereas seven sera in the RA cohort, and one SLE patient, contained AHA reactivity specifically against IgG4-F(ab′)2s. Of the anti-IgG1/IgG4 double-positive sera, some probably contain two separate AHA responses against IgG1- and IgG4-F(ab′)2s, whereas others contain one AHA response that is cross-reactive. Inhibition experiments with F(ab′)2s generated from IgG1, IgG2, and IgG4 therapeutic Abs (Fig. 6A–D) further illustrate that cross-reactivity can play a role in anti–pepsin-F(ab′)2 reactivity. In an HD with AHA reactivity against pepsin-IgG1-, IgG3-, and IgG4-F(ab′)2s, the anti–IgG4-F(ab′)2 reactivity was more efficiently inhibited with IgG1-F(ab′)2s than with IgG4-F(ab′)2s. AHA reactivity against pepsin-IgG2-F(ab′)2s did not cross-react with other IgG subclasses, as it was only inhibited by IgG2-F(ab′)2s (Fig. 6B).
To further confirm that the anti-IgG1 and anti-IgG4 hinge reactivities were directed against the C termini identified in the mass spectrometric analyses and not against alternative pepsin cleavage sites, that is, between V240 and F241 or between F241 and F242 (20), or against the IdeS cleavage site, inhibition experiments were performed with synthetic hexamer or heptamer peptide analogs of the C-terminal amino acid stretches of the identified and alternative hinge cleavage sites (Table I). Only peptide analogs of the IgG1-F(ab′)2 C terminus identified by mass spectrometry (i.e., GPAPEL) were able to inhibit anti–pepsin-IgG1-F(ab′)2 reactivity (Fig. 6E). Anti–pepsin-IgG2-F(ab′)2 reactivity could only be inhibited by an IgG2-derived peptide corresponding to one of the two C termini identified by mass spectrometry. This experiment therefore also revealed that this reactivity is dependent on the presence of a phenylalanine at the C terminus of the IgG2-F(ab′)2 (Fig. 6F).
By ELISA (Fig. 6G, 6H) or SPR (Fig. 6I, 6J), we confirmed that the anti-IgG4 response in an RA patient can target pepsin-cleaved IgG4 specifically; whereas healthy serum with IgG4 reactivity was blocked by either IgG1- or IgG4-F(ab′)2s (Fig. 6G, 6I), the single anti-IgG4+ RA sera were only blocked by IgG4-F(ab′)2s (Fig. 6H, 6J). The same results were seen when synthetic peptide analogs of the IgG1 and IgG4 hinges were used as inhibitors instead of F(ab′)2s (Fig. 6K, 6L).
Cleavage of IgG1, IgG3, and IgG4 with IdeS results in F(ab′)2s with longer exposed C-terminal amino acid residues compared with cleavage with pepsin (Fig. 1B), but the number of amino acid differences between the subclasses is the same. One amino acid differs between IgG1 and IgG3 and two differ between IgG1 and IgG4 (CPPCPAPELLG in IgG1 versus CPRCPAPELLG in IgG3 and CPSCPAPEFLG in IgG4), although in this case the C termini are the same. In all three cohorts most AHA+ sera showed AHA reactivity against both IdeS-IgG1- and IdeS-IgG3-F(ab′)2s, whereas anti–IdeS-IgG1 single positivity was more prevalent than anti-IgG1/IgG4 double positivity (Fig. 7). Significant correlations were found between levels of AHA reactivity against IdeS-IgG1- and IgG3-F(ab′)2s, where the C-terminal amino acid residues exposed in the hinge after proteolysis are more similar than between IgG1- and IgG4-F(ab′)2s. Anti–IdeS-IgG1-F(ab′)2 reactivity could not be compared with reactivity against IgG2, because no AHA reactivity was found against IdeS-IgG2-F(ab′)2s.
In sera with AHA reactivity against IdeS-F(ab′)2s the levels of AHAs measured against IgG1-F(ab′)2s were about equal to the levels measured against IgG3-F(ab′)2s (the samples cluster around the diagonal line in Fig. 7A–C). Levels of AHA reactivity against IgG4-F(ab′)2s are generally lower (Fig. 7D–F). This suggests that for reactivity against IdeS-F(ab′)2s, IgG1 hinge and/or IgG3 hinge are the primary targets of AHAs, and reactivity against IgG4 hinge may represent cross-reactivity.
AHAs, autoantibodies that target cryptic epitopes exposed in the hinge of IgG molecules after cleavage by proteases, are found both in healthy individuals and at higher levels in patients with RA (13, 14). Recently, using an RIA that only detects high-affinity Abs, our group found AHAs specifically targeting the pepsin-cleaved hinge of IgG4 in a subset of RA patients (15), suggesting that AHAs can be very specific and associated with disease.
In the present study, we developed a more sensitive ELISA, retaining the high specificity we observed in the RIA (15) by using anti-biotin F(ab′)2 fragments that were immobilized not via direct coating but via binding to biotinylated HSA, thereby retaining their native structure (16) and ensuring unidirectional presentation of the F(ab′)2 targets to the AHAs.
We show that IgG subclass– and protease-specific AHA responses are present in the healthy population and in patients with the autoantibody-mediated diseases RA and SLE. We demonstrate that AHAs specifically recognize various hinge epitopes depending on the IgG subclass and the protease that are used for creating the F(ab′)2 targets bound by the AHAs. Only when the C-terminal amino acid residues exposed in the hinge after proteolysis were similar did AHAs cross-react against F(ab′)2 originating from other IgG subclasses. By using two proteases, pepsin and IdeS, that cleave the hinge at locations just two amino acids apart in IgG1, IgG3, and IgG4, we found that AHAs specifically recognize F(ab′)2 targets generated by different proteases. Reactivity against IdeS-cleaved IgG1, IgG3, and IgG4 and pepsin-cleaved IgG of all four IgG subclasses was found in HDs, established SLE patients, and generally more frequently and at higher levels in patients with early active RA. In agreement with earlier research (28), we did not detect AHAs binding to IdeS-cleaved IgG2. The relative resistance to proteolytic cleavage of IgG2 is a probable factor in this (28) by limiting the availability of IgG- F(ab′)2 products in vivo to mount an AHA response against. We did however find specific AHA reactivity against pepsin-cleaved IgG2, suggesting that proteases are active in vivo that can cleave IgG2 at the pepsin cleavage site. To our knowledge the current study is the first to show that pepsin cleaves IgG2 symmetrically between F241 and L242, and asymmetrically between V240 and F241 on one side and between F241 and L242 on the other side (Supplemental Fig. 1B). In inhibition assays with synthetic peptide analogs of the IgG2 hinge terminating at V240 or F241 we showed that anti–pepsin-IgG2-F(ab′)2 reactivity is dependent on exposure of F241. This explains why in a previous study (28) anti-IgG2 hinge peptide reactivity was not observed, because the most downstream peptide analog that was used to test for AHA reactivity terminated at V240.
AHAs targeting pepsin-cleaved IgG1 and IgG3 that were cross-reactive with pepsin-cleaved IgG4 were found both in patients as well as in HDs. Although IdeS has been shown to very specifically cleave IgG between amino acids G236 and G237 for IgG1 and IgG4, and between A235 and G237 for IgG2 (G236 is deleted in IgG2) (18), producing an intact F(ab′)2 and an intact Fc fragment, cleavage of IgG1 by pepsin was recently shown to produce F(ab′)2 cleaved at one of three possible sites: a major product resulting from cleavage between L234 and L235, as previously reported by Diemel et al. (9), but also two minor products corresponding to cleavage between V240 and F241, or between F241 and F242 (20), generating F(ab′)2 with different C termini. Analysis by native mass spectrometry of our F(ab′)2 targets showed that the anti-biotin IgG1 had been cleaved between L234 and L235, and inhibition experiments performed with the peptide analogs representing the three possible pepsin IgG1 cleavage sites showed that anti–pepsin-IgG1-Fab2 reactivity was only inhibited by the peptide analog of this pepsin cleavage site (i.e., GPAPEL) for most of the samples tested, and never with peptide analogs of the alternative cleavage sites. Interestingly, not all sera with anti–pepsin-IgG1-Fab2 reactivity could be inhibited with this peptide (nor any of the other peptides). This finding, combined with the finding that not all sera are cross-reactive for pepsin-IgG3-Fab2 (Fig. 5), suggests a role of the core hinge, which differs between IgG1 and IgG3, as part of the primary hinge epitope for a subset of samples. This also implies that Abs binding to pepsin-cleaved IgG1 represent multiple anti-hinge reactivities.
We confirmed our previous finding (15) that AHAs exclusively targeting pepsin-cleaved IgG4 are virtually RA specific and are not cross-reactive with F(ab′)2s generated from other IgG subclasses. We identified the exact cleavage site for pepsin in IgG4 (Supplemental Fig. 1C) and show that a peptide analog of this cleavage site can specifically inhibit the anti–pepsin-IgG4-F(ab′)2 reactivity. Importantly, using the more sensitive ELISA described in this study, more than one third of RA patients were found positive for AHA reactivity against pepsin-IgG4-F(ab′)2s, whereas only 6% of HDs and 5% of SLE patients had AHAs binding this target.
Many proteases can cleave the IgG hinge (9–11). Considering that most HDs tested positive for reactivity against at least one of eight F(ab′)2 targets generated in this study, it seems likely that AHAs are part of the immune system of most individuals. This is in line with previous findings by Brezski et al. (29). We found AHA positivity more frequently with the AHA ELISA developed in this study compared with the previously described RIA, which is likely caused by differences in assay sensitivity, including for higher and lower affinity Abs, with the RIA being biased toward high-affinity Abs. The variable frequency of AHAs in previous studies is probably also caused by differences in assay sensitivity. Assays using directly coated F(ab′)2 fragments for AHA detection may be less specific (16), but inhibition experiments can be used to establish specificity. An important factor in binding of AHAs may be the conformational stability of the exposed hinge. We show that AHAs binding to cleaved IgG2 bind much more distally from the stabilizing core hinge than do AHAs binding to IgG1, 3, or 4. That our single-strand synthetic peptide hinge analogs inhibit anti-IgG2 reactivity at much lower peptide concentrations (micromolar versus millimolar) compared with anti-IgG1, 2, or 4 reactivity suggests that a stable conformation of the epitope is more critical for the latter reactivities.
The highly restricted specificity we observed in the present study suggests that AHAs develop in response to C-terminal amino acid residues exposed after IgG cleavage. This antigenic stimulus can theoretically occur during an infection with microbes producing IgG-cleaving proteases (30). Brezski et al. (29, 31) suggested that AHAs may play a role in the host defense system by binding to the proteolytically cleaved IgG hinge, thereby restoring Fc-mediated Ab effector functions. This would be beneficial in case of infection or when antitumor Abs are cleaved by tumor-related proteases (30). However, when IgG is cleaved by proteases in the inflamed joint of an RA patient, AHAs could lead to formation of immune complexes and activation of complement, exacerbating inflammation (11, 32). In RA, AHAs may crosslink proteolytically cleaved disease-specific autoantibodies such as ACPAs to form immune complexes, potentiating ACPA-mediated inflammation. Recent reports showing immune complex formation between rheumatoid factors and ACPAs support this possibility (33–35).
The lack of correlated AHA reactivity against F(ab′)2 targets generated from the same IgG subclass with pepsin and IdeS suggests that these AHA responses have a separate origin and developed in a different context. Considering that IdeS is a bacterial protease and that pepsin cleaves IgG at the same site where MMP-7 cleaves IgG, we might suggest that anti–IdeS-cleaved IgG responses develop during (bacterial) infection whereas anti–pepsin-cleaved IgG responses develop in the context of inflammation and tissue remodeling that take place in the joints of an RA patient under the influence of proteases such as MMPs. The increased protease activity in RA could induce a considerable antigenic stimulus of hinge-cleaved IgG molecules, explaining the higher AHA levels in RA patients compared with HDs and SLE patients. The fact that 59% of the SLE patients were on immunosuppressive treatment (the RA patients had not started treatment) could also have had an effect on the levels in SLE patients. However, comparing AHA levels against the F(ab′)2 targets between SLE patients with or without immunosuppressive treatment did not show statistically significant differences (data not shown). Because chronic inflammation is an important feature of RA and prolonged antigenic stimulation leads to IgG4 formation (36), the finding that AHAs against pepsin-IgG4-F(ab′)2s are specific for RA suggests that they are a marker of such processes. In RA, the effect of IgG4 hinge–specific AHAs binding to cleaved IgG4 could be especially detrimental by antagonizing the anti-inflammatory properties of IgG4. It was recently shown that AHAs against pepsin-IgG4-F(ab′)2s could trigger in vitro complement activation following binding to IgG4-F(ab′)2s (15).
In conclusion, AHAs target specific IgG subclass– and protease-restricted hinge neoepitopes and are readily detected with our sensitive AHA ELISAs in most HDs and SLE patients, but at elevated frequencies and at higher levels in RA patients. The restricted specificity of AHAs suggests that different AHA responses develop under distinct inflammatory or infectious conditions and may be markers of, and participants in, such processes. In the proteolytic environment of the RA joint, AHAs may promote pathogenic immune complex formation, leading to further tissue destruction and inflammation. Further research relating the various AHA reactivities to different immunological processes should identify specific AHAs as useful markers for disease prediction, diagnosis, and evaluation of treatment.
This work was supported by ZonMw, the Netherlands Organisation for Health Research and Development, in the program 2Treat (Grant 436001001).
The online version of this article contains supplemental material.
Abbreviations used in this article:
anti-citrullinated protein Ab
human serum albumin
IgG-degrading enzyme of Streptococcus pyogenes
IgG-degrading enzyme of Streptococcus pyogenes–generated F(ab′)2
systemic lupus erythematosus
surface plasmon resonance
surface plasmon resonance imaging.
The authors have no financial conflicts of interest.