Shared VH1-46 gene usage has been described in B cells reacting to desmoglein 3 (Dsg3) in the autoimmune disease pemphigus vulgaris (PV), as well as B cells responding to rotavirus capsid protein VP6. In both diseases, VH1-46 B cells bearing few to no somatic mutations can recognize the disease Ag. This intriguing connection between an autoimmune response to self-antigen and an immune response to foreign Ag prompted us to investigate whether VH1-46 B cells may be predisposed to Dsg3-VP6 cross-reactivity. Focused testing of VH1-46 mAbs previously isolated from PV and rotavirus-exposed individuals indicates that cross-reactivity is rare, found in only one of seven VH1-46 IgG clonotypes. High-throughput screening of IgG B cell repertoires from two PV patients identified no additional cross-reactive clonotypes. Screening of IgM B cell repertoires from one non-PV and three PV patients identified specific cross-reactive Abs in one PV patient, but notably all six cross-reactive clonotypes used VH1-46. Site-directed mutagenesis studies indicate that amino acid residues predisposing VH1-46 Abs to Dsg3 reactivity reside in CDR2. However, somatic mutations only rarely promote Dsg3-VP6 cross-reactivity; most mutations abolish VP6 and/or Dsg3 reactivity. Nevertheless, functional testing identified two cross-reactive VH1-46 Abs that both disrupt keratinocyte adhesion and inhibit rotavirus replication, indicating the potential for VH1-46 Abs to have both pathologic autoimmune and protective immune functions. Taken together, these studies suggest that certain VH1-46 B cell populations may be predisposed to Dsg3-VP6 cross-reactivity, but multiple mechanisms prevent the onset of autoimmunity after rotavirus exposure.
To combat the numerous foreign insults a typical human would encounter on a daily basis, there must be a diverse B cell repertoire. The heterogeneity of the human Ab repertoire can be as high as 1011 specificities per individual (1, 2). The trade-off for a diverse B cell repertoire is that autoreactivity toward self-antigens may occur, leading to autoimmunity. Desmoglein 3 (Dsg3) is a desmosomal cadherin responsible for mediating intercellular adhesion in stratified squamous epithelia. It is the autoantigen targeted in pemphigus vulgaris (PV) (3), a disease in which autoantibodies to Dsg3 disrupt epidermal adhesion in the skin and mucous membranes, causing potentially fatal blistering. Dsg3-reactive autoantibodies are both necessary and sufficient to cause an acantholytic phenotype in PV IgG passive transfer mouse models (4, 5). Blister formation is independent of complement and can be induced even by monovalent Fab or single-chain variable fragment (scFv) Abs (6–8), which cause endocytosis and loss of keratinocyte cell surface Dsg3 (9), underscoring the direct pathogenic effects resulting from binding of Ab-variable regions to Dsg3.
We previously characterized the B cell repertoires of four PV patients with active disease and discovered common utilization of the VH1-46 Ab gene segment by anti-Dsg3 B cells across these unrelated patients (10). Anti-Dsg3 VH1-46 B cells had relatively few somatic mutations overall and required few to none of these somatic mutations to bind Dsg3. Interestingly, VH1-46 gene usage has also been observed in the immune response to rotavirus infection, specifically to the rotavirus VP6 intermediate capsid protein (11, 12), and similarly VH1-46 VP6-reactive B cells had few to no somatic mutations (13, 14). In both disease conditions, amino acid residues in or near the H chain CDR2 were shown to be critical for Dsg3 or VP6 reactivity (10, 15).
The compelling similarity of features in the B cell response to self and foreign Ag prompted us to determine whether VH1-46 B cells may be predisposed to cross-react to Dsg3 and rotavirus VP6. To address this question, we performed both focused testing and high-throughput screening of B cell repertoires to identify cross-reactive Abs, as well as site-directed mutagenesis studies to investigate the sequence features that promote reactivity toward Dsg3 and/or VP6.
Materials and Methods
Patient recruitment and characteristics
Peripheral blood was collected from individuals following informed consent using a protocol approved by the Institutional Review Board of the University of Pennsylvania. The diagnosis of PV was established by typical clinical presentation, histology, and ELISA and/or immunofluorescence (IF). A summary of the patient characteristics appears in Supplemental Table I.
Ab phage display
PBMCs were isolated from 50 to 60 ml of peripheral blood via Ficoll (Sigma-Aldrich). RNA was isolated from PBMCs using the RNeasy Midi kit (Qiagen). cDNA amplification was carried out using the SuperScript first-strand system (Thermo Fisher Scientific). The primers sets used to amplify all expressed H and L chains were generated as described (16). PCR reactions were gel electrophoresed on a 2% agarose NuSieve 3:1 gel (Lonza), and bands were imaged using SYBR Safe DNA gel stain (Thermo Fisher Scientific). Bands were purified using a Wizard SV gel and PCR clean-up system (Promega), and DNA concentrations were quantitated using Low Mass DNA Ladder (Thermo Fisher Scientific). The L and H chains were assembled via overlap PCR, digested with SfiI (Promega), and ligated into the pComb3x vector (16) to generate the Ab phage display library. Libraries were electroporated into XL1-Blue (Stratagene) and titered and sequenced to determine library diversity.
To enrich for Ag-specific clones, libraries were panned as previously described (10), using ELISA plates as described below, with the following modifications. Phages (≥1012) were incubated in ELISA wells containing the Ag of interest for 2 h at either room temperature (RT) or 4°C. Wells were washed five times for 2–5 min each using TBS-Ca2+ plus 0.1% Tween 20 after the first round of panning, and 10 times for 2–5 min each for panning rounds two through four. Bound phages were eluted with 76 mM citric acid, amplified, and repanned for up to four rounds to enrich for Ag-specific clones. Approximately 40 colonies were picked for sequencing from the third and fourth rounds of panning, and unique clones were produced as soluble scFv in TOP10F′ for characterization. Standard screening was to test for reactivity against Dsg3, VP6, and BSA as an irrelevant Ag. mAbs that were positive for all three Ags were additionally tested against Hep2 ELISA; a positive result on this assay indicates a polyreactive Ab whereas a negative result was designated as a multireactive Ab. To determine somatic mutations and V(D)J segments used by each Ab, sequences were analyzed using IMGT/V-QUEST (17, 18) and Vbase2 (19).
cDNA was reverse transcribed from various human tissue RNA samples (Clontech Laboratories) using the High Capacity RNA-to-cDNA kit (Applied Biosystems). Quantitative PCR (qPCR) studies were run and analyzed on a ViiA 7 real-rime PCR system (Applied Biosystems) using the universal EXPRESS SYBR GreenER qPCR Supermix (Thermo Fisher Scientific). Primer sets are as follows (forward/reverse): Dsg3, 5′-TTCCTGATCACATGTCGGGC-3′, 5′-CACCAGTGAGTTTGAGGCACT-3′; Stro-1, 5′-TTGCCAGAGCCAACGTCAAG-3′, 5′-CGGCGCTGATCAGGTTGTTT-3′; CD90, 5′-AAGACCCCAGTCCAGATCCAG-3′, 5′-TGCTGGTATTCTCATGGCGG-3′; and β-actin, 5′-AGAGCTACGAGCTGCCTGAC-3′, 5′-AGCACTGTGTTGGCGTACAG-3′. Transcript abundance in each tissue sample was quantitated by the change-in-cycling threshold (ΔΔCt) method after normalization utilizing β-actin as a control gene. Reactions were run on an agarose gel (Lonza). Biological replicates were run in triplicate, with three technical replicates per biological replicate.
Production of rotavirus double-layered particles
Rhesus rotavirus double-layered particles (DLPs) were purified by isopycnic centrifugation in cesium chloride at a density of 1.38 g/cm3 as described previously (20). DLPs were dialyzed against 50 mM Tris (pH 7.4), 100 mM NaCl, and 0.1 mM EDTA and were stored at 4°C in the presence of protease inhibitors (Roche). Protein/RNA concentration of the preparations was determined by UV spectrophotometry (OD280). Protein quality was assessed by SDS-PAGE in 10% acrylamide gels and SilverQuest stain (Life Technologies). Particle integrity was assessed by transmission electron microscopy following staining with 1% uranyl acetate.
Dsg3 (Euroimmun) and Hep2 (IBL International) ELISAs were performed according to manufacturers’ directions, except in cases where mAb binding took place at 4°C overnight. Some mAbs were tested as unpurified bacterial lysates in case of inefficient production. To make VP6 and BSA ELISA plates, 5 μg/ml of each protein (DLPs or purified BSA) was diluted in PBS-Ca (Thermo Fisher Scientific) and incubated in 96-well plates (Corning) at 4°C overnight. Plates were washed with TBS-Ca2+ with 1% Tween 20 (Bio-Rad) before blocking with TBS-Ca2+ plus 3% milk (Bio-Rad, Sigma-Aldrich) for 1 h at RT. VP6/BSA ELISAs were then carried out according to the standard protocol. In some cases, plates were developed with high sensitivity tetramethylbenzidine reagent (BioLegend, Thermo Fisher Scientific). In all assays, the Dsg3 control was clone (D31)2/29 (8), the VP6 control was RV6-26, and the negative control was a mAb against an irrelevant Ag.
Ab mutation experiments
Somatic mutations identified in each mAb were reverted to the corresponding germline (GL) nucleotide based on IMGT/V-QUEST and/or Vbase2 as described, with more than one GL clone produced when more than one prediction was possible (10). GL L chains and H chains were synthesized as GeneArt strings (Thermo Fisher) and amplified with the primer set 5′-GGGCCCAGGCGGCCGAGCTC-3′, 5′-GGGCCGGCCTGGCCACTAGTGACCGATGGGCCCTTGGTGGAAGCTGAGGAGACGGTGACC-3′ and then digested with SfiI (Roche). Digested DNA was purified using the Wizard SV gel and PCR clean-up system (Promega) and ligated into the pComb3x vector using T4 DNA ligase (Thermo Fisher Scientific). Point mutations in various mAbs were carried out using the QuikChange Lightning multi site-directed mutagenesis kit (Agilent Technologies).
Normal human skin was acquired from dermatologic surgery procedures using Institutional Review Board–reviewed protocols and sectioned onto glass slides (Thermo Fisher Scientific). Monkey esophagus slides were purchased from Scimedx. Slides were blocked with TBS-Ca2+ (Bio-Rad) plus 1% BSA (Sigma-Aldrich) at RT for 30 min. Slides were washed with TBS-Ca2+ three times and incubated with scFv diluted in TBS-Ca2+ plus 1% BSA at RT for 1 h. Slides were washed as above and incubated with 1:100 rat anti-hemagglutinin Ab (3F10; Roche) diluted in TBS-Ca2+ plus 1% BSA for 1 h at RT. Slides were washed as above and incubated with 1:200 Alexa Fluor 594 donkey anti-rat IgG mAb (Thermo Fisher Scientific) for 30 min at RT. Slides were washed as above, fixed with 95% ethanol, and mounted with mounting medium (KBL). IF was visualized with an Olympus BX61 microscope and images were acquired using SlideBook 4.2 software (Olympus) and a Hamamatsu Orca ER camera.
Keratinocyte dissociation assay
The keratinocyte dissociation assay was performed as described (21). Human primary keratinocytes were derived from neonatal foreskin using Institutional Review Board–reviewed protocols, then seeded into 12-well tissue culture plates in DK-SFM media (Thermo Fisher Scientific). Confluent wells were incubated with mAbs diluted in DK-SFM media supplemented with 1.2 mM CaCl2 for 6 h at 37°C. For Abs not reactive to Dsg1, exfoliative toxin A (Toxin Technology) was added for a final concentration of 1 μg/ml to mAb mixtures during the last 2 h of the incubation. Wells were washed with PBS-Ca2+ and incubated with Dispase (Roche) at 37°C for 30 min. Cell monolayers were washed with PBS-Ca2+ and subjected to five rounds of pipetting with a 1 ml pipettor (Rainin). Cell monolayers were stained with Crystal Violet (Sigma-Aldrich), and cell fragments were counted using ImageJ.
Rotavirus replication assay
Lipofectin (Thermo Fisher Scientific) was incubated 15% (v/v) with serum-free Eagle’s MEM (American Type Culture Collection) at RT for 30 min. Rotavirus DLPs were added to the lipofectin-treated media at 2 μg/ml and incubated for 40 min at RT. Abs were diluted 1:4 starting at 25 μg/ml in serum-free EMEM, added 1:1 to the DLP-lipofectin mixture, and incubated at 37°C for 1 h. Forty microliters of DLP-Ab mixture was added to 100% confluent monkey renal MA104 cells (American Type Culture Collection) in a 96-well plate (Corning) and incubated at 37°C for 4 h. Forty microliters of 20% FBS and 2% penicillin/streptomycin (Thermo Fisher Scientific) was added to wells and incubated at 37°C overnight.
Cells were washed with PBS-Ca2+ (Thermo Fisher Scientific) and fixed with ice-cold methanol for 15 min at −20°C. Fifty microliters of 1:500 polyclonal goat anti-rotavirus Ab (Fitzgerald) was added and incubated at 37°C for 1 h. Cells were washed with PBS-Ca2+ and incubated with 50 μl of donkey anti-goat HRP secondary Ab (Abcam) at 1:1000 for 1 h at 37°C. Cells were washed with PBS-Ca2+ and stained with the Pierce diaminobenzidine substrate kit (Thermo Fisher Scientific). Cells were washed with PBS-Ca2+ and wells were imaged on an EVOS FL auto cell imaging system (Thermo Fisher Scientific). Rotavirus foci were quantitated using ImageJ and the immunohistochemistry image analysis toolbox. Percentage reduction was calculated based on the difference in number of foci in each Ab condition compared with the no Ab control. Ab conditions were run in duplicate and imaged at least once per well.
Error bars in all figures indicate SEM. Cutoff values were determined by calculating 3 SD above the mean of the negative controls in each experimental replicate.
Focused testing of previously isolated VH1-46 IgG mAbs from PV and rotavirus-exposed individuals reveals rare cross-reactivity to Dsg3 and VP6
In both the autoimmune disease PV and rotavirus infection, VH1-46 mAbs with few to no somatic mutations have been shown to bind the disease Ag [Dsg3 in PV (10) and VP6 in rotavirus (14)]. This finding prompted us to investigate whether VH1-46 mAbs might cross-react to Dsg3 and VP6. We first performed ELISA binding assays against Dsg3, VP6, and an irrelevant Ag (BSA) using five VH1-46 anti-Dsg3 clonotypes previously isolated from four PV patients with active disease (10), as well as two VH1-46 anti-VP6 clonotypes (RV6-25 and RV6-26) previously isolated from a rotavirus-exposed individual (22). One of five somatically mutated (SM) PV-derived VH1-46 mAbs (clone 4.2) demonstrated cross-reactivity to VP6, whereas neither RV6-25 nor RV6-26 demonstrated cross-reactivity to Dsg3 (Fig. 1A).
Because VH1-46 predominance was previously observed in VP6-reactive naive (i.e., unmutated) IgM B cells but VH1-46 enrichment did not persist in somatically mutated IgG memory B cells (15), we reverted somatic mutations in the VH1-46 clonotypes to their GL VDJ sequences to determine whether the corresponding unmutated VH1-46 mAbs might demonstrate increased cross-reactivity. After GL reversion, neither RV6-25 nor 6-26 bound to Dsg3; two of the five anti-Dsg3 VH1-46 clonotypes (PVE4-8 GL and 4.2 GL1) exhibited borderline but negative cross-reactivity to VP6, with the lower end of the SEM range for these measurements falling below the cutoff value for VP6 reactivity (Fig. 1B).
High-throughput screening of patient IgG and IgM libraries confirms rare VH1-46 cross-reactivity to Dsg3 and VP6
To more comprehensively determine whether cross-reactive Ab H chains can bind both Dsg3 and VP6, we used Ab phage display to screen peripheral blood IgG and IgM B cell repertoires. As previously reported, IgG B cell libraries from two PV patients (PV1, PV3, and PV3a) identified anti-Dsg3 clonotypes that include VH1-46 (8, 10, 23), of which only one clonotype (4.2) was cross-reactive (Fig. 1A). We screened these libraries, alternating between Dsg3 and VP6 Ags, to enrich for cross-reactive clones. However, cross-panning identified no additional cross-reactive clones from these three libraries. Single VP6 Ag panning of these same libraries indicated that anti-VP6 IgG B cell repertoires were dominated by one VH4-34 clonotype in PV1 and two VH4-39 clonotypes in PV3 (Supplemental Table II), perhaps explaining the lack of IgG cross-reactivity observed in these patients.
The IgM B cell repertoire is formed in the bone marrow, before exposure to peripheral Ag. We find that Dsg3 is not expressed in the bone marrow or secondary lymphoid organs, but it is expressed in thymic epithelia and stratified squamous epithelia of skin (Fig. 2A). Bone marrow stromal cells were present in the sample as evidenced by the expression of CD90 and Stro-1 (24) (Fig. 2B), indicating that despite the Aire-regulated expression of Dsg3 in thymic stromal epithelia (25), an equivalent population in the bone marrow does not express Dsg3. These data indicate that B cells maturing in the bone marrow and emerging into the periphery would likely not encounter sufficient Dsg3 Ag to undergo clonal deletion or become anergic. Therefore, using combinatorial libraries, Dsg3-reactive IgM H chains should theoretically be found in all individuals, including normal individuals. As discussed above, because VH1-46 predominance of the VP6 repertoire occurs in naive IgM but not IgG B cells, we reasoned that VH1-46 VP6 reactivity and hence cross-reactivity may be more common in IgM B cell libraries. We thus produced and screened combinatorial IgM libraries from two PV patients with active disease (PV8, PV16), one PV patient in remission (PV1c), and a normal healthy individual (NH1).
Sequencing of the unselected IgM libraries from each patient indicated a diversity of VH gene family usage, with only two VH1-46 H chains in a total of 110 clones analyzed across the four libraries (data not shown). Dsg3-based selection revealed VH1-46 gene usage by anti-Dsg3 Abs in all IgM PV patient libraries tested, plus a diversity of additional anti-Dsg3 VH gene usage in each patient (Supplemental Table III). Notably, Dsg3-based screening of the IgM library from the patient in remission (PV1c) identified eight unique Dsg3-reactive clonotypes, four of which used VH1-46 (Supplemental Table III). Two of these four VH1-46 clonotypes cross-reacted with VP6; furthermore, cross-panning of the PV1c IgM library against Dsg3 and VP6 identified five additional cross-reactive clonotypes, all five of which used VH1-46, indicating enrichment of VH1-46 cross-reactive IgM H chains in this individual. ELISA and IF using purified mAbs confirmed that six of the nine VH1-46 clones from PV1c IgM bound to Dsg3 and VP6 (Fig. 3A). In contrast, cross-panning of IgM libraries against Dsg3 and VP6 from two PV patients with active disease (PV8, PV16) and a presumably rotavirus-exposed but otherwise normal healthy individual (NH1) identified one VH1-69, four VH3-07/3-11/3-15/3-30, and 1 VH3-23 cross-reactive clonotypes, respectively (Supplemental Table III). However, all showed multireactivity and/or polyreactivity based on positive BSA with and without Hep2 ELISA (Fig. 3B), and none bound to Dsg3 expressed in human skin by IF, indicating that Dsg3 reactivity detected by ELISA in these cases was nonspecific.
VH1-46 IgM cross-reactivity in the PV1c library does not result in corresponding IgG cross-reactivity, because screening of the PV1c library against Dsg3 previously identified no anti-Dsg3 clones, consistent with the negative Dsg3 ELISA index value in this patient (23). Screening of the PV1c IgG library against VP6 identified two clonotypes using VH1-02 and VH4-04 (Supplemental Table II). Overall, the VP6-reactive IgG repertoire identified across both of these patients was VH4 family predominant, which is consistent with prior reports indicating that this is the most common VH family used in the rotavirus repertoire (11).
In summary, Dsg3-VP6 cross-reactive IgM H chains were identified in one PV patient in remission, in whom all six of the cross-reactive clonotypes identified used VH1-46. Cross-reactive IgM H chains in two PV patients with active disease and one normal healthy individual were also identified, although these IgM H chains did not use VH1-46, were multireactive or polyreactive to other Ags, and did not specifically bind to keratinocyte cell surface Dsg3 by IF. High-throughput screening did not identify any additional cross-reactive IgG clonotypes in any of the libraries tested. These data indicate that specific Ab cross-reactivity to Dsg3 and VP6 is rare, but when it occurs, it is restricted to VH1-46 IgM B cells.
Somatic mutation analysis of VH1-46 mAbs reveals largely divergent patterns of amino acid reactivity to Dsg3 and VP6
We next sought to identify sequence regions and features that confer Ab cross-reactivity to Dsg3 and VP6. VH1-46 predominance in Ab H chain gene usage suggests that residues in the H chain CDR1 or CDR2 may be most important. VH1-46, but not the closely related Ab gene segment VH1-02, is enriched in cross-reactive Abs. Alignment of VH1-02 and VH1-46 amino acid sequences indicates that these two VH genes differ most in their CDR2 and flanking residues (Fig. 4A). To determine whether CDR2 residues are important for VH1-46 Ab reactivity to Dsg3, we mutated VH1-46 CDR2 and/or the +1 and −1 flanking residues to the VH1-02 sequence in PVE4-8, a mAb that binds Dsg3 after reversion of somatic mutations to their GL VDJ sequences (10). Replacement of the VH1-46 CDR2 with the VH1-02 CDR2, as well as the VH1-02 CDR2 plus its +1 and −1 flanking residues, abolished PVE4-8 reactivity to Dsg3 (Fig. 4B), indicating that the three CDR2 residues that differ between VH1-46 and VH1-02 are critical for Dsg3 reactivity in this mAb clonotype. Mutation of only the +1 residue from S to N has no effect on PVE4-8 Ab reactivity, whereas mutation of the −1 residue from I to R/W results in an ∼25-fold decrease in relative affinity.
These data support our previous findings that VH1-46–specific CDR2 residues are necessary for VH1-46 Ab reactivity to Dsg3 (10). Similarly, previous studies have shown that amino acid mutations in and around the VH1-46 CDR2 determine VP6 reactivity (15). We therefore conducted mutagenesis experiments to characterize the H chain CDR2 somatic mutation patterns that could lead to cross-reactivity. Somatic mutations from either of two VP6-reactive VH1-46 mAbs, RV6-25 and RV6-26, which were previously shown to promote VP6 reactivity (15), were introduced into three different VH1-46 anti-Dsg3 mAb backbones, for a total of six permutations (Fig. 5A): 1) an SM VH1-46 mAb that cross-reacts to Dsg3 and VP6 (4.2 SM); 2) a GL VH1-46 mAb that reacts to neither Dsg3 nor VP6 (F779 GL2); and 3) a GL VH1-46 mAb that binds Dsg3 and has borderline negative reactivity to VP6 (PVE 4-8 GL).
For the first mAb (4.2 SM), mutagenesis of two amino acids in the H chain CDR2 (H-CDR2) to the respective RV6-25 residues reduced mAb binding to Dsg3 and abolished binding to VP6. Mutagenesis of the H-CDR2 plus the −1 and +4 flanking positions resulted in a complete loss of binding to both Dsg3 and VP6 (Fig. 5B). Mutagenesis of 4.2 SM to RV6-26 residues resulted in a complete loss of binding to VP6, but did not markedly affect Dsg3 reactivity (Fig. 5C). These data indicate that just a few alterations in CDR2 somatic mutations can abolish cross-reactivity of 4.2.
Insertion of RV6-25 or RV6-26 residues into F779 GL2 did not increase its affinity for either Dsg3 or VP6 above cutoff OD values (Fig. 5D, 5E), indicating that these mutations are not sufficient for VP6 reactivity in this VH1-46 mAb.
When RV6-25 residues were inserted into PVE4-8GL, the ability of this mutant to bind Dsg3 was completely lost, and no effect on VP6 reactivity was observed (Fig. 5F). Insertion of a single RV6-26 CDR2 mutation into PVE4-8 GL did not significantly affect reactivity to either Ag (Fig. 5G, diamond markers). However, when the H-CDR2 plus the −1 residue was mutated to the respective RV6-26 residues, there was an increase in reactivity to both Dsg3 and VP6 (Fig. 5G, triangle markers). Upon subsequent mutagenesis of the +1 and +5 residue to the relevant RV6-26 amino acid, reactivity to VP6 was abolished without a marked change in Dsg3 reactivity (Fig. 5G, square markers). A summary of the mutagenesis studies is shown in Table I. Taken together, these studies indicate that somatic mutations only rarely promote Ab cross-reactivity to Dsg3 and VP6.
|VH1-46 Ab .||.||.||Mutations Added .|
|Baseline Ag Reactivity .||RV6-25 .||RV6-26 .|
|F779 GL||Dsg3||−||No change||No change|
|VP6||−||No change||No change|
|VH1-46 Ab .||.||.||Mutations Added .|
|Baseline Ag Reactivity .||RV6-25 .||RV6-26 .|
|F779 GL||Dsg3||−||No change||No change|
|VP6||−||No change||No change|
A subset of cross-reactive VH1-46 mAbs can both inhibit rotavirus replication and disrupt keratinocyte adhesion
The finding of VH1-46 cross-reactivity to Dsg3 and VP6 prompted us to determine whether Ab reactivity leads to pathophysiologic effects against these two antigenic targets. To test the functional consequences of VP6 reactivity, we evaluated the ability of cross-reactive mAbs to inhibit rotavirus replication in a spreading infection assay (26). RV6-26 inhibits rotavirus replication by binding the transcriptional pore of rotavirus DLPs, which blocks the egress of nascent viral mRNAs (27). We tested all six cross-reactive IgM VH1-46 mAbs identified, plus 4.2 IgG by incubating scFv mAbs with rotavirus DLPs prior to transfection into MA104 host cells and counting the resulting number of viral foci. Of the seven VH1-46 mAbs tested in this assay, two (PV1c IgM DVDV-7 and DVDV-8) demonstrated a marked reduction in the number of rotavirus foci, comparable in potency to RV6-26 (Fig. 6A). To determine whether these Abs have pathogenicity relevant to PV, we performed a keratinocyte dissociation assay to measure the ability of these mAbs to disrupt adhesion of primary human keratinocytes (21). PV1cIgM DVDV-7 and DVDV-8 both increased keratinocyte dissociation relative to the negative PBS control, with PV1c IgM DVDV-7 showing pathogenicity comparable to (D3)3c/9 and (D3)2/29 PV mAb positive controls (Fig. 6B).
In the present study, we show that despite similar characteristics in the B cell response to the self-antigen Dsg3 and the foreign rotavirus Ag VP6, namely, shared VH1-46 gene usage and the requirement for few to no somatic mutations to bind the disease Ag, cross-reactivity of VH1-46 Abs to Dsg3 and VP6 is rare. The rarity of VH1-46 Ab cross-reactivity was confirmed through both focused testing of VH1-46 Abs previously isolated from four PV patients and one rotavirus-exposed individual (Fig. 1), which showed that one of seven VH1-46 clonotypes was cross-reactive, as well as high-throughput screening of combinatorial IgG and IgM libraries from PV patients in various stages of disease and one rotavirus-exposed but otherwise healthy individual, which indicated that only one of these four individuals showed specifically cross-reactive B cell clonotypes to Dsg3 and VP6 (Supplemental Table III). Although VH1-46 gene usage is observed in only 2% of the circulating repertoire in humans (28), consistent with our finding of VH1-46 gene usage in 2 out of 110 unselected clones in the unselected libraries from four individuals, all specifically cross-reactive clonotypes identified used VH1-46, indicating that cross-reactivity is enriched in VH1-46 IgM B cells.
Subsequent studies to investigate the reasons for the rarity of cross-reactive clones indicate that somatic mutations only rarely promote both Dsg3 and VP6 reactivity, and more often abolish reactivity to either one or both Ags (Fig. 5). Previously we had shown that acidic amino acid residues in the H chain CDR2 and CDR3 confer Dsg3 reactivity in VH1-46 clonotypes (10). This pattern is in contrast to the positively charged residues that are observed in the anti-VP6 RV6-26 paratope, which facilitate binding to the acidic rotavirus transcriptional pore (27). The precise epitope in Dsg3 bound by human VH1-46 Abs has not yet been determined. However, AK23, a pathogenic mouse PV mAb that utilizes murine VH1-53, which is the closest human homolog of VH1-46 (10), has been shown to bind to a 4-aa epitope in the N terminus of mouse Dsg3 (V3, K7, P8, D59), of which the first three residues including the lysine at position 7 are most important for AK23 binding (29). Taken together, these data suggest that cross-reactivity is uncommon due to the differing biochemical characteristics that promote Dsg3 versus VP6 binding, with acidic residues favoring Dsg3 reactivity and basic residues favoring VP6 reactivity.
Despite the rarity of Dsg3-VP6 cross-reactivity in the B cell repertoire, our studies identified that such cross-reactivity is largely restricted to VH1-46 B cells. Although the precise mechanism by which unmutated VH1-46 H chains may be predisposed to bind both Dsg3 and VP6 remains unknown, presumably certain sequence features in VH1-46 facilitate binding to Dsg3, and other distinct sequence features favor binding to VP6. These features could be seemingly nonspecific but confer a subtle advantage to binding Ag that leads to VH1-46 bias in the initial B cell response. For example, in the human Ab response to influenza, the hydrophobic VH1-69 CDR2 mediates interaction with the hydrophobic hemagglutinin stem, which is thought to explain the VH1-69–predominant B cell response observed in several individuals (30, 31). Additionally, cationic H chains, notably VH3 family members, are predominant in Abs reacting to anionic RBC or platelet Ags (32, 33). Previous studies identified polyreactive IgM from early B cell precursors, which frequently used VH3 family genes and had positively charged CDR3 sequences (34). In these studies, polyreactivity was defined based on reactivity to a panel of anionic Ags, including nucleic acids, insulin, and LPS, as well as Hep2 reactivity. In three of the four individuals for whom we screened IgM libraries, we identified multireactive or polyreactive H chains that bound Dsg3 and VP6, as well as BSA (isoelectric point = 4.7) or Hep2, and used VH3 family genes. Thus, polyreactivity is an additional, albeit nonspecific, mechanism for IgM B cell cross-reactivity to Dsg3 and VP6. Ultimately, most autoreactive or polyreactive B cells are eliminated from the B cell repertoire at various tolerance checkpoints during B cell maturation due to receptor editing, somatic mutation away from autoreactivity, or other mechanisms that prevent the maturation of autoreactive IgG class-switched B cells (34, 35). Our finding that Dsg3-VP6 cross-reactive clonotypes are not identified after high-throughput screening of IgG B cell repertoires, as well as previously published data indicating that anti-Dsg3 IgG clonotypes are not found in non-PV individuals (23), supports that these mechanisms are largely effective to prevent anti-Dsg3 autoimmunity. Nevertheless, the identification of 4.2 IgG indicates that Dsg3-VP6 cross-reactivity in the IgG compartment is possible and hence could be a potential trigger of autoimmunity in some individuals.
Interestingly, of the six cross-reactive VH1-46 IgM H chains identified, two were shown to inhibit rotavirus replication, and these same Abs also disrupt keratinocyte adhesion (Fig. 6). These data indicate that cross-reactive VH1-46 Abs have the potential to confer protection against rotavirus infection, but at the same time cause pathogenic autoimmune effects on skin. Even in the absence of pathogenicity by these clones, the activation of a cross-reactive B cell could subsequently result in presentation of Dsg3 autoantigen to Dsg3-reactive T cells, which have been shown to be detectable and dysregulated in certain HLA-susceptible individuals (36, 37). Thus, cross-reactive B cells originally activated by binding foreign VP6 Ag could either directly cause the PV phenotype or could stimulate Dsg3-reactive T cells to trigger a broader anti-Dsg3 B cell response, leading to PV.
Notably, Dsg3 and VP6 do not share any known structural homology at the protein level. Well-substantiated examples of direct Ab cross-reactivity to foreign and self-antigen due to molecular mimicry have only rarely been described; the best characterized example is the association of Campylobacter jejuni with the development of Guillain–Barré syndrome (38) due to Abs that cross-react between human ganglioside GM1 and bacterial lipooligosaccharide (39). The induction of autoimmunity in these studies is not 100% penetrant; only 4 of 10 rabbits sensitized with bacterial lipooligosaccharide demonstrate limb weakness, and large-scale epidemiologic studies indicate that C. jejuni exposure increases the risk of Guillain–Barré syndrome 100-fold, from 0.3 to 30 per 100,000, but still the vast majority of humans infected with C. jejuni do not develop autoimmunity. Additionally, Abs targeting N-acetyl glucosamine or M protein can interact with cardiac myosin (40), and anti-streptopain Abs can cross-react with peptides present in vimentin (41), which may contribute to the onset of rheumatic heart disease. In fogo selvagem, an endemic form of pemphigus foliaceus, anti-Dsg1 Abs can cross-react to sand fly salivary Ag LJM11 (42), and immunization of mice with LJM11 has been shown to protect against leishmaniasis (43), a disease that is prevalent in geographical regions affected by fogo selvagem. However, no known structural homology between Dsg1 and LJM11 has been found, so the precise mechanism for Ab cross-reactivity in fogo selvagem remains unclear.
To our knowledge, this is the first study to investigate shared VH gene usage as a determinant of Ab cross-reactivity to self and foreign Ag. The concept of shared VH gene usage as a basis for triggering autoimmunity has several differences compared with the molecular mimicry model, and perhaps better explains why only a small percentage of individuals exposed to a particular foreign agent develop autoimmunity, and conversely why cross-reactive Abs are not found in all individuals. The shared VH gene usage model indicates that two Ags initially stimulate similar B cell responses due to subtle biochemical differences in VH genes that favor Ag binding, although as the immune reaction proceeds, divergent evolution of the B cell lineages to foreign and self-antigen likely occurs, and secondary autoantigen-specific B cell responses can be stimulated, such that at the time of autoimmune disease diagnosis the vast majority of autoantibodies are likely only autoantigen-reactive and no longer cross-reactive. Although beyond the scope of this study, this hypothesis could be investigated by isolating Dsg3- and VP6-reactive B cell lineages in pemphigus patients and performing deep sequencing of the B cell repertoire to identify common ancestors. Ultimately such studies may further elucidate mechanisms of autoimmunity in PV.
We thank Ningguo Feng and Harry Greenberg for consultation on rotavirus replication assays.
This work was supported in part by National Institute of Arthritis and Musculoskeletal and Skin Diseases/National Institutes of Health Grants R56-AR064220 and P30-AR057217 (to A.S.P.), T32-AR007465 (to M.J.C. and E.M.M.), F31-AR066456 (to M.J.C.), and F30-AR065870 (to E.M.M.); Deutsche Forschungsgemeinschaft Grants EL711/1-1 (to C.T.E.) and HA6736/1-1 and HA6736/2-1 (to C.M.H.); Section of Medicine at the University of Luebeck Grant J03-2015 (to C.M.H.); and National Institute of Allergy and Infectious Diseases/National Institutes of Health Grants R01-AI116815, R21-AI113402, and R21-AI119588 (to C.E.B. and S.M.M.) and R21-AI083574 (to G.S. and J.E.C.).
The contents are solely the responsibility of the authors and do not necessarily represent the official views of the National Institute of Arthritis and Musculoskeletal and Skin Diseases, the National Institute of Allergy and Infectious Diseases, or the National Institutes of Health.
The online version of this article contains supplemental material.
Abbreviations used in this article:
single-chain variable fragment
The authors have no financial conflicts of interest.