Walnut antigens may trigger autoantibody development in pemphigus vulgaris via “hit-and-run” mechanism

Abstract

Background

Environmental factors as well as genetic predisposition are known to be critical for the development of autoimmunity. However, the environmental agents that trigger autoimmune responses have remained elusive. One possible explanation is the “hit-and-run” mechanism where the inciting antigens that initiate autoimmune responses are not present at the time of overt autoimmune disease.

Objective

Following our previous findings that some allergens may incite autoimmune responses, we have investigated the potential role of environmental allergens in triggering autoantibody development in patients with an autoimmune skin disease, pemphigus vulgaris (PV).

Methods

Revertant/germline monoclonal antibodies (mAbs) (with mutations on variable regions of heavy and light chains reverted to germline forms) of eight anti-desmoglein 3 (Dsg3) pathogenic mAbs from PV patients were tested for reactivity against a panel of possible allergens, including insects, pollens, epithelia, fungi and food antigens.

Results

All the PV germline mAbs were reactive with antigens from walnut including a well-known allergen, Jug r 2, and an uncharacterized 85 kD protein component. Sera from PV patients contained significantly higher levels of anti-Dsg3 autoAbs than walnut-specific antibodies, suggesting that the autoreactive B cell response in PV may be initially triggered by walnut antigens, but is subsequently driven by Dsg3.

Conclusion

Our findings suggest that walnut antigens/allergens may initiate autoantibody development in PV via a “hit-and-run” mechanism. The revertant/germline mAb approach may provide a paradigm for the etiological study of other allergic and autoimmune diseases.

Graphical Abstract

Capsule summary

Environmental antigens/allergens may trigger the development of autoimmune pemphigus vulgaris (PV). With future identification of additional PV associated food antigens, dietary restrictions may be recommended for the prevention and disease management of PV.

INTRODUCTION

Pemphigus vulgaris (PV) is a life-threatening autoimmune blistering disease that affects the skin and mucosa.¹ In PV, autoantibody (autoAb) binding to keratinocyte adhesion proteins desmoglein (Dsg) 1 and 3 leads to acantholysis and intraepidermal clefting histologically and blister formation clinically. AutoAb-mediated autoimmune diseases, including PV afflict more than 2.5% of general populations,² resulting in significant morbidity and mortality.³ While genetics plays an important role in determining the risk of autoimmunity, environmental triggers are also necessary for disease development.⁴ However, the environmental agents that initiate autoAb production are largely unknown. One obstacle to identifying the inciting antigens in autoimmune diseases concerns the “hit-and-run” mechanism, originally proposed for autoimmune diseases triggered by infectious microbes, whereby the inciting antigens might not be present by the time of overt autoimmune diseases.^5–8 According to this mechanism, the cross-reactivity between an infectious antigen and autoantigen could result in a long-lasting immune response even if the pathogen was cleared, as the autoantigen would drive the subsequent autoimmune response. Whether exposure to noninfectious environmental antigens or allergens can initiate autoAb responses through a similar mechanism remains unclear.

An important feature of pathogenic anti-Dsg3 autoAbs in PV is that they are primarily of the IgG4 isotype, suggesting that type 2 immune responses play an important role in pemphigus pathogenesis.^{9, 10} Indeed, Dsg3 specific Th2 cells are frequently found in patients with PV.^{11, 12} Additionally, patients with acute onset PV have high levels of Dsg3-specific IgG4 and IgE autoAbs,^{13, 14} further supporting the concept the PV is a Th2-driven disease. Given that allergens are potent inducers of Th2 responses,¹⁵ it is possible that exposure to environmental allergens may contribute to the development of autoAb production in genetically-susceptible individuals. In support of this hypothesis, our group has previously demonstrated that autoAbs from subjects with fogo selvagem, an endemic form of pemphigus foliaceus, cross-react with a salivary gland protein from indigenous sand flies, indicating that an environmental allergen may trigger the production of pathogenic anti-Dsg1 autoAb.^{16, 17} Whether environmental allergens can precipitate autoAb formation in subjects with non-endemic PV is unknown.

During humoral immune responses, interactions between antigen-specific T cells and B cells in germinal centers of lymphoid organs result in the production of antibodies with increasing affinity for antigen.¹⁸ This process, known as affinity maturation, is the result of somatic mutations in immunoglobulin heavy (H) and light (L) chain genes.¹⁹ While somatic hypermutation (SHM) is necessary for the generation of high-affinity antibodies against foreign antigens, it can also lead to autoAb formation.^{20, 21} Reactivity to autoantigens is acquired during SHM,^20–23 suggesting that autoreactive B cells may arise from normal immune responses directed against non-self antigens. To characterize the specificity of the naïve B cells that ultimately give rise to autoreactive B cells, we and others have used the strategy of generating revertant/germline mAbs from autoreactive mAbs that have undergone SHM. In this technique, the mutated immunoglobulin H and L chain genes in autoreactive mAbs are reverted to their germline sequences.^{16, 20, 24–26} Thus, the specificity of a germline mAb represents that of the naïve B cell from which autoreactive B cells originated. This approach has been used to investigate the inciting antigens for autoAb development in PV and systemic lupus erythematosus (SLE).^{20, 24–26} Prior studies of anti-Dsg3 mAbs indicated that their germline counterparts had diminished Dsg3 reactivity^{25, 26} and did not react to autoantigens from Hep-2 cells,²⁶ suggesting that foreign antigens may be the initial drivers of autoAb production. To further investigate this possibility, we generated revertant/germline forms of eight pathogenic anti-Dsg3 mAb derived from PV patients,^25–28 and then screened the specificity of these PV germline mAbs against a panel of environmental antigens/allergens. Surprisingly, we found that all PV germline mAbs reacted almost exclusively with walnut antigen extract. Further examination revealed that the PV germline mAbs specifically bound to the allergen Jug r 2, as well as a previously uncharacterized 85 kD protein in walnut extract. Sera from PV patients contained significantly higher levels of anti-Dsg3 autoAb than walnut-specific antibodies, suggesting that while walnut antigen may be the initial stimulus, the selection of autoreactive B cells is subsequently driven by reactivity to Dsg3. Taken together, our results provide evidence that walnut allergens may precipitate anti-Dsg3 autoAb production in genetically susceptible individuals through a “hit-and-run” mechanism.

METHODS

Generation and production of revertant/germline mAbs from pathogenic mAbs derived from PV and SLE patients.

The germline sequences of the H and L chain of the PV pathogenic mAbs were from published work of other investigators.^25–28 The anti-dsDNA mAbs from SLE patients were previously published,^29–32 and their H and L chain sequences were summarized by Rahman et al.³³ These SLE mAbs were compared to the immunoglobulin database using the National Center for Biotechnology Information (IgBlast)³⁴ to determine their germline sequences. The constructs of each scFv pathogenic or germline mAb were synthesized (Genscript, NJ) and subcloned into our modified pComb3xSS vector.³⁵ These scFv mAbs were expressed in Top 10 F’ (Invitrogen) and purified using either Nickel column (Qiagen, CA) or anti-FLAG magnetic beads (Sigma, MO) according to manufacturers’ instructions. The information of the original pathogenic PV and anti-dsDNA mAbs and their revertant/germline mAbs is listed in Table I and Table II. The negative control anti-tetanus toxoid scFv mAb TT1 was described before.³⁵

TABLE I.

List of pathogenic PV mAbs and the corresponding revertant/germline mAbs

Original publication	Pathogenic mAb	PV revertant/germline mAbs
Payne et al.27	3c/9	3c/9-R
Yeh et al.28	F706	F706-R
Di Zenzo et al.25	PVA224	PVA224-R
	PVB28	PVB28-R
	PVB124	PVB124-R
Cho et al.26	F779	F779-R
	PVE4–8	PVE4–8-R
	4.2	4.2-R

TABLE II.

Germline mAbs of anti-dsDNA mAbs originating from SLE patients

Origin	SLE anti-dsDNA mAb	SLE germline mAb
Manheimer-Lory et al.29	I-2a	I-2a-R
Song et al.32	KSUG	KSUG-R
Paul et al.30	KS3	KS3-R
Winkler et al.31	32.B9	32.B9-R

Exogenous environmental antigens/allergens.

Representative environmental antigen extracts from five major groups of exogenous antigens/allergens were purchased from Greer Labs (Lenoir, NC).

ELISA.

The specificity of purified scFv mAbs and sera from patients and controls were tested by ELISA as previous described.^{16, 35} Briefly, for testing the reactivity of PV mAbs to exogenous antigens and Dsg3, Stripwell Microplates (Corning, MA) were coated with 30ng/well of each antigen extract from Greer Labs or purified human recombinant Dsg3, and incubated overnight at 4°C. All ELISA experiments were conducted using Tris buffer saline (TBS) system. Washing buffer contained 10mM TBS, 0.05% Tween-20, and 5mM CaCl₂. The blocking buffer consisted of washing buffer plus 1% BSA. All scFv mAbs, sera from patients and controls, secondary Abs were diluted in blocking buffer. The secondary Ab for scFv mAbs detection was HRP conjugated M2 anti-FLAG (Sigma). The bound serum antibodies were detected by HRP conjugated anti-human IgG1 (HP6001), IgG4 (HP6023), and IgM (SA-DA4) Abs, respectively (SouthernBiotech, Birmingham, AL). The plates were developed using the Pierce^™ TMB Substrate (Pierce, Rockford, lL).

Western blot, expression and purification of recombinant walnut proteins.

Detailed methods are provided in the Supplementary Appendix 1.

Statistical analysis.

Comparisons between two groups were done using Welch’s t-test to deal with unequal variances. Though the data tended to be positively skewed, transformation of the data was not attempted as the sample sizes were large enough for the test to be robust to non-normality.

Human subjects.

All human studies have been approved by the University of North Carolina at Chapel Hill institutional review board.

RESULTS

PV germline mAbs recognize walnut extract.

To identify the exogenous antigens that potentially incite autoAb production, we generated revertant/germline forms of eight pathogenic anti-Dsg3 mAbs derived from PV patients (Table I).^25–28 The H chain V region sequences of PVB124 and its germline counterpart (PVB124-R) are shown in Figure E1 as an example, where the nucleotide mutations that resulted in amino acid changes (marked in red in Figure E1) were reverted to their germline forms. We then screened for binding of these PV germline mAbs against a panel of 31 common environmental antigens including pollens, insects, animal epithelia, fungi and foods. To reduce the chance of false postive results, the scFv (single-chain fragment variable) mAbs were all tested at 0.25 ng/μl, a concentration much lower than 2–5 ng/μl, a range used in a study of PV scFv mAbs to determine whether they react to autoantigens from Hep-2 cells.²⁶ Agreeing with a previous report,²⁶ two of PV germline mAbs (PVE4–8-R and 4.2-R) preserved their reactivity to Dsg3 (Figure 1). In contrast, the other six PV germline mAbs did not significantly react with Dsg3 (Figure 1), suggesting that Dsg3 was not the inciting antigen for these pathogenic autoAb. Surprisingly, all PV germline mAbs showed strong reactivity to walnut extract (Figure 1). None of the PV germline mAbs demonstrated significant reactivity to any other antigen extracts except for mAb PVE4–8-R and 4.2-R, which showed weak reactivity to cockroach, mosquito, peanut and soybean extracts (Figure 1). No reactivity was observed for the negative control mAb TT1, which is specific for tetanus toxoid (Figure 1).³⁵ These findings suggest that walnut antigens may be the inciting antigens for the activation of naive pathogenic B cell precursors in PV patients.

FIG 1.

PV germline mAbs derived from eight pathogenic PV mAbs all recognized protein(s) from walnut extract. ELISA results of PV germline mAbs to all tested exogenous antigens are represented by heat map. TT1 mAb served as a negative control. The color scale of the ELISA O.D. values is shown on the right. The scale applies to all heat maps in this report.

To confirm that the binding of walnut antigen(s) by PV germline mAbs was specific, we generated revertant/germline mAbs from four well-characterized anti-double stranded (ds)DNA mAbs derived from SLE patients (Table II).^29–32 We then tested the reactivity of these SLE germline mAbs against the same panel of exogenous antigens as shown in Figure 1. The SLE germline mAbs did not react to any of the antigen extracts (Figure 2), indicating that recognition of walnut antigen(s) was specific to PV germline mAbs.

FIG 2.

In contrast to PV germline mAbs derived from PV patients, germline form mAbs of four anti-dsDNA mAbs from SLE patients did not recognize protein(s) from walnut extract and other tested exogenous antigens. TT1 and a PV germline mAb (PVE4–8-R) were included as negative and positive controls, respectively.

PV germline mAbs recognize Jug r 2 and methyltransferase-like protein in walnut extract.

There are four major allergenic components from walnut (Jug r 1–4).³⁶ To identify the walnut protein component(s) recognized by PV germline mAbs, we produced recombinant proteins of these four components. Jug r 4 is post-translationally cleaved, and we expressed the two fragments separately (Jug r 4N and Jug r 4C). The reactivity of PV germline mAbs to these five recombinant proteins was tested by ELISA (Figure 3A). All PV germline mAbs strongly reacted to Jug r 2, but only weakly reacted to Jug r 1, Jug r 3, and Jug r 4. Western blot analysis also verified binding of PV germline mAbs to recombinant Jug r 2, but not to the other recombinant walnut allergens (Figure 3B). We were not able to detect binding of PV germline mAbs to the 47kD Jug r 2 isoform in walnut extract (Figure 3C), possibly due to proteolysis during extract production or storage.³⁷

FIG 3.

ELISA results showing that all PV germline mAbs derived from pathogenic mAbs recognized Jug r 2 and methyltransferase-like proteins (ML) from walnut. A, All eight germline PV mAbs strongly reacted to recombinant Jug r 2, but not other known walnut components. PVE4–8-R and 4.2-R also reacted more strongly to hDsg3 than the other germline PV mAbs. B, The specificity of these mAbs for recombinant Jug r 2 was verified by Western blot. Representative Western blot results of two PV germline mAbs (3c/9-R and PVB124-R) are shown. These recombinant proteins were also probed with anti-His mAb as positive controls (bottom). C, The PV germline mAbs recognized a ~85kD protein component from walnut extract. Western blot results of two representative PV germline mAbs (3c/9-R and PVB124-R) and negative control mAb (TT1) are shown. Mass spectrometry identified this protein as an uncharacterized ML. D, All PV germline mAbs reacted to recombinant ML in ELISA (upper panel) and Western blot (lower panel).

As shown in the western blot of walnut extract, PV germline mAbs also recognized a dominant ~85kD protein band (Figure 3C). There is no known walnut allergenic component that matches the molecular weight of the ~85kD protein. Therefore, we subjected this protein to mass spectrometry and identified it as a methyltransferase-like protein (ML) (NCBI Reference Sequence: XP_018837618.1). This protein has not been previously identified as a walnut allergen, but an ortholog has been described as an allergen (Sal k 3) in Russian Thistle pollen.³⁸ To confirm the reactivity of PV germline mAbs to this protein, we generated recombinant ML protein and assessed the binding of PV germline mAbs. Both ELISA and Western blot results indicated PV germline mAbs, but not control mAb TT1, recognized recombinant ML protein (Figure 3D).

The apparent recognition of both Jug r 2 and ML by PV germline mAbs suggests that the PV germline mAbs may cross-react with these proteins. To investigate the potential cross-reactivity of PV germline mAbs, we conducted competition assays using recombinant Jug r 2 and ML. Recombinant Jug r 2 inhibited the binding of PV germline mAbs to ML and walnut extract in a dose dependent manner (Figure 4A). Similarly, ML also inhibited the binding of PV germline mAbs to Jug r 2 and walnut extract (Figure 4B). Thus, PV germline mAbs cross-reacted with both Jug r 2 and ML in walnut extract. These two molecules are not homologs and sequence analysis found no apparent local similarity between them. In addition, there is no primary sequence similarity among Dsg3, Jug r 2 and ML. However, computational analysis of the crystal structure of Dsg3 identified at least one conformational epitope with similarity to linear epitopes on Jug r 2 and ML (data not shown). Further characterization of potential cross-reactive epitopes by epitope mapping studies is currently underway.

FIG 4.

Dsg3-recognizing PV germline mAbs cross-reacted to walnut proteins. A, Jug r 2 inhibited the binding of PV revertant/germline mAbs to immobilized ML and walnut extract. B, ML inhibited the binding of PV revertant/germline mAbs to immobilized Jug r 2 and walnut extract. C, Jug r 2 (left panel) and ML (middle panel), but not Jug r 4N (right panel), inhibited the binding of 4.2-R and PVE4–8-R to Dsg3 in a dose dependent manner. Error bars represent SD.

Most of the PV germline mAbs had weak reactivity to Dsg3 autoantigen, but two germline mAbs, namely 4.2-R and PVE4–8-R, retained their affinity for Dsg3 (Figure 1).²⁶ It is possible that the B cell precursors for these pathogenic PV mAbs were cross-reactive to both endogenous and exogenous antigens. We therefore determined whether Jug r 2 or ML inhibited binding of 4.2-R and PVE4–8-R to Dsg3. We found that both recombinant Jug r 2 (Figure 4C, left panel) and ML (Figure 4C, middle panel), but not Jug r 4N (Figure 4C, right panel), inhibited binding of germline PV mAbs to Dsg3 in a dose-dependent manner (Figure 4C). These findings suggest that some pathogenic autoAbs in PV patients may originate from naïve B cells that cross-react with both walnut antigens and Dsg3.

Pathogenic PV mAbs recognize walnut antigens, but with reduced affinity compared to Dsg3.

The apparent cross-reactivity of 4.2-R and PVE4–8-R to both Dsg3 and walnut antigens suggests that Dsg3 and walnut antigens may share cross-reactive epitopes. We assessed the cross-reactivity of all the pathogenic PV mAbs to Jug r 2 and ML. As expected, all pathogenic mAbs showed strong reactivity to Dsg3^25–28 (Figure 5A). Furthermore, pathogenic PV mAbs also recognized Jug r 2 and ML, but not Jug r 3 (Figure 5A). We next compared the affinities of pathogenic-revertant mAb pairs for walnut antigens. We found that all 8 pathogenic PV mAbs had higher binding to Dsg3 than their germline counterparts as reported previously^{18, 19} (represented by 3c/9, 4.2, and F779 pairs; Figure 5B, lower panel), but 6 of them had comparable (represented by F779 pair) or reduced (3c/9 and 4.2 pairs) binding to Jug r 2 compared to their germline counterparts (Figure 5B, upper panel). Affinity titration of these mAb pairs for Jug r 2 confirmed the binding results (Figure 5C). These findings indicate that antigen selection during affinity maturation (a process by which antibodies are produced with increased antigen-binding affinity during the course of an immune response)³⁹ increases reactivity to Dsg3 but not to walnut antigens.

FIG 5.

Pathogenic PV mAbs’ reactivity to walnut antigens. A, Pathogenic PV mAbs also recognized Jug r 2 and ML, but not Jug r 3, the control walnut protein. B, While they had increased reactivity to hDsg3 compared to the corresponding germline mAbs (lower panel), two of mAbs had decreased reactivity to Jug r 2 (3c/9 and 4.2) (upper panel). C, Compared to germline PV mAbs (open circles), their corresponding mutant PV mAbs (closed circles) had either lower affinity (3c/9 and 4.2, upper and middle panels) or similar affinities for Jug r 2 (represented by F779, lower panel).

Antigen selection of pathogenic B cells in patients with PV is driven by Dsg3 rather than walnut antigen.

As demonstrated in Figure 5A, all pathogenic PV mAbs cross-reacted with Dsg3 and walnut antigens. However, pathogenic mAbs had higher affinity for Dsg3, and germline mAbs generally had higher affinity for walnut antigens (Figure 5B and 5C). These findings suggest that while walnut may be the initial antigen recognized by autoreactive B cells, the affinity maturation of pathogenic autoAb is driven by Dsg3. In such a case, one would expect that as anti-Dsg3 antibody titers rise in PV patients, walnut-specific antibody levels remain comparable to those in controls. To investigate this, we measured Dsg3- and walnut-specific IgG1 and IgG4 levels in PV patients and healthy controls. As expected, PV patients had significantly higher anti-Dsg3 IgG1 and IgG4 levels than healthy controls (HC) (Figure 6A). In contrast, walnut-specific IgG1 and IgG4 were found at low levels in sera from PV patients, and were comparable to levels found in HC (Figure 6A). We further tested the IgG1 and IgG4 levels of PV patients against Jug r 2 and ML in comparison to HC. As shown in Figure E2, the IgG1 (left panel) and IgG4 (right panel) Abs of PV patients against Jug r 2 and ML were not significantly different from those of HC. Additionally, the levels of both IgG1 and IgG4 Abs against Jug r 2 and ML were significantly lower than those against Dsg3 among PV patients (Figure 6B), suggesting that anti-Dsg3 IgG1 and IgG4 autoAbs in PV patients did not cross-react to these two walnut antigens. Also, we measured the correlation between anti-Dsg3 and anti-walnut Abs in PV patients. There was no correlation between the levels of anti-Dsg3 IgG1 and either anti-Jug r 2 IgG1 (r=0.2352, p=0.1076) or anti-ML IgG1 (r=0.1457, p=0.3232). There was a weak correlation between anti-Dsg3 and anti-Jug r 2 IgG4 (r=0.3174, p=0.028), but not between anti-Dsg3 and anti-ML IgG4 (r=0.1161, p=0.4319). Overall, these findings suggest that anti-Dsg3 IgG1 and IgG4 Abs from PV patients do not significantly cross-react with walnut antigens. To determine if recent or ongoing exposure to walnut antigen was possibly driving autoAb production in PV patients, we measured anti-walnut IgM levels in serum. PV patients had lower levels of anti-walnut IgM compared to HC (Figure 6C), indicating that acute exposure was unlikely to be contributing to autoAb production. Thus, while walnut antigens may initially activate anti-Dsg3 autoAb precursor B cells, they do not appear to participate in driving the autoAb response in PV patients, which is consistent with our results using PV pathogenic mAbs (Figure 5B and C). These data suggest that affinity maturation (likely driven by Dsg3) may shift the cross-reactivity of B cell precursors (against both foreign and self-antigens) away from walnut antigens during PV disease progression.

FIG 6.

PV patients (n=48) had distinct levels of antibodies against Dsg3 and walnut antigens. A, PV patients had significantly higher levels of IgG1 (left panel) and IgG4 (right panel) antibodies against Dsg3 than healthy controls (HC, n=48), but the levels of antibodies against walnut extract were similar in both groups. B, Among PV patients, anti-Dsg3 IgG1 (left panel) and IgG4 (right panel) levels were significantly higher (**, p<0.001) than either anti-Jug r 2 or anti-ML IgG1 and IgG4 levels. C, PV patients (n=48) had significant lower levels of IgM anti-walnut than that of healthy controls (HC, n=48).

DISCUSSION

In this investigation, by studying the specificity of eight revertant/germline mAbs derived from pathogenic PV mAbs, we showed that all germline mAb recognize two protein components, Jug r 2 and ML, from walnut (Figure 3). This provides evidence that walnut could be an environmental antigen that triggers the development of anti-Dsg3 pathogenic autoAbs and ensuing PV among genetically susceptible individuals. However, PV sera have very low levels of Abs against walnut extract (Figure 6), suggesting that the initial immune response to walnut antigen may have diminished by the time of disease onset. This is similar to the “hit-and-run” theory proposed for autoimmune diseases triggered by infectious microbes.^5–8

It is well known that infectious agents are usually implicated in the development of autoimmune diseases,^{8, 40–45} but no infectious antigen has been positively identified as a risk factor in pemphigus thus far.⁴⁶ Prompted by our previous finding that an indigenous non-infectious environmental antigen may trigger the autoAb response in endemic pemphigus,^{9, 16, 17,35} we sought to understand whether non-infectious environmental antigens may play a role in triggering the autoAb development in non-endemic PV using the revertant/germline mAb approach. This approach has been previously utilized to assess whether pathogenic mAbs originate from autoreactive naïve B cells in SLE and PV.^{20, 24–26} Interestingly, binding to Dsg3 was lost^{25, 26} or reduced²⁶ when pathogenic autoAbs were reverted to their germline forms, suggesting that anti-Dsg3 autoAb might be derived from non-autoreactive B cell precursors.²⁵

However, the antigen targets for germline PV mAb were not further explored. In the current study, we found that all tested germline PV mAbs recognized walnut antigens (Jug r 2 and ML), regardless of their varied reactivity to Dsg3 autoantigen (Figure 3).^{25, 26} This suggests that walnut proteins may be exogenous antigens that activate the naïve B cells in individuals genetically predisposed to PV, leading to the subsequent autoAb development and disease onset.

Our findings also suggest that walnut antigens may not be involved in driving the pathogenic autoAb development during PV disease progression. In fact, the IgG1 and IgG4 Abs against walnut antigens are present at much lower levels than those against Dsg3 in PV sera (Figure 6B). The autoreactive B cells in PV may evolve from walnut-specific naïve B cells that are also cross-reactive with Dsg3. During the early stages of PV progression, selection of cross-reactive B cells by endogenous Dsg3 would result in increased specificity for Dsg3 with a resultant decrease in specificity to walnut. It is likely that walnut-specific IgM is lower because walnut-specific IgM is converted into Dsg3-specific IgM or IgG. This scenario is also supported by our results where two pathogenic mAbs’ affinities for Jug r 2 were weaker than those of their PV germline counterparts (Figure 5B and 5C). We propose that precursors of PV autoreactive B cells are initially activated (“hit”) by walnut antigens. However, the subsequent selection of autoreactive B cells is driven by Dsg3, and thus walnut-specific immune responses eventually fade (“run”) as autoAb levels rise and PV develops. However, “run’ does not necessary mean that PV patients lose the reactivity to walnut antigens. It only means that PV patients did not show significantly higher reactivity to walnut antigens than normal individuals.

Whereas it seems unusual that the intake of a dietary walnut antigen via gastrointestinal epithelial cells may incite autoimmune responses to skin, similar cases can be found. For example, a well-known small intestine autoimmune disorder, celiac disease,⁴⁷ and the associated autoimmune skin manifestation dermatitis herpetiformis^{48, 49} are caused by the ingestion of the food antigen gluten. PV differs from these two disease conditions in the development of pathogenic autoAbs as constant exposure to walnut may not be required for the anti-Dsg3 autoAb evolution once the anti-Dsg3 autoAb precursor naïve B cells are primed. While case reports suggest that some dietary factors, including plant-derived thiols and tannins, may impact PV development⁵⁰, large comprehensive studies investigating the role of diet in PV are lacking. Accordingly, we are not aware of any data regarding walnut consumption in PV patients. Jug r 2 from walnut is a well-known allergen and thus our finding may suggest that food or other exogenous allergens can also be triggers for autoimmune diseases, in addition to allergic diseases. Moreover, the seemingly unrelated allergy and autoimmunity may have communications between the processes of their development.⁵¹ For example, allergic and autoimmune conditions can coincide in some patients, implicating that the two abnormalities share a common pathophysiologic mechanism.⁵¹ Investigations have demonstrated an association between allergic and autoimmune diseases because of the intrinsic dysfunction of the immune system.^{52, 53}

There are potential limitations in our study. All eight pathogenic mAbs were generated from peripheral blood samples of PV patients and may not well represent the circulating disease-inducing pathogenic Abs in sera of pemphigus patients because of sampling bias.^{26, 54} In addition, autoAb-associated B cells or plasma cells are more abundant in lesional or perilesional skin in pemphigus.⁵⁵ Further studies using both pathogenic and revertant/germline mAbs generated from these tissues may better represent the PV-associated autoAbs in the circulation of pemphigus patients.

In conclusion, we have discovered that walnut antigens may incite autoAb development in persons genetically predisposed to PV. Our investigation demonstrates that the revertant/germline mAbs can be an effective tool to identify exogenous antigens that trigger autoAb production in PV and other autoimmune diseases. Our results also suggest that walnut components may not only be a risk allergen for allergic individuals, but may also trigger an autoimmune disease. PV patients or individuals at risk may benefit from this finding by limiting the exposure to PV-associated environmental antigens to evade possible disease development or relapse.

Key Messages.

• Common food antigens/allergens not only induce allergic diseases, but may also trigger the development of autoimmune diseases in genetically susceptible individuals.

• Revertant/germline monoclonal antibodies are a unique tool to identify environmental antigens/allergens for the initiation of allergic and/or autoimmune diseases.

Abbreviations

PV

Pemphigus vulgaris

mAb

Monoclonal antibody

AutoAb

Autoantibody

Dsg

Desmoglein

SLE

Systemic lupus erythematosus

ML

Methyltransferase like protein

HC

Healthy control