Identification of a primary antigenic target of epitope spreading in endemic pemphigus foliaceus

Abstract

Epitope spreading is an important mechanism for the development of autoantibodies (autoAbs) in autoimmune diseases. The study of epitope spreading in human autoimmune diseases is limited due to the major challenge of identifying the initial/primary target epitopes on autoantigens in autoimmune diseases. We have been studying the development of autoAbs in an endemic human autoimmune disease, Brazilian pemphigus foliaceus (or Fogo Selvagem (FS)). Our previous findings demonstrated that patients before (i.e. preclinical) and at the onset of FS have antibody (Ab) responses against other keratinocyte adhesion molecules in addition to the main target autoantigen of FS, desmoglein 1 (Dsg1), and anti-Dsg1 monoclonal Abs (mAbs) cross-reacted with an environmental antigen LJM11, a sand fly saliva protein. Since sand fly is prevalent in FS endemic regions, individuals in these regions could develop Abs against LJM11. The anti-LJM11 Abs could recognize different epitopes on LJM11, including an epitope that shares the structure similarity with an epitope on Dsg1 autoantigen. Thus, Ab response against this epitope on LJM11 could be the initial autoAb response detected in individuals in FS endemic regions, including those who eventually developed FS. Accordingly, this LJM11 and Dsg1 cross-reactive epitope on Dsg1 could be the primary target of the autoimmune response in FS. This investigation aimed to determine whether the autoAb responses against keratinocyte adhesion molecules are linked and originate from the immune response to LJM11. The anti-Dsg1 mAbs from preclinical FS and FS individuals were employed to determine their specificity or cross-reactivity to LJM11 and keratinocyte adhesion molecules. The cross-reactive epitopes on autoantigens were mapped. Our results indicate that all tested mAbs cross-reacted with LJM11 and keratinocyte adhesion molecules, and we identified an epitope on these keratinocyte adhesion molecules which is mimicked by LJM11. Thus, the cross-reactivity could be the mechanism by which the immune response against an environmental antigen triggers the initial autoAb responses. Epitope spreading leads to the pathogenic autoAb development and ensuing FS among genetically susceptible individuals.

1. Introduction

Epitope spreading (ES) is an important mechanism for the diversification of immune responses to increase the efficiency of protective functions of a host immune system against non-self antigens [1–3]. However, the immune responses induced by the primary epitopes on antigenic targets, such as viruses, may expand to include epitopes on autoantigens during the inflammatory process [3]. In this regard, ES may trigger the autoimmune responses that lead to the autoimmune diseases among genetically susceptible individuals [1, 4]. ES and its pathogenic role have been documented in animal models of both T cell and B cell mediated autoimmune diseases [3, 5] since it was first reported using experimental allergic encephalomyelitis (EAE) mouse model [6]. By contrast, investigations on etiological and pathogenic roles of ES in human autoimmune diseases are limited, mainly because ES may take place before the time of diagnosis [1] and ES experiments cannot be conducted in humans to provide direct step-wise evidence of ES [4]. It is also difficult to identify the primary target epitope on an autoantigen or determine when the initial autoimmune response is started in a human autoimmune disease [1, 3, 7]. Using a human endemic pemphigus foliaceus (PF), also known as Fogo Selvagem (FS), as a model, it has been demonstrated that serum antibodies (Abs) from patients before the onset of FS (pre-FS) and after the onset of FS respectively recognize mainly C-terminal and N-terminal domains on FS autoantigen, desmoglein (Dsg) 1 [8]. This is an important example of ES in human autoimmune diseases. But which epitope or domain is the primary target of autoimmune response in FS is yet to be determined.

The development of human autoimmune diseases is one of the fundamental enigmas of immunology, with many genetic and environmental factors contributing to the etiology and pathogenesis [9–12]. Groups of genes, such as MHC and non-MHC have been identified to contribute to genetic susceptibility to autoimmune diseases [13]. There is also evidence that environmental risk factors also play a role in the development of autoimmunity. For example, the concordance of autoimmune diseases in monozygotic twins is 12–67% [14], and is as low as 24% for systemic lupus erythematosus (SLE) [15], emphasizing the environmental factors in the etiology of autoimmune diseases. However, the inciting environmental antigens remain elusive in the majority of human autoimmune diseases. Infectious agents have been suggested to be linked to the development of autoimmune diseases [16–22]. Autoimmune skin diseases, such as pemphigus, have not been connected to infectious antigen in terms of risk factors [23]. The main question about the cause of pemphigus is whether the development of autoreactive B cells is triggered by endogenous or exogenous antigens. Though no autoimmune disease has positively linked to any relatively rare infection [12], the relevant role of exogenous antigens in the development of the human autoimmune diseases is known in several diseases, such as rheumatoid arthritis (RA) and FS. In RA, citrullinated autoantibodies (autoAbs) cross-react with synovial and bacterial gut antigens [24–26]. In FS, non-infectious environmental antigens may play a significant role in its etiology, as autoAbs from FS patients, pre-FS, and normal controls from FS endemic regions, cross-react with a sand fly salivary gland antigen LJM11 and keratinocyte adhesion molecule Dsg1 [27–29]. Insect (wasp) bites were also reported to lead to an autoimmune neurological condition, neuromyotonia [30].

The epidermal integrity of the skin is maintained by four desmogleins (Dsg1, Dsg2, Dsg3 and Dsg4) and three desmocollins (Dsc) (Dsc1, Dsc2 and Dsc3), as well as E-cadherin (Ecad) which is structurally and functionally similar to desmosomal cadherins [31–34]. We have reported that FS patients and the FS endemic control individuals have autoAbs against all these keratinocyte adhesion molecules [35, 36]. The significantly elevated levels of autoAbs among normal individuals in FS endemic regions compared to normal control individuals living in US [36, 37] suggest the environmental difference between these two geographic regions and may explain at least in part, why FS is endemic in its endemic regions. The strong correlations between the autoAbs against different keratinocyte adhesion molecules among FS endemic controls and FS patients [35, 36] may be due to the cross-reactivity of the same population of autoAbs among these individuals. The correlations are consistent with the fact that these cadherin molecules belong to the same desmosome superfamily and share approximately 30% sequence identity [38]. The presence of autoAbs among endemic control individuals also suggests chronic immune responses to a common environment antigen(s) present in FS endemic regions. Based on our previous investigations, LJM11 sand fly salivary gland antigen is one of these environmental antigens, as FS patients have Abs [27, 37] against LJM11 and anti-Dsg1 IgG4 mAbs from FS patients cross-react with LJM11 [28].

In an effort to find the mechanism by which immune responses to environmental antigens, such as LJM11 lead to the autoimmunity in FS, we have examined whether anti-cadherin autoAb development is linked to environmental antigens in FS endemic regions. We specifically wanted to answer the following questions. First, does the wide presence of autoAbs against different skin adhesion molecules among FS patients and FS endemic controls attribute to the cross-reactivity of the same population of autoAbs? Because of the polyclonal nature of the serum Abs, this question can only be definitively answered using mAbs from those individuals to determine whether a mAb cross-reacts to multiple antigens. From FS patients and a pre-FS individual, we generated and re-produced our published mAbs [28, 39] to examine their cross-reactivity to cadherin adhesion molecules. Secondly, do these autoAbs also cross-react with LJM11 exogenous antigen? If yes, with which epitope(s) on the cadherin adhesion molecules the autoAbs from FS and pre-FS individuals cross-react? Since cross-reactivity or molecular mimicry may be the first step in ES leading to the initiation of autoimmune responses [1], the cross-reactive epitope(s) could be the initial or primary target of autoreactive immune response among FS patients that links the autoAb development to exogenous antigen LJM11. These cross-reactive Abs initiate the autoimmune responses in individuals in FS endemic regions and direct the subsequent development of pathogenic autoAb responses through ES among genetically predisposed individuals, leading to the onset of FS. The answer to this question is of special interest in light of the finding that the immune response to LJM11 is developed prior to anti-Dsg1 response [37]. We mapped the epitopes on these adhesion molecules to identify the potential cross-reactive epitopes, and our results indicate that the mAbs from either FS patients or the pre-FS individual cross-react with all tested keratinocyte adhesion molecules and LJM11. Epitope characterization of these adhesion molecules revealed an epitope on ectodomain (EC) 2 of the cadherin molecules via which all tested mAbs recognize these adhesion molecules. This epitope resides between the pathogenic epitopes on Dsg1 that we identified lately [40]. These results strongly suggest that the development of autoAbs in individuals living in FS endemic regions is the result of the immune response to environmental antigens, such as LJM11, and ES is the main mechanism by which an exogenous antigen triggers the pathogenic autoimmune response in FS.

2. Materials and Methods

2.1. Serum samples and mAb generation from FS patients and a pre-FS individual.

Immortalized B cells and serum samples from FS patients and a pre-FS individual in this investigation have been described before [8, 27, 28, 39]. FS38 (female), FS45 and FS46 (males) developed FS when they were 39, 22, and 19 years old, respectively. These studies are approved by the institutional review board from the University of North Carolina at Chapel Hill. Six anti-Dsg1 IgG4 mAbs from FS patients generated previously using either hybridoma or Ab phage display (APD) method were selected for this investigation [28]. Two IgG1 mAbs (1C11 and 2G8) were generated from Epstein-Barr virus immortalized peripheral blood lymphocytes from FS patients using hybridoma method and selected against Dsg1 as described [39]. Two additional anti-Dsg1 IgG1 mAbs from FS patients, as well as three anti-Dsg1 IgG1 mAbs from a pre-FS individual, were selected from the pool of mAbs generated using hybridoma method before [39]. The genes of H and L chains of the mAbs generated using hybridoma method were synthesized (Genscript, NJ) as single chain variable fragment (scFv) and subcloned into the modified pComb3XSS vector as described [28]. All these anti-Dsg1 scFv mAbs (listed in Table 1) were produced accord to the method described [27, 28, 41] and purified by either Nickel column (Qiagen, CA) or anti-FLAG magnetic beads (Sigma, MO) according to manufacturers’ instructions.

Table 1.

List of mAbs from FS and pre-FS individuals.

Source	Subclass	Name	Method	VH gene	VL gene
FS	IgG1	1C11a	Hybridoma	IGHV1-18	IGKV1-33
		2G8a	Hybridoma	IGHV3-30-3	IGKV1-39
		FS7-2F8b	Hybridoma	IGHV3-7	IGKV1D-39
		FS33-2E6b	Hybridoma	IGHV3-11	IGKV3-20
	IgG4	4E4b	Hybridoma	IGHV3-48	IGKV1D-33
		2D11b	Hybridoma	IGHV2-5	IGLV6-57
		GCDS2c	APD	IGHV5-51	IGKV2-28
		GCDS5c	APD	IGHV3-9	IGKV3-20
		JLDO34c	APD	IGHV1-18	IGLV2-11
		JLDO-L5c	APD	IGHV1-69	IGLV2-8
Pre-FS	IgG1	Pre-3H9b	Hybridoma	IGHV4-39	IGKV4-1
		Pre-5D3b	Hybridoma	IGHV3-23	IGKV2D-28
		Pre-5G6b	Hybridoma	IGHV3-30-3	IGKV4-1

^aNewly generated mAbs.

^bThose mAbs were originally reported using hybridoma method [39}.

^cThose mAb were originally reported using APD method [28}.

2.2. Production and purification of human recombinant keratinocyte cadherins and sand fly proteins.

Human ectodomains of recombinant keratinocyte cadherins, including Dsg1, Dsg2, Dsg3, Dsg4, Dsc1, Dsc2, Dsc3, Ecad, sand fly LJM11 and LJL143 proteins were produced and purified as described [27, 36, 39, 42, 43].

2.3. ELISA and Western blot using mAbs.

The reactivity of the scFv mAbs were tested by ELISA as previous described [27, 28, 44]. Negative control scFv (TT1) was described before [28]. HRP conjugated M2 anti-FLAG mAb (Sigma, MO) was used as secondary Ab for detection of the binding of purified scFv mAbs to different antigens by either ELISA or Western blot. The final concentrations of all scFv mAbs were 0.25 ng/μl. Sera from FS and pre-FS individuals were diluted 1:100 and IgG1 and IgG4 Ab levels were determined using HRP conjugated anti-human IgG1 (HP6001) and IgG4 (HP6023) secondary mAbs (SouthernBiotech, AL). For Western blot, equal amounts of all cadherin proteins were resolved by electrophoresis through 10% SDS-PAGE gels and transferred to NC membrane. The membrane was incubated with scFv mAbs, followed by M2 anti-FLAG-HRP (Sigma, MO).

2.4. Epitope characterization

Conserved fragments among cadherin proteins were identified by aligning the amino acid sequences of these molecules using Vector NTI (Invitrogen) (Supplementary Fig 1). The peptides corresponding to the homologous regions were synthesized (Genscript, NJ). Competition assays were conducted using these peptides as inhibitors to determine whether a peptide blocks the binding of a mAb to a cadherin protein or LJM11. To test the direct binding of the mAbs to different peptides, we synthesized N-terminal biotinylated peptides (Genscript, NJ) with a linker (GSGSGSGS) between biotin and each peptide. The peptides were solubilized according to manufacturer’s recommendations and diluted in ELISA buffer (Tris buffered saline with 0.05% Tween-20 and 1% BSA). The plates (96-well half-area, Corning) were coated with 20 ng/well of each tested mAbs and incubated overnight. After blocking (1% BSA in Tris buffered saline) and washing, 50 μl of each biotin labeled peptide solution (10 ng/μl) was added to mAb coated plates and incubated for an hour at room temperature. After washing, HRP conjugated streptavidin (SouthernBiotech, AL) was added to each well and incubated for half an hour. After wash, TMB (ThermoFisher) substrate was added to each well and O.D. values determined.

2.5. Molecular modeling of human Dsg1

Molecular modeling of human Dsg1 was performed at the R. L. Juliano Structural Bioinformatics Core Facility at the University of North Carolina at Chapel Hill. Structural templates were identified using the HHpred Fold Recognition Server [45, 46] and the extracellular domains of the Dsg1 model were built using the MODELLER software package [47]. The structural templates used for modeling were: 1) PDB ID 3Q2W (mouse N-cadherin ectodomain), 2) PDB ID 3Q2V [mouse E-cad ecotodomain [48], 3) PDB ID 5EQX (human Dsg3), and 4) PDB ID 5IRY [human Dsc1 (PUBMED: 27298358) (Supplementary Fig 1 and 2).

2.6. Statistical analysis.

For each Ab subclass examined, the difference between a mAb’s reactivity towards an antigen of interest and that towards a control was analyzed using t-test. As all pairwise comparisons were planned in advance, no multiple-comparison correction was used.

3. Results

3.1. The autoreactive mAbs in FS patients and a pre-FS individual are cross-reactive to multiple cadherin molecules.

We have reported that most FS patients have high levels of IgG1 and IgG4 autoAbs against skin adhesion molecules, specifically keratinocyte cadherins among FS patients, and strong correlations between the levels of the autoAbs to each cadherin autoantigens [36]. These correlations raise the possibility that these autoAbs belong to the same population of Abs that cross-react to these different adhesion molecules. To test this possibility, we determined the cross-reactivity of newly generated anti-Dsg1 mAbs and a panel of selected anti-Dsg1 mAbs generated in our laboratory before [28, 39]. All mAbs used in this investigation and their relevant information are listed in Table 1.

In FS the main pathogenic autoAbs are of IgG4 subclass [49] and we first tested the specificity of the selected FS IgG4 mAbs. As shown in Fig 1A and Supplementary Fig 3, all tested IgG4 mAbs cross-reacted with all examined cadherin adhesion molecules except for negative control mAb TT1, which was specific for tetanus toxoid [28]. Similarly, all tested FS IgG1 mAbs also recognized these cadherin proteins (Fig 1B and Supplementary Fig 4). These results suggest that the wide presence of IgG1 and IgG4 autoAbs against multiple cadherin proteins observed [36] could be due to the cross-reactivity of the same population of autoAbs in FS patients. Since normal individuals in FS endemic regions also have IgG1 autoAbs against multiple cadherins other than Dsg1 [35, 36], pre-FS individuals (can be considered as normal control as they have no FS by the time of their blood draw) should also have autoAbs that cross-react with multiple cadherins. The anti-Dsg1 mAbs from a pre-FS individual [29] were tested for their cross-reactivity. As shown in Fig 1C and Supplementary Fig 5, these mAbs also recognized cadherin proteins other than Dsg1. All the above examined reactivity towards cadherins was significant compared that towards the control (p < 0.01). Western blot results of these mAbs from FS or pre-FS to cadherins confirmed the reactivity of these mAbs (Fig 1D). Western blot results also suggest that most mAbs react to the cadherin proteins via a linear epitope(s). Two tested IgG4 mAb did not recognize or only weakly recognize these cadherins (Fig 1D), suggesting that these two mAbs do not bind linear epitope on Dsg1. As has been shown for E4E mAb [27], they are likely bind to Dsg1 via a conformational epitope which comprises part of the linear epitope and other discontinuous amino acid residues on Dsg1.

Fig 1.

The cross-reactivity of anti-Dsg1 IgG mAbs from FS and pre-FS individuals with other cadherins. A. Anti-Dsg1 IgG4 mAbs from FS patients react with all tested cadherin adhesion molecules. B. Anti-Dsg1 IgG1 mAbs cross-react with cadherins in addition to Dsg1. C. Anti-Dsg1 IgG1 mAbs from a pre-FS individual also cross-react with cadherins. D. Western blot results indicate that most anti-Dsg1 IgG mAbs from FS and pre-FS individuals react with tested Dsg and Dsc molecules. Two IgG4 mAbs (4E4 and JLDO-L5) do not (or only weakly) react to cadherins.

3.2. Cadherin protein-recognizing mAbs also cross-react with environmental antigen LJM11.

We have reported that an environmental antigen, sand fly salivary gland component LJM11, could be an inciting antigen associated with the development of autoAb responses because FS IgG4 anti-Dsg1 mAbs cross-react with LJM11 antigen [27, 28]. In addition, the higher levels of IgE anti-LJM11 compared to anti-Dsg1 among pre-FS individuals suggest that LJM11 environmental antigen could be the initial trigger for the subsequent development of anti-Dsg1 autoAbs [37]. We then tested the possibility that these cadherin-cross-reactive mAbs also react to LJM11. The LJM11 reactivity of the representative mAbs is shown in Fig 2A. All tested IgG1 mAbs from FS patients (left panel) and a pre-FS individual (right panel), but not control mAb TT1, cross-reacted with LJM11, in much the same way as IgG4 mAbs (GCDS2 and JLDO34, Fig 2A left panel) from FS patients [28]. All the reactivity was significant compared to the control (p < 0.01). We also tested the reactivity of these mAbs to an irrelevant antigen (tetanus toxoid) and another sand fly salivary gland protein LJL143 [43]. As shown in Fig 2B, it is apparent that these FS associated mAbs did not react to either tetanus toxoid (left panel) or LJL143 (right panel). TT1 mAb only reacted to tetanus toxoid (Fig 2B, left panel). These results indicate that the binding of these tested mAbs to LJM11 is specific.

Fig 2.

Anti-Dsg1 FS and pre-FS mAbs react to LJM11 antigen, but not irrelevant antigen. A. IgG mAb from FS patients (left panel) and a pre-FS individual (right panel) all react to LJM11. Tetanus toxoid-specific mAb TT1 does not react to LJM11. B. The anti-Dsg1 mAbs from pre-FS and FS do not react with irrelevant tetanus toxoid antigen (except tetanus toxoid-specific mAb TT1, left panel) or another sand fly salivary gland antigen LJL143 (right panel).

3.3. Anti-Dsg1 mAbs recognize a cross-reactive epitope conserved among cadherin molecules.

These cadherin adhesion molecules belong to the cadherin family and share high sequence homology [38]. The alignments of each pair of these tested cadherins revealed that the similarities between these cadherin molecules are approximately between 30 to 50% (Supplementary Table 1). The fact that all tested mAbs from FS and pre-FS individuals reacted with keratinocyte adhesion molecules (Fig 1, Supplementary Fig 3, 4, and 5) suggests that these mAbs may bind to a highly conserved region(s) on these molecules. The alignment of the primary sequences of these molecules identified six highly conserved regions. Their relative positions on the model of Dsg1 are shown in Supplementary Fig 1. These amino acid residues are all exposed on the surface of the Dsg1 molecule and are directly accessible to Ab binding. Two of these peptides reside within the EC1 domain, three in EC2 and one in EC4, and their sequences are listed in Table 2. We conducted competition assay using six peptides that were designed and synthesized according to the homologous sequences on Dsg1, and tested whether any of these peptides inhibits the binding of the mAbs to Dsg1. The inhibition results indicated that only peptide Dsg1–3 significantly inhibited the binding of the mAbs to Dsg1 in a dose dependent manner (Fig 3A). These mAbs included IgG1, IgG4 mAbs from FS patients, and IgG1 mAbs from a pre-FS individual. The other five peptides did not inhibit the binding of these mAbs to Dsg1 (data not shown). We also tested the direct binding of the mAbs to these peptides. Consistent with the competition results, all tested FS and pre-FS mAbs reacted to Dsg1–3; the reactivity of Dsg1–3 was significantly higher than that of any other Dsg1 peptides and an irrelevant peptide (p<0.01) (Supplementary Fig 6). These FS and pre-FS mAbs also weakly reacted to Dsg1–5 and Dsg1–6 but were not significantly different from those to Dsg1–1, Dsg1–2, Dsg1–4 in binding (Supplementary Fig 6).

Table 2.

Dsg1 peptide sequences.

Name	Sequence	Domain
Dsg1-1	QKTGEINITSIVDREVT	EC1
Dsg1-2	VLDINDNPPVFSM	EC1
Dsg1-3	TLVMILNATDADEP	EC2
Dsg1-4	NFLDREQYGQYALAV	EC2
Dsg1-5	CNIKILDVNDNIPYM	EC2
Dsg1-6	RTCTGTINI	EC4

Fig 3.

Dsg1–3 peptide inhibits the binding of anti-Dsg1 mAbs to cadherins and LJM11 in a dose dependent manner. A. Dsg1–3 inhibits the binding of IgG4 (FS) and IgG1 (FS and pre-FS) mAbs to Dsg1. B. The reaction of representative mAbs from FS IgG4 (JLDO-L5) and a pre-FS IgG1 (Pre-5D3) to cadherins is blocked by Dsg1–3 peptide. C. The recognition of LJM11 by FS and pre-FS anti-Dsg1 mAb is also blocked by Dsg1–3 peptide.

The inhibition of the binding to Dsg1 by peptide Dsg1–3 and the recognition of Dsg1–3 by mAbs from FS and pre-FS individuals strongly suggest that peptide Dsg1–3 represents the epitope via which these mAbs bind to Dsg1. We then tested the ability of Dsg1–3 to block the binding of these mAbs to other cadherins. As shown in Fig 3B, Dsg1–3 inhibited the binding of the mAbs to Dsg3, Dsg4, and Dsc2 in a dose dependent manner and the inhibition was comparable to that for Dsg1. Dsg1–3 also inhibited the interactions between FS mAbs to Dsg2, Dsc1, and Dsc3 in a similar manner (data not shown). This further confirmed that amino acid residues on Dsg1 represented by peptide Dsg1–3 is the epitope that these FS mAbs recognize on these cadherins. This epitope is located between the pathogenic epitopes on Dsg1 that we recently identified [40] (Supplementary Fig 2).

3.4. The peptide representing cross-reactive epitope also blocks the binding of autoreactive mAbs to LJM11

Since these mAbs all cross-reacted with the FS environmental antigen LJM11, the epitope on Dsg1 that these mAbs recognize could be the cross-reactive epitope that shares its structural similarity with LJM11. Hence Dsg1–3 should also be able to block the binding of these mAbs to LJM11. As shown in Fig 3C, the binding of representative IgG4 and IgG1 mAbs from FS patients and a pre-FS individual was blocked by peptide Dsg1–3 in a dose dependent manner, confirming that this epitope on Dsg1 is the cross-reactive epitope that links the autoAbs in FS with the Abs against LJM11.

To determine whether the homologs of this epitope on other cadherin proteins are also the epitopes recognized by these mAbs, we designed two Dsg1–3 homologous peptides based on the corresponding amino acid sequences of Dsg4 and Dsc2. These two peptides were named Dsg4–3 (TLVVKLCATDADEE) and Dsc2–3 (TTVGQVCATDKDEP), respectively (Fig 4A). Both peptides also significantly inhibited the binding of representative mAbs from binding to Dsg1 (Fig 4B, upper panel) and LJM11 (Fig 4B, lower panel) in a dose dependent manner. The competitive effect of Dsg1–3 peptides was also validated using western blot where this peptide inhibited the binding of a FS (FS7–2F8) and a pre-FS (Pre-3H9) mAbs to Dsg1 and Dsg3 in a dose dependent manner (Fig 4C). The data demonstrate that mAbs from FS patients and a pre-FS individual cross-react to multiple skin adhesion molecules via the epitope represented by Dsg1–3. The peptide competition and western blot results suggest that this epitope is linear on Dsg1. These tested mAbs do not recognize LJM11 in western blot, which is consistent with our previous report that these mAbs bind to a conformational epitope on LJM11 [27]. Sequence analysis could not identify any homologous region between LJM11 and any cadherins.

Fig 4.

Homologs of Dsg1–3 peptide also inhibit the bindings of anti-Dsg1 mAbs to cadherins and LJM11. A. The sequences of two cadherin homologs (Dsg4–3 and Dsc2–3) of Dsg1–3 peptide. B. Like Dsg1–3 (left group in each plot), Dsg4–3 (middle group) or Dsc2–3 (right group) peptide also competes with selected mAbs’ binding to Dsg1 (upper panel) or LJM11 (lower panel) in a dose dependent manner. C. Dsg1–3 peptide inhibits the binding of FS7–2F8 (left panel) or Pre-3H9 (right panel) mAb to Dsg1 (top panel) or Dsg3 (lower panel) in a dose dependent manner in western blot.

3.5. The parallel IgG1 and divergent IgG4 responses against Dsg1 and LJM11

Our previous investigation found that the IgE response to LJM11 occurs before that to Dsg1 among pre-FS individuals [37], suggesting that Ab response to environmental antigen LJM11 may trigger the autoimmune response among the pre-FS individuals. The finding that autoreactive mAbs from the pre-FS patients all cross-reacted to LJM11 prompted us to determine whether the development of serum Abs before and at the onset of FS is in agreement with the findings from pre-FS and FS mAbs. We then analyzed the levels of anti-Dsg1 and anti-LJM11 IgG1 and IgG4 Abs in FS patients from whom we had collected serum samples at different time points before and at the onset of FS. The Ab levels from these individuals could reveal the progression of the specificity and isotypes/subclasses of the Abs over time. The IgG1 and IgG4 Ab levels against either Dsg1 or LJM11 from three representative individuals (FS38, FS 45 and FS46) are shown in Fig 5. Anti-Dsg1 IgG1 levels decreased over time (blue lines), while anti-Dsg1 IgG4 levels increased, reaching their highest levels at FS onset (red lines). The trend of the anti-LJM11 IgG1 levels (green lines) is similar to that of the anti-Dsg1 IgG1 levels. The levels of anti-LJM11 IgG4 (teal lines) slowly declined over time, suggesting the divergent development of the IgG4 Abs against Dsg1 and LJM11 antigens. The divergence of IgG4 responses to Dsg1 and LJM11 also implies that the epitopes recognized by IgG4 anti-Dsg1 pathogenic autoAbs at FS onset were not the cross-reactive epitope on Dsg1 that was identified by cross-reactive mAbs in FS and pre-FS individuals (Fig 1), which is another evidence of ES in IgG4 autoAb development in FS. Since the Abs were assayed separately against two different antigens, the O.D. values do not reflect the quantitative differences between the levels of anti-Dsg1 and anti-LJM11 Abs. However, the similar trends of the anti-Dsg1 and anti-LJM11 IgG1 Abs over time suggest a developmental association between these two Ab responses.

Fig 5.

The development of IgG1 and IgG4 anti-Dsg1 and anti-LJM11 Abs in pre-FS individuals before and at FS onset. The year of each individual’s FS onset is indicated in red.

4. Discussion

Due to the heterogeneity of autoimmune diseases, it is difficult to assess the environmental factors that contribute to the development of a given autoimmune disease. Autoimmune diseases are rarely endemic in particular, and it is difficult to assess their etiological commonality and further dissect their causes. Using an endemic autoimmune skin disease model (FS), our previous studies have found that sand fly salivary gland antigen LJM11 may trigger the development of FS. This report detailed the identification of the initial or primary target of the autoimmune response on Dsg1 autoantigen in FS. This finding represents a major step towards the understanding of the mechanism by which an environmental antigen, such as LJM11, may trigger the development of an autoimmune disease. It also provides further evidence that immune response to environmental risk antigens, such as LJM11, initiates the autoimmune response in FS. The findings that all tested autoreactive mAbs from either pre-FS or FS individuals cross-reacted to LJM11 and these mAbs also recognized other cadherin molecules have enabled us to identify the cross-reactive epitope on EC2 of Dsg1 molecule via the homologous sequence fragments shared among these cadherins. The epitope is most likely to be the primary antigenic target of autoimmune responses in FS based on our previous finding that IgE immune response to LJM11 is developed prior to anti-Dsg1 [37]. The development of autoAbs in individuals is likely the result of ES from the cross-reactive epitope on LJM11 to the epitope on autoantigens which structurally mimic that on exogenous antigens. The Abs directed against cross-reactive epitope on Dsg1 represent the initial wave of autoAb responses in pre-FS individuals that eventually leads to the development of pathogenic autoAbs via intramolecular ES (Fig 6 and see below).

Fig 6.

ES in the generation of anti-Dsg1 autoAb response. The initial immune responses to LJM11 leads to the production of Abs against the cross-reactive epitope (green) on LJM11. Molecular mimicry mediated intermolecular ES results in the initial autoimmune response against cross-reactive epitope on Dsg1. Subsequent ES on Dsg1 generates pathogenic anti-Dsg1 autoAbs against pathogenic epitopes (orange) on EC1 or EC2 domains of Dsg1 in FS patients, while autoAbs resulting from ES towards epitopes on EC3, EC4, and EC5 are not pathogenic.

ES in autoimmunity provides important clues for dissecting the disease origination and progression, and knowledge of the responsible epitopes on an autoantigens in a given autoimmune disease would aid the disease diagnosis and development of precision medical treatment [1, 2]. The mechanisms of ES in B cell mediated autoimmune diseases [2] and autoimmune skin diseases [1] have been thoroughly reviewed. Some ES phenomena were well studied in animal models, such as a mouse model of multiple sclerosis, EAE, in which the immune responses to the initial exogenous antigenic targets spread to other epitopes on autoantigens [1]. In Ab-mediated autoimmune disease SLE, immunization of mice with SmB peptide derived from SLE dominant Sm antigen also induced autoAbs reactive to other spliceosomal proteins [50]. In human, ES may be the most logical explanation for the clustering or association of different autoimmune diseases, such as in autoimmune skin diseases [1, 2, 4]. Most ES instances in human autoimmune skin diseases are found in the disease progression or transition [1, 4], such as in the case of bullous pemphigoid [51]. Another example is seen during the disease transition of pemphigus vulgaris (PV) from mucosal PV (mPV) to mucocutaneous PV (mcPV). Patients with mPV present with disease localized to the mucosal tissues and autoAbs to Dsg3, while patients with mcPV have disease affecting both the mucosal and cutaneous tissues and harbor autoAbs to both Dsg3 and Dsg1 [52, 53]. Most PV patients begin their clinical course with mPV [54–57], and the majority of patients will have disease progress to involve both mucosa and skin (mcPV). The transition from mPV to mcPV is marked by the development of autoAbs to Dsg1 in addition to Dsg3 [52, 53, 58, 59]. It is suggested that intramolecular and intermolecular ES underlines the mechanism for the acquisition of anti-Dsg1 autoAbs [1]. However, how the autoAb development is triggered or which epitope on Dsg1 is the primary target of the immune response in PV is still unclear.

In this investigation we have identified the primary target epitope of autoAb responses in endemic PF. Since most instances of ES of immune responses against exogenous antigens leading to the autoimmunity concerns autoimmunity induced by infectious agents, such as viruses [60], it reveals a mechanism by which ES of immune responses against non-infectious environmental antigens could incite the autoimmune responses in an autoimmune diseases. In FS, LJM11 triggered autoimmune responses obviously involve both physical association dependent and independent ES [1, 61] as shown in the model proposed in Fig 6. First involved is the Ab response against LJM11, which results from the prevalent presence of sand fly in FS endemic regions [62, 63]. The Abs directed against the cross-reactive epitope on LJM11 are the initial autoAbs responsible for responses against keratinocyte autoantigens, such as Dsg1 (Fig 6). The cross-reactivity mediated ES might not be considered as an “authentic ES”, but it is a first step leading to ES to other molecules, including autoantigens in autoimmunity [1]. Molecular mimicry or cross-reactivity mediated intermolecular ES in animal models is well established [1, 2], such as in B cell-mediated [64] or T cell-mediated [60] autoimmunity. Subsequently intramolecular ES on Dsg1 of autoimmune responses in pre-FS could move towards both N and C terminal of the Dsg1 (Fig 6, lower panel). The onset of FS is seen only when the ES spreads the autoAb responses to the pathogenic epitopes on EC1 and/or EC2 domains of Dsg1. It should be pointed out that LJM11 from sand fly may not be the only environmental antigen that could trigger the development of FS [29]. The insect bites have been found to be a major risk factor for FS development in FS endemic regions [62]. These hematophagous insects include the vectors for onchocerciasis (black flies) and Chagas disease (kissing bugs), in addition to leishmaniasis (sand flies). Patients with these insect transmitted diseases have anti-Dsg1 autoAbs [63], suggesting the possible cross-reactivity between Dsg1 and antigens from these insects. The identification of LJM11 of sand fly saliva as a possible trigger for FS is by no means excluding the possibility of other environmental triggers for FS development [29]. Other exogenous antigens (such as those from other insects which are prevalent in FS endemic regions) could have cross-reactive epitopes that activate the naïve B cells in genetically susceptible individuals, leading to the development of pathogenic autoAbs via ES and FS onset. To identify those other possible environmental triggers for FS is the future direction of our investigations.

It has been well documented that the pathogenic or disease inducing autoAbs in pemphigus bind to calcium-dependent conformational epitopes on the autoantigens [8, 65–67], while the linear epitopes on the autoantigens have not been reported to be pathogenic. The Dsg1 and LJM11 cross-reactive epitope identified in our current study is linear, and this cross-reactive epitope does not match or overlap with the amino acid residues of the conformational epitopes on EC1 and EC2 domains of Dsg1 targeted by pathogenic IgG4 Abs from FS patients [40] (Supplementary Fig 2). The pathogenic autoAbs in FS may evolve from further autoimmune responses against different epitopes via ES on Dsg1. It is possible that the ES may reach either N terminal (EC1 and 2) or C terminal (EC3–5) of Dsg1 molecule (Fig 6). The ES towards EC3–5 domains does not induce FS, such as in pre-FS individuals. But the ES to the pathogenic epitopes on EC1 and 2 domains of Dsg1 results in full blown FS as demonstrated [8, 40]. Moreover, the mAbs from the pre-FS individual were generated from his peripheral blood sample collected four years before his FS onset (when he had no active FS), providing further evidence that this cross-reactive epitope is not pathogenic. It should be noted that unlike certain antigens or pathogens, such as viruses, which can directly trigger a disease process (and thus are “pathogenic”), only autoreactive mAbs that could directly induce skin tissue damage in mouse model (such as skin blisters) and/or keratinocyte disassociation are termed as “pathogenic” in autoimmune blistering diseases, such as in pemphigus [66, 68, 69]. Autoreactive mAbs that could not induce skin tissue damage or keratinocyte disassociation, such as the mAbs that target the cross-reactive epitope identified in this investigation, are considered as “non-pathogenic”.

It seems surprising that all tested mAbs from FS patients and a pre-FS individual bind to the cadherin proteins via this particular epitope on EC2 domains of these cadherins. This is consistent with our previous reports that FS patients have IgG autoAbs to all these tested cadherins and the levels of IgG (especially IgG1) autoAbs to each cadherins are significantly correlated [36]. It also further indicates that the anti-cadherin autoAbs detected from FS patients could be the same autoAbs that cross-react to these different cadherin molecules. Similar findings have also been reported where autoAbs to other keratinocyte autoantigens, such as Dsc1, Dsg4 and Ecad, have been detected in pemphigus [70]. An explanation can be found in the main hypothesis. First, the insect bites have been found to be a major risk factor for FS development in FS endemic regions [62]. The constant exposure to the insect bites implies that Ab responses against these exogenous insect antigens can be induced among these FS susceptible individuals with predominant specificity against insect antigens (e.g. LJM11). Second, the autoreactive mAbs in this study were all selected and generated based on their reactivity to Dsg1. All these mAbs were originated from peripheral blood lymphocytes of FS and pre-FS individuals, and usually represent the specificity of memory B cells, such as in pemphigus [68] (not terminally differentiated Ab secreting plasma cells). They may not well represent disease inducing pathogenic autoAbs located in close proximity to affected tissues [71]; they likely represent the specificity of B cell receptors in the early stages of ES, which cross-react with the shared epitope. Therefore, the sampling process was highly biased towards the mAbs that react to both Dsg1 and LJM11 (i.e. via the cross-reactive epitope). This may be the reason that all tested mAbs reacted to both Dsg1 and LJM11. This is consistent with our hypothesis that the cross-reactivity is the underlying mechanism for the initiation of FS in its endemic regions.

5. Conclusions

This investigation has identified the primary antigenic target on Dsg1 autoantigen in FS, and the mechanism by which a non-infectious environmental antigen could trigger the development of FS. As the primary target on an autoantigen is difficult to identify due to the impossibility of conducting ES experiment on humans, our findings provide a useful approach to facilitate the discovery of the primary targets on autoantigens of other autoimmune diseases. How an autoimmune disease is triggered and the mechanism by which the role of ES in autoimmune disease progression, exacerbation, relapse, or transition can be investigated. Novel diagnosis and precision medical treatments to block the ES facilitated development of autoimmune diseases can be developed accordingly.

• An environmental antigen could trigger the development of endemic pemphigus foliaceus.

• Epitope spreading could lead to the development of autoantibodies in endemic pemphigus foliaceus.

• It is difficult to identify the primary target of epitope spreading in human autoimmune diseases.

• We identified a primary antigenic target of epitope spreading in endemic pemphigus foliaceus.

• Cross-reactivity between an environmental antigen and autoantigens could initiate autoantibody development.

Abbreviations

ES

Epitope spreading

FS

Fogo Selvagem

Dsg

Desmoglein

Dsc

Desmocollin

Ecad

E-cadherin

EC

Ectodomain

Ab

Antibody

autoAb

Autoantibody

mAb

Monoclonal Ab

APD

Ab phage display

PF

Pemphigus foliaceus

PV

Pemphigus vulgaris

mPV

Mucosal PV

mcPV

Mucocutaneous PV

EAE

Experimental allergic encephalomyelitis

SLE

Systemic lupus erythematosus

RA

Rheumatoid arthritis

scFv

Single chain variable fragment