Research ArticleAutoimmunity

Laminin 511 is a target antigen in autoimmune pancreatitis

See allHide authors and affiliations

Science Translational Medicine  08 Aug 2018:
Vol. 10, Issue 453, eaaq0997
DOI: 10.1126/scitranslmed.aaq0997

Pancreatic perturbation

Autoimmune pancreatitis (AIP) is difficult to diagnose and can sometimes be confused with pancreatic cancer, which presents with similar symptoms. AIP is an inflammatory disease involving elevated IgG4, but the target autoantigen(s) is unidentified. This group’s previous work pointed to the extracellular matrix, and now, Shiokawa et al. show that a truncated form of laminin 511 may be a major autoantigen in AIP. They observed that half of AIP patients they analyzed had anti–laminin 511 antibodies, which were absent in healthy controls. Patient pancreatic tissues were positive for laminin 511, and immunization of mice with this protein induced AIP-like symptoms. These results reveal an autoimmune target in this disease and one day may aid AIP diagnosis.

Abstract

Autoimmune pancreatitis (AIP), a major manifestation of immunoglobulin G4–related disease (IgG4-RD), is an immune-mediated disorder, but the target autoantigens are still unknown. We previously reported that IgG in patients with AIP induces pancreatic injuries in mice by binding the extracellular matrix (ECM). In the current study, we identified an autoantibody against laminin 511-E8, a truncated laminin 511, one of the ECM proteins, in patients with AIP. Anti–laminin 511-E8 IgG was present in 26 of 51 AIP patients (51.0%), but only in 2 of 122 controls (1.6%), by enzyme-linked immunosorbent assay. Because truncated forms of other laminin family members in other organs have been reported, we confirmed that truncated forms of laminin 511 also exist in human and mouse pancreas. Histologic studies with patient pancreatic tissues showed colocalization of patient IgG and laminin 511. Immunization of mice with human laminin 511-E8 induced antibodies and pancreatic injury, fulfilling the pathologic criteria for human AIP. Four of 25 AIP patients without laminin 511-E8 antibodies had antibodies against integrin α6β1, a laminin 511 ligand. AIP patients with laminin 511-E8 antibodies exhibited distinctive clinical features, as the frequencies of malignancies or allergic diseases were significantly lower in patients with laminin 511-E8 antibodies than in those without. The discovery of these autoantibodies should aid in the understanding of AIP pathophysiology and possibly improve the diagnosis of AIP.

INTRODUCTION

Immunoglobulin G4–related disease (IgG4-RD) is an immune-mediated disorder characterized by elevated serum IgG4 concentrations and tumor-like swelling of the involved organs with IgG4+ lymphoplasmacytic infiltration (1). These features were first reported in patients with autoimmune pancreatitis (AIP) (24). The presence of several autoantibodies (510) and a favorable response to steroids or anti-CD20 antibody rituximab (1) indicate an autoimmune mechanism. Seropositivities of the reported autoantibodies, however, are not frequent in patients with AIP (510), and their specificities and pathologic roles remain controversial. Thus, the identification of a specific autoantigen could elucidate the pathophysiology of AIP.

We previously demonstrated that injecting AIP patient IgG into neonatal mice induces pancreatic injury (11), and IgG bound to the basement membrane of the pancreatic acini. Thus, AIP patients may have autoantibodies that recognize molecules within the extracellular matrix (ECM) involved in cell-ECM adhesion, as in mucous membrane pemphigoid (12).

Among various ECM proteins in pancreatic tissue (13), laminin subfamily proteins are present as heterotrimers comprising one of five α chains (α1 to α5), one of three β chains (β1 to β3), and one of three γ chains (γ1 to γ3; fig. S1) (14). Among these, laminin 511, comprising α5, β1, and γ1 chains, plays an important role in cell-ECM adhesion by binding to integrin α6β1 or α3β1 (15). The phenotype of pancreas-specific integrin knockout mice supports the importance of laminin/integrin binding in cell-ECM adhesion in the pancreatic tissue (16, 17).

Here, we screened previously identified pancreatic ECM proteins with sera from AIP patients to identify autoantigens involved in AIP. Our findings provide evidence that a truncated form of laminin 511 that retains full capability for binding to integrins α6β1 and α3β1 (18) is a target antigen in patients with AIP.

RESULTS

Detection of laminin 511-E8 autoantibodies in AIP patients

Our previous study suggested that patients with AIP have antibodies against some antigens in the ECM of the pancreas (11). Therefore, we hypothesized that the antibodies target an ECM component. To identify the autoantigen, antigen screening was first performed by Western blot analysis and immunosorbent column chromatography experiments. We used human and mouse pancreas extracts as samples and AIP patient IgG as the antibody. Using these methods, however, we detected no specific bands (fig. S2). These results could be due to changes in antigenicity during the denaturation process, as is the case of antigens involved in heparin-induced thrombocytopenia (19) and mucous membrane pemphigoid (20), or to the inability of these methods to solubilize the antigen (21).

Therefore, we performed an enzyme-linked immunosorbent assay (ELISA) with previously identified pancreatic ECM proteins (22) including laminins 511-FL, 521-FL, 511-E8, and 521-E8. We also checked the reactivity to other laminin subfamily E8s (laminins 111-E8, 211-E8, and 332-E8; fig. S1). In general, laminins exist as nonprocessed (full-length) or processed forms in tissues (20, 2325), and these truncated laminins (E8s) were produced by enzymatic digestion with pancreatic elastase and contain the active integrin-binding site (14). Thus, laminin E8s serve as functionally minimal forms that retain the full capability for binding to integrins (18). Laminin 511-E8 consists of the C-terminal end of the α5, β1, and γ1 chains. Here, we confirmed that C-terminal processed forms of laminin 511 exist in normal mouse and human pancreas (fig. S3).

As a training group, we examined antibodies from 10 AIP patients and 10 controls whose sera had been used in our previous report (11). We found that 6 of 10 AIP patients had IgG antibodies against laminin 511-E8 based on a cutoff value [optical density units (OD)] of the mean + 3 SDs of the control sera (fig. S4A), whereas none of the controls had IgG antibodies reactive to laminin 511-E8 (P = 0.011). In contrast, none of the AIP patient sera reacted with laminin 511-FL or other ECM proteins tested, except for one patient who was complicated by IgG4-related kidney disease and had antibodies to laminins 511-FL, 521-FL, and 521-E8 (fig. S4A). The increased detection of IgG antibodies against laminin 511-E8 in the six patients was ablated when the sera were preincubated with laminin 511-E8, but not laminin 511-FL (fig. S4B).

We then examined dominant IgG subclasses to laminin 511-E8 and found that all six AIP patients with IgG antibodies against laminin 511-E8 had IgG1 antibodies. One patient complicated by IgG4-related kidney disease also had IgG4 antibodies (fig. S4C). We also examined IgA, IgM, and IgE antibodies to laminin 511-E8. One patient without IgG antibodies against laminin 511-E8 had IgM antibodies (fig. S5A). None of the patient sera reacted with laminin 511-E8 by Western blot analysis (fig. S6), suggesting that the antigenic epitope depends on its conformation, consistent with the results of the antigen screening by Western blot analysis.

To further confirm the presence of anti–laminin 511-E8 antibodies, we used ELISA for laminin 511-E8, laminin 511-FL, laminin 521-E8, laminin 521-FL, collagen IV, and fibronectin, which are abundantly present in pancreatic tissue, and validated the results in another cohort of 41 AIP patients and 112 controls (92 patients with cancer, autoimmunity, or other diseases and 20 healthy controls). IgG antibodies against laminin 511-E8 were detected in 20 of 41 AIP patients but in only 2 of 112 controls (P < 0.001; Fig. 1). In contrast to AIP patients, the positivity of the two controls did not change when preincubated with laminin 511-E8, suggesting a nonspecific reaction. As for other ECM proteins tested, only three AIP patients had IgG antibodies against laminin 521-E8, one patient against fibronectin, one control against laminin 511-FL, and one control against fibronectin (Fig. 1).

Fig. 1 Detection by ELISA of laminin 511-E8 antibodies in AIP patient sera.

Serum antibodies against laminin 511-E8, laminin 511-FL, laminin 521-E8, laminin 521-FL, collagen IV, and fibronectin were quantified by ELISA in a validation cohort. Forty-one AIP patients and 112 controls (20 normal and 92 diseased) were examined. The cutoff value (OD) was defined as the mean + 3 SDs of the control sera. The dashed line indicates the cutoff value.

When the training and validation groups were combined, IgG antibodies against laminin 511-E8 were present in 26 of 51 (51.0%) AIP patients and in 2 of 122 (1.6%) controls (P < 0.001). Only one AIP patient had anti–laminin 511-FL antibodies.

We analyzed the correlation between the OD values of anti–laminin 511-E8 antibody and serum IgG4 concentrations in patients with anti–laminin 511-E8 antibody (training group + validation group) and found the positive correlation between the two values [correlation coefficient (r) = 0.63, P = 0.0115]. Notably, the two patients with low serum complement concentrations had the highest titers for anti–laminin 511-E8 antibody (AIP1 and AIP11 in table S1).

Binding of AIP patient IgG to laminin in human and mouse pancreatic tissues

We confirmed that normal mouse and human pancreas expressed laminin and laminins α5, β1, and γ1 by immunofluorescence histology (fig. S7, A and B), consistent with the Western blot analysis (fig. S3). Each chain of laminin 511 is expressed and overlaid with laminin staining (merging as yellow; fig. S7, A and B).

To ascertain whether patient IgG binds to laminin in the pancreatic tissue of AIP patients, we examined resected pancreas of AIP patients (n = 5). Immunohistochemistry for human IgG revealed linear and interstitial staining at the base of the pancreatic acini in AIP patients in addition to typical staining of IgG-positive plasma cells (Fig. 2A). Immunohistochemistry using a commercial anti-laminin antibody revealed a staining pattern similar to IgG staining in AIP tissue, and dual immunofluorescence studies showed colocalization of IgG/IgG1 with laminin at the base of the pancreatic acini of AIP patients (Fig. 2B). In contrast, no IgG staining was observed in the nonaffected area of the same patient (Fig. 2C). We quantified the merged area by ImageJ (fig. S8). As a result, 73 and 91% of laminin merged with human IgG and human IgG1 in AIP tissue (n = 5), respectively. In contrast, only 3.2 and 2.8% of laminin merged with human IgG and IgG1 in normal pancreatic tissue (n = 5), respectively (P < 0.05 and P < 0.05, respectively).

Fig. 2 Colocalization of laminin and IgG in AIP patient pancreatic tissue.

Immunohistochemical studies of pancreatic tissues from AIP patients. (A) Top: Hematoxylin and eosin (H&E) staining (left) and immunohistochemical staining for IgG (middle) and laminin (right) in affected tissue sections from an AIP patient (patient numbers: AIP52 to AIP56 in table S3). Scale bars, 200 μm. Bottom: High-power magnification of the upper panels. Scale bars, 20 μm. (B) Immunofluorescence staining of the affected acini in the AIP patient. Top: Staining for IgG (green), laminin (red), and a merged image. Bottom: Immunofluorescence staining for IgG1 (green), laminin (red), and a merged image. Scale bars, 20 μm. (C) Immunofluorescence staining of the unaffected acini in the AIP patient. Scale bars, 20 μm. Left lower white boxes are magnified images of the dashed line boxes. Sections were stained with polyclonal anti-mouse/human laminin antibody produced against the protein purified from the basement membrane of Engelbreth-Holm-Swarm sarcoma, and the antibody reacts with several laminin family proteins of mouse and human, including human laminin 511-E8, but does not react with other ECM proteins examined.

Previously, we reported that AIP patient IgG binds to pancreatic ECM and exhibits pathogenic activities in neonatal mouse pancreas (11). We therefore examined whether exogenously administered IgG of AIP patients with laminin 511-E8 antibodies (n = 5) targets laminin 511 in mice (Fig. 3, A and B). As expected, dual immunofluorescence studies revealed colocalization of patient IgG/IgG1 with laminin in mouse pancreas ECM. Preincubation of patient IgG with laminin 511-E8, but not laminin 511-FL (Fig. 3C), abolished IgG staining, indicating the binding of patient IgG to mouse pancreatic laminin 511-E8. We quantified the merged area by ImageJ (fig. S8). As a result, 90% of mouse laminin merged with patient IgG1 (n = 5). In contrast, only 2.1% of mouse laminin merged with control IgG1 (n = 5; P < 0.05). These findings indicate that AIP patient IgG and IgG1 bind to laminin 511-E8 in the pancreas.

Fig. 3 Colocalization of AIP patient IgG and laminin in mouse pancreatic tissue by passive transfer of patient IgG.

Immunofluorescence studies were performed 12 hours after subcutaneous injection of control IgG or AIP patient IgG. (A) Immunofluorescence staining for human IgG in mouse pancreatic tissue sections. Top right three panels show staining for patient IgG (green), amylase (red), and a merged image. Bottom right three panels show staining for patient IgG (green), laminin (red), and a merged image. Scale bars, 20 μm. (B) Immunofluorescence staining for human IgG1 and laminin in mouse pancreas. Staining for a patient’s IgG1 (green), laminin (red), and a merged image are shown. Scale bars, 20 μm. (C) When AIP patient IgG was preincubated with laminin 511-E8 or 511-FL, IgG staining with laminin (left panel) was abolished only by preincubation with laminin 511-E8 (middle panel), but not by preincubation with laminin 511-FL (right panel). Scale bars, 20 μm. Representative photos are shown.

Induction of AIP-like lesions in mice by immunization with human laminin 511-E8

To ascertain whether a pathogenic autoantibody targets the truncated laminin 511 in AIP, we injected human laminin 511-E8, human laminin 511-FL, 521-FL, or ovalbumin into BALB/c mice. Eight-week-old male BALB/c mice were immunized with these proteins emulsified in complete Freund’s adjuvant (CFA) subcutaneously on day 0 and boosted on days 28 and 56. Mice were sacrificed 28 days after the third immunization, and various organs were removed for analysis. ELISA confirmed the generation of anti–laminin 511-E8 antibodies in all mice immunized with laminin 511-E8, but not ovalbumin (fig. S9A). Notably, only pancreatic tissues and salivary glands exhibited injury in all mice immunized with laminin 511-E8, whereas no mice injected with human laminin 511-FL, laminin 521-FL, or ovalbumin showed any change (Fig. 4A and fig. S9B). Histologically, acinar atrophy, fibrosis, dense infiltration of plasma cells (CD138+) and lymphocytes (CD45R+ or CD3+), and obliterative phlebitis were observed in the pancreas (Fig. 4, A to C). We also observed infiltration (>10 cells per high-power field) of plasma cells positive for IgG1, the mouse counterpart of human IgG4. Murine IgG1 and human IgG4 are considered to be analogous because of their biological and functional similarities. For example, similar to human IgG4, murine IgG1 does not activate the complement system through the classical pathway (26, 27). These histologic findings fulfill the pathologic criteria of the international consensus for human AIP [at least three of the following: (i) periductal lymphoplasmacytic infiltrate, (ii) obliterative phlebitis, (iii) storiform fibrosis, and (iv) abundant IgG4+ cells (>10 cells per high-power field) (28, 29)]. In the salivary gland, we observed only infiltration of lymphocytes and plasma cells around the duct (fig. S9B). Thus, immunization with human laminin 511-E8 recapitulated the development of chronic pancreatitis similar to human AIP in mice.

Fig. 4 Induction of AIP-like lesions in mice by immunization with human laminin 511-E8.

BALB/c mice were immunized with human laminin 511-E8 (right panels) or ovalbumin (left panels) once a month for 3 months. H&E staining (A); immunohistochemical staining for CD3, CD45R, CD138, and IgG1 (B); and elastica van Gieson staining (C) of mouse pancreatic tissue sections obtained 28 days after the last immunization are shown. Among various organs (brain, salivary gland, thyroid, heart, lung, gallbladder, bile duct, pancreas, kidney, intestine, bladder, prostate, aorta, skeletal muscles, and skin), only the pancreas and salivary gland (fig. S9B) exhibited injury in all mice injected with laminin 511-E8 (n = 5), whereas none of the ovalbumin mice (n = 5) had any lesions (five of five versus zero of five, P < 0.005). Subepithelial space enlargement in bile and pancreatic duct with no epithelial changes, typical features of AIP, were also observed (A, bottom panels, black arrows). IgG1, a counterpart of human IgG4, was also detected (B, bottom right panel). Scale bars, 50 μm (A) and 20 μm (B).

Detection of integrin α6β1 autoantibodies in AIP patients negative for laminin 511-E8 autoantibodies

In human mucous membrane pemphigoid, several autoantigens compose the hemidesmosome (BP180, BP230, integrin α6β4, type VII collagen, and laminin 332) (12). Because the ligands for laminin 511-E8 are integrins α6β1 and α3β1 on the cell surface (15), we hypothesized that integrin α6β1 or α3β1 could also be autoantigens in AIP. Sera from 4 of 25 AIP patients without laminin 511-E8 antibodies were positive for anti–integrin α6β1 antibodies, whereas none of the 26 AIP patients with laminin 511-E8 autoantibodies or the 122 controls had anti–integrin α6β1 antibodies (Fig. 5). None of the 51 AIP patients or the 122 controls had anti–integrin α3β1 antibodies. Thus, integrin α6β1 could be an autoantigen in a limited number of AIP patients, supporting the concept that laminin 511-E8 and integrin α6β1 binding is a target of the autoantibodies in AIP.

Fig. 5 Detection of the integrin α6β1 antibody in a limited number of AIP patients negative for the anti–laminin 511-E8 antibody.

The ELISA results of the 51 AIP patients and 122 controls are shown. The cutoff value (OD) was defined as the mean + 3 SDs of the control sera. The dashed line indicates the cutoff value.

Clinical relevance of the presence of laminin 511-E8 antibodies

We compared the clinical characteristics of AIP patients with and without laminin 511-E8 antibodies (Table 1 and table S2). The frequency of malignancies and allergic diseases was significantly lower in patients with laminin 511-E8 antibodies than in those without (P = 0.0017 and 0.0043, respectively). None of the patients with these antibodies had malignancy, whereas 8 of 25 patients without these antibodies had malignancy. Only 3 of 26 AIP patients with laminin 511-E8 antibodies had allergic diseases, whereas nearly half of the patients without these antibodies had allergic diseases. An imaging study revealed that pancreatic head involvement was significantly less frequent in patients with laminin 511-E8 antibodies than in those without (Table 1 and fig. S10).

Table 1 Clinical differences between AIP patients positive and negative for anti–laminin 511-E8 antibodies.

Malignancy includes those diagnosed at the same time of or after AIP diagnosis. Hypocomplement includes low C3, C4, or CH50. Allergy includes asthma, atopic dermatitis, drug allergy, food allergy, and contrast media hypersensitivity. Values are medians for age, serum IgG, and serum IgG4.

View this table:

In addition, we compared serum samples of five patients (see the Supplementary Materials) with laminin 511-E8 antibodies before and after steroid treatment. The laminin 511-E8 antibody titer decreased to under the cutoff value in association with a decrease of serum IgG4 concentration and improvement of the pancreatic image (fig. S11).

DISCUSSION

More than 50% of AIP patients tested in the present study had laminin 511-E8 antibodies, as detected by ELISA. In remarkable contrast, only 2 of 122 controls had these antibodies. IgG was colocalized with laminin 511 in the pancreas of AIP patients. When injected into mice, IgG of AIP patients with these antibodies colocalized with laminin 511 in the pancreas. Furthermore, immunization of mice with human laminin 511-E8 induced pancreatic lesions fulfilling the pathologic criteria for human AIP. These findings suggest that the truncated form of laminin 511 is a target antigen in AIP.

Although half of the AIP patients had laminin 511-E8 antibodies, laminin 511-FL antibodies were detected in only one patient. Moreover, in contrast to immunization of mice with human laminin 511-E8, immunization with human laminin 511-FL did not induce pancreatic injury. The reason for this is unknown. One possibility is that the antigenic epitope is cryptic in the full-length form and thus cannot be recognized by B cells. Laminins can be degraded by enzymes in vitro (30, 31). In vivo, laminin 332 abundant in the dermis is synthesized initially as full length that undergoes specific processing to a smaller form in ECM (20, 24). Similar degradations of laminin 511 may have induced a new epitope for the antibody production.

We found autoantibodies against integrin α6β1, a ligand for laminin 511, in 4 of 25 AIP patients who were negative for the laminin 511-E8 antibody. Laminin-integrin binding has important functions in the adhesion of the cells with basement membranes and subsequent signaling in the cells. How these antibodies exert pathogenic effects, however, remains unknown. One possibility is that the antibody acts as a blocking antibody against laminin 511/integrin α6β1 binding. Alternatively, analogous to human mucous membrane pemphigoid (12), the pathogenic effect may be mediated by a mechanism involving complement activation. In this regard, we previously observed that, in a passive IgG transfer model, the pancreatic injury induced by IgG from AIP patients is accompanied by complement deposition in the lesion (11).

Notably, although the elevated serum concentration of IgG4 relative to controls was more prominent than that of IgG1 in patients with AIP, we found specific IgG1 antibodies against laminin 511-E8 in more than half of the patients but IgG4 antibodies in only one patient, suggesting an increase of non–laminin-specific IgG4 antibodies. The reason for the presence of the specific IgG1 antibody rather than the IgG4 antibody and increase of nonspecific IgG4 is unclear at present. However, we previously reported that B cell activating factor (BAFF) production is enhanced in patients with IgG4-RD, and BAFF activates B cells and induces IgG4 production in B cells of patients with IgG4-RD (32). In addition, laminin 511-E8 is known to bind BAFF (33). Thus, it is possible that the anti–laminin 511-E8 antibody modulates these pathways to enhance IgG4 production from B cells.

AIP patients with laminin 511-E8 antibodies had distinctive clinical features. First, the frequency of malignancy in patients with these antibodies was significantly lower than in those without. In mucous membrane pemphigoid, the frequency of cancer is very high in patients with laminin 332 antibodies compared to those without (34). Accordingly, AIP patients with paraneoplastic features may have distinct autoantibodies that remain unknown. Second, AIP is frequently associated with various allergic diseases (35). Here, the frequency of allergic diseases in patients with these antibodies was significantly lower than in those without. The data suggest a different pathophysiology between AIP with and without anti–laminin 511-E8 antibodies. Third, pancreas body or tail lesions were significantly more frequent in AIP patients with laminin 511-E8 antibodies than in those without. This might be due to a different distribution of the autoantigens within the pancreas. AIP may therefore comprise different subtypes with different autoantigens. However, the sample size in this study was relatively small, and retrospective studies are more prone to bias than prospective studies. Thus, to clarify differences of the clinical characteristics between AIP patients with and without anti–laminin 511-E8 antibody more precisely, a prospective study with a large number of patients needs to be performed in future studies.

In conclusion, we identified autoantibodies against laminin 511-E8 in many AIP patients and autoantibodies against integrin α6β1, a ligand for laminin 511, in a limited number of patients, confirming that AIP is an autoimmune disease. Because both administration of AIP patient IgG in our previous study (11) and immunization with laminin 511-E8 in this study induced the dissociation of pancreatic acini in mice, these autoantibodies may exert pathogenic activities by targeting laminin 511/integrin α6β1 binding. The very high specificity of this antibody for AIP patients suggests that this antibody could possibly be a good diagnostic marker for AIP. In addition, the laminin 511-E8 antibody titer decreased to under the cutoff value in association with improvement of the pancreatic image, also suggesting that this antibody could possibly be a good marker for therapeutic effect. Because a considerable number of patients did not have either of the two antibodies identified in this study, further studies are required to find other autoantigens in AIP. Finally, our patients were all Japanese. So, whether anti–laminin 511-E8 antibodies are also present in patients with AIP in western countries needs to be examined in future studies.

MATERIALS AND METHODS

Study design

Studies involving animals (protocol number Med Kyo 17197) were approved by the Institutional Animal Ethics Committee of Kyoto University. Human studies (protocol number E1716) were performed according to the Declaration of Helsinki and approved by the Institutional Review Board of Kyoto University Hospital. All patients and controls provided written informed consent.

The main aim of this study was to explore the target autoantigens that contribute to the pathogenesis of AIP. Five experimental or clinical studies were performed: (i) Autoantigen screening by Western blot analysis and immunosorbent column chromatography experiments using human and mouse pancreas extracts as samples and AIP patient IgG as the antibody, and by ELISA using previously identified various ECM proteins as samples and AIP patient IgG as the antibody. (ii) The candidate protein was validated with sera from another cohort of 41 AIP patients and 112 controls by ELISA. (iii) Colocalization of AIP patient IgG with the candidate autoantigen was examined in the pancreas of AIP patients and mice injected with AIP patient IgG. (iv) To examine its pathogenicity, we immunized mice with the candidate autoantigen. (v) Last, we analyzed the clinical relevance of the presence of laminin 511-E8 antibodies. Exact numbers for each experiment are included below and in the figure legends. The investigators were not blinded when conducting or evaluating the experiments. We chose plasma of training samples from our previous study, which had been consecutively obtained (11). In the validation group, sample sizes were chosen from consecutive patients or controls for feasibility and provided a reasonable degree of statistical power for comparison of the two groups. All in vivo and in vitro studies were carried out at a minimum of triplicate independent experiments. Primary data are located in table S5.

Patients

Serum samples were obtained from 51 consecutive patients at Kyoto University who met the comprehensive diagnostic criteria for IgG4-RD 2011 (28) or the diagnostic criteria of AIP (29) between January 2010 and November 2016. We performed endoscopic ultrasound–guided fine needle aspiration (EUS-FNA) for all the patients in our AIP cohort and excluded pancreatic cancer. Moreover, no AIP patient developed pancreatic cancer during the follow-up period. These serum and tissue samples were obtained at diagnosis of AIP before steroid treatment. To detect the pancreatic lesions, all the AIP patients were examined by enhanced computed tomography (CT), positron emission tomography–CT, magnetic resonance imaging, and EUS, and the enhanced CT images were scanned by 1 mm slice thickness. We checked the laminin 511-E8 antibody and serum IgG4 concentrations and CT scan at nearly the same time both before and after steroid treatment. Blood samples were also collected from 5 of 51 patients after treatment with prednisolone. Serum samples were obtained from 25 healthy controls and 87 disease controls (49 patients with various histologically proven cancers and 39 noncancer patients who met the criteria for each disease). Sera from 10 of 51 AIP patients and from 10 of 122 controls were used in our previous study (11), and those sera were used as a training group in this study. All serum samples were stored at −80°C. An overview of the clinical characteristics of the patients and controls for these serum studies is provided in table S1. Pancreatic tissues of five AIP patients suspected to have pancreatic cancer and thus operated on were used for histologic analysis in Fig. 3. These five patients fulfilled the histopathologic criteria for AIP (30). Tissue from four normal human pancreases used for Western blot analysis was obtained from patients whose pancreas was operated on for other diseases. An overview of the clinical characteristics of the patients and controls for histologic and Western blot studies is provided in table S3.

Western blot analysis

Normal human pancreatic or mouse pancreatic extracts and mouse pancreatic stellate cells (PSCs) were boiled in Laemmli’s sample buffer with or without 5% mercaptoethanol, fractionated on 4 to 15% SDS-polyacrylamide gels (Bio-Rad), and transferred to nitrocellulose membranes according to standard protocols. After blocking with 10% skim milk, the blots were incubated with primary antibodies. The primary antibodies used were patient or control sera (1:1000), anti-mouse/human C-terminal one-third end of laminin α5 (1:1000; LSBio), anti-mouse/human N-terminal end of laminin α5 (1:1000; Abcam), anti-mouse/human C-terminal end of laminin β1 (1:1000; Abcam), anti-mouse/human C-terminal end of laminin γ1 (1:1000; Thermo Fisher Scientific), and anti-mouse/human β-actin (1:4000; Abcam). Secondary antibodies were peroxidase-conjugated anti-human IgG (1:4000; Abcam), anti-mouse IgG (GE Healthcare), or anti-rabbit IgG (GE Healthcare).

Mouse PSC isolation

Mouse PSC isolation was performed as described previously (36). Briefly, the pancreas was dissected and minced. After centrifugation, to remove excess fat, the tissue was resuspended in a solution containing collagenase P (Sigma) and incubated at 37°C for 20 min. Digested tissue was centrifuged. The pellet was resuspended in Gey’s balanced salt solution (GBSS), layered underneath GBSS, and centrifuged. PSCs were collected from the fuzzy band at the interface and plated. We ensured that PSCs had a fibroblast-like form and were α smooth muscle actin– and vimentin-positive and CK19-negative as determined by microscopy. The isolated cells were used for experiments after three passages.

Immunosorbent column chromatography

IgG1 was purified from the sera (10 ml) of patient 1 and control 1 (table S1) using CaptureSelect IgG1 (Hu) affinity matrix (Thermo Fisher Scientific) according to the manufacturer’s instructions. The purified IgG1 (10 mg/g of gel) was coupled with cyanogen bromide–activated Sepharose 4B beads (GE Healthcare) according to the manufacturer’s instructions. The IgG1 Sepharose 4B beads (10 ml of gel matrix) were then poured into a column (GE Healthcare). Pancreas extracts from 8-week-old male mice (SLC) were applied to the column for 20 min. The column was then washed four times with 20 ml of tris-buffered saline containing 0.1% NP-40. The antigen bound to the column was eluted with 0.1 M glycine (pH 3.0). The eluates were dialyzed against tris-buffered saline containing 0.05% NP-40, concentrated to 0.5 mg/ml using an Amicon Centriprep concentrator (Merck Millipore), and then boiled in Laemmli’s sample buffer. The samples were fractionated on 4 to 15% SDS-polyacrylamide gels (Bio-Rad) and stained with silver staining.

Enzyme-linked immunosorbent assay

Human recombinant ECM proteins were purchased; laminin 511-E8 (Nippi), laminin 511 full length (FL; BioLamina), laminin 521–FL (BioLamina), collagen IV–FL (Abcam), collagen I (BioVision), collagen alpha-1(XVIII) (MyBioSource), fibronectin (TAKARA), nidogen-1 (R&D Systems), nidogen-2 (R&D Systems), fibulin-1C (R&D Systems), endorepellin (R&D Systems), SMOC-1 (R&D Systems), agrin (R&D Systems), collagen 6a1 (MyBioSource), collagen 18a1 (MyBioSource), annexin A2 (Abcam), integrin α6β1 (R&D Systems), and integrin α3β1 (R&D Systems) were used. Human recombinant laminins 521-E8, 33 2-E8, 211-E8, and 111-E8 were generated in-house (table S4). Laminin E8s are truncated laminins comprising the C-terminal region of the α, β, and γ chains (fig. S1) (30). Purity of laminin 511-E8 was confirmed using Coomassie brilliant blue staining after SDS–polyacrylamide gel electrophoresis. Serum antibodies against candidate ECM proteins were quantified using the ELISA Starter Accessory Kit (Bethyl Laboratories) according to the manufacturer’s instructions. Briefly, microtiter plates were coated with 100 μl of candidate proteins (2 μg/ml) for 1 hour at room temperature. After washing the plates five times with tris-buffered saline containing 0.05% Tween 20 (wash solution), they were coated with tris-buffered saline containing 1% bovine serum albumin and, after five washes with wash solution, incubated with 100 μl of diluted patient serum (1:20) for 30 min at room temperature. After five washes with wash solution, the plates were incubated with 100 μl of goat anti-human IgG antibody conjugated with horseradish peroxidase (1:4000; Abcam) at room temperature for 1 hour. After five washes with wash solution, bound reactants were detected by 3-min incubation with 3,3′,5,5′- tetramethylbenzidine. Absorbance was determined at 450 nm. Control wells that were not coated with candidate proteins were also used as a negative control for each serum studied. All assays were performed in triplicate, and the specific binding of the serum antibody to candidate proteins was calculated by subtracting the mean absorbance of the control wells from the mean absorbance of the candidate protein-coated wells. To examine IgG subclasses, we used anti-human IgG1, IgG2, IgG3, and IgG4 antibodies conjugated with horseradish peroxidase (1:2000; The Binding Site) as the secondary antibody. To examine IgA, IgM, and IgE antibodies, we used anti-human IgA, IgM, and IgE antibody conjugated with horseradish peroxidase (1:4000; Bethyl Laboratories) as the secondary antibody.

Immunization study

Eight-week-old male BALB/c mice were immunized with 120 μg of laminin 511-E8, 511-FL, or 521-FL emulsified in 100 μl of phosphate-buffered saline and equal volumes of CFA subcutaneously at four sites on the back on day 0 and boosted on days 28 and 56. Control mice were injected with ovalbumin in the same manner. Mice were sacrificed 28 days after the third immunization, and various organs (brain, salivary gland, thyroid, heart, lung, gallbladder, bile duct, pancreas, kidney, intestine, bladder, prostate, aorta, skeletal muscles, and skin) were removed.

Animal studies

Passive transfer of AIP patient IgG to neonatal mice was performed as described previously (11). Briefly, IgG fractions purified from the sera of patients or controls were injected subcutaneously into neonatal BALB/c mice in a single dose of 10 to 20 mg of human IgG per gram body weight in 200 μl of phosphate-buffered saline. Mice were sacrificed 12 hours after IgG injection; the pancreas was immediately removed and evaluated histologically. For the blocking study, AIP patient IgG was incubated with human laminin 511-E8 (100 μg) or laminin 511-FL (100 μg) for 30 min and injected into the mice.

Histologic evaluation

The organs were immediately removed from the mice and placed in formalin for at least 1 day before embedding in paraffin. Organs were cut into 5-μm-thick coronal sections, and every 10th section of each entire organ was stained with H&E. Five randomly selected sections of the pancreas were stained with elastica van Gieson stain.

Immunohistochemical study

The immunohistochemical study was performed according to standard methods for mouse and human tissue sections. The primary antibodies used were anti-mouse CD3 (1:100; Abcam), anti-mouse CD45R (1:100; Abcam), anti-mouse CD138 (1:100; Abcam), anti-mouse IgG1 (1:100; Abcam), anti-human IgG1 (1:100; The Binding Site), anti-mouse/human amylase (1:100; Abcam), and anti-human IgG (1:100; Abcam) after antigen retrieval with 10 mM citrate buffer (pH 6.0) and anti-human IgG (1:50; Agilent Technologies) and anti-mouse/human laminin (1:100; Abcam) after antigen retrieval with proteinase K. This polyclonal anti-mouse/human laminin antibody is produced against the protein purified from the basement membrane of Engelbreth-Holm-Swarm sarcoma (37), and the antibody reacts with several laminin family proteins of mouse and human, including human laminin 511-E8, but does not react with other ECM proteins examined.

Sections were incubated with primary antibodies at room temperature for 60 min. For detection, the Universal Dako LSAB+ Kit (Agilent Technologies) was used according to the manufacturer’s instructions. Detection times were equally standardized for all sections.

Immunofluorescence study

Immunofluorescence studies were performed using standard methods for mouse and human tissue sections. For direct immunofluorescence, fluorescein isothiocyanate–conjugated antibody specific to anti-human IgG (1:50; Agilent Technologies) and anti-human IgG1 (1:50; The Binding Site) was used. For indirect immunofluorescence, the primary antibodies used were anti-mouse/human laminin α5 (Merck), anti-mouse/human laminin β1 (Abcam), anti-mouse/human laminin γ1 (Abcam), and anti-mouse/human laminin (1:100; Abcam). As secondary antibodies, Alexa Fluor 488 anti-rabbit IgG (1:100; Thermo Fisher Scientific), Alexa Fluor 594 anti-mouse IgG (1:100; Thermo Fisher Scientific), Alexa Fluor 594 anti-rabbit IgG (1:100; Thermo Fisher Scientific), and Alexa Fluor 594 anti-rat IgG (1:100; Thermo Fisher Scientific) were used.

Statistical analysis

Differences were assessed using Student’s t test for continuous data, and the χ2 test and Fisher’s exact test for categorical data. Statistical analysis was performed using two-tailed statistical tests with the Statistical Package for the Social Sciences [JMP version 13 (SAS)]. A P value of less than 0.05 was considered statistically significant.

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/10/453/eaaq0997/DC1

Fig. S1. Schematic representations of the laminin full-length (laminin-FL) and laminin E8 isoforms.

Fig. S2. Screening of antigens in AIP.

Fig. S3. Detection of C-terminal fragments of α, β, and γ chains of laminin 511 in mouse and human pancreas by Western blot analysis.

Fig. S4. Detection by ELISA of antibody against human laminin 511-E8 in the sera of AIP patients.

Fig. S5. IgA, IgM, and IgE antibodies against human laminin 511-E8 in the sera of AIP patients and titration of sera against laminin 511-E8.

Fig. S6. No reaction of patient sera with laminin 511-E8 by Western blot analysis.

Fig. S7. Expression and colocalization of laminin 511 and each chain of laminin in normal mouse and human pancreatic tissue.

Fig. S8. Quantification of the merged area in the pancreatic tissue of AIP patients and mouse pancreas of passive transfer model.

Fig. S9. Induction of antibody against human laminin 511-E8 and salivary lesions by immunization with human laminin 511-E8.

Fig. S10. Relationship between pancreatic image and positivity for the laminin 511-E8 antibody.

Fig. S11. Changes in antibody titers against laminin 511-E8 by steroid treatment.

Table S1. Clinical information on all patients and controls.

Table S2. Clinical differences among AIP patients with anti–laminin 511-E8 antibody, anti–integrin α6β1 antibody, or neither antibody.

Table S3. Clinical information on five AIP patients and controls whose pancreatic tissues were used in experiments.

Table S4. Antigens used for ELISA in this study.

Table S5. Primary data.

REFERENCES AND NOTES

Acknowledgments: We thank K.S. for supplying recombinant laminins. We also thank T.M. for providing control serum with collagen diseases. Funding: This work was supported by the Japan Society for the Promotion of Science KAKENHI (16K19339); the Health and Labour Sciences Research Grants for Research on Intractable Diseases from the Ministry of Health, Labour, and Welfare, Japan; the Practical Research Project for Rare/Intractable Diseases Grant in Japan; the Agency for Medical Research and Development (17im0210801h0002); and the Takeda Science Foundation. Author contributions: T.C., Y.K., K.S., and M.S. conceived this manuscript and designed and supervised the studies. M.S. gathered the data with the help of T.K., T. Tomono, and K.K. M.S. analyzed the data. N.K. and H.Y. checked the statistical analysis. T. Tsuruyama checked pathological analysis. M.S. wrote the first draft of the manuscript and revised it with considerable input from T. Morita, S.M., Y. Sogabe, N.K., T. Matsumori, A.M., Y.N., T.U., M.T., Y.Y., Y. Sakuma, T. Maruno, N.U., T. Tsuruyama, T. Mimori, and H.S. Competing interests: K.S. is a founder and shareholder of Nippi Inc. Kyoto University has filed patents related to this manuscript: PCT application PCT5439, entitled “Identifying the antigen of IgG4-related disease and measurement of the antibody,” with T.C., Y.K., and M.S. as inventors. All other authors declare that they have no competing interests. Data and materials availability: All data associated with this study are present in the paper or the Supplementary Materials.
View Abstract

Navigate This Article