Investigation of immune complexes formed by mitochondrial antigens containing a new lipoylated site in sera of primary biliary cholangitis patients

Summary

Primary biliary cholangitis (PBC) is characterized by the presence of serum anti‐mitochondrial autoantibodies (AMAs). To date, four antigens among the 2‐oxo‐acid dehydrogenase complex family, which commonly have lipoyl domains as an epitope, have been identified as AMA‐corresponding antigens (AMA‐antigens). It has recently been reported that AMAs react more strongly with certain chemically modified mimics than with the native lipoyl domains in AMA‐antigens. Moreover, high concentrations of circulating immune complexes (ICs) in PBC patients have been reported. However, the existence of ICs formed by AMAs and their antigens has not been reported to date. We hypothesized that AMAs and their antigens formed ICs in PBC sera, and analyzed sera of PBC and four autoimmune diseases (Sjögren's syndrome, systemic lupus erythematosus, systemic scleroderma, and rheumatoid arthritis) using immune complexome analysis, in which ICs are separated from serum and are identified by nano‐liquid chromatography‐tandem mass spectrometry. To correctly assign MS/MS spectra to peptide sequences, we used a protein‐search algorithm that including lipoylation and certain xenobiotic modifications. We found three AMA‐antigens, the E2 subunit of the pyruvate dehydrogenase complex (PDC‐E2), the E2 subunit of the 2‐oxo‐glutarate dehydrogenase complex (OGDC‐E2) and dihydrolipoamide dehydrogenase binding protein (E3BP), by detecting peptides containing lipoylation and xenobiotic modifications from PBC sera. Although the lipoylated sites of these peptides were different from the well‐known sites, abnormal lipoylation and xenobiotic modification may lead to production of AMAs and the formation ICs. Further investigation of the lipoylated sites, xenobiotic modifications, and IC formation will lead to deepen our understanding of PBC pathogenesis.

Primary biliary cholangitis (PBC) is characterized by the presence of serum anti‐mitochondrial autoantibodies (AMAs) against the 2‐oxo‐acid dehydrogenase complex family; however, the immune complexes (ICs) formed by AMAs and the antigens has not been reported to date and their in vivo antigenicity is not fully clear. Immune complexome analysis identified three AMA‐antigens to be incorporated into ICs from PBC sera, the E2 subunit of the pyruvate dehydrogenase complex (PDC‐E2), the E2 subunit of the 2‐oxo‐glutarate dehydrogenase complex (OGDC‐E2), and dihydrolipoamide dehydrogenase binding protein (E3BP), by detecting peptides containing lipoylation or xenobiotic modifications. The lipoylated sites of these peptides were different from the well‐known sites; therefore, abnormal lipoylation and xenobiotic modification may lead to production of AMAs and the formation ICs

Introduction

Primary biliary cholangitis (PBC) is a chronic inflammatory autoimmune liver disease characterized by chronic destructive cholangitis associated with selective destruction of intrahepatic small bile ducts [1, 2]. PBC predominantly affects middle‐aged to older women and is often associated with other autoimmune diseases [e.g. systemic scleroderma (SSc), Sjögren’s syndrome (SS), systemic lupus erythematosus (SLE) and rheumatoid arthritis (RA)] [3, 4]. This disease is characterized by the presence of serum anti‐mitochondrial autoantibodies (AMAs) in 90–95% of patients, which can be detected before the appearance of disease symptoms [5, 6, 7]. However, their titer does not correlate with disease severity, and it is unknown whether AMAs are associated with the pathology of PBC [8].

The autoantigens of AMAs (AMA‐antigens) have been identified as members of the 2‐oxo‐acid dehydrogenase complex (2‐OADC) family, including the E2 subunit of the pyruvate dehydrogenase complex (PDC‐E2), the E2 subunit of the branched chain 2‐oxo acid dehydrogenase complex (BCOADC‐E2), the E2 subunit of the 2‐oxo‐glutarate dehydrogenase complex (OGDC‐E2) and dihydrolipoamide dehydrogenase binding protein (E3BP) [7, 9]. All these proteins possess lipoyl domain(s) that contain a lysine residue(s) modified with lipoic acid (LA) and react with AMAs [10, 11]. It has recently been proposed that xenobiotic modification of the native lipoyl domains of AMA‐antigens may lead to loss of self‐tolerance and production of AMAs in PBC [12]. Notably, it was reported that the lipoyl domain of a PDC‐E2 peptide conjugated to 6,8‐bis(acetylthio)octanoic acid (SAc), 8‐(acetylthio)octanoic acid (OASAc), 6,8‐bis(propionylthio)octanoic acid (SCOEt) and 2‐octynoic acid (2‐OA) displayed highly specific reactivity to AMA [12, 13, 14].

A role for autoantibodies in the pathogenesis of autoimmune diseases has been reported; antibodies can traverse cell membranes and interact with intracellular proteins and subsequently induce apoptosis [15]. In the case of PBC, penetration of AMAs into cells has been demonstrated [16]. Furthermore, autoantibodies against intracellular proteins could contribute to tissue injury by forming immune complexes (ICs) that contain the intracellular proteins released from dying cells [15]. Although the presence of high concentrations of circulating ICs in PBC patients has been reported [17, 18], the existence of ICs formed by AMA and AMA‐antigens and the identity of IC‐incorporated antigens (IC‐antigens) has not been reported to date.

We previously developed an ‘immune complexome analysis’ capable of identifying specific antigens in ICs in biological fluids. The method uses IC‐capturing beads and nano‐liquid chromatography–tandem mass spectrometry (nano‐LC‐MS/MS) to comprehensively identify and profile IC‐antigens [19, 20, 21, 22, 23]. To confirm the presence of ICs formed by AMA‐antigens in PBC sera, we analyzed sera from patients with PBC as well as four other typical autoimmune diseases (SS, SLE, RA and SSc) using immune complexome analysis. Putative IC‐antigens were identified using a protein‐search algorithm that included post‐translational modification (i.e. lipoylation and aforementioned xenobiotic modification) to correctly assign MS/MS spectra into peptide sequences.

Materials and methods

Patients

Serum samples were collected from 53 female patients; PBC (n = 16; 44–87 years), SS (n = 10; 35–71 years), SLE (n = seven; 33–48 years), SSc (n = six; 57–78 years), RA (n = 14; 22–77 years). Clinical data for these patients are summarized in Table 1.

Table 1.

Characteristics of the patients with PBC and control

Characteristics	PBC	Other autoimmune diseasesa	Normal range
Number of patients	16 (all female)	37 (all female)	–
Age, years	66 (44–87)	54 (22–78)	–
Clinical stage† (I/II)	8/8	–	–
AMA (+/–)	15/1	Not performed	–
AMA titer (index)	123 (6–358)	–	< 7
PBC treatment (mg/day)	UDCA (300–1200), n = 12
	UDCA (600–900) + BF (400), n = 3
	UDCA (600) + BF (600) + PSL (5), n = 1	–
IgG (mg/dl)	1412 (754–2290)	1501 (604–2935)	861–1747
IgM (mg/dl)	218 (60–443)	112 (20–267)	50–269
AST (U/l)	31 (16–74)	21 (12–34)	13–30
ALT (U/l)	21 (12–71)	17 (8–52)	7–23
ALP (U/l)	327 (165–845)	223 (113–513)	106–322
M2BPGi (COI)	1.3 (0.2–3.6)	–	< 1

Data are counts or means and (range). ^aOther autoimmune diseases: RA (n = 14); SLE (n = 7); SS (n = 10); SSc (n = 6). ^†Clinical stage I: Scheuer’s stages 1 or 2 in liver biopsy or unknown histological stage without signs of portal hypertension or liver cirrhosis; clinical stage II: Scheuer’s stages 3 or 4 in liver biopsy or any histological stage with signs of portal hypertension or liver cirrhosis but without jaundice (total bilirubin < 2 mg/dl). ALP = alkaline phosphatase; ALT = alanine amino transferase; AMA = anti‐mitochondrial autoantibodies; AST = aspartate amino transferase; BF = bezafibrate; PBC = primary biliary cholangitis; M2BPGi = Mac‐2 binding protein glycosylation isomer; PSL = prednisolone; UDCA = ursodeoxycholic acid.

Patients were diagnosed with PBC if they met at least two of the following internationally accepted criteria [24]: biochemical evidence of cholestasis, based mainly on alkaline phosphatase elevation; presence of serum anti‐mitochondrial antibodies; histological evidence of nonsuppurative destructive cholangitis; and destruction of interlobular bile ducts. PBC patients were classified into the following two clinical stages (I and II) based on the liver biopsy and/or clinical manifestations [25]: clinical stage I (early stage) – Scheuer’s stages 1 or 2 or unknown histological stage without any signs indicating portal hypertension or liver cirrhosis; clinical stage II (late stage without jaundice) – Scheuer’s stages 3 or 4 or any histological stage with signs indicating portal hypertension or liver cirrhosis, but without jaundice (total bilirubin < 2 mg/dl). The antibody (AMA) titers to mitochondrial antigens M2 (PDC‐E2, BCOADC‐E2, OGDC‐E2) were determined using an enzyme‐linked immunosorbent assay (ELISA) kit (Mesacup‐2; MBL, Nagoya, Japan). Serum Mac‐2 binding protein glycosylation isomer (M2BPGi) level was directly measured with the HISCL M2BPGi reagent kit (Sysmex, Kobe, Japan) using an automatic immunoanalyzer HISCL‐5000 (Sysmex, Hyogo, Japan) [26]. M2BPGi levels were indexed using the following equation: cut‐off index [(COI = (M2BPGi) sample – (M2BPGi) NC/(M2BPGi) PC – (M2BPGi) NC], where (M2BPGi) sample represents the M2BPGi count of the serum sample, PC is the positive control and NC is the negative control. The positive control was supplied as a calibration solution standardized to yield a COI of 1·0. The positive control was supplied as a calibration solution. Moreover, each patient of the four autoimmune disease groups was diagnosed based on classification criteria from the American College of Rheumatology for RA, SLE and SSc, or the criteria for SS proposed by the American–European Consensus Group.

PBC sample collection and diagnoses were performed at Nagasaki Medical Center and serum samples of the other diseases were collected at Nagasaki University Hospital. All experiments were performed in accordance with the principles established by the Helsinki Declaration and with the approval of the institutional ethics committees of the Graduate School of Biomedical Sciences, Nagasaki University (approval no. 34). All patients provided written informed consent for their participation in this study.

Immune complexome analysis

A schematic diagram of this method is shown in Fig. 1. The analytical procedure was performed according to our previous study [21]. ICs were purified using three types of IC‐capturing beads (PureProteome^TM protein G‐coated magnetic beads and protein A‐coated magnetic beads, and Proceptor^TM‐sepharose beads). Each bead type (40 μl) was incubated with 10 μl of each patient’s serum diluted with 90 μl phosphate‐buffered saline (PBS) for 30 min with gentle mixing, then the liquid was removed. Further processes for isolation of antigen and tryptic digestion are described in the Supporting information.

Fig. 1.

A schematic overview of immune complexome analysis. Immune complexes (ICs) are separated and purified from serum by IC‐capturing beads (PureProteome^TM protein G‐coated magnetic beads, protein A‐coated magnetic beads and Proceptor^TM‐sepharose beads). The captured ICs are digested by papain, which selectively cleaves immunoglobulin at the heavy chain hinge region into Fab and Fc fragments, and the supernatant including IC‐antigens and Fab proteins are digested by trypsin for nano‐liquid chromatography‐tandem mass spectrometry (nano‐LC‐MS/MS) measurement. IC‐antigens are identified by database analysis with a protein‐search algorithm including post‐translational modification (i.e. lipoylation and xenobiotic modification).

The tryptic digests (peptides) (5 μl) were injected into an MS/MS instrument (Q‐Exactive MS^TM; Thermo Fisher Scientific, Waltham, MA USA) equipped with a nano‐LC system (EASY‐nLC^TM; Thermo Fisher Scientific). MS/MS spectra of digested peptides of IC‐antigens were searched against theoretical MS/MS spectra of four AMA‐antigens extracted from the public nonredundant protein database (UniProt Knowledgebase; human, 2015.01.29 download). We added differential modifications of lysine residues by lipoic acid (LA), 6,8‐bis(acetylthio)octanoic acid (SAc), 8‐(acetylthio)octanoic acid (OASAc), 6,8‐bis(propionylthio)octanoic acid (SCOEt) and 2‐octynoic acid (2‐OA) to the protein database search algorithm. Additionally, we considered modifications with reductive alkylated lipoic acid (RALA), because LA was thought to be reductively alkylated by dithiothreitol and iodoacetamide used during tryptic digestion. We obtained each molecular weight by subtracting H₂O from the molecular weight because the chemicals (LA, SAc, OASAc, SCOEt and 2‐OA) attack the ε‐amino group of lysine and form an amide bond. Chemical structures and molecular weights are shown in Table 2. Further details concerning this analysis are provided in the Supporting information.

Table 2.

Chemical structure and molecular weight of lipoic acid and xenobiotics

Structure	Name	Abbreviation	Molecular weight
	Lipoic acid	LA	+ 188·033 Da
	6,8‐bis(acetylthio)octanoic acid	SAc	+ 274·070 Da
	8‐(acetylthio)octanoic acid	OASAc	+ 200·087 Da
	6,8‐bis(propionylthio)octanoic acid	SCOEt	+ 302·101 Da
	2‐octynoic acid	2‐OA	+ 122·073 Da
	Reductive alkylated lipoic acid	RALA	+ 304·092 Da

Statistical analysis

Statistical analysis was performed using JMP^® software (SAS Institute Inc., Cary, NC, USA) to identify significant differences in laboratory data shown in Table 1; AMA titer, IgG, IgM, aspartate amino transferase (AST), alanine amino transferase (ALT), alkaline phosphatase (ALP) and M2BPGi between the groups positive or negative for AMA‐antigens. Here, patients considered positive for AMA‐antigens are defined by our detection of more than one AMA‐antigen in the patient’s sample. Based on the affinity differences toward Ig type among the three types of purification bead (PureProteome^TM protein G‐ or protein A‐magnetic beads or Proceptor^TM‐sepharose beads), we compared the IgG or IgM concentrations in the two groups obtained using each bead. The differences in laboratory data were assessed using Student’s t‐test. Statistical significance was defined as P < 0·05.

Results

Among the four AMA‐antigens (PDC‐E2, BCOADC‐E2, OGDC‐E2 and E3BP), we found that PDC‐E2, OGDC‐E2 and E3BP were IC‐antigens in PBC patients and other patients (Table 3). Three different peptides derived from OGDC‐E2 without or with modifications (LA, OASAc) were preferentially detected in PBC patients when using protein A‐beads, while three different peptides derived from E3BP without or with modifications (LA, SCOEt) were preferentially detected in PBC patients when using protein G‐beads. Among the three peptides for each AMA‐antigen, the most frequently detected peptides from OGDC‐E2 and E3BP were EAFLKKHNLK and LTESKSTVPHAYATADCDLGAVLK, respectively. One non‐modified peptide derived from PDC‐E2 (QTIPHYYLSIDVNMGEVLLVR) was preferentially detected in PBC patients when using Proceptor^TM‐sepharose beads. MS and MS/MS data of these peptides are presented in the Supporting information, Fig. S1.

Table 3.

Summary of AMA‐antigens and peptide sequences in ICs isolated from sera with each autoimmune disease using protein A, protein G and Proceptor^TM beads

Accession	Description	Position	Identified sequences	Modifications	Protein A					Protein G					ProceptorTM
					PBC n = 16	RA n = 14	SLE n = 7	SS n = 10	SSc n = 6	PBC n = 16	RA n = 14	SLE n = 7	SS n = 10	SSc n = 6	PBC n = 16	RA n = 14	SLE n = 7	SS n = 10	SSc n = 6
P36957	OGDC‐E2	268–277	EAFLKKHNLK	LA [K273; K277] or LA [K272; K273; K277]	5	1			1
		238–265	LKEAQNTCAMLTTFNEIDMSNIQEMRAR	–							2
		145–165	KTGAAPAKAKPAEAPAAAAPK	OASAc [K154; K165]; LA [K145; K152]											1
O00330	E3BP	291–314	LTESKSTVPHAYATADCDLGAVLK	LA [K295]	2	3				7	2	1	1		2	5	4	3	2
		59–78	ILMPSLSPTMEEGNIVKWLK	SCOEt [K78]	1
		342–379	QMPDVNVSWDGEGPKQLPFIDISVAVATDKGLLTPIIK	–													1
P10515	PDC‐E2	441–461	QTIPHYYLSIDVNMGEVLLVR	–						1	1				3	2		1	1

AMA = anti‐mitochondrial autoantibody; E3BP = dihydrolipoamide dehydrogenase binding protein; IC = immune complex; LA = lipoic acid; OASAc = 8‐(acetylthio)octanoic acid; OGDC = oxo‐glutarate dehydrogenase complex; PBC = primary biliary cholangitis; PDC = pyruvate dehydrogenase complex; RA = rheumatoid arthritis; SCOEt = 6,8‐bis(propionylthio)octanoic acid; SLE = systemic lupus erythematosus; SS = Sjögren's syndrome; SSc = systemic scleroderma.

In addition, we determined if there is a significant difference in laboratory data (AMA titer, IgG, IgM, AST, ALT, ALP or M2BPGi) between the groups that were positive or negative for AMA‐antigens in PBC patients. Significantly lower concentrations of IgM and IgG (isolated using protein A‐ or protein G‐beads, respectively) were observed in the PBC patients that were positive for AMA‐antigens in our analysis than in the negative patients (Fig. 2). Conversely, there was no difference in AMA titer, AST, ALT, ALP and M2BPGi between the two groups (data not shown).　In addition, there was no difference in detection frequency of AMA‐antigens between clinical stages I and II (data not shown).

Fig. 2.

Statistical analysis of immunoglobulin (Ig)G and IgM concentrations in the groups that are positive or negative for anti‐mitochondrial autoantibody (AMA)‐antigens in primary biliary cholangitis (PBC). Statistical significance was defined as P < 0·05. (a) Concentrations of IgG in the AMA‐antigen positive (n = 8) and negative (n = 8) groups isolated using protein G‐beads. (b) Concentrations of IgM in the AMA‐antigen‐positive (n = 7) and ‐negative (n = 9) groups isolated using protein A‐beads.

Discussion

The AMA‐antigens (PDC‐E2, BCOADC‐E2, OGDC‐E2 and E3BP) are among the mitochondrial proteins related to cellular energy metabolism. These proteins commonly have lipoyl domains that are believed to include immunodominant epitopes in AMAs [10, 27]. Some studies have demonstrated that sera from AMA‐positive PBC patients react more strongly with a number of chemically modified mimics (SAc, OASAc, SCOEt and 2‐OA) than with native lipoylated PDC‐E2 [12, 13, 14]. Although AMAs and their antigens have been investigated in detail, the existence of ICs formed by AMAs and their antigens has not been reported.

We hypothesized that ICs are formed by AMAs and AMA‐antigens in PBC patients. We used an original proteomic approach by obtaining MS/MS spectra of digested peptides of IC‐antigens and searching them against a protein database consisting of theoretical MS/MS spectra for four AMA‐antigens with and without lipoylation or other xenobiotic modifications. Among the four AMA‐antigens, we identified PDC‐E2, OGDC‐E2 and E3BP (Table 3); thus, these AMA‐antigens are thought to form ICs in the blood. Moreover, BCOADC‐E2 was not identified. Mutimer et al. reported that the frequency of AMA against each AMA‐antigens in PBC sera as follows; PDC‐E2 (94%), OGDC‐E2 (88%), E3BP (94%) and BCOADC‐E2 (53%) [28]. Our method did not detect BCOADC‐E2 possibly because the amount of ICs that are formed by BCOADC‐E2 and its corresponding may be relatively small. Tryptic digests (peptides) of OGDC‐E2 and E3BP with lipoylation and/or xenobiotic modification were detected, while those of PDC‐E2 without modification were detected. The peptides (position 268–277 and 145–165 for OGDC‐E2, 291–314 and 59–78 for E3BP) were not detected when we used a protein‐search algorithm that did not account for modifications. This shows that the protein‐search algorithm including lipoylation or xenobiotic modifications of lysine residues may increase the performance in searching MS/MS spectra against databases of AMA‐antigens. Among the four AMA‐antigens, E3BP and PDC‐E2 are highly homologous. Thus, it is difficult to distinguish between E3BP and PDC‐E2 by immunoassay [28, 29], which is also supported by BLAST search results showing the sequence similarity (score, 375; query cover, 95%; E‐value, 3e‐91). In contrast, our method identified these AMA‐antigens with detecting each unique peptide of them by MS/MS analysis and distinguished between them. Actually, our analysis found that the patients in which E3BP were detected were clearly different from those in which PDC‐E2 were detected, which also suggests that IC formation of E3BP occurs not only in PBC but also in other autoimmune diseases.

There was a difference in peptide detection frequency between diseases according to type of bead used for collecting ICs from sera. Protein G and protein A are proteins of bacterial origin and both have specificity to the Fc region of immunoglobulins. PureProteome^TM protein G has a high affinity to IgG₁–IgG₄ and protein A has a high or moderate affinity to IgG₁, IgG₂, IgG₄, IgM, IgA, IgD and IgE. Proceptor^TM‐beads use a protein that is isolated from Raji cells and binds to complement; thus, these beads can capture ICs that are formed by IgM and IgG and are bound to complement. In other words, protein G‐ or protein A‐beads collect ICs with and without complement, while Proceptor^TM‐beads selectively collect ICs with complement. When protein G‐ or protein A‐beads were used, the detection frequency of AMA‐antigens in PBC was higher than in other autoimmune diseases (Table 3). In contrast, when Proceptor^TM‐beads were used, the detection frequency in PBC was lower than in the other autoimmune diseases (Table 3). The different tendency of ICs detected by our method between PBC and non‐PBC patients may suggest that the pathology of PBC is related to ICs without complement that were characteristically detected in PBC patients. Absence of complement bound to ICs in PBC patients was partly supported by our analysis of the same serum samples using human proteome database instead of four AMA‐antigens. In this analysis, we found that C4b was not detected in PBC patients at all but was detected in 13 of 37 non‐PBC patients (data not shown). C4b is an important factor in complement system [30]. Deposition of ICs on tissue and initiation of immune cascade lead to tissue injury and pathogenesis in various disease [31]. Binding of complement to ICs in circulation improves IC solubilization and prevents the IC deposition [32]. Our data are not direct evidence to show that ICs containing AMA‐antigens bind to complement in PBC patients; however, it may show that the work of complement for ICs is different between PBC and other autoimmune diseases. The role of complement is complex, and this aspect does not fully explain our finding. Therefore, further studies are needed to ensure that the difference is clinically relevant.

Lipoylation is a rare post‐translational modification; however, it naturally occurs in AMA‐antigens and is required for the enzymatic activities of these proteins [33, 34, 35]. To date, only a few lipoylated proteins, including AMA‐antigens, have been identified in mammals [33, 35]. It has been reported that AMA‐antigens have specific lipoylated domains, comprising the lysine residue modified with LA; moreover, all have an immunodominant epitope containing an ExDKA (glutamic acid ‐E‐, x, aspartic acid, ‐D‐, lysine ‐K‐ and alanine‐A‐) motif [11]. However, it is somewhat unusual that an immune response occurs against such ‘normal’ lipoylated domains [11]. Rowland et al. reported that Western blot analyses of lipoylated motifs have not provided evidence for additional lipoylated proteins and peptides; however, mass spectrometric analyses are expected to be able to identify additional peptides [35]. The lipoylated peptides detected in our study (Table 3) did not contain well‐known lipoylated residues. Our results indicate the identification of new lipoylated sites on the AMA‐antigens. Lipoylation on these novel sites may not occur under normal circumstances, and abnormal lipoylation sites may cause IC‐formation and be involved in PBC pathogenesis. In addition to lipoylated peptides, OGDC‐E2 and E3BP were identified by peptides containing lysine residues modified by OASAc and SCOEt, respectively (Table 3). These xenobiotic modifications were only detected in PBC patients. The OGDC‐E2 peptide (KTGAAPAKAKPAEAPAAAAPK) modified with OASAc was only detected in one AMA‐negative PBC patient using Proceptor^TM‐beads. The ELISA kit that we used determines AMA titers to mixed mitochondrial antigens M2 (PDC‐E2, BCOADC‐E2, OGDC‐E2). However, this ELISA kit may not be able to detect antibodies against antigens with such xenobiotic modifications, even though these antibodies exist in the patient. Although the frequency of these peptides modified by OASAc and SCOEt was very low, this result shows that AMA‐antigens may be modified with these chemical molecules and may specifically form ICs in PBC.

We also determined statistically significant differences in laboratory data (AMA titer, IgG, IgM, AST, ALT, ALP or M2BPGi) between the groups that were positive and negative for AMA‐antigens in PBC patients. There was no significant difference in AMA titers between the two groups. Some studies have shown that immunological measurement of antibodies results in the more selective detection of free antibodies than antibodies forming ICs [36, 37, 38]. Therefore, the ELISA kit used here can technically determine free AMA, but not AMA‐forming ICs. It can be expected that AMA‐antigen detection and AMA titer do not correlate. A high serum level of IgG and IgM is a characteristic observation in PBC [39]. Among these two types of Ig, protein G‐beads have affinity to IgG, protein A‐beads and Proceptor^TM‐beads have affinity to IgG and IgM. Based on their affinity differences, we assessed the differences in IgG or IgM concentration between the groups that were positive or negative for AMA‐antigens identified using each bead. We found statistically significant differences in IgG or IgM concentrations between the two groups when we used protein G‐ or protein A‐beads, respectively (Fig. 2). IC formation expected in patients positive for AMA‐antigens may lead to observed decreases in free antibodies. Conversely, there was no difference in IgG concentration between the two groups when we used protein A‐beads, as well as in IgG and IgM concentrations, when we used Proceptor^TM‐beads (data not shown). In general, IC formation and its deposition leads to inflammation via interactions with leukocytes or complement and the consequent inflammation and tissue damage may lead to fibrosis. M2BPGi is a liver fibrosis marker [40, 41], and ALP, ALT and AST are markers of liver damage; however, no differences were observed in these levels between the two groups. These results showed that IC does not directly lead to liver fibrosis and liver injury. In addition, we compared the detection frequency of AMA‐antigens between clinical stages I and II. However, there was no difference between the two groups.

The results of this study show that there are ICs formed by AMA‐antigens (PDC‐E2, OGDC‐E2 and E3BP) in PBC sera. These antigens were identified by peptide modification analysis of lipoylation and the other xenobiotic modifications. Although the lipoylated sites identified by our method were different from the well‐known sites, abnormal lipoylation may cause IC‐formation and may be involved in the pathogenesis of PBC. In addition, lack of affinity of complement to ICs may affect PBC pathology. We expect that further investigation of the lipoylated sites, xenobiotic modifications and IC formation may lead to further elucidation of PBC pathology.

Author contributions

N. A., M. N. and K. O. researched the literature and conceived the study. K. O. was involved in protocol development and gaining ethical approval. M. N., M. T., H. N., A. K. and K. T. were involved in patient recruitment. N. A., N. K., M. N., N. K. and K. O. were involved in data analysis. N. A. and K. O. wrote the first draft of the manuscript. All authors reviewed and edited the manuscript and approved the final version of the manuscript.

Disclosures

None to report.

Supporting information

Fig. S1. MS and MS/MS data of identified peptides.

Fig. S2. Venn diagram showing the reproducibility of this method. Overlap number (ratio) of identified proteins from HeLa lysate peptides (n = 3).

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.