Anti-mitochondrial antibodies (AMAs) have long been recognized as a serological hallmark of primary biliary cirrhosis (PBC). Although high titers of immunoglobulin (Ig)A AMAs are found in bile, saliva, and urine of patients, a pathogenic role for this antibody has remained elusive. Functional studies of this IgA in general have been impeded by low quantities of antibody and the inability to recover antigen-specific IgA in dimeric form. Using a newly defined synthetic group A. Streptococcus derived peptide, we purified large quantities of dimeric and monomeric IgA from patient sera. The purified IgA was incubated with Madine-Darby canine kidney (MDCK) cells transfected with the human polymeric Ig receptor (pIgR) and the cells studied by flow cytometric analysis for binding of carboxyfluorescein conjugated VAD-fmk peptide to activated caspase enzymes. A total of 87% of PBC patients that were anti-PDC-E2 positive had serum IgA that increased caspase activation in MDCK-pIgR+ cells compared to serum-derived IgA from controls with a maximum reaction 48 hours after addition of IgA. The titer of anti-PDC-E2 IgA among the PBC patients strongly correlated with caspase activation (cc = 0.88). Pre-absorption of the IgA using recombinant 2-oxo-acid dehydrogenase complex significantly diminished this activation. IgG from the same PBC patients did not induce caspase activation. These data suggest that during transcytosis through pIgR-positive cells, exposure to PDC-E2-specific dimeric IgA results in the initiation of caspase activation. In conclusion, we propose that due to an even greater concentration of dimeric IgA in biliary and mucosal secretions, constant transcytosis would render the exposed cells more susceptible to apoptosis resulting in subsequent bile duct damage. (HEPATOLOGY 2004;39:1415–1422.)
Primary biliary cirrhosis (PBC) is an autoimmune liver disease manifested by chronic nonsuppurative destructive cholangitis and the presence of antimitochondrial antibodies (AMAs).1, 2 Although AMAs are found in more than 90% of patients with PBC, there has not been a specific pathogenic role defined for these autoantibodies, which often are detectable for several years before the onset of clinical disease.3 Similarly, although infiltrating T cells have been found in the small bile ducts of PBC, the mechanisms involved in the destruction of biliary epithelial cells remain enigmatic.4 Notwithstanding this uncertainty, the immune response in PBC and the classification of effector cells and molecules associated with disease strongly suggests that PBC is a mucosal disease.5 In fact, patients with PBC often have concurrent Sjögren syndrome.5 Thus, PBC can be considered a generalized epithelitis.
One of the common features of biliary and salivary epithelia is the expression of polymeric immunoglobulin receptors (pIgR) involved in the transcytosis of IgA across epithelium. IgA is the major immunoglobulin isotype involved in mucosal immunity. The concentration of IgA in serum is much lower than IgG, but the amount of total body IgA exceeds that of IgG due to local production in mucosal tissues, including the liver. IgA AMAs are readily detected in the bile, saliva, and even the urine of patients with PBC.6–8 Previous work involving confocal microscopy studies of Madine-Darby canine kidney (MDCK) cells expressing pIgR9, 10 has demonstrated that AMA IgA co-localizes, and thus potentially interacts with, PDC-E2 in the cytoplasm of these cells. A limitation of the studies to date has been the absence of functional effects of IgA AMAs on pIgR positive cells, due in large part to the difficulty in obtaining enough polymeric IgA. We therefore have taken advantage of a new methodology for isolating intact polymeric IgA from sera. We report herein that IgA, but not IgG AMAs, induce caspase activation in MDCK-pIgR+ cells. These results argue not only for an important clinical role of IgA autoantibodies in PBC, but also for a pathogenic function of AMAs in disease.
Streptococcal derived IgA binding peptide (Sap)11 is composed of 50 amino acids (aa) derived from the M protein of Streptococcus pyogenes. It is known to not only be an important virulence factor12 but has also been shown to specifically bind the Fc portion of IgA.13, 14 The Sap peptide sequence is YYALSDAKEEEPRYKALRGENQDLREKERKYQDKIKKLEEKEKNLEKKSC, which corresponds to aa 35–83 of protein Sir22 (M22) with an additional C-terminal cysteine. This efficient peptide binds IgA while retaining its full biological features.11 We also confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and immunoblotting that purified IgA contains both monomeric and dimeric forms, with a ratio similar to that seen in serum IgA. Because of its efficient binding of IgA, this method is considerably better than using the more conventional Jacalin columns. Similarly, IgG antibodies were purified from the sera of the same PBC patients using Hitrap protein G column (Amersham Biosciences, Piscataway, NJ). Sera were diluted 1:3 in phosphate-buffered saline (PBS) and applied to the column using the Prime FPLC system (Amersham Biosciences). After washing with PBS, IgGs were eluted by 100 mM Glycine buffer pH 2.7. The eluted IgGs were immediately neutralized by 1 M Tris-HCl buffer pH 8.0. Purified IgGs were concentrated, dialyzed, and sterilized as described in the purification of IgA.
We studied a total of 36 subjects: 12 normal controls (C1–12), eight patients with PBC that were anti-PDC-E2 antibody positive (P1–8), four patients with PBC that were anti-PDC-E2 antibody negative but positive for BCOADC-E2 or OGDC-E2 (N1–4), six patients with PBC that were AMA negative (A1–6), and six liver disease controls including two patients with primary sclerosing cholangitis (PSC) (L1–2) and four patients with nonalcoholic steatohepatitis (NASH) (L3–6). All samples were collected after informed consent and approval by the Institutional Review Board; none of these randomly selected samples had been previously studied. Briefly, 1.8 ml of human plasma was centrifuged at 14,000 rpm for 20 minutes and diluted 1:3 with PBS. After filtration through a 0.2 μm filter, diluted plasma was applied to a 1-ml Hitrup NHS-column with 5 mg of immobilized Sap. The bound IgA was eluted by 0.1 M sodium acetate, pH 4.0, followed by immediate neutralization with 1 M Tris-HCl buffer, pH 8.0. All purification steps were performed using an ÄKTA prime FPLC system (Amersham Biosciences). IgA was concentrated using an Amicon Ultra Centrifugal Device (Millipore Corporation, Bedford, MA) and then dialyzed against serum free culture medium, followed by sterilization using 0.2 μm filter. The final concentration of IgA ranged from 5 to 8 mg/ml. IgA reactivity to PDC-E2, BCKD-E2, and OGDC-E2 was determined in triplicate by enzyme-linked immunosorbent assay (ELISA). Purified IgA fractions were adjusted to a concentration of 1 mg/ml, diluted to 1:1000, and read at 405 nm. Titer was expressed at OD at 405 nm.
Induction of Apoptosis.
The MDCK-pIgR+ cell line was cultured in BD Falcon™ Cell Culture Inserts (BD Biosciences Discovery Labware, Bedford, MA) using a transwell system. Media consisted of DMEM supplemented with 10% fetal calf serum (FCS).9, 10 The inserts were composed of thin membranes of polyethylene terephthalate with 1.0 μm pores, 1.6 × 106 pores/cm2. The cell cultures were maintained at 37°C with 5% CO2 in a humid incubator. After growth to confluency was achieved, serial dilutions of the Ig preparation to be tested were inoculated in the lower chamber. These test Igs included highly enriched preparations of IgA, IgG, and sera depleted of IgA (i.e., IgM and IgG enriched). Cells were harvested for analysis 48 hours later. Additional controls for these experiments included the use of wild-type MDCK cells, i.e., without pIgR transfection. Finally, to further confirm the importance of receptor mediated uptake of IgA, Hela and T2 cells were also used as additional controls.
Antigenic Specificity of Apoptosis.
Purified IgA from both patients and controls was absorbed with five different antigens to demonstrate specificity: α recombinant PDC-E2, BCOADC-E2, OGDC-E2, pMIT3,15 the hybrid recombinant protein of three E2 components, or an irrelevant control antigen. Briefly, recombinant proteins were expressed in an E. Coli strain, DH-5α, as glutathione S-transferase (GST)-fusion proteins. A pellet collected from 500 ml of E. Coli culture was sonicated in lysis buffer (1% triton-X and 1% Tween 20 in PBS) and the soluble fraction applied to 2 ml of PBS saturated Glutathione-Agarose gel (Sigma, St. Louis, MO). This agarose gel, combining each recombinant protein via GST-glutathione interaction, was packed in a polypropylene column (Pierce, Rockford, IL) and directly used to absorb IgA. Purified IgA was applied to the column five times and the loss of reactivity against each 2-OADC-E2 was confirmed by ELISA.
To define the events involved in the apoptosis of the pIgR MDCK cells, carboxyfluorescein labeled VAD-fmk (FAM-VAD-fmk) (Cell Technology Inc, Minneapolis, MN), which specifically binds to activated caspases, was used as a probe. Briefly, FAM-VAD-fmk was dispersed into the upper chamber of the transwell at the final concentration of 10 μM. After a 1 hour incubation at 37°C with 5% CO2 in a humidified incubator, cells were washed twice with PBS supplemented with 2.5% bovine serum albumin (BSA). After detachment, using 0.05% Trypsin-0.53 mM EDTA (GIBCO™, Invitrogen Corporation, Carlsbad, CA), cells were washed and analyzed by FACScan (BD Biosciences, San Jose, CA). To clarify which caspases were activated, probes specific for activated caspase 3, caspase 8, and caspase 9 (Cell Technology Inc.) were used (FAM-DEVD [asp-glu-val-asp]-Fml, F-LETD [leu-glu-thr-asp], F-LEHD [leu-glu-his-asp], respectively).
Caspase activation was evaluated using the geometric mean value of fluorescence. Thus, activation was expressed as: caspase activation index = [(geometric mean value of IgA treated cells – geometric mean value of nontreated cells)/geometric mean value of nontreated cells] × 100 (%). Student t test was used for the statistical comparison of each group. In all experiments, positive and negative controls were included.
Caspase Activation of MDCK Cells.
MDCK-pIgR+ cells treated with IgA from patients with PBC demonstrated increased caspase activation (Fig. 1). In contrast, IgA purified from control sera showed no increase, nor did wild-type MDCK cells respond to IgA from patients. This caspase activation peaked at 48 hours after exposure to antibody (Fig. 1A), and maximal activation of caspases occurred at 1 mg/ml of IgA (Fig. 1B). All subsequent experiments were performed using this concentration of antibody for 48 hours of incubation time. We defined activation as a value lying three SD outside the mean activity of cells exposed to control IgA from normal patients. Using this cutoff, and focusing on PBC patients who were anti-PDC-E2 antibody positive, IgA derived from 7/8 (87.5%) patients increased caspase activation. In contrast, no activation was found using IgA purified from six AMA negative PBC patients, three anti-PDC-E2 antibody negative PBC patients, six liver control patients, or 12 control sera (Fig. 2). Enhanced caspase activation was limited to patients with anti-PDC-E2 IgA and required pIgR expression. For example, IgA did not enhance caspase activation in HeLa or T2 cells (Fig. 3).
Correlation Between Caspase Activation and Anti-2-OADC IgA Titer.
For each AMA and patient sera, the titer of purified IgA against each E2 component of 2-OADC, and the degree of caspase activation in treated MDCK-pIgR+ cells, IgA titer was measured at an absorbance of 405 nm using each recombinant protein as antigen (Table 1). Interestingly, the titer of anti-PDC-E2 IgA strongly correlated with the degree of caspase activation index (correlation index = 0.883, P = .0001) (Fig. 4A). The titers of IgA anti-OGDC-E2 and anti-BCOADC-E2 were much lower and only very weak correlations between OGDC/BCOADC-E2 IgA antibody titers and caspase activation were observed (correlation index = 0.365, –0.1, respectively. P = .2436, .7567, respectively) (Fig. 4B, C). Enhanced caspase activation in MDCK-pIgR cells is dependent on the titer of anti-PDC-E2 IgA.
AMA Specificity of Caspase Activation.
To confirm that caspase activation was due specifically to IgA AMAs, purified IgA were absorbed with recombinant 2-OADC proteins to deplete antimitochondrial reactivity and the loss of antimitochondrial reactivity confirmed by ELISA (data not shown). Absorption of IgAs from AMA and patients with pMIT3, which removed nearly all antimitochondrial reactivity, abrogated the increased caspase activation when incubated with MDCK-pIgR+ cells (Fig. 5). Looking at the autoantigens individually, IgA from AMA and patients absorbed with PDC-E2 decreased caspase activation compared to unabsorbed IgA. Absorption with BCOADC-E2 or OGDC-E2 elicited only a modest decrease in caspase activation. Absorption by the control protein, GST, did not alter caspase activation. Hence, caspase activation in pIgR MDCK cells required IgA that was PDC-E2 specific (Fig. 5). Furthermore, use of the same concentration of IgG, purified from the same patients and run in parallel, did not increase caspase activation (Fig. 6). As IgM is also known to be taken up by the pIgR, albeit much less efficiently than IgA, we also investigated the effect of IgM AMA. Sera, depleted of IgA, that still contained IgM (and noncaspase activating IgG) were added to the culture so that the final concentration of human IgM was 1 mg/ml. Data from these studies showed that sera containing IgM did not increase caspase activation as IgA did, indicating the exclusive role of IgA in the activation of caspases (Fig. 6). However, we should note that due to the decreased efficiency of IgM uptake by pIgR, a longer incubation period might be necessary to achieve effects similar to IgA. However, such an increase in incubation time would greatly increase background caspase activation in these cells.
Specificity of Caspase Enzymes Activated by AMA-IgA.
The probe used in the above experiments, FAM-VAD-fmk, detects the overall activation of several caspases. To identify which caspase enzymes were activated in the IgA treated MDCK-pIgR+ cells, caspase specific FAM-peptides were used: FAM-DEVD-fmk for caspase 3, FAM-LETD-fmk for caspase 8, and FAM-LEHD-fmk for caspase 9. Caspases 3 and 9 were preferentially activated by AMA positive IgA in MDCK-pIgR+ cells (Fig. 7).
One paradox in autoimmunity has been the presence of high titer autoantibodies directed at intracytoplasmic antigens.2 The apparent difficulty of autoantibodies coming into contact with intracellular autoantigens has fostered the argument that many such autoantibodies are merely epiphenomena and thus have no pathogenic role. Thus, there has been considerable debate for nearly 40 years regarding a pathogenic role for AMAs in PBC. Although the antimitochondrial response is both highly directed and specific to patients with PBC, there has not been any correlation with disease activity and antibody levels, isotypes, or specificity. Further, there has been no direct demonstration of a pathogenic role for antibody. Most studies have focused on IgG AMA.16 Attempts to define a pathogenic role for antibody, particularly IgA, have been stymied because IgA is difficult to isolate, particularly when enriched for antigen-specificity. The development of recombinant IgA was not beneficial to this work due to the low yield of expression systems.10 New technology using the Sap peptide for the isolation and purification of functionally intact IgA has enabled new studies concerning the role of this Ig isotype in mucosal disease.11 For the assay system, we chose the pIgR expressing MDCK cell model because a significant body of data already exists on the use of IgA against HIV and Rota virus in infected cells.17–19 This provided a well-studied and validated methodology.
Using these newly developed IgA reagents, we demonstrate herein a specific proapoptotic effect of IgA AMA in vitro. Such activity is found in the majority of patients with PBC and strongly correlates with the IgA titer to PDC-E2. Although the magnitude of activation is relatively small, it is nonetheless highly statistically significant. This relatively small magnitude takes on additional significance because the destruction of biliary cells in patients with PBC is itself a slow process and it would be unlikely that the presence of IgA AMA would cause an immediate destruction of biliary cells. Rather, we believe the kinetics of biliary cell destruction and repair are an ongoing process involving the kinetics of not only the titers of local IgA AMA, but also the degree by which it is uptaken and interferes with cell function. The presence of the polymeric Ig receptor on human bile duct cells is critical to this activity because use of wild-type MDCK, Hela, and T2 cells demonstrated no such IgA activity. Moreover, the lack of IgG induced activation is likely due to the inability of this isotype to enter the cells under study. While IgM is capable of pIgR uptake, lower efficiency of transport into the cell coupled with the lower affinity of this isotype for PDC-E2 and a short exposure time may preclude caspase activation in our culture system. We also note the issue of AMA-negative PBC20 and suggest that the destructive process must be by other means. However, while serum AMA is absent, this does not rule out the local production of IgA AMA. We also note case reports of patients with PBC that are serum IgA deficient.21 Such patients may have IgM AMAs, which may participate in the pathogenesis of mucosal lesions; pathogenic CD4 and CD8 T cells may also play a role.22–25 Hence, we do not mean to imply that IgA AMA are the sole pathogenic mechanism involved in biliary cell destruction. In fact, we believe that the biliary cell is an innocent victim of an orchestrated immune response involving several limbs of the immune system.2 The destruction of biliary cells probably involves the recognition of an intact autoepitope recognized by CD4 and CD8 cells, as well as by autoantibody; this appears to be unique to biliary cells.2, 26, 27
Two main pathways exist for caspase activation: activation via cell surface death receptors or via intracellular stress. It is intriguing to speculate that during transcytosis anti-PDC-E2 IgA might induce cellular stress by interacting with newly synthesized PDC-E2 migrating from the endoplasmic reticulum (ER) to the mitochondria. However, given the dependence of IgA AMA induction of caspase activity on cell membrane expression of pIgR, it is conceivable that caspase activation by IgA AMA is mediated via a cell surface receptor. The prolonged time delay in peak caspase activation is more typical of intracellular stress. Activation of caspase 8 is associated with cell surface receptor mediated apoptosis, whereas caspase 3 and 9 activation may occur in either pathway. The relatively greater increased activation of caspase 3 and 9 compared to caspase 8 also suggests increased caspase activation was due to intracellular stress and was not cell surface receptor mediated. After its activation, caspase 3 may activate other caspases, such as caspase 8. Of course, for some cell surface receptors, activation of caspase 10 rather than caspase 8 may predominate. Increased caspase activation by IgA AMA is most likely due to increased intracellular stress.
We also note that caspase 3 and caspase 9 are preferentially activated by IgA AMAs. Disruption of mitochondrial function induces cytochrome c release that in turn activates caspase 9, which rapidly activates caspase 3. Although this is an in vitro manipulated cell system, the data suggest that IgA AMAs are one factor responsible for the destruction of biliary epithelial cells. These results provide a plausible scenario for the mechanism of tissue damage based on the transit of IgA through intrahepatic biliary cells. Once inside the cell, anti-PDC-E2 antibody could bind to nascent PDC-E2 during transport to the mitochondria initiating the activation of the caspase enzymes with ensuing apoptosis, either by depletion of PDC-E2, or through a danger signal stimulated by a buildup of intracellular immune complexes. Thus, there is a constant low-grade destruction of biliary cells. Over time, the functional reserve of the tissue to replace itself is lost, perhaps because of loss of the local stem cell population, or because of synergistic activity of other immune effectors including CD4 and CD8 lymphocytes.22–25 Eventually, the ducts are destroyed and the liver becomes cirrhotic. Although AMAs of the IgA isotype are not found in every patient's bile, saliva, or urine, we have been impressed by the significant frequency of detection in fluids that are difficult to analyze because of volume, dilution issues, the presence of detergents, and proteases. We analyzed in the past the secretory immune system in a patient with PBC who was serum IgA deficient and found that this patient had IgA AMA in both their saliva and bile. IgA has been implicated in the pathogenesis of several autoimmune diseases, including linear IgA disease (LAD), intercellular IgA dermatosis (IAD), IgA pemphigus, dermatitis herpetiformis (DH), Henoch-Schonlein purpura (HSP), and IgA nephropathy. Mechanisms in these IgA-dependent diseases range from immune complex deposition (IgA nephropathy and HSP) to IgA-mediated damage in LAD, IgA pemphigus, and dermatitis herpetiformis.28, 29 Interestingly, there have been several receptors described for IgA, some of which are unique to epithelial cells such as those lining the mucosal tissues. The receptors so far described include s-IgAR,30 ASPGR,31 pIgR,32, 33 FcR or CD89,34 α-galactosyltransferase,35 mesangial cell receptor,36 and a recently described novel receptor for IgA expressed by human intestinal epithelial cells.37 While each of these receptors have been shown to bind IgA, the nature of the intracellular signals induced upon binding IgA and the role of each of these receptors in internalization of IgA has yet to be defined. We mention this because the concept that autoantibodies can penetrate living cells has been controversial but provocative.38–41 What is still not understood is the differing rates of progression of disease in patients with apparently similar levels of IgA. It may be that there is micro-heterogeneity in the antibody response so that some molecules are more efficient in caspase activation. In the data reported herein, caspase induction was seen with anti-PDC-E2 antibodies only and did not correlate with autoantibodies to OGDC-E2 or BCOADC-E2. It is possible that these differences reflect a much higher titer of autoantibodies to PDC-E2 compared to OGDC-E2 or BCOADC-E2 not only in the sera, but also in the secretions in PBC.7, 8, 42, 43 It would be interesting to prospectively study patients ranked on their capacity to induce caspase activation in this system and determine whether this predicts their progression of disease.