Peter A. Berg

The antigen reacting with complement-fixing antibodies in the sera of patients with primary biliary cirrhosis was localized predominantly in the mitochondrial fraction of tissue homogenates obtained by differential centrifugation. Purified mitochondrial preparations had a high content of the antigen whereas purified lysosomes failed to fix complement with PBC sera. Analysis of a number of fractionation experiments showed a high correlation between antigen content and the mitochondrial enzyme succinic dehydrogenase in all fractions. There was much poorer correlation with lysosomal and micrososomal enzyme markers. The patterns of staining obtained with a fluorescein conjugate of IgG from a PBC patient closely paralleled those obtained with a histochemical method for the demonstration of succinic dehydrogenase, further confirming the mitochondrial localization of the antigen. Staining was brightest in cells containing mitochondria with well-developed cristae. Studies on mitochondria fragmented by osmotic lysis, hexane, lysolecithin, and ultrasound suggest that the antigen is associated with the mitochondrial inner membranes.

Sera from patients with primary biliary cirrhosis reacted with four major bands in beef heart mitochondria and ATPase extract when analyzed by immunoblot after sodium dodecyl sulfate–polyacrylamide gel electrophoresis. These four immunologically reactive bands corresponded to protein bands with molecular weights of about (a) 80,000; (b) 63,000; (c) 56,000; and (d) 43,000 to 46,000. An additional immunoreactive band was found with some high–titered primary biliary cirrhosis sera at 36,000. No association with any ATPase subunits was found, except for band c which migrated between the α– and β–subunit of ATPase. Most ATPase fractions did not contain this band c, indicating that M2 determinants, as defined by immunoblot, are not identical with any ATPase subunit. Species and nonspecies–specific determinants of M2 were identified using mitochondria from rat liver and human heart and liver. Antigenic bands a, c and d were nonspecies–specific. Band b and e occurred only in beef heart. An additional determinant at about 38,000 was detected using human heart and liver mitochondria. Primary biliary cirrhosis sera showed a typical reaction with two protein bands of Escherichia coli , one at about 85,000 to 90,000 and the other at 60,000. Antibodies against both determinants could be absorbed with submitochondrial particles of beef heart showing that E. coli shares cross–reacting determinants with mitochondria. Sera from 56 primary biliary cirrhosis patients were tested using beef heart mitochondria. Fifty–four sera reacted positively with the M2 antigen in the ELISA, and these sera reacted with at least 1 of the 4 major antigenic bands: 85% were positive with band a and/or band b; 15% detected exclusively band c and none of the latter fixed complement. Antimitochondrial antibodypositive sera possessing other specificities (anti-M1, -M3, -M5, -M6 and -M7) were negative. A correlation with the precipitating antimitochondrial antibody system was not found.

The isolation of a marker antigen from rat liver and pancreas tissue, which reacts with antibodies in a subtype of autoimmune hepatitis by complement fixation test, enzyme-linked immunosorbent assay and Western blot, is described. The liver-pancreas antigen could be detected in tissue from different human or animal organs, but liver and pancreas yielded the highest activity. A highly specific antigen fraction was obtained by gel filtration and ion exchange chromatography with a 100,000 g supernatant from rat liver tissue, and this preparation was shown to be devoid of nuclear, mitochondrial and microsomal antigens and cytokeratin 8 and 18, as demonstrated by appropriate marker antibodies. These data and absorption studies with cell organelles indicate that liver-pancreas antigen is a cytosolic protein. By Western blotting, two major epitopes at molecular weights 52 kD and 48 kD could be visualized. Sera from 175 patients previously shown to have high complement-fixing activity to a nonpurified liver-pancreas antigen fraction were further analyzed. All were positive by enzyme-linked immunosorbent assay with the purified liver-pancreas antigen fractions, and 111 were also positive by Western blot. Eighty-six sera reacted with the 52-kD determinant, 33 with the 48-kD determinant and 2 with both determinants. In 117 of the 175 patients, antibody to liver-pancreas antigen was associated with other autoantibodies known to characterize subgroups of autoimmune hepatitis. Thus 19 patients had antibodies to nuclei and 96 to actin but none to liver-kidney microsomes, hereby suggesting that antibody to liverpancreas antigen may define another subgroup of autoimmune hepatitis. Only 38 (9%) of 425 patients with autoimmune hepatitis types 1 (lupoid hepatitis) and 2 (liver-kidney microsome 1 antibody-positive hepatitis) and 4 (3%) of 128 patients with chronic hepatitis B and C were positive for the antibody to liver-pancreas antigen as tested by enzyme-linked immunosorbent assay. None of the 196 patients with nonhepatic disorders had this antibody type. A classification of three different subgroups of autoimmune hepatitis is proposed: type 1a, lupoid hepatitis (antinuclear antibody positive with or without actin antibody); type 1b, only actin antibody positive; type 2, liver-kidney microsome 1 antibody positive; and type 3, liver-pancreas antigen antibody positive (with or without other autoantibodies). With this classification, group 3 comprises around 30%. (Hepatology 1993;18:1-9.

Antibodies against neutrophils have been detected in sera from patients with primary sclerosing cholangitis and inflammatory bowel diseases either by immunofluorescence or by enzyme-linked immunosorbent assay. To assess primary sclerosing cholangitis-specific antibodies, we examined sera from 30 patients with clinically and morphologically well-established primary sclerosing cholangitis by Western blotting against neutrophils and compared these results with those obtained by testing sera from patients with inflammatory bowel diseases. By Western blot using sonified neutrophils, 24 (80%) of 30 primary sclerosing cholangitis sera were positive. Five antigenic determinants at 95, 60, 55, 40 and 30 kD were visualized. Twenty-eight of the primary sclerosing cholangitis sera also showed the characteristic perinuclear fluorescence pattern by immunofluorescence on neutrophils. Thus a serological diagnosis of primary sclerosing cholangitis could be made in 80% of patients based on these two methods. In contrast, only 9% of 23 patients with ulcerative colitis and 10% of 60 patients with Crohn's disease were positive by Western blot, and these patients also showed positive perinuclear fluorescence pattern by immunofluorescence, suggesting an overlap between inflammatory bowel diseases and primary sclerosing cholangitis. Although some patients with classical primary biliary cirrhosis and autoimmune chronic active hepatitis had antibodies against primary sclerosing cholangitis epitopes, none of the patients with obstructive bile duct disorders, collagen diseases, Wegener's granulomatosis or other hepatic and nonhepatic disorders were positive by Western blot, indicating the specificity of these five primary sclerosing cholangitis-related neutrophilic epitopes.