Autoantibodies to BCOADC-E2 in Patients With Primary Biliary Cirrhosis Recognize a Conformational Epitope

PATRICK S. C. LEUNG,' DAVID T. Cm?AbiG,' R. M.m W ~ J N , ~ SANGHOON CHA,' DEAN J. DANNEII, ' AFTAB ANSAKI,' Ross L. cOk>i)E12,' AND M. ERIC GERSHWIN'

Primary biliary cirrhosis (PBC) is an autoimmune dis- ease of liver associated with a unique serologic response to mitochondrial autoantigens. Many of the autoanti- gens recognized by autoantibodies in PBC are members of the 2-0x0-acid dehydrogenase complex. The two major autoantigens are the E2 component of the pyruvate de- hydrogenase complex (PDC-E2) and the E2 component of the branched chain 2-0x0-acid dehydrogenase com- plex (BCOADC-E2). The autoantibody response to PDC- E2 has been mapped to one immunodominant epitope, which consists of both linear and conformational compo- nents. The presence of a single immunodominant epi- tope in PDC-E2 is unusual when contrasted to the im- mune response to autoantigens in other human autoimmune diseases. We have mapped the epitope rec- ognized by antimitochondrial autoantibodies (AMA) spe- cific to BCOADC-E2 in patients with PBC by taking advantage of the full-length bovine BCOADC-E2 comple- mentary DNA (cDNA) and a series of expression clones spanning the entire molecule. Reactivity to the various expression clones was studied by immunoblotting, en- zyme-linked immunosorbent assay (ELISA), as well as selective absorption of patient sera by expressed protein fragments. Autoantibodies to BCOADC-E2 map within peptides spanning amino acid residues 1 to 227 of the mature protein; our data demonstrate that the epitope is dependent on conformation and includes the lipoic

Abbreviations: AhlA, antimitocliondnal antibodies: PM', primary hiliary PDC-E2. pyruvate dehydrogenase complex E2: H('OAIX-EZ.

chain 2-oxo-acid dehydrogenase complex E2; cDNA. complenientary DNA: IPTG. isopropylthiogalactosid~; SDS-PAGE, sodium dodticyl siilrate piilyacrylamidr gel clectrophoresis: ELISA. enzyme-linked iinnlui1~isorbent assay

From the 'Division of lilieumatology. N l e r ~ and Clinical Ininiunology. University of Chlifornia. Ihvis. CA; 'Departnient or Riochemistry, Ilnivcrsity of Texas, Southwestern Medical Center, Dallas. TX: "Departnicmt of' Genetics and Molecular Medicine. Emory I J n i i t y School of lqedicine, Atlanta. GA; the "Department of Patholob!, Emor nivrrsity. Atlanta. GA: and the ',De- partment of Microbiology, hlonash University. Clayton. Victoria 3168. Auslra- lia.

Received November 28. 1994 This study was supported by NIH grants DI< 39.588 and A1 31535 and the

Liver Tissue Procurcwicnl a n d Distribution Systeni Grant 1-DK-6-227.1 Dr Cha's current address is the Ilepartnicnt or Riology and Technology.

Kangweon National University. ('hoonchwn, South Korea. Address reprint requests ti): Patrick S. ('. Leung, Phl). Division dRheuiii ; l-

toloCy. r'dlwgy and Clinical Immunolomv. L n i w r s i t y of ('alifornia at Davis, Ti3 192. School of Medicine, D a v i s ~ (:A 95616

Copyright c ' ~ 19116 by the American Association for the Study or 1,iv.t.r Diseases.

02iO-9 139/95/2202-00 11)$3.00/0

acid binding region. However, only the full-length clone (amino acid residue 1 to 421) is sufficient to remove all detectable BCOADC-E2 reactivity. Moreover, the ab- sence of lipoic acid on the recombinant polypeptides used in this study indicates that antibody binding to BCOADC-E2 is not dependent on the presence of lipoic acid. (HEPATOLOGY 1995; 22:505-513.)

Antimitochondrial autoantibodies (AVA) have been long recognized as a characteristic feature of primary biliary cirrhosis (PBC).' The major autoantigens recog- nized by these autoantibodies a re members of the 2-0x0-acid dehydrogenase comples, including the py- ruvate dehydrogenase complex E2 (PDC-EB 1, the branched chain 2-0x0-acid dehydrogenase complex E2 ( BCOADC-E2 1, the 2-0x0-glutarate dehydrogenase complex E2, the E l subunits of the pyruvate dehydro- genase complex and protein X of PDC. As for many autoantigens, the structure and antigenicity of the ma- jor PBC autoantigens are conserved among different species.'." Of these autoantigens, PDC-E2 is most com- monly recognized by serum antibodies of patients with PBC, and much of the research in PBC has focused on determining its reactivity with AMA. Sera from pa- tients with PBC inhibit catalytic function of PDC-E2,".' and the immunodominant epitope has been localized to a stretch of 93 amino acids within the inner lipoic acid domain of the enzyme? However, in addition to antibodies to PDC-E2, approximately 60(Z of patients with PBC are also immunoreactive with BCOADC-E2. Indeed, in some studies, there are 10% to 20% of pa- tients with PBC who react to BCOADC-EB but have no detectable reactivity to PDC-E2."".'

The BCOADC complex catalyzes the oxidative decar- boxylation of alpha keto acids derived by transamina- tion of the branched-chain amino acids, valine, leucine, and isoleucine.9 This complex consists of three catalytic components: a branched-chain 2-0x0-acid decarboxyl- ase ( E l ) , a dihydrolipoyl transacylase tE2), and a dihy- drolipoyl dehydrogenase iE3).'" Of these components, autoantibodies in PBC are limited to the dihydrolipoyl transacylase or BCOADC-E2. The complementary DNA (cDNA) of both the human and bovine BCOADC- E2 have been cloned,"" and primary structures de- rived from sequence cDNA demonstrate tha t the ma- ture BCOADC-E2 is similar to PDC-E2 in having a lipoic acid binding domain, dihydrolipoyl dehydroge-

505

506 LEUNG ET AL HEPATOI.O(;Y August 1995

nase (E3) binding domains, an inner E2 core, and hinge regions.' Confonnationally, the BCOADC is organized around a structural core consisting of E2 subunits to which E l , E3, and specific kinases are attached. The E2 subunit consists of an inner E2 core in a high assem- bled octahedral symmetry.12 This structure is very sim- ilar for other members of the 2-0x0 acid dehydrogenase complex. 13.14

The identification of the E2 components of the 2-0x0- acid dehydrogenase complexes as the mitochondria1 au- toantigens in PBC and the characterization of the PDC- E2 immunodominant epitope as the lipoyl domain, together with the distinctively similar subunit struc- tures among the 2-0x0-acid dehydrogenase complexes, has suggested that the lipoyl domain is a common epi- tope. However, the E2 proteins of each complex are recognized by a defined population of autoantibodies and cross-reactive antibodies that recognize several 2- 0x0-acid dehydrogenase E2 proteins have not been demonstrated. To examine the epitope recognized by antibodies specific to BCOADC-E2, we have taken ad- vantage of the full length bovine BCOADC-E2 cDNA and mapped the epitope recognized by autoantibodies using a series of expression clones spanning the entire molecule.

MATERIALS AND METHODS

Sera. Sera from 13 patients with PBC were chosen for this study of epitope mapping. These sera have been previously studied in detail by immunoblotting against beef heart mito- chondria." Nine sera were shown to have antibodies to bo- vine BCOADC-E2 by immunoblotting. The remaining four sera were immunoreactive to PDC-E2, but did not react to BCOADC-E2. In addition, sera from patients with progres- sive sclerosing cholangitis (n = 3), chronic active hepatitis (n = 3), and normal volunteers ( n = 6) were used throughout as negative controls. Rabbit anti-BCOADC-E2 antibodies were obtained as described p r e v i ~ u s l y . ' ~ ~ ' ~

cDNA Clones. A recombinant plasmid pkkbE2-llCD9 con- taining the mature protein of bovine BCOADC-E2 was ex- pressed in plasmid pKK233-2. '' Expression subclones of BCOADC-E2 were constructed from pkkbE2-11, a cDNA clone encoding the full-length bovine BCOADC-E2 cDNA.'"'~ Recombinant clones containing amino acid residues 114-42 1 and 208-421 were obtained by digesting the pkkbE2-11 with PstI. The digested DNA was relegated and transformed into Escherichia coli. Ampicillin-resistant clones were selected and further analyzed by plasmid DNA isolation and re- striction enzyme digestion. Other expression subclones were generated by Ba131 deletion. Briefly, the 2283 bp HindIII- Hind111 fragment of pkkbE2-11 was subjected to Ba131 diges- tion. The Bal31-digested DNA was blunt ended with T4 DNA polymerase and then digested at a unique internal KpnI site. The digested DNA was then separated on agarose gel and fragments of 350 to 550 bp isolated. These fragments possess a blunt 5' end and a 3' KpnI site; they were ligated with a derivative of pkkbE2, which had been digested with PstI, blunt ended, and digested with KpnI. The ligated DNA were then transformed into competent E coli. Ampicillin-resistant colonies were chosen for further DNA isolation and restric- tion enzyme digestion.lfi The identity of each expression clone was confirmed by nucleotide sequencing using the dideoxy method."

In addition to these fragments, the lipoic acid domain of BCOADC-E2 was expressed i n pMal, a maltose binding vec- tor (New Englands Biolabs, Beverly, MA). The lipoic acid domain of BCOADC-E2 was amplified by the polymerase chain reaction; primers were designed such that codons corre- sponding to amino acid residues 1 to 84 of the mature protein were amplified. The amplified DNA contained a StrcI site a t the 5' end and a stop codon with PstI site a t the 3' end. The amplified DNA was ligated into the StuI site and PstI site of the pMal vector. Similarly, an expression clone containing amino acid residues 1 to 115 of the BCOADC-E2 was cloned into the StuVEcoRI site of pMal; expression clones containing amino acid residues 1 to 227 and 84 to 227 of mature BCOADC-€32 were expressed in the EcoRI site of pMal. Ampi- cillin-resistant transformants were sequenced. Thus, the clones studied included plasmids containing amino acid resi- dues: 1 to 421, 1 to 84, 1 to 115, 1 to 227, 84 to 227, 115 to 421, 132 to 421, 167 to 421. 179 to 421, 183 to 421, 188 to 421,208 to 421,209 to 421 (Fig. 1). As controls, an expression clone of the human PDC-E2 lipoyl domain was prepared as

In addition. a n irrelevant expression clone of a ra t liver cDNA specific for the alloantigen F was used throughout.

Expression of cDNA Clones. Competent cells of E coli XL1 blue were przpared as described'' and transformed with plas- mids of the BCOADC-E2 cDNAs. The transformed cells were grown at 37°C overnight in Luria-Bertani media containing 25 pg/mL ampicillin, diluted 1:lO. and induced with 2 mmol/ L of isopropylthiogalactoside (IPTG) for 4 hours. Cells were harvested and analyzed by sodium dodecyl sulfate polyacryl- amide gel electrophoresis (SDS-PAGE) as described below. The presence of induced proteins was compared with wild- type plasmid controls. Expression of recombinant peptides was detected by rabbit anti-BCOADC-E2 antibodies1"I6 or sera from patients with PBC as described below.

Incorporation of Lipoic Acid into E Coli. DL-[2-'"H.I lipoic acid (1.95 mCi/mg or 401 mCi/mmol) was prepared by a cata- lyzed exchange procedure involving metalation of lipoic acid with lithium a t C-2'0.21 followed by treatment with tritiated water in dimethylformamide." The product was high-perfor- mance liquid chromatography purified and stored in aqueous ethanol (1:l) solution a t 20°C. E coli containing the appro- priate plasmids were grown in LB medium containing 25 pg/ mL ampicillin. Lipoic acid incorporation was performed by the addition of 10 pg of DL-I.2-"H1 lipoic acid to the culture at the same time that IPTG was added. Extracts of the induced culture were resolved on SDS-PAGE, stained with Coomassie Blue. treated with Enlightening (Du Pont-New England Nu- clear, Boston, MA), dried, and exposed to x-ray film to detect the presence of lipoylated proteins. Alternatively, the pMal lipoic acid domains were expressed in the presence of "1- lipoic acid, purified by amylose resin, and assayed via scintil- lation counting.

Absorption of Antibody Reactivity. The specificity of the BCOADC-E2 expression clone was studied by absorption of PBC sera with crude lysates of E COIL containing either plas- mid pKKbE2-11DC9, or its expression clone deriv a t ' ives or a plasmid control. Briefly, PBC sera at 1:lOOO dilution were incubated overnight with 10 mg/mL of sonicated lysates of the expression clones. The absorbed sera was spun for 2 min- utes at 10,OOOg and probed against a blot of beef heart mito- chondria."' To study the inhibition of anti-RCOADC-E2 auto- antibody activity by the expression BCOADC-E2 subclones, PBC sera were diluted a t lo-,'' to 10 ' with 3!4 milk in PBS and thereafter, 10 mg of sonicated lysates of each of the ex- pression clones were added per milliliter of the diluted sera.

97 -

66 -

45 -

In addition. t h e following combinations of lysatcs wcrc' also includccl; thcy were 11 to 84) I (84 to 227). (1 to 115) - (84

t (115 to 321 1. and ( 1 to 227) + (115 to 421). Absorption w a s carried out a t 4'C overnight on a rocker. The ahsorhed sera were then p i ~ h e d against the maturci BCOAL)C-E2 clone by hoth imniunohlotting and enzyme-linked immunosor1)eiit assay (ET,TSA). As controls, scra w c w absorbed with I P T G induced F: c d i XL1 blue containing an expvession clone of P D C E 2 . the plasniid itself or a n irrc>Icvant clone, thc F allo- an ti gen . ' "

SDS-PAGE wid Imtnunobloffin,q. Beef hear t mitochondria

;icrylaniide gel. Briefly. 200 pg of beef hcxart mitochondria were boiled fhr 5 minutes in SDS-sample Imffcr and scpa- rated el(:ctrophoreticaIly on a 10% acrylamide gel. The re- solved protein was transferred onto nitrocellulose f i l ters. Tlircc-millimeter s t r ips were cut a n d t h e strips were first blocked with Y; nonfat dry milk in PHS ("blotto") for 30 minutcs a t room tenipcwiture heforc. incubating Lvith sera ( 1:1000 dilution 1 for 1 hour. The filters were then washed with I'BS containing 0.05';; Tween for three times. 10 min- utcs each. and then foljowcd by ~ l l ~ l l h ~ i t ~ ~ J l l with I'"-labc.lcd ant i -human immunoglohulin t Amerskiam. 11,) fbr 1 hour. The strips \vvcre then ivashcd 21s before a n d exposed to x-ray film. Ibactivity to mitoc:hondrial proteins in nhsorbcd and unnh- sorbed sera u'as scored.'.! Sirnilarly, lysatcs from IPTG-in- duccd B(.X)ALK:-EB cyxess ion cloncss were resolved under denatur ing conditions by 10'zi SDS-PAGE. Thc resolvd pro- teins wei'e transfcrrcd onto nitroct~llulosc filters. iIft.er hlocking with hlotto. thc filters wei'c1 incubated with PBC

to 227,, ( 1 10 84) .. (84 10 227) + (115 to 4211. I1 to 11.5)

W E ~ C pr~>p:~red':' i1lld resolvcd by SDS-PAGE 011 a lo!; ~ 0 1 . ~ -

sc:ra t h a t had b c w i previously absorbed with E coli SL1 blue lysatc. 'I'he reaction conditions ha\~c bcc~n clescribccl in de- tail.'.' '

ELISA. The reactivity of BCOAD(.:-E2 reactive PBC: sera, non-RCOAD('-EX reactive PI3C sc~ra. and normal s t w were tested against t h e B('OAlWE2 expression clones. f!: coli ly- sa tes of 1 PT( -i n d uced DCOAL)C-E2 e x pre ssi oi 1 clones were suspended in carbonate huI'fbr a n d coated onto microtiter platc:s a t 100 pg!ml at 4Y: overnight. After blocking with 1C.i hov inc se r u 111 a1 bun1 i n in P 1 <S. the plat c k s wc'rc i n cu ba ted \vith sera ( 1:lOOO dilution) fbr I hour at room tempcraturc a n d then washcid with PRS-Tiveen. t h w c times. Thencc~. 100 p I , of perox i d asc c( )nj uga t.cd nn ti-h 11 in a 11 i i n in u n ogl o hul i n I 1:3000 dilution) r'l'ago, Inc.. Rurlingamc~. CAI was addcd, t h e plates incubated for 1 hour at room tempera tur (~ . and washcd as abovc~. Licactivity was detected by 2.2' azinohis 13-etliyl- henzthiazolinc> sulfonic acid 1 as substrate."'

RESULTS Expression of cDNA CZones. Lysat.cs of the RCOADC-

E2 expression clones and the wild-type plasinid cont.ro1 were resolved by SIX-PAGE, and the presence of fu- sion protoins detected by Coomassic Blue staining (not shown ). Plasmid pKKbE2-11DC9 expresses the mature BCOADC-EB as a 52-kd band on SDS-gel. The molecu- lar weights of the expressing subclones coiwlatcd with the prcdictcd molecular weight of the cDNA inserts (Table I and the respective determined nucleotide se- quence ol' each clone coiwsponded to thcl known se- quence and location of RCOADC-EB. The expression of'

TABLE 1 . Molecular Weights of Recombinant BCOADC-EQ Peptides (Recombinant Clones)

Molecular Weight tkd) SDS-PAGE Kesiduc Domain

52 52 59

68

60 3 6 3 1 2X 27 26.5 26 24.,5 24.5

1-42 1 I-x.1 1 - 1 1 .-, 1-22 7 :;.

1;.&"7 115-421 192-121 167-42 1 l i9 -42 1 1 83- 1 2 1 1 Xl j - , lX 1 2ox-42 I 209-421

Mat urc protcin Lipoyl domain I,ipoyl domain and part of E:l

Lipoyl domain. E3 domain. and

E3 and par t of' inner E2 core G 3 binding and inner E2 core E3 binding and inner E2 core Hinge region and inner E2 core Inner E2 core Inncr E2 core I n n c ~ E2 core'

Inner E2 core Inner E2 COW

don1 ai n

part of inner EP core

BCOADC-E2 by immunoblotting. Only the full-length clone (amino acid residues 1 to 421) and expression clone containing amino acid residues 1 to 84, 1 to 115, 1 to 227 , and 84 to 227 were positive (Table 2). Lysates from the E 3 binding site and E2 inner core expression clones, including expression clones of amino acid resi- dues 115 to 421, 132 to 421, 167 to 421, 179 to 421, 183 to 421, 188 to 421, 208 to 421, 209 to 421, and 210 to 421 were not reactive to PBC sera. Similarly by ELISA. nine of nine inimunoblot-positive BCOADC

B C D E

NOTE. The recombinant pcsptidcs of BCOAL)C-E2 \verc expressed

''. The recombinant clonc of amino acid residues 1-84. 1-1 15. 1 - i n pKK239-2.

227. and 81-22; were esprvswd in p h 1 .

recombinant peptides was also detected by rabbit aiiti- BCOADC-E2 antibodies by immunoblotting (Fig. 1 ).

Specificity of the BCOADC-E2 Expression Clone. PBC sera that had been preabsorbed with lysates of pKKbE2-11DC9, a n expression clone of the mature bo- vine BCOADC-E2, lose imniunoreactivity against the mitochondrial BCOADC-E2 band. However, PBC sera tha t has been absorbed with the control E coli XL1 blue lysate retain their reactivities against the BCOADC- E2 protein. Normal sera, a s expected, do not show any imniunoreactivity to mitochondrial proteins (Fig. 2 1. Fi- nally, PBC sera absorhod with a human PDC-E2 clone lose reactivity against the 70-kd PDC-E2 band but re- tain reactivity to BWADC-E2. Similarly, sera ab- sorbed with the F alloantigen still retained reactivity to both PDC-E2 and HCOADC-E2 (data not shown). Taken together. these experiments demonstrate tha t the recombinant protein expressed from the full-length BCOADC-E2 clone contains all epitopes normally rec- ognized by sera autoantibodies to BCOADC-E2 of PBC patients in imniunoblotting.

Detection of Lipoic Acid in Recornbinant BCOADC- E2. Using DL-[Z-:'Hl lipoic acid a s a bracer, the recom- binant proteins of BCOADC-E2 were resolved on SDS- PAGE and exposed to x-ray films. The recombinant proteins were clearly seen on the gel; however, the pres- ence oflipoic acid was not found on x-ray films. Scintil- lation counting of the recombinant lipoic domain in pMal showed that. recombinant BCOADC-E2 did not contain the "H-labeled co-factor a s in the nonlipoylated control. Thus, the recombinant proteins do not contain "H-lipoic acid.

Reactiuity of PBC Sera Against Expression Subclones of BCOADC-E2. PBC sera tha t had been previously preabsorbed with E coli lysates were tested for their immunoreactivities against expression clones of

70

52

FIG. 2. Absorption of BCOADC-E2 reactivity by recombinant clone pKKbEZ-llDC9. Beef heart mitochondria were resolved by SDS-PAGE, transferred onto nitrocellulose. and probed with sera from PRC patients. Note tha t PBC sera tha t had bcvn absorbed only with E coli reacted to both the 70-kd PDC-ES band and the 52-kd BCOAI)C-E2 I~and (lane A). Sera tha t had been absorbed with the pKKbES-llD(Xl expression clone reacted to the 70-kd PT)C-E2 band but not the 52-kd BCOADC-E2 band (lane R ) . A sera that only rc- acted with the 52-kd RCOADC-E2 band (lane C ) lost all its reactivity after absorption with the recombinant clone pkkhE2-11DC9 (lane

J. Normal sera do not. react ( lane E ).

HEPAT~I,O(;Y Vol. 22. No. 2. 1995

TABLE 2. Reactivity of PBC (n 7 - 9) Sera With BCOADC-E2 Expression Clones

Clones lmmunoreactivity Amino Acid

Residues Domain Immunoblot*t ELISA?

1-421 1-84 1-11:?

1-227

84-227 115-421

132-421

167-421

179-421 183-421 188-421 208-421 209-421 Negative

Full length Lipoic Lipoic and part of

E 3 I ~ p o i c and E3 and

part of E2 core

E3 Binding and inner E2 core

E3 Binding and inner E2 core

Hingc, repon and inner E2 core

Inner E2 core Inner E2 core Inner E2 core Inner E2 core Inner E2 core F antigen

0.894 2 0.096 0.55 2 0.056

0.769 x 0.095

0.812 2 0.065

0.275 + 0.031 0.075 i 0.015

0.063 t 0.011

0.066 ? 0.011

0.053 + 0.009 0.051 I 0.008 0.048 -c 0.008 0.042 -c 0.009 0.041 2 0.009 0.041 i 0.009

NOTE. Immunoreactivity was scored with 9 I'BC sera ( a t 1:1000 dilution) tha t have been previously shown to react with BCOADC- E2.

* + = positive reactivity; - = negative reactivity ?None of the non-BCOADC-reactive sera or control sera reacted.

sera (1:1,000 dilution) reacted to the full-length BCOADC-E2 clone by ELISA. In contrast, none of the other BCOADC-E2 expression clones react (Fig. 3). Further, none of the non-BCOADC reactive PBC sera nor do the normal control sera react to either of these clones. Finally, a s an additional control, none of the nine PBC sera reacted to an irrelevant F alloantigen (Table 2).

Absorption of PBC Sera With BCOADC-E2 Expression Subclones. The ability of BCOADC subclones to re- move reactivity in PBC sera were studied by immu- noblotting and ELISA. Absorption of BCOADC-positive PBC sera a t 1:1,000 dilution with a crude lysate of the mature BCOADC expression clone removed all reactiv- ity against itself by immunoblotting. In contrast, when nine PBC sera, diluted 1:1,000, were absorbed against an expression clone containing residues 1 to 84 of BCOADC-E2, all the sera retained reactivity against the full-length protein. If absorptions were performed a t higher dilutions of sera, one in nine sera lost reactiv- ity against the full-length protein at a serum dilution of 10 and four of nine lost reactivity at a serum dilu- tion of 10 ' (Table 3). Sequence analysis shows tha t lipoic acid domain of BCOADC-E2 is contained within amino acid residues 1 to 84 of the molecule (Table 4). If absorption was performed against a recombinant protein containing residues 1 to 115, then reactivity was lost in one of nine, three of nine, and eight of nine of the sera when examined a t dilutions ranging from 10 '' to respectively. If a recombinant protein

LEUNG ET AL 509

spanning residues 1 to 227 was used to absorb antibody reactivity, then four of nine sera diluted 1:1,000 lost reactivity. Eight of nine sera diluted to 1:10-4 and eight of nine sera diluted to 10 lost reactivity after absorp- tion. Recombinant proteins spanning region 84 to 227 could remove reactivity be absorption in only two of nine sera diluted to 1:lO '.

Likewise, reactivity of PBC sera with BCOADC-E2 expression subclones were studied by ELISA. At dilution to 10 ' dilution, the full-length clone absorbed out 95% to 1009% of the total reactivity, whereas the expression clone containing amino acid residues 1 to 84 partially reduced the reactivity by 29.7%. to 50.9%,. Expression clones of amino acid residues 1 to 115, 1 to 227 were able to remove reactivity of PBC sera by 36.1% to 67.53% and 50.8% to >go%), respectively, against the full-length BCOADC-E2 clone a t to

sera dilution. Interestingly, a combination of ei- ther expression clone 1 to 115 and 115 to 421, as well a s 1 to 227 and 115 to 421, were capable of absorbing >74.1%' and 90% of PBC sera reactivity a t dilu- tion, respectively. Recombinant clones 1 to 84 and 84 to 227, 1 to 115 and 84 to 227 absorbed reactivity of PBC sera against mature BCOADC-E2 at 54.65% and 75.3%, respectively. Finally 1 to 84, 84 to 227, and 115 to 421 together absorbed only 45%, of the reactivity at 10 sera dilation. All other expression clones and the control pKK233-2 were not capable of absorbing any of the anti-BCOADC antibodies (Table 3).

DISCUSSION Epitope mapping of autoantibodies including discus-

sions of linear, discontinuous, and conformational epi- topes have provided valuable insights into the specifici- ties of anti-self In PBC the data on epitope mapping of AMA have hitherto been available only for autoantibodies directed against PDC-E2.8j27 Using a re- combinant clone of the mature bovine BCOADC-E2Ifi and

Llpoyl Domaln €3 Blndlnp Inner €2 Core

NH2 PSI1 Kpnl COOH - 4 *

AA Resldue 1-421 + 1-84 1-115 1-227 84-227 t

1 1 5 - 4 2 1 . 1 3 2 - 4 2 1 . 1 6 7 - 4 2 1 . 1 7 9 - 4 2 1 . 1 8 3 - 4 2 1 - 1 8 8 - 4 2 1 . 2 0 8 - 4 2 1 - 2 0 9 - 4 2 1 .

FIG. 3. Immunoreactivity of BCOADC-E2 expression clones. The diagram of recombinant clones is aligned with the structural domains of bovine RCOADC-E2. The main structural domains are shown in empty boxes and the filled boxes represent the hinge regions between the domains. Lys is the location of the lipoyl lysine residue 44. Re- striction sites ofHind 111, PsfI. and KpnI are shown. NH, and COOH arc the amino and carboxy termini, respectively, A and - indicate positive and negative reactivity, respectively.

510 LEUNG ET AL HEPATOLOGY August 1995

TABLE 3. Absorption of Anti-BCOADC-E2 Antibodies by BCOADC-E2 Expression Clones

Clones/Residue

lmmunoblot (No. Absorbed' Total)* ELISA ("r Absorbed r SD)t n - 9*

Domain 10 lo- ' 10 10 10-3 lo-' 10 %*

pKK233-2 1-421 1-84 1-115

1-227

84-227 1-84 T 84-227 1-115 + 84-227 1-84 + 84-227 + 115-227 1-115 + 115-421 1-327 7 115-421 115-421

132-421

167-421

179-421 183-421 188-421 208-421 209-42 1 Clone F

Control plasinid Full length Lipoyl domain Lipoyl domain + part of

E3 binding domain Lipoyl domain + E3

binding domain + part of E2 domain

E3 binding and inner

E3 binding and inner

Hinge regon and inner

Inner E2 core Inner E2 core Inner E2 core Inner E2 core Inner EZ core Negative control

E2 core

E2 cnre

E2 core

0/9 919 0/9

1/9

419 019 019 1 I9 1/9 119 419

019

019

019 019 019 019 0/9 019 0/9

019 919 119

319

819 019 1 I9 319 119 319 819

019

019

019 019 019 0/9 019 019 019

019 919 419

819

819 219 419 819 419 619 919

019

019

019 019 019 019 0/9 019 019

019 919 919

919

919 419 919 919 919 919 919

019

019

019 019 019 019 019 019 019

0 95

29.73 z 0.21

36.1 -c 1.19

50.8 z 1.10 8.1 -c 1.10

30.85 z 1.15 66.7 -c 0.60 34.4 5 1.60

34.03 -c 3.10 75.53 L 24

0

0

0 0 0 0 0 0 0

0 99

36.5 1 2.23

55.26 + 0.95

62 L 0.70 19.75 + 1.35 40.05 i 1.25

71.7 z 0.30 44 + 1.83

50.3 f 1.20 83.56 2 6.31

0

0

0 0 0 0 0 0 0

0 100

50.9 + 2.21

67.53 z 2.23

31.15 ? 0.25 54.6 2 3.65 75.3 -c 2.30 45.1 2 1.40 74.1 z 5.90

,.go

0

0

0 0 0 0 0 0 0

NOTE. PRC sera ( n = 9 ) were absorbed with E coli expression clones of BCOADC-E2 and tested for remaining activity against the mature BCOADC-E2 protein.

Non-BCOADC-E2 reactive sera from PBC patients and negative control sera did not react. + Percent ahsorbed -C SD =: O.D. (absorbed with pkk233.2) - O.D. (absorbed by BCOADC-E2)/O.D. (absorbed with pkk233-2) x 100C;.

its derivatives, we studied the antibody reactivity of PBC sera against the str-uctural domains of BCOADC-E2. PBC sera reacted strongly to the mature BCOADC-E2 clone by immunoblotting as well as ELISA, whereas clones bearing only the E3 binding site and E2 inner core failed to react. However, a clone containing amino acid residues 1 to 84 (lipoyl domain) also reacted but to a lesser intensity than the full-length clone, suggesting

TABLE 4. Amino Acid Comparison Between the Lipoyl Domains of Human PDC-E2, Human BCOADC-E2,

and Bovine BCOADC-E2

~ ~ ~ ~ ~~ ~~~~~~~

NOTE. Identical sequences a re shown in bold. Conservative sub-

'. Indicates the lysine rcsidues where thc lipoic acid is bound. stitutions are underlined.

that an epitope resides within the lipoyl domain. To fur- ther characterize the reactivity of the lipoyl domain, we absorbed a panel of PBC sera (from 10 to 10 ' dilution) with the crude lysates of the lipoyl domain expression clone and tested for their reactivity against the mature expression clone. Our results show that the lipoyl domain clone was able to absorb out reactivity against the mature BCOADC-E2 clone only at a very high dilution of the sera (lo-')); five of nine of the sera still reacted at 10 dilution. It is therefore likely that the lipoyl domain, and its flanking sequences, constitute the major conforma- tional epitope in BCOADC-E2. Because a clone con- taining recombinant peptides 115 to 421 did not react to PBC sera, it suggests that the inner E2 core and the E3 domain by themselves do not have any antibody binding activity.

Expression clones of peptide 1 to 84, 1 to 11 5, and 1 to 227 were only able to partially absorb AMA reactivity against the mature BCOADC-E2. Interestingly, the longer the peptide from the N-terminus to aa residues 1 to 227, the more effective it is in AMA binding ability tie., 1 to 227 >1 to 115 >1 to 84 in antibody binding), indicating a synergistic effect in AMA recognition within aa residues 1 to 227. However, only aa residues 1 to 421 can absorb out all AMA reactivity at lo - , '

HEPATO~,OC;Y Vol. 22 , No. 2. 1995 LEUNG ET AL 511

dilution of PBC sera (Table 3). Another peptide con- taining amino acids 84 to 227 of BCOADC-E2, however, is weak in AMA recognition even at 10 sera dilution. Moreover, absorption data of peptide combinations 1 to 84 and 84 to 227, 1 to 115, and 84 to 227 are very similar to that of 1 to 84 only and 1 to 115 only, respec- tively. A combination of peptide 1 to 227 and 115 to 421 is more effective than 1 to 227 itself in removing AMA activity against mature BCOADC-E2.

Our observation that combination of peptides cannot remove all AMA reactivities further support our thesis that proper folding of the protein is critical in the pre- sentation of epitopes for recognition by anti-BCOADC- E2 antibodies. It therefore suggests that the BCOADC- E2 epitope is highly conformational and most likely consists of a major epitope in the lipoic acid binding domain and its immediate flanking regions and other minor epitopes in the E3 domain and in the E2 inner core domain, which contribute to the proper folding and synergistic presentation of epitopes. Because individ- ual reactive peptides individually were unable to re- move all sera reactivity against the original clone, the proper folding of these domains is thus important in antibody recognition. We note that we have made use of the conserved structure of BCOADC-E2 among mammalian species in our study. This has allowed us to use the available bovine expression clones to deter- mine the autoantibody reactivity to bovine BCOADC- E2. Although the nucleotide and amino acid sequence of human and bovine BCOADC-E2 are nearly identi- cal,'"16 it remains a possibility that additional B cell epitopes may be detectable if human homologue of BCOADC-E2 is used in such mapping studies.

In contrast to T-cell epitopes, which are short linear peptides, B-cell epitopes may be more complex as evi- denced by the recent mapping of a number of B-cell epitopes. For example, the histidyl t-RNA transferase of the myositis-associated anti-Jo-1 autoimmune re- sponse contains both linear and conformational epi- topes.28 Multiple epitopes, including linear and discon- tinuous epitopes, have also been mapped to the p70 (Ku) antigens using sera from patients with systemic lupus erythematosus.29 In the human La autoantigen, conformational epitopes within the helical core and the RNP consensus sequence were identified.3o A discontin- uous immunodominant epitope also has been defined in the NH2-terminus of the LdSS-B ribonucleoprotein autoantigen.26 The U1 snRNP-associated B B ' polypep- tide of Sm autoantibodies in systemic lupus erythema- tosus has six epitopes, some of which are conforma- tional. '' Multiple epitopes have been demonstrated in the 70-kd autoantigen of the U1 ribonucleoprotein com- plex.32 Most relevant, it has also been demonstrated that the autoepitope of PDC-E2 lies within a conforma- tional lipoyl domain of the enzyme.,3" Based on molecu- lar modeling of antibodies reacting with antigens,34 9070 or more of B-cell epitopes are thought to be confor- mational. The epitopes detected by BCOADC-E2 auto- antibodies are somewhat different from those found in other autoantigens."fi.'H'32 In BCOADC-E2, there are no

clearly defined discrete linear epitopes. Instead, there is a longer less defined region that extends fkom amino acid residues 1 to 115 of the molecule. Moreover, within these 115 amino acid residues, a certain degree of im- munoreactivity is clearly present in the first 84 resi- dues where the lipoyl domain resides. However, the classification of this region as a separate epitope is dif- ficult and somewhat arbitrary, based on the boundaries of the particular expression clones used in this study. This pattern of autoantibody reactivity, i.e., reactivity with a continuous long stretch of amino acid residues, has its closest analog to PDC-E2":' and is different from the pattern detected in other antigens.""."-"2

Highly conserved molecules are common targets of autoantibodies. For example, antibodies from patients with systemic lupus erythcmatosus react to histones, DNA, and a number of nuclear proteins, all of which are highly conserved among different species. Antipro- liferating cell nuclear antigen (PCNA cyclin 1 antibodies from patients with systemic lupus erythematosus react with proliferating cell nuclear antigen from mamma- lian cells, protozoa, and polytene chromosome of insects and plant^.^"^'" Interestingly, the proliferating cell nu- clear antigen autoantibodies recognize a particular conserved structure that represents only a part of the overall primary structure. Similar observations have been observed in the La autoantigen,2".:'9 in which a conserved structure is the dominant epitope.

Recent studies in PBC have shown that the lipoic acid binding domain, also a highly conserved structure, constitutes the immunodominant epitope of PDC-E2.' Substitution of the lysine residue within the lipoic do- main with other amino acids did not markedly aSSect antibody recognition,I8 suggesting that the overall con- served structure of the lipoyl domain is important for antibody binding. Other studies also demonstrate that lipoic acid alone is not responsible for antibody bind- ing,40 although the unique peptide backbone in the li- poic domain is important for its antigenicity. This issue is ongoing4'-48 and will require additional data for reso- lution. Therefore, the relative importance of each com- ponent to antibody binding remains to be established.

Although the lipoyl domain is common among the E2 subunits of the mitochondria1 autoantigens, the only cross-reactivity is seen between protein X and PDC- E2; there is no crossreactivity a t the level of the lipoyl domains.49 Interestingly, PDC-E2 specific combinato- rial human autoantibodies recognize a conformational epitope in the inner lipoyl of PDC-E2 but not BCOADC- E2.j' These data and our current report suggest that it is likely that each of the 2-oxo-acid dehydrogenase enzymes has its own unique antigenicity with respect to structural requirement for antibody binding despit.e the overall similarity in their lipoyl domain and sub- unit structures. Recently, Chen et a15' reported that antibodies to PDC-E2 recognized two forms of PDC-E2 in immunoblotting; two bands of 74 kd and 70 kd were recognized in unboiled sample, whereas only one 74- kd band was detected in boiled PDC-E2. It is therefore likely that these two forms represent the different pre-

512 LEUNG ET AL HEPATOLOGY August 1995

sentation of the molecule on the blot; perhaps lipoic acid is obstructed in one form but not the other. In addition, the significance of the autoantigedantibody recognition and their possible role in pathogenesis still remains to be elucidated, but it is noteworthy that PDC-E2 has been uniquely detected on the periluminal surface of biliary epithelial cells of patients with PBC.52 However, currently no data are available on the distri- bution of BCOADC-E2 in the biliary epithelium of PBC patients.

Acknowledgments: We thank Kiyomi Yoshida, Yvonne Chan, Joseph Lam, and Keith Wong for their technical help and Theresa Andreozzi for preparing the manuscript.

REFERENCES 1. Kaplan MM. Primary biliary cirrhosis. Adv Intern Med

2. Van de Water J, Surh CD, Leung PS, Krams SM, Fregeau D, Davis P, Coppel R, e t al. Molecular definitions, autoepitopes, and enzymatic activities of the mitochondrial autoantigens of primary biliary cirrhosis. Seminar Liver Dis 1989;9:132-137.

3. James OF, Yeaman SJ, Bassendine MF. Molecular aspects of the M2 autoantigens in primary biliary cirrhosis: What a difference a year makes. HEPATOLOGY 1989; 10:247-251.

4. Mackay IR, Gershwin ME. Primary biliary cirrhosis: considera- tions on pathogenesis based on identification of the M2 autoanti- gens. Springer Semin Immunopathol 1990; 12:lOl-119.

5. Mackay IR, Gershwin ME. Molecular basis of mitochondrial auto- reactivity in primary biliary cirrhosis. Immunol Today 1989; 10:

6. Van de Water J , Davis P, Ansari A, Danner D, Fregeau D, Leung P, Coppel R, et al. Autoantibodies of primary biliary cirrhosis (PBC) recognize dihydrolipoamide acetyltransferase and inhibit enzyme function. J Immunol 1988; 141:2321-2324.

7. Fregeau DR, Davis PA, Danner DJ, Ansari A, Coppel RL, Dick- son ER, Gershwin ME. Antimitochondrial antibodies of primary biliary cirrhosis recognize dihydrolipoamide acyltransferase and inhibit enzyme function of the branched chain a-keto acid dehy- drogenase complex. J Immunol 1989; 142:3815-3820.

8. Surh CD, Coppel R, Gershwin ME. Structural requirement for autoreactivity on human pyruvate dehydrogenase-E2, the major autoantigen of primary biliary cirrhosis. Implication for a confor- mational autoepitope. J Immunol 1990; 144:3367-3374.

9. Griffin TA, Lau KS, Chuang DT. Characterization and conserva- tion of the inner E2 core domain structure of branched-chain a- keto acid dehydrogenase complex from bovine liver: construction of cDNA clone encoding the entire transacylase (E2b) precursor. J Biol Chem 1988;263:14008-14014.

10. Pettit FH, Yeaman SJ, Reed LJ. Purification and characteriza- tion of branched chain a-keto acid dehydrogenase complex of bovine kidney. Proc Natl Acad Sci USA 1978;75:4881-4885.

11. Danner JJ , Litwer S, Herring WJ, Pruckler J. Construction and nucleotide sequence of a cDNA encoding the full length prepro- tein for human branched chain acetyltransferase. J Biol Chem

12. Chuang DT, Hu C-WC, Lau S, Markovitz PJ, Cox RP. Subunit structure of the dihydrolipoyl transacylase component of branched chain a-keto acid dehydrogenase complex of bovine liver. J Biol Chem 1985;260:13779-13786.

13. Pate1 MS, Roche TE. Molecular Biology and Biochemistry of py- ruvate dehydrogenase complexes. FASEB J 1990; 4:3224-3233.

14. Perham RN. Domains, motifs and linkers in 2-0x0-acid dehydro- genase multienzyme complex: a paradigm in the design of a multifunctional protein. Biochemistry 1991; 3023501-8512.

15. Griffin TA, Wynn RM, Chuang DT. Expression and assembly of mature apotransacylase (E2b) of bovine branched chain a-keto

1987;32:359-377.

315-318.

1989;264:7742-7746.

acid dehydrogenase complex in Escherichia coli. J Biol Chem 1990~265:12104-12110.

16. Griffin TA, Chuang DT. Genetic reconstruction and characteriza- tion of the recombinant transacylase (E2b) component of bovine branched chain a-keto acid dehydrogenase complex. J Biol Chem

17. Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain terminating inhibitors. Proc Natl Acad Sci USA 1990; 74:5463- 5467.

18. Leung PSC, Iwayama T, Coppel RL, Gershwin ME. Site-directed mutagenesis of lysine within the immunodominant autoepitope of PDC-E2. HEPATOLOGY 1990; 12:1321-1328.

19. Gershwin ME, Coppel RL, Bearer E, Peterson MG, Sturgess A, Mackay IR. Molecular cloning of the liver specific rat F antigen. J Immunol 1987; 139:3828-3833.

20. Pfeffer PE, Silbert LS. a-anions of carboxylic acids. I. Effect of hexamethylphosphoramide on metalation and alkylation. J Org Chem 1970;35:262-264.

21. Pfeffer PE, Silbert LS, Chirinko JM. a-Anions of carboxylic acids. Ii. The formation and alkylation of a-methylated aliphatic acids. J Org Chem 1972;37:451-458.

22. Pfeffer PE, Kinsel E, Silbert LS. a-Anions of carboxylic acids. V. A simple high yield presentation of a-alkylhydracrylic acids and a-alkylacrylic acids. J Org Chem 1972;37:1256-1258.

23. Fregeau DR, Roche TE, Davis PA, Coppel R, Gershwin ME. Pri- mary biliary cirrhosis. Inhibition of pyruvate dehydrogenase complex activity by autoantibodies specific for E l alpha, a non- lipoic containing mitochondrial enzyme. J Immunol 1990; 144:

24. Van de Water J , Copper A, Surh CD, Coppel R, Danner D, Ansari A, Dickson R, et al. Detection of autoantibodies to recombinant mitochondrial proteins in patients with primary biliary cirrhosis. New Eng J Med 1989;320:1277-1380.

25. Leung PSC, Van de Water J , Coppel RL, Gershwin ME. Molecu- lar characterization of the mitochondrial autoantigens in pri- mary biliary cirrhosis. Immunol Res 1991; 10:518-527.

26. McNeilage LJ, Umapathysivam K, Macmillan E, Guidolin A, Whittingham S, Gordon T. Definition of a discontinuous immu- nodominant epitope at the NH2 terminus of the La/SS-B ribo- nucleoprotein autoantigen. J Clin Invest 1992; 89:1652-1656.

27. Van de Water J , Gershwin ME, Leung P, Coppel RL. The auto- epitope of the 74kd mitochondrial autoantigen of primary biliary cirrhosis corresponds to the functional site of dihydrolipoamide acetyltransferase. J Exp Med 1988; 167:1791-1799.

28. Ramsden DA, Chen J , Miller FW, Misener V, Bernstein RM, Siminovitch KA, Tsui FWL. Epitope mapping of cloned human autoantigen, histidyl tRNA synthetase. J Immunol 1989; 143:

29. Reeves WH, Pirtani A, Chou CH, Ng T, Nicastri C, Roeder RG, Sthoeger ZM. Epitopes of the p70 and p80 (Ku) lupus autoanti- gens. J Immunol 1991; 146:2678-2701.

30. Chambers JC, Kenan D, Martin BJ, Keene JD. Genomic struc- ture and amino acid sequence domains of the human La autoan- tigen. J Biol Chem 1988;203:18043-18051.

31. Rokeach LA, Jannatipour M, Hoch SO. Heterologous expression and epitope mapping of a human small nuclear ribonucleoprotein associated Sm-BB' autoantigen. J Immunol 1990; 144:1015- 1022.

32. Cram DS, Fisicaro N, Coppel RL, Whittingham S , Harrison LC. Mapping of multiple B cell epitopes on the 70-kilodalton autoan- tigen of the U1 ribonucleoprotein complex. J Immunoll990; 145:

33. Rowley MJ, McNeilage LJ, Armstrong JM, Mackay IR. Inhibi- tory antibody to a conformational epitope of the pyruvate dehy- drogenase complex, the major antigen of primary biliary cirrho- sis. Clin Immunol Immunopathol 1991;60:356-370.

34. Blundell TL, Sibanda BL, Sternberg MJE, Thornton JM. Knowl- edge-based prediction of protein structures and the design of molecules. Nature 1987; 326:347-352.

35. Takashi Y, Deng J-S, Tan EM. A nuclear antigen associated with cell proliferation and blast transformation: its distribution in synchronized cells. J Exp Med 1981; 154:1899-1909.

36. Olins DE, Olins AL, Colcheiro LH, Tan EM. PCNNcyclin in the ciliate Euplotes eurystomus: localization in the replication band and in micronuclei. J Cell Biol 1989; 109:1399-1409.

1990;265:13174-13180.

1671-1676.

2267-2272.

630-635.

37

38

:39

10

41

12

43

1 4

Aniahis Jhl. Aniahis DC'. Kaburaki J . Stollar HT). The prcsencc of iiii antigen reactive \villi :I human auto;intil)ody in Trictiosia pubrsccns 1 Diptera: S c i a r i d w ~ and its ;issociation \vith certain transcriptionally active rogions of the gc~nomct. (Ihromosoma 1990;99: 102- 110. Suzuka 1. Daicloji H. Matsuoka h4, Kudownki K, 'I'aknsaki Y. Kiikancb PK, Moriuchi '1'. Gene for proliferating ccll nuclear anti- gen (DNA polymerase sigma auxiliary protein) is prctsent in both m;imm:ili;in and higher plant gcnomes. Proc Natl Acad Sci LJSA 1989; 86::3 189-:3 193. Chan EKL. Tnn EM. I lunian autoantibody rcwtive cpitopcs of SS-IVI,n are highly conscrvetl i n comparison with epitopes rccog- n i z d hy inurine monoclonal ant.ihodies. J Exp h1C.d 1985: 166: 162i-1640. Flannery ( iR , Uui~ougtis AK, ButlcLr P, Chelliirh .J. Hainilton- Slill(.r ,I. I$riinifilt W, Uaum H. Ant.iniitochondria1 antihodios in primary hiliary cirrhosis rccognize both specific peplitle and shared ellitopes of the b12 family of antigcsns. HEP.\TOILM ;Y I 989: io::iio-3.;4. Stcnicrowicz K. Floufl:. hloller B. Witterhrink ('. Koclloff A. Kein- hiirdt H . Frcudenl)crg &I. et al. Arc nntimitochondri;ll an1il)odies in primary hiliary cirrhosis induced hy R ( rough~-mutants of cntcrot)actc,ri~~cp~~e'~ Lancet 1988: i i : 1166-1 170. Linden horn-Fotinos J . Rarim H. Ucrg 1'A. hiitochondrial antihod- ies i n primary hiliary cirrhosi pecies and nonspcvies specific dcL term in an t s of M', an t i p n . I1 li P)Y~OLO(.: Y 1985; 5 : 763-769. Fusscy SPhl. Ali S. Guest J K . James OFW. Rassendine hlF, Yeaman S.J. lieactivity of primary biliary cirrhosis SCW with Escltc~rrchrn coli dihydrolipoamide acetyltransfc~rase ( b 2 p 1 : Characterization o f thc main immunogenic region. I'roc Natl Acad Sci L!SA 1!190;Ri:31-18T-3991 Tuaillon N. Andw c', I3riand .J-P. Penner E, Mullcr S. A lipoyl synthetic octnpeptide of di1iycirolipo;rmidc acetyltransrerase spc- cilically recognized by anti-Mil:! autoantibodies in primary hiliary cirrhosis. .I Iminunol 1992: 118:445-150.

15. 41i ST. (;urat .JK. Isoliitivn and ehariicteriz;rtion of lipoylatcd and unlipoylntcd domains ol' t!ie E2p subunit ol' the pyruvate dohydrogenas;c~ comples of E,sc~/ic~rrc~/iici c,oli. Riocheni ,J 1990:251:139-145.

36. .41i S1'. Xloir A.JC;, Ashton I ' R ~ Kngcl I'C'. Guts t .JR. Octanoyla- tion ol'the lipoyl domains of the pyriivat e dehvdlogc.ri;-ise complcs i n a lipoyl-deficient strain of Esch(~r-ichin ( , o / i . hlol hlicrohiol 1990: 1:94.'3-1-)50.

17. Quinn J . Dianiond AG. Masters AK, Urookfieltl L)E, Wnllis NG. Yeainan SJ. Expression and lipoylntion i n Esc~/ir~rr(hio coli of the inner lipoyl domain ol'thcs E:! coniponc)nt of the human pyruvntc dehvdrogcnasc~ complex. Riocheni J 1993;289:81-85.

18. Quinn .J . L)i;rmond AG. Palmer .JhI. Rassendinc, MF. .James OFW. Yeaman SJ. Lipoylated and unlipoylatcd domains of l i u - man t'l)C:-K2 :IS autoantigcns i n primary h i l i ~ i y cirrho nificance of lipoate attaclimcwl. I~P:I):VIY)~,(.)(;Y 1993; 1 1991,

.19. Surh CD. Roclic TE, Danncr D J , Ansari A. Coppel RI,. Printliville T. Dickson ER, et al. Antiinitochondrial autoantibodies i n pri- mary hiliary cirrhosis recognize cross-rractive rpitopccs) on pro- tein S and dihydr(~li~)oarnide acetyltransfcrasc of pyruvate dehy- drogenas;cx complex. HEP.v~( ) I . ( )(:Y 1989: 10:127-133.

50. c'ha S. Leung I'SC. Ck~ppel RI.. Van de Water .J. Ansari A, Gersh- win ME. Mctcrogcneity of comhinatorial hunian autoantibodies iigainst P D C - U a n d hiliary epithelial cells in paticsnts with pri- mary biliary cirrhosis. kk11.4TOI,0(;Y 199.1;20:571-58:3.

51. Chcsn QY. Kowley MJ. Mackay Ili. A n t i t d y t o two forms 01' dihydroliponniide acetyltransferascfi (PDC-EP) i n primary hiliary cirrhosis. Liver 1993; 13: 130- 135.

52. Van de Water .J. Turchany J . 1,cuiig PSC. Lakc .J, hlunoz S. Surh CD. Coppel R, cst al. Molecular mimicry in primary hiliary cirrhosis: evidcnce for biliary epithelial expression o f :i molcculc cross-reactive with pyruvate dehydrogennsc complcs-E2. .I C'lin Invc5t 1993; 91:2653-2664.