Genomic Structure and Amino Acid Sequence

Vol. 263,No. 34, Issue of December 5, pp. 18043-18061,1988 Printed in U.S.A.

THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc

Genomic Structure andAmino Acid Sequence Domains of the Human La Autoantigen” (Received for publication, May 4,1988)

Jasemine C. Chambers$, Daniel Kenan, Barbara J. Martin$, andJack D. Keenen From the Department of Microbiology and Immunology, Duke University Medical Center, Durham, North Carolina 27710

and Steitz, 1979; Rinke and Steitz, 1982; Chambers et al.., 1983), as well as some viral RNAs such as VA-RNAs, EBERRNAs (Rosa et al., 1981; Francoeur and Mathews, 1982), and leader RNAs of some negative-strand viruses (Kurilla and Keene, 1983; Kurilla et al., 1984; Wiluszand Keene, 1984). In addition, La protein can be cross-linked to small RNAs using ultraviolet light (van Eckelen et al., 1982; Chan and Tan, 1987).Due to its nuclear localization and transientassociation with RNA polymerase 111transcripts, La protein was thought to be involved in the biogenesis of these RNAs. The nature of the interactions between the La protein and small RNAs and its possible functions in RNA synthesis have been reviewed (Keene et al., 1987a, 1987b). Although it is clear that autoimmune diseases are associated with an abnormal reaction of the immune system against oneself, the origins of autoimmunity and autoantibodies are not well understood. Furthermore, it is not clear whether certain determinants on these proteins are more antigenic than others and whether epitopes vary among patients. This is of interest because autoantigens are often components of RNP complexes, and the structuralrelationship between continuous anddiscontinuous (conformational) epitopes on these complexes has not been examined. For example, it is not known under what circumstances the immune system recognizes individual components of the RNP complex versus the entire complex in the development of a humoral autoimmune response. In a recent study on the evolutionary origins of various autoreactive determinants (Kieber-Emmons and Kohler, 1986), it was shown that theimmune system did not produce La protein was originally defined by its reactivity with autoantibodies reactive with all potential antigenic sites. Inautoantibodies from patients with Sjogren’s syndrome and stead, autoreactive determinants appear to reside on the sursystemic lupus erythematosus (SLE)’ (Mattioli and Reichlin, faces of proteins and to show evolutionary variability. The 1974; Tan, 1982). The human La (or SS-B) antigen appears localization of autoimmune epitopes on histones H1 andH2B by immunoprecipitation and immunoblotting to be a single was previously demonstrated by immunoblotting of large fragphosphoprotein of 46 to 50 kDa (Habets et al., 1983; Venables ments produced by specific proteolytic or chemical cleavage et al., 1983; Stefano, 1984). of the proteins (Hardin and Thomas, 1983). It was suggested Many small RNAs of mammalian cells associate with the that autoantibodies to histones specifically recognizeselected La antigen, including 7 S RNA, 5 S RNA, and tRNA (Lerner segments of individual histones which might occupy more * These studies were supported by grants from the National Insti- exposed positions in intact chromatin. In an effort to understand further the role of the La protein tutes of Health and theArthritis Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. in RNA metabolism and in the origins of autoimmunity, we This article must therefore be hereby marked “aduertisernent” in have characterized human genomic and cDNA clones (Chamaccordance with 18 U.S.C. Section 1734 solelyto indicate this fact. bers and Keene, 1985). Here we report the amino acid seThe nkleotide sequence(s) reported in thispaper has been submitted quence and genomic structure of the La protein. In addition, to the GenBankTM/EMBL Data Bank withaccessionnumber(s) we have identified at least three antigenic epitopes on La 504205. $Present address: Laboratory of Molecular Genetics, Bldg. 36, protein and predicted regions of the protein involved in RNA binding based upon structural similarities with other RNARm. 4A01, NINCDS, NIH, Bethesda, MD 20892. § Present address: Dept. of Neurology, University of Pennsylvania binding proteins.

La isan autoimmune RNA-binding protein of 47 kDa that plays a role in the transcriptionof RNA polymerase 111. Both genomic and complementary DNAs were isolated that encompass the coding sequence of the human La molecule. The genomic clones encompass 11 exons and a putative G/C-rich promoter upstream of the mRNA start site. The cDNA sequence encodes a protein of 408 amino acids and canbe divided into two structural domains based upon amino acid content and protease sensitivity. An unusually long stretch of 130 amino acids, much of which was predicted to form a stable a-helix, was found near themiddle of the protein between the two domains. A ribonucleoprotein (RNP) consensus sequence was found just NH2-terminal to the long a-helix. TheRNP consensus sequence is split into two exons by the fifth intron. Expression of three separate fragments of the La protein in Escherichia coli showed that a strongly autoimmune-reactiveportion resides in the fragment containing the RNPconsensus sequence and most of the long a-helical core. Autoantibodies from La patientsalso reacted with the terminal regions of the protein, but the extent of reactivity variedamong patients. Differences in reactivity of autoantibodies toeach portion of La proteinmay reflect an evolution of recognition of different epitopes during the development of the autoimmune response. These findings support an antigen-driven mechanism for autoimmune reactivity.

School of Medicine, Philadelphia, PA 19107. 7 Pew Scholar in the Biomedical Sciences. MATERIALSANDMETHODS The abbreviations used are: SLE, systemic lupus erythematosus; SDS, sodium dodecyl sulfate; PIPES, 1,4-piperazinediethanesulfonic Cells and Enzymes-HeLa cells, strains of Escherichia coli, and acid; RNP, ribonucleoprotein. bacteriophage were prepared as previously described (Chambers and

18043

Structure of Human La Protein

18044

Keene, 1985). Restriction enzymes and sequencing materials were obtained from IBI and Bethesda Research Laboratories. A genomic libraryprepared from human fetal DNA (Lawn et al., 1978) was obtained from Dr. Tom Maniatis (HarvardUniversity). Human chromosome libraries were obtained from Dr. Marvin Van Dilla of the Los Alamos Laboratories. VA-RNA expressed in the T7 in vitro synthesis system was a gift of Dr. Dana Fowlkes (University of North Carolina, Chapel Hill) and was designed to produce truncated transcripts which terminate at 3"uridylates similar to the natural product in adenovirus-infected cells. Southern and Northern Blot Analysis-High molecular weight DNA was isolated from cultured cells, digested with various restriction enzymes, and electrophoresed in nondenaturing1% agarose gels. Polyadenylated ortotal cRNA was prepared from HeLa cells as described (Chambers and Keene, 1985), denatured in 2.2 M formaldehyde and 50% formamide, and fractionated by electrophoresis in 1% agarose gels containing 2.2 M formaldehyde. Denatured DNA and RNA were transferred electrophoretically toNytran membranes (Schleicher & Schuell, Inc.) and hybridized to 3ZP-labeled cDNA probes following nick translation. Hybridization was a t 42 "C for 24 h using 50% formamide, 5 X SSPE (0.9 M NaCI, 50 mM NaH2P04,5 mM EDTA, pH 7.4), 5 X Denhardt's solution, 0.1% SDS, and 100 pg/ ml sheared salmon spermDNA. The blots were subsequently washed four times with 0.2 X SSC (0.03 M NaC1, 0.03 M sodium citrate, pH 7.4) and 0.1% sodium dodecyl sulfate (SDS) a t either 37 "C for low stringency hybridization acrossspecies or at65 "C for high stringency hybridization and air-dried for autoradiography. DNA Sequencing-DNA fragments cloned in M13 or plasmid vectors were sequenced using either the dideoxynucleotide method (Sanger et al., 1980) or the chemical method (Maxam and Gilbert, 1977). Primer Extension-Synthetic oligonucleotides representing the cDNA sequence were 5'-end labeled with [ Y - ~ ~ P I A T using P T4 polynucleotide kinase and were used for primer extension. Polyadeynlated RNA (10 pg) from HeLa cells was hybridized with 1.5 mg of labeled primer in 15 pl of buffer containing 400 mM NaCl and 10 mM PIPES (pH6.7) a t 51 "C for 3 h. The annealed RNA-primer mixture was adjusted to 50 pl containing 100 mM Tris (pH8.3), 10 mM MgC12, 10 mM dithiothreitol,a 0.5 mM concentrationeach of the four deoxynucleoside triphosphates, and 10 units of reverse transcriptase. The reaction mixture was incubated at 42 "C for 1 h and extracted with phenol/chloroform before precipitation with ethanol. Extension products were analyzed on 8 M urea, 6% polyacrylamide gels. Immunoblotting-After separationon6% polyacrylamide/SDS gels, protein species were electrophoretically transferred to nitrocellulose as described (Towbin et al., 1970). Filters were blocked with A Blotto,reactedwith antiserumand 1251-Staphylococcus-protein (Query and Keene, 1987). Fusion Protein Preparation-Fusion proteins were prepared using cell lysis and centrifugation as described (Adam et al., 1986; Query and Keene, 1987). E. coli Y1089 was lysogenized with various XgtllLa constructions and grown in LB medium a t 30 "C to ODW of 0.5. The cultures were induced by transfer to 42 "C for 15 min, followed to 3 mM final by addition of isopropyl-j3-D-thiogalactopyranoside concentration. Cultures were incubated a t 38 "C for 2 h. After harvesting and lysis, the fusion proteins were recovered by centrifugation. Computer Analysis-Sequences were edited and analyzed on an IBM PC using the Beckman Microgenie sequence software (Queen and Korn, 1984).

B

A

23.1, 9.9 L

6.61 4.4,

2.0-

C

1 2 3 4 5 6 7 8 ~

-F

23.1-

-

9.9 P

6.6

4.4-

I

.

., v

2.0r

FIG. 1. Southern blot analysis of mammalian cell DNA following digestion with various restriction endonucleases. Panel A, HeLa cell DNA probed with 5' '/3 of La cDNA; panel B, HeLa cell DNA probed with 3' 2/3 of La cDNA; panel C, lanes I and 2, human (HeLa); lanes 3 and 4, mouse (L929); lanes 5 and 6, hamster (BHK); and lanes 7 and 8, bovine (MDBK) cell DNA probed with 3' 2/3 of human La cDNA. Samples were digested with EcoRI (lanes I, 3,5, and 7)or Hind111 (lanes2,4,6, and 8). The blot was washed at 37 "C, but all bands were also visible after washing a t 65 "C.

RESULTS

Organization of the Human La Gene-Two DNA fragments representing the 5' l/3 and 3' */3 of the La cDNA (Chambers and Keene, 1985), but overlapping by 100 nucleotides, were radiolabeled by nick translation and used for Southern blot analysis. Similar patterns of reactive bands were observed with both probes when analyzing various restriction enzyme digestions of HeLa cell DNA (Fig. 1, A and B ) . It appears that approximately three or four copies of the La gene may reside in human DNA. The 3'-derived cDNA probe, which included the 3"noncoding region (Fig. lB), detected fewer restriction fragments than the 5'-probe(Fig. LA), suggesting that the 3"region may have diverged among the La genes. For example, pseudogenes of La may have retained 5'-cDNA sequences but lostsequences homologous to the3"probe. All

bands that did react with the3"probe were subsets of those detected with the5"probe. When human LacDNA was used to analyze genomic DNA from other mammalian species, distinct bands were observed in mouse, hamster, and bovine (Fig. IC). The blot shown in Fig. 1C was washed a t 37 "C (low stringency), but all bands remained visible, although weaker in lanes 3-8, after washing a t 65 "C. Thus, it appears that the La gene sequence is conserved among these species but may exist in fewer copies than in the human genome. Human La cDNAs were used to screen human genomic libraries produced in X Charon phage (Lawn etal., 1978). One clone contained a region of DNA representing the 6.6-kb EcoRI band which is a member of a doublet (Fig. L4). DNA

Structure of Human La Protein La cDUA

Organization of the Human La Gm

-90

16.81

ALa2-l A La 262

14.41

-

E E

-50

CCGGCGGCGC

14.61

E E

E E E

7

-

18045

-30

E

CTOTGGGCCG

-60

-80

-70

TGGGAGGTOG

AGTCGTTGCT GTTGCTGTTT GTGAGCCTGT GGCGCGGCTT

-40

Y

-20

40

GMCCTTAM

QATAGCCGTA

15 ATG GCT G M M T GGT GAT AAT C I A AAG M e t Ala G I ” Asn G l y Asp A s ” G l u l y s

E 7

H 1 Kb

E=EcoRIsite

m = Cxm seqlencc

FIG. 2. Physical map of the human La gene. Two overlapping clones were isolated from an Hue111 or AluI partially digested DNA library from human fetal DNA. The approximate locations of the 11 exons are shown as solid boxes. The sizes of major EcoRI fragments I, 11, and 111 as detected by Southern blot analysis in Fig. 1are shown in brackets.

. 390

.

I 405

375

OAT OCAACTCTT GAT GAC I T A A M GAA TGG TTA GAA GAT A M GOT Pro Thr Aep Ala Thr Leu Asp A.p I l e L y s G l u Trp Leu G 1 u ASP Lym G l y

T T C CCAACT

Pho 435

360

420

C U OTA CTA M T AT1 CAO ATG AGA AGA ACA TTGCAT M A OCA sequencing of this clone showed a long stretch of 400 nucleoO h V a l LOU A s n Ilc O l n net Arg A r g Thv L e u His L y s Ala tides that were similar to the cDNA coding sequence (see 510 495 ATT GAA TCTGCT M G A M TTT GTA GAG ACC CCT below) but contained numerous mismatches and closed readI l c Glu S e r A h Lys Lys Phe Val Glu Thr Pro ing frames. The 6.6-kb band was not part of the active La 540 525 5 x oac CAO MG TAC AM OM ACA GAC CTG c t A ATA CTT TTC AAG GAC GAT TAC TTT gene (Fig. 2). Thus, it appeared that this EcoRI fragment Oly O h Ly. Tyr Ly. Glu Thr Asp Leu Leu Ile Leu Phe L y s Asp Asp Tyr Phe represents an inactive pseudogene of the La protein. 615 UlO sa5 600 Two other genomic clones were also isolated and were found GCC AMAM M T G M GAA AGA M A C M AAT M A GTG G M GCT M A I T A AGIGCT A l a Ly. Lys h n Olu G l u A r q Lys G l n Aan L y s V a l G l u Ala Ly. u.L A r q Ala to contain overlapping restriction enzyme patterns (Fig. 2). 645 660 675 630 Nucleotide sequence analysis of these clones revealed open AM CAO OAO C M O M OCA M A C A I AAG TTA GAA GAA GAT GCT G M ATG M A TCT Lyo O h Olu O l n Olu 11. L y s G l n L y s Leu Glu G l u l a p Ala Glu net Lye Ser reading frames corresponding to the La cDNA. Thus, the 720 705 690 EcoRI fragments shown in Fig. 1, A and B, of 6.8,4.6, and 4.4 CTA ou OM MO ATT OGA TGC TTG CTG AM rm TCG GOT GAT TTA GAT GAT CAG WU G l U O1U Ly. 11. G l y Cy. Leu Leu Lys Phe Ser G l y Asp Leu Amp Amp G l n kb appeared to represent an active human La gene. Based 780 73s 750 upon the finding that the 3’ cDNA probe hybridized in situ 7 6 5 ACC TOT AOA O M GATTTACAC ATA CTTTTC TCA M T CAT GOT G M ATA M A TOG Ibr W . Arg Olu AOp Leu His Ile Leu Phs Ser n.A Him G l y G l u Xle L y s Trp to human chromosome 2: we also screened a human chro810 826 v795 mosome 2 library and isolated genomic clones representing ATA OAC TTC OTC AGA GOA GCA A M GAG OGG ATA ATT CTA TTT AM GAA AM GCC 11. Amp Ph. Val Arg Gly Ala Lym G l u G l y 11. 11. Lou Ph. Ly. Glu L y e Ala EcoRI fragments of 6.8, 4.6, and 4.4 kb. a40 855 885 Sequence of the Human La cDNA-OverlappingcDNA 8 7 0 M G O M GCA TTG GGT M A GCC AM GATOCA AAT M T GGT AAC CTA CAA TTA AGO Lys Glu A h L e u G l y L y s A l l Lys Amp A1. A a n Amn Gly A m L e u G l n Leu A r q clones were isolated as described (Chambers and Keene, 945 930 eo0 915 1985), sequenced on both strands, and found to contain a M C AM O M GTG ACT TGG G M GTA CTA G M GGA GAG OTG GAA A M G M OCA CTG A s n Lys GlU V a l Ihr Tlp G l U Val Lou G l U Gly G l U V a l Qlu LYS G l u Ala Lou single open reading frame of 1224 nucleotides (Fig. 3). The 960 975 990 total length of La cDNA was 1620 nucleotides excluding the M O AM ATA ATA GAA GAC C M C M O M TCC CTA M C A M TOG M G TCA A M GGT 11. Ile G l u ASP G l n G l n G l u S8r Leu A m Ly. Irp LYE SCr L y S Gly Ly. Ly. (po1y)A tail. In addition, the 5’-end contained an AUG codon 1020 1005 103s 1050 within a consensus sequence for ribosomal initiation (Kozak, COT AGA TTT A M GGA M A GGA M G GOT M T AM OCTGCC CAG CCT GGG TCT GOT Arg Arg Phe L y s G l yL Y S Gly L y s G l y A m Ly. Ala A h Gln Pro G l y Se? G l y 1983). The predicted amino acid sequence encompasses a 1065 loa0 1095 protein of 408 amino acids and amolecular mass of 46.8 kDa. AM GGA A M GTA CAG TTT CAG GGC AAG A M ACG M A TTT GCT AGT GATGAT GAA LYS V a l Gln Phc G l n G l Y Lys Lys Thr Lys Phe A l a Scr Asp Asp G 1 u LYS G l Y A single base sequencing error deleted a C residue at position Y 1110 1125 1x 0 1155 292 and resulted in our previous mis-assignment of a termiCAT GAT GAA CAT GAT G M AAT GOT OCA ACT GGA CCT GTG A M AGA GCA AGA GAA His ASP G l u His ASP G l U ASn Gly Ala I h r G l y Pro V a l LYS A r g A l a A r q G l u nation codon of an earlier cDNA clone (Chambers and Keene, 1985). Northern blot analysis of HeLa cell RNA, using the two 1230 1250 1240 1250 1210 major overlapping cDNA clones of 0.4 and 1.4 kb usedin Fig. GGA GAC CAG T AGTTT AGTAMCCM GTTTTTATTC ATTTTAAATA GGTTTTAAAC G l y Asp G l n 1, revealed a single band of 1.8 kb (Fig. 4, A and B ) . This 1280 1290 1300 1310 1320 1330 same result was obtained using poly(A)-selected (Fig. 4, lanes GACTTTTGTT TGCGGGGCTT TTAMAGGM AACCGAITTA GGTCCACTTC AATGTCCACC 1 ) as well as total cell RNA (Fig. 1, lanes 2 and 3 ) . These 1340 1350 1360 1370 1380 1390 TGTGAGAMG G A M M T T T T TTTGTTGTTT M C T T G T C T T TTTGTTATGC AAATGAGATT findings suggest that the cDNA clones account for the full length 1.8-kb mRNA when 100 to 200 nucleotides of (po1y)A are included. 1500 1490 1460 1480 1470 1510 Primer Extension Analysis of La mRNA-To precisely asTCTTCCATTA MTTGCCTTT GTMTATGAG AATGTATTAG TACAAACTM M CT-T sign the transcription start site of the La mRNA, oligonucle1520 1530 15IO ATATACTATA ToI*blbAGC “ M A A M A otide primers derived from near the 5’-end of the cDNA sequence were usedfor primer extension of isolated HeLa cell FIG. 3. Nucleotide and predicted amino acid sequence of mRNA. The primers are shown with solid underlines within human La cDNA. Large boxed region, extended RNP consensus the genomic sequence of exons 1 and 2 in Fig. 5A. The 5’- motif3; small box, core RNP consensus sequence described by Adam proximal primer resided in the predicted 5”noncoding region et al. (1986). Arrowheads, intron-exon junctions. of the cDNA (exon l ) , and the distal primer overlapped the

v

J. Hozier, J. C. Chambers, and J. D. Keene, unpublished data.

C. C. Query, R. C. Bentley, and J. D. Keene, submitted for publication.

Structure of Human La Protein

18046

human H-ras proto-oncogene (Ishii et al., 1985a); however, the spacing of the H-ras sequence (-83 nucleotides) differs from that of the La gene. The H-ras and epidermal growth 1 2 3 1 2 3 factor genes may not be of the housekeeping type because they are involved in growth control. In addition to the three elements discussed above, there are two upstream regions a t -230 and at -295 that form short direct repeats anddyad symmetry, respectively (Fig. 5A). The significance of these regions is notknown a t present, but they may represent another promoterelement. Comparison of the cDNA sequence with the genomic sequence using both restriction enzyme mapping and sequence analysis indicated the presence of 10 introns in the La gene were sequenced (Figs. 2 and 3). The intron-exon junctions that contained motifs that matched the standard mammalian consensus splice sites (data not shown). One intron resided in the 5”noncoding region of the LamRNA (Fig. 5A). Structural Featuresof the La Protein-The predicted amino acid sequence of the La protein (Fig. 3) shows that approximately 40% of the amino acids are charged, with equal proportions of acidic and basic amino acids. No striking charge distributions were noted,andthe predicted molecule was generallyhydrophilic. Chanet al. (1986) have previously FIG. 4. Northern blot analysis of HeLa cell mRNA probed demonstrated that the La protein canbe split into approxiwith the 5‘ l/j (panel A ) and the 3’ Vi (panel B ) of the La cDNA. Lanes 1, 10 p g of poly(A) selected RNA; lanes 2, 10 pg; lanes mate halves by a number of proteases. One half contains all of the methionine residues and the other containsall of the 3, 20 pg of total HeLa cell RNA. phosphoamino acids. Based on the La sequence, we predict that proteolytic cleavage occurs in theregion between position predicted initiationcodon (exon 2). As shown inFig. 5B, both 224 and 248, where many basic and acidic residues are found. primers produced extension products of a heterogeneous na- Under these circumstances,all six methionine residues would ture, but predominant bands were evident. In addition, the reside in the NHz-terminalpeptide, and the C-terminalpeppatterns of extension productswere identical for both primers, tide would contain allof the phosphoaminoacids. Six serines and the lengthsof the products indicated that they extendedand five threonines are found in this segment. Two-dimento the same point on the RNA template. The three major sional gel analysisdetects between five and sevencharge bands observed (indicated by arrows in Fig. 5) are likely to forms of the La protein (data not shown), most of which are represent the major initiation sites for mRNA transcription. related to phosphorylation (Francoeur et al., 1985). Phosphoof proThe presence of multiple initiation sites is typical serine is the predominant phosphoamino acid in the La promoters lacking the TATA element (Melton al., et 1986). tein, and phosphotyrosine has not been detected (Pizer etal., Fig. 5A shows vertical arrows to indicate the locations of 1983). the predominant primer extension products that were located The predicted secondary structure of the La protein (Fig. in the upstreamgenomic sequence. We conclude that the 5’7) is predominantly a-helical (Chou and Fasman, 1974; Garend and -start sitesof the La mRNAs are5 to 25 nucleotides beyond the 5’ end of the cDNA clones shown in Fig. 3. Thus, nier et al., 1978). All predicted helices were further analyzed the La cDNA as shown in Fig. 3 encompasses the complete for evidence of stability conferredby amphipathicity, salt bridging (Sundaralingham et al.., 1987), and aminoacid usage open reading frame of the La protein. The full length La at the helix termini (Richardson and Richardson, 1988). An protein was translated in vitro using an SP-&derived tranunusually long helix-rich region is predicted near the center script and comigrated with HeLa cell La protein in SDSof the protein which, if continuous, would span 38 helical polyacrylamide gels (data not shown). turns. The following evidence suggests that portions of this Structural Features of the La Gene-Three notableuppredicted helix could exist ina stable conformation. 1) Helical stream sequences with similarities to known regulatory elewheel and helical net analyses show that much of the Cments are apparent in Fig. 5A (Breathnach and Chambon, 1981). A consensus CCAAT box a t position -135 is identical terminal half of the long helixis amphipathic,suggesting that to sequences shown to bind multisubunit proteins (Chodosh i t is capable of stable hydrophobic interaction with the preet al., 1988) and is followed by an Spl consensus “G/C box” sumed hydrophobic core of the C-terminal structuraldomain. a t -110. In addition, examination of the genomic sequences 2) A large number of salt bridges could be formedthroughout upstream of the La mRNA start sitesshows the presence of the putativehelix, with a density of 0.84 salt bridges per turn a G/C-rich repeat a t -30 to -46 (Fig. 5A) and the absenceof overall and local densities greater than1.0 in some portions. a TATA box. Furthermore, a predominance of CG dinucleo- Only molecules with long, stable, solvent-exposed helices, tides is evident in this region. G/C-rich promoters are char- such as calmodulin, troponin C, and myosin heavy chains, acteristic of “housekeeping genes” as opposed to promoters have been shown to have salt bridge densities approaching that are thought be to of a more “regulated type” (Dynan and 1.0 per turn (Sundaralingam et al., 1987; Landschulz et al., Tjian, 1985; McKnightandTjian, 1986). The function of 1988). Three positions in the helix-rich region have a low been local salt bridge density, coincident with predicted turns in these sequences as promoters of La transcription has not the helix, and these are represented as discontinuities in the investigated. Examples of known genes with sequence similarity to the long helix in Fig. 7. 3) Computer searchesperformed with the La -30 to -45 sequence are shown in Fig. 6. The strongest La sequence revealed sequence similarity to the troponin C similarity in these examples is to the upstreamregion of the family (40% identical over 40 residues, La amino acids 219-

A

B

Structure of Human La Protein

A

18047

I

. ) 4""" TTTGAGAACTACTACAACACGATGTGTGTGTGTGTATTCACACACATTTTTGG~TATCT """

2

CGCCGTCCAGAAATCCTTAAAAATAAATGTTTATACTAACACAAATCAGCAGCCATTCCC

(- 130) (-110) CTACAACCTGGAGTACTTTTTAAC~~~T~TTTGGGATT~~GGCGGGA~CCAGGGT

GCTGCCGGCGGCGCTGGGAGGTGGAGTCGTTGCTGTTGCTGTTTGTGAGCCTGTGGCGCG GCTTCTGTGGGCCGGAACCTT 4 AATGGTGAT...

TGAGTAA........

exon 1 I

ATAGCCGTAATGGCTGAA e Met Ala Glu I exon 2

1

I

intron 1

,,

Asn GlyAsp

FIG.5. Analysis of the mRNA start site and upstreamsequences of the human Lagene. Panel A, the transcriptional initiation sites, as determined by primer extension (panel B ) , are indicated by vertical arrows. Oligonucleotides used as primers are underlined by solid lines. The G/C-rich region, of similarity to other promoters, that lacks TATA sequences is underlined by shaded bar. Horizontal arrows indicate dyad symmetry (broken line) and direct repeat (solid line). Panel B, DNA sequence using upstream primer and extension products and using upstream primer (lane I ) and downstream primer (lane 2 ) .

258) and to the myosin heavy chain family (34% identical, resistant domains of the La protein are bridged by a long, with gapping,over 131 residues, La amino acids 202-302), solvent-exposed helix. Calmodulin and troponin C are both both of which are highly a-helical. cleaved by trypsin within analogous solvent-exposed helices The similarityof myosin heavy chain and La protein is not(Anderrson et al., 1983; Drakenberg et al., 1987), and a trypsin well preserved at the level of the regular heptadrepeats cleavage site of troponin C lies inthe region of similarity with characteristic of coiled-coil structures. Many proteins with the La protein. Thus, we presume that the putative long ahelical structure are similar to myosin by computer search helix in the La protein contains sites sensitive to numerous (Doolittle, 1986), and we interpret this similarity as further serine proteases. evidence that a long a-helix exists in the central region of the Sequence similarity tothe"RNP consensus"sequence La protein. Troponin C is a dumbbell-shaped protein with (Adam et al., 1986) was evident in the human La protein a t globular NH2-terminal and C-terminal domains separated by amino acids 150 to 160 (Figs. 3 and 7). The similarity over a a long a-helix in which three helical turns are fully exposed region of 80 amino acids for La protein is not as strong as to the solvent (Herzberg and James, 1985). The region of that of 25 other known RNA-binding proteins, although the similarity with the La protein corresponds exactly with the most strongly conserved residues are present in the La prothat a region flankingan beginning of the troponinC solvent-exposed region, and 7 of tein?We haverecentlyshown the first 10solvent-exposed residues are identical with the Laextended R N P consensus motif of the 70-kDa U1RNAsequence. Therefore,itis possible that the two protease- binding protein issufficient for specific binding to U1RNA.3

Structure Laof Human

18048

""_

Promoter

Protein

panel a, lane 7), and ovalbumin (Fig. 8, panel a, lane 8 ) . Equivalent amountsof fusion proteins were loaded on three otheridentical gels which were subsequentlyblottedonto Human Lagene I I nitrocellulose filters (Fig. 8, panels b, c, and d ) . The blotswere I I -63 I reacted with either antibodies and1251-StaphylococcusA proHuman H-ras proto-oncogene I G $ G tein or with32P-labeled VA-RNA (Fig.8, panel d ) . I I -?3 When antibodies specific for @-galactosidase were used as Human EGF receptor T G C y! G an internal standard to compare therelative levels of fusion I I I protein in each lane, each La-containing protein, as well as I -58 @-galactosidase (Fig. 8, panel b, lune 6 ) , was reactive. When Human ptubulin CT GAT C A I I the sera of the La autoimmune specificity from six patients I I -?3 I were combined and immunoblotted, strikingdifferences were Mouse Thy 1.2-1 T C;CT T F observed among the different portions of the La protein(Fig. I 8, panel c). It appeared that the middle portion, fragment C, I -2s Mouse Thy 1.2-2 TTT~T TA GG contained a major epitope of the La antigen.It is also evident, 1- - - - -1 however, that fragments A and D were immunoreactive. pFIG. 6. Nucleotide sequences of the upstream G/C-richshort Galactosidase did not react with sera from the autoimmune repeat of the La gene and other similar genes. Sequence data patients as shown previously (Chambers andKeene, 1985). It were from Ishii et al. (1985a) for H-ras, Ishii et al. (198513) for epidermal growth factor receptor, Lee et al. (1983) for @-tubulin,and is important to note that immunoreactivity of any protein, including La, could differ when present as a fusion protein Ingraham and Evans (1986) for thy 2.1 genes. due to steric interference. In several studies to date using La protein, the immunoreactivity of authentic La and the La Protease-rensltive reglo" fusion proteins was indistinguishable (St. Clair et al., 1988a, Melhlonlne-contalnlng domaln Phosphorylated domaln 1988b). We conclude that atleast three distinct autoimmune RNP Core Camensue Sequetre ' &' epitopes reside on the Lamolecule and that themiddle third, 4 which contains the extended R N P motif and muchof the long N- 4 C stretch of a-helix (Fig. 7), may contain an immunodominant epitope. -A"C . D To study the RNA-binding activity of these portionsof the I B I La protein, we synthesized VA-RNA and @-globinmRNA i n uitro using the T7 and the SP-6 RNA polymerase systems, I I I 100 200 300 400 respectively. When VA-RNA was used in the "Northwestern blot" assay, the HeLa cell La protein of 47 kDa was labeled a-helix non-helical backbone clusters of basic residues (Fig. 8, panel d, lane I ) as shown previously (van Eckelen et FIG. 7. Predicted secondary structure of the human Lapro- al., 1982). In addition, a small amount of binding occurred as tein (Garnier et al., 1978).The hatched bones represent regions predicted to adopt an a-helical structure, and the thick solid line a smear of bands of up to100 kDa insize. These may indicate represents the non-helical backbone. The segments marked A, B, C, the presence of other HeLa cell proteins that can bind VAprobably notdimers of Laprotein because and D represent regions of the La protein that were expressed as 0- RNAbutare galactosidase fusion proteins. The similarity with the myosin heavy Western blots have never demonstrated a La protein species chain family extended from La amino acids 202 to 332 and corre- of this size. Even under these high stringency salt washes, a sponded to the "short S2" region of myosin heavy chain (Strehler et al., 1986). The similarity with troponin C extended from La amino very small amount ( 4 0 % ) of binding of the @-globinRNA acids 219 to 258 and corresponded tothe D/E helix bridging the NH2- was also observed withthe HeLa cell La protein (not shown). and C-terminal domains of troponin C (Herzherg and James, 1985). As shown in Fig. 8, lane 3 of panel d, the B fragment of the 2/3 of the molLa protein, representing the carboxyl-terminal ecule and containing the RNP consensus sequence, bound the Thus, we predict that the RNP consensus region of La interRNA most strongly. No RNA binding was detectable with acts with RNA. It is interesting to note that an intron falls within the apparent RNP consensussequence (Fig. 3). Thus, fragment A, the amino-terminal3/' of the La protein. Similar this RNP decamer sequence appears to be encoded on two results were obtainedusing a solutionbindingassayand immunoprecipitation (data not shown). separate exons. Although the La B fragment was able to bindVA-RNA in Regions of Antigenic Recognition andR N A Binding-As an the Northwesternfilterbinding assay,whenexpressed as initial approach to understanding the properties of autoimdecreased mune recognition and RNA binding by the La protein, we separate halves, La C and LaD, the binding affinity constructed four recombinant subfragments under control of significantly(Fig. 8, paneld, lanes 4 and 5 ) . These filter the lac Z promoter in X-phage lysogens. As shown in Fig. 7, binding resultswere also obtained in competition experiments three approximatelyequal portions were expressed separately in which unlabeled VA, @-globinmRNA, and tRNAs were so that the large blocks of central a-helix were transected a t added during the blotting procedure. Thus, the binding of a predicted turn (LaA, La C, and La D). In addition, a larger VA-RNA to the recombinant La protein fragments appeared fragment representing the carboxyl-terminalY 3 of the mole- t o be specific. Low amounts of binding to control RNAswas occasionallyobserved, however, suggesting that the RNAcule (La B)was also expressed. Fusion proteins of La protein expressed in E. coli Y1089 binding portion of La proteinmay also have alow affinity for lysogens were analyzed by polyacrylamide-SDS gel electro- other RNA molecules. This was not unexpected because the phoresis (Fig. 8, panel a). Approximately equivalent amounts sequence specificity of RNA recognition by La protein may of partially purified fusion proteins were loaded in each lane be limited to terminal uridylate residues (Rosa et al., 1981; Reddy et al., 1983; Stefano, 1984; Francoeur et al., 1985; Keene and stained with Coomassie Blue dye. Control proteins inwashes (0.5 M ) clude a HeLa cell extract (Fig. 8, panel a, lane 1 ), P-galacto- et al., 1987a, 1987b). Highstringencysalt VA-RNA. sidase (Fig. 8, panel a, lane 6 ) ,bovine serum albumin (Fig. 8, eluted the control RNAs but not the

:+ +: -?9 CCG?'GCCCTGCCqACCC :

:

:

---

I

I

I

Structure of Human La Protein

a

b

18049

d

C

1 2 3 4 5 6 17 20 3 4 5 6 17 28 3 4 5 6 7 0

9 7-

68-

47-

1 2 3 4 5 6 7 8

--

* 97-

9 7-

97-

68-

68-

6&

47-

47-

4 7 "

e

c

*

0

FIG.8. Analysis of antibody- and RNA-binding activities of immobilized La fusion proteins. La fusion proteins were prepared using cell lysis and centrifugation (Query andKeene, 1987). Protein samples were electrophoresed in 6% polyacrylamide-SDS gels. Lane I, 100 pg of total HeLa cell proteins; lanes 2-5, 10 pg of La fusion proteins A, B, C, D, respectively; lane 6, 5 pg of @-galactosidase; lanes 7 and 8, 10 pg of bovine serum albumin and ovalbumin, respectively. Panel a, gel stained with Coomassie Blue; panels b and c, gels electroblotted onto nitrocellulose and probed with @galactosidaseand Laantibodies, respectively, followed by ''sI-labeled protein A; panel d, gel electroblotted and probed with "P-labeled VA-RNA synthesized in vitro from a cloned VA gene (m241-32) using T 7 RNA polymerase; binding conditions were as described by van Eckelen et al. (1982) except by each of the four fusion that the finalwash was in 0.2 M NaCI; panel e, the portionof the La protein represented proteins (A, B, C, and D). LA FUSION PROTEIN

serumconcentrationsbutarenot rigorously quantitative. More precise quantitation of these differences will require a A B C D refined analysis by optical density determinations, antibody depletion, and antigencompetition experiments (St. Clair et e 0 I al., 1988a, 1988b). Additional experiments of this type are in 2 progress and essentiallyconfirm these findingsfrommore a LA 3 than 100 serum samples. By using amicrotiter assay, however, PATIENT 4 0 fragment A has proved more strongly immunoreactive than 5 0 e shown here with nitrocellulose.4 This approach may prove 6 usefulforclinical correlations of epitoperecognition with FIG.9. Immunodot blot analysis of expressed fragmentsA- progression of disease or co-occurrence with other autoimD of human La protein using sera from six patientsthat were mune specificities. We conclude fromthese data that antisera mixed in Fig. 8. panel c. Purified fusion proteins were spotted and from patients of the La specificity recognize three separate tested asdescribed by Chambers and Keene(1985) and by Query and regions of the protein and immunodominant regions appear Keene (1987). to exist. From allof our studies to date, it seems that fragment C, encompassinga highly charged portion of thecentral The dataof Fig. 8, panel c, indicate that patientsof the La helical core, contains the most common and strongestepitope. autoimmune specificity produce anitibodies that react preDISCUSSION dominantly with the middle, strongly a-helical portionof the La protein. However, other regions of the molecule were also Data presented in this study suggest that antigenic and reactive. Because amixture of serum sampleswas used in the RNA-binding domains appear to reside in a highlyhelical and antibody binding assays, we further examined individual La positively charged region of the La molecule. At least three sera using a direct immuno dot blot assay similar to that separate autoimmune epitopes are present on molecule. the It described previously (Chambers and Keene,1985). As shown is likely that each thirdof the molecule contains more than in Fig. 9, purified fusion proteins representing fragments A, one epitope. Some of the epitopes, especially the assembled B, C, and D were spotted ontonitrocellulose strips and reacted topographic sites (Benzofsky, 1985), may have been disrupted with serum from each of the six patients used in Fig. 8. As or enhanced inreactivity by our choice of subcloning sites in expected, each of the serareacted to some extent withall four generatingthethree domains. It isalso possible that Bfusion proteins, although longer exposures were sometimes galactosidase can influence immunoreactivity of an autoanrequired to detect fragments A and D. However, four of the tigen.We cannotdetermineintheseassayswhetherthe six serum samples reactedmost strongly with fusion protein C, whereas patients 2 and 3 reactedmorestronglywith 'E. W. St. Clair, D. S. Pisetsky, and J. D. Keene, unpublished fragment D. Theseresults were identical over a range of data.

a

*

*

18050

Structure of Human La Protein

different reactivities of the Laepitopes were due to differences in the titersor the affinities of different autoantibodies. Also, we cannot distinguish continuous (linear) from discontinuous (conformational) antigenic determinants. However, it is clear that autoimmune sera showed distinct differences in their reactivities to the three domains of the protein. We are currently investigating whether the apparent immunodominance of certain regions of La protein is related to theclinical profiles of the patients. When this work was initiated, it was not clear that autoantigenic proteins could be derived by screening cDNA libraries with sera from autoimmune patients, in part, because the role of discontinuous epitopes in autoimmune reactivity was unknown. It became evident that linear peptide sequences are autoantigenic because various short fragments can be isolated by cDNA cloning using autoantibodies (Chambers and Keene, 1985). Whittingham et al. (1987) have also reported the isolation of partial La cDNAs. Analysis of several different autoantigens cloned in our laboratory has indicated that multiple epitopes reside along these Query and Keene, 1987; proteins (Fresco et al., 1987; Deutscher et al., 1988).5 Although our molecular dissection is limited by the large size of the expressed fragments, we conclude that the RNAbinding portion of the La molecule appears to overlap with the two most immunoreactive segments. It is possible that regions of proteins that bind to nucleic acids possess characteristics that tend to be moreimmunogenic during the autoimmune response. However, both the C and D regions of La protein appeared to be required for efficient RNA binding. In producing the C and Dsubclones, fragment B was transected in a region of predicted turns that linked two predicted ahelices. Recent evidence from our laboratory indicates that a region of 124 amino acids flanking the RNP consensus sequence is required for specific binding of a 70-kDa protein to U1 RNA.3 Comparative studies among several RNA-associated proteinshave revealed an extended RNP consensus motif of 80 amino acids that may be involved directly in RNA recognition. In addition, Bugler et al. (1987) reported that the ribosomal RNA-binding protein, nucleolin, contains the RNP consensus octamer within an RNA-binding fragment. Data presented here for La protein and cross-linking studies reported by Chanand Tan (1987) are compatible with the suggestion that the region containing the extended RNP consensus motif is involved in RNA recognition by La protein. Immediately C-terminal to the end of the extended RNP motif is a remarkable region in which basic residues are positioned approximately every four residues over seven helical turns (Fig. 7, residues 185 to 208). Most of these residues are in positions unfavorable for salt bridging. Thus, they are available to form a narrow band of basicity wrapping halfway around the helix over seven helical turns. This basic helical configuration may bea candidatefor an RNA-binding region. A similar basic region is seen from position 266 to 287, although thissecond domain contains only five basic residues per seven helical turns. The distribution of introns in relation to putative structural domains suggests that the La protein has a modular organization, containing a series of structurallydistinct regions encoded byseparate exons. The extended RNP motif, each of the clusters of basic residues, and the region sharing similarity with the troponinC solvent-exposed helix are all closely bounded by introns (Fig. 3). Various parameters have been demonstrated to be useful in predicting antigenic determinants in proteins. These include local average hydrophilicity (Hopp and Woods, 1981), segS. D. Deutscher, J. B. Harley, and J. D. Keene, unpublished data.

mental mobility (Westhof et al., 1984), and accessibility to large probes (antibody domains) (Novotny et al., 1986). Previously, we used antibody selection of various expressed cDNA fragments of La protein to approximate autoantigenic regions (Chambers and Keene, 1985). In the present study, we have used recombinant DNA constructions to generate specific fragments of the La protein for the approximation of autoimmune epitopes. This approach allowed the dissection of the protein into four separate portionsbut remains limited in the ability to assign absolute binding values. However, it is clear that patients do not react equally to all regions of the La protein. Nonetheless, in our experience, any of the four fragments was sufficiently immunoreactive to allow an absolute diagnosis of every La serum tested. Significant differences in the immunoreactivities among sera from patients demonstrates thepotential of this approachfor subcategorization of autoimmune specificities. This also opens the possibility that immune reaction to theLa protein may evolve in recognition of epitopes throughout the course of disease. Followinginitial recognition of a single immunoreactive portion of a selfprotein, for example, as the result of cross-reactivity to an infectious agent or breakdown of tolerance, the authentic protein may be presented and recognized as non-self. As an immune complex involving a single initial epitope, the autoantigenic protein may undergo an expansion of epitopes and a progressive evolution of antibody recognition along different domains of the protein (Query and Keene, 1987). Acknowledgments-We thank Jane Richardson and David Richardson for their advice with computer modeling of protein structure and close associates David Pisetsky and Bill St. Clairforhelpful discussions during the course of this work. REFERENCES Adam, S. A., Nakagawa, T., Swanson, M. S., Woodruff, T. K., and Dreyfuss, G. (1986) Mol. Cell. Biol. 6,2932-2943 Andersson, A., Forsen, S., Thulin, E., and Vogel, H. J. (1983) Biochemistry 22,2309-2313 Benzofsky, J. A. (1985) Science 229,932-940 Breathnach, C., and Chambon, P. (1981) Annu. Rev. Biochem. 50, 349-388 Bugler, B., Bourbon, H., Lapeyre, B., Wallace, M. O., Chang, J.-H., Amalric, F., and Olson, M. 0.J. (1987) J. Biol. Chem. 262,1092210925 Chambers, J. C., and Keene, J. D. (1985) Proc. Natl. Acad. Sci. s. A . 82,2115-2119 Chambers. J. C.. Kurilla. M.G., and Keene, J. D. (1983) J. Bioi. Chem. 258,11438-11441 Chan, E. K. L., and Tan, E. M. (1987) Mol. Cell. Biol. 7 , 2588-2591 Chan, E. K. L., Francoeur, A. M., and Tan, E. M. (1986) J. Zmmunol. 136,3744-3749 Chodosh, L. A., Baldwin, A. S., Carthew, R. W., and Sharp, P. A. (1988) Cell 5 3 , 11-24 Chou, P. Y., and Fasman, G . D. (1974) Biochemistry 13,222-245 Deutscher, S. D., Harley, J. B., and Keene, J. D. (1988) Proc. Natl. Acad. Sci. U. S. A., in press Doolittle, R. F. (1986) Of URFS and ORFS: A Primer on How to Analyze Derived Amino Acid Sequences, p. 60, University Science Books, Mill Valley, CA Drakenbere. T., Forsen. S., Thulin, E., and Vogel, H. J. (1987) J. Biol. Che;. 262,672-678 Dynan, W. S., and Tjian, R. (1985) Nature (Lond.)316, 774-778 Francoeur, A. M., and Mathews, M. B. (1982) Proc. Natl. Acad. sci. U. S. A . 79,6772-6776 Francoeur, A. M., Chan, E. K. L., Garrels, J. I., and Mathews, M. B. (1985) Mol. Cell. Biol. 5 , 586-590 Fresco, L. D., Kurilla, M. G., and Keene, J. D. (1987) Mol. Cell. Biol. 7,1148-1155 Gamier, J., Osguthorpe, D. J., and Robson, B. (1978) Mol. Biol. 120, 97-120 Habets, W. J., den Brok, J. H., Boerbooms, A. M. Th., van de Putte, L. B. A,, and van Venrooij, W. J. (1983) EMBO J. 2, 1625-1631

u.

La Structure of Hu!man Hardin, J. A., and Thomas, J. 0. (1983) Proc. Natl. Acad. Sci. U. S. A. 80,7410-7414 Herzberg, O., and James, M. N. B. (1985) Nature (Lond.) 3 1 3 , 653659 Hopp, T. P., and Woods, K. R. (1981) Proc. Natl. Acad. Sci. U. S. A. 78,3824-3828 Ingraham, H. A., and Evans, G. A. (1986) Mol.Cell.Biol. 6, 29232931 Ishii, S., Merlino, G. T., and Pastan, I. (1985a) Science 230, 13781380 Ishii, S., Xuy, Y., Stratton, R. H., Roe, B. A., Marling, G. T., and Pastan, I. (1985b) Proc. Natl. Acad. Sci. U. S. A. 82, 4920-4924 Keene, J. D., Chambers, J . C., and Martin, B. J. (1987a) in DNAProtein Interactions and Gene Regulation (Thompson, E. B., and Papaconstaniou, J., eds) pp. 107-116, University of Texas Press, Galveston Keene, J. D., Deutscher, S. L., Kenan, D., and Kelekar, A. (1987b) Mol. Biol. Rep. 12, 235-238 Kieber-Emmons, T., and Kohler, H. (1986) Proc. Natl. Acad. Sci. U. S. A. 83,2521-2526 Kozak, M. (1983) Microbiol. Reu. 47, 1-45 Kurilla, M. G., and Keene, J. D. (1983) Cell 34,837-845 Kurilla, M. G., Cabradilla, C. D., Holloway, B. P., and Keene, J. D. (1984) Virology 50, 773-778 Landschulz, W. H., Johnson, P. F., and McKnight, S. L. (1988) Science 240, 1759-1764 Lawn, R. M., Fritsch, E. F., Parker, R. C., Blake, G., and Maniatis, T. (1978) Cell 15, 1157-1174 Lee, M. G.-S., Lewis, S. A., Wilde, C.D., and Cowan, N. J. (1983) Cell 33,477-487 Lerner, M. R., and Steitz, J. A. (1979) Proc. Natl. Acad. Sci. U. S. A . 76,5495-5499 Mattioli, M., and Reichlin, M. (1974) Arthritis Rheum. 17,421-430 Maxam, A. M., and Gilbert, W. (1977) Proc. Natl. Acad. Sci. U. S. A. 74,560-564 McKnight, S., and Tjian, R. (1986) Cell 46, 795-805 Melton, D. W., McEwan, C., McKie, A. B., and Reid, A. M. (1986) Cell 44,319-328

Protein

18051

Novotny, J., Handschumacher, M., Haber, E., Bruccoleri, R.E., Carlson, W. B., Farming, D.W., Smith, J. A., and Rose,G.D. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 226-230 Pizer, L. I., Deng, J. S., Stenberg, R. M., and Tan, E. M. (1983) Mol. Cell. Biol. 3, 1235-1245 Queen, C., and Korn, C. J. (1984) Nucleic Acids Res. 12,589-599 Query, C. C., and Keene, J. D. (1987) Cell 51, 211-220 Reddy, R., Henning, D., Tan, E., and Busch, H. (1983)J. Biol. Chem. 258,8352-8356 Richardson, J. S., and Richardson, D.C. (1988) Science 240, 16481652 Rinke, J., and Steitz, J. A. (1982) Cell 29, 149-159 Rosa, M. D., Gottlieb, E., Lerner, M. R., and Steitz, J. A. (1981) Mol. Cell. Biol. 1 , 785-796 St. Clair, E. W., Pisetsky, D. S., Reich, C. F., Chambers, J. C., and Keene, J. D. (1988a) Arthritis Rheum. 31,506-514 St. Clair, E. W., Pisetsky, D. S., Reich, C. F., and Keene, J. D. (198813) J. Immunol., in press Sanger, F., Coulson, A. R., Barrell, B. G., Smith, A. J. H., and Roe, B. A. (1980) J. Mol. Biol. 143, 161-178 Stefano, J. E. (1984) Cell 36, 145-154 Strehler, E. E., Strehler-Page, M.-A., Perriard, J.-C., Periasamy, M., and Nadal-Ginard, B. (1986) J. Mol. Biol. 190, 291-317 Sundaralingham, M., Sekharadu, Y. C., Yathindra, N., and Ravichandran, V. (1987) Proteins 2 , 64-71 Tan, E. M. (1982) Adu. Zmmunol. 33, 167-240 Towbin, H., Staehelin, T., and Gordon, J. (1970) Proc. Natl. Acad. Sci. U. S. A. 76, 4350-4354 van Eckelen, C., Buijtels, H., Linne, T., Ohlsson, R., Philipson, L., and van Venrooij, W. J. (1982) Nucleic Acids Res. 10, 3039-3052 Venables, P. J. W., Smith, P. R., and Maini, R. N. (1983) Clin. Exp. Zmmunol. 54,731-738 Westhof, E., Altschuh, D., Moras, D., Bloomer, A. C., Mondragon, A., Klug, A,, and Van Regenmortel, M. H. V. (1984)Nature (Lord.) 311,123-126 Wilusz, J., and Keene, J. D. (1984) Virology 135, 65-73 Whittingham, S., Naselli, G., McNeilage, L. J., Coppel, R. L., and Sturgess, A. D. (1987) Lancet ii, 1-3