Eur. J. Immunol. 1990.20: 2127-2131
Kenth Gustafssono, Mikael Karlsson", Leif AnderssonO and Rikard HolmdahP Department of Animal Breeding and Genetics, Biomedical Centero, Swedish University of Agricultural Sciences and Department of Medical and Physiological Chemistryv, Biomedical Center, Uppsala, University, Uppsala
I-A restriction of collagen-induced autoimmunity
Structures on the LA molecule predisposing for susceptibility to type I1 collagen-induced autoimmune arthritis* The susceptibility to type I1 collagen (CI1)-induced arthritis (CIA) in mice is profoundly influenced by major histocompatibility complex (MHC) class I1 genes in the H-2 region. Analyses of MHC-congenic strains on the B10 background show that only strains developing an anti-CII antibody response after immunization with autologous CII develop arthritis after induction with CII from various species. The susceptible haplotypes have been found to be H-2q, H-2', H-zW3and H-2w17.In addition, these haplotypes respond to different patterns of CII derived from various species suggesting that T cell receptors and CII peptides interact. In contrast, certain haplotypes closely related to H-2q, such as the H-2p and H-2w5 haplotypes, are resistant to induction of CIA and are nonresponders to CII. We have earlier shown that a critical structure on the I-Ap molecule determines the susceptibility differences between the p and q haplotypes. We have now determined the structure of exon 2 of the Ap as well as some of the A, genes of the remaining haplotypes in the p, q and r families. The sequences show similarities between the CIA-susceptible haplotypes in the Ap C-terminal part and the A, N-terminal part of the first domains forming a large part of the antigenic peptide-binding site. Among the wild mouse-derived haplotypes, the w5 haplotype showed an Ag sequence identical to that of the p haplotype consistent with its nonresponder nature to CII immunization.These findings suggest that (a) structures shared between different class I1 molecules are of importance for the susceptibility to disease in mouse strains and (b) most likely recognition of different CII peptides is important for development of disease.
lar interaction between (a) a certain class I1 molecule on APC, (b) a certain autoantigenic peptide or a class of The potential genetic influence of polymorphic MHC structurally related peptides and (c) certain sets of TcR on class I1 molecules on autoimmune diseases has during autoreactive Tcells is of critical importance for susceptibilrecent years been highlighted by the finding of strong ity to RA. However, at present the nature of the autoanassociations between certain MHC class I1 haplotypes and tigens and of the autoreactive T cells involved is unclear. diseases such as insulin-dependent diabetes and rheuma- Moreover, there is limited knowledge on the biological role toid arthritis (RA). Development of RA has long been of such an interaction. Although this interaction might be a known to be weakly associated with the DR4 specificity [l]. critical step in the pathogenesis, it is likely that other It is now clear that a much stronger linkage is present with genetic and environmental influences are necessary for the certain subtypes of the DR4 and DR1 haplotypes and development of disease. sequence information reveals that critical residues for RA susceptibility are located at positions 67,70,71 and 74 in the To answer these questions animal models will be of utmost DR-B molecule . These positions are found in the importance. One such model is the collagen-induced proposed antigenic peptide recognition site (ARS) of the arthritis (CIA) which can be induced with cartilage-specific class I1 molecule .This finding indicates that trimolecu- type I1 collagen (CII) in mice, rats and apes [4-61 .There are several similarities between RA and CIA . First, a similar histopathology of the joints develops with occurrence of both marginal erosions and pannus formation as well as massive polymorphonuclear cell infiltration in acute lesions [I 85661 and edema formation in the synovium. Second, autoimune * This work was supported by the Swedish Medical Research reactions to both CII and IgG (rheumatoid factors) occur in Council, Swedish Research Council for Forestry and Agricul- both diseases.Third, susceptibility t o both RA and CIA are ture, Swedish Cancer Society, Craaford foundation, King Gus- influenced by gender, female sex hormones and pregnancy. tav V's 80-years foundation and Riksforbundet mot Rheuma- Lastly, both CIA and RA are associated with certain MHC tism. haplotypes [2, 81. Correspondence: Kenth Gustafsson, Department of Animal Breeding and Genetics, Swedish University of Agricultural Sciences, Biomedical Center, Box 596, S-75124 Uppsala, Sweden Abbreviations: ARS: Antigen-peptide recognition site CIA: Type I1 collagen-induced arthritis CII: Type I1 collagen RA: Rheumatoid arthritis 0 VCH Verlagsgesellschaft mbH, D-6940 Weinheim, 1990
In the mouse, the susceptibility to arthritis after immunization with heterologous CII is restricted to the H-24 and H-2' haplotypes .With recombinant strains the susceptibility in the H-2'7 haplotype has been mapped to the I-A locus in the MHC class I1 region [S]. The susceptibility to CIA of strains expressing H-2'7 is valuable for the analysis since the q haplotype is a member of a closely related family 0014-2980/90/0909-2127$3.50+ .25/0
K. Gustafsson, M. Karlsson, L. Andersson and R. Holmdahl
of haplotypes consisting of q and p, expressed by inbred strains, but also of w3, w5 and w17 derived from wild mice captured around the world [lo] .We have recently described a minor structural difference between the H-2q haplotype and the H-2p haplotype which may explain the difference in susceptibilityto CIA in these haplotypes; the difference is restricted to four amino acids in the Ap molecule (residues 85,86,88 and 89; [ll]). Expression of I-E did not influence CIA susceptibility or antibody responses as shown with experiments with F1 animals. Thus, only a dominant influence of AS can explain the MHC class I1 association of CIA development and CII responsiveness. We have now extended this analysis by sequencing the A, first domain exon of the A: and A t genes and the first domain exon of the A;, Ar3 and Ar5 genes. Together with earlier published sequences, this information is compared with the responder status of B10 congenic strains expressingthese haplotypes with respect to susceptibility to CIA and development of anti-CII antibody responses after immunization with heterologous as well as autologous
2 Materials and methods 2.1 Mice BlORIII, B10D2, BlOS, B10 and BlOG mice, obtained from Jackson Laboratories Inc., Bar Harbor, ME, and BlOP, B10CAS2, B10SAA48 and BlOKEA5 mice, from Prof. Jan Klein,Tubingen, FRG,were kept and bred in the animal unit at the Biomedical Center, Uppsala, Sweden. 2.2 Polymerase chain reaction (PCR)analysis and DNA sequencing
Eur. J. Immunol. 1990. 20: 2127-2131
Synthesizer (Foster City, CA). Four different primers were used for the amplification of exon 2 coding for the p chain first domain; LA1, 5'-ATAGGATCcTGACCGCGTCCGTCCGCAG-3'; LA2,5'-ATAGGATCCCGCGCTCACCAAG-3'; LA4,5'-GAGAAlTCACCAAGCCGCCGCAGGGA-3'; and for the amplification of cDNA coding for the a chain first domain the following primers were used; LA20, 5'-TGAGGATCCGAAGACGACATTGAGGCCGA-3'; LA21, 5'-GTAGAATTCACATTGGTAGCTGGGGTG-3'.The PCR was carried out using 2.5 U of thermostable Taq polymerase (PerkinElmer, Cetus, Emeryville, CA) according to the manufacturer's recommendation in a final volume of 100 pl. The amplification was done with 1 pg genomic DNA and approximately0.1 pg cDNA, respectively,and oligonucleotide primers at a concentration of 1 pM. The amplification from genomic DNA was carried out in two steps such that primers LA1 and LA2 were utilized in the first step from which 1 pl of the resulting amplificationwas transferred to a new round of amplification using primers LA1 and LA4 (LA4 being internally located with respect to LA2). Denaturation was carried out at 94 "C for 30 s, annealing at 68°C for 1min and extension at 72°C for 2min. The amplified products were cloned into M13mp18 and/or M13mp19 and the nucleotide sequence determined using the chain termination method .The sequence of at least two different M13 clones was determined in order to detect PCR misincorporations.
3 Results and discussion 3.1 H-2linkage of CIA susceptibility and CII autoantibody response
We have earlier investigated the H-2 association of the autoimmune response and development of arthritis after immunization with heterologous as well as with autologous Genomic DNA samples from BlORIII, BlOCAS2, CII . These data are schematically depicted in Table 1. B10SAA48 and BlOKEAS strains were prepared from liver Immunization of B10 congenic strains with autologous CII tissue as previously described .Total RNA samples from led to development of arthritis and/or anti-CII autoantibodBlOQ, BlOCAS2, B10SAA48 and BlOP strains were pre- ies in H-2', H-2q, H-2W3and H-2w17strains but not in H-2P, pared from spleen tissue as previously described . H-zW5,H-2b, H-2d or H-2Sstrains . Immunization with mRNA samples were prepared from total RNA using various heterologous CII induced more severe arthritis in oligo(dT) coupled to magnetic beads [Dynabeads strains with H-2s and H-2I haplotypes and, in addition, oligo(dT)B, Dynal AS, Oslo, Norway]. Approximately induced an anti-CII autoantibody response but no arthritis 100 pg of total RNA was mixed with 150 pg of oligo(dT) in some strains with other haplotypes [15, 161. These beads in 0.5 M LiC1, 10 mM Tns-HCI (pH 7 . 9 , 1 mM findings suggest that the induction of arthritis is critically EDTA and 0.2% SDS in a total volume of 200 pl. After dependent on activation of auto-CII-reactiveTcellsbut that 2 min of incubation at room temperature the beads were activation of hetero-CII-reactive T cells may enhance the separated on an MPC-E magnetic test tube rack for 1 min severity and elevate the autoantibody response to CII. (Dynal AS) and the SN was discarded. The beads were Analysis of the anti-CII T cell response in DBA/1 (H-29) subsequently washed twice in 200 pl 0.15 M LiCI, 10 mM mice has shown that Tcells to heterologousCII recognized a Tris-HC1(pH 7.5) and 1mMEDTA.The boundmRNAwas major immunogenic epitope common to many different eluted in 25 p12 mM EDTA at 65 "C for 2 min and separated heterologous CII (chicken, bovine, human, rat) but not to immediately from the beads in the magnetic rack for 1min. mouse CII .In contrast, the anti-CII T cell response in cDNA was synthesized using 5 p1 of the eluted mRNA in BlORIII (H-2') mice is most likely directed to CII derived 50 mM KCl, 10 mM Tris-HC1(pH 8.3), 4 pM MgC12,20 mM from a more limited number of species (pig, bovine; ), dNTP, 20 U RNAsin (Promega Corp. Madison,WI), 90 U suggesting that the hetero-CII-reactiveT cells in the two pd(N)6 hexamer (Pharmacia LKB Biotechnology, Upp- haplotypes recognize different CII peptides. Finally, the T sala, Sweden) and 10 U AMV reverse transcriptase (Prom- cell recognition of autologous CII seems to occur in both ega Corp.) in a total volume of 25 pl. H-29 and H-2' and, at least in DBA/1 (H-29 mice, involves recognition of epitopes other than those recognized by T Oligonucleotide primers for PCR analysis were synthe- cells recognizing heterologous CII. Thus, at least three sized on an Applied Biosystems 380A Oligonucleotide different clonal populations of T cells recognize different
Eur. J. Immunol. 1990. 20: 2127-2131
I-A restriction of collagen-induced autoimmunity
Table 1. MHC linkage of susceptibility to collagen-induced autoimmune arthritisa)
a) Summary of the MHC linkage of susceptibility to CIA and I-A I-E development of anti-CII autoantibody response from an ***I++ ***I++ ***I++ ***I+ experiment earlier described in DBAll 9 ***/+ ***/+ **/+ ***Idetail . Incidence of arthriN F m 9 **I++ **I+++ **I++ -I+ tis 15 weeks after immunization BlOG 9 with CII derived from various BlOP P P -/+ 4-/species. Occurrence of arthritis BlOKEA5 w5 P -/+ -/-/+ -/induced with heterologous CII B10SAA48 w3 w3 *I -/+ -/+ **I+ **I+ -/+ has been recorded after macros*I **I+ w17 BlOCAS2 -/+ **I+ copic examination and arthritis ND ***I+ + r r BlOFUII -/+ 4induced with mouse CII with 4+ B10 b -/+ 4both macroscopic and micros-/B10D2 d copic examination. Incidence of arthritis is given by 51%-100% arthritis (***), 11%-50% arthritis (**), 1%-10% arthritis (*), and 0% arthritis (-). Mean levelsof anti-CII autoantibody in serum are expressed by >250 kg/ml (+++), 25-249 pg/ml (++), 1-24 pg/ml (+), and < 1 pglrnl (-).
Incidence of arthritis after immunization with ChickenCII BovineCII Rat CII MouseCII
CII peptides bound to the presently investigated H-2 haplotypes. 3.2 Structures of I-A molecules of importance for anti-CII autoimmunity
We have sequenced the first domain exons of the A, and Ag chains of several haplotypes related to the q,p family and other haplotypes (r) conferring susceptibility t o CIA (Fig. 1).The deduced amino acid sequences, together with sequences from other known A, and Ag molecules, are shown in Fig. 2.We can now confirm our earlier prediction that the A: and A: are identical, further emphasizing the importance of the four-amino acid differences, at positions 85, 86, 88 and 89, of which position 86 is predicted to constitute part of the ARS between the P chains of the same haplotypes [111. Of particular interest is the A' molecule since H-2' haplotype mice (BlORIII) are also susceptible to CIA in spite of no previously known relationships between the r and q class I1 molecules. However, there is a striking similarity between the predicted A'and Aq molecules in the regions Ag68-96 and A,13-68 (three variant positions between q and r) which form a large part of the ARS including the earlier described structure which differs between q and p. In contrast, the remaining part of the ARS shows extensive differences between r and q. In addition, the As molecule shows similarities to the Aq molecule although the s molecule has two threonin residues in the P-pleated sheat which may be of importance for the binding of a collagen peptide. BlOS mice do not respond t o CII immunization , while H-2S haplotype strains with other genetic backgrounds have been shown to develop arthritis (the Biozzi strain; ) or an anti-CII antibody response (the SJL strain; ). Moreover, the B10 (H-2b) strain responds to bovine CII  and has a similar sequence at residues 68-96 of the Ag chain to that of the r haplotype. It islikely that the different CII peptides bind to the A*molecule with a similar, but not identical, structural orientation, as they bind to the AS molecule. However, we have not ruled out the possibility that I-E molecules, expressed in BlORIII, may contribute to CII responsiveness
An DNA sequences 8 7 8 9 10 11 1 2 13 14 15 15 17 18 19 20 Blffi(q) CAT T I C GlG UX CAD TrG AAG 01: Oag Toc TAC TTC ACC AAC G Blop(p) ............................................. BIOKEAS (w5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . TA. . . . . . . 8 1 OSAA4B(w3). . . . . . . . . CA. ... ..C ........................... BlORlll (r) TAC ... ..C ...... .C. . . . . . . . . . . . . . . . . . .
21 2 2 23 24 25 26 27 28 2 9 30 31 32 3 3 3 4 3 5 BlOG (4) f f i CAD Coc ATA 0 3 4 TCT GTG AAC AGA TAC ATC TAC AAC CG WG Blop lP) ............................................. BlOKEM (ws) B j o S A A 4 8 ( ~ 3 ).......................... . -1. . . . . . . . . . . . . . . . B(ORIII (r) . . . . . . . . . . . . ..G ... .C. .....................
37 38 09G TGG GIG
39 4 0 Coc T I C
4 2 43 AQ2 W
GlG 01: G4G
47 4 8 TAC Coc
B ~ O K E A(w5) ~
6 1 OSAA48(w3). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . BlORlll (r) ... .AC C.. ... .A. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51 1 2 5 3 54 5 6 5 8 57 6 B 69 60 6 i 6 2 63 6 4 6 5 BlOG(q) AM: G4G CIO (ro CCO CCA Gu: QX Glc. TAC TGG AAC AGC ffi CCG BlOP lP) ............................................. B10KEA5 ( ~ 5 ) BlOSAA48(w3). BlORlll (r) ..................... C.. . . . . . . .AC ..T . A 0 f.
88 6 7 88 G4G ATC CTO
ow ..................... .....................
BlOG(q) wo Blop (p) BIOKE& (M) 8 1 OSAA48(w3). . . . . . . . . . . . . . . . . . . . . BlORlll (r) TAC * * * . . . . . . . . . . . .
81 82 83 84 85 BlOG(q) CAC AAC TAC Wyi (ro BlOP (p) .A. B10KEA5 (w5) . . . . . . . . . . . . .A. BlOSAA48(w3) 1.. . . . . . . . . . . . . BlORlll (r) ...............
... ... . . . . . . . . . . . . . . . . . ... ... . . . . . . . . . . . . . . . . . . ... ... . . . . . . . . . . . . . . . . . . ... ... c . . . . . . . . . . . . . . . .
76 77 GAC f f i
7 8 7 9 60 GTG TOC AGA
18 19 CAG TCT
OAO T I 3
37 38 TAT GTG
GlG WG AM: CAC AM: AC. AC. C.. C..
... ... ... ...
14 16 ATA GTr
-C- C... ...
(bl A, DNA sequences 7 BlOG(q1 C B10P
10 11 GTA 01: TTC
1 6 17 GTA TAT
23 24 26 20 27 26 29 3 0 31 3 2 BlOG (q) ATT 01: CAD TAT ACA CAT GAA TTT GAT OClT & l o p (p) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
39 4 0 4 1 4 2 43 4 4 4 5 4 6 4 7 4 8 4 9 5 0 5 3 52 5 3 54 BlOG(q) OIY: T I G GAT AAG AAG GG ACT GTC TGG AT0 CTT 0.7 G4G TTT CC CAA ............................................. Blop(p)...
66 BlOG(q1 TrG B1Dp
5 8 59 5 0 TTT G X Coc
61 62 C M CGT
8 4 66 66 67 68 CTA CAA AAC ATA OCT
69 70 ACA OOA
71 72 73 74 B l f f i (9) AAA CAC AAC T I G
76 I33 W2
7 8 7 9 60 81 ACT AAG KG TCA
82 83 AAT TT
Figure 1. (a) Nucleotide sequences of the Ap first domain exons from BlOG (H-2q), BlOP (H-29, BlOKEA5 (H-2w5),B10SAA48 (H-Zw3)and BlORIII (H-2'). (b) Nucleotide sequences of the A, first domain exons from BlOG (H-24) and BlOP (H-29. Numbers denote deduced amino acid positions in the mature protein.
K. Gustafsson, M. Karlsson, L. Anderson and R. Holmdahl
although the finding that four different I-E-deficient haplotypes (q,w17, b, s; ) respond to CII suggests a dominant role of the A' molecule for collagen presentation.
Ag AMINO ACID SEQUENCES
Eur. J. Immunol. 1990.20: 2127-2131
Sequence analysis of genes coding for Ap molecules derived from wild mice also show similarities with the A: chain. First, the predicted AF5 chain, expressed by BlOKEAS mice resistant to CIA, was identical to AP corroborating the suggested importance of the proposet critical restriction site in comparison with the A; chain. Second, the AF3 molecule, expressed by B10SAA48 mice which responded to mouse CII, both with arthritis and anti-CII antibody responses, but only with limited responses to the heterologous CII, displayed large similarities with the A; chain (five exchanges). Among the variant amino acids it has a proline at position 86 and a phenylalanine at position 11 in the P-pleated sheet in similarity with the A' molecule. Third, the chain, expressed by BPOCAS2 mice which develop arthritis after immunization with bovine and chick CII and an antibody response after immunization with mouse CII, was more similar to q than p in the C-terminal part of the molecule.
3 1 32 33 34 35 36 37 38 39 40 41 42 4 3 44 4 5 46 47 48 49 50 5 1 5 2 53 54 55 56 57 58 5 9 60 I Y N R E E W V R F D S D V G E Y R A V T E L G R P D A E Y
P v5 w3 w17
3.3 Concluding remarks
f nod 9
W5 w3 w17 S
d b h12
lb) &AMINO ACID SEQUENCES
Figure 2. Deduced amino acid sequences corresponding to the Ag (a) and A, (b) first domain exons of various haplotypes.For the Ap sequences, the s, k, u and f , b , d , p and q [ll], nod , bm12  and w17  sequences have been published earlier whereas w3, w5 and r derive from the present investigation. Of the A, sequences,the s, k and r , b, d, u and f  and part of the q  sequences have been published earlier whereas q and p derive from the present investigation. Numbers denote deduced amino acid positions in the mature protein.
The immune response and arthritis development after immunization with heterologous and autologous CII in mice do not only involve activation of several different Tcell clones but also many different CII peptides which may display varying degrees of immunodominance in the different mouse strains. It is therefore surprising that the immune response to heterologous CII and to autologous CII as well as the development of CIA are all linked to the H-2q and H-2' haplotypes although with some variation in the response to heterologous CII. A structural explanation for these phenomena might be the similarities found in a large part of the antigenic peptide-binding site (Ap68-96 and AJ3-68) of these haplotypes. Analyses of the Ag chain gene from two other mouse haplotypes (w3 and w17) responding to mouse CII after CII immunization further emphasize the importance of this part of the I-A molecules. Thus, the Aw3 chain displayed similarities with both A; and A; wkile the chain showed similarities preferentially in the C-terminal part. Amino acid position 85-89 of the Ap molecule appears to be of critical importance for the immune recognition of CII peptides since we have now found two nonresponder mouse haplotypes (p and w5) which differ only in that region as compared with the Aq molecule. Comparisons of a large number of Ag sequences from different haplotypes show that the 85-89 region contains a relatively high degree of variability (Fig. 2). The three-dimensional structure of position 85-96 in the Ap molecule is difficult to interprete from the HLA-A2 crystal since there are large sequence differences in this region between class I and class I1 molecules. However, this region seems to have an important function for the binding of peptides to class11 molecules; it is closely interacting with parts of the A, chain, and it displays a large degree of polymorphism but is unlikely to form an a helix . A possible explanation for the structural similarity between the class I1 alleles permitting a CII immune response might be that CII peptides, have a common structural requirement for binding to the class I1 molecules, irrespective of whether they originate from different parts of the CII molecule or from different species. The extremely conserved nature and the repetitive tripeptide structures
Eur. J. Immunol. 1990.20: 2127-2131
building the triple helical part of this protein are in line with such an assumption. In addition, due to the common occurrence of prolines in the structure it is possible that CII peptides do not form an a helical structure in the antigenic peptide site but will tend to form a triple helical structure. The structural similarities found on different I-A molecules restricting the susceptibility to CIA, in spite of the complicated pattern of immune recognition leading to this disease in the different mouse strains, may cast some light over mechanisms involved in human RA. Also in RA a common structure has been found on a number of DR B molecules strongly associated with RA . The RA is a syndrome based on a number of clinical criteria and to explain the specificity of the disease implied by the suggested MHC association it is more likely that the binding of a class of peptides, rather than that of a single specific antigenic peptide to the ARS, is one of the critical steps for development of the disease. In CIA we suggest this peptide is derived from collagen and therefore this possibility should also be investigated in RA. Received May 18, 1990.
4 References 1 Stastny, I?, N. Engl. J. Med. 1978. 298: 869. 2 Gregersen, P. K., Silver, J. and Winchester, R. J., Arthritis Rheum. 1987. 30: 1205. 3 Brown, J. H., Jardetzky, T., Saper, M. A., Samraoui, B., Bjorkrnan, P. J. and Wiley, D. C., Nature 1988. 332: 845. 4 Courtenay, J. S., Dallman, M. J., Dayan, A. D., Martin, A. and Mosedal, B., Nature 1980. 283: 666. 5 Trentharn, D. E.,Townes, A. S. and Kang, A. H., J. Exp. Med. 1977. 146: 857. 6 Yoo,T. J., Kim, S.Y., Stuart, J. M., Floyd, R. A., Olson, G. A., Cremer, M. A. and Kang, A. H., J. Exp. Med. 1988. 168: 777. 7 Holmdahl, R., Int. Rev. Immunol. 1988. 4: 1.
I-A restriction of collagen-induced autoimmunity
8 Wooley, P. H., Luthra, H. S., Stuart, J. M. and David, C. S., J. Exp. Med. 1981. 154: 688. 9 Wooley, P. H., Luthra, H. S., Griffiths, M. M., Stuart, .I. M., Huse, A. and David, C. S., J. Immunol. 1985. 135: 2443. 10 Peck, A. B., Darby, B. and Wakeland, E. K., J. Zmmunol. 1983. 131: 2432. 11 Holmdahl, R., Karlsson, M., Anderson, M. E., Rask, L. and Andersson, L., Proc. Natl. Acad. Sci. USA 1989. 86: 9475. 12 Chyrnczynski, P. and Sacchi, N., Anal. Biochem. 1987. 162: 156. 13 Sanger, R., Nicklen, S. and Coulson, A. R., Proc. Natl. Acad. Sci. USA 1977. 74: 5463. 14 Holmdahl, R., Jansson, L., Gullberg, D., Rubin, K., Forsberg, P. 0. and Klareskog, L., Clin. Exp. Zmmunol. 1985. 62: 639. 15 Holmdahl, R., Jansson, L., Anderson, M. and Larsson, E., Immunology 1988. 65: 305. 16 Holmdahl, R., Klareskog, L., Andersson, M. and Hansen, C., Immunogenetics 1986. 24: 84. 17 Anderson, M. and Holmdahl, R., Eur. J. Immunol. 1990.20: 1061. 18 Bouvet, J. I?, Couderc, J., Bouthillier,Y., Franc, B., Decreusefond, C. and Mouton, D., J. Immunol. 1989. 143: 1537. 19 Nowack, H., Hahn, E. and Timpl, R., Zmmunology 1975.29: 621. 20 Vu,T. H.,Tacchini-Cottier, F. M., Day, C. E., Begovich, A. B. and Jones, P. P., J. Immunol. 1988. 141: 3654. 21 Wakeland, E. K., Boehme, S. and She, J. X., Immunol. Rev. 1990. 113: 207. 22 Estess, I?, Begovich, A. B., Koo, M., Jones, I? P. and McDevitt, H. O., Proc. Natl. Acad. Sci. USA 1986. 83: 3594. 23 Larhammar, D., Hammerling, U., Denaro, M., Lund, T., Flavell, R. A., Rask, L. and Peterson, P. A., Cell 1983. 34: 179. 24 Malissen, M., Hunkapiller,T. and Hood, L., Science 1983.221: 750. 25 Acha-Orbea, H. and McDevitt, H. O., Proc. Natl. Acad. Sci. USA 1987. 84: 2435. 26 McIntyre, K. R. and Seidman, J. G., Nature 1984. 308: 551. 27 Davis, C. B., Mitchell, D. J.,Wraith, D. C.,Todd, J. A., Zamvil, S. S., McDevitt, H. O., Steinman, L. and Jones, P. I?, J. Immunol. 1989. 143: 2083. 28 Landais, D., Matthes, H., Benoist, C. and Mathis, D., Proc. Natl. Acad. Sci. USA 1985. 82: 2930. 29 Benoist, C. O., Mathis, D. J., Kanter, M. R.,Williams,V. E . , I1 and McDevitt, H. O., Cell 1983. 34: 169.