STRUCTURE AND DIVERSITY OF HLAmB27

0022-1767/90/14410-4036$02.00/0

THEJOURNAL OF IMMUNOLOGY

Vol. 144.4038-4045. No. 10. May 15. 1990 Prtnted tn U.S.A.

Copyright 0 1990 by The American Association of Immunologists

STRUCTUREANDDIVERSITYOFHLAmB27-SPECIFIC

T CELLEPITOPES

Mutants Mimicking HLA-B27 Subtype Polymorphism' Analysis with Site-Directed V~CTORCALVO, SUSANA ROJO, DANIELLOPEZ, B E G O ~ ~GALOCHA, A JOSE A. LOPEZ DE CASTRO~

AND

From the Department of Immunology, FundacionJimenez Diaz, Consejo Superior d e Inuestigaciones Cientificas, Avda, Reyes Catolicos. 2. 28040-Madrid,Spain

HLA-B27 subtype polymorphism is amenable to surface glycoproteins, encoded in humans by the HLAdifferential recognition by CTL, Site-directedmuta- A, B, and C loci, that control recognition of modified (i.e., genesis was used to construct a series of HLA-B27 infected) self cells, and of foreign cells, by CTL. Recogmutants reproducing mostof the changes occurring nition of infected cells by CTL is accomplished through in the natural subtypes. The reactivity of 21 anti- the presentation of viral peptides bound to class I moleHLA-B27 CTL clones was examined withthese mu- cules (1).TCR specifically recognize the complex between tants to address three issues concerning the alloreactive response against HLA-B27: 1) diversity of the self-HLA molecule and thepeptide. The three-dimensional structure of a class I Ag, HLAclonotypic specificities,2) structural featuresof the A2, (2, 3) shows that thepolymorphic a1 and a2regions epitopes recognized by these clones, and 3)role of individual positions in the differential recognition of the molecule are folded to form a groove which is of HLA-B27 subtypes. Virtually all CTL clones dis- thought to be the foreign Ag binding site. Polymorphism played unique reaction patterns with the mutants, at this site modulates the specificity of peptide binding indicating a corresponding diversity of epitopes. and the interaction with the TCR. Phenomenologically, However, theseshare somemolecular features, allorecognition implies just recognition of a n alloantigen such as certain amino acid residues and related molecule. Nevertheless, it is thought that T cell allorecoglocations. Individual mutations induced complex ef-nition might also involve presentation of endogenous fects on multiple B27-specificCTL epitopes, reveal- peptides. This idea is fostered by the location of many ing some of their very precise stereochemical con- substitutions affecting T cell allorecognition in the pepstraints. An important feature of HLA-B27 subtype tide binding groove, sometimes at positions that do not polymorphism is that every individual change was appear to be easily accessible to direct contact with the relevant, altering recognition by many CTL clones. TCR (3).In addition, there is increasing suggestive eviAlthough the specific set affected by each mutation dence that peptides can modulate recognition by allowas partially different, the global numberof clones affected by most changes was very similar. This reactive T cells (4-9). Among class I Ag, HLA-B27 has special interest besuggests that theantigenic profile ofany given subit is strongly linked to the development of ankyloscause type is not dominated by one particular change but ing spondylitis (10, 1 1). Studies in several laboratories is uniquelydefined by its corresponding set of changes. An exception was the change a t position have demonstrated the existence of at least six subtypes 152, which totally abrogated recognition by all 20 of this Ag in the human population. They are designated anti-B*2705CTL clones. This effect decisively influ-as B*2701 to B*2706and differ from each other by just ences the profound differencesin T cell recognition a few amino acid changes, that arelocated in the peptide between B*2705and thetwo subtypes, B*2704and binding site of the molecule (1 2).Because of its location, B*2706,carrying this change. Theresults are com- subtype polymorphism induces significant, sometimes patible with the idea that HLA-B27 allorecognition critical, differences inCTL recognition, so that B27 submay involve multiple peptides boundto thealloan- types are, inspite of their structuralsimilarity, functiontigen onthe cell surface. ally distinct alleles. MHC class I Ag include the highly polymorphic cell Received for publication November 10, 1989. Accepted for publication February 27, 1990. The costs of publication of this article were defrayed in part by the payment of page charges. This article must thereforebe hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I This work was supported in part by grants from INSALUD-F.I.S. (87/ 888: 88/ 1806)and fromComision Interministerial de CienciaTecnologia y (PB87/0347). V.C. is a fellow of the Ministry for Education and Science. S . R . and D.L. are fellows of the F.1.S. Address correspondenceand reprint requeststo Dr. Jose A. Lopez de Castro. Department of Immunology, Avda, Reyes Catolicos. 2. 28040Madrid. Spain. The contributionto this articleby V. C. and S. R. is equal, and their orderof authorship is arbitrary.

Previous studies in our laboratory have used the structurally characterized subtypes to analyze the clonal diversity of the alloreactive CTL response againstHLA-B27 and to define some structural parameters of the determinants recognized by these CTL (13-17). An obvious limitation of these studiesis that subtypes usually differ by more than one aminoacid residue. In this report, we haveaddressed the issue of the immunologic relevance of individual positions, among those that arepolymorphic in the HLA-B27 family. This was done by using site-directed mutagenesis to analyze their role in modulating CTL allorecognition at a clonal level. We have also examined the interactive effects of polymorphism at pairs of spatially close positions by the

4038

STRUCTURE ANDDIVERSITY

OF T CELL EPITOPES IN HLA-B27

4039

mAb a n d FMF analyses. Cell surface expression of HLA-B27 use of double mutations. The results indicate that all molecules wasmeasuredwith mAb ME1 (anti-HLA-B27 + B7 + positions examined deeply influence T cell recognition. Bw22) (28) andW6/32 (anti-HLA class I monomorphic determinant) They alsoshow the extraordinary diversity of allospecific (29)by using 100pl of a 1: 100dilution of ascites fluid for bothmAb, CTL epitopes on the HLA-B27 molecule and further de- exactly as describedelsewhere (14). Phenotyping of Tcells was carried out by the same procedure. except that 10 pl of FITC-conjufine theirtopologic relationships and structural features.gated anti-CD3. -CD4, and -CD8 mAb (Coulter Corporation, Hialeah. FL) were used. MATERIALSANDMETHODS

Cell lines. Mycoplasma free, human LCL.3 used a s stimulators in MLC and as target cells were cultured without antibiotics as described elsewhere (14).HMy2.CIR cells, used a s recepients for transfections. were grown in Dulbecco's modified Eagle's medium (Flow Laboratories. Middlesex. UK) containing 10% heat-inactivated FCS (GIBCO Laboratories, Paisley, UK). 2 mM L-Glutamine (Flow Laboratories), and no antibiotics. These cells (a kind gift of Dr. P. Cresswell, Duke University, Durham, NC) are a class I-deficient mutant derived from the human plasma cell leukemia line LICR.LON.HMy2 (18). Site-directed mutagenesis of HLA-B27. A 6.5-kb ECOR I DNA fragment containing the HLA-B.2705 gene (19) (a gift of Dr. H. Coppin. University of Toulouse. Toulouse. France) was used in this study. I t was inserted in pBR322 vector or in the SmaI site of pUCl8 after filling in the EcoR I ends of the HLA-B27 insert with E. coli polymerase I Klenow fragment. Oligonucleotide-mediated site-specific mutagenesis wasperformed in dsDNA plasmid pUC 18 (formost of the mutants) or pBR322 (for the El 52 mutant] by using a previously described procedure (20).adapted for HLA-B27. Briefly. purified supercoiled plasmid DNA was digested separately with Hind 111 to linearize the plasmid in the vector sequence, and with Sma I (for mutations in the a1 coding sequence) or Kpn I (for mutations in 02) to gap theregion in the HLA-B27 gene to be mutated. The resultant molecules were mixed a t equimolar amounts, heat-denatured and renatured in the presence of the corresponding oligonucleotide a t a 20-fold molar excess and at10°C belowits melting temperature. for 2 h. Polymerization was subsequently carried out by primer extension withE. coli DNA polymerase I Klenow fragment and the plasmid was closed with T4 DNA ligase a t 14°C for 15 h (21). The ligation mixture wasused to transform competentE. coli DH5 (22).Colonies harboring mutants were detected by hybridization with the mutagenic oligonucleotide (23). Plasmid DNA from these colonies was digested with appropriate restriction enzymes, usuallySma I, BgZ II and/or K p n 1. and analyzed by Southern blot hybridization (24) to assure correct localization of the mutating oligonucleotide used as primer to the appropriate region of the HLA-B27 gene. DNA from primary colonies was used for retransformation ofDH5 bacteria until colony homogeneity (usually two to threetimes) to obtain purified mutant plasmid DNA. All enzymes were purchased from Boehringer (Mannheim, FRG) or New England Biolabs (Beverly, MA) and used as recommended by the manufacturer. Confirmation of the intended mutation was obtained by DNA sequence analysis using the procedure of Maxam and Gilbert (25). after subcloning the appropriate HLA-B27 fragments intopUC12 or pUC18 vectors. The N63 mutant was sequenced by the dideoxy chain-termination method, after subcloning into M13 m p l 9 vector (26). using Sequenase (USB, Cleveland, USA) as indicated by the manufacturers. For most mutants, the complete coding sequences of crl and a2 were obtained to assure the absence of undesired extra mutations. For the N63,180. N77180. and El52 mutants the coding sequences of only the corresponding exon, a1or a2, wasdetermined. DNA-mediated genetransfer. Plasmids containing HLA-B.2705, B.2702 or the mutant B27 genes, a s well as pSV2neo. containing the selection neo gene, were linearized by digestion with EcoR I (for pUC vectors] or Hind 111 (for pBR322).A mixture containing 10 pg of linearized 827 construct and 1 pg of linearized pSV2neo were cotransfected into lo7HMy2.CIR cells in 1ml of 20 mM HEPES buffer, containing137 mMNaC1. 5 mM KC1, 0.7 mM Na2HP04, 6 mM dextrose. pH 7.05. by electroporation (27) at 950 pF and 300V. Cells were subsequently maintained in 20 ml ofDMEM culture medium containing 10%FCS and 2 mM glutamine for 72 h . After this time, they wereseededintwo 24-well platescontaining 1 ml/well of culture medium supplemented with 1,5 mg/ml G418 (GIBCO, Paisley, UK). Surviving cells in each proliferating well were screened for HLA-B27 expression by FMF analysis. Populations showing a homogeneous FMF profile and high HLA-B27 expression were selected for furtheruse. T cell cloning. MLC and T cell cloning wereperformed as described elsewhere (14). T cell clones were grown with 20 IU/ml of rIL-2 (Hoffmann-La Roche Inc., Nutley. N J ) as described (14)."Cr-release cytotoxicity assays were performed a t various E:T ratios, exactly as described (14). Abbreviations usedin this paper: LCL. lymphoblastoid cell line:FMF. flow microfluorometry.

RESULTSANDDISCUSSION

A series of alloreactive, HLA-Ba7-specific. CTL clones were analyzed for their reactivity with the natural subtypes of this Ag. a s well a s with various mutants reproducing changes occurring in the subtypes. The purpose of this study wasthreefold. First, to outline the diversity of clonotypic specificities generated in allogeneic CTL responsesagainst HLA-B27. Second, to define some structuralparameters of the correspondingepitopes. Third, to establish the role of those positions that are polymorphic amongsubtypesin modulating HLA-B27 antigenicity. Fine speciftcity of HLA-BZ7-specific CTL clones. Twenty B27-specific CTL clones were isolated by limiting dilution ofPBMC from various B27- responder individuals. stimulated in vitro with B*2705+LCL (Table I). Cells were cloned after primary or secondary MLC and were selected on the basis of their capacity to lyse B*2705+ HMy2.CIR transfectant target cells, but not the corresponding cells transfected only with pSV2neo. at anE:T ratio of 4: 1. Various responder donors were used to minimize a possible bias towards particular reaction patterns of the CTL clones due to responder phenotype. In addition,a CTL clone, 64.8P, derived from another B27donor against B*2704, was included in this study. Its specificity has been described elsewhere (17).All clones have homogeneous CD3+ CD8'CD4- phenotypes a s established by FMF analysis. CTL clones were tested for their cytotoxicity towards a panel ofLCL, including those expressing all six known HLA-B27 subtypes. A s shown in Table 11, the clones raised against B*2705 could be classified, according to their reactivity with these subtypes, into six reaction patterns (Table 11, A to F). Each group included clones with similar fine specificity by this criterium. Nevertheless, some heterogeneity was evident among CTL clones within a given group in: a) their lytic efficiency towards B+2705+ targets: b) their relative cytotoxicity towards target cells expressing different B27 subtypes: andc) the occasional cross-reactions of some CTL clones, such a s CTL 23 and 102DRF, with other HLA Ag. The main features of the reaction patterns shown in Table 11, for CTL clones raised against B*2705, are the following: 1) all 20 clones fail to recognize the B*2704 and B+2706 subtypes, with the same two changes at positions 77 and 152,regardless of their reactivity with other subtypesand theresponder from whom they arose. This feature contrasts with the reaction pattern of the anti-B+2704CTL 64.8P (Table 11); 2) 19 of 20 clones fail to recognize either B*2701, B*2702. or both. These two alleles differ from B*2705 only within the segment 7481 and 77-81, respectively; and 3) the B*2703 subtype. differing from B*2705 only at position 59, is recognized by 7 (or 35%)of 20 CTL clones. These data reveal two levels of structural relatedness among the corresponding B27-specific CTL epitopes. First. virtually all epitopes involve residue 152 and one

4040

STRUCTURE AND DIVERSITY OF T CELL EPITOPES IN HLA-B27 TABLE I Responder/stimulator Combinations usedfor deriving CTL clones

Stlmulator

CTL Clone

Responder

LG15(HLA-A32;B*2705:DRl)

PA(HLA-A24,~33;B35,39:DRw6)

LG15

RSS(HLA-A3,24;B*2705.7) GM(HLA-A1,24:87,8:C7)

LG15

BG(HLA-A2;B5.7;C7:DR.4)

LG15

AE(HLA-Al;B8.18;DR3)

KNE(HLA-A1.2;B*2704,8;DR2.3)

2 7 23 26 28 29 40" 46" 64DRF 72DRF 1O2DRF 172DRF 139DRD" 202DRD" 2 2DRD" 1 GM7" 5A2 17A2 B54" G36 64.8P

Obtained by cloning after primary MLC.CTL clones without superscript were obtained after secondary 64.8P was obtained after tertiary MLC (17). a

MLC. CTL

TABLE 11 Reactfon patterns ofalloreactive C T L clones against8*2705 with HLA-827 subtypes" Reactlon Pattern Clone

CTL

8,2705 -

__ 812701 (74. 77.

A

B

46 GM7 5A2

C

2'

D

28 172DRF' 139DRD 202DRD

E

72DRF

23' 40 64DRF 102DRFC 1 2 2DRD 17A2 B54 G36

+++* +++

++ + + ++ ++ +

++

7 26 29

+++ ++ ++ ++ + ++ + +++ ++ +++

64.8P

+++ +++

F

81Ib

8.2702 (77. 80.81)

8.2703 (59) (77.

812704 152)

-

-

-

++

-

-

-

-

+ + ++ ++ ++ +

-

-

-

-

+++ +

-

+++

-

++

++

+

-

+

+++

-

+

-

8,2706 (77. 114. 116. 152) -

-

-

-

-

-

-

++

+

The reaction patternof anti-Bs2704 CTL 64.8P (17) is also included for comparison. The amino acid sequence positions in which 827 eachsubtype differs from Be2705 are given in parentheses. See also Table Ill. CTL 23 cross-reacts witha HLA-DR2 subset (14);CTL 2 and lO2DRF cross-react withHLA-B40* (B*4002),a subtype of HLA-B40. Cross-reactions between HLA-B27 and B40* have been observed with other anti-B27CTL clones and have been discussed elsewhere (13. 14). Percent specific "Cr release a s measured in representative targetLCL from each subtypea t a nE:T ratio of 1:1. The of the clones, following code was used: -(0 to 10%);+( 11 to 30%);++(31to 50%); +++(51 to 100%). The reaction pattern that is, their relative killing with the various subtypes, was maintained at E:T an ratio of 5:1 (data not shown). 'The assignment ofCTL 172DRF to reaction pattern D is tentative, as its lysis of B*2702+ LCL is very close to background levels. However, it killsvery efficiently B*2702+ transfectant HMy2.CIR cells. a

or more residues within the segment 74-81. This indi- morphism influences, in a very drastic manner, recogcates the relevance of such positions for allorecognition nition of HLA-B27 by a set ofCTL clones from multiple significance of HLA-B27 of B*2705, a s previously noted on the basis of a similar individuals. This underlines the analysis carried out with a single donor (16).Second, the subtypes as functionally distinct alleles. similar finespecificity of CTL clones within each reaction HLA-B27 mutant molecules. Eight site-directed HLApattern group means that the joint effect of the changes B27 mutants were constructed to replace,individually or in each of the natural subtypes is similar for all such at two positions, the amino acid residues occurring in B*2705 by those in other subtypes (Table 111). An addiclones. In addition, the results indicate that all subtype poly- tional mutant was alsoconstructed in which residue63,

CELL OF TEPITOPES

STRUCTUREANDDIVERSITY TABLE 111

Amino acid substitutions inHLA-B27 subtypes and mutants Amino Acid Sequence Positiona Molecule

Subtype B.2705 B.2701 B.2702 Be2703 8.2704 B.2706 Mutantb N63 Y74 N77 577 I80 Y74N77 N77180 N77A81 E152

59

63

74

77

80

81

114

116

152

Y

E

D N N

L A A

H

D

V

-

D Y -

T

-

-

H

-

-

-

D

Y

E E

-

-

-

-

-

"

_

-

-

N

-

-

"

-

"

"

I

s - "

S

-

-

-

-

-

y " " -

_

N

s

"

-

" I

-

Y

-

-

-

N N

-

N

-

I -

-

-

-

-

-

A

-

-

-

-

-

-

-

-

IN HLA-B27

404 1

B27 molecule were selected a s target cells for additional experiments. All transfectant cells used expressed similar levels ofHLA-B27 to those ofLCL as estimated by FMF analysis (Fig. 2). Expression was homogenous and stable as assessed by periodic FMF analysis. The CTL clones described in Table I1 were tested for their lytic ability against thepanel of transfectant target cells expressing HLA-B*2705, B*2702, and the mutant B27 molecules described in Table 111. The effect of each

E

"The following single letter code for amino acids was used: A:Ala. C:Cys. D:Asp. E:Glu. F:Phe, G:Gly, H:His.I:Ile,K:Lys. L:Leu. M:Met, N:Asn. P:Pro. 9:Gln. R:Arg. S:Ser, T:Thr, V:Val. W:Trp. Y:Tyr. Dashes indicate identity with 8.2705at thecorresponding positions. 'Mutants were designatedby the single letter code of the amino acid(s] introduced followed by the corresponding position number(s1.

Figure 1 . Spatial location of the mutated residues in the HLA-Be2705 of HLAmolecule. The figureis based on the three-dimensional structure regions. A2 (2)and showsa top view of the a1 and a2

which is either Glu or Asn in class I HLA molecules of known structure (30), waschangedtoitsalternative form. This residue is invariant among HLA-B27 subtypes but is spatially close to residue 59, changed in B*2703. On the basis of the three-dimensional model of HLA-A2 (2, 3),all mutated positionsare located in thea-helices of the foreign Ag-binding site. They appear to be pointing into the groove, although the precise orientation of the side chains could vary among different class I Ag and, thus, be different in HLA-B27. Residues 59 and 63 a s well a s residues 74, 77, and80 are related to each other 3.6 amino acid residues. by the &-helicalturn, that spans Roughly, these mutations are located near three of the four ends of the site (Fig. 1). B*2705, B*2702, and each of the mutant genes were transfected into HMy2.CIR cells, and subpopulations of each transfectant line expressing high (and a s close to one another as possible) levels of the corresponding HLA-

flUOlt8CtRCt

IHTtl8lTT (lOl.t8Clltl

Figure 2. FMF analysis showing the expression of HLA-B27 molecules on human cells. The ME1 mAb (seeMaterials and Methods)was used in ( C )in each panel was these experiments. The negative fluorescence peak separately measured after incubatingcells with only the second antibody. The two upper panels show the FMF profile of the B*2705+LCL LG15. used a s a reference, and ofHMy2.CIR cells transfected only with the pSV2neo plasmid. as a negative control. The remaining panels show the FMF profile of the HMy2.CIR cells transfected with B12705. Bt2702. or the various HLA-B27 mutants as indicatedineachpanel. A similar analysis (data not shown] was carried out with the mAb.W6/32 to account for the possibility that some mutations could alter theME1 determinant. The relative expression levels of HLA-827 and mutant molecules. as measured with this antibody, were the same as with ME1. suggesting that the ME1 determinant was not affectedby any of the mutations. The background signal with W6/32. a s estimated in the pSV2neo control, was I HLA weakly positive, reflecting alow expression of the endogenous class Ag in the HMy2.CIR cells.

4042

STRUCTURE AND DIVERSITY OF T CELL EPITOPES IN H L A - B ~ ~

mutation on recognition by the anti-B*2705 CTL clones

There are two possibilities to explain such diversity. First, many structurally different clonotypic receptors ing features of the B27-allospecific CTL epitopes, as de- could interact with the alloantigen. The observed diverscribed below. sity would reflect the many different types of intermolecClonal diversity. Most CTL clones analyzed possessed ular contactsthat could be established between the HLAdifferent specificity when tested against the panel of B27 molecule and multiple TCR. Indeed, we have anamutants (Fig. 3).regardless of their reactionpattern with lyzed the structure of the VP genes expressed by 12 of HLA-B27 subtypes. Theonly exceptions were CTL 23 and the 21 CTL clones used in this study and have shown 64DRF, which were indistinguishable in this analysis. that they are all different (33).Second, the clonal diverThese two clones, however, differ inthat CTL 23, but not sity could reflect, perhaps to a large extent, recognition 64DRF, cross-reacts with HLA-DR2 (see Table 11). This of multiple peptides constitutively bound to HLA-B27. indicates that eachclone recognizes a structurally differ- Although we favor this second possibility, our data to not ent epitope and shows the extraordinary complexity of allow to distinguish between both alternatives. Furtherthe allogeneic CTL response against HLA-B*2705.Other more, they are not mutually exclusive, as it is possible studies suggest similar diversity in responses against that not all the HLA-B27 molecules on the cell surface HLA-A2 (31, 32). have bound peptide. In addition, a same MHC + peptide is shown in Fig. 3. These resultsreveal several outstand-

Figure 3. Cytotoxicity of anti-HLA€312705 CTL clonesagainst HMy2.CIR transfectants expressing b 2 7 0 2 or the specified HLA-B27 mutants. Results are expressed as the percentage of relative lysis in whichthe lysis of each individual target cell is expressed as a percentage of the specific lysis of the B.2705' HMy2.CIR transfectant at the same E:T ratio. Data are meansof duplicate or triplicate experiments. except for CTL B54.7. and 23. for which it was not possible to raise sufficient cell numbers for multiple assays. Results were obtainedfor all target cells at 0.5: 1,1: 1, and 2: 1E:T ratios but. for simplicity. only those at an E:T ratio of 2:l are represented in this figure. Experimental values were highly reproducible and the relative killing efficiency of the various mutants by each of the CTL clones was kept at the three E:T ratios examined. In evaluating the effect of each mutation on the various CTL clones examined, relative killing efficiency values were grouped in four ranges: 130%. Figures to 70%. 70% within each of these ranges were considered equivalent to oneanother. CTL clones are grouped in panels A to F according tothe corresponding reaction patterns with the B27 subtypes, a s shown in Table 11. The percentage specific lysis of the B*2705+ transfectant target cells at this E:T ratio ranged from 40% to 100%. depending on the CTL clones. The relative LCL CH(HLAlysis of the 8.2703' A31.32;6=2703,18) is alsoincluded for comparison, B*2703is designated hereas H59 to emphasize the natureof its single substitution. For CH cells,therelative lysis is referred to the specific lysis of the B.2705' LCL LG15 (HLA-A32; B.2705) at a n E:T ratio of 1: 1. This ranged from13% to 86%.dependtng on the CTL clones.

PERCENTAGE

RELATIVE

LYSIS

STRUCTURE DIVERSITY AND

OF T CELL EPITOPES

complex can be recognized by multiple TCR (34, 35) and the predicted surface of the MHC-peptide complex (36, 37) is large enough to allow TCR binding in different registers. The reactivity of the CTL clones towards the mutants accounts,ingeneral, for theirreaction patterns with HLA-B27 subtypes. Lack of recognition of B*2704 and B*2706 by anti-B*2705 CTL clones, is related to the change at residue 152 (seebelow). Most CTL from groups A and B (Table 11) were affected by the presence of or Alasl (CTL 46). which are present inboth B*2701 and B*2702, or by an accumulation of various changes from these subtypes, a s in B54 (Fig. 3).Similarly, most clones in groupsD and E, that recognize B*2702but not B*2701, were affected by the changeto Tyr74rspecific of the latter, or by the double change of Tyr74plus CTL 72DRF, that recognized B*2701 and B*2702, was not significantly affected by any of the changes in thesetwo subtypes. There are a few CTL clones whose reactivity with the panel of mutants does not explainby itself their reaction patterns with subtypes. Thus, CTL 2 cross-reacts with B*2701 (Table 11). but it does not recognize any of the double mutants (Fig. 3).CTL 28, 172DRF and, somewhat less clearly,CTL 26 present a n opposite situation inthat they do not recognize Bs2701. a s expressed in LCL, whereas they lyse transfectant cells expressing double mutants that contain two of the three changes in that subtype, such a s Y74N77 and N77A81. This has two possible explanations. First, for the effectsobserved, be it recognition of B*2701 (CTL 2). or lack of it, (CTL 28, 172DRF. and 26). the simultaneous presence of the three changes at residues 74, 77, and 81 is required. Second, these clones depend. for recognition of B*2701, on the particular targetcell in which this molecule is expressed. Experiments are in progress to discern between these alternatives. Significance of individual changes. Every single position that was mutated had significant effects on CTL recognition (Table IV). Two levels of influence were clearly distinguished. First, the mutation at residue 152 totally abrogated recognition by all the anti-B*2705CTL clones examined. Second,all other mutations drastically affected 35%to 55%of the CTL clones (relative cytotoxicity, referred to that against the wild type: 70%) by any of the mutations (Table IV). Thus. theoverall effect of any of these mutations on the

4043

IN HLA-B27

particular set ofCTL clones analyzed was analogous to one another. This suggests that the antigenic profile of any of the HLA-B27 subtypes carrying multiple substitutions, is notdetermined by any single changebut, rather, by the setof changes in each subtype. The magnitude of the effects observed with the El52 mutant indicates that polymorphism at thisposition accounts, by itself, for the conspicuous immunologic disparity between B*2705 and B*2704 or B*2706, a s evaluated with T cells. However, the additional changes in these subtypes, such as that atposition 77, almost certainly contribute to determine their specific antigenic profiles (12). The relevance of residue 152 for T cell recognition has also been demonstrated in HLA-A3 (38, 39) and HLA-A2 (31, 32) although, for the latter, the effects on changing this position may be somewhat less overriding. It is likely that residue 152 is involved in critical intermolecular contacts, whose disruption abrogate TCR binding for most CTL clones. Whether these contacts are established directly with the TCR or with putative peptides interacting with theHLA-B27 (or other class I) molecule(s) remains to be determined. This idea is also supportedby recent results showing that a variety of changes at position 152 in HLA-A2, some of them much more conservative than the naturally occurring change of Val to Glu, all have drastic effects on CTL recognition (32).

Differential effects of substitutions at one and at spatially close positions. The two different mutations,

from Asp to Asn or Ser, at position 77 have different effects on recognition by some CTL clones. This is most clearly seen in GM7, CTL 2, 139DRD, and CTL 26 (Fig. 3). However, the global effects of these two changes on the whole set of clones analyzed is similar (Table IV). This result indicatesthat not only the location, but also the nature of the change, is relevant in modulating CTL recognition. An analogous conclusion has been recently obtained from a studyinvolving other polymorphic positions in HLA-A2 (32). For most CTL clones, the particular mutations introduced at spatially close positions induced drastically different effects on recognition. A s mentioned above, positions 59 and 63, 74 and 77, 77 and 80, and 77 and 81 are all located in a-helices and are spatially close. Although the exact orientation of their side chains is unknown, these residues are not in different helical faces, a s they are spaced approximately the spanof the a-helical turn, that relates residues i and i 3 or i 4. Yet. the changes introduced at positions 59 and 63 have clearly different effects on CTL 23,64DRF, 212DRD. 17A2, B54, TABLE IV 202DRD. and26.Similarly, the effects of a specific Effect of mutations on anti-B*2705CTL clones change at position 77 areclearly different from those at Relative Cytotoxicity" residue 74 and/or 80 in 212DRD. 17A2, CTL 40, G36, Mutation