Alloreactive Cytotoxic T Cells Recognize MHC Class I Antigen Without Peptide Specificity

Home , Cytotoxic T cell

0022-1767/91/1476-1765$02.00/0 THEJOURNAL OF IMMUNOLCGY Vol. 147. 1765-1772. No. 6. September 15. 1991 Copyright 0 1991 by The Amerlcan Association of ImmUnOlOgiStS Printed In U.S.A.

ALLOREACTIVE CYTOTOXIC T CELLS RECOGNIZE MHC CLASS I ANTIGEN WITHOUT PEPTIDE SPECIFICITY

ARNO MULLBACHER," ANN B. HILL,2* ROBERT V. BLANDEN,* WILLIAM B. COWDEN,* NICOLAS J. C. KING,' AND RON THA HLA* From 'Vlral Immunology, Division of Cell Biology, John CurtinSchool of Medical Research, AustralianNational University, Canberra, A.C.T. 2601, Australia and 'Department of Anatomy, University of Sydney, N.S. W.. Australia

In this report, experiments are described to differ- capable of forming biologically meaningful interactions entiate between three potential models of class I with theTCR. One, whichis in essence the sameas that MHC allorecognition, namely 1)recognition of pep- observed for self-restricted TCR/class I interaction. was tide-free MHC, 2) peptide-MHC-specific recognition, proposed initially by Matzinger and Bevan (2) and has and 3) peptide-MHC-nonspecific recognition. Using been extended by Kourilsky and Claverie (3).Thus, indi- a nucleoprotein peptide (NPP) with a sequence de- vidual alloreactive Tc cells recognize all0 MHC class I Downloaded from rived from influenza virus nucleoprotein with high molecules with a specific endogenous peptide. Evidence affinity for Kd class I MHC molecules, it is shown for the presentationof endogenous peptidesis numerous, that target cells rapidly become lysable by Kd-NPP i.e., Tc cell responses against HY, (4)oz-Microglobulin T self-restricted cytotoxic (Tc)cells, and retain suf- (5). and other minor Ag (6).Support for this model has ficient Kd-NPP complexes for at least 72 h. Kd-specific alloreactive Tc cells at the clonal and poly- come recently with the finding by Heath et al. (7)which clonal level do not show decreased lysis of Kd-bear- showed that murine alloreactive Tc cloneslysed human http://www.jimmunol.org/ ing targets in the continuous long term (48 h) cells transfected with the relevant murine class I MHC presence of NPP. Kd-stimulatorcells modified with Ag only in the presenceof murine peptides. NPP are able to induce potent Kd-NPP-specificself- In contrast, the difficulty in obtaining allorestricted, restricted Tc cells, however Kd-NPPstimulator cells virus-specific Tc cells (8).the failure to detect minor H- do not generate Kd-NPP specific alloreactive Tc cells specific alloreactive Tc cells in split clone-limiting dilu- from CBA and B1O.A (5R)mouse strains as tested tion assays using targets from congenic mouse strains by limiting dilution split clone experiments. Human (9).and thelack of tissue specificity in allograft rejection cells infected with the vaccinia virus recombinant (10) argue againstpeptide-specific allorecognition. coding for the murine Kd class I MHC Ag can be Evidence that different interactions are in operation in by guest on July 13, 2017 lysed by murine Kd-specificalloreactive Tc cells. In self-restricted recognition vs allorecognition is indicated addition the rateof reemergenceof alloreactive and by studies on selected mutantsof class I Ag involving the self-restricted Tc cell epitopes on virally infected @-sheet and a-helices that affect allorecognition differ- target cells that had their cell-surface class I MHC entlyfrom self-restricted recognition (11) and experi- Ag removed is identical. These results are consist- ments showing that a lymphoma mutant, unable to ex- ent with model 3 namely that the majority of Tc precursor and effector cells recognize class I MHC press normallevels of class I MHC Ag because of a defect Ag without peptide specificity. in endogenous peptide production, can be recognized by alloreactive T cells but is not recognized by minor H Ag- specific, self MHC-restricted T cells(1 2). Recognition of target cells by CD8+ T lymphocytes has An alternative is that the epitope recognized is deter- been well characterized inself MHC restricted responses. mined by the conformation of the polymorphic part of The epitoperecognized is a complex of a self MHC class I the MHC class I molecule per se, irrespective of the pep- molecule plus a foreign peptide (1).Although the phenom- tide bound. A modified version of this model would be enon of recognition by T cells of foreign MHC (allorecog- that allorecognition is seeing peptide-free class I MHC, nition) has been investigated for many years, our under-estimated to be in the order of 0.3% of cell-surface ex- standing of what is precisely recognized in such interac- pressed classI molecules (13). Evidence that peptide-free tions is much more limited. However, a priori, there exist class I molecules can be recognized as allogeneic mole- two basic options regarding the formof epitopes that are cules hasbeen shown recentlyby Elliot and Eisen (14). Recognition of all0 MHC irrespective of peptide would Received for publication April 11, 1991. be consistent with the model for restrictive recognition Accepted for publication June 18. 199 1. vs allorecognition as proposed by Langman (15). The The costs of publication of this article were defrayed in part by the payment Of page charges. This article must thereforebe hereby marked results presented here suggest that the majority of allo- advertisement in accordance with 18U.S.C. Section 1734 solely to indi- reactive Tc clones recognize all0 MHC class I per se, but cate this fact. ' Address correspondence and reprint requestsDr. to Arno Mdllbacher, do not exclude the possibility that a small minority of Viral Immunology. Division of Cell Biology. John Curtain School of Medi- such clones are specific for endogenous peptides. cal Research, Australian National University, Canberra. A. c. T. 2601. Australia. 3Abbreviations used inthis paper: Tc cells, cytotoxic T cells: VV. Present address: ICRF, University of Oxford, Institute of Molecular vaccinia virus: NPP. nucleoprotein peptide: TGM, thioglycollate-induced Medicine. John Radciiff Hospital, Oxford, England. B/c peritoneal macrophages. 1765 1766 ALLOREACTIVE CYTOTOXIC T CELL RECOGNITION

MATERIALS AND METHODS 10000 7, Animals. Mouse strains CBA/H (KkDk)(CBA), BIO.A (5R), (KbDd) (5R). C3H.H-2" (KdDk)(OH], and BALB/c (KdDd)(B/c) were obtained from the breeding establishment of the John Curtin School, Can- berra, Australia. Only females older than 6 wk were used. Viruses. A strain influenza viruses A/WSN (HlN1) and A/JAP (H2N2) were grown and titrated as has been described elsewhere (16). The recombinant VV coding for the MHC class I Kd molecule (VV-Kd)and the TK- control VV or wild-type VV WR (VV-WR)were grown and titrated asdescribed (1 7). NPP. The R- peptide of influenza A virus 147-158 [ 18) was a generous gift of Peptide Technology, Sydney, Australia. Iodination was performed by the lactoperoxidase method, essentially as fol- lows- 10 rg of peptide was labeled using 0.5 mCi of 1251 in PBS (pH 7.2) by the addition of lactoperoxidase (2 pg) and hydrogen peroxide (30 ng) for 2 min. Three further additionsof hydrogen peroxide were made at 2-min intervals and the reaction was terminated by the addition of BSA (3%w/v in PBS,pH 7.2, and 0.1% (w/v) sodium azide). Free iodine was separated from the peptide by Sephadex G- 25 chromatography. Immunization. Animals were immunized with either lo7plaque- forming units of VV-WR or lo3 hemagglutinating units of A/strain influenza virusA/WSN. 100 Generation of effector cells. Spleen cells frommice immunized 6 I ,1 1 10 100 days previously were used for primary VV-immune Tc cells.Cultures Downloaded from for the generation of secondary influenza-immune Tc cells have been described in detail elsewhere (16). For the generation of allo- TIME (h) reactive Tccells. 8 X 1O7 responder splenocytes were cocultured with 4 X lo7 irradiated (2000 r) allogeneic stimulator cells for 5 days in Figure 1. Kinetics of uptake of labeled NPP. P815 cells [squares) medium with or without NPP as indicated in the text. or B/c TGM [circles)were either left untreated [closed symbols) or treated Generation ofeffector cellsin microcultures. Alloreactive Tc cells with mitomycin C [open symbols) and incubated wlth IOw4 M 1251-labeled were generated in cocultures of 5 x lo3 red cell-depleted responder peptide. Cell-associated radioactivity was measured over a 48-h period. splenocytes with 2 X lo5 irradiated (2000r) red cell-depleted stimu- Line without symbols gives background radioactivity in the absence of http://www.jimmunol.org/ lator splenocytes. The culture medium (F15. GIBCO, Grand Island, cells. NY) was supplemented with 5% FCS plus M 2-ME (0.2 ml/well) and was conditioned with 10% Con A spleen supernatant (161. In individual cultures, NPP waspresent at M concentration. After 5 days in culture, plates were spun at 1500rpm for 3 min. culture supernatant was discarded, and cells were resuspended in 0.2 ml 1 fresh medium and split into two 0.1-ml aliquots. Target cells.TGM, P815 cells [H-zd) L929cells (H-2k) methylchol- anthrene-induced fibrosarcoma cell lines MC57 (H-zb], HTG (H- 2KdDb).and OH (H-2KdDk)and the human cell line 143-8 were

infected with vaccinia or influenza virus labeled with "Cr as de- by guest on July 13, 2017 scribed in detail elsewhere (1 6, 17, 19).Target cells were incubated before or after infection for different times with NPP as detailed in results. E Papaln treatment of target cells.The method has been described cn (20). 5 51Cr-release cytotoxicity assay. The methods used for cell line 8 and macrophage targets have been describedin detail elsewhere (1 6, 19). The durationof the assayswere 6 h. The percent specific lysis was calculated using the formula: % specific lysis = [(experimental release - medium release)/(maximum release - medium release)) X 100. Data given for bulk cultures are the meansof triplicates. SEM were always <5%and are omittedfor clarity. Significance was determined by Student's t-test.

RESULTS Kinetics uptake and retention peptide. To assess -...... ,,. of of 1 10 100 the role of peptides associated with classI MHC molecules in recognition by alloreactive Tc cells, NPP was used to saturate available Kd on targetcells. The kinetics of K:T uptake of I'25-labeled,Kd-restricted, synthetic peptide Figure 2. Peptide-modified target cell lysis by influenza-immune Tc (NPP) with a sequence derived from the nuclear protein cells. Lysis by secondary in vitro A/WSN-immune. Kd-restrictedTc cells of PE15 cells modified with M NPP for 1 h and left for 72 h ID), 48 h of influenza virus, by dividing and nondividing-Kd bear- (A]. 24 h (0).or 1 h (0)before assay, or unmodified (0).Assay duration ing target cellsis shown in Figure 1. Plateau levels were was 6 h. reached at 8 h in both dividing or mitomycin C-treated P815 cells or nondividing B/c TGM, thus indicating the suggests that under the conditions employed here, NPP NPP-saturated available Kd (free of endogenous peptide) remains associated with Kd insufficient quantity for by 8 h. To investigate the retentionof NPP on Kd, nondi- optimal Tc cell recognition for at least 72 h. viding B/c TGM wereincubated with M NPP for 1 h, Continuous presence of NPP does not alter allorecog- washed, and left for up to 72 h before being tested for nition. The previous experiments show that Kd can be susceptibility to lysisby Kd-restricted, A/WSN influenza- modified by NPP and that thismodification is stable for immune Tc cells(Fig. 2).No differences could be observed long periodsof time. We therefore investigated the effect in target cell lysis for up to 72 h after labeling. This of continuouspresence ofNPP at M concentration ALL OREAC TIVE CYTOTOXIC T CELL ALLOREACTIVECYTOTOXICT RECOGNITION 1767

loo /A I’ r/”2 100 80 Figure 3. Peptide-modified P815target cell lysis by Tccells. Lysis of targetsin- v, 60 60 fected with A/WSN (0)left uninfected (W) or v) culturedincontinuous presence of M 5 NPP over 72 h (A). 48 h (A),24 h (0).1 h (0) 8 40 by secondary influenza-immune OH effector cells (A1 or 5R anti-B/c alloreactive Tc cells (B).Aisay duration was6 h.

40120 01 I 4 o! I 1 10 100 1 10 100

K:T K:T for up to 72 h before the addition of effector cells, on splenicresponder cells cocultured with irradiated B/c target cell lysis by anti-Kd alloreactive and Kd-restricted, splenic stimulator cells in microtiterwells. After 5 days’ influenza-immune Tc cells(Fig. 3A). Kd-restricted, influ- culture the individual cultures were split and tested on enza virus-immune Tc cellslysed P8 15 targetsmodified normal P8 15 and P8 15 incubated for h48 with 1 0-4 M Downloaded from with NPP equally whether the NPP was present for 1 or NPP.NPP was replenished after 24 h and 1 h before 72 h.A/WSN influenza virus-infected targets were lysed assay.The results of two independentrepresentative to a greater extent in this experiment, possiblyas a result experiments are shown in Figure 5. In no case did we of contributions of other viral peptidesrecognized by this observe clones that significantlylysed unmodified P8 15 polyclonal Tc cell population. Kd-specific alloreactive Tc more than NPP-modified P815. In addition,the same cells from 5R (KbDd)anti-B/c (KdDd)lysed the panel of observation was made when Kd-specific alloreactive Tc http://www.jimmunol.org/ targets approximately equally (Fig. 3B).Similar observa- cellswere generated by differenthaplotype combinations were obtained with the nondividingB/c TGM targets tions, i.e.,CBA (KkDk) anti-OH(KdDk) (data not shown). (Fig. 4). Identical results were obtained with Kd-specific Kd modified with NPP can stimulate syngeneic but alloreactive Tc cells generated with different haplotype not allogeneic NPP-specgic responses. The possibility combinations, i.e., CBA (KkDk) anti-OH (KdDk) (data not that Kd-allospecific Tc cells are cross-reactive (with re- shown). spect to Kd-bound peptides) at the target cell level, but Limiting dilution split clone analyses on NPP satu- ration modvied targets. To test thepossibility that long recognize Kd-peptide specifically at induction was tested in split clone experiments. We tested if B/c-irradiated incubation of target cells at high NPP concentration may by guest on July 13, 2017 allow NPP to displace endogenous peptides (bound toKd) splenocytes can present K~-NPPspecifically in a synge- to such an extentas to lose threshold avidity for certain neicsystem. B/c splenocytes werepulsed with M low affinity Kd-reactive clones specific for endogenous NPP for 24 h, infected with influenza virus A/WSN. or peptides, limiting dilution split clone experiments were left uninfected. Splenocytes of A/JAP-primed B/c mice undertaken.Effector cells were generated from 5R were added for 5 days as a source of NPP-specific, Kd- restricted memory Tc cells. The effector cellswere tested forlysis on uninfected, NPP-modified and A/WSN-in- fectedP815 targets (Fig. 6). NPP-modified stimulators induced NPP-specific Kd-restricted effector Tc cells with potency equal to virus-infected stimulators.We have also shown that NPP pulsed stimulator cells are specifically recognized by syngeneic cells in vivo.4Thus Kc‘-NPPcan be seen by resting memory and naive, syngeneic, splenic T cells in a specific manner. Using the same stimulator cell conditions we tested for generationof Kd-NPP-specific alloreactive clonesin a limiting dilution split clone experiment. A total of 2000 5R anti-B/c-NPP clones (determined from a precursor frequency analysis) were tested on 48-h NPP-modified P815 vs unmodified targets (Fig. 7). No P815-NPP (Kd-NPP)specific clones were found. The same experiment was performed usinga different strain combination (CBA anti-OH) with the same results being obtained (data not shown). The failure to detectKd-NPP- specific clones in two separate strain combinationscon- 1 IO 100 taining differentMHC and background genes reduces the possibility of cross-tolerance or “immunodominance” ef- K:T fects as the sole explanation of the results. Figure 4. Peptide-modified B/c TGM target cell lysis by Tc cells. Leg- end as for Figure3. 4M~llbacher,A., and R. Tha Hla. Manuscript in preparation. 1768 ALLOREACTIVECYTOTOXIC T CELL RECOGNITION

loo 7 80

Figure 5. Limiting dilution split clone analysis. Kd-specific alloreactive effector 60 cells were generated in microculture wells e ,,, 60 0 (96 wells/panel) by coculture of 5 x IO3 5R 5 e responders with 2 x lo5 B/c-irradiated P 8ol stimulator cells. Contents of wells were split and tested for lysis of 1 x IO' P815 and 1 x lo' P815 cultured for 48 h in the continuous presence of M NPP.

o! ...,..,..,..,.. 0 20 40 80 60 1 0

P815-NPP P015-NPP

100 sis to investigate Kd expression of these two VV-Kd-in- 1 fected target cells, a two- to threefold lower expression was found in 143-B target cells than in MC57 (data not

shown). These data indicate that mouse-specific endog- Downloaded from enous peptides were not essential for recognition by Kd- specific murinealloreactive Tc, butdo not exclude a role for endogenous peptides shared by mouse and human. VV-Kd recombinant virus can induce Kd-alloreactive Tc cell responses of equal magnitude to Kd allogeneic

splenocytes. Vaccinia is a cytopathic virus which inhib- http://www.jimmunol.org/ its host cell protein synthesis soon after infection. Kd encoded by vaccinia should thereforebe exposed predom- 40 inantly to vaccinia-encoded peptides rather than endogenous mouse peptides. To examine the effects of this scenario on Kd-specificalloreactive Tc induction, 5R re- 20 sponder splenocytes were stimulated in vitro with either B/c or VV-Kd-infected 5Rsplenocytes. Effector cells were tested for lytic activity on HTG (Kd)targets or L929 (H-2")

and MC57 (H-2b)target cells infected with either VV-Kd, by guest on July 13, 2017 control VV, or left uninfected (Table 11). 5R anti-B/c ef- u NPP AlWSN fector cells lysed MC57-and L929 VV-Kd-infectedtargets specifically, although, as shown above, cross-reactivity STIMULATORS on VV-infected and uninfected targets was alsoobserved. FLgure 6. Influenza-specific target cell lysis by in vitro NPP-boosted Lysis of HTG targets washigher than L929-VV-Kdwhich effector cells. Lysis of P815 cells infected with A/WSN influenza (a), in turn was higher than MC57-VV-Kd-infectedtargets, modified with lo-' M NPP (W). or unmodified (0) by B/c A/WSN-immune reflecting the Kd Ag concentration as determined by FACS Tc cells generated by boosting influenza-primed memory cells in vitro with B/c stimulator splenocytes either left uninfected. modified with loe4 analysis (data not shown). 5R effectors generated in re- M NPP, orinfected with A/WSN. Values taken are froma fourfold titration sponse to VV-Kd-infected5R cells gave results similar to curve at an E:T ratio of 10:1. Duration of assay was6 h. those stimulated with B/c, thus indicating no effect at- tributable to vaccinia infectionon availability of endog- Kd-expressed ina human cell line can be recognized enous mouse cell peptides. by murine Kd-alloreactive Tc. To further test if endogenous mouse peptide-specific recognition occurs in murine Limiting dilution split clone experiments on murine T cell alloreactivity, we infected murine and human tar- and human target cells infected with the VV-Kdrecom- get cells with a VV recombinant coding for Kd class I MHC binant virus. The question regarding murine peptide- (VV-Kd).The humantarget cells could not provide mouse- specific allorecognition was also addressed using a lim- specific endogenous peptides. Kd-specificalloreactive Tc iting dilution split clone approach. 5R anti-B/c cultures cells (5R anti-B/c) lysed Kd-bearing HTG and OH targets were set up underlimiting dilution conditions, and effec- very efficiently (Table I). Significant cross-reactivity on tor clones were split and tested on MC57 VV-Kd vs 143- control MC57 targets, uninfected or vaccinia-infected, B W-Kdtargets (Fig. 8). was observed. However, threefold higher lysis was ob- Murine targets infected with VV-Kd were more suscep- tained on targets infected with the Kd-coding recombi- tible to lysis by anti-Kd alloreactive Tc cell clones than nant virus. Importantly, however, Kd-specific lysis oc- the human 143-B VV-Kd-infected target cells. However, curred with the humancell line 143-B infected with VV- a large number of clones significantly lysed both targets Kd. Kd-restricted VV-immune Tc cells (OH anti-VV) lysed suggesting no role for murine-specific peptides. Quanti- MC57 VV-Kd-infected targets threefold more efficiently tative analysis of Kd expression on the target cells re- than 143-B VV-Kd-targets,suggesting that 143-B VV-Kd vealed that MC57 targets expressed Kd at higher concen- expresses less Kd than MC57-VV-Kd.Using FACS analy- trations than 143-B (data not shown), thus indicating a ALLOREACTIVE CYTOTOXIC T CELL RECOGNITION 1769

20 Figure 7. Limiting dilutionsplit clone 4:L analysis of NPP-Kd-stimulated alloreactive Tc cells. Kd-NPP-speclfic alloreactlve effec- 60 d 40 20 SO 100 torcells were generated In microculture wells by coculture of 5 X IO3 5R responder cells with2 x 1 O5 B/c stimulator cells in the presence of M NPP. Contents of wells were split and tested for lysis on PB 15 and PB15-48h NPP ( M) modified targets. Eachpanel contains the results of 192 80 wells. Duration of assay was6 h. I"" i 0- "r.. Downloaded from 4o/f;, 4o/f;, , , , , , . , 9 20

0 0 40 20 60 80 1 % LYSIS Pa15 http://www.jimmunol.org/ TABLE I VV-Kd-infected murine andhuman target cell lysis by murine alloreactive anti-Kd Tccells and VV-spectffcKd- restricted Tc cells % Specific 51CrRelease

Effectors E:T HTC OH MC57 (H-zb) 143-8(human] (KdDb) (KdDh) U U U VV VV-Kd U VV VV-Kd 5R anti-B/cb(anti-Kd) 87 89 30 52 63 82 11 6 36 1081 69 24 26 662 5 17

3 39 955 9 34 2 1 6 by guest on July 13, 2017 OH anti-VV' (Kd-restricted) 90 11 40 12 20 8314 19 61 30 9 29 8 14 7512 12 35 10 5 815 4 46 6 5 16 "Mean percent "Cr release from target cells over a 6-h period. Spontaneous release ranged from 10 to 13%.Means of triplicates are given: SEof the mean were never>3%. In vitro generatedKd-specific alloreactive Tc cells. Primary in vivo VV-immune Tc cells.

TABLE I1 Induction and recognitionofallo-Kd expressed by VV-Kd recombinant '% Speciflc 51Cr Release' Effectors E:T HTG MC57 (H-2b) L929 (H-2') (KdDb) U VV VV-Kd U VV VV-Kd U 5R anti-B/cb(anti-Kd) 30 41 37 73 40 47 72 87 IO 17 14 34 19 22 60 79 5 2 3 5 15 45 6 7 49

5R anti-VV-Kd' 30 52 50 68 48 55 68 83 10 28 26 42 29 37 61 68 3 8 7 20 IO 14 53 44 As for Table1. Spontaneous release ranged from9 to 17%. As for Table I. Primary in vitro cultureof 5R splenocytes stimulated with 5R splenocytes infected withVV-Kd quantitative contribution of Kd to differences in lysis of absence of NPP lyse target cells infected withVV-Kd more murine and humantarget cells. efficiently if such cells are pretreated withNPP. Table I11 NPP-enhances VV-expressed Kdas recognized by al- showstwo such experiments. Kd-alloreactiveTc cells loreactive Tc cells. Recent results suggest that exoge- lysedVV-Kd-infected, NPP-treated murine and human nous peptides that are able to bind to MHC class I Ag targets twofold more efficiently than untreated VV-Kd- facilitate thecell surface expression and or the stabilityinfected targets. This result implies thatNPP complexed of these molecules (21).Therefore we tested the proposi- with Kd formed an appropriate targetcell epitope for Kd- tion that Kd-specific alloreactive Tc cells generated the in specific alloreactive Tccell recognition, even thoughNPP 1770 ALLOREACTIVE CYTOTOXIC T CELL RECOGNITION

40 100 1

00 D 0 ? 1 j O. > 20

0 I 0 10 20 30 Downloaded from 2 3 4 5 6

143-8 VV-Kd ASSAY TIME (h) Figure 8. Limiting dilution split clone analysis of murine and human Figure 9. Kinetics of reemergence of T cell epitopes. Lysis of influenza Kd-expressing target cells. 5R anti-B/c alloreactive Tc cells generated in virus-infected targets treated with papain (open symbols) or left un- microculture wells were tested by splitting contents of wells for lysis of treated (closed symbols] 3, and4. 5 h before alloreactive (CBA anti-B/c)

MC57 and 143-Btarget cells infected with VV-Kd. Panel represents 284 (squares]or influenza-immune (circles)effector were added. All values http://www.jimmunol.org/ wells tested. Assayduration 6 h. are from a fourfold titration curve.

TABLE 111 lysed untreated control targetseffectively at anE:T ratio Lysis of murine L929-and human 143-8 target cells infected with VV-Kdin the presence or absence of NPP by Kd-specficalloreactlue of 3:l giving 65%lysis at 3 h to 84% lysis at 5 h. Papain- Tc cells treatment of cells reduced the lysis to 11 % at 3 h and '36 Specific 51CrRelease from L929 30% at 5 h. Influenzavirus-immune Tc cells gave a Expt. 1 Effectors E:T Targetsa similar picture, althoughoverall lower lysis was obtained. vv VV-K~ VVINPP VV-K~INPP Most importantly,however, the slopes of the lines, a

CBA anti-B/cb 4330 67 53 80 measure of the rateof killing, was the same for both the by guest on July 13, 2017 10 28 48 36 57 Kd-alloreactiveTc cells and the influenza-immune Tc 3 2 26 2 3 18 32 Bjcanti-A/WSN'27 26 23 30 72 cells. 10 14 10 17 17 50 35 7 7 25 DISCUSSION % Specific "Cr Release from 143-8 Expt. 2 Targetsa The experiments described in this paperan are attempt VV VV-Kd VV-Kd/NPPUVV-Kd VV to differentiate between three possible models of alloreactive Tc cell recognition of class I MHC. Namely, a) CBA anti-OHb 30 10 11 46 57 10 224 4 38 onlypeptide-free MHC is recognized, b) recognition is 32 0 9 14 specific for an epitope formed by a complex of an all0 "As for Table 1. Spontaneous release ranged from 12 to 18%. class I MHC molecule plus an endogenous peptide, anal- As for Table 1. 'Secondary in vitro culture of B/c A/WSN-immune memory spieno- ogous to MHC-restricted responses, or c)allogeneic class cytes stimulated with A/WSN-infected B/c splenocytes. I MHC molecules complexed with endogenous peptides are recognized by Tc cells regardlessof which particular was not present during the induction of the response, peptide is present. thus suggesting a lack of specificity for peptide in allo- We could test the firstpossibility by using a synthetic recognition. peptide (NPP) resembling the sequenceof the nucleopro- Kinetics of emergence of Kd-influenza and alloreac- tein of influenza virus, which hasa very high affinity for tiue Kd epitopes on papatn-treated target cells.Because Kd and which forms with Kd a dominant epitope recog- influenza virus is cytopathic and inhibitscell host protein nized by Kd-restricted influenza-immune Tc cells (18). synthesis soon after infection, influenza-specified pep- The initial experiments showed that target cells express- tides should be much more prevalent than endogenous ing Kd reached plateau levels of NPP binding by 8 h and mouse cell peptides in Kd-peptidecomplexes produced retained NPP for at least 72 h. Thus the continuous subsequent to infection. Therefore, pre-existing Kd-pep- presence of NPP over a 48-h period should saturate all tidecomplexes on P815 target cells were removed by available Kd (free of endogenous peptides) estimated to be papain treatment1 h after infection with influenza virus less than 0.3%of all cell-surface Kd (13).However, poly- and the treated cells were used as target cells for Kd- clonal Kd-alloreactive Tc cell populations did not show alloreactive and Kd-restricted influenza-immune Tc cells decreasedlysis of peptide-modified targetscompared (Fig. 9). Lysis of target cells was assayed 3, 4, and 5 h with unmodified controls.Clonal analysis using the split- after the additionof effector cells. Kd-alloreactive Tc cellsclone limiting dilution technique also did not reveal a ALLOREACTIVECYTOTOXIC RECOGNITIONT CELL 1771 single Kd-reactive clone that was unable to lyse NPP- VV-specific Kd-restricted recognition. This again points modified Kd targets, although it is known that individualto the possibility of Kd alloreactive Tc cells without pep- alloreactive Tccell clones are heterogeneous with respect tide specificity. to their ability to lyse target cells withlow levels of class Second, the results presented in Table111 are in agree- I MHC cell-surfaceexpression (22). Thus these data ment with recent reports (21) indicatingexogenously that strongly suggest that peptide-free class I molecules are appliedpeptides facilitate MHC class I cell-surface not the principleligand for alloreactive Tc cells. expression and stability. L929 and human 143-B cells We then askedif the Kd-NPP epitope can berecognized infected with VV-Kd and pretreated withNPP were more during the inductionof an alloreactive Tc cell response. susceptible to lysis by CBA anti-Kd-alloreactive Tc cells The immunogenicity of Kd-NPP was clearly established, than VV-Kd-infected targets. Under these conditions it is as self Kd-restricted influenza-primed splenocytes could reasonable to propose that the increased lysis was caused be stimulated in vitro with NPP-armed splenocytes to by increased expression of Kd complexed with NPP, a T differentiate intoKd-NPP-specific Tc cells. Employing the cell epitope not present during induction of the Tc cell same conditions with NPP-armed splenocytes we tested response. in limiting dilution a total of 2000 alloreactive 5R (Kb,Dd) Finally, influenza virus-infected target cells that had anti-Kd Tc cell precursors. The results were unambigu- their class I MHC cell-surface molecules removed by pa- Kd-NPP specific clones could be detected. The ous: no pain cleavage displayed an identical rate of reemergence same experiment was performed usingCBA splenocytes of all0 and self-restricted T cell epitopes. Because influ- as responders, to test thepossibility that thymicpositive enzainfection inhibits host cell proteinsynthesis, it selection or cross-tolerance toMHC or background Ag in Downloaded from 5R skewed the T cell repertoire in such a way that Kd- would be expected that class I MHC reexpressed on pa- NPP recognition is reduced. Again it was found that Kd- pain-treated influenza-infected cellswould be complexed NPP specific alloreactive Tc cells were not induced imply-predominantly with influenza-specified peptides. Thus ing thatrecognition of Kd by alloreactive Tc cellswas not these results also point to thepossibility that MHC class specific for peptide. I-influenza peptide complexesare recognized by alloreactive peptide-nonspecific Tc cells.

Three further resultsprovide additional evidencecom- http://www.jimmunol.org/ patible with peptide-nonspecific allorecognition. First,Kd There are two possible explanations as to why allore- expressed in human 143B cells by infection with recom- cognition differsin its molecular model fromself-re- binant VV encoding Kd (VV-Kd) can render these cells stricted recognition. First, it isnow established that self susceptible to lysis by murine alloreactive Tc cells from MHC must be boundby TCR during thymic development both polyclonal and limitingdilution monoclonal cul- of T cells (positive selection), but the specificity of this tures. although the level of lysis was about threefold eventwith respect to peptides is unknown (26). Self lower than on VV-Kd-infected MC57 murine target cells. tolerance mechanisms then delete or control T cells re- This difference could be explained by a combination of active to normal self MHC epitopes (negative selection) the observed cross-reactivity of 5R anti-B/c (anti-Kd)ef- (26). but permit the survivalof T cells with affinity for by guest on July 13, 2017 fectors on MC57 [KbDb)targets (TablesI and II), a two- to all0 MHC. There is no a priori reason to believe that threefold concentration differenceof cell-surface Kd mol- peptide specificity is essential with respect to all0 MHC ecules evident by FACS analysis and possibly reduced recognition. Furthermore,two recent studies indicate efficiency of lymphocyte function-associatedAg-intercel- that it is possible for alloreactive Tc cells to recognize lular adhesion molecule interactions across the species peptide-freeallo class I MHC molecules (12,14). The barrier (23). The same targets showeda threefold differ- present results suggest that peptide-free MHC is not es- ence in lysis by VV-immune Kd-restricted effectors that sential for allorecognition and that peptide specificity is recognize vaccinia-derived peptides. These results sug- oftenirrelevant to all0 MHC binding by the TCR. In gested that rather than alack of endogenous mouse- contrast, because of self tolerance, self MHC-restricted specificpeptides, the concentration of cell-surface Kd Tc can bindto self class I moleculeswith sufficient expression wasa limiting factor (24, 25) in theavidity of affinity to trigger activation, only when a foreign peptide Tc cell binding required for effector function of Tc cells against VV-Kd-infected 143B target cells. Such quanti- occupies the cleft in theMHC molecule. tative differencescould also have contributed to the find- Second, the results are also inagreement with the ings by Heath et al. (7)which were interpreted qualita- mechanism proposed by Langman andCohn for alloreac- tively as evidence for peptide-specific recognition by al- tivity by the one receptor dual recognition model of TCR loreactive Tc cells.In this latter case, additionof murine [ 15).In this model, one chainof the TCR-aP is selected in peptides to target cells may have increased the quantity the thymus forrestricted recognition of self MHC (e.g., 01- of murine class I MHC expressed by human target cells chain) and variants of theother chain (e.g.,P-chain) transfected with a murine classI gene (21) (see Table111 randomly associate with the first chain togive a family discussed furtherbelow). of heterodimers,which constitute the self-MHC-re- Using the same VV-Kd recombinant virus to infect5R stricted repertoire. The rules of self tolerance apply as (KbDd)stimulator cells we were able to induce potent outlined above. This model permits all0 MHC recognition alloreactive Kd-specific 5R Tc cellsthat lysed uninfected by the random repertoire of the second chain (e.g., p) HTG (Kd,Db)targets. Because VV infection inhibits host regardless of peptide specificity. cell protein synthesis, whichwould reduce availability of Acknowledgments. We would like to thank Drs Klaus endogenousmurine peptides, the VV-encoded Kd pro- Eichmann. Jean Langhorne and MarkusSimon for crit- duced in infected 5R stimulatorcells may have been ically reading the manuscript and Gabriele Prosch and predominantly exposed to VV peptides as indicated by Michelle Caulfield for excellent secretarial assistance. 1772 ALLOREACTIVECYTOTOXIC T CELL RECOGNITION

REFERENCES 14. Elliot, T. J., and H. N. Eisen. 1990. Cytotoxic T lymphocytes recognize a reconstituted class I histocompatibility antigen (HLA-A2) as 1. Yewdell, J. W.. and J. R. Bennink. 1990. The binarylogic of antigen an allogeneic target molecule. Proc. Natl. Acad. Scl. USA 98:5213. processing and presentation to T cells. Cell62203. 15. Langman. R. E. 1989. The Immune System, Academic Press, San 2. Matzinger, P., and M. J. Bevan. 1977.Why do so many lymphocytes Diego,pp. 111-118. respond to major histocompatibility antigens?Cell. Irnrnunol. 29: 1. 16. Mullbacher. A. 1984. Hyperthermia and the generation and activity 3. Kourilsky, P.. and J. M. Claverie. 1989. MHC restriction, alloreac- of murineinfluenza-immune cytotoxic T cells in uftro. J. Vfrol. tivity. and thymic education: a common link?Cell 56:327. 52:928. 4. Simpson, E., and R. D. Gordon. 1977. Responsivenessto H-Y antigen 17. Coupar, B. E. H., M. E. Andrew, D. B. Boyle, and R. V. Blanden. Ir gene complementation and target cell specificity. Irnrnunol. Reu. 1986. Immune responses to H-2 Kd antigen expressed by recombi- 25:59. nant vaccinia virus. Proc.Natl. Acad. Scl. USA 83~7879. 5. Rammensee. H. G.,P. J. Robinson, A. Crisanti. and M. J. Bevan. 18. Bodmer. H., R. M. Pemberton. J. B. Rothbard, andB. A. Askonas. 1986. Restricted recognition of &-microglobulin by cytotoxic T lym- 1988. Enhanced recognition of a modified peptide antigen by cyto- phocytes. Nature 319:502. toxic T cells specific for influenza nucleoproteins.Cell 52253. 6. Bevan, M. J. 1975. Themajor histocompatibility complex determines 19.Mtillbacher, A.,R. V. Blanden, and M. Brenan.1983. Neonatal susceptibility to cytotoxic T cells directed against minor histocom- tolerance to MHC antigens alters Ir gene control of the cytotoxic T patibility antigens. J. Exp. Med. 142: 1349. cell response to vaccinia virus.J. Exp. Med. J57:1324. 7. Heath, W. R., M. E. Hurd, F. R. Carbone, andL. A. Sherman. 1989. 20. Thomas, K.. H. D. Engers, J. C. Cerottini, and K. T. Brunner. 1976. Peptide-dependent recognition of H-2Kb by alloreactive cytotoxic T Enzymatic removal of H-2 alloantigens from the surfaceof P8 15-(x2) lymphocytes. Nature 341:749. mouse tumor cells,,Eur.J. Irnrnunol. 6:257. 8. Bennink. J. R., and P. C. Doherty. 1978. T cell population specifi- 21. Townsend, A., C. Ohlen, J. Bastin, H. G. Ljunggren, L. Foster, and cally depleted of alloactivity potential cannot be induced to lyse H-2 K. Karre. 1989.Association of class I major histocompatibility heavy different virus-infected targetcells. J. Exp. Med. 148: 128. and light chains induced by viral peptides. Nature 340:443. 9.Langhorne. J.. and K. FischerLindahl. 1982. Role of non-H-2 22. Shimonkevitz, R.. B. Luescher, J. C. Cerottini, and H.R. Mac- antigens in the cytotoxic T cell response to allogeneic H-2. Eur. J. Donald. 1985. Clonal analysis of cytolytic T lymphocyte-mediated lysis of target cells with induceable antigen expression: correlation Irnrnunol. 12:lOl. between antigen density and requirement for Lyt-2/3 function. J.

10. Lafferty, K. J., S. J. Prowse, C. Simeonovic. and H. S. Warren. Downloaded from Irnrnunol. 135:892. 1983. Immunology of tissue transplantation: a return to the passen- 23. Johnston, S. C., M. L. Deistin. M. L. Hibbs. and T. A. Springer. ger leukocyte concept. Annu. Rev. Irnrnunol. 1:143. 1990. On the species specificity of the interaction of LFA-1 with 11. Mattson. D. H.. N. Shimojo, E. P. Cowan. J. J. Baskin, R. V. Turner, intracellular adhesionmolecules. J. Irnrnunol. 145: 1 181. B. D. Shvetsky. J. E. Coligan, W. L. Maloy, and W. E. Biddison. 24.Blanden, R. V.. P. D. Hodgkin, A. Hill, V. G. Sinickas. and A. 1989. Differential effectsof amino acid substitutions in the @-sheet Mullbacher. 1987. Quantitative considerations of T cell activation floor and 01-2helix of HLA-A2 on recognition by alloreactive andviral and self tolerance. Irnrnunol. Rev. 98~75. peptide-specific cytotoxic T lymphocytes.J. Imrnunol. 143: J 101. 25. MacDonald, H. R.. A. L. Glasebrook. C. Bron, A. Kelso, and J. C. 12. Ohlen, C., J. Bastin. H. Ljunggren, L. Foster, E. Wolpert. Cerottini. 1982. Clonal heterogeneity in the functional requirement G. G. http://www.jimmunol.org/ Klein, A. R. M. Townsend. and K. Karre. 1990. Resistance toH-2- forLyt-2/3 molecules on cytolytic Tlymphocytes (CTL): possible restricted but not to allo-H-2-specific graft and cytotoxic T lympho- implications forthe affinityof CTL antigen receptors.Irnrnunol. Reu. cyte responses in lymphoma mutant.J. Irnrnunol. J45:52. 68:89. 13.Chen, B. P., and P. Parham.1989. Direct binding of influenza 26. Marrack, P., and J. Kappler. 1988. TheT-cell repertoire for antigen peptides to class I HLA molecules. Nature 337:743. and MHC. Irnrnunol. Today 9:308. by guest on July 13, 2017