Antigen-specific Deletion of Cloned T Cells Using Peptide-Toxin Conjugate Complexed with Purified Class I 1 Major-Histocompatibility Complex Antigen ”

In a previous report, we showed that cloned T cella incubated with soluble, cognate major histocompatibility complex (MHC) II-peptide complex internalized the peptide moiety of the complex. Here, we report antigenspecific deletion of cloned T cells by treatment with soluble,  cognate MHC II-(peptide-toxin)  complexes. Toxin (doxorubicin or mycophenolic acid) was attached to synthetic AcMBP(l-14)Ala4 peptide, an analog of the natural acetylated NH2-terminal segment, AcMBP(1-14), of rat myelin basic protein (MBP). IAk-restricted, AcMBP(1-14)-specific AJ1.2 and 4R3.9 cloned murine T cells were killed by IAk-(AcMBP(l-14)Ala4-toxin). No killing resulted from incubating AJ1.2 and 4R3.9 cells with irrelevant MHC II-(peptide-toxin) or treating IEkrestricted, pigeon cytochrome c-specific k E 7 cloned murine T cells with IAk-(AcMBP(l-14)Ala4-toxin). T cell receptor-mediated T cell uptake  of  the  peptide-toxin moiety of relevant complex was blocked by anti-T cell receptor-df3 antibody and by excess toxin-free complex. LDw determinations revealed that cognate MHC II-(peptide-toxin) killed T cells much more effectively than did peptide-toxin conjugate alone. Finally, T cell uptake of peptide-toxin and intracellular release of toxin occurred after incubation with relevant MHC II-(peptide-toxin) containing radiolabeled toxin. These findings, which provide the first evidence that cloned T cells can be deleted with soluble, cognate “ I C II-(peptide-toxin) complexes, may have significant clinical relevance for antigen-specific therapy of autoimmune or other T cellmediated diseases.

chromatography; AcMBP(1-14)Ala4, etc., analogs of the acetylated NHz-terminal segment of rat myelin basic protein; OVA(324-3 3 6 ) C y~~~~, ovalbumin peptide analog. 9829; Fax: 415-368-9829. consisting of the stimulatory antigenic peptide combined with molecules of the restricting MHC class I1 antigen (1-5). Anumber of accessory molecules on both the APC and T cell also interact during the presentation of antigen (6,7). Although this model of antigen presentation is well accepted, the fate of the components of the TCR-(MHC II-peptide) complex following antigen stimulation and disengagement of the APC and T cell is unknown. Recent results from our laboratory indicated that interaction of purified relevant MHC II-peptide complexes with TCRs on cloned T cells in vitro leads to T cell internalization of only the peptide moiety of the complex (8). Based on this novel observation, the present study was undertaken to determine if cloned T cells could be deleted in vitro, using purified complexes of cognate MHC II-(peptide-toxin) containing an intracellularly cleavable peptide-toxin bond.
The association of T cells and particular MHC class I1 molecules with several autoimmune diseases has been well established (9-12) and provides a rational basis for development of antigen-specific therapies. Toxin conjugates of acetylcholine receptor using the plant toxin, ricin, have been employed for selective in vitro elimination of specific lymphocytes involved in triggering and progression of experimental autoimmune myasthenia gravis (EAMG) in animals (13,14). Recently, in vivo treatment of EAMG with acetylcholine receptor conjugated with another plant toxin, gelonin, also has been reported (15). Moreover, antigen-specific and MHC II-restricted T cell killing in vitro requiring lZ5I-labeled antigen and a nonspecific soluble factor of adherent cells has also been reported (16).
In contrast to large toxin molecules like ricin and gelonin, the smaller doxorubicin and mycophenolic acid molecules act intracellularly by intercalating into DNA (17) or inhibiting DNA synthesis (la), respectively. Doxorubicin, a glycoside antibiotic (19) and analog of daunorubicin which differs from the latter by a single hydroxylation site (201, is a n important anticancer agent (21). DNA has been considered to be the primary target for the cytotoxic action of this drug on susceptible cells (22). Mycophenolic acid is a novel immunosuppressive agent (23) distinct from cyclosporin A and FK506 in activity (24). In this study, doxorubicin or mycophenolic acid was coupled to the COOH terminus of MBP peptide analogs via an intracellularly cleavable disulfide or ester linkage, respectively. The peptidedoxorubicin or peptide-mycophenolic acid conjugate was combined with purified MHC I1 molecules, and the resulting MHC II-(peptide-toxin) complexes were used for antigen-specific deletion of cloned T cells in uitro.
MATERIALS AND METHODS Cells, Antibodies, and Chemicals-Murine T cell clones, AJ1.2 and 4R3.9, which respond to IAk-AcMBP(1-14) complexes, were obtained from the Laboratory of Dr. Patricia Jones, Stanford University, Stanford, CA. Cloned A.E7 T cells restricted for IEk in association with a peptide segment of pigeon cytochrome c (pCyt C(81-104)) were obtained from the laboratory of Dr. R. H. Schwartz. The hybridoma cell lines, T Cell Deletion by Soluble MHC I I -( P e p~i~-~~i n~ Complex 95 10-2.16, which produces monoclonal antibodies (IgG2b) against murine IAk, and MK-D6, which produces monoclonal antibodies (IgG2a) against murine LAd, were obtained from American Type Culture Collection, &&ville, MD. Affinity-purified hamster anti-TCR-cr/B monoclonal antibody (H57-595) and purified hamster I& (standard isotype 1 6 %~ antibody), specific for the 2 , 4 ,~t~t r o p h e n y l group were p m b s e d from Pharmingen, Inc., San Diego, CA. Doxorubicin hydrochloride, collidine, 2,2-dithiodipyridine (Aldrithiol), and diisopropylethylamine were purchased from Aldrich. 2-Iminothiolane was purchased from Pierce Chemical Co. and mycophenolic acid, bromoacetic acid N-hydroxysuccinimide ester, and dimethylsulfoxide were purchased from Sigma.
&r 48 h of stimulation, the cells were split 1:19 into complete RPMI medium c o n t~n g 20 unitdml of recombinant interleukin-2. Before stimulation, residual APCs were removed by subjecting cloned T cells to a 19% metrizamide density gradient centfigation, followed by two washes in RPMI 1640 medium. Cultured A.E7 T cells were stimulated similarly every 20 days with freshly irradiated splenocytes from BIO.A mice and pigeon cytochrome c (pCyt C) at a final concentration of 5 p~. Prior to use in experiments, the cells were subjected twice to 19% me-de density gradient centrifugation, washed in RPMI 1640, and resuspended in medium containing 10% fetal bovine serum.

( O G t buffer. Finally, the I A k or
LAd molecules were eluted with 20 m~ phosphate buffer, pH 11, containing 0.1 M NaCI, 1% OG, 0.02% sodium azide and 1 n m phenylmethylsulfonyl fluoride. Each fraction was neutralized with 1 M acetic acid to a final acetate concentration of 12 m~, and the I A k or I A d was then concentrated using an Amicon Centriprep-10 concentrator. Affinity-purified I A k and I A d molecules were characterized by 12% SDS-polyacrylamide gel el~rophoresis and silver staining. Synthesis and Purification of Peptides-The rat MBP peptide analogs, AcMBP(1-14)Ala4 and A~MBP(l-l4)Ala~Cys'~, with the sequences, AC-ASQARF'SQRHGSKY-NH~ and Ac-ASQARPSQRHGSKC-N H z , respectively; AcMBP(1-14)Ala3Ala4Alas and AcMBp(1-14)Al~~-Ala4Ala6Cys14 with the sequences, Ac-ASAARASQRHGSKY-NH2 and Ac-ASAARASQRHGSKC-NH,, respectively; and the ovalbumin peptide analog, OVA(32&336)Cy~~~~, representing the sequence, SQAVHAA-~~C -~z , were synthesized by the standard solid phase methodology using side-chain protected Fmoc (N-(9-fluorenylmethoxyca~onyl) amino acids and an Applied Biosystems 431A automated peptide synthesizer. The deprotected, crude carboxyl-terminal peptide amides were purified by reverse-phase HPLC, and the homogeneity and identity of the purified peptides were confirmed by mass spectroscopic analysis.
Synthesis and htrification of Peptide-Doxorubicin Conjugate Doxorubicin hydrochloride (20 mg, 34.4 p o l ) was dissolved in 1 ml of dry Me2S0 containing 100 pl of collidine. %n mg (72 pmol) of 2-iminothiolane and 75 mg (340 p o l ) of 2,2-dithiodipyridine were dissolved in 1 ml of dry Me2S0. The latter solution was added dropwise with stirring on a vortex mixer to the doxorubicin solution. After 6 h at room temperature and in the dark, the reaction mixture containing N-[4-(2-pyridylMithiobutyrimidoldoxorubicin intermediate was directly purified by reverse-phase HPLC on a preparative C18 column using linear gradient elution (solvent A: 0.1% aqueous trifluoroacetic; solvent B 0.1% t~u o~c e t i c acid in 70% aqueous acetonitrile). The structure of the intermediate product was confirmed by mass spectroscopy. Lyophilized, pure N-[4-(2-pyridyl)dithiobutyrimido]doxorubicin (2.3 mg) was dissolved in 0.5 ml of degassed water in a 15-ml polypropylene centrifuge tube. To this solution, 1 mg in 0.5 ml of degassed water of HPLCpure synthetic MBP or OVA peptide (containing cysteine at the COOH terminus) was added with vortex mixing. After 6 h at mom temperature in the dark, the disulfide-linked peptide-doxorubicin conjugate was purified by gradient elution reversephase HPLC on a preparative C18 column as described above. The structure and homogeneity of the peptide-doxorubicin conjugates were confirmed by mass spectroscopic analysis.
Synthesis and Purification of Peptide-Mycophenolic Acid

Con-
jugate-Mycophenolic acid (32 mg, 100 p o l ) and 12 mg (50 w o l ) of bromoacetic acid N-hydroxysuccinimide ester were dissolved in 200 pl of dry Me2S0 containing 9 r;t of d~sop~pylethylamine. After standing overnight at room temperature, the solution was added with vortex mixing to a solution of 5 mg 12.5 p o l ) of AcMBP(1-14)Ala4 or AcMBP(1-14)Ala3Ala4Ala6, in 100 pl of Me2S0 containing 50 pl (278 pmol) diisopropylethylamine. f i r 2 h at room temperature, the mycophenolyloxyacetylated peptide was precipitated by addition of 50 volumes of dry ethyl acetate, and the precipitate was collected by centrifugation, washed several times with ethyl acetate, and dried. The dried residue was dissolved in 0.1% t~uoroacetic acid and immediately applied to a preparative C18 reverse-phase column. Gradient HPLC elution (solvent A 0.1% aqueous trifluoroacetic acid; solvent 3 0.1% trifluoroacetic acid in 70% aqueous acetonitrile) was performed to obtain pure peptide-mycophenolic acid conjugate (with mycophenolyloxyacetic acid attached at the Ne-amino group of the position 13 lysine residue of MBP peptide analogs). Finally, the structure and homogeneity of the puriiied MBP peptide-mycophenolic acid conjugate were confirmed by mass spectroscopy. Preparation of L 4 -~~e p t~~-l l b~i n ) Complexes-Minity-purified IAk or I A d (100 pg) was incubated with a 50-fold molar excess (167 pg) of AcMBP(l-14)Ala4Cys14-doxorubicin or AcMBP(l-14)Ala4-mycophenolic acid conjugate at 37 "C for 48 h in a total volume of 1 ml. The excess unbound peptide-toxin conjugate was removed by dialyzing the complex three times against 3 liters of RPMI 1640 medium at 4 "C. The absence of free peptide-toxin in the final preparation was confirmed by TLC analysis as described elsewhere (27,28). For control experiments, an equivalent amount of peptide-toxin was incubated and dialyzed under identical conditions in the absence of I A k or I A d molecules.
Assay for Proliferation of T Cells Exposed to IA-(Peptide-llbxin) Complexes-The dialyzed IA-(peptide-toxin) complexes were mixed with 1 x lo6 cloned AT1.2, 4R3.9, or A.E7 T cells in a total volume of 1 ml in a 15-ml polypropylene tube and incubated at 37 "C for 24 h. Unbound IA-(peptide-~~n) complexes were removed by washing the cells three times with 15 ml of T cell medium (RPMI 1640 containing 10% fetal bovine serum, 2 n m L-glutamine, 10 n m HEPES, and 10 unitdml penicillidstreptomycin). The cells were examined at this stage for percent viability by staining with acridine orange/ethidium bromide (0.0003%/0.001%) and viewing under an epifluorescence microscope. For the proliferation assay, washed T cells were plated in triplicate in a 96-well microtiter plate at a cell density of 106 cells/200 pl per well in the presence of 5 unitdrnl recombinant interleukin-2. &r 72 h of incubation, 100 pl of cell supernatant was removed, and 10 pl of 5 m g / d MTT solution prepared in PBS were added to each well. The plate was incubated for 4 h at 37 "C, and 150 pl of acidhopropanol (1.7 ml of concentrated HCI in 500 ml of isopropanol) were added. The reaction mixture in each well was mixed thoroughly, and the absorbance was measured at 650 nm in a Molecular Devices V , , plate reader. R a d i~a~~i n g of the Doxorubicin Moiety of Pept~.Doxorubicin Con-~~a~B o l~n -H u n t e r reagent (500 pCil was converted to the carboxylic acid hydrazide by incubation with excess hydrazine for 16 h at room temperature. The excess hydrazine was removed by evaporating the reaction mixture under a stream of argon gas at room temperature. To the dried residue, 1 ml of 0.1 M hydrochloric acid containing 30% sodium chloride was added, and the product was extracted with 14 ml of ethyl acetate. The extract was then treated with 1 ml of 10% sodium bicarbonate containing 20% NaCl. The organic layer was recovered, trifluoroacetic acid was added to a final concentration of 0.01%' and the resultant *26-I-labeled Bolton-Hunter hydrazide reagent was stored at 4 "C until used. For radiolabeling, ethyl acetate was removed from the 126-I-labeled Bolton-Hunter hydrazide reagent by evaporation under a stream of argon gas at room temperature. Peptide-doxorubicin conjugate was dissolved in acetic acid, added to the dry radiolabeling agent, and incubated at room temperature for 6 h. The excess reagent was then removed by ethyl acetate precipitation of the peptide-toxin conjugate. The dried peptjde-lZ6-I-doxo~bicin conjugate was redissolved in acetic acid and reprecipitated with ethyl acetate. The final precipitate was washed with heptane, dried, and stored at 4 "C.

Complex"& equivalent amount of radiolabeled ~k -~A~~l -1 4 )~a 4 C~~4 -1~-I -d o x o r u b i~) or LAk-(& M B~1 -l~~a S A t a 4~a s C y s 1 4 -1~~I~d o x o~b i~~
complex (19,000 cpm) was incubated with lo7 4R3.9 T cells in a tot$ volume of0.5 ml of RPMI 1640 medium at 37 "C for 5 h. Following incubation, the cells were washed three times with 10 d of PBS, suspended in 300 $ of PBS, and counted in a y counter. To this suspension, 800 p l of acetonitrile containing 10 $ of ~ch.loroacetic acid were added. The samples were mixed well and centrifuged at 100,000 x g. The supernatant was recovered, the pellet was re-extracted with 1 ml of 80% aqueous acetanitrile containing 0.1% trichloroacetic acid, and the combined supernatants were vacuum-dried and dissolved in 5 $ of aqueous acetonitrile. Samples of the trichloroacetic acid extract were then applied onto a silica gel TLC plate and developed with the upper phase of butanol: acetic acid:water in the ratio of 10:1:3. The developed TZC plate was subjected to autoradiography, and the extent of intracellular disulfide bond cleavage was estimated by the percent of total counts recovered from the plate at the Rp value (0.9) for the cleavage product, 4mercap- RESULTS ANR D~S C U~~O~ ~~e T cell clones, AJ1.2 and 4B.9, were selected for their recognition of IAk complexes containing the Nwz&mninal nonapeptide segment, AcMBP(l-9), of rat MBP which is acetylated at the NHz terminus (29). These cell lies also proliferate when stimulated with APC-associated IAk complexed with   or the MBP peptide analog, AcMBP(1-14)Ala4, but not with the analog, A~B F ( l -1 4 )~a 3~a 4 A l a 6 , although both peptide analogs were found to bind equally well to Uk.
The natural sequence o f A~P~1 -1 4 ) contains a lysine residue at position 4. Replacement of this lysine residue by alanine r e s u h in an increased binding of the peptide analog to LAk (30) which may explain the heteroclitic response observed for ia vitro s t~u l a t~o n of cloned T cells with this analog (31).
Two toxin molecules, doxorubicin and mycophenolic acid, were covalently linked to the Synthetic peptide analog, A~MBPfl-l4h%la~Cys~~. For covalent attachment of these pep tides to doxorubicin, the tyrosine residue at position 14 was substituted with cysteine during peptide synthesis, The threestep synthesis of peptide-doxorubicin conjugate containing an intracellul~ly cleavable disulfide linkage between the peptide and doxorubicin moieties is shown in Fig. 1. In the first step, a free s u l~y~~ group was generated at the amino group of doxorubicin h y d r~~o n d e by reaction with 2 -~n o~o l a n e . The resulting sulfhydryl group was converted in the second step to a ~~o p~d y l group, using 2 , Z ' -~t~o~p~~n e .  Peptide-mycophenolic acid conjugate containing a n intracellularly cleavable ester bond between peptide and toxin was prepared in a two-step synthesis (Fig. 2). In the first step, an active N-hydroxysuccinimide ester derivative of mycophenolic acid was generated by reaction with bromoacetic acid N-hydroxysuccinimide ester. In the second step, the active ester product, mycophenolyloxyacetic acid N-hydroxysuccinimide ester, was then linked by amide bond formation to the Ne-amino group of the position 13 lysine residue of AcMBP(1-14)Ala4. The resulting peptide-mycophenolic acid conjugate was purified by reverse-phase HPLC, and the structure was confirmed by mass spectroscopy.
The toxin conjugates of the MBP and OVA peptide analogs were complexed with affinity-purified I A k and I A d , respectively, and the resulting purified complexes were incubated with the T cell clones. Only T cells incubated with cognate IAk-(peptidedoxorubicin) or IAk-(peptide-mycophenolic acid) complexes were killed (Fig. 3). Untreated T cells and T cells incubated with either purified I A k alone, I A k complexed with unmodified AcMBP(1-14)Ala4, or IAd-(OVA(324-336)Cys336-doxorubicin) complex were unaffected.
The specificity of T cell killing by MHC 11-(peptide-toxin) was further demonstrated in three different experiments the data of which are presented in Fig. 4. To demonstate that the killing of T cells is mediated by the T cell receptor, 4W.9 cloned T cells were incubated with IAk- (AcMBP(1-14)Ala4Cys14-doxorubicin) complex in the presence of H57-595 anti-TCR-dP monoclonal antibody. The killing of T cells was substantially reduced in the

IA'-(peptide-toxin) complex required for 50% killing of AJ1.2 T cells
The LD50 values for toxin alone and peptide-toxin conjugate were calculated from the results of dose response studies, and the LD50 values for IAk-(peptide-toxin) complexes were estimated from the data presented in Fig. 3. Ten days aRer antigen pulsing, 1 x 106AJ1.2 T cells were incubated at 37 "C for 24 h with either free doxorubicin (0.00-10 PM), purified A~MBP(l-l4)Ala~Cys~~-doxorubicin conjugate (0.00-100 p~) , or purified AcMBP(l-14)Ala4-mycophenolic acid conjugate (0.00plate at a cell density of lo5 celldwell per 200 pl of medium containing 100 PM). Treated cells were plated in triplicate in a 96-well microtiter 5 unitdml recombinant interleukin-2. Proliferation was measured after 72 h by the MlT colorimetric assay. presence of the anti-TCR-a@ antibody (Fig. 4, lane 3 ) . In a control experiment, a n equivalent amount of isotype-matched hamster IgG did not show any inhibition of T cell killing (Fig.  4, lane 4 ) , demonstrating that the binding of the relevant MHC 11-(peptide-toxin) complex and the uptake of the peptide-toxin moiety were TCR-mediated. Similarly, in a competition assay, T cell killing was inhibited by incubating the 4W.9 T cells with IAk- (AcMBP( 1-14)Ala4Cys'4-doxorubicin) complex in the presence of a %fold molar excess of IAk-AcMBP(1-14)Ala4) complex containing no conjugated toxin (Fig. 4, lane 5). Finally, to demonstrate that irrelevant T cells are not killed by similar concentrations of these MHC 11-(peptide-toxin) compIexes, A.E7

TABLE I1 T cell internalization and intracellular cleavage of the peptide-toxin moiety of IAk-(AcMBP(I-14)Ala4-Cys1*-126-I doxorubicin)
4R3.9 T cells were incubated with complexes at 37 "C for 5 h. After incubation, the T cells were washed, radioactivity was measured by counting, and the percent of total applied radioactivity associated with the cells was calculated. The T cells were then lysed with acetonitrildtrichloracetic acid and the lysate was centrifuged at 100,000 x g. The resultant (trichloracetic acid) extract was analyzed by TLC, and the amount of radioactivity at the Rf for free Pmercaptobutyrimidodoxorubicin was measured and used to calculate the percent cleavage of the T cell-internalized peptidedoxorubicin conjugate.  Fig. 4, lanes 6, 7, and 8, no significant killing of AX7 cells was observed under identical experimental conditions. The concentration of doxorubicin, AcMBPf 1-14)Ala4CysI4doxorubicin conjugate, or AcMBP(1-14)Ala4-mycophenolic acid conjugate required for 50% killing (LD50) of the T cell clones was determined in a dose-response study. These were compared with an estimation of the corresponding LD5o for T cell killing with the cognate IAk-(AcMBP(1-14)Ala4Cys14-doxombicin) or IAk-(AcMBP(l-14)Ala4-mycophenolic acid) complex (see Fig. 3 for data representation). As shown in Table I, 50% killing of AJ1.2 T cells was observed after 24 h of incubation with relevant IAk-(peptide-toxin) complex at a concentration less than or equal to 0.3 w. As expected, incubation ofAJ1.2 T cells with uncomplexed peptide-doxorubicin or peptide-mycophenolic acid conjugate resulted in T cell killing, but achieving 50% cell killing under identical incubation conditions required 16 times (5 pd or 160 times (50 w) the concentration of the respective IAk-(peptide-doxorubicin) or IAk-(peptide-mycophenolic acid) complex. The concentration of free, lipophilic doxorubicin required for 50% cell killing was also analyzed and found to be 0.5

Complex T cell-associated
w. The LDs0 of free mycophenolic acid could not be accurately determined as a result of the insolubility of mycophenolic acid in aqueous medium. The internalization by T cells of peptide-toxin molecules from ternary TCR-(MHC 11-(peptide-toxin)) complex was demonstrated using radiolabed doxorubicin. A~MBP(l-14)Ala~Cys'~doxorubicin or AcMBP(l-14)Ala3Ala4Ala6Cys'4-doxorubicin conjugate was radiolabeled at the ketone carbonyl group of the doxorubicin moiety by reaction with lz5-I-1abeled Bolton-Hunter hydrazide reagent, prepared as described under "Materials and Methods." Cloned 4R3.9 T cells were incubated with IAk- (AcMBP(1-14)Ala4Cys14-'25-I-doxorubicin) or IAk-(AcMBP (l-14)~a3Ala4~a6Cys14-'25-I-doxorubicin) complexes a t 37 " c for 5 h. The treated T cells were washed and lysed with acetonitrile/trichloroacetic acid in order to extract peptides. The amount of radioactivity recovered in the trichloroacetic acid extract indicated that 44% of the total radioactivity applied in the form of relevant L4k- (AcMBP(1-14)Ala4Cys14-1z5-I-doxorubicin) complex was internalized by the T cells (Table   11). In a control experiment in which cells were exposed to irrelevant IAk- (AcMBP( 1-14)Ala3Ala4Ala6Cys14-125-I-doxorubicin) complex, the cell-associated radioactivity, as determined by trichloroacetic acid extraction, was only 5.9% of the total applied radioactivity.
Disulfide and ester bonds in prodrugs are generally known to be cleaved after uptake into target cells, and intracellular release of peptides from cell-internalized substances has been described (33, 34). To determine if free 4-mercaptobutyrimidodoxorubicin was released by intracellular disulfide bond cleavage of the internalized peptide-doxorubicin moiety from T cell-bound IAk-(peptide-125-I-doxorubicin) complex, the trichlo-roacetic acid extracts were subjected to TLC. The bands at RF values corresponding to peptide-125-I-doxorubicin and free Iz5-I-4-rnercaptobutyrimidodoxorubicin were excised and counted. As shown in Table 11, 92.5% of the internalized peptide-doxorubicin conjugate was cleaved intracellularly at the disulfide bond joining toxin with peptide.
The killing of T and B lymphocytes by toxin conjugated to intact antigen molecule has been reported (13,14), and the deletion of mature T cells by this method can be used to treat autoimmune diseases in animal models (15). The deletion of autoreactive T cells by soluble MHC 11-(peptide-toxin) conjugates in which the toxin is linked to a T cell epitope of the antigen by an intracellularly cleavable linker may provide further specificity. The data may have relevance for development of therapies aimed at deletion of antigen-specific T helper cells in autoimmune and other T helper cell-mediated diseases.