Chemical identification of a low abundance lysozyme peptide family bound to I-Ak histocompatibility molecules.

The processing by antigen-presenting cells (APC) of the protein hen egg-white lysozyme (HEL) results in the selection of a number of peptide families by the class II major histocompatibility complex (MHC) molecule, I-A(k). Some of these families are expressed in very small amounts, in the order of a few picomoles/10(9) APC. We detected these peptides from an extract of class II MHC molecules by using monoclonal anti-peptide antibodies to capture the MHC-bound peptides prior to their examination by HPLC tandem mass spectrometry. Here, we have identified several members of a family of peptides encompassing residues 20-35, which represent less than 1% of the total HEL peptides. Binding analysis indicated that the core segment of the family was represented by residues 24-32 (SLGNWVCAA). Asn-27 (shown in boldface) is the main MHC-binding residue, mapped as interacting with the P4 pocket of the I-A(k) molecule. Analysis of several T cell hybridomas indicated that three residues contacted the T cell receptor: Tyr-23 (P-1), Leu-25 (P3), and Trp-28 (P5). The HEL peptides isolated from the APC extract were sulfated on Tyr-23, but further analysis showed that this modification did not occur physiologically but took place during the peptide isolation.

The processing by antigen-presenting cells (APC) of the protein hen egg-white lysozyme (HEL) results in the selection of a number of peptide families by the class II major histocompatibility complex (MHC) molecule, I-A k . Some of these families are expressed in very small amounts, in the order of a few picomoles/10 9 APC. We detected these peptides from an extract of class II MHC molecules by using monoclonal anti-peptide antibodies to capture the MHC-bound peptides prior to their examination by HPLC tandem mass spectrometry. Here, we have identified several members of a family of peptides encompassing residues 20 -35, which represent less than 1% of the total HEL peptides. Binding analysis indicated that the core segment of the family was represented by residues 24 -32 (SLGNWVCAA). Asn-27 (shown in boldface) is the main MHC-binding residue, mapped as interacting with the P4 pocket of the I-A k molecule. Analysis of several T cell hybridomas indicated that three residues contacted the T cell receptor: Tyr-23 (P؊1), Leu-25 (P3), and Trp-28 (P5). The HEL peptides isolated from the APC extract were sulfated on Tyr-23, but further analysis showed that this modification did not occur physiologically but took place during the peptide isolation.
To understand the nature of the CD4 T cell response to peptides in a protein antigen, it is important to explain the basis for peptide selection during processing. It requires, therefore, the identification and quantification of the peptides that are selected by the class II MHC 1 molecules of the antigenpresenting cells (APC). There are two major difficulties with the identification of the class II MHC-bound peptides. The first is that peptides are selected as families, which comprise many members; a family is characterized by having a core sequence that interacts with the P1 to P9 anchoring sites of the MHC binding groove and flanking residues (1)(2)(3)(4)(5)(6)(7). The flanking residues vary greatly in length and determine the number of members in a family. The flanking residues contribute to the affinity of binding to class II molecules and to the half-life of the peptide-MHC complex on the surface of APC (7)(8)(9)(10). Additionally, flanking residues can have a marked effect on the T cell recognition (11,12). The length of the peptides and their size heterogeneity, however, make it difficult to purify them for analysis as well as to predict binding motifs. A second difficulty with the analysis of class II-bound peptides is that their amounts vary greatly among the various families, by as much as 250-fold (13). Thus, characterization of these heterogeneous peptide families is particularly difficult.
Studying the HEL protein as an antigen, we identified most of the segments selected for presentation by the I-A k molecule (14). Most of the HEL-specific T cell response was directed to the HEL epitopes: HEL- (18 -33), HEL-(31-47), HEL-(48 -63), and HEL-(115-129) (14). In a previous study, we identified the naturally processed peptides of the HEL-(48 -63) family (DG-STDYGILQINSRWW) (4), the most abundant epitope, occupying up to 10 -20% of the total class II MHC molecules (15). However, we had to develop a peptide immunoaffinity capture technique to isolate and characterize the lower abundance HEL peptide families (3). This approach consisted of purifying and concentrating the HEL peptides from a class II MHC peptide extract by using an anti-HEL peptide-specific monoclonal antibody (mAb) followed by their identification by HPLC/electrospray tandem mass spectrometry (ESI-MS). This method allowed the identification and quantification of the HEL-(31-47 peptide family (AAKFESNFNTQATNRNT) (3). The anti-HEL peptide-specific antibodies were also used by applying ELISA methods to quantitate the peptides in the class II MHC peptide extract (13).
In this article we report our studies of a second HEL peptide family that stimulates T cells, although it is represented in very low amounts on the APC class II molecules, i.e. ϳ100 copies of 20 -35-mer peptide/10 5 I-A k molecules (13). Despite the 20 -35 family occurring in extremely small amounts, we were able to characterize five members by using the antipeptide capture technique.

MATERIALS AND METHODS
Cell Lines-The murine B cell lymphoma lines M12.C3.F6 (M12-A k ) (16) expressing class II I-A k molecules and the M12.C3.F6 cell line transfected with a membrane form of HEL (M12-A k mHEL) were used (17). Three other C3F6 cell lines expressing mutations of HEL were generated as described previously (18). The lines contained alanine substitutions for Ser-24, Asn-27, or Cys-30. All cell lines were cultured in Dulbecco's modified Eagle's medium supplemented with 5% heat-* This work was supported by grants from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 1 The abbreviations used are: MHC, major histocompatibility complex; HEL, hen egg-white lysozyme; M12-A k mHEL, cell line transfected with a membrane form of HEL; ESI-MS, electrospray tandem mass spectrometry; MS/MS, tandem mass spectrometry; CID, collision-induced dissociation; EAY, acidic random polymer-containing tyrosine, poly(Glu-Ala-Tyr); PAPS, 3Ј-phosphoadenosine 5Ј-phosphosulfate; TPST, tyrosylprotein sulfotransferase; APC, antigen-presenting cell; mAb, monoclonal antibody(s); ELISA, enzyme-linked immunosorbent assay; OVA, ovalbumin; IL-2, interleukin 2; TCR, T cell receptor. inactivated normal calf serum and were tested for levels of I-A k and HEL before their analysis.
Production of mAb-A mAb (15B7-1B5, IgG 1 ) that recognizes the HEL-(18 -33) epitope (DNYRGYSLGNWVCAAK) was generated as described previously (13). (HEL- (18 -33) had been identified previously with synthetic peptides as one segment that stimulated T cells.) This mAb recognizes the free HEL peptide in solution or bound to an ELISA plate. Briefly, the anti-HEL- (18 -33) peptide mAb was generated by repeated immunizations of CB.17 mice with synthetic HEL- (18 -33) peptide coupled to OVA protein using bromoacetyl succinimide (Sigma). A cysteine was added to the carboxyl terminus of the peptide to facilitate coupling to OVA, and a serine was substituted for cysteine at residue 30. Mice were injected subcutaneously with 400 g of HEL- (18 -33) peptide-OVA protein in complete Freund's adjuvant (Difco, Detroit, MI) followed by two subcutaneous injections in incomplete Freund's adjuvant (400 g of HEL- (18 -33) peptide-OVA/immunization) at 2-week intervals. Finally, 3 days before B cell fusion, the mice were boosted intravenously with 100 g of HEL-(18 -33) peptide-OVA in sterile saline solution. Spleen cells from immunized mice were fused with the myeloma fusion partner P3 ϫ 63.Ag8 using polyethylene glycol 1500 (Roche Molecular Biochemicals) by standard procedures. The B cell hybridomas were selected by testing supernatants in ELISA plates coated with the HEL-(18 -33) peptide, and the positive supernatants were screened immediately by binding to free HEL- (18 -33) peptide in a competitive ELISA (13). The 15B7-1B5 B hybridoma was selected and injected into pristane (Sigma)-treated SCID (severe combined immunodeficient) mice. The 15B7-1B5 mAb was purified from ascites using protein A-Sepharose (Sigma).
Isolation of I-A k -associated Peptides-Class II I-A k molecules were isolated from ϳ0.8 -3.2 ϫ 10 10 (ϳ20 -60 liters of cell culture) B lymphoma cells M12-A k mHEL or control B cell line M12-A k fed with exogenous HEL by immunoaffinity chromatography, as described previously (3,13), but with some modifications to improve the peptide isolation method. Briefly, the APCs were lysed with 40 mM MEGA-8/MEGA-9 detergents (Sigma) in phosphate-buffered saline in the presence of enzyme inhibitors (1 mM phenylmethylsulfonyl fluoride, 10 mM iodoacetamide, and 20 g/ml leupeptin). The suspension was centrifuged at 8,000 rpm for 30 min at 4°C. The I-A k molecules were isolated from the supernatant using the anti-I-A k mAb 40 F, which recognizes the I-A k ␤-chain (19). Specifically, the supernatant was incubated with mAb 40 F coupled to cyanogen bromide-activated Sepharose beads (Sigma) overnight at 4°C. Then, the Sepharose beads were loaded into a chromatography column (Bio-Rad) and washed extensively with a large volume of phosphate-buffered saline at pH 7.2 until the OD readings at 280 nm and 214 nm of the washing solution were equal to the OD reading of the blank solution; the excess salt in the column was removed with 5-10 column volumes of Milli-Q water. After the washes, the 40 F mAb beads were transferred to another chromatography column (Bio-Rad) but without the porous polymer bed support. Glass wool fiber was placed in the column opening to retain the beads, and the I-A k ⅐peptide complexes were subsequently eluted with 0.1% trifluoroacetic acid. The replacement of the bed support was made to minimize the loss of peptides. To recover all of the I-A k material, the flow-through was incubated with new 40 F mAb beads again, and the class II molecules were eluted as mentioned above. Peptides were separated from I-A k molecules by filtration through a Centripep (YM-10) centrifugal filter device (Amicon, Beverly, MA). The peptide extract was dried down and reconstituted in a volume of 5 ml with 0.1 M Tris-HCl, pH 7.4, and incubated with 0.8 -1.1 ml of a 50% v/v of 15B7-1B5 mAb (anti-HEL- (18 -33)) coupled to cyanogen bromide-activated Sepharose beads (Sigma) overnight at 4°C. After an overnight rotation, the beads were loaded into a chromatography column (Bio-Rad) and washed extensively with a large volume of phosphate-buffered saline, and the excess salt in the column was removed with distilled water. After the washes, the 15B7-1B5 beads were transferred into a 5-ml syringe (BD Pharmingen). Glass wool fiber was placed in the syringe opening to retain the beads. The bound peptides were eluted with 0.1% trifluoroacetic acid into prelubricated tubes (Costar, Corning Glass). Four fractions were collected at 1.0 ml/fraction, and each fraction was dried down and reconstituted in a total volume of 20 -60 l of a 2% acetronitrile, 0.6% acetic acid and analyzed by mass spectrometry. To ensure the complete removal of the peptide, the extract from the I-A k was neutralized with 1.0 M Tris-HCl, pH 7.4, and passed two to four more times through the 15B7-1B5 mAb column. The HEL peptides were eluted from the mAb column with 0.1% trifluoroacetic acid as described above. The recovery of peptide was estimated using synthetic peptides to range from 50 -75%.
Analysis by Mass Spectrometry-Analysis of the class II I-A k peptides and synthetic peptides were carried out using the HPLC Finnigan LCQ Classic and LCQ Deca electrospray tandem mass spectrometers (Finnigan, San Jose, CA). The peptide samples were reconstituted in 20 -60 l of 2% acetronitrile, 0.6% acetic acid solution, and 5-15 l of the sample were injected into the capillary reverse-phase HPLC (a Zorbac C18 0.3 ϫ 25 mm column; Micro-Tech, Sunnyvale, CA) connected on-line to the electrospray mass spectrometer. A flow rate of 4.5 l/min into the electrospray source was maintained with either a 600-MS Waters HPLC system (for the LCQ classic) or a Micro-Tech Scientific Ultra Plus HPLC system (for the LCQ Deca). The gradient used was as follows: solvent A (acetic acid in water) was kept constant at 0.6%; solvent B (acetronitrile) was kept constant at 2% for the first 5 min, increased to 15% in the next 5 min, and then to 60% in the next 50 min. The capillary temperature was kept at 150°C. The scan range used was m/z 700 -1400, and three "microscans" were averaged to give one scan. Acquisition was started 10 min after the commencement of the LCQ run. For collision-induced dissociation (CID) experiments, precursor ions were selected with a width of m/z 2.5 and activated using collision energies ranging between 25 and 35% of the maximum collision energy. For the LC and CID analyses, 200 fmol of the synthetic peptides was used. Sequence analysis was performed using an automated protein data base sequencing program (SEQUEST; John Yates, University of Washington, Seattle) on an ICIS work station (Finnegan), which compares the experimental production of the mass spectra to the predicted ion mass spectra.
Peptide Binding Assays-I-A k molecules expressed in baculovirus were used in a binding competition assay with a standard radiolabeled peptide (20). I-A k molecules were expressed containing a peptide from the invariant chain, the CLIP peptide, tethered to a linker segment having a thrombin-sensitive sequence. The purified molecules were treated with thrombin at the same time that a standard I-125 radiolabeled HEL peptide was added. (The peptide was HEL-(52-61) with an added tyrosine at the amino terminus: YDYGILQINSR). The time of incubation was 18 -24 h; all reactions were carried out at pH 5.5. The amounts of peptide bound about 20 -30% of input cpm. Variable amounts of nonlabeled peptides were tested in the reactions, and the amounts that inhibited 50% (IC 50 ) of the binding of the standard radiolabeled peptide to I-A k were estimated.
Generation of T Cell Hybridomas and T Cell Assay-T cell hybridomas were generated by immunizing B10.BR mice with 10 nmol of HEL protein emulsified with complete Freund's adjuvant (Difco) in the footpad (14). A cell suspension was prepared from the draining lymph nodes at day 7 after immunization. After 72 h of restimulation in vitro with HEL-(18 -33) peptide, T cells were fused with the thymoma cell line BW5147␣ Ϫ ␤ Ϫ . T hybridoma lines specific to the HEL-(18 -33) peptide were selected on the basis of peptide sensitivity and subcloned by limiting dilution. T hybridoma assays were performed in 96-well tissue culture plates. T cell hybridomas were added at 10 5 cells/well and incubated with the B cell lymphoma line M12-A k at 5 ϫ 10 4 cells/well in the presence of different doses of HEL protein or test peptide in a total volume of 200 l; the mixture was then incubated for 20 h at 37°C with 5% CO 2 . Subsequently, each well was assayed for levels of IL-2 using the IL-2-dependent cell line, CTLL. In a total volume of 200 l, 100 l of supernatant was transferred to 20,000 CTLL cells and cultured for 18 h. In the last 8 h, the cells were pulsed with 0.5 Ci/well of Studies on Tyrosine Sulfation of HEL Peptides-For a control on the finding of sulfation of the peptide, I-A k molecules containing a synthetic peptide were added to a control cell lysate from M12-A k cells. I-A k molecules expressed in baculovirus (20) or isolated from APC (21) were used to make the HEL-(20 -35) peptide⅐I-A k complex in vitro. Similar results were obtained with both I-A k preparations. I-A k molecules (300 g) were incubated with 75 nmol of synthetic HEL- (20 -35) peptide in the presence of 50 mM L-lysophosphatidylcholine and 1 M 2-N-morpholinoethane sulfonic acid at pH 5.5. The mix reaction was incubated for 7 days at room temperature. The HEL-(20 -35)⅐I-A k complex was added into a control M12-A k cellular lysate obtained from ϳ10 10 M12-A k cell lysed with 100 ml of MEGA 8/MEGA 9 solubilization buffer (10 8 cell equivalents/1 ml of lysate). The isolation of the HEL-(20 -35) peptide and its MS analysis was performed as described above.

RESULTS AND DISCUSSION
To isolate and quantify the naturally processed forms of the HEL-(18 -33) epitope, we used the 15B7-1B5 specific to the HEL peptide sequence from residues 18 -33: DNYRGYSLGN-WVCAAK (13). This mAb recognizes the HEL-(18 -33) peptide both in solution and bound to an ELISA plate and had been used to determine the abundance of this HEL peptide on I-A k molecules isolated from an APC line that expresses a membrane form of HEL, M12-A k mHEL. The HEL-(18 -33) epitope (13) was expressed in low abundance on the M12-A k mHEL APC, i.e. about 9 pmol of peptide/5.6 ϫ 10 6 APC. In this study, we isolated the natural sequence, represented mostly by the 20 -35 peptide, and identified the amino acid residues responsible for anchoring the peptide to the I-A k molecules as well as those residues that interact with the T cell receptor (TCR). We call attention to the finding that the peptide was found sulfated, but this change took place following its isolation.
Identification of Naturally Processed Forms of the HEL- (18 -33) Epitope-We first extracted all of the peptides bound to class II molecule I-A k of the M12.A k mHEL line. Then the members of the HEL-(18 -33) peptide family were isolated and concentrated by immunoaffinity chromatography using the 15B7-1B5 mAb specific to the peptide- (18 -33). The eluted peptides were analyzed by mass spectrometry. In the mass chromatogram of the purified peptides, we observed several peptides eluting at retention times between 28 and 30 min.  (Fig. 1). The most abundant member of this HEL family was peptide- (20 -35). The amino acid sequence of the HEL-(20 -35) peptide was determined by MS/MS analysis; a search conducted in the protein sequence data base, using the SEQUEST program, identified this sequence as corresponding to the HEL peptide.
In addition to these HEL peptide masses, we observed three doubly charged ions with 40 m/z units higher that the HEL peptides and with identical retention times. The m/z values of these ions were 972.5 2ϩ , 1030.5 2ϩ , and 1087.5 2ϩ (Fig. 1). This m/z difference indicates that the HEL peptides were likely post-translationally modified by either a phosphate or sulfate group, either of which would add a mass of 80 Da to the peptide or protein (for doubly charged ions these modifications change the m/z of peptides by 40). Although we used larger m/z mass ranges for many of our experiments (i.e. m/z 700 -1400) than that shown in Fig. 1, we found no signals of greater intensity than those of the putative sulfated peptides and the peptides themselves.
The HEL- (20 -35) Epitope Is Modified by a Sulfate Group-The amino acid sequence of the HEL-(20 -35) epitope (YRGYS-LGNWVCAAKFE) contains three potential residues to be modified: the tyrosines at positions 20 and 23, which could be modified by either sulfation or phosphorylation, and a serine residue at position 24, which could be phosphorylated. We ruled out the phosphorylation at serine 24 as a possible modification because the recognition of the anti-HEL-(18 -33) mAb (15B7-1B5) used to isolate this HEL peptide family depends on an unmodified serine 24 (data not shown). Therefore, the only two possible modification sites on the HEL-(20 -35) epitope are the two tyrosine residues at positions 20 and 23.
To identify the nature of the modification present on the naturally processed HEL-(20 -35) epitope, we isolated from the M12-A k mHEL line the HEL-(20 -35) peptide family. The most abundant member of this family (HEL-(20 -35) peptide) was submitted to CID in the ion trap. The resulting product-ion spectrum was compared with those of synthetic sulfated or phosphorylated HEL-(20 -35) peptides. The product-ion spectra of the naturally processed HEL peptide and the synthetic modified peptides are very similar (Fig. 2). In contrast, the CID spectrum of the synthetic phosphorylated HEL-(20 -35) peptide is significantly different. This observation strongly indicates that the HEL-(20 -35) peptide isolated from APC was sulfated (not phosphorylated) on tyrosine. In subsequent MS experiments, we identified the tyrosine at position 23 as the amino acid modified by the sulfate group. Synthetic peptides with sulfated tyrosine at either residue 20 or 23 were added to the natural peptide and submitted for analysis; the addition of the Tyr-20 sulfated peptide resulted in two peaks (retention times of 28.6 and 29.5 min) that were base line-resolved, whereas addition of the Tyr-23 sulfated peptide only yielded one peak (29.5 min).
Interestingly, when we analyzed the mass spectrum of the synthetic HEL-(20 -35) peptide that was sulfated at Tyr-23, we observed two ions with m/z values of 932.5 2ϩ and 972.5 2ϩ . Those m/z values correspond to the unsulfated and sulfated forms of the HEL-(20 -35) peptide, respectively. This facile loss of sulfur trioxide that accompanies electrospray ionization of a sulfated peptide was described previously (23,24) and it is due to the weak bond between the sulfate group and peptide (25) and the thermochemical stability of SO 3 . Furthermore, there were a considerable number of fragment ions generated in the CID of the synthetic phosphorylated peptide, which were not present either in the product-ion spectrum of the naturally processed peptides or in that of the synthetically sulfated peptides (Fig. 2). The loss of SO 3 from sulfated peptides occurs so readily that the typical fragmentation of the peptide chain to give b and y ions, for example, is not competitive. The stronger linkage to phosphate, however, allows the other fragmentations to compete with the loss of the phosphate group. Others have pointed out this contrast between sulfated and phospho- rylated peptides (23)(24)(25), and the difference appears to be general.
The loss of 98 atomic mass units (either H 3 PO 4 or HPO 3 and H 2 O) that is observed for the phosphorylated standard is often viewed as difficult or impossible from peptides that contain phosphorylated Tyr, as exemplified in a recent report (26). Nevertheless, there is ample precedent for this loss from singly charged phosphopeptides, beginning with some of the earliest reports of matrix-assisted laser desorption ionization (MALDI)/MS of these materials (27)(28)(29). The precedent that is most relevant to our observations is that reported by DeGnore and Qin (30), who showed that doubly charged 20-mer peptides phosphorylated on Tyr can lose 98 Da in CID to produce abun-dant product ions. It is likely that the loss takes place in two steps, the first being that of HPO 3 and the second that of H 2 O. This loss of 98 Da, however, is not general for peptides that contain phosphotyrosine, and the rules governing its occurrence are not known.
We emphasize that the mass spectrum of the naturally processed HEL- (20 -35) peptide family also showed the sulfated and unsulfated versions of the HEL- (20 -35), HEL- (19 -35), and HEL-(18 -35) peptides (Fig. 1). All of these sulfated and unsulfated forms showed identical retention times in the HPLC preceding the MS, indicating that the signal for the unsulfated HEL peptides in this mass spectrum was because of the loss of the sulfur trioxide from the sulfated peptide during the ionization process in the mass spectrometer and not the presence of both sulfated and unsulfated HEL peptides in the original sample.
Some of the low abundance ions could not be detected because the (chemical) noise level in certain m/z regions of the naturally processed species' spectra was poorer than that of the standard. Nevertheless, the agreement between spectra and   Fig. 7. Ϫ, no response or very poor response at the highest peptide concentration tested (10 M). ND, not determined.
a The T cell hybridomas tested were generated and selected using the HEL- (18 -33) peptide. Therefore, the recognition of the Ala-substituted peptides at positions 34 and 35 was positive as expected. To know whether these two amino acids (Phe-34 and Asp-35) play a role in the T cell recognition of peptide-(20 -35), we generated T cell hybridomas specific to HEL- (20 -35) peptide, all of them (10 of 10) recognized the Alasubstituted peptides-(20 -35) Ala-34 and Ala-35, indicating that these two amino acid are not T cell contact residues.
the sequence coverage add confidence that the naturally processed peptide is sulfated HEL- (10 -35).
In experiments not shown, we also examined HEL peptides from the M12.C3.F6 line cultured in the presence of exogenous HEL. We identified HEL- (20 -35), HEL- (19 -35), and HEL- (18 -35) as observed in the M12.A k mHEL APCs experiment (Fig. 1). The most abundant member of this HEL peptide family was the peptide HEL- (20 -35). We also identified peptide- (20 -36) and -(20 -37) but in lesser abundance. All of the HEL peptides carried a post-translational modification of 80 atomic mass units, as indicated by the two signals separated by 40 m/z in the mass spectra.
Tyrosine Sulfation of the HEL- (20 -35) Epitope is an Artifact Generated during the Peptide Isolation Procedure-Tyrosine sulfation is a post-translational modification of secretory, plasma membrane, and lysosomal proteins occurring in all multicellular eukaryotic organisms (31,32). This modification is catalyzed by the TPST, an integral membrane glycoprotein residing in the trans-Golgi network (33). TPST uses PAPS as the sulfate donor in the sulfate transfer reaction. Based on the Trans-Golgi localization of TPST, we were surprised by the isolation of sulfated HEL peptides from APC fed with exogenous HEL protein. Therefore, we evaluated whether the tyrosine sulfation of this HEL peptide was occurring physiologically or was generated during the peptide isolation procedure. We formed in vitro the unsulfated HEL-(20 -35) peptide⅐I-A k complex. When the 20 -35 peptide was isolated from this com-plex and examined by MS, we found the correct mass of the unsulfated peptide to be 932.9 2ϩ (Fig. 3A). We next added the complex into a control M12-A k cellular lysate, at the cell equivalent concentration used to obtain the previous results. The mass spectrum of the recovered HEL peptide showed the expected mass ion for the sulfated HEL- (20 -35) peptide, m/z of 972.5 (Fig. 3B). The presence of sulfate group in the HEL peptide was confirmed by MS/MS analysis. Indeed, other studies using radioactive PAPS to evaluate the TPST activity in the M12-A k cellular lysate indicated the presence of such activity in the extract (Fig. 4). (We also examined, using HPLC/MS, whether the HEL protein isolated from the APCs M12-A k mHEL or from M12-A k fed with exogenous HEL was sulfated on tyrosine residues. We did not find any evidence of sulfation on either or both protein samples, indicating that under physiological conditions in APC, the HEL protein was not sulfated.) Finally, we generated a panel of T cell hybridomas that specifically recognize sulfated peptide. These T hybridomas, developed by immunizing mice with sulfated HEL-(20 -35) peptide, were selected in vitro for reactivity to the sulfated HEL peptide. We used these T hybridomas as a read-out system to evaluate whether the sulfated HEL peptides were presented on APC. None of these T cells reacted with APC cultured in HEL or in APCs, which express the HEL protein as a membranebound protein, confirming that under physiological conditions epitope-(20 -35) was not sulfated. Thus, the tyrosine sulfation of the HEL-(20 -35) family of peptides was generated during the peptide isolation procedure, resulting from the presence of TPST in the extract.
Binding Studies-The binding of the most abundant peptide, HEL- (20 -35), to I-A k molecules was tested. In 17 different experiments, the IC 50 averaged 1.4 M. Although the degree of binding varied about 30% depending on the preparation, it was highly consistent within experiments in which various mutant peptides were tested. Fig. 5 summarizes experiments testing HEL- (20 -35) peptide in which several residues were changed to alanines in order to see the effects on MHC binding. Notably, changing Asn-27 to Ala resulted in a drop in binding of 200%, an average from six different experiments. No other amino acid when changed to alanine reduced binding significantly. However, a change of Cys-30 to Ala increased binding by 72%.
The crystal structure of I-A k containing the HEL-(48 -62) peptide indicated the interactions of the peptide amino acid residues with the various binding sites on I-A k , as well as the relationship between amino acids that served to anchor the peptide to I-A k with those that stimulated the T cell (34). The central or core sequence occupied the stretch from P1 to P9. An acidic residue at P1, Asp-52, was favored for strong binding as well as for peptide selection. At P4, small hydrophobic residues or asparagines were favored. As with most MHCdisplayed peptides, amino acids that contacted the TCR were positioned at P2, P5, and P9.
Extrapolating this information to peptide-(20 -35), we mapped the P1 to P9 stretch to residues 24 -32 (Tables I and  II). As will be noted below, changing Leu-25 and Trp-28 to Ala did not affect binding but reduced entirely the T cell response; these two residues were TCR contact residues. In residues 24 -32, Leu-25 is at P2, a TCR contact, Asn-27 at P4, the main MHC contact, and Trp-28 at P5, another TCR contact. P1 is occupied by Ser-24, contributing little to binding energy. Fig. 6 shows the results of one experiment in which the binding of HEL-(20 -35) was compared with HEL-(24 -32); there was a drop in binding from 2.1 to 5.6 M. Testing the HEL-(24 -32) peptide, we found that a Ser-24 to Ala change reduced binding, whereas the change of Asn-27 to Ala resulted in a complete lack of interaction.
We also examined the selection of the HEL-(20 -35) family from cell lines expressing HEL in which either the Ser-24 or the Asn-27 had been changed to alanines. The cell lines were tested for their presentation of the chemically dominant peptide-(48 -63) or -(20 -35) (Fig. 7). Very clearly, the presentation of peptide-(48 -63) was not affected by the changes, which was not an unexpected result. However, the changes in either residue 24 or 27 profoundly affected presentation. (Note below, however, that such changes did not affect T cell responses.) Not shown are the results of changing Cys-30 to Ala, which had no effect on presentation.
In summary, these results, shown in Figs. 5-7, indicate the important role of Asn-27 in binding as well as in peptide selection from processing. They also indicate the importance of Ser-24 and, as indicated in Fig. 6, the contribution of residues outside the central core to binding strength. That is to say, peptide-(20 -35) bound better than - (24 -32). Previously, we had established that residues outside the central core, as in the case of peptide-(52-61), contributed to binding strength, as a result of main chain interactions with residues in the binding site of I-A k (8,34).
T Cell Contact Residues of the HEL- (20 -35) Epitope-To identify the TCR contact residues involved in the recognition of the HEL-(20 -35) epitope, we evaluated the response of 11 HEL-(20 -35)-specific T cell hybridomas to Ala-substituted peptides. The T cell recognition of all the T hybridomas was completely abolished when the Trp at position 28 was changed to Ala. A similar response was observed for the Ala-substituted HEL-(20 -35) Ala-25 peptide ( Fig. 8 and Table I). On the basis that the HEL-(20 -35) Ala-25 and Ala-28 peptides bind to I-A k molecule as well as the HEL-(20 -35) peptide, we can state with confidence that residues Leu-25 (P2) and Trp (P5) are obligatory TCR contacts. The Ala substitution of Tyr-23 (PϪ1) had an effect on T cell recognition; 6 of 11 T cell hybridomas did not respond to the HEL-(20 -35) Ala-23 peptide, indicating that residues outside the core peptide (HEL-(24 -32)) can influence the T cell recognition (11). A significant number of T cells (7 of 11) was sensitive to the change of Valϭ29 (P6) to Ala. This change did not reduce the peptide binding to the class II molecule I-A k , suggesting that this amino acid substitution on the anchor residue P6 (Val) can alter the display of the HEL- (20 -35) peptide affecting the T cell recognition. The capacity of anchor residues to influence the T cell responses has been observed previously by other groups and by us (for example, see Refs. 35 and 36). Cys-30 to Ala improved the T cell recognition. The Ala-27 substitution that affected binding had minimal effect in the T cell response; in assays in which the peptide is present for the entire period, there is little effect unless the impairment in binding is quite pronounced. Most of the T hybridomas tested responded better to Ala-substituted HEL-(20 -35) Ala-30 peptide than the HEL- (20 -35). An explanation for this observation is the Ala change at position 30 (P7), which increased the binding of the peptide to I-A k (Fig. 5). COMMENTS We report the isolation and identification of a minor HEL epitope, using an antibody capture method followed by its analysis using ESI-MS. The identification of this HEL peptide family confirmed the usefulness of this method to isolate, concentrate, and analyze peptides displayed in low abundance on APC (3). This peptide would have been difficult to identify with current approaches that directly examine for the peptide by MS combining with biological activity. Indeed, the binding affinity of the peptide makes it difficult to detect from diluted mixtures comprising several molecular species. Some of the critical problems with the peptide isolation from class II molecules stem from the fact that a wide range of peptides with different degrees of affinities, lengths, and compositions binds to the class II molecules. Moreover, these isolated HEL peptides are presented as a group or family with several members, which adds to the difficulty of identifying all of them. The conspicuous observation reported in this study, that the tyrosine sulfation of the HEL- (20 -35) peptides was generated during the peptide isolation procedure, emphasizes the need to establish that posttranslational modification are present on naturally processed peptides isolated from APCs.
We are starting to gain an understanding of the chemical basis for peptide selection by class II MHC molecules during the intracellular processing of the protein (8,37,38) (Table II). This understanding is fundamental in determining the degree of specificity of MHC molecules for peptides and the relationship between the major selected peptides and their binding and sequence motifs (8).
The most abundant peptide families selected by I-A k , both from natural autologous proteins, and HEL, are made of sequences that contain acidic residues that interact with the P1 site (forming a favorable ion pair with an Arg-32 residue from the ␣-chain). The acidic peptide residue is usually situated four or five residues from the first amino-terminal amino acid (17). With HEL, the major peptide family, 48 -63, is selected always in one register (39); this selection of epitope-(48 -63) depends on a vary favorable Asp-52 plus other anchor residues that, without contributing much to binding strength, favor the interaction of the entire peptide.
At present, two epitopes found in much lesser amounts have been identified by us, 20 -35 in this study and 31-45 in the previous reports (3,18), with both having similar features (Table II). The HEL-(20 -35) segment binds with much less strength to I-A k than the dominant -(48 -61); its most favorable residue, contributing both to the selection and binding, is an Asn, mapped at P4. The structural analysis of I-A k molecule made by Fremont's laboratory (34) and the analysis of natural peptides (8) indicated that the P4 pocket could tolerate small hydrophobic residues but also showed a preference for asparagines. The asparagines could be favored by their interactions with a Glu-␤74 residue (34). We establish here that peptides having an Asn that fits into P4, but lacking a strong acidic residue at P1, bind with less strength and are also selected at much lesser amounts than those with the favorable P1 acidic residue. Thus, definite chemical distinctions between strong and weak peptides are emerging as these studies progress. Finally, it is notable that epitope- (20 -35) and -(48 -61) are selected independently of each other, in accordance with previous results with epitope-(31-47) and -(48 -61) (18); the presence of one or the other does not affect the amount expressed. We had found previously that deletion of segment-(48 -63) did not affect the presentations of the other peptides. Here, as shown in Fig. 7, the impairment of the 20 -35 did not affect the presentation of 48 -63.