Essential for Optimal Peptide Loading Antigen Processing-Associated ...

Recruitment of MHC Class I Molecules by Tapasin into the Transporter Associated with Antigen Processing-Associated Complex Is Essential for Optimal Peptide Loading Pamela Tan,* Harald Kropshofer,† Ofer Mandelboim,‡ Nadja Bulbuc,* Gu¨nter J. Ha¨mmerling,* and Frank Momburg1* The ER protein tapasin (Tpn) forms a bridge between MHC class I H chain (HC)/␤2-microglobulin and the TAP peptide transporter. The function of this TAP-associated complex was unclear because it was reported that soluble Tpn that has lost TAP interaction would be fully competent in terms of peptide loading and Ag presentation. We found, however, that only wild-type human Tpn (hTpn), but not three soluble hTpn variants, a transmembrane domain point mutant of hTpn (L4103 F), wild-type mouse Tpn, nor a mouse-human Tpn hybrid, fully up-regulated peptide-dependent Bw4 epitopes when expressed in Tpn-deficient .220.B*4402 cells. Consistent with suboptimal peptide loading, the t1/2 of class I molecules was considerably reduced in the presence of soluble hTpn, hTpn-L410F, and murine Tpn. Furthermore, eluted peptide spectra and the class I-mediated inhibition of NK clones showed distinct differences to the hTpn transfectant. Only wild-type hTpn efficiently recruited HC and calreticulin (Crt) into complexes with TAP and endoplasmic reticulum p57 (ERp57). The L410F mutant was defective in TAP association, but bound to class I molecules, Crt, and ERp57. Mouse Tpn associated with human TAP and ERp57 on the one hand, and with HC and Crt on the other, but failed to recruit normal amounts of HLA class I molecules into the TAP complex. We conclude that the loading with peptides conferring high stability requires the Tpn-mediated introduction of HC into the TAP complex, whereas the mere interaction with Tpn is not sufficient. The Journal of Immunology, 2002, 168: 1950 –1960.

D

estined for presentation to cytotoxic T cells, MHC class I molecules are loaded with peptides that mostly originate from the degradation of cytosolic protein Ags and that are translocated across the endoplasmic reticulum (ER)2 membrane by the TAP peptide transporter (1, 2). During their assembly and maturation in the ER, class I molecules transiently interact with a number of auxiliary molecules, including the calcium-regulated chaperones calnexin (Cnx) and calreticulin (Crt), the thioredoxin family member ERp57(ER60), the ER glycoprotein tapasin (Tpn), and with TAP (3). Early after its biosynthesis into the ER membrane, the class I H chain (HC) binds to the lectin-like chaperone Cnx, followed by association with ␤2-microglobulin (␤2m) (4, 5). Subsequently, the HC/␤2m dimer exchanges Cnx for another lectin-like chaperone, Crt (6, 7). In mouse cells, the interaction of class I HC with Cnx was also reported to persist after binding of ␤2m until late stages of class I maturation in the ER (4, 8, 9). Association of Crt with class I HC is not detected in the absence of ␤2m (6), nor in the absence of Tpn (10 –12).

*Department of Molecular Immunology, German Cancer Research Center (Deutsches Krebsforschungszentrum), Heidelberg, Germany; †Basel Institute for Immunology, Basel, Switzerland; and ‡Department of Immunology, Hebrew University, EinKarem, Jerusalem, Israel Received for publication September 17, 2001. Accepted for publication December 6, 2001. 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 U.S.C. Section 1734 solely to indicate this fact. 1 Address correspondence and reprint requests to Dr. Frank Momburg, Department of Molecular Immunology, Deutsches Krebsforschungszentrum (G0400), Im Neuenheimer Feld 280, 69120 Heidelberg, Germany. E-mail address: [email protected]

Abbreviations used in this paper: ER, endoplasmic reticulum; ␤2m, ␤2-microglobulin; BFA, brefeldin A; Cnx, calnexin; Crt, calreticulin; D-PBS, Dulbecco’s PBS; Endo H, endoglycosidase H; HC, MHC class I H chain; Tpn, tapasin; hTpn, human Tpn; KIR, killer cell Ig-like receptor; m∧hTpn, mouse-human chimera; MFI, mean fluorescence intensity; MS, mass spectrometry; mTpn, mouse Tpn; TM, transmembrane.

2

Copyright © 2002 by The American Association of Immunologists

The thiol oxidoreductase ERp57 interacts with class I HC in the Cnx-associated stage and joins class I/␤2m dimers into the TAPassociated complex (10, 13–15). ERp57 is thought to mediate the formation/isomerization of disulfide bonds and thereby ensure a proper folding of class I HC in conjunction with Cnx and Crt (3, 15). In the absence of ␤2m or Tpn, ERp57 does not associate with HC, Crt, or TAP (10, 11). While an association of Tpn with HC/␤2m has also been observed in TAP1/2-deficient cells (6, 16), the prime function of Tpn is, however, to bridge HC/␤2m, Crt, and ERp57 to the peptide transporter (6, 17–22). Similar to Crt, Tpn requires HC/␤2m dimers for association with class I molecules. In human cells lacking expression of ␤2m, only small amounts of class I HC can be detected in association with TAP (11, 20), whereas in ␤2m-deficient mouse cells, a HC:Crt:Tpn:TAP subcomplex seems to assemble more readily (13). Specific peptide is required to dissociate class I molecules from the Tpn:TAP complex (8, 23, 24) as well as from Tpn in TAP-deficient cells (16). Therefore, a lack of suitable peptides can lead to prolonged association with TAP (9, 25, 26). A large proportion of the total amount of HC/␤2m dimers in the ER is found in complexes with TAP (21, 27). Early studies using the mutant human B cell line .220 have demonstrated a down-regulation of HLA-A and HLA-B products on the cell surface that considerably varied among HLA alloforms (28). Due to aberrant splicing of a mutated Tpn gene, .220 cells express only minute quantities of a nonfunctional, N-terminally truncated Tpn molecule, but no wild-type Tpn (6, 29). In .220 cells, class I HC associates normally with Cnx, whereas only trace amounts of class I HC form a complex with TAP (17, 30). The functional consequences of Tpn deficiency in .220 cells are unstable class I molecules and diminished ER egress due to a partial failure of loading with stabilizing peptides (30, 31) as well as impaired Ag presentation, which could be restored by transfection of wild-type Tpn (18, 32). Equivalent results were obtained with a 0022-1767/02/$02.00

The Journal of Immunology H-2Dd mutant that fails to assemble with Tpn (21). By contrast, wildtype Kb molecules expressed in .220 or in Tpn-deficient insect cells were found to exit the ER more rapidly than in the presence of Tpn (12, 33). For different HLA alleles, a hierarchy was noted in the dependence on Tpn for loading of peptides recognized by CTL (32), and also mouse class I alleles differed in the amounts of peptide-loaded forms that were detectable on the cell surface in the absence of Tpn (34). Thus, from the studies employing .220 transfectants, it can be concluded that Tpn is not absolutely required for the loading of TAPdependent peptides. Recent investigations provided, however, evidence for a quantitative and qualitative shaping of the class I-bound peptide repertoire by Tpn (12, 35, 36). Available evidence suggests that mouse Tpn (mTpn) is able to functionally replace human Tpn (hTpn) in HLA-transfected .220 cells (37, 38). On the other hand, certain human class I alloforms insufficiently assemble with peptides when expressed in presentation-competent mouse cells (32, 38). The structural reasons for this distinct species incompatibility of mTpn remained, however, elusive. The analysis of Tpn-deficient mice clearly demonstrated that Tpn is indispensable for a proper function of the class I Ag presentation pathway (35, 39). Tpn mutant mice show strongly reduced class I surface expression and stability of surface class I molecules. The presentation of cytosolic Ags by Tpn⫺/⫺ cells can be significantly impaired (35). Not only were defects in the development of CD8⫹ T cells and immune responses against some viruses noted, but also an altered NK cell repertoire was observed in Tpn⫺/⫺ mice (35, 39). Deletion of the N-terminal 50 residues ablated a proper interaction of Tpn with class I HC, Crt, and ERp57 (11). Another study suggested the full competence of Tpn lacking the transmembrane (TM) and cytoplasmic domains in terms of rescuing peptide loading and surface expression of class I molecules in .220.B8 cells, although the interaction with TAP was abolished (40). If soluble Tpn could generally replace membrane-anchored Tpn, this result would, however, challenge the physiological significance of the TAP:Tpn complex. In this study, we investigated the function of wild-type hTpn, soluble or point-mutated hTpn, as well as mTpn during the peptide loading of HLA-B*4402 molecules. Only wild-type hTpn was able to facilitate binding of an optimized spectrum of peptides due to distinct failures of all other Tpn variants to mediate assembly of the TAP-associated loading complex.

Materials and Methods Cloning of human and mouse Tpn cDNAs, and generation of mutant Tpn cDNA constructs hTpn cDNA was cloned from a human leukocyte Marathon-Ready cDNA library (Clontech, Heidelberg, Germany) using as 5⬘ primer the Marathon AP1 primer and as 3⬘ primer the Tpn-specific oligonucleotide 5⬘-CAGAG ATGATGGTGGCTTCC-3⬘. The PCR product was cloned into the EcoRI site of pCR2.1 (Invitrogen, Groningen, The Netherlands) and sequenced. The coding sequence differs from the published hTpn sequence AF009510 (18) at codons 161 (GTT instead of GTC) and 260 (ACA instead of AGA). mTpn cDNA was cloned from a BALB/c spleen Marathon-Ready cDNA library (Clontech) using as 5⬘ primer the Marathon AP2 primer and as 3⬘ primer the Tpn-specific oligonucleotide 5⬘-TGGAGGGAGAAGAAGAG AAGAAGG-3⬘. The PCR product was cloned into the EcoRV site of pBluescript II KS⫹ (Stratagene, Amsterdam, The Netherlands) and sequenced. The coding sequence differs from the published murine Tpn (mTpn) sequence AF043943 (37) at codons 17 (GTC instead of TTC), 79 (GTG instead of CTG), 274 (GCT instead of GGT), and 419 (CTA instead of CTG). A chimeric mouse-human Tpn cDNA was constructed by using the unique XhoI site in the mTpn cDNA, which is conserved in the hTpn sequence. For hTpn lacking the C-terminal 33 or 49 residues, StuI or filled-up BstEII sites were ligated with filled-up XbaI (SpeI) polylinker sites, to create stop codons directly after hTpn residues G395 or T379, respectively. For the hTpn-⌬C53KDEL construct, the hTpn Eco47III site was ligated into a vector with an ER retention sequence, creating the C-terminal sequence NSEKDEL* after hTpn residue S375. To generate the hTpn point mutant L410F, an oligonucleotide containing TTC instead of CTG at codon

1951 410 was inserted between the unique StuI and AccI sites. All Tpn constructs were cloned into the episomal expression vector pREP4 (Invitrogen).

Cell lines and transfections The Tpn-deficient, human B-lymphoblastoid mutant cell line LCL .220 (28) was kindly provided by R. DeMars (University of Wisconsin, Madison, WI). HLA-B*4402-transfected .220 cells were granted by C.-A. Peh and J. McCluskey (Melbourne University, Melbourne, Australia). The .220.B*4402 cells were transfected with Tpn cDNAs constructs in pREP4 by electroporation at 220 V, 4 pulses, 4 ms, using a BTX electroporator (BTX, San Diego, CA). Following selection, transfectants were maintained in RPMI 1640 (Life Technologies, Freiburg, Germany) supplemented with 0.3 mg/ml hygromycin B (Roche Diagnostics, Mannheim, Germany), 0.5 mg/ml geneticin (Sigma, Taufkirchen, Germany), 2 mM glutamine, 0.2% glucose, and 10% FCS (Biochrom, Berlin, Germany).

Antibodies Rabbit antisera against N-terminal residues 2–20 (Rb␣PAV) and C-terminal residues 418 – 428 (Rb␣STC) of hTpn were raised against the keyhole limpet hemocyanin-coupled, cysteine-modified peptides PAVIECWFVE DASGKGLAK-C and C-STCKDSKKKAE, respectively, by G. Moldenhauer (DKFZ, Heidelberg, Germany). Rabbit anti-mTpn N-terminal peptide antiserum, Ra2223 (41), was a gift of T. Hansen (Washington University, St. Louis, MO). Mouse mAbs specific for Crt (FMC75), ERp57 (MaP.Erp57), and anti-Crt rabbit antiserum (SPA-600) were from StressGen/Biomol (Hamburg, Germany). Mouse anti-TAP1 mAb 148.3 (42) was provided by R. Tampé (University of Marburg, Marburg, Germany), and rabbit antiserum raised against the cytoplasmic domain of human TAP1 fused to GST (43) was a gift of J. Neefjes (Netherlands Cancer Center, Amsterdam, The Netherlands). The Cnx-specific mAb AF8 (5) was provided by M. Brenner (Harvard Medical School, Boston, MA). mAbs HC10, preferentially reacting with ␤2m-free HLA-B and HLA-C HC (44), and W6/32 recognizing HLA-A, B, C HC/␤2m dimers (45) as well as the ¨ 109 (46) and TT4-A20 (47) have been deHLA-Bw4-specific mAbs TU scribed. The Bw4 mAb T116-5-28 was from Saxon Europe (Newmarket, Suffolk, U.K.).

Immunoprecipitation Cells were lysed in TBS (150 mM Tris-HCl, 50 mM NaCl, 5 mM MgCl2, pH 7.4) supplemented with 1% digitonin (Riedel-de Haen, Seelze, Germany) and Complete protease inhibitor cocktail (Roche) for 2 h at 4°C. After centrifugation at 15,000 ⫻ g to remove nuclei, samples were incubated for 1 h at 4°C with Ab prebound to protein A-Sepharose beads (Amersham Pharmacia, Freiburg, Germany), or protein A/G PLUS agarose beads (Santa Cruz Biotechnology, Heidelberg, Germany) in case of MaP.Erp57. Beads were washed thrice with TBS/0.1% digitonin, followed by heating at 95°C for 5 min with SDS-containing sample buffer and analysis by 10% SDS-PAGE. For the determination of class I HC/␤2m stability in lysates, cells were lysed in TBS containing 1% Nonidet P-40 (Sigma). The postnuclear supernatant was incubated on ice or at 37°C for 1 h before immunoprecipitation with W6/32-conjugated Sepharose 4B beads.

Western blot analysis To determine the expression of transfected Tpn constructs and TAP, cells were lysed in TBS containing 1% Nonidet P-40 for 30 min at 4°C. To analyze the ER egress of Tpn constructs, samples were digested overnight with 10 mU endoglycosidase H (Endo H; Roche) at 37°C in 50 mM NaAc, pH 5.5. Samples were separated on 10% SDS-PAGE gels and transferred onto polyvinylidene difluoride membranes (Millipore, Eschborn, Germany). Membranes were blocked with 1% skim milk powder (Sigma) in PBS with 0.05% Tween 20 (Sigma) and incubated with the primary Ab for 1 h. After extensive washing with PBS, membranes were incubated with goat anti-mouse IgG or anti-rabbit IgG HRP-conjugated second Abs at appropriate dilutions (Dianova, Hamburg, Germany) for 1 h. Following the development with SuperSignal West Dura ECL substrate (Pierce/KMF, Sankt Augustin, Germany), bands were visualized and quantified using the Lumi-Imager system and LumiAnalyst 3.x software (Roche).

Flow cytometry For surface immunofluorescence stainings, 1–2 ⫻ 106 cells were washed once in ice-cold Dulbecco’s PBS (D-PBS)/0.05% NaN3/2% FCS, followed by incubation with a saturating amount of the primary Ab for 30 min at 4°C. After three washes with D-PBS, cells were incubated with goat anti-mouse IgG-FITC (Dianova) for 30 min at 4°C. Cells were washed twice and resuspended in 300 ␮l D-PBS/0.1% propidium iodide (Sigma). FACS analysis was performed using FACScan flow cytometers (BD Biosciences, Heidelberg,

1952

OPTIMIZATION OF PEPTIDE LOADING IN THE TAPASIN-TAP COMPLEX

Germany). To measure the stability of surface class I molecules, cells were cultured for 10 h in the presence of 20 ␮g/ml brefeldin A (Sigma) before staining with W6/32. Loading with the natural HLA-B44 ligands EBNA-6 657– 666 VEITPYKPTW (see “SYFPEITHI” database, http://syfpeithi.bmiheidelberg.com), Hsp90 427– 436 AEDKENYKKF, or the HLA-A11-binding peptide ASYDKAKLK as control was performed at a peptide concentration of 100 ␮M in PBS/0.1% NaN3 for 2 h at room temperature prior to extensive washing to remove excess exogenous peptide and staining with the Bw4¨ 109. Peptides were synthesized by -moc chemspecific mAbs TT4-A20 or TU istry in the peptide synthesis unit of the DKFZ and ⬎95% pure by mass spectrometrical analysis.

Pulse-chase radiolabeling Cells were preincubated for 1 h with either the proteasome inhibitor lactacystin (Calbiochem, Schwalbach, Germany) at a concentration of 100 ␮M or DMSO for control. After washing with warm D-PBS, cells were starved in methionine/cysteine-free RPMI (ICN, Eschwege, Germany) supplemented with 3% dialyzed FCS for 1 h at 37°C. In case of the lactacystintreated probe, fresh lactacystin (100 ␮M) was added after each washing step. Pulse radiolabeling of the cells was performed with 1 mCi [35S]Met/ Cys Pro-mix (Amersham Pharmacia) per 5 ⫻ 106 cells/ml for 10 min. After washing with warm RPMI 1640, cells were chased in complete RPMI 1640/10% FCS. Harvested aliquots were washed with ice-cold D-PBS and lysed in TBS/1% Nonidet P-40. Postnuclear supernatants were subjected to immunoprecipitation with W6/32-conjugated Sepharose beads. Precipitates were analyzed by 10% SDS-PAGE and autoradiography. Class I HC bands were quantified using the Lumi-Imager.

Analysis of class I-associated peptides HLA-B*4402/Cw1-associated self peptides were analyzed by matrix-assisted laser desorption/ionization mass spectrometry (MS), as described (48). In brief, 1–2 ⫻ 107 cells were lysed in 1% n-octylglucoside (Sigma), and HLA molecules were precipitated with W6/32-conjugated Sepharose beads. Beads were transferred into an Ultrafree ultrafiltration tube with a 5-kDa cutoff (Millipore, Volketswil, Switzerland), washed with 10 vol of 20 mM Tris, pH 7.8, 0.1% Zwittergent 3-12 (Calbiochem) and 20 vol of bidestilled water. Peptides were eluted by incubation in 0.1% trifluoroacetic acid in bidestilled water for 30 min, followed by ultrafiltration and lyophilization. Spectra were recorded on a Reflex III mass spectrometer (Bruker Analytik, Rheinstetten, Germany) and collected by averaging the ion signals from 50 –100 individual laser shots.

Isolation of NK clones and cytotoxicity assays NK cell lines and clones were isolated from PBL using the human NK cell isolation kit and the autoMACS instrument (Miltenyi Biotech, Rishon Le Zion, Israel). NK cells were cultured as previously described (49). The cytotoxic activity of NK cells against the various targets was assayed in 5-h 35 S release assays, as described (49). The specific release was calculated as follows: percentage of lysis ⫽ ((cpm experimental well ⫺ cpm spontaneous release)/(cpm maximal lysis ⫺ cpm spontaneous release)) ⫻ 100. For all target cells used, the spontaneous release in medium was 15–16% of the maximal release in 0.1 M NaOH.

Results Generation and expression of Tpn mutants To assess the function of membrane-anchored vs soluble Tpn for the loading of class I molecules with peptides, we generated Cterminal deletion constructs on the basis of a hTpn cDNA clone and expressed them stably in .220.B*4402 cells. In the hTpn-⌬C49 and hTpn-⌬C33 constructs, the mature protein of 428 residues was truncated after aa 379 and 395, respectively, which are located above the TM domain (18). These Tpn truncations are thus expected to be soluble and secreted. In the construct Tpn-⌬C53KDEL, a KDEL ER retention signal (50) was engineered C-terminal of hTpn residue 375. In the point mutant hTpn-L410F, a leucine residue that participates in a putative leucine zipper motif (L396/ L403/L410/L417) in the TM domain was substituted by phenylalanine. In addition, mTpn as well as a mouse-human chimera (m∧hTpn) containing the N-terminal 149 aa of mTpn fused to the corresponding C-terminal portion of hTpn were expressed in .220.B*4402 cells. The Tpn variants used in this study are schematically depicted in Fig. 1A.

FIGURE 1. Expression and ER retention of hTpn, hTpn mutants (⌬C33, ⌬C49, ⌬C53KDEL, L410F), mTpn, and a mouse-human Tpn chimeric molecule (m∧hTpn) in Tpn-deficient .220 cells. Failure of the hTpnL410F TM domain point mutant to stabilize TAP. A, Schematic representation of the wild-type hTpn (18), mTpn (37), and mutant Tpn variants used in this study. B, Nonidet P-40 lysates from 106 .220.B*4402 cells transfected with the indicated Tpn constructs were separated on 10% SDSPAGE gels, and Western blots were probed with hTpn N- or C-terminal peptide antisera (Rb␣PAV, Rb␣STC) or with the anti-mTpn serum Ra2223. C, To monitor the ER retention of Tpn variants, lysates of 106 Tpn-transfected .220.B*4402 cells were subjected to digestion with Endo H overnight (⫹ Endo H) or mock treated (⫺ Endo H). Immunoblots were probed with the Rb␣PAV antiserum. Endo H-sensitive forms display a lower apparent m.w. D, Lysates from 10, 5, and 2.5 ⫻ 106 .220.B*4402 cells transfected with the indicated Tpn forms were separated by 10% SDS-PAGE and electrotransferred. Blots were stained anti-TAP1 mAb 148.3, or a Cnx-specific antiserum as titration control, respectively.

The Journal of Immunology The expression of transfected wild-type hTpn, Tpn truncations, and point mutants was monitored on immunoblots using peptide antisera generated against hTpn sequences 2–20 (Rb␣PAV) or 418 – 428 (Rb␣STC), respectively. These blots demonstrated similar levels of expression of the constructs (Fig. 1B). We noted that the Rb␣PAV serum recognized mTpn somewhat less efficiently than hTpn, probably because four residues in mTpn 2–20 vary from the respective hTpn sequence. Therefore, an immunoblot of the same number of .220.B*4402/mTpn cells with the genuine anti-mTpn (1–20) serum Ra2223 (41) is shown in addition. Lysates of selected Tpn transfectants were subjected to Endo H treatment to monitor ER retention (Fig. 1C). Wild-type hTpn and mTpn molecules were fully Endo H sensitive, consistent with complete ER retention. Unexpectedly, a fraction of Tpn-L410F molecules reproducibly displayed Endo H resistance. This suggests that, in addition to the canonical KKAE ER retention sequence at the C terminus of Tpn (18, 50), a signal within the TM domain may play a role in its ER retention. Soluble Tpn-⌬C33 molecules in the total cellular lysate showed partial Endo H resistance, indicating release into the secretory pathway. The KDEL-tagged soluble Tpn construct remained, however, fully Endo H sensitive, proving that the KDEL retention signal was functional. Tpn TM point mutant pinpoints an interaction site with TAP Next, we assessed the capacity of hTpn, mTpn as well as hTpnL410F, mTpn to stabilize steady state levels of the TAP peptide transporter (11, 40). Immunoblots of whole cell lysates of equal numbers of .220.B*4402 transfectants were probed with the TAP1-specific mAb 148.3, or anti-Cnx as a loading control. In cells transfected with mTpn (Fig. 1D, see also Fig. 5B) and m∧hTpn (data not shown), we detected 80 –100% of the TAP levels present in hTpn transfectants, whereas in the presence of the Tpn-L410F mutant only ⬃10% of the normal TAP1 amounts were detected (Fig. 1D). Thus, the point mutation apparently impaired an interaction of Tpn and TAP located in the membrane-spanning domain. As shown in Fig. 5B (anti-TAP1 blot), hTpn-L410F transfectants retained a residual capacity to stabilize TAP as compared with Tpn-deficient .220 cells. hTpn mutants and mTpn entail a reduced formation of Bw4 determinants The surface expression of class I molecules on .220.B*4402 Tpn transfectants was analyzed using mAb W6/32 recognizing a monomorphic HC/␤2m combinatorial determinant on all HLA-A, B, C products, including the transfected B*4402 and the endogenously expressed Cw1 molecules (28, 32, 45, 51). The level of HLA-B*4402/ Cw1 expression in the absence of functional Tpn is indicated by the stainings of .220.B*4402/pREP4 control transfectants. This staining is mostly due to B*4402 molecules since .220/pREP4 or .220 cells were significantly less strongly labeled with W6/32 (data not shown). On .220.B*4402 cells transfected with hTpn-⌬C49, hTpn-⌬C33 (Fig. 2A), hTpn-⌬C53KDEL (Fig. 2C), hTpn-L410F, mTpn, and m∧hTpn (Fig. 2B), the labeling with W6/32 was increased to levels that were similar to .220.B*4402/hTpn cells, suggesting that all Tpn constructs were equally competent to promote the formation of stable, presumably peptide-loaded HC/␤2m dimers that migrate to the cell surface. Apparently, the low amounts of TAP that were detected in the presence of hTpn-⌬C49, hTpn-⌬C33 (data not shown), and hTpn-L410F (Fig. 1D) were not limiting during the assembly of enhanced amounts of W6/32-reactive HC/␤2m dimers. Next, we examined the cell surface expression of B*4402 molecules using three Abs reactive with Bw4 public determinants that are controlled by the polymorphic residues 77 and 80 – 83 in the HC ␣1 domain (52). The Bw4 epitope recognized ¨ 109 has been reported to be sensitive to the peptide by mAb TU

1953 cargo of HLA-B*5101 molecules (53). Except for a slightly weaker staining of .220.B*4402/hTpn-⌬C49 cells, hTpn mutants and mTpn strongly induced the Bw4 epitope recognized by T116-5-28 to levels that equalled or even exceeded wild-type Tpn (Fig. 2, A and B), thus confirming the results obtained with ¨ 109 reW6/32. In clear contrast, stainings with TT4-A20 and TU vealed distinct defects of soluble hTpn constructs to generate Bw4 determinants (Fig. 2A). This failure could not be rescued by ER ¨ 109 retention of soluble Tpn, as indicated by the strongly reduced TU staining of the hTpn-⌬C53KDEL transfectant (Fig. 2C). Also, the hTpn-L410F point mutant, mTpn, and m∧hTpn showed pro¨ 109 nouncedly reduced stainings with TT4-A20 (Fig. 2B) and TU (data not shown). From the observed defect of the mouse-human Tpn hydrid, we conclude that the N-terminal 149 residues of mTpn are responsible for the failure of mTpn to produce wild-type amounts of these Bw4 epitopes. Lower stability of class I molecules in the presence of Tpn mutants and mTpn Using biochemical assays, we analyzed the stability of class I molecules in lysates and on the cell surface, which is a measure for the stabilizing properties of the spectrum of loaded peptides. First, we determined the thermostability of W6/32-precipitable class I molecules in lysates (51). Quantification of Western blots stained with HC-10 revealed that 75% of HC/␤2m dimers from .220.B*4402/ hTpn cells survived an incubation at 37°C for 1 h (Fig. 3A). In contrast, only 30% and 40% of HC/␤2m dimers in lysates of .220.B*4402/hTpn-⌬C49 and .220.B*4402/mTpn cells, respectively, survived this treatment, suggesting suboptimal stabilization by endogenous peptides in these cases. In another approach, various Tpn transfectants were treated with brefeldin A (BFA) that inhibits the egress of newly synthesized class I molecules from the ER (54). Mean fluorescence intensities of TT4-A20 stainings following a 10-h incubation in the presence or absence of BFA were compared (Fig. 3B). BFA treatment reduced the mean fluorescence intensity (MFI) of surface-stained class I molecules on .220.B*4402/ hTpn cells by 27%. In contrast, on hTpn-⌬C49, hTpn-⌬C33, mTpn, and hTpn-L410F transfectants, the MFI was reduced by 38 – 41%, pointing to a lower average stability and survival of surface B*4402 molecules. This finding was further substantiated in a pulse-chase experiment with metabolically labeled Tpn transfectants (Fig. 3C). Quantification of class I molecules that were precipitable with W6/32 from Nonidet P-40 lysates after chase periods up to 18 h showed considerable differences in their t1/2. While metabolically labeled class I molecules had an approximate t1/2 of 20 h in .220.B*4402/Tpn cells, the t1/2 were reduced to 4 – 8 h in .220.B*4402/mTpn, m∧hTpn, hTpn-⌬C49, and hTpn-L410F cells. Class I molecules in Tpn-deficient .220.B*4402/pREP4 cells showed the lowest stability with an approximate t1/2 of 2 h. Peptide- and Tpn-dependent ER egress of class I molecules A more rapid ER egress of HLA-B8 molecules in the presence of soluble Tpn was reported previously (40). Since premature ER release might be accompanied by an incomplete loading of class I molecules with peptides, we examined whether in the presence of Tpn mutants HLA-B*4402 and -Cw1 molecules were more readily allowed to leave the ER without peptide (or loaded with suboptimal peptide) than in the presence of wild-type Tpn. To this end, we analyzed the acquisition of Endo H resistance of class I HC, being indicative of transport to and beyond the Golgi apparatus, in the presence of the proteasome-specific inhibitor lactacystin, which is known to cause a

1954

OPTIMIZATION OF PEPTIDE LOADING IN THE TAPASIN-TAP COMPLEX

FIGURE 2. Surface expression of class I molecules in Tpn transfectants. A, FACS stainings of .220.B*4402 cells transfected with wild-type hTpn, soluble hTpn-⌬C33 or -⌬C49, or the empty pREP4 vector, with the pan-HLA-A, B, C class I mAb W6/32 and with the HLA-Bw4 supertype-reactive mAbs ¨ 109, TT4-A20, and T116-5-28. B, Stainings of .220.B*4402 cells transfected with wild-type hTpn, mTpn, the m∧hTpn chimera, or the TM domain point TU mutant hTpn-L410F. C, Stainings of .220.B*4402 cells transfected with wild-type hTpn or the ER-retained soluble hTpn-⌬C53KDEL mutant. Control stainings with second Ab only are indicated by dashed lines. Stainings in each of the columns were performed in the same experiment. Vertical lines denote the MFI of the pREP4 and the hTpn transfectants, respectively. Variations in these MFI can be accounted for by the use of cytofluorometers with different instrument settings.

reduced class I maturation due to limited supply with cytosolic peptides (55). In the absence of lactacystin, we noted similar kinetics in the formation of Endo H-resistant class I molecules expressed in .220.B*4402/hTpn, hTpn-⌬C49, and mTpn cells (Fig. 3D, left). The maturation of class I in .220.B*4402/pREP4 cells was significantly slower. In the presence of lactacystin, the overall amounts of W6/32precipitable HC/␤2m dimers and their maturation kinetics were concomitantly reduced in all transfectants (Fig. 3D, right). Under conditions of limited peptide supply, the lower stability of HC/␤2m dimers in ⌬C49-, mTpn-, and particularly in pREP4-transfected cells was apparent already after 120 min of chase, as indicated by decreased amounts of W6/32-precipitated material. Thus, this experiment provided no indication for an increased peptide-independent or more

rapid ER egress of the studied class I alloforms in the presence of mTpn or hTpn mutants. Altered class I-associated peptide repertoire in the presence of mTpn and hTpn mutants To investigate whether differential peptide loading in the presence of soluble hTpn mutants leads to the defects in the formation of Bw4 determinants, we incubated .220.B*4402 cells expressing wild-type hTpn, hTpn-⌬C49, or hTpn-⌬C33 with known B44binding peptides before surface labeling with either TT4-A20 or ¨ 109. Incubation with exogenous B44 ligands significantly inTU creased the MFI of TT4-A20 staining of hTpn-⌬C49 cells as compared with cells incubated with an irrelevant A11-binding peptide

The Journal of Immunology

FIGURE 3. Tpn mutants confer lower stability and shorter t1/2 to class I HC/␤2m dimers, but do not entail peptide-independent ER egress. A, The thermostability of HC/␤2m in lysates was analyzed by immunoprecipitation with W6/32 from Nonidet P-40 lysates from 2 ⫻ 106 .220.B*4402/ pREP4, hTpn, ⌬C49, or mTpn cells following incubation for 1 h at 4°C or 37°C. B, After culturing the indicated .220.B*4402 Tpn transfectants for 10 h in the presence or absence of 20 ␮g/ml BFA, surface staining with W6/32 was performed. Results are presented as percentage of reduction of the MFI in the presence of BFA compared with the control. SEM of two independent experiments are shown. C, .220.B*4402 cells transfected with the indicated Tpn variants were metabolically labeled for 10 min and chased for 1, 4, 7, or 18 h. Following lysis in 1% Nonidet P-40, HLA class I/␤2m dimers were immunoprecipitated with W6/32. Autoradiograms were quantified using a Lumi-Imager. D, Maturation of class I/␤2m dimers was analyzed for the indicated .220.B*4402 Tpn transfectants following treatment with 100 ␮M lactacystin (2 h, 37°C) to deplete for proteasomally produced peptides, or DMSO for control. Following metabolical labeling

1955 (Fig. 4A, left). The B44-binding peptides were also able to augment ¨ 109 of both hTpn-⌬C49 and hTpn-⌬C33 transfecthe staining by TU tants (Fig. 4A, right). These findings indicate that the epitopes recog¨ 109 mAbs are influenced by HLAnized by TT4-A20 and TU B*4402-bound peptides, and suggest that B*4402 molecules expressed in the presence of soluble Tpn constructs contain altered peptide spectra because they appear to be more receptive for exogenous peptides than B*4402 in the presence of wild-type Tpn. Previous studies indicated that the recognition of HLA-C and Bw4 alloforms by NK cell inhibitory receptors of the killer cell Ig-like receptor (KIR) family is influenced by the class I-bound peptide with particular importance of charged amino acid side chains at position 8 in 9-mer ligands (56, 57). To provide further evidence for differential peptide loading of class I molecules in the presence of hTpn vs mutant hTpn or mTpn, we analyzed the capacity of our transfectants to inhibit killing by 64 freshly isolated human NK clones, of which representative examples are shown in Fig. 4B. For comparison, class I-deficient .221 cells as well as .221/Cw3 and .221/Cw4 transfectants were included. Some NK clones were inhibited by .221/Cw3, but not by .221/Cw4 cells, suggesting that they employed KIR-2DL2 or KIR-2DL3 receptors (58), which should also be inhibitable by HLA-Cw1 molecules expressed on the .220 cells. Other NK clones that were inhibited by .220.B*4402/Tpn transfectants, but not by .221/Cw3 or .221/ Cw4, probably engaged the Bw4-reactive KIR-3DL1 receptor (58). The pREP4 control transfectant was efficiently killed by essentially all NK clones consistent with the comparably low surface staining with W6/32. Most NK clones were equally well inhibited by .220.B*4402 cells transfected with hTpn, hTpn-⌬C33, hTpn⌬C49, or mTpn. A number of clones displayed, however, clear differences in their inhibition by Tpn transfectants. Clones 13 and 14 are examples for clones that were more efficiently inhibited by transfectants expressing ⌬C33/⌬C49 (and mTpn) than by the hTpn transfectant. On the other hand, clones 31 and 49 were not significantly inhibited by the mTpn transfectant, but were blocked by the transfectants expressing full-length or soluble hTpn. These results suggest that certain NK clones were sensitive to specific conformational changes in HLA-peptide complexes expressed in the presence of the different Tpn variants. In the light of similar total amounts of surface class I molecules (see Fig. 2A), this is consistent with the concept that peptide loading of B*4402 and Cw1 molecules is differentially modulated by the Tpn constructs investigated in this study. To gain direct insight into the composition of class I-associated peptide spectra, peptides were acid eluted from immunopurified class I molecules and analyzed by MS (Fig. 4C). On immunoblots probed with HC-10, we determined the total amount of HLA-Cw1 in .220 cells to be ⬍5% of HLA-B*4402 and Cw1 molecules expressed in .220.B*4402 cells (data not shown). Thus, the vast majority of eluted peptides were derived from B*4402 molecules. Peptides from .220.B*4402/hTpn cells showed a broad distribution in the range of 950-1350 Da with a relative abundance of masses between 1100 and 1200 Da. In the spectrum derived from .220.B*4402/ mTpn cells, peptides in the range of 950-1050 Da were more prominent than in the profile from hTpn-expressing cells, whereas peptide species at about 1150 Da were less abundant. Independent repeats

for 10 min, cells were chased for the indicated periods. W6/32 immunoprecipitates from Nonidet P-40 lysates were digested with 10 mU Endo H overnight and separated by 10% SDS-PAGE. Endo H-resistant (EHR) and -sensitive (EHS) class I form were quantified and are presented as EHR/EHS ratio. 䡺, .220.B*4402/pREP4; f, .220.B*4402/hTpn; ⽧, .220.B*4402/hTpn-⌬C49; E, .220.B*4402/mTpn.

1956

OPTIMIZATION OF PEPTIDE LOADING IN THE TAPASIN-TAP COMPLEX

FIGURE 4. Different peptide spectra are loaded in the presence of hTpn vs hTpn mutants or mTpn. A, 2 ⫻ 106 cells per sample of .220.B*4402/hTpn, .220.B*4402/hTpn-⌬C33, and .220.B*4402/hTpn-⌬C49 cells were incubated for 2 h in the presence of the HLA-B44 ligands VEITPYKPTW (V10W) or ¨ 109. The MFI AEDKENYKKF (A10F), or an irrelevant A11-binding peptide for control. Cells were stained with the Bw4-specific mAbs TT4-A20 and TU obtained with V10W and A10F is given as ratio to the MFI with the control peptide. B, Freshly isolated NK clones (3 ⫻ 105 cells/well) were used in a 5-h 35S release assay with 1 ⫻ 105 .220.B*4402 Tpn transfectants as target cells (E:T ratio ⫽ 3:1). For control, .221, .221/Cw3, and .221/Cw4 cells were run in parallel. For all target cells used, the spontaneous release was 15–16%. C, Matrix-assisted laser desorption/ionization MS profiles of self peptides eluted from HLA class molecules that were immunopurified with W6/32 from 2 ⫻ 107 .220.B*4402/hTpn, .220.B*4402/mTpn, and .220.B*4402/hTpnL410F cells, respectively. As a control, Tris-conjugated Sepharose beads were acid eluted (mock eluate). Mass ranges containing differentially expressed peptide peaks are marked by boxes.

confirmed the differences between these spectra. In another MS analysis, the peptide profile also produced from .220.B*4402/⌬C49 cells showed a high frequency of peptides in the 950- to 1050-Da range (data not shown). We noted a few prominent peaks of relatively high masses that were underrepresented in the mTpn profile and, on the other hand, a few peptide species of relatively low masses that were underrepresented in the hTpn profile. In the peptide profile eluted from .220.B*4402/hTpn-L410F cells, some prominent peaks of the hTpn profile were missing, while the overall distribution of masses was quite similar. Proper assembly of the human TAP-associated loading complex requires wild-type hTpn To investigate the molecular basis for the defective formation of Bw4 epitopes, we analyzed complexes containing ER components of the class I pathway by coprecipitation experiments that are summarized in Table I. In Fig. 5A, Crt and Crt-associated molecules were immunoprecipitated from digitonin lysates of equal amounts of .220.B*4402 transfectants expressing hTpn, mTpn, hTpn⌬C49, hTpn-⌬C33, hTpn-⌬C53KDEL, or hTpn-L410F, respectively. Western blots of SDS-PAGE-separated precipitates were

probed with Abs specific for Crt, TAP1, ERp57, class I HC, and Tpn. The Crt immunoblot showed similar amounts of Crt, proving that precipitations from all cell lines had been equally efficient. Striking differences were visible for the Crt-associated proteins. Only in the presence of hTpn, all components of the human TAPassociated complex could efficiently be coprecipitated with Crt. Little TAP1 and ERp57, but normal or only slightly reduced amounts of HC and Tpn were associated with Crt in the presence of hTpn-L410F. Also in mTpn transfectants, Crt:HC and Crt:Tpn complexes were readily detectable. However, the quantities of TAP1 coprecipitated through Crt were clearly diminished, and Crt: ERp57 complexes were undetectable in this and repeating experiments. In cells expressing soluble hTpn variants, Crt:HC complexes were strongly reduced or absent, and no Tpn, TAP1, and ERp57 were detected in Crt immunoprecipitates. Thus, defective peptide loading in the presence of hTpn mutants and mTpn seems to be correlated with distinct failures during the formation of a complete HC:Crt:Tpn:ERp57:TAP complex. Consistent with the Crt coprecipitation, coimmunoprecipitations via anti-TAP1 Abs revealed an efficient association of TAP with Tpn,

The Journal of Immunology

1957

Table I. Association of components of the class I ER processing machinery in the presence of wild-type Tpn and Tpn mutants Tpn Molecule Expressed in .220.B*4402 Cells Coprecipitated Components

Crt3HC TAP3HC ERp573HC Crt3Tpn TAP3Tpn ERp573Tpn TAP3Crt ERp573Crt TAP3ERp57 TAP3Cnx a b c

hTpn

mTpn

hTpn-⌬C33

hTpn-L410F

Fig.

⫹ ⫹⫹ ⫹⫹ ⫹ ⫹ ⫹⫹ ⫹⫹ ⫹⫹ ⫹⫹ (⫹)

⫹ (⫹) (⫹) ⫹ ⫹ ⫹⫹ (⫹) (⫹) ⫹⫹ ⫹

(⫹) ND ⫺ ⫺ ND ⫺ ⫺b (⫹) ⫺c ND

⫹ (⫹) ⫹ ⫹ (⫹) ⫹ (⫹) ⫹ (⫹) ND

5A 5B 5C 5A 5B 5C 5B 5C 5B 5D

a

Strength of signals in immunoblots of Fig. 5, A–D; ⫹⫹, very strong to strong; ⫹, intermediate; (⫹), weak; ⫺, negative. Fig. 5A. Fig. 5C.

ERp57, Crt, and HC only in lysates of .220.B*4402/hTpn cells (Fig. 5B). Low amounts of these components could be coisolated with TAP1 in the presence of hTpn-L410F, which is probably due to the

reduced TAP levels present in this transfectant (compare Fig. 1D). In digitonin lysates of .220.B*4402/mTpn cells, normal amounts of Tpn and ERp57 were found to be associated with TAP, whereas the

FIGURE 5. Only full-length hTpn mediates assembly of the TAP-associated peptide-loading complex in large amounts. A, Crt-associated components were analyzed by immunoprecipitation of 2 ⫻ 106 digitonin-lysed .220.B*4402 Tpn transfectants using a Crt antiserum. Immunoblots of SDS-PAGEseparated proteins were probed with mAbs specific for Crt (FMC75), ERp57 (MaP.Erp57), TAP1 (148.3), and HLA-B, C HC (HC-10), respectively. The anti-mTpn serum Ra2223 and anti-hTpn serum Rb␣PAV were used to detect mTpn and hTpn proteins, respectively. B, TAP-associated proteins from the indicated .220.B*4402 transfectants were analyzed by immunoprecipitation from 1% digitonin lysates of 2 ⫻ 106 cells with a rabbit antiserum against the cytoplasmic domain of human TAP1 and immunoblot using the same Abs as in A, except for the Tpn blot that was probed with biotinylated Rb␣PAV serum. C, ERp57-associated proteins were examined by precipitation from 2 ⫻ 106 digitonin-lysed .220.B*4402 transfectants with MaP.ERp57, followed by immunoblot using anti-TAP1 mAb 148.3, rabbit anti-Crt, Ra2223 or Rb␣PAV anti-Tpn antisera, and HC-10, respectively. D, TAP-associated Cnx and class I HC in .220.B*4402/hTpn and mTpn cells were analyzed by immunoprecipitation with TAP1-C antiserum from digitonin lysates of 3 ⫻ 106 cells. Immunoblots were probed with the Cnx-specific mAb AF8 and with HC-10, respectively.

1958

OPTIMIZATION OF PEPTIDE LOADING IN THE TAPASIN-TAP COMPLEX

complex formation of TAP with Crt and HC was significantly disturbed. Thus, although mTpn recruited ERp57 into the TAP complex and independently participated in a Tpn:Crt:HC complex (see Fig. 5A), it apparently did not properly bridge HC and Crt to TAP. In good agreement with these results, the coprecipitation via antiERp57 Abs showed that in .220.B*4402/hTpn or mTpn transfectants, considerable amounts of TAP1 were associated with ERp57 (Fig. 5C). In the presence of hTpn-⌬C33, hTpn-L410F, or in the vector control transfectant, ERp57:TAP complexes were not formed or below the threshold of detection. The assembly of ERp57 with Crt was clearly reduced in the presence of mTpn, as compared with hTpn, and barely detectable in ⌬C33 and pREP4 control transfectants. In the ERp57 precipitate from .220.B*4402/hTpn-L410F cells, only a slight reduction of coisolated Crt was observed. The fact that Crt was more readily coprecipitated with anti-ERp57 than vice versa (Fig. 5A) may be due to differences in the relative portions of these ER housekeeping proteins that are engaged with the respective partner under steady state conditions. Similar, or in other experiments even increased amounts of mTpn coprecipitated with ERp57 as compared with hTpn, whereas the interaction of hTpn-L410F with ERp57 appeared slightly reduced (Fig. 5C). Only small quantities of HC could be coisolated with ERp57 from .220.B*4402/mTpn cells. These findings would be consistent with the hypothesis that mTpn fails to recruit normal amounts of HC and Crt into a properly assembled Tpn:ERp57:TAP complex. In hTpn-L410F cells, the amount of HC associated with ERp57 appeared only slightly reduced as compared with wild-type hTpn cells. As no concomitant association with TAP was detected, this result suggests that, independently of TAP, ERp57:HC(:hTpnL410F:Crt) complexes stably assembled. HC and TAP were, however, not detectably associated with ERp57 in .220.B*4402/pREP4 and .220.B*4402/hTpn-⌬C33 cells, clearly showing that membraneanchored Tpn is required to mediate assembly of ERp57 with HC, and additionally, that an intact TM region of Tpn is required to associate this complex with TAP. Recently, it was reported that in human cells a Tpn:TAP subcomplex is first associated with Cnx, which is then released upon binding of HC and Crt (22). We asked whether a prevailing association of Cnx with the TAP complex may be correlated with the apparent failure of mTpn to bridge HC and Crt to TAP. As shown in Fig. 5D, TAP1 immunoprecipitates of .220.B*4402/mTpn and hTpn cells indeed revealed that larger amounts of Cnx are TAP associated under steady state conditions in mTpn- as compared with hTpn-expressing transfectants. Thus, a partial inability of mTpn to facilitate the displacement of Cnx from human TAP:Tpn complexes may contribute to the defective adaptor function.

Discussion In the present study, we describe the defective function of hTpn mutants as well as wild-type mTpn in the loading of HLA-B*4402 molecules with peptides that confer high stability and contribute to the formation of the Bw4 supertypic determinant. Peptide presentation by this class I allele was previously shown to critically depend on Tpn (32). In the absence of Tpn, B*4402 molecules are poorly expressed on the cell surface (32, 38; this study). The three soluble hTpn-⌬C mutants examined in this study up-regulated W6/32 surface staining to the same extent that was observed with full-length hTpn, suggesting that soluble Tpn variants actively supported peptide loading and ER egress of HLA-B*4402 (and -Cw1) HC-␤2m dimers (Fig. 2). In keeping with published work regarding .220.B8 cells (40), we found that our hTpn-⌬C mutants also up-regulated W6/32 staining in .220.B8 as well as in .220 cells (P. Tan and F. Momburg, unpublished results). We did not observe, however, faster maturation kinetics of class I molecules in the presence of

soluble Tpn. Soluble Tpn could not be coprecipitated from digitonin lysates using Crt Abs (Fig. 5A) or ␤2m Abs (unpublished result). It can thus be concluded that the interaction of soluble Tpn with HC is weak and probably short-lived, even if soluble Tpn was retained in the ER by means of a KDEL sequence. Nevertheless, soluble hTpn variants retained an albeit reduced capacity to facilitate the formation of precipitable HC:Crt complexes that, however, did not contain detectable amounts of ERp57 (Fig. 5, A and C). The quantities of coprecipitated HC decreased with the length of the C-terminal truncation, being the largest with the hTpn-⌬C33 construct that contains the entire connecting sequence between the membrane-proximal Ig-like domain and the TM domain (59). This suggests that the connecting sequence of Tpn may directly or indirectly contribute to the recruitment of Crt to HC that was previously shown to require the N-terminal 50 residues of Tpn (11). Through an unknown mechanism, the type I glycoprotein Tpn increases steady state levels of the heterodimeric TAP peptide transporter, and thereby increases quantitatively the peptide transport capacity of TAP (11, 40). At present, it can only be speculated whether this stabilization is due to the permanent presence of up to four Tpn molecules (18) per TAP1/TAP2 heterodimer, or whether Tpn facilitates the assembly of newly synthesized TAP1 and TAP2 subunits in a chaperone-like fashion. The novel TM point mutant L410F has essentially lost the capacity of wild-type Tpn to elevate TAP1 steady state levels. Since this mutation interrupts a putative leucine-zipper motif in the hTpn TM, such a motif may play a role in the interaction between the TM of hTpn and TMs to be identified within the multimembrane-spanning TAP1 and TAP2 molecules. The partial suspension of ER retention of the L410F mutant may have contributed to the failing stabilization of TAP. The formation of complexes between Crt, hTpn-L410F, and HC (Fig. 5A) suggests, however, that the TM mutation did not significantly affect the interaction of Tpn with HC. Findings that soluble class I molecules can bind to the TAP complex (8, 9) support the view that an interaction of the TM domain of HC with that of Tpn may not be critical for the assembly of HC and Tpn. Human .220.B*4402 cells transfected with mTpn exhibited another kind of assembly defect. Also, mTpn fully reconstituted the class I surface expression measured by the pan-HLA Ab W6/32 (Fig. 2). Furthermore, it enabled the formation of subcomplexes containing Crt together with HC and Tpn (Fig. 5A). We show in this study that ERp57 very efficiently associated with TAP:mTpn complexes independent of the poor integration of HC and Crt into this complex (Fig. 5B). This is in good agreement with the finding that class I-independent ERp57:Tpn:TAP complexes can assemble in ␤2m-deficient Daudi cells (22). We noted a strong interaction of mTpn with human TAP, resulting in the full up-regulation of TAP steady state levels. To our surprise, this association of mTpn with hTAP did not result in a similarly productive accumulation of HC and Crt in the TAP complex (Fig. 5, B and D). In mTpn-transfected cells, we detected elevated amounts of Cnx associated with TAP as compared with hTpn transfectants. Cnx was recently shown to be part of a precursor TAP:Tpn:ERp57 complex, preceding the entry of Crt and HC into the human TAP complex and the coincident release of Cnx (22). Alternatively, TAP:Cnx complexes may in part represent nonfunctional dead-end products. Whereas earlier work has noted the concomitant presence of Cnx and HC in murine TAP/Tpn complexes (9, 21), a recent study of mouse cells demonstrated that the Tpn-dependent assembly of HC with ERp57, Tpn, and TAP was correlated with the association of Crt rather than that of Cnx (60). These findings (22, 60) support a model in which the association of Cnx and of Crt:HC with TAP is mutually exclusive. Therefore, it seems likely that TAP:Cnx complexes and TAP:HC complexes observed in TAP immunoprecipitates (Fig. 5D)

The Journal of Immunology belonged to separate TAP subpopulations, and second, that the strongly reduced TAP association of B*4402/Crt in the presence of mTpn was due to an abnormally low dissociation rate of human Cnx, possibly caused by a partial inability of mTpn to release human Cnx. The poor recruitment of B*4402 into the human TAP complex by mTpn is in contrast with a prolonged association of B*4402 with the murine TAP complex that was recently described for B*4402-transfected mouse J26 cells (38). This prolonged TAP association and the diminished surface expression of B*4402 could be normalized by introduction of hTpn (38). In addition, our own preliminary data on Kd- or Kk-transduced .220 transfectants suggest that mTpn is superior to hTpn in facilitating Kd- or Kk-restricted Ag presentation by this human cell line (M. Lobigs, J. C. S. M. R., A. N. U., Canberra, and F. Momburg, unpublished results). Thus, not only species-specific constraints in the interaction between Tpn and ER chaperones, but also between Tpn and HC may influence the assembly of the TAP complex and successful peptide loading. This partial species barrier should be kept in mind when analyzing the efficiency of Ag presentation by HLAtransgenic mouse cells, or conversely, by H-2-transfected human cells. The three soluble hTpn variants, the hTpn TM point mutant, and mTpn failed to assemble a complete TAP-associated complex for different reasons. Nevertheless, the functional outcome was similar, as none of these Tpn forms imparted a full-fledged induction ¨ 109 as of the Bw4 epitopes detected by mAbs TT4-A20 or TU compared with wild-type hTpn. In accordance with Peh and colleagues (38), a defect to build up the Bw4 supertypic determinant was, however, not detected using mAb T116-5-28. This suggests that in the presence of wild-type hTpn, subtle conformational changes are induced in class I HC ␣1 helix around residues 80 – 83, which appear to be recognized by some Bw4-specific Abs, but not by others (52, 61). Similar observations were made when .220.B8 cells transfected with wild-type or soluble hTpn variants were examined with a panel of Bw6-reactive Abs (P. Tan and F. Momburg, unpublished results). Analysis with TT4-A20 revealed that the failure of mTpn to elicit the full expression of the Bw4 determinant is a function of the N-terminal 149 mTpn residues comprising two of the four predicted ectodomains of Tpn (18, 37, 59). Since 35 aa differ between hTpn and the m∧hTpn hybrid, additional studies are required to exactly delineate those mTpn residues that confer the incompatibility with the human class I assembly machinery. Clearly, putative conformational changes depicted by Bw4 reagents in the presence of hTpn mutants or mTpn escaped detection when labeling with the monomorphic class I Ab W6/32 that requires HC/␤2m heterodimers, but is less sensitive to the peptide cargo (51). This explains the previous conclusion that soluble Tpn may be fully functional (40). The serological findings with cells expressing hTpn mutants or mTpn were complemented by biochemical assays that demonstrated the inferior thermostability of class I molecules in lysates, the reduced stability of surface class I molecules in the presence of BFA, as well as the reduced t1/2 of cohorts of metabolically labeled class I molecules. Consistently, an inferior thermostability of W6/ 32-precipitated HLA-B8 molecules was noted in .220.B8 cells transfected with soluble hTpn constructs (P. Tan and F. Momburg, unpublished results). These findings all pointed to qualitative differences in the peptide cargo loaded onto class I molecules. This view is supported by the result that B*4402-binding peptides were ¨ 109 and TT4-A20 on able to enhance surface staining by TU .220.B*4402/hTpn-⌬C33 or ⌬C49, but not on .220.B*4402/hTpn cells (Fig. 4A), suggesting that peptide-dependent HLA-Bw4 epitopes were incompletely formed in the presence of soluble Tpn. This is ¨ 109 epitope was consistent with an earlier study showing that the TU

1959 strongly influenced by residues at position 8 of HLA-B*5101-binding 9-mer peptides (53). The situation appears similar to H-2Ld and other mouse class I molecules displaying more open, peptide-receptive forms on the surface of .220 cells as compared with Tpn-positive .221 cells (34). Furthermore, it has been shown that negatively or positively charged residues at peptide position 8 interfere with the recognition by Bw4-specific KIR-3DL1 inhibitory NK receptors (56). Our result that certain NK clones can be differentially inhibited by hTpn transfectants on the one hand, and mTpn or ⌬C33(⌬C49) transfectants on the other hand (Fig. 4B), suggests that these clones may be sensitive to Tpn-dependent peptides. In accordance with recent studies by the McCluskey group, who analyzed Kb- and B*2705-associated peptide spectra in the presence and absence of Tpn (12, 36), we noted a considerable overlap in the spectra derived from .220.B*4402 cells transfected with hTpn, mTpn, or hTpn-L410F. The MS analysis of class I-eluted peptides, however, revealed the presence of individual peptides that were bound in the presence of hTpn and not detected in the presence of mTpn and hTpn mutants, or vice versa. Certainly, this does not contradict the previous finding that particular peptides are equally well presented to T cells by .220.B*4402/hTpn and .220.B*4402/mTpn cells (38). Since the t1/2 of HLA class I molecules were more dramatically influenced by Tpn mutants or mTpn than the peptide profiles, it seems possible that the differences in the peptide profiles tend to be underestimated due to the preferential potential loss of low affinity ligands during the immunoisolation of class I molecules. There is evidence for an enhanced exchange of class I-binding peptides in the ER (62). It remains to be elucidated whether under physiological conditions class I molecules regularly enter the TAP-associated complex in the unloaded form, or whether the Tpn:TAP complex predominantly serves as a peptide editor for class I molecules that may enter the complex with prebound peptides of suboptimal affinity. The latter scenario would resemble peptide editing in the HLA-DR/HLA-DM system (63). Taken together, our findings indicate a predominant function of the TAPassociated complex for the loading of HLA class I molecules with peptides conferring optimal stability. While Tpn is in principle able to modify the peptide cargo in the absence of TAP association, it plays out its crucial role to the full extent only by means of its capacity to recruit class I molecules into the complex with TAP.

Acknowledgments We are grateful to Drs. J. McCluskey, C.-A. Peh, R. DeMars, G. Moldenhauer, J. Neefjes, R. Tampé, and M. Brenner for their generous gifts of cells and Abs, and to Dr. M. Lobigs for helpful discussions.

References 1. Pamer, E., and P. Cresswell. 1998. Mechanisms of MHC class I-restricted antigen processing. Annu. Rev. Immunol. 16:323. 2. Momburg, F., and G. J. Ha¨ mmerling. 1998. Generation and TAP-mediated transport of peptides for major histocompatibility complex class I molecules. Adv. Immunol. 68:191. 3. Cresswell, P., N. Bangia, T. Dick, and G. Diedrich. 1999. The nature of the MHC class I peptide loading complex. Immunol. Rev. 172:21. 4. Degen, E., and D. B. Williams. 1991. Participation of a novel 88-kD protein in the biogenesis of murine class I histocompatibility molecules. J. Cell Biol. 112: 1099. 5. Hochstenbach, F., V. David, S. Watkins, and M. B. Brenner. 1992. Endoplasmic reticulum resident protein of 90 kilodaltons associates with the T- and B-cell antigen receptors and major histocompatibility complex antigens during their assembly. Proc. Natl. Acad. Sci. USA 89:4734. 6. Sadasivan, B., P. J. Lehner, B. Ortmann, T. Spies, and P. Cresswell. 1996. Roles for calreticulin and a novel glycoprotein, tapasin, in the interaction of MHC class I molecules with TAP. Immunity 5:103. 7. Van Leeuwen, J. E., and K. P. Kearse. 1996. Deglucosylation of N-linked glycans is an important step in the dissociation of calreticulin-class I-TAP complexes. Proc. Natl. Acad. Sci. USA 93:13997. 8. Carreno, B. M., J. C. Solheim, M. Harris, I. Stroynowski, J. M. Connolly, and T. H. Hansen. 1995. TAP associates with a unique class I conformation, whereas calnexin associates with multiple class I forms in mouse and man. J. Immunol. 155:4726.

1960

OPTIMIZATION OF PEPTIDE LOADING IN THE TAPASIN-TAP COMPLEX

9. Suh, W. K., E. K. Mitchell, Y. Yang, P. A. Peterson, G. L. Waneck, and D. B. Williams. 1996. MHC class I molecules form ternary complexes with calnexin and TAP and undergo peptide-regulated interaction with TAP via their extracellular domains. J. Exp. Med. 184:337. 10. Hughes, E. A., and P. Cresswell. 1998. The thiol oxidoreductase ERp57 is a component of the MHC class I peptide-loading complex. Curr. Biol. 8:709. 11. Bangia, N., P. J. Lehner, E. A. Hughes, M. Surman, and P. Cresswell. 1999. The N-terminal region of tapasin is required to stabilize the MHC class I loading complex. Eur. J. Immunol. 29:1858. 12. Barnden, M. J., A. W. Purcell, J. J. Gorman, and J. McCluskey. 2000. Tapasinmediated retention and optimization of peptide ligands during the assembly of class I molecules. J. Immunol. 165:322. 13. Lindquist, J. A., O. N. Jensen, M. Mann, and G. J. Ha¨ mmerling. 1998. ER-60, a chaperone with thiol-dependent reductase activity involved in MHC class I assembly. EMBO J. 17:2186. 14. Morrice, N. A., and S. J. Powis. 1998. A role for the thiol-dependent reductase ERp57 in the assembly of MHC class I molecules. Curr. Biol. 8:713. 15. Farmery, M. R., S. Allen, A. J. Allen, and N. J. Bulleid. 2000. The role of ERp57 in disulfide bond formation during the assembly of major histocompatibility complex class I in a synchronized semipermeabilized cell translation system. J. Biol. Chem. 275:14933. 16. Paulsson, K. M., P. O. Anderson, S. Chen, H. O. Sjo¨ gren, H. G. Ljunggren, P. Wang, and S. Li. 2001. Assembly of tapasin-associated MHC class I in the absence of the transporter associated with antigen processing (TAP). Int. Immunol. 13:23. 17. Grandea, A. G., III, M. J. Androlewicz, R. S. Athwal, D. E. Geraghty, and T. Spies. 1995. Dependence of peptide binding by MHC class I molecules on their interaction with TAP. Science 270:105. 18. Ortmann, B., J. Copeman, P. J. Lehner, B. Sadasivan, J. A. Herberg, A. G. Grandea, S. R. Riddell, R. Tampe´ , T. Spies, J. Trowsdale, and P. Cresswell. 1997. A critical role for tapasin in the assembly and function of multimeric MHC class I-TAP complexes. Science 277:1306. 19. Li, S., H. O. Sjo¨ gren, U. Hellman, R. F. Pettersson, and P. Wang. 1997. Cloning and functional characterization of a subunit of the transporter associated with antigen processing. Proc. Natl. Acad. Sci. USA 94:8708. 20. Solheim, J. C., M. R. Harris, C. S. Kindle, and T. H. Hansen. 1997. Prominence of ␤2-microglobulin, class I heavy chain conformation, and tapasin in the interactions of class I heavy chain with calreticulin and the transporter associated with antigen processing. J. Immunol. 158:2236. 21. Suh, W. K., M. A. Derby, M. F. Cohen-Doyle, G. J. Schoenhals, K. Fru¨ h, J. A. Berzofsky, and D. B. Williams. 1999. Interaction of murine MHC class I molecules with tapasin and TAP enhances peptide loading and involves the heavy chain ␣3 domain. J. Immunol. 162:1530. 22. Diedrich, G., N. Bangia, M. Pan, and P. Cresswell. 2001. A role for calnexin in the assembly of the MHC class I loading complex in the endoplasmic reticulum. J. Immunol. 166:1703. 23. Ortmann, B., M. J. Androlewicz, and P. Cresswell. 1994. MHC class I/␤2-microglobulin complexes associate with TAP transporters before peptide binding. Nature 368:864. 24. Suh, W. K., M. F. Cohen-Doyle, K. Fru¨ h, K. Wang, P. A. Peterson, and D. B. Williams. 1994. Interaction of MHC class I molecules with the transporter associated with antigen processing. Science 264:1322. 25. Neisig, A., C. J. Melief, and J. Neefjes. 1998. Reduced cell surface expression of HLA-C molecules correlates with restricted peptide binding and stable TAP interaction. J. Immunol. 160:171. 26. Knittler, M. R., K. Gu¨ low, A. Seelig, and J. C. Howard. 1998. MHC class I molecules compete in the endoplasmic reticulum for access to transporter associated with antigen processing. J. Immunol. 161:5967. 27. Owen, B. A., and L. R. Pease. 1999. TAP association influences the conformation of nascent MHC class I molecules. J. Immunol. 162:4677. 28. Greenwood, R., Y. Shimizu, G. S. Sekhon, and R. DeMars. 1994. Novel allelespecific, post-translational reduction in HLA class I surface expression in a mutant human B cell line. J. Immunol. 153:5525. 29. Copeman, J., N. Bangia, J. C. Cross, and P. Cresswell. 1998. Elucidation of the genetic basis of the antigen presentation defects in the mutant cell line .220 reveals polymorphism and alternative splicing of the tapasin gene. Eur. J. Immunol. 28:3783. 30. Grandea, A. G., III, P. J. Lehner, P. Cresswell, and T. Spies. 1997. Regulation of MHC class I heterodimer stability and interaction with TAP by tapasin. Immunogenetics 46:477. 31. Lewis, J. W., A. Sewell, D. Price, and T. Elliott. 1998. HLA-A*0201 presents TAP-dependent peptide epitopes to cytotoxic T lymphocytes in the absence of tapasin. Eur. J. Immunol. 28:3214. 32. Peh, C. A., S. R. Burrows, M. Barnden, R. Khanna, P. Cresswell, D. J. Moss, and J. McCluskey. 1998. HLA-B27-restricted antigen presentation in the absence of tapasin reveals polymorphism in mechanisms of HLA class I peptide loading. Immunity 8:531. 33. Schoenhals, G. J., R. M. Krishna, A. G. Grandea, III, T. Spies, P. A. Peterson, Y. Yang, and K. Fru¨ h. 1999. Retention of empty MHC class I molecules by tapasin is essential to reconstitute antigen presentation in invertebrate cells. EMBO J. 18:743. 34. Myers, N. B., M. R. Harris, J. M. Connolly, L. Lybarger, Y. Y. Yu, and T. H. Hansen. 2000. Kb, Kd, and Ld molecules share common tapasin dependencies as determined using a novel epitope tag. J. Immunol. 165:5656. 35. Garbi, N., P. Tan, A. D. Diehl, B. J. Chambers, H. G. Ljunggren, F. Momburg, and G. J. Ha¨ mmerling. 2000. Impaired immune responses and altered peptide repertoire in tapasin-deficient mice. Nat. Immun. 1:234.

36. Purcell, A. W., J. J. Gorman, M. Garcia-Peydro, A. Paradela, S. R. Burrows, G. H. Talbo, N. Laham, C. A. Peh, E. C. Reynolds, J. A. Lo´ pez de Castro, and J. McCluskey. 2001. Quantitative and qualitative influences of tapasin on the class I peptide repertoire. J. Immunol. 166:1016. 37. Grandea, A. G., III, P. G. Comber, S. E. Wenderfer, G. Schoenhals, K. Fru¨ h, J. J. Monaco, and T. Spies. 1998. Sequence, linkage to H2-K, and function of mouse tapasin in MHC class I assembly. Immunogenetics 48:260. 38. Peh, C. A., N. Laham, S. R. Burrows, Y. Zhu, and J. McCluskey. 2000. Distinct functions of tapasin revealed by polymorphism in MHC class I peptide loading. J. Immunol. 164:292. 39. Grandea, A. G., III, T. N. Golovina, S. E. Hamilton, V. Sriram, T. Spies, R. R. Brutkiewicz, J. T. Harty, L. C. Eisenlohr, and L. Van Kaer. 2000. Impaired assembly yet normal trafficking of MHC class I molecules in tapasin mutant mice. Immunity 13:213. 40. Lehner, P. J., M. J. Surman, and P. Cresswell. 1998. Soluble tapasin restores MHC class I expression and function in the tapasin-negative cell line .220. Immunity 8:221. 41. Yu, Y. Y., H. R. Turnquist, N. B. Myers, G. K. Balendiran, T. H. Hansen, and J. C. Solheim. 1999. An extensive region of an MHC class I ␣2 domain loop influences interaction with the assembly complex. J. Immunol. 163:4427. 42. Van Endert, P. M., R. Tampe´ , T. H. Meyer, R. Tisch, J. F. Bach, and H. O. McDevitt. 1994. A sequential model for peptide binding and transport by the transporters associated with antigen processing. Immunity 1:491. 43. Nijenhuis, M., S. Schmitt, E. A. Armandola, R. Obst, J. Brunner, and G. J. Ha¨ mmerling. 1996. Identification of a contact region for peptide on the TAP1 chain of the transporter associated with antigen processing. J. Immunol. 156:2186. 44. Stam, N. J., H. Spits, and H. L. Ploegh. 1986. Monoclonal antibodies raised against denatured HLA-B locus heavy chains permit biochemical characterization of certain HLA-C locus products. J. Immunol. 137:2299. 45. Parham, P., C. J. Barnstable, and W. F. Bodmer. 1979. Use of a monoclonal antibody (W6/32) in structural studies of HLA-A,B,C antigens. J. Immunol. 123:342. 46. Mu¨ ller, C., H. Herbst, B. Uchanska-Ziegler, A. Ziegler, F. Schunter, I. Steiert, C. Muller, and P. Wernet. 1985. Characterization of a monoclonal anti-Bw4 antibody (Tu¨ 109): evidence for similar epitopes on the Bw4 and Bw6 antigens. Hum. Immunol. 14:333. 47. Tahara, T., S. Y. Yang, R. Khan, S. Abish, G. J. Ha¨ mmerling, and U. Ha¨ mmerling. 1990. HLA antibody responses in HLA class I transgenic mice. Immunogenetics 32:351. 48. Kropshofer, H., A. B. Vogt, L. J. Stern, and G. J. Ha¨ mmerling. 1995. Self-release of CLIP in peptide loading of HLA-DR molecules. Science 270:1357. 49. Mandelboim, O., H. T. Reyburn, M. Vales-Gomez, L. Pazmany, M. Colonna, G. Borsellino, and J. L. Strominger. 1996. Protection from lysis by natural killer cells of group 1 and 2 specificity is mediated by residue 80 in human histocompatibility leukocyte antigen C alleles and also occurs with empty major histocompatibility complex molecules. J. Exp. Med. 184:913. 50. Nilsson, T., and G. Warren. 1994. Retention and retrieval in the endoplasmic reticulum and the Golgi apparatus. Curr. Opin. Cell Biol. 6:517. 51. Neefjes, J. J., G. J. Ha¨ mmerling, and F. Momburg. 1993. Folding and assembly of major histocompatibility complex class I heterodimers in the endoplasmic reticulum of intact cells precedes the binding of peptide. J. Exp. Med. 178:1971. 52. Mu¨ ller, C. A., G. Engler-Blum, V. Gekeler, I. Steiert, E. Weiss, and H. Schmidt. 1989. Genetic and serological heterogeneity of the supertypic HLA-B locus specificities Bw4 and Bw6. Immunogenetics 30:200. 53. Takamiya, Y., T. Sakaguchi, K. Miwa, and M. Takiguchi. 1996. Role of HLAB*5101 binding nonamer peptides in formation of the HLA-Bw4 public epitope. Int. Immunol. 8:1027. 54. Yewdell, J. W., and J. R. Bennink. 1989. Brefeldin A specifically inhibits presentation of protein antigens to cytotoxic T lymphocytes. Science 244:1072. 55. Bai, A., and J. Forman. 1997. The effect of the proteasome inhibitor lactacystin on the presentation of transporter associated with antigen processing (TAP)-dependent and TAP-independent peptide epitopes by class I molecules. J. Immunol. 159:2139. 56. Peruzzi, M., K. C. Parker, E. O. Long, and M. S. Malnati. 1996. Peptide sequence requirements for the recognition of HLA-B*2705 by specific natural killer cells. J. Immunol. 157:3350. 57. Mandelboim, O., S. B. Wilson, M. Vales-Gomez, H. T. Reyburn, and J. L. Strominger. 1997. Self and viral peptides can initiate lysis by autologous natural killer cells. Proc. Natl. Acad. Sci. USA 94:4604. 58. Long, E. O., and S. Rajagopalan. 2000. HLA class I recognition by killer cell Ig-like receptors. Semin. Immunol. 12:101. 59. Herberg, J. A., J. Sgouros, T. Jones, J. Copeman, S. J. Humphray, D. Sheer, P. Cresswell, S. Beck, and J. Trowsdale. 1998. Genomic analysis of the tapasin gene, located close to the TAP loci in the MHC. Eur. J. Immunol. 28:459. 60. Harris, M. R., L. Lybarger, Y. Y. Yu, N. B. Myers, and T. H. Hansen. 2001. Association of ERp57 with mouse MHC class I molecules is tapasin dependent and mimics that of calreticulin and not calnexin. J. Immunol. 166:6686. 61. Lutz, C. T., K. D. Smith, N. S. Greazel, B. E. Mace, D. A. Jensen, J. A. McCutcheon, and N. E. Goeken. 1994. Bw4-reactive and Bw6-reactive antibodies recognize multiple distinct HLA structures that partially overlap in the ␣1 helix. J. Immunol. 153:4099. 62. Sijts, A. J., and E. G. Pamer. 1997. Enhanced intracellular dissociation of major histocompatibility complex class I-associated peptides: a mechanism for optimizing the spectrum of cell surface-presented cytotoxic T lymphocyte epitopes. J. Exp. Med. 185:1403. 63. Vogt, A. B., S. O. Arndt, G. J. Ha¨ mmerling, and H. Kropshofer. 1999. Quality control of MHC class II associated peptides by HLA-DM/H2-M. Semin. Immunol. 11:391.