Delineation of the HLA-DR Region and the Residues Involved in the Association with the Cytoskeleton*

Whereas the association of major histocompatibility complex (MHC) class II molecules with the cytoskeleton and their recruitment into lipid rafts play a critical role during cognate T/antigen-presenting cell interactions, MHC class II-induced signals, regions, and residues involved in their association and recruitment have not yet been fully deciphered. In this study, we show that oligomerization of HLA-DR molecules induces their association with the cytoskeleton and their recruitment into lipid rafts. The association of oligomerized HLA-DR molecules with the cytoskeleton and their recruitment into lipid rafts occur independently. Furthermore, the association with the cytoskeleton is HLA-DR-specific, since oligomerization of HLA-DP triggers its recruitment only into lipid rafts. HLA-DR molecules devoid of both (cid:1) and (cid:2) cytoplasmic tails did not associate with This DR TM was cloned into the The cDNA coding for the truncated DR (cid:1) chain has alanine mutants, first fragment from pBS KS 3 (cid:2) DR (cid:2) 008.14 a DR (cid:2) 3 (cid:2) BamHI:ClaI.D primer, ClaI restriction sites, cloning alanines.

Major histocompatibility complex (MHC) 1 class II molecules (HLA-DR, HLA-DP, and HLA-DQ in humans and I-A and I-E in mice) are polymorphic transmembrane heterodimers formed by the noncovalent association of a 36-kDa ␣-chain and a 29-kDa ␤-chain. They are constitutively expressed on B lymphocytes, dendritic cells, macrophages, the so-called professional antigen-presenting cells, and to some extent they are inducible on many other cell types. Recognition of the MHC class IIpeptide complex by the T cell receptor results in a signal transduction cascade both in T cells (1) and antigen-presenting cells (2). Engagement of MHC class II molecules with their natural ligands, superantigens, CD4, and LAG-3, or with specific antibodies is known to trigger signals leading to homo-and heterotypic aggregations (3,4), proliferation and differentiation of B lymphocytes (5-7), dendritic cell maturation (8), production of cytokines and inflammatory mediators (9 -11), and, under certain conditions, even programmed cell death (12,13). The cellular responses to MHC class II-mediated signaling depend upon the cell type (14), the developmental state, and/or the physiological status of the interacting cell (12,13) as well as the presence of other costimulatory molecules (15). There is also a growing body of evidence suggesting that the spatial organization of MHC class II in the plasma membrane as well as structural features play a critical role in eliciting effector responses.
Recent studies have shown that 3-20% of HLA-DR molecules constitutively reside in the lipid rafts of various antigen-presenting cells (16,17). Both signaling and antigen presentation are greatly enhanced by the incorporation of HLA-DR molecules into lipid rafts (18 -20). These specific microdomains of the plasma membrane are rich in cholesterol and glycosphingolipids (21). They can be isolated based on their insolubility in certain nonionic detergents (e.g. Triton X-100) at low temperature and their low buoyant density in sucrose gradients (22,23). Lipid rafts are residences for a number of proteins such as glycosylphosphatidylinositol-anchored proteins, Src family tyrosine kinases, Csk, and other signaling molecules such as protein kinase C, phosphatidylinositol 3-kinase, phospholipase C-␥ (24), PAG, and LAT as well as T cell receptor, B cell receptor (BCR), and Fc⑀RI (25)(26)(27).
Ligation of HLA-DR with specific bivalent Abs induced tyrosine phosphorylation of many substrates and cell-cell adhesion (16). Both events are dependent on the integrity of the lipid raft (16). Unlike HLA-DR, ligation of HLA-DP with specific mAbs induces neither tyrosine phosphorylation nor homotypic adhe-* This work was supported by grants from the Arthritis Society of Canada and the Canadian Arthritis Network (to W. M.) and by grants from the Canadian Institutes of Health Research (CIHR) and CRS Inc. (to W. M. and J. T.). 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.
¶ Supported by studentships from the Fonds pour la Formation de Chercheurs et l'Aide à la Recherche.
ʈ Supported by a fellowship award from Canadian Arthritis Network (CAN sion (28). It is thus possible that the differential signaling through the HLA-DP isotype is due to differential localization in the plasma membrane following engagement. Whereas there is no consensus as to the requirements for MHC class IImediated signaling, it is well established that MHC class IIinduced tyrosine phosphorylation is not affected by the substitution or deletion of the cytoplasmic domains of the ␣ and ␤ chains (29,30). However, the intracellular tail of the MHC class II ␤ chain is required for the translocation of the protein kinase C␣ and protein kinase C␤II isoforms (29,(31)(32)(33). Clearly, the cytoplasmic domains are involved in Ag presentation, since truncated MHC class II molecules have a compromised Ag presentation capacity (15,34,35), but their intact localization into lipid rafts demonstrates the unresponsiveness of lipid raft populations of HLA-DR in Ag presentation (16,33). In addition to their localization and recruitment into lipid rafts, MHC class II molecules also associate with the cytoskeleton (36 -39). Such association seems to be mediated by the actin cytoskeleton (37,40), and actin microfilaments may act as regulators of the movement and capping of HLA-DR receptors (38). They may also be involved in MHC class II signaling, since both cytoskeleton-stabilized membrane domains and lipid rafts provide sites at which localized signal transduction can occur because of the changed phosphorylation status caused by local kinases and phosphatases (18,41).
In this study, we investigated the impact of oligomerization of HLA-DR and -DP isotypes on their recruitment into lipid rafts and their association with the cytoskeleton and the importance of the cytoplasmic domains of the ␣ and ␤ chains in these responses. Our results show that cytoplasmic tails of HLA-DR were required for their association with the cytoskeleton unlike their recruitment into lipid rafts, where cytoplasmic tails were not required. Residues at positions Gly 226 -His 227 -Ser 228 of the ␤ chain play a critical role in the cytoskeleton association. Furthermore, HLA-DP partitioned into lipid rafts but not to the cytoskeleton, suggesting that MHC class II isotype-specific signaling is linked to their differential association with the cytoskeleton.

EXPERIMENTAL PROCEDURES
Antibodies and Reagents-The following antibodies were used: mAb L243 (mouse IgG2a that recognizes a conformational epitope on HLA-DR; ATCC, Manassas, VA), mAb DA6.147 (mouse IgG1 that recognizes the C-terminal intracellular tail of the HLA-DR ␣ chain; a generous gift from Dr. P. Cresswell, Yale University), mAb B7/21 (mouse IgG1 that recognizes a conformational epitope of HLA-DP; a generous gift from Dr. M. Tremblay, Infectious Diseases Research Centre, Université Laval), 8C12 and 6E3 (mouse IgG2a, anti-SEB and anti-SEA, respectively), generated in our laboratory). Rabbit polyclonal anti-HLA-DR ␣-chain was a generous gift from Dr. R. Sékaly (Department of Microbiology and Immunology, Université de Montréal). The secondary antibodies included goat anti-mouse IgG (H ϩ L) (Jackson Immunoresearch Laboratories Inc.), goat anti-mouse IgG coupled to HRP, and goat anti-rabbit IgG coupled to HRP (Santa Cruz Biotechnology, Inc., Santa Cruz, CA). Methyl-␤-cyclodextrin (M␤CD) and HRP-conjugated cholera toxin B subunit were purchased from Sigma, and latrunculin B (LB) was from Calbiochem.
cDNAs and Mutagenesis-cDNAs coding for HLA-DR␣ (42) and HLA-DR␤ 0101 (DR␤008) (43) as well as DPw2 ␣ and ␤ cDNAs cloned from 45.1 cells were a kind gift from Dr. Eric O. Long (National Institutes of Health). DP␣ (42) was cloned in the RSV.5 vector (44), and DP␤ (43) was cloned in the RSV.7 vector, a fusion between RSV.3 (44) and pHEBO (45). All mutations were introduced by the PCR overlap extension method (46). The sequence of the mutagenic oligonucleotides is available upon request. The HLA-DR␤ chain cDNA coding for a truncated protein was obtained from Dr. R. Sékaly (47). This DR␤TM cDNA was cloned into the RSV.3 vector. The cDNA coding for the truncated DR␣ chain has been described previously (47,48). To construct the alanine mutants, a first fragment was amplified from pBS KS 3ЈDR␤008.14 using a DR␤ 3Ј BamHI:ClaI.D primer, which harbors BamHI and ClaI restriction sites, for subsequent cloning steps and a mutagenic primer coding for three successive alanines. A second reaction with pBS KS 3ЈDR␤008.14 was performed using a complementary mutagenic primer and the universal primer. Following the overlap reaction, the PCR product was subcloned into the StyI and ClaI sites of pBS KS 3ЈDR␤008.14. The DNA sequence was confirmed by sequencing. The final cloning step was accomplished by cloning the cDNA fragments into the BamHI site of the SR␣ puro vector. This strategy was used for all alanine scan mutants except DR␤ 4AAA. To construct the DR␤ 4AAA mutant, the first fragment was amplified from pBS KS 5ЈDR␤008.7 using the reverse primer and a mutagenic primer coding for the GHS:AAA mutation. A second reaction was performed with pBS KS 5ЈDR␤008.7 using a complementary mutagenic primer and the universal primer. Following the overlap reaction, the PCR product was subcloned into the StyI sites of pBS KS 5ЈDR␤008.7. The DNA sequence was confirmed by sequencing. cDNA fragments were introduced into the BamHI site of the SR␣ puro vector.
Cell Treatment and Stimulation-BJAB cells (10 7 cells/ml) were either pretreated with 10 mM M␤CD for 10 min at 37°C to disrupt rafts or with 10 M LB for 45 min at 37°C to destabilize actin microfilaments, as indicated. Treatment with M␤CD or LB did not affect the interaction of HLA-DR with the specific anti-HLA-DR mAb L243 as determined by fluorescence-activated cell sorting analysis, and cell viability was not compromised as measured by a trypan blue dye exclusion assay (data not shown). Cells were incubated with 6E3 isotype control mAb or anti-HLA-DR mAb L243 (0.5 g/10 6 cells) for 30 min at 4°C and cross-linked with goat anti-mouse IgG (1 g/10 6 cells) for 5 min at 37°C or 20 min at 4°C before lysis.
Sucrose Gradients and Western Blotting-Cells (10 7 ) were lysed in 400 l of ice-cold TNE buffer (10 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA) containing 1% Triton X-100 and a mixture of protease inhibitors (Roche Applied Science) for 30 min on ice. Lysates were mixed with an equal volume of 85% sucrose in TNE and placed in SW60Ti centrifuge tubes. The samples were overlaid with 2.4 ml of 35% sucrose and 1 ml of 5% sucrose in TNE and centrifuged for 16 h at 35,000 rpm at 4°C. Eleven fractions (380 l) were collected, beginning at the top of the tube. Fraction 12, which had a high density insoluble pellet containing the cytoskeleton (49), was extensively washed with TNE. The low density insoluble fractions (2-4) of the HeLa cells (lipid rafts) were recovered from the 35-5% interface, mixed with three volumes of TNE buffer, and centrifuged at 39,000 rpm for 4 h at 4°C. Aliquots of each fraction were mixed with 6ϫ Laemmli buffer containing 6% 2-mercaptoethanol and heated for 5 min at 95°C. Samples were resolved by SDS-PAGE, and proteins were transferred onto polyvinylidene difluoride membranes (Millipore Corp., Bedford, MA). After blocking with 5% skim milk and 0.1% Tween 20 in TBS, the polyvinylidene difluoride membranes were incubated with the polyclonal anti-HLA-DR-␣ chain, the mAb DA6.147, or the mAb XD5, washed extensively, and subjected to chemiluminescent detection with appropriate HRP-conjugated anti-IgG antibodies and ECL (PerkinElmer Life Sciences). To detect the GM1 rafts marker, 10-l aliquots of each fraction were dotted on a polyvinylidene difluoride membrane, which was then incubated with HRP-conjugated cholera toxin B subunit and developed with chemiluminescent reagents as described above.

Cross-linking of HLA-DR Induces Their Recruitment into Both Lipid Rafts and the High Density Insoluble Pellet Fraction in BJAB Cells in a Temperature-independent Fashion-We
recently reported that 3-5% of HLA-DR molecules are constitutively present in the lipid rafts of normal B cells and different B cell lines and that ligation of these molecules with specific mAbs alone did not lead to the recruitment of additional HLA-DRs into these compartments (16). To determine whether the localization of these molecules could be modified proportionate to the extent of their oligomerization, BJAB cells were stimulated for 30 min at 4°C with anti-HLA-DR mAb L243 followed by medium or goat anti-mouse IgG antibodies for an additional 5 min at 37°C. As expected, slight amounts of HLA-DR molecules are constitutively present in the lipid rafts of BJAB cells. GM1 ganglioside, a well known constituent of the lipid rafts, was recovered in fractions 3 and 4, indicating that our conditions were appropriate for separating raft and nonraft fractions. Anti-HLA-DR mAb L243 alone had little if any effect on the recruitment of HLA-DR molecules into lipid rafts (Fig. 1). In contrast, cross-linking of anti-HLA-DR mAbs with goat antimouse IgG antibodies induced a substantial recruitment of HLA-DR into the raft compartment (Fig. 1). Increasing the concentration of the secondary antibody resulted in the appearance of HLA-DR molecules in the high density insoluble pellet fraction. Kinetics experiments indicated that the translocation of HLA-DR molecules to the pellet fraction occurred after 5 min of cross-linking (data not shown).
Since MHC class II and BCR molecules are very similar with respect to their induced signaling (24, 50) and since maximal BCR translocation and stable residency within lipid rafts occur at 4°C (2, 51), we looked at whether the recruitment of HLA-DR into these different compartments could occur at 4°C. As shown in Fig. 2, the recruitment of HLA-DR molecules into the raft compartment and their recovery in the high density insoluble pellet occurred to the same extent at 37 and 4°C. These results indicate that the levels of HLA-DR oligomerization affect their localization in the plasma membrane compartments and that the response is totally temperatureindependent.
M␤CD Inhibits the Recruitment of HLA-DR into Lipid Rafts but Not into the High Density Insoluble Pellet in BJAB Cells-A previous study showed that the recovery of HLA-DR in lipid rafts is prevented by M␤CD (16), a drug that disrupts raft integrity by extracting plasma membrane cholesterol, confirming its presence in lipid rafts (52). We looked at whether the recovery of cross-linked HLA-DR in the high density insoluble pellet could be affected by the integrity of lipid rafts and whether recruitment of HLA-DR molecules into lipid rafts is a step that is required prior to their association with the high density insoluble pellet. Cells were pretreated with M␤CD followed by stimulation with anti-HLA-DR mAb L243 and goat anti-mouse IgG antibodies. Fig. 3, A and B, shows that, as expected, the M␤CD treatment resulted in the complete disappearance of HLA-DR from the low density fractions with a corresponding increase in the soluble fractions. In contrast, the recovery of HLA-DR from the high density insoluble pellet fraction (Fig. 3B) was not affected by this treatment. Thus, the integrity of lipid rafts is not necessary for the recruitment of HLA-DR into the high density insoluble pellet, and it is highly likely that the recruitment of HLA-DR into these two different membrane compartments occurs independently.
LB Affects the Association of HLA-DR Molecules with the Cytoskeleton but Not Their Recruitment into Lipid Rafts-We hypothesized that the recruitment of HLA-DR into the high density insoluble pellet occurred because of their association with cytoskeleton components, since it was previously demonstrated that HLA-DR associates with the cytoskeleton (36 -39) and co-precipitates with actin (37,40). Moreover, the cytoskeleton is recovered in the high density insoluble pellet fraction of sucrose density gradients (49). To confirm the association of HLA-DR with the cytoskeleton in the high density insoluble pellet, to verify the role of actin polymerization in this association, and to demonstrate the independence of HLA-DR recruitment into different cell compartments, cells were pretreated with 10 M LB, an inhibitor of actin polymerization (53), prior to the oligomerization of HLA-DR. As shown in Fig.  3C, the LB pretreatment greatly reduced the amount of HLA-DR recovered in the high density insoluble pellet. However, the same treatment had no effect in the recruitment of HLA-DR into lipid rafts. These results confirm the association of HLA-DR with cytoskeleton in the high density insoluble fraction, the requirement of actin polymerization for this association, and the independence of the recruitment of these molecules into lipid rafts with respect to their association with cytoskeleton.
The Localization of HLA-DR in Transfected HeLa Cells-We extended our analysis to HeLa cells transfected with HLA-DR␣ and HLA-DR␤ (HLA-DRWT/WT), and we analyzed the localization of HLA-DR molecules in these cells. HLA-DRWT/WT molecules were present constitutively at very low levels in lipid rafts (Fig. 4). Their ligation with anti-HLA-DR mAb L243 slightly increased the amount of raft-associated HLA-DR molecules. Cross-linking of anti-HLA-DR mAb L243 with secondary antibodies resulted in significant recruitment of HLA-DR molecules into lipid rafts and greatly enhanced their association with the cytoskeleton. The HLA-DR molecules thus act in a similar fashion to HLA-DR molecules in BJAB cells with respect to raft recruitment and the association with the cytoskeleton (Fig. 4).
The Localization of HLA-DP in Transfected HeLa Cells-Recent studies demonstrated that differential signaling was mediated by HLA-DR and HLA-DP (28,54), suggesting a difference in the localization of these molecules in membrane compartments. To investigate the behavior of HLA-DP in the plasma membrane, we analyzed the localization of HLA-DP molecules in transfected HeLa cells. Like HLA-DR, HLA-DP molecules were constitutively present in lipid rafts at very low levels but not in the cytoskeleton fraction (Fig. 5). Unlike HLA-DR, the oligomerization of HLA-DP with mAb B7/21 cross-linked with secondary anti-IgG antibodies resulted in their recruitment to lipid rafts only. The failure of HLA-DP to associate with cytoskeleton is not due to the specificity of the mAbs, because similar results were obtained when other anti-HLA-DP/DR (IVA12) mAbs were used. Similar results were also observed when HLA-DP molecules were cross-linked in BJAB cells (data not shown).
The Cytoplasmic Domains of HLA-DR Are Required for Their Association with the Cytoskeleton-Recent studies showed that the cytoplasmic tails of the ␣and ␤-chains are not required for the constitutive localization of HLA-DR in lipid rafts (16,33). To determine whether these domains are involved in the association of HLA-DR with the cytoskeleton and their recruitment into lipid rafts, HeLa cells expressing HLA-DR molecules devoid of ␣ and ␤ cytoplasmic domains (DR TM/TM) were left unstimulated or stimulated with cross-linked Abs. The partitioning was analyzed as described above. Like HLA-DR WT/ WT, an association with the cytoskeleton could not be detected in the unstimulated cells (Fig. 6). However, unlike HLA-DR WT/WT, cross-linked HLA-DR TM/TM did not partition in the cytoskeleton (Fig. 6). These results suggest that the ␣or ␤-chain cytoplasmic domain or both are able to link HLA-DR to the cytoskeleton. To determine which of the tails is critical, the distribution of cross-linked HLA-DRs with truncated cytoplasmic domains of the ␣ (DR TM/WT) or ␤ chain (DR WT/TM) were analyzed as described above. As shown in Fig. 7, both crosslinked HLA-DR TM/WT and WT/TM partitioned into lipid rafts as well as into the cytoskeleton, indicating that cytoplasmic domain of both the chains mediated the association of HLA-DR with the cytoskeleton.
Glycine, Histidine, and Serine (Gly 226 , His 227 , and Ser 228 ) Residues in the ␤ Chain Cytoplasmic Domain Are Involved in the HLA-DR Association with the Cytoskeleton-Several studies point to the predominance of the ␤ chain in signaling events (29,30,(55)(56)(57). We therefore focused on the ␤ chain to determine which residues in its cytoplasmic tail are involved in this association. Based on the above results showing that oligomerization of HLA-DP failed to induce HLA-DP association with FIG. 3. M␤CD abolishes the association of HLA-DR molecules with lipid rafts without affecting their association with the cytoskeleton. Latrunculin B treatment affects the association of HLA-DR molecules with the cytoskeleton. Human BJAB cells were pretreated or not pretreated with 10 mM M␤CD for 10 min at 37°C (A and B), stimulated with isotype control mAb (A) or with the anti-DR mAb L243 (B) for 30 min on ice, and then cross-linked for 5 min at 37°C with (10 g/10 7 cells) goat anti-mouse secondary antibodies. C, BJAB cells were pretreated or not pretreated for 45 min at 37°C with 10 M latrunculin B before stimulation and cross-linking as described above. Cells were then lysed in 1% Triton X-100 and fractionated as described elsewhere. Raft samples from 2.5 ϫ 10 6 cell equivalents (fractions 3 and 4), pellet samples from 2.5 ϫ 10 6 cell equivalents (fraction 12), and soluble samples from 2.5 ϫ 10 5 cell equivalents (fractions 10 and 11) were resolved by SDS-PAGE and analyzed by immunoblotting using anti-HLA-DR ␣ chain mAb DA6.147 as described in the legend to Fig. 1 the cytoskeleton, we first used HeLa cells expressing chimerical HLA-DR ␤ (in which the cytoplasmic domain was replaced by that of HLA-DP) and tailless ␣-chains (HLA-DR ␣TM/␤DP) ( Table I). As shown in Fig. 8, oligomerization of HLA-DR induced significant recruitment of HLA-DR into lipid rafts but failed to induce any association with the cytoskeleton, supporting the involvement of specific residues in the association with the cytoskeleton. We  (Table I). In all transfectants, and like HLA-DR WT/WT molecules, mutant HLA-DR molecules were constitutively present in lipid rafts and absent from the cytoskeleton (Fig. 9). Oligomerization of HLA-DR molecules with L243 mAb and secondary antibodies resulted in the recruitment of all mutant HLA-DR molecules into lipid rafts and their recovery from the cytoskeleton except for HLA-DR ␣TM/␤4AAA, which localized only very slightly with the cytoskeleton (Fig. 9). Three residues in the ␤ chain cytoplasmic domain, glycine-histidine-serine (Gly 226 , His 227 , and Ser 228 ), are thus involved in the association of HLA-DR with the cytoskeleton. DISCUSSION Dimerization or even oligomerization is often a necessary step in the functioning of some cell surface receptors (58).  5. Oligomerization of HLA-DP molecules with the specific mAb B7/21 results in their recruitment into lipid rafts but not into the cytoskeleton in transfected HeLa cells. HeLa-transfected cells (10 7 cells) that express HLA-DP molecules were stimulated for 30 min at 4°C with isotype control mAb (1 g/10 6 cells) (A) or anti-HLA-DP mAb B7/21 (1 g/10 6 cells) (B) and cross-linked by goat anti-mouse IgG antibodies (1 g/10 6 cells) for 5 min at 37°C. Cells were then lysed with 1% Triton X-100 and fractionated as described previously. Raft (R) samples from 7 ϫ 10 6 cell equivalents (pool of fractions 2-4), soluble samples from 2 ϫ 10 6 cell equivalents (fractions 9 -11), and pellet samples from 7 ϫ 10 6 cell equivalents (fraction 12) were resolved by SDS-PAGE and subjected to immunoblot analysis with XD5 antibody. C, analysis of the expression of HLA-DP in transfected HeLa cells by flow cytometry using B7/21 mAb. Moldovan et al. (59) demonstrated that dimer formation is essential for the coligand and coreceptor functions of CD4 in T cell activation. CD4 oligomers have also been detected on T cells (60). The cross-linking of HLA-DR molecules with antibodies may thus mimic HLA-DR-peptide complex-driven oligomerization by CD4 oligomers, which is essential for signal transduction. We (16) and others (17,(61)(62)(63) have reported that 3-20% of HLA-DR molecules are constitutively located in the raft compartment of normal human tonsillar B cells, human B cell lines (including Epstein-Barr virus-positive cells), human dendritic cells, normal human monocytes, and monocytic cell lines. Ligation of HLA-DR by mAbs alone does not increase recruitment into lipid rafts (16). Data presented in this study show that oligomerization of HLA-DR molecules leads to a significant increase in recruitment of HLA-DR molecules into lipid rafts and triggers their association with cytoskeleton. Whereas MHC class II and BCR molecules share many similarities with respect to their induced signals (24,50) and whereas maximal BCR recruitment into rafts occurs at 4°C (2, 51), the recruitment of HLA-DR into both lipid rafts and the cytoskeleton was an energy-independent event and occurred to the same extent at 37 and 4°C. The association of HLA-DR with lipid rafts was abolished by M␤CD but not by latrunculin B, whereas the translocation of HLA-DR to the cytoskeleton was decreased by latrunculin B but not by M␤CD. Cross-linked HLA-DR molecules can thus associate with the cytoskeleton independently of their localization in lipid rafts and vice versa. Previous studies on the localization/recruitment of MHC class II molecules in lipid rafts failed to detect partitioning to the cytoskeleton fraction (33,63). As shown in Fig. 1, a high level of oligomerization is required for the translocation of MHC class II molecules to the cytoskeleton, a level probably not reached in other studies (33,63).
HLA-DP molecules, like HLA-DR, are recruited into the lipid raft compartment following oligomerization. However, the oligomerization of HLA-DP molecules did not result in their as- sociation with the cytoskeleton, indicating that the association of oligomerized MHC class II molecules with the cytoskeleton is isotype-specific. It was recently shown that HLA-DR and HLA-DP differ in their ability to transduce signals (28,54). Since the cytoskeleton plays a role in receptor signaling, MHC class II isotype-specific signaling may in part be the consequence of their differential association with the cytoskeleton.
We then addressed the issue of which structural features of HLA-DR molecules are responsible for their association with lipid rafts and the cytoskeleton. Having previously demonstrated that the constitutive localization of HLA-DR in lipid rafts is unaffected by the deletion of the cytoplasmic domain of both the ␣ and ␤ chains (16), we now show that oligomerized HLA-DR TM/TM molecules translocate into lipid rafts. Our results are in agreement with those of Bécart et al. (33), who showed that truncation of the intracytoplasmic domains of I-A does not impair their recruitment into lipid rafts. However, recent data from Dolan et al. (64) using selected clones derived from the same cell types (I-A-transfected M12.C3 cells or SaI/A k cells) as Becart et al. (33) appear to indicate that cytoplasmic domain truncation may alter the constitutive association of I-A molecules with plasma membrane lipid rafts. The differences between these results probably come from the difference in the behavior of the used clones as compared with transfected cell lines and/or from the methods used to prepare cell lysates for the density gradient fractionation.
Our results demonstrate that truncated HLA-DR TM/TM molecules do not associate with the cytoskeleton following their oligomerization, and either chain cytoplasmic domain can independently mediate the association of HLA-DR with the cytoskeleton. In contrast, earlier studies provided evidence that truncated MHC class II molecules are still able to associate with the cytoskeleton. With the "discovery" of lipid rafts, one may wonder whether all reported MHC class II associations with the cytoskeleton are really with the cytoskeleton or lipid rafts or both. Indeed, both the interaction of receptors with the cytoskeleton and with lipid rafts are biochemically defined in part by their resistance to disruption by nonionic detergents such as Triton X-100 (21,41). The results reported here clearly show that truncated MHC class II molecules can relocalize to lipid rafts but not to the cytoskeleton. Thus, in the case of the earlier studies, the truncated MHC class II molecules were probably associated with lipid rafts (37). The reported associations with the cytoskeleton of wild-type MHC class II molecules most probably reflect both their recruitment into lipid rafts and their recruitment into the cytoskeleton (36,37,39).
We identified a sequence motif in three residues that conferred a distinct functional identity to the ␤-chain through the association of HLA-DRs with the cytoskeleton. The function of the IA␤ cytoplasmic domain in Ag presentation both in vitro and in vivo (56) and in MHC class II signaling (29) is disturbed by mutations of amino acid residues Gly 227 -Pro 228 of the I-A ␤-chain. These two residues correspond to the Gly 226 and His 227 residues of the human DR1␤ chain. Harton and Bishop (57) showed that Gly 227 and Pro 228 are required for wild-type I-A signaling levels in mutant IA k -transfected cells, although Gln, a residue found in the I-E␤ cytoplasmic domain, can replace Pro 228 . Unlike HLA-DR, the HLA-DP ␤-chain has Val 226 and Gln 227 at these positions, and cross-linked HLA-DPs do not associate with the cytoskeleton. It thus appears that Gln 227 is not enough for the association of MHC class II molecules with the cytoskeleton. The three residues in the HLA-DR ␤ chain may function as an interaction site for adaptors or cytoskeleton proteins. It is possible that HLA-DR molecules are associated directly or indirectly with membrane actin, as suggested by the fact that actin co-immunoprecipitates with MHC class II molecules (37,40). This association may favor their linkage to polymerized cytoskeletal actin following oligomerization. Actin microfilaments co-localize underneath HLA-DR receptor caps on the surface of B cells, whereas F-actin microfilaments control the capping of HLA-DR in B cells (38). Although the three ␤-chain residues involved in the association of HLA-DR molecules with the cytoskeleton are absent from the ␣ chain, oligomerized HLA-DR molecules devoid of the ␤ chain are still able to associate with cytoskeleton, suggesting that other residues on the ␣ chain can also mediate the HLA-DR cytoskeleton association. Studies are currently under way to address this issue.
The mobility of MHC class II molecules in the plasma membrane, which involves changes in their membrane localization and organization, are important for the immunological functioning of these molecules as their accumulation at the contact zone between antigen-presenting cells and T cells initiated by agonist MHC class II-peptide complexes (65). Truncating I-A k cytoplasmic domains seems to have a limited effect on class II lateral diffusion following binding of Fab or intact anti-I-A mAb, although more I-A molecules appear mobile when Fabs are used to measure the mobility of I-A (66). HLA-DR is thus constitutively mobile and is not associated with the cytoskeleton. Dimerization of HLA-DR molecules by mAbs alone keeps them mobile, and only oligomerization results in their association with the cytoskeleton, which limits their mobility in the plasma membrane. This reduction in mobility is probably a critical step for an efficient interaction of T cell receptor and MHC class II molecules in the immunological synapse, since FIG. 8. Substitution of the cytoplasmic HLA-DR ␤-chain for HLA-DP ␤-chain does not rescue the association of truncated HLA-DR with the cytoskeleton. HeLa cells transfected with HLA-DR, which has a truncated ␣ cytoplasmic chain and a ␤ cytoplasmic chain, replaced with the ␤ cytoplasmic chain of HLA-DP (HLA-DR ␣TM/␤DP) were stimulated for 30 min at 4°C with isotype control mAb (1 g/10 6 cells) (A) or anti-HLA-DR mAb L243 followed by secondary antibody anti-IgG (B), lysed with 1% Triton X-100, and fractionated as described previously. Raft (R) samples from 7 ϫ 10 6 cell equivalents (pool of fractions 2-4), soluble samples from 2 ϫ 10 6 cell equivalents (fractions 9-11), and pellet samples from 7 ϫ 10 6 cell equivalents (fraction 12) were resolved by SDS-PAGE and analyzed by immunoblotting using anti-HLA-DR␣ polyclonal Ab. C, analysis of the expression of the chimera HLA-DR in transfected HeLa cells by flow cytometry. , and HeLa DR ␣TM/␤5AAA (E) were stimulated for 30 min at 4°C with isotype control mAb or with anti-HLA-DR mAb L243 followed by secondary antibody anti-IgG for 5 min at 37°C, lysed with 1% Triton X-100, and fractionated as described elsewhere. Raft (R) samples from 7 ϫ 10 6 cell equivalents (pool of fractions 2-4), soluble samples from 2 ϫ 10 6 cell equivalents (fractions 9-11), and pellet samples from 7 ϫ 10 6 cell equivalents (fraction 12) were resolved by SDS-PAGE and analyzed by immunoblotting using anti-HLA-DR␣ polyclonal Ab.
CD4 and irrelevant MHC class II peptides are excluded from the center and move to the periphery of the immunological synapse (67).