A Defective Viral Superantigen-Presenting Phenotype in HLA-DR Transfectants Is Corrected by CIITA1
http://www.100md.com
免疫学杂志 2005年第12期
Abstract
Activation of T lymphocytes by mouse mammary tumor virus superantigen (vSAg) requires binding to MHC class II molecules. The subcellular location where functional interactions occur between MHC class II molecules and vSAgs is still a matter of debate. To gain further insight into this issue, we have used human epithelial HeLa cells expressing HLA-DR1. Surprisingly, the human cells were unable to present transfected vSAg7 or vSAg9 to a series of murine T cell hybridomas. The defect is not related to a lack of vSAg processing, because these cells can indirectly activate T cells after coculture in the presence of B lymphocytes. However, after IFN- treatment, the HeLa DR1+ cells became apt at directly presenting the vSAg. Furthermore, transfection of CIITA was sufficient to restore presentation. Reconstitution experiments demonstrated the necessity of coexpressing HLA-DM and invariant chain (Ii) for efficient vSAg presentation. Interestingly, inclusion of a dileucine motif in the DR cytoplasmic tail bypassed the need for HLA-DM expression and allowed the efficient presentation of vSAg7 in the presence of Ii. A similar trafficking signal was included in vSAg7 by replacing its cytoplasmic tail with the one of Ii. However, sorting of this chimeric Ii/vSAg molecule to the endocytic pathway completely abolished both its indirect and direct presentation. Together, our results suggest that functional vSAgs-DR complexes form after the very late stages of class II maturation, most probably at the cell surface.
Introduction
Superantigens (SAgs)3 are small bacterial (bSAgs) or viral (vSAgs) proteins that bind MHC class II molecules and activate up to 20% of the T cell repertoire in a V-specific manner (for reviews, see Refs. 1, 2, 3). SAgs have long been implicated in various diseases as well as in autoimmune conditions (4, 5). They have proven to be useful tools for the appreciation of structural heterogeneity among class II molecules at the surface of APCs (6).
MHC class II heterodimers assemble in the endoplasmic reticulum (ER), where they swiftly associate with the invariant chain (Ii) (for reviews, see Refs. 7, 8, 9). This chaperone occupies the peptide binding groove of the newly synthesized class II molecules, preventing the binding of endogenous ER peptides and polypeptides (10, 11, 12). Ii is a type II membrane protein containing endosomal targeting signals through which the class II/Ii complex gets delivered to peptide-rich endosomes (13, 14, 15). In contrast to B cells, DCs from mice deficient for Ii express high levels of class II molecules at their surface (16, 17). Also, in the absence of Ii, class II molecules are efficiently expressed at the plasma membrane of transfected cells (18, 19). However, under these conditions, intracellular trafficking of class II molecules varies between cell lines (20).
Once in the endocytic pathway, proteinases sequentially digest Ii, except for the CLIP fragment, which remains in the peptide binding groove of the MHC class II molecule (reviewed in Ref. 21). The exchange of CLIP for an antigenic peptide that will be presented at the cell surface is promoted by HLA-DM (DM) (22, 23). This nonclassical class II molecule exhibits very little polymorphism and resides in the endocytic pathway (24, 25, 26). A cytoplasmic YXXL sorting motif allows DM to accumulate principally in late vesicles (27, 28, 29). The activity of DM is not restricted to CLIP removal, because numerous studies have highlighted its capacity to edit the global peptide repertoire by favoring those of higher affinity (30, 31, 32, 33). Beyond its effect on peptide editing, DM could also play an important direct role in fine-tuning the conformation of class II molecules (34, 35, 36).
Bacterial SAgs do not require processing and bind MHC class II molecules outside their peptide binding groove (37, 38, 39). However, it is now well established that the class II-bound peptides modulate the interaction of staphylococcal enterotoxin A, toxic shock syndrome toxin-1, and Mycoplasma arthritidis mitogen with their receptor (40, 41, 42, 43, 44, 45). Accordingly, chaperones involved in peptide binding to class II molecules can have a profound effect on SAg presentation (34, 46, 47). Bacterial SAgs bind to MHC class II molecules - and/or -chains in numerous fashions, and the three-dimensional structure of many complexes revealed direct contacts between SAgs and the peptide (48, 49, 50, 51)
The best-studied vSAgs are those produced by mouse mammary tumor virus (MMTV). Nevertheless, their crystal structure remains to be determined. MMTV is a type B retrovirus, transmitted horizontally through the milk of infected mothers or vertically once integrated in the germline (reviewed in Ref. 52). They produce vSAgs that are necessary for viral infectivity and virus spread (53, 54). These are type II transmembrane proteins encoded by an open reading frame in the 3' long terminal repeat (55, 56, 57, 58, 59). The mature MTV vSAg is a 45-kDa glycoprotein that needs to be activated through proteolytic processing (60). The nature of the active fragment remains ill defined; redundancy has been reported regarding the potency of the furin cleavage sites in activating vSAgs (61). Also, cathepsin L was shown to liberate a 27-kDa fragment, which could explain the activity of vSAgs in furin-deficient cells or after mutagenesis of the furin recognition motifs (62, 63). In addition to activating the vSAg, processing would explain the paracrine transfer of a soluble active C-terminal fragment to class II-positive APCs (64, 65, 66).
The exact binding site(s) of MMTV SAg on class II molecules remains obscure (67, 68, 69). Also, the intracellular trafficking of vSAgs and the site of class II encounter are still matters of debate. A thorough biochemical analysis of their maturation allowed Grigg et al. (70) to conclude that vSAgs reach the plasma membrane via the default exocytic pathway, independently of MHC class II expression. Class II-vSAg complexes were found exclusively at the plasma membrane and not in the endocytic pathway or the ER. Their observations imply an exclusive association of the vSAg with mature, peptide-loaded, SDS-stable class II molecules. Accordingly, it was later proposed that successful presentation of vSAg7 is conditional to the existence of a diverse, class II-bound peptide repertoire (71). Indeed, endogenous vSAg-induced T cell deletion was not taking place in modified mice expressing class II molecules almost exclusively loaded with CLIP (DM knockout (KO)) or loaded with a covalently linked peptide (E52–68).
These results contrast with those obtained in vitro by Hsu et al. (72), which suggest that class II-vSAg interactions occur in the ER through a CLIP-like region of vSAg occupying the peptide binding groove. In accordance with such an interaction early in the biosynthetic pathway, they reported that a DM-deficient human B cell line was as potent as its wild-type counterpart in stimulating T cells upon transfection of a vSAg.
In this study we show that DR-transfected cell lines differ dramatically in their ability to present MMTV vSAgs and that the mere expression of class II molecules is not sufficient to support presentation to T cell hybridomas. We report that CIITA expression rescues the presentation of vSAg7 by HeLa cells and that the presentation-competent phenotype is dependent on the coexpression of DR, Ii, and DM. The need for class II molecules to access the endocytic pathway coupled to the inactivation of vSAgs in these compartments support a model in which both ligands traffic independently inside the cell. Also, functional complexes probably form at the plasma membrane and only after class II molecules have acquired some presentation-permissive peptides.
Materials and Methods
Approximately 2 x 106 HeLa DR1 vSAg7 cells were cultured in medium supplemented with various concentrations of recombinant human IFN- (0.5–5 ng/ml; BD Pharmingen) for 5 days before T cell stimulation assays.
Results
HLA-DR1 expression by human epithelial cells does not support vSAg presentation
Transfected mouse fibroblasts expressing HLA-DR molecules are capable of vSAg presentation (95). Moreover, the responsive human V repertoire was similar to that described in the mouse (74). In the process of characterizing the molecular interactions between DR and SAgs, we generated vSAg7+ populations of DR1-transfected HeLa cells (HeLa DR1 vSAg7). To our surprise, these transfectants could not present the vSAg7 to the murine T cell hybridomas Kmls 13.11 (mV6) and KR3+ (mV8; Fig. 1A). The negative result obtained with HeLa was unexpected, because expression of the exact same class II cDNAs in DAP cells triggered a strong T cell response. We used three different APC:hybridoma ratios, and control DAP DR1 vSAg7 cells were effective at all concentrations.
We addressed the possibility that the MTV vSAg expression or activity was much reduced in human cells compared with mouse cells. Because the vSAg7-specific VS7 mAb staining was negative on both populations when analyzed by flow cytometry (data not shown), we capitalized on the ability of vSAg-producing cells to transfer the activity to vSAg-negative APCs (65, 66). Fig. 1B shows that in the presence of vSAg7-negative CH12 murine B cells, HeLa DR1 MTV7 cells can trigger a T stimulation as efficient as DAP DR1 vSAg7 cells. Although the exact mechanism remains to be characterized, such indirect presentation is probably the result of a successful paracrine transfer (64, 65, 66), and this strongly suggests that not only do HeLa cells express the vSAg, but it is also processed adequately.
We extended our initial observation to vSAg9 produced by MMTV9 (96). New independent transfections were obtained for both vSAg7 and vSAg9, and clones were picked. Compared with DR+ DAP cells expressing vSAg7 or -9, none of the vSAg+ HeLa clones could stimulate the specific hybridomas (Fig. 1C), whereas all triggered an efficient response in the indirect coculture assay (data not shown). These results confirm that HeLa DR1 cells cannot present vSAgs.
To rule out that this lack of presentation was due to the epithelial nature of HeLa cells and to determine whether this phenomenon could be observed in other cell types, we used the DR1+ M1 fibroblasts called 4N5 (88). Fig. 1C (left panel) shows that 4N5 cells stably transfected with vSAg7 were unable to stimulate the Kmls 13.11 T cell hybridoma.
A trivial explanation for the defect observed with HeLa would be that these cells do not express sufficient levels of class II molecules (97). However, surface stainings with the pan-DR L-243 mAb showed that both DAP and HeLa cells express significant and comparable levels of class II molecules (Fig. 1D). In addition, an independent vSAg+ DAP cell population expressing much lower levels of class II molecules triggers an efficient response from those T cell hybridomas (data not shown). Interestingly, although the DR expression is quantitatively very similar, the 2.06 mAb staining is much stronger on DAP cells (Fig. 1D). The 2.06 mAb binds to the -chain of all human class II isotypes, but would not recognize class II superdimers (98). The significance of this result is not clear, but it supports the idea that class II molecules are qualitatively somewhat different among various cell lines (6, 46, 89).
The presentation of vSAgs was suggested to be dependent on a diverse repertoire of class II-bound peptides (71). Because a fraction of the class II molecules in HeLa and DAP cells are bound with large polypeptides in the absence of Ii, we verified whether HeLa cells express some SDS-stable compact heterodimers containing short peptides and migrating at 55 kDa on immunoblots (10, 46, 99). Fig. 2A shows that nonboiled lysates prepared from both HeLa DR1 and DAP DR1 cells contain stable dimers recognized by the DR-specific XD5 mAb. Although such analysis does not give insights into the diversity of the peptide bound to DR, it rules out the absence of compact forms as a possible explanation for the behavior of HeLa cells. Importantly, DAP, HeLa, and M1 cells transfected with DR1 are all very efficient in stimulating the V8+ I1B3 murine T cell hybridoma in the presence of the bSAg staphylococcal enterotoxin B (SEB), ruling out a general Ag presentation defect (Fig. 2B).
IFN- rescues endogenous vSAg presentation by HeLa cells
The ability to present MTV SAgs is certainly not restricted to the mouse system, because class II-positive monkey (COS) and hamster (CHO) cells have been successfully used in the past (62, 66). Thus, we specifically addressed the role of the class II Ag processing machinery in vSAg presentation. The class II pathway is constitutively active only in professional APCs, but is strongly inducible in nearly all cells by IFN-. This pleiotropic cytokine affects all aspects of Ag processing and presentation, because it regulates the expression of >200 genes (100). We treated the inept HeLa DR1 vSAg7 cells for 5 days with IFN- and analyzed class II expression by flow cytometry (Fig. 3A). High levels of endogenous HLA-DM and -DQ were observed, whereas HLA-DR was clearly up-regulated due to expression of the homozygous endogenous HLA-DRB* 0102 genes (101). These cells were capable of endogenous vSAg7 presentation to Kmls 13.11 and KR3+ T cell hybridomas (Fig. 3B and data not shown). This increased presentation could be caused by up-regulation of critical molecules involved in either the class II maturation or the processing of vSAg. Although the latter possibility cannot be formally excluded, it does not concur with the fact that HeLa DR1 vSAg7 cells do not need this cytokine to produce functional vSAg7 and that IFN- did not increase presentation in the presence of BJAB (Fig. 3C).
Transfection of CIITA allows vSAg presentation by HeLa cells
As a first step in deciphering the mechanism of IFN- action and knowing that in its life cycle the MMTV primarily infects B lymphocytes, we tested endogenous presentation via the human Raji B cell line (54, 102, 103). These cells express all the class II processing machinery under control of the CIITA. The expression of this essential coactivator is constitutive in professional APCs and is induced by IFN- in almost all other cells (104). The vSAg7-encoding construct was electroporated in Raji cells, and total mRNA from the stably transfected clone 2H8 was analyzed for SAg expression. Fig. 4A shows that overall, the B cell transfectant expresses much less vSAg7-specific mRNA than do HeLa DR1 cells. However, functional assays showed that Raji cells are fully capable of presenting the vSAg7 to the two T cell hybridomas (Fig. 4B). Viral SAg surface expression was detected on Raji 2H8 by flow cytometry using the VS7 mAb (Fig. 4C). We were unable to detect vSAg7 expression on HeLa cells. However, other potent cell lines present vSAg7, but do not stain with the VS7 mAb (see below), suggesting that many factors contribute to the display of the epitope. Taken together, these results suggest that besides class II molecules, expression of additional genes under the control of CIITA is required for vSAg presentation.
To narrow down the list of molecules capable of restoring efficient presentation, we supertransfected CIITA into vSAg7-expressing HeLa cells. These cells became fully apt at presenting the endogenous vSAg to Kmls 13.11 (Fig. 4D). To relate to our HLA-DR transfectants and given the fact that CIITA up-regulates the expression of all classical class II isotypes (Fig. 3A), it became important to assess the relative importance of DR for the presentation of vSAg7 by HeLa CIITA cells. The anti-DR mAb L-243 almost completely inhibited T cell activation in a dose-dependent manner (Fig. 4E). DP- and DQ-specific mAbs did not affect the level of stimulation, confirming that these isotypes do not efficiently present vSAgs (95) (data not shown). Taken together, these results suggest a direct implication of MHC class II Ag processing machinery in vSAg presentation.
Coexpression of Ii and HLA-DM is sufficient to restore vSAg presentation by HeLa DR1+ cells
CIITA was shown to affect the expression of a large panel of genes implicated in diverse functions, such as cell migration, differentiation, and proliferation (105). Although many different mechanisms could potentially contribute to the CIITA-mediated enhanced vSAg presentation, we focused our efforts on the role of the intracellular class II chaperones. HLA-DM and Ii are strongly up-regulated by CIITA and play a crucial role in the trafficking of class II molecules and in the selection of their peptide repertoire. First, we stably transfected DM into HeLa DR1 vSAg7 (HeLa DR1 DM vSAg7). Even though vSAg7 transfer to BJAB remained high, this new population could not directly activate the two T cell hybridomas (Fig. 5, A and B). Similar results were obtained using cells transfected with Ii and sorted for high levels of CLIP surface expression (HeLa DR1 Ii vSAg7; Fig. 5C). However, when Ii and DM were cotransfected (HeLa DR1 Ii DM vSAg7), the stimulation increased dramatically for both T cell hybridomas used (Fig. 5A). An efficient stimulation was also observed using independent transfections where vSAg7 and DM were cotransfected into HeLa DR1 Ii cells (data not shown). Together, these results strongly point to a role of class II-bound peptides in vSAg presentation.
An ectopic endosomal sorting motif in the tail of DR rescues vSAg presentation
In light of the functional role of HLA-DM, we hypothesized that class II molecules must travel to the endocytic compartments to acquire the capacity to present vSAgs. Sorting of class II to endosomes is assured by Ii, which bears two leucine-based motifs in its cytoplasmic tail (15, 106). We reasoned that the need for Ii could be bypassed by introducing such a signal in the tail of HLA-DR. To achieve this, we replaced the cytoplasmic tail of DR by that of HLA-DO. This chimeric molecule (DR1-LL) contains a functional dileucine motif, which results in accumulation not only at the plasma membrane, but also in lysosome-associated membrane protein-1+ intracellular compartments (77). However, HeLa DR1-LL vSAg7 cells did not stimulate T cell hybridomas despite the efficient transfer of the activity to a B cell line (Fig. 6). Interestingly, coexpression of Ii in these cells allowed the efficient direct presentation of vSAg7. These cells express class II levels similar to their nonstimulating counterpart (Fig. 6A). Also, the cell surface staining with CerCLIP.1 demonstrates that they express copious amounts of CLIP-class II complexes. These results confirm that class II molecules must travel to late endocytic compartments before acquiring the ability to present vSAgs.
Sorting vSAg7 to the endocytic pathway abrogates its activity
Given the importance of targeting the class II molecules to the endocytic pathway, we investigated the possibility of increasing the efficiency of presentation by sorting the vSAg7 to endosomes. Just like Ii, the endogenous MMTV superantigens are type II transmembrane proteins (59). Thus, we swapped the cytoplasmic tail of vSAg for that of Iip33 or p35 and tested the capacity of these chimeric molecules to activate T cells. For these experiments, we used DAP cells, which are fully competent for both direct presentation and transfer of vSAgs. Surprisingly, the stably transfected p33 cyto-vSAg7 and p35 cyto-vSAg7 chimeric molecules could not stimulate T cells even when cocultured in the presence of BJAB (Fig. 7A). Using the Ii-specific Pin.1 mAb on immunoblots, we found that the chimeric molecules migrated at the expected Mr of 45 kDa, and no degradation products were detected (Fig. 7B and data not shown). The chimeric molecules were almost undetectable by flow cytometry and fluorescence microscopy, suggesting that the vSAg portion imparts instability to the constructs (data not shown).
To circumvent the problem of low protein expression, we turned to transient overexpression in human HEK 293T cells. Two days after transfection in the absence of class II molecules, cells were split and tested in a functional transfer assay (data not shown) or were analyzed by fluorescence microscopy and flow cytometry (Fig. 7, C and D). In contrast to wild-type vSAg7, the chimeric molecules were not transferred from 293T cells to BJAB. Pin.1 staining showed that the chimeric p33 cyto-vSAg7 molecule was expressed and transported to punctate vesicles scattered in the cytoplasm. Importantly, to demonstrate the specificity of this distribution, we tested a similar construct between Iip35 and vSAg7 (p35 cyto-vSAg7). As expected from the fact that this Ii isoform is retained in the ER in the absence of DR (107), the staining was diffuse and did not show any defined vesicles (Fig. 7C). Together, these results suggest that the small amount of vSAg destined for T cell stimulation must not be sorted to the endocytic pathway even in the presence of class II molecules.
Discussion
Structure-function aspects of SAgs have been mostly studied in the past using MHC class II-transfected cells. For example, by analogy to the hierarchy described for the H-2 gene products, murine DAP cells have been used to demonstrate isotypic variations in the capacity of human class II molecules to present vSAgs (95). These cells do not express Ii or DM, but still present vSAgs very efficiently when transfected with HLA-DR. However, the situation was very different when we used a series of HeLa cell transfectants for study of the vSAg binding site on those exact same class II molecules. In such conditions, we found that the human epithelial cell line is incapable of presenting endogenous vSAgs. It has been reported that MMTV preferentially infects immature DCs as well as B cells, and that both APCs lead to V-specific T cell activation (108). Accordingly, when the prototypic human B cell line Raji was tested for endogenous vSAg7 presentation, efficient T cell stimulation was measured. We identified CIITA as being responsible for activating the class II processing pathway and for the efficient vSAg presentation. This effect was mainly due to the up-regulation of HLA-DM and Ii. From our results, we cannot rule out additional CIITA-mediated effects on the processing or sorting of vSAgs, because this transactivator was shown to modulate the expression of >50 genes in B cells (105). However, given the fact that all other CIITA-negative vSAg+ HeLa transfectants very efficiently transferred the activity, this hypothesis is unlikely. Comparing the ability of primary B cells from CIITA– vs CIITA+class II– mice to transfer vSAg might shed light on this issue.
The presentation efficiency of HeLa CIITA and Raji most certainly relies in part on their very high levels of surface class II compared with the DR/DM/Ii transfectants. Also, the relative expressions of class II, Ii, and DM is probably optimal under CIITA control compared with the situation in HeLa cells where transfected molecules are transcribed from a variety of promoters and by different vectors integrated randomly at various copy numbers.
Our results clearly showed that the expression of DM and Ii was sufficient to rescue vSAg presentation in HeLa DR1 cells. This finding is reminiscent of the important peptide/class II conformation for the binding of bacterial SAgs (44). For example, TSST-1 binding to class II is greatly augmented by the transfection of Ii and DM in DR+ T2 cells (41, 46). Indeed, Wen et al. (42, 43) demonstrated unambiguously that TSST-1 binding is affected by the class II-bound peptides. Interestingly, it was recently shown that the presentation of both vSAgs and TSST-1 is diminished in mice deficient for DM or Ii, suggesting that a diverse class II peptide repertoire is needed for these SAgs to bind APCs (71). At first glance, our results are in line with these in vivo studies. At the surface of chaperone-negative HeLa DR cells, the class II molecules are loaded with large ER polypeptides that could interfere with binding of vSAgs (46). However, polypeptides are also found in class II molecules expressed at the surface of DAP cells (10). Nevertheless, in both cell lines there are clearly some SDS-stable compact forms migrating at 55 kDa, which most certainly correspond to class II molecules loaded with some relatively short peptides (Fig. 2). Cell type differences may explain the fact that mouse cells display some permissive peptide(s) or polypeptide(s) allowing vSAg presentation. Mass spectrometry analyses will give clues about the degree of heterogeneity of these class II ligands.
The need for a diverse peptide repertoire could explain why coexpression of Ii, which displaces large polypeptides found in the ER, did not restore vSAg presentation in DR+ HeLa cells. For DM KO mice, this configuration results in the display of class II molecules bound almost exclusively with the CLIP peptide. Introducing DM in these cells could restore presentation through CLIP displacement and the binding of vSAg-permissive peptides. However, Hsu et al. (72) reported that DM-negative B cells expressing copious amounts of DR3-CLIP complexes are still capable of presenting endogenous vSAgs. Also, the surprising finding reported here that a dileucine motif in the tail of DR could restore presentation warrants evaluating alternative explanations for the need of displacing CLIP. Indeed, the majority, if not the totality, of class II molecules on the HeLa DR1-LL Ii cells are loaded with the CLIP peptide and present vSAgs. Because both lines are expressing DR-CLIP at their surface, what differentiates these complexes between HeLa DR Ii and HeLa DR1-LL Ii cells? Possibly the isoform of CLIP found on HeLa DR Ii cells (or on APCs from DM KO mice) is not permissive for vSAg binding. CLIP is usually found as a nested set of peptides on class II molecules, and N- or C-terminal extensions outside the peptide binding groove have been shown to inhibit SEB binding to class II molecules (109). HeLa cells, which are devoid of cathepsin S (110), may generate a form of CLIP incompatible with vSAg binding. This may explain why replacing in Ii the core CLIP sequence with a variety of defined DR-binding motifs did not restore vSAg binding (data not shown). To allow presentation, nonpermissive CLIPs must be either displaced by DM or, in the context of the DR1-LL construct, be trimmed to competent forms by lysosomal exopeptidases. We have previously shown that this chimeric DR1-LL accumulates in large quantities in endocytic vesicles containing markers of late endosomes (77). Despite a prolonged stay in these vesicles, these molecules recycle, because they are found in large amounts at the cell surface. HPLC analysis of the peptides found in DR vs DR1-LL would help resolve the issue of CLIP heterogeneity. Alternatively to the trimming of CLIP, the trafficking of DR1-LL through late vesicles might allow binding of some functionally relevant peptides after long-standing exposure to acidic pH. Such a mechanism could explain the indispensable need for Ii, which will incorporate in DR the more easily exchangeable CLIP peptide. In this context, it will be interesting to test the presentation of vSAg by HeLa cells expressing Ii and an allele such as DR4, which releases CLIP spontaneously (111). Whether DR1-LL could also bypass the need for DM in the presentation of DM-dependent nominal Ags remains to be tested.
Although its influence seems clear in many experimental systems, it is still possible that DM plays a role beyond its peptide-editing activity. Looking at bacterial superantigen, Ab, or TCR recognition of class II molecules, recent evidence strongly suggests that DM might directly influence the structure of class II molecules (34, 35, 36). However, the mechanism underlying this new DM attribute remains unknown. In the case of vSAg presentation by HeLa cells, a contact-induced, peptide-independent role of DM on the structure of DR would be difficult to reconcile with the fact that an LL motif on DR compensates for the absence of the nonclassical class II chaperone. Also, the cell type dependency for DM is intriguing. If one considers a role for DM on the structure of class II molecules, then what could be compensating for DM and Ii in DAP, COS, and CHO cells, which all stimulate T cells upon transfection of vSAgs and DR? The most likely explanation is that in these permissive cells, HLA-DR could undergo a conformation modification mediated by some enzymes, tetraspans, chaperones, or severe pH conditions. For example, as opposed to HeLa cells, murine cells allow class II molecules to accumulate in the endocytic pathway in the absence of Ii (20). In nonpermissive cells, DM could affect the sorting of DR by dragging some molecules to the highly acidic lysosomes before their dissociation. Nevertheless, our observations describe the essential role of the endocytic pathway in the presentation of endogenous vSAgs. These results are in line with those reported by Golovkina et al. (71) showing the importance of peptides in vSAg presentation and those of Grigg et al. (70) showing that vSAgs interact with mature class II molecules and only at the cell surface.
Our results contrast with those of Hsu et al. (72) describing a DM-independent vSAg presentation pathway. The straightforward model that they proposed suggests that vSAgs bare an essential and specific motif that interacts with newly formed class II heterodimers in a CLIP-like manner. We cannot formally rule out that DR and vSAgs associate in the ER of HeLa cells in the absence of Ii and DM, but these complexes are nonfunctional. Also, vSAg molecules following DR (especially DR-LL) to endosomes would probably be degraded, as previously suggested (70). Indeed, replacing the cytoplasmic tail of vSAg for that of Iip33 completely abolished the ability of the vSAg to stimulate T cells in direct or transfer assays (Fig. 7). This last result suggests that if vSAgs gain access to the endocytic pathway, their passage must be swift and is probably not compatible with their intercalation in a putative class II-Ii-MTV nonamer. In furin-negative cells, it was proposed that alternative cleavage by cathepsins could generate an active form of vSAg (62, 63). HEK 293T cells are furin negative, and the efficient transfer of vSAg is in accordance with the existence of alternative cleavage sites. However, the fact that a soluble fragment is not released from cells expressing the chimeric Ii-vSAg suggests that the cleavage does not occur in endosomes. Alternatively, the transferred C-terminal part of the open reading frame product may need to remain associated to the N-terminal part of the vSAg and be re-expressed at the surface, a step that is precluded by the endosomal retention motif in our chimera. Finally, the binding of vSAg in an Ii-like fashion would imply a direct competition between the two molecules for class II binding. However, DAP cells expressing high levels of Ii20 (which lacks endosomal sorting motifs and stays associated with DR from the ER to the cell surface) stimulate T cells as efficiently as their Ii-negative counterpart (J. Thibodeau and R.-P. Sékaly, unpublished observations). In conclusion, our results support a model in which vSAgs travel on their own to the plasma membrane, where they encounter mature class II molecules. Binding and presentation of the vSAg depend on the nature of the stably bound class II peptides, which are acquired in various compartments and may be homogeneous under certain experimental conditions.
Acknowledgments
We thank Alexandre Brunet for helping with the characterization of chimeric molecules. We also thank Drs. F. Denis, G. Winslow, O. Bakke, D. L. Hardie, I. C. MacLennan, P. Cresswell, E. O. Long, P. Belhumeur, J. Kappler, P. Marrack, L. K. Denzin, and E. Cohen for plasmids, Abs, hybridomas, and cell lines.
Footnotes
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 This work was supported by a grant from the Canadian Institutes for Health Research. R.P.S. holds a Canadian Research Chair in human immunology. J.T. is supported by a senior fellowship from the Fonds de la Recherche en Santé du Québec.
2 Address correspondence and reprint requests to Dr. Jacques Thibodeau, Laboratoire d’Immunologie Moléculaire, Département de Microbiologie et Immunologie, Faculté de Médecine, Université de Montréal, CP 6128, Succursale Centre-ville, Montréal, Québec, Canada. E-mail address: jacques.thibodeau{at}umontreal.ca
3 Abbreviations used in this paper: SAg, superantigen; bSAg, bacterial SAg; DC, dendritic cell; ER, endoplasmic reticulum; Ii, invariant chain; KO, knockout; MMTV, mouse mammary tumor virus; MTV7, mammary tumor virus superantigen 7; SEB, staphylococcal enterotoxin B; vSAg7, viral SAg7.
Received for publication February 10, 2005. Accepted for publication April 12, 2005.
References
Papageorgiou, A. C., K. R. Acharya. 2000. Microbial superantigens: from structure to function. Trends Microbiol. 8: 369-375.
Marrack, P., J. Kappler. 1990. The staphylococcal enterotoxins and their relatives. Science 248: 705-711.
Fraser, J., V. Arcus, P. Kong, E. Baker, T. Proft. 2000. Superantigens: powerful modifiers of the immune system. Mol. Med. Today 6: 125-132.
Kotzin, B. L., D. Y. M. Leung, J. Kappler, P. Marrack. 1993. Superantigens and their potential role in human disease. Adv. Immunol. 54: 99-166.
Llewelyn, M., J. Cohen. 2002. Superantigens: microbial agents that corrupt immunity. Lancet Infect. Dis. 2: 156-162.
Thibodeau, J., I. Cloutier, P. M. Lavoie, N. Labrecque, W. Mourad, T. Jardetzky, R.-P. Sekaly. 1994. Subsets of HLA-DR1 molecules defined by SEB and TSST-1 binding. Science 266: 1874-1878.
Castellino, F., G. M. Zhong, R. N. Germain. 1997. Antigen presentation by MHC class II molecules: invariant chain function, protein trafficking, and the molecular basis of diverse determinant capture. Hum. Immunol. 54: 159-169.(Georges A. Azar*, Rafick-)
Activation of T lymphocytes by mouse mammary tumor virus superantigen (vSAg) requires binding to MHC class II molecules. The subcellular location where functional interactions occur between MHC class II molecules and vSAgs is still a matter of debate. To gain further insight into this issue, we have used human epithelial HeLa cells expressing HLA-DR1. Surprisingly, the human cells were unable to present transfected vSAg7 or vSAg9 to a series of murine T cell hybridomas. The defect is not related to a lack of vSAg processing, because these cells can indirectly activate T cells after coculture in the presence of B lymphocytes. However, after IFN- treatment, the HeLa DR1+ cells became apt at directly presenting the vSAg. Furthermore, transfection of CIITA was sufficient to restore presentation. Reconstitution experiments demonstrated the necessity of coexpressing HLA-DM and invariant chain (Ii) for efficient vSAg presentation. Interestingly, inclusion of a dileucine motif in the DR cytoplasmic tail bypassed the need for HLA-DM expression and allowed the efficient presentation of vSAg7 in the presence of Ii. A similar trafficking signal was included in vSAg7 by replacing its cytoplasmic tail with the one of Ii. However, sorting of this chimeric Ii/vSAg molecule to the endocytic pathway completely abolished both its indirect and direct presentation. Together, our results suggest that functional vSAgs-DR complexes form after the very late stages of class II maturation, most probably at the cell surface.
Introduction
Superantigens (SAgs)3 are small bacterial (bSAgs) or viral (vSAgs) proteins that bind MHC class II molecules and activate up to 20% of the T cell repertoire in a V-specific manner (for reviews, see Refs. 1, 2, 3). SAgs have long been implicated in various diseases as well as in autoimmune conditions (4, 5). They have proven to be useful tools for the appreciation of structural heterogeneity among class II molecules at the surface of APCs (6).
MHC class II heterodimers assemble in the endoplasmic reticulum (ER), where they swiftly associate with the invariant chain (Ii) (for reviews, see Refs. 7, 8, 9). This chaperone occupies the peptide binding groove of the newly synthesized class II molecules, preventing the binding of endogenous ER peptides and polypeptides (10, 11, 12). Ii is a type II membrane protein containing endosomal targeting signals through which the class II/Ii complex gets delivered to peptide-rich endosomes (13, 14, 15). In contrast to B cells, DCs from mice deficient for Ii express high levels of class II molecules at their surface (16, 17). Also, in the absence of Ii, class II molecules are efficiently expressed at the plasma membrane of transfected cells (18, 19). However, under these conditions, intracellular trafficking of class II molecules varies between cell lines (20).
Once in the endocytic pathway, proteinases sequentially digest Ii, except for the CLIP fragment, which remains in the peptide binding groove of the MHC class II molecule (reviewed in Ref. 21). The exchange of CLIP for an antigenic peptide that will be presented at the cell surface is promoted by HLA-DM (DM) (22, 23). This nonclassical class II molecule exhibits very little polymorphism and resides in the endocytic pathway (24, 25, 26). A cytoplasmic YXXL sorting motif allows DM to accumulate principally in late vesicles (27, 28, 29). The activity of DM is not restricted to CLIP removal, because numerous studies have highlighted its capacity to edit the global peptide repertoire by favoring those of higher affinity (30, 31, 32, 33). Beyond its effect on peptide editing, DM could also play an important direct role in fine-tuning the conformation of class II molecules (34, 35, 36).
Bacterial SAgs do not require processing and bind MHC class II molecules outside their peptide binding groove (37, 38, 39). However, it is now well established that the class II-bound peptides modulate the interaction of staphylococcal enterotoxin A, toxic shock syndrome toxin-1, and Mycoplasma arthritidis mitogen with their receptor (40, 41, 42, 43, 44, 45). Accordingly, chaperones involved in peptide binding to class II molecules can have a profound effect on SAg presentation (34, 46, 47). Bacterial SAgs bind to MHC class II molecules - and/or -chains in numerous fashions, and the three-dimensional structure of many complexes revealed direct contacts between SAgs and the peptide (48, 49, 50, 51)
The best-studied vSAgs are those produced by mouse mammary tumor virus (MMTV). Nevertheless, their crystal structure remains to be determined. MMTV is a type B retrovirus, transmitted horizontally through the milk of infected mothers or vertically once integrated in the germline (reviewed in Ref. 52). They produce vSAgs that are necessary for viral infectivity and virus spread (53, 54). These are type II transmembrane proteins encoded by an open reading frame in the 3' long terminal repeat (55, 56, 57, 58, 59). The mature MTV vSAg is a 45-kDa glycoprotein that needs to be activated through proteolytic processing (60). The nature of the active fragment remains ill defined; redundancy has been reported regarding the potency of the furin cleavage sites in activating vSAgs (61). Also, cathepsin L was shown to liberate a 27-kDa fragment, which could explain the activity of vSAgs in furin-deficient cells or after mutagenesis of the furin recognition motifs (62, 63). In addition to activating the vSAg, processing would explain the paracrine transfer of a soluble active C-terminal fragment to class II-positive APCs (64, 65, 66).
The exact binding site(s) of MMTV SAg on class II molecules remains obscure (67, 68, 69). Also, the intracellular trafficking of vSAgs and the site of class II encounter are still matters of debate. A thorough biochemical analysis of their maturation allowed Grigg et al. (70) to conclude that vSAgs reach the plasma membrane via the default exocytic pathway, independently of MHC class II expression. Class II-vSAg complexes were found exclusively at the plasma membrane and not in the endocytic pathway or the ER. Their observations imply an exclusive association of the vSAg with mature, peptide-loaded, SDS-stable class II molecules. Accordingly, it was later proposed that successful presentation of vSAg7 is conditional to the existence of a diverse, class II-bound peptide repertoire (71). Indeed, endogenous vSAg-induced T cell deletion was not taking place in modified mice expressing class II molecules almost exclusively loaded with CLIP (DM knockout (KO)) or loaded with a covalently linked peptide (E52–68).
These results contrast with those obtained in vitro by Hsu et al. (72), which suggest that class II-vSAg interactions occur in the ER through a CLIP-like region of vSAg occupying the peptide binding groove. In accordance with such an interaction early in the biosynthetic pathway, they reported that a DM-deficient human B cell line was as potent as its wild-type counterpart in stimulating T cells upon transfection of a vSAg.
In this study we show that DR-transfected cell lines differ dramatically in their ability to present MMTV vSAgs and that the mere expression of class II molecules is not sufficient to support presentation to T cell hybridomas. We report that CIITA expression rescues the presentation of vSAg7 by HeLa cells and that the presentation-competent phenotype is dependent on the coexpression of DR, Ii, and DM. The need for class II molecules to access the endocytic pathway coupled to the inactivation of vSAgs in these compartments support a model in which both ligands traffic independently inside the cell. Also, functional complexes probably form at the plasma membrane and only after class II molecules have acquired some presentation-permissive peptides.
Materials and Methods
Approximately 2 x 106 HeLa DR1 vSAg7 cells were cultured in medium supplemented with various concentrations of recombinant human IFN- (0.5–5 ng/ml; BD Pharmingen) for 5 days before T cell stimulation assays.
Results
HLA-DR1 expression by human epithelial cells does not support vSAg presentation
Transfected mouse fibroblasts expressing HLA-DR molecules are capable of vSAg presentation (95). Moreover, the responsive human V repertoire was similar to that described in the mouse (74). In the process of characterizing the molecular interactions between DR and SAgs, we generated vSAg7+ populations of DR1-transfected HeLa cells (HeLa DR1 vSAg7). To our surprise, these transfectants could not present the vSAg7 to the murine T cell hybridomas Kmls 13.11 (mV6) and KR3+ (mV8; Fig. 1A). The negative result obtained with HeLa was unexpected, because expression of the exact same class II cDNAs in DAP cells triggered a strong T cell response. We used three different APC:hybridoma ratios, and control DAP DR1 vSAg7 cells were effective at all concentrations.
We addressed the possibility that the MTV vSAg expression or activity was much reduced in human cells compared with mouse cells. Because the vSAg7-specific VS7 mAb staining was negative on both populations when analyzed by flow cytometry (data not shown), we capitalized on the ability of vSAg-producing cells to transfer the activity to vSAg-negative APCs (65, 66). Fig. 1B shows that in the presence of vSAg7-negative CH12 murine B cells, HeLa DR1 MTV7 cells can trigger a T stimulation as efficient as DAP DR1 vSAg7 cells. Although the exact mechanism remains to be characterized, such indirect presentation is probably the result of a successful paracrine transfer (64, 65, 66), and this strongly suggests that not only do HeLa cells express the vSAg, but it is also processed adequately.
We extended our initial observation to vSAg9 produced by MMTV9 (96). New independent transfections were obtained for both vSAg7 and vSAg9, and clones were picked. Compared with DR+ DAP cells expressing vSAg7 or -9, none of the vSAg+ HeLa clones could stimulate the specific hybridomas (Fig. 1C), whereas all triggered an efficient response in the indirect coculture assay (data not shown). These results confirm that HeLa DR1 cells cannot present vSAgs.
To rule out that this lack of presentation was due to the epithelial nature of HeLa cells and to determine whether this phenomenon could be observed in other cell types, we used the DR1+ M1 fibroblasts called 4N5 (88). Fig. 1C (left panel) shows that 4N5 cells stably transfected with vSAg7 were unable to stimulate the Kmls 13.11 T cell hybridoma.
A trivial explanation for the defect observed with HeLa would be that these cells do not express sufficient levels of class II molecules (97). However, surface stainings with the pan-DR L-243 mAb showed that both DAP and HeLa cells express significant and comparable levels of class II molecules (Fig. 1D). In addition, an independent vSAg+ DAP cell population expressing much lower levels of class II molecules triggers an efficient response from those T cell hybridomas (data not shown). Interestingly, although the DR expression is quantitatively very similar, the 2.06 mAb staining is much stronger on DAP cells (Fig. 1D). The 2.06 mAb binds to the -chain of all human class II isotypes, but would not recognize class II superdimers (98). The significance of this result is not clear, but it supports the idea that class II molecules are qualitatively somewhat different among various cell lines (6, 46, 89).
The presentation of vSAgs was suggested to be dependent on a diverse repertoire of class II-bound peptides (71). Because a fraction of the class II molecules in HeLa and DAP cells are bound with large polypeptides in the absence of Ii, we verified whether HeLa cells express some SDS-stable compact heterodimers containing short peptides and migrating at 55 kDa on immunoblots (10, 46, 99). Fig. 2A shows that nonboiled lysates prepared from both HeLa DR1 and DAP DR1 cells contain stable dimers recognized by the DR-specific XD5 mAb. Although such analysis does not give insights into the diversity of the peptide bound to DR, it rules out the absence of compact forms as a possible explanation for the behavior of HeLa cells. Importantly, DAP, HeLa, and M1 cells transfected with DR1 are all very efficient in stimulating the V8+ I1B3 murine T cell hybridoma in the presence of the bSAg staphylococcal enterotoxin B (SEB), ruling out a general Ag presentation defect (Fig. 2B).
IFN- rescues endogenous vSAg presentation by HeLa cells
The ability to present MTV SAgs is certainly not restricted to the mouse system, because class II-positive monkey (COS) and hamster (CHO) cells have been successfully used in the past (62, 66). Thus, we specifically addressed the role of the class II Ag processing machinery in vSAg presentation. The class II pathway is constitutively active only in professional APCs, but is strongly inducible in nearly all cells by IFN-. This pleiotropic cytokine affects all aspects of Ag processing and presentation, because it regulates the expression of >200 genes (100). We treated the inept HeLa DR1 vSAg7 cells for 5 days with IFN- and analyzed class II expression by flow cytometry (Fig. 3A). High levels of endogenous HLA-DM and -DQ were observed, whereas HLA-DR was clearly up-regulated due to expression of the homozygous endogenous HLA-DRB* 0102 genes (101). These cells were capable of endogenous vSAg7 presentation to Kmls 13.11 and KR3+ T cell hybridomas (Fig. 3B and data not shown). This increased presentation could be caused by up-regulation of critical molecules involved in either the class II maturation or the processing of vSAg. Although the latter possibility cannot be formally excluded, it does not concur with the fact that HeLa DR1 vSAg7 cells do not need this cytokine to produce functional vSAg7 and that IFN- did not increase presentation in the presence of BJAB (Fig. 3C).
Transfection of CIITA allows vSAg presentation by HeLa cells
As a first step in deciphering the mechanism of IFN- action and knowing that in its life cycle the MMTV primarily infects B lymphocytes, we tested endogenous presentation via the human Raji B cell line (54, 102, 103). These cells express all the class II processing machinery under control of the CIITA. The expression of this essential coactivator is constitutive in professional APCs and is induced by IFN- in almost all other cells (104). The vSAg7-encoding construct was electroporated in Raji cells, and total mRNA from the stably transfected clone 2H8 was analyzed for SAg expression. Fig. 4A shows that overall, the B cell transfectant expresses much less vSAg7-specific mRNA than do HeLa DR1 cells. However, functional assays showed that Raji cells are fully capable of presenting the vSAg7 to the two T cell hybridomas (Fig. 4B). Viral SAg surface expression was detected on Raji 2H8 by flow cytometry using the VS7 mAb (Fig. 4C). We were unable to detect vSAg7 expression on HeLa cells. However, other potent cell lines present vSAg7, but do not stain with the VS7 mAb (see below), suggesting that many factors contribute to the display of the epitope. Taken together, these results suggest that besides class II molecules, expression of additional genes under the control of CIITA is required for vSAg presentation.
To narrow down the list of molecules capable of restoring efficient presentation, we supertransfected CIITA into vSAg7-expressing HeLa cells. These cells became fully apt at presenting the endogenous vSAg to Kmls 13.11 (Fig. 4D). To relate to our HLA-DR transfectants and given the fact that CIITA up-regulates the expression of all classical class II isotypes (Fig. 3A), it became important to assess the relative importance of DR for the presentation of vSAg7 by HeLa CIITA cells. The anti-DR mAb L-243 almost completely inhibited T cell activation in a dose-dependent manner (Fig. 4E). DP- and DQ-specific mAbs did not affect the level of stimulation, confirming that these isotypes do not efficiently present vSAgs (95) (data not shown). Taken together, these results suggest a direct implication of MHC class II Ag processing machinery in vSAg presentation.
Coexpression of Ii and HLA-DM is sufficient to restore vSAg presentation by HeLa DR1+ cells
CIITA was shown to affect the expression of a large panel of genes implicated in diverse functions, such as cell migration, differentiation, and proliferation (105). Although many different mechanisms could potentially contribute to the CIITA-mediated enhanced vSAg presentation, we focused our efforts on the role of the intracellular class II chaperones. HLA-DM and Ii are strongly up-regulated by CIITA and play a crucial role in the trafficking of class II molecules and in the selection of their peptide repertoire. First, we stably transfected DM into HeLa DR1 vSAg7 (HeLa DR1 DM vSAg7). Even though vSAg7 transfer to BJAB remained high, this new population could not directly activate the two T cell hybridomas (Fig. 5, A and B). Similar results were obtained using cells transfected with Ii and sorted for high levels of CLIP surface expression (HeLa DR1 Ii vSAg7; Fig. 5C). However, when Ii and DM were cotransfected (HeLa DR1 Ii DM vSAg7), the stimulation increased dramatically for both T cell hybridomas used (Fig. 5A). An efficient stimulation was also observed using independent transfections where vSAg7 and DM were cotransfected into HeLa DR1 Ii cells (data not shown). Together, these results strongly point to a role of class II-bound peptides in vSAg presentation.
An ectopic endosomal sorting motif in the tail of DR rescues vSAg presentation
In light of the functional role of HLA-DM, we hypothesized that class II molecules must travel to the endocytic compartments to acquire the capacity to present vSAgs. Sorting of class II to endosomes is assured by Ii, which bears two leucine-based motifs in its cytoplasmic tail (15, 106). We reasoned that the need for Ii could be bypassed by introducing such a signal in the tail of HLA-DR. To achieve this, we replaced the cytoplasmic tail of DR by that of HLA-DO. This chimeric molecule (DR1-LL) contains a functional dileucine motif, which results in accumulation not only at the plasma membrane, but also in lysosome-associated membrane protein-1+ intracellular compartments (77). However, HeLa DR1-LL vSAg7 cells did not stimulate T cell hybridomas despite the efficient transfer of the activity to a B cell line (Fig. 6). Interestingly, coexpression of Ii in these cells allowed the efficient direct presentation of vSAg7. These cells express class II levels similar to their nonstimulating counterpart (Fig. 6A). Also, the cell surface staining with CerCLIP.1 demonstrates that they express copious amounts of CLIP-class II complexes. These results confirm that class II molecules must travel to late endocytic compartments before acquiring the ability to present vSAgs.
Sorting vSAg7 to the endocytic pathway abrogates its activity
Given the importance of targeting the class II molecules to the endocytic pathway, we investigated the possibility of increasing the efficiency of presentation by sorting the vSAg7 to endosomes. Just like Ii, the endogenous MMTV superantigens are type II transmembrane proteins (59). Thus, we swapped the cytoplasmic tail of vSAg for that of Iip33 or p35 and tested the capacity of these chimeric molecules to activate T cells. For these experiments, we used DAP cells, which are fully competent for both direct presentation and transfer of vSAgs. Surprisingly, the stably transfected p33 cyto-vSAg7 and p35 cyto-vSAg7 chimeric molecules could not stimulate T cells even when cocultured in the presence of BJAB (Fig. 7A). Using the Ii-specific Pin.1 mAb on immunoblots, we found that the chimeric molecules migrated at the expected Mr of 45 kDa, and no degradation products were detected (Fig. 7B and data not shown). The chimeric molecules were almost undetectable by flow cytometry and fluorescence microscopy, suggesting that the vSAg portion imparts instability to the constructs (data not shown).
To circumvent the problem of low protein expression, we turned to transient overexpression in human HEK 293T cells. Two days after transfection in the absence of class II molecules, cells were split and tested in a functional transfer assay (data not shown) or were analyzed by fluorescence microscopy and flow cytometry (Fig. 7, C and D). In contrast to wild-type vSAg7, the chimeric molecules were not transferred from 293T cells to BJAB. Pin.1 staining showed that the chimeric p33 cyto-vSAg7 molecule was expressed and transported to punctate vesicles scattered in the cytoplasm. Importantly, to demonstrate the specificity of this distribution, we tested a similar construct between Iip35 and vSAg7 (p35 cyto-vSAg7). As expected from the fact that this Ii isoform is retained in the ER in the absence of DR (107), the staining was diffuse and did not show any defined vesicles (Fig. 7C). Together, these results suggest that the small amount of vSAg destined for T cell stimulation must not be sorted to the endocytic pathway even in the presence of class II molecules.
Discussion
Structure-function aspects of SAgs have been mostly studied in the past using MHC class II-transfected cells. For example, by analogy to the hierarchy described for the H-2 gene products, murine DAP cells have been used to demonstrate isotypic variations in the capacity of human class II molecules to present vSAgs (95). These cells do not express Ii or DM, but still present vSAgs very efficiently when transfected with HLA-DR. However, the situation was very different when we used a series of HeLa cell transfectants for study of the vSAg binding site on those exact same class II molecules. In such conditions, we found that the human epithelial cell line is incapable of presenting endogenous vSAgs. It has been reported that MMTV preferentially infects immature DCs as well as B cells, and that both APCs lead to V-specific T cell activation (108). Accordingly, when the prototypic human B cell line Raji was tested for endogenous vSAg7 presentation, efficient T cell stimulation was measured. We identified CIITA as being responsible for activating the class II processing pathway and for the efficient vSAg presentation. This effect was mainly due to the up-regulation of HLA-DM and Ii. From our results, we cannot rule out additional CIITA-mediated effects on the processing or sorting of vSAgs, because this transactivator was shown to modulate the expression of >50 genes in B cells (105). However, given the fact that all other CIITA-negative vSAg+ HeLa transfectants very efficiently transferred the activity, this hypothesis is unlikely. Comparing the ability of primary B cells from CIITA– vs CIITA+class II– mice to transfer vSAg might shed light on this issue.
The presentation efficiency of HeLa CIITA and Raji most certainly relies in part on their very high levels of surface class II compared with the DR/DM/Ii transfectants. Also, the relative expressions of class II, Ii, and DM is probably optimal under CIITA control compared with the situation in HeLa cells where transfected molecules are transcribed from a variety of promoters and by different vectors integrated randomly at various copy numbers.
Our results clearly showed that the expression of DM and Ii was sufficient to rescue vSAg presentation in HeLa DR1 cells. This finding is reminiscent of the important peptide/class II conformation for the binding of bacterial SAgs (44). For example, TSST-1 binding to class II is greatly augmented by the transfection of Ii and DM in DR+ T2 cells (41, 46). Indeed, Wen et al. (42, 43) demonstrated unambiguously that TSST-1 binding is affected by the class II-bound peptides. Interestingly, it was recently shown that the presentation of both vSAgs and TSST-1 is diminished in mice deficient for DM or Ii, suggesting that a diverse class II peptide repertoire is needed for these SAgs to bind APCs (71). At first glance, our results are in line with these in vivo studies. At the surface of chaperone-negative HeLa DR cells, the class II molecules are loaded with large ER polypeptides that could interfere with binding of vSAgs (46). However, polypeptides are also found in class II molecules expressed at the surface of DAP cells (10). Nevertheless, in both cell lines there are clearly some SDS-stable compact forms migrating at 55 kDa, which most certainly correspond to class II molecules loaded with some relatively short peptides (Fig. 2). Cell type differences may explain the fact that mouse cells display some permissive peptide(s) or polypeptide(s) allowing vSAg presentation. Mass spectrometry analyses will give clues about the degree of heterogeneity of these class II ligands.
The need for a diverse peptide repertoire could explain why coexpression of Ii, which displaces large polypeptides found in the ER, did not restore vSAg presentation in DR+ HeLa cells. For DM KO mice, this configuration results in the display of class II molecules bound almost exclusively with the CLIP peptide. Introducing DM in these cells could restore presentation through CLIP displacement and the binding of vSAg-permissive peptides. However, Hsu et al. (72) reported that DM-negative B cells expressing copious amounts of DR3-CLIP complexes are still capable of presenting endogenous vSAgs. Also, the surprising finding reported here that a dileucine motif in the tail of DR could restore presentation warrants evaluating alternative explanations for the need of displacing CLIP. Indeed, the majority, if not the totality, of class II molecules on the HeLa DR1-LL Ii cells are loaded with the CLIP peptide and present vSAgs. Because both lines are expressing DR-CLIP at their surface, what differentiates these complexes between HeLa DR Ii and HeLa DR1-LL Ii cells? Possibly the isoform of CLIP found on HeLa DR Ii cells (or on APCs from DM KO mice) is not permissive for vSAg binding. CLIP is usually found as a nested set of peptides on class II molecules, and N- or C-terminal extensions outside the peptide binding groove have been shown to inhibit SEB binding to class II molecules (109). HeLa cells, which are devoid of cathepsin S (110), may generate a form of CLIP incompatible with vSAg binding. This may explain why replacing in Ii the core CLIP sequence with a variety of defined DR-binding motifs did not restore vSAg binding (data not shown). To allow presentation, nonpermissive CLIPs must be either displaced by DM or, in the context of the DR1-LL construct, be trimmed to competent forms by lysosomal exopeptidases. We have previously shown that this chimeric DR1-LL accumulates in large quantities in endocytic vesicles containing markers of late endosomes (77). Despite a prolonged stay in these vesicles, these molecules recycle, because they are found in large amounts at the cell surface. HPLC analysis of the peptides found in DR vs DR1-LL would help resolve the issue of CLIP heterogeneity. Alternatively to the trimming of CLIP, the trafficking of DR1-LL through late vesicles might allow binding of some functionally relevant peptides after long-standing exposure to acidic pH. Such a mechanism could explain the indispensable need for Ii, which will incorporate in DR the more easily exchangeable CLIP peptide. In this context, it will be interesting to test the presentation of vSAg by HeLa cells expressing Ii and an allele such as DR4, which releases CLIP spontaneously (111). Whether DR1-LL could also bypass the need for DM in the presentation of DM-dependent nominal Ags remains to be tested.
Although its influence seems clear in many experimental systems, it is still possible that DM plays a role beyond its peptide-editing activity. Looking at bacterial superantigen, Ab, or TCR recognition of class II molecules, recent evidence strongly suggests that DM might directly influence the structure of class II molecules (34, 35, 36). However, the mechanism underlying this new DM attribute remains unknown. In the case of vSAg presentation by HeLa cells, a contact-induced, peptide-independent role of DM on the structure of DR would be difficult to reconcile with the fact that an LL motif on DR compensates for the absence of the nonclassical class II chaperone. Also, the cell type dependency for DM is intriguing. If one considers a role for DM on the structure of class II molecules, then what could be compensating for DM and Ii in DAP, COS, and CHO cells, which all stimulate T cells upon transfection of vSAgs and DR? The most likely explanation is that in these permissive cells, HLA-DR could undergo a conformation modification mediated by some enzymes, tetraspans, chaperones, or severe pH conditions. For example, as opposed to HeLa cells, murine cells allow class II molecules to accumulate in the endocytic pathway in the absence of Ii (20). In nonpermissive cells, DM could affect the sorting of DR by dragging some molecules to the highly acidic lysosomes before their dissociation. Nevertheless, our observations describe the essential role of the endocytic pathway in the presentation of endogenous vSAgs. These results are in line with those reported by Golovkina et al. (71) showing the importance of peptides in vSAg presentation and those of Grigg et al. (70) showing that vSAgs interact with mature class II molecules and only at the cell surface.
Our results contrast with those of Hsu et al. (72) describing a DM-independent vSAg presentation pathway. The straightforward model that they proposed suggests that vSAgs bare an essential and specific motif that interacts with newly formed class II heterodimers in a CLIP-like manner. We cannot formally rule out that DR and vSAgs associate in the ER of HeLa cells in the absence of Ii and DM, but these complexes are nonfunctional. Also, vSAg molecules following DR (especially DR-LL) to endosomes would probably be degraded, as previously suggested (70). Indeed, replacing the cytoplasmic tail of vSAg for that of Iip33 completely abolished the ability of the vSAg to stimulate T cells in direct or transfer assays (Fig. 7). This last result suggests that if vSAgs gain access to the endocytic pathway, their passage must be swift and is probably not compatible with their intercalation in a putative class II-Ii-MTV nonamer. In furin-negative cells, it was proposed that alternative cleavage by cathepsins could generate an active form of vSAg (62, 63). HEK 293T cells are furin negative, and the efficient transfer of vSAg is in accordance with the existence of alternative cleavage sites. However, the fact that a soluble fragment is not released from cells expressing the chimeric Ii-vSAg suggests that the cleavage does not occur in endosomes. Alternatively, the transferred C-terminal part of the open reading frame product may need to remain associated to the N-terminal part of the vSAg and be re-expressed at the surface, a step that is precluded by the endosomal retention motif in our chimera. Finally, the binding of vSAg in an Ii-like fashion would imply a direct competition between the two molecules for class II binding. However, DAP cells expressing high levels of Ii20 (which lacks endosomal sorting motifs and stays associated with DR from the ER to the cell surface) stimulate T cells as efficiently as their Ii-negative counterpart (J. Thibodeau and R.-P. Sékaly, unpublished observations). In conclusion, our results support a model in which vSAgs travel on their own to the plasma membrane, where they encounter mature class II molecules. Binding and presentation of the vSAg depend on the nature of the stably bound class II peptides, which are acquired in various compartments and may be homogeneous under certain experimental conditions.
Acknowledgments
We thank Alexandre Brunet for helping with the characterization of chimeric molecules. We also thank Drs. F. Denis, G. Winslow, O. Bakke, D. L. Hardie, I. C. MacLennan, P. Cresswell, E. O. Long, P. Belhumeur, J. Kappler, P. Marrack, L. K. Denzin, and E. Cohen for plasmids, Abs, hybridomas, and cell lines.
Footnotes
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 This work was supported by a grant from the Canadian Institutes for Health Research. R.P.S. holds a Canadian Research Chair in human immunology. J.T. is supported by a senior fellowship from the Fonds de la Recherche en Santé du Québec.
2 Address correspondence and reprint requests to Dr. Jacques Thibodeau, Laboratoire d’Immunologie Moléculaire, Département de Microbiologie et Immunologie, Faculté de Médecine, Université de Montréal, CP 6128, Succursale Centre-ville, Montréal, Québec, Canada. E-mail address: jacques.thibodeau{at}umontreal.ca
3 Abbreviations used in this paper: SAg, superantigen; bSAg, bacterial SAg; DC, dendritic cell; ER, endoplasmic reticulum; Ii, invariant chain; KO, knockout; MMTV, mouse mammary tumor virus; MTV7, mammary tumor virus superantigen 7; SEB, staphylococcal enterotoxin B; vSAg7, viral SAg7.
Received for publication February 10, 2005. Accepted for publication April 12, 2005.
References
Papageorgiou, A. C., K. R. Acharya. 2000. Microbial superantigens: from structure to function. Trends Microbiol. 8: 369-375.
Marrack, P., J. Kappler. 1990. The staphylococcal enterotoxins and their relatives. Science 248: 705-711.
Fraser, J., V. Arcus, P. Kong, E. Baker, T. Proft. 2000. Superantigens: powerful modifiers of the immune system. Mol. Med. Today 6: 125-132.
Kotzin, B. L., D. Y. M. Leung, J. Kappler, P. Marrack. 1993. Superantigens and their potential role in human disease. Adv. Immunol. 54: 99-166.
Llewelyn, M., J. Cohen. 2002. Superantigens: microbial agents that corrupt immunity. Lancet Infect. Dis. 2: 156-162.
Thibodeau, J., I. Cloutier, P. M. Lavoie, N. Labrecque, W. Mourad, T. Jardetzky, R.-P. Sekaly. 1994. Subsets of HLA-DR1 molecules defined by SEB and TSST-1 binding. Science 266: 1874-1878.
Castellino, F., G. M. Zhong, R. N. Germain. 1997. Antigen presentation by MHC class II molecules: invariant chain function, protein trafficking, and the molecular basis of diverse determinant capture. Hum. Immunol. 54: 159-169.(Georges A. Azar*, Rafick-)