CD8 Binding to MHC Class I Molecules Is Influenced by T Cell Maturation and Glycosylation

Immunity, Vol. 15, 1051–1061, December, 2001, Copyright 2001 by Cell Press

CD8 Binding to MHC Class I Molecules Is Influenced by T Cell Maturation and Glycosylation Mark A. Daniels,1 Lesley Devine,2 Joseph D. Miller,3 Janice M. Moser,4 Aron E. Lukacher,4 John D. Altman,3 Paula Kavathas,2 Kristin A. Hogquist,1 and Stephen C. Jameson1,5 1 Center for Immunology Department of Lab Medicine and Pathology University of Minnesota Medical School Minneapolis, Minnesota 55455 2 Department of Laboratory Medicine Section of Immunobiology Yale University School of Medicine New Haven, Connecticut 06520 3 Emory Vaccine Center and Department of Microbiology and Immunology 4 Department of Pathology Emory University School of Medicine Atlanta, Georgia 30322

CD8 serves both as an adhesion molecule for class I MHC molecules and as a coreceptor with the TCR for T cell activation. Here we study the developmental regulation of CD8-mediated binding to noncognate peptide/MHC ligands (i.e., those not bound by the TCR). We show that CD8’s ability to bind soluble class I MHC tetramers and to mediate T cell adhesion under shear flow conditions diminishes as double-positive thymocytes mature into CD8ⴙ T cells. Furthermore, we provide evidence that this decreased CD8 binding results from increased T cell sialylation upon T cell maturation. These data suggest that CD8’s ability to interact with class I MHC is not fixed and is developmentally regulated through the T cell’s glycosylation state.

to be upregulated during T cell activation (Kane and Mescher, 1993; Luescher et al., 1995; O’Rourke et al., 1990). In addition to its role in mature T cell responses, a variety of experiments suggest that CD8 interactions (both cognate and noncognate) can play an important role in positive and negative selection of class I restricted thymocytes (Fung Leung et al., 1991; Ingold et al., 1991; Kirberg et al., 1994; Matechak et al., 1996). The affinity of mouse CD8 for MHC class I is low (Garcia et al., 1996; Kern et al., 1998), precluding analysis of this binding using soluble monomeric MHC ligands on live cells. This has lead to the recent use of tetramerized forms of MHC class I molecules to study interactions between CD8 and class I MHC (Bosselut et al., 2000) and between CD8, TCR, and class I MHC (Block et al., 2001; Daniels and Jameson, 2000) by flow cytometry. Intriguingly, the report of Bosselut et al. (2000) suggested TCR-independent binding of class I MHC tetramers to CD8 on thymocytes, while we and others have observed little evidence for CD8 binding to class I tetramers in the absence of a cognate TCR interaction on mature T cells (Altman et al., 1996; Busch et al., 1998; Daniels and Jameson, 2000; Murali-Krishna et al., 1998). In order to explain this discrepancy, we sought to determine whether binding of class I MHC tetramers to CD8 changes during T cell maturation. Here we show the ability of CD8 to bind class I MHC ligands diminishes as T cells develop from immature double-positive (DP) to mature CD8 SP cells. Further experiments suggested that a critical factor in this maturation effect is the glycosylation state of the developing thymocyte—specifically that sialylation inhibits CD8’s ability to bind to class I ligands. We also show that these effects drastically alter the ability of CD8 to mediate T cell adhesion to class I molecules under shear flow conditions.

Introduction

Results

CD8 is involved in recognition of class I MHC molecules by developing thymocytes and mature cytotoxic T lymphocytes (CTL) (Janeway, 1992; Miceli and Parnes, 1993; Zamoyska, 1994). On murine T cells, CD8 is typically a heterodimer consisting of ␣ and ␤ chains, although it can be expressed as an ␣␣ homodimer on a distinct subset of lymphocytes (Zamoyska, 1994). CD8 can act as a coreceptor, participating in the “cognate” recognition of peptide-MHC I by the TCR (Janeway, 1992; Miceli and Parnes, 1993; Zamoyska, 1994). Such coreceptor functions of CD8 may involve both stabilization of the TCR/MHC-peptide complex (Garcia et al., 1996) and signal amplification by virtue of CD8 association with Lck (Miceli and Parnes, 1993; Thome et al., 1996), LAT (Bosselut et al., 1999), and lipid rafts (Arcaro et al., 2000). Alternatively, CD8 can bind class I molecules independently of the TCR (“noncognate” class I binding) and in this way can act as an adhesion molecule (Norment et al., 1988), a property which has been shown

The Nature of the Noncognate Tetramer/CD8 Interaction Bosselut and colleagues showed that H-2Dk tetramers can bind thymocytes through direct interaction with CD8␣␤ (Bosselut et al., 2000). We had not observed similar H-2Kb tetramer binding to CD8 alone in our studies of mature peripheral CD8⫹ T cells, although a role for CD8 in TCR cognate interactions could clearly be seen (Daniels and Jameson, 2000). To account for these discordant results, we considered that either the MHC allele and/or the developmental stage of the T cell influenced the capacity of CD8 to bind noncognate MHC class I. Accordingly, we stained B6 thymocytes with a titration of increasing concentrations of Dk/MT and Kb/OVA tetramers. Both tetramers showed a dose-dependant staining of bulk thymocytes (mostly CD4⫹8⫹ DP thymocytes) but not of CD4⫹8⫺ (CD4SP) thymocytes (Figure 1A). Dk/MT staining was consistently higher than that seen with Kb/OVA tetramer, and since it has previously been shown there is a direct correlation between the level of tetramer staining and the affinity for its ligand (Crawford et al.,

Summary

5

Correspondence: [email protected]

Immunity 1052

Figure 1. The Nature of the Noncognate Tetramer/CD8 Interaction (A and B) B6 thymocytes were stained with increasing concentrations of Kb/OVA (䉬) and Dk/MT (䊏) tetramers in the absence (A) and presence (B) of CD4 and CD8 (53-6.7) antibodies. In (B), the staining is shown for the CD4⫹8⫹ DP population. Dk/MT tetramer staining on CD4⫹ SP cells (䉱) is shown in both panels as the negative control. (C–E) B6 thymocytes were stained with Kb/OVA (C), Dk/MT (D), and Kb/SIY (E) tetramers, with or without CD4 and CD8 antibodies. Tetramer staining is shown for the total thymocyte population (mostly DP thymocytes) without antibodies (dotted histogram) or with the antiCD8 antibodies 53-6.7 (open) or with CTCD8a (black). Tetramer staining of CD4⫹ SP cells (gray) is shown as a negative control.

1998; Daniels and Jameson, 2000), this may mean that CD8 has a higher affinity for Dk than Kb. Nevertheless, thymocytes demonstrated detectable binding to tetramers of both allelic forms. Thymocyte subsets are classically defined using CD4 and CD8 antibodies. Therefore, to determine the thymocyte populations which bound these tetramers, we first needed to test the impact of CD8 antibodies on tetramer binding. In previous work, we showed that some CD8␣ and CD8␤ antibodies block cognate tetramer staining, while others have no effect or even enhance specific binding (Daniels and Jameson, 2000). Indeed, the CD8␣specific antibody 53-6.7, which we and others have shown stabilizes TCR/CD8/MHC interactions (Daniels and Jameson, 2000; Luescher et al., 1995), dramatically enhanced binding of both Dk and Kb tetramers while preserving the hierarchy of binding seen in the absence of CD8 antibodies (Figures 1B–1E). In contrast to the enhancing effect of 53-6.7, the CD8␣-specific CT-CD8a antibody inhibited tetramer staining (Figures 1C–1E). Likewise, anti-CD8␤ antibodies KT-112 enhanced noncognate binding to CD8, and 53-5.8 blocked noncognate CD8 binding (data not shown). These patterns of augmentation and inhibition are identical to those we reported for tetramer binding to TCR-specific T cells (Daniels and Jameson, 2000). It should be stressed, though,

that the T cells studied here are polyclonal thymocytes and, as such, tetramer binding is unlikely to involve any contribution by a specific TCR. Indeed, we also found that such staining was observed on TCR␣⫺/⫺ DP thymocytes, which cannot express a mature ␣␤ TCR (see below), and on COS fibroblast cells transfected with mouse CD8␣␤ (L.D. and P.K., unpublished data). Noncognate Tetramer Binding Decreases with Thymocyte Development The ability of thymocytes to bind noncognate Kb-based tetramers is in stark contrast to what we have previously shown for peripheral CD8⫹ T cells, in which we showed that tetramer staining faithfully reflected TCR fine specificity, even in the presence of saturating amounts of enhancing anti-CD8 antibodies (Daniels and Jameson, 2000). Consequently, we explored whether noncognate MHC I/CD8 binding changed during T cell development. B6 thymocytes were incubated with anti-CD4, antiCD8␣ (53-6.7), and Kb/OVA tetramer and gated as shown in Figure 2A. CD4 SP thymocytes did not bind class I MHC tetramers and were used to establish the negative control (Figures 2B–2F). DN thymocytes were unable to bind Kb/OVA tetramers as would be expected because they lack CD8 expression (Figure 2B). Conversely, DP thymocytes displayed substantial binding of Kb/OVA tet-

Glycosylation Affects CD8 Binding to Class I MHC 1053

Figure 2. Noncognate Tetramer Binding Decreases with Thymocyte Development B6 thymocytes were stained with CD4, CD8 (53-6.7) antibodies, and Kb/OVA tetramer. (A) The thymocyte subsets defined by CD4 and CD8 staining are listed as “B” through “F”. (B–F) Tetramer staining in each of these subsets is indicated (open histogram) with staining of the CD4 SP subset acting as negative control (gray histogram).

ramer (Figure 2C). As DP thymocytes go through selection, they downregulate both coreceptors, becoming CD4loCD8lo, and this population bound the Kb/OVA tetramer with much less efficiency than DP thymocytes (Figure 2D), perhaps due to their lower expression of CD8. Similarly, CD4⫹8lo intermediate thymocytes were unable to efficiently bind Kb/OVA tetramer (Figure 2E). Remarkably, CD4⫺, 8ⴙ (CD8 SP) bound tetramer poorly (Figure 2F), despite the fact that these cells display at least as much CD8 on their cell surface as DP thymocytes (Figure 2A). These data suggest developmental regulation of CD8/class I MHC interactions. It was formally possible that the change in Kb tetramer staining of B6 thymocytes during T cell maturation was due to Kb-specific negative or positive selection. This was ruled out by examination of Kb/OVA tetramer staining of thymocytes from H-2d and H-2k haplotype mice (Figure 3). The lower tetramer staining on mature CD8 T cells when compared to immature CD8 DP cells was similar in all of these strains. Furthermore, similar staining profiles were obtained using other Kb or Dk/MT tetramers regardless of the H-2 haplotype of the mouse strain, although, as in Figure 1, Dk tetramer staining was generally more intense than Kb tetramer staining (data not shown). These data suggest the binding is independent of TCR repertoire and/or peptide/MHC-specific in-

teractions, as would be expected for CD8 noncognate class I binding. Positive Selection of DP Thymocytes Diminishes CD8/Class I Binding Next we asked whether CD8/class I interactions were altered by thymic positive selection of the DP population. Hence we compared the noncognate tetramer staining of DP thymocytes from three mouse strains: TCR␣⫺/⫺, OT-I, and OT-I TAP⫺/⫺. DP thymocytes develop in TCR␣⫺/⫺ mice but cannot undergo positive selection, since they lack a mature ␣␤ TCR (Mombaerts et al., 1992). At the other extreme, positive selection among OT-I T cells on a B6 background is highly efficient, due to the transgenic ␣␤ receptor. In contrast, OT-I CD8⫹ T cell positive selection is markedly reduced in the TAP⫺/⫺ background, due to the lack of selecting class I MHC ligands (Hogquist et al., 1997). Intriguingly, Dk/MT tetramer staining intensity followed the series OT-I ⬍ OT-I TAP⫺/⫺ ⬍ TCR␣⫺/⫺ (Figure 4A), and a similar hierarchy was seen using Kb/SIY tetramers (data not shown). In similar experiments, tetramer staining of wildtype B6 DP thymocytes was similar to that of TCR␣⫺/⫺ DPs (data not shown). These data suggested first, that CD8 binding to noncognate ligands was not equivalent for all DP cells and second, that tetramer staining inten-

Immunity 1054

Figure 3. Noncognate Tetramer Binding Is Not Influenced by Thymocyte MHC Haplotype Thymocytes and mature lymph node T cells from B6 (A), BALB/c (B), and C3H (C) mice were stained with Kb/OVA tetramer in the presence of antibodies to CD4 and CD8␣ (53-6.7). In each case, Kb/OVA tetramer staining is show for DP (open histogram), CD8⫹ SP (dotted histogram), and CD4⫹ SP thymocytes (gray histogram), and for CD8⫹ lymph node cells (black histogram).

sity inversely correlated with the induction of thymic positive selection among the DP populations. Similar patterns of tetramer staining were observed with DP thymocytes from two other class I-restricted TCR transgenic strains (H-Y and 2C) and a class II MHC restricted TCR transgenic strain (DO11.10): each transgenic DP population stained less intensely than wild-type thymocytes with noncognate tetramers (data not shown). Again, these data suggest that positive selection negatively influences the ability of CD8 to bind noncognate class I directly. How might thymic maturation in general and positive selection in particular alter CD8’s ability to engage class I MHC? One well-established change that occurs as DP thymocytes undergo positive selection and maturation is an increase in levels of surface TCR (Robey and Fowlkes, 1994). We see this progression in the analysis of the three strains above with CD3 levels following the hierarchy TCR␣⫺/⫺ ⬍ OT-I TAP⫺/⫺ ⬍OT-I, i.e., opposite to the hierarchy for tetramer staining (Figure 4B). One could hypothesize that increased TCR expression levels might have a detrimental effect on direct CD8/class I interactions, perhaps by physically restraining CD8’s ability to bind noncognate ligands. Another change reported to occur during thymic selection is increased cell surface sialylation. This is reflected by decreased staining with PNA as T cells mature (Fowlkes et al., 1980; Reisner et al., 1976). PNA is a galactose binding lectin which can bind unsialylated Gal␤1-3GalNAc␣-Ser/Thr (core-1) O-glycans (Pereira et al., 1976; Varki et al., 1999). More specifically, it has been shown that CD8 itself undergoes changes in glycosylation during development in the thymus and following

Figure 4. DP Populations from Different Strains of Mice Show Variability in CD8 Binding to Noncognate Class I Tetramers Thymocytes from TCR␣⫺/⫺ (black histogram), OT-I (open), and OT-I TAP⫺/⫺ (gray) mice were stained with Dk/MT tetramers (A), anti-CD3⑀ antibody (B), and/or PNA (C) simultaneously with CD4 and CD8 (536.7) antibodies. Staining is shown for CD4⫹8⫹ DP thymocytes in each case.

activation of mature CD8 T cells, and at least some of these differences relate to sialylation (Casabo et al., 1994; Wu et al., 1996). Indeed, PNA staining shows a hierarchy (albeit subtle) in the order OT-I ⬍ OT-I TAP⫺/⫺ ⬍ TCR␣⫺/⫺, matching the hierarchy of tetramer staining (Figure 4C). Therefore, further experiments were performed to test whether TCR levels and/or surface sialylation contributed to the changes in CD8/class I binding observed during thymocyte development.

The Role of Cell Surface TCR Expression Level on CD8/Class I Binding There was an impressive inverse correlation between TCR levels and noncognate tetramer staining (Figure 4). If elevated TCR expression inhibited CD8/class I noncognate binding, we would expect that downregulation of surface TCRs on mature thymocytes would lead to an increase in CD8 binding. To test this, OT-I thymocytes were induced to downregulate TCR levels by antigendriven internalization, and their binding to Dk/MT tetra-

Glycosylation Affects CD8 Binding to Class I MHC 1055

CD8⫹ T cells was observed after neuraminidase treatment, reflecting desialylation (Figures 6A and 6B). PNA staining also increased on neuraminidase-treated immature DP (Figures 6A and 6B), as has been reported previously (Wu et al., 1996), indicating some low-level cell surface sialylation on this population. Remarkably, neuraminidase treatment led to a dramatic increase in noncognate class I tetramer staining among the CD8⫹ populations. Staining of desialylated CD8⫹ lymph node cells with Dk/MT tetramer reached levels similar to those on untreated DP thymocytes (Figure 6C), while neuraminidase treatment also enhanced tetramer binding on the DP population (Figure 6D). Similar results were obtained with both Kb/OVA and Kb/SIY tetramers (data not shown). Two controls indicate that this tetramer binding involves CD8 engagement rather than some form of nonspecific binding to desialylated lymphocytes. First, tetramer staining did not appear on the CD4 SP population after desialylation (Figure 6E). Second, tetramer staining on neuraminidase-treated CD8⫹ mature cells and DP thymocytes was completely blocked by the anti-CD8␣ antibody, CTCD8a (Figures 6C and 6D). Together, these data suggest a role for sialylation in the ability of CD8 to bind to MHC class I. Figure 5. Induced TCR Downregulation Does Not Affect CD8/Class I Interactions OT-I and B6 thymocytes were cultured alone or in the presence of 10 ␮M OVA peptide. (A and B) The cells were stained for CD4, CD8, and CD3⑀. CD3⑀ (A) and CD8 (B) staining is shown for OT-I CD8SP after OVAp (open histogram) or control (dotted) culture and for OVAp-treated B6 DP cells (gray). (C) The same populations of cells were stained for CD4, CD8 (53-6.7), and Dk/MT tetramer. Staining of OVAp-treated B6 DP thymocytes is shown but was identical to untreated B6 DPs (data not shown).

mers was measured (Figure 5). Using this approach, TCR expression on OT-I CD8⫹ SP thymocytes was decreased to levels seen on B6 DP thymocytes (Figure 5A), while CD8 expression levels were unchanged (Figure 5B). However, this extensive TCR downregulation resulted in little, if any, enhancement of CD8 binding to noncognate class I tetramers (Figure 5C). Similar results were obtained when OT-I DP populations (which also showed extensive TCR downregulation) were analyzed (data not shown). These data imply that while TCR expression per se may influence CD8 noncognate binding, it does not fully account for the change in binding we observed with T cell maturation and among DPs from different mouse strains. The Role of T Cell Glycosylation in CD8/Class I Interactions To test whether differential sialylation patterns affect noncognate tetramer binding, thymocytes and lymph node T cells from B6 mice were treated with neuraminidase, which cleaves sialic acid residues from oligosaccharides. As controls, we followed the appearance of PNA staining and loss of staining with the S7 antibody (which recognizes a sialylation-dependent epitope on CD43) following neuraminidase treatment (Figures 6A and 6B). A marked change in PNA and S7 staining on lymph node

T Cell Sialylation State Regulates CD8-Mediated Adhesion to Class I under Shear Flow Conditions Although our data using tetramer staining indicated modulation of CD8/class interactions with development and extent of glycosylation, it was not clear that these measurements would have any consequences for CD8 function. One well-characterized functional property of CD8 is its ability to mediate cell adhesion to class I molecules displayed on APC or artificial surfaces (Luescher et al., 1995; Norment et al., 1988; O’Rourke et al., 1990; Sun et al., 1995). Hence, we compared the ability of mature and immature thymocytes to adhere to immobilized MHC class I ligands. To do this using defined conditions, we allowed cells to interact with plate-bound MHC class I ligands and then subjected them to increasing shear flow stresses. Cell adhesion was viewed by phase contrast microscopy and videotaped to determine the fraction of cells that remain adhered to the plate with increasing flow. To establish the suitability of this approach, we first tested binding of CD8⫹ OT-I lymph node T cells to plates coated with Kb/OVA or streptavidin (SA). As cells were subjected to maximum shear flow conditions, 88% of the CD8⫹ OT-I LN remained adhered to the Kb/OVAcoated plate, while there were no cells left on the SAcoated plate (Figure 7A). Furthermore, this binding was completely inhibited by the anti-CD8␣ antibody CTCD8a (Figure 7A), in keeping with its effects on Kb/OVA tetramer binding to OT-I cells (Daniels and Jameson, 2000), suggesting that the cell adhesion involved CD8. We next tested whether CD8-mediated noncognate class I adhesion could be measured in the same way. Consistent with their intense staining by class I tetramers (Figure 4), TCR␣⫺/⫺ DP thymocytes adhered well to immobilized Dk/MT molecules but not to the control (SA alone) plate (Figure 7B). In stark contrast, mature CD8⫹ OT-I lymph node T cells were unable to adhere to the Dk surface (Figure 7C). This lack of adhesion is

Immunity 1056

Figure 6. Desialylation Increases CD8/Class I Interactions on Mature T Cells Thymocytes and lymph node cells from B6 mice were cultured in the presence and absence of neuraminidase and were then stained for CD4 and CD8 (53-6.7) together with FITC-PNA (A), FITC-S7 (B) antibody. Staining of the CD4⫹8⫹ DP and lymph node CD8⫹ T cells with or without neuraminidase treatment is shown as indicated. (C–E) In parallel, control (dotted histogram) or neuraminidase-treated cells (open and gray histograms) were stained with Dk/MT tetramer in the presence of anti-CD8 antibodies 53-6.7 (open and dotted histograms) or CTCD8a (gray histogram) plus anti-CD4. Tetramer staining is shown for lymph node CD8⫹ cells (C), DP thymocytes (D), and CD4⫹ SP thymocytes (E).

consistent with poor noncognate tetramer binding to mature OT-I T cells (Figure 5). Thus, the adhesion assay accurately reflected the differences in CD8/class I binding determined by flow cytometry. In the same experiments, we addressed whether adhesion could be modified by desialylation. Neuraminidase treatment had a marginal enhancing effect on the efficient TCR␣⫺/⫺ DP thymocyte binding to Dk/MT (Figure 7B) but had a dramatic effect on OT-I T cell adhesion to the same type of surface: approximately 60% of neuraminidase-treated OT-I cells were still bound to the Dk/MT surface under maximum shear flow, compared to 0% of the untreated OT-I T cell population (Figure 7C). Indeed, adhesion of neuraminidase-treated OT-I cells resembled that of the untreated TCR␣⫺/⫺ DP thymocytes on the Dk/MT surface (compare Figures 7B and 7C). We did observe slightly elevated adhesion of neuraminidase-treated OT-I cells to the control SA plate under low shear flow conditions (Figure 7C), perhaps reflective of some nonspecific adhesion, but this adhesion was lost within the first step of increasing shear flow force, implying very weak binding. We confirmed that the adhesion of neuraminidasetreated TCR␣⫺/⫺ thymocytes to the Dk/MT surface was mediated by CD8 by pretreating with the blocking anti-

CD8 antibody CTCD8a, which reduced adhesion to background (SA control) levels (Figure 7D). This was not a nonspecific effect of coating the cells with antibody, since anti-CD4 antibody did not inhibit binding of these cells (Figure 7D). Similarly, the CTCD8a antibody blocked binding of desialylated OT-I cells to Dk/MT (data not shown). Analogous results were obtained when cell adhesion to Kb/SIY and Kb/P815 complexes was studied, suggesting that this is not a unique feature of binding to Dk/MT (data not shown). These data indicate that the changes in CD8 which occur with thymocyte development and surface sialylation state have a significant impact on the ability of CD8 to act as an adhesion molecule for class I MHC. Discussion In this study, we show that CD8 binding to noncognate MHC class I molecules is regulated by the developmental state of the T cell, specifically that such binding decreases as cells mature from the DP stage. How is the CD8/class I interaction regulated during development? Initially, we explored the possibility that TCR expression levels negatively regulate the ability of CD8 to bind (noncognate) class I ligands. While our

Glycosylation Affects CD8 Binding to Class I MHC 1057

Figure 7. Influence of T Cell Development and Glycosylation on CD8-Mediated Adhesion to Class I Ligands (A) OT-I CD8⫹ LN cells were allowed to adhere to plates coated with streptavidin (SA) (filled circles) or SA and bio-Kb/OVA (OVA/Kb) (open squares and filled triangles). In some cases, the cells were pretreated with CTCD8a antibody (filled triangles). The cells were subject to increasing flow as indicated by the schematic at the bottom of the figure; flow was initiated (arrowhead) and increased in five increments over 50 s (steps) to a maximum force of ⵑ20 dynes cm2 . Adherent cells were enumerated by video microscopy, and data is shown as the fraction of initially adherent T cells remaining at each video frame (ⵑ1 s intervals). (B and C) Similar experiments were performed with TCR␣⫺/⫺ (B) thymocytes and OT-I CD8⫹ LN cells (C). In both cases, binding to plates coated with SA (circles) or SA plus Dk/MT (squares) was measured using control (filled symbols) or neuraminidase-treated cells (open symbols). (D) Neuraminidase-treated TCR␣⫺/⫺ thymocytes were allowed to adhere to SA (filled circles) or SA plus Dk/MT coated plates (squares and open circles). Prior to adhesion, some of the cells were incubated with antiCD4 (open squares) or anti-CD8 (CTCD8a) (open circles) TCR␣⫺/⫺ thymocytes.

experiments using antigen-induced TCR downregulation argue against this effect (Figure 5), we cannot rule out other changes which may have been induced by T cell activation and could mask a role for the TCR. A far more impressive correlation was observed when the glycosylation state of the developing T cells was considered. Glycosylation patterns on CD8 T cells are linked to their maturation, their activation state, and their potential for survival as naı¨ve T cells (Casabo et al., 1994; Harrington et al., 2000; Priatel et al., 2000; Wu et al., 1996). In both mice and humans, immature thymocytes and activated effectors demonstrate low levels of sialylation, while naı¨ve T cells are highly sialylated (Figure 5) (Baum et al., 1996; Casabo et al., 1994; Galvan et al., 2000; Gillespie et al., 1993; Kono et al., 1997; Priatel et al., 2000; Wu et al., 1996). This is very well characterized for the changes in sialylation of Gal␤1-3GalNAc␣-Ser/Thr (core-1 O-linked glycans), which is mediated by the enzyme ST3Gal-1. ST3Gal-1 expression is developmentally regulated, being absent in immature DP thymocytes and upregulated during maturation into mature SP T cells, and it is the activity of ST3Gal-1 which blocks

the PNA binding site on mature T cells (Baum et al., 1996; Galvan et al., 2000; Gillespie et al., 1993; Kono et al., 1997; Pereira et al., 1976; Priatel et al., 2000). We show here that desialylation of naı¨ve mature CD8⫹ cells enhances the ability of CD8 to bind noncognate class I ligands. Similar conclusions were reached by Moody et al. in a report published while this manuscript was in press (Moody et al., 2001a). Similar to our results, these authors observed that CD8 binding to noncognate class I tetramers could be observed on DP cells but not CD8 SP thymocytes or peripheral T cells, and that this developmental change was regulated by cell surface sialylation levels. These authors also showed that mature CD8 SP T cells from ST3Gal-1⫺/⫺ mice showed enhanced noncognate tetramer binding, suggesting an important role for sialylation of core-1 O-linked glycans in regulating CD8 binding. How might the sialylation of O-linked glycans influence CD8 binding? There is evidence for extensive O-linked glycosylation of the “stalk” region of CD8, which may play a role in membrane protein-protein interactions and/or membrane mobility of CD8, which con-

Immunity 1058

tribute indirectly to more efficient or more stable class I binding (Casabo et al., 1994; Moody et al., 2001a; Rudd et al., 2001; Rudd et al., 1999). We propose that this involves a critical contribution from the CD8␤ chain. Similar to the findings of others (Bosselut et al., 2000; Moody et al., 2001a), we find that noncognate tetramer staining requires expression of CD8␤ on thymocytes (data not shown). Furthermore, CD8␤ sialylation is altered during development and CD8 T cell activation (Casabo et al., 1994). Recent experimental evidence argues that the CD8␤ ectodomain may bind class I most proximal to APC membrane (Devine et al., 1999), requiring CD8␤ to be extended, while at the same time CD8 ectodomains must pivot to engage the class I ligand (Kern et al., 1998), requiring flexibility of the molecule. The extent of sialylation on the CD8␤ stalk may therefore be critical for optimal CD8 binding. At the same time, our desialylation experiments did not completely eliminate the difference in class I tetramer staining observed between immature DPs and mature CD8 T cells, as was evident from the greater staining on DP thymocytes than mature cells when both are treated with neuraminidase (Figure 6). Our experiments involved the neuraminidase from Vibrio cholerae, which has a fairly broad specificity cleaving ␣2-3, ␣2-6, and ␣2-8 sialic acid linkages. However, on certain cell-free substrates this enzyme shows a preference for ␣2-3 linkages (Corfield et al., 1983), leaving open the possibility that certain sialic acid linkages may not be efficiently removed in our experiments. Furthermore, Moody et al. found that ST3Gal-1⫺/⫺ CD8 SP T cells did not bind noncognate tetramer as well as neuraminidase-treated cells (Moody et al., 2001a). These data suggest that numerous sialic acid linkages could be involved in regulating CD8 binding. Indeed, there is evidence that ␣2-6 sialylation is also developmentally regulated in the thymus (Baum et al., 1996). Alternatively, there are other differences in glycosylation that may play a role in mediating the effect. For example, it is known that core-2 O-linked glycans are expressed on cell surface proteins of immature thymocytes and that sialylation of core-1 by ST3Gal-I in mature T cells inhibits the core-2 pathway, presumably through a mechanism that involves competition for substrate (Priatel et al., 2000). Such altered patterns of glycosylation would not be reversed by neuraminidase treatment and may account for the differences between noncognate CD8 binding to immature and mature CD8⫹ T cells which persist after desialylation. More sophisticated experiments will be required to test the roles of specific glycosylation patterns in addition to those induced by general desialylation. Lastly, our data do not rule out a role for changes in glycosylation of other T cell surface molecules. Differential glycosylation may also have a consequence of changing the lateral mobility or packing on the cell surface in general. Indeed, it has been proposed that Galectin-3 plays just such a role in regulating T cell activation (Demetriou et al., 2001). We also demonstrated that CD8/class I binding could be both enhanced and inhibited with appropriate CD8 antibodies, similar to the effects we reported previously for cognate interactions (Daniels and Jameson, 2000). Thus, the antibody 53-6.7 augmented and CT-CD8␣ inhibited CD8 noncognate class I interactions. We and

others showed that cognate class I binding was also enhanced with 53-6.7 Fab fragments, indicating that this effect is not due to CD8 aggregation (Daniels and Jameson, 2000; Luescher et al., 1995). The epitopes for 53-6.7 and CT-CD8␣ are distinct, as measured by antibody binding competition (M.A.D. and S.C.J., unpublished data) and by CD8␣ mutation analysis (L.D. and P.K., unpublished data), and this may relate to their very different effects on tetramer binding. However, it is important to note that the apparent developmental regulation of CD8 noncognate binding was not dependent on 53-6.7 binding: this antibody was not used in the adhesion assays, yet we still observed profound differences in CD8-mediated adhesion by immature DP versus mature CD8⫹ T cell populations (Figure 7). Our data revealed an inverse correlation between DP positive selection and CD8-class I binding (Figure 4). It was somewhat surprising that the tetramer staining of OT-I TAP⫺/⫺ DP cells was not as intense as that on TCR␣⫺/⫺, given the severe block in OT-I T cell positive selection in the former population. We note, however, that OT-I TAP⫺/⫺ DP cells do express low levels of CD69 (data not shown), suggesting that these cells have undergone some partial events in positive selection (although insufficient to induce complete maturation). This is in keeping with a model recently proposed by Wong and colleagues (Wong et al., 2001). In aggregate then, our results suggest that CD8 noncognate class I binding is downregulated by positive selection. What might be the functional role of regulated CD8 noncognate class I binding in T cell development? We showed in this report that the noncognate interaction was sufficient to cause CD8-mediated cell adhesion. In analogous studies using activated CTL, it was shown that CD8 adhesion to noncognate class I enhanced effector functions and triggered other adhesion molecule interactions (Anel et al., 1996; O’Rourke and Mescher, 1993; O’Rourke et al., 1990). Similarly, efficient CD8/class I binding could enhance immature thymocyte adhesion to thymic epithelial cells. Moreover, optimal CD8 binding at the DP stage might concentrate potential selecting MHC ligands for examination by the TCR by retaining class I molecules in the region of T cell thymic epithelial cell contact—a feature which could compensate for the low levels of TCR expression at this stage. It is noteworthy that the immature DP stage, at which we observe most efficient CD8 binding to noncognate class I ligands, is also characterized by heightened sensitivity to low-affinity TCR ligands (Davey et al., 1998; Lucas et al., 1999). Therefore, decreasing the ability of CD8 to bind class I as T cells mature may contribute to terminating the interaction with positively selecting self-peptide/MHC ligands and thus could play a role in regulating self-tolerance or survival. Indeed, there is evidence that there may be a specific glycosylation requirement for CD8 T cell survival, as demonstrated by profound loss of peripheral CD8⫹ T cells in the ST3Gal-I⫺/⫺ mice (Priatel et al., 2000). At present it is not clear whether this is mediated by an effect on CD8/class I interactions, but the fact that the loss of mature CD4⫹ T cells in ST3Gal-1⫺/⫺ mice is much less prominent supports this hypothesis. Since CD8 interactions are critically involved in mediating positive and negative selection of class I-restricted T cells, it is difficult to determine the relative contribution

Glycosylation Affects CD8 Binding to Class I MHC 1059

of CD8 acting as an adhesion molecule versus its acting as a coreceptor with the TCR. Ironically, the most compelling evidence for a role for noncognate class I interactions in thymic development comes from analysis of class II-restricted TCR transgenic T cells which aberrantly committed to the CD8 lineage. For example, in CD4-deficient mice bearing the AND transgenic TCR, thymocytes develop into the CD8 lineage. This process involved TCR interaction with specific class II alleles but also required exposure to class I MHC molecules (Matechak et al., 1996). This latter interaction was not allele specific, leading to the conclusion that noncognate CD8/class I MHC interactions were critical to T cell maturation in this model (Matechak et al., 1996). Similar interpretations were arrived at by other groups using different model systems (Kirberg et al., 1994; Yasutomo et al., 2000). These data support the concept that noncognate coreceptor interactions may be important in successful positive selection and/or lineage commitment. Our data imply that the ability of CD8 to undergo such interactions changes during thymic differentiation and may be part of the dynamic “tuning” required for development of a tolerant but useful T cell repertoire. Very recent reports from other groups also indicated efficient CD8/class I binding by immature DP thymocytes (Moody et al., 2001b; Tsujimura et al., 2001), although these initial studies did not correlate this CD8 binding pattern with thymic selection, nor did they determine the basis for altered CD8 binding during development. Likewise, there is little data on the mechanism by which CD8 binding is upregulated during mature CD8 T cell activation; it should be pointed out that there have been no reports of noncognate tetramer interactions on activated CD8 effector populations, although this population is desialylated compared to naı¨ve CD8 T cells (Harrington et al., 2000; Priatel et al., 2000). Furthermore, noncognate MHC binding by CD8 on effector CTL requires TCR-induced activation (Anel et al., 1996; Luescher et al., 1995; O’Rourke and Mescher, 1993; O’Rourke et al., 1990), while this was evidently not required for the binding by immature DP thymocytes we study here. Whether this indicates different mechanisms for upregulated CD8 binding in development versus mature T cell activation will require further investigation. Our data highlight the growing awareness that glycosylation of lymphocytes plays a pivotal role in regulating their activity (Lowe, 2001; Rudd et al., 1999, 2001). In this study we demonstrate the role of sialylation in developmental regulation of class I binding by CD8⫹ T cells. However, similar changes in glycosylation occur during development of CD4 T cells, and it will be interesting to determine whether there is similar developmental regulation of CD4 binding to class II. Experimental Procedures Mice and Cells C57BL/6 (H 2b), C3H (H 2k), Balb/c (H 2d), and TCR␣-deficient mice were obtained from Jackson Labs (Bar Harbor, ME). OT-I and TAP-1⫺/⫺ OT-I mice (Hogquist et al., 1997) were bred in our colony. All mice were generated and maintained under specific pathogenfree conditions. Thymi and major lymph nodes were harvested, and a single-cell suspension was generated as previously described and was, in some cases, enriched for CD8⫹ T cells by negative selection

using a CD8 Cellect column (Cytovax Technologies, Edmonton, Alberta, Canada). MHC Tetramers Tetrameric forms of Kb with the OVA peptide (SIINFEKL) or SIY peptide (SIYRYYGL) and Dk bound to the MT389-397 peptide (RRLGRTLLL) have been described previously (Daniels and Jameson, 2000; Lukacher et al., 1999). Production of MHC I/␤2m/peptide multimers followed protocols described previously (Altman et al., 1996; Busch et al., 1998; Daniels and Jameson, 2000). Briefly, MHC class I molecules with the Bir A recognition sequence at the carboxyl terminus and human ␤2m molecules were transformed and overexpressed in E. coli. The synthesized proteins were purified from inclusion bodies, denatured, mixed with the appropriate peptide, and the mixture was allowed to renature in suitable oxido-reductive conditions over 48 hr. Following dialysis and concentration, the folded monomer complexes were purified via FPLC on a Superdex 200 column (Pharmacia Biotech Inc., Piscataway, NJ). The monomers were then biotinylated using Bir A (Auidity, Denver, CO) and repurified via FPLC. Following concentration of the appropriate sized fractions, the biotinylated MHC I/peptide complexes were mixed with SA-PE (Molecular Probes, Portland, OR) that had been purified by FPLC to remove aggregates, at a 4:1 molar ratio. Multimers were used without further purification except where noted. Multimers were tested over a range of doses for flow cytometry and were typically used at 25 ␮g/ml. Antibodies and Flow Cytometry Cells (1 ⫻ 106 ) were stained in 100 ␮l of FACS buffer (PBS, 1% fetal calf serum and 3 mM azide) or RP-10 with 3 mM azide on ice at the indicated times, usually 2 hr. Cells were incubated in cold FACS buffer for 30 min prior to addition of cold staining reagents, and plates were prechilled and kept on ice throughout staining. Typically, MHC multimers (25 ␮g/ml) and anti-CD8 antibodies (50 ␮g/ml unless stated otherwise) were added simultaneously. The following antiCD8␣ antibodies were used: 53-6.7 (Pharmingen, San Diego, CA) and CT-CD8a (Caltag, Burlingame, CA). Cells were analyzed using a Becton Dickinson FACSCaliber (Becton Dickinson, San Jose, CA). Data were analyzed using both CellQuest (Becton Dickinson) and FlowJo (TreeStar, San Carlos, CA) software. TCR Downregulation Thymocytes from B6 or OT-I TCR transgenic mice were incubated in RP10 media at 37⬚C in the presence or absence of 10 ␮M OVAp (SIINFEKL) for 30 min, after which time the cells were spun into a pellet and incubated an additional 3.5 hr. Cells were washed and chilled in FACS buffer and stained for CD4, CD8, CD3⑀ (Pharmingen, San Diego, CA), and MHC I/peptide tetramers as described. Neuraminidase Treatment Cells were prepared as described above and incubated at 37⬚C at a concentration of 2 ⫻ 106 cells per ml in RPMI in the presence or absence of 1 ␮l (0.016 units) type II neuraminidase from Vibrio cholerae (Sigma, St. Louis, MO) per million cells for 20 min. This enzyme cleaves ␣2-3, ␣2-6, and ␣2-8 sialic acid linkages, and the dose used was determined (by titration) as sufficient to induce maximal PNA staining of mature thymocytes. Cells were then washed three times in RPMI 1640 with 10% FCS and resuspended in adhesion (RPMI 1640 with 0.5% BSA) or FACS buffer. Parallel Plate Flow Chamber and Adhesion Assay Streptavidin (Molecular Probes, Eugene, OR) (125 ␮l of 125 ␮g/ml solution) was adsorbed to a defined area on 35 mm Falcon petri dishes (No. 1008) overnight at 4⬚C. The plates were washed three times with PBS, then incubated for a minimum of 4 hr in PBS with or without biotinylated peptide/MHC I (125 ␮l of 10 ␮g/ml solution) complexes. The plates were then blocked with PBS with 2% BSA for at least 1 hr. Lymphocyte adhesion under shear stress was examined with a parallel plate flow chamber obtained from Glycotech (Rockville, MD) as previously described (Walcheck et al., 1996a, 1996b). Briefly, plates coated with the specified peptide/MHC I complexes were attached to the flow chamber containing a 5-mm-wide gasket. Shear

Immunity 1060

flow was generated and controlled with an automated syringe pump (Harvard Apparatus, Natick, MA) attached to the outlet side of the flow chamber. Cells at a concentration of 2 ⫻ 106 /ml in RPMI 1640 with 0.5% BSA were injected into the flow chamber and allowed to settle on the plate for 2 min, then flow rate was increased stepwise every 10 s (by a programmed set of flow rates) over the course of 50 s in increments of 1.28 ml/min (ⵑ4 dynes/cm2 shear stress) to a maximum of 6.4 ml/min (corresponding to ⵑ20 dynes/cm2 ). Lymphocyte adhesion was visualized with an Olympus IX-70 inverted microscope (Melville, NY) equipped with a Panasonic color CCD camera (Secaucus, NJ), recorded to VHS video tape using a Panasonic S-VHS professional deck, digitized, and compressed using Adobe Premiere software. The compressed movies were analyzed using NIH Image (v1.62). The number of cells that remained bound was determined relative to the number of cells that accumulated during the 2 min of zero flow and are displayed as fraction bound per frame of the compressed movie (ⵑ1 frame per s real time). Data are shown for the last 5 s of zero flow and the full 50 s of increasing flow rates. Acknowledgments We thank Royce Robinson-Lawler and Bruce Walcheck for help in establishing the adhesion assays, and Matt Mescher and members of the Jameson and Hogquist labs for helpful discussions. We also wish to thank Ellis Reinherz and Jamey Marth for communicating their findings ahead of publication. This work was supported in part by the following grants from the National Institutes of Health: R01 AI38903 (S.C.J.) and R01 CA48115 (P.K.). M.A.D. was supported by an NIH Immunology Training Grant (T32 AI07313). Received September 24, 2001; revised November 20, 2001. References Altman, J.D., Moss, P.A.H., Goulder, P.J.R., Barouch, D.H., McHeyzer-Williams, M.G., Bell, J.I., McMichael, A.J., and Davis, M.M. (1996). Phenotypic analysis of antigen-specific T lymphocytes. Science 274, 94–96. Anel, A., O’Rourke, A.M., Kleinfeld, A.M., and Mescher, M.F. (1996). T cell receptor and CD8-dependent tyrosine phosphorylation events in cytotoxic T lymphocytes: activation of p56lck by CD8 binding to class I protein. Eur. J. Immunol. 26, 2310–2319.

alpha(2-6)-linked sialic acids in glycoproteins. Biochim. Biophys. Acta 744, 121–126. Crawford, F., Kozono, H., White, J., Marrack, P., and Kappler, J. (1998). Detection of antigen-specific T cells with multivalent soluble class II MHC covalent peptide complexes. Immunity 8, 675–682. Daniels, M.A., and Jameson, S.C. (2000). Critical role for CD8 in T cell receptor binding and activation by peptide/major histocompatibility complex multimers. J. Exp. Med. 191, 335–346. Davey, G.M., Schober, S.L., Endrizzi, B.T., Dutcher, A.K., Jameson, S.C., and Hogquist, K.A. (1998). Pre-selection thymocytes are more sensitive to TCR stimulation than mature T cells. J. Exp. Med. 188, 1867–1874. Demetriou, M., Granovsky, M., Quaggin, S., and Dennis, J.W. (2001). Negative regulation of T-cell activation and autoimmunity by Mgat5 N-glycosylation. Nature 409, 733–739. Devine, L., Sun, J., Barr, M.R., and Kavathas, P.B. (1999). Orientation of the Ig domains of CD8 alpha beta relative to MHC class I. J. Immunol. 162, 846–851. Fowlkes, B.J., Waxdal, M.J., Sharrow, S.O., Thomas, C.A., 3rd, Asofsky, R., and Mathieson, B.J. (1980). Differential binding of fluorescein-labeled lectins to mouse thymocytes: subsets revealed by flow microfluorometry. J. Immunol. 125, 623–630. Fung Leung, W.P., Schilham, M.W., Rahemtulla, A., Kundig, T.M., Vollenweider, M., Potter, J., van Ewijk, W., and Mak, T.W. (1991). CD8 is needed for development of cytotoxic T cells but not helper T cells. Cell 65, 443–449. Galvan, M., Tsuboi, S., Fukuda, M., and Baum, L.G. (2000). Expression of a specific glycosyltransferase enzyme regulates T cell death mediated by galectin-1. J. Biol. Chem. 275, 16730–16737. Garcia, K.C., Scott, C.A., Brunmark, A., Carbone, F.R., Peterson, P.A., Wilson, I.A., and Teyton, L. (1996). CD8 enhances formation of stable T-cell receptor/MHC class I molecule complexes. Nature 384, 577–581. Gillespie, W., Paulson, J.C., Kelm, S., Pang, M., and Baum, L.G. (1993). Regulation of alpha 2,3-sialyltransferase expression correlates with conversion of peanut agglutinin (PNA)⫹ to PNA⫺ phenotype in developing thymocytes. J. Biol. Chem. 268, 3801–3804. Harrington, L.E., Galvan, M., Baum, L.G., Altman, J.D., and Ahmed, R. (2000). Differentiating between memory and effector CD8 T cells by altered expression of cell surface O-glycans. J. Exp. Med. 191, 1241–1246.

Arcaro, A., Gregoire, C., Boucheron, N., Stotz, S., Palmer, E., Malissen, B., and Luescher, I.F. (2000). Essential role of CD8 palmitoylation in CD8 coreceptor function. J. Immunol. 165, 2068–2076.

Hogquist, K.A., Tomlinson, A.J., Kieper, W.C., McGargill, M.A., Hart, M.C., Naylor, S., and Jameson, S.C. (1997). Identification of a naturally occurring ligand for thymic positive selection. Immunity 6, 389–399.

Baum, L.G., Derbin, K., Perillo, N.L., Wu, T., Pang, M., and Uittenbogaart, C. (1996). Characterization of terminal sialic acid linkages on human thymocytes. Correlation between lectin-binding phenotype and sialyltransferase expression. J. Biol. Chem. 271, 10793–10799.

Ingold, A.L., Landel, C., Knall, C., Evans, G.A., and Potter, T.A. (1991). Coengagement of CD8 with the T cell receptor is required for negative selection. Nature 352, 721–723.

Block, M.S., Johnson, A.J., Mendez-Fernandez, Y., and Pease, L.R. (2001). Monomeric class I molecules mediate TCR/CD3⑀/CD8 interaction on the surface of T cells. J. Immunol. 167, 821–826. Bosselut, R., Zhang, W., Ashe, J.M., Kopacz, J.L., Samelson, L.E., and Singer, A. (1999). Association of the adaptor molecule LAT with CD4 and CD8 coreceptors identifies a new coreceptor function in T cell receptor signal transduction. J. Exp. Med. 190, 1517–1526. Bosselut, R., Kubo, S., Guinter, T., Kopacz, J.L., Altman, J.D., Feigenbaum, L., and Singer, A. (2000). Role of CD8beta domains in CD8 coreceptor function: importance for MHC I binding, signaling, and positive selection of CD8⫹ T cells in the thymus. Immunity 12, 409–418. Busch, D.H., Pilip, I.M., Vijh, S., and Pamer, E.G. (1998). Coordinate regulation of complex T cell populations responding to bacterial infection. Immunity 8, 353–362.

Janeway, C.A., Jr. (1992). The T cell receptor as a multicomponent signalling machine: CD4/CD8 coreceptors and CD45 in T cell activation. Annu. Rev. Immunol. 10, 645–674. Kane, K.P., and Mescher, M.F. (1993). Activation of CD8-dependent cytotoxic T lymphocyte adhesion and degranulation by peptide class I antigen complexes. J. Immunol. 150, 4788–4797. Kern, P.S., Teng, M.K., Smolyar, A., Liu, J.H., Liu, J., Hussey, R.E., Spoerl, R., Chang, H.C., Reinherz, E.L., and Wang, J.H. (1998). Structural basis of CD8 coreceptor function revealed by crystallographic analysis of a murine CD8␣␣ ectodomain fragment in complex with H-2Kb. Immunity 9, 519–530. Kirberg, J., Baron, A., Jakob, S., Rolink, A., Karjalainen, K., and von Boehmer, H. (1994). Thymic selection of CD8⫹ single positive cells with a class II major histocompatibility complex-restricted receptor. J. Exp. Med. 180, 25–34.

Casabo, L.G., Mamalaki, C., Kioussis, D., and Zamoyska, R. (1994). T cell activation results in physical modification of the mouse CD8 beta chain. J. Immunol. 152, 397–404.

Kono, M., Ohyama, Y., Lee, Y.C., Hamamoto, T., Kojima, N., and Tsuji, S. (1997). Mouse beta-galactoside alpha 2,3-sialyltransferases: comparison of in vitro substrate specificities and tissue specific expression. Glycobiology 7, 469–479.

Corfield, A.P., Higa, H., Paulson, J.C., and Schauer, R. (1983). The specificity of viral and bacterial sialidases for alpha(2-3)- and

Lowe, J.B. (2001). Glycosylation, immunity, and autoimmunity. Cell 104, 809–812.

Glycosylation Affects CD8 Binding to Class I MHC 1061

Lucas, B., Stefanova, I., Yasutomo, K., Dautigny, N., and Germain, R.N. (1999). Divergent changes in the sensitivity of maturing T cells to structurally related ligands underlies formation of a useful T cell repertoire. Immunity 10, 367–376. Luescher, I.F., Vivier, E., Layer, A., Mahiou, J., Godeau, F., Malissen, B., and Romero, P. (1995). CD8 modulation of T-cell antigen receptor-ligand interactions on living cytotoxic T lymphocytes. Nature 373, 353–356. Lukacher, A.E., Moser, J.M., Hadley, A., and Altman, J.D. (1999). Visualization of polyoma virus-specific CD8⫹ T cells in vivo during infection and tumor rejection. J. Immunol. 163, 3369–3378. Matechak, E.O., Killeen, N., Hedrick, S.M., and Fowlkes, B.J. (1996). MHC class II-specific T cells can develop in the CD8 lineage when CD4 is absent. Immunity 4, 337–347. Miceli, M.C., and Parnes, J.R. (1993). Role of CD4 and CD8 in T cell activation and differentiation. Adv. Immunol. 53, 59–122. Mombaerts, P., Clarke, A.R., Rudnicki, M.A., Iacomini, J., Itohara, S., Lafaille, J.J., Wang, L., Ichikawa, Y., Jaenisch, R., Hooper, M.L., and Tonegawa, S. (1992). Mutations in T-cell antigen receptor genes alpha and beta block thymocyte development at different stages. Nature 360, 225–231. Moody, A.M., Chui, D., Reche, P.A., Priatel, J.J., Marth, J.D., and Reinherz, E.L. (2001a). Developmentally regulated glycosylation of the CD8␣␤ coreceptor stalk modulates ligands binding. Cell 107, 501–512. Moody, A.M., Xiong, Y., Chang, H.C., and Reinherz, E.L. (2001b). The CD8alpha beta coreceptor on double-positive thymocytes binds with differing affinities to the products of distinct class I MHC loci. Eur. J. Immunol. 31, 2791–2799. Murali-Krishna, K., Altman, J.D., Suresh, M., Sourdive, D.J., Zajac, A.J., Miller, J.D., Slansky, J., and Ahmed, R. (1998). Counting antigen-specific CD8 T cells: a reevaluation of bystander activation during viral infection. Immunity 8, 177–187. Norment, A.M., Salter, R.D., Parham, P., Engelhard, V.H., and Littman, D.R. (1988). Cell-cell adhesion mediated by CD8 and MHC class I molecules. Nature 336, 79–81. O’Rourke, A.M., and Mescher, M.F. (1993). The roles of CD8 in cytotoxic T lymphocyte function. Immunol. Today 14, 183–188. O’Rourke, A.M., Rogers, J., and Mescher, M.F. (1990). Activated CD8 binding to class I protein mediated by the T-cell receptor results in signalling. Nature 346, 187–189. Pereira, M.E., Kabat, E.A., Lotan, R., and Sharon, N. (1976). Immunochemical studies on the specificity of the peanut (Arachis hypogaea) agglutinin. Carbohydr. Res. 51, 107–118. Priatel, J.J., Chui, D., Hiraoka, N., Simmons, C.J., Richardson, K.B., Page, D.M., Fukuda, M., Varki, N.M., and Marth, J.D. (2000). The ST3Gal-I sialyltransferase controls CD8⫹ T lymphocyte homeostasis by modulating O-glycan biosynthesis. Immunity 12, 273–283. Reisner, Y., Linker-Isreali, M., and Sharon, N. (1976). Separation of mouse thymocytes into two subpopulations by the use of peanut agglutinin. Cell. Immunol. 25, 129–134. Robey, E., and Fowlkes, B.J. (1994). Selective events in T cell development. Annu. Rev. Immunol. 12, 675–705. Rudd, P.M., Wormald, M.R., Stanfield, R.L., Huang, M., Mattsson, N., Speir, J.A., DiGennaro, J.A., Fetrow, J.S., Dwek, R.A., and Wilson, I.A. (1999). Roles for glycosylation of cell surface receptors involved in cellular immune recognition. J. Mol. Biol. 293, 351–366. Rudd, P.M., Elliott, T., Cresswell, P., Wilson, I.A., and Dwek, R.A. (2001). Glycosylation and the immune system. Science 291, 2370– 2376. Sun, J., Leahy, D.J., and Kavathas, P.B. (1995). Interaction between CD8 and major histocompatibility complex (MHC) class I mediated by multiple contact surfaces that include the alpha 2 and alpha 3 domains of MHC class I. J. Exp. Med. 182, 1275–1280. Thome, M., Germain, V., DiSanto, J.P., and Acuto, O. (1996). The p56lck SH2 domain mediates recruitment of CD8/p56lck to the activated T cell receptor/CD3/zeta complex. Eur. J. Immunol. 26, 2093– 2100. Tsujimura, K., Obata, Y., Matsudaira, Y., Ozeki, S., Yoshikawa, K.,

Saga, S., and Takahashi, T. (2001). The binding of thymus leukemia (tl) antigen tetramers to normal intestinal intraepithelial lymphocytes and thymocytes. J. Immunol. 167, 759–764. Varki, A., Cummings, R., Esko, J., Freeze, H., Hart, G., and Marth, J., eds. (1999). Essentials of Glycobiology (Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press). Walcheck, B., Kahn, J., Fisher, J.M., Wang, B.B., Fisk, R.S., Payan, D.G., Feehan, C., Betageri, R., Darlak, K., Spatola, A.F., and Kishimoto, T.K. (1996a). Neutrophil rolling altered by inhibition of L-selectin shedding in vitro. Nature 380, 720–723. Walcheck, B., Moore, K.L., McEver, R.P., and Kishimoto, T.K. (1996b). Neutrophil-neutrophil interactions under hydrodynamic shear stress involve L-selectin and PSGL-1. A mechanism that amplifies initial leukocyte accumulation of P-selectin in vitro. J. Clin. Invest. 98, 1081–1087. Wong, P., Barton, G.M., Forbush, K.A., and Rudensky, A.Y. (2001). Dynamic tuning of T cell reactivity by self-peptide-major histocompatibility complex ligands. J. Exp. Med. 193, 1179–1187. Wu, W., Harley, P.H., Punt, J.A., Sharrow, S.O., and Kearse, K.P. (1996). Identification of CD8 as a peanut agglutinin (PNA) receptor molecule on immature thymocytes. J. Exp. Med. 184, 759–764. Yasutomo, K., Doyle, C., Miele, L., Fuchs, C., and Germain, R.N. (2000). The duration of antigen receptor signalling determines CD4⫹ versus CD8⫹ T-cell lineage fate. Nature 404, 506–510. Zamoyska, R. (1994). The CD8 coreceptor revisited: One chain good, two chains better. Immunity 1, 243–246.