Dynamic Association of CD45 with Detergent-Insoluble Microdomains in T Lymphocytes¹

The transmembrane protein tyrosine phosphatase CD45 is expressed on the surface of all nucleated cells of hematopoietic origin. It is a highly glycosylated protein and abundantly expressed, constituting up to 10% of the total cell surface protein. Using CD45-deficient cell lines and CD45 gene-disrupted mice, it has been shown that CD45 functions as an important regulator of maturation and activation pathways in lymphocytes (1, 2, 3). The importance of the phosphatase activity of CD45 in lymphocyte activation has been confirmed by the finding that transfection of chimeric proteins containing the cytoplasmic domains of CD45 into cells deficient in CD45 expression restores normal T cell signaling events (4, 5). Downstream substrates of CD45 have been identified as the protein tyrosine kinases p56^lck and p59^fyn. It has been demonstrated that CD45 can regulate p56^lck and p59^fyn activity by dephosphorylation of their negative regulatory sites both in vitro and in vivo (6, 7, 8). Little is known about the regulation of the phosphatase activity of CD45 itself; however, there is evidence to suggest that dimerization and phosphorylation may be involved (9, 10, 11).

After TCR complex ligation, the signaling cascade is initiated by the tyrosine phosphorylation of specific motifs called immunoreceptor tyrosine-based activation motifs found on the CD3 ζ homodimer by p56^lck and/or p59^fyn. This allows for the recruitment of the Syk family tyrosine kinase ZAP70 to the TCR complex by the binding of its tandem Src homology 2 domains to the phosphorylated immunoreceptor tyrosine-based activation motifs on CD3 ζ. Once at the membrane, ZAP70 can now phosphorylate the linker for activation of T cells (LAT),⁴ which links TCR triggering to a number of downstream pathways, including the Ras pathway (12), ultimately leading to activation of transcription factors and gene expression.

There have been many reports implicating membrane microdomains or lipid rafts in TCR activation as a number of key signaling components have been shown to localize to these regions of the cell membrane (13, 14, 15, 16). Lipid rafts have been described as gel-like, liquid-ordered regions within the more fluid, liquid-disordered areas of glycerophospholipids (17). They are tightly packed regions of the plasma membrane, highly concentrated in cholesterol and sphingolipid, and are therefore insoluble in nonionic detergents, which has led to them also being referred to as detergent-resistant membranes (DRMs). It is thought that DRMs represent in vitro rafts because DRMs appear to be derived from lipid rafts. Due to their relative buoyancy, DRMs can be isolated in low-density, buoyant fractions upon sucrose density centrifugation (18). As well as being enriched in certain lipids, a number of proteins are also concentrated in rafts largely as a result of modification with GPI anchors or acylation, which pack well into the ordered lipid environment. By virtue of their lipid modifications, p56^lck and LAT have been found to localize to DRMs (19, 20). Biochemical studies have suggested that CD45 can associate with Brij 58 DRMs (14) and DRMs prepared from the T lymphoblastoid cell line CEM after Triton X-100 extraction (21). In contrast, other researchers have concluded that CD45 is primarily excluded from Triton X-100 DRMs prepared from Jurkat T cells and TCR-expressing BW5147 cells (13, 19, 22, 23). This result has been confirmed by confocal microscopy studies in which it was shown that CD45 is excluded from rafts upon aggregation of the lipid rafts with cholera toxin (CTx) (16). Here we describe the cholesterol-dependent association of LAT with CD45 in CD45 immunoprecipitates and show that CD45 is indeed present in DRMs after extraction with Triton X-100. We find that the amount of CD45 present in DRMs is reduced upon T cell stimulation, suggesting the DRM association of CD45 is dependent upon the activation state of the cell.

BW5147 (BW) and Yac-1 T lymphoma cells were obtained from American Type Culture Collection (Manassas, VA) and were grown in DMEM supplemented with 8% defined and supplemented calf serum (HyClone, Logan, UT). BWT200 (CD45⁻) and BW revertant (BWRev) cell lines were obtained from Dr. R. Hyman (The Salk Institute, La Jolla, CA). Murine cytotoxic T cell clone AB.1 has been previously described (24). Thymocytes were freshly prepared from thymi obtained from B6 mice and used the same day. Anti-CD45 cytoplasmic domain-specific antisera 95k (10) and J37 were prepared in our laboratory. The CD45-specific mAb I3/2 was provided by Dr. I. Trowbridge (The Salk Institute). Anti-LAT was purchased from Upstate Biotechnology (Lake Placid, NY) and anti-p56^lck was obtained from BD Biosciences (Lexington, KY). The hybridomas producing the anti-CD45RB mAb MB23G2, the anti-CD3ε mAb 145-2C11, and the anti-LFA-1 mAb M17/5.2 were obtained from American Type Culture Collection.

Cells were lysed in 0.5% Triton X-100, 1% Brij 97, or 1% Brij 58 in 10 mM Tris-HCl containing 150 mM NaCl, 0.02% NaN₃, and 0.2 mM PMSF. CD45 was subjected to immunoprecipitation from postnuclear lysates prepared with the indicated detergent by incubation with anti-CD45 mAbs (I3/2) conjugated to Sepharose 4B for 2 h at 4°C. Immunoprecipitates were washed four times in lysis buffer. Proteins coprecipitating with CD45 were resolved by SDS-PAGE and identified by immunoblotting.

For some experiments, cells were treated with MCD (Sigma-Aldrich, St. Louis, MO) to disrupt lipid rafts. In general, we found MCD to be highly toxic to T cells, so treatment was limited to a subset of T cell lines. As an initial set of experiments, cells were treated with different concentration of MCD ranging from 2 to 20 mM in the absence of serum for 30 min at 37°C, after which cell viability was assessed by trypan blue exclusion. Cells were then put into culture in the presence of serum and assessed for viability and cell number 24 h later. Only conditions that allowed for cell viability comparable to mock treatment were used for these studies. We found that 10 mM MCD was the highest concentration that could be used for the BW5147 and Yac-1 cells and allow for essentially complete recovery after treatment.

Wells of a Falcon 3912 96-well plate were incubated overnight at 4°C with 50 μl of 10 μg/ml I3/2 anti-CD45 Ab. The wells were washed with PBS and blocked with 2% BSA in PBS for 45 min at 37°C. The wells were again washed and 2 × 10⁴ Yac-1 cells were added to each well of the plate and incubated at 37°C. Cells were examined under an inverted light microscope 20–30 min after addition to the plate.

To preserve the DRMs, all steps were conducted at 4°C, and all detergent and sucrose solutions were prepared in 10 mM Tris-HCl containing 150 mM NaCl and 0.2 mM PMSF. Cells were lysed in 0.5% or 1% Triton X-100 before mixing with an equal volume of 85% sucrose. This was then overlaid with a stepwise gradient of 35–5% sucrose. After centrifugation at 160,000 × g for a minimum of 5 h, fractions were collected from the top of the gradient. At this time, fractions containing buoyant DRMs were identified by GM1 content. Briefly, 1 μl of each fraction was spotted onto a nitrocellulose filter and allowed to dry. The filter was then probed with CTx coupled to HRP (Sigma-Aldrich) to reveal those fractions containing the ganglioside G_M1, a raft marker. For gel analysis of the individual fractions from the sucrose gradient, 30 μl of each fraction was mixed with 2× reducing sample buffer and loaded onto a 4–12% gradient gel, transferred to immobilon, and probed with the indicated Ab.

For some experiments it was desirable to concentrate the detergent-insoluble material from each fraction. This was done by diluting the fractions threefold with the original lysis buffer to dilute the sucrose, followed by centrifugation at 19,000 × g. The resulting insoluble pellet was washed one time in the appropriate lysis buffer to remove any remaining soluble material and pelleted by centrifugation to recover the insoluble material. The insoluble proteins were then solubilized in reducing sample buffer, resolved by SDS-PAGE, and identified by Western blotting. In all cell types studied, the distribution of CD45RB across the sucrose gradient was found to be identical with that of total CD45 as determined using a pan-isoform-specific antisera. In addition, CD45 detection using anti-CD45RB Abs allows better definition of bands for quantitation, and therefore in some experiments the CD45RB isoform was used as a marker for CD45.

Anti-CD3 stimulation of Yac-1 cells was essentially as described previously (25). Briefly, cells were incubated with 10 μg/ml 145-2C11 for 15 min on ice followed by 5 μg/ml rabbit anti-hamster IgG over a period of 60 min at 37°C. After washing the cells to remove the Abs, DRMs were then prepared and the amount of CD45 and Lck found to localize to DRMs was determined by optical densitometry using NIH Imager 1.62, expressed as a percentage of that detected in resting cells.

It has previously been shown that p56^lck cannot be found in CD45 immunoprecipitates in the presence of Triton X-100 but can be detected in immunoprecipitates prepared in digitonin (26). After lysis of BW or BW/T200⁻ (CD45-negative) cells in various nonionic detergents, CD45 was immunoprecipitated and the presence of p56^lck in each immunoprecipitate was determined. Although substantially more p56^lck is found associated with CD45 in 1% Brij 97 and Brij 58, p56^lck is detected in CD45 immunoprecipitates prepared after extraction in 0.5% Triton X-100, albeit at considerably reduced levels compared with other detergents (Fig. 1 A). No p56^lck is immunoprecipitated with anti-CD45 Abs from the CD45-negative cell line in any detergent.

Surprisingly, LAT was also found to coimmunoprecipitate with CD45 in each of the detergents used (Fig. 1,A, lower panel). Like p56^lck, the amount of LAT associating with CD45 differed considerably depending upon the detergent used for extraction. However, in contrast to what was seen with p56^lck, a significant amount of LAT can be detected in CD45 immunoprecipitates prepared in the presence of Triton X-100. Conversely, much less LAT was found to associate with CD45 when Brij 97 was used compared with each of the other detergents. Upon immunoblotting for CD45, it can be seen that less CD45 is immunoprecipitated after lysis in Brij 58, but there is still a considerable amount of both p56^lck and LAT in the immunoprecipitate. This suggests that Brij 58 provides the optimal conditions in which to preserve the association of CD45 with these two proteins. No LAT can be detected in CD45 immunoprecipitates isolated from the CD45-negative cell line BW/T200⁻ (Fig. 1,A). Anti-LFA-1 immunoprecipitates were performed as a control and no Lck was detected in these immunoprecipitates using the three different detergents (Fig. 1 B).

The association of CD45 with LAT is not restricted to BW cells. Using another T lymphoma cell line, Yac-1, it is also possible to detect LAT in CD45 immunoprecipitates prepared in either Triton X-100 or Brij 58 (Fig. 1 C). The association is again much stronger after Brij 58 extraction; however, a significant amount of LAT is also seen in the CD45 immunoprecipitates in the presence of Triton X-100.

Previously, it has been shown that both p56^lck and LAT are targeted to DRMs (19, 20). A powerful tool in the characterization of lipid rafts is the use of MCD, which selectively sequesters cholesterol, thereby disrupting the DRMs (27). Therefore, to determine whether the association between CD45 and LAT is a cholesterol-dependent interaction, cells were pretreated with MCD and lysed in Triton X-100 or Brij 97 before immunoprecipitation with anti-CD45 Abs. Upon treatment of the cells with MCD, the amount of p56^lck found to coimmunoprecipitate with CD45 did not significantly change after lysis in either Triton X-100 or Brij 97 (Fig. 2). MCD also had no effect on the low amount of LAT found to associate with CD45 in immunoprecipitates prepared in Brij 97 (Fig. 2). However, the coimmunoprecipitation of LAT with CD45 is abolished after MCD treatment and extraction with Triton X-100. Equivalent amounts of CD45 were immunoprecipitated in the presence of each of the detergents as determined by immunoblotting (data not shown).

It has been reported that CD45 can be found in DRMs isolated in the presence of Brij 58 (13), but not when Triton X-100 is used as the extracting detergent (13, 19, 22, 23). Because we found the interaction between CD45 and LAT is cholesterol dependent, we next wanted to determine whether CD45 was localized to DRMs. BW cells were lysed in Triton X-100 and DRMs were isolated. Fractions containing DRMs were identified by CTx reactivity and LAT immunoblotting (Fig. 3,A). Surprisingly, CD45 can be detected, though not enriched, in DRMs isolated when 0.5% Triton X-100 was used as detergent (Fig. 3,A). Optical densitometry analysis of DRM-containing fractions revealed that ∼5% of total CD45 can be found in Triton X-100 lipid rafts (Fig. 3 B). This compares to ∼20% of total p56^lck associating with Triton X-100 rafts in the same preparation (data not shown).

Lipid rafts were also isolated from the nontransformed, Ag-specific CTL clone AB.1. When a sample of the total protein isolated from each fraction was subjected to immunoblotting, it is clear that CD45 and Lck are detected in the buoyant fractions, whereas CD3ε is not detected in these buoyant fractions (Fig. 4,A, top panels). To confirm that the CD45 found in the raft fractions is indeed Triton X-100 insoluble, the Triton X-100-insoluble protein was recovered from each fraction. To accomplish this, each fraction was diluted in buffer containing 0.5% Triton X-100 and subjected to centrifugation. The insoluble pellet was washed and solubilized in reducing sample buffer and analyzed by immunoblotting. In the middle panels of Fig. 4,A, it can clearly be seen that CD45 and Lck, but not CD3ε, are insoluble and are therefore enriched in the raft fractions after centrifugation. In contrast, when the soluble protein remaining in the supernatant is analyzed for CD45 and Lck, they are no longer detected in the raft fractions, but are detected in the higher density fractions where the Triton X-100-soluble material is predicted to be found (Fig. 4 A, bottom panels). These results confirm that the CD45 localized to lipid rafts is indeed Triton X-100 insoluble.

We examined the localization of CD45 in a number of different cell lines by making use of the additional precipitation step to enrich for the Triton X-100-insoluble material found in each fraction. It is clear that CD45 can also be detected in the lipid rafts isolated by this method from thymocytes (Fig. 4,B). Note that there is no CD45 detected in the high-density fractions because we are only examining the Triton X-100-insoluble protein. However, in some cells we can detect a significant amount of CD45 (Fig. 4 B), p56^lck, and GM1 (data not shown) that is Triton X-100 insoluble but is present in the high-density sucrose fraction, the significance of which is not understood. The detection of Triton X-100-insoluble proteins in the high-density fractions seems to vary by the cells used for analysis rather than by extraction procedure (S. E. Edmonds and H. L. Ostergaard, unpublished observation). This does not appear to be the direct result of cytoskeletal associations because disruption of the actin cytoskeleton with cytochalasin D or Latrunculin A does not release the insoluble material from the high-density fractions (data not shown).

It has been proposed that the concentration of Triton X-100 used for detergent extraction may influence the stringency of DRM preparations, i.e., DRMs could be contaminated with non-DRM proteins if a low concentration of Triton X-100 is used (28). Here we show that similar amounts of CD45 are detected in the rafts prepared from Yac-1 cells after extraction with 1% Triton X-100 (Fig. 4 B), suggesting that 0.5% Triton X-100 is not limiting in these cells.

To confirm that the association of CD45 with DRMs was cholesterol dependent, cells were treated with MCD and the amount of CD45 associating with Triton X-100 DRMs was examined. MCD treatment had no effect on the ability to recover lipid rafts as defined by GM1 content determined by CTx reactivity (Fig. 5,A). Although roughly the same amount of raft material was isolated as determined by CTx blotting, significantly decreased levels of the characterized raft proteins Lck and LAT were observed (Fig. 5,B). CD45 was also no longer associated with the DRM fraction after MCD pretreatment (Fig. 5 B). Cell viability after MCD treatment was confirmed by the ability of the cells to exclude trypan blue and to survive and divide after culture (data not shown).

We have previously demonstrated that when cells are plated on immobilized Abs specific for CD45, the cells undergo a dramatic change in cell morphology (29). The cells spread out and flatten on the plate and become very difficult to visualize under the microscope. We decided to make use of this CD45-initatied signal to determine whether MCD has any impact on CD45-regulated events, further establishing a biological link between CD45 and lipid rafts. When Yac-1 cells were plated on immobilized anti-CD45, the cells, as previously demonstrated (29), underwent dramatic cell spreading (Fig. 5,C). Interestingly, when Yac-1 cells were first treated with 10 mM MCD, the cells did not undergo any visible cell spreading (Fig. 5 C). This result suggests that CD45 is somehow dependent on lipid rafts for the observed cell spreading, although it does not directly confirm that it is the CD45 found in lipid rafts that is important for the cell spreading.

To investigate the association of CD45 with LAT, we made use of a BWRev cell line that expresses a truncated form of CD45 in which very little of the cytoplasmic region of CD45 is expressed (30) and is therefore not an active protein tyrosine phosphatase (6). When the amount of LAT associating with immunoprecipitated CD45 from BWRev cells after lysis in Triton X-100 is compared with that of BW cells, it can be seen that there is little difference in the amount of LAT associating with CD45 (Fig. 6,A). Equivalent amounts of CD45 are immunoprecipitated from each cell line. LAT does not appear to be a direct substrate of CD45 in that no hyperphosphorylation of LAT can be detected by immunoblotting with anti-phosphotyrosine in lysates, CD45, or LAT immunoprecipitates prepared from the BWRev cell line (data not shown). To determine whether the cytoplasmic region of CD45 has any influence on the targeting of CD45 to lipid rafts, we prepared rafts from BWRev cells after extraction with Triton X-100 (Fig. 6 B). It can be seen that a significant amount of the CD45 from BWRev cells is associated with the DRM-containing fractions. Optical densitometry revealed that ∼15% of total CD45 was present in the DRM-containing fractions. Fractions containing the DRMs were confirmed by immunoblotting for LAT (data not shown). Interestingly, we could not detect any association between CD45 and p56^lck in CD45 immunoprecipitates prepared from BWRev cells after extraction in Brij58, which suggests that the cytoplasmic region of CD45 is critical in mediating this interaction (data not shown).

To determine whether stimulation of T cells had any effect on the amount of CD45 found to associate with DRMs, Yac-1 cells were stimulated by CD3 cross-linking for up to 60 min, before raft isolation. Maximal tyrosine phosphorylation was found to occur after 2 min of stimulation (Fig. 7,A). The number of tyrosine-phosphorylated proteins and the degree of phosphorylation of certain proteins were then reduced at later time points, reaching near basal levels at around 60 min. After stimulation of the cells for 2 min, the amount of CD45 present in the DRMs is very similar to that seen in resting cells (Fig. 7,B). However, after longer periods of stimulation, CD45 begins to be excluded from the DRMs, with only ∼25–55% of the amount seen in resting cells present after 60 min of stimulation. In contrast to that seen with CD45, the amount of p56^lck in the DRMs remains relatively constant throughout the period of stimulation (Fig. 7,B). The two experiments presented in Fig. 7 show the range of results from four different experiments. Therefore, it appears that CD45 remains in the rafts during the induction of tyrosine phosphorylation and begins to leave the rafts as the overall level of tyrosine phosphorylation decreases after stimulation.

It has previously been shown that p56^lck can associate with CD45 and that this association is stabilized in certain detergents (26, 31). Here we show that p56^lck is indeed present in CD45 immunoprecipitates prepared in the presence of Brij 58 and to a lesser extent Brij 97. However, very little p56^lck coimmunoprecipitates with CD45 after extraction with Triton X-100. We also show that LAT is also present in CD45 immunoprecipitates, and again this is dependent upon the detergent used for extraction. In contrast to p56^lck, a significant amount of LAT associates with CD45 in the presence of Triton X-100. This suggests that p56^lck, CD45, and LAT may not associate together in a larger complex, but there are different CD45 pools associating with different proteins. This possibility is supported by a recent study showing that LAT and Lck can be found in distinct lipid rafts prepared with Brij 58 (32).

Using the BWRev cell line, which expresses a truncated form of CD45 missing the cytoplasmic region, we show that the association between LAT and CD45 is preserved. Therefore, the association of LAT with CD45 is unlikely to be mediated by the cytoplasmic domain of CD45, but an interaction between the transmembrane regions of the two proteins cannot be discounted. However, the interaction between CD45 and LAT would appear to be cholesterol dependent because MCD eliminates the association. Therefore, we propose that the association between CD45 and LAT is unlikely to be a direct interaction but is probably by virtue of the association of the two molecules with detergent-insoluble microdomains. This is in contrast to the predominant interaction between p56^lck and CD45, which is unaffected by MCD and is therefore unlikely to be DRM dependent. This is consistent with previous reports showing that recombinant p56^lck can bind to recombinant cytoplasmic region of CD45 in vitro (33).

There has been much speculation as to whether CD45 is associated with DRMs. Based on previous studies of Jurkat T cells using confocal microscopy of CTx-patched rafts (16) and the isolation of Triton X-100-insoluble buoyant fractions (13, 19), it has been suggested that CD45 is excluded from lipid rafts. However, using Brij 58 extraction of proteins from thymocytes, there is a weak association of CD45 with DRMs (14). Here we show that CD45 can be detected in Triton X-100 DRMs. Furthermore, the association of CD45 with Triton X-100 rafts is cholesterol dependent in that it is abolished by pretreatment of the cells with the DRM-disrupting agent MCD. Therefore, CD45 is not completely excluded from DRMs, but because only 5% of total CD45 is localized to DRMs, it is also not enriched in DRMs. However, due to the abundant expression of CD45, 5% of total CD45 represents a large number of CD45 molecules being present in DRMs, which would likely be sufficient to regulate tyrosine phosphorylation within DRMs.

Extensive research has been performed to establish structural criteria for proteins associated with DRMs. Proteins which are dual acylated, such as p56^lck and LAT, are thought to pack well into the ordered lipid environment of the DRMs, whereas transmembrane proteins are considered too bulky and therefore difficult to accommodate in the DRM (34). Although LAT is a transmembrane protein, it is still DRM associated. Mutation of palmitoylation sites of LAT has been shown to abolish its association with DRMs (20). Therefore, it would appear that lipid modification is the overriding structural characteristic allowing LAT to localize to DRMs. No acylation has been demonstrated for CD45, although it has been reported that CD45 possesses an unusual sphingolipid-like modification, which may promote the association of CD45 with DRMs (35). However, it is now emerging that lipid modification may not be a required characteristic in order for proteins to be localized to DRMs. It has been shown that other non-lipid-containing transmembrane proteins such as CD44 can be DRM associated (36).

There is much conjecture as to the role of lipid rafts in T cell activation. It has been postulated that rafts act as signaling platforms that become clustered upon T cell engagement with APCs (28, 37). When a T cell encounters Ag presented on APCs or model membranes, molecules at the contact site segregate into a defined supramolecular activation cluster (SMAC), with TCR and protein kinase C-θ forming a central cluster and LFA-1 forming an outer cluster (38, 39). Although CD45 is essential for the initiation of signaling events, it has been suggested that CD45 can also be a negative regulator of signaling (40) and is predicted to be excluded from the SMAC, in part because of the large size of the external domain (41). Consistent with this prediction, it has been reported that CD45 is indeed excluded from the SMAC (42, 43). However, another study found that CD45 is initially excluded from the SMAC, but over time a small portion of CD45 migrates to a region in close proximity to the SMAC, suggesting that CD45 may play a role in regulating the activity of molecules within the SMAC (44). Interestingly, the amount of total CD45 that is found in close proximity to the SMAC appears to be similar to the proportion of CD45 that we find within lipid rafts. Recently it was shown that CD45, along with CD3, CD4/8, talin, and protein kinase C-θ, polarizes at the interface between Ag-independent T cell and dendritic cell conjugates (45) and that these Ag-independent interactions can initiate signals such as Ca²⁺ flux and tyrosine phosphorylation within the T cell (45, 46). That this occurs in the absence of Ag or MHC (45) suggests that molecules besides the TCR are likely mediating these signals. We have found that Abs to the external domain of CD45 induce tyrosine phosphorylation and cytoskeletal changes, demonstrating that CD45 is capable of signaling (47). Interestingly, MCD disrupts the CD45-triggered cytoskeletal changes (Fig. 5 C), implying that induction of CD45-initated signals is raft dependent. It will be of interest to determine whether raft-associated CD45 is one source of the Ag-independent signals observed in T cells bound to dendritic cells.

It remains possible that there is no functional relationship between lipid rafts and SMACs. It is plausible that clustering of rafts concentrates a number of signaling molecules at the site of contact, thereby leading to some signaling events, but does not form any higher order structure within the SMAC. The rafts may therefore serve as a source of activated kinases to carry out signals initiated through the clustered TCR complexes. Interestingly, it has been shown that the negative regulator of Src family kinases, Csk, is also localized to DRMs by virtue of its association with the raft-associated adapter protein Cbp/phosphoprotein associated with glycosphingolipid-enriched microdomains (PAG) (48). Cbp/PAG becomes rapidly dephosphorylated after T cell stimulation, leading to the release of Csk (49). CD45 therefore may be localized to lipid rafts to counter the negative regulatory properties of Csk on Src family kinases. It is also possible that CD45 dephosphorylates Cbp/PAG, leading to Csk release, thereby favoring dephosphorylation of the negative regulatory site of Lck, upon T cell activation. CD45 appears to remain in rafts during the period of induction of tyrosine phosphorylation (Fig. 7), consistent with a positive regulatory role in activation. CD45 may then move out of the rafts to allow the basal phosphorylation to be reestablished and may help explain why cells are refractory to induction of tyrosine phosphorylation if they are stimulated shortly (around 1 h) after activation.

Clearly, the precise role of lipid rafts in T cell function needs to be elucidated. First, it appears that the structural criteria for the localization of molecules to rafts is not definitive. The association of any molecule with lipid rafts cannot be ruled out based on the absence of lipid moieties. Second, it would appear that rafts do play a major role in T cell signaling. However, whether rafts are involved in the formation of the SMAC and the role(s) rafts play in initiating the membrane-proximal events remain to be elucidated. Here we clearly demonstrate that CD45 is not excluded from DRMs but, unlike that seen with p56^lck, the association of CD45 with DRMs, and possibly lipid rafts, is dynamic. This now opens the way for the dissection of the function of CD45 in lipid rafts in the context of T cell activation.

We thank Barbara Simon for excellent technical support, Dr. Kevin Kane and Troy Baldwin for critical reading of the manuscript, and Dr. Andrew Shaw for helpful discussion.