Novel Redistribution of an Intracellular Pool of CD45 Accompanies T Cell Activation*

The major tyrosine phosphatase activity against angiotensin detected in membranes of the antigen-spe- cific T cell hybridoma 2B4 is contained in the cytoplasmic tail of the CD45 molecule. When these cells are stimulated with either an antibody directed against the T cell antigen receptor or an activating anti-Thy-1 antibody, there is a rapid redistribution of CD45 in the cells. The redistribution can be observed in two ways: morphology and subcellular fractionation. Morpho- logic examination of resting cells reveals intense CD45 staining of the Golgi as well as surface staining. Upon activation the Golgi is rapidly cleared of CD45. This redistribution is specific for CD45 and is not observed for an intrinsic Golgi protein, mannosidase 11, or a protein traversing the secretory pathway, the T cell receptor. In activated cells, in contrast to resting cells, approximately 30% of the total cellular CD45 is pre-cipitated either at 280 X g or at 200,000 X g through a 2.2 M sucrose cushion after cell homogenization. This fraction is not accessible to cell surface labeling. CD45 redistribution does not require hydrolysis of phosphatidylinositides and cannot be reproduced by the addition of phorbol ester and calcium ionophore. It does require the presence of an intact functional T cell receptor on the cell surface. These studies suggest that the residence time of CD45 within an intracellular organelle can be acutely regulated by a signal mediated via the T cell receptor. This regulation may control access of this phosphatase to critical substrates.

The major tyrosine phosphatase activity against angiotensin detected in membranes of the antigen-specific T cell hybridoma 2B4 is contained in the cytoplasmic tail of the CD45 molecule. When these cells are stimulated with either an antibody directed against the T cell antigen receptor or an activating anti-Thy-1 antibody, there is a rapid redistribution of CD45 in the cells. The redistribution can be observed in two ways: morphology and subcellular fractionation. Morphologic examination of resting cells reveals intense CD45 staining of the Golgi as well as surface staining. Upon activation the Golgi is rapidly cleared of CD45. This redistribution is specific for CD45 and is not observed for an intrinsic Golgi protein, mannosidase 11, or a protein traversing the secretory pathway, the T cell receptor. In activated cells, in contrast to resting cells, approximately 30% of the total cellular CD45 is precipitated either a t 280 X g or at 200,000 X g through a 2.2 M sucrose cushion after cell homogenization. This fraction is not accessible to cell surface labeling. CD45 redistribution does not require hydrolysis of phosphatidylinositides and cannot be reproduced by the addition of phorbol ester and calcium ionophore. It does require the presence of an intact functional T cell receptor on the cell surface. These studies suggest that the residence time of CD45 within an intracellular organelle can be acutely regulated by a signal mediated via the T cell receptor. This regulation may control access of this phosphatase to critical substrates.
The regulation of protein phosphorylation on tyrosine residues is believed to play a crucial role in intracellular signal transduction pathways that regulate cellular activation and differentiation Cooper, 1985, 1986). Although the identification and biochemical characterization of tyrosine kinases have advanced rapidly over the past several years, relatively less is known about tyrosine phosphatases. Considerable effort is currently being directed toward the purification and characterization of cytosolic and membrane protein tyrosine phosphatases (PTPases)' (Lim-Tung and Reed, * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. j: Recipient of a postdoctoral fellowship from the Arthritis Foundation.
The abbreviations used are: PTPase(s), protein tyrosine phosphatase(s); SDS, sodium dodecyl sulfate; Hepes, 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride; PBS, phosphate-buffered saline; ER, endoplasmic reticulum; TCR, T cell antigen receptor. Shriner and Brautigan, 1984;Tonks et al., 1988aTonks et al., , 1988bJones et al., 1989). Recently, a major advance in our understanding of tyrosine phosphatases and their potential role in transmembrane signaling has been reported. Tonks et al. (1988aTonks et al. ( , 1988b have isolated and sequenced a molecular clone encoding a tyrosine phosphatase from human placenta. When the sequence was compared with known protein sequences a striking similarity to the cytoplasmic domain of the common leukocyte surface antigen CD45 was noted (Charbonneau et al., 1988). This was followed by the demonstration of PTPase activity of CD45 (Tonks et al,, 1988c(Tonks et al,, , 1990. These findings raise the possibility of a direct role of tyrosine phosphatases in surface activation events. CD45 is a so-called leukocyte common antigen and is expressed in cells of the hematopoietic lineage including lymphoid and myeloid cells (Trowbridge, 1978;Thomas, 1989). One of its striking characteristics is the lineage-specific expression of various isoforms. The cell type-specific expression of CD45 variants demonstrates an exquisite level of differentiation (Thomas, 1989). Thus B cells express a different form than T cells, and different subsets of T cells express different CD45 forms. These varieties arise as a result of alternative splicing such that various combinations of three exons are found in the mRNA encoding this protein (Saga et al., 1987). This alternative splicing results in the expression of isoform-specific epitopes and altered biochemical characteristics. The latter is manifested by apparent molecular masses, as determined by SDS-polyacrylamide gel electrophoresis, that vary between 180 and 200 kDa. Much of this heterogeneity is the result of isoform-specific glycosylation differences although there are also differences in the core polypeptide size.
Recent studies have demonstrated the efficacy of anti-CD45 antibodies in modulating T cell activation (Ledbetter et al., 1988). These studies emphasized the potential importance of the structural proximity of CD45 to the T cell antigen receptor for the observed effects on cellular activation. The recent reports pointing to a possible requirement for CD45 for T cell activation emphasized further the importance of tyrosine phosphatase activity (Pingel and Thomas, 1989;Koretzky et al., 1990). As with kinases, establishing the mechanisms underlying the regulation of tyrosine phosphatase function poses a major challenge.
Phosphatases can be envisioned to be regulated by altering their intrinsic activity, by inhibitors and/or activators, or by determining their access to substrates. The final control could be accomplished by altering the subcellular localization of the phosphatase. In this paper we report that CD45, which is the major membrane tyrosine phosphatase in a T cell hybridoma, undergoes a novel and rapid intracellular redistribution in response to cell activation.
Cells and Antibodies-The T cell hybridoma 2B4 cells and LSTRA cells were maintained continuously under the conditions described previously (Samelson et al., 1985a;Hellstrom et al., 1979). The 6deficient T cell line 21.2.2, {-deficient T cell line MA5.8, and 7deficient T cell line EV3 were derived by repetitive subcloning of 2B4 cells (Sussman et aL, 1988) and were kindly provided by Dr. J. Ashwell (NCI). The cell line 2A7 was obtained by transfection of MA5.8 cells with the { chain gene. The src-transfected 2B4 cell line was obtained by retroviral transfection. A2B4-2 is a monoclonal mouse IgG2a that reacts with a clonotypic determinant of the 2B4-a chain (Samelson et al., 1983). 145-2C11 (2Cll) is a hamster monoclonal antibody that binds the murine CD3-e chain (Leo et al., 1987). G7 (rat IgG2C) is an activatable monoclonal antibody that binds a nonpolymorphic determinant on mouse Thy-1 (Gunter et al., 1984). Either of two nonactivating antibodies was used as a control, 3Hll an anti-Thy-1 antibody or 10.2.16, an antibody to Aka:Akp. Anti-CD45 monoclonal antibodies used were M1/9.3.4 (M1/9) (Springer et al., 1978) and 55.10.1. Rabbit anti-mouse mannosidase I1 antibody was a generous gift from Dr.
Phosphatase Assays-Phosphotyrosyl [Vals]-angiotensin I1 phosphatase assays were carried out using two methods. In most of the experiments, phosphotyrosyl [Vals]-angiotensin I1 phosphatase activity was measured by the release of 32P04 from the 32P-labeled [Val5]angiotensin I1 peptides. The phosphatase reaction took place in a volume of 50 pl containing 50 mM Hepes, pH 7.4,0.5% Triton X-100, 1 mM PMSF, 10 pg/ml leupeptin, and 10 pg/ml aprotinin. After preincubation of the buffer and cellular phosphatase source for 5 min at 30 "C, the reaction was initiated by the addition of [32P]phosphotyrosyl [Val6]-angiotensin 11. The reaction was terminated at varying times from 0 to 30 s by the addition of a 5% activated charcoal suspension (550 pl). Addition of the charcoal solution to the reaction mixture quenched the phosphatase activity immediately. The activated charcoal absorbed 32P-labeled [Val5]-angiotensin I1 peptides but could not bind the released phosphate. Therefore, after charcoal particles were removed by centrifugation, aliquots of the supernatant (200 pl) were counted for radioactivity in 10 ml of scintillation fluid. The second method of assaying phosphotyrosyl [Vals]-angiotensin I1 phosphatase activity assessed the loss of "P-labeled peptides by a phosphocellulose binding assay.
Preparation of Cell Lysates and Subcellular Fractions-The cells were solubilized in a lysis buffer of 0.5% Triton X-100,50 mM Hepes, pH 7.4, with 1 mM PMSF, 10 pg/ml leupeptin, and 10 pg/ml aprotinin, and cell lysates (postnuclear supernatants) were prepared by centrifugation at 12,000 X g for 15 min. For preparation of subcellular fractions, the 2B4 cells were washed with ice-cold PBS and then resuspended in 50 mM Hepes, pH 7.4, containing a 0.5 mM MgC12, 1 mM PMSF, 10 pg/ml leupeptin, and 10 pg/ml aprotinin for 10 min at 4 "C. Subsequently the cells were homogenized with 20-30 strokes of a tight fitting Dounce homogenizer. After restoration to isotonicity by the addition of a one-third volume of 600 mM NaCl, the homogenate was centrifuged at 280 X g. The 280 X g supernatant was made 5 mM in EDTA and was centrifuged at 150,000 X g for 45 min. The membrane pellet was lysed in 0.5% Triton X-100, 50 mM Hepes, pH 7.4, containing the above protease inhibitors. In some experiments the membrane pellet was solubilized in the above lysis buffer containing 150 mM NaCl. The 150,000 X g supernatant (cytosolic fractions) was made in 0.5% Triton X-100.
Sucrose Density Fractionation-The Dounce homogenate of the 2B4 cells (after restoration of the isotonicity) was prepared as described above. The homogenate (1.0 ml) was gently layered onto 4 ml of 2.2 M sucrose in 50 mM Hepes, pH 7.4, 150 mM NaC1, containing 0.5 mM MgCl,, 1 mM PMSF, 10 +g/ml leupeptin, and 10 pg/ml aprotinin. After centrifugation for 25 min at 200,000 X g, the supernatant was removed. The 200,000 X g pellet was washed once with 50 mM Hepes, pH 7.4, 150 mM NaC1, containing 0.5 mM MgC12, 1 mM PMSF, 10 pg/ml leupeptin, and 10 pg/ml aprotinin to remove sucrose. We will refer to this fraction as the 200 suc-pellet. The 200 suc-pellet was subsequently solubilized with 0.5% Triton X-100 in the above buffer (without MgCl,).
Labeling, Fractionation, Immunoprecipitation, and Electrophoresis-For steady-state metabolic labeling, the 2B4 cells (106/ml) were labeled for 8 h with ["sS]methionine in methionine-free RPMI 1640 medium supplemented with 1 pg/ml methionine and 10% fetal calf serum. After labeling, cells were washed with medium containing methionine and preincubated for 30 min at 37 "C either with G7 ascites (1:50 dilution) or with irrelevant ascites (1:50 dilution) as a control. After activation, preparation of total cell lysates and lysates of the 200 suc-pellet were carried out as described previously. Immunoprecipitations were performed with the indicated monoclonal antibodies or rabbit antisera absorbed to protein G-agarose. Immunoprecipitates were washed three times in Hepes buffer containing 0.1% Triton X-100 and washed once for 5 min at room temperature in the above buffer containing 0.1% SDS, 0.1% deoxycholate. Bound antigen was eluted from the beads by boiling in SDS sample buffer with 3% (v/v) 2-mercaptoethanol. Purifiedproteins were analyzed by SDS-polyacrylamide gel electrophoresis on one-dimensional acrylamide gels (7.5%) using the buffer systems of Laemmli as described by Samelson et al. (1985b). Surface iodination of the 2B4 cells was performed using lactoperoxidase-catalyzed reactions described by Samelson et al. (1985b).
Immunofluorescence-Cells were grown on 12-mm glass coverslips overnight and then incubated for 30 min at 37 "C either with anti-Thy-1, G7 ascites (1:50 dilution), or with irrelevant ascites (1:50 dilution) as a control. Alternatively, cells were incubated for 30 min at 37 "C on a glass coverslip that had been coated with either 2Cll or an irrelevant antibody. Cells were then washed with PBS and fixed with 2% formaldehyde in PBS for 15 min at room temperature. After permeabilization with 0.1% saponin, 1% bovine serum albumin in PBS, cells were incubated with biotin-labeled M1/9 and anti-mannosidase I1 antibody for 1 h at room temperature. After washing off the unbound antibodies, cells were incubated for 45 min at room temperature with rhodamine-avidin and fluorescein-conjugated goat anti-rabbit IgG. After washing twice with 4 X SSC for 2 min, coverslips were mounted on glass slides using Fluoromount G (Southern Biotechnology Associates, Birmingham, AL) and examined with an inverted microscope (ICM 405, Cal Zeiss, Inc., Thornwood, NY).
Electron Microscopy-The ultrastructural localization of CD45 and CD3-t was studied in 2B4 cells by immunoperoxidase electron microscopy using biotin-labeled antibodies, M1/9 and 500A2. Cells were grown on 12-mm glass coverslips overnight and incubated for 30 min at 37 "C either with anti-Thy-1, G7 ascites (1:50 dilution) or with irrelevant ascites (1:50 dilution) as a control. Cells were washed with PBS, fixed with formaldehyde, permeabilized, incubated with biotinlabeled M1/9 and 500A2 followed by peroxidase-conjugated streptavidin, and reacted with diaminobenzidine hydrochloride/H202. The cells were then prepared for electron microscopy as described by Yuan et al. (1987).

CD45 Is the Major Membrane Tyrosine Phosphatase in T
Cell Membranes-T cell tyrosine phosphatase (PTPase) activity was detected using tyrosine-phosphorylated [Val5] -angiotensin I1 as a substrate. The results obtained using the activated charcoal method were similar t o those obtained using the P-81 phosphocellulose paper binding method (Table  CD45 Redistribution on T Cell Activation 12.008 I). Detergent lysates of 2B4 T hybridoma cells dephosphorylated the substrate readily. Phosphatase activity curves were linear with respect to lysate concentration. Furthermore, a t all concentrations of lysate, the specific release of '"P-labeled phosphate from phosphotyrosyl [Va15]-angiotensin I1 increased linearly during the reaction periods employed. When these cells were separat,ed into membrane and cytosolic fractions, more than 80% of the total cellular phosphotyrosyl angiotensin I1 phosphatase activity was found associated with the particulate fraction. All of the PTPase activity against angiotensin is found in the Triton X-100 cell lysate supernatant after a 280 X g centrifugation. The specific activity of the PTPase activity in the membrane and cytosolic fractions was 2.9 and 0.48 nmol of PO, released/min/mg of protein, respectively. Moreover, fractionation of both the intact 2B4 cells and fresh membranes using Triton X-114 phase separation (Bordier, 1981) indicated that the membrane-associated tyrosine phosphatase activity partitioned into the detergent phase (data not shown), suggesting that the enzyme is a relatively hydrophobic protein.
The recent identification of CD45 as a tyrosine phosphatase led us to examine whether any of the PTPase activity detected in these cells was attributable to this molecule (Tonks et al., 1988c(Tonks et al., , 1990. A rat monoclonal antibody, M1/9, directed against murine CD45, was used to precipitate CD45 from solubilized membranes of 2B4 cells. The remaining PTPase activity was compared with a control immunodepletion. At saturating concentrations of anti-CD45, between 60 and 70% of angiotensin I1 PTPase activity was specifically removed from membrane lysates of control 2B4 cells (see Fig. 3A, open and closed circles). T o ensure that the antibody did not inactivate the activity, PTPase was measured in the immunoprecipitate. Although no activity was detected in control precipitates, PTPase activity was recovered quantitatively on the CD45 beads (Fig. 1B). The M1/9 antibody appeared to be specific for CD45 as determined by SDS-polyacrylamide gel electrophoresis analysis of radioiodinated surface proteins immunoprecipitated from 2B4 cells with the M1/9 antibody. Only one labeled band of 180-kDa corresponding to the molecular mass of CD45 was isolated (Fig. 1A).
Effect of T Cell Activation on Membrane Tyrosine Phosphatase Activity-We have shown previously that activation of T cells with a variety of agents leads to the rapid tyrosine phosphorylation of the { chain of the T cell receptor as well as of other cellular substrates (Samelson et al., 1986;  (0) or with control antibody-conjugated beads (M). Tyrosine phosphatase assays were performed using 4 PM phosphotyrosyl [Val']-angiotensin I1 as a substrate. The amount of phosphate released is calculated relative to the amount of total solubilized cell protein used for immunoprecipitation. Klausner et al., 1987;Baniyash et al., 1988;Hsi et al., 1989). Because activation results in the rapid change of tyrosine phosphate content of a variety of proteins, we were interested in determining whether activation resulted in any alteration in CD45 phosphatase activity. T cells were activated with an anti-Thy-1 monoclonal antibody termed G7 (Gunter et al., 1984). This antibody has been demonstrated previously to activate 2B4 T cells  and does not require presenting cells to achieve activation. T cells were treated with either G7 or a variety of monoclonal antibodies which fail to activate the cells. Membranes were prepared from a 280 X g supernatant of Dounce homogenized cells and assayed for angiotensin tyrosine phosphatase activity. We noted a 20% drop in the specific activity of phosphate release only after activation with G7 (Fig. 2B). When antibody was added to isolated 2B4 membranes no change in phosphatase activity was observed (data not shown). If cells were instead lysed directly in Triton X-100 and assayed, G7 stimulation resulted in no change in phosphatase activity (Fig. 2 A ) . The missing activity in the membrane fraction could be accounted for by examining the 280 X g pellet obtained after Dounce homogenization of the cells. When this fraction was assayed after solubilization in Triton X-100 a significant increase in phosphatase activity was observed after G7 activation. This effect of cell activation on the PTPase activity could be observed more clearly after centrifugation through 2.0 or 2.2 M sucrose at 200,000 X g (Fig. 2C). The 200,000 X g sucrose pellet (200 suc-pellet) showed virtually no phosphatase activity in unactivated cells whereas after activation, up to 20% of total cellular angiotensin phosphatase activity was found in this fraction. These changes were complete within minutes of the addition of G7.
We next asked whether the redistributed PTPase activity included CD45. When membranes were assayed for PTPase activity, membranes from activated cells contained about 20-25% less total activity (Fig. 3A). After depletion with M1/9 antibody beads, the PTPase activity from control and activated membranes was nearly identical (Fig. 3A). This demonstrated that the vast majority of the redistributed PTPase activity was CD45. T o demonstrate this directly the pelleted  PTPase activity was subjected to immunodepletion with the M1/9 beads. This resulted in 100% depletion of PTPase activity (Fig. 3B). Thus, all of the redistributed PTPase activity was attributable to CD45. Activation Results in Rapid Depletion of CD45 from the Golgi Region-We next examined the intracellular distribution of CD45 by immunofluorescence microscopy in 2B4 cells. These cells displayed diffuse surface staining under conditions in which they were not permeabilized. When permeabilized with saponin an intracellular staining pattern was observed. This intracellular CD45 was found in a structure that colocalized with the region of the cell containing the Golgi apparatus (Fig. 4). The latter was identified with a polyclonal antibody against the Golgi enzyme mannosidase 11. Little if any CD45 was found in the ER and nuclear envelope, and its distribution did not overlap that of antibodies directed against resident proteins of the ER (data not shown). When the T cells were activated either with G7 or with 2C11-coated coverslips, the morphology of the cell changed rapidly as the cell rounded up and became swollen. The Golgi apparatus was now found over the nucleus in the center of the cell and remained heavily stained by the mannosidase I1 antibodies. However, the strong Golgi staining region of CD45 seen in control cells became more diffuse in the area of the Golgi. These morphological changes were not observed in cells treated with 3Hl1, nonstirnulatable anti-Thy-1 antibody (data not shown).
The intracellular distribution of CD45 was examined at the ultrastructural level using immunoperoxidase staining. As shown in Figs. 5A and 6A all cisternae of the Golgi apparatus were labeled by anti-CD45 with no detectable labeling of the ER or nuclear envelope. This pattern can be contrasted to the intracellular localization of the TCR which showed a more common pattern of proteins in transit through the secretory pathway. Thus, staining with either an anti-a or an anti-CD3-6 monoclonal antibody revealed ER, partial Golgi staining, and peripheral vesicle staining. In addition, significant lysosomal staining was observed (Fig. 5B). When the cells were examined by electron microscopy after G7 activation, there was a dramatic loss of the Golgi staining pattern (Fig.  6B). In contrast to the rapid loss of Golgi staining of CD45 upon activation, no activation-induced change in the intracellular staining pattern of the TCR-e chain was seen in the same cells (Fig. 6D).
Metabolic pulse-chase studies failed to demonstrate any differences in the rate of synthesis or carbohydrate processing of CD45 between control and activated cells (data not shown). The latter was measured by the rate of acquisition of resistance to endoglycosidase H and of susceptibility to neuraminidase as assessed by two-dimensional gel electrophoresis. The disappearance of the heavy pan-Golgi staining by CD45 after activation was not a result of inhibition of CD45 biosynthesis with subsequent clearing of the secretory pathway. We can achieve clearing of the Golgi staining pattern by inhibition of new protein synthesis but only after several hours of treatment with cycloheximide. Concomitant with this, we observed a loss of the activation-induced redistribution of CD45 to the 200 suc-pellet. The half-time for the loss of redistribution (measured by phosphatase activity) was approximately 2 h. Less than 5% of total cellular mannosidase 11 was detected in the 200 suc-pellet both before and after cell activation (see min a t 37 "C. After activation, membrane lysates were immunodepleted either with anti-CD45 antibody-conjugated beads (W) or with control antibody-conjugated beads (0) as a mock depletion. Membrane lysates from an equal number of 2B4 cells, treated with irrelevant antibody for 30 min a t 37 "C, were immunodepleted either with anti-CD45 antibody-conjugated beads (0) or with control antibodyconjugated beads (0). The supernatants were subjected to phosphatase assays using 4.8 PM phosphotyrosyl [Val5]-angiotensin I1 as a substrate. B , 2B4 cells were treated with G7 for 30 min a t 37 "C. After activation lysates of the 200 suc-pellets were prepared as described under "Experimental Procedures" and were immunodepleted either with anti-CD45-conjugated beads (W) or with control antibody-conjugated beads (0) as a mock depletion. Lysates of the 200 suc-pellets from an equal number of 2B4 cells, treated with irrelevant antibody for 30 min a t 37 "C, were immunodepleted either with anti-CD45 antibody-conjugated beads ( 0 ) or with control antibody-conjugated beads (0). Phosphatase assays were performed using 1 PM phosphotyrosyl [Val"]-angiotensin I1 as a substrate. Fig. 7 B ) . Thus, the internal pool of CD45 which was detected after activation did not represent the coprecipitation of the This was demonstrated by surface iodination of resting 2B4 cells followed by activation with G7 and fractionation of the 200 suc-pellet (Fig. 7A). The 200 suc-pellet had virtually no radioactivity as determined by y counting and no labeled immunoprecipitable CD45. If cells were first activated and then subjected to surface iodination, again no labeled CD45 was detected in the 200 suc-pellet. In contrast, when these cells were labeled overnight with ["S]methionine, we could readily detect CD45 in the 200 suc-pellet after G7 activation (Fig. 7B). Although a small amount of CD45 could be found in this fraction from unactivated 2B4 cells, much greater amounts could be recovered after activation. Approximately 30% of the total labeled CD45 redistributed to the 200 sucpellet after activation. This fit well with the percentage found in this fraction when assessed by phosphatase activity. Thus, no change in specific activity of the phosphatase was observed during this redistribution. These findings suggested that the activation-induced redistributed CD45 was derived from a pool of the phosphatase which was inaccessible to external labeling. This differed from observations on neutrophils in which an increase in cell surface expression of CD45 was observed during activation (Lacal et al., 1988).
Role of the T Cell Antigen Receptor in CD45 Redistribution-We addressed the question of the role of the TCR in CD45 redistribution in this hybridoma by examining variants of 2B4 that lack surface TCR molecules because of defined genetic defects. 21.2.2 cells are 2B4 variants that lack expression of TCR / 3 chains in which cell surface TCR expression can be reconstituted by the introduction of the /3 chain gene.
When these cells were fractionated to look for CD45 redistri- pool. A, 2B4 cells were surface iodinated as described under "Experimental Procedures." Cells were then incubated either with GT or with irrelevant antibody (ctrl) for 30 min a t 37 "C. After activation, total cell lysates and lysates of the 200 suc-pellet were prepared as described under "Experimental Procedures," subjected to immunoprecipitation either with M1/9.3.4 (anti-CD45 antibody) or with A2B4-2. One-dimensional gel electrophoresis under reducing condition was performed using 7.5% acrylamide. B, steady-state metabolic labeling of the 2B4 cells was performed as described under "Experimental Procedures." After activation with either G7 or irrelevant antibody (ctrl), total cell lysates and lysates of the 200 suc-pellet were prepared as described previously. Immunoprecipitations were carried out using M1/9.3.4 (CD45) or anti-mannosidase I1 antibody (rnan.ll), or irrelevant antibody (ctrl) and analyzed by reducing SDS-polyacrylamide gel electrophoresis (7.5% acrylamide). bution, two interesting differences from parental 2B4 cells emerged: 1) G7 induced no CD45 redistribution; and 2) the unstimulated cells were indistinguishable, in terms of CD45 distribution, from the activated, receptor-positive parental cell (Fig. 8). T o confirm this observation we examined another 2B4 variant, MA5.8, which expresses no { chain of the TCR. This cell possesses about 3-5% of the number of cell surface receptors present on 2B4, but these are abnormal in terms of both structure (they lack 5') and function. As in the 21.2.2 cells these cells demonstrated no redistribution of CD45 in response to G7, and whether stimulated or not they have a distribution of CD45 indistinguishable from activated 2B4 cells. When the MA5.8 cells are reconstituted structurally with a cDNA clone encoding { surface receptor expression is restored (Weissman et al., 1989). When these cells (2A7 cells) were evaluated for their distribution of CD45 a partial reconstitution of the 2B4 phenotype was observed. { expression in these transfected cells appears to be unstable, and when the experiments show in Fig. 8 were performed approximately 30% of the cells in the population were TCR negative as judged by fluorescence-activated cell sorting. When we correct the CD45 signal for this subpopulation we can calculate that the { transformants have resulted in the reconstitution of about 80% of the CD45 response seen in 2B4 cells.
Signaling Pathways and the Redistribution of CD45"Multiple signaling pathways emanating from the TCR have been defined. These include the breakdown of phosphoinositides and the activation of one or more tyrosine kinases. Activation of protein kinase C and a rise in intracellular calcium (at least in part) result from the former pathway. The consequences of this pathway can be mimicked by the addition of phorbol esters and calcium ionophores. Incubation of 2B4 cells with 100 pg/ml phorbol myristate acetate or with 1 p M ionomycin or with a combination of these did not result in redistribution of CD45. We have demonstrated recently that the tyrosine kinase pathway can be mimicked partially (in terms of substrate phosphorylation) by transfecting 2B4 cells with the cDNA encoding v-src (O'Shea et al., 1991). Transfection of 2B4 cells with v-src did not result in the redistribution of CD45. Incubation of 2B4 cells with 0.5 mM dibutyryl CAMP, an analog of CAMP, another common intracellular second messenger, also failed to stimulate the redistribution of CD45. In addition, none of these maneuvers abrogated the structural response to G7 stimulation. Recently we have identified variants of 2B4 which are deficient in the 9 chain of the TCR (Mercep et at, 1988). These cells break down little if any phosphatidylinositides in response to activating signals.

on T Cell Activation
When one of these lines, EV3, was tested for its ability to redistribute CD45 we found this response to be intact (see Fig. 8). Thus, although EV3 had less total CD45 PTPase activity than 2B4, a marked change in the distribution of CD45 was observed after treatment with G7. Thus, the 7 chain of the TCR does not appear to be required for this phenomenon.

DISCUSSION
Early Cellular Events of T Cell Activation-The studies reported here provide a new and unusual addition to our identification of the early events of T cell activation. A variety of phenomena have been described over the past several years which take place within seconds to minutes of the stimulation of surface molecules on T cells (Samelson et al., 1985aOettgen et ab, 1985;Gardner et al., 1989). These include the breakdown of phosphatidylinositides, external calcium influx, release of internal calcium stores into the cytosol, the activation of one or more tyrosine kinases, and the phosphorylation of a variety of cellular substrates by both serine and tyrosine kinases. It is still unclear exactly how these multiple signals are translated into the complex pattern of gene expression and altered proliferative status which characterizes the activated T cell. In addition to these early biochemical events, the stimulated T cell undergoes a marked and rapid morphological change that includes swelling, microvillus formation, and reorientation of the Golgi apparatus as Kupfer and Dennert (1984) observed in cloned cytotoxic T lymphocytes. Recently, Lee and co-workers (1988) reported that a peri-Golgi accumulation of spectrin rapidly reorganizes beneath the plasma membrane upon activation of a T cell hybridoma. It is not yet apparent how the currently described biochemical pathways relate to these structural changes. To this list of structural changes we can add the redistribution of at least a portion of the cellular population of CD45. Direct stimulation of the T cell receptor was effective in inducing the change.
The ability to change the intracellular distribution of CD45 acutely in response to the stimulatory anti-Thy-1 antibody clearly depends upon the expression of functional T cell receptors on the cell surface. This was demonstrated by studying a variety of mutants (or variants) of the 2B4 hybridoma. In cells that failed to express parental levels of full receptor complexes because of the absence of 5; a, p, or 6, no effect of the antibody was seen. When the receptor complex was reconstituted in these cells by cDNA or gene transfection, the CD45 redistribution phenomenon was restored. Interestingly, in cells that lack the newly described 9 chain, CD45 redistribution did take place. These cells failed to break down phosphatidylinositides or to activate protein kinase C in response to G7 stimulation (Mercep et al., 1988). Thus, the redistribution of CD45 in 7-negative cells is consistent with our failure to mimic CD45 redistribution with added phorbol esters and calcium ionophores. A surprising aspect of the phenotype of the T cells lacking surface TCR was that CD45 appeared constitutively redistributed, as assessed by subcellular fractionation. This observation is reminiscent of the recent findings concerning spectrin redistribution in which a TCR-negative variant of the T cell hybridoma demonstrated constitutive redistribution to the plasma membrane, yielding a phenotype indistinguishable from the activated parental TCR-positive cells (Lee et al., 1988). These observations suggest the intriguing possibility that at least in T cell hybridomas, the presence of a functional surface TCR complex may exert a tonic signal that affects the structure of the cell.
Regulating the Residence Time of a Protein within a n Organelle-Perhaps the most surprising finding of these studies is the nature of the change of the intracellular distribution of CD45 upon activation. Immunoelectron microscopy reveals intense staining of the plasma membrane and of the entire Golgi. Little if any staining is observed in the endoplasmic reticulum. The steady-state distribution of components of the TCR can be compared and contrasted with that of CD45 in these cells. For the former, one observes staining of the ER and only spotty staining of the Golgi as well as staining of peripheral organelles. Pulse-chase studies of the carbohydrate processing of CD45 in these cells revealed that the half-time for the acquisition of endoglycosidase H resistance was approximately 10 min, consistent with rapid export from the ER. A rough estimate of the residence time in the Golgi was obtained by observing the loss of Golgi staining by immunofluorescence microscopy after the addition of cycloheximide. Although imprecise, we estimated a half-time of about 2 h. We measured the overall half-life of the total cellular pool of CD45 directly by pulse-chase studies and determined a halflife of 8-9 h. These values were consistent with 25-30% of the total CD45 being present in the Golgi during steady state. This number fit well with the fraction of CD45 which redistributed to the 200 suc-pellet after activation (see below). Within minutes of activation, the Golgi staining pattern seen by immunofluorescence disappeared. The immunoelectron microscopy confirmed that this represented a "clearing out" of the Golgi of CD45. One limitation of our ability to quantitate this is the nonquantitative nature of horseradish peroxidase-based immunoelectron microscopy. In contrast, the Golgi staining with anti-TCR antibodies showed no diminution upon activation. There was no inhibition of CD45 biosynthesis to explain this. We examined the rate of sialylation of the CD45 in both control and activated cells and found no differences. Thus it appears that a signal generated via the TCR releases CD45 that is retained in the Golgi.
All of our data are consistent with the conclusion that the population of CD45 that redistributes to the 200 suc-pellet after activation is the population that resides in the Golgi in the unactivated cell. This conclusion is strengthened by the observation that the gradual loss of Golgi staining after cycloheximide correlated with the loss of redistributed CD45. The failure to label CD45 in the 200 suc-pellet by surface iodination of either pre-or postactivated cells points against the population of CD45 in the plasma membrane being involved in the redistribution. What remains unclear is how the clearing of the Golgi relates to the subcellular fractionation results. Is the fraction of CD45 which redistributes to the 200 SUCpellet derived from the Golgi? What is the location of the "redistributed CD45 in the intact cell? Ultrastructural localization of intracellular CD45 after activation shows intense staining of some vesicular structures around the Golgi and in the trans-Golgi region, however we do not know whether this accounts quantitatively for the loss of Golgi CD45. Thus, it remains to be determined just where this Golgi CD45 goes. We suspect that the ability to precipitate via centrifugation this fraction of CD45 represents the association of the membranes containing the redistributed CD45 with cytoskeletal structures.
The determinants of the residence time of membrane proteins within the organelles of the secretory pathway are the object of intense interest in cell biology. Much of this interest has been focused on the mechanisms for the retention of proteins within specific organelles. A variety of mechanisms can be responsible for retention including binding to retention proteins, transport incompetence, and sorting (Klausner, 1989). Progress is beginning to be made in defining possible signals within proteins responsible for their retention in the ER (Nilsson et al., 1989). Residence questions have been raised largely to address the problems of establishing and maintaining the identity of distinct organelles within the secretory pathway. However, the role of retention in the ER in preventing the expression of incomplete oligomeric complexes on the cell surface represents a use of organelle residence in regulating cellular functions. The rapid loss of Golgi residence of CD45 in response to receptor-mediated signaling now suggests that the residence of a molecule in organelles along the secretory pathway can be regulated acutely by surface-generated signals.
Speculations on the Functional Implication of the Redistribution of CD45"After many years of intense interest in the role of protein tyrosine kinases it is now becoming clear that PTPases play important regulatory roles in cell signaling (Hunter, 1989;Tonks and Charbonneau, 1989). Interest in CD45 as one of the major membrane PTPases of T cells has been driven by at least two types of observations. First CD45 "looks" like a receptor and thus raises the possibility of ligandregulated membrane PTPase activity. Second are a variety of observations on the potential functional roles of CD45 in T cell activation. These include the effects of anti-CD45 antibodies on T cell activation (Ledbetter et al., 1988) and the finding that T cell variants that lack surface CD45 are incapable of TCR-mediated signal transduction (Pingel and Thomas, 1989). Furthermore, Koretzky et al. (1990) demonstrated recently the essential role of CD45 for coupling T cell antigen receptor to the phosphatidylinositol pathway. We do not know what the basal level of PTPase activity of CD45 is within T cells. However, the extremely high activity of this (and other) phosphatases in uitro suggests strongly that its activity must be regulated to allow tyrosine phosphorylation. Two ways to regulate this PTPase are by controlling its enzymatic activity and by controlling its spatial access to substrates. Structural regulation of surface CD45 in T cells has been suggested by observing changes in the mobility of CD45 in the membrane upon treatment of cells with phorbol esters and the finding that CD45 interacts with fodrin in a manner that changes with activation (Bourguignon et al., 1985). The receptor-mediated redistribution of the internal pool of CD45 observed in 2B4 T cell hybridoma can be thought of in two ways. First, the cell may move this population of CD45 into a compartment in which it now functions in important dephosphorylations.
Alternatively, CD45 may function tonically in the Golgi, and the removal from this organelle may function then to allow tyrosine phosphorylations of structures near or associated with the Golgi. These two possibilities are not mutually exclusive. The latter possibility is intriguing because it might explain the unusually long residence time of newly synthesized CD45 in the Golgi. This residence, coupled with the half-life of the protein in these cells, resulted in about 30% of total CD45 (and about 20% of total membrane angiotensin PTPase activity) being present in the Golgi. Is the Golgi involved in cellular signaling? The relatively recent report of the intracellular localization of one form of CAMP-dependent kinase to the Golgi region raised this possibility (Nigg et al., 1985). At this point we can only speculate about such signaling functions for this organelle. However, the specific association of the Golgi with the centrioles might give this organelle a role in the regulation of the centrioles via cytoplasmically oriented kinases and phosphatases. Perhaps the examination of tyrosine phosphorylations in or around the Golgi and correlation of changes with CD45 redistribution would point to the possible functional effects of the phenomenon reported in this study.