(Received for publication, September 11, 1996, and in revised form, February 28, 1997)
From the McGill Cancer Centre and the Departments of
§ Biochemistry,
Medicine, and ** Oncology, McGill
University, Montreal, Quebec H3G 1Y6, Canada and the
Departments of Medicine and Oncology,
Montreal General Hospital, Montréal,
Quebec H3G 1A4, Canada
To further understand the functional interactions between CD45 and p56lck in T-cells, we stably reconstituted their expression in a nonlymphoid system. The results of our analyses demonstrated that CD45 could dephosphorylate tyrosine 505 of p56lck in NIH 3T3 fibroblasts. As is the case for T-cells, removal of the unique domain of p56lck interfered with dephosphorylation of tyrosine 505 in fibroblasts, further stressing the importance of this region in the interactions between CD45 and p56lck. The ability of CD45 to dephosphorylate tyrosine 505 in NIH 3T3 cells was also greatly influenced by the catalytic activity of p56lck. Indeed, whereas CD45 provoked dephosphorylation of kinase-defective Lck molecules in this system, it failed to stably dephosphorylate kinase-active p56lck polypeptides. Finally, our studies showed that CD45 was also able to inhibit the oncogenic potential of a constitutively activated version of p56lck in NIH 3T3 cells. This effect did not require the Lck unique domain and apparently resulted from selective dephosphorylation of substrates of activated p56lck in fibroblasts. In addition to providing insights into the nature and regulation of the interactions between CD45 and p56lck in T-cells, these results indicated that CD45 clearly has the capacity to both positively and negatively regulate p56lck-mediated functions in vivo.
p56lck is a lymphocyte-specific member of the Src family of tyrosine-protein kinases (Refs. 1-3; reviewed in Refs. 4 and 5). Like the other members of the Src family, it bears several independent structural domains that include, from the amino terminus to the carboxyl terminus: (i) sites of myristylation (glycine 2) and palmitylation (cysteines 3 and/or 5), involved in targetting to cellular membranes (Ref. 6; reviewed in Ref. 7); (ii) a unique domain of roughly 50 amino acids, which mediates binding to the CD4 and CD8 T-cell surface antigens (8-12); (iii) Src homology (SH)1 3 and SH2 domains, capable of interacting with proline-rich motifs and phosphotyrosine-containing sequences, respectively (reviewed in Ref. 13); (iv) a kinase domain, including sites for ATP-binding, phosphotransfer (lysine 273) and autophosphorylation (tyrosine 394) and (v) a negative regulatory domain, encompassing the major site of in vivo tyrosine phosphorylation, tyrosine 505.
The catalytic activity of p56lck is primarily regulated by phosphorylation of tyrosines 394 and 505 (reviewed in Ref. 5). Phosphorylation at tyrosine 394 activates the catalytic function of p56lck, by provoking a conformational change in the kinase domain (14). Conversely, phosphorylation at tyrosine 505 inhibits the enzymatic activity of Lck (15-17), presumably by allowing an intramolecular interaction between the carboxyl-terminal end and the SH2 domain of the enzyme. Accumulating data indicate that tyrosine 505 phosphorylation is mediated by p50csk, a tyrosine-protein kinase expressed in all cell types (17-19; reviewed in Ref. 20). However, tyrosine 505 can also be a site of autophosphorylation. This possibility is suggested by the findings that p56lck could undergo phosphorylation at tyrosine 505 in bacteria (which lack endogenous tyrosine-protein kinases) or during in vitro kinase reactions (21, 22). Nonetheless, autophosphorylation may not be a prominent component of tyrosine 505 phosphorylation in mammalian cells, because a kinase-defective version of p56lck (lysine 273 to arginine 273 (Arg273) Lck) was still extensively phosphorylated at this site in transfected NIH 3T3 fibroblasts (19).
Contrary to Lck polypeptides expressed in NIH 3T3 cells, those isolated from T-lymphocytes are poorly phosphorylated at tyrosine 505 (23). This difference is seemingly consequent to the action of CD45, a transmembrane protein-tyrosine phosphatase selectively expressed in nucleated hemopoietic cells (for a review, see Ref. 24). This notion is supported by the observation that Lck polypeptides immunoprecipitated from CD45-deficient T-cells exhibited a marked (8-10-fold) increase in tyrosine 505 occupancy, when compared with p56lck molecules recovered from their CD45-positive counterparts (25-29). In contrast, expression of CD45 had little or no effect on the tyrosine phosphorylation of two other Src family kinases, p59fynT and p60c-src (26). The lack of constitutive dephosphorylation of tyrosine 505 in CD45-negative T-cells is thought to explain the inability of these cells to become activated upon stimulation with antigen or anti-T-cell receptor antibodies (30-33).
Over the past few years, significant efforts have been directed toward understanding the mechanism(s) by which CD45 causes selective dephosphorylation of p56lck in T-cells. Because a small fraction of Lck molecules can be co-immunoprecipitated with CD45 in mild detergent lysates of T-cells (34, 35), it is likely that these two molecules are in close proximity in the cell. Interestingly, we demonstrated that deletion of the unique domain of p56lck or its replacement by the equivalent sequence from another Src family kinase (p59fynT) significantly reduced dephosphorylation at tyrosine 505 in a CD45-positive T-cell line (23). This finding implied that the unique domain contains elements that favor the functional interactions between CD45 and p56lck. Because the unique domain of p56lck can associate with the cytoplasmic portion of CD45 in vitro (36), it is conceivable that Lck directly associates with CD45 in vivo via these sequences. However, it is also plausible that other molecules participate in this interaction. One proposed candidate is the CD45-associated protein, a hemopoietic cell-specific polypeptide having the capacity to bind both CD45 and p56lck (37-40).
NIH 3T3 mouse fibroblasts and -2 packaging cells
were grown in
-minimal essential medium (41) containing 10% fetal
calf serum and antibiotics. Derivatives expressing the neomycin
phosphotransferase were grown in the additional presence of G418 (0.5 mg/ml), whereas cells expressing the puromycin resistance gene were
propagated in medium supplemented with puromycin (1 µg/ml).
Anti-mouse CD45 rat monoclonal antibody (mAb) M1.89.18.7 was obtained from the American Type Culture Collection. Purified antibodies were used in our experiments. Rabbit anti-CD45 polyclonal antibodies were generated against the carboxyl-terminal tail of CD45 and were kindly provided by Dr. Phil Branton (McGill University, Montreal, Quebec, Canada). Rabbit antisera directed against amino acids 2-148 of p56lck (42) or against the SH3 domain of p56lck (23) were reported previously. Rabbit antisera directed against annexin II and GTPase-activating protein (GAP) of p21ras were generously provided by Drs. Tony Hunter (The Salk Institute, La Jolla, CA) and Tony Pawson (Mount Sinai Hospital Research Institute, Toronto, Ontario, Canada), respectively. Anti-cortactin mAbs were kindly provided by Dr. Tom Parsons (University of Virginia, Charlottesville, VA). Affinity-purified rabbit anti-phosphotyrosine antibodies were generated in our laboratory,2 whereas anti-phosphotyrosine mAb 4G10 was purchased from Upstate Biotechnology Inc. (Lake Placid, NY).
Expression of CD45 in NIH 3T3 CellsA cDNA encoding the
R0 (T200) isoform of mouse CD45 (cloned in the expression vector
pHApr-1-neo) was kindly provided by Dr. Pauline Johnson (University
of British Columbia, Vancouver, British Columbia, Canada). This
construct was stably introduced in NIH 3T3 cells by calcium phosphate
precipitation (43). Monoclonal G418-resistant cell lines were
established by limiting dilution and screened by fluorescence-activated
cell sorter (FACS) analysis using mAb M1.89.18.7 and fluorescein
isothiocyanate-conjugated goat anti-rat IgG.
cDNAs
encoding wild-type p56lck, lysine 273 to arginine 273 (Arg273) p56lck, an Arg273 Lck variant
lacking amino acids 16-62 of the unique domain (16-62R273 Lck), or
tyrosine 505 to phenylalanine 505 (Phe505) Lck were
described elsewhere (15, 23). An lck cDNA containing the
Phe505 mutation, in addition to a deletion of amino acids
16-62, was created by standard recombinant DNA technology. These
cDNAs were inserted in the multiple cloning site of the retroviral
expression vector pBabePuro (44), which also contains the puromycin
resistance gene (puro). The retroviral constructs were
stably transfected in
-2 packaging cells by calcium phosphate
precipitation (43), and retroviral supernatants were used to infect NIH
3T3 cells expressing either CD45 or the neomycin phosphotransferase
alone. Cells expressing the various Lck polypeptides were selected by growth in puromycin-containing medium in the continuous presence of
G418.
Immunoprecipitations and immunoblots were conducted as described elsewhere (8), with the exception of CD45 immunoprecipitations, which were performed using mAb M1.89.18.7 (10 µg) coupled to protein G-Sepharose (Pharmacia Biotech Inc.). Following immunoblotting, immunoreactive products were detected by autoradiography and quantitated using a PhosphorImager (BAS 2000, Fuji). Peptide mapping with cyanogen bromide was also performed as reported previously (19).
Pervanadate TreatmentTo inhibit protein-tyrosine phosphatase activity, cells were treated for 10 min at 37 °C with the protein-tyrosine phosphatase inhibitor pervanadate (1:50, v/v) as described elsewhere (19). Lck polypeptides were recovered by immunoprecipitation and subjected to further analyses as outlined above.
Cell FractionationCell fractionation studies were conducted according to a previously described protocol (23).
Cell Transformation AssaysTo examine focus formation, NIH 3T3 derivatives (103 or 104 cells) were seeded in 6-well Costar plates with 105 neomycin-resistant NIH 3T3 cells. Foci were counted after 9 days of growth. For growth in soft agar, 2 × 104 cells were plated in medium containing 0.3% agar as described previously (15). Colony formation was monitored for 2 weeks. All assays were done in duplicate and were repeated at least three times.
To further dissect the regulation of p56lck
by CD45 in T-cells, we wished to recreate the expression of these two
molecules in a nonlymphoid mammalian cell system. To this end, mouse
NIH 3T3 fibroblasts were chosen, because they do not normally express either polypeptide. As a first step, NIH 3T3 variants expressing the
isoform of CD45 predominantly contained in mature T-cells, CD45 R0/T200
(24), were generated as outlined under "Materials and Methods." Two
CD45-positive NIH 3T3 clones (clones 2 and 31) were selected for
further studies. As depicted in Fig. 1 (A and B), FACS analyses showed that these two cell lines expressed
easily appreciable amounts of CD45 at their surface. To ensure that
full-length CD45 polypeptides were expressed in these cells, CD45 was
immunoprecipitated with mAb M1.89.18.7 and subsequently immunoblotted
with a polyclonal rabbit anti-CD45 serum (Fig. 1C). This
experiment revealed that the two clones contained a ~180-kDa
immunoreactive polypeptide (lanes 2 and 3) that
comigrated with the CD45 molecule immunoprecipitated from the mouse
T-cell line BI-141 (lane 4). The CD45-positive fibroblasts
expressed approximately 10 times lower amounts of CD45 than BI-141
T-cells.
CD45 Can Induce Dephosphorylation of p56lck in NIH 3T3 Cells
Next, CD45-positive and CD45-negative NIH 3T3 cells were infected with retroviruses encoding either wild-type or kinase-defective (Arg273) p56lck molecules. Kinase-defective Lck molecules were included in these studies to eliminate the possibility of compensatory autophosphorylation at tyrosine 505, as discussed elsewhere (19, 23). Although the results reported herein primarily concerned CD45-positive clone 31, similar results were obtained with clone 2 (data not shown).
The impact of CD45 on the tyrosine phosphorylation of p56lck in
NIH 3T3 cells was first examined (Fig. 2). Lck
polypeptides were recovered by immunoprecipitation, and their
phosphotyrosine content was assessed by immunoblotting with
anti-phosphotyrosine mAb 4G10 (Fig. 2A, top
panel). Moreover, the abundance of p56lck was verified by
immunoblotting of parallel anti-Lck immunoprecipitates with an
antiserum generated against the Lck SH3 region (middle panel). This was taken into consideration when determining the relative extents of Lck tyrosine phosphorylation in these various cell
lines. Finally, the levels of CD45 were monitored by immunoblotting of
anti-CD45 immunoprecipitates with rabbit anti-CD45 antibodies (bottom panel).
In keeping with our previous report (19), wild-type p56lck (Fig. 2A, lane 2) and Arg273 p56lck (lane 3) were tyrosine phosphorylated to similar extents in NIH 3T3 cells lacking CD45. However, although tyrosine phosphorylation of wild-type p56lck was not decreased by expression of CD45 (lane 4), the phosphotyrosine content of kinase-inactive Lck molecules was reduced ~5-fold (lane 5). The differential impact of CD45 on wild-type and Arg273 p56lck was not caused by variations in the expression levels of either Lck or CD45, because the two cell lines contained comparable quantities of these two products (middle and bottom panels).
To identify the sites of tyrosine phosphorylation on these Lck molecules, peptide mapping studies were conducted using cyanogen bromide. Cells were metabolically labeled with 32Pi, and Lck polypeptides were recovered by immunoprecipitation. Following cleavage with cyanogen bromide, phosphorylated peptides were resolved by gel electrophoresis and detected by autoradiography (Fig. 2B). Great care was taken to ensure that cleavage of p56lck by cyanogen bromide was complete and that equivalent amounts of Lck peptides were applied in each lane.
Cleavage of p56lck with cyanogen bromide results in three possible phosphopeptides: C1 (28 kDa), which contains potential amino-terminal sites of serine, threonine, and tyrosine phosphorylation; C2 (10 kDa), which bears tyrosine 394, the major site of in vitro autophosphorylation; and C3 (4 kDa), which carries tyrosine 505, the major site of in vivo tyrosine phosphorylation (45-47). In CD45-negative fibroblasts, both wild-type (lane 1) and kinase-defective (lane 2) p56lck were extensively phosphorylated within the tyrosine 505-containing C3 fragment, in agreement with earlier reports (19). As expected from the anti-phosphotyrosine immunoblot (Fig. 2A), expression of CD45 did not noticeably influence the extent of tyrosine 505 phosphorylation of wild-type Lck (Fig. 2B, lane 3). In contrast though, it caused a marked reduction of carboxyl-terminal phosphorylation of Arg273 Lck (lane 4). Although phosphorylation within the amino-terminal C1 fragment is not observed in Fig. 2B, it could be seen in longer autoradiographic exposures of this gel. Phosphorylation of the C1 peptide was not affected by CD45 expression (data not shown).
To verify that CD45 reduced the phosphorylation of kinase-inactive Lck
through its phosphatase activity and not through other mechanisms such
as steric hindrance, we tested the influence of pervanadate, a potent
protein-tyrosine phosphatase inhibitor (Fig. 3). Cells
were treated with pervanadate for 10 min and then processed for
anti-Lck immunoprecipitations as described above. Anti-phosphotyrosine immunoblotting of these immunoprecipitates demonstrated that the protein-tyrosine phosphatase inhibitor caused a marked (~8-fold) increase in the phosphotyrosine content of Arg273 Lck in
CD45-positive NIH 3T3 cells (Fig. 3A, compare lanes
2 and 4). Cyanogen bromide cleavage analysis (Fig.
3B) established that pervanadate restored tyrosine 505 phosphate occupancy of Arg273 Lck (lane 4). In
addition, the phosphatase inhibitor provoked phosphorylation of both
wild-type (lane 3) and kinase-defective (lane 4)
p56lck within the tyrosine 394-containing C2 fragment, in
keeping with earlier studies (19, 48, 49). Although the basis for
phosphorylation of kinase-inactive Lck molecules at the
"autophosphorylation" site has not yet been elucidated, this
finding suggests that another cellular tyrosine-protein kinase can
transphosphorylate tyrosine 394 in vivo (19, 48,
49).
The Unique Domain of p56lck Is Necessary for CD45-mediated Dephosphorylation of Tyrosine 505 in NIH 3T3 Cells
To verify that
dephosphorylation of tyrosine 505 in NIH 3T3 cells occurred via a
mechanism similar to that existing in T-cells, we examined whether the
unique domain of p56lck was also necessary for the effect of
CD45 in fibroblasts. Arg273 Lck polypeptides lacking
residues 16-62 (16-62R273 Lck) were introduced in CD45-negative
and CD45-positive NIH 3T3 cells. Anti-phosphotyrosine immunoblotting of
anti-Lck immunoprecipitates (Fig. 4A,
top panel) established that
16-62R273 Lck (lane
3) was tyrosine phosphorylated to nearly the same extent as
Arg273 p56lck (lane 1) in CD45-negative
cells. In cells expressing CD45, however, the tyrosine phosphorylation
of
16-62R273 Lck (lane 4) was only reduced ~1.5-fold,
in comparison with the ~5-fold decrease noted above for
Arg273 Lck (lane 2; Fig. 2A,
lanes 3 and 5). Complementary peptide mapping analyses confirmed that, unlike Arg273 Lck (Fig.
4B, lane 2),
16-62R273 Lck remained
prominently phosphorylated at tyrosine 505 in CD45-positive cells
(lane 4).
Because CD45 is a membrane-bound phosphatase, we wished to ensure that
the association of p56lck with membranes in NIH 3T3 cells was
not altered by deletion of the unique region. After incubation in
hypotonic buffer, cells were mechanically broken in a Dounce
homogenizer. Lysates were then fractionated by differential
centrifugation, and Lck polypeptides were recovered from the
particulate (P100) and cytosolic (S100) fractions by
immunoprecipitation. The abundance of Lck in each fraction was
determined by anti-Lck immunoblotting. These studies demonstrated that
as is the case for Arg273 p56lck (Fig
4C, lanes 1 and 3), more than 80% of
16-62R273 Lck was localized to the P100 fraction of either
CD45-negative (lane 5) or CD45-positive (lane 7)
NIH 3T3 cells.
Introduction of a constitutively
activated version of p56lck (tyrosine 505 to phenylalanine 505 p56lck) in NIH 3T3 cells causes marked tyrosine phosphorylation
of several intracellular proteins (15, 16). This biochemical
modification leads to oncogenic cellular transformation, with its
characteristic morphological alterations and the ability to form foci
in monolayers and grow in semi-solid medium. To test whether CD45 could
also regulate activated p56lck molecules, CD45-positive NIH 3T3
cells were infected with retroviruses encoding Phe505
p56lck. CD45-negative neomycin-resistant NIH 3T3 cells (Neo)
were used as control recipient. Immunoblot analyses showed that all
infected cells expressed comparable amounts of Phe505
p56lck (data not shown; see Fig. 7A, bottom
panel).
Whereas fibroblasts expressing either the neomycin phosphotranferase
alone (Fig. 5A) or CD45 alone (Fig.
5B) were flat and possessed short plasma membrane
extensions, those containing Phe505 p56lck without
CD45 (Fig. 5C) were rounded and refractile and displayed multiple neuronal-like processes (15, 16). By contrast, cells expressing Phe505 p56lck and CD45 (Fig.
5D) exhibited an intermediate morphology, being less
rounded, less refractile, and showing fewer processes than cells
containing activated Lck alone (Fig. 5C). The ability of these cells to form foci in monolayer cultures and colonies in soft
agar was also assessed (Table I). Like cells containing Phe505 Lck alone, cells expressing Phe505 Lck
and CD45 could form foci in monolayers. However, the foci were ~five
times smaller than those generated by Phe505 Lck-expressing
cells. Furthermore, although fibroblasts expressing Phe505
p56lck formed large colonies in soft agar, cells expressing
Phe505 Lck and CD45 failed to grow under this
condition.
|
To elucidate the mechanism by which CD45 prevented oncogenic
transformation by Phe505 p56lck, we studied its
impact on the accumulation of phosphotyrosine-containing proteins (Fig.
6). Anti-phosphotyrosine immunoblotting of total cell
lysates demonstrated that cells expressing Phe505 Lck and
CD45 (Fig. 6A, lane 4) contained lower amounts of
phosphotyrosine-containing proteins than cells expressing
Phe505 p56lck alone (lane 3). This
diminution especially affected polypeptides of 140, 120, and 36 kDa.
Individual substrates were also recovered by immunoprecipitation with
specific antibodies, and their phosphotyrosine content was determined
by anti-phosphotyrosine immunoblotting (Figs. 6, B-D). In
keeping with the analysis of total cell lysates (Fig. 6A),
tyrosine phosphorylation of the 36-kDa annexin II was absent in cells
expressing Phe505 Lck and CD45 (Fig. 6B,
lane 4) in contrast to cells containing Phe505
Lck alone (lane 3). However, CD45 had little or no impact on the extent of tyrosine phosphorylation of cortactin (Fig.
6C), as well as on that of GAP and its associated p190 and
p62 (Fig. 6D). We were not able to determine the identity of
the 140- and 120-kDa substrates that appeared to be tightly regulated
by CD45 in Phe505 p56lck-expressing cells (Fig.
6A, lane 4).
We wanted to determine whether the effect of CD45 on Phe505 Lck-expressing cells was due to dephosphorylation of Phe505 p56lck or to dephosphorylation of downstream targets. To this end, the phosphotyrosine content of Phe505 Lck was first examined by anti-phosphotyrosine immunoblotting of anti-Lck immunoprecipitates (Fig. 7A). This study showed that expression of CD45 reduced the extent of tyrosine phosphorylation of Phe505 p56lck by ~2-fold. To ascertain whether these changes were due to dephosphorylation of tyrosine 394, the positive regulatory site of p56lck, peptide mapping studies were conducted (Fig. 7B). These analyses failed to show any reduction in phosphorylation of the C2 fragment of p56lck, which contains tyrosine 394. Although a small decrease in C1 phosphorylation could be seen in CD45-positive cells in this experiment (Fig. 7B, lane 2), a similar change was not observed in other experiments (data not shown).
These findings raised the possibility that CD45 did not act by
dephosphorylating Phe505 p56lck but rather by
dephosphorylating its substrates. To help support this idea, the impact
of CD45 on transformation by a variant of Phe505 Lck
lacking the unique domain was tested (Fig. 8 and Table
II). Because the unique domain plays an important role
in the CD45-mediated dephosphorylation of p56lck (this report
and Ref. 23), we reasoned that removal of this domain should have no
impact on the effect of CD45 if it were due to dephosphorylation of
downstream substrates. Thus, CD45-positive and CD45-negative NIH 3T3
cells were infected with retroviruses encoding 16-62F505 Lck. Cells
containing
16-62F505 Lck without CD45 were morphologically
transformed in a manner analogous to Phe505 Lck-expressing
cells (Table II and data not shown). Moreover, these cells were capable
of growing in soft agar, albeit with a slightly lower efficiency than
Phe505 Lck-expressing cells (Table II).
Anti-phosphotyrosine immunoblotting of total cell lysates showed that
16-62F505 Lck-expressing cells also possessed elevated levels of
phosphotyrosine-containing proteins (Fig. 8, lane 5).
|
As was the case for cells containing Phe505 Lck and CD45,
cells expressing 16-62F505 Lck and CD45 exhibited a morphology that was intermediate between those of Neo cells and
16-62F505
Lck-expressing cells (Table II and data not shown). Furthermore, they
were unable to grow in soft agar (Table II). CD45 also reduced the
accumulation of phosphotyrosine-containing proteins in
16-62F505
Lck-expressing cells (Fig. 8, lane 6) in manner similar to
that observed in cells expressing Phe505 p56lck
(lane 4). Hence, these data indicated that CD45 inhibited
Phe505 Lck-mediated transformation by a process independent
of the Lck domain.
Herein, we report the first successful attempt at reconstituting the CD45-mediated regulation of p56lck in a nonlymphoid system. The results of our experiments showed that expression of the R0 isoform of CD45 in NIH 3T3 fibroblasts caused a ~5-fold decrease in the extent of tyrosine 505 phosphorylation of kinase-defective (Arg273) p56lck molecules. This effect was most likely due to the phosphatase activity of CD45, because treatment with pervanadate, a protein-tyrosine phosphatase inhibitor, restored phosphorylation of the carboxyl-terminal tyrosine of Lck in CD45-positive NIH 3T3 cells. As demonstrated for BI-141 T-cells (23), the unique region of p56lck was also found to play an important role in the CD45-induced dephosphorylation of tyrosine 505 in NIH 3T3 cells. Thus, dephosphorylation of the inhibitory tyrosine of p56lck by CD45 can occur in a nonlymphoid cellular system by a mechanism similar to that taking place in T-cells.
These findings implied that no other lymphoid-specific components are absolutely necessary for the CD45-mediated dephosphorylation of p56lck. Specifically, they also demonstrated that CD45-associated protein, a protein that can associate with both CD45 and p56lck (37-41) and is not expressed in NIH 3T3 cells,2 is not critical for the action of CD45 on p56lck. This idea was further supported by our finding that enforced expression of CD45-associated protein in NIH 3T3 cells did not modify the ability of CD45 to dephosphorylate wild-type, Arg273, or Phe505 Lck molecules.2 Nevertheless, it is likely that additional features or processes enhance the ability of CD45 to dephosphorylate tyrosine 505 in T-cells. Indeed, although CD45 provoked dephosphorylation of kinase-inactive Lck molecules in fibroblasts, it failed to stably dephosphorylate wild-type Lck polypeptides. In contrast, the two forms of Lck were equally well dephosphorylated in T-cells (23). The higher amounts of CD45 typically present in T-cells may explain this difference. Possibly, greater amounts of CD45 are needed for stable dephosphorylation of kinase-active Lck polypeptides in vivo. Alternatively, the presence of other yet unidentified molecules interacting with CD45, Lck or both, may contribute to more efficient dephosphorylation of Lck in T-cells.
These findings provided clear evidence that the catalytic activity of p56lck can influence the ability of CD45 to stably dephosphorylate tyrosine 505. It is conceivable that the enzymatic function of p56lck antagonizes the effect of CD45 by allowing autophosphorylation at tyrosine 505 (21, 22). Alternatively, the enzymatic activity of Lck may facilitate the recruitment of p50csk, the tyrosine-protein kinase normally responsible for phosphorylating tyrosine 505 (Refs. 17-19). Lastly, it is possible that active Lck molecules can modify the function of CD45 and reduce its ability to dephosphorylate tyrosine 505. Regardless of the mechanism underlying this phenomenon, these observations raised the interesting possibility that activation of p56lck may diminish the capacity of CD45 to dephosphorylate tyrosine 505 in T-cells, thereby providing a potential negative feedback mechanism. This concept may explain the earlier finding that CD4-mediated activation of p56lck in T-cells led to a paradoxical augmentation of tyrosine 505 phosphorylation, in addition to the expected increase in tyrosine 394 phosphorylation (50, 51).
The impact of CD45 on an activated version of p56lck (Phe505 Lck) was also evaluated. Contrary to cells containing Phe505 p56lck alone, cells harboring Phe505 p56lck and CD45 exhibited a less transformed morphology, formed smaller foci in monolayer cultures, and were unable to grow in soft agar. These biological effects were accompanied by a noticeable reduction in the extent of tyrosine protein phosphorylation induced by Phe505 p56lck. Several findings indicated that the impact of CD45 in this context was primarily caused by dephosphorylation of downstream signaling targets rather than dephosphorylation of Phe505 p56lck. First, the phosphotyrosine content of Phe505 p56lck was only minimally reduced (less than ~2-fold) in CD45-positive NIH 3T3 cells. Second, CD45 did not cause a global reduction of tyrosine protein phosphorylation in Phe505 Lck-expressing cells. Instead, it seemed to regulate a specific set of tyrosine phosphorylation substrates. These included annexin II, as well as polypeptides of 140 and 120 kDa of yet undetermined identity. In contrast, CD45 had no or little effect on substrates such as cortactin and GAP, as well as the GAP-associated p190 and p62. Third, removal of the Lck unique domain had no consequence on the ability of CD45 to inhibit transformation by Phe505 Lck. Because the unique region was crucial for CD45-mediated dephosphorylation of Lck in other systems (this report and Ref. 23), we feel that this observation provided a most compelling indication that CD45 reduced transformation by Phe505 Lck by acting on downstream signaling events.
These results suggested that CD45 may also have the ability to negatively regulate Lck-mediated signals in T-cells. This notion is consistent with the observation that antibody-mediated co-aggregation of CD45 and the T-cell antigen receptor prevented T-cell receptor-induced intracellular tyrosine protein phosphorylation and T-cell activation (52). Similarly, it supports the finding that some CD45-negative T-cell lines (such as YAC-N1) contained elevated levels of phosphotyrosine-containing proteins prior to T-cell receptor stimulation (29, 53). Hence, CD45 may provide inhibitory signals in T-cells, in addition to its aforementioned positive regulatory role in unstimulated T-cells. Obviously, if these two opposite effects were to occur physiologically, it is expected that they would be differentially regulated at the various stages of T-cell activation.
In summary, we have shown that the ability of CD45 to dephosphorylate the inhibitory tyrosine of p56lck can be reconstituted in a nonlymphoid system, implying that this functional interaction does not require the participation of any other lymphoid-specific component. Unexpectedly, we also found that the catalytic activity of Lck can influence the dephosphorylation of tyrosine 505 by CD45. This observation suggested that physiological changes in the enzymatic activity of p56lck in T-cells may modify its regulation by CD45. Finally, evidence was adduced that CD45 has the capacity to inhibit the impact of Lck activation in vivo, most probably by dephosphorylating downstream signaling targets. In addition to clarifying the basis for the interactions between CD45 and p56lck in T-cells, our experiments permitted us to uncover previously unappreciated mechanisms involved in regulating these interactions.
We thank Pauline Johnson for provision of the expression construct for CD45, Phil Branton for anti-CD45 antibodies, Tony Hunter for the anti-annexin II serum, Tom Parsons for anti-cortactin antibodies, and Tony Pawson for the anti-GAP serum. We also thank the members of our laboratory for critical reading of the manuscript.