Autophosphorylation of Src and Yes Blocks Their Inactivation by Csk Phosphorylation

Home , Autophosphorylation

Oncogene (1998) 17, 1587 ± 1595  1998 Stockton Press All rights reserved 0950 ± 9232/98 $12.00 http://www.stockton-press.co.uk/onc Autophosphorylation of Src and Yes blocks their inactivation by Csk phosphorylation

Gongqin Sun, Ajay K Sharma and Raymond JA Budde

Department of Neuro-Oncology, Box 316, University of Texas, MD Anderson Cancer Center, Houston, Texas 77030, USA

Csk phosphorylates Src family protein tyrosine kinases proto-oncogenes, since their activation by mutation on a tyrosine residue near their C-terminus and down- leads to transformation of the host cells (Courtneidge, regulates their activity. We previously observed that this 1994). Elevation of the enzymatic activity of some regulation requires a stoichiometric ratio of Csk : Src in members has been associated with many types of a time-independent manner. In this report we examined human cancer (Levitzki, 1996). In tumor cells where this unusual kinetic behavior and found it to be caused by Src activity is elevated, the increase in its activity is Src autophosphorylation. First, pre-incubation of Src mostly the result of increased speci®c activity (Bolen et with ATP-Mg led to time-dependent autophosphoryla- al., 1987; Talamonti et al., 1993). Furthermore, no tion of Src, activation of its kinase activity and loss of its mutation has been identi®ed in humans that is ability to be inactivated by Csk. However, the autopho- responsible for the increased speci®c activity. These sphorylated Src can still be phosphorylated by Csk. The ®ndings suggest that elevation of Src activity in human SH2 binding site for phospho-Tyr of this hyperactive and tumor cells is due to a disruption of its regulation. For doubly phosphorylated form of Src is not accessible. this reason, understanding the regulation of Src family Second, dephosphorylation of autophosphorylated Src by PTKs may be one of the keys in understanding and protein tyrosine phosphatase 1B allowed Src to be designing eective treatment for cancers that have inactivated by Csk. Third, protein tyrosine phosphatase elevated activity of Src family PTKs. 1B preferentially dephosphorylates the Src autopho- A variety of stimuli, such as various growth factors, sphorylation site and allows for Src regulation by Csk. UV radiation and oxidative stress can directly or Finally, Yes, another member of the Src family, was also indirectly modulate Src activity. The ability to respond only partially inactivated when a sub-stoichiometric to these signals depends on a complex multi-domain amount of Csk was used. Mutation of the tyrosine structure. All PTKs in the Src family have an N- autophosphorylation site of Yes to a phenylalanine terminal acylation motif that is required for their resulted in a mutant Yes enzyme that can be fully membrane association, a unique region that is inactivated by a sub-stoichiometric amount of Csk in a heterologous in the family, two Src homology time-dependent manner. These results demonstrate that domains (SH3 and SH2) that mediate protein-protein Csk phosphorylation inactivates Src and Yes only when interactions, an SH1 catalytic domain which carries out they are not previously autophosphorylated and Src the phosphorylation reaction, and a regulatory Tyr (Yr) autophosphorylation can block the inactivation by Csk phosphorylation motif located at the C-terminal phosphorylation. This conclusion suggests a dynamic region. Another ubiquitous feature among Src family model for the regulation of the Src family protein PTKs is the presence of a Tyr autophosphorylation site tyrosine kinases, which is discussed in the context of (Ya) within the activation loop located in the catalytic previously reported observations on the regulation of Src domain (Smith et al., 1993). At the molecular level, family protein tyrosine kinases. several mechanisms that modulate Src activity have been discovered, including binding to other proteins Keywords: Src and Yes regulation; Csk; protein through the SH2 and SH3 domains, phosphorylation tyrosine phosphatase 1B; autophosphorylation of multiple Ser/Thr residues, Tyr phosphorylation by receptor PTKs, autophosphorylation and Csk phosphorylation (Brown and Cooper, 1996). The major regulatory mechanism is apparently phosphorylation of

Introduction Yr by Csk, which results in inactivation of Src (Cooper et al., 1986). In csk knockout mice, Src is mostly Protein tyrosine kinases (PTKs) of the Src family play unphosphorylated on the C-terminal regulatory Tyr important roles in cellular signal transduction (Brickell, and has elevated kinase activity (Imamoto and 1992). There are nine members: Src, Yes, Fyn, Yrk, Soriano, 1993; Nada et al., 1993). Src expressed in

Fgr, Hck, Lyn, Lck and Blk. Src, Yes and Fyn are yeast is minimally phosphorylated on Yr and highly expressed ubiquitously, while the other six are active, while co-expression of Csk led to phosphoryla- expressed in restricted cell types, particularly in tion of Yr and inactivation of Src (Superti-Furga et al., haematopoietic cells. These PTKs are key switches in 1993). These ®ndings establish Csk as the negative regulating many cellular signal transduction pathways regulator of Src family PTKs. (Thomas and Brugge, 1997). Most of these enzymes are In this communication, we examined the kinetics of Src inactivation by Csk and found that autophosphorylation of Src and Yes blocks their inactivation but not phosphorylation by Csk. This ®nding indicates Correspondence: RJA Budde Received 12 February 1998; revised 23 April 1998; accepted 23 April that autophosphorylation plays a greater role in 1998 regulating Src and Yes activity than previously Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1588 understood and suggests that an altered level of required a stoichiometric amount of Csk to achieve autophosphorylation may be the reason for the maximal inactivation of Src. Second, prior phosphor- paradoxically activated Src in malignant human cell ylation by Csk (pre-incubation with Csk and ATP-Mg) lines with high levels of active Csk. does not lead to signi®cantly higher levels of inactivation of Src than without prior phosphorylation

by Csk (incubation with no MgCl2). This apparent time Results independence is further demonstrated in Figure 1b. At the selected Csk to Src ratios of 1 : 5 and 1 : 2, In order to better understand the regulation of Src by approximately 30% and 60% of Src was inactivated Csk and identify factors that may aect this regulation, and this was independent of the pre-incubation time. we reconstituted this system in vitro with puri®ed Incubation of Src with ATP-MG in the absence of Csk recombinant enzymes. In the process, we observed that (Csk to Src ratio of 0 : 1) resulted in activation of Src the inactivation of Src by Csk does not follow a as a result of its autophosphorylation. These observa- normal enzymatically catalyzed process. Such observations were not a function of the particular enzyme tions are demonstrated by pre-incubating Src with ATP samples used or methods of their expression and and dierent amounts of Csk in the presence or isolation. Data consistent with these observations

absence of MgCl2 for 30 min before the Src activity have been reported repeatedly in the literature with was assayed with carboxymethylated-maleylated, re- dierent members of the Src family (Okada and duced lysozyme (RCM-L) (Figure 1a). In both Nakagawa, 1989; Ruzzene et al., 1994; Koegl et al., treatments, Src activity decreased with the increase in 1994) and Csk from dierent sources. Such consistency the amount of Csk added to the assay. While these suggested that these observations are a re¯ection of the results con®rmed that Csk was able to inactivate Src, intrinsic properties of Src regulation by Csk phosphor- two points in these results were unexpected. First, it ylation. These observations suggested that Src inactivation by Csk phosphorylation is not a simple ATP-Mg- dependent enzymatic conversion of active to inactive Src. There appears to be other factor(s) aecting this process. There are three possibilities that are consistent with these observations: (1) Csk inactivation of Src requires the formation of a stable Src-Csk complex after phosphorylation; (2) Csk is quickly inactivated by incubation with Src and/or ATP-Mg and thus unable to inactivate additional molecules of Src; and (3) Src is quickly transformed upon incubation with Csk and/or ATP-Mg into a form that can not be inactivated by Csk. The ®rst possibility would be consistent with the high stoichiometry of Csk required and the lack of time-dependent inactivation of Src. Furthermore, it was reported that Csk can form a complex with members of the Src family (Ruzzene et al., 1994; Bougeret et al., 1996). However, evidence from the literature and our own results eliminated this possibility. First, attempts by others (Sabe et al., 1992) and us (data not shown) have failed to detect strong and stable complexes between Csk and Src. Second, Src isolated after pre-incubation with Csk and ATP-Mg remain inactivated (Stover et al., 1994). Third, the ability of Csk deletion mutants (deletions in SH2 or SH3 domains) to inactivate Src correlates with the kinase activity of the mutants (data not shown). These results indicate that Src inactivation by Csk is a phosphorylation dependent process and does not require the formation of a stable Csk-Src complex. We also examined and eliminated the possibility of Csk being inactivated by incubation with Src and/or ATP- Mg. Csk undergoes autophosphorylation to a low Figure 1 Unusual kinetic behaviors of Src inactivation by Csk. extent (Amrein et al., 1995) and this does not (a) Dosage-dependent inactivation of Src by Csk. Src signi®cantly aect its activity (Oetken et al., 1994; 71 (0.13 mgml ) was pre-incubated for 30 min at 308C with ATP Sun et al., 1997). Furthermore, Csk is neither and dierent amount of Csk in the presence or absence of MgCl2 (10 mM). After pre-incubation, Src activity was assayed with phosphorylated nor inactivated by Src even when a RCM-lysozyme as the substrate. The concentration of MgCl2 was near stoichiometric amount of Src was used (Sun et al., adjusted after pre-incubation so that all reactions have the same 1997). The elimination of the ®rst two possibilities ®nal MgCl2 concentration in the Src activity assay. (b) Time leaves the third one as the only plausible explanation course of Src inactivation by Csk phosphorylation. Reaction and we further examined this possibility. PTKs of the conditions are as in (a) except the pre-incubations were carried out for dierent times. The Csk to Src molar ratios are given on Src family undergo autophosphorylation and are the graph activated by this autophosphorylation (Smith et al., Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1589 1993). Consistent with this possibility is the recent that the ability of Src to be inactivated by Csk is suggestion that autophosphorylation could activate Src inversely correlated with the level of Src autopho- that was inactivated by Csk phosphorylation (Xu et al., sphorylation. 1997; Sicheri et al., 1997). To examine the eect of Src autophosphorylation on inactivation by Csk, we examined the correlation between the time courses of Src autophosphorylation and the ability of Src to be inactivated by Csk (Figure 2). In this experiment, Src was incubated with 32 [g- P]ATP and MgCl2 for dierent times and then the proteins were separated by SDS ± PAGE. The phosphorylated proteins were visualized by autoradiography. Src autophosphorylation increased with incubation time and reached over 0.8 mol phosphate/ mol of Src within 1 h. Concomitant with the increase in Src autophosphorylation, the ability of Src to be inactivated by Csk decreased from 80% to about 20%. Longer incubation resulted in complete autophosphorylation and total loss of inactivation by Csk. Pre- incubation of Src with ATP or MgCl2 alone did not result in any autophosphorylation and had no eect on its inactivation by Csk. The correlation between Src autophosphorylation and lack of Csk inactivation strongly suggests that autophosphorylation leads to a conformation of Src that cannot be inactivated by Csk phosphorylation. Figure 2 Inverse relationship of Src autophosphorylation and its inactivation by Csk. Src (0.2 mg) was incubated with 0.1 mM ATP To further verify this hypothesis, we tested if 71 (40 000 d.p.m. pmol ) and 5 mM MgCl2 for dierent times and dephosphorylation of the autophosphorylation site the autophosphorylation of Src was determined by autoradio- could reverse the eect of autophosphorylation. graphy and scintillation counting of excised Src protein band. (a) Protein tyrosine phosphatase (PTP) 1B is a soluble Autoradiogram of Src autophosphorylation. (b) Stoichiometry of enzyme located in the cytoplasm and has a broad Src autophosphorylation and Src inactivation by Csk. Src substrate speci®city (Tonks et al., 1988a,b). We autophosphorylated as above but with non-radioactive ATP for dierent time was assayed in the presence or absence of Csk examined whether it could dephosphorylate autopho- (Csk : Src=3 : 1). Percentage of Src inactivation by Csk was sphorylated Src. Src that was 32P-labeled by autopho- calculated as described in the Material and methods section sphorylation was incubated with recombinant PTP 1B and EDTA (to inhibit further autophosphorylation by Src) and the level of 32P-label remaining within Src after the phosphatase treatment was detected by SDS ± PAGE, autoradiography and quanti®ed by liquid scintillation counting (Figure 3). Incubation with PTP 1B led to a time dependent loss of phosphate from Src, with approximately 80% of the label being removed within 20 min. We next investigated the eect of PTP 1B dephosphorylation on the ability of Src to be inactivated by Csk. As shown in Table 1, approximately 10% of autophosphorylated Src is inactivated by Csk (control). After treatment with PTP 1B to dephosphorylate the autophosphorylation site, over 50% the Src activity could be inactivated by Csk. When 1 mM ammonium molybdate (a potent PTP 1B inhibitor that does not aect the Csk or Src kinase activity, data not shown) is included in the PTP 1B treatment to prevent Src dephosphorylation, the majority of Src remained in a state that could not be inactivated by Csk. This result further demonstrated

Table 1 Eect of PTB 1B treatment on the Csk inactivation of Figure 3 Dephosphorylation of autophosphorylated Src by autophosphorylated Src. Autophosphorylated Src was treated with protein tyrosine phosphatase 1B. Autophosphorylated Src PTP 1B as indicated and the ability of Csk to inactivate such treated (0.2 mg) prepared as described in Figure 2 was incubated with Src was determined GST-PTP 1B (50 ng) in the presence of 10 mM Na2 EDTA for Src activity inactivated by Csk (%) dierent times and the proteins in the reactions were separated by Experiment Control +PTP 1B +PTP 1B+molybdate SDS ± PAGE. The phosphorylation level of Src was visualized by autoradiography and quantitated by scintillation counting of the 1 11 53 23 excised Src protein band. The phosphorylation level was 2 8 58 14 normalized to the initial level Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1590 The above conclusion predicts that if the autopho- PTP 1B enables Src to be inactivated by a sub- sphorylation of Src is kept at a minimal level, it would stoichiometric amount of Csk. be inactivated by a sub-stoichiometric amount of Csk. At this point, it was not clear if autophosphoryla- As demonstrated in Figure 3, PTP 1B readily depho- tion blocks phosphorylation by Csk or allows sphorylates the autophosphorylation site of Src, but it phosphorylation by Csk but blocks the inactivation was reported not to dephosphorylate the phosphory- of Src. To address this issue, we performed peptide

lated Yr (Somani et al., 1997). We took advantage of mapping of Src after various treatments (Figure 5). In this apparent speci®city of PTP 1B to examine how the treatment a, incubation with Csk and ATP-Mg led to presence of PTP 1B would change Csk inactivation of a loss of 80% of Src activity and an incorporation of Src when a sub-stoichiometric amount of Csk was used approximately 0.8 mol of phosphate per mol of Src. (Figure 4). For this study, a Csk : Src ratio of 1 : 10 was Peptide mapping indicates that phosphate label is

used. Src was pre-incubated in buer alone, or in the speci®cally on Yr in the 4 kD peptide fragment. presence of ATP-Mg, ATP-Mg+Csk, ATP- Incubation of this Csk-inactivated Src in treatment a Mg+Csk+PTP 1B or ATP-Mg+PTP 1B for 1 h with PTP 1B (treatment b) led to a partial recovery of before Src activity was assayed with RCM-L as the the initial Src activity and a loss of about 50% of the substrate. Prior to determining the activity of Src, phosphate, indicating that PTP 1B could depho- ammonium molybdate (1 mM) was added to inhibit the sphorylate the Csk phosphorylation site. Incubation activity of PTP 1B. Incubation with buer alone did of Src with ATP-Mg (treatment c) led to an activation not change the Src activity. Incubation with ATP-Mg of Src activity of about 60% and an incorporation of led to an increase in Src activity of approximately 60% as the result of autophosphorylation. When both ATP- Mg and Csk were included in the pre-incubation, Src activity did not change signi®cantly compared to the control. This is consistent with minimal inactivation of Src activity observed when a sub-stoichiometric amount of Csk is used as described earlier (Figure 1). Pre-incubation with ATP-Mg, Csk and PTP 1B led to a decrease in Src activity of 80% of the initial. The presence of ATP-Mg+PTP 1B in the pre-incubation did not change the Src activity signi®cantly, indicating an equilibrium between autophosphorylation with activation and dephosphorylation by PTP 1B (com- pare lane 2 vs 5). These results suggests that PTP 1B reverses Src autophosphorylation and that PTP 1B does not directly aect Src activity other than reversing its autophosphorylation. PTP 1B does not aect Csk activity (data not shown). Together these results indicate that by reversing Src autophosphorylation,

Figure 5 Eect of various phosphorylation and dephosphorylation treatments on the phosphorylation status and activity of Src. Src (0.2 mg) was incubated with 0.6 mg Csk (treatment a) or alone 32 (treatment c) in phosphorylation buer (0.1 mM [g- P]ATP 71 (40 000 d.p.m. pmol ), 12 mM MgCl2,75mM EPPS-NaOH (pH 8.0), 5% glycerol, 0.005% Triton X-100, 0.05% 2- Figure 4 Inactivation of Src by a sub-stoichiometric amount of mercaptoethanol) at 308C for 30 min. For treatment b and d, Csk in the presence of protein tyrosine phosphatase 1B. Src Src pre-treated as in a and c were incubated with 50 ng PTP 1B (0.12 mg per reaction) was incubated with dierent combinations and 10 mM EDTA for 15 min, respectively. For treatment e, Src of the following: Csk (12 ng per reaction), ATP (0.1 mM), MgCl2 sample pre-treated as in c was further incubated with Csk (0.6 mg) (10 mM) and PTP 1B (50 ng) in the kinase assay buer for 30 min for 30 min. For treatment f, Src sample from treatment e was before the Src activity was assayed. In the Src activity assay, incubated with PTP 1B and EDTA for 15 min. 10% of the 1mM ammonium molybdate was included in all assays to inhibit sample from each treatment was withdrawn to determine Src PTP 1B that is present in some treatments. We have kinase activity, and the rest was fractionated by SDS ± PAGE. independently veri®ed that ammonium molybdate at this The phosphorylation level of Src was determined by Cenrenkov concentration completely inhibits PTP 1B but does not inhibit scintillation counting of the excised Src protein bands, which were the activity of Src and Csk then used for peptide mapping Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1591 approximately 1 mol of phosphate per mol of Src. Based on the data presented above, it is clear that Peptide mapping indicates that the phosphate label autophosphorylation of Src blocks its regulation by was speci®cally on the 10 kD peptide fragment Csk phosphorylation. To further test this model and containing Ya. Treatment of autophosphorylated Src investigate whether this applies to other members of in treatment c with PTP 1B (treatment d) reversed the the Src family of PTKs, we compared the in vitro activating eect of autophosphorylation and led to a inactivation of wild-type and a Phe autophosphoryla- loss of over 85% of the label. The more extensive tion mutant of human Yes. Yes is a Src family PTK dephosphorylation of Ya over Yr (treatment bvsd) that is closest to Src in primary structure. We have con®rms the kinetic selectivity of PTP 1B for Ya. previously expressed human Yes in a bacterial Further incubation of autophosphorylated Src in expression system and shown that this enzyme can be treatment c with Csk (treatment e) did not result in activated by Csk phosphorylation when a stoichio- any inactivation of Src and led to a total incorpora- metric amount of Csk was used (Sun and Budde, tion of over 1.6 mol of phosphate per mol of Src. 1997b). We generated a mutant Yes enzyme in which

Peptide mapping indicates that Yr was also phos- the autophosphorylation site, Tyr-426 (Ya), was phorylated, to a similar level as Ya. This result changed to a phenylalanine (Fa). To ensure that the indicates that Csk phosphorylates but does not mutation does not change the property of the enzyme inactivate autophosphorylated Src. Treatment of Src globally, we compared the thermal stability pro®les of Ya Fa phosphorylated on both Ya and Yr from treatment e the wild-type, Yes and the mutant, Yes , enzymes with PTP 1B (treatment f) led to a loss of over 75% of (Figure 7a). Both enzymes are stable when incubated activity of the Src phosphorylated on both Ya and Yr, for 10 min under 258C and start to denature when not only reversing the activating eect of autopho- incubated at above 258C. A 10 min incubation sphorylation, but also leading to inactivation of Src by inactivated both enzymes by 50% at 348C and 100% the prior phosphorylation on Yr. This treatment at 418C. The identical thermal inactivation pro®les of resulted in a loss of about 60% of the phosphate. the two enzymes indicate that the Ya to Fa mutation

Peptide mapping indicates that Ya becomes completely does not aect the overall structure of the enzyme. We Ya dephosphorylated while majority of Yr remains then examined the time course of inactivation of Yes phosphorylated, further demonstrating the speci®city and YesFa by Csk. In this experiment, a Csk : Yes ratio of PTP 1B for Ya. From these studies, we drew two of 1 : 20 was chosen. As shown in Figure 7b, the level conclusions: ®rst, Src autophosphorylation does not of inactivation of YesYa reached approximately 50% in block Yr phosphorylation by Csk, but blocks the the ®rst 10 min and leveled o. Longer incubation up inactivating eect of Yr phosphorylation; second, PTP to 40 min did not result in further inactivation. The

1B speci®cally dephosphorylates Ya compared to Yr. time-dependent inactivation in the ®rst 10 min is likely Since the doubly phosphorylated form is hyperac- the result of slower autophosphorylation of bacterially tive, similar to the Ya-phosphorylated form, an important question is if the phospho-Yr is bound to the SH2 domain in this form. This question can be directly addressed by resolution of the tertiary structure of this form. We indirectly addressed this question by asking if the SH2 domain in this form is available to bind phospho-Tyr-Agarose. Phospho-Tyr- Agarose has been used for the anity puri®cation of a number of SH2 containing proteins (Koegl et al., 32 1994). Src was labeled with P-phosphate on Ya,Yr or both and phospho-Tyr Agarose adsorption was performed on all three forms (Figure 6). For Src that was phosphorylated on Ya (stoichiometry of phosphorylation =0.95 mol phosphate per mole of Src), approximately 80% was precipitated by phospho-Tyr- Agarose (lane 2 vs 1), while for Src phosphorylated on

Yr (stoichiometry of phosphorylation =0.99 mol phosphate per mole of Src), only 13% was precipitated (lane 4 vs 3). This result shows that the SH2 domain is accessible for phospho-Tyr-Agarose binding in the phospho-Ya form but not in the phospho-Yr form. For the form that was phosphorylated on both Figure 6 Accessibility of SH2 domain for phospho-Tyr-Agarose Y and Y (stoichiometry of phosphorylation =2.2 mol a r binding. Src that was phosphorylated on Ya (lanes 1 and 2), Yr phosphate per mole of Src), approximately 6% was (lanes 3 and 4) or both Ya and Yr (lanes 5 and 6) were prepared precipitated by phospho-Try-Agarose, indicating that as in Figure 5 (treatments a, c and e). One ®fth of each sample the SH2 domain binding site for phospho-Tyr is not was directly applied to SDS ± PAGE (lanes 1, 3 and 5, respectively). The rest was mixed with equal volume of accessible for phospho-Tyr-Agarose in the doubly phospho-Tyr-Agarose suspension in the reaction buer and phosphorylated form. This conclusion is consistent incubated on ice for 10 min. The beads were then pelleted by with the observation that phospho-Ya is preferably centrifugation at 4000 r.p.m. for 1 min, and the proteins dephosphorylated by PTP 1B over phospho-Yr in the associated with the beads were analysed by SDS ± PAGE (lanes doubly phosphorylated form. These two observations 2, 4 and 6). The sample applied to all lanes was equivalent to one ®fth of the original sample (3.5 pmol). The 32P-labeling on Src together suggests that phospho-Yr is bound to SH2 was determined by liquid scintillation counting of the excised Src domain in the doubly phosphorylated form. band Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1592

expressed Yes enzyme due to its lower Vmax than that of Src, which was expressed in a eukaryotic system. In contrast to the inactivation pattern of YesYa, YesFa was inactivated virtually completely in a time dependent manner in 40 min. This result further demonstrates that the lack of time-dependent complete inactivation of YesYa by Csk is the result of autophosphorylation. Furthermore, these results indicate that the autophosphorylation that blocks Csk inactivation is on the main autophosphorylation site within the activation loop of Yes and Src, although other sites have been reported to also undergo autophosphorylation in the Src family (Osusky et al., 1995; Barker et al., 1995; Jullien et al., 1994). This result also indicates that autophosphorylation will likely have the same function in modulating Csk regulation of other PTKs in the Src family.

Discussion

Many lines of evidence support the model that Csk phosphorylates the Src family PTKs and down regulates their activity. However attempts to recon- stitute the reaction have not been fully successful. It was reported by several independent laboratories (Okada and Nakagawa, 1989; Ruzzene et al., 1994; Koegl et al., 1994; Xu et al., 1997) that a Csk : Src molar ratio of more than one was required to achieve a reasonable level of Src inactivation. In the present study we found that this apparent ineciency is the result of Src autophosphorylation. We demonstrated that after Src is autophosphorylated, it can be phosphorylated but not inactivated by Csk. When autophosphorylation is kept to a minimum or Figure 7 Kinetics of wild type and autophosphorylation mutant prevented by inclusion of a Y -speci®c PTP or Ya a human Yes. (a) Thermal inactivation of GST-Yes and GST- mutating the Y to phenylalanine, Src and Yes can YesFa. The enzymes in kinase assay buer were incubated at a dierent temperatures for 10 min, cooled on ice and the activities be eectively inactivated by a sub-stoichiometric were assayed with polyE4Y as the substrate. The assays were amount of Csk in a time-dependent manner. These carried out as described (Sun and Budde, 1997b). (b) Inactivation ®ndings indicate that autophosphorylation not only of wild-type and mutant Yes by Csk phosphorylation. Yes directly activates Src, but is also a key element in its enzymes (2 mg per reaction) were incubated with Csk (0.1 mg per regulation by Csk phosphorylation. These results reaction), ATP (0.1 mM) and MgCl2 (10 mM) for dierent time and then the activity of Yes was assayed with RCM-lysozyme as provide a better understanding of how Src activity

the substrate is controlled by its phosphorylation state on Ya and

Yr. A comprehensive model based on the ®ndings in this report and consistent with observations in the literature is shown in Figure 8. In this model, the phosphoryla-

tion of only Ya and Yr is considered. Once autopho-

sphorylated on Ya, Src becomes hyperactive regardless

of the phosphorylation status of Yr, although Yr can still be phosphorylated to form the doubly phosphorylated form. In support of this model, Lck that is

phosphorylated on both Ya and Yr and hyperactive was recently isolated from Jurkat human leukemic T cells (Hardwick and Sefton, 1997). Csk phosphorylation can only inactivate Src that is not autophosphorylated. Two dierent protein tyrosine phosphatases are responsible for the dephosphorylation of the two phosphorylation sites, respectively. The conversion of

the inactive Yr-phosphorylated form to the hyperactive doubly phosphorylated form by autophosphorylation Figure 8 Regulation of Src activity by reversible autopho- (a trans-mechanism) has not been observed by us. The sphorylation and Csk phosphorylation. The autophosphorylation two hyperactive forms may be especially relevant to site is indicated as Y and Csk phosphorylation site is indicated as a understanding the oncogenic potential of Src in Yr. PTP x is the protein tyrosine phosphatase that speci®cally dephosphorylate Yr tumorigenesis. Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1593

The conclusion in this paper is consistent with our preference of PTP 1B for phospho-Ya is likely the current knowledge of how Src is inactivated by Yr result of phospho-Yr being bound to the SH2 domain phosphorylation, and further expands the current and thus not accessible to PTP 1B. Further study is model. A comparison of crystal structures of Src required to verify this role in vivo and understand the inactivated by Yr phosphorylation (Xu et al., 1997), basis for the preference for PTP 1B. Several PTPs have with other activated PTKs, including Lck (Yamaguchi been proposed to speci®cally dephosphorylate the Yr of and Hendrickson, 1996) and insulin receptor kinase Src, including SH2 domain-containing PTP (Somani et (Hubbard, 1997) suggests that the inactivation caused al., 1997) and trans-membrane PTP a (Zheng et al., by Yr-phosphorylation of Src is executed by the 1992) and PTP l (Chappel et al., 1997). disposition of Helix C located in the ATP-binding Consistent with this model is the observation that in lobe of the catalytic domain. Helix C contains the normal untransformed cells, Src is fully phosphorylated conserved Glu-310, which is essential for the activity of on Yr but not on Ya. Activation of Src during mitosis kinases. In the active kinases, the side chain of this is accompanied by the dephosphorylation of Yr residue projects into the catalytic cleft and forms a salt (Bagrodia et al., 1991). However, Src activation and bridge with Lys-295, a residue important for binding transforming potential correlate directly to the ATP. In the structure of the inactive Src, however, autophosphorylation status, not to the phosphoryla-

Helix C moves away from the catalytic cleft and Glu- tion level of Yr (Jove et al., 1989; Iba et al., 1985; 310 faces outward where it is stabilized by a salt-bridge Cheng et al., 1988). These pieces of evidence support with Arg-385. Two groups of interactions are our conclusion that autophosphorylation of Src family responsible for the disposition of Helix C: the SH2- PTKs overrides their regulation by Csk phosphoryla- phospho-Yr binding dependent interactions of the C- tion. end of Helix C with the linker between the SH2 and This model also provides a more plausible explana- SH1 catalytic domain, and the interaction between the tion for several observations that have been reported in N-end of Helix C to the beginning segment of the the literature. First, the autophosphorylation mutant of activation loop. It has been proposed that removal of Src (Tyr to Phe) has much reduced transforming either one of these two interactions would result in potential than the wild type enzyme although this rearrangement of Helix C and activation of Src. It is mutation does not greatly aect the activity of the expected that phosphorylation of Ya would change the enzyme (Kmiecik and Shalloway, 1987). Our current conformation of the activation loop and remove the model would propose that the major dierence in the constraints on Helix C from the N-end. Our results wild-type and the mutant enzymes is not the enzymatic veri®ed this hypothesis and demonstrated that phos- activity but the ability of the wild-type Src to phorylation of Ya does lead to activation of Src even autophosphorylate and escape Csk inactivation, while when Yr is also phosphorylated. the mutant enzyme, incapable of autophosphorylation, The results raise the question what happens to the will be inactivated by Csk and thus have reduced interaction between SH2 and the phospho-Yr in the transforming potential. Second, in colorectal cancer doubly phosphorylated and hyper-activated Src? Two cell lines with activated Src, Csk is highly expressed experiments in this current study addressed this and fully active (Li et al., 1996) and furthermore the question. First, the phospho-Yr in doubly phosphory- level of Csk expression corresponds to the elevated lated Src appears not accessible for dephosphorylation levels of Src and Yes activities (Watanabe et al., 1995). by PTP 1B. Second, the SH2 domain in the doubly These ®ndings led to the suggestion that Csk may not phosphorylated Src is not available to bind to have an anti-oncogenic role to play through the phospho-Tyr-Agarose. These two observations to- negative regulation of Src family kinases in colorectal gether suggest that phospho-Yr is still bound to the carcinogenesis. It was recently reported (Chedin et al., SH2 domain in the doubly phosphorylated form. 1997) that human breast tumor tissues contain two Although our results demonstrated that this form of major soluble and highly active PTKs, one is Src and Src is hyperactive, similar to autophosphorylated Src, the other is Csk. Relative to these observations in both these two forms are expected to have dierent the colorectal and breast cancer tissue and cell lines, properties, such as their ability to interact with our model would suggest that Src and Yes are phospho-Tyr containing proteins. autophosphorylated and thus are able to escape the Besides proposing a far more important role for inactivation by Csk phosphorylation. A direct exam- autophosphorylation in controlling Src activity, this ination of Src activity, phosphorylation level on Ya and model also positions the PTP that speci®cally depho- Yr, Csk protein level and kinase activity in tumor sphorylates the Src autophosphorylation site at a tissues and cell lines will further test this model. critical point in the regulation of Src activity and function. The importance of this enzyme is demonstrated by two experiments in this study. First, in the Materials and methods absence of such a PTP activity, inactivation of Src requires a stoichiometric amount of Csk. In the Enzymes and other chemicals presence of PTP 1B activity, Src can be eectively Recombinant Src (3900 U mg71 with polyE Y as substrate) inactivated by a sub-stoichiometric amount of Csk. 4 was expressed in and puri®ed from baculovirus-infected Second, Src that is phosphorylated on both Y and Y a r insect cells (Budde et al., 1993). Csk (130 U mg71 with are hyperactive, and dephosphorylation of Ya by this polyE4Y as substrate) and protein tyrosine phosphatase 1B phosphatase converts Src to the inactive form that is (16 000 U mg71 with [32P]RCM-lysozyme as substrate) only phosphorylated on Yr. Our data presented in this were expressed as a glutathione S-transferase (GST) study indicate that PTP 1B is able to serve this fusion protein in Escherichia coli and puri®ed as described function in that it selectively dephosphorylates Ya. The (Sun and Budde, 1995). The Csk used in this study was Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1594 tagged with the 10-amino acid strep-tag at the C-terminus fractionated by SDS ± PAGE and the phosphorylation and puri®ed to apparent homogeneity. The E. coli level of Src visualized by autoradiography and quanti®ed expressed Csk has a speci®c activity similar to that by liquid scintillation counting of the excised Src protein puri®ed from the brain. The PTP 1B was in the form of band. To determine the eect of PTP 1B treatment on the a full length GST-PTP 1B fusion protein. GST-PTP 1B ability of Src to be inactivated by Csk phosphorylation, the phosphatase activity is readily lost through one freeze-thaw Src sample treated as described above was assayed in the cycle but stable for at least 6 months when stored in 50% presence of excess Csk (Csk : Src=3 : 1 in molar ratio) and glycerol at 7208C. One unit of PTP 1B activity is de®ned 1mM ammonium molybdate to inhibit PTP 1B. astheamountofenzymethatcatalyzesthereleaseof 1 nmole of phosphate per min from [32P]RCM-L. Phospho- Inactivation of Src by Csk phosphorylation Tyr-Agarose was purchased from Sigma Chemical Co. To quantify the inactivation of Src by Csk, Src activity was assayed using RCM-L as substrate in the presence of excess Protein tyrosine kinase and protein tyrosine phosphatase activity Csk (Csk : Src=3 : 1 in molar ratio). Since RCM-L is a far assays superior substrate for Src than for Csk (dierence in

Routine PTK activity assays were carried out as described Vmax/Km ratio4100-fold), Src activity can be accurately previously (Sun and Budde, 1997a). The PTP activity was determined without removing Csk from the assay. The determined by measuring the ability to dephosphorylate percentage of inactivation of Src by Csk is calculated as [32P]RCM-L, that was phosphorylated by Src as described follows: (Src activity7Src activity in the presence of Csk)/ (Tonks et al., 1991). The assay was conducted in a 50 ml Src activity6100. volume that included 50 mM Tris-Cl (pH 7.5), 0.05% b- mercaptoethanol, 0.1mgml71 bovine serum albumin, Cyanogen bromide digestion and peptide mapping of Src recombinant GST-PTP 1B, and the [32P]RCM-L substrate (1000 d.p.m. pmol71). The substrate was added last to After various treatments, Src was precipitated with 10% initiate the reaction, which was then incubated at 308Cfor trichloroacetic acid and fractionated on SDS ± PAGE and 10 min. The reaction was stopped by the addition of 10 ml the Src protein bands were excised. Cyanogen bromide of 100% trichloroacetic acid (w/v). The mixture was digestion and peptide mapping were performed as incubated on ice for 5 min and centrifuged 14 000 g for described (Jove et al., 1989). 5 min to pellet the protein from the mixture. An aliquot of 30 ml was removed from the supernatant for liquid Generation and expression of YesFa scintillation counting to determine the amount of phosphate released from [32P]RCM-L. The mutation of Tyr-426 to Phe-426 was introduced into the Yes coding sequence in the plasmid pGEX-Yes with PCR and other standard molecular biology methods Autophosphorylation of Src (Sambrook et al., 1989). The PCR ampli®ed portion was To determine the time course of Src autophosphorylation, sequenced and the designed mutation was con®rmed. The puri®ed Src protein was incubated at 308C with 0.2 mM resultant plasmid pGEX-YesFa was used to transform 71 ATP (10 000 d.p.m. pmol ), 10 mM MgCl2 in 75 mM DH5a competent cells for protein expression. The mutant EPPS-NaOH (pH 8.0), 5% glycerol, 0.005% Triton X- fusion protein, GST-YesFa, was expressed and puri®ed as 100 and 0.05% b-mercaptoethanol. At the indicated time, previously described for the wild-type enzyme (Sun and aliquots were removed and mixed with an equal volume of Budde, 1997b). sample buer and prepared for SDS ± PAGE. Phosphory- lated proteins were identi®ed by autoradiography after SDS ± PAGE. The phosphorylation level of Src was quanti®ed by liquid scintillation counting of the excised Src protein band. The stoichiometry of autophosphoryla- Abbreviations tion was calculated by dividing the mole of phosphate GST, glutathione S-transferase; PTK, protein tyrosine incorporated by the mole of Src protein loaded onto the kinase; RCM-L, carboxymethylated-maleylated, reduced gel. lysozyme; PTP, protein tyrosine phosphatase; SH domain, Src homology domain.

Dephosphorylation of Src by PTP 1B Src was autophosphorylated by pre-incubating with Acknowledgements [g-32P]ATP-Mg as described above. The dephosphorylation This research was supported by National Cancer Institute of the autophosphorylated Src was performed by incuba- grant CA53617 and the John S Dunn Foundation

tion with 10 U of PTP 1B in the presence of 10 mM Na2 (Houston, TX). DNA sequencing was performed by the EDTA for the indicated time. The proteins were MDACC Sequencing Core Facility (Grant CA16672).

References

Amrein KE, Takacs B, Steiger M, Molnos J, Flint NA and Bougret C, Delaunay T, Romero F, Jullien P, Sabe H, Burn P. (1995). Proc. Natl. Acad. Sci. USA, 92, 1048 ± Hanafusa H, Benarous R and Fischer S. (1996). J. Biol. 1052. Chem., 271, 7265 ± 7472. Bagrodia S, Chackalaparampil I, Kmiecik TE and Shalloway Brickell PM. (1992). Crit. Rev. Oncogenesis, 3, 401 ± 446. D. (1991). Nature, 349, 172 ± 175. Brown MT and Cooper JA. (1996). Biochim. Biophys. Acta, BarkerSC,KasselDB,WeiglD,HuangX,LutherMAand 1287, 121 ± 149. Knight WB. (1995). Biochemistry, 34, 14843 ± 14851. Budde RJA, Ramdas L and Ke S. (1993). Prep. Biochem., 23, Bolen JB, Veillette A, Schwartz AM, Deseau V and Rosen N. 493 ± 515. (1987). Oncogene Res., 1, 149 ± 168. Regulation of Src, Yes by auto- and trans-phosphorylation GSunet al 1595 Chappel J, Ross PR, Abu-Amer Y, Shaw A and Teitelbaum Sambrook J, Fritsch EF and Maniatis T. (1989). Molec. SL. (1997). J. Cell. Biochem., 67, 432 ± 438. Cloning. Cold Spring Harbor Laboratory Press. Chedin M, Filhol O, Duminy C, Bolla M, Benistant C, Sicheri F, Moare® I and Kuriyan J. (1997). Nature, 385, Roche S, Chambaz EM and Cochet C. (1997). Carcino- 602 ± 609. gen., 18, 1463 ± 1472. Smith JA, Francis SH and Corbin JD. (1993). Mol. Cell. Cheng SH, Piwnica-Worms H, Harvey RW, Roberts TM and Biochem., 127/128, 51 ± 70. Smith AE. (1988). Mol. Cell. Biol., 8, 1736 ± 1747. Somani A-K, Bignon JS, Mills GB, Siminovitch KA and Cooper JA, Gould KL, Cartwright CA and Hunter T. Branch DR. (1997). J. Biol. Chem., 272, 21113 ± 21119. (1986). Science, 231, 1431 ± 1434. Stover DR, Leibetanz J and Lydon NB. (1994). J. Biol. Courtneidge SA. (1994). Cancer Biol., 5, 239 ± 246. Chem., 269, 26885 ± 26889. Hardwick JS and Sefton BM. (1997). J. Biol. Chem., 272, Sun G and Budde RJA. (1995). Anal. Biochem., 231, 458 ± 25429 ± 25432. 460. Hubbard SR. (1997). EMBO J., 16, 5572 ± 5581. Sun G and Budde RJA. (1997a). Biochemistry, 36, 2139 ± Iba H, Cross FR, Garber EA and Hanafusa H. (1985). Mol. 2146. Cell. Biol., 5, 1058 ± 1066. Sun G and Budde RJA. (1997b). Arch. Biochem. Biophys., Imamoto A and Soriano P. (1993). Cell, 73, 1117 ± 1124. 345, 135 ± 142. Jove R, Hanafusa T, Hamaguchi M and Hanafusa H. (1989). Sun G, Ke S and Budde RJA. (1997). Arch. Biochem. Oncogene Res., 5, 49 ± 60. Biophys., 343, 194 ± 200. Jullien P, Bougeret C, Camoin L, Bodeus M, Durand H, Superti-Furga G, Fumagalli S, Koegl M, Courtneidge SA Disanto JP, Fischer S and Benarous R. (1994). Eur. J. and Draetta G. (1993). EMBO J., 12, 2625 ± 2634. Biochem., 224, 589 ± 596. Talamonti MS, Roh MS, Curley SA and Gallick GE. (1993). Kmiecik TE and Shalloway D. (1987). Cell, 49, 65 ± 73. J. Clin. Invest., 91, 53 ± 60. KoeglM,KyptaRM,BermanM,AlitaloKandCourtneidge Thomas SM and Brugge JS. (1997). Annu. Rev. Cell Dev. SA. (1994). Biochem. J., 302, 737 ± 744. Biol., 13, 513 ± 609. Levitzki A. (1996). Anti-Cancer Drug Design, 11, 175 ± 182. Tonks NK, Diltz CD and Fischer EH. (1998a). J. Biol. Li S, Ke S and Budde RJA. (1996). Cell Biol. Int., 20, 723 ± Chem., 263, 6722 ± 6730. 729. Tonks NK, Diltz CD and Fischer EH. (1988b). J. Biol. Nada S, Yagi T, Takeda H, Tokunaga T, Okada M and Chem., 263, 6731 ± 6737. Aizawa S. (1993). Cell, 73, 1125 ± 1135. Tonks NK, Diltz CD and Fischer EH. (1991). Methods Oetken C, Couture C, Bergman M, Bonnefoy-Berard N, Enzymol., 201, 427 ± 442. Williams S, Alitalo K, Burn P and Mustelin T. (1994). Watanabe N, Matsuda S, Kuramochi S, Tsuzuku J, Oncogene, 9, 1625 ± 1631. Yamamoto T and Endo K. (1995). Japan. J. Clin. Okada M and Nakagawa H. (1989). J. Biol. Chem., 264, Oncol., 25, 5±9. 20886 ± 20893. Xu W, Harrison SC and Eck MJ. (1997). Nature, 385, 595 ± Osusky M, Taylor SJ and Shalloway D. (1995). J. Biol. 602. Chem., 270, 25729 ± 25732. Yamaguchi H and Hendrickson WA. (1996). Nature, 384, Ruzzene M, James P, Brunati AM, Donella-Deana A and 484 ± 489. Pinna LA. (1994). J. Biol. Chem., 269, 15885 ± 15891. Zheng XM, Wang Y and Pallen CJ. (1992). Nature, 359, Sabe H, Knudsen B, Okada M, Nada S, Nakagawa H and 336 ± 339. Hanafusa H. (1992). Proc. Natl. Acad. Sci. USA, 89, 2190 ± 2194.