Oncogene (2000) 19, 3961 ± 3970 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc The activation loop in Lck regulates oncogenic potential by inhibiting basal activity and restricting substrate speci®city

Lorraine E Laham1, Nishit Mukhopadhyay1 and Thomas M Roberts*,1

1Department of Cancer Biology, Dana-Farber Cancer Institute, 44 Binney Street, Boston, Massachusetts, MA 02115, USA

The activities of Src-family non-receptor 1996) and SF-NRTKs (Schindler et al., 1999; Sicheri et are regulated by structural changes that alter the al., 1997; Xu et al., 1997, 1999; Yamaguchi and orientation of key residues within the catalytic domain. Hendrickson, 1996) complement earlier mutagenesis In this study, we investigate the e€ects of activation loop studies that revealed the importance of the activation mutations on regulation of the -speci®c kinase loop tyrosine for activity. The catalytic domain consists Lck (p56lck). Substitution of 5 ± 7 residues amino terminal of two globular lobes and enzymatic activity is to the conserved activation loop tyrosine (Y394) increases regulated through conformational changes that a€ect kinase activity and oncogenic potential regardless of positioning of residues critical for ATP binding, g- regulatory C-terminal tail levels (Y505), phosphate transfer, and substrate interaction. Struc- while most mutations in the 13 residues carboxyl to Y394 tures of SF-NRTKs suggest that stabilization of the decrease kinase activity. Phosphorylation of the C- regulatory changes hinge on the phosphorylation state terminal regulatory tail is carried out by the cytosolic of the activation loop tyrosine residue. Negative Csk and we ®nd that mutations upstream regulation results from misalignment of the catalytic or downstream of Y394 or mutation of Y394 do not a€ect residues in the N-terminal lobe of the kinase where in the level of Y505 phosphorylation. In addition, we report the unphosphorylated Src and Hck , the that mutations on either side of Y394 a€ect substrate catalytic lysine residue (K295 in Src) is displaced from speci®city in vivo. We conclude that the high degree of the correct orientation necessary for electrophillic conservation across the entire activation loop of Src- attack on ATP and g-phosphate transfer. The family kinases is critical for normal regulation of kinase structures appear to be stabilized by steric interactions activity and oncogenicity as well as substrate selection. of residues at the beginning of the activation loop with Oncogene (2000) 19, 3961 ± 3970. the N-terminal lobe of the kinase domain. Inactive conformations of the kinases also appear to block Keywords: Lck; tyrosine kinase; activation loop; sub- access to the substrate . Phosphorylation of strate speci®city the activation loop tyrosine induces and stabilizes conformational changes that shift the N-terminal lobe residues into an active conformation and allows access Introduction to the substrate binding site. Although a number of mutational studies have The activities of Src-family non-receptor tyrosine shown the importance of the activation loop tyrosine, kinases [SF-NRTKs] are upregulated following phos- relatively few studies have proven the regulatory phorylation of a conserved tyrosine residue (Abraham contributions of the remaining activation loop residues. and Veillette, 1990; Cartwright et al., 1987; Kmiecik The structural analysis studies of kinases in complex and Shalloway, 1987; Marth et al., 1988; Piwnica- with inhibitor peptides suggest that this region may Worms et al., 1987) within the activation loop also be important for substrate selection (Hubbard, subdomain (Johnson et al., 1996). Phosphorylation in 1997; Madhusudan et al., 1994; Zheng et al., 1993). p60src occurs primarily through an intermolecular Several naturally occurring activation loop mutations mechanism involving another molecule of Src or a in the Ret receptor (Hofstra et al., 1994) and Btk di€erent tyrosine kinase (Cooper and MacAuley, 1988; (Sideras and Smith, 1995) tyrosine kinases a€ect Osusky et al., 1995). Sequence in this region is highly activation while mutation at a di€erent residue in the conserved among SF-NRTKs but diverges between Kit receptor alters substrate selection (Piao et al., kinase subfamilies (Figure 1a) and previous mutational 1996). Substitution of the loop in Lck with a studies focused primarily on the tyrosine residue itself. homologous domain from the Ser/Thr kinase c-raf A Tyr to Phe substitution does not destroy kinase decreased autophosphorylation and altered substrate activity of p60src in vitro, however, removal in SF- speci®city (Carrera et al., 1995). NRTKs decreases both kinase activity in vivo and the In this study, we have examined the contributions of ability to transform cells (Abraham and Veillette, 1990; activation loop residues in regulating the activity of Kmiecik and Shalloway, 1987; Piwnica-Worms et al., Lck in vivo (Figure 1b). We have made single and triple 1987). amino acid substitutions throughout the activation Recent crystal structure analyses of receptor (Hub- loop (Ala381 ± Ala408) with speci®c amino acids chosen bard, 1997; Hubbard et al., 1994; Mohammadi et al., to mimic those found in other kinases to retain activity or to add charged residues to maximize changes in substrate speci®city (Figure 1c). Mutants were then *Correspondence: TM Roberts expressed in bacteria or mammalian cells and e€ects on Received 8 May 2000; revised 5 June 2000; accepted 13 June 2000 kinase regulation were examined. With exception to the Effects of activation loop mutations on Lck regulation LE Laham et al 3962 SH2 interactions at this site. Mutant enzymes isolated from bacterial and mammalian expression systems were assayed using both the wildtype (Wt) and activating (Y505F) backgrounds. Since we were concerned about the e€ects of these mutations on the oncogenic potential of Lck as well its biochemical behavior, all mammalian measurements were made in murine ®broblasts rather than in transformed lines.

Mutations N-terminal to Y394 can activate the kinase Crystal structure analyses of receptor tyrosine kinases and SF-NRTKs suggest that amino acids near the beginning of the activation loop are involved with the regulation of kinase activity (Hubbard et al., 1994; Mohammadi et al., 1996; Schindler et al., 1999; Sicheri et al., 1997; Xu et al., 1997, 1999; Yamaguchi and Hendrickson, 1996). To investigate the contributions of this region to kinase regulation, we have analysed the e€ects of mutations at two sites in the activation loop (Leu388 and Glu390) on the activity of Wt Lck or the activated variant, Y505F, expressed in both bacteria and mammalian cells. Because bacteria do not express endogenous tyrosine kinases, no PTyr are observed in the bacterial lysates in the absence of exogenous Lck expression (data not shown). When Lck is expressed in the bacteria, both Wt and Y505F mutant proteins exhibit similar kinase activities as judged by autokinase activity and by overall levels of PTyr on bacterial substrates because the bacteria lack expres- sion of the down regulatory kinase Csk (Figure 2a). Figure 1 (a) Conservation of the activation loop sequence in SF- Mutations at Leu to Asn (L N) and at Glu to NRTKs. The activation loop in SF-NRTKs is highly conserved 388 388 390 but diverges in other tyrosine kinases. Residues in Lck that were Thr (E390T) have little a€ect on activity, both in the Wt mutated in this study are shaded. (b) A schematic drawing of Lck and Y505F backgrounds. Mutation of each site to identifying the location of amino acids and domains that bulkier (Met; E390M) or charged residues (Asp; L388D) regulate kinase activity. Speci®c substitutions made in the measurably reduces kinase activity towards bacterial activation loop of Lck are summarized in (c). Sequences for the human homologs were obtained from GenBank with numbering proteins (as does the double mutant L388D/E390M). for Src (Xu et al., 1997), Lck (Yamaguchi and Hendrickson, The same experiment can also be used to analyse the 1996), Hck (Sicheri et al., 1997), FGFR (Mohammadi et al., mutants for di€erences in substrate speci®city by 1996), and IRK (Hubbard et al., 1994) as referenced in previous comparing the proteins tyrosine phosphorylated in crystal structure studies the total lysates. An activation loop mutation was determined to a€ect substrate speci®city if the PTyr pattern for the mutant di€ered from that of the Wt activating tyrosine residue itself, no other amino acids and Y505F lysates, either by adding new proteins to within the loop have been reported to be essential for the phosphorylation pattern or by decreasing tyrosine the binding or catalysis of ATP, yet we ®nd that many phosphorylation on a subset of the proteins. Despite of the substitutions have signi®cant e€ects on kinase their apparent distance from the reported substrate activity and cellular transformation potential. We binding pocket, mutations at both Leu388 and Glu390 propose that the high degree of activation loop have subtle e€ects on speci®city, decreasing the PTyr sequence conservation in SF-NRTKs is important both content on some proteins while not a€ecting others, for substrate speci®city and regulation of kinase depending on the amino acid substitution. Substitution activity. of a charged residue (Asp) at Leu388 has more a€ect on substrate selection and overall activity than substitu- tion to the neutral amino acid (Asn). Similar results Results were observed for the mutations at Glu390 where substitution to Met has more a€ect on activity than We have made a series of amino acid mutations does the substitution to Thr. Notably, the only throughout the activation loop of Lck to assess the previously reported mutations in this portion of the contributions of the region on regulation or kinase activation loop of any tyrosine kinase were sponta- activity and substrate selection (Figure 1). Single amino neous activating mutations in the Kit receptor at acid substitutions were chosen either to resemble Asp816, analogous to position Leu388 in Lck. In these residues found in other kinases or to add charged mutants, the Asp residue was substituted with either residues to maximize the potential e€ects on substrate Val or Tyr, which also led to altered substrate selection. In the case of changes immediately down- speci®city (Piao et al., 1996). stream of the activating tyrosine Y394, triplet substitu- To examine the e€ects of these substitutions on tions were made to maximize disruption of potential kinase activity in the presence of the downregulatory

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3963

Figure 2 Analysis of mutations in the N-terminus of the activation loop. (a) Expression in bacteria. Wt Lck, Lck containing activation loop mutations below Y394 (L388D, L388N, E390M, E390T, and L388D/E390M), and corresponding Y505F double mutants were subcloned into the prokaryotic expression vector pGEX-2T. Uninduced bacterial total lysates normalized for Lck levels were analysed for di€erences in PTyr content by Western blotting using anti-PTyr (4G10). Blots were stripped and then re-probed using anti-Lck polyclonal sera. The position of the gst-Lck protein is indicated. Di€erences in substrate phosphorylation patterns due to Lck proteolysis have been ruled out by using both N- and C-terminal Lck peptide anti-sera to identify breakdown products. (b) Expression in mammalian cells. Drug-resistant pools of BalbC 3T3 ®broblasts expressing either Wt Lck, Lck containing activation loop mutations N-terminal to Y394 (L388D, L388N, E390M, E390T, and L388D/E390M), and corresponding Y505F double mutants were analysed for di€erences in PTyr content by Western blotting. Total lysates normalized for Lck levels were fractionated by SDS ± PAGE, transferred to polyvinylidene di¯uoride membrane, and incubated with monoclonal PTyr antibodies. The blot was ®rst probed for PTyr, stripped, and then re-blotted with polyclonal anti-Lck. (c) Phase contrast photos (106) of the ®broblast monolayers used to prepare the Western blots shown in (b)

kinase Csk, the mutant Lck enzymes were expressed in while mutations that decrease kinase activity or change ®broblasts. Previous studies have reported that Wt Lck substrate selection would be revealed using the is tyrosine phosphorylated only at Y505 in ®broblasts, activated Y505F background. The levels of PTyr which serves to downregulate the via an associated with the E390M and E390T single mutants intramolecular SH2 interaction (Abraham and Veill- were similar to that of Wt Lck, presumably due to Csk ette, 1990; Amrein and Sefton, 1988; Gervais et al., downregulation at Y505. However, the L388D, L388N, 1993; Marth et al., 1988; Weil and Veillette, 1994). In and L388D/E390M mutants display increased PTyr contrast to the activity observed with expression in levels compared to Wt Lck and have increased abilities bacteria, Wt Lck is inactive towards proteins in to phosphorylate cellular proteins (more easily visible mammalian cells while Y505F Lck is constitutively on longer exposures). When these mutants are placed active and transforming (Figure 2b). Mutations that in a Y505F background, they display activities that activate Lck would be apparent in the Wt background approach that of the single Y505F mutant. The subtle

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3964 di€erences in substrate speci®city observed when the YCPK/Y505F and YNQQ/Y505F mutants were able to L388D, E390M, and L388D/E390M mutants were ex- phosphorylate proteins more readily than the YGQQ/ pressed in bacteria are not detectable upon expression Y505F and Y394F/Y505F mutants (Figure 3b). The in ®broblasts. overall PTyr content in cells expressing the Lck Y394- The ability of a given Lck mutant to direct downstream mutants is much lower than that in the phosphorylation of cellular proteins correlates well single Y505F mutant. Moreover, in the Y505F back- with the degree of transformed morphology in the ground, the mutants all display much lower levels of ®broblast monolayers (Figure 2c). Thus, the L388N tyrosine phosphorylation on Lck (presumably at Y394) cells exhibited the greatest increase in kinase activity in and the decrease in PTyr content correlates with the Wt background (Figure 2b), and, not surprisingly, decreased ability to phosphorylate endogenous pro- the L388N cells appeared to be partially transformed, teins. Correspondingly, as the level of phosphorylation having a more rounded phenotype (Figure 2c). on endogenous proteins decreases, the degree of Similarly, most of the mutants show slightly decreased transformed, more rounded phenotype also decreases kinase activity in a Y505F background and also display when assayed in the Y505F background (Figure 3c). a level of morphological transformation which is The di€erences in Lck activity in vivo do not result slightly down from the Y505F single mutant, with the from a loss of intrinsic kinase activity as all the L388D/E390M/Y505F triple mutant having the least mutants are kinase active using enolase as a substrate transformed morphology. in vitro (Figure 3d). Indeed Y394 phosphorylation is not necessary for in vitro kinase activity as the Y394Fand the Y F/Y F Lck mutants isolated from either Mutations immediately downstream of Y decrease 394 505 394 mammalian cells (Figure 3d) or insect cells (data not kinase activity but do not affect Y phosphorylation 505 shown) are still active. Several studies have suggested that the highly con- As shown in Figures 3b and 4a, substitution of the served sequence within the activation loop of SF- three amino acids downstream of Y394 (TAR) to CPK, NRTKs is important for binding to the downregula- GQQ, and NQQ does not a€ect the level of Lck Y505 tory kinase Csk via its SH2 domain (Bougeret et al., phosphorylation when expressed in ®broblasts as the 1996; Songyang et al., 1994). Current models based on Wt Lck, the Y394-downstream mutants, and Y394F Lck crystal structure analyses of SH2 domains show that all contain similar levels of PTyr. Many previous in the activating PTyr and three amino acids immediately vivo labeling and phosphate mapping studies have C-terminal to the phosphorylated tyrosine residue are shown that both Wt and Y394F Lck expressed in most important for determining SH2 binding speci®city ®broblasts only contain tyrosine phosphorylation at (Eck et al., 1993; Waksman et al., 1993). Therefore, we the C-terminal downregulatory site Y505 (Abraham and constructed mutants in the three amino acids following Veillette, 1990; Amrein and Sefton, 1988; Gervais et Y394 and analysed the mutant proteins ®rst for e€ects al., 1993; Weil and Veillette, 1994). When Y505 has on autophosphorylation and kinase activity and then been either mutated or deleted, then Y394 is the only subsequently for e€ects on downregulation by Csk. We site of tyrosine phosphorylation (Abraham and Veill- made three mutations, one which changed the wild ette, 1990; Amrein and Sefton, 1988; Gervais et al., type sequence Tyr-Thr-Ala-Arg (YTAR) to Tyr-Cys- 1993; Weil and Veillette, 1994). To eliminate the Pro-Lys (YCPK), the sequence found in the Drosophila possibility of any background autophosphorylation at Src64 homolog, and two others, Tyr-Gly-Gln-Gln Y394, the PTyr were re-examined using constructs (YGQQ) and Tyr-Asn-Gln-Gln (YNQQ), which were containing an additional Lys273 to Arg mutation chosen to minimize the chance of binding either Csk or (K273R), which has been shown to inactivate the other SH2 containing proteins. kinase domain by disrupting the phosphate-transfer Autophosphorylation of Lck occurs primarily via an reaction. As shown in Figure 4b, these mutants do not intermolecular trans-phosphorylation mechanism in decrease the levels of tyrosine phosphorylation on vivo when expressed in bacterial systems (Jullien et immunoprecipitated Lck and mutation of Y505 to Phe al., 1994). To analyse the e€ects of activation loop abolishes all phosphorylation. These results suggest mutations on autophosphorylation and subsequent that the primary site of phosphorylation in the mutants activation of the kinase, Lck Y394F and Y394-down- is Y505, and mutations at or downstream of Y394 do stream mutants were expressed in both bacterial and not a€ect downregulation by Csk phosphorylation. mammalian expression systems in combination with Y F mutations. As expected, Wt and Y F Lck 505 505 Mutations near the carboxyl end of the activation loop exhibited similar levels of kinase activity when alter substrate selection and kinase activity expressed in bacterial cells that lack Csk expression (Figure 3a). Mutation of the three amino acids From the crystal structures of SF-NRTKs, the downstream of Y394 from TAR to either CPK, GQQ, carboxy terminal end of the activation loop is or NQQ decreased the ability of Lck to phosphorylate postulated to interact with substrate peptides, stabilize endogenous bacterial substrates. The fact that similar the active conformation of the catalytic loop, and kinase activities are observed for the mutants in the function in autoinhibition by blocking the peptide presence or absence of an additional Y505F mutation binding pocket. We have made a large number of suggests that the decreases in activity are not the result mutations in this region and ®nd that the high degree of downregulation via phosphorylation of Y505. of conservation is essential for retaining normal When the same mutants are placed in ®broblasts in kinase activity. Bacterially expressed Lck variants combination with the Y505F background, the Y394- containing mutations at amino acids Phe402 (to Ala downstream mutants have decreased abilities to or Ile), Pro403 (to Ser or Thr), Lys405 (to Asp, Phe, phosphorylate endogenous proteins. Speci®cally, the or Gly), Trp406 (to Tyr), and Thr407 (to Asp or Arg)

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3965

Figure 3 Y394-downstream mutations disrupt phosphorylation of the activation loop and decrease kinase activity. (a) Expression in bacteria. Wt Lck, Lck containing activation loop mutations at or immediately downstream of Y394 (Y394F, YCPK, YGQQ, YNQQ), and corresponding Y505F double mutants were subcloned into the prokaryotic expression vector pGEX-2T. Uninduced bacterial total lysates were analysed for di€erences in PTyr content and Lck as described in Figure 2. (b) Expression in mammalian cells. Mouse ®broblasts expressing Wt Lck, Lck containing activation loop mutations at or immediately downstream to Y394 (Y394F, YCPK, YGQQ, YNQQ), and corresponding Y505F double mutants were analysed for di€erences in PTyr and Lck as described in Figure 2. (c) Phase contrast photos (106) of the ®broblast monolayers used to prepare the Western blots shown in (b). (d) Lck autokinase and in vitro kinase assays. Lck was immunoprecipitated from the same ®broblast total lysates shown in (b), divided into thirds and then analysed for either in vitro autokinase activity, in vitro kinase activity (with enolase as substrate), or levels of PTyr and Lck by Western blotting were inactive as measured by the ability to PTyr on proteins normally phosphorylated by Wt phosphorylate proteins in vivo (data not shown). Lck and not through phosphorylation of new The inactivity of several mutants was further veri®ed proteins with di€erences being much more dramatic by comparing the ability of each mutant to transpho- than those observed for mutations amino terminal to sphorylate itself using in vitro kinase assays. The Y394. The activity of the Lys401 to Ser (K401S) and K405D, K405G, T407D, and T407R mutants all Thr407 to Ser (T407S) mutants are most similar to Wt exhibited autokinase activities that were 500 ± 1200 and appear to have primarily lost phosphorylation of times less than that of Wt Lck (data not shown). a 31 ± 33 kDa doublet while substitution of Thr407 to However, the Lck mutants in this region that retained either Ala (T407A) or Met (T407M) disrupts mutually activity showed signi®cant changes in substrate exclusive protein . The Gly399 to Ala phosphorylation patterns when expressed in bacteria (G399A) mutant displays a unique phosphorylation (Figure 5a). The changes result mainly from a loss in pattern from the Lys401 and Thr407 mutants.

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3966 selection. However, no systematic mutational analysis has been reported and structural analyses have been dicult since the region is not well ordered in most crystals. In this study, we analysed the e€ects of a series of amino acid mutations across the activation loop in Lck on the regulation of kinase activity and substrate selection. Our mutagenesis results veri®ed a number of predictions from crystallographic studies for the role of the conserved activation loop in enzyme regulation. From structural analyses, regulation of kinase activity in SF-NRTKs occurs either from stabilization of the active conformation through phosphorylation of the activation loop Tyr or stabilization of the inactive conformation by SH3/linker and SH2/C-terminal Tyr interactions. Our results indicate that, in addition to these mechanisms, SF-NRTKs exist in a basal inactive state that is stabilized through intramolecular interac- tions involving the amino terminal region of the activation loop and most likely with the amino terminal lobe of the catalytic domain. We show that mutations in the conserved amino terminal region of the activation loop (L388D, L388N, and L388D/E390M) activate Lck in vivo, coincident with increased levels of PTyr on the kinase, but have little e€ect on substrate speci®city. Mutation of a nearby residue, Ile389 to Ala, also increased the basal activity of Lck expressed in ®broblasts whereas the mutation Asn392 to Thr did not

Figure 4 Y394-downstream mutants do not a€ect Y505 phos- (LE Laham, unpublished results). Disruption of this phorylation levels. (a) Kinase active (k+) and (b) kinase inactive region partially activates Lck by relieving the auto- (k7) Lck proteins were immunoprecipitated from lysates inhibitory mechanism and subsequent phosphorylation prepared from drug-resistant Balb C 3T3 ®broblast pools and then analysed for di€erences in PTyr content. After determining of either the activation loop tyrosine (Y394) or the the levels of PTyr in the Lck immunoprecipitations, blots were regulatory tail tyrosine (Y505) shifts the kinase into a stripped and reprobed using anti-Lck sera. Kinase inactive more strongly active or inactive conformation, respec- constructs were generated by making K273R mutations in the tively. These ®ndings con®rm the hypotheses suggested catalytic domain of Lck from recent crystals for Src and Hck that this region stabilizes an inactive conformation of the kinase by displacing helix aC from an enzymatically favorable The mutant proteins were then expressed in conformation (Schindler et al., 1999; Sicheri et al., ®broblasts and their activities were compared to that 1997; Xu et al., 1997, 1999). Di€erences in primary of Wt and Y505F Lck (Figure 5b). The Gly399, Lys401, sequence for this region of the activation loop may and Thr407 single mutants have PTyr levels similar to re¯ect di€erences in the level of negative regulation that of Wt Lck, indicating that they are regulated by required for di€erent classes of tyrosine kinases. Y505 phosphorylation in a manner similar to wild type. The central region of the activation loop contains When these mutants were expressed in a Y505F the structurally important Y394 residue which when background, the order of activity levels mimicked that phosphorylated induces a rearrangement in the loop observed in bacteria with the K401S mutant being most near Leu385-Arg387 that stabilizes the enzyme's active similar to Y505F followed by the T407A and G399A conformation. Amino acid sequences around the mutants. In most cases, changes in overall levels of tyrosine residue are highly conserved in all mammalian tyrosine phosphorylation in cells expressing the mutant Src homologs except the myeloid-speci®c member Fgr. proteins correlate well with the changes in cellular Although the sequence in Fgr diverges, it still is transformation in the ®broblast monolayer (Figure 5c). eciently downregulated by Csk (Ruzzene et al., 1994). However, the similarity in PTyr blotting patterns We ®nd that mutation of Y394 and the three amino between Y505F and G399A/Y505F stands in contrast to acids immediately downstream (TAR) does not a€ect the relatively large di€erence in the morphologies of C-terminal phosphorylation of Lck in vivo. However, cells expressing these proteins. This suggests that the the mutations do decrease kinases activity and revert G399A/Y505F mutant has lost the ability to phosphor- the morphological transformed phenotype of Y505F ylate a key substrate necessary for changes in mutants with the extent of reversion correlating with transformation. the levels of decreased Y394 phosphorylation. Although some of these mutants are less active than their Wt counterparts, di€erences in substrate speci®city are Discussion more easily recognized than with the mutants near the beginning of the activation loop. As all of our mutants The activation loop in tyrosine kinases is believed to be are active towards exogenous substrate when immuno- important for regulation of kinase activity via con- precipitated but are not active in bacteria or formation stabilization and determination of substrate mammalian cells, we believe that the three amino acids

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3967

Figure 5 Analysis of amino acid mutations at the C-terminus of the activation loop in Lck. (a) Expression in bacteria. Wt Lck, Lck containing activation loop mutations C-terminal to Y394 (G399A, K401S, T407A, T407M, T407S), and the corresponding Y505F double mutants were subcloned into the prokaryotic expression vector pGEX-2T. Uninduced bacterial total lysates were analysed for di€erences in PTyr content and Lck as described in Figure 2. (b) Expression in mammalian cells. Mouse ®broblasts expressing Wt Lck, Lck containing activation loop mutations C-terminal to Y394 (G399A, K401S, T407A, T407M, T407S), and the corresponding Y505F double mutants were analysed for di€erences in PTyr content by Western blotting as described in Figure 2. (c) Phase contrast photos (106) of the ®broblast monolayers used to prepare the Western blots shown in (b)

immediately downstream of Y394 are important for the the active kinase, phosphorylated Y394 interacts with trans-autophosphorylation in vivo. The defects in trans- Arg363 and Arg387 that induces a shift of the entire autophosphorylation are presumably masked when the activation loop away from the substrate binding enzyme is arti®cially concentrated upon immunopreci- pocket. This may explain, in part, the lack of activity pitation. for the carboxy terminal end mutants in vivo since they Substitution of amino acids near the carboxyl end of contain lower levels of Y394 phosphorylation. However, the activation loop has the most signi®cant e€ect on the four mutants in this region analysed in vitro kinase activity and substrate speci®city. Most muta- (K405D, K405G, T407D, and T407R) were also inactive tions in this region decrease or abolish kinase activity upon immunoprecipitation. Despite the lack of recog- when measured using in vivo and in vitro assays. The nition in earlier SF-NRTK structural studies, our lack of activity is particularly surprising since most of results suggest that the amino acids near the carboxy the residues in the loop do not appear to be making terminal end of the activation loop are essential for signi®cant contributions for stabilizing the kinase into catalytic function, possibly by stabilization of the active or inactive conformations with exceptions of the active conformation Y394 positioning or proper place- Y394 and T407 residues (Schindler et al., 1999; Sicheri et ment of peptide substrates. al., 1997; Xu et al., 1997, 1999; Yamaguchi and For the activation loop mutants near its carboxyl Hendrickson, 1996). In the inactive kinase, T407 end that retained kinase activity, di€erences in interacts with R387 near the beginning of the activation substrate selection are clearly evident. Crystal structure loop and helps to stabilize the displacement of helix analysis of the activated -peptide aC. Therefore, mutations at this site might be expected substrate complex indicated that this region of the to activate the kinase by partially relieving the loop's activation loop (L1171-M1176 in IRK; F402-T407 in Lck) autoinhibitory steric hindrance mechanism, but this forms part of two adjacent hydrophobic pockets that was not found to be the case as mutations at T407 bind to the Y+1 and Y+3 residues on the substrate decreased or abolished kinase activity. In addition, in (Hubbard, 1997). Substitution at T407 in Lck had the

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3968 most signi®cant e€ects on substrate selection when Restriction and modifying enzymes were purchased from expressed in both bacterial and mammalian systems. N.E. BioLabs (Beverly, MA, USA) except where noted. Prior to construction of mutations within the activation loop of Notably, the mutants at T407 seem to be unable to phosphorylate available substrates rather than gaining Lck, the 3' 800 bp HindIII ± EcoRI fragment of synk was + the ability to phosphorylate new substrates. This is cloned into the vector pGEM3zf (Promega, Madison, WI, USA). A new silent mutation was generated by substituting a consistent with the idea that the loop is playing a key BssHII ± S®I double-stranded oligo (5' cg cgc gag ggc gcc aag role in substrate interactions but that other nearby ttt ccc att aag tgg acg gcc cca g 3') into the backbone that regions are also important for determining substrate created a NarI restriction enzyme site at nucleotides 1198 ± selection. Interestingly, substitution of an adjacent 1203 to facilitate mutagenesis of the region. residue, W406Y, destroyed Lck's ability to phosphor- To create the speci®c activation loop mutations at, or ylate exogenous substrates when expressed in bacteria immediately C-terminal to Y394, double-stranded NheI±NarI yet the mutant retained ability to phosphorylate itself oligos (Y394F:5'cta gca cgc ctc att gag gac aac gag ttc aca gcg (Mukhopadhyay, unpublished work). Several addi- cgc gag gg3'; YCPK: 5'cta gca cgc ctc att gag gac aac gag tac tgc cct aag gag gg3';YGQQ: 5'cta gca cgc ctc att gag gac aac tional mutations further away from the T407 residue, G A and K S, have altered substrate speci®city that gag tac gga cag cag gag gg3';YNQQ: 5'cta gca cgc ctc att gag 399 401 gac aac gag tac aac cag cag gag gg3') were subcloned into the is more pronounced with expression in bacteria than NheI±NarI digested synk-pGEM3zf+ backbone. To generate mammalian cells. These mutants also display di€er- the activation loop mutations N-terminal to Y , double- ences in ®broblast monolayer morphologies from that 394 stranded NheI±BssHII oligos (L388D: 5'cta gca cgc gat att of Y505F to suggest the subtle loss of key substrate gag gac aac gag tac aca g3'; L388N: 5'cta gca cgc aat att gag phosphorylations in mammalian cells. We believe that gac aac gag tac aca g3'; E390M: 5'cta gca cgc ctc att atg gac this region, in coordination with the region near T407, aac gag tac aca g3'; E390T: 5'cta gca cgc ctc att acc gac aac may be making speci®city contacts with substrate gag tac aca g3'; L388D/E390M: 5'cta gca cgc gat att atg gac residues, although we cannot rule out that the changes aac gag tac aca g3') were subcloned into the NheI±BssHII are indirect due to a local disruption of activation loop digested synk-pGEM3zf+ backbone. To generate the point mutations at K , double-stranded NarI±S®I oligos structure. 405 (K405D: 5'c gcc aag ttt ccc att gac tgg acg gcc cca g3'; Surprisingly, the conservative substitutions at F402 in K405F: 5'c gcc aag ttt ccc att ttc tgg acg gcc cca g3'; K405G: Lck (analogous to the insulin receptor residue 5'c gcc aag ttt ccc att ggc tgg acg gcc cca g3') were subcloned hypothesized to interact with the substrate Y+1 into the NarI±S®I digested synk-pGEM3zf+ backbone. Full- residue) destroyed activity. The inactivity of the F402, length Lck cDNA was re-generated by ligation of the NarI+ P403, and K405 mutants suggests that the high degree of Wt and mutated 800 bp HindIII ± EcoRI cDNA fragments sequence conservation in this region is essential to back with the synk-pGEX-2T HindIII ± EcoRI plasmid back- retain enzyme activity and probably not important for bone. To generate the remaining amino acid mutations determining minor di€erences in substrate speci®city. downstream of Y394, double-stranded HincII ± S®I fragments Analysis of the active insulin receptor structure were PCR generated, digested with HincII and S®I, and then suggests that conformation of this region is essential subcloned into the HincII and S®I digested synk-pGEX2T backbone. The upstream PCR oligo at the HincII site was for determining speci®city where the conserved Pro 5'tgc tct aga cat ggc agc cca aat tgc3' while the downstream (P403 in Lck) orients the phenolic ring of the substrate PCR oligos containing the point mutations are summarized Tyr residue. In our mutants, substitution of any as follows: G399A: 5'cgg ggt acc gta gtt aat ggc ctc tgg ggc cgt residues in the immediate region of the conserved Pro cca ctt aat ggg aaa ctt ggc ggc ctc3'; K401S: 5'cgg ggt acc gta may still allow substrates to interact with the kinase gtt aat ggc ctc tgg ggc cgt cca ctt aat ggg aaa gct ggc tcc3'; but inactivate the enzyme due to misalignment of the F402I: 5'cgg ggt acc gta gtt aat ggc ctc tgg ggc cgt cca ctt aat ggg gat ctt gcc3'; F A: 5'cgg ggt acc gta gtt aat ggc ctc tgg proline relative to Y394. 402 Taken as a whole our results point to the high level ggc cgt cca ctt aat ggg tgc ctt ggc3'; P403S: 5'cgg ggt acc gta of importance of virtually every residue in the gtt aat ggc ctc tgg ggc cgt cca ctt aat tga aaa ctt3'; P403T: activation loop. This highly conserved element not 5'cgg ggt acc gta gtt aat ggc ctc tgg ggc cgt cca ctt aat tgt aaa ctt3'; W Y: 5'cgg ggt acc gta gtt aat ggc ctc tgg ggc cgt gta only helps to hold the basal enzyme activity to a 406 ctt aat 3'; T407A: 5'cgg ggt acc gta gtt aat ggc ctc tgg ggc tgc minimum but also regulates the activity and substrate cca ctt aat3'; T D:5'cgg ggt acc gta gtt aat ggc ctc tgg ggc speci®city of the activated enzyme. While none of the 407 gtc cca ctt aat3'; T407M: 5'cgg ggt acc gta gtt ggc ctc tgg ggc mutations that we made at the C-terminus of the loop cat cca ctt aat3'; T407R: 5'cgg ggt acc gta gtt aat ggc ctc tgg which changed substrate speci®city retained full ggc cct cca ctt aat3'; T407S: 5'cgg ggt acc gta gtt aat aac ctc oncogenic potential, it is quite possible that mutations tgg ggc tga cca ctt aat3'. To generate the activating C- can be made in this region which will dramatically terminal mutations at Y505 in Lck, a 3' 287 bp KpnI±EcoRI change the oncogenic nature of Lck by altering fragment of synk/Y505F was substituted for the fragment in substrate speci®city while retaining catalytic activity. the wildtype (Wt) and the activation loop mutant constructs. To verify that all clones obtained the correct mutations, one strand of the cDNA fragments was sequenced using the chain termination method with Sequenase 2.0 (Amersham, Cleve- land, OH, USA). All oligos were generated either by the Materials and methods DFCI core facility or purchased from Integrated DNA Technologies (Coralville, IA, USA). For construction of Construction of bacterial and retroviral vectors Y505F, YCPK, YGQQ, YNQQ and Y394F mutants in The synthetic human Lck cDNA (synk), cloned into mammalian expression vectors, Wt Lck and mutant cDNA's prokaryotic expression vector pGEX-2T (Pharmacia Biotech were subcloned into pLNCX (Clontech, Palo Alto, CA, Inc., Piscataway, NJ, USA), contains silent mutations that USA). The vector was ®rst modi®ed by adding a BglII insert unique restriction enzyme sites every 20 ± 40 nucleotides restriction enzyme site into the multiple cloning site at the and was used for all Lck plasmid constructions (Carrera et HindIII site. Wt and mutant Lck cDNA fragments were al., 1995; AC Carrera and TM Roberts, unpublished results). generated by digestion of the pGEX-2T clones with NcoI and

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3969 EcoRI and the fragments were ligated after blunt-end acids 39 ± 64 of human Lck; a generous gift from Dr Chris generation to phosphorylated double-stranded BclI oligos Rudd, Dana-Farber Cancer Institute, Boston, MA, USA) or (5'gtg atc ac3') to generate a consensus 5' Kozak sequence. a rabbit polyclonal anti-C-terminal Lck peptide sera Fragments generated by BclI digestion were subcloned into (generated against amino acids 476 ± 505 of human Lck; the newly generated BglII site on pLNCX. To construct the Santa Cruz Biotechnology, Santa Cruz, CA, USA). Blots remaining constructs, Lck HindIII ± EcoRI fragments con- were stripped in (62 mM Tris-HCl pH 6.8, 2% SDS, 100 mM taining the activation loop mutations were subcloned into the b-mercaptoethanol) at 508C for 30 min and then reprobed Wt Lck/pLNCX or Y505F/pLNCX backbones. Kinase after blocking in (10 mM Tris-HCl pH 8.0, 120 mM NaCl, inactive Lck constructs were generated by creating K273R 0.5% Tween-20, 5% non-fat dry milk) for 60 min at room mutations in the pLNCX expression vectors by substituting temperature or overnight at 48C. the 739 bp BglII ± SalI fragments from K273R synk (Carrera Lck was immunoprecipitated from lysates by rocking 90' at et al., 1993) into each of the Wt, Y394F, YCPK, YNQQ, and 48C using 1 : 1 (v/v) mixtures of rabbit polyclonal sera Y505F pLNCX mutants. generated against the N-terminal Lck peptide (obtained from Dr Chris Rudd) and the C-terminal Lck peptide (Santa Cruz Biotechnology) (5 ml antibody mixture/mg lysate) after lysates Cell culture, retroviral infections, and lysate preparations were normalized for total protein content. Complexes were Balb C 3T3 ®broblasts were grown in Dulbecco's modi®ed collected by additional incubation for 30 ± 45 min with Eagle's media supplemented with 10% (v/v) calf serum, hydrated protein A sepharose (Pharmacia) and washed three penicillin, and streptomycin sulfate (Gibco-BRL, Gaithers- times with NP-40 lysis bu€er containing 1 mM NaVO4. burg, MD, USA). All cells were grown at 378Cin8%CO2. Precipitations were then either resuspended in gel sample Supernatants from the retroviral packaging cell line bu€er for SDS ± PAGE analysis or divided into thirds for pBOSC23 (Pear et al., 1993) were harvested 48 ± 72 h after analysis by in vitro kinase assays and Western blotting. To CaPO4-mediated transfection of 10 mg Wt or mutant Lck- determine Lck levels, immunoprecipitations were Western pLNCX plasmids. Balb C cells (0.56106) were infected with blotted using either mouse monoclonal anti-Lck N-terminal 1/10 ± 1/200th volume of the retroviral supernatants and peptide sera (obtained from either Transduction Labs stable pools were generated from geneticin (G418)-resistant (Lexington, KY, USA) (generated against amino acids 1 ± colonies (Gibco-BRL). Stable drug-resistant pools were lysed 191 of human Lck) or PharMingen (San Diego, CA, USA) in ice-cold NP-40 bu€er (20 mM Tris-HCl pH 8.0, 137 mM (generated against amino acids 70 ± 94 of human Lck). NaCl, 10% glycerol, 1% Nonidet P-40) supplemented with Di€erences in PTyr content for both were analysed using protease and phosphatase inhibitors (1 mM phenylmethylsul- anti-PTyr (4G10) (Upstate Biotechnology Incorporated). All fonyl ¯uoride, 1 mg/ml pepstatin, 1 mg/ml leupeptin, 0.1 TIU/ blots were visualized with ECL substrate (Amersham, ml aprotinin, 1 mM NaVO4,25mM b-glycerol phosphate). Arlington Heights, IL, USA) and exposed to ®lm after Supernatants were collected after centrifugation (14 000 g, incubation with either anti-rabbit IgG or anti-mouse IgG 5 min, 48C). Lysates were normalized for total protein conjugated horseradish peroxidase (Amersham). content using the BioRad protein assay (BioRad, Richmond, For analysis of Lck autokinase activity, immunoprecipi- CA, USA) and then analysed for Lck and PTyr content by tates were washed once more with kinase bu€er (50 mM Tris- SDS ± PAGE and immunoblotted as indicated below. To HCl pH 7.4, 10 mM MnCl2), resuspended in 50 ml kinase generate Lck bacterial lysates, overnight JM109 bacterial bu€er supplemented with 10 mCi of [g-32P]ATP (3000 Ci/mM) cultures were diluted 1 : 10 in LB containing ampicillin (Dupont-NEN, Boston, MA, USA), and incubated for 15 ± (50 mg/ml) and grown at 378C for four additional hours. 20 min at room temperature with occasional mixing. Uninduced bacterial pellets were lysed on ice in 1/10th Reactions were washed once with 1 ml kinase bu€er and volume (20 mM Tris-HCl pH 8.0, 0.5% Nonidet P-40, resuspended in 16gel sample bu€er prior to SDS ± PAGE. 100 mM NaCl, 1 mM EDTA) supplemented with protease For analysis of in vitro kinase activity, immunoprecipitates and phosphatase inhibitors (1 mM NaVO4,1mM phenyl- were washed once with kinase bu€er, resuspended in 40 mlof methylsulfonyl ¯uoride, 25 mM b-glycerol phosphate, 1 mg/ml kinase bu€er supplemented with (10 mM ATP, 6 mg acid pepstatin, 1 mg/ml leupeptin, 0.1 TIU/ml aprotinin) followed treated enolase (Sigma), 5 mCi of [g-32P]ATP), and incubated by bath sonication (three 10 s pulses). All protease and at room temperature for 15 ± 20 min with occasional shaking phosphatase inhibitors were obtained from Sigma Chemical prior to dilution with 36 gel sample bu€er and analysis by Corporation (St. Louis, MO, USA). Supernatants were SDS ± PAGE. All gels were dried and then exposed to cleared by centrifugation (14 000 g, 5 min). Lysates were autoradiography ®lm at room temperature for 5 ± 60 min in diluted with SDS ± PAGE sample bu€er and then analysed by the absence of an intensifying screen. standard SDS ± PAGE followed by transfer to Immobilin-P membranes (Millipore, Bedford, MA, USA) and immuno- blotted as indicated below. Acknowledgments We are grateful to Chris Rudd for generously providing the Immunoblot analysis and immunoprecipitation polyclonal anti-Lck sera. We would also like to thank Toni Equivalent amounts of total lysate protein obtained from Jun, Fred King, Paul Rose, Kathryn Campbell and Mike either uninduced bacterial cultures (20 mg) or Balb C 3T3 Eck for their helpful discussions during the completion of pools (75 mg) were analysed by SDS ± PAGE and transferred this work and Joanne Chan, Ole Gjoerup, Joan Brugge, onto Immobilin-P membrane. To determine the levels of Lck and Sugata Sarkar for their critical reading of the manu- protein, blots were probed with either a rabbit polyclonal script. This work was supported by a National Institutes of anti-N-terminal Lck peptide sera (generated against amino Health Grant CA43803-10 (TM Roberts).

References

Abraham N and Veillette A. (1990). Mol. Cell. Biol., 10, Bougeret C, Delaunay T, Romero F, Jullien P, Sabe H, 5197 ± 5206. Hanafusa H, Benarous R and Fischer S. (1996). J. Biol. Amrein KE and Sefton BM. (1988). Proc. Natl. Acad. Sci. Chem., 271, 7465 ± 7472. USA, 85, 4247 ± 4251.

Oncogene Effects of activation loop mutations on Lck regulation LE Laham et al 3970 Carrera AC, Alexandrov K and Roberts TM. (1993). Proc. Osusky M, Taylor SJ and Shalloway D. (1995). J. Biol. Natl. Acad. Sci. USA, 90, 442 ± 446. Chem., 270, 25729 ± 25732. Carrera AC, Borlado LR, Gonzalez-Garcia A, Roberts TM Pear WS, Nolan GP, Scott ML and Baltimore D. (1993). and Martinez AC. (1995). Oncogene, 10, 2379 ± 2386. Proc. Nalt. Acad. Sci. USA, 90, 8392 ± 8396. Cartwright CA, Eckhart W, Simon S and Kaplan PL. (1987). Piao X, Paulson R, Van der Geer P, Pawson T and Bernstein Cell, 49, 83 ± 91. A. (1996). Proc. Natl. Acad. Sci. USA, 93, 14665 ± 14669. Cooper JA and MacAuley A. (1988). Proc. Natl. Acad. Sci. Piwnica-WormsH,SaundersKB,RobertsTM,SmithAE USA, 85, 4232 ± 4236. and Cheng SH. (1987). Cell, 49, 75 ± 82. Eck MJ, Shoelson SE and Harrison S. (1993). Nature, 362, Ruzzene M, James P, Brunati AM, Donella-Deana A and 87 ± 91. Pinna LA. (1994). J. Biol. Chem., 269, 15885 ± 15891. Gervais FG, Chow LML, Lee JM, Branton PE and Veillette Schindler T, Sicheri F, Pico A, Gazit A, Levitzki A and A. (1993). Mol. Cell. Biol., 13, 7112 ± 7121. Kuriyan J. (1999). Mol. Cell, 3, 639 ± 648. Hofstra RMW, Landsvater RM, Cecchini I, Stulp RP, Sicheri F, Moare® I and Kuriyan J. (1997). Nature, 385, Stelwagen T, Luo Y, Paasini B, Hoppener JWM, van 602 ± 607. Amstel HKP, Romeo G, Lips CJM and Buys CHCM. Sideras P and Smith CI. (1995). Adv. Immun., 59, 135 ± 223. (1994). Nature, 367, 375 ± 376. Songyang Z, Shoelson SE, McGlade J, Olivier P, Pawson T, Hubbard SR. (1997). EMBO J., 16, 5572 ± 5581. Bustelo XR, Barbacid M, Sabe H, Hanafusa H, Yi T, Ren Hubbard SR, Wei L, Ellis L and Hendrickson WA. (1994). R, Baltimore D, Ratnofsky S, Feldman RA and Cantley Nature, 372, 746 ± 754. LC. (1994). Mol. Cell. Biol., 14, 2777 ± 2785. Johnson LN, Noble MEM and Owen DJ. (1996). Cell, 85, Waksman G, Shoelson SE, Pant N, Cowburn D and Kuriyan 149 ± 158. J. (1993). Cell, 72, 779 ± 790. Jullien P, Bougeret C, Camoin L, Bodeus M, Durand H, Weil R and Veillette A. (1994). J. Biol. Chem., 269, 22830 ± Disanto JP, Fischer S and Benarous R. (1994). Eur. J. 22838. Biochem., 224, 589 ± 596. Xu W, Doshi A, Lei M, Eck MJ and Harrison SC. (1999). Kmiecik TE and Shalloway D. (1987). Cell, 49, 65 ± 73. Mol. Cell, 3, 629 ± 638. Madhusudan, Trafny EA, Xuong N-H, Adams JA, Eyeck Xu W, Harrison SC and Eck M. (1997). Nature, 385, 595 ± LFT, Taylor SS and Sowadski JM. (1994). Protein Sci., 3, 602. 176 ± 187. Yamaguchi H and Hendrickson W. (1996). Nature, 384, Marth JD, Cooper JA, King CS, Ziegler SF, Tinker DA, 484 ± 489. Overell RW, Krebs EG and Perlmutter RM. (1988). Mol. ZhengJ,KnightonDR,EyeckLFT,KarlssonR,XuongN- Cell. Biol., 8, 540 ± 550. H, Taylor SS and Sowadski JM. (1993). Biochemistry, 32, Mohammadi M, Schlessinger J and Hubbard SR. (1996). 2154 ± 2161. Cell, 86, 577 ± 587.

Oncogene