View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

Research Article 651

Structural analysis of the -specific Lck in complex with non-selective and Src family selective kinase inhibitors Xiaotian Zhu*, Joseph L Kim, John R Newcomb, Paul E Rose, David R Stover, Leticia M Toledo, Huilin Zhao and Kurt A Morgenstern*

Background: The lymphocyte-specific kinase Lck is a member of the Src family Address: Kinetix Pharmaceuticals, Inc., 200 Boston of non-receptor . Lck catalyzes the initial of Avenue, Medford, MA 02155, USA. T- receptor components that is necessary for signal transduction and T-cell *Corresponding authors. activation. On the basis of both biochemical and genetic studies, Lck is E-mail: [email protected] considered an attractive cell-specific target for the design of novel T-cell [email protected] immunosuppressants. To date, the lack of detailed structural information on the All other authors are listed alphabetically. mode of inhibitor binding to Lck has limited the discovery of novel Lck inhibitors. Key words: AMP-PNP, kinase, Lck, PP2, Src family, Results: We report here the high-resolution crystal structures of an activated staurosporine, X-ray structure Lck kinase domain in complex with three structurally distinct ATP-competitive 13 April 1999 inhibitors: AMP-PNP (a non-selective, non-hydrolyzable ATP analog); Received: Revisions requested: 22 April 1999 staurosporine (a potent but non-selective kinase inhibitor); and PP2 Revisions received: 30 April 1999 (a potent Src family selective protein inhibitor). Comparison of Accepted: 4 May 1999 these structures reveals subtle but important structural changes at the ATP-. Furthermore, PP2 is found to access a deep, hydrophobic Published: 1 June 1999 pocket near the ATP-binding cleft of the ; this binding pocket is not Structure June 1999, 7:651–661 occupied by either AMP-PNP or staurosporine. http://biomednet.com/elecref/0969212600700651

Conclusions: The potency of staurosporine against Lck derives in part from an © Elsevier Science Ltd ISSN 0969-2126 induced movement of the glycine-rich loop of the enzyme upon binding of this ligand, which maximizes the van der Waals interactions present in the complex. In contrast, PP2 binds tightly and selectively to Lck and other Src family kinases by making additional contacts in a deep, hydrophobic pocket adjacent to the ATP-binding site; the amino acid composition of this pocket is unique to Src family kinases. The structures of these Lck complexes offer useful structural insights as they demonstrate that kinase selectivity can be achieved with small-molecule inhibitors that exploit subtle topological differences among protein kinases.

Introduction tyrosine residues in the ITAMs by Lck is required for the T-cell activation is a complex process that results from the binding of tandem ZAP-70 Src homology 2 (SH2) domains integrated activation of multiple signal transduction path- [14–16]. Colocalization of ZAP-70 and Lck to the TCR ways [1–3]. One of the earliest T-cell signaling events ζ-subunit–CD4/8 complex facilitates the Lck-mediated observed upon interaction of the T-cell receptor (TCR) activation of ZAP-70 and subsequent ZAP-70 autophos- with its ligand, is the CD4/CD8-dependent activation of phorylation [17–21]. Activated Lck and ZAP-70 perpetu- lymphocyte-specific kinase (Lck), a member of the non- ate the TCR signaling cascade by providing additional receptor Src family of tyrosine kinases [4–8]. Lck phospho- docking sites for other kinases that contain SH2 domains rylates and activates a number of substrates necessary for (including Fyn, Syk and Itk), adaptor (including TCR signaling [9]. Perhaps the best understood activity of SLP-76, SHC, LAT, FyB and Grap), and transducing ele- Lck is the phosphorylation of immunoreceptor tyrosine- ments (including PLCγ, PI3-kinase and Rac/Rho) [2,3,22]. based activation motifs (ITAMs) in the TCR ζ subunit Biochemical information is then transmitted down multi- [4,6,9]. The extent of ζ chain ITAM phosphorylation dic- ple signaling pathways, including the Ras/mitogen-acti- tates the threshold for ligand-mediated TCR signaling and vated pathway, the phosphatidylinositol T-cell activation [10,11]. Phosphorylated ITAMs serve as pathway, and the Rho/Rac pathway [2]. Among other high-affinity docking sites for the recruitment of additional effects, TCR signaling up-regulates the transcription and signaling factors, in particular the Syk family tyrosine translation of interleukin-2 (IL-2) and IL-2 receptors, kinase ZAP-70 [12,13]. The dual phosphorylation of which are prerequisites for T-cell proliferation. 652 Structure 1999, Vol 7 No 6

Genetic studies have demonstrated that Lck expression is previously reported further establishes the structural basis restricted to . Loss of Lck expression in human for the high potency and poor selectivity of this inhibitor. Jurkat T cells results in a loss of signaling in response to TCR ligation [23,24]. In addition, inactivation of the lck The role of Src family members Lck and Fyn in TCR acti- gene, or expression of dominant negative transgenes in vation has been studied with two related Src kinase mice, results in early arrest of thymocyte maturation inhibitors, PP1 and PP2 [46]. PP1 and PP2 are claimed to [25–27]. These and other biochemical studies have impli- selectively inhibit Lck and c-Src in vitro at concentrations cated Lck as an essential early mediator of the TCR signal- much lower than those required to inhibit ZAP-70, Janus ing pathway. Lck therefore represents an attractive target kinase 2 (JAK2), epidermal (EGFR) for therapeutic intervention in T-cell-mediated disorders, kinase and PKA [46]. These compounds also inhibit anti- such as autoimmune diseases and transplant rejection. CD3-induced protein tyrosine phosphorylation and subse- quent IL-2 gene activation in T lymphocytes [46]. Thus, it Lck is a modular protein consisting of a C-terminal catalytic appears that PP1 and PP2 dissect a component of TCR sig- domain, a single SH2 and SH3 domain, and a unique N-ter- naling not distinguished by other immunosuppressive minal region. The N-terminal region is involved in anchor- drugs, such as cyclosporin and FK-506. We present here the ing Lck to CD4/8 through Zn2+ coordination with conserved structural basis for the potency and selectivity of these com- cysteine residues present in both proteins [28,29]. The pounds with the crystal structure of PP2 bound to Lck. The activity of Lck is regulated by the autophosphorylation of structure will be a useful tool in the design of specific Lck Tyr394 located in the catalytic domain activation loop [30] inhibitors and should allow for the tailoring of the pharma- and by the phosphorylation of Tyr505 by C-terminal Src cokinetic properties of inhibitors such as PP1 and PP2. kinase (CSK) [31–33]. Further understanding of the regula- tion of Lck has been provided by the crystal structures of Results and discussion two other Src family protein kinases, c-Src and Hck [34–36]. AMP-PNP binding to the Lck catalytic domain From these structures it can be inferred that the SH2 and To provide a structural basis for understanding the interac- SH3 domains function in part to negatively regulate Lck tions of ATP-competitive inhibitors with Lck, the Lck cat- activity by forming intramolecular contacts that stabilize the alytic domain was cocrystallized with the non-hydrolyzable catalytic domain in an inactive conformation [37]. The SH2 ATP analog adenyl imidodiphosphate (AMP-PNP). Consis- domain binds to phosphorylated Tyr505 and the SH3 tent with structures of other protein kinases in complex domain associates with a proline-containing motif in a hinge with ATP analogs [47–49], AMP-PNP binds in the cleft region connecting the SH2 and catalytic domains [34–36]. between the N- and C-terminal lobes of Lck, with a pair of Release of these intramolecular regulatory constraints by conserved hydrogen bonds formed between the adenine the dephosphorylation of Tyr505 [38] and/or the presence base and the backbone of the kinase linker region of competing SH3/SH2 ligands [39] results in the autophos- (Figures 1a and 2a). The γ-phosphate of AMP-PNP is disor- phorylation of Tyr394 in the activation loop and a catalyti- dered in the binary complex, perhaps due to the absence of cally active kinase [19]. A structural basis for Lck activation a substrate peptide or divalent cations. In the ternary com- has been previously elucidated from the crystal structure of plexes of PKA with bound ATP and a substrate peptide an autophosphorylated Lck catalytic domain [40]. inhibitor [47,48] and tyrosine kinase (IRK) with AMP-PNP and a substrate peptide [49], the bound Protein kinases have been implicated as potential targets for peptides appear to help anchor the γ-phosphate of ATP to a variety of clinical applications. Efforts to understand the the enzyme. Only small conformational changes are molecular constraints necessary to achieve inhibitor potency observed in the Lck–AMP-PNP complex relative to the and selectivity have been aided by the availability of an previously reported apo Lck structure [40]. However, increasing number of crystal structures of different protein Ser323 undergoes a conformational change in the ribose- kinases in complex with ATP-competitive inhibitors. One binding pocket of Lck that appears to be important for such inhibitor is staurosporine, an alkaloid that has been AMP-PNP binding. In the apo Lck structure, Ser323 previously shown to inhibit a broad range of tyrosine and adopts two partially occupied conformations: one conforma- serine/threonine kinases with nanomolar potency [41]. tion results in a hydrogen bond between Ser323 and Asp326 Crystal structures of staurosporine bound to the serine/thre- and the other results in Ser323 hydrogen bonding to the onine kinases protein kinase A (PKA) and cyclin-depen- backbone carbonyl of Asp368. In both conformations, the dent kinase 2 (CDK2) elucidated the binding mode of this Ser323 Oγ points away from the ATP-binding cleft and inhibitor [42,43] (reviewed in [44]). A similar binding mode faces the C-terminal lobe. In the Lck–AMP-PNP structure, has been reported in a recently solved structure of the tyro- the sidechain of Ser323 is rotated more than 100° about χ1 sine kinase CSK in complex with staurosporine [45]. We and forms a hydrogen bond with the ribose oxygen of AMP- present here crystal structures of Lck in complex with stau- PNP (Oγ–O2′ distance 2.7 Å). Ser323 of Lck is conserved rosporine obtained from both soaking and cocrystallization among all known Src family tyrosine kinases with the experiments. Comparison of these two complexes and those exception of Blk, which contains a cysteine at this position. Research Article Lck in complex with non-selective and Src family selective kinase inhibitors Zhu et al. 653

Figure 1

Electron-density maps of ligands bound to Lck. The 2Fo–Fc electron- Hydrogen bonds formed between the ligands and the kinase linker density maps are contoured at 1σ. The linker region between the N- region are represented by magenta dashed lines. Atoms are colored by and C-terminal lobes of the Lck kinase domain is shown to the left of type: carbon, green; oxygen, red; nitrogen, blue; sulfur, yellow. the bound ligands: (a) AMP-PNP; (b) staurosporine; (c) PP2.

Staurosporine binding to Lck contribute a total of 78 van der Waals contacts to the bound Structures of staurosporine bound to Lck were determined inhibitor (Figure 3a). The majority of these contacts are to both from apo Lck crystals soaked with inhibitor and from the fused carbazole moiety of staurosporine, which spans a cocrystals derived from a preformed Lck–staurosporine plane of approximately 15 × 11 Å2. In contrast, the glyco- complex (Figures 3a and b). Cocrystallization of this sidic group of staurosporine spans only 6 Å in a direction Lck–staurosporine complex produced a new crystal form, perpendicular to the plane of the carbazole ring system. which contains different crystal-packing interactions than Approximately half of the van der Waals interactions result observed in the apo crystal form. In both Lck–stau- from a large movement of the glycine-rich loop of Lck, rosporine complexes the inhibitor occupies the which is induced by staurosporine binding. This movement ATP-binding site and forms three hydrogen bonds with is most prominent for the β1 strand of the glycine-rich loop the enzyme. The NH and keto oxygen of the lactam ring and is best described as a rotation of this loop about the β2 of staurosporine make a pair of hydrogen bonds with the strand coupled with a small translation towards the C-termi- carbonyl oxygen of Glu317 and the backbone NH of nal domain (Figure 4a). The backbone of strand β1 moves Met319, similar to those formed by the adenine ring of towards the bound ligand by approximately 1.7 Å in the ATP (Figures 1b and 2b). The third hydrogen bond, soaked Lck–staurosporine complex. In the cocrystallized which occurs in the ribose-binding pocket of Lck, appears Lck–staurosporine complex, movement of β1 is even more to be different in the two complexes. In the Lck–stau- pronounced bringing this strand 2.4 Å nearer to the C-ter- rosporine complex derived from soaking, the methylamino minal lobe. In contrast to the apo crystals used for soaking nitrogen of the glycosidic ring participates in a hydrogen experiments, cocrystals of the Lck–staurosporine complex bond with Ser323 (N–Oγ distance 2.9 Å). This interaction do not contain lattice contacts in the glycine-rich loop. This is similar to the hydrogen bond observed between Ser323 is likely to explain the relative difference in the position of and the ribose 2′-hydroxyl group in the Lck–AMP-PNP this loop in the Lck–staurosporine structures. These two complex. In contrast, the methylamino nitrogen of stau- structures may well represent snapshots of the association rosporine in the cocrystallized complex forms a hydrogen between Lck and staurosporine in solution. bond with the carbonyl oxygen of Ala368 from the catalytic loop of Lck (N–O distance 3.2 Å). Although the Ser323 Oγ Superimposition of the two Lck–staurosporine complexes is within hydrogen-bonding distance of the methylamino using the C-terminal domain results in a root mean square nitrogen of staurosporine in this complex as well (2.7 Å), (rms) deviation of 0.28 Å between 172 common Cα atoms. the geometry is not optimal for a hydrogen bond. Comparison of the superimposed structures shows that the largest positional differences occur in the highly flexible Staurosporine also makes extensive van der Waals contacts glycine-rich loop of the enzyme (Figure 4a). Interestingly, with Lck. Seven residues from the N-terminal lobe although the bound staurosporine molecules superimpose (Leu251, Gly252, Val259, Ala271, Lys273, Thr316 and well in the two structures, the plane of the carbazole moiety Tyr318) and six residues from the C-terminal lobe (Met319, is rotated slightly towards the N-terminal lobe of Lck in the Gly322, Ser323, Ala368, Leu371 and Asp382) of Lck complex derived from soaking. The net result of this 654 Structure 1999, Vol 7 No 6

Figure 2 rotation is a 0.8 Å shift of the glycon moiety of staurosporine towards the glycine-rich loop of the N-terminal lobe. Exam- ination of the two enzyme–inhibitor structures provides a (a) Ala271 Thr316 ready explanation for this shift. The closest contact between staurosporine and the glycine-rich loop in the two structures Glu317 is a CH–O interaction between the glycosidic oxygen of Tyr318 Val259 staurosporine and the Cα atom of Gly252. This CH–O dis- NH2 tance is 3.5 Å in the cocrystallized complex and 3.6 Å in the Met319 N N Leu251 complex derived from soaking. Because the glycine-rich N N loop has not rotated as far towards the C-terminal domain in O O O Leu371 the soaked Lck–staurosporine complex, the bound ligand HO OPO PNH must rotate slightly in order to maintain this interaction. As O O HO a result of this rotation, the glycon methylamino group of staurosporine is shifted away from the carbonyl oxygen of Ala368 in this complex, resulting in an NH–O distance of 4.1 Å as compared to 3.2 Å in the cocrystallized complex. Ser323 Crystal structures of the protein kinases CDK2, PKA and (b) Ala271 Thr316 CSK in complex with staurosporine have also been reported [42,43,45]. The same hydrogen-bonding pattern observed between the lactam ring of staurosporine and the Glu317 Tyr318 linker region of Lck is observed in each of these crystal structures. Inspection of the PKA–staurosporine and N Lys273 CDK2–staurosporine complexes (CSK coordinates not Met319 O available) reveals that the CH–O interaction described above for the two Lck–staurosporine structures is present Val259 Leu371 as the closest contact between staurosporine and the N N O Leu251 glycine-rich loop in these structures as well. Not only are

OCH3 Gly382 the CH–O distances constant in the four complexes + (3.5 Å), but the geometry of the interaction is similar. This H3CH2N Asp382 type of CH–O interaction is well documented in small- Asp326 molecule crystal structures [50] and has also been observed Ala368 in other biomolecular complexes [51]. Interestingly, both Ser323 PKA and CDK2 undergo conformational changes in the glycine-rich loop upon staurosporine binding and, (c) Ala271 although this loop contains additional conformational dif- Thr316 Ile314 ferences in the CDK2 complex (Figure 4b), the interaction between the glycine Cα atom and the glycosidic oxygen is Glu317 Cl Lys273 maintained. This emphasizes the importance of this inter- Tyr318 action in complexes between staurosporine and NH2 Glu288 serine/threonine and tyrosine kinases. We propose that Met319 N N this interaction is critical for the potency of staurosporine N N Val259 binding to the ATP-binding site of these kinases. Leu371 Leu251 In both the Lck and CSK complexes a single hydrogen bond is formed between staurosporine and the ribose- binding pocket [45]. In comparison, staurosporine has been observed to form two hydrogen bonds with this Ser323 pocket in CDK2 and PKA [42,43]. In these two complexes Structure rotation about the C–N bond of the methylamino sub- stituent allows the amine nitrogen to hydrogen bond to Schematic representation of the hydrogen-bond interactions and van both a carbonyl oxygen from the catalytic loop and a der Waals contacts between Lck and bound ligands: (a) AMP-PNP; sidechain from the ribose-binding pocket. Of the two (b) staurosporine; (c) PP2. Hydrogen bonds are represented with dashed lines. The residues of Lck in contact with the bound ligand methylamino hydrogen-bonding interactions observed in are shown. our Lck–staurosporine complexes, the contact with the carbonyl oxygen of Ala368 most closely resembles the Research Article Lck in complex with non-selective and Src family selective kinase inhibitors Zhu et al. 655

Figure 3

Interactions of (a,b) staurosporine and (c,d) PP2 with Lck at the ATP-binding cleft. The residues of Lck in contact with the bound ligand are shown in yellow in (a) and (c). Parts (b) and (d) show the surface curvature of Lck when bound to staurosporine and PP2, respectively. The most convex parts of the molecular surface are colored green, the most concave regions gray and the most planar parts are in white.

interactions observed in the PKA and CDK2 complexes. PP2 binding to Lck The distance between this carbonyl oxygen and the Cα PP2 has been reported to be a potent Src family selective atom of Gly252 is 10.3 Å and 9.6 Å, respectively, for the tyrosine kinase inhibitor [46]. This compound inhibits Lck soaked and cocrystallized Lck–staurosporine complexes. with an IC50 of 4 nM and Fyn with an IC50 of 5 nM. PP2 is µ The equivalent distances in the PKA and CDK2 struc- slightly less potent against EGFR (IC50 = 0.45 M) and µ tures are 9.0 Å and 8.8 Å, respectively. This indicates that inactive against ZAP-70 (IC50 = 100 M) [46]. PP1, an the Lck–staurosporine cocrystal structure is more likely to analog of PP2, shows approximately the same inhibitory represent a physiologically relevant structure, with a activity against Lck [46]. To determine the structural basis slightly more open conformation than those of PKA and for this selectivity, PP2 was cocrystallized with the kinase CDK2 in complex with staurosporine. domain of Lck. The structure of this complex reveals that

Figure 4

Structural comparison of protein kinases bound to staurosporine. (a) Superposition of the structures of apo Lck (orange), the Lck–staurosporine complex derived from the soaking experiment (green) and the Lck–staurosporine complex obtained from cocrystallization (purple). The superposition is based on the kinase domain of Lck. The ligand from the soaking experiment is shown in cyan and staurosporine from the cocrystallization experiment is shown in purple. (b) Superposition of Lck (green), CDK2 (cyan) and PKA (yellow) in complex with staurosporine (purple). The structure alignment is based on the bound ligands. The Lck–staurosporine cocrystallization complex contains a loop conformation that is intermediate between the more open and closed positions observed in the CDK2 and PKA complexes. 656 Structure 1999, Vol 7 No 6

Figure 5

Structure-based sequence alignment of Lck, β1 β2 β3 ZAP-70, EGFR and PKA. Conserved residues Lck 225 -QTQKPQKPWWEDEWEVPRETLKLVERLGAGQFGEVWMGYY-----NGHTKVAVKSLKQG are highlighted in yellow. The amino acids in the ZAP-70 -YSDPEE-LKDKKLFLKRDNLLIADIELGCGNFGSVRQGVYRMRK--KQIDVAIKVLKQG hydrophobic pocket where PP2 binds are EGFR -LTPSGEAPNQALLRILKETEFKKIKVLGSGAFGTVYKGLWIPEGEKVKIPVAIKELREA highlighted in cyan; Tyr394 on the activation PKA AKEDFLKKWESPAQNTAHLDQFERIKTLGTGSFGRVMLVKHKETG----NHYAMKILDKQK loop is highlighted in magenta. The linker and catalytic regions are labeled. The positions of αC β4 β5 linker αD secondary structure elements are indicated with arrows (β strands) and rectangles (α helices). Lck 279 SMS---PDAFLAEANLMKQLQHQRLVRLYAVVTQEPIYIITEYMENGSLVDFLKTPSGIK ZAP-70 TEKAD-TEEMMREAQIMHQLDNPYIVRLIGVCQAEALMLVMEMAGGGPLHKFLVGKR-EE EGFR TSPKA-NKEILDEAYVMASVDNPHVCRLLGICLTSTVQLITQLMPFGCLLDYVREHK-DN PKA VVKLKQIEHTLNEKRILQAVNFPFLVKLEFSFKDNSNLYMVMEYVPG-GEMFSHLRRIGR

αE catalytic β6 β8

Lck 335 LTINKLLDMAAQIAEGMAFIEERNYIHRDLRAANILVSDTLSCKIADFGLARLI--EDNE ZAP-70 IPVSNVAELLHQVSMGMKYLEEKNFVHRDLAARNVLLVNRHYAKISDFGLSKALGADDSY EGFR IGSQYLLNWCVQIAKGMNYLEDRRLVHRDLAARNVLVKTPQHVKITDFGLAKLLGAEEKE PKA FSEPHARFYAAQIVLTFEYLHSLDLIYRDLKPENLLIDQQGYIQVTDFGFAKRV--KGRT

αEF αF αG

Lck 394 YTAREGAKFPIKWTAPEAINYGTFT-IKSDVWSFGILLTEIVTHGRI-PYPGMTNPEVIQ ZAP-70 YTARSAGKWPLKWYAPECINFRKFS-SRSDVWSYGVTMWEALSYGQK-PYKKMKGPEVMA EGFR YHAE-GGKVPIKWMALESILHRIYT-HQSDVWSYGVTVWELMTFGSK-PYDGIPASEISS PKA WTLC----GTPEYLAPEIILSKGYNK-AVDWWALGVLIYEMAAGYPP--FFADQPIQIYE

αH αI

Lck 451 NLERG--YRMVRPDNCPEELYQLMRLCWKER—PEDRPTFDYLRSVLEDFFT------ZAP-70 FIEQG--KRMECPPECPPELYALMSDCWIYK—WEDRPDFLTVEQRMRACYY------EGFR ILEKG--ERLPQPPICTIDVYMIMVKCWMID—ADSRPKFRELIIEFSKMARDPQR---YL PKA KIVSG---KVRFPSHFSSDLKDLLRNLLQVD-LTKRFGNLKNGVNDIKNHKWFATTD-WI

Lck 500 ATEG--QYQ-PQP------ZAP-70 SLAS--KVEGP-PGSTQKAEAACA------EGFR VIQGDERMHLPSPTDSNFYRALMDEEDMDDVVDADEYLIPQQGFFSSPSTSRTPLLSSLS PKA AIYQR-KVEAPFIPKFKGPGDTSNFDDYEEEEIRVSINEKCGKEFSEF------Structure

PP2 binds in the ATP-binding site and induces little of the pyrazolopyrimidine contributes four van der Waals global conformational change in the enzyme. Superimposi- contacts to the complex. This substituent is located at the tion of the Lck–PP2 and Lck–AMP-PNP structures yields entrance of the ATP-binding pocket and contacts residues an overall rms difference of 0.27 Å for 278 Cα atoms. The from both the N- and C-terminal lobes of Lck. pyrazolopyrimidine ring of PP2 occupies a similar position in the Lck ATP-binding cleft as the adenine ring of AMP- The hydrophobic pocket occupied by the 3-(4- PNP (Figures 1c and 2c). This binding mode places the 3- chlorophenyl) substituent of PP2 is defined by residues (4-chlorophenyl) substituent of PP2 in a hydrophobic Thr316, Ile314, Met292, Glu288 and Lys273. The exact pocket adjacent to the ATP-binding cleft (Figures 3c and composition of this pocket appears to be unique to the Src 3d). PP2 forms three hydrogen bonds with Lck, two of family (Figure 5). For instance, Thr316, which is located at which are similar to those found in the Lck–AMP-PNP the entrance of the hydrophobic pocket, is not conserved and Lck–staurosporine structures (Figures 1c and 2c). in other tyrosine kinase families. ZAP-70 contains a These hydrogen bonds are between the 4-amino group of methionine at this position, which is likely to block access PP2 and the backbone carbonyl of Glu317, and between of this pocket to PP2-like inhibitors. This observation is µ the N5 of PP2 and the backbone NH of Met319. The consistent with the 100 M IC50 previously reported for third hydrogen bond, formed between the 4-amino group PP2 against ZAP-70 [46]. Like Lck, the EGFR kinase has of PP2 and the sidechain hydroxyl group of Thr316, is a threonine at the entrance of the hydrophobic pocket and µ unique in the structures reported here. The two conserved is inhibited moderately by PP2 (IC50 = 0.45 M) [46]. The hydrogen bonds in the Lck–PP2 complex are relatively hydrophobic pocket in EGFR differs from the Src kinases long, with distances between donor and acceptor atoms of by having a leucine at the position equivalent to Ile314 in approximately 3.2 Å. PP2 also makes 38 van der Waals Lck. In the Lck–PP2 complex, Ile314 contacts the 4-chloro interactions with Lck. Of these contacts, 19 involve the 3- substituent of the 3-phenyl ring. The presence of a leucine (4-chlorophenyl) substituent, which is deeply buried at this position in EGFR could partially account for the inside the hydrophobic pocket. The tert-butyl substituent weaker inhibition of this by PP2. Research Article Lck in complex with non-selective and Src family selective kinase inhibitors Zhu et al. 657

Figure 6

The chemical structures of PP1, PP2 and 4-amino-1,3-diphenyl-pyrrolo[3,4-d]pyrimidines. CH3 Cl R1

NH2 NH2 NH 2 4 3 4 4 5 N 3 3 2 5 N 5 N N2 6 N2 N N1 6 N1 6 N N N1 7 7 7

R2

4-Amino-1,3-diphenyl- PP1 PP2 pyrrolo[3,4d]pyrimidines Structure

The structure of the Lck–PP2 complex helps to explain Comparison of ligand binding to Lck the structure activity relationships of a series of 4-amino- Superimposition of the Lck–AMP-PNP, Lck–stauro- 1,3-diphenyl-pyrrolo[3,4d]pyrimidines that show a high sporine and Lck–PP2 structures highlights similar features degree of specificity towards c-Src [52]. The molecular of inhibitor binding to the enzyme. The aromatic ring structures of these compounds are analogous to PP2, but systems of the bound inhibitors occupy similar positions in have a phenyl ring at the N1 position of the pyrrole the adenine-binding pocket, as do their hydrogen-bond instead of a tert-butyl group (Figure 6). A wide variety of donors and acceptors (Figure 7). This results in a hydro- polar moieties are well tolerated on this phenyl ring. The gen-bonding pattern to the backbone carbonyl of Glu317 amino acid identity of the active sites of Lck and Src and the amide of Met319 that is conserved in all three (defined as a 10 Å radius around ATP) is 89%. The structures (Figures 1 and 2). amino acid composition of the ribose binding pocket within the Src family is completely conserved, whereas Staurosporine makes significantly more interactions with the hydrophobic pocket is less conserved. Superimposi- the glycine-rich loop of Lck than does either AMP-PNP or tion of several of these compounds on the Lck–PP2 PP2. The majority of these interactions are with residues complex indicates that the polar groups at the N1-phenyl that are highly conserved among protein kinases. These ring of pyrrolopyrimidines can interact favorably with include Leu251, Gly252, Val259, Ala271, Lys273, Gly322 hydrophilic residues in the ribose-binding pocket and Leu371, residues that are either absolutely or highly (Ser343 and Asp345), while the 3-phenyl group occupies conserved among known tyrosine kinase sequences. In the same region of the hydrophobic pocket as the 3-(4- contrast, PP2 makes a number of interactions with chlorophenyl) group of PP2. residues that are specific to the Src family kinases by

Figure 7

Comparison of the ligand positions in the Lck complexes based on a superposition of the Cα atoms of Lck. (a) AMP-PNP (purple) and staurosporine (white). (b) AMP-PNP (purple) and PP2 (white). 658 Structure 1999, Vol 7 No 6

Figure 8 other kinases as well and has been exploited in the discov- ery of specific inhibitors. For example, structures of (a) 100 fibroblast growth factor receptor (FGFR) and p38 mitogen-activated protein (MAP) kinases bound with spe- cific inhibitors show that the inhibitors gain both potency and specificity by placing substituents in this hydrophobic pocket [53–56]. However, the exact position and topology 50 of the hydrophobic pockets of Lck, FGFR, p38 and other kinases are likely to be defined not only by sequence, but

Inhibition (%) by additional factors such as activation state or relative positioning of the kinase N- and C-terminal domains. This diversity around the ATP-binding site provides opportuni- ties for the discovery or design of potent, selective, small- 0 0.001 0.01 0.1 1 10 molecule inhibitors for specific protein kinases. Compound (µM) (b) Staurosporine PP2 Inhibition of Lck activity and T-cell receptor signaling (µM) 0.02 0.2 2.0 0.2 2.0 20 In the crystallographic studies presented here, the catalytic Anti-CD3 +++++++– domain of Lck was used as a substitute for the full-length kDa p110 98 protein. Previous studies have demonstrated that the Lck p70 catalytic domain can be expressed and isolated as a consti- p56 64 50 tutively active enzyme [40]. Nevertheless, a detailed com- 36 parison of the catalytic activities of Lck in the full-length 30 and truncated forms has not been reported. To provide a TCR ζ p23 basis for the physiological relevance of our crystallographic TCR ζ p21 16 studies, the IC50 values of staurosporine and PP2 were 6 measured against these two forms of Lck. The full-length and catalytic domains of Lck displayed comparable specific (c) 100 activities (10 and 15 nmoles/min/nM enzyme, respectively) when assayed using a poly-Glu-Tyr substrate. Further-

more, staurosporine (IC50 full-length = 34 nM, kinase domain = 40 nM) and PP2 (IC50 full-length = 19 nM, kinase domain = 20 nM) each inhibited both the full- length and truncated forms of Lck to a similar extent in an 50 autophosphorylation assay (Figure 8a).

The effect of staurosporine and PP2 on the Lck-medi- ated phosphorylation of TCR ζ chain and IL-2 produc- tion in human T cells was also investigated. Both Inhibition of IL-2 secretion (%) PP2 Staurosporine inhibitors showed a dose-dependent inhibition of Lck- dependent phosphorylation of the TCR ζ chain (p23), 0 0.00010.001 0.01 0.1 1 10 100 and also inhibited the phosphorylation of a 70 kDa µ protein which is likely to be ZAP-70 (Figure 8b). Compound ( M) Structure Figure 8c, shows that staurosporine (IC50 = 60 nM) and PP2 (IC50 = 600 nM) also exhibited dose-dependent inhi- Inhibition of Lck activity. (a) Enzymatic assay. IC50 titration curves for the Lck catalytic domain (squares) and the nearly full-length enzyme bition of IL-2 production in human T-cell cultures. The with SH2 and SH3 regulatory domains (circles). The Lck proteins were results of these in vitro and cellular studies suggest that titrated with staurosporine (open symbols) and PP2 (filled symbols). the catalytic domain of Lck is a valid substitute for the (b) Endogenous protein phosphorylation assay. PP2 and staurosporine full-length Lck as a molecular target for the development inhibit TCR-induced increases in phosphotyrosine incorporation into of new immunosuppressive therapeutic agents. the TCR ζ chain (p23) and a 70 kDa protein (see Materials and methods section for details). (c) Cellular assay. PP2 (filled diamonds) and staurosporine (filled triangles) inhibit TCR-induced IL-2 secretion Biological implications from human peripheral blood lymphocytes. The molecular targets of currently used immunosup- pressive drugs such as FK-506 and cyclosporine are broadly expressed in many different tissues and cell accessing a hydrophobic pocket neighboring the adenine- types. The non-immunosuppressant toxicity profiles of binding region of Lck. This hydrophobic pocket exists in such drugs can be traced to the inhibition of their targets Research Article Lck in complex with non-selective and Src family selective kinase inhibitors Zhu et al. 659

in non-lymphoid tissues. Targeting lymphocyte-specific at 0.5 mg/ml and α-thrombin added at a 1:1000 ratio (w/w). A pro- kinase (Lck) for the development of novel immunosup- tease inhibitor cocktail was then added and the protein sample was incubated for 30 min at 25°C. The inhibition of thrombin was confirmed pressive drugs has promise, as this enzyme is selectively in a spectrophotometric assay as described [57,58]. The cleaved GST expressed in T cells and natural killer (NK) cells. Thus, and Lck were separated by anion exchange chromatography essentially agents that selectively inhibit Lck could lead to T-cell- as described for the separation of Lck phosphorylation species [40]. specific immunosuppression with improved therapeutic The pooled fraction of Lck was then concentrated in a centriprep-10 windows and broader clinical potential. and size fractionated on a column of superdex-75. The monomeric frac- tion appeared homogeneous by sodium dodecyl sulfate (SDS) and native polyacrylamide gel electrophoresis (PAGE). In the past few years much progress has been made in the design of selective kinase inhibitors. It has been estab- Structure determination lished that highly specific ATP-competitive inhibitors can Crystals of the Lck kinase domain in complex with AMP-PNP/Mg (5 mM) were grown from 1.6 M ammonium sulfate in 0.1 M Bis-Tris be obtained against a number of different kinases with (pH 6.5) by the hanging-drop method. These crystals are isomorphous clinical utility in oncology. In the present study, a com- to the apo Lck crystals [40]. Crystals of apo Lck were obtained under parison of several ligated Lck structures has provided the same conditions as described above by microseeding the apo valuable insight into the mode of binding of non-selective protein sample with the crystals of Lck–AMP-PNP. These crystals were subsequently soaked for three days in a solution containing 1.6 M and Src family selective inhibitors. The structure of Lck ammonium sulfate, 0.1 M Bis-Tris (pH 6.5) and 0.3 mM staurosporine. in complex with a non-hydrolyzable analog of ATP, Cocrystals of the Lck–staurosporine complex were grown from 0.1 M

AMP-PNP, is likely to represent a conformation of Lck Tris (pH 8.5), 0.25 M Li2SO4, 20% polyethylene glycol (PEG) 6000 by when ATP is bound prior to the binding of substrates and the hanging-drop method. Diffraction-quality crystals were obtained after one round of macroseeding. The crystals belong to space group phosphotransfer. Analysis of the Lck–staurosporine × × 3 P212121 with unit-cell dimensions 61.5 69.0 73.7 Å and contain complex reveals that binding of this inhibitor to Lck and one complex per asymmetric unit. other kinases induces a conformational change in the glycine-rich loop, which helps to maximize van der Crystals of Lck–AMP-PNP and Lck–staurosporine (soaked) were equili- Waals interactions. This conformational change is medi- brated against a solution containing 1.6 M ammonium sulfate, 0.1 M Bis-Tris and 20% ethylene glycol and frozen at 100K for data collec- ated by a CH–O interaction that appears to be a common tion. Diffraction data for the crystals of Lck–AMP-PNP were collected at component of kinase–staurosporine complexes. The non- the X4A beamline at Brookhaven National Laboratory using an Raxis-IV selectivity of staurosporine may be explained by interac- image-plate detector. Diffraction data for the Lck–PP2 and Lck–stau- tions with residues that are highly conserved in the rosporine cocrystals were collected on an Raxis-II image-plate detector mounted on the RU300 generator. Lck–PP2 crystals were equilibrated ATP-binding cleft. In contrast, the Src-selective inhibitor as above prior to freezing, whereas the Lck–staurosporine cocrystals PP2 binds to Lck by accessing a hydrophobic pocket, the were transferred to a solution containing 0.05 M Tris (pH 8.5), 0.025 M composition of which is unique to the Src family. The Li2SO4, 20% PEG 6000 and 15% PEG 400 prior to freezing. All data structures of these Lck complexes offer useful structural were processed using the HKL software package [59]. insights as they demonstrate binding modes that make dif- The structure of the Lck–staurosporine cocrystallization complex was ferential use of various regions of the ATP-binding cleft. solved by molecular replacement using the program AmoRe [60]. The Furthermore, these complexes indicate that kinase selec- apo Lck structure was used as a search model. The initial molecular tivity can be achieved with small-molecule inhibitors that replacement solution was subject to rigid-body and positional refine- ment using X-PLOR [59] (Molecular Simulations, Inc.) exploit subtle topological differences or sequence substitu- tions among protein kinases. Bound ligands were identified using the difference Fourier method phased by the structure of apo Lck [40]. Model building of the protein and inhibitor into electron-density maps was performed using the Materials and methods graphic program Quanta (Molecular Simulations, Inc.) and the struc- Construct design, protein expression and protein purification tures were refined using X-PLOR [61]. The graphic figures were made Full-length Lck cDNA (a gift from T Roberts, DFCI) was used as a tem- using the programs GRASP [62] and SETOR [63]. A summary of the plate for the amplification of a 879 fragment encoding amino diffraction data and refinement statistics are given in Table 1. acid residues 225–509 of the Lck catalytic domain using the poly- merase chain reaction (PCR). The PCR product was cloned into the Kinase activity assays Bam HI and Eco RI sites of the plasmid vector pFastBac1(Gibco/BRL) Protein kinase activity was measured in two different in vitro assays. In modified to contain the coding region for glutathione-S- the first assay, the kinase of interest was incubated with [33P]-ATP in a (GST) and a thrombin cleavage site upstream of the multiple cloning 96-well plate previously coated with substrate (i.e. poly-Glu-Tyr 4:1) and site. Recombinant baculovirus was obtained using the Bac-to-Bac the kinase activity determined in a Microbeta, Wallac Top-Count expression system (Gibco/BRL). After two rounds of amplification in (Packard Instruments). In the second assay, protein kinase autophos- Sf9 insect cells (Spodoptera frugiperda) cultured in Hink’s modification phorylation was examined. GST-fused Lck proteins consisting of either of Graces media, the virus was used to infect High Five insect cells the kinase domain (residues 225–509) or nearly full-length sequences (Trichoplusia ni) grown in Ex-cell 405 media for protein production. (residues 66–509) were incubated in 10 mM Mg2+, 25 mM Tris 7.5, 1 mM DTT, 1 µM ATP (10 µCi/ml [33P]-ATP) for 5 min at room tempera- Recombinant GST–Lck residues (225–509) was purified from bac- ture with the indicated concentration of compounds. The reaction was culovirus cells essentially as previously described [40], except that the stopped by the addition of one volume of 10% trichloroacetic acid first step involved fractionating cell lysates on glutathione sepharose (TCA) and filtered through a millipore filter plate. After three washes with (Pharmacia Biotech). The GST–Lck bound to the resin was eluted with 200 µl 10% TCA, 50 µl of scintillation cocktail was added to each well 30 mM glutathione and cleaved overnight at 4°C with the fusion protein and the plate was read in a microbeta scintillation counter (Wallach). 660 Structure 1999, Vol 7 No 6

Table 1

Diffraction data and refinement statistics.

‡ Resolution (Å) Rsym* Completeness Unique Observed R factor (%) Rfree Rmsd (%) bonds (Å)/angles (°) AMP-PNP 1.6 5.9 (14.5%) 98 (96%) 38,529 258,235 20.0 23.0 0.018/2.1 Staurosporine 1† 2.0 5.9 (12.1%) 99 (100%) 20,010 161,934 19.5 23.7 0.019/2.0 Staurosporine 2† 2.2 7.2 (20.2%) 96 (94%) 16,145 53,567 21.4 26.2 0.017/2.2 PP2 2.0 8.2 (19.8%) 93 (90%) 19,946 53,618 18.9 25.4 0.018/1.8

*Rsym values in parentheses are for the highest resolution shell. water molecules added. The N- and C-terminal segments (GS-225–230 †Staurosporine 1 and 2 are derived from the soaking and cocrystallization and 502–509, where GS represents the two extra residues left from the σ experiments, respectively. Data with Fobs greater than 1 were used in the thrombin cleavage) are disordered, however. In the staurosporine 2 structural refinement. In all the structures except for that of staurosporine structure, residues 235–501 and water molecules are included in the final 2, residues 231–501 of Lck are included in the final refined model with refined model. ‡Rmsd, root mean square deviation.

T-cell activation References Whole blood was obtained from normal donors and human peripheral 1. Alberola-Ila, J., Takaki, S., Kerner, J.D. & Perlmutter, R.M. (1997). blood lymphocytes (hPBL) were isolated by ficol-hypaque density cen- Differential signaling by lymphocyte receptors. Annu. Rev. trifugation. T cells were then purified from the hPBL by negative selec- Immunol. 15, 125-154. tion using an R&D column following the manufacturers directions (R&D 2. Berridge, M.J. (1997). Lymphocyte activation in health and disease. Systems, Minneapolis, MN). A 96-well flat-bottomed plate was coated Crit. Rev. Immunol. 17, 155-178. 3. Qian, D. & Weiss, A. (1997). antigen receptor signal with 10 µg/ml of goat antimouse (GAM) IgG (Caltag, Burlingame, CA) 1 transduction. Curr. Opin. Cell Biol. 9, 205-212. in phosphate buffered saline (PBS) overnight at 4°C. The GAM-coated 4. Barber, E.K., Dasgupta, J.D., Schlossman, S.F., Trevillyan, J.M. & Rudd, plate was flicked out and anti-CD3 mAb (UCHT-1, Coulter/Immunotech, C.E. (1989). The CD4 and CD8 are coupled to a protein- Miami, FL) added at a concentration of 0.2 µg/ml in AIMV medium tyrosine kinase (p56lck) that phosphorylates the CD3 complex. Proc. (Gibco, Grand Island, NY) for 3 h at 37°C. Purified T cells were pre- Natl Acad. Sci. USA 86, 3277-3281. incubated at 1 × 105/well in AIMV with or without compound for 30 min 5. Wong, J., Straus, D. & Chan, A.C. (1998). Genetic evidence of a role then transferred to the anti-CD3 capture plate. Finally, anti-CD28 for Lck in T-cell receptor function independent or downstream of ZAP- (Pharmingen, San Diego, CA) in AIMV (150 ng/ml final concentration) 70/Syk protein tyrosine kinases. Mol. Cell Biol. 18, 2855-2866. was added to each well. Cells were incubated for 20 h at 37°C in 5% 6. Thome, M., Germain, V., DiSanto, J.P. & Acuto, O. (1996). The p56lck CO then supernatants were tested by enzyme-linked immunosorbant SH2 domain mediates recruitment of CD8/p56lck to the activated T 2 cell receptor/CD3/zeta complex. Eur. J. Immunol. 26, 2093-2100. (ELISA) for levels (Endogen, Woburn, MA). 7. Chu, K. & Littman, D.R. (1994). Requirement for kinase activity of CD4-associated p56lck in antibody-triggered T cell signal Phosphotyrosine western blotting transduction. J. Biol. Chem. 269, 24095-24101. Jurkat (ATCC, Manassas, VA) cells (1 × 107) in RPMI-1640 (Gibco, 8. Chalupny, N.J., Ledbetter, J.A. & Kavathas, P. (1991). Association of Grand Island, NY) containing 10% fetal calf serum (FCS) (Sigma, St CD8 with p56lck is required for early T cell signaling events. EMBO J. Louis, MO) were incubated with or without anti-CD3 mAb (UCHT-1, 10, 1201-1207. 10 µg/ml) for 15 min on ice. Cells were washed in cold PBS then incu- 9. van Oers, N.S., Killeen, N. & Weiss, A. (1996). Lck regulates the bated with or without GAM IgG (10 µg/ml) in RPMI-1640 containing tyrosine phosphorylation of the T cell receptor subunits and ZAP-70 in 1 183 10% FCS for 15 min on ice. Cells were transferred to a 37°C water murine thymocytes. J. Exp. Med. , 1053-1062. 10. Kersh, E.N., Shaw, A.S. & Allen, P.M. (1998). Fidelity of T cell bath for 1 min. Stimulation was stopped by the addition of five volumes µ activation through multistep T cell receptor zeta phosphorylation. of cold PBS containing 200 M sodium orthovanadate. Cells were Science 281, 572-575. spun down and lysed in 150 mM Tris/10 mM HEPES buffer (pH 7.3), 11. Madrenas, J., Wange, R.L., Wang, J.L., Isakov, N., Samelson, L.E. & containing 1% Triton X-100 and complete protease inhibitor cocktail Germain, R.N. (1995). Zeta phosphorylation without ZAP-70 activation (Boehringer Mannheim, Germany) for 30 min on ice. Whole cell lysates induced by TCR antagonists or partial agonists. Science 267, 515-518. (2 × 106/cell equivalents per lane) were separated by 14% reducing 12. Straus, D.B. & Weiss, A. (1993). The CD3 chains of the T cell antigen SDS-PAGE and transferred to PVDF membrane. Blots were probed receptor associate with the ZAP-70 tyrosine kinase and are tyrosine with antiphosphotyrosine (4G10, Upstate Biotechnology, Inc., Saranac phosphorylated after receptor stimulation. J. Exp. Med. 178, 1523-1530. Lake, NY) and developed using ECL-plus following the manufacturers 13. Wange, R.L., Kong, A.N. & Samelson, L.E. (1992). A tyrosine- phosphorylated 70 kDa protein binds a photoaffinity analogue of ATP directions (Amersham, Arlington Heights, IL). and associates with both the zeta chain and CD3 components of the activated T cell antigen receptor. J. Biol. Chem. 267, 11685-11688. Accession numbers 14. Chan, A.C., Irving, B.A., Fraser, J.D. & Weiss, A. (1991). The zeta The coordinates will be deposited in the . chain is associated with a tyrosine kinase and upon T-cell antigen receptor stimulation associates with ZAP-70, a 70 kDa tyrosine phosphoprotein. Proc. Natl Acad. Sci. USA 88, 9166-9170. Acknowledgements 15. Isakov, N., Wange, R.L., Burgess, W.H., Watts, J.D., Aebersold, R. & We would like to acknowledge Christina Boucher, Tom Daniell and David Samelson, L.E. (1995). ZAP-70 binding specificity to T cell receptor Carney for their assistance with the Lck enzymatic assays and Michael tyrosine-based activation motifs: the tandem SH2 domains of ZAP-70 Begley for his diligence in Lck purification. We would especially like to thank bind distinct tyrosine-based activation motifs with varying affinity. David Armistead and Nicholas Lydon for their unwavering support during J. Exp. Med. 181, 375-380. the course of this work and for their helpful discussions and editorial com- 16. Wange, R.L., Malek, S.N., Desiderio, S. & Samelson, L.E. (1993). ments during preparation of the manuscript. We would also like to thank Tandem SH2 domains of ZAP-70 bind to T cell antigen receptor zeta Craig Ogata and coworkers for their support during data collection at the and CD3 epsilon from activated Jurkat T cells. J. Biol. Chem. beamline X4A at the Brookhaven National Laboratory. 268, 19797-19801. Research Article Lck in complex with non-selective and Src family selective kinase inhibitors Zhu et al. 661

17. Chan, A.C., et al., & Kurosaki, T. (1995). Activation of ZAP-70 kinase 41. Meggio, D., et al., & Furet, P. (1995). Different susceptibility of protein activity by phosphorylation of tyrosine 493 is required for lymphocyte kinases to staurosporine inhibition. Kinetic studies and molecular antigen receptor function. EMBO J. 14, 24499-24508. bases for the resistance of protein kinase CK2. Eur. J. Biochem. 18. Neumeister, E.N., Zhu, Y., Richard, S., Terhorst, C., Chan, A.C. & 234, 317-322. Shaw, A.S. (1995). Binding of ZAP-70 to phosphorylated T-cell 42. Prade, L., Engh, R.A., Girod, A., Kinzel, V., Huber, R. & Bossemeyer, D. receptor zeta and eta enhances its autophosphorylation and (1997). Staurosporine-induced conformational changes of cAMP- generates specific binding sites for SH2 domain-containing proteins. dependent protein kinase catalytic subunit explain inhibitory potential. Mol. Cell Biol. 15, 3171-3178. Structure 5, 1627-1637. 19. Watts, J.D., Affolter, M., Krebs, D.L., Wange, R.L., Samelson, L.E. & 43. Lawrie, A.M., Noble, M.E., Tunnah, P., Brown, N.R., Johnson, L.N. & Aebersold, R. (1994). Identification by electrospray ionization mass Endicott, J.A. (1997). Protein kinase inhibition by staurosporine spectrometry of the sites of tyrosine phosphorylation induced in revealed in details of the molecular interaction with CDK2. Nat. Struct. activated Jurkat T cells on the protein tyrosine kinase ZAP-70. J. Biol. Biol. 4, 796-801. Chem. 269, 29520-29529. 44. Toledo, L.M. & Lydon, N.B. (1997). Structures of staurosporine bound 20. LoGrasso, P.V., Hawkins, J., Frank, L.J., Wisniewski, D. & Marcy, A. to CDK2 and cAPK — new tools for structure-based design of protein (1996). Mechanism of activation for Zap-70 catalytic activity. Proc. kinase inhibitors. Structure 5, 1551-1556. Natl Acad. Sci. USA 93, 12165-12170. 45. Lamers, M.B., Antson, A.A., Hubbard, R.E., Scott, R.K. & Williams, 21. Wange, R.L., Guitian, R., Isakov, N., Watts, J.D., Aebersold, R. & D.H. (1999). Structure of the protein tyrosine kinase domain of C- Samelson, L.E. (1995). Activating and inhibitory mutations in adjacent terminal Src kinase (CSK) in complex with staurosporine. J. Mol. Biol. in the kinase domain of ZAP-70. J. Biol. Chem. 285, 713-725. 270, 18730-18733. 46. Hanke, J.H., et al., & Connelly, P.A. (1996). Discovery of a novel, 22. Rudd, C.E. (1999). Adaptors and molecular scaffolds in immune cell potent, and Src family-selective tyrosine kinase inhibitor. Study of Lck- signaling. Cell 96, 5-8. and FynT-dependent T cell activation. J. Biol. Chem. 271, 695-701. 23. Straus, D.B. & Weiss, A. (1992). Genetic evidence for the involvement 47. Knighton, D.R. et. al., & Sowadski, J.M. (1991). Crystal structure of the of the Lck tyrosine kinase in signal transduction through the T cell catalytic subunit of cyclic adenosine monophosphate-dependent antigen receptor. Cell 70, 585-593. protein kinase. Science 253, 407-414. 24. Henning, S.W. & Cantrell, D.A. (1998). p56lck signals for regulating 48. Knighton, D.R., et al., & Sowadski, J.M. (1991). Structure of a peptide thymocyte development can be distinguished by their dependency on inhibitor bound to the catalytic subunit of cyclic adenosine Rho function. J. Exp. Med. 188, 931-939. monophosphate-dependent protein kinase. Science 253, 414-420. 25. Levin, S.D., Abraham, K.M., Anderson, S.J., Forbush, K.A. & Perlmutter, 49. Hubbard, S.R. (1997). Crystal structure of the activated insulin R.M. (1993). The protein tyrosine kinase p56lck regulates thymocyte receptor tyrosine kinase in complex with peptide substrate and ATP development independently of its interaction with CD4 and CD8 analog. EMBO J. 16, 5572-5581. coreceptors J. Exp. Med. 178, 245-255. Erratum appears in J. Exp. 50. Kuduva, S.S., Craig, D.C., Nangia, A. & Gautam, R. (1999). Med. (1993) 178 p.1135. Cubanecarboxylic acids. Crystal engineering considerations and the 26. Molina, T.J., et al., & Veillette A. (1992). Profound block in thymocyte role of C–H O hydrogen bonds in determining O–H O networks. development in mice lacking p56lck. Nature 357, 161-164. J. Am. Chem. Soc. 121, 1936-1944. 27. Molina, T.J., Bachmann, M.F., Kundig, T.M., Zinkernagel, R.M. & Mak, 51. Kim, Y.H., Buchholz, M.A., Chrest, F.J. & Nordin, A.A. (1994). Up- T.W. (1993). Peripheral T cells in mice lacking p56lck do not express regulation of c-myc induces the of the murine significant antiviral effector functions. J. Immunol. 151, 699-706. homologues of p34cdc2 and cyclin-dependent kinase-2 in T 28. Lin, R.S., Rodriguez, C., Veillette, A. & Lodish, H.F. (1998). Zinc is lymphocytes. J. Immunol. 152, 4328-4333. essential for binding of p56(lck) to CD4 and CD8alpha. J. Biol. Chem. 52. Traxler, P. (1998). Tyrosine kinase inhibitors in cancer treatment (Part 273, 32878-32882. II). Expert Opinion in Therapeutic Patents 8, 1599-1625. 29. Huse, M., Eck, M.J. & Harrison, S.C. (1998). A Zn2+ ion links the 53. Mohammadi, et al., & Hubbard, S.R. (1998). Crystal structure of an cytoplasmic tail of CD4 and the N-terminal region of Lck. J. Biol. angiogenesis inhibitor bound to the FGF receptor tyrosine kinase Chem. 273, 18729-18733. domain. EMBO J. 17, 5896-5904. 30. Watts, J.D., et al., Aebersold, R. (1992). Purification and initial 54. Tong, L., et al., & Pargellis, C.A. (1997). A highly specific inhibitor of characterization of the lymphocyte-specific protein-tyrosyl kinase human p38 MAP kinase binds in the ATP pocket. Nat. Struct. Biol. p56lck from a baculovirus expression system. J. Biol. Chem. 4, 311-316. 267, 901-907. 55. Wilson, K.P., et al., & Su, M.S.S. (1997). The structural basis for the 31. Gervais, F.G., Chow, L.M., Lee, J.M., Branton, P.E. & Veillette, A. specificity of pyridinylimidazole inhibitors of p38 MAP kinase. Chem. (1993). The SH2 domain is required for stable phosphorylation of Biol. 4, 423-431. p56lck at tyrosine 505, the negative regulatory site. Mol. Cell Biol. 56. Wang, Z., et al., & Goldsmith, E.J. (1998). Structural basis of inhibitor 13, 7112-7121. selectivity in MAP kinases. Structure 6, 1117-1128. 32. Bougeret, C., et al., & Fischer, S. (1996). Detection of a physical and 57. Morgenstern, K.A., et al., & Kisiel, W. (1994). Complementary DNA functional interaction between Csk and Lck which involves the SH2 cloning and kinetic characterization of a novel intracellular serine domain of Csk and is mediated by autophosphorylation of Lck on proteinase inhibitor: mechanism of action with trypsin and factor Xa as tyrosine 394. J. Biol. Chem. 271, 7465-7472. model proteinases. Biochemistry 33, 3432-3441. 33. Jullien, P., et al., & Benarous, R. (1994). Tyr394 and Tyr505 are 58. Morgenstern, K.A., et al., & Thomson, J.A. (1997). Polynucleotide autophosphorylated in recombinant Lck protein-tyrosine kinase modulation of the protease, nucleoside triphosphatase, and helicase expressed in Escherichia coli. Eur. J. Biochem. 224, 589-596. activities of a hepatitis C virus NS3–NS4A complex isolated from 34. Williams, J.C., et al., & Wierenga, R.K. (1997). The 2.35 Å crystal transfected COS cells. J. Virol. 71, 3767-3775. structure of the inactivated form of chicken Src: a dynamic molecule 59. Otwinowski, Z. (1993) Oscillation data reduction program. In Data with multiple regulatory interactions. J. Mol. Biol. 274, 757-775. collection and Processing, Proceedings of the National Study 35. Sicheri, F., Moarefi, I. & Kuriyan, J. (1997). Crystal structure of the Src Weekend. (Sawyer, L., Issacs, N. & Bailey, S.W., eds), pp.55-62, family tyrosine kinase Hck. Nature 385, 602-609. Daresbury Laboratory, Warrington, Uk. 36. Xu, W., Harrison, S.C. & Eck, M.J. (1997). Three-dimensional structure 60. Navaza, J. (1994). AmoRe: an automated package for molecular of the tyrosine kinase c-Src. Nature 385, 595-602. replacement. Acta Crystallogr. A 50, 157-163. 37. Reynolds, P.J., Hurley, T.R. & Sefton, B.M. (1992). Functional analysis 61. Brünger, A.T., Krukowski, A. & Erickson, J.W. (1990). Slow-cooling of the SH2 and SH3 domains of the Lck tyrosine protein kinase. protocols for crystallographic refinement by simulated annealing. Acta Oncogene 7, 1949-1955. Crystallogr. A 46, 585-593. 38. Mustelin, T. & Altman, A. (1990). Dephosphorylation and activation of 62. Nicholls, A., Sharp, K.A. & Honig, B. (1991). Protein folding and the T cell tyrosine kinase pp56lck by the leukocyte common antigen association: insights from the interfacial and thermodynamic (CD45). Oncogene 5, 809-813. properties of hydrocarbons. Proteins 11, 281-296. 39. Moarefi, I., et al., & Miller, W.T. (1997). Activation of the Src-family 63. Evans, S.V. (1993). SETOR: hardware-lighted three-dimensional solid tyrosine kinase Hck by SH3 domain displacement. Nature 385, 650-653. model representations of macromolecules. J. Mol. Graph. 11, 134-138. 40. Yamaguchi, H. & Hendrickson, W.A. (1996). Structural basis for activation of human lymphocyte kinase Lck upon tyrosine phosphorylation. Nature 384, 484-489.