Biochem. J. (2014) 460, 211–222 (Printed in Great Britain)

211

doi:10.1042/BJ20131139

*Pfizer Worldwide Research, Eastern Point Road, Groton, CT 06340, U.S.A. †Pfizer Worldwide Research, 200 Cambridge Park Drive, Cambridge, MA 02140, U.S.A.

ITK (interleukin-2-inducible T-cell kinase) is a critical component of signal transduction in T-cells and has a well-validated role in their proliferation, cytokine release and chemotaxis. ITK is an attractive target for the treatment of T-cell-mediated inflammatory diseases. In the present study we describe the discovery of kinase inhibitors that preferentially bind to an allosteric pocket of ITK. The novel ITK allosteric site was characterized by NMR, surface plasmon resonance, isothermal titration calorimetry, enzymology and X-ray crystallography. Initial screening hits bound to both the allosteric pocket and the ATP site. Successful lead optimization was achieved by improving the contribution of the allosteric component to the overall inhibition. NMR competition experiments demonstrated that the dual-site binders showed higher affinity for the allosteric site compared with

the ATP site. Moreover, an optimized inhibitor displayed noncompetitive inhibition with respect to ATP as shown by steadystate enzyme kinetics. The activity of the isolated kinase domain and auto-activation of the full-length enzyme were inhibited with similar potency. However, inhibition of the activated full-length enzyme was weaker, presumably because the allosteric site is altered when ITK becomes activated. An optimized lead showed exquisite kinome selectivity and is efficacious in human whole blood and proximal cell-based assays.

INTRODUCTION

multi-faceted role of ITK in T-cell signalling makes ITK an attractive target for therapeutic intervention for the treatment of autoimmune and allergic diseases. As the expression of ITK is restricted to Th2 cells, ITK inhibitors are less likely to interfere with the natural immune response to viral or Th1 cell-mediated pathogens. ITK shares a similar multi-domain architecture with the other members of the TEC family non-receptor tyrosine kinases [BTK (Bruton’s tyrosine kinase), BMX, TXK and TEC], possessing an N-terminal pleckstrin homology domain, followed by TEChomology, SH3 (Src homology 3), SH2 (Src homology 2) and kinase domains [10]. Activation of ITK is thought to involve phosphorylation of Tyr512 in the activation loop of the kinase domain by the Src family kinase Lck, followed by autophosphorylation of Tyr180 within the SH3 domain [11,12]. Biochemical studies with full-length and truncated ITK reagents have shown phosphorylation of Tyr512 , but not Tyr180 , is critical for ITK catalytic activity in cells [12]. ITK contains a phenylalanine (Phe435 ) as the gatekeeper residue in the ATP-binding site of its kinase domain, whereas all the other TEC family kinases possess a smaller threonine residue at this position. Protein kinases are attractive pharmacological targets for many human diseases. Most of the reported kinase inhibitors compete

ITK (interleukin-2-inducible T-cell kinase) is a TEC family kinase that is expressed in a number of immune cells, including T-cells, mast cells, NK and NKT cells [1]. The significance of ITK in T-cell development is shown by the reduced numbers of αβ Tcells, NK cells and NKT cells and increased numbers of γ δ Tcells and innate CD4 + and CD8 + T-cells in ITK-knockout mice [2]. As a positive regulator of T-cell signalling, ITK activates phospholipase Cγ 1 which signals through NFAT (nuclear factor of activated T-cells) to drive the transcription of various cytokines [3]. ITK is present in naive T-cells and is required for T-cell proliferation and IL-2 (interleukin-2) production. Among the various TEC family kinases, ITK is the only kinase that is expressed in Th2 cells, driving the production of cytokines, IL-4, IL-5 and IL-13 [4–6]. In vitro, Th17-lineage T-cells that lack ITK have reduced IL-17A expression which is thought to be a consequence of a blunted Ca2 + flux [7]. Regulation of T-cell-mediated adhesion requires ITK, which up-regulates LFA-1 (lymphocyte function-associated antigen 1) integrin and its clustering [8]. ITK has also been shown to be involved in linking chemokine signalling pathways to the cytoskeleton which ultimately regulates T-cell trafficking [6,9]. The important and

Key words: allosteric, drug discovery, inhibition mechanism, interleukin-2-inducible T-cell kinase (ITK), kinase inhibitor, structure-based drug design.

Abbreviations: BTK, Bruton’s tyrosine kinase; DDR1, discoidin domain receptor tyrosine kinase 1; IL, interleukin; ITC, isothermal titration calorimetry; ITK, interleukin-2-inducible T-cell kinase; ITK-FL, ITK full-length; ITK-KD, ITK kinase domain; MEK, mitogen-activated protein kinase/extracellular-signalregulated kinase kinase; p[NH]ppA, adenosine 5 -[β,γ-imido]triphosphate; SH2, Src homology 2; SH3, Src homology 3; SPR, surface plasmon resonance; STD, saturation transfer difference. 1 Correspondence may be addressed to these authors (email [email protected], [email protected] or [email protected]). The structural co-ordinates reported for the protein structures have been deposited in the PDB under the accession codes 4M0Y, 4M0Z, 4M12, 4M13, 4M14 and 4M15.  c The Authors Journal compilation  c 2014 Biochemical Society

Biochemical Journal

Seungil HAN*1 , Robert M. CZERWINSKI†, Nicole L. CASPERS*, David C. LIMBURG*, WeiDong DING*, Hong WANG*, Jeffrey F. OHREN*, Francis RAJAMOHAN*, Thomas J. MCLELLAN*, Ray UNWALLA*, Chulho CHOI*, Mihir D. PARIKH*, Nilufer SETH†, Jason EDMONDS†, Chris PHILLIPS*, Subarna SHAKYA†, Xin LI†, Vikki SPAULDING†, Samantha HUGHES*, Andrew COOK*, Colin ROBINSON*, John P. MATHIAS†, Iva NAVRATILOVA†*, Quintus G. MEDLEY†, David R. ANDERSON*, Ravi G. KURUMBAIL*1 and Ann AULABAUGH*1

www.biochemj.org

Selectively targeting an inactive conformation of interleukin-2-inducible T-cell kinase by allosteric inhibitors

212

S. Han and others

with ATP for its binding site, but the generally conserved nature of this site creates a challenge with respect to selectivity and safety. Although a number of potent ATP-competitive inhibitors have been reported for ITK, none of these efforts, to our knowledge, have resulted in a clinical candidate for this target [10]. The main drawback of the reported reversible ATP competitive ITK inhibitors is the lack of kinome selectivity and poor pharmacokinetic properties. Those few compounds which have shown a high specificity for ITK have been plagued by chemotypespecific liabilities such as metabolic instability, cellular toxicity and reactive metabolites. Although the majority of known protein kinase inhibitors are ATP competitive, several examples of small-molecule inhibitors which bind partly or entirely outside the conserved ATP-binding site have also been reported [13,14]. Non-competitive or allosteric inhibitors present an opportunity for improved kinome selectivity and higher biochemical efficiency in the cellular environment. There have been a few attempts to identify non-competitive ATP inhibitors of ITK as well. Exploration of bipartite ITK inhibitors that make contact with both the conserved ATP site and a less conserved region near the activation loop has been unsuccessful, in part due to the bulky and rigid gatekeeper residue Phe435 . ITK is unlikely to adopt a ‘DFG-out’ conformation, as pharmacophores such as biarylureas and biarylamides which are known to induce this conformational change in other kinases are weak inhibitors of ITK in biochemical assays [15]. Previously, based on empirical screening by an SPR (surface plasmon resonance)-based biophysical method, we reported the discovery of dual-site binders of ITK that bind simultaneously at the ATP site and an allosteric site [16]. In order to gain insights into structural features that permit dual-site binding of inhibitors, we have solved multiple crystal structures of ITK in complex with allosteric inhibitors. In the present study we describe the discovery of an allosteric binding pocket in the non-active conformation of ITK and its novel binding features. Our biochemical, biophysical and structural studies provide insights for the design of allosteric ITK inhibitors with improved potency and kinome selectivity. The optimized lead compound, compound 9, is efficacious in multiple cell-based assays.

MATERIALS AND METHODS Cloning, expression and purification of ITK

The kinase domains of human ITK (residues 354–620, K596R) and residues 348–620 were cloned into the pFastBac system with an N-terminal His6 cleavable tag. Recombinant baculovirus was produced using the Bac-to-Bac baculovirus expression system (Invitrogen). Protein was expressed by infection of Spodoptera frugiperda Sf9 insect cells with recombinant baculovirus. ITK was purified from cell lysate using nickel affinity chromatography [NiNTA (Ni2 + -nitrilotriacetate), Qiagen]. The His6 -tagged protein was eluted with imidazole and dialysed overnight at 4 ◦ C with 1:100 (w/w) AcTEV (Invitrogen) protease to remove the tag. The protein was further purified by ion exchange over a 5 ml HiTrap Q HP column (GE Healthcare), followed by gel filtration over a Superdex 200 16/60 column (GE Healthcare). The final protein buffer contained 25 mM Tris/HCl, 50 mM NaCl and 2 mM DTT, pH 8.6, and the protein was concentrated to approximately 3.5 mg/ml. Full-length human ITK was cloned in the same fashion with an N-terminal His6 tag. It was purified by nickel affinity and size exclusion chromatography. To stabilize the protein, 20 % (v/v)  c The Authors Journal compilation  c 2014 Biochemical Society

glycerol and 0.1 % Tween 20 were added to the final protein buffer. Crystallization and data collection of ITK

Crystals of ITK-KD (ITK kinase domain) were grown by the hanging-drop vapour diffusion method at room temperature (22◦ C). Purified ITK was incubated with 0.5–1 mM inhibitor for approximately 30 min prior to crystallization trials. Crystallization drops were set up with 1 μl of protein plus 2 μl of reservoir containing approximately 12 % PEG 3350, 0.1 M magnesium acetate and 0.1 M Hepes, pH 7.2. Crystals grew overnight and were flash frozen in liquid nitrogen after transfer to a cryoprotectant solution containing mother liquor plus 20 % (v/v) glycerol. Data were collected on beamline 17-ID of the Advanced Photon Source (APS) and ID23–1 of the European Synchrotron Radiation Facility (ESRF). Crystals typically diffracted between 1.4 and 2.0 Å (1 Å = 0.1 nm) and belonged to space group P21 . Structure determination

Data were processed using the HKL2000 software suite [17]. The structures of ITK–inhibitor complexes were solved by molecular replacement methods with the CCP4 version of PHASER [18], using the apo ITK-KD structure (PDB code 1SNX) as a search model. After molecular replacement, maximum likelihoodbased refinement of the atomic position and temperature factors were performed with autoBUSTER [19] and the atomic model was built with the program COOT [20]. The stereochemical quality of the final model was assessed with PROCHECK [21]. Crystallographic statistics for the final models are shown in Supplementary Table S1 (at http://www.biochemj.org/ bj/460/bj4600211add.htm). Figures were prepared with PyMOL (http://www.pymol.org). LANCE® assay

ITK LANCE® assays were run at room temperature for 60 min at 2 nM enzyme and 400 nM biotinylated peptide substrate (BioKKVVALYDYMPMN-OH), 1 mM ATP or 5 μM ATP (ATP apparent K m ), 20 mM Hepes, pH 7.2, 10 mM MgCl2 , 0.1 % BSA, 0.0 025 % Brij-25 and 1 mM 2-mercaptoethanol. Reaction mixtures were stopped with 15 mM EDTA and 250 mM NaCl, incubated for 30 min with 1 nM europium-labelled anti-pTyr antibody and 5 μg/ml streptavidin–APC (allophycocyanin) and the time-resolved fluorescence was read at 665 nm/615 nm. Enzyme kinetics

The enzymatic activity of the ITK-KD and ITK-FL (ITK fulllength) enzymes were determined at 25 ◦ C using the coupled PK (pyruvate kinase)/LDH (lactate dehydrogenase) assay as described previously [22]. Data were analysed using Sigmaplot 2000 Enzyme Kinetics Module (SPSS Scientific). Cell-based assays

IP-1 (inositol phosphate 1) ELISA assays to evaluate activation of phospholipase C by ITK were run according to the manufacturer’s instructions with the Cis-Bio IP-One Tb kit using Jurkat T-cells. Cells were starved for 30 min, resuspended in stimulation buffer provided with the kit for 30 min and incubated with 0.25 μg/ml anti-CD3 for 70 min. Cells were lysed and detection reagent

Allosteric inhibitors of ITK

added, then incubated for 1 h in the dark at room temperature and time-resolved fluorescence was read at 665 and 620 nm. Human whole blood was collected in BD Vacutainer Collection tubes containing sodium heparin. Serially diluted compounds were incubated for 1 h in 0.2 ml of a 1:1 mixture of whole blood and RPMI for 1 h in a 37 ◦ C incubator adjusted to 5 % CO2 in 96well flat-bottom plates. Blood was stimulated with 0.06 μg/ml anti-CD3 (BD Biosciences) and anti-CD28 (BD Biosciences) at 37 ◦ C for 22–24 h. IL-2 measurements were performed using the MSD ultrasensitive IL-2 kit. Research using human blood was carried out in accordance with the Declaration of Helsinki (2013) of the World Medical Association and taken by consent from healthy human donors under the IRB protocol number 201065670. NMR competition experiments

The NMR STD (saturation transfer difference) competition experiment spectra were collected at 25 ◦ C on a Bruker Avance 600 MHz spectrometer equipped with a 1.7 ml cryoprobe. The sample initially contained 4 μM ITK-KD348–620 , 300 μM compound 1 or 300 μM ADP in 25 mM d-Tris, 150 mM NaCl, 5 mM MgCl2 and 0.01 % Triton X-100, pH 7.5 buffer. Competition experiments were performed by addition of 100 μM (final concentration) of the potent ATP-site binder BMS-509744 (K d = 8 nM) [23,24] to the ITK/compound 1 complex or by addition of compound 1 (300 μM final) to the ITK–ADP complex. SPR experiments

Sensorchip surfaces were prepared on a Biacore T100 instrument (GE Healthcare). The ITK proteins were captured at a flow rate of 10 μl/min on to an anti-His monoclonal (R&D Systems) surface prepared through standard amine coupling to a CM5 sensor chip (GE Healthcare), and immobilized by cross-linking using NHS/EDC [N-hydroxysuccinimide/N-ethyl-N  -(3-dimethylaminopropyl)carbodi-imide] (10 s) at levels of 16 000 response units for ITK-FL and 4500 response units for ITKKD. The ITK inhibitors were injected over the immobilized proteins for a period of 60 s, dissociation was measured for 500 s with a flow rate of 30 μl/min using the following assay running buffer: 25 mM Hepes, pH 7.5, 100 mM NaCl, 5 mM MgCl2 , 1 mM L-cysteine, 3 % DMSO and 0.02 % Tween 20 at 25 ◦ C. The affinity (K d ) and binding kinetic parameters for the inhibitors were determined by global fitting the sensorgrams for concentrations below 1 μM to a one-site or two-site kinetic model with BIAevaluation software (GE Healthcare). ITC (isothermal titration calorimetry)

Isothermal titration experiments were performed using a high precision AutoITC200 titration calorimeter (MicroCal). The ITK solution (9–10 μM) in the calorimetric cell at 25 ◦ C was titrated with the ligand dissolved in the same buffer. RESULTS Dual-site binding of ITK ligands

One of the early dual-site binders that we had previously identified by SPR screening, compound 1, is shown in Scheme 1. It has an IC50 of 170 nM for the isolated kinase domain, measured at ATP concentration equal to its apparent K m (50 μM). To characterize the binding mode of compound 1 and its analogues, proton NMR STD competition experiments were performed using ITK-KD and ITK-FL. The binding was evaluated in the presence and absence

Table 1

Compound 9 8 7 5 1

213

SPR binding affinities of dual-site inhibitors for ITK-KD and ITK-FL Presence of 1 mM p[NH]ppA

ITK-KD

K d1 a * (μM)

K d2 * (μM)

K d1 * (μM)

ITK-FLnon-activated

K d2 * (μM)

− − + − + − + − +

0.2 0.2 0.3 0.9 1.7 6.4 5.7 0.35 0.24

>10.0 >5.0 n.a.† >5.0 n.a.† >10.0 n.a.† 8.2 n.a.†

0.3 0.7 1.1 0.8 1.4 11.3 5.1 1 0.6

>5 >5 n.a.† >5 n.a.† >15 n.a.† >100 n.a.†

*K d1 : dissociation constant for binding to the allosteric pocket; K d2 : dissociation constant for binding to the ATP pocket. † Not applicable, one class of site observed.

of the previously characterized potent ATP-site binder BMS509744 as a probe [10]. These experiments revealed dual site binding of compound 1 to ITK-KD, as only partial displacement of compound 1 was observed in the presence of a saturating concentration of the probe (Figure 1A). Similar dual-site binding was observed by SPR competition experiments, with K d values of 240 nM for the allosteric pocket and 8.2 μM for the ATP site (Figure 1B and Table 1). We repeated the biophysical studies with ITK-FL to probe the contribution of the other ITK domains for ligand binding. ITK-FL has ∼10-fold tighter binding affinity for ATP (apparent K m = 5 μM). Compound 1 displayed an IC50 of 2.6 μM for ITKFL (evaluated at 5 μM ATP) which is approximately 15-fold lower compared with that for the ITK-KD. The observed biochemical IC50 for ITK-FL is consistent with the dissociation constant (K d ) evaluated by SPR (1 μM) (Table 1). Discovery of an allosteric binding pocket in the ITK-KD

Six allosteric inhibitor-bound crystal structures were determined using unphosphorylated human ITK-KD, all of which were refined to acceptable R-values and geometry (Supplementary Table S1). The introduction of the surface mutation of Lys596 to an arginine residue at the C-terminal lobe contributes to a significant enhancement in the diffraction quality of the ITK crystals in comparison with the wild-type crystals. The mutated arginine side chain forms multiple interactions with the amide backbone of Asp601 and the side chains of Asp402 and Glu405 from a neighbouring molecule in the crystal. All six structures reveal a fully-ordered inactive conformation of the activation loop which forms two short α-helices, thus occluding access to the protein substrate-binding site (Figure 2A). To accommodate the distorted conformation of the activation loop, helix C rotates away from the N-terminal lobe. As a result, the highly conserved Glu406 side chain is shifted away from the active site and is unable to form a critical ion pair with the conserved catalytic Lys391 . Instead, the Glu406 side chain is stabilized by a bidentate hydrogen bond with Arg505 (Figure 2B). This inactive ‘αC helix-out’ conformation resembles the co-crystal structure of BMS-509744 bound to an ITK triple mutant, which was crystallized under completely different conditions [24]. Importantly, the major difference in the structures of ITK between the allosteric inhibitor complex and the BMS-509744 complex is the orientation of the unphosphorylated Tyr512 side chain, which is the critical residue for activation. In the structure of ITK complexed with allosteric inhibitor, Tyr512 points into the active site by forming hydrogen bonds with Asp482 and Arg486 side chains, whereas the equivalent  c The Authors Journal compilation  c 2014 Biochemical Society

214

Scheme 1

S. Han and others

Development of an SPR hit into ITK allosteric inhibitors

See the text for details.

tyrosine side chain in the BMS-509744 complex points towards the solvent [24]. The crystal structure of ITK complexed with compound 1 clearly shows the dual binding of the inhibitor at the known nucleotide-binding site and at a novel allosteric pocket (Figure 3A). The allosteric binding pocket is adjacent to but nonoverlapping with the ATP-binding site. The pocket is defined by the activation loop, the C-helix and loops before and after the C-helix in the N-terminal lobe of the kinase domain. One wall of the allosteric binding pocket is mostly hydrophobic, formed by the side chains of Phe403 , Ala407 , Met410 , Leu421 , Phe435 , Phe501 , Met503 and Phe506 . In contrast, the opposite side comprises several backbone polar groups of Met410 , Leu413 , Val419 , Leu421 and Asp500 as well as the side chain hydroxy of Ser499 (Figure 3B). Comparison of the ligand-bound structure with the apo structure [25] reveals a significant conformational change of the Chelix upon binding of compound 1. The C-helix swings out approximately 36◦ away from the N-terminal lobe to create an allosteric pocket. Importantly, the backbone carbonyl of Met410 at the C-terminal end of the C-helix is displaced by 2.5 Å and forms hydrogen bonds with the amide group of compound 1 and the Leu413 amide nitrogen (Figure 3B). The urea moiety is anchored by multiple hydrogen bonds to the amide backbones of Val419 and Asp500 and the side chain hydroxy of Ser499 . The pyrazole ring is sandwiched between the side chains of Leu421 and Phe501 . The naphthyl group is almost orthogonal to the  c The Authors Journal compilation  c 2014 Biochemical Society

pyrazole ring and is deeply buried within the hydrophobic region of the pocket. The aromatic side chain of Phe403 forms a strong π–π stacking interaction with the naphthyl ring of the ligand (∼4.0 Å between their centroids). The crystal structure also reveals an unoccupied hydrophobic tunnel adjacent to the C7 atom of the naphthyl moiety. This tunnel is composed of Trp356 , Val357 , Phe403 , Ile404 , Ala407 , Met411 , Leu421 , Val424 and Leu426 (Figure 3). At the ATP-binding pocket, the urea and amide moieties of compound 1 form an extensive hydrogen-bonded network with protein atoms from the ITK hinge. The urea moiety is involved in hydrogen bond interaction with the backbone carbonyls of Met438 and Glu439 . The amide group is stabilized by hydrogen bonds with the backbone carbonyl of Glu436 and the backbone amide of Met438 , resembling the conserved interaction of the six-amino group of the adenine ring in the hinge region (Figure 3C). The pyrazole ring is sandwiched between two hydrophobic residues, Ile369 and Leu489 . The naphthyl ring, tilted by 39◦ relative to the pyrazole, makes van der Waals contacts with side chains of Ile369 and Val377 and the backbone of Gly370 of the glycine-rich loop (Figure 3D). No detectable co-operativity of binding between the two pockets is observed as compound 1 has similar binding affinities (K d ) for the allosteric pocket in the presence or absence of p[NH]ppA (adenosine 5 -[β,γ -imido]triphosphate) (Table 1). A similar profile was observed when the ATP site was blocked with a high-affinity ligand [16]. ITC experiments showed

Allosteric inhibitors of ITK

215

formed by the urea-carboxamide moiety at the hinge region. However, the entropic portion (TS) is unfavourable for the ATP site, whereas it is slightly negative for the allosteric site, reflecting displacement of bound water molecules from the allosteric site. Computational analysis by a method called ‘SiteMap’ also shows a significant contribution of hydrophobic interaction in the allosteric pocket compared with the ATP site (Supplementary Table S2 at http://www.biochemj.org/bj/460/bj4600211add.htm) [26].

Structure–activity relationship of the allosteric inhibitor

Figure 1

Two-site binding of compound 1 to ITK

(A) Aromatic region of competition NMR STD spectra. Bottom spectrum: 1D STD spectrum of 300 μM compound 1 with ITK-KD. Top spectrum: STD spectrum of 300 μM compound 1 with ITK-KD after the addition of 100 μM BMS-509744, a potent ATP site binder (K i = 8 nM) demonstrating incomplete displacement of compound 1 in the presence of saturating BMS-509744. (B) Binding of compound 1 to ITK-KD monitored on a BIAcore T100 demonstrating 2:1 binding in the absence (left) and 1:1 binding in the presence (right) of 1 mM p[NH]ppA (AMPPNP).

that the ligand has a higher affinity for the allosteric site compared with the ATP site (Figure 3E and Supplementary Figure S1 at http://www.biochemj.org/bj/460/bj4600211add.htm). The enthalpic component is more favourable for the ATP site presumably because of multiple hydrogen bond interactions

Figure 2

Following the discovery of the ITK allosteric pocket, we embarked on a programme of structure-based drug design to optimize the binding at the allosteric site. Initially, we explored different trajectories from the naphthyl ring of compound 1 with a methoxy substituent. The vast majority of these modifications (compounds 2, 3, 4 and 6) resulted in loss of potency for ITK-FL (Scheme 1). In contrast, methoxy substitution at the 7 position (compound 5) maintained the biochemical inhibition of the catalytic activity of both ITK-FL and ITK-KD. This is consistent with the structural data revealing a ‘hydrophobic tunnel’ emanating from the 7 position of the naphthyl group, the only accessible trajectory that might allow for further optimization. The crystal structure of ITK in complex with compound 5 shows the binding of inhibitor at both the allosteric and the nucleotide binding pockets. In the allosteric binding pocket, compound 5 adopts a similar binding mode to that of compound 1. The methoxy side chain at the 7 position makes van der Waals contacts with the conserved Trp356 and projects into the Met411 side chain. Interestingly, the side chain of Met411 assumes dual conformations in the complex, suggesting that the methoxy group can be further modified to form a more optimal interaction with Met411 . Concurrently, the side chain of Trp356 swings toward the narrow hydrophobic channel to create a space for the alternate side chain conformation of Met411 , resulting in a complete closure of the hydrophobic channel from bulk solvent (Figure 4A). It

Crystal structure of ITK complexed with compound 9 in the presence of ADP

(A) Overall structure of ITK (green) bound to compound 9 (magenta) and ADP (yellow). Tyr512 in the activation loop (red) is shown in stick rendering. (B) Close-up view of the occupation of the allosteric pocket by compound 9 and superimposition on human apo ITK (grey) showing a conformational change in the C-helix. All residues mentioned in the text are rendered as sticks and hydrogen bonds are shown as dashed lines.  c The Authors Journal compilation  c 2014 Biochemical Society

216

Figure 3

S. Han and others

Compound 1 binding at both allosteric and ATP sites

(A) View of the ITK active site surface (pink) looking down through the N-terminus. The opening of the narrow hydrophobic channel is highlighted by a black arrow. (B) Interaction of compound 1 in the allosteric pocket of the ITK (stereo). Compound 1 is shown in stick rendering with cyan carbons and hydrogen bonds as dashes. (C) Interaction of compound 1 in the ATP site, named as ‘pose 1’. (D) Compound 1 interaction in ATP site showing P-loop residues involved in van der Waals contact. (E) Histogram showing changes in G , H , and TS upon compound 1 binding at the allosteric or the ATP pocket.

has previously been shown that the two conserved residues, Trp356 and Met411 , play an important role in the mechanism of regulation of the TEC family members [27,28]. Mutation of either residue resulted in inactivation of ITK, showing that interactions involving these residues are vital for catalytic activity. Therefore the displacement of these two residues by compound 5 could be responsible for the inhibition of ITK catalytic activity. Surprisingly, the binding mode of compound 5 at the ATPbinding site (‘pose 2’) is completely different from that of compound 1 (‘pose 1’). The additional methoxy group at the 7 position on the naphthyl ring causes a steric clash with the glycine-rich loop in the binding mode observed with compound 1. As a result, the whole inhibitor molecule rotates ∼120◦ so that the urea moiety of compound 5 forms hydrogen bonds to the backbone amide of Met438 in the hinge region (Figure 4B). The amide carbonyl group is stabilized by a hydrogen bond with the conserved Lys391 side chain. The naphthyl ring, tilted by 37◦ relative to the pyrazole, makes a van der Waals contact with Ile369 side chain of the glycine-rich loop and is partially exposed to solvent, suffering a high desolvation penalty (Figure 4C). Nonequivalence of the energetics of the two binding sites is also supported by differences in the B-factor for the two binding modes of compound 5. The average B-factor of the bound inhibitor at the ATP site is 1.5-fold higher than that of the inhibitor at the  c The Authors Journal compilation  c 2014 Biochemical Society

allosteric pocket (Supplementary Table S1). The higher B-factors and the exposed non-polar surface of compound 5 at the ATP site suggest that the ATP site for compound 5 is a lower affinity site. Structure-based optimization of the allosteric inhibitor

To improve the potency further, a series of analogues of compound 5 were synthesized by elongating the alkoxy chain to ethoxy (compound 7), propoxy (compound 8) or isopropoxy (compound 9) moieties (Scheme 1). The ethoxy and propoxy analogues are more potent in biochemical assays using the ITK-KD with an IC50 of 0.04 μM for compound 7 and 0.02 μM for compound 8 at 50 μM ATP respectively. In the crystal structure of compound 5 bound to ITK, the methoxy moiety makes van der Waals contacts with the glycine-rich loop in the ATP-binding pocket, suggesting that the longer alkyl group might lead to a steric clash with the glycine-rich loop. As predicted, we see no evidence of binding of compounds 7 and 8 at the ATP site. Instead, these analogues bind only at the allosteric binding site, adopting similar binding modes as compounds 1 and 5. In the allosteric pocket, the Met411 side chain swings out toward solvent to avoid steric clash with the bulkier alkyl group (Figure 4D). The mouth of the hydrophobic channel is concurrently closed by the side chain movement of Trp356 . Furthermore, in the compound 8-bound structure, Val424

Allosteric inhibitors of ITK

Figure 4

217

Crystal structures of ITK complexed with compounds 5, 7 and 8

(A) Compound 5 (orange) binding in allosteric pocket. Upon compound 5 binding, Met411 and Trp356 in the compound 1 complex (green) undergo significant conformational changes shown in magenta. Molecular surface of the ITK in complex with compound 1 is shown in light blue. (B) Compound 5 interaction in the hinge region of ATP site, named as ‘pose 2’. (C) Molecular surface of ITK in complex with compound 5 in ATP site showing the van der Waals contact with Ile369 . (D) Details of the interactions of the ethoxy 7 (pink) and propoxy 8 (blue) in the allosteric site of the ITK (stereo). The Met411 and Trp356 in the compound 1-bound structure are shown in grey. (E) Comparison of inhibitor binding affinities (as measured by SPR) for the allosteric site (orange bars) and ATP site (blue bars) as a function of the alkoxy moiety at position 7.

adopts a different side chain rotamer to avoid steric clashes with the bulky propoxy moiety (Figure 4D). Biophysical studies of ITK-KD by SPR show that the bulky alkyl moieties at the 7 position have little effect on their binding affinities to the ATP site (Figure 4E). In contrast, they progressively improve the binding affinities to the allosteric site. In addition, binding studies with the inactive ITK-FL show a similar trend in the binding affinities of the 7 position alkoxy ligands for the allosteric and ATP sites (Table 1). These data suggest that the size and shape of the allosteric pockets are similar in the inactive ITK-FL and the ITK-KD. The improved potency that we observed in 7 position analogues prompted us to design compound 9 wherein the propoxy moiety of compound 8 was replaced with an isopropoxy moiety (Scheme 1). Compound 9 inhibits ITK-KD with an IC50 of 20 nM even at 1 mM ATP. The crystal structure shows that it is deeply buried in the allosteric hydrophobic pocket and the isopropoxy moiety is involved in hydrophobic interactions with Trp356 , Leu421 , Ala407 , Val424 and Met411 (Figure 5). Unexpectedly, we also observed weak and discontinuous electron density for compound 9 at the ATP site. This suggests that compound 9 retains some binding affinity for the ATP site. However, it is likely to be a weaker binding site as suggested by the higher B-factor of the ligand (2.2-fold higher compared with the allosteric binding site) (Supplementary Table S1). The Ser371 backbone flanked by two glycine residues

Figure 5 Details of the interactions of compound 9 (magenta) in the ITK allosteric pocket and ATP site All of the residues mentioned in the text are labelled and shown in stick rendering (cyan).

in the glycine-rich loop is pushed away from the ATP site by approximately 1.0 Å to fit the bulky and rigid isopropoxy moiety. When compound 9 is complexed with ITK in the presence of ADP, the electron density unambiguously reveals concurrent  c The Authors Journal compilation  c 2014 Biochemical Society

218

S. Han and others

Table 2 LANCE® IC50 and steady-state inhibition constants for the allosteric inhibitors for ITK-KD Compound

K is * (μM)

K ii * (μM)

Inhibition pattern

LANCE® IC50 † (μM)

9 8 7 5 1

0.057 + − 0.011 0.057 + − 0.009 0.102 + − 0.015 1.194 + − 0.097 0.464 + − 0.120

0.061 + − 0.008 0.083 + − 0.012 0.098 + − 0.009 1.14 + − 0.08 1.72 + − 1.07

Non-competitive Non-competitive Non-competitive Non-competitive Non-competitive‡

0.03 0.03 0.04 0.2 0.17

*K is : the inhibition constant for inhibitor binding to free enzyme; K ii: the inhibition constant for inhibitor binding to the enzyme–substrate complex. † ATP concentration set at the ITK-KD apparent K m , 50 μM. ‡ Linear intersecting non-competitive inhibition.

binding of ADP and compound 9 (Supplementary Figure S2 at http://www.biochemj.org/bj/460/bj4600211add.htm). The ADP molecule is bound in a location comparable with that seen in other ATP-bound protein kinases, whereas compound 9 remains bound in the adjacent allosteric binding pocket separated by a gatekeeper residue, Phe435 (Figure 2). As shown in Table 2, increasing alkoxy side chain length progressively leads to an improvement in potency with a maximal effect of 5-fold for compound 9. Kinase selectivity of ITK allosteric inhibitors

The kinase selectivity of ITK inhibitors was evaluated by profiling them in functional assays against a subset of the human kinome available at Invitrogen. At a concentration of 10 μM (comparable with the potency of compound 1 for ITK), compound 1 shows >50 % inhibition of 12 kinases out of the 39 kinases that were tested (Supplementary Table S3 at http://www.biochemj.org/bj/460/bj4600211add.htm). The small methoxy substitution at the 7 position of the naphthalene (compound 5) had no effect on the kinase selectivity profile. However, incorporation of longer alkyl ethers (compounds 7, 8 and 9) progressively improved the kinome selectivity with compound 9 inhibiting (>50 %) only three kinases out of 39 (TrkA, Abl and Aurora-A). Further evaluation showed that compound 9 has an IC50 of 4.0, 20.0 and 9.7 μM for Trk-A, Abl and Aurora-A respectively. To confirm the kinome selectivity of compound 9, we also profiled it in the broad DiscoveRx KinomeScan competitive binding assay that included 451 kinases [29]. At a concentration of 10 μM, compound 9 inhibited only three kinases >65 %. It showed significant binding to the pseudokinase (JH2) domain of TYK2, DDR1 (discoidin domain receptor tyrosine kinase 1) and the non-phosphorylated ABL1 H396P mutant (Figure 6A and Supplementary Table S4 at http://www.biochemj.org/ bj/460/bj4600211add.htm). Interestingly, compound 9 was completely ineffective in displacing a competitive probe from the ATP site of ITK in the DiscoveRx panel. This observation further supports our conclusion that binding of compound 9 at the allosteric site is primarily responsible for its inhibition of ITK functional activity. In parallel, the selectivity of compound 9 was further evaluated in a broad kinase panel using biochemical assays. We screened against 284 kinases in the Carna Biosciences panel at a concentration of 10 μM of compound 9. Consistent with the results from the DiscoverRx KinomeScan, compound 9 showed >65 % inhibition of only three kinases (TrkA, TrkB  c The Authors Journal compilation  c 2014 Biochemical Society

and TrkC) apart from ITK (Figure 6B and Supplementary Table S5 at http://www.biochemj.org/bj/460/bj4600211add.htm). Furthermore, compound 9 shows exquisite selectivity against other members of the TEC subfamily (BMX, BTK, TEC and TXK, Supplementary Table S6 at http://www.biochemj.org/ bj/460/bj4600211add.htm). Mechanism of inhibition

In order to gain further insights into the mechanism of inhibition, we applied steady-state kinetic methods. When profiled using ITK-KD, compound 1 and all other analogues displayed noncompetitive inhibition with respect to ATP (Table 2). Secondary plots of the intercept against inhibitor concentration show a slight parabolic nature at higher inhibitor concentration (>5× K is ), suggestive of higher order stoichiometry. This is consistent with initial binding at the allosteric site, followed by binding at the ATP site as the concentration of the ligand is increased. The inhibition constants for free enzyme (K is ) are comparable with those of the enzyme–substrate complex (K ii ), which suggests that binding at the allosteric pocket is sufficient for inhibition of the kinase activity. However, when evaluated with ITK-FL, the inhibition kinetics are more complex with non-linearity observed in the reciprocal plots at high ATP concentration (>ATP K m ) and in the secondary intercept plots at high inhibitor concentration (>5× K is ) (Figure 7). Mixed-mode inhibition patterns are observed for compounds 1 and 9 consistent with linear intersecting non-competitive inhibition with respect to ATP (Figure 7) [30]. In contrast, competitive inhibition was observed for compounds 5, 7 and 8 for ITK-FL (Supplementary Table S7 at http://www.biochemj.org/bj/460/bj4600211add.htm). To determine whether the phosphorylation state affected the inhibition pattern, ITK-FL was pre-incubated with ATP prior to the steady-state kinetics experiments. We observed a decrease in the apparent inhibitor potencies, although the overall inhibition pattern was maintained (Supplementary Table S7). Allosteric inhibition is dependent on phosphorylation state and regulatory domains

Since ITK-FL has a high binding affinity for ATP (apparent K m = 5 μM), we routinely screened our compounds at low and high ATP concentrations in order to get a realistic estimate of their cellular potencies and mechanism of inhibition. Based on the Cheng–Prusoff equation, we expect a right shift of ∼50-fold in potency for a competitive inhibitor in going from 5 μM to 1 mM ATP, whereas comparable potencies are predicted for a non-competitive inhibitor. However, we observed a 10-fold shift in the potency of the allosteric compounds for non-activated ITK-FL in the LANCE® endpoint assay (Supplementary Table S7). We suspected that this apparent discrepancy could be due to changes in the allosteric site brought about by autophosphorylation of ITK during the course of our experiment. To evaluate this, we next monitored the effect of the inhibitors on the lag phase of ITK-FL auto-phosphorylation and activation. The lag phase of non-activated ITK-FL increased with increasing inhibitor concentration at both low (0.02 mM) and high (0.2 mM) ATP concentrations (Figures 8A and 8B respectively). We used the DynaFit program to fit the progress curves to the integrated rate equations for the model that includes inhibition of non-activated ITK-FL (see the Supplementary Materials and methods at http://www.biochemj.org/bj/460/bj4600211add.htm) to determine the inhibition constant for the onset of activity,

Allosteric inhibitors of ITK

Figure 6

219

Kinome selectivity of compound 9

(A) KinomeScan selectivity tree for compound 9; data provided in Supplementary Table S4 (at http://www.biochemj.org/bj/460/bj4600211add.htm). Of 451 profiled kinases, only TYK2, DDR1 and ABL1 H369P mutant demonstrated >65 % probe displacement at 10 μM ligand, indicated by the red circles. (B) Kinome selectivity tree for compound 9 from Carna Biosciences; data provided in Supplementary Table S5 (at http://www.biochemj.org/bj/460/bj4600211add.htm). Of 284 profiled kinases, only ITK, TrkA, TrkB and TrkC demonstrated >60 % inhibition at 10 μM, indicated by the red circles. Images were generated using software from Carna Biosciences and a TREEspotTM Software Tool from DiscoveRx Corporation and reprinted with permission from KINOMEscan.

K i autoact . The resulting K i autoact values are similar to the K is and K ii values obtained for the ITK-KD (Figure 8C and Supplementary Table S8 at http://www.biochemj.org/bj/460/bj4600211add.htm), implying that the allosteric site in the non-activated ITKFL resembles that in the ITK-KD. The K is values for the inhibition of phosphorylated ITK-FL are approximately 6-fold weaker compared with that of non-activated ITK-FL (Figure 8D). These results suggest that the phosphorylation of ITK-FL distorts the allosteric site, leading to a suboptimal binding site. Similar potency changes that are dependent upon the precise kinase phosphorylation state have also been reported for the MEK1 (mitogen-activated protein kinase/extracellular-signal-

regulated kinase kinase 1) allosteric inhibitors PD0325901 and U0126 [31].

Cellular potencies of allosteric ITK inhibitors

To evaluate the biological activity of the ITK inhibitors, compounds were profiled in a whole cell assay (IP-One) that measures the activation of phospholipase Cγ 1 in human Jurkat T-cells stimulated with an anti-CD3 antibody. Compound 9 has an IC50 of 5.1 μM in this assay. In comparison, the ATP-competitive control compound BMS-509744 has an IC50  c The Authors Journal compilation  c 2014 Biochemical Society

220

Figure 7

S. Han and others

Mechanism of inhibition of compound 1 and compound 9 in ITK-FL

(A) Double reciprocal plots of the velocity against ATP concentration for inhibition of 25 nM ITK-FL by compound 1 at four fixed inhibitor concentrations: 0 μM (䊊), 1.25 μM (䊉), 2.5 μM (䊐) and 5.0 μM (䊏). (B) Compound 9 inhibition of ITK-FL. Concentrations used: 0 μM (䊊), 0.05 μM (䊉), 0.10 μM (䊐), 0.20 μM (䊏), 0.40 μM () and 0.80 μM (䉱).

of 1.5 μM. The cellular potency of the allosteric compounds in general correlates with the IC50 values obtained in the LANCE® assay with ITK-FL (Supplementary Figure S3 at http://www.biochemj.org/bj/460/bj4600211add.htm). Additionally, compound 9 was profiled in human whole blood stimulated by the antigens anti-CD3 and anti-CD28 for inhibition of IL2. It has an IC50 of 5.0 μM in the whole blood assay, whereas BMS-509744 has an IC50 of 1.1 μM. These studies demonstrate the importance of ITK in the signalling events that lead to the production of pro-inflammatory cytokines and the utility of compound 9 as a starting point for further optimization of the allosteric ITK inhibitors. DISCUSSION

Over the last decade, the pharmaceutical industry has been highly successful in developing protein kinase inhibitors as therapies for several diseases. To date, we have seen the emergence of ∼25 small molecule kinase inhibitor drugs in the marketplace. All of the approved kinase drugs target the highly conserved ATP site of kinases. This raises a potential toxicity concern because of off-target activities of the compounds. Although the selectivity consideration may not be as critical for oncology indications, and polypharmacology in fact could be even an advantage in some cases, development of kinase inhibitors for chronic indications demands exquisite kinome selectivity. Hence, there has been continued interest in discovering allosteric kinase inhibitors which potentially could be more specific for a target of interest. Moreover, allosteric kinase inhibitors, which are noncompetitive or uncompetitive with ATP, are also likely to have inherently higher biochemical efficiencies because they do not  c The Authors Journal compilation  c 2014 Biochemical Society

have to overcome the high cellular ATP concentration. There are several examples of known allosteric kinase inhibitors that target Akt, PDK1 (phosphoinositide-dependent kinase 1), JNK1 (c-Jun N-terminal kinase 1), MEK and BCR-ABL1 [32]. In these cases, investigators have been able to exploit highly druggable binding sites outside the ATP pocket irrespective of whether they are located on the kinase domain or elsewhere. In general, discovery and optimization of allosteric inhibitors is a challenging task. This problem is even more acute for kinase targets because of a number of additional factors including the construct used in the assays (i.e. kinase domain compared with full-length), the activation or phosphorylation state of the kinase, and the specific assay format. Our initial ITK lead (compound 1) is a dual-site binder that simultaneously binds at the ATP site and an adjacent allosteric site. Hence, endpoint biochemical assays were inadequate for complete mechanistic understanding and lead optimization. We established a battery of biophysical assays such as SPR, NMR and ITC in order to deconvolute the binding affinities at the two sites. In addition, we also brought forward steady-state kinetics and progress curve analysis in order to assess the potency of compounds for different activation states of ITK. The compounds described in the present study are the first reported ITK inhibitors that bind to and stabilize an inactive conformation of ITK through occupation of an allosteric pocket. A computational approach called ‘SiteMap’ allowed us to assess the druggability of the allosteric and ATP sites of ITK. The druggability score called ‘Dscore’ includes terms such as number of site points found for the site, degree of enclosure of the site and the hydrophilic contributions from the site. An average threshold value of DScore>1.1 has been suggested as a good indicator for a druggable site [26]. The allosteric site with a DScore of 1.2 could be classified as a potential druggable site and compares favourably with that of the ATP site score of 1.1. More importantly, the hydrophobic contribution, which is perhaps the most important driver for potency, was found to be ∼2-fold higher for the allosteric site (Supplementary Table S2). The crystallography data show that compound 9 binds to a novel allosteric pocket in ITK adjacent to but non-overlapping with the ATP-binding site. In the allosteric pocket, the binding of the ligand is dominated by hydrophobic interactions formed by the naphthyl ring of the inhibitor. Additionally, the urea and the amide moieties of the inhibitor form specific hydrogen bonding interactions. Importantly, the ITK crystal structure with compound 1 revealed a distinct hydrophobic tunnel exposing the C7 atom of the naphthyl ring. This unique tunnel is sufficiently wide to accommodate alkyl derivatives at the 7 position of the naphthyl moiety and provides an avenue for optimization of both selectivity and potency. The alkyl ether substitution at the 7 position promotes hydrophobic interaction in the allosteric pocket. In contrast, the allosteric inhibitors show two distinct binding modes (‘pose 1’ and ‘pose 2’) at the ATP-binding site, depending on the precise alkyl substitution at the 7 position of the naphthyl ring. Evaluation of binding affinities by SPR and NMR show that the ATP site is a lower-affinity site for compound 9, consistent with its weak electron density at the ATP site. Previously, Ohren et al. [33] had reported allosteric inhibitors of MEK1 and MEK2 exemplified by PD-318088. These inhibitors are non-competitive with MgATP and bind at an allosteric binding site of MEK1 and MEK2. They stabilize an inactive conformation of MEKs that not only inhibit the kinase activity of MEKs, but also prevent their activation by upstream kinases. An overlay of the crystal structures of MEK1 and ITK (Supplementary Figure S4 at http://www.biochemj.org/bj/460/bj4600211add.htm) shows that the allosteric sites in these two kinases do not overlap exactly, although the C-helix is pushed out in both cases.

Allosteric inhibitors of ITK

Figure 8

221

Evaluation of auto-activation and kinetic parameters for ITK-FL and ITK-KD

(A) Effect of compound 9 on ITK-FL activation at 0.02 mM ATP. (B) Effect of compound 9 on ITK-FL activation at 0.2 mM ATP. The lag phase is longer with increasing inhibitor concentration. Comparison of inhibition constant K is values for inactive and active ITK proteins as a function of the alkoxy substituent at position 7. (C) Comparison of K is for ITK-KD (blue) and K i autoact for auto-activation of inactive ITK-FL (red). (D) Comparison of K i autoact for inactive ITK-FL auto-activation (red) and K is for activated ITK-FL (pFL, green).

There have been several reports of protein kinase inhibitors that contain the dual functionalities of urea and carboxamide attached to a small heterocyclic scaffold. For example, very potent IKK-β [IκB (inhibitor of nuclear factor κB) kinase β] inhibitors containing a pyrazole or thiophene core have been reported [34]. These inhibitors typically bind at the ATP site of kinases in potentially two different poses as shown in Figures 3(C) and 4(B). The network of hydrogen bonding interactions formed by the urea and the carboxamide moieties at the ATP site of kinases allow these inhibitors to bind a large number of protein kinases. As shown in Supplementary Table S3, compound 1 is a promiscuous kinase inhibitor that shows >50 % inhibition of ∼30 % of the kinases in the Invitrogen mini kinase panel. It is likely that in the majority of these cases, the inhibition of protein kinase activity is due to competitive displacement of ATP from the hinge region of kinases. However, in the case of ITK, the measured inhibitory potency of compound 1 is a composite of its binding affinities at the ATP site and the allosteric site as shown earlier. In contrast, compound 9 is an exquisitely selective protein kinase inhibitor as shown by selectivity profiling in binding assays as well as by functional assays. The incorporation of a branched side chain ether at the 7 position of the naphthyl group results in significant attenuation of the binding affinity of this scaffold to the vast majority of kinases. Most likely, this is caused by the steric clash of the isopropyl substituent with the glycine loop of kinases at the ATP site. All of the compounds described in the present study bind to the allosteric site of the ITK-KD and ITK-FL with a higher affinity compared with the ATP site (Table 1 and Supplementary Figure S1). Evaluation of steady-state kinetics shows that all of the inhibitors are non-competitive with MgATP when profiled against the kinase domain of ITK. In contrast, mixed-mode or competitive inhibition is observed against activated ITK-FL. This apparent

complication arises because of a multitude of factors. The ATPbinding affinities of ITK-KD and ITK-FL are different. Although the kinase domain is homogenous in terms of its activation state, the full-length enzyme is heterogeneously phosphorylated during the time course of the assay. The isolated kinase domain has less intrinsic kinase activity compared with the full-length enzyme [27,35]. Finally, the size and the shape of the allosteric site are dictated by the phosphorylation state of ITK-FL. As a consequence, differential binding affinities for the allosteric inhibitors are observed (Supplementary Tables S7 and S8 at http://www.biochemj.org/bj/460/bj4600211add.htm). Using the crystal structure of the isolated ITK-KD, we were able to generate compounds that potently inhibited the ITK-KD in enzymatic assays, but were less potent against the activated full-length enzyme. As described above, the precise size and shape of the allosteric site is not only dependent on the presence of regulatory domains (SH2 and SH3), but also on the phosphorylation state of the enzyme. The compounds that we have discovered are more potent for the inactive form of ITK (KD or FL) compared with the activated enzyme. In conclusion, we report the first known allosteric inhibitors of ITK, an important target for autoimmune diseases. By the application of a battery of biophysical and biochemical assays, we have fully characterized the allosteric binding site of ITK. Through structure-based drug design, we have transformed a promiscuous ATP site kinase inhibitor scaffold into a highly selective allosteric inhibitor of ITK that is non-competitive with ATP and highly selective for the inactive form of ITK. The optimized compound (compound 9) inhibits cytokine production in human whole blood and Jurkat T-cells and holds excellent promise for further optimization of ADME (adsorption, distribution, metabolism and excretion) and physico-chemical properties.  c The Authors Journal compilation  c 2014 Biochemical Society

222

S. Han and others

AUTHOR CONTRIBUTION Seungil Han, Nicole Caspers, Jeffrey Ohren, Francis Rajamohan, Andrew Cook, Colin Robinson, Chris Phillips and Ravi Kurumbail were involved in the crystallographic studies. Robert Czerwinski and Ann Aulabaugh performed the enzyme kinetics studies and analysis. David Limburg, Chulho Choi, Mihir Parikh, John Mathias and David Anderson were involved in the chemical synthesis of the compounds described. Ray Unwalla did the computational and druggability analysis. WeiDong Ding, Colin Robinson, Samantha Hughes, Iva Navratilova and Ann Aulabaugh carried out SPR studies and analysis. Hong Wang was responsible for the NMR studies. Thomas McLellan was involved in the mass spectrometric characterization of protein reagents. Ann Aulabaugh carried out the isothermal titration calorimetry studies and analysis. Nilufer Seth, Jason Edmonds, Subarna Shakya, Xin Li, Vikki Spaulding and Quintus Medley were responsible for biochemical and cell-based assays. Seungil Han, Ann Aulabaugh and Ravi Kurumbail wrote the paper.

ACKNOWLEDGEMENTS We thank Victoria Wong, Marie Anderson, Kerry Kelleher, Jefry Shields and Laura Lin for technical support and Kieran Geoghegan, Suvit Thaisrivongs and Xiayang Qiu for insightful discussions before submission.

FUNDING This work was supported by internal R&D funding within Pfizer.

REFERENCES 1 Berg, L. J., Finkelstein, L. D., Lucas, J. A. and Schwartzberg, P. L. (2005) Tec family kinases in T lymphocyte development and function. Annu. Rev. Immunol. 23, 549–600 CrossRef PubMed 2 Prince, A. L., Yin, C. C., Enos, M. E., Felices, M. and Berg, L. J. (2009) The Tec kinases Itk and Rlk regulate conventional versus innate T-cell development. Immunol. Rev. 228, 115–131 CrossRef PubMed 3 Andreotti, A. H., Schwartzberg, P. L., Joseph, R. E. and Berg, L. J. (2010) T-cell signaling regulated by the Tec family kinase, Itk. Cold Spring Harb. Perspect. Biol. 2, a002287 CrossRef PubMed 4 Fowell, D. J., Shinkai, K., Liao, X. C., Beebe, A. M., Coffman, R. L., Littman, D. R. and Locksley, R. M. (1999) Impaired NFATc translocation and failure of Th2 development in Itk-deficient CD4 + T cells. Immunity 11, 399–409 CrossRef PubMed 5 Au-Yeung, B. B., Katzman, S. D. and Fowell, D. J. (2006) Cutting edge: Itk-dependent signals required for CD4 + T cells to exert, but not gain, Th2 effector function. J. Immunol. 176, 3895–3899 PubMed 6 Sahu, N., Mueller, C., Fischer, A. and August, A. (2008) Differential sensitivity to Itk kinase signals for T helper 2 cytokine production and chemokine-mediated migration. J. Immunol. 180, 3833–3838 PubMed 7 Gomez-Rodriguez, J., Sahu, N., Handon, R., Davidson, T. S., Anderson, S. M., Kirby, M. R., August, A. and Schwartzberg, P. L. (2009) Differential expression of interleukin-17A and -17F is coupled to T cell receptor signaling via inducible T cell kinase. Immunity 31, 587–597 CrossRef PubMed 8 Finkelstein, L. D., Shimizu, Y. and Schwartzberg, P. L. (2005) Tec kinases regulate TCR-mediated recruitment of signaling molecules and integrin-dependent cell adhesion. J. Immunol. 175, 5923–5930 PubMed 9 Takesono, A., Horai, R., Mandai, M., Dombroski, D. and Schwartzberg, P. L. (2004) Requirement for Tec kinases in chemokine-induced migration and activation of Cdc42 and Rac. Curr. Biol. 14, 917–922 CrossRef PubMed 10 Sahu, N. and August, A. (2009) ITK inhibitors in inflammation and immune-mediated disorders. Curr. Top. Med. Chem. 9, 690–703 CrossRef PubMed 11 Heyeck, S. D., Wilcox, H. M., Bunnell, S. C. and Berg, L. J. (1997) Lck phosphorylates the activation loop tyrosine of the Itk kinase domain and activates Itk kinase activity. J. Biol. Chem. 272, 25401–25408 CrossRef PubMed 12 Wilcox, H. M. and Berg, L. J. (2003) Itk phosphorylation sites are required for functional activity in primary T cells. J. Biol. Chem. 278, 37112–37121 CrossRef PubMed 13 Eglen, R. M. and Reisine, T. (2010) Human kinome drug discovery and the emerging importance of atypical allosteric inhibitors. Expert Opin. Drug Discov. 5, 277–290 CrossRef PubMed Received 26 August 2013/28 February 2014; accepted 4 March 2014 Published as BJ Immediate Publication 4 March 2014, doi:10.1042/BJ20131139

 c The Authors Journal compilation  c 2014 Biochemical Society

14 Harrison, S., Das, K., Karim, F., Maclean, D. and Mendel, D. (2008) Non-ATP-competitive kinase inhibitors – enhancing selectivity through new inhibition strategies. Expert Opin. Drug Discov. 3, 761–774 CrossRef PubMed 15 Bamborough, P., Brown, M. J., Christopher, J. A., Chung, C. W. and Mellor, G. W. (2011) Selectivity of kinase inhibitor fragments. J. Med. Chem. 54, 5131–5143 CrossRef PubMed 16 Navratilova, I., Macdonald, G., Robinson, C., Hughes, S., Mathias, J., Phillips, C. and Cook, A. (2011) Biosensor-based approach to the identification of protein kinase ligands with dual-site modes of action. J. Biomol. Screen. 17, 183–193 CrossRef PubMed 17 Otwinowski, Z. and Minor, W. (1997) Processing of X-ray diffraction data collected in oscillation mode. Methods Enzymol. 276, 307–326 CrossRef 18 McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. and Read, R. J. (2007) Phaser crystallographic software. J. Appl. Cystal. 40, 658–674 CrossRef 19 Blanc, E., Roversi, P., Vonrhein, C., Flensburg, C., Lea, S. M. and Bricogne, G. (2004) Refinement of severely incomplete structures with maximum likelihood in BUSTER-TNT. Acta Crystallogr. D Biol. Crystallogr. 60, 2210–2221 CrossRef PubMed 20 Emsley, P. and Cowtan, K. (2004) Coot: model-building tools for molecular graphics. Acta Crystallogr. D Biol. Crystallogr. 60, 2126–2132 CrossRef PubMed 21 Laskowski, R. A., Moss, D. S. and Thornton, J. M. (1993) Main-chain bond lengths and bond angles in protein structures. J. Mol. Biol. 231, 1049–1067 CrossRef PubMed 22 Czerwinski, R., Aulabaugh, A., Greco, R. M., Olland, S., Malakian, K., Wolfrom, S., Lin, L., Kriz, R., Stahl, M., Huang, Y. et al. (2005) Characterization of protein kinase C theta activation loop autophosphorylation and the kinase domain catalytic mechanism. Biochemistry 44, 9563–9573 CrossRef PubMed 23 Das, J., Furch, J. A., Liu, C., Moquin, R. V., Lin, J., Spergel, S. H., McIntyre, K. W., Shuster, D. J., O’Day, K. D., Penhallow, B. et al. (2006) Discovery and SAR of 2-amino-5-(thioaryl)thiazoles as potent and selective Itk inhibitors. Bioorg. Med. Chem. Lett. 16, 3706–3712 CrossRef PubMed 24 Kutach, A. K., Villasenor, A. G., Lam, D., Belunis, C., Janson, C., Lok, S., Hong, L. N., Liu, C. M., Deval, J., Novak, T. J. et al. (2010) Crystal structures of IL-2-inducible T cell kinase complexed with inhibitors: insights into rational drug design and activity regulation. Chem. Biol. Drug Des. 76, 154–163 CrossRef PubMed 25 Brown, K., Long, J. M., Vial, S. C., Dedi, N., Dunster, N. J., Renwick, S. B., Tanner, A. J., Frantz, J. D., Fleming, M. A. and Cheetham, G. M. (2004) Crystal structures of interleukin-2 tyrosine kinase and their implications for the design of selective inhibitors. J. Biol. Chem. 279, 18727–18732 CrossRef PubMed 26 Halgren, T. A. (2009) Identifying and characterizing binding sites and assessing druggability. J. Chem. Inf. Model. 49, 377–389 CrossRef PubMed 27 Joseph, R. E., Min, L. and Andreotti, A. H. (2007) The linker between SH2 and kinase domains positively regulates catalysis of the Tec family kinases. Biochemistry 46, 5455–5462 CrossRef PubMed 28 Guo, S., Wahl, M. I. and Witte, O. N. (2006) Mutational analysis of the SH2-kinase linker region of Bruton’s tyrosine kinase defines alternative modes of regulation for cytoplasmic tyrosine kinase families. Int. Immunol. 18, 79–87 CrossRef PubMed 29 Fabian, M. A., Biggs, 3rd, W. H., Treiber, D. K., Atteridge, C. E., Azimioara, M. D., Benedetti, M. G., Carter, T. A., Ciceri, P., Edeen, P. T., Floyd, M. et al. (2005) A small molecule-kinase interaction map for clinical kinase inhibitors. Nat. Biotechnol. 23, 329–336 CrossRef PubMed 30 Segel, I. H. (1975) In Enzyme Kinetics: Behavior and Analysis of Rapid Equilibrium and Steady-state Enzyme Systems, Wiley Interscience 31 Sheth, P. R., Liu, Y., Hesson, T., Zhao, J., Vilenchik, L., Liu, Y. H., Mayhood, T. W. and Le, H. V. (2011) Fully activated MEK1 exhibits compromised affinity for binding of allosteric inhibitors U0126 and PD0325901. Biochemistry 50, 7964–7976 CrossRef PubMed 32 Gavrin, L. K. and Saiah, E. (2013) Approaches to discover non-ATP site kinase inhibitors. MedChemComm 4, 41–51 CrossRef 33 Ohren, J. F., Chen, H., Pavlovsky, A., Whitehead, C., Zhang, E., Kuffa, P., Yan, C., McConnell, P., Spessard, C., Banotai, C. et al. (2004) Structures of human MAP kinase kinase 1 (MEK1) and MEK2 describe novel noncompetitive kinase inhibition. Nat. Struct. Mol. Biol. 11, 1192–1197 CrossRef PubMed 34 Karin, M., Yamamoto, Y. and Wang, Q. M. (2004) The IKK NF-κB system: a treasure trove for drug development. Nat. Rev. Drug Discov. 3, 17–26 CrossRef PubMed 35 Hawkins, J. and Marcy, A. (2001) Characterization of Itk tyrosine kinase: contribution of noncatalytic domains to enzymatic activity. Protein Expr. Purif. 22, 211–219 CrossRef PubMed

Biochem. J. (2014) 460, 211–222 (Printed in Great Britain)

doi:10.1042/BJ20131139

SUPPLEMENTARY ONLINE DATA

Selectively targeting an inactive conformation of interleukin-2-inducible T-cell kinase by allosteric inhibitors Seungil HAN*1 , Robert M. CZERWINSKI†, Nicole L. CASPERS*, David C. LIMBURG*, WeiDong DING*, Hong WANG*, Jeffrey F. OHREN*, Francis RAJAMOHAN*, Thomas J. MCLELLAN*, Ray UNWALLA*, Chulho CHOI*, Mihir D. PARIKH*, Nilufer SETH†, Jason EDMONDS†, Chris PHILLIPS*, Subarna SHAKYA†, Xin LI†, Vikki SPAULDING†, Samantha HUGHES*, Andrew COOK*, Colin ROBINSON*, John P. MATHIAS†, Iva NAVRATILOVA†*, Quintus G. MEDLEY†, David R. ANDERSON*, Ravi G. KURUMBAIL*1 and Ann AULABAUGH*1 *Pfizer Worldwide Research, Eastern Point Road, Groton, CT 06340, U.S.A. †Pfizer Worldwide Research, 200 Cambridge Park Drive, Cambridge, MA 02140, U.S.A.

The Supplementary Materials and Methods section is available online at http://www.biochemj.org/bj/460/bj4600211add.htm

Figure S1

Typical ITC curves of ITK-KD348–620 with different inhibitors

(A) ITK-KD348–620 and the ATP-site binder BMS-509744, (B) ITK-KD348–620 and the dual-site inhibitor 1 and (C) titration of compound 1 into the ITK-KD348–620 /BMS-509744 complex. The cell contained 8.5 μM ITK-KD348–620 in the assay buffer, and the syringe contained 100 or 200 μM of compound in the same buffer. Top panels: the raw power traces after baseline correction. Bottom panels: data after peak integrations and concentration normalization. The continuous line in the bottom panels of (A) and (C) is the least-squares fit of the data to a single binding site model, and the continuous line in the bottom panel of (B) is the least-squares fit of the data to a two-site binding model.

1 Correspondence may be addressed to these authors (email [email protected], [email protected] or [email protected]). The structural co-ordinates reported for the protein structures have been deposited in the PDB under the accession codes 4M0Y, 4M0Z, 4M12, 4M13, 4M14 and 4M15.

 c The Authors Journal compilation  c 2014 Biochemical Society

S. Han and others

Figure S3 Allosteric inhibitors block CD3-induced signalling and IP-1 production in Jurkat cells Data are shown as the mean of duplicate measurements, and fit to the following equation % inh = (% inh − offset) × x n /(IC50 n + [cmpd]n ) + offset, where n is the Hill coefficient.

Figure S4

Allosteric site comparison between ITK and MEK1

Compound 9 in ITK structure is shown in magenta and PD318 088 in teal (PDB code 1S9J).

Figure S2 at 3σ

The (F o − F c ) omit maps at ATP and allosteric sites, contoured

(A) Compound 1 coloured in cyan, (B) compound 5 in orange, (C) compound 7 in pink, (D) compound 8 in navy, (E) compound 9 in magenta and (F) compound 9 in magenta and ADP in yellow.

 c The Authors Journal compilation  c 2014 Biochemical Society

Allosteric inhibitors of ITK Table S1

X-ray crystallographic data collection and refinement statistics

Each dataset was collected from one crystal. Numbers in parentheses represent the highest resolution shell. n.a., not applicable.

PDB code Data collection Space group Cell dimensions a, b, c (A˚) α, β, γ (◦ ) Resolution (A˚) R merge I/σ I Completeness (%) Redundancy Refinement Resolution (A˚) Number of reflections R work /R free (%) Number of non-H atoms Protein Ligand/ion Water B -factor (A˚2 ) Protein Ligand (ATP/allosteric site) Water RMSDs Bond lengths (A˚) Bond angles (◦ )

ITK–compound 1

ITK–compound 5

ITK–compound 7

ITK–compound 8

ITK–compound 9

ITK–compound 9 + ADP

4M0Y

4M0Z

4M12

4M13

4M14

4M15

P 21

P 21

P 21

P 21

P 21

P 21

41.1, 69.1, 50.0 90, 106.6, 90 1.7 (1.79–1.70) 0.09 (0.45) 10.8 (2.8) 98.9 (91.4) 3.1 (2.5)

40.4, 68.6, 49.6 90, 107.0, 90 2.0 (2.03–2.0) 0.08 (0.43) 10.3 (2.1) 99.3 (97.9) 3.2 (2.8)

40.2, 68.9, 49.5 90, 106.6, 90 2.15 (2.19–2.15) 0.10 (0.39) 10.1 (2.1) 98.6 (98.5) 2.8 (2.6)

40.5, 68.9, 49.5 90, 106.9, 90 1.85 (1.88–1.85) 0.06 (0.47) 18.3 (2.2) 95.9 (99.9) 3.5 (3.5)

40.4, 68.9, 49.5 90, 106.8, 90 1.55 (1.555–1.55) 0.05 (0.32) 15.6 (3.0) 95.4 (69.8) 3.5 (2.7)

40.6, 69.0, 49.4 90, 106.8, 90 1.52 (1.59–1.52) 0.05 (0.26) 15.8 (3.3) 96.0 (71.8) 3.4 (2.4)

30.0-1.7 28 868 19.0/21.7

19.3-2.0 17 467 16.4/22.1

18.6-2.15 13 896 18.3/22.8

19.8-1.85 21 364 18.7/23.2

35.4-1.55 35 998 16.5/18.9

39.0-1.52 38 324 17.2/18.6

2119 44 162

2128 48 209

2129 25 122

2129 26 189

2117 52 234

2117 53 202

23.1 28.4/15.7 33.4

23.8 32.5/22.0 33.1

31.2 n.a./24.1 35.0

30.2 n.a./23.9 38.6

22.0 32.3/14.7 35.0

22.2 33.0/14.5 34.5

0.010 0.94

0.010 1.04

0.010 1.07

0.010 1.01

0.010 1.05

0.01 1.01

Table S2 SiteMap analysis to predict the druggability of the allosteric and ATP sites of ITK Properties

ATP site

Allosteric site

SiteScore DScore Volume Enclosure Hydrophobic Hydrophillic

1.1 1.10 359 0.81 1.2 0.97

1.2 1.24 179 0.96 2.5 0.84

Table S3

Kinase selectivity of ITK allosteric inhibitors assessed by Invitrogen SelectScreen® Kinase Profiling

These ligands were profiled against 39 kinases at 10 μM concentration of ligands at Invitrogen using standard biochemical assays at an ATP concentration corresponding to their K m for ATP. Compound

Number of kinases inhibited >50 % at 10 μM concentration

Kinases inhibited >50 % inhibition at 10 μM concentration

1 5 7 8 9

12 12 5 4 3

EGFR, GSK-3b, MAP4K4, LCK, MST2, CDK2/cyclin A, TRKA, AURA, KDR, JAK3, SRC, ABL1 EGFR, GSK-3b, MET, MAP4K4, LCK, CDK2/cyclin A, TRKA, AURA, MST4, TA02, SRC, ABL1 MAP4K4, TRKA, AURA, SRC, ABL1 TRKA, AURA, SRC, ABL1 TRKA, AURA, ABL1

 c The Authors Journal compilation  c 2014 Biochemical Society

S. Han and others Table S4 Kinase selectivity of compound 9 as assessed by DiscoveRx KinomeScan

Table S4

Compound 9 was screened at 10 μM and the results for the primary screen binding interactions are reported as percentage of control binding, where lower numbers indicate stronger hits in the matrix.

Ambit gene symbol

Percentage of control binding

CDK11 CDK2 CDK3 CDK4-cyclinD1 CDK4-cyclinD3 CDK5 CDK7 CDK8 CDK9 CDKL1 CDKL2 CDKL3 CDKL5 CHEK1 CHEK2 CIT CLK1 CLK2 CLK3 CLK4 CSF1R CSF1R-autoinhibited CSK CSNK1A1 CSNK1A1L CSNK1D CSNK1E CSNK1G1 CSNK1G2 CSNK1G3 CSNK2A1 CSNK2A2 CTK DAPK1 DAPK2 ERN1 FAK FER FES FGFR1 FGFR2 FGFR3 FGFR3(G697C) FGFR4 FGR FLT1 FLT3 FLT3(D835H) FLT3(D835Y) FLT3(ITD) FLT3(K663Q) FLT3(N841I) FLT3(R834Q) FLT3-autoinhibited FLT4 FRK FYN GAK GCN2(Kin.Dom.2,S808G) GRK1 GRK4 GRK7 GSK3A GSK3B HASPIN HCK HIPK1 HIPK2 HIPK3

67 86 94 100 100 100 100 76 100 100 36 100 100 89 100 94 87 100 78 97 75 100 88 100 63 100 93 97 86 100 100 100 100 95 100 100 100 100 99 100 86 100 91 93 80 100 65 58 40 81 48 41 97 100 87 85 89 86 99 100 100 88 89 71 100 80 100 100 70

Ambit gene symbol

Percentage of control binding

AAK1 ABL1(E255K)-phosphorylated ABL1(F317I)-nonphosphorylated ABL1(F317I)-phosphorylated ABL1(F317L)-nonphosphorylated ABL1(F317L)-phosphorylated ABL1(H396P)-nonphosphorylated ABL1(H396P)-phosphorylated ABL1(M351T)-phosphorylated ABL1(Q252H)-nonphosphorylated ABL1(Q252H)-phosphorylated ABL1(T315I)-nonphosphorylated ABL1(T315I)-phosphorylated ABL1(Y253F)-phosphorylated ABL1-nonphosphorylated ABL1-phosphorylated ABL2 ACVR1 ACVR1B ACVR2A ACVR2B ACVRL1 ADCK3 ADCK4 AKT1 AKT2 AKT3 ALK AMPK-alpha1 AMPK-alpha2 ANKK1 ARK5 ASK1 ASK2 AURKA AURKB AURKC AXL BIKE BLK BMPR1A BMPR1B BMPR2 BMX BRAF BRAF(V600E) BRK BRSK1 BRSK2 BTK BUB1 CAMK1 CAMK1D CAMK1G CAMK2A CAMK2B CAMK2D CAMK2G CAMK4 CAMKK1 CAMKK2 CASK CDC2L1 CDC2L2 CDC2L5

99 70 91 100 71 100 16 81 70 56 100 36 100 67 36 76 66 100 80 100 100 90 100 95 100 93 100 100 84 100 71 100 76 97 100 64 58 78 67 70 86 100 100 100 100 100 92 100 100 100 100 87 89 100 100 100 100 100 100 100 100 92 91 100 100

 c The Authors Journal compilation  c 2014 Biochemical Society

Continued

Allosteric inhibitors of ITK Table S4

Continued

Table S4

Continued

Ambit gene symbol

Percentage of control binding

Ambit gene symbol

Percentage of control binding

HIPK4 HPK1 HUNK ICK IGF1R IKK-alpha IKK-beta IKK-epsilon INSR INSRR IRAK1 IRAK3 IRAK4 ITK JAK1(JH1domain-catalytic) JAK1(JH2domain-pseudokinase) JAK2(JH1domain-catalytic) MERTK MET MET(M1250T) MET(Y1235D) MINK MKK7 MKNK1 MKNK2 MLCK MLK1 MLK2 MLK3 MRCKA MRCKB MST1 MST1R MST2 MST3 MST4 MTOR MUSK MYLK MYLK2 MYLK4 MYO3A MYO3B NDR1 NDR2 NEK1 NEK11 NEK2 NEK3 NEK4 NEK5 NEK6 NEK7 NEK9 NIM1 NLK OSR1 p38α p38β p38δ p38γ PAK1 PAK2 PAK3 PAK4 PAK6 PAK7 PCTK1 PRKD1

93 100 98 100 82 86 91 100 84 100 100 100 100 100 96 47 56 87 87 85 78 100 100 98 100 90 89 99 100 100 100 90 100 100 100 97 87 89 100 95 69 95 100 100 100 100 100 98 100 100 90 100 80 100 100 100 100 100 87 100 100 87 84 91 97 100 84 100 100

PRKD2 PRKD3 PRKG1 PRKG2 PRKR PRKX PRP4 PYK2 QSK RAF1 RET RET(M918T) RET(V804L) RET(V804M) RIOK1 RIOK2 RIOK3 RIPK1 RIPK2 RIPK4 RIPK5 ROCK1 ROCK2 ROS1 RPS6KA4(Kin.Dom.1-N-terminal) RPS6KA4(Kin.Dom.2-C-terminal) RPS6KA5(Kin.Dom.1-N-terminal) RPS6KA5(Kin.Dom.2-C-terminal) RSK1(Kin.Dom.1-N-terminal) RSK1(Kin.Dom.2-C-terminal) RSK2(Kin.Dom.1-N-terminal) RSK2(Kin.Dom.2-C-terminal) RSK3(Kin.Dom.1-N-terminal) RSK3(Kin.Dom.2-C-terminal) RSK4(Kin.Dom.1-N-terminal) RSK4(Kin.Dom.2-C-terminal) S6K1 SBK1 SGK SgK110 SGK3 SIK SIK2 SLK SNARK SNRK SRC SRMS SRPK1 SRPK2 ZAK ZAP70

97 90 96 100 100 100 67 77 77 100 100 85 93 95 66 88 86 83 100 100 83 100 100 100 100 100 100 100 74 100 97 100 100 76 94 100 100 100 100 87 100 100 100 81 100 100 82 100 90 93 57 95

 c The Authors Journal compilation  c 2014 Biochemical Society

S. Han and others Table S5 Kinase selectivity of compound 9 as assessed by Carna Biosciences

Table S5

Compound 9 was screened at 10 μM against a panel of kinases (total of 284 kinases) and the results for the functional assays are reported as percentage inhibition, where higher numbers indicate stronger inhibition.

Carna Biosciences kinase symbol

Percentage inhibition

DCAMKL2 DDR1 DDR2 DLK_Cascade DYRK1A DYRK1B DYRK2 DYRK3 EEF2K EGFR EGFR(d746-750/T790M) EGFR(T790M/L858R) EPHA1 EPHA2 EPHA3 EPHA4 EPHA5 EPHA6 EPHA7 EPHA8 EPHB1 EPHB2 EPHB3 EPHB4 Erk1 Erk2 Erk5 FAK FER FES FGFR1 FGFR2 FGFR3 FGFR4 FGR FLT1 MAP3K4_Cascade MAP3K5_Cascade MAP4K2 MAPKAPK2 MAPKAPK3 MAPKAPK5 MARK1 MARK2 MARK3 MARK4 MELK MER MET MGC42105 MINK MLK1_Cascade MLK2_Cascade MLK3_Cascade MNK1 MNK2 MOS_Cascade MRCKα MRCKβ MSK1 MSK2 MSSK1 MST1 MST2 MST3 MST4 MUSK NDR1 NDR2 NEK1

− 2.3 45.6 26.0 − 4.4 1.3 5.1 − 8.8 − 8.7 − 1.3 17.0 10.5 10.4 − 2.6 − 2.3 − 4.4 − 8.5 − 2.1 10.6 4.8 0.9 − 4.4 − 3.4 − 7.0 − 3.7 − 3.0 − 7.1 − 12.0 0.2 5.2 − 1.5 27.7 51.5 26.9 − 0.3 19.9 21.2 2.3 − 1.7 4.8 − 10.4 − 1.3 − 4.5 0.3 − 5.5 − 2.0 2.0 2.7 16.1 0.3 − 7.5 4.7 17.8 5.7 1.8 − 9.0 − 2.7 − 4.8 − 4.2 − 3.2 − 0.4 − 4.8 − 1.2 7.4 6.9 − 3.4 − 0.1 4.4 − 3.9 − 3.3 − 2.0

Carna Biosciences kinase symbol

Percentage inhibition

ABL ABL(E255K) ABL(T315I) ACK AKT1 AKT2 AKT3 ALK ALK(F1174L) ALK(L1196M) ALK(R1275Q) EML4-ALK NPM1-ALK AMPKα1/β1/γ 1 AMPKα2/β1/γ 1 ARG AurA AurA/TPX2 AurB AurC AXL BLK BMX BRAF_Cascade BRAF(V600E)_Cascade BRK BRSK1 BRSK2 BTK CaMK1α CaMK1δ CaMK2α CaMK2β CaMK2γ CaMK2δ CaMK4 CDC2/CycB1 CDC7/ASK CDK2/CycA2 CDK2/CycE1 CDK3/CycE1 CDK4/CycD3 CDK5/p25 CDK6/CycD3 CDK7/CycH/MAT1 CDK9/CycT1 CGK2 CHK1 CHK2 CK1α CK1γ 1 CK1γ 2 CK1γ 3 CK1δ CK1ε CK2α1/β CK2α2/β CLK1 CLK2 CLK3 COT_Cascade CRIK CSK DAPK1

33.3 8.6 7.9 7.1 − 1.9 − 2.6 − 3.1 2.0 1.7 3.2 1.9 1.1 2.4 − 3.9 0.0 54.4 48.8 − 2.9 5.9 8.5 20.5 16.0 16.1 6.9 18.5 3.2 − 3.5 1.9 7.3 − 0.6 2.3 − 3.8 − 4.1 − 2.1 − 5.0 − 3.9 − 1.5 − 10.4 0.0 0.1 − 2.7 − 0.8 0.1 2.8 − 5.9 − 5.9 − 5.4 − 9.4 0.0 − 4.1 − 5.3 − 2.8 − 6.4 − 7.7 − 0.9 − 5.4 − 9.0 16.0 − 4.4 − 3.2 − 2.8 − 6.6 3.0 − 2.5

 c The Authors Journal compilation  c 2014 Biochemical Society

Continued

Allosteric inhibitors of ITK Table S5

Continued

Table S6

Carna Biosciences kinase symbol

Percentage inhibition

NEK2 NEK4 NEK6 NEK7 NEK9 NuaK1 NuaK2 p38α p38β p38γ p38δ p70S6K p70S6Kβ PAK1 PAK2 PAK4 PAK5 RSK4 SGK SGK2 SGK3 SIK skMLCK SLK SPHK1 SPHK2 SRC SRM SRPK1 SRPK2 SYK TAK1-TAB1_Cascade TAOK2 TBK1 TEC TIE2 TNIK TNK1 TRKA TRKB TRKC TSSK1 TSSK2 TSSK3 TXK TYK2 TYRO3 WNK1 WNK2 WNK3 YES ZAP70

− 4.4 − 4.4 − 2.6 − 8.4 − 0.2 2.9 3.4 − 2.1 − 3.2 − 1.8 − 2.8 − 1.3 − 3.4 − 7.9 10.9 3.4 4.9 − 3.6 − 5.3 − 3.0 − 10.5 − 3.5 1.2 3.9 − 1.2 2.5 17.0 − 1.8 6.4 − 3.2 15.1 1.7 − 0.7 − 3.3 12.7 − 7.1 14.5 5.1 78.3 63.8 60.8 − 1.5 − 2.8 − 0.9 2.1 2.9 3.6 − 3.3 − 4.1 − 4.3 22.9 2.7

Selectivity profile of compound 9 against TEC family kinases

IC50 values were generated at the ATP concentration corresponding to the respective K m for ATP. Kinase

Compound 9 IC50

BMX BTK ITK TEC TXK

>50 μM >50 μM 1 μM >50 μM >50 μM

 c The Authors Journal compilation  c 2014 Biochemical Society

S. Han and others Table S7

®

LANCE IC50 and steady-state inhibition constants for the allosteric inhibitors for ITK-FL

NC, non-competitive; C, competitive; ATP is the varied substrate. Equations used to derive the kinetic parameters are listed in the Supplementary Materials and Methods section. K is : the inhibition constant for inhibitor binding to free enzyme; K ii: the inhibition constant for inhibitor binding to the enzyme–substrate complex. n.a., not applicable. LANCE IC50 Compound

ATP pre-incubation

K is (μM)

K ii (μM)

Inhibition pattern

K is (FL)/K is (KD)

5 μM ATP†

1 mM ATP†

9

− + − + − + − + +

0.236 + − 0.022 0.678 + − 0.124 0.284 + − 0.014 0.345 + − 0.028 0.413 + − 0.031 0.626 + − 0.049 2.24 + − 0.28 3.04 + − 0.34 0.95 + − 0.26

1.20 + − 0.104 1.50 + − 0.124 n.a. n.a. n.a. n.a. n.a. n.a. 2.0 + − 0.5

NC* NC C C C C C C NC*

4.1 12 5.0 6.0 4.0 6.1 7.8 10.6

0.54

4.99

0.84

7.12

0.89

9.30

2.70

23.0

2.64

12.4

8 7 5 1

*Linear intersecting non-competitive inhibition. † ATP concentration in the assay.

Table S8 inhibitors

ITK-FL auto-activation inhibition constants for allosteric

The kinetic model used to derive K i autoact is listed in the Supplementary Materials and Methods section. Compound

K i autoact (μM)

9 8 7 5

0.026 + − 0.001 0.051 + − 0.002 0.165 + − 0.008 0.610 + − 0.017

Received 26 August 2013/28 February 2014; accepted 4 March 2014 Published as BJ Immediate Publication 4 March 2014, doi:10.1042/BJ20131139

 c The Authors Journal compilation  c 2014 Biochemical Society