Development of Selective Covalent Janus Kinase 3 Inhibitors

Abstract

The Janus Kinases (JAKs) and their downstream effectors Signal Transducer and Activator of Transcription proteins (STATs) form a critical immune cell signaling circuit, which is of fundamental importance in innate immunity, inflammation and hematopoiesis and dysregulation is frequently observed in immune disease and cancer. The high degree of structural conservation of the JAK ATP binding pockets has posed a considerable challenge to medicinal chemists seeking to develop highly selective inhibitors as pharmacological probes and as clinical drugs. Here we report the discovery and optimization of 2,4-substituted pyrimidines as covalent JAK3 inhibitors that exploit a unique cysteine (Cys909) residue in JAK3. Investigation of structure-activity-relationship (SAR) utilizing biochemical and transformed Ba/F3 cellular assays resulted in identification of potent and selective inhibitors such as compounds 9 and 45. A 2.9 Å co-crystal structure of JAK3 in complex with 9 confirms the covalent interaction. Compound 9 exhibited decent pharmacokinetic properties and is suitable for use in vivo. These inhibitors provide a set of useful tools to pharmacologically interrogate JAK3-dependent biology.

TOC image

Introduction

The Janus Kinases (JAKs) and their downstream effectors Signal Transducer and Activator of Transcription proteins (STATs) form a critical immune cell signaling circuit.¹ There are four JAK family members in mammals, JAK1, JAK2, JAK3 and tyrosine kinase 2 (TYK2), each of which can bind to distinct cytokines and/or growth factor receptors. As the receptors that act upstream of JAKs are dimeric or multimeric, more than one JAK molecule is recruited to each receptor. Ligand binding to the receptor results in a conformation change that brings two JAK molecules close together allowing them to trans-phosphorylate and activate each other. Once active, JAKs catalyze the phosphorylation of tyrosine residues in the receptor allowing the recruitment of specific STATs which are in turn phosphorylated by the JAK. This promotes STAT dimerization which induces translocation to the nucleus and activation of gene regulatory programs.² JAK/STAT signaling is of fundamental importance in innate immunity, inflammation and hematopoiesis and dysregulation is frequently observed in immune disease and cancer.³ The JAKs have been the subject of extensive drug discovery efforts and numerous small molecule ATP-competitive inhibitors have been developed including ruxolitinib (1, JAK1/2 inhibitor)⁴ approved as an anti-myelofibrosis drug, tofacitinib (2, pan-JAK inhibitor)⁵ approved for rheumatoid arthritis, and several others that are currently undergoing clinical investigation.⁶ The high degree of structural conservation of the JAK ATP binding pockets has posed a considerable challenge to medicinal chemists seeking to develop highly selective inhibitors as pharmacological probes and as clinical drugs.⁷

Among the JAK family JAK3 exhibits predominant expression in the hematopoietic system in contrast to other JAKs which are broadly expressed in a variety of cell types.^2b JAK3 is required for signaling by cytokines, including interleukin 2 (IL-2), IL-4, IL-7, IL-9, IL-15 and IL-21, that act via receptors that contain the common gamma chain (γ_c) cytokine receptor subunit.⁸ In addition to the γ_c subunit, most of these receptors contain a second unique α subunit that determines cytokine specificity. In these receptors, JAK3 binds to the γ_c subunit and JAK1 to the α subunit. The IL-2 Receptor (IL-2R) and IL-15R are exceptions to this as they can form trimeric complexes through the recruitment of α and β chains. In this case the β subunit binds JAK1; the α subunit, while involved in cytokine binding, does not directly bind to a JAK.⁹ Signaling via the γ_c receptors is essential for the normal development and function of the immune system. Loss-of-function mutations in either γ_c or JAK3 in humans results in severe combined immunodeficiency caused by defective development of T- and NK cells. The role for JAK3 in T and natural killer (NK) cell development was subsequently confirmed by deletion of Jak3 in mice; JAK3 knockouts also lacked normal numbers of B cells indicating a more important role for JAK3 in B cell development in mice than in humans.^{2b, 10} On the basis of JAK3’s essential function in immune signaling and its restricted expression in hematopoietic tissues, JAK3 has been pursued as a target for the treatment of autoimmune and inflammatory diseases. JAK3 has also been explored as a potential anti-cancer target due to aberrant activation in several lymphoproliferative disorders;¹¹ for example, mutations in JAK3 have been found in leukemias such as T-cell acute lymphoblastic leukemia (T-ALL) and T-cell prolymphocytic leukemia (T-PLL).¹² The A572V activating mutation in the kinase domain of JAK3 was discovered in acute megakaryoblastic leukemia and natural killer/T-cell lymphoma, and has been demonstrated to be capable of forming tumors in mice.¹³ Recently JAK3 mutations were also found to promote programmed death ligand 1 (PD-L1) induction in lung cancer cells an in the tumor immune microenvironments.¹⁴

The most obvious strategy to develop small-molecule JAK3 inhibitors is to target the catalytic ATP-binding site of the JAK3 kinase domain (JH1). Numerous ATP-competitive JAK3 inhibitors have been developed (Figure 1A).¹⁵ Tofacitinib (2) was developed as a specific JAK3 inhibitor to prevent organ-transplant rejection but subsequent studies revealed that it is also a potent inhibitor of JAK1 and JAK2.^{5, 16} The combination of compound 2’s excellent potency, selectivity and pharmacological properties have made it a favored inhibitor to interrogate JAK kinase activity in numerous biological models.^{13b, 17} In our efforts to develop a more selective Jak3 inhibitor, we noted that among JAK family members JAK3 is unique in having a cysteine residue at the gatekeeper-plus-7 (GK+7) position.¹⁸ This residue is Cys909 in human JAK, and it is structurally equivalent to cysteine residues in epidermal growth factor receptor (EGFR) and Bruton’s tyrosine kinase (BTK) that have been successfully targeted by covalent kinase inhibitors that are now approved drugs. Afatinib¹⁹ (6) targets Cys797 in EGFR and ibrutinib²⁰ (7) targets Cys481 in BTK. We and others have also developed mutant-selective inhibitors of the drug-resistant EGFR T790M that react covalently with Cys797. WZ4002²¹ (8) is a pyrimidine-based inhibitor with an acrylamide “warhead”²² that potently inhibits EGFR T790M. Recently three other groups have also reported the development of covalent JAK3 inhibitors. In 2013 Winssinger et al. reported a covalent probe which selectively labeled JAK3, EGFR and ERBB2.²³ Taunton et.al reported the development of ‘reversible-covalent’ JAK3 inhibitors incorporating a cyanoacrylamide warhead which inhibited JAK3 in a biochemical kinase assay with an IC₅₀ of 100 nM while sparing other JAKs up to a concentration of 10 μM.²⁴ Goedken et.al reported several tricyclic covalent JAK3 inhibitors with optimized inhibitors exhibiting biochemical IC₅₀’s of 7 nM, good kinome selectivity and confirmed covalent binding to Cys909 using kinetic and mass-spectrometry studies.²⁵ Here we report a medicinal campaign to derive potent disubstituted pyrimidine-based inhibitors that exploit an acrylamide electrophile to form a covalent bond to Cys909. Optimization was guided by cellular assays using Translocation ETS Leukemia protein (Tel)-JAK²⁶ fusion kinase transformed Ba/F3 cells that provided an efficient and reliable measure of the ability of compounds to inhibit JAK signaling in cellular context.

Figure 1.

Structures of exemplary JAK3 inhibitors (2–5) and lead compounds (9, 45 and 50) in this work.

Results

Identification of Compound 9 as a Potent JAK3 Inhibitor

Amongst the 11 kinases that possess a cysteine at the GK+7 position only JAK3 and mitogen-activated protein kinase kinase 7 (MAP2K7) have a methionine gatekeeper, which is also present in the drug-resistant EGFR^T790M that is targeted by the pyrimidine-based inhibitor WZ4002 (8). Compound 8 is a much less potent inhibitor of cellular JAK3 kinase activity (JAK3 Kd = 150 nM, TEL-JAK3 Ba/F3 IC₅₀ = 2.82 μM) compared to T790M EGFR (EGFR^L858R/T790M Ba/F3 IC₅₀ = 8 nM).²¹ Through the syntheses of a small collection of analogs of 8 we discovered that relatively subtle changes could result in a dramatic improvement in JAK3 inhibitory potency relative to EGFR^T790M. For example, 9 where the ether linkage of 8 is replaced with an aminomethylene (-NHCH₂-) linkage resulted in a great improvement in JAK3 inhibition in a fixed time-point Z’-lyte enzymatic assay (Life Technology, SelectScreen, IC₅₀ = 4.8 nM) (Scheme 1A).

Scheme 1.

^α i) DIEA, 1,4-dioxane, RT; ii) TFA, 2-BuOH, 100 °C; iii) Raney nickel, H₂, MeOH; iv) acryloyl chloride, sat. NaHCO₃, THF, 0 °C~RT.

To broadly assess its kinase selectivity, 9 was profiled against a diverse panel of 456 kinases (DiscoveRX, KinomeScan²⁷) using an in vitro ATP-site competition binding assay at a concentration of 1.0 μM. Compound 9 exhibited good overall kinase selectivity with an S(5) selectivity score, defined as the percentage of kinases with scores less than 5 (S(5))^27b, of 0.02. The results suggested that 9 most potently inhibits JAK3 and identified fms-related tyrosine kinase 3 (FLT3) and several tyrosine protein kinase (TEC)-family kinases as being potential off-targets (Figure 2). Enzymatic assays using the Z’-lyte or LanthaScreen²⁸ formats confirmed enzymatic inhibition of FLT3 (IC₅₀ = 13 nM), TTK protein kinase (TTK, IC₅₀ = 49 nM), BLK proto-oncogene (BLK, IC₅₀ = 157 nM) and tyrosine protein kinase TXK (TXK, IC₅₀ = 36 nM). Compound 9 showed very low inhibition scores for other JAKs and wild-type (WT) EGFR, which is consistent with the over 180-fold higher IC₅₀s against JAK1, JAK2, TYK2 and EGFR^WT (IC₅₀s = 896, 1050, > 10000 and 409 nM respectively). As BTK and ITK have important functions in B-cell and T-cell signaling pathways, we confirmed that compound 9 possesses over 165-fold higher IC₅₀s for BTK or ITK (IC₅₀s = 794 and 1070 nM respectively) (Table S1).

Figure 2.

KinomeScan kinase selectivity profiles for Compound 9. Compound 9 were profiled at a concentration of 1 μM against a diverse panel of more than 456 kinases and mutants. Scores for primary screen hits were reported as a percent of the DMSO control (% control). The lower the Kd is likely to be, such that scores of zero represent strong hits. Scores are related to the probability of being a hit but are not strictly an affinity measurement.

As enzymatic potencies sometimes do not translate into cellular inhibition, the ability of 9 to inhibit the proliferation of kinase-transformed Ba/F3 cells was evaluated. Ba/F3 cells are a murine pre-B cell that can readily be transformed with activated kinases to allow for growth in the absence of IL-3, and are frequently used to evaluate the activity of compounds against kinases of interest in a cellular context.²⁹ We utilized JAK1, JAK2 and JAK3 dependent Ba/F3 cell lines, where the JH1 domain of each JAK is fused with the oligomerization domain of TEL which results in constitutive tyrosine kinase activity and confers IL-3 independent proliferation.²⁶ In addition, we engineered a TYK2 Ba/F3 cell line whose proliferation is driven by constitutively activated TYK2 (TYK2^E957D).³⁰ As a further control, we also used a Ba/F3 cell line transformed by TEL- Abelson murine leukemia viral oncogene homolog (ABL). To enable a direct comparison with the commonly used JAK inhibitors we profiled reported compounds 1–5,^{4–5, 15b, 15g, 31} and 12³² against this panel of Ba/F3 cells. Compound 1 exhibited the most potent inhibition of JAK1 and JAK2 Ba/F3 cells, 2 exhibited most potent inhibition of JAK3 Ba/F3 cells and 3 exhibited most selective inhibition of JAK3 (Table S2). Overall the potency and selectivity of these inhibitors are consistent with their reported properties. Consistent with the biochemical assays, 9 selectively inhibited the proliferation of JAK3-dependent Ba/F3 cells (IC₅₀ = 69 nM) relative to other JAK-dependent Ba/F3 cells, for which there was no antiproliferative effect at concentrations below 3.0 μM (Table 2). The general antiproliferative activity that appears at concentrations of approximately 3.0 μM could be due to inhibition of other kinases such as TTK (aka Mps1, IC₅₀ = 49 nM) as inhibition of this kinase has been reported to decrease cancer cell viability.³³

Table 2.

SAR of R₁

ID	R1	JAK3 IC50 (nM)a	Ba/F3 cellular IC50s (nM)b					MLM T1/2 (min)c
			TEL-JAK1	TEL-JAK2	TEL-JAK3	TYK2 (E957D)	TEL-ABL
9		4.8	3124±116	3194±619	69±13	2266±490	3047±116	4.7
14		46	3871	6175	1305	3756	5257
15		170
16		1.2	513	1013	11	1320	2077
17		0.9	821	993	15	599	1558
18		2.0	378	692	20	874	1932
19		20	1059	2685	75	992	3470
20		4.6	3182	5981	125	1318	2824
21		1.3	1857±581	3704±410	7±0.1	2353±201	13705±194	5.2
22		1.4	3287	2910	29	1542	4494	3
23		0.9	5195±3001	9087±4895	11±3.3	6613±2650	22410±4752	3
24		5.2	>5E4	>5E4	478	>5E4	>5E4
25		20	>5E4	>5E4	99	>5E4	>5E4
26		3.6	2004	3290	23	4666	8769	3.4
27		7.4	2484	4321	35	4702	2326
28		6.2	2346	2916	54	4171	5477
29		24	17440	30770	5452	17300	27680
30		99	20330	30780	9802	24780	17270
31		1600	6792	9695	12430	8170	11150

ID	R1	JAK3 IC50 (nM)	Ba/F3 cellular IC50s (nM)					MLM T1/2 (min)
			TEL-JAK1	TEL-JAK2	TEL-JAK3	TYK2 (E957D)	TEL-ABL
32		0.6	2043	3878	10	1429	4055	5.1
33		1.7	5198	8487	109	6502	11030	8.3
34		2.9	2792	3993	114	2492	3998	3.5
35		1.4	1686	3290	21	994	3136	10.4
36		<0.5	1780	3164	11	759	4374	6.4
37		0.7	2143	2762	16	5581	11820	1.5
38		<0.5	2797	2776	28	1099	5286	3
39		<0.5	>5E4	>5E4	143	>5E4	>5E4	14.3
40		0.5	3506	27680	110	17260	21820	3.4
41		1.1	4553	6223	67	3043	10610	4.5
42		0.6	8223	19500	119	4998	18660	7.2
43		<0.5	3129	6960	17	1887	5398	1.8
44		<0.5	3462	6439	24	2227	6830	1.9
45		<0.5	7501±460	31133±17012	19±8	>5E4	>5E4	3.3
46		<0.5	3137±468	6211±638	22±2	2291±1342	11169±2688	1.8
47		<0.5	2370±494	9081±3534	39±5	1576±631	13157±6948	1.3
48		0.7	23510±13265	>5E4	45±32	31496±41386	>5E4	2.6
49		0.6	3065	9098	124	3227	5973	1.8
50		<0.5	3758±1863	7177±2847	34±7	3510±1281	8822±126	10.8
51		0.6	5744	7287	92	4090	8369	4.8

^aEnzymatic IC₅₀s obtained with Z’-Lyte activity assays.

^bAll cellular IC₅₀s were obtained with 8-point titrations and triplicates, the key compounds were tested in at least 3 independent experiments and the standard deviations were shown, compound 9 was always used as a control.

^cMouse liver microsomal half-life.

Co-crystal Structure of JAK3-Compound 9 Complex

To investigate the structural basis for achieving selectivity for JAK3 we solved the co-crystal structure of the JAK3 kinase domain in complex with 9 at a resolution of 2.9 Å (Figure 3). In this structure (PDB 4Z16), the anilinopyrimidine moiety of 9 makes the expected bidentate hinge hydrogen bonds with Leu905, and continuous electron density is observed between the acrylamide warhead and Cys909, indicative of covalent bond formation. The kinase exhibits an active conformation, with the DFG-motif in the inward position and both tyrosines 980 and 981 in the activation loop phosphorylated. The orientation and interactions of the anilinopyrimidine portion of the compound closely resembles the interactions observed in the EGFR^T790M-8 co-crystal structure (PDB 3IKA)²¹ (Figure S1). In addition to the hinge hydrogen bonds, the chlorine is directed towards the gatekeeper methionine (Met902 in the present structure) and the methoxy substituent extends toward the sidechain of Tyr904 in the hinge region. Interestingly, the “linker” segment that attaches the acrylamide adopts a distinctly different conformation as compared with the corresponding region of 8 in complex with EGFR^T790M. The phenyl group in the linker pivots down and is in van der Waals contact with Leu956 in the floor of the ATP-binding cleft. This orientation is likely enabled by the longer aminomethylene moiety in 9, as compared with the ether linkage in compound 8. In addition, the acrylamide amide is positioned to hydrogen bond with the carbonyl of Arg953, also in the floor of the binding pocket.

Figure 3.

X-ray co-crystal structure of JAK3-9 (PDB 4Z16). Compound 9 (yellow sticks) covalently binds to Cys909 at the GK+7 position of JAK3 (blue ribbons). The hydrogen bonds between the anilinopyrimidine moiety of 9 and Leu905 in the JAK3 hinge region are indicated.

Structure-Activity Relationship (SAR) Studies Based on Compound 9

The syntheses of 9 was readily achieved with high-overall-yield in four steps starting from 2,4,5-trichloropyrimidine. The 4-chloride group was substituted with a 3-nitrobenzyl amine under basic conditions, followed by the substitution of the 2-chloride with a 2-methoxy-4-(4-methylpiperazin-1-yl)aniline under acidic conditions to give 11. The nitro group of 11 was reduced using hydrogenation and the resulted aniline was acrylated to afford 9 (Scheme 1B). In order to elucidate the structural requirements to achieve potent and selective inhibition of JAK3 we prepared approximately 70 analogs. To approach this optimization in a systematic fashion, this chemotype was divided into three moieties: tail (R₁), arm (R₂) and core (R₃), and each of these moieties were varied sequentially. The 2,4-disubstited aniline tail moiety of 9 was replaced with different anilines, aliphatic amines or 4-aminopyrazoles; the 4-acrylamidobenzyl arm of 9 was substituted with functional groups at the double bond, phenyl ring or benzyl position; the pyrimidine core of 9 was modified at 5- or 6-position, or was replaced with a variety of bicyclic cores. All the analogs were profiled with the same JAK3 enzymatic assay and JAKs and ABL dependent Ba/F3 cell line assays. Additionally, the in vitro metabolic stability of some analogs was measured using mouse liver microsomes (MLM) to calculate half-lives (T_1/2).³⁴

The first series of analogs focused on modification of the tail moiety (R₁) (Table 2). Substitutions at the 5-postion of the aniline (14, 15) resulted in a dramatic loss of potency. In contrast, removing the 2-methoxy group (16) enhanced potency over 5-fold against JAK3, however, it also increased the cytotoxicity toward other transformed Ba/F3 cell lines. Substitutions at the 3-position did not mitigate the cytotoxicity (17–19), nor did introduction of a fluorine at the 2-position (20). Compound 21 with a 4-morpholinylaniline tail potently inhibited JAK3 despite not having the 2-methoxy substitution. Furthermore, 21 showed comparable cellular selectivity and MLM T_1/2 as 9 (T_1/2 of 4.7 min.). However, 21 only showed moderate selectivity in biochemical assays, with 10~20-fold higher IC₅₀s against JAK1, JAK2 or TTK (Table 1). Adding a 2- or 3-fluoro (22, 23) did not improve the kinase selectivity of 21 and also did not improve the MLM stability. 24 which contains two fluorine substituents at the 2- and 5-positions showed good selectivity but decreased potency against JAK3. 25, with 3,5-difluorine substitutions, exhibited improved potency compared to 24 with good selectivity in cells, even though it showed moderate selectivity in enzymatic assays. Based on 25 several analogs (26–28) were made and exhibited increased potency, but none of them were as selective as 25. Aliphatic tails resulted in complete loss of potency (29–31). 32 with a 1-methyl-1H-pyrazol-3-amine tail showed increased potency, similar selectivity and MLM stability compared with 9; it also showed a 300-fold selectivity window over TTK. Addition of a 4-methyl group to the pyrazole ring (33) enhanced the MLM stability but decreased the potency over 10-fold. The isoxazol-4-amine tail in 34 is disfavored in terms of both potency and stability. Modifications were then focused on the 1-methyl group of the pyrazole tail. Replacement of the 1-methyl with a 1-difluoromethyl (35) or a 1-difluoroethyl (36) maintained the potency and enhanced the MLM stability. However, bulky and hydrophobic substituents resulted in decreased MLM stability (37, 38). 39 with a 2-substituted N-methylacetamide exhibited noticeably improved MLM stability (T_1/2 of 14.3 min) and favorable selectivity albeit with only moderate potency against JAK3. Nevertheless, other 2-substituted N-alkylacetamide groups (40–42) failed to improve the potency or stability. Several analogs elaborated with a 2-substituted ethyl group (43–45) showed good potency but decreased MLM stability. To our surprise, 45 with a 2-methoxyethyl exhibited an IC₅₀ of 19 nM against JAK3 Ba/F3 cells, and over 390-fold selectivity over other Ba/F3 cell lines. 45 showed excellent selectivity with JAK3 IC₅₀ of less than 0.5 nM in fixed end point enzymatic assay, and over 70-fold higher IC₅₀ against JAK1 and over 100-fold higher IC₅₀s against JAK2, TYK2, TTK, BTK or ITK (Table 1, S1). 45 was profiled against a diverse panel of 468 kinases (DiscoveRX, KinomeScan²⁷) at a concentration of 100 nM; the compound exhibited excellent overall kinase selectivity with selectivity scores, defined as the percentage of kinases with scores less than 5 (S(5))^27b, of 0.01 (Figure 4). The results suggested 45 inhibits JAK3 most potently, and aurora A (AURKA) was likely to be the only off-target which we confirmed to be inhibited by 45 with IC₅₀ of 43 nM. 45 showed relatively poor MLM stability with T_1/2 of 3.3 min. Thus, several analogs with different 2-alkoxyethyl groups (46–48) were made to solve this problem. To our disappointment, all changes did not improve the MLM stability. Dimethyl substitution of the 2-methoxyethyl (49) is disfavored with respect to MLM stability as well as potency and selectivity. 50, with a hydroxyl group in the tail, showed 3-fold longer MLM T_1/2 compared with 45. Although 50 was not as selective as 45 in Ba/F3 cell lines, it was more potent than 9 with comparable selectivity in Ba/F3 cells. Further modification of 50 failed to improve the selectivity but led to decreased potency and MLM stability (51). In this series of analogs, 45 and 50 stood out with good selectivity or MLM stability, furthermore, we found 50 was actually one of the MLM metabolites of 45 (Figure S2).

Table 1.

Enzymatic IC₅₀s of key compounds ^a

Cmpd ID	IC50s (nM)
	JAK1	JAK2	JAK3	TYK2	TTK
9	896	1050	4.8	>10000	49
21	24.4	17.6	1.3		28.6
25	495	351	20		355
32	49	107	0.6		180
45	35	51	<0.5	407	212
50	11	32	<0.5	140	143
85	743	421	2.3	3120

^aIC₅₀s against JAK-family kinases were obtained with Z’-Lyte activity assays while IC₅₀s against TTK were obtained with LanthaScreen binding assays.

Figure 4.

KinomeScan kinase selectivity profiles for Compound 45. Compound 45 were profiled at a concentration of 100 nM against a diverse panel of more than 468 kinases and mutants. Scores for primary screen hits are reported as a percent of the DMSO control (% control). The lower the Kd is likely to be, such that scores of zero represent strong hits. Scores are related to the probability of a hit but are not strictly an affinity measurement.

After optimization of the tail moiety, the SAR of the arm (R₂) was investigated (Table 3). Replacing the acrylamide with a propionamide (52) resulted in over an 80-fold decrease of potency against JAK3. Installation of a methyl group at α-position of the acrylamide (53) resulted in significant loss of potency and a shorter MLM T_1/2. Polar substitutions at the β-position of the acrylamide (54–56) also resulted in loss of potency. Addition of a fluorine ortho to the acrylamide in the phenyl ring (57) also decrease the potency dramatically. Switching the acrylamide from meta- to para- on the phenyl ring decreased the potency considerably in combination with the aniline tail of 9 (58), but did not affect the potency when used in conjunction with other tails (59–62). This change did however decrease the selectivity in some cases (59, 62). Analogs with a para-acrylamide generally showed comparable or diminished MLM stability. Finally, adding a methyl at the benzyl position (63, 64) reduced both potency and selectivity. In summary, the modifications of R₂ failed to further improve our JAK3 inhibitors. Several reported inhibitors with similar acrylamide warheads (6, 7, 8 and AZD9291³⁵ (65)) were also profiled and compared, none of them showed good inhibition against JAK3 Ba/F3 cells with IC₅₀s over 1.0 μM (Table S2).

Table 3.

SAR of R₂

ID	R2	R1	JAK3 IC50 (nM)a	Ba/F3 cellular IC50s (nM)b					MLM T1/2 (min)c
				TEL-JAK1	TEL-JAK2	TEL-JAK3	TYK2 (E957D)	TEL-ABL
52		T9	390	4948	9354	7938	8870	7594
53		T45	7.8	3825	9824	947	15090	23780	2.1
54		T9	245
55		T9	64	1823	17240	14650	12680	26660
56		T9	258
57		T32	4.4	3624	2326	102	1598	4123
58		T9	83	3974	4996	2363	5452	4020	6
59		T21	0.87	701	874	17	778	459
60		T32	0.58	2008 ±340	4733 ±326	18±5	754± 298	2140 ±519	3.1
61		T39	2.0	7422	10120	465	6857	7811	21.3
62		T45	1.3	3117	5575	131	3549	3021	2.8
63		T9	6.6	5034	6039	439	4215	6782
64		T45	1.6	3173	7731	57	5193	9942	2.6

^aEnzymatic IC₅₀s obtained with Z’-Lyte activity assays.

^bAll cellular IC₅₀s were obtained with 8-point titrations and triplicates, the key compounds were tested in at least 3 independent experiments and the standard deviations were shown, compound 9 was always used as a control.

^cMouse liver microsomal half-life.

Simultaneously we investigated the SAR of the core moiety (R₃) (Table 4). We found that a thioether linkage at the 4-postion (66, 67) slightly increased the MLM stability but decreased the potency, and a tertiary amine linkage (-NMe-) decreased the potency even more dramatically (68). Next we investigated the effects of various substituents at the 5-position. We evaluated substituents that possessed electron withdrawing groups such as chlorine (69–72), electron donating groups (73–75) or simply a hydrogen atom (76). However, only 70 with a 5-bromo group showed good potency among these analogs. A 4,6-disubstituted pyrimidine core (77) also led to substantial loss of potency. Next we tried to cyclize the 4,5-substituents into a fused 5-membered or 6-membered ring (78–82), and found that only 78 with a pyrrolo[2,3-d]pyrimidine core showed good potency against JAK3, but with a slightly decreased MLM stability. 81 with a pyrimido[4,5-d]pyrimidinone core possessed a good IC₅₀ against JAK3 in the biochemical assay, but mediocre potency in the TEL-JAK3-transformed Ba/F3 cells. In the end, modifications at the 6-position of the pyrimidine core (83, 84) were fairly disfavored. These results suggested that 5-chlorine is the most suitable, and the secondary amine (-NH-) is a satisfactory linkage at 4-postion; some bicyclic cores did improve the inhibitor modestly, and modifications at 6-position are not tolerated.

Table 4.

SAR of R₃

ID	R3	R1	JAK3 IC50 (nM)a	Ba/F3 cellular IC50s (nM)b					MLM T1/2 (min)c
				TEL-JAK1	TEL-JAK2	TEL-JAK3	TYK2 (E957D)	TEL-ABL
66		T9	12	2324	3101	433	1318	2848
67		T32	1.2	2854	3870	33	1122	11040	6.9
68		T32	0.7	3490	4238	112	3943	5538	3.7
69		T9	30	11860	11870	755	5874	8890
70		T9	2.3	2480	3005	107	2005	3022
71		T9	18	3640	2426	307	1836	6036
72		T9	2570
73		T9	31	3822	4094	1089	3163	6185
74		T9	31
75		T9	220
76		T9	222
77		T16	9.9	25100 ± 610	29895 ± 329	1338 ± 61	15435 ± 2167	32710 ± 976
78		T16	0.7	1315	2733	8	4458	2989	3.7
79		T16	101	26720	>5E4	4947	24630	>5E4
80		T16	17	3457	4047	210	3327	2883
81		T16	0.9	10076 ±1106	6728 ±1180	159± 11	3503 ±81	12551 ±4680
82		T16	7.0	25120	8398	144	5948	4922
83		T9	499	4268	5975	2563	3212	5098
84		T9	2720	7890	5306	6232	4950	6671

^aEnzymatic IC₅₀s obtained with Z’-Lyte activity assays.

^bAll cellular IC₅₀s were obtained with 8-point titrations and triplicates, the key compounds were tested in at least 3 independent experiments and the standard deviations were shown, compound 9 was always used as a control.

^cMouse liver microsomal half-life.

Validation of the Target-Engagement and the Covalent Binding

Based on the SAR results above, we selected 9, 45 and 50 as our optimal JAK3 inhibitors, and investigated their inhibitory activity on JAK3-dependent signaling. As STAT5 has been reported as a direct substrate of TEL-JAK3 in Ba/F3 cells,²⁶ we examined the effect of our JAK3 inhibitors on STAT5 phosphorylation in this context. As we expected, STAT5 was constitutively phosphorylated in the cells treated with DMSO (Figure S3). Compound 2 at 300 nM almost fully inhibited the phosphorylation of STAT5 (p-STAT5), while 9 at 300 nM completely abolished p-STAT5, and so did 45 or 50 at 100 nM. These results are consistent with the IC₅₀s in JAK-transformed Ba/F3 proliferation assays. In order to confirm that 45 and 50 are bonafide covalent inhibitors in analogy to compound 9 we performed cellular ‘wash-out’ experiments. JAK3 Ba/F3 cells were treated with these four inhibitors at various concentrations for 3 h, the cells were washed extensively with PBS and then allowed to recover for 4 h. Western blots of the cellular lysates revealed that 9, 45 or 50 were capable of sustained inhibition of p-STAT5 after the washout, whereas the reversible inhibitor 2 was not (Figure 5A). To monitor the degree of JAK3 ‘target engagement’, a biotinylated version of 45 (85) was designed with a biotin tethered via a flexible PEG linker at the tip of 45’s tail moiety (Figure 5C). We confirmed that 85 maintained similar biochemical potency and selectivity for JAK3 (Table 1). 85 showed selective but weak potency against JAK3 Ba/F3 cells with an IC₅₀ of 1.3 μM (Table S2) presumably due to poor membrane permeability. We demonstrated that streptavidin mediated pulldown of 85 in cell lysates allowed for efficient recovery of JAK3 as assessed by western blotting. Consistent with the wash-out results, 85 strongly labeled JAK3 when cells were treated with the reversible inhibitor 2 but not when cells were treated with acrylamide modified inhibitors 9, 45 or 50 (Figure 5B). Compounds 9, 45 and 50 were further demonstrated to be covalent inhibitors using electrospray ionization mass spectrometry, where incubation of recombinant JAK3 kinase domain with inhibitor resulted in addition of the expected molecular weight (Figure 6). Subsequent protease digestion and LC/MS² experiments identified a single modified peptide LVMEYLPSGC*LR (C*, cysteine labeled by 9, 45 or 50, JAK3 residues 900–911) and localized the site of modification to Cys909 (Figure S4), in agreement with the JAK3/9 co-crystal structure. Cumulatively these results provide strong evidence that 9, 45 and 50 are all irreversible, covalent inhibitors and that Cys909 of JAK3 is the only labeled site.

Figure 5.

Compounds 9, 45, and 50 are covalent, irreversible JAK3 inhibitors. TEL-JAK3 Ba/F3 cells were treated with 2, 9 (1.0 μM), 45 or 50 (300 nM) for 3 h, washed extensively with PBS, allowed to recover for 4 h, then lysed and subjected to western blot for phospho-STAT5 and total STAT5 (A). TEL-JAK3 Ba/F3 cells were treated 2, 9 (1.0 μM), 45 or 50 (300 nM) for 3 h, and the resulting cell lysates were treated with 85 (C) (1.0 μM, 1 h), followed by pull-down with streptavidin beads and immunoblotting with anti-JAK3 antibody (B).

Figure 6.

Compound 9, 45, or 50 react quantitatively with JAK3. Raw (A, C, E, G) and deconvoluted (B, D, F, H) mass-spectra obtained for JAK3 after treatment with DMSO (A, B), 9 (C, D), 45 (E, F), or 50 (G, H).

Validation in Cells with Endogenous JAK3

Compounds 9, 45 and 50 were further evaluated for their ability to inhibit JAK3 kinase activity in a variety of other cell types. We evaluated the ability of 9, 45 or 50 to inhibit signaling following stimulation of primary murine bone marrow derived macrophage (BMDMs) with a panel of 4 different cytokines based on their requirement for different cytokines (Figure 7). Among these cytokines, only signaling via IL-4, whose receptor binds JAK1 and JAK3,³⁶ is dependent on JAK3. In contrast, granulocyte-macrophage colony-stimulating factor (GM-CSF), which signals via JAK2, and IL-10 and interferon beta (IFNβ), which signal via JAK1 and TYK2, are JAK3 independent.³⁷ 9, 45 or 50 completely inhibited IL-4 induced p-STAT6 at a concentration of 500 nM and only partially inhibited IFNβ-induced p-STAT1 at a concentration of 5.0 μM. None of our inhibitors detectably reduced the p-STAT5 level stimulated by GM-CSF or p-STAT3 induced by IL-10 at a concentration of 5.0 μM. While demonstrating excellent selectivity in Ba/F3 cellular assays, these inhibitors showed decreased but still decent selectivity in primary cells. The washout experiments with 45 or 50 in BMDMs cells exhibited consistent results as in TEL-JAK3 Ba/F3 cells, and suggested that the inhibitory effect lasted for at least 24 h after the washout (Figure S5). In a similar way, 9, 45 and 50 exhibited consistent potency and selectivity for JAK3 in leukemia cancer cell lines: T-ALL1 and OCL-AML5 (Figure 8). In T-ALL1 cells, IL-2 induced STAT5 phosphorylation was completely inhibited by 9 at a concentration of 1.0 μM or by 45 or 50 at a concentration of 100 nM (Figure 8A). However, in OCL-AML5 cells the GM-CSF induced STAT5 phosphorylation was maintained in the presence of the inhibitors up to a concentration of 10 μM (Figure 8B). In contrast, previously reported pan-JAK inhibitors 1 and 2 could abolish p-STAT5 in both stimulated cell lines at 1.0 μM. Based on these results compounds 9, 45 and 50 are capable of efficiently inhibiting JAK3 and specifically blocking the JAK3-dependent signaling pathway in human cells at sub-micromolar concentrations.

Figure 7.

Compound 9, 45 or 50 potently and selectively inhibited JAK3-dependent signaling in BMDMs cells. BMDMs cells were pre-incubated with the indicated concentrations of 9 (A), 45 (B) or 50 (C) for 3 h. Cells were then stimulated with the indicated cytokines (IL-4 at 10 ng/mL, IFNβ at 500 U/mL, GM-CSF at 10 ng/mL, and IL-10 at 100 ng/mL) for 30 min then lysed. The levels of tyrosine phosphorylation on the appropriate STAT protein were determined by immunoblotting. Levels of total ERK1/2 were also examined to show equal loading. The signaling cascades were described in D.

Figure 8.

Compound 9, 45 and 50 potently and selectively inhibited JAK3-dependent p-STAT5 in TALL-1 cells whereas pan-JAK inhibitors 1 and 2 also potently inhibited JAK2-dependent p-STAT5 in OCL-AML5 cells. TALL-1 (A) or OCL-AML5 (B) cells were pre-incubated with the indicated concentrations of indicated inhibitors for 3 h. Cells were then stimulated with 10 ng/mL of IL-2 (A) or 5 ng of GM-CSF (B) for 30 min then lysed. The p-STAT5 levels were determined by immunoblotting. Levels of total STAT5 were also examined to show equal loading. The signaling cascades were described in C.

Mouse Pharmacokinetic Properties and In Vivo Studies

We evaluated the pharmacokinetic properties of 9 in mice following intravenous and oral delivery. Compound 9 demonstrated reasonable pharmacokinetic properties, with moderate T_1/2 of 1.4 h, area under the curve (AUC) value of 795 ng*hr/mL following a 10 mg/Kg oral dose and good oral bioavailability of 66% (Table 5). This suggests that 9 may be a suitable probe for future murine efficacy studies. We evaluated the ability of compound 9 to alter immune cell numbers in genetically engineered models (GEM) of lung adenocarcinoma in mice. Immune cell numbers in mice harboring tumors driven by Kras^G12D;Lkb1^L/L–³⁸ were assessed in tumor-associated (lung) as well as general immune cell populations (spleen) upon treatment with compound 9. After oral administration with 9 (75 mpk, QD) for 8 days, the numbers of B or T lymphocytes in the tumor-bearing lungs and spleens of treated mice was not affected, however, the number of NK cells was reduced (Figure S6). Similarly, the number of tumor-associated macrophages and neutrophils in the lung of treated mice were not affected. During the short duration of this study, we did not detect any effect on the tumor growth.

Table 5.

Pharmacokinetic properties of 9

ID	route	Dose (mg/kg)	Tmax (h)a	Cmax (ng/mL)b	AUClast (h*ng/mL)c	T1/2 (h)d	CL (mL/min/kg)e	Vss (L/kg)f	F (%)g
9	IV	2	–	158.56	241.30	1.40	135.88	18.38	–
	PO	10	0.5	204.25	795.29	–	–	–	66

^aTime for peak plasma concentrations.

^bPeak plasma concentrations.

^cArea under the concentration time curve.

^dTerminal half-life.

^ePlasma Clearance.

^fVolume in steady state.

^gBioavailability.

Discussion

Building on our covalent T790M EGFR inhibitor 8 we designed and synthesized new analogs aiming to target the analogous cysteine residue in JAK3. Among those analogs compound 9 was identified as a potent and selective JAK3 inhibitor compared with current reversible JAK inhibitors. We determined a 2.9 Å co-crystal structure of JAK3 with 9, which revealed that 9 covalently binds to Cys909 of JAK3 as expected. With 9 as a lead, we designed and synthesized dozens of analogs to intensively study the SAR with regard to each moiety of this scaffold. These analogs were tested in enzymatic assays for their potency against JAK3 and profiled against a panel of transformed Ba/F3 cell lines, for their potency and selectivity in a cellular context. Based on these studies we discovered that a 2-methoxy group in the aniline tail of 9 is disfavored by JAK3 presumably due to a disfavored interaction with the bulky side chain of Tyr904 located at the GK+2 position. Hydrophobic or less polar tails are better for selectivity for JAK3 presumably because they have weaker non-covalent affinity for the kinase thereby allowing covalent bond formation to be a more important contributor to the potency. The acrylamidobenzyl arm moiety of 9 is significant for the potency against JAK3 and sensitive to changes, its specific length and flexibility are critical to enable the inhibitors to efficiently form a covalent bond with Cys909 of JAK3. The H-bond between the acrylamide of 9 and the carbonyl of Arg953 fixed the conformation of the acrylamide and left no space to accommodate substituents. The chlorine atom in the pyrimidine core is responsible for both potency and selectivity with its interaction with Met902 of JAK3. Modifications at the 6-position of the pyrimidine core were fairly disfavored as they collided with the carbonyl of Glu903 3.3Å away. Some of these analogs were also evaluated for their MLM stability, the results indicated that more lipophilic analogs with aliphatic non-polar substituents tend to have a shorter half-life, maybe resulting from their enhanced binding with phospholipid in microsomes.³⁹

Among these analogs 9, 45 and 50 stood out with overall favorable potency, selectivity and MLM stability. They all exhibited efficient and selective inhibition of JAK-dependent signaling within different contexts. So far 45 is our most selective JAK3 inhibitor with at least 70-fold selectivity over other JAKs in biochemically assays, at least 390-fold selectivity in JAK-transformed Ba/F3 cells, and decent selectivity in primary cells. In addition to the evidence provided by JAK3-9 co-crystal structure, all three inhibitors were demonstrated to covalently modify Cys909 of JAK3 based upon wash-out and pull-down experiments and mass spectrometry. We developed a biotinylated probe 85 which was demonstrated to be a useful reagent for establishing “target-engagement” in cellular assays. We also found that a short-term treatment with high dose of compound 9 in mice with lung tumors resulted in a decrease in the number of NK cells suggesting that the compounds are active in vivo. Further investigations will be required to elucidate the basis for this phenomena and to assess the activity of covalent JAK3 inhibitors in relevant disease models.

In conclusion, the covalent JAK3 inhibitors 9, 45, 50, and the biotinylated probe 85 provide a set of useful tools to pharmacologically interrogate JAK3-dependent biology. Compound 45 is a highly potent and selective biochemical and cellular inhibitor of JAK3 and is an ideal reagent for cell biological studies. Compound 9 combines favorable target profile with good pharmacokinetic properties and is ready to be used in future animal studies. Finally, the structure-activity relationships with respect to JAK3 combined with the co-crystal structure will serve as an excellent foundation for future JAK3 inhibitor development efforts.

Experimental Section

Chemistry

Unless otherwise noted, reagents and solvents were obtained from commercial suppliers and were used without further purification. ¹H NMR spectra were recorded on 600 MHz (Varian AS600), and chemical shifts are reported in parts per million (ppm, δ) downfield from tetramethylsilane (TMS). Coupling constants (J) are reported in Hz. Spin multiplicities are described as s (singlet), br (broad singlet), d (doublet), t (triplet), q (quartet), and m (multiplet). Mass spectra were obtained on a Waters Micromass ZQ instrument. Preparative HPLC was performed on a Waters Sunfire C18 column (19 × 50 mm, 5μM) using a gradient of 15–95% methanol in water containing 0.05% trifluoroacetic acid (TFA) over 22 min (28 min run time) at a flow rate of 20 mL/min. Purities of assayed compounds were in all cases greater than 95%, as determined by reverse-phase HPLC analysis.

2,5-Dichloro-N-(3-nitrobenzyl)pyrimidin-4-amine (10)

2,4,5-Trichloropyrimidine (112 μL, 1.0 mmol), (3-nitrophenyl)methanamine hydrochloride salt (227 mg, 1.2 mmol), and N,N-diisopropylethylamine (DIEA, 530 μL, 3.0 mmol) were combined in dioxane (5 mL) and stirred overnight. The mixture was then diluted with ethyl acetate and washed with water and brine, dried over Na₂SO₄, filtered and concentrated. The crude product was purified by column chromatography (hexane:ethyl acetate = 1:1) to yield 270 mg (80%) of 10 as a white solid. ¹H NMR (600 MHz, CDCl₃) δ 8.21 (s, 1H), 8.18 (d, J = 8.4 Hz, 1H), 8.10 (s, 1H), 7.71 (d, J = 7.8 Hz, 1H), 7.56 (dd, J = 7.8, 7.8 Hz, 1H), 5.94 (br, 1H), 4.84 (d, J = 6.0 Hz, 1H). MS (ESI) m/z 299 (M+H)⁺.

5-Chloro-N²-(2-methoxy-4-(4-methylpiperazin-1-yl)phenyl)-N⁴-(3-nitrobenzyl)pyrimidine-2,4-diamine (11)

To 10 (150 mg, 0.5 mmol) and 2-methoxy-4-(4-methylpiperazin-1-yl)aniline (166 mg, 0.75 mmol) in sec-butanol (5 mL) was added trifluoroacetic acid (57 μL, 0.75 mmol) and the mixture was stirred overnight at 100 °C. The mixture was then concentrated, neutralized with ammonia in methanol and purified by column chromatography (dichloromethane:methanol = 10:1) to yield 184 mg (76%) of 11 as a pale-yellow solid. ¹H NMR (600 MHz, CD₃OD) δ 8.23 (s, 1H), 8.09 (d, J = 8.4 Hz, 1H), 7.84 (s, 1H), 7.70 (d, J = 7.8 Hz, 1H), 7.65 (d, J = 9.0 Hz, 1H), 7.54 (dd, J = 7.8, 7.8 Hz, 1H), 6.60 (d, J = 3.0 Hz, 1H), 6.36 (dd, J = 9.0, 3.0 Hz, 1H), 4.75 (s, 1H), 3.83 (s, 3H), 3.15 (m, 4H), 2.64 (m, 4H), 2.36 (s, 3H). MS (ESI) m/z 484 (M+H)⁺.

N-(3-(((5-chloro-2-((2-methoxy-4-(4-methylpiperazin-1-yl)phenyl)amino)pyrimidin-4-yl)amino)methyl)phenyl)acrylamide (9)

To 11 (97 mg, 0.2 mmol) in MeOH (20 mL) was added 1 mL Raney nickel suspension in MeOH. The reaction mixture was stirred for 3 h under 1 atm of hydrogen. The mixture was then filtered with Celite, and the filtrate was concentrated and dried under vacuum to give a crude product as a white solid. To the obtained white solid in DMF (2 mL) was added DIEA (53 μL, 0.3 mmol), the stirred mixture was then cooled to −60 °C, and acryloyl chloride (17.8 μL, 0.22 mmol) was added dropwise. The reaction mixture was stirred at −60 °C for 10 min, allowed to recover to RT (room temperature) gradually in 30 min, and purified by reverse phase HPLC to give 88 mg (TFA salt, 71% for 2 steps) of 9 as a white solid. ¹H NMR (600 MHz, DMSO-d₆) δ 10.07 (s, 1H), 7.82 (s, 1H), 7.73 (m, 1H), 7.59 (s, 1H), 7.52 (d, J = 8.4 Hz, 1H), 7.47 (d, J = 7.8 Hz, 1H), 7.29 (s, 1H), 7.18 (dd, J = 7.8, 7.8 Hz, 1H), 6.90 (d, J = 7.2 Hz, 1H), 6.48 (s, 1H), 6.44 (dd, J = 16.8, 10.2 Hz, 1H), 6.31 (d, J = 8.4 Hz, 1H), 6.25 (d, J = 16.8 Hz, 1H), 5.75 (d, J = 10.2 Hz, 1H), 4.56 (d, J = 6.0 Hz, 2H), 3.78 (s, 3H), 3.04 (m, 4H), 3.01 (m, 4H), 2.22 (s, 3H). MS (ESI) m/z 508 (M+H)⁺.

Compound 14–64, 66–85 were synthesized with same procedures as 9, 3 and 8 were synthesized as reported.^{19, 31} Compound 1, 2, 6, 7, 12, 13 and 65 were from Selleckchem, 4 was from Santa Cruz, and 5 was from Millipore.

Enzymatic assays

The enzymatic activities against JAK1, JAK2, JAK3, TYK2, FLT3, BLK, TXK, BTK, ITK and AURKA were tested in Z’-Lyte assays with ATP concentrations of Km for each kinases, the activities against TTK were tested in LanthaScreen binding assays. All the protocols are available from Life Technologies.⁴⁰

Protein expression and purification

Human JAK3 kinase domain (residues 811–1124) was expressed and purified as described previously⁴¹ except that JAK3 was co-expressed with human c-SRC (residues 86–536) to achieve more consistent phosphorylation of JAK3.

Crystallization and structure determination

Crystals of JAK3 in complex with 9 were prepared using the hanging drop vapor diffusion method; JAK3 at a concentration of 5 mg/mL was incubated for 30 min with 500 μM compound 9, then added to an equal volume of well solution (0.1 M bis-tris pH 6.5, 0.2 M ammonium sulfate, 16% PEG 3350, 5 mM TCEP) and equilibrated over well solution at 4 °C. Crystals were looped into cryoprotectant solution (0.1 M bis-tris pH 6.5, 0.2 M ammonium sulfate, 20 % PEG 3350, 15% glycerol, 5 mM TCEP) for 30–60 sec, then flash-frozen in liquid nitrogen. X-ray diffraction data were collected at APS beamline 24-ID-E (NE-CAT), processed with XDS,⁴² and scaled with Scala.⁴³ The structure was solved by molecular replacement with Phaser⁴⁴ using the JAK3 kinase domain as a search model (PDB 1YVJ).⁴¹ Manual refitting of the crystallographic model was performed with Coot⁴⁵ and refinement was performed with Phenix⁴⁶ and BUSTER.⁴⁷ Topology and parameter files for compound 9 were generated with PRODRG.⁴⁸

Mass spectrometry

JAK3 protein (5 μg) was labeled with a 10-fold excess of inhibitor or DMSO for 30 minutes at 4 °C. After labeling, proteins were desalted using 0.5 mL Zeba spin desalting columns (Thermo Fisher Scientific). Analysis of intact protein was performed essentially as described.⁴⁹ Briefly, ~5 μg protein was injected onto a self-packed reversed-phase column (500 μm inner diameter, 5 cm of POROS 50R2 resin). After washing to remove salts, protein was eluted with an HPLC gradient (0%–100% B in 1 min, A = 0.2 M acetic acid in water, B = 0.2 M acetic acid in acetonitrile, flow rate = 10 μL/min) into a linear ion trap mass spectrometer (LTQ, Thermo Fisher Scientific, San Jose, CA). Data were acquired in profile mode scanning m/z 300–2000. Mass spectra were deconvoluted using MagTran software (version 1.03b2).⁵⁰ Sites of covalent modification were identified using a “bottom-up” strategy. Desalted proteins were reduced with tris(2-carboxyethyl)phosphine (10 mM, 10 min room temperature), alkylated with iodoacetamide (20 mM, 30 min room temperature in dark), and digested with trypsin for 4 h at 37 °C. Digests were analyzed by nanoLC-ESI-MS as described with modifications.⁵¹ Peptides were loaded onto a self-packed pre-column (4 cm POROS10R2), resolved on an analytical column (30 μm I.D., packed with 12 cm C18) and eluted into the mass spectrometer (LTQ Orbitrap XL, Thermo Fisher Scientific) using an HPLC gradient (Waters NanoAcquity, Milford, MA; 0%–35% B in 60 min, A = 0.2 M acetic acid in water, B = 0.2 M acetic acid in acetonitrile, flow rate = ~30 nL/min). The instrument was operated in data dependent mode such that the top 10 most abundant precursors were subjected to MS/MS (electron multiplier detection, relative collision energy 35%, q = 0.25). Raw data files were converted to .mgf using in-house software⁵² and searched using Mascot 2.2.1 against a forward-reverse human refseq database. Search parameters specified variable oxidation of methionine, variable inhibitor modification (9, 45, or 50) of cysteine, and fixed carbamidomethylation of cysteine (i.e. cysteines are considered carbamidomethylated or inhibitor labeled). To confirm labeling sites, internally calibrated HCD spectra (image current detection, resolution at m/z 200 = 15,000, relative collision energy 35%) were acquired using an Orbitrap Fusion mass spectrometer (Thermo Fisher Scientific). For these experiments, LC parameters were similar to those described above, except that the analytical column was packed with 50 cm of C18 and peptides were eluted with a gradient of 0–35% B in 90 min.

Mouse liver microsomal stability

The MLM assays were previously reported and are commercially available from Scripps Florida.³⁴

Ba/F3 cell viability assays

Ba/F3 derivatives expressing various oncogenic fusion kinases, TEL-JAK1, TEL-JAK2, TEL-JAK3 and TEL-ABL were described previously.²⁶ Ba/F3 cells transformed by TYK2^E957D were generated as previously described.³⁰ These cells were maintained in RPMI-1640 medium (GIBCO) supplemented with 10% fetal bovine serum (FBS) (Sigma-Aldrich) and penicillin/streptomycin (Invitrogen). For cell viability assays, the cells were plated at a density of 10,000 cells per well in a 96-well white plate and incubated with DMSO or increasing concentrations of drugs. At 72 h after the initiation of treatment, the relative cell viability was determined using the Cell Titer Glo assay (Promega) and reported as a percentage of the DMSO control. Concentration values for 50% inhibition (IC₅₀) of cell viability were determined with GraphPad Prism software.

Immunoblotting analysis and washout experiment

Ba/F3 cells transformed by TEL-JAK3 were incubated with JAK3 inhibitors or DMSO for 3 h. Then whole-cell lysates were prepared in RIPA buffer (Cell Signaling) with FOCUS™, ProteaseArrest™ (G-Biosciences) and Phosphatase Inhibitor Cocktail Set II (EMD Millipore). Western blotting analyses were conducted after separation by SDS-PAGE electrophoresis and transfer to nitrocellulose membranes. Immunoblotting was performed with each of specific antibodies to STAT5, phospho-STAT5 (Tyr694), α-tubulin (Cell Signaling Tech) and JAK3 (Santa Cruz, #sc-513). For washout experiment, cells were incubated with JAK3 inhibitors or DMSO for 3 h, washed with PBS three times, maintained in RPMI-1640 medium with 10% of FBS for 4 h, and then harvested and the resulting lysates were analyzed. Primary bone marrow derived macrophages (BMDMs) were prepared as described.⁵³ Cells were maintained on bacterial grade DMEM supplemented with 10% heat-inactivated fetal bovine serum, 2 mM L-glutamine, 100 U/ml of penicillin G, 100 μg/ml of streptomycin, 0.25 μg/ml of amphotericin (Invitrogen), and 5 ng/ml of macrophage colony-stimulating factor (M-CSF; PreProTech) for 7 days. Cells were then replated onto tissue culture grade plastic and used the following day. Following stimulation cells were lsyed into SDS sample buffer and run on 10% polyacrylamide gels according to standard techniques. Proteins were transferred onto nitrocellulose membranes and blotted with antibodies against total ERK1/2, p-Y641 STAT6, p-Y701 STAT1, pY701 STAT3 or pY694 STAT5 (all from Cell Signaling Tech). The experiments in cancer cell lines were done with similar procedures.

Streptavidin pulldown experiment

Ba/F3 cells transformed by TEL-JAK3 were incubated with JAK3 inhibitors or DMSO for 3 h, and then lysed with Pierce IP lysis buffer (Thermo Scientific) with Halt™ protease inhibitor cocktail (Thermo Scientific). The lysates were treated with 85 (5 μM) at 4 °C overnight, and then further incubated at room temperature for 3 h. The solution was mixed with Streptavidin beads (25 μL, Sigma) and incubated with rotation at 4 °C for 2 h. The beads were washed five times using the lysis buffer with 4 M urea (1 mL), and then boiled in the presence of Lammli sample buffer with 2-mercaptoethanol. The eluted proteins were immunoblotted with each of specific antibodies.

In Vivo Pharmacokinetic Studies

Male Swiss albino mice were dosed via tail vein (intravenous, 0.1% v/v Tween 80, 0.5% w/v NaCMC in water at a dose of 10 mg/kg) or via oral gavage (suspensions in 5% NMP, 5% solutol HS in normal saline intravenously via tail vein at a dose of 2 mg/kg). Blood samples were collected at 0.08, 0.25, 0.5, 1, 2, 4, 8 and 24 h (i.v.) and at Predose, 0.25, 0.5, 1, 2, 4, 6, 8 and 24 h (p.o.). Plasma samples were separated by centrifugation of whole blood and stored below −70°C until bioanalysis. All samples were processed for analysis by protein precipitation using acetonitrile and analyzed with fit-for-purpose LC/MS/MS method (LLOQ, 1.06 ng/mL). Pharmacokinetic parameters were calculated using the non-compartmental analysis tool of WinNonlin Enterprise software (version 6.3).

Mouse treatment studies

Kras^G12D;Lkb1^L/L mice were intrathoracically injected with adenovirus expressing Cre recombinase (2.5×10⁷ PFU/10 μl) at 6 weeks of age to initiate tumor formation.^38b All experimental mice were maintained on a mixed genetic background (C57Bl/6, Balb-c, and S129). MR imaging and tumor quantification were performed as described previously.^38b Compound 9 and vehicle controls were given per os once daily at a 75 mpk dose.

Total mouse lung cell and tumor infiltrating immune cell characterization was performed as previously described.⁵⁴ Splenocytes were isolated similarly without collagenase treatment. The antibodies used for immune analysis are listed in supplementary methods. Acquisition of eight color samples was performed on a BD Canto II cytometer equipped with Diva software and analyzed using Flowjo. All in vivo experiments performed in the PK studies and this study were approved by the Animal Care and Use Committee of the Dana-Farber Cancer Institute.