Development of a Covalent Inhibitor of c-Jun N-Terminal Protein Kinase (JNK) 2/3 with Selectivity over JNK1

Abstract

The c-Jun N-terminal kinases (JNKs) are members of the mitogen-activated protein kinase (MAPK) family, which includes JNK1–JNK3. Interestingly, JNK1 and JNK2 show opposing functions, with JNK2 activity favoring cell survival and JNK1 stimulating apoptosis. Isoform-selective small molecule inhibitors of JNK1 or JNK2 would be useful as pharmacological probes but have been difficult to develop due to the similarity of their ATP binding pockets. Here, we describe the discovery of a covalent inhibitor YL5084, the first such inhibitor that displays selectivity for JNK2 over JNK1. We demonstrated that YL5084 forms a covalent bond with Cys116 of JNK2, exhibits a 20-fold higher K_inact/K_I compared to that of JNK1, and engages JNK2 in cells. However, YL5084 exhibited JNK2-independent antiproliferative effects in multiple myeloma cells, suggesting the existence of additional targets relevant in this context. Thus, although not fully optimized, YL5084 represents a useful chemical starting point for the future development of JNK2-selective chemical probes.

Graphical Abstract

INTRODUCTION

c-Jun N-terminal kinases (JNKs) are key signal transducers in the MAPK signaling pathways that are essential for relaying extracellular stimuli to an appropriate cellular response.¹ JNK kinases have three isoforms: JNK1 and JNK2, which are widely expressed, and JNK3, which is predominantly expressed in the brain.² While JNK3 has been investigated for its role in neurodegenerative disorders, the biological functions of JNK1 and JNK2 are multifaceted and emerging evidence suggests that they may possess opposing functions.^3,4 For example, Liu et al. reported that JNK1, but not JNK2, is essential for TNFα-induced apoptosis.³ Similarly, a lack of JNK1 in fibroblast cells results in resistance to UV-induced cell death. In contrast, JNK2-deficient cells display enhanced sensitivity to UV irradiation.⁴ In cell lines derived from multiple myeloma (MM), JNK2 is constitutively activated and suppresses JNK1-mediated apoptosis.⁵ In melanoma-derived cell lines, activation of JNK2 and suppression of JNK1 positively correlate with cell proliferation and invasiveness.⁶ Proteasome inhibitor-induced apoptosis in cancer cells requires JNK1, whereas JNK2 inhibits apoptosis under these conditions.⁷ Collectively, these studies suggest that JNK1 is pro-apoptotic, whereas JNK2 is pro-survival but in a context-dependent manner. This framework supports developing JNK2-selective inhibitors to enable investigation of JNK2-specific pharmacology.

Several pan and isoform-selective JNK inhibitors have advanced as far as clinical trials, but without favorable outcomes.^8–11 Designing an isoform-selective JNK inhibitor is challenging due to the high sequence identity especially within the ATP binding pocket where JNK1 and JNK2 share 98% amino acid identity and differ by only two amino acids (Met77 vs Leu77 and Ile106 vs Leu106), whereas JNK2 and JNK3 differ by only one amino acid (Leu77 vs Met115) (Figure S1).^9,12 A majority of the reports on JNK isoform-selective inhibitors have focused on JNK1 and JNK3 due to their distinct biology and tissue-specific expression.¹³ Previously, our lab reported the first covalent pan-JNK inhibitor, JNK-IN-8, that targets a cysteine residue near the ATP binding pocket that is conserved in JNK1–JNK3.¹⁴ Here we document our optimization efforts to exploit subtle differences in the ATP binding pocket to develop a new series of isoform-selective covalent JNK inhibitors. Through these efforts, we arrived at YL5084, an inhibitor that exhibits selectivity for JNK2 over JNK1.

RESULTS AND DISCUSSION

YL5084 Selectively Inhibits JNK2 and JNK3 over JNK1.

We developed YL5084 through systematic modification of JNK-IN-8, a pan-JNK inhibitor that covalently targets a cysteine conserved in JNK1–JNK3 (Figure 1). A previous structure–activity relationship (SAR) study of JNK-IN-8 yielded a slightly more potent compound, JNK-IN-11, in which a 4-phenylpyrazolo[1,5-a]pyridine replaces the pyridine group present in JNK-IN-8.¹⁴ Interestingly, one of its regioisomers, THZ-3–60-1, exhibited selectivity for JNK2 versus JNK1 in a fixed time point biochemical kinase assay. Unfortunately, THZ-3–60-1 exhibited poor kinome-wide selectivity when tested against a panel of more than 400 kinases and displayed strong binding to EGFR, CK1, DDR1, CDK7, and Aurora A and B (Table S1). In a parallel project that aimed to improve the kinase selectivity of THZ1, a covalent inhibitor of CDK7, CDK12, and CDK13, we discovered that replacing the phenyl with a (R)-3-amino piperidine group (THZ531) improved selectivity for CDK12 and CDK13 while also improving overall kinome selectivity (Figure 1).¹⁵

Figure 1.

Evolution of YL5084 starting from JNK-IN-8 and THZ531.

Building on this discovery, we installed a series of saturated rings, including piperidine, pyrrolidine, azepane, and cyclohexane, as the replacement for the phenyl ring of THZ-3–60-1. As shown in Table 1, all of these replacements and their enantiomers retained JNK2 inhibitory activity. Among them, YL2056 displayed the highest selectivity for JNK2 versus JNK1, and in contrast, its enantiomer YL2012 retained activity on JNK1 and thus significantly diminished selectivity. Therefore, we fixed (R)-3-aminopyrrolidine and explored various substitutions to the pyrimidine and pyrazolo[1,5-a]pyridine, which resulted in the discovery of YL5129, YL5188, and YL5189. YL5189 demonstrated that there is room for modification at C6 of pyrazolo[1,5-a]pyridine, while YL5129 and YL5188 showed diminished activity, suggesting that positions of 5-pyrimidine and C4 of pyrazolo[1,5-a]pyridine are less permissive of modification. Furthermore, the introduction of a halogen to the phenyl group was tolerated for activity against both JNK1 and JNK2; however, the selectivity was compromised (YL5130, YL5198, and YL5199).

Table 1.

Fixed Time Point IC₅₀ Values on JNK1 and JNK2 Inhibition for YL2056 and Its Analogues (1 h incubation)

Compound	I	X	R1	R2	R3	IC50 (nM)a		Selectivity
						JNK1	JNK2	JNK1/JNK2
YL1035		H	H		H	53±5	5±0	10
YL1042		H	H		H	470±51	17±2	28
YL2056		H	H		H	166±35	5±1	33
YL2012		H	H		H	10±1	2±1	5
YL2048		H	H		H	387±31	17±6	23
YL2079		H	H		H	256±8	28±1	9
YL1128		H	H		H	475±20	44±2	11
YL5129		Cl	H		H	3047±142	483±6	6
YL5188		H	Me		H	1140±125	142±7	8
YL5189		H	H		Me	18±2	1±0	18
YL5130		H	H		H	67±3	4±0	17
YL5198		H	H		H	74±4	4±1	19
YL5199		H	H		H	62±2	7±1	9

^aValues are means ± SEM. The experiments were performed in duplicate.

To confirm that these compounds maintain the ability to form a covalent bond with Cys116 of JNK2, we analyzed the purified JNK2 protein by LC/MS after incubation with YL2056 for 60 min at room temperature and observed a mass shift consistent with addition of a single molecule of YL2056. To confirm the modification site, the labeled protein was digested with trypsin and peptides were analyzed by capillary electrophoresis-MS/MS. A database search revealed exclusive modification on JNK2 Cys116 (Figure 2).

Figure 2.

Mass labeling studies of recombinant JNK2 protein with the covalent JNK2 inhibitor YL2056 (room temperature, 60 min incubation).

Crystal Structure of JNK2 Bound to YL2056.

To elucidate the structural basis for selectivity, we obtained crystal structures of JNK2 (residues 4–364) bound to AMP [Protein Data Bank (PDB) entry 7N8T] or YL2056 (PDB entry 8ELC). The crystals diffracted to 1.6 and 2.0 Å, respectively, with excellent collection statistics (Table S2). The structures were determined using molecular replacement, starting from PDB entry 3NPC as a search model.¹⁶ The final models were refined to an R_free value of 21% for each.

Comparison of the adenosine monophosphate (AMP) and inhibitor-bound structures showed a large shift in the activation loop upon binding of YL2056. Additionally, changes were observed in the C-terminal loop (Figure 3A). The movement of the activation loop appears to be driven, in part, by hydrophobic interactions between Tyr185 and a portion of the covalent linker of YL2056. On the basis of this movement, it is likely that the conformational dynamics of the activation loop contribute to YL2056 binding.

Figure 3.

Structure of JNK2 bound to YL2056. (A) Comparison of the overall architectures of AMP (green) and YL2056-bound (red) X-ray structures. Large changes were seen in the activation loop and C-terminal loop. (B) F_o − F_c electron density map contoured at 2.5σ. (C) Schematic of ligand–protein interactions.

The density for YL2056 and residues surrounding the binding pocket was of high quality, allowing for modeling of inhibitor interactions with a high degree of confidence (as shown in Figure 3B). There was also strong continuous density extending from Cys116, confirming covalent bond formation. Hydrogen bonds were observed with hinge residue Met111 and the pyrazolopyrimidine with Lys55 in the back pocket (as shown in Figure 3C). We also noted that the portion of the ligand bound to the back pocket showed lower B factors, averaging around 43 Ų for the phenyl-dihydropyrazolopyridine compared to around 79 Ų for the remainder of the ligand. We therefore maintained the phenyl-dihydropyrazolopyridine and optimized other parts of the ligand.

Development of YL5084 with an Improved Selectivity.

We investigated the selectivity of YL2056 against a panel of >400 kinases using KINOMEscan at a concentration of 1 μM, which revealed JNKs as the highest-scoring targets in addition to p38α, CK1γ, and PIKFYVE (Figure S1 and Table S3). To complement this in vitro profiling, we also performed in-cell-based profiling by KiNativ in which pretreated SUM159 cells with 1 μM YL2056 were lysed and then chased by a nonspecific ATP or ADP-biotin probe that covalently labels kinases on the conserved catalytic lysine.¹⁷ As shown in Table S4, JNK2 was significantly protected from the ATP or ADP-biotin probe and p38α, CK1γ, and PIKFYVE are main off-targets among 237 detected kinases profiled by KiNativ. While available crystal structures of both CK1γ and p38α do not show a targetable cysteine in the ATP pocket, a sequence alignment of PIKFYVE (PIP5K3) with PIP4K2A, another phosphatidyl inositol kinase that has been previously targeted with covalent compounds, suggests cysteine 1970 could be targeted in PIKFYVE (Figure S2A).¹⁸ Furthermore, as a characteristic phenomenon resulting from PIKFYVE inhibition,¹⁹ vacuole formation in MDA-MB-231 was induced by YL2056, but not its reversible counterpart (YL2056R), suggesting functional inhibition of PIKFYVE in a cellular context (Figure S2B). Vacuole formation as a consequence of PIKFYVE inhibition would confound the biology of cellular JNK2 inhibition and makes YL2056 unsuitable as a chemical probe. To engineer out PIKFYVE inhibition, we undertook further chemical modifications and introduced a “flag methyl” that is known as a selectivity determinant for JNK-IN-8¹⁵ into the pyrrolidine ring in YL2056. Thus, four diastereomers (YL5084, YL6017, YL6038, and YL6016) were synthesized. As shown in Table 2, YL5084 retained the best selectivity albeit with a reduced potency. We also tested the inhibitory activity of YL5084 against JNK3, the JNK isoform that is predominantly expressed in the brain, in a commercial Z′-LYTE assay from Invitrogen (Figure S2C). The results showed that although YL5084 inhibited JNK3 with an IC₅₀ value of 84 ± 10 nM, it has a relatively lower maximum inhibition indicating YL5084 shows preferential inhibition against JNK2. We have also profiled the metabolic stability and intrinsic reactivity of YL5084.²⁰ The result showed that YL5084 has moderate metabolic stability in human and mice microsomes with half-lives of 16 and 11 min, respectively (Table S5). The half-life of adduct formation with glutathione (GSH t_1/2) is 46 min with afatinib and ibrutinib as reference controls (Table S6).

Table 2.

Fixed Time Point IC₅₀ Values for JNK1 and JNK2 Inhibition for YL5084 and Its Analogues (1 h incubation)

Compound	I	IC50 (nM)a		Selectivity
		JNK1	JNK2	JNK1/JNK2
YL5084		2173±90	70±1	31
YL6016		160±2	15±3	11
YL6017		237±16	27±3	9
YL6038		89±15	5±1	18

^aValues are means ± SEM. The experiments were performed in duplicate.

Compared with YL2056, YL5084 displayed weaker inhibition against PIKFYVE (44% inhibition at 1 μM) in KINOMEscan. In a KdELECT binding assay, the K_D values of YL2056 and YL5084 against PIKFYVE were 270 and 5000 nM, respectively, further corroborating the improved selectivity over PIKFYVE (Figure S2D). To further characterize PIKFYVE engagement in a more quantitative way, we tested JNK2 inhibitors in a PIKFYVE NanoBRET assay that uses inhibitor-induced quenching of bioluminescence resonance energy transfer (BRET) between a fluorophore-labeled ligand and luciferase-tagged target as a read-out for on-target binding.²¹ The results showed that YL2056 binds to PIKFYVE more strongly than its reversible counterpart (YL2056R), further corroborating PIKFYVE as a covalent off-target of YL2056 (Figure 4A). Interestingly, YL5084 displayed a significantly lower level of binding for PIKFYVE in the KINOMEscan and KiNativ assays compared to YL2056 (Figure 4B and Tables S7 and S8). In addition, up to a concentration of 4 μM, YL5084 did not induce vacuole formation, a phenotype known to result from PIKFYVE inhibition (Figure S2E). It is noteworthy that p38α was also identified as YL2056 off-target in the KiNativ assay; thus, we examined YL5084 together with its reversible counterpart with the p38α SelectScreen kinase assay. As shown in panels A and B of Figure S3, we observed that both YL5084 and YL5084R show a low nanomolar IC₅₀ of ~15 nM in biochemical assays and similar activity in NanoBRET engagement assays, indicating that p38α is most likely noncovalently targeted by YL5084.

Figure 4.

YL5084 showed improved kinome selectivity in cells. (A) PIKFYVE NanoBRET assay. Transfected HEK293T cells were treated with YL2056, YL2056R, YL5084, and PIKFYVE reference inhibitor Apilimod at the indicated concentrations for 3 h. Reported as means ± SD for three biological replicates. (B) KiNativ data from MM.1S cells showing the JNK family kinase is selectively targeted by YL5084 at a concentration of 1 μM.

Having engineered out PIKFYVE activity and keeping p38α noncovalent binding in mind, we further characterized the kinetic properties of YL5084, specifically, the potency of reversible binding (K_I) and the rate of irreversible reaction (k_inact).²² This was done using an in vitro system that utilizes a sulfonamido-oxine (Sox)-based chemosensor to detect peptide phosphorylation.²³ In this assay, the Sox chromophore is incorporated as a modified amino acid within the peptide substrate. When the peptide substrate is phosphorylated, it binds to Mg²⁺, capable of yielding a fluorescent signal. This analysis confirmed that YL5084 is selective for JNK2 versus JNK1 whereas JNK-IN-8 is almost equipotent against both (Figure 5A and Figure S4). The kinetic parameter k_inact/K_I for YL5084 was 335 and 7166 M⁻¹ s⁻¹ for JNK1 and JNK2, respectively, reflecting the tighter binding and faster covalent reaction with JNK2 (Figure 5B). Taken together, these results indicate that YL5084 has selectivity for JNK2 over JNK1, and an improved off-target profile, with no PIKFYVE activity. Thus, we advanced YL5084 into cell-based evaluations.

Figure 5.

Kinetic analysis of YL5084 against JNK1 and JNK2. (A) Biochemical activities of JNK1 and JNK2 with variable YL5084 concentrations shows a greater response for JNK2 to probe. (B) Data from panel A were processed to obtain k_inact/K_I values.

YL5084 Selectively Binds JNK2 over JNK1 in Cells.

To verify that in vitro selectivity between JNK2 and JNK1 is maintained in a cellular context, we performed a cell-based pull-down competition assay with biotin-JNK-IN-7, an established probe for pull down of nonspecific JNKs (Figure S5).²⁴ Multiple myeloma MM.1S cells were treated with YL5084, JNK-IN-8, or DMSO for 6 h, followed by whole cellular lysate extraction. The lysates were then incubated with 1 μM biotin probe that competes with YL5084 for covalent JNK1 and JNK2 binding. Western blotting was used to evaluate cellular target engagement. Western blotting for JNK1 and JNK2 showed that YL5084 strongly engages JNK2 at ~500 nM but fails to efficiently engage JNK1 even at a high concentration of 2 μM. As a control, JNK-IN-8 engaged both JNK1 and JNK2 at the same concentration (Figure 6A). In addition, we applied intracellular NanoBRET as an orthogonal assay for target engagement.²⁵ As shown in Figure 6B, YL5084 retains JNK2 activity similar to that of JNK-IN-8 but significantly less engagement of JNK1 in HEK293T cells (Figure 6B). These data demonstrate that YL5084 binds to JNK2 in cells, with selectivity over JNK1 that mirrors the in vitro results.

Figure 6.

YL5084 selectively engages JNK2 in cells. (A) Pretreatment with YL5084 or JNK-IN-8 for 6 h blocks pull down of JNK1 and JNK2 by biotin-JNK-IN-7 in MM cells. (B) Cellular JNK1 and JNK2 engagement in HEK293T cells exposed to YL5084, YL5084R, or JNK-IN-8. Reported as means ± SD for three biological replicates.

Mechanisms of YL5084 Selectivity for JNK2 and JNK3.

To understand the mechanisms of selectivity of YL5084 for JNK2 and JNK3 over JNK1, we performed computational studies. We focused on the differences in residues involved in ligand interactions among the highly conserved primary sequences of JNK1–JNK3 (Figure S6). We found that Leu106, which is conserved in JNK2 and JNK3 but not in JNK1, interacts with the phenyl of YL2056 in our crystal structure. To study the impact of this difference on ligand binding, we created homology models of JNK1 and JNK3 based on the X-ray structure of JNK2:YL2056 using Prime.²⁶ We docked YL5084 into the JNK2 structure using Glide²⁷ and found that the top-scoring pose was similar to the binding mode of YL2056 in our X-ray structure (Figure S7). We then calculated the binding energy of this pose in the contexts of JNK1 and JNK3 using Glide.²⁷ We found an increase in binding energy from −8.61 kcal/mol in JNK2 and −8.56 kcal/mol in JNK3 to −6.1 kcal/mol in JNK1, consistent with selectivity for JNK2 and JNK3 over JNK1. In addition, when we docked YL5084 into the JNK1 model, the top-scoring pose did not occupy the back pocket in JNK1 likely due to clashes with Ile106 (Figure S8).

The interaction of the back pocket with the ligand also involves the backbone of Val54 in JNK2. However, the analogous residue in JNK1 and JNK3 is a larger isoleucine. This difference could suggest that the main chain is more constrained in JNK1 and JNK3 than in JNK2 because bulkier residues on the opposite side of the main chain do not allow motion to accommodate the ligand. To test this idea, we compared X-ray structures of the three isoforms. The structure of JNK2:Y2056 shows that, relative to the JNK2:AMP structure, the β sheet of Val54 moves approximately 0.6 Å to accommodate the ligand (Figure S9A). Superposition of JNK1 and JNK3 structures bound to ligands that do not occupy their back pockets (PDB entries 3ELJ²⁸ and 4KKE²⁹) with the JNK2:YL2056 structure shows a 0.8 Å difference in the main chain position away from the ligand binding pocket, supporting our hypothesis (Figure S9B).

We also studied the molecular dynamics of JNK1–JNK3 using computational simulation. We performed 500 ns molecular dynamics simulations in Desmond (Desmond Molecular Dynamics System, D. E. Shaw Research, New York, NY, 2021; Maestro-Desmond Interoperability Tools, Schrödinger, New York, NY, 2021) for all three proteins bound noncovalently to YL5084, based on the assumption that selectivity is conferred by noncovalent interactions. These simulations showed preferential interactions between the inhibitor and JNK2 over JNK1 and JNK3 based on an MM-GBSA binding energy calculation³⁰ (Figure S10A,B). Consistently, the root-mean-square deviation (rmsd) of the ligand in JNK1 and JNK3 is large, at ~4 Å, compared to that of JNK2, which showed an average rmsd of 2 Å. We also noted that ligands bound to JNK2 were more likely to occupy a geometry that placed the covalent warhead in a reactive trajectory to Cys116 of JNK2 than the analogous cysteines in JNK1 and JNK3 (Figure S10C,D).

Further analysis of the protein’s conformations during simulation revealed that the P and activation loops were highly flexible, in comparison to other regions of the protein (Figure 7A,B). The P loop in JNK2 was found to be closer to the ligand than in JNK1, with JNK3 exhibiting an intermediate behavior (Figure S11A,B). These dynamics may be influenced by sequence differences at positions 106 and 54, as previously discussed. However, other residues may also contribute. An analysis of the simulation trajectory of JNK1 revealed that Arg50, a residue near the P loop, forms a salt bridge with Glu109 (Figure S12A,B). This interaction cannot occur in JNK2 because it contains an isoleucine (Ile50) at the analogous structural position, which cannot interact in the same way. This led to the hypothesis that the Arg50–Glu109 interaction may influence the flexibility of the P loop. To test this idea, we repeated the simulation for JNK2 with an I50R mutation. In this model, Arg50 strongly interacted with Glu109 as expected (Figure S13A) and we observed a substantial shift in the P loop away from the ligand (Figure S13B). This suggests that the Arg50–Glu109 interaction in JNK1 likely alters P loop dynamics in a way that is unfavorable for interactions with this class of ligand.

Figure 7.

Molecular dynamics (MD) simulation of JNKs with YL5084. (A) Root-mean-square fluctuation (rmsf) plot obtained from a 500 ns MD simulation showing that the P and activation loops are highly flexible. (B) Overlay of average simulated structures of the P and activation loops. JNK1 loops are colored red, JNK2 loops green, and JNK3 loops blue. Compound YL5084 is colored yellow. Cys116 is colored black. For better visualization, the less mobile portions are shown in white and come from the avereage JNK2 model.

YL5084 Impairs Proliferation and Induces Apoptosis in MM Cells in a JNK2-Independent Manner.

JNK2 plays critical roles in the unfolded protein response and maintenance of MM cell survival, and shRNA-mediated knockdown of JNK2 causes apoptotic cell death.⁵ We therefore evaluated the potential antiproliferative activity of JNK inhibitors in the multiple myeloma (MM) cell lines. YL5084 displayed dose-dependent antiproliferative effects with GR₅₀ values of 200–300 nM, while its reversible counterpart, YL5084R, displayed 15-fold weaker activity (Figure 8A).³¹ Exposure to YL5084 induced apoptosis as evidenced by an increased level of PARP cleavage, an increased level of cleavage of caspase 3, and an increased level of Annexin V/PI staining (Figure S14). However, JNK-IN-8, a pan-JNK inhibitor, failed to induce apoptotic cell death to the same level. In Bortezomib-resistant ANBL6 MM cells,³² YL5084 overcomes Bortezomib resistance with IC₅₀ values of 120 and 96 nM in parental ANBL6 and Bortezomib-resistant ANBL6 cells, respectively, while JNK-IN-8 showed minimal effect (Figure S15A).

Figure 8.

Antiproliferative effect of YL5084 unlikely due to JNK inhibition alone. (A) Antiproliferative effects of YL5084, YL5084R, and JNK-IN-8 in MM.1S cells determined with a growth rate inhibition assay. The Y-axis presents the calculated GR₅₀ values in micromolar. Experiments were conducted with three biological replicates. (B) Antiproliferative effects of YL5084 in MM.1S cells stably overexpressing JNK2-WT and JNK2-C116S. Experiments were conducted with three biological replicates. (C) Antiproliferative effects of YL5084 with or without 10 μM CC930 pretreatment in MM.1S cells. Experiments were conducted with three biological replicates. (D) Antiproliferative effects of YL5084, CC930, and VX745 in HAP1 cells. Experiments were conducted with three biological replicates. (E) Antiproliferative effects of YL5084 in JNK2-knockout HAP1 cells and parental cells. Experiments were conducted with three biological replicates.

To investigate whether the observed antiproliferative effects of YL5084 were on-target for JNK2, we performed additional pharmacological and genetic experiments. First, we tested the antiproliferative activity of CC930, a reversible JNK2-selective inhibitor developed as an antifibrotic agent.¹¹ In contrast to YL5084, CC930 did not exhibit antiproliferative activity even at micromolar concentrations (Figure S15B). YL5084 also shows a similar antiproliferative effect in MM.1S cells stably overexpressing wild type JNK2 or mutant JNK2 (C116S) (Figure 8B). In addition, pretreatment with 10 μM CC930 did not protect cells from the YL5084-mediated cell killing effect (Figure 8C). A similar cell killing effect has also been observed in HAP1 cells (Figure 8D). We further investigated whether p38α, a confirmed additional target of YL5084, could be contributing by performing studies with the selective p38α inhibitor VX745.³³ Interestingly, single-agent treatment with VX745 and co-treatment with CC930 and VX745 were not toxic, indicating that cytotoxicity is not due to p38α inhibition or p38α/JNK2 dual inhibition (Figure S15C). To further validate the JNK2 dependency in cancer cells, we generated a JNK2 knockout using a haploid cell line, HAP1, which prevents an unedited allele from masking a phenotype. YL5084 exhibits equipotent antiproliferative activity in HAP1 parental cells and JNK2-knockout cell lines (Figure 8E). Therefore, the observed antiproliferative and pro-apoptotic activity of YL5084 seems to result from intracellular targets other than JNK2 and p38α. Furthermore, these results suggest that JNK2 may not be a therapeutic target in multiple myeloma, which is consistent with CRISPR and RNAi data sets from the DepMap database.

CONCLUSION

We have presented the development and characterization of a series of covalent inhibitors that exhibit kinetic selectivity for JNK2 over JNK1. Starting from a pan-JNK inhibitor JNK-IN-8 as a lead, we designed and synthesized dozens of analogues and tested them in biochemical and cellular assays. We identified several analogues derived from the 2-phenylpyrazolo[1,5-a]pyridine scaffold that exhibit >20-fold selectivity for JNK2 over JNK1 as assessed through measurement of k_inact/K_I. Moreover, we obtained a high-resolution co-crystal structure of JNK2 in complex with YL2056, which demonstrated the covalent binding to Cys116 of JNK2. Further optimization on YL2056 led to the discovery of YL5084, which is a highly potent and selective JNK2 inhibitor over JNK1 in both biochemical and cellular assays. Although YL5084 demonstrates antiproliferative activity in MM.1S cells, we could not prove its dependence on JNK2 inhibition or its potential target p38α, suggesting that YL5084 might have off-targets. Therefore, a follow-up study to probe its intracellular covalent target landscape is currently ongoing. Nevertheless, the development of YL5084 is a useful advance on the road to JNK inhibitor development and represents a viable lead for further medicinal chemistry optimization.

CHEMISTRY

YL5084 and its analogues were synthesized as shown in Scheme 1. A Sonogashira reaction between 2,4-dichloropyrimidines (1) and substituted 1-ethynylbenzenes afforded compound 2, which was converted to intermediate 3 through a [3+2] cycloaddition with N-aminopyridines. Boc-protected 3-aminopyrrolidine or its variants were then coupled with intermediate 3 to provide intermediate 4, which was then deprotected and subjected to either a one-step amide coupling reaction with (E)-4-[4-(dimethylamino)but-2-enamido]benzoic acid or a four-step reaction of amide coupling (compound 5), reduction (compound 6), acrylamide formation, and dimethylamine substitution, leading to the final products.

Scheme 1. Synthesis of YL5084 and Its Analogues^a.

^aReagents and conditions: (a) PdCl₂(PPh₃)₂, CuI, Et₃N, DMF, 30 °C, overnight; (b) substituted N-aminopyridines, DBU, MeCN, 50 °C, overnight; (c) Boc-protected 3-aminopyrrolidine or its variants, NMP, DIEA, 140 °C; (d) (i) 4 N HCl/dioxane, MeOH or TFA/DCM; (ii) 4-nitrobenzoyl chloride, Py; (g) (i) 4 N HCl/dioxane, MeOH, (ii) TCFH, NMI, DMF, room temperature; (e) SnCl₂, ethyl acetate/MeOH, 80 °C; (f) (i) (E)-4-bromobut-2-enoyl chloride, DIEA, MeCN, 0 °C, (ii) dimethylamine in THF, room temperature.

EXPERIMENTAL SECTION

Compound Synthesis.

Unless otherwise noted, reagents and solvents were obtained from commercial suppliers and used without further purification. ¹H NMR spectra were recorded on 400 or 500 MHz (Mega Hertz, Bruker A500), and chemical shifts are reported in parts per million (δ) downfield from tetramethylsilane (TMS). Coupling constants (J) are reported in hertz. Spin multiplicities are denoted as s (singlet), br (broad singlet), d (doublet), t (triplet), q (quartet), and m (multiplet). Mass spectra were recorded on a Waters Micromass ZQ instrument. Preparative HPLC was performed on a Waters Sunfire C18 column (19 mm × 50 mm, 5 μM) using a gradient of 15% to 95% methanol in water or acetonitrile in water containing 0.05% trifluoroacetic acid (TFA) over 22 min (28 min run time) or 35 min (45 min run time) at a flow rate of 20 mL/min. All of the final compounds reported were >95% pure on the basis of ¹H nuclear magnetic resonance (NMR) and liquid chromatography/mass spectrometry (LC/MS) (Figure S16).

Synthesis of Compound YL5084.

2-Chloro-4-(phenylethynyl)pyrimidine.

To a solution of 2,4-dichloropyrimidine (600 mg, 4 mmol) and ethynylbenzene (500 mg, 4.8 mmol) in DMF (10 mL) were added PdCl₂(PPh₃)₂ (14 mg, 0.02 mmol), CuI (7.6 mg, 0.04 mmol), and Et₃N (5.5 mL, 40 mmol) under a N₂ atmosphere. The reaction mixture was stirred at 30 °C overnight. After cooling to room temperature, the reaction mixture was diluted with water (50 mL), extracted with ethyl acetate, dried over Na₂SO₄, and concentrated in vacuo. The residue was purified by column chromatography on silica gel (1:4 ethyl acetate/hexane) to give the title compound (640 mg, 74%). LC/MS (ESI): m/z 215 [M + H]⁺.

3-(2-Chloropyrimidin-4-yl)-2-phenylpyrazolo[1, 5-a]pyridine.

To a solution of 2-chloro-4-(phenylethynyl)pyrimidine (640 mg, 3 mmol) and 1-aminopyridinium iodide (800 mg, 3.6 mmol) in MeCN (10 mL) was added DBU (544 mg, 3.6 mmol) at 0 °C. The reaction mixture was stirred at 50 °C overnight. After cooling to room temperature, the reaction mixture was diluted with water (200 mL), and the precipitated solid was filtered to give the title compound (700 mg, 76%). LC/MS (ESI): m/z 307 (M + H)⁺.

tert-Butyl (3S,4S)-3-Methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carboxylate.

A mixture of 3-(2-chloropyrimidin-4-yl)-2-phenylpyrazolo[1,5-a]pyridine (1.70 g, 5.54 mmol), (3S,4S)-tert-butyl 3-amino-4-methylpyrrolidine-1-carboxylate (1.12 g, 5.54 mmol), Pd₂(dba)₃ (0.30 g, 0.34 mmol), BINAP (0.34 g, 0.56 mmol), and K₃PO₄ (3.50 g, 16.62 mmol) in 1,4-dioxane (60 mL) was degassed with nitrogen three times and then heated at 100 °C under a nitrogen atmosphere for 18 h. The reaction mixture was cooled and partitioned between ethyl acetate (100 mL) and water (100 mL). The organic phase was separated, and the aqueous phase was extracted with ethyl acetate (100 mL). The combined organic layers were washed with brine (100 mL*3), dried over Na₂SO₄, filtered, and concentrated in vacuo. The residue was purified with a silica gel column (MeOH/DCM, 0–15%) to afford the title compound as a white solid (3.3 g, 89.2%). LC/MS (ESI): m/z 471.4 [M + H]⁺.

N-[(3S,4S)-4-Methylpyrrolidin-3-yl]-4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-amine.

A mixture of (3S,4S)-tert-butyl 3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carboxylate (3.0 g, 6.4 mmol) and a HCl solution in 1,4-dioxane (4 M, 25 mL) was stirred at room temperature for 3 h. The reaction mixture was partitioned between ethyl acetate (100 mL) and water (100 mL). The aqueous phase was extracted with ethyl acetate (100 mL). The combined organic layers were washed with brine (3 × 100 mL), dried over Na₂SO₄, filtered, and concentrated in vacuo. The residue was purified with a silica gel column (MeOH/DCM, 0–15%) to afford the title compound as a yellow solid (1.5 g, 57.7%). LC/MS (ESI): m/z 371 [M + H]⁺.

(E)-4-(Dimethylamino)-N-[4-((3S,4S)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL5084).

A mixture of N-[(3S,4S)-4-methylpyrrolidin-3-yl]-4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-amine (0.90 g, 2.43 mmol), (E)-4-[4-(dimethylamino)but-2-enamido]benzoic acid (1.20 g, 4.86 mmol), NMI (1.00 g, 12.10 mmol), and TCFH (1.35 g, 4.86 mol) in DMF (22 mL) was stirred at 20 °C for 2 h. The reaction mixture was poured into water (300 mL), and the resulting precipitate was collected by filtration, washed with water, and purified with a silica gel column (MeOH/DCM, 0–15%) and preparative HPLC (0.05% TFA in MeOH/H₂O, 0–100%) to afford the title product as a white solid (0.56 g, 32.2%). ¹H NMR (400 MHz, methanol-d₄): δ 8.66–8.60 (m, 1H), 8.59–8.33 (m, 1H), 8.09–7.98 (m, 1H), 7.81–7.72 (m, 2H), 7.64–7.43 (m, 8H), 7.14–7.05 (m, 1H), 7.01–6.87 (m, 1H), 6.43 (s, 1H), 6.36–6.24 (m, 1H), 4.76–4.44 (m, 1H), 4.00–3.79 (m, 1H), 3.75–3.64 (m, 1H), 3.62–3.42 (m, 2H), 3.26–3.15 (m, 2H), 2.79–2.46 (m, 1H), 2.32 (d, J = 6.9 Hz, 6H), 1.08 (dd, J = 51.5, 6.9 Hz, 3H). ¹³C NMR (126 MHz, DMSO-d₆): δ 168.08, 168.00, 163.47, 162.25, 162.06, 153.28, 141.98, 140.62, 139.82, 133.38, 131.32, 131.30, 129.33, 129.31, 129.05, 128.72, 128.50, 128.27, 128.24, 126.44, 125.69, 118.46, 118.43, 113.94, 107.95, 59.70, 53.92, 53.57, 51.78, 45.14, 33.84, 11.95. LC/MS (ESI): m/z 601 [M + H]⁺.

Synthesis of Compound YL5084R.

4-(Dimethylamino)-N-[4-((3S,4S)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]butanamide (YL5084R).

To a solution of (4-aminophenyl)((3S,4S)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone (8 mg, 0.016 mmol) and DIEA (9 μL, 0.05 mmol) in anhydrous THF (1 mL) was added 4-chlorobutanoyl chloride dropwise at 0 °C until the reaction had reached completion. Then an excess of dimethylamine (2 N in dioxane) was added. The mixture was stirred at room temperature for 1 h and then concentrated in vacuo at 50 °C. The residue was purified by preparative HPLC (0.05% TFA in MeOH/H₂O, 0–100%) to give the title compound (1.4 mg, 15%). ¹H NMR (500 MHz, DMSO-d₆): δ 10.05 (s, 1H), 8.84–8.64 (m, 1H), 8.65–7.89 (m, 2H), 7.68–7.26 (m, 1H), 7.13–6.92 (m, 1H), 4.76–4.13 (m, 1H), 3.84–3.44 (m, 4H), 2.33–2.22 (m, 4H), 2.18–2.08 (m, 6H), 1.83–1.50 (m, 2H), 1.49–1.06 (m, 1H), 0.89 (d, J = 6.8 Hz, 3H). LC/MS (ESI): m/z 603 [M + H]⁺.

Synthesis of Compound YL2056.

tert-Butyl (S)-3-{[4-(2-Phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carboxylate.

To a solution of 3-(2-chloropyrimidin-4-yl)-2-phenylpyrazolo[1,5-a]pyridine (200 mg, 0.65 mmol) and tert-butyl (S)-3-aminopyrrolidine-1-carboxylate (186 mg, 1.3 mmol) in NMP (3 mL) was added DIEA (0.35 mL, 2 mmol). The reaction mixture was stirred at 140 °C overnight. The mixture was diluted with water (100 mL), extracted with ethyl acetate, washed with brine, dried over Na₂SO₄, and concentrated in vacuo. The residue was used in the next step without any purification. LC/MS (ESI): m/z 457 [M + H]⁺.

(S)-4-(2-Phenylpyrazolo[1,5-a]pyridin-3-yl)-N-(pyrrolidin-3-yl)pyrimidin-2-amine.

To a solution of tert-butyl (S)-3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carboxylate (crude from last step) in dioxane (5 mL) was added 5 mL of HCl (4 N in dioxane). The reaction mixture was stirred at room temperature overnight and then diluted with ethyl acetate. The residue was filtered to give the title compound as a HCl salt (160 mg, 63% for two steps). LC/MS (ESI): m/z 357 [M + H]⁺.

(S)-(4-Nitrophenyl)(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone.

To a solution of (S)-4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)-N-(pyrrolidin-3-yl)pyrimidin-2-amine (90 mg, 0.25 mmol) in pyridine (3 mL) was added 4-nitrobenzoyl chloride (70 mg, 0.38 mmol). The reaction mixture was stirred at room temperature overnight and then concentrated in vacuo. The residue was redissolved in water (200 mL) and then extracted with a 1:3 (v/v) isopropanol/chloroform mixture. The combined organic layer was concentrated in vacuo to give the crude product, which was used in the next step without any purification. LC/MS (ESI): m/z 506 [M + H]⁺.

(S)-(4-Aminophenyl)(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone.

To a solution of (S)-(4-nitrophenyl)(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone (crude from the last step) in 5 mL of a 1:1 (v/v) ethyl acetate/methanol mixture was added SnCl₂ (380 mg, 8 mmol). The reaction mixture was stirred at 80 °C for 2 h. After cooling to room temperature, the reaction mixture was diluted with saturated aqueous Na₂CO₃. The resulting mixture was extracted with a 1:3 (v/v) isopropanol/chloroform mixture. The combined organic layer was concentrated in vacuo and then purified by preparative HPLC (0.05% TFA in MeOH/H₂O, 0–100%) to give the title compound (68 mg, 45% for two steps) as a TFA salt. LC/MS (ESI): m/z 476 [M + H]⁺.

(S,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL2056).

To a solution of (S)-(4-aminophenyl)(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone (27 mg, 0.057 mmol) and DIEA (28 μL, 0.17 mmol) in anhydrous THF (1 mL) was added (E)-4-bromobut-2-enoyl chloride dropwise at 0 °C until the reaction had reached completion. Then an excess of dimethylamine (2 N in dioxane) was added. The mixture was stirred at room temperature for 1 h and then concentrated in vacuo at 50 °C. The residue was purified by preparative HPLC (0.05% TFA in MeOH/H₂O, 0–100%) to give the title compound (30 mg, 73%). ¹H NMR (500 MHz, methanol-d₄): δ 8.86–8.34 (m, 2H), 8.09–8.03 (m, 1H), 8.01–7.85 (m, 1H), 7.81–7.35 (m, 9H), 7.33–7.18 (m, 1H), 6.98–6.80 (m, 1H), 6.78–6.48 (m, 2H), 4.76–4.43 (m, 1H), 4.07–3.93 (m, 3H), 3.91–3.58 (m, 3H), 2.95 (d, J = 8.4 Hz, 6H), 2.49–2.34 (m, 1H), 2.31–2.13 (m, 1H). ¹³C NMR (126 MHz, DMSO-d₆): δ 168.23, 162.37, 159.19, 158.84, 158.56, 158.29, 158.02, 140.34, 140.28, 133.14, 132.19, 132.00, 131.91, 131.83, 129.53, 129.47, 129.27 129.17, 128.84, 128.77, 128.49, 128.43, 118.93, 118.91, 114.94, 114.78, 107.67, 56.96, 54.41, 51.41, 47.38, 42.27, 34.51. LC/MS (ESI): m/z 587 [M + H]⁺.

Synthesis of Compound YL2056R.

(S)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]butanamide (YL2056R).

To a solution of (S)-(4-aminophenyl)(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidin-1-yl)methanone (29 mg, 0.062 mmol) and DIEA (33 μL, 0.19 mmol) in anhydrous THF (1 mL) was added 4-chlorobutanoyl chloride dropwise at 0 °C until the reaction had reached completion. Then an excess of dimethylamine (2 N in dioxane) was added. The mixture was stirred at room temperature for 1 h and then concentrated in vacuo at 50 °C. The residue was purified by preparative HPLC (0.05% TFA in MeOH/H₂O, 0–100%) to give the title compound (24 mg, 66%). ¹H NMR (500 MHz, DMSO-d₆): δ 10.22 (s, 1H), 9.42 (s, 1H), 8.92–7.84 (m, 3H), 7.72–7.40 (m, 10H), 7.24–7.09 (m, 1H), 6.34 (s, 1H), 3.91–3.62 (m, 2H), 3.14–2.99 (m, 2H), 2.79 (m, 6H), 2.58–2.52 (m, 3H), 2.47–2.34 (m, 2H), 2.28–1.80 (m, 4H). LC/MS (ESI): m/ z 590 [M + H]⁺.

All other compounds were prepared from intermediates 4 in a fashion analogous to that of YL2056 (supplemental experimental procedure).

(S,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}piperidine-1-carbonyl)phenyl]but-2-enamide (YL1035).

¹H NMR (500 MHz, DMSO-d₆): δ 10.20 (s, 1H), 8.77 (dt, J = 6.9, 1.1 Hz, 1H), 8.52–8.35 (m, 1H), 8.03 (d, J = 5.5 Hz, 1H), 7.72–7.55 (m, 4H), 7.54–7.43 (m, 4H), 7.36 (d, J = 8.2 Hz, 2H), 7.19–7.06 (m, 2H), 6.86–6.70 (m, 1H), 6.46 (d, J = 15.4 Hz, 1H), 6.30 (d, J = 5.5 Hz, 1H), 3.93–3.71 (m, 4H), 3.26–3.05 (m, 3H), 2.83 (s, 6H), 2.11–1.81 (m, 2H), 1.74–1.51 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 171.09, 169.13, 162.07, 162.00, 158.81, 158.54, 158.27, 158.00, 140.13, 139.68, 133.06, 131.97, 131.84, 129.39, 129.00, 128.61, 127.89, 118.72, 117.67, 115.30, 110.66, 106.39, 56.75, 54.74, 47.26, 42.04, 35.20, 34.33. LC/MS (ESI): m/z 601 [M + H]⁺.

(R,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}piperidine-1-carbonyl)phenyl]but-2-enamide (YL1042).

¹H NMR (500 MHz, DMSO-d₆): δ 10.20 (s, 1H), 8.77 (dt, J = 6.9, 1.1 Hz, 1H), 8.55–8.35 (m, 1H), 8.03 (d, J = 5.5 Hz, 1H), 7.69–7.56 (m, 4H), 7.52–7.46 (m, 4H), 7.36 (d, J = 8.1 Hz, 2H), 7.24–7.06 (m, 2H), 6.83–6.69 (m, 1H), 6.46 (d, J = 15.3 Hz, 1H), 6.31 (d, J = 5.4 Hz, 1H), 3.98–3.76 (m, 4H), 3.29–3.04 (m, 3H), 2.83 (s, 6H), 2.11–1.79 (m, 2H), 1.75–1.50 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 171.10, 169.18, 162.08, 162.07, 159.02, 158.75, 158.48, 158.21, 140.21, 139.77, 133.00, 131.98, 131.85, 131.19, 129.41, 129.07, 128.64, 128.24, 127.89, 118.85, 118.79, 117.62, 115.27, 112.93, 110.66, 106.99, 106.31, 56.75, 47.29, 42.02, 40.11, 35.21. LC/MS (ESI): m/z 601 [M + H]⁺.

(R,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL2012).

¹H NMR (500 MHz, DMSO-d₆): δ 10.53 (s, 1H), 8.96–8.72 (m, 1H), 8.66–8.17 (m, 2H), 8.17–7.31 (m, 11H), 7.28–6.96 (m, 1H), 6.83–6.64 (m, 1H), 6.53–6.20 (m, 2H), 4.53–4.26 (m, 1H), 4.03–3.70 (m, 6H), 2.80 (s, 6H), 2.31–1.88 (m, 2H). LC/MS (ESI): m/z 587 [M + H]⁺.

(R,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}azepane-1-carbonyl)phenyl]but-2-enamide (YL2048).

¹H NMR (500 MHz, DMSO-d₆): δ 10.54 (s, 1H), 10.44–9.78 (m, 1H), 9.00–8.52 (m, 1H), 8.46–7.86 (m, 2H), 7.80–7.37 (m, 11H), 7.30–7.15 (m, 1H), 6.77 (m, 1H), 6.55–6.02 (m, 1H), 4.97–3.69 (m, 5H), 3.18–3.07 (m, 2H), 2.80 (s, 6H), 2.06–1.43 (m, 6H). ¹³C NMR (126 MHz, DMSO-d₆): δ 170.44, 162.36, 158.77, 158.51, 158.24, 157.97, 151.33, 149.65, 140.55, 140.25, 132.91, 132.16, 131.99, 131.93, 129.55, 129.45, 129.41, 128.89, 128.77, 127.55, 119.18, 117.94, 115.53, 115.29, 106.93, 106.37, 56.93, 53.72, 42.24, 41.98, 34.49, 18.17, 16.81, 12.57. LC/MS (ESI): m/z 615 [M + H]⁺.

(S,E)-4-(Dimethylamino)-N-[4-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}azepane-1-carbonyl)phenyl]but-2-enamide (YL2079).

¹H NMR (500 MHz, DMSO-d₆): δ 10.53 (s, 1H), 10.37–9.79 (m, 1H), 9.01–8.45 (m, 1H), 8.43–7.85 (m, 2H), 7.85–7.04 (m, 12H), 6.89–6.60 (m, 1H), 6.55–6.00 (m, 1H), 4.63–3.84 (m, 5H), 3.26–3.03 (m, 2H), 2.80 (s, 6H), 2.10–1.33 (m, 6H). ¹³C NMR (126 MHz, DMSO-d₆): δ 162.42, 158.93, 158.65, 158.38, 158.10, 140.70, 140.34, 139.25, 132.84, 132.21, 132.83, 132.03, 129.89, 129.60, 129.57, 129.33, 128.98, 128.85, 127.60, 119.24, 117.75, 115.54, 115.41, 106.85, 106.25, 56.99, 53.78, 42.30, 42.04, 34.54, 18.21, 16.84, 12.61. LC/MS (ESI): m/z 615 [M + H]⁺.

(E)-4-[4-(Dimethylamino)but-2-enamido]-N-(3-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}cyclohexyl)benzamide (YL1128).

¹H NMR (500 MHz, DMSO-d₆): δ 10.57 (s, 1H), 10.46–10.03 (m, 1H), 9.06–8.78 (m, 1H), 8.49–8.38 (m, 3H), 8.16–7.99 (m, 1H), 7.92–7.44 (m, 10H), 6.84–6.65 (m, 1H), 6.58–6.22 (m, 2H), 3.98–3.94 (m, 2H), 3.19–3.16 (m, 2H), 2.81 (s, 6H), 2.30–1.74 (m, 8H). ¹³C NMR (126 MHz, DMSO-d₆): δ 164.99, 162.41, 158.88, 158.63, 158.37, 158.11, 141.40, 140.45, 133.04, 132.23, 132.00, 129.78, 129.53, 129.29, 128.91, 128.79, 128.40, 128.31, 118.73, 118.45, 116.07, 112.96, 106.47, 56.81, 53.61, 42.04, 41.88, 34.34, 18.07, 16.76, 12.38. LC/MS (ESI): m/z 615 [M + H]⁺.

(S,E)-N-[4-(3-{[5-chloro-4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]-4-(dimethylamino)but-2-enamide (YL5129).

¹H NMR (500 MHz, DMSO-d₆): δ 10.52 (s, 1H), 9.87 (s, 1H), 8.88–8.75 (m, 1H), 8.51–8.32 (m, 1H), 7.86 (s, 1H), 7.73 (d, J = 8.5 Hz, 2H), 7.66–7.24 (m, 8H), 7.12–6.97 (m, 1H), 6.84–6.71 (m, 1H), 6.48 (d, J = 15.2 Hz, 1H), 4.70–4.04 (m, 3H), 3.96 (d, J = 7.2 Hz, 2H), 3.44–3.36 (m, 2H), 2.82 (s, 6H), 2.16–1.84 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.98, 162.31, 162.18, 158.16, 157.89, 140.12, 133.04, 132.09, 131.91, 131.84, 129.68, 129.28, 128.97, 128.62, 128.56, 128.30, 128.27, 127.59, 125.69, 125.57, 118.71, 118.15, 117.55, 117.33, 115.79, 113.54, 56.84, 56.83, 54.17, 47.25, 44.41, 42.14, 29.08. LC/MS (ESI): m/z 621 [M + H]⁺.

(S,E)-4-(Dimethylamino)-N-{4-[3-({4-[2-(4-fluorophenyl)pyrazolo[1,5-a]pyridin-3-yl]pyrimidin-2-yl}amino)pyrrolidine-1-carbonyl]phenyl}but-2-enamide (YL5130).

¹H NMR (500 MHz, DMSO-d₆): δ 10.51 (d, J = 17.4 Hz, 1H), 9.87 (s, 1H), 8.93–8.74 (m, 1H), 8.18–7.44 (m, 9H), 7.41–7.26 (m, 2H), 7.19–7.06 (m, 1H), 6.82–6.66 (m, 1H), 6.54–6.17 (m, 2H), 4.54–4.27 (m, 2H), 4.02–3.65 (m, 5H), 2.81 (s, 6H), 2.24–1.96 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.94, 163.43, 162.16, 161.48, 158.05, 157.80, 152.39, 149.25, 140.13, 132.07, 131.84, 131.56, 131.52, 131.50, 131.47, 131.45, 129.65, 129.15, 128.33, 128.28, 118.72, 118.68, 115.66, 115.48, 114.19, 114.18, 107.93, 54.48, 51.56, 51.21, 47.30, 42.12, 29.33. LC/MS (ESI): m/z 605 [M + H]⁺.

(S,E)-4-(Dimethylamino)-N-[4-(3-{[4-(4-methyl-2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL5188).

¹H NMR (500 MHz, DMSO-d₆): δ 10.51 (s, 1H), 9.85 (s, 1H), 8.70–8.53 (m, 1H), 8.35–8.12 (m, 1H), 7.77–7.65 (m, 3H), 7.56–7.25 (m, 6H), 7.13–6.74 (m, 3H), 6.65–6.45 (m, 2H), 4.61–4.18 (m, 2H), 4.04–3.86 (m, 3H), 3.44–3.30 (m, 2H), 2.82 (s, 6H), 2.26–1.85 (m, 5H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.99, 167.91, 162.16, 161.32, 158.54, 158.26, 157.98, 157.71, 157.14, 150.60, 140.09, 132.57, 132.08, 131.84, 128.43, 128.34, 128.29, 128.23, 128.17, 126.78, 126.72, 124.93, 118.65, 117.37, 113.37, 112.88, 109.41, 56.82, 54.49, 51.45, 47.27, 42.13, 31.66, 29.16. LC/MS (ESI): m/z 601 [M + H]⁺.

(S,E)-4-(Dimethylamino)-N-[4-(3-{[4-(6-methyl-2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL5189).

¹H NMR (500 MHz, DMSO-d₆): δ 10.52 (s, 1H), 9.87 (s, 1H), 8.80–8.60 (m, 1H), 8.26–7.93 (m, 2H), 7.79–7.65 (m, 2H), 7.62–7.22 (m, 8H), 6.85–6.67 (m, 1H), 6.53–6.12 (m, 2H), 4.61–4.20 (m, 1H), 4.06–3.65 (m, 6H), 2.81 (s, 6H), 2.39 (s, 3H), 2.27–1.92 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.92, 162.16, 158.53, 158.25, 157.98, 157.71, 140.16, 138.56, 133.32, 132.08, 131.82, 129.37, 129.31, 128.90, 128.80, 128.60, 128.55, 128.38, 128.30, 128.27, 127.16, 127.11, 127.04, 118.73, 118.65, 117.51, 115.16, 107.57, 56.80, 54.40, 51.24, 47.26, 42.12, 29.29, 17.46. LC/MS (ESI): m/z 601 [M + H]⁺.

(S,E)-N-{4-[3-({4-[2-(3-Chlorophenyl)pyrazolo[1,5-a]pyridin-3-yl]pyrimidin-2-yl}amino)pyrrolidine-1-carbonyl]phenyl}-4-(dimethylamino)but-2-enamide (YL5198).

¹H NMR (500 MHz, DMSO-d₆): δ 10.53 (s, 1H), 9.81 (s, 1H), 8.92–8.80 (m, 1H), 8.74–7.82 (m, 2H), 7.82–7.39 (m, 9H), 7.33–7.04 (m, 1H), 6.96–6.66 (m, 1H), 6.60–5.75 (m, 2H), 4.51–4.18 (m, 1H), 4.00–3.84 (m, 4H), 3.46–3.39 (m, 2H), 2.80 (s, 6H), 2.26–1.91 (m, 2H). ¹³C NMR (126 MHz, CDCl₃): δ 167.93, 162.16, 158.50, 158.23, 157.96, 157.70, 140.14, 135.32, 133.18, 132.07, 131.84, 130.48, 129.25, 128.89, 128.79, 128.34, 128.31, 128.17, 128.10, 127.00, 118.71, 117.69, 115.31, 114.52, 114.46, 107.97, 56.81, 54.36, 51.49, 47.28, 42.12, 29.35. LC/MS (ESI): m/z 621 [M + H]⁺.

(S,E)-N-{4-[3-({4-[2-(4-Chlorophenyl)pyrazolo[1,5-a]pyridin-3-yl]pyrimidin-2-yl}amino)pyrrolidine-1-carbonyl]phenyl}-4-(dimethylamino)but-2-enamide (YL5199).

¹H NMR (500 MHz, DMSO-d₆): δ 10.53 (s, 1H), 9.85 (s, 1H), 8.97–8.78 (m, 1H), 8.26–7.86 (m, 2H), 7.85–7.40 (m, 9H), 7.29–7.02 (m, 1H), 6.87–6.65 (m, 1H), 6.54–6.11 (m, 2H), 4.54–4.21 (m, 1H), 4.05–3.64 (m, 6H), 2.80 (s, 6H), 2.25–1.89 (m, 2H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.92, 162.15, 160.60, 158.52, 158.25, 157.98, 157.70, 155.86, 152.31, 140.13, 133.71, 133.70, 132.06, 131.82, 131.18, 131.13, 129.27, 128.67, 128.33, 128.27, 126.98, 118.68, 117.54, 115.14, 114.47, 114.38, 107.96, 107.94, 56.80, 54.38, 51.23, 47.27, 42.11, 29.31. LC/MS (ESI): m/z 621 [M + H]⁺.

(E)-4-(Dimethylamino)-N-[4-((3R,4S)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL6016).

¹H NMR (500 MHz, DMSO-d₆): δ 10.53 (s, 1H), 9.84 (s, 1H), 9.00–8.33 (m, 1H), 8.28–7.87 (m, 2H), 7.79–7.32 (m, 11H), 7.27–7.09 (m, 1H), 6.96–6.67 (m, 1H), 6.70–6.21 (m, 1H), 4.29–3.69 (m, 7H), 2.80 (s, 6H), 1.44–0.70 (m, 4H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.97, 167.91, 162.17, 161.15, 160.91, 158.54, 158.27, 158.00, 157.73, 140.20, 133.26, 132.10, 132.08, 131.84, 131.60, 131.54, 129.38, 129.30, 128.59, 128.38, 128.35, 118.79, 118.71, 117.61, 114.37, 107.80, 56.81, 55.24, 54.20, 50.72, 42.13, 35.98, 15.89. LC/MS (ESI): m/z 601 [M + H]⁺.

(E)-4-(Dimethylamino)-N-[4-((3R,4R)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL6017).

¹H NMR (500 MHz, DMSO-d₆): δ 10.52 (s, 1H), 9.88 (s, 1H), 8.94–8.78 (m, 1H), 8.74–8.02 (m, 3H), 7.82–7.42 (m, 9H), 7.33–7.09 (m, 1H), 6.87–6.68 (m, 1H), 6.57–6.15 (m, 2H), 4.71–4.34 (m, 2H), 4.04–3.70 (m, 5H), 2.81 (s, 6H), 1.28–0.84 (m, 4H). ¹³C NMR (126 MHz, DMSO-d₆): δ 167.97, 162.18, 158.56, 158.29, 158.02, 157.75, 140.16, 140.03, 133.21, 132.10, 131.85,131.78, 131.75, 129.38, 129.35, 128.96, 128.92, 128.62, 128.57, 128.33, 118.75, 118.72, 117.69, 115.31, 114.44, 114.40, 107.70, 56.82, 53.90, 51.32, 42.13, 33.93, 11.92. LC/MS (ESI): m/z 601 [M + H]⁺.

(E)-4-(Dimethylamino)-N-[4-((3S,4R)-3-methyl-4-{[4-(2-phenylpyrazolo[1,5-a]pyridin-3-yl)pyrimidin-2-yl]amino}pyrrolidine-1-carbonyl)phenyl]but-2-enamide (YL6038).

¹H NMR (500 MHz, DMSO-d₆): δ 10.51 (s, 1H), 9.84 (s, 1H), 9.00–8.73 (m, 1H), 7.86–7.32 (m, 12H), 7.21–6.99 (m, 1H), 6.84–6.62 (m, 1H), 6.55–5.97 (m, 2H), 4.27–3.68 (m, 5H), 3.28–3.06 (m, 2H), 2.81 (s, 6H), 1.34–0.72 (m, 4H). ¹³C NMR (126 MHz, DMSO-d₆): δ 168.00, 167.93, 162.20, 161.06, 161.03, 158.53, 158.26, 157.99, 157.73, 140.21, 139.96, 133.31, 132.11, 132.10, 131.86, 131.62, 131.56, 129.39, 129.31, 128.61, 128.41, 128.37, 118.82, 118.74, 114.35, 107.85, 56.84, 55.25, 53.87, 50.76, 42.16, 35.99, 15.91. LC/MS (ESI): m/z 601 [M + H]⁺.

Cell Culture.

MM1.S cells (ATCC, CRL-2974) and RPMI-8226 cells (ATCC, CCL-155) were both grown in RPMI-1640 supplemented with 10% FBS and 100 μL/mL penicillin-streptomycin. HEK293T cells (ATCC, CRL-3216) and MDA-MB-231 cells (ATCC, HTB-26) were grown in DMEM supplemented with 10% FBS and 100 μL/mL penicillin-streptomycin. Lentivectors (pHAGE-N-FLAG-HA-IRES-PURO backbone) encoding full length wild type or mutant JNK2 (C116S) with an N-terminal FLAG-HA tag were used for virus packaging in HEK293T cells. The filtered virus was used to transduce the MM1.S cells in the presense of 8 μg/mL Polybrene. Then transduced MM1.S cells were further selected by 0.5 μg/mL puromycin for 3 days. All cell lines were cultured at 37 °C in a humidified chamber in the presence of 5% CO₂, unless otherwise noted. All cells were mycoplasma tested upon initial receipt.

Immunoblotting.

Whole cell lysates for immunoblotting were prepared by pelleting cells from each cell line at 4 °C (1200 rpm) for 5 min using a Sorvall Legend centrifuge (Thermo Fisher Scientific). The resulting cell pellets were washed once with ice-cold 1× PBS and then resuspended in lysis buffer containing 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 1% NP-40, 0.1% SDS, and 5 mM EDTA with protease (Roche, catalog no. 04 693 159 001) and phosphatase inhibitors (Roche, catalog no. 04 906 837 001). Whole cell lysates were collected and snap-frozen in liquid nitrogen before being stored at −80 °C. Protein concentrations were determined by using the Pierce BCA protein assay kit (Life Technologies, catalog no. 23225). Whole cell lysates were loaded into Bolt 4–12% bis-tris gels (Invitrogen, Carlsbad, CA) and separated by electrophoresis at 90 V for 2 h. The gels were then transferred onto a nitrocellulose membrane (Bio-Rad, catalog no. 162–0115) and blocked by incubation with Odyssey blocking buffer (TBS). Membranes were probed using antibodies raised against the indicated proteins. Appropriate IR-labeled secondary antibodies were used, and LI-COR Odyssey CLx was used for detection. The following primary antibodies were used: JNK1 (Cell Signaling, catalog no. 3708S), JNK2 (Cell Signaling, catalog no. 9258S), JNKs (Cell Signaling, catalog no. 9252S), HA (Cell Signaling, catalog no. 3724S), Flag (Sigma, catalog no. F1804), β-actin (Cell Signaling, catalog no. 3700S), and vinculin (Cell Signaling, catalog no. 13901S).

Pull-Down Experiments.

Cells were treated with inhibitors or DMSO for 6 h. Following treatment, cells were washed twice with cold PBS and then lysed in lysis buffer [50 mM Hepes (pH 7.4), 150 mM NaCl, 1% Nonidet P40 substitute, 5 mM EDTA, 1 mM DTT, and protease/phosphatase cocktails]. Following clearance, lysates were treated with biotin-JNK-IN-7 at 1 μM overnight at 4 °C. Lysates were further incubated at room temperature for 2 h to increase the efficiency of covalent bond formation. Lysates were then incubated with streptavidin agarose beads (Thermo Scientific, catalog no. 20349) for pull down for an additional 2–3 h at 4 °C. Agarose beads were washed six times with lysis buffer and then boiled in 2× sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) loading buffer for 10 min at 95 °C. SDS–PAGE-resolved precipitated proteins were probed for the indicated proteins.

Vacuole Formation.

Live MDA-MB-231 cells were seeded in dishes and the next day treated with small molecule inhibitors or DMSO based on experimental design. After 3 days, cells were imaged by bright-field microscopy with 20× magnification. YL5084 retains the antiproliferative activity and cytotoxicity without causing the creation of many vacuoles

Establishment of JNK2-Hnockout Cells.

The HAP1-Cas9 single-cell clone was established using a lentivector that encodes Cas9 with BFP as a fluorescence marker. pMCB320 lentivector (Addgene, catalog no. 89359) containing the following guide RNA were used to generate JNK2-knockout HAP1 cells: 5′-GTGCATTTGATACAGTTCTT-3′ or 5′-GACTTACCAAACAATCCCTT-3′. After transduction for 3 days, transduced cells were selected by 1 μg/mL puromycin for an additional 3 days. The knockout efficiency of single-cell clones was verified by Western blotting.

Enzymatic Assays.

The inhibitory activities against JNK1–JNK3 were tested in Z′-Lyte assays with ATP concentrations of 100 μM. The inhibitory activities against p38a kinase were tested in Z′-Lyte assays with ATP concentrations at the K_m level. The inhibitory activities against PIKFYVE (PIP5K3) were determined by the KdELECT assay. All of the protocols are available from Life Technologies and Eurofins.

PhosphoSens Kinetic Assay.

k_inact/K_i measurements were obtained from kinetic enzyme activity data using a purified system comprising ATP, a recombinant enzyme, and a Sox-based chemosensor substrate.³⁴ Enzymes were obtained from Carna Biosciences (JNK1, catalog no. 04–163; JNK2, catalog no. 04–164), and the substrate sensor was obtained from AssayQuant (catalog nos. CSKS AQT0365 and CSKS-AQT0365K). All components were equilibrated to 25 °C prior to setting up reactions in 384-well microplates (PerkinElmer, catalog no. 6008280). Final working concentrations in reactions were 2 nM enzyme, 2 μM substrate sensors, and 20 μM ATP in assay buffer [50 mM HEPES (pH 7.5), 20 μM ATP, 1.2 mM DTT, 0.012% Brij-35, 0.52 mM EGTA, 10 mM MgCl₂, 1% glycerol, and 0.2 mg/mL BSA]. Buffer, substrate, ATP, DTT, and EGTA were premixed in assay wells, and then the reaction was started by adding enzyme and inhibitors. During the reaction, plates were sealed using optically clear adhesive film (TopSealA-Plus plate seal, PerkinElmer, catalog no. 6050185) to prevent evaporation. Fluorescence intensity (excitation and emission at 360 and 485 nm, respectively) measurements were read kinetically every 0.5 min for 120 min, with Filter Cube #113 in a Synergy Neo2 plate reader (Biotek, Agilent). GraphPad Prism (version 9.01) was used to analyze data. Experiments were repeated three times, with results expressed as means ± SD. The observed rate for inhibition, k_obs, was obtained by plotting the product signal at various concentrations of inhibitors versus time to fit a nonlinear association curve {product signal = v₀[1 − exp(−k_obs × time)]/k_obs}. To obtain values for k_inact and K_i, the k_obs values were replotted for each inhibitor concentration and fitted to the equation k_obs = k_inact[inhibitor]/(K_i + [inhibitor]).

Kinome Profiling.

Kinome profiling was performed using KinomeScan ScanMAX at a compound concentration of 1 μM. Data are reported in the Supporting Information. Protocols are available from DiscoverX.

KiNativ Cellular-Based Profiling.

KiNativ cellular-based profiling was performed using KiNativ technology by ActivX as previously described.¹⁷ SUM159 cells or MM.1S cells were plated in 15 cm dishes and treated for YL2056 or YL5084 at a compound concentration of 1 μM. After treatment, treated cells were collected, washed with PBS thrice, and immediately frozen in liquid nitrogen. KiNativ analysis was conducted at ActivX.

Proliferation Assay.

Two-dimensional proliferation assays were conducted as previously described.³⁵ For 72 h studies for a single agent, MM.1S or RPMI-8226 cells were plated in 96-well plates at a density of 4000 cells/well in fresh medium. Twenty-four hours later, cells were treated with inhibitors at the indicated concentrations or DMSO for 72 h. For synergy studies, MM.1S cell lines or RPMI-8226 cells were seeded in 96-well plates at a density of 4000 cells/well in fresh medium. Twenty-four hours later, cells were treated with YL5084 with or without Bortezomib at the indicated concentrations for 72 h. Antiproliferative effects of compounds were assessed using CellTiter Glo as described in the product manual by luminescence measurements on an Envision plate reader. The GR values were calculated using the online GR calculator (http://www.grcalculator.org/grtutorial/Home.html), according to the methodology recently described.^31,36

Mass Spectrometry Analysis.

JNK2 protein (5 μg) was treated with DMSO or a 5-fold molar excess of YL2056 for 1 h at room temperature and analyzed by LC/MS using an HPLC system (Shimadzu, Marlborough, MA) interfaced with an LTQ ion trap mass spectrometer (ThermoFisher Scientific, San Jose, CA). After injection, proteins were desalted for 4 min on column with 100% A and then 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). The mass spectrometer was programmed to acquire full scan mass spectra (m/z 300–2000) in profile mode (spray voltage of 4.5 kV). Mass spectra were deconvoluted using MagTran version 1.03b2.³⁷

JNK1/2 and PIKfyve NanoBRET Target Engagement Assay.

JNK1/JNK2 and PIKfyve intracellular kinase engagement assays were performed in HEK293T cells using the NanoBRET TE intracellular kinase assay kit (Promega, catalog no. N2500). HEK293T cells were resuspended at a density of 2.5 × 10⁵ cells/mL in DMEM without antibiotics and transfected with 0.5 μg of NanoLuc-JNK1 (Promega, catalog no. NV170A), NanoLuc-JNK2 vector (Promega, catalog no. NV171A), or PIKFYVE-NanoLuc fusion vector (Promega, catalog no. NV4041) together with 4.5 μg of carrier DNA vector (Promega, catalog no. E4881) by reverse transfection. After transfection for 24 h, cells were resuspended at a density of 2.5 × 10⁵ cells/mL in Opti-MEM supplemented with 1% FBS, replated in the 384-well plate, and pretreated with inhibitors at the indicated concentrations for 3 h using a JANUS liquid handler workstation (PerkinElmer). Then K-5 tracers (Promega, catalog no. N2500) or K-8 tracers (Promega, catalog no. N2620) were added at a final concenration of 130 nM in cells for JNK family proteins or PIKFYVE engagement assessment. After incubation for 1 h, the substrate and extracellular tracer inhibitor were added to generate the NanoBRET signal. Dual filtered luminescence at 610 nm (acceptor emission) and 450 nm (donor emission) was collected with a CLARIOstar plate reader (BMG Labtech). The relative BRET ratio was calculated as follows, and IC₅₀ values were generated in GraphPad Prism version 9.

$$\text{mBRET}\text{ratio}=\left[\frac{\text{acceptor}\text{channel}}{\text{donor}\text{channel}}-\frac{\text{acceptor}\text{channel}(\text{no}\text{ligand})}{\text{donor}\text{channel}(\text{no}\text{ligand})}\right]\times 1000$$

JNK2 Purification.

The codon-optimized (for Escherichia coli) JNK2 kinase domain (amino acids 4–364) fused with a TEV protease-cleavable N-terminal six-His tag was cloned into plasmid pMCSG49 using Gibson assembly. The plasmid introduced into YGT cells (BL21DE3 cells transformed with YopH on a streptomycin-resistant plasmid and GroEL/TriggerFactor on a chloramphenicol-resistant plasmid). Cells were grown in LB medium supplemented with kanamycin, streptomycin, and chloramphenicol at 37 °C until an OD₆₀₀ of ~1.0 was reached and then induced with 1 mM IPTG for 12 h at 16 °C. Cells were harvested by centrifugation at 6000g for 10 min at 4 °C and stored at −80 °C until purification. Cell pellets were resuspended in buffer A [50 mM Tris (pH 8.0), 500 mM NaCl, 10 mM imidazole, 5 mM β-ME, 0.02% n-octyl β-d-glucopyranoside, and 5% glycerol] supplemented with protease inhibitors benzamidine (15 μg/mL) and PMSF (1 mM) and lysed with four cycles of homogenization at 8000 psi in ice using an EmulsiFlex-C3 homogenizer (AVESTIN, Inc.). The lysate was clarified by centrifugation at 18000 rpm and 4 °C for 1 h. The supernatant was collected and filtered through a 0.22 μm filter and loaded onto 5 mL Bio-Scale Mini cartridges filled with Profinity Ni-charged IMAC resin (Bio-Rad Laboratories, Inc., Hercules, CA). After washing for 10 column volumes, His-tagged JNK2 kinase was eluted with buffer B [50 mM Tris (pH 8.0), 500 mM NaCl, 300 mM imidazole, 5 mM β-ME, 0.02% n-octyl β-d-glucopyranoside, and 5% glycerol] and desalted into buffer C [50 mM Tris (pH 8.0), 50 mM NaCl, 5 mM β-ME, 0.02% n-octyl β-d-glucopyranoside, and 5% glycerol]. The typical protein yield was 30 mg/L. The His tag was removed using TEV protease at a ratio of 1:20 relative to JNK2 overnight at 4 °C. Final purification of JNK kinase was performed using size exclusion chromatography in buffer D [50 mM Hepes (pH 7.5), 100 mM NaCl, 5 mM TCEP, 0.02% n-octyl β-d-glucopyranoside, 10 mM MgCl₂, and 5% glycerol].

JNK2 X-ray Crystallography.

The JNK kinase domain was concentrated to 26 mg/mL, and 2 mM ADP was added; then the mixture was incubated for 1 h at room temperature before crystallization. Crystals were obtained through hanging drop vapor diffusion at 4 °C in crystallization buffer [0.2 M potassium phosphate dibasic and 20% (w/v) polyethylene glycol 3350] and optimized with 10% Tween 80 as an additive. Crystals grew for 5 days, were harvested, and were frozen with 30% glycerol. For structures with YL2056, the AMP-JNK2 crystals were soaked in 5 mM YL2056 for 6 h before being frozen. Data were collected at beamline 19-ID of Advanced Photon Source and scaled using HKL3000.³⁸ A molecular replacement solution was obtained using Phaser³⁹ with PDB entry 4MXO⁴⁰ as the search model. Model building and refinement were carried out using Coot and Phenix.⁴¹

Computational Methodology.

Docking was done using Glide²⁷ software (Schrödinger suite, version 2022–3). The receptor was prepared via the Protein Preparation Wizard²⁶ to refine bond orders, fill in missing side chains and loops, add and optimize hydrogen bond networks, and assign force-field atom types. Protein minimization was carried out using the OPLS4⁴² force field. Epik⁴³ was employed to generate the het states at pH 7.4, and PROPKA⁴⁴ was used to generate the protonation states of residues. The LigPrep module was used to prepare ligands, including generation of protonation states and generating stereoisomers and tautomers. Docking was performed in the Standard Precision (SP) scoring mode. Homology models of JNK1 (Uniprot entry P45983) and JNK3 (Uniprot entry P53779) were prepared using the Prime module and the docked YL5084:JNK2 model as a template. To assess the stability and flexibility of the ligand–receptor complex, a 500 ns molecular dynamics simulation was conducted using Desmond software (Schrödinger version 2022–3). The binding energy was calculated using the Prime MM-GBSA method,³⁰ which includes protein–ligand van der Waals contacts, electrostatic interactions, ligand desolvation, and internal strain (ligand and protein) energies using the VSGB2.0 implicit solvent model with the OPLS2005 force field.⁴⁵ The trajectories were visualized using VMD,⁴⁶ and graphs were plotted using GraphPad (GraphPad Software, La Jolla, CA).

Liver Microsome Stability and GSH Reactivity Studies.

The liver microsome stability studies were conducted by WuXi AppTec Co., Ltd. Human microsomes (Corning, catalog no. 452117) and mouse microsomes (Xenotech, catalog no. M1000) were used for microstability tests with a final concentration of 0.5 mg of protein/mL in the presence of 1 μM compounds. After incubation at the indicated time points, samples were then subjected to LC/MS/MS analysis.

The GSH reactivity and liver microsome stability studies were conducted by WuXi AppTec Co., Ltd. The GSH reactivity was examined in 100 mM potassium phosphate buffer (pH 7.4) in the presence of 5 mM GSH and 1 μM compound at 37 °C at the indicated time points. Cold 3% AcOH in H₂O including 200 ng/mL Tolbutamide and 200 ng/mL Labetalol as an internal standard was used as the stop solution. All samples were then subjected to LC/MS/MS quantificiation.

ABBREVIATIONS

ADCK

AarF domain-containing kinase

BINAP

2,2-bis(diphenylphosphino)-1,1′-binaphthyl

Boc

tert-butyloxycarbonyl

BRET

bioluminescence resonance energy transfer

CDK

cyclin-dependent kinase

CuI

copper(I) iodide

CSNK1

casein kinase 1

DDR1

discoidin domain receptor tyrosine kinase 1

DIEA

N,N-diisopropylethylamine

DMF

dimethylformamide

EGFR

epidermal growth factor receptor

Et₃N

triethylamine

ER

endoplasmic reticulum

HPLC

high-performance liquid chromatography

JNK

c-Jun N-terminal protein kinase

MAPK

mitogen-activated protein kinase

MM

multiple myeloma

MM-GBSA

molecular mechanics generalized Born surface area continuun solvation

NMP

N-methyl-2-pyrrolidone

NMI

N-methylimidazole

PARP

poly-ADP-ribose polymerase

Py

pyridine

rmsf

root-mean-square fluctuation

SD

standard deviation

Sox

sulfonamido-oxine

TCFH

N,N,N′,N′-tetramethylchloroform amidinium hexafluorophosphate

SnCl₂

tin(II) chloride

TFA

trifluoroacetic acid

THF

tetrahydrofuran

T3P

propylphosphonic anhydride

UPR

unfolded protein response