Discovery of Pyrazolopyridine Derivatives as HPK1 Inhibitors

Abstract

In spite of the great success of immune checkpoint inhibitors in immune-oncology therapy, an urgent need still exists to identify alternative approaches to broaden the scope of therapeutic coverage. Hematopoietic progenitor kinase 1 (HPK1), also known as MAP4K1, functions as a negative regulator of activation signals generated by the T cell antigen receptor. Herein we report the discovery of novel pyrazolopyridine derivatives as selective inhibitors of HPK1. The structure–activity relationship campaign led to the discovery of compound 16, which has shown promising enzymatic and cellular potency with encouraging kinome selectivity. The outstanding pharmacokinetic profiles of 16 in rats and monkeys supported further evaluations of its efficacy and safety in preclinical models.

Immuno-oncology (IO) therapy has transformed the cancer treatment landscape and fueled hope for long-term survivorship and even cure in some cancers.¹ However, the response rates to IO therapy vary widely across tumor types. Thus, there is a need to develop alternative approaches to broaden the scope of therapeutic coverage.² One such approach is targeting the enzymes that can modulate the immune response. This approach, when used in combination with immune checkpoint inhibitors, could potentially result in a synergistic effect and thus improve the response rate and reduce the resistance developed by tumor cells. Hematopoietic progenitor kinase 1 (HPK1), also known as MAP4K1, functions as a negative regulator of activation signals generated by the T cell antigen receptor (TCR).³⁻⁹ HPK1 acts through phosphorylation of SLP76 and has an immunosuppressive function across a variety of cell types, including CD4⁺ and CD8⁺ T cells as well as dendritic cells. Loss of HPK1 kinase function in HPK1 kinase-dead (HPK1.kd) knockin mice enhanced T cell receptor signaling and cytokine secretion in a T-cell-intrinsic manner.^10,11 A synergistic effect in controlling tumor growth was also observed when anti-PD-L1 was used to treat HPK1.kd mice.¹⁰ These results, coupled with emerging studies of small-molecule HPK1 inhibitors,¹²⁻¹⁸ suggest that HPK1 could be an excellent drug target for enhancing antitumor immunity. Herein we describe the design and development of novel pyrazolopyridine derivatives as ATP-competitive inhibitors of HPK1.

The program commenced with a high-throughput screen (HTS) of Incyte’s internal compound collection against the kinase domain of HPK1. Binding affinity K_i was determined by incubating the tested compound with HPK1 kinase domain followed by measuring the homogeneous time-resolved fluorescence (HTRF) of the resulting solution. Compound A (Figure 1) with HPK1 K_i = 91 nM was identified from an earlier fibroblast growth factor receptor (FGFR) inhibitor program.¹⁹ Initial efforts focused on improving the HPK1 potency while attenuating the activity against other kinases. However, structure–activity relationship (SAR) efforts to replace the 2,6-difluoro-3,5-dimethoxyphenyl group, which is known to be specifically important for FGFR binding, led to significant loss of binding affinity against HPK1. Further replacing the pyrazolopyrimidine scaffold with other bicyclic structures such as pyrazolopyridine did not result in any improvement (e.g., compound 1 in Figure 1, K_i = 1151 nM). These unsuccessful attempts prompted us to redesign the topography of the molecules.

Figure 1.

Discovery of the pyrazolopyridine scaffold.

Our design was guided by compound B, a compound derived from another series identified by another HTS hit from Incyte’s internal compound library. Its structure suggested that migration of the top phenyl ring to the 5-position might potentially be tolerated. This modification resulted in the discovery of compound 2 with moderate binding affinity (K_i = 223 nM). In order to reduce the molecular planarity of 2, twisting the conformation of the biaryl system by incorporation of a fluoro group at the 2′-postion to obtain 3 (Table 1) provided a remarkable 8-fold improvement in binding affinity (K_i = 28 nM). To our delight, micromolar cellular potency of 3 (IC₅₀ = 2744 nM) was observed in Jurkat cells as determined by inhibition of phospho-SLP76. Encouraged by this success, we next examined a small library of 2′,6′-disubstituted compounds, of which fluoromethoxy analogue 4 rapidly emerged as a promising lead with single-digit binding affinity (K_i = 3.0 nM) and improved cellular potency (IC₅₀ = 395 nM).

Table 1. SAR of 3–8.

compd	X1	X2	R1	HPK1 Ki (nM)	p-SLP76 IC50 (nM)	IntCl (L h–1 kg–1)
3	CH	N	H	28	2744	ND
4	CH	N	OMe	3.0	395	>1.1
5	N	N	OMe	2.5	258	>1.1
6	N	CH	OMe	<1.0	144	1.1
7	N	CH	OMe	2.0	148	0.8
8	N	N	OMe	12	640	0.7

With the promising improvement of HPK1 potency, our attention turned to exploring additional bicyclic scaffolds adopting a similar conformation, as exemplified by 5 and 6. Both of these compounds were potent in the binding assay, with K_i values of 2.5 and <1.0 nM, respectively. Further profiling in the cellular assay showed that 1H-pyrazolo[3,4-c]pyridine 6 was almost 2-fold more potent (IC₅₀ = 144 nM) compared to 4 and 5. A similar trend was also observed with 1-methylpyrazol-4-yl analogues, as compound 7 (IC₅₀ = 148 nM) was significantly more potent than 8 (IC₅₀ = 640 nM). Having become a milestone of the program, 7 served as a prototype for the following round of the SAR campaign at the R₂ position in light of its relatively lower molecular weight and lower human intrinsic clearance (IntCl) (Table 1).

While many additional molecules with a variety of R₂ groups were prepared, finding the right balance between HPK1 cellular potency and physicochemical properties remained a challenge. We hypothesized that the frequently observed high clearance was related to the high lipophilicity of the molecule. Thus, our exploration then shifted to the introduction of polar substituents on the phenyl ring. Consequently, several lead compounds were discovered, of which benzylic amine 9 (Table 2) was the most promising, with significantly improved human intrinsic clearance (<0.5 L h^–1 kg^–1). Although initial kinase profiling of 9 revealed submicromolar potency against several other kinases, its good HPK1 cellular potency (IC₅₀ = 219 nM) warranted further investigation of the benzylic amine moiety. The goal was to maintain the desired HPK1 potency and at the same time improve the kinome selectivity. Moving forward, we routinely monitored enzymatic activities against the kinases TrkA, FLT3, and KIT as surrogates for overall selectivity.

Table 2. SAR of 9–12.

compd	R1	HPK1 Ki (nM)	p-SLP76 IC50 (nM)	IntCl (L h–1 kg–1)	Caco-2 Pm (10–6 cm/s)	CYP3A4 IC50 (μM)	TrkA IC50 (nM)a	FLT3 IC50 (nM)a	KIT IC50 (nM)a
9	OMe	3.0	219	<0.5	0.2		330	260	857
10	F	5.4	117	<0.5	0.7	>25	2600	366	1078
11	CF3	3.5	128	<0.5	1.6	>25	959	2233	1976
12	Me	2.7	91	<0.5	1.1	>25	1624	1507	7471

^aIn the presence of 1 mM ATP.

In an expanded examination of substituents at the 6′-position of the phenyl ring, we discovered that the dihedral angle of the biaryl system could play an important role in improving the kinome selectivity. We hypothesized that the biaryl moiety occupied a tight area of the ATP binding pocket and that small changes in the conformation could lead to selectivity even in the absence of amino acid differences in this region among kinases. In particular, difluoro compound 10 afforded a remarkable improvement in TrkA selectivity (IC₅₀ = 2600 nM) but still suffered from undesired FLT3 potency (IC₅₀ = 366 nM). Replacing the F group at the R₁ position by a CF₃ group (compound 11) provided a 6-fold boost in FLT3 selectivity (IC₅₀ = 2233 nM) over 10. However, subsequent in vitro profiling of 11 revealed significant time-dependent inhibition (TDI) against CYP 2C19. Finally, the TDI liability was suppressed by switching to the methyl analogue (compound 12). Similar to 11, 12 achieved IC₅₀ values of >1 μM against all three kinases along with improved HPK1 cellular potency (IC₅₀ = 91 nM). Compound 12 also displayed a balanced physicochemical profile with low human intrinsic clearance (<0.5 L h^–1 kg^–1) and moderate Caco-2 permeability (1.1 × 10^–6 cm/s).

On the basis of its promising in vitro profile, 12 was evaluated in a rat pharmacokinetic (PK) study. In spite of its low human intrinsic clearance, the compound showed high in vivo clearance (7.1 L h^–1 kg^–1) in rats (Table 3). Studies of drug metabolism by incubation of several related compounds with rat CYP450 isozymes suggested that the high clearance of 12 arose from CYP 1A1 metabolism, which is localized to rat lung tissue and is generally minimal in humans. As a result, cynomolgus monkey (cyno) PK was employed to assess compounds for further preclinical development. In contrast to rat PK, compound 12 delivered moderate in vivo clearance (1.2 L h^–1 kg^–1) in monkeys. However, given its low oral bioavailability (7%), there still remained a need for further optimization.

Table 3. SAR of 12–16.

			in vitro ADME			rat ivd	rat poe	cyno ivf			cyno poe
compd	HPK1 Ki (nM)	p-SLP76 IC50 (nM)	IntCL (L h–1 kg–1)	Caco-2 Pmb	fu (%)c	CL (L h–1 kg–1)	AUC (nM h)	CL (L h–1 kg–1)	Vdss (L h–1 kg–1)	T1/2 (h)	AUC (nM h)	F (%)
12	2.7	91	<0.5	1.1	41	7.1	456	1.2	4.6	5.5	759	7
13a	4.0	94	<0.5	0.2	49	5.6	799	1.1	7.6	8.6	3930	50
14a	4.6	138	<0.5	0.4	50	4.8	1460	0.54	4.6	8.9	7840	55
15	5.0	137	<0.5	1.4	61	6.1	761	0.66	3.9	6.3	1640	16
16	4.1	146	0.5	3.1	43	6.1	1100	0.95	4.0	6.0	9770	96

^aSingle enantiomer.

^bIn units of 10^–6 cm/s.

^cFraction unbound in human plasma.

^d1.5 mg/kg dosing.

^e3.0 mg/kg dosing.

^f1.0 mg/kg dosing.

One of the approaches examined was to incorporate a methyl group at the benzylic position (compound 13) to block the potential metabolic soft spot. The distomer of 13 is less active in the p-SLP76 cell assay. As a result, oral exposure of 13 (3.9 μM h) displayed 4-fold increase in monkeys compared to 12 (0.76 μM h). To further improve the cyno PK profile, cyclization of the benzylic side chain (compound 14) was designed to reduce the number of rotatable bonds, which could limit the permeability and metabolic stability. This modification significantly diminished the in vivo metabolism and drove a nearly 2-fold boost in oral exposure of cyno PK (7.8 μM h). The distomer of 14 is more active in the p-SLP76 cell assay, but it is much less selectivie against the general kinome.

An alternative strategy was employed to elaborate on the methylamine moiety in an attempt to block potential metabolic pathways such as dealkylation. Ethyl compound 15 showed reduced in vivo clearance (0.66 L h^–1 kg^–1) compared to 12 while affording a small enhancement in Caco-2 permeability (1.4 × 10^–6 cm/s). These properties together delivered moderate oral exposure (1.6 μM h) but still low bioavailability (16%) in monkeys. To address the low bioavailability, we hypothesized that reducing the number of hydrogen-bond donors of the molecule could significantly enhance its permeability. SAR studies of tertiary amines culminated in the discovery of azetidine 16, which had similar HPK1 potency (IC₅₀ = 146 nM) and intrinsic clearance (0.5 L h^–1 kg^–1) compared to 15 but displayed higher Caco-2 permeability (3.1 × 10^–6 cm/s). Accordingly, 16 provided a good half-life (6 h), high oral exposure (9.8 μMh), and excellent bioavailability (96%) in cyno PK following oral dosing at 3 mg/kg.

Having demonstrated an overall attractive profile, 16 was selected for further in vitro characterization. The compound was found to be potent in the human-whole-blood p-SLP76 assay (IC₅₀ = 302 nM). Its general kinase selectivity was determined by kinase assay screen at Reaction Biology Corporation. Among 371 kinases measured, only 12 kinases were projected to have IC₅₀ values below 1 μM. The molecule also did not exhibit significant activity against a nonkinase CEREP panel of over 50 receptors, ion channels, and transporters. Assessment of the potential for drug–drug interactions (DDIs) indicated a low overall risk, as IC₅₀ values of >15 μM were observed against multiple CYP450 isozymes (1A2, 2B6, 2C19, 2C8, 2C9, 2D6, and 3A4). Moreover, a low percent inhibition (9% at 5 μM) in the hERG patch clamp assay indicated limited cardiovascular risk.

The binding mode of 16 was studied based on a cocrystal structure of a related compound from the series (undisclosed). The docking of 16 to the HPK1 kinase domain shows hinge binding through the pyrazolopyridine ring, where the pendent pyrazole lies coplanar, enabling the hydrophobic interaction with Leu23 (Figure 2). The azetidine amine is protonated under physiological conditions and makes a hydrogen-bonding interaction with Asn142. Tyr28 provides a hydrophobic cleft and possibly π stacking with the phenyl ring (3.9 Å distance). The SAR from the structures (Tables 1 and 2) shows that the methyl group not only is important in fixing the biaryl torsion but also occupies a small hydrophobic dent at the bottom of the kinase pocket surrounded by Leu144 and Ala154, contributing to the overall activity. The pyridine nitrogen engages in a water-bridged hydrogen bond with Asp155, which leads to the superior potency of the 1H-pyrazolo[3,4-c]pyridine core compared to others (Table 1).

Figure 2.

Proposed binding model of 16 with HPK1.

In summary, we designed and identified a novel series of pyrazolopyridine derivatives as selective HPK1 inhibitors, starting from HTS hit A. The SAR campaign led to the discovery of compound 16, which has shown promising enzymatic and cellular potency with encouraging kinome selectivity.

Glossary

Abbreviations

SLP76

lymphocyte cytosolic protein 2

TrkA

tropomyosin receptor kinase A

FLT3

fms-like tyrosine kinase 3

ADME

absorption, distribution, metabolism, excretion

hERG

human ether a-go-go

iv

intravenous

po

per os

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmedchemlett.2c00238.

• Materials and methods; experimental procedures and analytical data for compounds 1–16; 2D NMR spectra for compounds 12 to 16; full kinase profile for compound 16; ADME assays; biological assays (PDF)

Author Contributions

Design/conception of study or data acquisition/analysis and/or interpretation: all authors. Drafting manuscript or providing critical review of content: all authors. Approval of final draft for publication: all authors. Accountable for all aspects of work: all authors.

This work was funded by Incyte Corporation.

The authors declare no competing financial interest.