Allosteric targeting of RIPK1: discovery of novel inhibitors via parallel virtual screening and structure-guided optimization

Abstract

Receptor-interacting serine/threonine protein-kinase 1 (RIPK1) is a critical signalling protein that regulates inflammation and cell death in response to TNF signalling. Inhibiting RIPK1 kinase activity prevents neuronal cell death in various animal models, making it a promising therapeutic target for neurodegenerative, inflammatory, and autoimmune disorders. To identify novel allosteric RIPK1 inhibitors, we used a parallel virtual screening strategy that employed structure-based pharmacophore, shape-based, and fuzzy pharmacophore similarity approaches. Structure-guided optimization enabled by X-ray crystallography led to the discovery of a potent and selective piperidinecarboxamide inhibitor with an acceptable pharmacokinetic (PK) profile and limited brain exposure. This work highlights the effectiveness of virtual screening, followed by structure-guided optimization, in identifying progressible allosteric kinase inhibitors.

---

Introduction

Receptor-interacting serine/threonine protein-kinase 1 (RIPK1) is recruited downstream of the TNF triggered signalling pathway and is a key regulator of cell death and inflammation.^1,2 Recent studies suggest that RIPK1 dependent activation in microglia and astrocytes triggers the production of many pro-inflammatory cytokines contributing to neuroinflammation, a key driver in the pathogenesis of Alzheimer's disease³ and other neurodegenerative disorders.⁴ This hypothesis is further supported by genetic and pharmacological evidence, which shows that knock-in of a kinase-dead form of RIPK1 or inhibition of the kinase activity of RIPK1 using a chemical probe reduces the amyloid burden, suppresses inflammatory cytokines, and alleviates memory deficits in transgenic (PS1) mouse models.^3,5,6 As a result, RIPK1 appears to be an attractive therapeutic target for the treatment of various inflammatory, autoimmune, and neurodegenerative disorders.⁷

Since the discovery of necrostatin-1 (Nec-1), the first reported small-molecule RIPK1 inhibitor,⁶ multiple RIPK1 inhibitors with diverse chemical scaffolds have been reported in both the scientific and patent literature (Fig. 1).^8,9 The crystal structure of Nec-1s (1), a stable analog of Nec-1, bound to RIPK1 revealed the inhibitor occupies an allosteric pocket adjacent to, but not overlapping with, the ATP binding site (PDB ID: 4ITH).¹⁰ This type III kinase inhibitor binding mode provides high selectivity across the kinome due to low sequence conservation in this region. In contrast, conventional type I and type II inhibitors, which both contain features that bind the ATP pocket, lack RIPK1 selectivity.¹¹

Fig. 1 Chemical structures of allosteric RIPK1 inhibitors.

Several potent and selective type III RIPK1 inhibitors that occupy the same allosteric binding site have progressed to clinical trials.^12–16 These include GSK2982772 (2), the first peripherally restricted RIPK1 inhibitor developed for the treatment of chronic inflammatory conditions,^17,18 and DNL758 (3) for treatment of cutaneous lupus erythematosus.¹³ However, both clinical candidates were discontinued after failing to meet their primary endpoints in Phase 2 trials.¹⁹ Additionally, brain penetrant RIPK1 inhibitors such as DNL788 (4) were clinically evaluated for neurological indications, including multiple sclerosis,²⁰ but failed to meet their primary endpoints in phase 2 trials.²¹

The benzoxazepinone scaffold, exemplified by 2, 3, 5, and 6, has emerged as a common chemotype among allosteric type III RIPK1 inhibitors. Notably, compounds 3, 5, and 6 appeared in the literature only after we started our discovery program. Unfortunately, this scaffold has several liabilities that pose challenges for clinical development, including high clearance due to demethylation of the N-methyl lactam and active P-glycoprotein mediated extrusion leading to a low brain-to-plasma ratio. Hence, our small molecule discovery effort was directed towards identifying novel inhibitor scaffolds that bound the same allosteric site but were structurally distinct from the benzoxazepinone scaffold.

To date, several virtual screening (VS) campaigns targeting the RIPK1 allosteric site have been reported. Various VS techniques, such as pharmacophore-based screening followed by docking,²² deep learning enabled VS followed by docking,²³ generative AI combined with pharmacophore and docking,²⁴ and docking based VS followed by MM-GBSA based filtering,²⁵ have successfully identified novel inhibitors; however, all these VS campaigns relied on a sequential VS approach, wherein multiple VS techniques were applied in a funnel-style workflow to select compounds for experimental screening.

To overcome the limitations inherent in sequential virtual screening, we employed parallel virtual screening to efficiently explore chemical space and find new chemical scaffolds. This strategy involves running multiple VS methods simultaneously and combining their results to prioritize compounds for experimental testing. By leveraging the strengths of each individual VS method and offsetting their individual limitations, parallel VS has been shown to improve enrichment rates and increase the likelihood of identifying diverse chemotypes.^26–28

This work demonstrates the value of parallel virtual screening for hit-finding, followed by structure-guided optimization, for accelerating the hit to lead optimization process in small molecule drug discovery.

Results and discussion

Virtual screening

We leveraged publicly available structural data on RIPK1 allosteric inhibitors along with the extensive chemical and bioactivity information on small molecule inhibitors when selecting virtual screening methods. While structure-based virtual screening using docking is a well-established and widely used approach, we opted to use ligand-based methods. These methods offer comparable enrichment and chemical diversity to docking, but with the added benefit of significantly lower computational cost.²⁹ To identify novel allosteric inhibitors that are structurally distinct, yet maintain a similar interaction pattern to known ligands, we employed a combination of 3D structure-based pharmacophore, molecular shape, and 2D similarity-based screening methods. Search queries are shown in Fig. 2 and details on each method are outlined in the following sections.

Fig. 2 A) Structure-based pharmacophore query generated using the X-ray structure of 2 bound to RIPK1 (PDB: 5TX5). The pharmacophore model consists of two ring aromatic features (orange), one hydrogen bond acceptor feature (cyan) and one hydrophobic feature (green). The excluded volume is not shown for clarity. B) Hybrid structure-based pharmacophore query generated from two distinct RIPK1i scaffold classes (PDB: 5TX5 and 4ITH). C) Shape query based on the X-ray conformation of 2 bound to RIPK1. Key chemical features are mapped as spheres. D) Feature tree-based representation of 2 used as a query for similarity searching.

The above techniques were used to search a database of approximately 11.4 million purchasable compounds compiled from nine chemical supplier catalogues. Compounds underwent in silico structural checks, standardization, and filtering based on widely accepted structural alerts, functional groups counts, physicochemical properties, and lead-likeness criteria.^30,31 Precompiled conformer databases were generated to feed into shape and pharmacophore-based screening.

Pharmacophore-based virtual screen. A structure-based pharmacophore model was developed in the MOE molecular modeling package³² using a structural overlay of the X-ray structures of 2, which contains a benzoxazepinone moiety (PDB ID: 5TX5),³³ and 1 (PDB ID: 4ITH)¹⁰ which has a hydantoin scaffold.

Key protein-ligand interactions in the RIPK1 structure with 2 bound were analysed and translated into a structure-based pharmacophore model to provide an abstract three-dimensional representation of the essential molecular features and their spatial arrangement required for target binding (Fig. 2A). Using the EHT pharmacophore scheme, a four-feature pharmacophore model was developed, consisting of two ring aromatic (RA) features, one hydrophobic (HYD), and one hydrogen bond acceptor (HBA) feature. To further enhance the specificity of the pharmacophore model, excluded volume spheres were added to define regions of steric hindrance imposed by protein atoms near the bound inhibitor. These excluded volume spheres represent physically inaccessible areas for the ligand and help refine the model by eliminating compounds consistent with the other pharmacophore features that would clash with the protein binding site.

Analysis of the RIPK1 X-ray structures with 1 and 2 bound revealed that lone pairs on the carbonyl oxygen atom of the triazole carboxamide linker of 2 and the ketone oxygen at the 4-position of the imidazolidine-2,4-dione ring of 1 can each form a hydrogen bond with the backbone NH of Asp 156 in two distinct orientations, relative to the lone pairs on the oxygen atoms of their respective chemical scaffolds. To account for this flexibility in the hydrogen bond geometry, a hybrid pharmacophore (Fig. 2B) was created by combining key pharmacophoric features from the benzoxazepinone and the hydantoin scaffolds. This hybrid pharmacophore captures the variability in the spatial orientation of the hydrogen bond acceptor, a critical feature for RIPK1 inhibition as supported by previous SAR studies.³⁴

These pharmacophore models were used as queries to search a multi-conformational database of commercial vendor compounds. A ranked hit list based on rmsdx score (a measure of RMSD between pharmacophoric features and matching ligand annotation points) with a cutoff ≥0.45 was used to define hits for each query.

Shape-based virtual screen. Shape-based virtual screening was performed using ROCS (rapid overlay of chemical structures) to identify compounds with similar 3D shape to known active molecules.³⁵ Unlike pharmacophore-based screening, ROCS-based screening uses a smooth Gaussian function to represent the molecular volume of a reference ligand, thereby creating a shape query for comparison. This representation can be used effectively to “scaffold hop” to novel chemical space and provide chemically distinct starting points for chemical elaboration.

Our shape query was defined using the X-ray structure of 2 bound to RIPK1 (PDB: 5TX5). To enhance specificity, key protein-ligand interactions observed from the co-crystal structure were mapped as color features to the shape volume to enable the identification of compounds that have similar shape and chemical features. This shape query was used to search our multi-conformational database of commercial compounds. Results were ranked using Tanimoto Combo score, a similarity metric that measures the overlap of shape and chemical (color) features between the query and database molecules, with a cut-off of ≥1.25 used to retain hits for further evaluation.

Similarity-based virtual screen. To complement our shape- and pharmacophore-based screening strategies, we incorporated FTrees (Feature Trees)³⁶ to provide a more direct similarity-based approach. This orthogonal computational technique leverages a fuzzy fingerprint-based similarity method to identify similar compounds that may be non-obvious relative to the query molecule's chemical structure. FTrees has been shown capable of identifying novel hit compounds through scaffold hopping.

Compound 2 served as our search query. All compounds with an FTrees similarity score above 0.85, a threshold that reflects a reasonable balance between structural diversity and similarity, were selected for follow-up.

Compound selection and experimental evaluation

Compound selection schemes following parallel screening often use data fusion algorithms such as sum rank, rank vote, sum score, Pareto ranking, and parallel selection to combine results from multiple VS methods. We employed the parallel selection approach due to its demonstrated superiority over other data fusion methods, as shown in previous benchmarking studies.³⁷ Accordingly, the top ranked compounds from each VS method were selected until the target number of compounds intended for experimental screening was reached. If a compound had already been selected by another method, the next highest-ranked compound from that method was chosen instead.^27,37

To ensure a balance between chemical diversity and quality in compound selection, we used Bemis–Murcko framework-based clustering³⁸ to select structurally diverse compounds. Additionally, the quantitative estimate of drug likeness (QED)³⁹ desirability score was used to prioritize compounds with favourable properties. This resulted in the selection of 236 diverse compounds for purchase and experimental testing.

Primary screening for hit identification was performed using the ADP-Glo™ kinase assay⁴⁰ at two fixed concentrations (10 μM and 50 μM). Biological activity was quantified as a percentage of the untreated control (PoC), representing the remaining kinase activity relative to the DMSO control. A compound was considered a primary hit if it showed a PoC ≤ 50 at 50 μM, demonstrated a dose-dependent response across the two concentrations tested, and exhibited no visible signs of precipitation at 10 μM compound concentration. Based on our hit-calling criteria, 23 primary hits were identified, corresponding to a hit-rate of approximately 10%. These hits were further evaluated by obtaining full dose–response curves to estimate their IC₅₀ values. This process resulted in 18 confirmed hits with IC₅₀ values ranging from 2 to 48 μM. A detailed list of the confirmed hits, excluding those identified from our proprietary internal compound collection, is provided in the SI (Table S1), along with associated data.

Structure-guided optimization of piperidine carboxamides

To structurally validate the binding modes and enable rational design, hit compounds with IC₅₀ < 20 μM were subjected to X-ray co-crystallography with RIPK1. These efforts quickly led to a 2.98 Å resolution RIPK1 co-structure with 7, a 16 μM inhibitor with a piperidine carboxamide core that is structurally distinct from the benzoxazepinone scaffold found in other RIPK1 inhibitors. Crystallographic details, data collection and refinement statistics are provided in the SI (Table S2). The 2Fo–Fc electron density map around ligand is provided in Fig. S1.

As anticipated, the co-crystal structure revealed that 7 binds to the allosteric pocket adjacent to the ATP-binding site, making no interactions with the kinase hinge motif. As shown in Fig. 3A, 7 binds to an inactive conformation of the kinase, wherein residues Asp156 and Leu157 of the DLG motif (commonly referred to as DFG in other kinases) adopt a “DFG-out” conformation⁴¹ and the αC helix displays an “αC-out” conformation characterized by the absence of the canonical ion pair between the catalytic Lys45 and αC Glu63. The amide carbonyl was found to make a hydrogen-bonding interaction with the backbone NH of Asp156, while the methylbenzyl group filled a mostly hydrophobic back pocket, interacting with the side chains of Met67, Leu70, Leu129, and Ser161 (Fig. 3B).

Fig. 3 A) X-ray structure of compound 7 (PDB ID: 9HY8) bound to the allosteric pocket of RIPK1, showing the key residues involved in interaction. B) Opportunities for optimization of the piperidine carboxamide scaffold in the RIPK1 allosteric pocket. The green arrows on the molecule indicate vectors for growing.

In addition, the phenylmethyl group forms a C–H–π interaction with the side chain of His136, part of the HRD motif belonging to the kinase catalytic loop. The hydrogen atom on the piperidine alpha carbon orients to form a weak C–H⋯S interaction with Met92, the gatekeeper residue. The 2-chloro atom of the phenyl ring projects into a sub-pocket behind the gatekeeper residue, while the 5-chloro atom extends toward the solvent exposed phosphate binding region. This dichlorophenyl group forms additional hydrophobic interactions with Ser25, Val31, Gly98, and Leu157. These structural insights guided the optimization of the piperidine carboxamide series.

Alongside the structural biology effort, hit-expansion using SAR-by-catalogue⁴² was also undertaken to rapidly evaluate the tractability of the piperidine carboxamide scaffold and to develop early SAR. A substructure search for the piperidine 4-carboxamide scaffold in the Enamine in-stock collection⁴³ was carried out, and 15 analogs were subsequently purchased and screened. Among these, compound 8, a methylpiperidine carboxamide analog, had an IC₅₀ of 930 nM, a 17-fold improvement in activity relative to 7. Additional optimization started with the exploration of substitutions on the phenyl ring and piperidine ring, as shown in Table 1.

Table 1 Kinase inhibition of piperidine carboxamides with substitutions on the phenyl and piperidine rings

Compound	R1	R2	R3	ADP-Glo IC50^a (nM)
a Mean ± standard deviation.
1	—	—	—	470 ± 150
7	Cl	Cl	H	14 000 ± 5100
8	H	H	Me	930
9	Cl	F	H	1200 ± 21
10	Br	F	H	540
11	Br	F	OMe	1500 ± 42
12	Br	F	F	64 ± 2
13		F	H	300 ± 96
14		F	H	490
15		F	H	370 ± 79
16		F	H	130 ± 24
17		F	H	110 ± 28
18		F	F	13 ± 2

---

Structural insights from the co-crystal structure with 7 highlighted opportunities for introducing polar heterocyclic groups (Fig. 3B) at the 5-position of the phenyl ring, oriented toward the solvent-exposed phosphate binding region. To explore this, several polar, monocyclic rings were constructed in silico using the R-group enumeration tool in the Schrödinger software suite.⁴⁴ These molecules were then evaluated using Glide SP docking.⁴⁵ We selected a set of promising pyridine, thiazole, and pyrazole heterocyclic analogs resulting from this exercise for synthesis. It was gratifying to see pyrazole analog 16 maintain reasonable activity while improving hLM stability (t_1/2 = 64 min). Combining the best substituents from this exercise resulted in 18, a compound with the best activity to date (IC₅₀ = 13 nM) without a loss in microsomal stability (hLM t_1/2 = 66 min).

Next, we focused on optimization of the opposite side of the piperidine carboxamide scaffold using a parallel library synthesis approach, probing the SAR for N-substitution of the amide, as well as substitutions at the benzyl methylene and benzene ring (Table 2). This required shifting to a cellular readout for the most active compounds, due to reaching the reliable detection limit of the ADP-Glo assay. Hence, compounds that displayed an IC₅₀ value ≤50 nM in this assay were also run in a cell assay using the human monocytic U937 cell line to obtain a readout of cellular necroptosis.⁴⁶

Table 2 Kinase activity, cellular activity, and microsomal stability for analogs substituted at amide and benzylic positions

Compound	R1	R2	R3	R3^′	ADP-Glo IC50^a (nM)	hU937 IC50^a (nM)	hLM t1/2 (min)
a Mean ± standard deviation, ND = not determined.
1	—	—	—	—	470 ± 150	700 ± 170	ND
2	—	—	—	—	ND	11 ± 0.5	ND
3	—	—	—	—	ND	14 ± 8	ND
4	—	—	—	—	ND	250 ± 98	ND
19	H	(S)Me	H	H	92	ND	57
20	H	(R)Me	H	H	6700	ND	ND
21	H	H	2-F	H	18 ± 5	93 ± 1	ND
22	H	H	3-F	H	17 ± 4	82 ± 1	ND
23	H	H	4-F	H	99 ± 13	ND	ND
24	H	H	2-Cl	H	23	130 ± 2	58
25	H	H	3-Cl	H	52	ND	29
26	H	H	4-Cl	H	110	ND	ND
27	Me	H	2-F	H	25	220 ± 1	17
28	H	(S)Me	2-F	H	15	230 ± 9	ND
29	H	H	2-F	3-F	13	74 ± 2	130
30	H	H	2-F	4-Cl	42	300 ± 5	15
31	H	H	2-F	4-F	38	230 ± 1	ND
32	H	H	2-F	5-F	15 ± 1	48 ± 1	59
33	H	H	2-F	6-F	16 ± 3	54 ± 1	120
34	H	H	3-F	4-F	65	ND	ND
35	H	H	3-F	5-F	22 ± 6	78 ± 2	150
36	H	(S)Me	2-F	6-cl	9 ± 1	45 ± 1	ND

---

To gain structural insights for compound design, 18 was docked to RIPK1 using the Glide SP protocol, with further refinement using Prime (OPLS3 and VSGB 2.0 implicit solvent model).⁴⁷ This refinement step accounts for induced-fit effects at the rotamer level within the binding pocket at a reduced computational cost. The docking model suggested extending from the benzylamide methylene linker region into a lipophilic pocket surrounded by Met67, Val68, and Leu76 (Fig. S2). Accordingly, a methyl group was introduced at the methylene position, resulting in two enantiomers with the S-enantiomer (19) approximately 72-fold more potent than the R-enantiomer (20), suggesting a key role for the chiral methyl group. While 19 retained reasonable potency and hLM stability, it showed an approximate 7-fold reduction in potency compared to 18.

Focus was then shifted to the benzyl ring, which occupies the allosteric hydrophobic pocket. We hypothesized that halogenating this ring could enhance metabolic stability, while maintaining or improving potency. Hence, mono-chloro and mono-fluoro substitutions were explored. Results showed that ortho and meta substitutions (compounds 21, 22, 24, 25) were tolerated, but did not provide an advantage relative to 18, while halogenation at the para position (compounds 23, 26) resulted in loss of potency.

Given the potential liability of hydrogen-bond donors to impact blood brain barrier permeability, we methylated the free N–H of the amide functionality of 21, resulting in compound 27. Unfortunately, this modification did not improve its cellular potency. In addition, the (S)Me analog (compound 28) was synthesized as a matched pair to 21, again with no improvement to its cellular potency.

We next explored the effect of di-halogen substitution on the benzyl ring. Several analogs with di-fluorinated substitution patterns, including 29 (2,3-difluoro), 33 (2,6-difluoro), and 35 (3,5-difluoro), retained enzymatic and cellular potency while improving microsomal stability.

Combining substitutions at the benzyl ring and methylene group revealed that a 2-fluoro, 6-chloro, (S)-methyl arrangement (compound 36) retained enzymatic and cellular potency. While we were hopeful that this combination would result in improved hLM stability, unfortunately 36 exhibited moderate in vitro CYP3A4 inhibition (IC₅₀ = 2.2 μM) and was not further profiled.

However, we were able to obtain an X-ray co-structure of 36 bound to the RIPK1 allosteric site at high resolution (2.29 Å), validating our binding mode hypothesis for the pyrazole-containing analogs and substitutions to the benzylic region. Crystallographic details, data collection and refinement statistics are provided in the SI (Table S3). The 2Fo–Fc electron density map around the ligand is provided in Fig. S3. Compound 36 retains the key hydrogen bonding interaction with the backbone NH of Asp156 in the DLG motif (Fig. 4A). The chiral methyl attached to the benzylamide fills a lipophilic pocket and forms hydrophobic interactions with Val76 and Leu70. The methyl group of 4-methyl pyrazole is solvent exposed, and the pyrazole ring sits within a hydrophobic cage, forming π-alkyl interactions with the alkyl side chains of Val31 and Leu157. In addition, the fluorine atom on the piperidine ring is positioned axially, contributing to weak polarization-enhanced C–H⋯S interactions with Met92 and Met67.

Fig. 4 A) X-ray structure of compound 36 bound to RIPK1. B) Structural rearrangement observed in the RIPK1 allosteric pocket. Cyan residues are from the X-ray structure of RIPK1 bound to compound 36, while magenta residues are from PDB entry 5TX5. Arrows indicate the positions where residues have shifted.

When analyzing our compound-bound X-ray structures of RIPK1, we noted a significant structural rearrangement (Fig. 4B) of residues Phe162 and Met67 compared to the publicly available structures containing Nec-1s (1) and GSK2982772 (2). Residue Phe162 is a part of the activation loop, immediately C-terminal to the DLG motif. This residue also participates in a single 3₁₀ helix turn that packs against the outwardly displaced αC helix (αC-out conformation), which contains Met67, located at the C-terminal base of the αC helix (Fig. 5A). In our X-ray structures, residues Phe162 and Met67 swap positions upon inhibitor binding (compounds 7 and 36). The piperidine ring in our scaffold likely pushes against the side chain atoms of Phe162, causing it to swap positions with Met67. This structural rearrangement allows the pocket to better accommodate the inhibitor.

Fig. 5 A) A 3₁₀ helix turn immediately following the DLG motif. The helix is stabilized by a hydrogen bond between the carbonyl oxygen of residue i and the amide hydrogen of residue i + 3. B) Phe-in, Phe-inter, and Phe-out conformations of Phe162 observed across RIPK1 X-ray structures.

This notable structural rearrangement prompted us to take a close look at other RIPK1 structures available in the PDB. Following sequence alignment-based 3D superposition, we observed that the conformations of Phe162 and Met67 vary depending on the inhibitor scaffold. Specifically, Phe162 can adopt one of three distinct conformations: Phe-in, Phe-intermediate, or Phe-out (Fig. 5B). An annotated list of these conformations across all RIPK1 structures available in the PDB is provided in the SI (Table S4). We observed that the lipophilic allosteric pocket has a slightly larger pocket volume in the Phe-in conformation compared to Phe-out conformation, since Phe162 is oriented away from the allosteric pocket. In contrast, the Phe-out conformation has a relatively small allosteric pocket; however, this shift opens an accessible pocket near residues Leu159 and Phe162, which are located at the entrance of the allosteric pocket (Fig. S4).

This finding offers a valuable strategy for inhibitor design, particularly for achieving selectivity and optimizing affinity. Hence, we recommend using ensemble or induced fit approaches that account for side chain flexibility during molecular docking.

Incorporation of sp² nitrogen into an aromatic ring is often a valuable strategy for optimizing metabolic stability by decreasing the electron density of the ring, making it less susceptible to oxidation. Accordingly, a nitrogen walk was carried out on the fluoro phenyl ring as shown in Table 3. We found that incorporating a nitrogen at the 4-position of the fluorophenyl ring (compound 38) resulted in improved metabolic stability while retaining biochemical potency. However, incorporating nitrogen at the 3- and 6-positions resulted in reduced activity.

Table 3 Enzyme activity, cellular activity, and microsomal stability of analogs with nitrogen incorporation in the fluorophenyl ring

Compound	X1	X2	X3	ADP-Glo IC50^a (nM)	hU937 IC50^a (nM)	hLM t1/2 (min)
a Mean ± standard deviation, ND = not determined.
37	N	CH	CH	72 ± 5	ND	ND
38	CH	N	CH	37 ± 5	210 ± 37	320
39	CH	CH	N	370	ND	ND

---

Pharmacokinetic and selectivity profiling of compound 38

Given the promising in vitro profile of 38, this compound was subjected to a pharmacokinetic (PK) study to evaluate its potential as an in vivo tool for pharmacodynamic studies in mice. 38 was administered to C57BL/6J mice via oral (PO, 5 mg kg⁻¹) and intravenous (IV, 1 mg kg⁻¹) routes, and its concentrations in plasma and brain were measured. The PK profile is summarized in Table 4.

Table 4 Pharmacokinetic properties of compound 38, estimated from a single-dose intravenous and oral PK study

Route	PK parameter	Value
IV (1 mg kg^−1)	Cl (mL h^−1 kg^−1)	0.7
	Vdss (L kg^−1)	1.46
	MRT (h)	1.98
	T1/2 (h)	1.86
PO (5 mg kg^−1)	Tmax (h)	0.5
	Cmax (μg mL^−1)	3.05
	T1/2 (h)	1.37
	AUC (μg h mL^−1)	6.3
	F (%)	45
	Brain-to-plasma ratio	0.16

---

Overall, 38 exhibited a favourable PK profile, achieving a maximal plasma concentration (C_max) of 3 μg mL⁻¹ at 0.5 hours following oral administration. It showed high plasma exposure (AUC = 3 μg h mL⁻¹), moderate plasma residence time (MRT = 1.98 h), and a moderate half-life (IV: t_1/2 = 1.86 h; PO: 1.37 h). 38 also demonstrated a medium volume of distribution at steady state (V_ss = 1.4 L kg⁻¹) and favourable bioavailability (F = 45%).

The selectivity profile of 38 was measured against a panel of 97 kinases that broadly represent the kinome at a concentration of 1 uM. Unsurprisingly, our data revealed that 38 is highly selective (Fig. 6 and Table S5), likely due to binding to an unconserved allosteric site.

Fig. 6 TREEspot interaction map showing the kinome profiling of 38 at a screening concentration of 1 μM against a panel of 97 kinases.

Conclusion

We used a parallel virtual screening approach combining pharmacophore, shape, and similarity-based methods that successfully identified novel allosteric RIPK1 inhibitors. X-ray crystallography provided critical structural insights, guiding the structure-based optimization of a promising piperidine carboxamide hit. Our efforts led to the discovery of 38, a compound with good selectivity, promising cell potency and good PK.

At the outset of our discovery program in 2017, no piperidine carboxamide scaffold-based inhibitor of RIPK1 had been disclosed. Contemporaneous with our work was the disclosure of GSK'547 in 2018,⁴⁸ which also contained the piperidine carboxamide scaffold. In addition, a 2022 publication reported the discovery of a 2,8-diazaspiro[4.5]decan-1-one inhibitor,²² which is a spirocyclic analog of the piperidine carboxamide class.

Future efforts on optimization of 38 would need to focus on improving cell potency and brain-to-plasma ratio before considering any in vivo efficacy studies. Presumably the shift in enzyme-to-cell potency and low brain exposure evident with 38 are due to active efflux from transporters such as P-glycoprotein and breast cancer resistance protein. Nevertheless, compound 38 remains a promising lead with good PK and selectivity profile. Further optimization of related compounds, as well as in vivo activity, will be described in a later publication.

Additionally, the solvent-exposed 4-methylpyrazole moiety in 38 offers an ideal exit vector for linker attachment, enabling the design of selective RIPK1 degraders. Such PROTAC degraders could serve as valuable chemical probes to investigate the scaffolding functions of RIPK1.

We also analyzed structural rearrangements in the allosteric pocket that were dependent on the inhibitor scaffold bound. We found that the positions of Phe162 and Met67 vary, and three major conformations of Phe162 were observed. We hope this insight will prove valuable to other researchers and recommend accounting for this flexibility in structure-based in silico studies.

Experimental

Computational chemistry

Compound library preparation. A compound collection for virtual screening was assembled from purchasable compound catalogs from nine commercial vendors. The SD tools utility available in MOE 2016.0802 (ref. 32) was to prepare the compound structures. This procedure included standardization of molecular structure, removal of salts, enumeration of tautomeric and protonation state at pH 7, and enumeration of stereoisomers. The Filter program within OMEGA 2.5.1 (ref. 49) was used to eliminate non-optimal compounds, including those with undesirable structural alerts, functional groups, physicochemical properties, and lead-likeness criteria.

3D multi-conformer database generation. A multi-conformer database was generated with OMEGA 2.5.1 (ref. 49) using default settings (maximum of 10 conformations per rotatable bonds, maximum of 200 conformers within an energy window of 10 kcal mol⁻¹ from the lowest-energy conformation, modified version of the MMFF94 force field, RMSD-based deduplication threshold of 0.5 Å). The parameter flag -fromCT was set to “true” to ensure initial 3D coordinates were generated using the input connection table.

Shape-based screening. The shape reference query was constructed using the X-ray conformation of GSK-2982772 (2) bound to RIPK1, retrieved from PDB (PDB ID: 5TX5). The multi-conformer database generated was searched with ROCS v3.2.2.2.³⁵ ROCS was run using a built-in ImplicitMillsDean colour force field, and all superpositions were optimized to maximize color overlap after the best shape overlay was identified. The hits were ranked based on an unweighted sum of their shape Tanimoto and the normalized color score to provide a single similarity metric, combo score.

Similarity-based screening. The 2D structure of GSK-2982772 (2) was used as the reference query for similarity searching against the compound library using FTrees.³⁶ Compounds with a similarity score below 0.8 were discarded and the top 500 molecules with the highest similarity score were selected for further analysis.

Pharmacophore-based screening. A structure-based pharmacophore model was developed using the MOE molecular modeling package³² with the EHT annotation scheme. The interactive query editor was used to construct a 3D structure-based pharmacophore guided by a structural overlay of the X-ray structures of 2, (PDB ID: 5TX5), and 1 (PDB ID: 4ITH). A four-feature pharmacophore consisting of two ring aromatic (RA) features, one hydrophobic (HYD), and one hydrogen bond acceptor (HBA) feature based on a systematic assessment of protein-ligand contacts. Hydrogen bond acceptor features were projected as directional vectors aligned with the hydrogen bond axis, of the H-bond acceptor heavy atom. Excluded volume spheres were added to represent steric constraints from protein atoms within 5 Å of the refined ligand. All pharmacophoric features were set to require explicit matching.

Finally, the generated pharmacophores were used as queries to perform virtual screening against previously generated multi-conformer database created with OMEGA. Hits from the database were ranked in order of their rmsdx score, a measure of how well pharmacophoric features matches the annotation points on the aligned ligand conformer.

Protein and ligand preparation for molecular modelling. Ligand preparation was performed using the LigPrep module in Schrödinger 2018-1.⁴⁴ This included addition of hydrogen atoms, assignment of bond order and partial charges, generation of relevant tautomeric and ionization states using Epik at pH 7.0 (±1.5), enumeration of stereoisomers, and the generation of high-quality 3D structures.

Preparation of protein structures was carried out using the Protein Preparation Wizard in Maestro.⁴⁴ This included addition of hydrogens, charge and bond order corrections, removal of extraneous molecules, correction of structural defects, addition of missing side chains, optimization of hydrogen bonding networks, and restrained minimization (RMSD of 0.3 Å) to relax strained bonds, angles, and clashes.

Molecular docking was carried out using the Glide⁴⁵ docking program under the Standard Precision (SP) mode. The receptor grid was defined as a rectangular box based on the mass centre of the bound ligand present in the prepared receptor structure. Hydrogen bond constraints to the backbone NH of Asp156 were defined during the grid generation protocol as a part of our design workflow.

Cloning, expression, and purification of RIPK1

N-terminal HIS-TEV-tagged human RIPK1(1-294) including 4 Cys to Ala mutations (C34A, C127A, C233A, and C240A) was cloned into pFastBac1.¹⁰ Bacmids from this vector were transfected in Sf9 insect cells to generate baculovirus. Protein from baculovirus was expressed in Sf21 cells in Sf-900 II SFM medium + 10% FCS at 26 °C for 64 h in 5 L Waves. Cells were pelleted down and stored at −80 °C for future use. All purification steps were carried out on ice or at 4 °C in presence of necrostatin-4.

For purification, frozen cells were removed from −80 °C storage, thawed and resuspended in lysis buffer (20 mM Tris pH 8, 150 mM NaCl, 5 mM DTT, 40 mM imidazole) supplemented with DNaseI and protease inhibitor tabs (Roche).

Cells were lysed by ultaturrax and centrifuged at 75 000 × g for 40 min. The cleared lysate was loaded on HIS-Trap column (Cytiva). HIS-Trap column was washed with lysis buffer containing 40 mM imidazole and eluted with a gradient to 500 mM imidazole.

RIPK1 eluted from HIS-Trap was incubated with TEV protease in dialysis buffer overnight to remove HIS-TEV tag and loaded on ResourceQ (Cytiva). Elution was performed in a 20CV gradient to 0.5 M NaCl. Elution fractions were pooled, concentrated and loaded on HiLoad S-200 26/60 equilibrated in SEC buffer (20 mM Tris pH 8, 150 mM NaCl, 5 mM DTT). Elution fractions were analyzed through tricine-SDS–PAGE, and relevant fractions were pooled, concentrated to >20 mg ml⁻¹ in presence of necrostatin-4, flash-frozen in liquid nitrogen and stored at −80 °C.

Crystallization, data collection and structure refinement

RIPK1 protein crystals complexed with compound 7 and 36 were grown at 4 °C using hanging drop vapor diffusion method with a protein concentration of 10 mg/ml and a protein to reservoir ratio of 1 : 1. Crystals with 36 and 7 appeared in reservoir conditions 22.5% w/v PEG 3350, 0.22 M NH₄I, 6.00 mM Gly3 and 29% w/v PEG 3350, 0.25 M NH₄I, 6.00 mM Gly3, respectively.

Co-crystals were directly flash cooled in liquid nitrogen to collect X-ray diffraction data. X-ray diffraction data was recorded at beamline X10SA (PXII) at the Swiss Light Source (SLS), Paul Scherrer Institute (PSI) in Switzerland. Data processing was done using XDS⁵⁰ and XSCALE.

The crystals belong to space group P2₁2₁2₁ with two monomers of RIPK1 in the asymmetric unit. The structures with compound 7 (PDB ID: 9HY8) and compound 36 (PDB ID: 9HY9) were solved and refined to a final resolution of 2.98 Å and 2.29 Å, respectively. Structures were solved by molecular replacement (MR) using the CCP4 program Phaser.⁵¹ The model was optimized using Coot⁵² and refined with REFMAC.⁵³ Final structural data statistics are compiled in Tables S3 and S4.

Chemistry

The synthesis of inhibitors described herein employed standard chemical transformations. Starting materials and reagents were purchased from commercial suppliers such as Sigma-Aldrich, Alfa Aesar, TCI, Combi-Block, Enamine, or Acros and were used without further purification unless otherwise indicated. Anhydrous solvents (e.g., THF, DMF, DMA, DMSO, MeOH, DCM, and toluene) were purchased from Sigma-Aldrich and used directly. Purification of final compounds by normal phase flash chromatography utilized a Biotage system applying Biotage SNAP columns with Biotage KP-Sil silica or Biotage Zip Si columns with Biotage KP-Sil silica or a Teledyne ISCO system with RediSep Rf normal phase silica cartridges. Some compounds were purified by preparative HPLC using a Waters Autopurify system with a Waters Xbridge Prep C18 5 μm OBD, 19 mm × 150 mm or 50 mm × 100 mm column and SQ detector mass spectrometer with ESI ionization. NMR spectra were recorded on Bruker instruments operating at 300, 500, or 600 MHz. NMR spectra were obtained as CDCl₃, or (CD₃)₂SO solutions (reported in ppm), using tetramethylsilane (0.00 ppm) or residual solvent (CDCl₃, 7.26 ppm; (CD₃)₂SO, 2.50 ppm) as the reference standard. Low-resolution mass spectra were obtained on either a Waters H class UPLC with a Waters Acquity UPLC BEH C18 1.7 μm, 2.1 mm × 50 mm column, UV detection between 200 and 400 nm, evaporating light scattering detection, and a SQ detector mass spectrometer with ESI ionization or a Water I class UPLC with a Waters Acquity UPLC CSH C18 1.7 μm, 2.1 mm × 50 mm column, UV detection at 254 and 290 nm, evaporating light scattering detection, and a SQ detector 2 mass spectrometer with ESI ionization. High-resolution mass spectra were obtained on a Waters Acquity I-Class UPLC coupled to a LTQ-Orbitrap Elite mass spectrometer. The injection volume was 5 μL. Chromatographic separation was performed on a Waters Acquity UPLC BEH C18 1.7 μm, 2.1 mm × 50 mm column, at a flow rate of 0.5 mL min⁻¹. The mobile phases were 0.1% acetic acid in water (solvent A) and 0.1% acetic acid in acetonitrile (solvent B). The gradient had a total run time of 18 minutes and was as follows: 0–2 minutes 5% B; 2–12 minutes from 5% to 65% B; 12–14 minutes from 65% to 95% B; 14–16 minutes at 95% B; 16–16.1 minutes from 95% to 5% B and 16.1–18 minutes at 5% B. The column temperature was kept at 40 °C. The samples were analyzed using the positive electrospray ionization (ESI) mode. The ESI source temperature was set at 375 °C, the capillary temperature at 320 °C and the electrospray voltage at 4.1 kV. Sheath and auxiliary gas were 45 arbitrary unit and 10 arbitrary unit, respectively. ¹H NMR spectra for intermediates, and compounds 9 to 39 are provided in Fig. S5 and S6. MS data for compounds compounds 9 to 39 are provided in Fig. S7 and HRMS data for 38 is provided in Fig. S8. HRMS were obtained on a Waters Acquity I-Class UPLC coupled to an LTQ-Orbitrap Elite mass spectrometer. The injection volume was 5 μL. Chromatographic separation was performed on a Waters Acquity UPLC BEH C18 1.7 μm, 2.1 mm × 50 mm column, at a flow rate of 0.5 mL min⁻¹. The mobile phases were 0.1% acetic acid in water (solvent A) and 0.1% acetic acid in acetonitrile (solvent B). The gradient had a total run time of 18 min and was as follows: 0–2 min 5% B; 2–12 min from 5 to 65% B; 12–14 min from 65 to 95% B; 14–16 min at 95% B; 16–16.1 min from 95 to 5% B; and 16.1–18 min at 5% B. The column temperature was kept at 40 °C. The samples were analyzed using the positive ESI mode. The ESI source temperature was set at 375 °C, the capillary temperature at 320 °C, and the electrospray voltage at 4.1 kV. Sheath and auxiliary gases were of 45 arbitrary unit and 10 arbitrary unit, respectively.

Procedure A. To the carboxylic acid (compounds A, D, H, or M) was added DMF (0.2 M), amine analogs (compounds B or N, 1.1 eq.), DIEA (2 eq.), and then HATU (1.5 eq.). The reaction was stirred at room temperature overnight. The reaction was poured into water, and the product was filtered off, washed with water and hexanes and dried over house vacuum to give the desired product, compounds C, J or O (ppt method). Alternatively the aqueous reaction mixture was acidified with 1 M HCl and extracted twice with EtOAc. Each organic layer was washed with 0.2 M NaOH, water, and brine, combined, dried over MgSO₄, and concentrated to give the desired product, compounds C, J, or O (liquid extraction method). Additional purification specified with each example (Scheme 1).

Scheme 1 Synthesis of compounds 9–37.

Procedure B. To Boc protected analogs (compounds C) dissolved in THF (0.2 M) and cooled in an ice bath was added 2 eq. of 4 M HCl in dioxane and the reaction allowed to warm to room temperature overnight. The reaction was concentrated to give the HCl salt as a white solid. To the HCl salt was added DMF (0.2 M), benzoic acid analogs (compounds D, 1.1 eq.), and DIEA (4 eq.) and the mixture stirred for 5 min. To the reaction was added HATU (1.5 eq.) and then stirred at room temperature overnight. The desired product was worked up by either the ppt or liquid extraction methods as described in procedure A to give compounds E. Additional purification specified with each example.

Procedure C. To aryl boromide analogs (compounds E, or J) was added Pd(dppf)Cl₂ (0.1 eq.), a solution of aryl boronate ester (or boronic acid) (compounds F or K, 1.3 eq.) in DMF (0.1 M), followed by 2 M Na₂CO₃ (2 eq.). The mixture was purged with N₂, the reaction vial sealed, and then in a dry block set at 80 °C. The reaction progression was followed by LCMS and if necessary additional reagents were added. Upon consumption of the starting material, the reactions were filtered and rinsed with DMF. The filtrate was cooled in an ice bath acidified with TFA and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 10–90%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the compound G or L. Additional purification specified with each example.

Procedure D. To a solution of ethyl ester (compound L, 1.3 g, 3.4 mmol) in THF (4 ml) and water (1.2 ml) was added LiOH (2 N, 3.4 ml, 6.8 mmol) dropwise and the resulting mixture was stirred at room temperature for 4 h. The reaction mixture was diluted with water and the reaction was quenched with the dropwise addition of HCl (0.21 ml, 6.8 mmol). The organic phase was extracted with DCM (×2), the combined organic layers were dried over anhydrous sodium sulfate, filtered and concentrated under reduced pressure to give compound M. Additional purification specified with each example.

Compound C1 (R3 = F, R4 = H, R5 = H, R6 = Me). tert-Butyl 4-fluoro-4-((4-methylbenzyl)carbamoyl)piperidine-1-carboxylate. Synthesized by procedure A and liquid extraction to give the title compound as a viscous oil with a yellow tint carboxylate (260 mg, 0.75 mmol, 85% yield). LCMS ES⁺ C₁₉H₂₇FN₂O₃, expected 350, found 373 [M + Na]⁺.

Compound C2 (R₃ = OMe, R₄ = H, R₅ = H, R₆ = Me). tert-Butyl 4-methoxy-4-((4-methylbenzyl)carbamoyl)piperidine-1-carboxylate. Synthesized by procedure A and liquid extraction to give the title compound as a viscous oil with a yellow tint (66 mg, 0.18 mmol, 20% yield). LCMS ES⁺: C₂₀H₃₀N₂O₄, expected 362, found 385 [M + Na]⁺.

Compound C3 (R₃ = F, R₄ = H, R₅ = H, R₆ = 3,5 F2). tert-Butyl 4-((3,5-difluorobenzyl)carbamoyl)-4-fluoropiperidine-1-carboxylate. Synthesized by procedure A and after liquid extraction was purified by flash chromoatography (0–70% EtOAc in hexanes) to give the title compound as a white solid. LCMS ES⁺: C₁₈H₂₃F₃N₂O₃, expected 372, found: 373 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.93–8.83 (m, 1H), 7.15–7.06 (m, 1H), 6.98–6.87 (m, 2H), 4.31 (d, J = 6.1 Hz, 2H), 3.89 (s, 2H), 2.98 (s, 2H), 1.96–1.73 (m, 4H), 1.41 (s, 9H).

Compound E1 (X = N, Y = CH, Z = CH, R₁ = Br, R₂ = F, R₃ = F, R₄ = H, R₅ = H, R₆ = 3,5 F2). 1-(6-Bromo-3-fluoropicolinoyl)-N-(3,5-difluorobenzyl)-4-fluoropiperidine-4-carboxamide. Synthesized similar to procedure B and liquid extracted to give the title compound (48 mg, 78%), as an off-white semisolid. LCMS (ES⁺) C₁₉H₁₆BrF₄N₃O₂: expected 474, found 475 [M + H]⁺.

Compound E2 (X = CH, Y = N, Z = CH, R₁ = Br, R₂ = F, R₃ = F, R₄ = H, R₅ = H, R₆ = 3,5 F2). 1-(2-Bromo-5-fluoroisonicotinoyl)-N-(3,5-difluorobenzyl)-4-fluoropiperidine-4-carboxamide. Synthesized similar to procedure B and liquid extracted to give the title compound (44 mg, 72%), as an off-white semisolid. LCMS (ES⁺) C₁₉H₁₆BrF₄N₃O₂: expected 474, found 475 [M + H]⁺.

Compound E3 (X = CH, Y = CH, Z = N, R₁ = Br, R₂ = F, R₃ = F, R₄ = H, R₅ = H, R₆ = 3,5 F2). 1-(5-Bromo-2-fluoronicotinoyl)-N-(3,5-difluorobenzyl)-4-fluoropiperidine-4-carboxamide. Synthesized similar to procedure B and liquid extracted to give the title compound (46 mg, 75%), as an off-white semisolid. LCMS (ES⁺) C₁₉H₁₆BrF₄N₃O₂: expected 474, found 475 [M + H]⁺.

Compound J. Ethyl 1-(5-bromo-2-fluorobenzoyl)-4-fluoropiperidine-4-carboxylate. Synthesized similar to procedure A. To a 4 mL vial of 5-bromo-2-fluorobenzoic acid (compound H 1.1 g, 5.2 mmol) and ethyl 4-fluoropiperidine-4-carboxylate hydrochloride (compound I, 1.0 g, 4.7 mmol) in DCM (24 ml) was added DIEA (2.5 ml, 14 mmol) and HATU (2.7 g, 7.1 mmol) sequentially. The reaction vial was sealed, and the mixture was stirred at room temperature for 4 hours. The reaction mixture was diluted with DCM and the aqueous layers were extracted with water (100 mL) and brine (100 mL), and the layers were separated. The organic layers were collected and dried over Na₂SO₄ and filtered under reduced pressure. The residue was dry loaded on silica gel and purified via flash chromatography (0–100% EtOAc in Hexanes) to give the title compound (1.5 g, 4.0 mmol, 84% yield) as a clear oil. The product eluted at approximately 45% EtOAc in hexanes. LCMS (ES⁺) C₁₅H₁₆BrF₂NO₃, expected 375/377, found 376/378 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 7.83–7.59 (m, 2H), 7.33 (t, J = 9.0 Hz, 1H), 4.45–4.37 (m, 1H), 4.21 (q, J = 7.1 Hz, 2H), 3.31–3.26 (m, 1H), 3.20–3.16 (m, 1H), 3.14–3.05 (m, 1H), 2.13–1.84 (m, 4H), 1.28–1.19 (m, 3H).

Compound L. Ethyl 4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxylate. Synthesis similar to procedure C. To a suspension of ethyl 1-(5-bromo-2-fluorobenzoyl)-4-fluoropiperidine-4-carboxylate (compound J, 1.6 g, 4.1 mmol) in toluene (25 ml) and ethanol (16 ml) were added 1-methyl-4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1H-pyrazole (compound K, 1.1 g, 5.4 mmol) and Na₂CO₃ (2 M, 4 ml, 8.2 mmol) and the resulting mixture was degassed with N₂ PdCl₂(dppf)(DCM) (0.34 g, 0.41 mmol) was added and the mixture was stirred at 80 °C overnight. The reaction mixture was diluted with ethyl acetate and filtered through celite. The organic layers were washed with water and brine, combined, dried over anhydrous magnesium sulfate, filtered, and concentrated under reduced pressure. The residue was purified via flash chromatography (0–100% EtOAc in Hexanes to give (1.2 g, 3.4 mmol, 83% yield) as a brown solid. The product eluted in two distinct peaks between 90–100% EtOAc. LCMS (ES⁺) C₁₉H₂₁F₂N₃O₃, expected 377, found 378 [M + H]⁺. 1H NMR (500 MHz, DMSO-d₆) δ 8.18 (s, 1H), 7.89 (s, 1H), 7.70–7.65 (m, 1H), 7.63 (s, 1H), 7.29 (t, J = 9.0 Hz, 1H), 4.49–4.41 (m, 1H), 4.20 (q, J = 7.1 Hz, 2H), 3.85 (s, 3H), 3.49–3.40 (m, 1H), 3.31–3.24 (m, 1H), 3.15–3.05 (m, 1H), 2.13–1.86 (m, 4H), 1.23 (t, J = 7.1 Hz, 3H).

Compound M. 4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxylic acid. Synthesized similar to procedure D. To a solution of ethyl 4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxylate (compound L, 1.3 g, 3.4 mmol) in THF (4 ml) and water (1.2 ml) was added LiOH (2 N, 3.4 ml, 6.8 mmol) dropwise and the resulting mixture was stirred at room temperature for 4 h. The reaction mixture was diluted with water and the reaction was quenched with the dropwise addition of HCl (0.208 ml, 6.84 mmol). The organic phase was extracted with DCM (×2), the combined organic layers were dried over anhydrous sodium sulfate, filtered and concentrated under reduced pressure to give the title compound (1.2 g, 3.3 mmol, 96% yield) as a brown solid. LCMS (ES⁺) C₁₇H₁₇F₂N₃O₃, expected 349, found 350 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 14–13 (1H not observed), 8.19 (s, 1H), 7.90 (s, 1H), 7.70–7.66 (m, 1H), 7.63 (s, 1H), 7.30 (t, J = 9.0 Hz, 1H), 4.45 (dt, J = 13.4, 4.2 Hz, 1H), 3.86 (s, 3H), 3.47–3.44 (m, 1H), 3.31–3.28 (m, 1H), 3.10 (t, J = 14.2 Hz, 1H), 2.08–1.81 (m, 4H).

1-(2,5-Dichlorobenzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (7). At the time of testing was purchased from Asinex, catalog number: BAS 11102405.

1-Benzoyl-4-methyl-N-(4-methylbenzyl)piperidine-4-carboxamide (8). At the time of testing was purchased from ChemDiv, catalog number: F709-0540.

1-(5-Chloro-2-fluorobenzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (9). Synthesized by procedure B, liquid extraction and purification by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 10–90%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₁H₂₂ClFN₂O₂, expected 388, found 389 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.31 (t, J = 6.0 Hz, 1H), 7.59–7.54 (m, 1H), 7.37 (t, J = 8.9 Hz, 1H), 7.11 (s, 4H), 4.51–4.41 (m, 1H), 4.26–4.16 (m, 2H), 3.48–3.37 (m, 2H), 3.14–3.02 (m, 1H), 2.84 (t, J = 12.6 Hz, 1H), 2.48–2.42 (m, 1H), 2.26 (s, 3H), 1.86–1.79 (m, 1H), 1.74–1.64 (m, 1H), 1.61–1.43 (m, 2H).

1-(5-Bromo-2-fluorobenzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (10). Synthesized by procedure B, liquid extraction and purification by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 10–90%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₁H₂₂BrFN₂O₂, expected 432/434, found 433/435 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.31 (t, J = 6.0 Hz, 1H), 7.71–7.66 (m, 1H), 7.31 (t, J = 9.0 Hz, 1H), 7.11 (s, 4H), 4.51–4.41 (m, 1H), 4.27–4.15 (m, 2H), 3.45–3.38 (m, 2H), 3.14–3.01 (m, 1H), 2.84 (t, J = 12.7 Hz, 1H), 2.48–2.42 (m, 1H), 2.26 (s, 3H), 1.86–1.78 (m, 1H), 1.74–1.64 (m, 1H), 1.62–1.45 (m, 2H).

1-(5-Bromo-2-fluorobenzoyl)-4-methoxy-N-(4-methylbenzyl)piperidine-4-carboxamide (11). Synthesized by procedure B, liquid extraction, and used further to give the desired product (74 mg, 0.160 mmol, 88% yield) as a light yellow solid. LCMS ES⁺, C₂₂H₂₄BrFN₂O, expected 462/464, found 463/465 [M + H]⁺. 1H NMR (500 MHz, DMSO-d₆) δ 8.53 (t, J = 6.3 Hz, 1H), 7.71–7.62 (m, 1H), 7.31 (t, J = 9.0 Hz, 1H), 7.11 (q, J = 8.1 Hz, 4H), 4.29–4.20 (m, 3H), 3.39–3.33 (m, 1H), 3.29–3.16 (m, 2H), 3.12 (s, 3H), 3.10–3.00 (m, 1H), 2.26 (s, 3H), 1.96–1.65 (m, 4H).

1-(5-Bromo-2-fluorobenzoyl)-4-fluoro-N-(4-methylbenzyl)piperidine-4-carboxamide (12). Synthesized by procedure B and was initially worked up by the ppt method with minimal recover. Additional product was recovered from the filtrate by the liquid extraction method. The desired products from each method were combined and further purified by flash chromatography (0 to 100% EtOAc in hexanes) to give the desired product (33 mg, 0.073 mmol, 9.1% yield) as a white solid. LCMS ES⁺, C₂₁H₂₁BrF₂N₂O₂ expected 450/452, found 451/453 [M + H]⁺. 1H NMR (500 MHz, DMSO-d₆) δ 8.84–8.72 (m, 1H), 7.79–7.59 (m, 1H), 7.32 (t, J = 9.1 Hz, 1H), 7.11 (s, 4H), 4.44 (d, J = 13.3 Hz, 1H), 4.31–4.20 (m, 2H), 3.42–3.36 (m, 2H), 3.30–3.24 (m, 1H), 3.12–3.00 (m, 1H), 2.26 (s, 3H), 2.13–1.87 (m, 3H), 1.85–1.72 (m, 1H).

1-(2-Fluoro-5-(pyridin-3-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (13). Synthesized by procedure C to give the title compound (16 mg, 67% yield) as a TFA salt. LCMS ES⁺ C₂₆H₂₆FN₃O₂ expected 431, found 432 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 9.03 (s, 1H), 8.69 (d, J = 5.0 Hz, 1H), 8.36 (s, 1H), 8.32 (t, J = 6.0 Hz, 1H), 7.93–7.88 (m, 1H), 7.74–7.66 (m, 1H), 7.48 (t, J = 8.9 Hz, 1H), 7.11 (s, 4H), 4.57–4.49 (m, 1H), 4.26–4.15 (m, 2H), 3.47–3.43 (m, 2H), 3.16–3.03 (m, 1H), 2.91–2.81 (m, 1H), 2.49–2.44 (m, 1H), 2.26 (s, 3H), 1.87–1.81 (m, 1H), 1.75–1.65 (m, 1H), 1.62–1.50 (m, 2H).

1-(2-Fluoro-5-(thiazol-5-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (14). Synthesized by procedure C to give the title compound (6.8 mg, 27% yield) as a TFA salt. LCMS ES⁺ C₂₄H₂₄FN₃O₂S, expected 437, found 438 [M + H]⁺. 1H NMR (600 MHz, DMSO-d6) δ 9.11 (s, 1H), 8.36 (s, 1H), 8.34–8.28 (m, 1H), 7.83–7.77 (m, 1H), 7.40 (t, J = 8.9 Hz, 1H), 7.11 (s, 4H), 4.56–4.47 (m, 1H), 4.25–4.16 (m, 2H), 3.52–3.44 (m, 2H), 3.15–3.03 (m, 1H), 2.90–2.80 (m, 1H), 2.49–2.42 (m, 1H), 2.26 (s, 3H), 1.87–1.81 (m, 1H), 1.75–1.63 (m, 1H), 1.63–1.47 (m, 2H).

1-(2-Fluoro-5-(1-methyl-1H-pyrazol-5-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (15). Synthesized by procedure C to give the title compound as a TFA salt. LCMS ES⁺ C₂₅H₂₇FN₄O₂, expected 434, found 435 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.36–8.28 (m, 1H), 7.68–7.63 (m, 1H), 7.48 (d, J = 2.0 Hz, 1H), 7.44 (t, J = 9.0 Hz, 1H), 7.11 (s, 4H), 6.45 (d, J = 1.9 Hz, 1H), 4.55–4.47 (m, 1H), 4.27–4.17 (m, 2H), 3.86 (s, 3H), 3.50 (m, 2H), 3.15–3.05 (m, 1H), 2.91–2.82 (m, 1H), 2.49–2.43 (m, 1H), 2.27 (s, 3H), 1.88–1.80 (m, 1H), 1.75–1.64 (m, 1H), 1.62–1.47 (m, 2H).

1-(2-Fluoro-5-(1H-pyrazol-4-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (16). Synthesized by procedure C using tert-butyl 4-(4,4,5,5-tetramethyl-1,3,2-dioxaborolan-2-yl)-1H-pyrazole-1-carboxylate as the aryl boronate ester where the boc protecting group decompose during the reaction to give the title compound (17 mg, 72% yield) as a TFA salt, white solid. LCMS ES⁺ C₂₄H₂₅FN₄O₂, expected 420, found 421 [M + H]⁺. 1H NMR (600 MHz, MeOD) δ 9.18 (t, J = 5.9 Hz, 1H), 8.96 (s, 2H), 8.59–8.54 (m, 1H), 8.52–8.42 (m, 1H), 8.13 (t, J = 9.0 Hz, 1H), 7.97 (s, 4H), 5.42–5.34 (m, 1H), 5.12–5.02 (m, 2H), 4.41–4.33 (m, 2H), 3.99–3.88 (m, 1H), 3.75–3.65 (m, 1H), 3.34–3.28 (m, 1H), 3.12 (s, 3H), 2.73–2.65 (m, 1H), 2.62–2.50 (m, 1H), 2.49–2.30 (m, 2H).

1-(2-Fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (17). Synthesized by procedure C to give the title compound (11 mg, 46% yield) as a TFA salt. LCMS ES⁺ C₂₅H₂₇FN₄O₂, expected 434, found 435 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.32 (t, J = 6.0 Hz, 1H), 8.18 (s, 1H), 7.89 (s, 1H), 7.68–7.62 (m, 1H), 7.27 (t, J = 9.0 Hz, 1H), 7.11 (s, 4H), 4.56–4.46 (m, 1H), 4.28–4.16 (m, 2H), 3.85 (s, 3H), 3.50–3.45 (m, 2H), 3.14–3.01 (m, 1H), 2.90–2.79 (m, 1H), 2.49–2.42 (m, 1H), 2.26 (s, 3H), 1.83 (dd, J = 13.6, 3.6 Hz, 1H), 1.68 (s, 1H), 1.62–1.45 (m, 2H).

4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(4-methylbenzyl)piperidine-4-carboxamide (18). Synthesized by procedure C to give the title compound (13 mg, 62% yield) as a TFA salt, brown solid. LCMS ES⁺ C₂₅H₂₆F₂N₄O₂, expected 452, found 453 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.81–8.75 (m, 1H), 8.18 (s, 1H), 7.89 (s, 1H), 7.70–7.65 (m, 1H), 7.29 (t, J = 9.0 Hz, 1H), 7.11 (s, 4H), 4.55–4.44 (m, 1H), 4.31–4.21 (m, 2H), 3.85 (s, 3H), 3.50–3.43 (m, 2H), 3.31–3.26 (m, 1H), 3.13–3.01 (m, 1H), 2.26 (s, 3H), 2.13–1.89 (m, 3H), 1.79 (s, 1H).

(S)-4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(1-phenylethyl)piperidine-4-carboxamide (19). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 40–80%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (16 mg, 46% yield). LCMS ES⁺ C₂₅H₂₆F₂N₄O₂, expected 452, found 453 [M + H]⁺.

(R)-4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(1-phenylethyl)piperidine-4-carboxamide (20). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 40–80%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (17 mg, 47%). LCMS ES⁺ C₂₅H₂₆F₂N₄O₂, expected 452, found 453 [M + H]⁺.

4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(2-fluorobenzyl)piperidine-4-carboxamide (21). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₄H₂₃F₃N₄O₂, expected 456, found 457 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.86–8.80 (m, 1H), 8.18 (s, 1H), 7.90 (s, 1H), 7.70–7.65 (m, 1H), 7.64–7.55 (m, 1H), 7.36–7.23 (m, 3H), 7.20–7.12 (m, 2H), 4.54–4.45 (m, 1H), 4.40–4.30 (m, 2H), 3.85 (s, 3H), 3.35–3.28 (m, 1H), 3.13–3.04 (m, 2H), 2.14–1.90 (m, 3H), 1.88–1.74 (m, 1H).

4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(3-fluorobenzyl)piperidine-4-carboxamide (22). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (15 mg, 48%). LCMS ES⁺ C₂₄H₂₃F₃N₄O₂, expected 456, found 457 [M + H]⁺.

4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(4-fluorobenzyl)piperidine-4-carboxamide (23). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (16 mg, 45%). LCMS ES⁺ C₂₄H₂₃F₃N₄O₂, expected 456, found 457 [M + H]⁺.

N-(2-Chlorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (24). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (5.9 mg, 16%). LCMS ES⁺ C₂₄H₂₃ClF₂N₄O₂, expected 472, found 473 [M + H]⁺.

N-(3-Chlorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (25). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (16 mg, 44%). LCMS ES⁺ C₂₄H₂₃ClF₂N₄O₂, expected 472, found 473 [M + H]⁺.

N-(4-Chlorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (26). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 10–90%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₄H₂₃ClF₂N₄O₂, expected 472, found 473 [M + H]⁺.

4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(2-fluorobenzyl)-N-methylpiperidine-4-carboxamide (27). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (9.5 mg, 26%). LCMS ES⁺ C₂₅H₂₅F₃N₄O₂, expected 470, found 471 [M + H]⁺.

(S)-4-Fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)-N-(1-(2-fluorophenyl)ethyl)piperidine-4-carboxamide (28). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₅H₂₅F₃N₄O₂, expected 470, found 471 [M + H]⁺.

N-(2,3-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (29). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (16 mg, 42%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺.

N-(4-Chloro-2-fluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (30). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (15 mg, 38%). LCMS ES⁺ C₂₄H₂₂ClF₃N₄O₂, expected 490, found 491 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.92–8.85 (m, 1H), 8.18 (s, 1H), 7.89 (s, 1H), 7.70–7.65 (m, 1H), 7.64–7.56 (m, 1H), 7.40–7.35 (m, 1H), 7.32–7.21 (m, 3H), 4.55–4.45 (m, 1H), 4.38–4.27 (m, 2H), 3.85 (s, 3H), 3.50–3.42 (m, 1H), 3.11–3.01 (m, 2H), 2.10–1.90 (m, 3H), 1.87–1.74 (m, 1H).

N-(2,4-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (31). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (14 mg, 38%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺.

N-(2,5-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (32). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (Mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (13 mg, 36%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.93–8.85 (m, 1H), 8.18 (s, 1H), 7.90 (s, 1H), 7.70–7.65 (m, 1H), 7.64–7.55 (m, 1H), 7.30 (t, J = 9.0 Hz, 1H), 7.26–7.21 (m, 1H), 7.18–7.11 (m, 1H), 7.07–7.01 (m, 1H), 4.55–4.46 (m, 1H), 4.40–4.28 (m, 2H), 3.86 (s, 3H), 3.50–3.46 (m, 1H), 3.13–3.04 (m, 2H), 2.13–1.91 (m, 3H), 1.89–1.76 (m, 1H).

N-(2,6-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (33). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (12 mg, 32%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺.

N-(3,4-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (34). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (14 mg, 37%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺.

N-(3,5-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (35). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound (15 mg, 41%). LCMS ES⁺ C₂₄H₂₂F₄N₄O₂, expected 474, found 475 [M + H]⁺.

(S)-N-(1-(2-Chloro-6-fluorophenyl)ethyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)benzoyl)piperidine-4-carboxamide (36). Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 12 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₅H₂₄ClF₃N₄O₂, expected 504, found 505 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.54–8.47 (m, 1H), 8.17 (d, J = 8.0 Hz, 1H), 7.88 (d, J = 8.1 Hz, 1H), 7.70–7.63 (m, 1H), 7.62–7.55 (m, 1H), 7.34–7.23 (m, 3H), 7.22–7.12 (m, 1H), 5.37–5.27 (m, 1H), 4.45 (dd, J = 28.4, 13.4 Hz, 1H), 3.85 (d, J = 5.7 Hz, 3H), 3.46–3.37 (m, 1H), 3.32–3.19 (m, 2H), 3.11–2.98 (m, 1H), 2.03–1.77 (m, 3H), 1.75–1.63 (m, 0H), 1.53–1.42 (m, 3H).

N-(3,5-Difluorobenzyl)-4-fluoro-1-(3-fluoro-6-(1-methyl-1H-pyrazol-4-yl)picolinoyl)piperidine-4-carboxamide (37). (6.4 mg, 0.013 mmol, 26% yield) as an off white solid. Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 20 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₃H₂₁F₄N₅O₂, expected 475, found 476 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.99–8.91 (m, 1H), 8.27 (s, 1H), 7.96 (s, 1H), 7.85 (t, J = 9.0 Hz, 1H), 7.82–7.77 (m, 1H), 7.10 (t, J = 9.2 Hz, 1H), 6.95 (d, J = 7.6 Hz, 2H), 4.53–4.45 (m, 1H), 4.33 (d, J = 6.1 Hz, 2H), 3.87 (s, 3H), 3.50–3.42 (m, 1H), 3.32–3.26 (m, 1H shoulder of water peak) 3.15–3.07 (m, 1H), 2.14–1.94 (m, 3H), 1.88–1.80 (m, 1H).

N-(3,5-Difluorobenzyl)-4-fluoro-1-(5-fluoro-2-(1-methyl-1H-pyrazol-4-yl)isonicotinoyl)piperidine-4-carboxamide (38). (12.3 mg, 0.026 mmol, 51.3% yield) as an off white solid. Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 20 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₃H₂₁F₄N₅O₂, expected 475, found 476 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.98–8.90 (m, 1H), 8.59 (s, 1H), 8.29 (s, 1H), 8.01 (s, 1H), 7.68 (d, J = 145.9 Hz, 1H), 7.10 (t, J = 9.5 Hz, 1H), 6.94 (d, J = 7.6 Hz, 2H), 4.53–4.45 (m, 1H), 4.38–4.27 (m, 2H), 3.88 (s, 3H), 3.52–3.43 (m, 1H), 3.37–3.34 (m, 1H shoulder of water), 3.15–3.06 (m, 1H), 2.13–1.92 (m, 3H), 1.89–1.77 (m, 1H). HRMS (ES⁺) C₂₃H₂₂F₄N₅O₂ calculated 476.1704 [M + H]⁺; found 476.1696 [M + H]⁺.

N-(3,5-Difluorobenzyl)-4-fluoro-1-(2-fluoro-5-(1-methyl-1H-pyrazol-4-yl)nicotinoyl)piperidine-4-carboxamide (39). (10.6 mg, 0.022 mmol, 44.2% yield) as an off white solid. Synthesized similar to procedure A and purified by mass-triggered preparative HPLC (mobile phase: A = 0.1% TFA/H₂O, B = 0.1% TFA/MeCN; gradient: B 30–70%; 20 min; column: XBridge C18, 5 μm, 19 mm × 150 mm) to give the title compound. LCMS ES⁺ C₂₃H₂₁F₄N₅O₂, expected 475, found 476 [M + H]⁺. 1H NMR (600 MHz, DMSO-d₆) δ 8.98–8.91 (m, 1H), 8.57 (s, 1H), 8.30 (s, 1H), 8.26 (s, 1H), 8.01 (s, 1H), 7.10 (t, J = 9.3 Hz, 1H), 6.94 (d, J = 7.6 Hz, 2H), 4.54–4.45 (m, 1H), 4.38–4.28 (m, 2H), 3.87 (s, 3H), 3.53–3.44 (m, 1H), 3.37–3.34 (m, 1H), 3.14–3.05 (m, 1H), 2.13–1.94 (m, 4H), 1.83 (sm, 1H).

ADP-Glo assay

RIPK1 autophosphorylation activity was measured biochemically using Promega's ADP-Glo Assay kit (Promega, V7001). The assay was performed in a 384-well, white, PerkinElmer, Optiplate (PerkinElmer, 6007290) in assay buffer containing 50 mM Hepes (pH 7.5) (Life Technologies, 15630080), 50 mM NaCl (Teknova, S0252), 30 mM MgCl2 (Ambion, AM9530G), 0.02% CHAPS (Sigma, C5070), 0.05% BSA (Sigma, A3059), and 1 mM DTT (GoldBio, DTT10). The final enzymatic reaction volume was 10 μL. 4 μL of GST tagged-RIPK1 (1-327) (Signal Chem, R07-11G-05) diluted in assay buffer was transferred to the assay plate for a final concentration of 5 nM. 4 μL of assay buffer, instead of RIPK1, were added to inhibitor control wells. Stock solutions of the test compounds were prepared in 100% DMSO (Sigma, D2650) and serially diluted 1 : 3 using 100% DMSO. Compounds were additionally diluted 40-fold in assay buffer, and 2 μL per well were transferred to the assay plate for a final DMSO concentration of 0.5% followed by a 10 minute incubation at room temperature. 4 μL of ATP diluted in assay buffer were transferred to the assay plate for a final concentration of 50 μM followed by a 60 minute incubation at room temperature.

The ADP-Glo Reagent was diluted to 1× in water with 10% CHAPS, and 5 μL per well were added to the assay plate followed by a 60 minute incubation at room temperature. The kinase detection reagent was diluted to 1× in water with 10% CHAPS, and 10 μL per well were added to the assay plate under subdued light followed by a 30-minute incubation at room temperature in the dark. The luminescence signal was measured using a BioTek Synergy Neo plate reader. IC₅₀ values were calculated using a four-parameter logistic curve fit using Genedata Screener software.

Necroptosis assay

U937 cells were routinely maintained in RPMI media (Gibco, 11875-093) supplemented with 10% heat-inactivated fetal bovine serum (Gibco, 16140-071) and 1× PenStrep (Life Technologies, 15140122) using a humidified incubator (37 °C, 5% CO₂, and ambient O₂).

In preparation for the necroptosis assay, cells were harvested and resuspended in phenol red free RPMI media (Gibco, 11835-030) supplemented with 10% fetal bovine serum (Sigma, F4135) and 1× PenStrep. Cells were seeded onto a 384-well white PerkinElmer Tissue Culture Plate (PerkinElmer, 6007680) at a density of 5000 cells per well in a volume of 40 μL. Included in the cell seeding suspension was Z-VAD (R&D systems, FMK001) and TNFalpha (Cell Sciences, CSI15659B) for final concentrations of 25 μM and 25 ng mL⁻¹, respectively, per 40 μL per well. Stock solutions of the test compounds were prepared in 100% DMSO (Sigma, Catalog #D2650) and serially diluted 1 : 3 using 100% DMSO. Compounds were additionally diluted 1 : 40 in culture medium, and 10 μL per well were transferred to the tissue culture plate. Following the compound addition the microplate was incubated at 37 °C for 20 hours. 20 μL per well of Promega's CellTiter Glo 2.0 (Promega, G9243) were added to the tissue culture plate. The tissue culture plate was then shaken on an orbital shaker at 500 RPM for 15 minutes at room temperature. Luminescence signal was measured using a PerkinElmer Envision plate reader. IC₅₀ values were calculated using a four-parameter logistic curve fit using Genedata Screener software.

Microsomal stability assay

Microsomal stability assay was conducted on a Beckmann Biomek FX^p laboratory automation system. The liver microsomal incubation mixture consisted of liver microsomes (0.5 mg microsomal protein per ml), the test compound (1 μM), MgCl2 (3 mM), EDTA (1 mM) in potassium phosphate buffer (100 mM, pH 7.4). Midazolam and Ketanserin were used as the assay control substrates. The reaction was initiated with the addition of an NADPH regeneration solution (1.3 mM NADPH) and maintained at 37 °C with shaking. At five time points ranging from 0 to 45 min, aliquots (50 μL) were removed and quenched with acetonitrile (100 μL) containing an internal standard (imipramine). After vortex and centrifugation, samples were analyzed by LC-MS/MS. Calculation of the in vitro T_1/2 and clearance was determined as described in literature.

CYP inhibition assay

CYP450 inhibition studies were performed using mixed-gender, pooled human liver microsomes (hLM) from BioIVT (Xtreme-200) at a final concentration of 0.1 mg mL⁻¹. Reactions were conducted in 50 mM phosphate buffer (pH 7.4) supplemented with 1 mM NADPH. Test compounds were prepared as 10-point titrations ranging from 10 μM to 0.5 nM, with a final DMSO concentration of 1%.

Each reaction consisted of a 5 μL aliquot of compound preincubated with 10 μL hLM for 30 minutes at room temperature. Substrates were pooled into two different cocktails for CYP3A4 as per FDA guidelines for CYP inhibition testing. Pool A contained 20 μM testosterone (CYP3A4), and Pool B contained 2 uM midazolam (CYP3A4). All substrates were used at concentrations approximating their Km.

Reactions proceeded for 45 minutes at room temperature with orbital shaking (IKA, MTS 2/4) at 300 rpm. Reactions were quenched by the addition of 60 μL of acetonitrile containg 0.1% formic acid and 3 μM internal standard (6β-hydroxytestosterone-d7), followed by centrifugation at 2250 × g for 20 minutes at 4 °C. Supernatant was analyzed using an Agilent RapidFire 365 mass spectrometry system coupled to a 6460 QQQ. Metabolites monitored were 6β-hydroxytestosterone and α-hydroxymidazolam.

Metabolite peak areas were quantified by area under the curve (AUC), normalized to the internal standard, and compared to no-HLM control wells to calculate percent of control activity (POC). %Inhibition was determined relative to POC, and IC₅₀ values were calculated using a four-parameter logistic regression model.

Kinome profiling

Compound 38 was assayed at 1 μM against 97 kinases that consistute the Eurofins scanEDGE panel, part of the KINOMEscan platform. Data are reported in Table S1. Protocols are available from Eurofins DiscoverX (https://www.eurofinsdiscovery.com/solution/kinomescan-technology).

Ethics statement

All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of Shanghai ChemPartner Co, Ltd. and approved by the Animal Ethics Committee of Institutional Animal Care and Use Committee (IACUC), Shanghai ChemPartner Co., Ltd.

Author contributions

JBC, RSKV, and RTL conceived and designed the study. MMH designed the chemistry experiments and conducted them with DEP. FGA and NJR performed data curation. JPB, CR, and HS performed the U937 necroptosis and ADP-Glo assays, while CR also carried out the CYP inhibition assay. KM, SJ, GK, and LN handled protein purification and X-ray crystallography. QAX and YJ performed the microsomal stability assay and oversaw the PK study. SG and YL assisted with data analysis and interpretation. RSK, MMH, and JBC drafted the manuscript with input from all co-authors. RTL, MGD, PJ, WJR, and JBC supervised the study. All authors have read and given approval to the final version of the manuscript.

Conflicts of interest

The authors declare no conflict of interest.

Data availability

Data for this article, including details on virtual screening hits, X-ray data and refinement statistics, annotation of Phe162 conformations, kinome profiling data, and MS and NMR data for compounds are available in the SI. See DOI: https://doi.org/10.1039/D5MD00317B.

Crystallographic data for 7 and 36 bound to RIPK1 have been deposited at the PDB under accession numbers 9HY8 and 9HY9, respectively, and can be obtained from https://www.rcsb.org/structure/9HY8 and https://www.rcsb.org/structure/9HY9.⁵⁴

Acknowledgements

This work was supported by the Robert A. and Renee E. Belfer Family Foundation. We acknowledge the Paul Scherrer Institut, Villigen, Switzerland for provision of synchrotron radiation beamtime at the SLS.

References

1. J. Yuan, P. Amin and D. Ofengeim, Nat. Rev. Neurosci., 2019, 20, 19–33.
2. D. Ofengeim and J. Yuan, Nat. Rev. Mol. Cell Biol., 2013, 14, 727–736.
3. D. Ofengeim, S. Mazzitelli, Y. Ito, J. P. DeWitt, L. Mifflin, C. Zou, S. Das, X. Adiconis, H. Chen, H. Zhu, M. A. Kelliher, J. Z. Levin and J. Yuan, Proc. Natl. Acad. Sci. U. S. A., 2017, 114, E8788–E8797.
4. C. Gao, J. Jiang, Y. Tan and S. Chen, Signal Transduction Targeted Ther., 2023, 8, 359.
5. B. Shutinoski, N. A. Alturki, D. Rijal, J. Bertin, P. J. Gough, M. G. Schlossmacher and S. Sad, Cell Death Differ., 2016, 23, 1628–1637.
6. A. Degterev, Z. Huang, M. Boyce, Y. Li, P. Jagtap, N. Mizushima, G. D. Cuny, T. J. Mitchison, M. A. Moskowitz and J. Yuan, Nat. Chem. Biol., 2005, 1, 112–119.
7. L. Mifflin, D. Ofengeim and J. Yuan, Nat. Rev. Drug Discovery, 2020, 19, 553–571.
8. K. Shi, J. Zhang, E. Zhou, J. Wang and Y. Wang, J. Med. Chem., 2022, 65, 14971–14999.
9. C. R. Gardner, K. A. Davies, Y. Zhang, M. Brzozowski, P. E. Czabotar, J. M. Murphy and G. Lessene, J. Med. Chem., 2023, 66, 2361–2385.
10. T. Xie, W. Peng, Y. Liu, C. Yan, J. Maki, A. Degterev, J. Yuan and Y. Shi, Structure, 2013, 21, 493–499.
11. A. Degterev, D. Ofengeim and J. Yuan, Proc. Natl. Acad. Sci. U. S. A., 2019, 116, 9714–9722.
12. J. Lickliter, S. Wang, W. Zhang, H. Zhu, J. Wang, C. Zhao, H. Shen and Y. Wang, Clin. Transl. Sci., 2023, 16, 1691–1703.
13. https://clinicaltrials.gov/study/NCT04781816.
14. A. Hincelin-Mery, X. Nicolas, C. Cantalloube, R. Pomponio, P. Lewanczyk, M. Benamor, D. Ofengeim, E. Krupka, J. Hsiao-Nakamoto, A. Eastenson and N. Atassi, Clin. Transl. Sci., 2024, 17, e13690.
15. A. L. A. Sun, J. D. Gillies, Y. Shen, H. Deng, F. Xue, Y. Ma and L. Song, Clin. Transl. Sci., 2024, 17, e13857.
16. https://clinicaltrials.gov/study/NCT05848258.
17. K. Weisel, N. E. Scott, D. J. Tompson, B. J. Votta, S. Madhavan, K. Povey, A. Wolstenholme, M. Simeoni, T. Rudo, L. Richards-Peterson, T. Sahota, J. G. Wang, J. Lich, J. Finger, A. Verticelli, M. Reilly, P. J. Gough, P. A. Harris, J. Bertin and M. L. Wang, Pharmacol. Res. Perspect., 2017, 5(6), e365.
18. K. Weisel, N. Scott, S. Berger, S. Wang, K. Brown, M. Powell, M. Broer, C. Watts, D. J. Tompson, S. W. Burriss, S. Hawkins, K. Abbott-Banner and P. P. Tak, BMJ Open Gastroenterol., 2021, 8(1), e000680.
19. https://www.biospace.com/denali-and-sanofi-pause-alzheimer-s-trial-and-pivot-to-another-drug.
20. https://clinicaltrials.gov/study/NCT05237284.
21. https://www.clinicaltrialsarena.com/news/sanofi-and-denalis-als-therapy-misses-primary-endpoint-in-phase-ii/.
22. A. Niu, L. Lin, D. Zhang, K. Jiang, D. Weng, W. Zhou and J. Wang, Bioorg. Med. Chem., 2022, 59, 116686.
23. B. Liu, L. Zhao, Y. Tan, X. Yao, H. Liu and Q. Zhang, ACS Chem. Neurosci., 2025, 16, 1617–1630.
24. Y. Li, L. Zhang, Y. Wang, J. Zou, R. Yang, X. Luo, C. Wu, W. Yang, C. Tian, H. Xu, F. Wang, X. Yang, L. Li and S. Yang, Nat. Commun., 2022, 13, 6891.
25. Y. Yu, Y. Hu, H. Yan, X. Zeng, H. Yang, L. Xu and R. Sheng, Drug Dev. Res., 2024, 85, e22235.
26. X. Zhang, J. B. Cross, J. Romero, A. Heifetz, E. Humphries, K. Hall, Y. Wu, S. Stucka, J. Zhang, H. Chandonnet, B. Lippa, M. D. Ryan and J. C. Baber, J. Comput.-Aided Mol. Des., 2018, 32, 573–582.
27. G. M. Sastry, V. S. Inakollu and W. Sherman, J. Chem. Inf. Model., 2013, 53, 1531–1542.
28. Y. Tanrikulu, B. Kruger and E. Proschak, Drug Discovery Today, 2013, 18, 358–364.
29. G. B. McGaughey, R. P. Sheridan, C. I. Bayly, J. C. Culberson, C. Kreatsoulas, S. Lindsley, V. Maiorov, J. F. Truchon and W. D. Cornell, J. Chem. Inf. Model., 2007, 47, 1504–1519.
30. C. G. Bologa, O. Ursu and T. I. Oprea, Methods Mol. Biol., 2019, 1939, 119–138.
31. M. M. Hann and T. I. Oprea, Curr. Opin. Chem. Biol., 2004, 8, 255–263.
32. Molecular Operating Environment (MOE), 2016, 0802 Chemical Computing Group ULC, 910–1010 Sherbrooke St. W., Montreal, QC H3A 2R7, 2025.
33. P. A. Harris, S. B. Berger, J. U. Jeong, R. Nagilla, D. Bandyopadhyay, N. Campobasso, C. A. Capriotti, J. A. Cox, L. Dare, X. Dong, P. M. Eidam, J. N. Finger, S. J. Hoffman, J. Kang, V. Kasparcova, B. W. King, R. Lehr, Y. Lan, L. K. Leister, J. D. Lich, T. T. MacDonald, N. A. Miller, M. T. Ouellette, C. S. Pao, A. Rahman, M. A. Reilly, A. R. Rendina, E. J. Rivera, M. C. Schaeffer, C. A. Sehon, R. R. Singhaus, H. H. Sun, B. A. Swift, R. D. Totoritis, A. Vossenkamper, P. Ward, D. D. Wisnoski, D. Zhang, R. W. Marquis, P. J. Gough and J. Bertin, J. Med. Chem., 2017, 60, 1247–1261.
34. https://www.massbio.org/news/member-news/investigating-the-sar-of-ripk1-inhibitors/.
35. T. S. Rush, 3rd, J. A. Grant, L. Mosyak and A. Nicholls, J. Med. Chem., 2005, 48, 1489–1495.
36. M. Rarey and J. S. Dixon, J. Comput.-Aided Mol. Des., 1998, 12, 471–490.
37. F. Svensson, A. Karlen and C. Skold, J. Chem. Inf. Model., 2012, 52, 225–232.
38. G. W. Bemis and M. A. Murcko, J. Med. Chem., 1996, 39, 2887–2893.
39. G. R. Bickerton, G. V. Paolini, J. Besnard, S. Muresan and A. L. Hopkins, Nat. Chem., 2012, 4, 90–98.
40. https://www.promega.com/-/media/files/resources/protcards/adp-glo-kinase-assay-quick-protocol.pdf?rev=19aef4612b064d43b0e7d604dcf164bd&sc_lang=en.
41. R. S. Vijayan, P. He, V. Modi, K. C. Duong-Ly, H. Ma, J. R. Peterson, R. L. Dunbrack, Jr. and R. M. Levy, J. Med. Chem., 2015, 58, 466–479.
42. M. N. Schulz, J. Landstrom, K. Bright and R. E. Hubbard, J. Comput.-Aided Mol. Des., 2011, 25, 611–620.
43. https://enamine.net/compound-collections/screening-collection.
44. Schrödinger Release 218-1, Schrödinger, LLC, New York, NY, 2025.
45. R. A. Friesner, J. L. Banks, R. B. Murphy, T. A. Halgren, J. J. Klicic, D. T. Mainz, M. P. Repasky, E. H. Knoll, M. Shelley, J. K. Perry, D. E. Shaw, P. Francis and P. S. Shenkin, J. Med. Chem., 2004, 47, 1739–1749.
46. S. B. Berger, P. Harris, R. Nagilla, V. Kasparcova, S. Hoffman, B. Swift, L. Dare, M. Schaeffer, C. Capriotti, M. Ouellette, B. W. King, D. Wisnoski, J. Cox, M. Reilly, R. W. Marquis, J. Bertin and P. J. Gough, Cell Death Discovery, 2015, 1, 15009.
47. J. Li, R. Abel, K. Zhu, Y. Cao, S. Zhao and R. A. Friesner, Proteins, 2011, 79, 2794–2812.
48. W. Wang, J. M. Marinis, A. M. Beal, S. Savadkar, Y. Wu, M. Khan, P. S. Taunk, N. Wu, W. Su, J. Wu, A. Ahsan, E. Kurz, T. Chen, I. Yaboh, F. Li, J. Gutierrez, B. Diskin, M. Hundeyin, M. Reilly, J. D. Lich, P. A. Harris, M. K. Mahajan, J. H. Thorpe, P. Nassau, J. E. Mosley, J. Leinwand, J. A. Kochen Rossi, A. Mishra, B. Aykut, M. Glacken, A. Ochi, N. Verma, J. I. Kim, V. Vasudevaraja, D. Adeegbe, C. Almonte, E. Bagdatlioglu, D. J. Cohen, K. K. Wong, J. Bertin and G. Miller, Cancer Cell, 2018, 34, 757–774 e757.
49. P. C. Hawkins, A. G. Skillman, G. L. Warren, B. A. Ellingson and M. T. Stahl, J. Chem. Inf. Model., 2010, 50, 572–584.
50. W. Kabsch, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2010, 66, 125–132.
51. M. D. Winn, C. C. Ballard, K. D. Cowtan, E. J. Dodson, P. Emsley, P. R. Evans, R. M. Keegan, E. B. Krissinel, A. G. Leslie, A. McCoy, S. J. McNicholas, G. N. Murshudov, N. S. Pannu, E. A. Potterton, H. R. Powell, R. J. Read, A. Vagin and K. S. Wilson, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2011, 67, 235–242.
52. P. Emsley and K. Cowtan, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2004, 60, 2126–2132.
53. R. A. Nicholls, F. Long and G. N. Murshudov, Acta Crystallogr., Sect. D: Biol. Crystallogr., 2012, 68, 404–417.
54. R. S. K. Vijayan, M. M. Hamilton, D. E. Pfaffinger, F. G. Alvarez, N. J. Reyna, J. P. Bardenhagen, H. Shepard, C. Rodriguez, S. Goodwani, Y. Lightfoot, K. Maskos, S. Johannsson, G. Kempf, Q. A. Xu, L. Neumann, Y. Jiang, M. G. Do, P. Jones, R. T. Lewis, W. J. Ray and J. B. Cross, Allosteric targeting of RIPK1: discovery of novel inhibitors via parallel virtual screening and structure-guided optimization, 2025.

---