Discovery of an Orally Bioavailable Inhibitor of Defective in Cullin Neddylation 1 (DCN1)-mediated Cullin Neddylation

Abstract

We previously reported the discovery, validation, and structure-activity relationships of a series of piperidinyl ureas that potently inhibit the DCN1-UBE2M interaction. We demonstrated that compound 7 inhibits both the DCN1-UBE2M protein-protein interaction and DCN1-mediated cullin neddylation in biochemical assays and reduces levels of steady-state cullin neddylation in a squamous carcinoma cell line harboring DCN1 amplification. Although compound 7 exhibits good solubility and permeability, it is rapidly metabolized in microsomal models (CL_int = 170 mL/min/kg). This work lays out the discovery of an orally bioavailable analog, NAcM-OPT (67). Compound 67 retains the favorable biochemical and cellular activity of compound 7 but is significantly more stable both in vitro and in vivo. Compound 67 is orally bioavailable, well tolerated in mice, and currently used to study the effects of acute pharmacologic inhibition of the DCN1-UBE2M interaction on the NEDD8/CUL pathway.

TOC image

Introduction

Ubiquitin-like protein (UBL) modification pathways have emerged as important targets for oncology drug discovery. Successful examples include clinical proteasome inhibitors (e.g., Bortezomib or Carfilzomib)², E3 inhibitors³, and the NEDD8 E1 inhibitor (Pevonedistat)^4–6. Here we present the discovery of an orally bioavailable small molecule inhibitor of the UBL NEDD8 co-E3 enzyme, defective in cullin neddylation 1 (DCN1). DCN1 is also known as DCUN1D1, DCNL1, or SCCRO (squamous cell carcinoma-related oncogene); we use “DCN1” hereafter. DCN1 is an oncogenic driver gene contained within a recurrent 3q26.3 chromosomal amplification that is common in squamous cell carcinomas (SCC) of the lung, head, and neck.^7–13 DCN1 amplification is associated with malignant transformation, metastasis, and poorer patient survival rates.^7–13

DCN1 is part of a hierarchical E1-E2-E3 enzymatic cascade that covalently attaches the UBL NEDD8 to target proteins¹⁴. The first step of this cascade is the ATP-dependent formation of an E1-NEDD8 thioester intermediate. Next, the NEDD8 is transferred from the E1 to a catalytic Cys on the E2. E3s then join the E2-NEDD8 complex and recruit the target protein. This complex consequently catalyzes the formation of an isopeptide bond connecting NEDD8’s C-terminus to the ɣ-amino group of a lysine side-chain on the target protein. DCN1 is a co-E3 that stimulates the neddylation reaction.^15–16 Cullins, the best-characterized substrates of the NEDD8 pathway, are a family of ubiquitin E3 ligases that are activated by neddylation^17–19. Cullins serve as the catalytic domain for approximately 300 human multi-protein CUL-RING E3 ligases (CRLs) that regulate ubiquitin (UB) conjugation.^{18, 20}

The CRLs control diverse biological processes and their dysfunction is implicated in human diseases, including cancer.^21–23 Thus, drug discovery efforts targeting the CRLs continues to grow.^24–27 One approach to inhibiting CRL activity is to block cullin neddylation.^28–31 Pevonedistat (MLN4924) is the most well characterized inhibitor of the NEDD8 pathway and is currently in phase II clinical trials for the treatment of cancer.^{6, 32–33} Pevonedistat is an adenosine sulfamate analogue that covalently inhibits the NEDD8 E1, thus inhibiting all neddylation.³⁴ The success of Pevonedistat has stimulated discovery of other inhibitors of the neddylation pathway. Virtual screening has yielded both reversible and covalent small molecule inhibitors of the NEDD8 E1.^35–38 Several rhodium(III) complexes also inhibit the NEDD8 E1, offering an alternative to small molecules.^39–40 Recently, an inhibitor of the COP9 signalosome, responsible for de-neddylation of the CRLs, has been reported and also displays anti-tumor activity.⁴¹ We, and others, have recently reported the discovery of inhibitors of DCN1.^42–44

While complete ablation of neddylation appears to be clinically efficacious,²⁹ it is also toxic. In an effort to maintain efficacy and reduce toxicity, we sought more specific ways to interrogate the neddylation pathway.⁴² DCN1 plays a role in tumor formation, maintenance, and regulation of the neddylation pathway.^{7–17, 45–46} Therefore, we sought to develop DCN1 inhibitors that may offer advantages and/or alternative specificities compared to pan-neddylation or de-neddylation inhibitors.^{1, 42–43}

We recently identified a class of piperidinyl ureas that inhibit the DCN1-UBE2M protein-protein interaction and thereby reduce steady-state neddylation of specific cullins.⁴² Subsequent iterative structure-enabled optimization of our high throughput screening hit (compound 2, Table 1, IC₅₀ = 7.6 μM) yielded compound 7 (IC₅₀ = 0.060 μM)¹ that was 100-fold more potent than our initial hit.^{1, 42} Our optimization targeting five sub-pockets within the DCN1-UBE2M binding site is reported in the companion manuscript.¹ The first three sub-pockets are the Ile, N-acetyl, and Leu, named for the UBE2M amino acid side chains that occupy these pockets. The last two sub-pockets are the urea and hinge pockets, named for the moieties from the chemotype that occupy them (Figure 1).⁴²

Table 1. Rapid metabolism of early lead DCN1-UBE2M inhibitors.

Values for the half-life and intrinsic clearance were determined at a concentration of 0.8 μM using mouse microsomes and are represented as the means plus or minus standard deviation from three replicates. Compound preparation and potency was previously reported.¹

No	R1	R2	t(1/2) (hr)	CLint (mL/min/kg)	No	R1	R2	t(1/2) (hr)	CLint (mL/min/kg)
1			< 0.1	> 500	6			0.150 ± 0.019	361 ± 45.1
2			0.591 ± 0.0525	91.4 ± 8.12	7			0.327 ± 0.010	165 ± 25.9
3			0.137 ± 0.021	394 ± 55.2	8			0.245 ± 0.014	220 ± 12.8
4			0.170 ± 0.016	318 ± 29.6	9			0.124 ± 0.006	434 ± 21.1
5			< 0.1	> 500	10			0.271 ± 0.014	199 ± 9.99

Figure 1.

Top: Chemical structure of 1. Bottom: Overlay of 1(teal): DCN1 (electrostatic potential surface) (PDB: 6BG5)¹ and UBE2M^NAc (magenta): DCN1 (PDB: 3TDU) X-ray co-crystal structures, highlighting regions targeted for optimization¹.

Examination of several X-ray co-crystal structures led us to identify four key drivers of potency. This finding is further exemplified by the 1:DCN1 X-ray co-structure shown in Figure 1 (PDB: 6BG5)¹. The four potency drivers are: (1) the 4-aminopiperidinyl urea core that positions the hydrophobic alky and aryl ends of the molecule into the Ile and hinge pockets; (2) a key H-bond between the urea aryl N-H and the backbone amide of Gln114 in the urea pocket; (3) the induced hinge pocket, which affords tight steric packing of hydrophobic residues around the m-CF₃ phenyl ring; and (4) the hydrophobic interactions of the methylbenzodioxole urea substituent within the Leu pocket (Figure 1).⁴² Compound 7 biochemically inhibits the DCN1-UBE2M protein-protein interaction and cullin neddylation.¹ This compound also suppresses steady-state levels of cullin neddylation in HCC95 cells, a squamous carcinoma (SCC) cell line highly expressing DCN1.¹ However, compound 7 is rapidly metabolized in microsomal and murine models. This report describes our continuing optimization efforts aimed at improving the clearance of compound 7. Our optimization strategy focused on 1) understanding the physical drivers for bioavailability (SPR); and 2) evaluating metabolic and toxic liabilities associated with the scaffold. This strategy has yielded a potent, orally bioavailable, and significantly more stable chemical probe, NAcM-OPT (67)⁴².

Results

Given the clear effect of compound 7 in cells¹, we examined whether this compound possessed appropriate absorption, distribution, metabolism, excretion, and toxicology (ADME/T) characteristics to block DCN1 activity using in vivo models. Compound 7 was moderately soluble (57 ± 1.4 μM) and highly cell permeable (1400 ± 39 × 10⁻⁶ cm/s)¹. However, compound 7 was unstable in mouse microsomes (CL_int = 170 mL/min/kg, Table 1). We tested several structurally similar analogs and found that they were also rapidly metabolized (CL_int > 150 mL/min/kg, Table 1). The most stable analog was the initial screening hit 2 (CL_int = 90 mL/min/kg). Therefore, it seems as though the structural modifications made to improve potency also reduced metabolic stability. In particular, the benzyl urea substitution that was critical for potency and specificity due to its interactions within the Leu pocket¹ (Figure 1) appears to increase rates of oxidative metabolism.

In order to ensure that the microsomal models accurately predict in vivo clearance and reveal any systemic toxicology exhibited by the series, we tested compound 7 in a single-dose intravenous (IV) pharmacokinetic (PK) study using three doses: 0.5, 1.0, and 1.5 mg/kg (Figure 2A). We did not observe any adverse reactions or compound-related side effects within 48 h after dosing (Supplementary Table 1). Compound 7 was rapidly cleared (t_1/2 = 0.14 h, Clplasma = 42 L/h/kg) at all doses, suggesting that the microsomal models correctly predict clearance of compound 7 in vivo. Therefore, we sought to mitigate the scaffold’s inherent metabolic liabilities using microsomal models as a predictor of in vivo clearance.

Figure 2. Characterization of compound metabolism.

A) Compound 7 plasma concentrations over 24 h following IV administration 0.5, 1.0, and 1.5 mg/kg doses. Plasma concentrations from time points taken after 2 h were below the limit of quantitation and are thus excluded from the graph. B) Summary of metabolite identifications studies using compound 5. For details, see Supplementary Table 2.

We utilzed a recursive process of analog design, synthesis, testing, and design refinement to optimize the series. We assessed biochemical potency using our previously reported TR-FRET assay,⁴² which measures the binding of full-length DCN1 to a constrained peptide corresponding to N-terminally acetylated UBE2M^{15–16, 42}. We used structurally directed modifications, as well as more empircal investigation of the moeity fitting into the induced fit hinge pocket to probe the steric and electronic requirements of binding^47–51. We assessed the metabolic stability of all analogs exhibiting reasonable potency in mouse microsomes. Analogs that retained potency and posessed significantly improved microsomal stability were profiled in murine models to determine in vivo toxicity, clearance, and bioavailability.

To further guide the design of metabolically stable analogs, we conducted metabolite identification (Met-ID) studies using a close structural analog of compound 7, compound 5 (Figure 2B). This study identified N-dealkylation (11) of the piperidine ring in the Ile targeting substituent and oxidation of the aromatic ring of the urea (12), targeting the hinge region, as the primary metabolic transformations (Figure 2B; Supplementary Table 2). Time-course metabolite formation studies suggest that these modifications take place sequentially; oxidation of the aromatic ring occurs first, followed by N-dealkylation (Supplementary Table 2). Although we were unable to identify the specific oxidized carbon on the aromatic ring, we hypothesize that the carbons ortho- or para- to the amino group are the most likely the sites of oxidation. This hypothesis is based on the electron donating nature of the amino substituent and on previous reports^52–53. Therefore, we targeted optimization of the Ile and hinge moieties. We used both a rational, structure-based, design and an empirical parallel chemistry approach to understand the key structural elements that drive the structure-activity relationship (SAR) and structure-property relationship (SPR) that define metabolic liabilities.

Following these findings, we sought to maintain potency and improve metabolic stability by leaving invariant the benzyl group (targeting the Leu pocket) of compound 7 and varying the alkyl side chain (targeting the Ile pocket) and aniline (targeting the hinge pocket) (Figure 1). We devised two synthetic routes that permitted late stage introduction of either substituent (Scheme 1; routes i, ii). Both routes were amenable to parallel synthesis.

Scheme 1. General synthetic routes for generating analogs.

(a) potassium carbonate, 2-iodopentane, acetonitrile, 80 °C, 16 h, 30-48% (b) benzylamine, sodium triacetoxyborohydride, acetic acid, dichloromethane, rt, 16 h, 90-100%; (c) R¹-isocyanate, N,N-diisopropylethylamine, dichloromethane, rt, 2 h, 40-80%; (d) R¹-amine, 1,1′-carbonyldiimidazole, tetrahydrofuran, 16 h, 30-55%; (e) 3-Fluoro-5-(trifluoromethyl)phenyl isocyanate, N,N-diisopropylethylamine, dichloromethane, rt, 2 h, 80-99%; (f) trifluoroacetic acid, dichloromethane, 0 °C - rt, 16 h, 95%; (g) R² (aldehyde or ketone), dibutyltin (IV) chloride, phenylsilane, tetrahydrofuran, rt, 16 h, 40-75%.

Synthesis of aniline analogs began by alkylation of 4-piperidone (13) with 2-iodopentane (Scheme 1; routes i)⁴² to generate ketone 14. Reductive amination of 14 with benzylamine furnished the secondary amine 15⁴². 15 was then either coupled directly with isocyanates or with anilines using cabonyldiimidazole (CDI) as a coupling agent (21 – 46, Tables 2 & 3)^54–57. Given the wider availability of anilines, we used the CDI-mediated reaction for our parallel chemistry. For this reaction, it was critical to allow CDI to react with the aniline prior to addition of the secondary amine 15. The optimized procedure required no work-up and direct purification by HPLC chromatography afforded moderate yields (30-55%) of the desired products. The scalability and ease of this synthetic route allowed us to prepare roughly 140 analogs variant in the aniline substituent.

Table 2. Optimization of the phenyl urea occupying the hinge pocket.

IC50 values were generated using our TR-FRET binding assay and are represented as the mean of three replicates with errors reported as the 95% confidence interval.

No	R4	IC50 (μM) (95% Clf	t(1/2) (hr)	CLint’ (mL/min/kg)	No	R4	IC50 (μM) (95% Cl)a	t(1/2) (hr)	CLint’ (mL/min/kg)
21		1.69 (0.640-4.46)	≤0.10	≥ 500	26		0.068 (0.049-0.095)	0.386 ± 0.021	148 ± 7.63
22		0.304 (0.239-0.386)	≥0.10	≥500	27		0.054 (0.044-0.066)	0.498 ± 0.013	115 ± 2.78
23		0.353 (0.288-0.434)	0.223 ± 0.021	256 ± 22.4	28		0.356 (0.241-0.527)	0.160 ± 0.015	356 ± 31.9
24		0.314 (0.229-0.432)	0.455 ± 0.012	125 ± 3.17	29		0.654 (0.445-0.961)	0.177 ± 0.003	323 ± 4.46
25		0.112 (0.082-0.152)	0.368 ± 0.027	156 ± 10.7	30		1.72 (1.23-2.41)	ND	ND

^a95% confidence intervals were generated based on non-linear regression using the R drc package with the four-parameter log-logistic function (LL2.4)⁵⁹. Values for the half-life and intrinsic clearance were determined at a concentration of 0.8 μM using mouse microsomes and are represented as the means plus or minus standard deviation from three replicates. ND = data not determined.

Table 3. Second round of optimization of the phenyl urea occupying the hinge pocket.

IC50 values were generated using our TR-FRET binding assay and are represented as the mean of three replicates with errors reported as the 95% confidence interval.

No	R4	IC50 (μM) (95% Cl)a	t(1/2) (hr)	CLint’ (mL/min/kg)	No	R4	IC50 (μM) (95% Cl)a	t(1/2) (hr)	CLint’ (mL/min/kg)
31		0.082 (0.060-0.112)	0.379 ± 0.027	151 ± 10.2	39		> 34	0.249 ± 0.015	230 ± 13.1
32		0.817 (0.666-1.00)	0.189 ± 0.013	302 ± 15.2	40		9.43 (6.16-14.5)	ND	ND
33		0.241 (0.212-0.273)	0.150 ± 0.013	381 ± 31.9	41		13.3 (11.7-15.1)	> 4	< 13.5
34		0.260 (0.202-0.333)	0.932 ± 0.056	61.3 ± 3.49	42		0.626 (0.338-1.16)	0.239 ± 0.025	240 ± 24.0
35		0.290 (0.240-0.350)	0.393 ± 0.027	145 ± 9.51	43		0.077 (0.044-0.133)	^ 0.10	≥500
36		0.058 (0.045-0.075)	0.196 ± 0.016	291 ± 22.4	44		0.990 (0.718-1.37)	< 0.10	> 500
37		0.481 (0.390-0.594)	<0.10	>500	45		0.018 (0.014-0.025)	<0.10	> 500
38		7.44 (6.28-8.81)	0.808 ± 0.062	66.9 ± 5.15	46		> 20	ND	ND

^a95% confidence intervals were generated based on non-linear regression using the R drc package with the four-parameter log-logistic function (LL2.4)⁵⁹. Values for the half-life and intrinsic clearance were determined at a concentration of 0.8 μM using mouse microsomes and are represented as the means plus or minus standard deviation from three replicates. ND = data not determined.

Synthesis of analogs targeting variation of the alkyl sidechain started with reductive amination of 1-N-boc-4-piperidone (16) with benzylamine to give 17. A limited range of anilines with electron-withdrawing substitutions provided appropriate activity and the potential for increased stability. Thus, the secondary amine 17 was coupled with 3-fluoro-5-(trifluoromethyl)phenyl isocyanate to give 18. After BOC de-protection, Lewis acid-mediated reductive amination⁵⁸ of 18 with either ketones or aldehydes, afforded the targeted analogs (47 – 61, Table 4) in moderate to good yields (40-75%). We used this route to prepare roughly 45 analogs variant in the alkyl side chain (Supplementary Dataset 1). Compounds prepared by parallel synthesis were characterized for purity (≥ 95% as characterized by UV/ELSD) and identity using UPLC-LRMS (Supplementary Dataset 1). Compounds that proved critical to our SAR analysis were further characterized using ¹H NMR and HRMS (Supplementary Information).

Table 4. Optimization of the alkyl piperidine.

IC50 values were generated using our TR-FRET binding assay and are represented as the mean of three replicates with errors reported as the 95% confidence interval.

No	R4	IC50 (μM) (95% cna	No	R4	IC50 (μM) (95% Cl)a	No	R4	IC50 (μM) (95% Cl)a
47		0.256 (0.095-0.685)	52		>28	57		>25
48		1.79 (1.36-2.36)	53		4.49 (3.86-5.21)	58		4.27 (2.27-8.04)
49		3.21 (2.12-4.85)	54		> 28	59		> 24
50		>24	55		>25	60		> 22
51		>24	56		>28	61		> 21

^a95% confidence intervals were generated based on non-linear regression using the R drc package with the four-parameter log-logistic function (LL2.4)⁵⁹.

Our studies revealed that the n-butyl alkyl sidechain was as potent but more metabolically stable than the 2-pentyl substitution (Table 5). Thus, we synthesized a final set of analogs starting from commercially available n-butyl piperidinone 19 (Scheme 1; route iii)⁴². This two-step procedure is readily scalable and has been used to prepare more than five grams of our top compound 67 with > 98% purity as characterized by UV/ELSD and NMR analysis⁴².

Table 5. Final optimization of n-butylpiperidines.

IC50 values were generated using our TR-FRET binding assay and are represented as the mean of three replicates with errors reported as the 95% confidence interval.

No	R4	IC50 (μM) (95% Cl)a	t(1/2) (hr)	CLint’ (mL/min/kg)	No	R4	IC50 (μM) (95% Cl)a	t(1/2) (hr)	CLint’ (mL/min/kg)
62		3.57 (2.87-4.43)	<0.10	>500	68		4.27 (2.95-6.16)	<0.10	>500
63		0.358 (0.248-0.515)	≥0.10	≥500	69		0.583 (0.437-0.777)	≥0.10	≥500
64		1.30 (0.947-1.78)	0.178 ± 0.002	305 ± 4.02	70		0.612 (0.266-1.41)	1.94 ± 0.189	27.9 ± 2.71
65		1.85 (1.48-2.31)	ND	ND	47		0.256 (0.095-0.685)	2.43 ± 0.531	22.3 ± 4.88
66		4.25 (2.72-6.64)	5.155 ± 0.015	349 ± 22.5	71		0.325 (0.214-0.496)	0.286 ± 0.022	189 ± 14.5
67		0.045 (0.013-0.153)	2.11 ± 0.336	25.6 ± 4.08	72		1.09 (0.925-1.28)	1.40 ± 0.059	38.7 ± 1.63

^a95% confidence intervals were generated based on non-linear regression using the R drc package with the four-parameter log-logistic function (LL2.4)⁵⁹. Values for the half-life and intrinsic clearance were determined at a concentration of 0.8 μM using mouse microsomes and are represented as the means plus or minus standard deviation from three replicates. ND = data not determined.

First, we explored the SAR/SPR of the aniline urea substituent targeting the hinge pocket. Our previous SAR studies and X-ray co-structures revealed that to permit deeper compound binding, the hinge pocket undergoes a structural rearrangement to produce a hydrophobic cavity composed of alkyl and aromatic residues (Ile86, Phe89, Val102, Ile105, Ala106, Phe109, Ala111, Phe117, Phe122, Phe164)^{1, 42}. Throughout all of our investigations, an aromatic ring with a small hydrophobic substituent proved optimal for filling this sub-pocket. Compound 7 (IC₅₀ = 0.060 μM), the most potent analog, utilizes a meta-trifluoromethylphenyl aniline that fits perfectly into the hinge sub-pocket¹. Previous work suggests that although meta- and para-substitutions were tolerated, modestly increased steric bulk or ortho-substitution eliminate binding¹. Thus, we used the parallel chemistry route to rapidly produce diverse analogs while paying particular attention to roughly isosteric para- and meta-substituted aniline rings (Table 2).

Initial exploration of mono-substituted phenyl ureas (Table 2, 21 – 30) revealed that only the para-bromo (26) and para-iodo (27) analogs were as potent as compound 7, with half maximal inhibitory concentrations (IC₅₀’s) of roughly 0.060 μM. All of the mono-substituted phenyl ureas were rapidly cleared in the microsomal assay (CL_int > 120 mL/min/kg). Halogen substituents were generally more potent than the corresponding alkyl substituents, suggesting that while hydrophobicity is necessary, the presence of an electron-withdrawing group, and potentially a hydrogen bond acceptor, is important for potency (25 – 27). This result suggested that we pursue the common strategy of decreasing the rate of metabolic oxidation of aromatic rings by decreasing their electron density. Indeed, such analogs were slightly more stable in microsomal models than compound 7 (CL_int = 170 mL/min/kg vs. between 120 and 160 mL/min/kg, Table 2).

In concordance with previous studies,¹ larger substitutions such as i-Pr (23, 29) and t-Bu (30) significantly reduced analog potency (Table 2, Supplementary Dataset 1). Similar to the para- series, the meta-substituted compounds’ potency appeared to improve with hydrophobic electron-withdrawing substituents. The meta-trifluoromethylphenyl (7, IC₅₀ = 0.060 μM) was more potent than the meta-chlorophenyl (22, IC₅₀ = 0.300 μM), a result that stands in stark contrast to the para-substituted analogs (p-CF₃ (24) IC₅₀ = 0.300 μM; p-Cl (25), IC₅₀ = 0.100 μM). The meta-i-Pr analog (23, IC₅₀ = 0.350 μM) was more potent than the para-i-Pr analog (29, IC₅₀ = 0.650 μM). The meta-fluoro analog (21) was less potent (IC₅₀ = 1.70 μM), likely due to its inability to fill the available pocket. Together, these results suggest that the meta-position affords slightly more room for increased steric bulk compared to the para-position. The results also highlight that subtle variations in size and electronic properties can significantly affect potency when there is induced fit in the binding site. The para-iodo analog (27, IC₅₀ = 0.050 μM) was the only compound that maintained the potency of compound 7 and increased stability (CL_int = 120 mL/min/kg). However, the observed increase in stability (CL_int = 170 mL/min/kg vs. 120 mL/min/kg) was not sufficient to render the compound suitable for in vivo studies. We next investigated the di-substituted phenyl ureas with the intent to identify a substituent with improved potency and stability.

The targeted di-substituted phenyl ureas (Table 3) were designed to solidify the trends observed for the mono-substituted analogs. That is, the analogs should incorporate small hydrophobic and electron-withdrawing substituents at either the ortho- or para-position. These substitutions appear to electronically and sterically block oxidation of the aromatic ring without negatively affecting potency. We first tested meta- and para-di-substituted analogs (31 – 35). These analogs retained moderate potency with the meta-nitro, para-chloro analog (31) exhibiting an IC₅₀ value below 0.100 μM. However, due to the potential toxicology associated with the nitro substituent⁶⁰, this compound was deprioritized. These compounds were also subject to rapid metabolism. The meta-trifluoromethyl, para-nitrile derivative 34 (IC₅₀ = 0.260 μM) was the only compound that was significantly more stable in the microsomal model than compound 7 (CL_int = 60 mL/min/kg vs. 120 mL/min/kg). Among the phenyl ureas with bis-meta substitution patterns, only the meta-trifluoromethyl, meta-fluoro analog 36 was equipotent to compound 7 (IC₅₀ = 0.060 μM). The corresponding meta-chloro, meta-fluoro analog 37 was significantly less potent, further emphasizing the tightly constrained nature of the SAR. All of the bis-meta analogs were rapidly metabolized.

We next examined ortho-, meta-di-substituted phenyl ureas. We observed reduced potency even in the presence of small ortho-substituents (Table 3, 38 – 41), which is consistent with previous SAR data¹. This series led to the emergence of the pyridine analog 41 - our first compound to exhibit good metabolic stability (CL_int < 13 mL/min/kg). Unfortunately, this compound, like other pyridine-containing analogs, exhibited significantly reduced potency (IC₅₀ = 13 μM, Supplementary Dataset 1)^{1, 42}. Examination of fused bicyclic aryl ureas revealed a general trend that the hydrophobic carbocyclic compounds were more potent than their heterocyclic counterparts (42 – 46). The 2-napthyl analog (46) afforded our first significant gain in potency (IC₅₀ = 0.020 μM). However, all of the fused bicyclic analogs were rapidly metabolized in the microsomal model (Table 3). This result illustrates a common phenomenon in medicinal chemistry: as one parameter is optimized, another in compromised. Based on these analogs, we hypothesized that reducing the electron density of the aniline ring would mitigate the oxidation of the aromatic ring, but the scope of the SAR would be narrow.

In parallel to our investigations of aniline analogs, we sought to address N-dealkylation of the alkyl-piperidine by exploring various substitutions occupying the Ile pocket. Initially, we fixed the aniline substituent as the meta-CF₃, meta-F, based on inhibitory potency. Using this base, we screened a range of analogs replacing the 2-pentyl piperidine substitution that targets the Ile sub-pocket (Figure 1). Our previous studies closely explored alkyl substituents at this position and revealed that an n-butyl group was preferred and branched alkyl chains improved potency¹. We have also previously prepared simple benzyl and aryl analogs. To expand our SAR, we explored several other alkyl, aryl, and heteroaryl substituents (Table 4, 47 – 61). Unfortunately, with the exception of the n-butyl analog 47 (IC₅₀ = 0.260 μM), all of the substituents led to significantly reduced potency (Table 4, Supplementary Dataset 1). Thus, we selected only compound 47 for further evaluation in the microsomal model.

Microsomal modeling revealed that compound 47 was significantly more stable than compound 7 (CL_int = 22 mL/min/kg vs. 170 mL/min/kg), with only a five-fold reduction in biochemical potency (IC₅₀ = 0.260 μM vs. 0.060 μM). Comparison of 47 to the 2-pentyl derivative 36 revealed that replacing the 2-pentyl substitution with an n-butyl substitution markedly increased stability (CL_int = 22 mL/min/kg vs. 290 mL/min/kg). Assuming this was a general trend, we examined other aryl ureas containing the n-butylpiperidine (Table 5). The di-substituted analogs were more stable than the corresponding mono-substituted aryl urea analogs. The di-substituted para-fluoro, meta-trifluoromethyl (70) or meta-fluoro, meta-trifluoromethyl (47) were more stable than the ortho-fluoro, meta-trifluoromethyl (69). This result supports our hypothesis that the site of oxidation is most likely the carbon para- to the amino substituent. Compound 67, which contains a meta-chloro, para-chloro phenyl urea afforded the best balance of potency and stability. Compound 67 (IC₅₀ = 0.050 μM) retains the biochemical potency of our optimized hit compound 7 but is significantly more stable in microsomal models with a calculated intrinsic clearance of 25 mL/min/kg.

In order to determine whether we had altered the landscape of primary metabolites or simply decreased their rate of formation, we conducted metabolite identification studies with 67 (Supplementary Table 3). These studies revealed that replacing the meta-trifluoromethyl group of the aniline with a meta-, para-dichloro group effectively decreased the rate of microsomal oxidation. However, oxidation of the aniline ring remained a minor metabolic pathway. N-dealkylation was now the predominant metabolism event. Additionally, we observed oxidation of the benzyl urea substituent targeting the Leu pocket. These results indicate that further optimization may yield even more stable compounds. We believed we had successfully increased simulated oxidative stability without negatively affecting potency, and therefore sought to characterize the toxicology and pharmacokinetic characteristics of compound 67 in murine models.

We first conducted a low-dose IV PK experiment to understand the gross toxicity and clearance of compound 67 (Figure 3A). Similar to compound 7, 67 failed to induce any measureable adverse reactions during the 48 h monitoring period following IV administration (Supplementary Table 4, 5). The observed C_max of compound 7 (Figure 2A) and 67 (Figure 3A) shows a non-linear increase from 0.5 to 1.5 mg/kg. Both compounds exhibit similar and large volumes of distribution (87.3 – 176 L/kg). The IV half-life of 67 is 3-4 hours, which is significantly longer than the half-life of compound 7 (t_1/2 = 0.14-0.2 h). The increased metabolic stability of compound 67 correlates with the results from the microsomal model, which showed that this compound has a six-fold lower clearance in mouse microsomes (Supplementary Table 6). Due to the comparatively shorter half-life and lower C_max, the AUC_last of compound 7 is much smaller than that of compound 67. Therefore, compound 67 is the preferred early lead for preliminary in vivo studies. Based on these results, we determined that it was unethical to conduct an oral dosing study for compound 7. However, we did conduct an oral PK experiment with compound 67 (Figure 3B).

Figure 3. Preliminary pharmacokinetic profiling of compound 67.

A) Time course of the drug plasma concentration over 24 h resulting from single dose intravenous administration of 67 at doses of 0.5, 1.0, and 1.5 mg/kg. B) Time course of the drug plasma concentration over 24 h resulting from single oral doses of 67 at 50, 100, and 200 mg/kg. C) Repeated dose-exposure studies to determine toxicological responses. Subjects were dosed orally with 200 mg/kg either once a day (QD) or twice a day (BID) and drug plasma concentration was measured 1 h after drug administration. D) Table comparing the in vitro and in vivo clearance data. Results support our hypothesis that the microsomal models predict in vivo clearance. Additional PK parameters are in Supplementary Table 6.

When dosed orally as a suspension in EtOH/PG/PEG/PBS (10/10/40/39; v/v/v/v; pH = 7.4) and 1% (w/v) 2-hydroxy-β-cyclodextrin (MW~1396 Da), compound 67 displayed a half-life of 3-5 h. This observation concurs with the IV half-life of 3-4 h and suggests that 1) the total exposure is driven by clearance and 2) levels of absorption are optimal. The observed C_max and AUC_last show a dose-dependent increase from 1.5 to 3 μM and 3.4 to 17.1 μM*h, respectively. The apparent volume of distribution and clearance are similar to those of IV. Based on these data, compound 67 has an oral bioavailability of 88% (calculated using IV dose at 1.5 mg/kg and PO dose at 50 mg/kg). This apparent bioavailability (Fin vivo) is close to what we predicted from in vitro clearance (F_in vitro = 78%), calculated by 1 - CL/Q_h⁶¹. Compound 67 did not cause a significant adverse reaction at any dose tested during the 48 h monitoring period following oral administration (Supplementary Table 7, 8).

In order to enable efficacy modeling in mice, we investigated the toxicology and pharmacokinetics of compound 67 during repeat-dose experiments over a period of 14 days using either once (QD) or twice (BID) daily schedule with a fixed oral dose of 200 mg/kg. Throughout the 14-day dosing period, no adverse reactions or compound-related side effects were observed (Supplementary Table 9, 10). Dosed animals in both groups experienced minor weight loss (3-5%) during the first 3 days, but resumed normal growth within 5 days (Supplementary Figure 1). In order to determine if there were significant changes in metabolism and clearance during the study period, drug plasma concentrations were monitored one hour post administration after each oral dose (Figure 3C). C_1h was 2.4 ± 0.93 μM (QD) and 3.3 ± 1.5 (BID), and was consistent with earlier data from our single oral dose study (C_1h = 3 μM at 200 mg/kg). Therefore, compound 67 is well tolerated, possesses good oral bioavailability, and is suitable for use in animal efficacy models.

The details of the biochemical selectivity and cellular activity of compound 67 (NAcM-OPT), along with an irreversible inhibitor (NAcM-COV), and a structurally related negative control (NAcM-NEG), have recently been disclosed.⁴² Briefly, compound 67 maintains the high level of selectivity for inhibition of DCN1/2 over the highly homologous remaining human DCN isoforms.⁴² Compound 67 effectively disrupts the DCN1-UBE2M interaction and UBE2M interactome in cells.⁴² In HCC95, a squamous carcinoma cell line highly expressing DCN1, either treatment with compound 67 or knockdown of DCN1 with shRNA selectively reduces steady-state levels of neddylated CUL1 and CUL3 while CUL2, CUL4A, and CUL5 are relatively unaffected.⁴² Compound 67 inhibits steady-state levels of neddylated CUL1 and CUL3 in a dose-dependent manner (Figure 4A). Consistent with our previous studies,⁴² treatment of HCC95 cells with compound 67 does not elicit gross substrate stabilization (Figure 4B). However, at longer exposures, which oversaturate MLN4924 treated protein levels making direct comparison no longer possible, compounds 67 elicits a dose-dependent increase in the known CUL1/CUL3 substrates NRF2 and p27 (Figure 4B) but not other known CUL1/CUL3 substrates Cylin E or p21 (Supplementary Figure 2).

Figure 4.

Western blot for inhibition of cellular neddylation and increased levels of substrates at 48 hours by DMSO, MLN4924 (positive control, 1 μM, single dose at T = 0), and compound 67 (indicated concentration, dosing at T = 0, and T = 24 hr). A) Dose-dependent inhibition of steady-state levels of CUL1 and CUL3 neddylation. B) Dose-dependent effects of 67 on steady-state levels of NRF2 and p27.

Discussion and Conclusions

In the present study, we employed both a rational, structure-based, design as well as an empirical chemistry approach to uncover the key structural elements critical for binding to DCN1 and driving compound metabolism. Our studies yielded compound 67, which is equipotent to earlier leads like compound 7 but is significantly more bioavailable. By developing short, scalable, synthetic routes, we rapidly prepared over 200 analogs (Supplementary Dataset 1) to select compounds with better PK profiles. A combinatory approach of microsomal modeling and in vivo PK allowed us to confidently conclude that the microsomal models correctly predict in vivo clearance. Metabolite identification studies focused our attention on the most rapidly metabolized moieties of the inhibitor series: the aniline and alkyl side chain of the piperidine. The structure-based design incorporated our previously described X-ray co-structures and SAR data^{1, 42} and allowed us to narrow our focus to the substituents that were most likely to yield potent molecules. In total, 38 of the newly synthesized analogs possessed potency values less than or equal to 360 nM in our TR-FRET binding assay (Supplementary Dataset 1). In parallel to the rational design approach, we employed an empirical approach to discover unexpectedly potent analogs, such as the bicyclic compound 45, which appears to bypass the stringent steric restrictions in the hinge sub-pocket, as suggested by our previous data. In the face of a known induced fit binding event, we believe it is imperative to keep testing the limits of the design model. The ease and scalability of the synthetic routes allowed us to produce ample material for both the PK and extended dosing studies. The synthetic routes will also allow us to produce sufficient quantities for future in vivo studies.

We observed two general SAR/SPR trends. First, the incorporation of small hydrophobic and electronegative substituents at either or both of the meta- and para-positions of the aniline ring electronically and sterically blocked oxidation of the aromatic ring without negatively affecting potency. Second, the SAR of the alkyl piperidine substituent was highly restricted only tolerating small hydrophobic substituents.¹ Subsequent metabolite identification studies using the optimized compound 67 revealed that although incorporating the di-chlorophenyl ring reduced the oxidation of the aniline substituent. N-dealkylation of the 4-aminopiperidine and oxidation of the benzyl urea substituent targeting the Leu pocket were still prevalent. Future synthetic efforts will focus on these regions as well as on accessing the currently unoccupied N-acetyl pocket, which likely affords the best opportunity for potency gains.¹⁶

We are currently using compound 67 and others to dissect the effects of acute pharmacologic inhibition of the DCN1-UBE2M interaction on the NEDD8/CUL pathway in both cellular and in vivo models. The data suggest that inhibiting this NEDD8 E2-E3 interaction has significantly different biological consequences from inhibition of the upstream NEDD8 E1 by Pevonedistat or other E1 inhibitors. Pevonedistat acts by irreversibly inhibiting the E1 enzyme that initiates the NEDD8 tri-enzyme cascade, completely blocking all neddylation. Compound 67 selectively reduces the steady-state neddylation of specific cullins. The most striking phenotypic difference is that unlike E1 inhibition, compound 67 does not grossly stabilize canonical Cullin-RING ligase substrates.⁴² However, compound 67 does elicit a minor and selective increase in the levels of two known CRL substrates, NRF2 and p27. Compound 67 also inhibits the anchorage-independent growth of HCC95 cells, a phenotypic consequence associated with DCN1-amplification and hallmark of cancer in this squamous cell carcinoma line.⁴² Future experiments, outside the scope of this work, will focus on understanding the mechanisms driving the observed selectivity and phenotypic consequences associated with acute DCN inhibition in cells and murine models. These results suggest that selective regulation of the neddylation pathway still influences tumor progression, at least in cells harboring DCN1 amplification. However, we are clear to point out that selective inhibition of the DCN1-UBE2M interaction is not grossly toxic to the proliferation of monolayer cultures. Given the high levels of DCN1 amplification exhibited in many human cancers,^{7, 11, 13} selective inhibition of DCN1-mediated neddylation could provide a more targeted therapy relative to the complete ablation of neddylation afforded by Pevonedistat or other NEDD8 E1 inhibitors. Recent reports of a second class of DCN1 inhibitors demonstrates a similar spectrum of activity.^43–44

In conclusion, the development of compound 67 permits a more selective and acute pharmacologic regulation of this highly complex and dynamic pathway that is already helping to discover previously unappreciated biological roles of DCN1-mediated neddylation,^{42, 62} and hopefully will provide further insight into tumor biology in areas with significant unmet clinical need.

Experimental Section

TR-FRET Assay^{1, 42}

The TR-FRET assay was carried out in black 384-well microtiter plates at a final volume of 20 μL per well. To screen library compounds, the assay cocktail was prepared as a mixture of 50 nM Biotin-DCN1, 20 nM Ac-UBE2M12-AlexaFluor488, 2.5 nM Tb-Streptavidin (ThermoFisher) in assay buffer (25 mM HEPES, 100 mM NaCl, 0.1% Triton X-100, 0.5 mM DTT, pH 7.5). The assay cocktail was then incubated for 1 h at room temperature and distributed using a WellMate instrument (Matrix). Compounds to be screened were added to assay plates from DMSO stock solutions by pin transfer using 50SS pins (V&P Scientific). The assay mixture was incubated for 1 h at room temperature prior to measuring the TR-FRET signal with a PHERAstar FS plate reader (BMG Labtech) equipped with excitation modules at 337 nm and emissions at 490 and 520 nm. We set the integration start to 100 μs and the integration time to 200 μs. The number of flashes was set to 100. The ratio of 520:490 was used as TR-FRET signal in calculations. Assay endpoints were normalized from 0% (DMSO only) to 100% inhibition (unlabeled competitor peptide) for hit selection and curve fitting. All compounds were tested in triplicate or more.

Liver Microsomal Stability

We mixed 0.633 mL of mouse liver microsome (20 mg/mL, female CD9 mice, Fisher Scientific, #NC9567486) with 0.051 ml of 0.5 M EDTA solution and 19.316 ml potassium phosphate buffer (0.1 M, pH 7.4, 37°C) to produce 20 ml of liver microsome solution. One part 10 mM DMSO compound stock was mixed with 4 parts acetonitrile to make 2 mM diluted compound stock in DMSO and acetonitrile. We added 29.1 μL diluted compound stock to 2.3 mL liver microsomal solution and ran the mixture through a vortex to make final microsomal solution with 0.8 μM compound. We dispensed 180 μL of the mixed solutions into a 96-well storage plate (pION Inc., MA, #110323). Five time points (0-4 hr) were taken for each microsome assay. For the 0 h time point, 450 μL pre-cooled (4 °C) internal standard (5 μM caffeine in methanol) was added to the deep wells before the reaction started. For other time points, 1.25 ml of microsome assay solution A (Fisher Scientific, #NC9255727) was combined with 0.25 ml of solution B (Fisher Scientific, #NC9016235) in 3.5 ml of potassium phosphate buffer (0.1 M, pH 7.4). We added 45 μL of this A+B solution to each of the 96 wells in the storage or reaction plate. The reaction plate was then sealed and incubated at 37 °C and shaken at a speed of 60 rpm. At 0.5, 1, 2, and 4 h, we added 450 μL of pre-cooled internal standard (4 μg/ml caffeine in methanol) to the respective wells in the reaction plate. The liquid was then transferred to a quenched plate. The quenched plate was centrifuged (Eppendorf, model 5810R) at 4000 rpm for 20 minutes at 4 °C. We transferred 200 μL of supernatant to a 96-well plate and analyzed using UPLC-MS (Waters Inc.). Selected ion recording (SIR) was used to detect the compounds and internal standard. The log peak area ratio (compound peak area / internal standard peak area) was plotted against time and the slope was used to calculate the elimination rate constant [k = (−2.303) * slope]. The half-life (h) was calculated as t (1/2) = 0.693 / k. All compounds were tested in triplicate.

Characterization of key compounds and intermediates

All moisture-sensitive reactions were performed using syringe-septum techniques under an atmosphere of dry N₂ unless otherwise noted. Prior to use, all glassware was dried in an oven at 120° C for at least 6 h or flame-dried under an atmosphere of dry nitrogen. Methylene chloride and acetonitrile were dried using an aluminum oxide column. Deuterated chloroform was stored over anhydrous potassium carbonate. Reactions were monitored by TLC analysis (pre-coated silica gel 60 F₂₅₄ plates, 250 μm layer thickness) and visualized using an ultraviolet lamp (254 nm) or by staining with either Vaughn’s reagent (4.8 g of (NH₄)₆Mo₇O₂₄·4H₂O and 0.2 g of Ce(SO₄)₂ in 100 mL of a 3.5 N H₂SO₄) or a potassium permanganate solution (1.5 g of KMnO₄ and 1.5 g of K₂CO₃ in 100 mL of a 0.1% NaOH solution). Unless otherwise specified, commercially available reagents were used as received. Flash column chromatography was performed using a Biotage Isolera One and Biotage KP-SIL SNAP cartridges. Melting points were acquired on Buchi Melting Point B-545. All NMR data was collected at room temperature in CDCl₃ or (CD₃)₂SO on a 400 or 500 MHz Bruker instrument. Chemical shifts (δ) are reported in parts per million (ppm) with internal CHCl₃ (δ 7.26 ppm for ¹H and 77.00 ppm for ¹³C), internal DMSO (δ 2.50 ppm for ¹H and 39.52 ppm for ¹³C), or internal TMS (δ 0.0 ppm for ¹H and 0.0 ppm for ¹³C) as the references. ¹H NMR data are reported as follows: chemical shift; multiplicity (s = singlet, bs = broad singlet, d = doublet, t = triplet, q = quartet, p = pentet, sext = sextet, sep = septet, m= multiplet, dd = doublet of doublets, dt = doublet of triplets, td = triplet of doublets, qd = quartet of doublets); coupling constant(s) (J) in Hertz (Hz); and integration. Purity was assessed by LC/MS/UV/ELSD using a Waters Acquity UPLC-MS with the purity being assigned as the average determined by UV/ELSD. All compounds were confirmed to ≥ 95% purity. Compounds that proved critical to our SAR analysis were further characterized using 1H NMR and HRMS.

N-benzyl-1-(pentan-2-yl)piperidin-4-amine (15)

We added acetic acid (2.11 mL, 36.9 mmol, 1.2 equiv) to a stirred solution of 14⁴² (5.20 g, 30.7 mmol, 1 equiv) and benzylamine (3.36 ml, 30.7 mmol, 1 equiv) in CH₂Cl₂ (50 mL). The resulting cloudy mixture was stirred at room temperature for 1 h and sodium triacetoxyborohydride (8.46 g, 39.9 mmol, 1.3 equiv) was added portion-wise over 10 min. The resulting heterogeneous mixture was stirred at room temperature overnight (ca. 16 h), quenched with a saturated aqueous solution of 1M sodium hydroxide, extracted with EtOAc (×2), dried (MgSO₄), filtered, and concentrated under reduced pressure to give 8.0 g (100% yield) of the desired product (15) as a pale yellow oil. ¹H NMR (500 MHz, Chloroform-d) δ 7.36 – 7.28 (m, 4H), 7.26 – 7.17 (m, 1H), 3.81 (s, 2H), 2.81 – 2.76 (m, 2H), 2.59 (sext, J = 6.6 Hz, 1H), 2.51 – 2.46 (m, J = 1H), 2.32 (t, J = 11.2 Hz, 1H), 2.19 (t, J = 11.3 Hz, 1H), 1.98 – 1.83 (m, 2H), 1.70 – 1.22 (m, 6H), 0.96 (d, J = 6.5 Hz, 3H), 0.90 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 140.8, 128.4, 128.0, 126.8, 59.0, 54.4, 50.7, 48.5, 45.7, 35.7, 33.1, 32.9, 20.2, 14.3, 14.1. HRMS (ESI+) m/z calcd for C₁₇H₂₉N₂ [M+H]⁺, 261.2331, found 261.2327.

tert-butyl 4-(benzylamino)piperidine-1-carboxylate (17)

We added acetic acid (2.76 mL, 48.2 mmol, 1.2 equiv) to a stirred solution of 1-Boc-4-piperidone (8.00 g, 40.2 mmol, 1 equiv) and benzylamine (4.39 mL, 40.2 mmol, 1 equiv) in dry CH₂Cl₂ (150 mL). The resulting cloudy mixture was stirred at room temperature for 1 h and sodium triacetoxyborohydride (11.4 g, 60.3 mmol, 1.5 equiv) was added portion-wise over 30 minutes. The resulting heterogeneous mixture was stirred at room temperature overnight (ca. 16 h), quenched with a saturated aqueous solution of NaHCO3, extracted with EtOAc (×2), dried (MgSO₄), filtered, and concentrated under reduced pressure. The crude mixture was purified by chromatography on SiO₂ (MeOH/CH₂Cl₂, 2% to 15%) to give 11.4 g (98% yield) of the desired product of a colorless oil. ¹H NMR (500 MHz, Chloroform-d) δ 7.37 – 7.32 (m, 4H), 7.30 – 7.23 (m, 1H), 4.20 – 3.90 (m, 3H), 3.85 (s, 2H), 2.85 – 2.75 (m, 2H), 2.72 – 2.67 (m, 1H), 1.89 (d, J = 12.8 Hz, 2H), 1.47 (s, 9H), 1.40 – 1.32 (m, 2H). ¹³C NMR (126 MHz, CDCl₃) δ 154.7, 139.3, 128.5, 128.2, 127.2, 79.4, 54.1, 50.4, 42.6, 42.1, 31.9, 28.4. HRMS (ESI+) m/z calcd for C₁₇H₂₇N₂O₂ [M+H]⁺, 291.2073, found 291.2067.

tert-butyl 4-(1-benzyl-3-(3-fluoro-5-(trifluoromethyl)phenyl)ureido)piperidine-1-carboxylate (18)

To a stirred solution of 17 (2.90 g, 9.99 mmol, 1 equiv) in CH₂Cl₂ (50 mL), we added 1-isocyanato-3-(trifluoromethyl)benzene (1.44 mL, 9.99 mmol, 1.1 equiv). The reaction mixture was stirred at room temperature overnight (ca. 16 h), diluted with EtOAc, washed with a saturated aqueous solution of NaHCO₃, brine, dried (MgSO₄), filtered, and concentrated under reduced pressure to give a pale yellow liquid. The crude mixture was purified by chromatography on SiO₂ (MeOH/CH₂Cl₂, 1% to 10%) to give 4.8 g (97% yield) of the desired product 18 as a white solid. ¹H NMR (400 MHz, Chloroform-d) δ 7.49 – 7.40 (m, 2H), 7.39 – 7.21 (m, 4H), 7.11 (s, 1H), 6.90 (d, J = 8.1 Hz, 1H), 6.62 (s, 1H), 4.60 – 4.52 (m, 1H), 4.48 (s, 2H), 4.36 – 4.01 (m, 2H), 2.90 – 2.65 (m, 2H), 1.79 (d, J = 12.3 Hz, 2H), 1.66 – 1.50 (q, J = 12.3 Hz, 2H), 1.44 (s, 9H). ¹³C NMR (101 MHz, Chloroform-d) δ 162.6 (d, J = 246.4 Hz), 155.0, 154.5, 141.4 (d, J = 11.3 Hz), 132.2 (qd, J = 33.5, 9.4 Hz), 129.4, 128.2, 126.0, 123.1 (q, J = 273.5 Hz), 111.7, 109.9 (d, J = 26.2 Hz), 106.7 (d, J = 25.0 Hz), 79.8, 53.0, 46.2, 43.2, 30.0, 28.4. HRMS (ESI−) m/z calcd for C₂₅H₂₈F₄N₃O₃ [M-H]⁻, 494.2067, found 494.2058.

General reductive amination procedure.⁵⁸

To a stirred solution of 1-benzyl-3-(3-fluoro-5-(trifluoromethyl)phenyl)-1-(piperidin-4-yl)urea (0.040 g, 0.101 mmol, 1 equiv) in tetrahydrofuran (1 ml), we added the aldehyde or ketone (0.202 mmol, 2 equiv) and dibutyltin (IV) chloride (1.53 mg, 0.005 mmol, 0.05 equiv). After stirring the heterogeneous mixture for 5 minutes at room temperature, we added phenylsilane (0.025 ml, 0.202 mmol, 2 equiv). The resulting solution was stirred at room temperature overnight (ca. 16 h). For some more sterically encumbered ketones, we observed less than 20% conversion after overnight reaction. For these ketones, we added an extra 3-4 equiv of ketone and 2 equiv of phenylsilane and the reaction was stirred at room temperature for an additional 24 h. After the starting material was completely consumed, the crude mixture was diluted with DMSO, filtered through a 0.45 micron filter, and subjected to HPLC chromatography.

General isocyanate addition

To a stirred solution of 1-benzylpiperidin-4-amine (0.250 g, 1.314 mmol, 1.2 equiv) in CH₂Cl₂ (6 mL), we added 1-isocyanato-3-(trifluoromethyl)benzene (1.53 mL, 1.10 mmol, 1 equiv) and N,N-diisopropylethylamine (0.286 mL, 1.64 mmol, 1.5 equiv). The resulting mixture was stirred at room temperature overnight (ca. 16 h), concentrated under reduced pressure, and purified by chromatography on SiO₂.

General CDI-mediated isocyanate addition

We added carbonyldiimidazole (0.026 g, 0.158 mmol, 1.3 equiv) to a stirred solution of aniline (0.146 mmol, 1.2 equiv) in tetrahydrofuran (1 ml). The mixture was stirred at room temperature for 3 h, followed by the addition of N-benzyl-1-butylpiperidin-4-amine (20) (0.030 g, 0.122 mmol, 1 equiv). The resulting solution was stirred at room temperature overnight (ca. 16 h). The crude mixture was diluted with DMSO, filtered through a 0.45 micron filter, and directly subjected to HPLC chromatography.

Reproduction of characterization from previously reported key compounds

1-benzyl-1-(1-(pentan-2-yl)piperidin-4-yl)-3-(3-(trifluoromethyl)phenyl)urea (7).⁴²

Note: characterization was previously reported⁴² but is reproduced below for convenience: ¹H NMR (500 MHz, Chloroform-d) δ 7.57 (s, 1H), 7.42 – 7.36 (m, 2H), 7.34-7.30 (m, 3H), 7.29 – 7.17 (m, 4H), 6.50 (s, 1H), 4.53 (s, 2H), 4.37 (tt, J = 12.1, 4.2 Hz, 1H), 2.84-2.80 (m, 2H), 2.57 (sext, J = 7.0 Hz, 1H), 2.42 (td, J = 11.6, 2.5 Hz, 1H), 2.27 (td, J = 11.6, 2.3 Hz, 1H), 1.83 – 1.76 (m, 2H), 1.69 (dqd, J = 24.1, 12.0, 3.8 Hz, 2H), 1.47 (dq, J = 12.5, 5.2 Hz, 1H), 1.41 – 1.13 (m, 3H), 0.95 (d, J = 6.6 Hz, 3H), 0.89 (t, J = 7.1 Hz, 3H). ¹³C NMR (126 MHz, CDCl₃) δ 155.4, 139.6, 137.5, 130.8 (q, J = 31.5 Hz), 129.1, 129.0, 127.1, 126.0, 123.9 (q, J = 273.4 Hz, 1C), 122.6, 119.2 (q, J = 3.8 Hz), 116.3 (q, J = 3.8 Hz), 58.8, 53.2, 49.5, 46.1, 46.0, 35.7, 30.6, 30.4, 20.1, 14.1, 14.0. m/z (APCI-pos) M+1 = 448.2. HRMS (ESI+) m/z calcd for C₂₅H₃₃F₃N₃O [M+H]⁺, 448.2575 found 448.2575.

1-benzyl-1-(1-butylpiperidin-4-yl)-3-(3,4-dichlorophenyl)urea. NAcM-OPT (67).⁴²

Note: characterization was previously reported⁴² but is reproduced below for convenience: m.p.: 118-120 °C. ¹H NMR (500 MHz, Chloroform-d) δ 7.42 – 7.39 (m, 3H), 7.36 – 7.29 (m, 3H), 7.21 (d, J = 8.7 Hz, 1H), 6.91 (d, J = 8.7 Hz, 1H), 6.27 (s, 1H), 4.49 (s, 2H), 4.44 – 4.38 (m, 1H), 2.99 (d, J = 11.2 Hz, 2H), 2.31 (t, J = 7.8 Hz, 2H), 2.05 (t, J = 11.7 Hz, 2H), 1.85 – 1.65 (m, 4H), 1.44 (p, J = 7.5 Hz, 2H), 1.30 (sext, J = 7.4 Hz, 2H), 0.90 (t, J = 7.4 Hz, 3H). ¹³C NMR (126 MHz, Chloroform-d) δ 155.3, 138.6, 137.4, 132.4, 130.1, 129.3, 128.0, 126.1, 125.9, 121.2, 118.8, 58.4, 53.1, 52.9, 46.1, 30.1, 29.4, 20.8, 14.0. HRMS (ESI+) m/z calcd for C₂₃H₃₀Cl₂N₃O [M+H]⁺ 434.1766, found 434.1761.

Abbreviations

DCN1

defective in cullin neddylation 1

UBE2M

NEDD8-conjugating enzyme Ubc12

SCC

Squamous cell carcinoma

CRL

cullin ring ligases

NEDD8

Neural Precursor Cell Expressed, Developmentally Down-Regulated 8

SAR

structure-activity relationship

SPR

structure-property relationship

TR-FRET

time-resolved fluorescence resonance energy transfer

HTS

high-throughput screen

t-Bu

tert-butyl

NAE

NEDD8-activating enzyme

RBX1

Ring-Box 1, E3 Ubiquitin Protein Ligase

CUL1

cullin 1

AF

AlexaFluor 488

IV

intravenous

PO

per os (by mouth)

CL_int

intrinsic clearence

C_max

maximum achieved concentration

t(1/2)

half-life (time taken for the concentration of analyte to fall to half its original value)

AUC_last

area under the plasma concentration-time curve from time zero to time of last measurable concentration