SAR studies of 4-acyl-1,6-dialkylpiperazin-2-one arenavirus cell entry inhibitors

Abstract

Old World (Africa) and New World (South America) arenaviruses are associated with human hemorrhagic fevers. Efforts to develop small molecule therapeutics have yielded several chemical series including the 4-acyl-1,6-dialkylpiperazin-2-ones. Herein, we describe an extensive exploration of this chemotype. In initial Phase I studies, R¹ and R⁴ scanning libraries were assayed to identify potent substituents against Old World (Lassa) virus. In subsequent Phase II studies, R⁶ substituents and iterative R¹, R⁴ and R⁶ substituent combinations were evaluated to obtain compounds with improved Lassa and New World (Machupo, Junin, and Tacaribe) arenavirus inhibitory activity, in vitro human liver microsome metabolic stability and aqueous solubility.

The Arenaviridae family of viruses is comprised of a large number of species, some of which are associated with acute hemorrhagic fevers (HF) in humans.^1–3 HF arenaviruses, including Lassa (LASV) and the New World species Machupo (MACV) or Junin (JUNV), represent a serious public health risk, especially in endemic regions of Africa and South America.^4–6 Due to the high mortality rates and limited therapeutic options these HF viruses are classified as Category A pathogens by the Centers for Disease Control (CDC).⁷ LASV, in particular, may be responsible for up to 300,000 human disease cases per year.^8–10 Mortality rates for hospitalized LASV HF patients are 15–20% and survivors often suffer permanent sequelae including bilateral hearing damage.^11–14 The nonspecific antiviral Ribavirin currently offers the sole therapeutic option, however, its use is restricted to high-risk patients due to its variable efficacy and potentially serious adverse effects.^15–17 The development of potent and specific agents to treat Lassa and other arenavirus hemorrhagic fevers is therefore, urgently needed.

To mitigate the significant safety and logistical challenges associated with utilizing Biosafety Level 4 (BSL4) arenaviruses in drug discovery, a number of surrogate BSL2 assays have been developed. For example, a screening campaign using the New World arenavirus Tacaribe virus (TCRV), which is a BSL2 Clade B New World arenavirus that is highly-related to Category BSL4 arenaviruses including JUNV and MACV, was used to identify a novel chemical series (including ST-294) of arenavirus inhibitors.¹⁸ Cell-based assays specifically targeting the arenavirus glycoprotein (GP) and cell entry, including pseudotyped virus and cell–cell membrane fusion assays, have also been used to identify or characterize the benzimidazole and 4-acyl-1,6-dialkylpiperazin-2-one chemical series, which share a common GP binding site and mechanism of action.^19–23

The initial report of the 4-acyl-1,6-dialkylpiperazin-2-one inhibitors identified a set of substituents at the R¹, R⁴ and R⁶ positions of the 4-acyl-piperazin-2-one core (Fig. 1) that exhibited low micromolar to submicromolar potency against LASV glycoprotein expressed in vesicular stomatitis virus (VSV) pseudotyped viruses (pLASV).^20,24,25 Additional investigation of the structure–activity relationships (SAR) of this chemical series identified the carbonyl groups as significant activity determinants and that the enantiomer possessing (S) configuration at C⁶ (R⁶ group attachment point) was approximately 15-fold more potent than the (R)-enantiomer.²⁶ The (S)-enantiomers of our initial 4-acyl-1,6-dialkylpiperazin-2-one compounds 1 (16G8) and 2 (17D1) represent eutomers with pLASV EC₅₀ values: rac-16G8 600 nM vs (S)-16G8 300 nM and rac-17D1 500 nM vs (S)-17D1 250 nM.^26,27 We therefore utilized the (S)-enantiomers of R⁶ benzyl, phenethyl and other substituents as starting points for further SAR exploration of this chemical series.

Fig. 1.

Initial 4-acyl-1,6-dialkylpiperazin-2-one starting points and optimization flow scheme.

In Phase I exploration R¹ and R⁴ scanning libraries were synthesized and screened for inhibition of LASV cell entry as determined in a membrane cell–cell fusion assay at 50 nM (Fig. 1). In Phase II studies, iterative SAR modifications at substituent positions R¹, R⁴ and R⁶ were evaluated in assays with pseudotyped pLASV, pMACV or pJUNV virions expressing LASV, MACV or JUNV glycoproteins, respectively, in order to identify broad-spectrum arenavirus cell entry inhibitors. Selected compounds were subsequently characterized in human liver microsome (HLM) metabolism and solubility assays (Fig. 1).

For R¹ scanning library compounds we devised a synthetic route enabling final step diversification at the R¹ position starting with an intermolecular Mitsunobu alkylation²⁸ of t-butyl nosylglycine with Boc protected aminoalcohol A to provide B (Scheme 1). The nosyl group was cleaved (thiophenol, K₂CO₃, DMF) to yield the secondary amine C, which was then coupled with the appropriate carboxylic acid R⁴COOH (EDCI, HOAt, 2,6-lutidine, DMF) to give the tertiary amide D. Hydrolysis of the t-butyl ester (LiOH, t-BuOH/H₂O/THF) followed by Bocdeprotection with TFA (CH₂Cl₂) provided the amino acid E as the TFA salt. Final step diversification at R¹ was completed through a one-pot reductive amination²⁹/EDCI-mediated ring closure starting with intermediate E.

Scheme 1.

Reagents and conditions: (i) PPh₃, DIAD, THF, under N₂, rt, 2 h; (ii) PhSH, K₂CO₃, DMF, under N₂, rt, 3 h; (iii) R⁴COOH, HOAt, 2,6-lutidine, EDCI, DMF, rt, 16 h; (iv) LiOH, t-BuOH/H₂O/THF, rt, 1 h, then 1 N aq. HCl; (v) TFA, CH₂Cl₂, rt; (vi) RCHO, NaBH(OAc)₃, Et₃N, ClCH₂CH₂Cl, rt, 18 h, then EDCI, NaHCO₃, CH₂Cl₂, rt, 8 h.

A synthetic route enabling final step diversification at the R⁴ position (Scheme 2) started with commercially available aminoalcohol G. The R¹ group was added using a stepwise reductive amination procedure with appropriate aldehyde entailing pre-formation of imine followed by NaBH₄ reduction (MeOH) to give intermediate H. Subsequent N-acylation by treatment of H with the acid chloride of nosylglycine (NaHCO₃, CH₂Cl₂) provided I. The 6-membered ring was closed using an intramolecular Mitsunobu alkylation of the sulfonamide²⁸ (PPh₃, DIAD, THF) to give J. Nosyl deprotection using thiophenol and K₂CO₃ (DMF) and subsequent acylation of the free amine with the appropriate carboxylic acid R⁴COOH (EDCI, HOAt, 2,6-lutidine, DMF) furnished compounds of type K. All compounds were purified by PTLC or flash chromatography and the purity and identity of all compounds was assessed by LC/MS.

Scheme 2.

Reagents and conditions: (i) RCHO, MeOH, rt, 1 h, then NaBH₄, 0 °C to rt, 2 h; (ii) 2-(2-nitrophenylsulfonamido)acetyl chloride, NaHCO₃, CH₂Cl₂, rt, 3 h; (iii) PPh₃, DIAD, THF, under N₂, rt, 2 h; (iv) PhSH, K₂CO₃, DMF, under N₂, rt, 3 h; (v) R⁴COOH, HOAt, 2,6-lutidine, EDCI, DMF, rt, 14 h.

At R¹ benzyl was favored over phenethyl or phenylpropyl substituents (Table 1, compare 6, 14 and 16), indicating a defined preference for the linker length at R¹. A single aryl ring demonstrated greater activity than naphthalene and biaryl groups (compare 6 with 10, 12, 13 and 15) and the identity of aryl substituents significantly impacted activity while their positional effects were relatively weaker. The incorporation of single methoxy or hydroxy groups revealed a preference for meta-substitution but were well tolerated at any ring position (4 vs. 8 and 9, 5 vs. 7). In contrast, halogen, phenoxy, cyano, nitro, or basic nitrogen substitutions in the ring significantly reduced activity at any position (see Supplementary data, Fig. S1). Overall, the results indicated significant flexibility for incorporation of small lipophilic substituents or polar alcohols on the R¹ benzyl ring.

Table 1.

SAR data of selected R¹ (left panel) and R⁴ (right panel) substituent Phase I scanning libraries indicating their % inhibition at 50nM in the LASV GPC cell–cell fusion assay.

Compd.	R1	% inh. 50 nM	Compd.	R4	% inh. 50 nM
3		86	17		66
4		82	18		61
5		81	19		59
6		71	20		58
7		64	21		56
8		61	22		54
9		60	23		50
10		58	24		45
11		53	25		40
12		51	26		26
13		36	27		18
14		34	28		15
15		33	29		14
16		31	30		0

As opposed to R¹ substituents, the R⁴ acyl substituents demonstrated SAR with strong preferences in both the identity and position of aryl functionality (Table 1, Supplementary data, Fig. S2). The two most active classes of aryl groups observed in the original piperazinone study (the 2-heteroaryl (17–19, 22) and the 4-substituted phenyl (20, 21, 23) groups) proved optimal in the R⁴ scanning library.²⁰ Of note, incorporation of an alkoxy or dimethylamino group at the 4-position of the phenyl ring resulted in a notable increase in potency for compounds 20, 21, and 23 while the absence of a 4-position substituent (28) and/ or the placement of a dimethylamino (29) or alkoxy (30) group at the 3-position resulted in dramatic losses in activity. The addition of a 5-methoxy group to the 2-indole substituent (compounds 17 vs. 26) also provided a notable increase in potency. In contrast, the addition of a single carbon linker (27) to the phenyl ring significantly decreased activity.

Following the individual R¹ and R⁴ replacements, a small combinatorial library was synthesized containing three of the most active R¹ substituents (compounds 3–5, Table 1) and R⁴ acyl groups (compounds 17, 18, and 20, Table 1). These compounds showed low-nanomolar inhibition of LASV GP fusion with no loss of activity when the more polar optimal substituents were combined in the same structure (compounds 31–33, Table 2) suggesting that optimization for lipophilic efficiency (LipE) may lead to potent compounds with improved drug-like properties. These initial scanning libraries allowed the identification of promising R¹ and R⁴ substituents as starting points to initiate optimization of broad-spectrum arenavirus inhibitory activity and drug-like properties.

Table 2.

Analogs incorporating combinations of the most active substituents from the R¹and R⁴ scanning libraries and their respective EC₅₀ values in the LASV GPC cell–cell fusion assay.

R4	R1
	Compd.	EC50 (nM)	Compd.	EC50 (nM)	Compd.	EC50 (nM)
	31	5	34	4	37	5
	32	8	35	7	38	7
	33	7	36	6	39	5

Phase II SAR exploration studies, including characterization of broad-spectrum arenavirus (pseudotyped pLASV, pMACV and pJUNV) activity, human liver microsome (HLM) metabolic stability and solubility, were initiated with the synthesis of analogs containing benzyl or phenethyl at R⁶, benzyl or m-methoxy benzyl at R¹ and 5-methoxyindole at R⁴. R⁶ phenethyl was found to exhibit approximately 2–4 fold greater potency over benzyl (Table 3, compare 35, 40, 41 and 42). However, R⁶ benzyl compounds exhibited greater solubility and metabolic stability in HLM assays. Thus, while 42 was 2–4 fold less potent in pLASV, pMACV and pJUNV assays than 41 it exhibited a substantially increased HLM half-life of 59 vs 10 min and solubility of 36 vs 4 μM.

Table 3.

Comparison of activity of R⁶ linker length and R¹ substituents. Activity is given as EC₅₀ (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (T_1/2 (min)) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	n	X	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
35	2	OMe	1.74	2.39	1.77	4,500	7.7	4.9
40	1	OMe	6.9	3.6	2.52	3,300	9.8	7.4
41	2	H	3.3	2.13	5.8	> 30,000	10	4.4
42	1	H	8.64	11.52	4.87	> 10,000	59	36.3

For Phase II SAR expansion we synthesized R⁶ (S) benzyl analogs with a limited number of R⁴ substituents (5-methoxy, 5-methyl, and 5-cyclopropyl indoles) to further explore R¹ SAR (80 analogs) as well as a limited number of R¹ substituents (benzyl, m-methoxybenzyl, m-isopropoxybenzyl, or m-difluoromethoxybenzyl) to explore additional R⁴ SAR (68 analogs), respectively. Novel R⁶ SAR was also explored utilizing a limited number of R¹ and R⁴ substituents (64 analogs). In total 212 additional analogs were synthesized in Phase II modifications at R¹, R⁴ and R⁶.

All Phase II analogs were first screened against pLASV and pMACV. Those exhibiting EC₅₀ values < 25 nM in both assays were subsequently tested against pJUNV. Of the 132 of compounds displaying < 25 nM potency in the pLASV assay, 118 were also < 25 nM against pMACV and of those 108 were also < 25 nM against pJUNV. Comparative EC₅₀ values and the range of activity spectrums for these Phase II analogs are illustrated in Fig. 2. While a number of interesting pseudotyped-specific SAR trends were observed, our focus was on the identification of compounds exhibiting broad spectrum activity with EC₅₀ values < 10 nM in all three assays (64 compounds).

Fig. 2.

A 3-dimensional plot of EC₅₀ values obtained in the pLASV, pMACV and pJUNV assays for the analogs exhibiting < 25 nM (dark circles) or < 10 nM (yellow circles) EC₅₀ values in each of the pseudtoyped virus assays.

To confirm that the activities observed in pseudotyped virus assays translate to the inhibition of replicative arenaviruses the activities of two compounds (35 and 42, Table 3) were initially tested and confirmed against BSL4 replicative LASV with EC₅₀ values of 33 and 113 nM, respectively. To extend this observation to a broader range of compounds we subsequently tested compounds exhibiting EC₅₀ values < 25 nM against pLASV, pMACV and pJUNV in assays with both pseudotyped (p-TCRV) and replicative Tacaribe (TCRV) virus, a BLS2 virus that is non-pathogenic to humans but highly related to JUNV.¹⁸ As shown in Fig. 3 compounds exhibiting < 25 nM EC₅₀ values against the BSL4 pseudotyped viruses also exhibited comparable activities against pTCRV. These data confirm comparable inhibition between the glycoprotein-mediated cell entry of both the BSL4 and BSL2 viruses. Furthermore, the data confirm the translation between pseudotyped and replicative virus inhibition (R² = 0.61, Fig. 3). Thus, the data further validate the selection of cell entry inhibitors based on their activities in pseudotyped virus assays.

Fig. 3.

Logarithmic plot of pTCRV vs replicative TCRV (immunostaining assay) EC₅₀ values (R² = 0.61) for analogs selected for EC₅₀ values < 25 nM against each of the pseudotyped viruses (pLASV pMACV and pJUNV) and HLM half-life data.

At R¹, SAR trends were observed in both antiviral activity and HLM stability (Table 4). Benzyl groups with less polar meta- or ortho- substituents are consistently potent and broad spectrum (compounds 40, 43, 44, 45, 48, and 49) with a particular preference for lipophilic ethers. Some polar R¹ meta-benzyl substituents groups including alcohols (compounds 52, 56, and 59), retain broad spectrum activity despite their increased polarity while others such as sulfonamides (54), and moderately polar groups like 1H-pyrazolyl (53) and alkyl esters (55) retain pMACV and pJUNV activity but lose activity against pLASV. Both carboxylic acids and amides (compounds 60 and 61) are not tolerated for any viral species tested (see data for more analogs in Supplementary data, Table S1).

Table 4.

Phase II SAR of R¹ substituents. Activities provided are EC₅₀ values (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (minutes) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	R1	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
40		6.9	3.6	2.52	3,300	9.8	7.4
42		8.64	11.52	4.87	> 10,000	59	36.3
43		2.92	2.08	3.04	> 10,000	37	5.8
44		1.33	1.54	1.91	41,000	12	6.9
45		2.37	2.88	3.78	49,000	8	5.7
46		0.94	1.85	0.75	> 30,000	59	7.2
47		5.8	> 25	12.23	> 30,000	72	5.6
48		5.45	2.85	1.56	> 30,000	25	11.3
49		5.56	6.26	11.16	> 100,000	27	24.2
50		0.733	1.73	2.68	5,000	36	10.21
51		4.44	1.32	3.14	> 30,000	13	9.2
52		2.81	3.17	6.91	79,500	11	48.8
53		22.1	1.38	1.65	> 30,000	N/A	6.8
54		> 25	5.71	3.27	N/A	N/A	12
55		23.6	5.4	1.8	> 30,000	N/A	6.9
56		2.42	2.50	5.38	10,200	18	17.2
57		3.7	2.88	N/A	> 30,000	5.3	3.9
58		2.7	2.03	2.6	> 100,000	13	7.5
59		4.09	1.01	1.85	39,000	29	30.2
60		> 25	> 25	N/A	N/A	N/A	N/A
61		> 25	> 25	N/A	N/A	N/A	N/A

Unfortunately, alkyl ethers at R¹ pose a significant metabolic liability as demonstrated by the relative HLM stability of compounds 42 and 43 vs. 40, 44 and 45. Stabilized ethers with oxidation-resistant alkyl groups (compounds 46 and 47) have more favorable microsome stability but only the m-difluoromethoxy (46) group maintains broadspectrum activity. The broad-spectrum nature of m-difluoromethoxy group is interesting in light of the poor pJUNV and pMACV activity of more lipophilic m-trifluoromethoxy (47) analog and the general preference for alcohols as R¹ substituents, as the difluoromethoxy group has been reported to act as a hydrogen-bond donor and potential isostere for alcohols.³⁰

Both 6-indolyl (50) and N-alkyl 6-indolyl groups (51, 57, 58) also demonstrate good broad-spectrum activity, but the highest activity is derived from the unalkylated analog 50. This is likely due to the hydrogen bonding of indole NeH, similar to the phenols and –OCF₂H groups. The unalkylated indoles also possess above average metabolic stability with 50 having an HLM half-life (T_1/2) of 36 min, while introduction of alkyl groups causes significant metabolic liability.

Novel R⁴ analogs were synthesized using the metabolically more stable R¹ groups benzyl and m-difluoromethoxy benzyl to further evaluate potential metabolic liabilities stemming from the R⁴ substituent (Table 5). The majority of Phase II R⁴ analogs were designed to explore substituents on the most potent aryl and heteroaryl R⁴ groups (indoles, phenyls, and quinolines) identified in the Phase I exploration, with the intention of further improving metabolic stability while retaining broad-spectrum potency. Although some sterically bulky 5-alkoxy indoles showed dramatic improvements in metabolic stability (compounds 62 and 63) broad-spectrum potency was lost. 5-methylindole and 5-cyclopropylindole substituents (compounds 68 and 69) demonstrated sufficiently increased metabolic stability over 5-methoxyindole analog (42), when paired with a benzyl R¹. Of note, compound 66 having 5-cyclopropylindole at R⁴ and m-difluoromethoxy benzyl at R¹ showed some improvement in metabolic stability but at the same time was slightly less potent than the 5-methoxyindole analog 46 (see Table 4).

Table 5.

Phase II SAR of R⁴ substituents. Activities provided are EC₅₀ values (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (T_1/2 (min)) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	R4	X	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
42		H	8.64	11.52	4.87	> 10,000	59	36.3
62		H	9.38	> 25	> 25	> 30,000	137	8.6
63		H	> 25	> 25	4.235	N/A	246	14.6
64		OCF2H	2.34	19.52	3.8	> 100,000	43	5.3
65		H	25.4	> 25	> 25	> 10,000	120	15.5
66		OCF2H	2.12	6.81	4.05	10,400	77	6.2
67		OCF2H	2.29	2.99	0.52	29,700	13	10
68		H	1.95	21.52	2.68	11,200	119	7.7
69		H	14.6	> 25	> 25	> 100,000	125	8.3
70		H	> 25	> 25	> 25	16,000	N/A	10.8
71		H	5.96	12.53	4.46	9,040	42	5.7
72		H	> 25	> 25	> 25	N/A	N/A	67.7
73		OCF2H	2.31	3.59	4.78	> 100,000	16	9.5
74		OMe	> 25	> 25	> 25	N/A	N/A	> 100
75		OCF2H	4.68	3.98	8.82	46,700	5	8.8
76		H	> 25	> 25	> 25	> 100,000	31	> 100

The 7-fold difference in potency against pLASV between 66 and 69 strongly suggested that it should be possible to pair 5-cyclopropylindole and a suitable R¹ group to obtain a broadly potent compound that maintained the cyclopropylindole’s metabolic stability gains. While other heteroaryl groups showed acceptable-to-good broad-spectrum potency, such as quinolines 67 and 73 and benzothiophene 71, none of them were found to endow stability or solubility advantages over the indole. Further, the wide range of 4-substituted phenyls synthesized showed significantly reduced broad-spectrum activity with few exceptions, e.g., the 4-piperidinylphenyl group (75), which while potent, exhibited an HLM half-life of only 5 min. Polar substituents such as sulfones (72, 74), nitriles (70), acetyl (65, 76) and a variety of other groups were not tolerated on either phenyl or heteroaryl rings at any position and lead to dramatically reduced potency (see Supplementary data, Table S2). 5-methyl and 5-cyclopropyl indoles were thus selected as the preferred R⁴ groups to generate R¹ and R⁶ combinatorial compounds optimized for stability and potency.

In parallel with expansion of R¹ and R⁴ SAR we explored additional R⁶ substituents whereby R⁴ (5-methoxy indole) and R¹ (benzyl, m-methoxybenzyl and m-difluoromethoxybenzyl) substituents were restricted and novel R⁶ substituent analogs synthesized. The R⁶ position is the most tolerant of different substitutions and a wide variety of either lipophilic or polar charged groups were synthesized using either Scheme 1, starting with the appropriate commercially available chiral aminoalcohol, or Scheme 3 with the intention of increasing solubility and metabolic stability concurrently. Thus, methyl (O-t-Bu)serinate (L) was first acylated with R¹COOH and EDCI (Scheme 3). Subsequent reduction of M using lithium aluminum hydride afforded aminoalcohol N. Acylation of the amine with nosylglycine acyl chloride and magnesium oxide gave intermediate O, which was cyclized under Mitsunobu conditions to afford piperazinone P. Treatment of the t-butyl ether with 4 N HCl in 1,4-dioxane gave free alcohol Q, which was first triflated using triflic anhydride and finally reacted with a desired amine NHR’R” to afford intermediate R. Finally, nosyl deprotection followed by the amide coupling with the appropriate carboxylic acid R⁴COOH furnished compounds of type S.

Scheme 3.

Reagents and conditions: (i) R¹COOH, EDCI, Et₃N, DCM, rt, 16 h; (ii) LAH, THF, reflux, 16 h; (iii) 2-(2-nitrophenylsulfonamido)acetyl chloride, MgO, 4:1 THF/H₂O, rt, 4 h; (iv) PPh₃, DIAD, THF, under N₂, rt, 16 h; (v) 4 N HCl/1,4-dioxane, 70 °C, 4 h; (vi) Tf₂O, Et₃N, DCM, under N₂, 0 °C, 1 h, then rt, 3 h; (vii) NHR’R”, Et₃N, THF, rt, 16 h; (viii) PhSH, K₂CO₃, DMF, under N₂, rt, 16 h; (ix) R⁴COOH, 2,6-lutidine, EDCI, DMF, rt, 4 h.

For non-polar R⁶ substituents (Table 6) potency tracks well with lipophilicity and the physical length of the group is unimportant (compounds 77, 81, 82, 83, and 84). For R⁶ analogs bearing a basic amine, a linker length of one carbon between the piperazinone and amine (e.g., 79 vs 94) is preferred for activity. Overall lipophilicity of the R⁶ amine is important for activity, as tertiary amines (90) with fewer carbons than piperidine, and thus lower lipophilicity, resulted in complete loss of potency. Larger amine groups themselves bearing polar substituents such as sulfones and sulfonamides (88, 91), nitriles (92), and even methoxy (87) groups all show significant losses in activity. Additionally, the basicity of the amine plays an important role in determining activity, as difluorinated piperidine analogs, which are more lipophilic yet less basic, exhibit decreased activity compared to the unfluorinated parent piperidine (compound 78 vs. 86 and 93). If the site of fluorination is sufficiently removed from the basic nitrogen, as in the 4-methylpiperidine (85) and its 4-trifluoromethyl analog (89), the inductive withdrawal effect on basicity is minimized and the increased lipophilicity makes 89 more potent.

Table 6.

Phase II SAR of R⁶ substituents. R⁶ groups bearing amines were synthesized according to Scheme 3. Activities provided are EC₅₀ values (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (T_1/2 (min)) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	R6	X	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
77		H	5.53	20.75	5.66	> 10,000	39	12.2
78		H	4.76	4.75	5.35	76,900	17	> 100
79		H	2.99	4.53	1.25	17,100	52	28.2
80		H	12.56	11.24	12.6	4,900	N/A	11.8
81		OMe	2.7	3.38	4.65	> 10,000	N/A	11.7
82		OMe	1.34	2.66	1.14	> 10,000	N/A	3.8
83		OMe	5.56	5.08	8.05	> 10,000	N/A	11.8
84		H	5.53	4.285	4.3	> 30,000	N/A	8.8
85		H	11.2	> 25	3.15	> 30,000	16	93.1
86		H	11.34	20.5	23.6	95,300	23	72.4
87		H	> 25	> 25	N/A	N/A	N/A	> 100
88		H	> 25	> 25	N/A	N/A	N/A	> 100
89		H	3.47	8.925	3.92	17,300	32.5	16.5
90		H	> 25	> 25	N/A	N/A	N/A	> 100
91		H	21.1	> 25	16.2	N/A	N/A	16.9
92		OCF2H	20.69	23.24	> 25	N/A	N/A	27.9
93		OCF2H	20.25	25.52	32.7	> 100,000	N/A	53.1
94		H	> 25	> 25	> 25	N/A	N/A	> 100

The HLM half-life of all compounds in this R⁶ exploration series was < 50 min, except 79, which exhibited a half-life of 52 min. Pairing the cyclopropylpiperidine of 79 with an R⁴ group that is less metabolically labile than 5-methoxy indole could provide further improvement in HLM stability. A number of additional R⁶ analogs did not exhibit broad-spectrum potency and/or improved HLM stability (see Supplementary data, Table S3).

Initial exploration at positions R3 and R5, as well as attempts to conformationally lock the position of the R⁴ or R¹ substituents (see Supplementary data, Table S4 and Fig. S3) were unable to identify suitably active chemical subseries. Thus, final analog series were synthesized to identify additional R¹ (Table 7, Supplementary data, Table S5) and R⁶ (Table 8) modifications in combination with the most stable, broadly potent complementary R⁴, R⁶ and R¹ groups. Combining preferred R¹ substituents with the 5-methylindole and 5-cyclopropylindole moieties gave a series of compounds with attractive broad-spectrum potency and improved stability.

Table 7.

Secondary Phase II synthesis and SAR of R¹ analogs combined with selected R⁴ and benzyl R⁶ groups. Activities provided are EC₅₀ values (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (T_1/2 (min)) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	R1	X	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
95			7.72	1.08	2.16	4,300	50	9.4
96			0.94	1.38	1.24	9,770	46	3.2
97			2.16	1.62	0.98	> 100,000	35	3.6
98			6.40	3.97	3.51	8,170	102	11.7
99			7.05	4.54	3.75	23,950	20	28.5
100			3.57	3.05	1.53	5,560	22	11.75
101			3.15	1.14	0.71	> 100,000	54	3.25
102			3.91	3.35	3.50	9,210	187	12
103			1.00	2.79	1.03	27,830	19	11.34
104			0.89	2.04	0.65	4,960	6.8	7.39

Table 8.

Secondary Phase II synthesis and SAR of R⁶ analogs combined with best R⁴ and benzyl R¹ groups. Activities provided are EC₅₀ values (nM) in the indicated pseudotyped virus assays. Cytotoxicity is given as CC₅₀ in Vero cells (7 days). HLM data is the observed half-life (T_1/2 (min)) in a human liver microsome assay. Aqueous solubility (μM) was measured at pH 7.4.

Compd.	R6	X	p-LASV	p-MACV	p-JUNV	CC50 (nM)	HLM	Solubility
105			11.87	17.18	> 25	7,800	N/A	12.30
106			10.50	15.58	2.33	3,400	156	7.70
107			15.10	5.09	5.55	5,500	N/A	12.86
108			4.13	11.32	3.64	24,450	23	12.35

The most active polar compounds in this set (Table 7, compounds 99, 100, 103, and 104) contain R¹ groups with benzyl alcohol substituents, as previously suggested by the Phase I and II R¹ explorations. While the benzyl alcohols proved to be metabolically labile m-alkyl substituted groups (96, 101) provided moderate HLM stability. Compound 102, containing the 6-indolyl group, demonstrated broad-spectrum potency and an excellent HLM half-life of 187 min. Unfortunately, additional combinations of selected metabolically stable R¹ (benzyl) and R⁴ groups (5-methylindole and 5-cyclopropylindole) with R⁶ amines intended to increase aqueous solubility did not prove advantageous for either solubility or broad-spectrum antiviral activity (Table 8, compounds 105–108).

In a final functional assay the prioritized compounds (Table 9) were also tested against pseudotyped virus expressing the GP of an attenuated JUNV strain (Candid-1). Each of these analogs exhibited decreased potencies against pJUNV (Candid-1) vs pJUNV, ranging from > 10- (98) to nearly 1000-fold (46). Interestingly, the attenuating mutation (F427I) in the GP of Junin (Candid-1) virus also exhibits resistance to benzimidazole cell entry inhibitors.^31,32 Consistent with earlier reports that the 4-acyl-1,6-dialkylpiperazin-2-one and benzimidazole chemical series share a common GP binding site and mechanism of action our results indicate that these distinct chemical series also share at least some resistance variants associated with attenuation.^22,23,31

Table 9.

Collection of prioritized compounds expressing potent, broad spectrum activity (< 10 nM EC₅₀ activity against pLASV, pMACV and pJUNV) and HLM metabolic half-life (T_1/2) > 45 min with indicated aqueous solubility (μM) at pH 7.4 and pJUNV (Candid-1) EC₅₀ activity.

	Compd.	p-LASV	p-MACV	p-JUNV	p-JUNV (Candid-1)	HLM	Solubility
	66	2.12	6.81	4.05	N/A	77	6.2
	46	0.94	1.85	0.75	710.3	59	7.2
	79	2.99	4.53	1.25	56.2	52	28.2
	95	7.72	1.08	2.16	272.9	50	9.4
	96	0.94	1.38	1.24	356	46	3.2
	98	6.40	3.97	3.51	48.4	102	11.7
	101	3.15	1.14	0.71	173.2	54	3.25
	102	3.91	3.35	3.5	181.5	187	12

In summary, the SAR exploration and optimization of 4-acyl-1,6-dialkylpiperazin-2-one chemical series was initiated with the (S)-enantiomers of compounds 1 and 2 exhibiting pLASV EC₅₀ values in the 250–300 nM range.²⁶ In the current study we generated analogs with up to ~100-fold greater potency including low to sub-nanomolar broad spectrum activity against Old (pLASV) and New World (pMACV, pJUNV and pTCRV) pseudotyped viruses that translate to activity against replicative Lassa and Tacaribe viruses. Attempts to substantially increase solubility while maintaining broad-spectrum potency and other drug like properties met with limited success. However, novel R¹, R⁴ and R⁶ substituent combinations that improved the HLM metabolic half-lives from < 10 min in our initial broad-spectrum compounds to > 45 min were identified for a subset of analogs. Those compounds exhibiting broad-spectrum arenavirus activity with EC₅₀ values ≤10 nM against each of the pseudotyped viruses and HLM half-lives of > 45 min (Table 9) may be prioritized for further evaluation and advancement.