Physicochemical Property-Driven Optimization of Diarylaniline Compounds as Potent HIV-1 Non-Nucleoside Reverse Transcriptase Inhibitors

Abstract

Using physicochemical property-driven optimization, twelve new diarylaniline compounds (DAANs) (7a–h, 11a–b and 12a–b) were designed and synthesized. Among them, compounds 12a–b not only showed high potency (EC₅₀ 0.96–4.92 nM) against both wild-type and drug-resistant viral strains with the lowest fold change (FC 0.91 and 5.13), but also displayed acceptable drug-like properties based on aqueous solubility and lipophilicity (LE > 0.3, LLE > 5, LELP < 10). The correlations between potency and physicochemical properties of these DAAN analogues are also described. Compounds 12a–b merit further development as potent clinical trial candidates against AIDS.

Non-nucleoside reverse transcriptase inhibitors (NNRTIs) with diverse structures are a key component of antiretroviral therapy (ART) for HIV infection and AIDS, because they exhibit high efficacy and low toxicity, as well as synergistic activity in combination with other anti-HIV drugs.^1,2 Two new-generation NNRTIs, etravirine (TMC125, 1a) and rilpivirine (TMC278, 1b) (Fig. 1), which were recently approved by the FDA for anti-AIDS therapy, have much better potency and pharmacological profiles than early NNRTIs, such as nevriapine, delavirdine, and efavirenz, and can efficiently inhibit a broad spectrum of drug-resistant viral strains.³ However, clinical trials revealed novel resistance mutations⁴ conferred against drugs 1a and 1b, which are both diarylpyrimidine (DAPY) compounds, similar to the early NNRTIs. However, these newly produced resistance mutations differ from those affecting the early NNRTIs and from each other, suggesting that a subtle structural difference between the drugs was sufficient to cause the occurrence of distinct HIV mutations. This discovery underscores the necessity for developing new NNRTI drugs with diverse scaffolds in order to provide more choices for AIDS treatment and overcome new resistance mutants. Accordingly, a number of new-generation NNRTI agents with diverse structures have been discovered⁵ and are currently undergoing preclinical and clinical trials.

Figure 1.

Next-generation NNRTI drugs, diarylaniline leads (DAANs), and new DAAN analogues

In our prior studies, several diarylanilines (DAANs) were identified as novel class of HIV-1 non-nucleoside reverse transcriptase inhibitor (NNRTI) agents with low nanomolar anti-HIV potency against wild-type and mutated viral strains^6,7, both comparable to and better than new-generation NNRTI drugs 1a and 1b. These DAANs are shown in Figure 1 as leads 2a and 2b. However, their poor aqueous solubility (< 1 μg/mL) resulted in very low absorption in vivo. To improve molecular aqueous solubility, several polar groups^8,9, including carboxyl, ester, amide, hydroxyl, and CF₃, were introduced at the R¹ group on the central phenyl ring, a point known to be modifiable for anti-HIV potency, while also associated with molecular physicochemical properties. These efforts led to the discovery of hydroxymethyl-DAAN 2c (Fig. 1) with high potency against wild-type and multi drug-resistant viral strains (EC₅₀ 0.53 nM and 0.4 nM, respectively) and improved aqueous solubility of 3.23 μg/mL at pH 7.4 and 20.9 μg/mL at pH 2.0. Unfortunately, 2c displayed low oral bioavailability (F% 6.10) in pharmacokinetics assays in vivo. Herein, we have again modified the DAAN compounds to identify potential drug candidates with balanced potency and a desirable absorption, distribution, metabolism, and excretion (ADME) profile.

To explore the correlations between potency and physicochemical properties associated with ADME profile, we continued to focus on the R¹ substituent on the central phenyl ring. In our newly designed series of DAAN analogues (7a–h, 11a–b, 12a–b), R¹ was altered to alkylamines or alkoxyethers with different shapes, lengths or volumes. After anti-HIV evaluations, the new active DAAN compounds were further assessed for multiple physicochemical properties, including aqueous solubility and lipophilicity, as estimated by log P. Apart from aqueous solubility, lipophilicity is another major physicochemical property that contributes to potency, affects compound solubility, determines the passive permeability of small molecules through biological membranes, impacts drug metabolism and pharmacokinetics, and influences adverse effects and compound-related toxicity. Most recently, new lipophicilic parameters, i.e., lipophilic efficiency (LE), lipophilic ligand efficiency (LLE),¹⁰ and ligand-efficiency-dependent lipophilicity (LELP),¹¹ have been proposed and applied in many medicinal chemistry programs^12,13,14 to efficiently guide lead optimization. Herein, the synthesis, anti-HIV potency, and assessments of multiple physicochemical properties of three series of new DAAN compounds (7a–h, 11a–b, 12a–b) are reported. The results will be helpful in guiding our further lead optimization aimed at the discovery of new clinical trial candidates as potent anti-AIDS drugs.

As shown in Scheme 1, target DAAN compounds 7a–h were prepared through a short synthetic route, starting from commercially available 4-hydroxy-3,5-dimethylbenzonitrile (3). The previously synthesized intermediate 5-chloro-N¹-(4-cyanophenyl)-4-methoxycarbonyl-2-nitroaniline (4)⁸ was coupled with 3 in the presence of potassium carbonate in DMF under 120 °C for 6 h to afford 2,4-diarylnitrobenzene 5. By using lithium borohydride (LiBH₄), the ester group on the central phenyl ring in 5 was reduced to a hydroxymethyl group in the key intermediate 6a. Subsequently, 6a was treated with 2,4,6-trichloro-[1,3,5]triazine followed by nucleophilic substitution with methylamine, cyclopropanamine, 3-aminopropan-1-ol, or 1-methyl-piperazine to produce the corresponding compounds 6b–e, respectively, with different alkylamines at the R¹ position. Alternatively, 6a was reacted with isopropanol or methanol in the presence of bismuth chloride (BiCl₃) to afford the corresponding alkoxymethyl-DAAN compounds 6f and 6g. Furthermore, the hydroxyl group in 6a was esterified with acetic anhydride to yield compound 6h. Finally, the nitro group on the central ring of 6a–h was reduced via catalytic hydrogenation in the presence of Pd-C (10%) in either EtOAc or anhydrous ethanol to furnish new DAAN compounds 7a–h. The structures of these new DAAN compounds were identified from proton NMR and MS spectra.¹⁵

Scheme 1.

i) K₂CO₃/DMF, 120 °C, 6 h; ii) LiBH₄, THF/MeOH, 0 °C, 7 h; iii) 2,4,6-trichloro-[1,3,5]triazine, DMF/CH₂Cl₂, rt, 4 h; iv) amine, THF, 0 °C, 0.5 h; v) ROH/BiCl₃, CH₂Cl₂/CCl₄, rt, 4 h; (vi) Ac₂O, 100 °C, Microwave, 5 min; vii) H₂/Pd-C in EtOAc or EtOH.

Newly synthesized DAAN compounds 7a–h were initially evaluated against wild-type HIV-1 (IIIB) replication in MT-2 cells in parallel with drug 1b. The data are presented in Table 1. As expected, most new DAANs, except 7e with a bulky N-methylpiperazinyl group at the R¹ position (EC₅₀ 170 nM), exhibited low nanomolar potency with EC₅₀ values ranging from 1.06 to 14 nM and high selective index (SI) values of 1,142 to 114,019. The new 7-series compounds were also evaluated against K103N/Y181C mutant-derived, NNRTI-resistant viral strain A17. However, their potencies against the wild-type viral strain were clearly reduced, as demonstrated by EC₅₀ values of greater than 33 to 2,000 nM.

Table 1.

Anti-HIV data of new DAANs against wild-type and resistant viral strains^a

	R1	EC50 (nM) IIIBb	CC50 (μM)	SI	EC50 (nM) A17c	FCd
7a	OH	1.07 ± 0.25	122	114,019	33.1 ± 4.05	30.9
7b	NHMe	9.94 ± 1.57	12.7	1,278	115 ± 3.10	11.6
7c	NHCH(CH2)2	10.6 ± 0.64	12.1	1,142	130 ± 2.87	12.3
7d	NH(CH2)3OH	6.10 ± 1.10	15.1	2,475	>2000	>328
7e	N(CH2CH2)2NMe	170 ± 5.60	96.9	570	>2000	>11.8
7f	OCHMe2	14.0 ± 1.10	89.4	6,386	315 ± 25.7	22.5
7g	OMe	3.54 ± 0.62	146	41,243	298 ± 133	84.2
7h	OAc	1.06 ± 0.01	113	106,604	374 ± 111	353
11a	OMe	5.74 ± 2.20	50.3	8,763	14.7 ± 0.74	2.6
12a	OMe	3.25 ± 0.23	>200	>61,538	2.95 ± 0.50	0.91
11b	OAc	0.82 ± 0.07	96.3	117,439	36.50 ± 1.78	44.5
12b	OAc	0.96 ± 0.12	44.8	46,667	4.92 ± 0.12	5.1
1b	TMC278	0.49 ± 0.17	90.7	185,102	9.03 ± 0.74	18.4

^aExperiments performed at least in triplicate in MT-2 cells and data presented as the mean ± SD.

^bHIV-1 wild-type virus.

^cDrug-resistant virus from NIH with mutated K103N and Y181C in the NNRTI binding pocket.

^dFold change resistance.

Based on previous SAR results,⁹ we then designed and synthesized two pairs of compounds 11a–b and 12a–b with a para-cyanovinyl and para-cyanoethyl (R²) group, respectively, on the phenoxy ring (C-ring), as shown in Scheme 2. Similarly to the preparation of 7g and 7h, methoxymethyl-DAAN 9 and acetoxymethyl-DAAN 10 were synthesized from N¹-(4-cyanophenyl)-5-(4′-cyanovinyl-2′,6′-dimethylphenoxy)-4-hydroxymethyl-2-nitroaniline (8).⁹ Subsequently, the nitro group in 9 and 10 was reduced with iron powder in the presence of NH₄Cl to afford corresponding para-cyanovinyl-DAAN compounds 11a and 11b,¹⁵ respectively, while the nitro group (R¹) and the conjugated double bond in the cyanovinyl group (R²) of 9 and 10 were reduced simultaneously using catalytic hydrogenation with Pd/C to produce para-cyanoethyl-DAAN compounds 12a and 12b.¹⁵ The two pairs of compounds, 11a–b and 12a–b, exhibited high potency against wild-type HIV-1 replication with sub- to low nanomolar EC₅₀ values ranging from 0.83 to 5.74 nM, and were as or more potent than 7g and 7h, regardless of whether R² was p-cyanovinyl or p-cyanoethyl. More importantly, compounds 12a–b showed high potency against resistant viral strain A17. Specifically, cyanoethyl-DAAN 12a (EC₅₀ 2.95 nM) was more potent than cyanovinyl-DAAN 11a (14.7 nM), while both were more potent than cyano-DAAN compound 7g (298 nM). Similar differences in potency were observed when comparing acetoxymethyl-DAAN compounds 12b (4.92 nM), 11b (36.5 nM), and 7h (374 nM). These results clearly demonstrate that the R² group on the phenoxy ring (C-ring) directly affects molecular potency against wild-type, as well as resistant, viral strains. A cyanoethyl group, which is more flexible due to its linearity, was more favorable than a cyanovinyl or cyano group. Notably, highly potent 12a and 12b had low fold change (FC) between A17 and wild-type IIIB virus with FC values of 0.91 and 5.13, respectively, much lower than that of 1b (FC 18.4) in the same assay.

Scheme 2.

i) ROH/BiCl₃, CH₂Cl₂/CCl₄, r.t. 4 h; ii) Ac₂O, 100 °C, Microwave, 5 min; iii) Fe, NH₄Cl, THF/MeOH/H₂O, reflux, 3 h; iv) H₂/Pd-C in EtOH.

Next, several physicochemical properties of newly generated DAANs (EC₅₀ < 11 nM) (7a–d, 7g–h, 11a–b, 12a–b) and drug 1b were assessed, and the resulting data are summarized in Table 2. Aqueous solubility was measured by HPLC at pH 2.0 and 7.4 to reflect the physiological conditions encountered by these compounds in stomach and plasma, respectively. As expected, alkylamine-DAAN compounds 7b, 7c, and 7d displayed greatly improved solubility at both pH 2.0 (263, 285, and 290 μg/mL, respectively) and pH 7.4 (159, 13, and 236 μg/mL, respectively) compared with drug 1b (pH 2.0, 74 μg/mL; pH 7.4, 0.29 μg/mL). Thus, the introduction of suitable alkylamino groups at the R¹ position could greatly improve the molecular aqueous solubility. Active compounds hydroxymethyl-DAAN 7a (R¹ = CH₂OH), methoxymethyl-DAANs 7g, 11a, and 12a (R¹ = CH₂OMe), and acetoxymethyl-DAANs 7h, 11b, and 12b (R¹ = CH₂OAc) also demonstrated improved aqueous solubility at pH 2.0 (1.7–9.10 μg/mL), but not at pH 7.4 (< 1 μg/mL). For oral drug candidates, better aqueous solubility at pH 2.0 is desirable to enhance absorbability in the stomach.¹⁸ Meanwhile, we observed that all 10 new active compounds had lower melting points than 1b. This difference might be explained by the assumption that the R¹ group on the central phenyl ring might disrupt molecular planarity and crystal packing,¹⁹ which could also enhance molecular aqueous solubility. To estimate molecular lipophilicity, the log P parameters of these active compounds were measured by HPLC at pH 7.4.⁹ The experimental log P values fell within an acceptable range of 1.80–4.40, which is consistent with the measured aqueous solubilities, and showed the same trend as the clog D values predicted by ACD software. Additionally, topological polar surface area (tPSA) parameters of all active compounds were calculated by ChemDraw Ultra 12.0 and met the criterion²⁰ of < 140 Ų for potential oral drug candidates.

Table 2.

Physicochemical parameters of new active DAAN compounds

	Aqueous solubility (μg/mL)a		mp °C	Log Pb pH 7.4	clogDc	tPSAd (Å2)	Lipophilic efficiency indices16,17
	pH 2.0	pH 7.4					pEC50e	LEf	LLEg	LELPh
7a	9.12	0.86	230–2	3.57	3.23	115.1	8.97	0.42	5.40	8.50
7b	263	159	202–3	1.86	3.72	106.9	8.00	0.37	6.14	5.03
7c	286	13.0	72–3	3.84	4.08	106.9	7.97	0.34	4.13	11.30
7d	290	236	76–8	1.81	3.31	127.1	8.21	0.34	6.40	5.32
7g	3.75	0.07	176–8	3.84	4.16	104.1	8.45	0.39	4.61	9.85
7h	16.7	0.20	161–3	4.40	4.12	121.2	8.97	0.38	4.57	11.60
11a	1.70	0.11	171–2	4.13	4.48	104.1	8.24	0.35	4.11	11.80
12a	1.82	0.66	104–5	3.25	3.87	104.1	8.49	0.36	5.24	9.03
11b	1.87	0.49	194–6	4.48	4.45	121.2	9.08	0.37	4.60	12.20
12b	2.63	1.86	164–6	2.84	3.84	121.2	9.02	0.36	6.18	7.92
1b	86.8	0.24	246–8	> 5	3.62	97.4	9.31	0.46	5.69i	7.87i

^a,bMeasured by HPLC in triplicate.

^cPredicted by ACD software.

^dTopological polar surface area predicted by ChemDraw Ultra 12.0.

^eNegative logarithm values of potency converted from experimental data against wild-type virus IIIB shown in Table 1.

^fCalculated by the formula −ΔG/HA_{(non-hydrogen atom)}, in which normalizing binding energy ΔG = −RT lnK_d, presuming EC₅₀ ≈ K_d.

^gCalculated by the formula pEC₅₀ − log P.

^hDefined as a ratio of measured log P and LE.

ⁱData calculated with clog D.

To estimate the possible ADME profiles and potential of our drug candidates, we focused next on their lipophilic indices, including lipophilic efficiency (LE), lipophilic ligand efficiency (LLE), and ligand-efficiency-dependent lipophilicity (LELP). Defined as the difference between the negative logarithm of the measured potency (pEC₅₀) and log P, LLE quantifies the contribution of lipophilicity to potency, whereas LELP¹⁷ correlates with ADME properties. Thus, compounds with a low LELP value would most likely have a high chance of passing all ADME and safety criteria,²¹ while compounds with high LELP values (typically > 10) would have a higher propensity to fail because of ADME and safety risks. Accordingly, lipophilic parameters of the new active DAANs were calculated from their experimental EC₅₀ and log P values by the formulas cited at the bottom of Table 2. Consequently, compounds 7a, 7b, 7d, 12a, and 12b met acceptable levels for all three ligand lipophilic-efficiency indices (LE > 0.3, LLE > 5, LELP < 10),²¹ while the remaining compounds in Table 2 did not, having either LLE values lower than 5 or LELP values higher than 10. Among the five promising compounds, 7b and 7d showed higher aqueous solubility at different pH conditions as well as lower log P and LELP values than the other compounds, suggesting better ADME profiles. On the other hand, the 7 series of DAANs were not efficient against resistant viral strain A17. Compounds 12a and 12b did show more balanced potency between the HIV-1 wild-type and resistant viral strains, as well as met acceptable lipophilic criteria, as determined by the LE, LLE and LELP indices. However, the aqueous solubility of both compounds was obviously lower than that of either 7b or 7d. Thus, to avoid the risk of potential oral absorption, these results suggested that the 12a–b pair requires additional optimization to improve aqueous solubility.

In summary, three series of new DAAN compounds (7a–h, 11a–b, and 12a–b) with interchangeable R¹ and R² modification/optimization were successfully synthesized as part of our ongoing anti-HIV NNRTI program. Our current physicochemical property-driven optimization resulted in the discovery of two promising compounds, 12a and 12b, with high potency against wild-type and drug-resistant viral strains, low nanomolar EC₅₀ values (0.96–4.92 nM), low fold change resistance (FC 0.91 and 5.13), and acceptable lipophilicity, as demonstrated by meeting acceptable values for all lipophilic parameters (LE > 0.3, LLE > 5, LELP < 10; log P < 5, tPSA < 140 Ų)²¹, even though their aqueous solubility needs further improvement. Optimization of the series 7 compounds revealed that (1) the presence of an alkylamine substituent at the R¹ position can greatly improve molecular aqueous solubility (see 7b–d) without loss of antiviral potency, (2) a bulky R¹ group can result in substantially impaired antiviral activity (see 7e), and (3) introducing an H-bond acceptor or donor at the R¹ position (such as 7a–b, 7d–g) might regulate the molecular lipophilicity to meet desired drug criteria. Our optimization efforts at the R² position on the phenoxy ring (C-ring) indicated that a more flexible and longer linear cyanovinyl (11 series) or, preferably, cyanoethyl (12 series) substituent, rather than a cyano group (7 series) is crucial for high potency against both wild-type and double-mutant drug-resistant viral strains (compare 7g–h, 11a–b, and 12a–b). Consequently, a number of compounds from this series are being considered for in vivo pharmacokinetic evaluation, and the results will be reported later.