Identification of Quinolinones as Antivirals against Venezuelan Equine Encephalitis Virus

ABSTRACT

Venezuelan equine encephalitis virus (VEEV) is a reemerging alphavirus that can cause encephalitis resulting in severe human morbidity and mortality. Using a high-throughput cell-based screen, we identified a quinolinone compound that protected against VEEV-induced cytopathic effects. Analysis of viral replication in cells identified several quinolinone compounds with potent inhibitory activity against vaccine and virulent strains of VEEV. These quinolinones also displayed inhibitory activity against additional alphaviruses, such as Mayaro virus and Ross River virus, although the potency was greatly reduced. Time-of-addition studies indicated that these compounds inhibit the early-to-mid stage of viral replication. Deep sequencing and reverse genetics studies identified two unique resistance mutations in the nsP2 gene (Y102S/C; stalk domain) that conferred VEEV resistance on this chemical series. Moreover, introduction of a K102Y mutation into the nsP2 gene enhanced the sensitivity of chikungunya virus (CHIKV) to this chemical series. Computational modeling of CHIKV and VEEV nsP2 identified a highly probable docking alignment for the quinolinone compounds that require a tyrosine residue at position 102 within the helicase stalk domain. These studies identified a class of compounds with antiviral activity against VEEV and other alphaviruses and provide further evidence that therapeutics targeting nsP2 may be useful against alphavirus infection.

INTRODUCTION

Alphaviruses are positive-strand RNA viruses in the family Togaviridae (1). The Alphavirus genus includes more than 30 species, most of which are maintained in nature in transmission cycles between mosquito vectors and vertebrate hosts that include humans, nonhuman primates, horses, birds, and rodents (2). Following transmission to humans, a variety of alphaviruses can cause serious disease that ranges from severe, life-threatening encephalitis to debilitating acute and chronic arthritis/arthralgia. For example, alphaviruses such as eastern (EEEV), western (WEEV), and Venezuelan (VEEV) equine encephalitis viruses have the potential to cause fatal neuroinvasive disease in humans, with case fatality rates ranging from 1% to >50% (3). Other alphaviruses, including chikungunya (CHIKV), Mayaro (MAYV), o’nyong-nyong (ONNV), Ross River (RRV), and Una (UNAV) viruses lead to incapacitating arthralgia that can result in long-term disability (4). Alphaviruses also have the capacity to cause explosive epidemics. In 1995, 75,000 to 100,000 human VEEV cases occurred in Venezuela and Colombia, and more than 300 fatal encephalitis cases were recorded (5). Since 2004, CHIKV has infected millions of people and expanded into Europe, Asia, the Pacific region, and the Americas (6–9). Joint pain, swelling, and stiffness, as well as tenosynovitis, can endure for months to years after CHIKV infection (10–15). As chronic CHIKV disease is debilitating, large epidemics have severe economic impact (16, 17). No approved vaccines or specific therapies are available to prevent or treat any disease in humans caused by alphaviruses. Thus, there is a pressing need to identify therapeutic targets and new classes of therapeutics that could be used against alphavirus infection.

The alphavirus genome is a positive-sense single-stranded RNA that is ∼12 kb in length and contains two open reading frames (ORFs) (18). The first ORF encodes four nonstructural proteins (nsP1 to nsP4) that form the replicase complex, which synthesizes the viral genomic and subgenomic RNA species. The second ORF, translated from the subgenomic viral mRNA, encodes a polyprotein that is posttranslationally processed into six proteins, including three major structural proteins termed capsid, E2, and E1. These ORFs are flanked by 5′ and 3′ untranslated regions (UTRs) and are separated by an internal noncoding region.

Alphaviruses enter cells via receptor-mediated endocytosis (19). Fusion of the viral and endosomal membranes, mediated by the viral E1 fusion peptide, results in the release of viral nucleocapsids into the cell cytosol (20–22). Following nucleocapsid disassembly, the nonstructural ORF of the viral genomic RNA is translated, thus forming the viral replicase, which synthesizes full-length negative-sense RNAs that serve as a template for synthesis of full-length and subgenomic positive-sense mRNAs (23). Translation of the subgenomic mRNA produces the structural polyprotein that is proteolytically cleaved by virus and host proteinases in a process that releases the viral capsid protein and permits trafficking of the E1 and E2 glycoproteins to the plasma membrane (24). The viral capsid protein nucleates the assembly of ribonucleoprotein complexes that interact with E1-E2 heterodimers, leading to the budding of new virus particles (25).

In previous studies, a high-throughput screening campaign of 348,140 compounds for antiviral activity against VEEV strain TC-83, a live attenuated vaccine strain, was completed (26). Here, we report the identification of a thiourea quinolinone compound (SRI-33394) from this screen with potent antiviral activity against VEEV. Limited chemical optimization was performed to identify more potent analogs in this chemical series. Time-of-addition studies with SRI-33394 and additional analogues suggest that these compounds inhibit the early-to-mid stage of the viral replication cycle prior to virion assembly and release. Consistent with these data, we identified two independent substitution mutations in the nsP2 gene (VEEV_TC-83; Y102C or Y102S) that conferred resistance to the quinolinones. CHIKV showed limited sensitivity to this compound series, and the virus naturally contains a lysine (K) residue at this position; however, mutation of this position (nsP2 K102Y) enhanced antiviral sensitivity. Molecular modeling identified a reasonable docking site within the helicase stalk domain that utilizes Y102 for pi-pi bonding between the Tyr residue and compound ring structure. This work has identified a class of compounds with antiviral activity against VEEV and other alphaviruses, and we suggest that therapeutics targeting nsP2 may be useful against alphavirus infection.

RESULTS

High-throughput screen identifies compound SRI-33394 with inhibitory activity against VEEV_TC-83.

As previously reported (26, 27), a total of 348,140 compounds from the NIH Molecular Libraries Small Molecule Repository (MLSMR) library were screened for the capacity to prevent a cytopathic effect (CPE) following infection of Vero cells with VEEV_TC83. The ability of each compound at a single dose (20 μM) to inhibit CPE caused by VEEV_TC83 infection was measured at 72 hpi. A total of 3,608 hits were identified with antiviral activity against VEEV, which was downselected to 453, based on their ability to inhibit viral replication by >30% (26). After data mining of the previously run high-throughput screen (HTS) against VEEV (27), a set of compounds (>60% inhibition) were tested against VEEV in a CPE-based dose response assay using normal human dermal fibroblasts (NHDFs). Active compounds (50% inhibitory concentration [IC₅₀] < 15 μM) identified in this CPE assay were further tested in a virus titer reduction (VTR) assay at 10 μM using NHDFs. SRI-33394, a quinolinone compound, showed an 8.98-log reduction in virus titer at this concentration, and the IC₉₀ was calculated to be 0.77 μM (Fig. 1A). Based on these data, SRI-33394 was selected as the lead compound and was prioritized for structure-activity-relationship (SAR) studies. To confirm the antiviral activity of SRI-33394 and any analogs resulting from the SAR studies against VEEV, compounds were tested in parallel assays for anti-VEEV potency and cytotoxicity using dose-response curves on NHDFs. In dose-response assays, compounds were added to NHDF cells at concentrations ranging from 100 to 0.19 μM 1 h prior to infection with VEEV_TC83. Supernatants were collected at 24 hpi, and titer was determined by limiting dilution plaque assays on confluent monolayers of Vero cells; effective concentration was reported as an IC₉₀ value. Compounds were also screened in a CPE assay that employed telomerized human foreskin fibroblasts that lack IRF3 (THF-ΔIRF3), for determination of an IC₅₀ value. In parallel, compounds were screened for cytotoxicity after 72 h of treatment on NHDF cells, using the Promega CellTiter-Glo luminescent cell viability assay.

FIG 1.

Structure-activity relationship studies with SRI-33394, a quinolinone with antiviral activity against VEEV. (A) Lead compound SRI-33394 has potent antiviral activity against VEEV. (B) Structure-activity-relationship (SAR) study to replace sulfur in SRI-33394 with oxygen. (C) Analogs with substitutions at R¹, R², R⁴, and R⁵. (D) SRI-36959 with cyclization of the ethyl-N,N-dimethyl moiety to pyrrolidines at R⁴. (E) SRI-34329 with hydrogen on the quinolinone ring switched to a methyl group. (F) Analogs with substitutions at R¹ to R⁵.

Chemical synthesis of quinolinone analogs.

Although the aqueous solubility at pH 7.4 of SRI-33394 was acceptable (31 μM), the liver microsomal stability was found to be poor (mouse liver microsomes [MLM] half-life [t_1/2] = 2 min, human liver microsomes [HLM] t_1/2 = 11 min) (Fig. 1A). Analogs were thus designed and synthesized, as summarized in Fig. 1, to maintain potency while increasing their drug-like properties, such as microsomal stability and solubility. A few representative examples are shown in Table 1 (detailed synthetic procedures are provided in Materials and Methods). In general, 2-substituted quinolinones (Table 1, entry 1) were synthesized according to the literature (28) and were subjected to reductive amination (29) to yield intermediates (Table 1, entry 2), which were subsequently condensed with substituted isocyanates or thioisocyanates to yield structurally diverse ureas (X = O) and thioureas (X = S) (Table 1, entries 2 to 11). Compounds with a VTR of >2.0 log at 10 μM concentration were further evaluated in a 10-point dose-response VTR assay (described above), which resulted in IC₉₀ values. Compounds with an IC₉₀ of <1.5 μM and a VTR of >2.0 log were then selected for in vitro studies to determine microsomal stability and solubility.

TABLE 1.

Antiviral data on quinolinones^a

^aReagents and conditions for synthesis of target compounds are shown in the diagram. (i) Alkyl iodide, K₂CO₃, dimethylformamide (DMF), room temperature (RT), 4 h; (ii) R⁴NH₂, NaBH₄, methanol (MeOH); (iii) R⁵NCX, RT, 18 h.

^bIC₅₀ determined by cytopathic effect (CPE) assay on THF-ΔIRF3 cells.

^cVTR, virus titer reduction.

^dIC₉₀ determined by VTR assay on normal human dermal fibroblasts (NHDFs). ND, not determined.

^eCC₅₀, 50% cytotoxic concentration of compounds determined using NHDFs.

The first SAR study included replacement of sulfur with oxygen in SRI-33394 (Fig. 1B). A set of 19 urea-containing analogs were synthesized and tested for their antiviral activity by VTR and CPE assays (some are represented as Table 1, entries 2 to 6). All 19 urea analogs were found to be inactive, as they did not exhibit in vitro antiviral activity (>2 log) in a VTR assay at 10 μM compound concentration and had an IC₅₀ of >25 μM in the CPE assay (Table 1, entries 2 to 6). Henceforward, the thiourea core was maintained for the rest of the SAR studies. In a systematic exploration, different areas of the molecule were modified, which resulted in various thiourea analogs with substitutions at R¹ to R⁵. R¹ and R² were kept as either hydrogen or methyl. SRI-33394, the original hit, possessed an N,N-diethylamino moiety at R⁴ and a methylfuran moiety at R⁵. One group of analogs included substituting different saturated and unsaturated heterocycles and N,N-dialkylamino moieties in place of R⁴ while keeping the methylfuran group intact at R⁵. Another group involved exploring substituted benzyl, phenyl, alkyl, and cycloalkyl groups in place of R⁵, keeping R⁴ as N,N-diethylamine constant. A total of 48 analogs of SRI-33394 were synthesized in these three areas (Fig. 1C), and some representative examples are shown in Table 1. We observed that removing both of the methyls on the aromatic quinolinone ring was detrimental to the activity, as seen in compound 3f (Table 1, entry 7). Switching R⁴ to a nonbasic moiety, such as a tetrahydrofuranyl (3g) or methylfuranyl (3h) group, decreased the VTR (Table 1, entries 8 and 9). The basic nitrogen-containing moiety at R⁴ is thus essential to maintain the activity. Switching to different substituents at R⁵ rendered the compounds inactive (data not shown). Hence, we maintained the methylfuran moiety constant at R⁵ in order to retain activity in subsequent analogs. Next, the cyclization of the ethyl-N,N-dimethyl moiety to pyrrolidines at R⁴ (Fig. 1D) was carried out to give compounds 3i and 3j (Table 1, entries 10 and 11). Incorporating N-methyl pyrrolidines at R⁴ in compound 3i (SRI-36959) maintained antiviral activity (IC₉₀ = 0.78 μM; VTR= 6.83 log), compared to HTS hit SRI-33394, whereas pyrrolidine-containing compound 3j (Table 1, entry 11) lost ∼2-fold antiviral potency (IC₉₀). It should be noted that monomethylation in 3i and 3j (R¹ = Me and R² = H) was also well tolerated. While these substitutions improved the aqueous solubility of 3i (SRI-36959) at pH 7.4 (46 μM), the microsomal stability was still poor (MLM t_1/2 = 2 min and HLM t_1/2 = 3 min). When the hydrogen on the quinolinone ring of SRI-33394 was switched to methyl (SRI-34329, compound 3k; Fig. 1E), the potency increased (Table 1, entry 12; IC₉₀ = 0.12 μM, VTR = 5.96 log). However, in vitro absorption, distribution, metabolism, and excretion (ADME) studies showed that the stability of the compound remained poor (MLM t_1/2 = 1 min, HLM t_1/2 = 8 min). A total of 10 analogs with different substitutions at R¹ to R⁵ were explored next (Fig. 1F). We found that while alkyl groups were tolerated (Table 1, entries 13 and 14), the best activity was observed with SRI-34329. All of the substitutions explored in R⁵ rendered the scaffold inactive or cytotoxic (Table 1, entries 15 to 17).

SRI-33394 has potent antiviral activity against both attenuated and virulent VEEV strains.

To further evaluate the antiviral activity of SRI-33394 and SRI-34329, these compounds were tested for anti-VEEV_TC83 activity using the previously described dose titration assay and cytotoxicity in both NHDF cells and Aedes albopictus C6/36 cells. Both compounds exhibited potent antiviral activity against VEEV_TC83 in NHDF cells (SRI-33394 = 0.77 μM and SRI-34329 = 0.12 μM) (Fig. 2A) and more limited antiviral activity in C6/36 cells (Fig. 2B). In these experiments, cytotoxicity (50% cytotoxic concentration [CC₅₀]) was detected at 16.7 μM for each compound (Fig. 2C). The analog SRI-34329 exhibited increased anti-VEEV activity compared to that of the parental compound, SRI-33394, in both cell types (Table 1). Collectively, these data demonstrate that the quinolinone compounds maintain antiviral activity against VEEV_TC83 during viral replication in disparate vertebrate and invertebrate host cells.

FIG 2.

Antiviral activity of SRI-33394 and SRI-34329. At 1 h prior to infection, (A) normal human dermal fibroblasts (NHDFs) or (B) Aedes albopictus C6/36 mosquito cells were pretreated with dimethyl sulfoxide (DMSO) or 2-fold serial dilutions of SRI-33394 or SRI-34329, ranging from 100 μM to 0.2 μM. Treated cells were infected with VEEV_TC-83 at a multiplicity of infection (MOI) of 1 PFU/cell. At 48 hpi, infectious virus in the supernatant was quantified by plaque assay, and the IC₉₀ was determined by nonlinear regression analysis. (C) For cytotoxicity assays, NHDFs were treated with dilutions of quinolinone compounds. At 24 h posttreatment, cells were analyzed using the CellTiter-Glo luminescent cell viability assay. Fifty percent cytotoxic concentration (CC₅₀) values were determined by nonlinear regression analysis. (D) NHDFs were pretreated for 1 h with DMSO, 10 μM SRI-33394, or 10 μM SRI-34329. Treated cells were then infected with VEEV_ZPC738 at an MOI of 3 PFU/cell. At 48 h postinfection (hpi), infectious virus in the supernatant was quantified by plaque assay. Data are combined from two independent experiments. Each bar represents the mean ± standard error of the mean (SEM). P values were determined by one-way analysis of variance (ANOVA) with Tukey’s multiple-comparison test. ****, P < 0.001.

VEEV_TC83 is an attenuated vaccine strain of VEEV generated by serial passage in tissue culture cells (30). To determine if SRI-33394 and SRI-34329 exhibit antiviral activity against a virulent strain of VEEV, we performed single-dose antiviral assays using VEEV_ZPC738, an enzootic subtype ID strain of VEEV (31). NHDFs were treated 1 h prior to infection with dimethyl sulfoxide (DMSO) or with 10 μM either SRI-33394 or SRI-34329 in DMSO. Treated NHDFs were inoculated with VEEV_ZPC738 (multiplicity of infection [MOI], 3 PFU/cell), and 2 h later, infection medium was removed, cells were washed, and fresh medium containing DMSO or 10 μM either SRI-33394 or SRI-34329 was added. At 48 hpi, the yields of infectious virus measured by plaque assay in culture supernatants of cells treated with SRI-33394 or SRI-34329 were reduced ∼10,000-fold compared with that in DMSO-treated cells (Fig. 2D). These data indicate that these quinolinone compounds have potent antiviral activity against attenuated and virulent strains of VEEV.

Breadth of SRI-33394 and SRI-34329 antiviral activity.

To evaluate the spectrum of antiviral activity of SRI-33394 and SRI-34329, we tested the compounds against a panel of alphaviruses in dose titration IC₉₀ assays. NHDFs were treated with SRI-33394 or SRI-34329 (100 to 0.19 μM) 1 h prior to infection with either CHIKV, MAYV, ONNV, RRV, or UNAV (MOI of 2 PFU/cell). The IC₉₀ values of SRI-33394 or SRI-34329 against VEEV TC-83 in NHDFs were 0.77 μM and 0.12 μM, respectively (Table 2). In contrast, against other arthritogenic alphaviruses IC₉₀ values were increased, resulting in values of >10 μM and >10 μM against CHIKV, >10 μM and 2.58 μM against MAYV, >10 μM and 2.70 μM against ONNV, and >10 μM and >10 μM against UNAV, for SRI-33394 and SRI-34329, respectively (Fig. 3A and B and Table 2). These results suggest that SRI-33394 and SRI-34329 have selective antiviral activity against VEEV.

TABLE 2.

Antiviral breadth against alphaviruses for quinolinone compounds

Alphavirus	IC90 (μM) for compound:
	SRI-33394	SRI-34329
CHIKV181/25	>10	>10
MAYVTRVL	>10	2.58
ONNV	>10	2.70
RRV	>10	2.52
UNAVMAC150	>10	>10
VEEVTC83	0.77	0.12

FIG 3.

Antiviral spectrum of quinolinones against selected alphaviruses. NHDFs were pretreated for 1 h with 2-fold serial dilutions of (A) SRI-33394 or (B) SRI-34329, ranging from 100 μM to 0.2 μM. Treated cells were then infected with CHIKV, MAYV, ONNV, RRV, or UNAV at an MOI of 3 PFU/cell. At 48 hpi, infectious virus in the supernatant was quantified by plaque assay.

Quinolinones inhibit an early step in viral replication cycle.

To begin to define the step of the VEEV replication cycle inhibited by the quinolinone family of compounds, we performed time-of-drug-addition studies with SRI-34329. NHDFs were inoculated with VEEV_TC-83 at 3 PFU/cell, 10 μM SRI-34329 was added to cultures at time points between −1 and 24 h postinfection (hpi), and yields of infectious virus in culture supernatants were quantified at 48 hpi by plaque assay. The addition of SRI-34329 within 12 h of virus inoculation potently reduced yields of infectious virus in culture supernatants (Fig. 4A). The addition of the compound at 24 hpi still reduced yields of infectious virus, but the effect was substantially reduced in magnitude (Fig. 4A). Consistent with these results, Northern blot analysis revealed that SRI-33394 and SRI-34329 inhibited the accumulation of both 49S genomic and 26S subgenomic viral RNAs in VEEV_TC-83-infected NHDFs (Fig. 4B), suggesting that quinolinones block viral RNA synthesis or a step in the replication cycle prior to RNA synthesis.

FIG 4.

Quinolinones inhibit VEEV_TC-83 replication at a postentry step. (A) NHDFs were infected with VEEV_TC-83 at an MOI of 3 PFU/cell and treated with DMSO or 10 μM SRI-34329 at the indicated time points. Infectious virus in the supernatant at 48 hpi was quantified by plaque assay. Data are derived from two independent experiments. **, P < 0.01; Mann-Whitney test. (B) NDHFs were pretreated for 1 h with DMSO, 10 μM SRI-33394, or 10 μM SRI-34329 and then infected with VEEV_TC-83 (MOI = 3 PFU/cell). At 12 hpi, total RNA was evaluated by Northern blotting analysis with digoxigenin (DIG)-labeled RNA probes specific for VEEV or human β-actin. Data shown are representative of two independent experiments.

Mapping of resistance mutations to quinolinones.

To determine whether the quinolinone compound family directly targets a VEEV-encoded function, the lead quinolinone compound SRI-34329 was used to select for drug-resistant virus. Triplicate series of VEEV_TC-83 were repeatedly passaged (in a blind manner) in NHDFs in the presence of DMSO or 10 μM SRI-34329 (Fig. 5A). At passage 3, one passage series (passage 3C) displayed evidence of resistance to SRI-34329 (Fig. 5B). To confirm this observation, the susceptibility of wild-type (WT) VEEV_TC-83, VEEV_TC-83 passaged in DMSO in parallel with the passage 3C resistant mutant (DMSO passage 3C VEEV_TC-83), and that of the potential passage-derived SRI-34329-resistant mutant (SRI-34329 passage 3C VEEV_TC-83), to restriction by treatment with SRI-34329 were compared. As shown in Fig. 5C, only SRI-34329 passage 3C VEEV_TC-83 displayed resistance, as yields of infectious virus were unaffected by the presence of 10 μM SRI-34329. In contrast, yields of infectious WT VEEV_TC-83 and DMSO passage 3C VEEV_TC-83 were potently reduced by 4 log with 10 μM SRI-34329 (Fig. 5C).

FIG 5.

Serial passage of VEEV_TC-83 in the presence of SRI-34329 yields an escape mutant. (A) VEEV_TC-83 was repeatedly passaged in triplicate through normal human dermal fibroblasts (NHDFs) in the presence of DMSO or 10 μM SRI-34329. (B) Titer of supernatant collected from the third passage of triplicate (3A to 3C) DMSO- or SRI-34329-treated cells was determined by plaque assay in the absence of drug. (C) NHDFs treated with DMSO or 10 μM SRI-34329 1 h prior to infection with VEEV_TC-83 or with replicate passage 3C samples from either DMSO or SRI-34329 serial passages at an MOI of 3 PFU/cell. At 48 hpi, virus in the supernatant was evaluated by plaque assay. P values were determined by unpaired t tests. **, P < 0.01; ***, P < 0.001.

Based on these results, we sequenced viral RNA isolated from stocks of WT VEEV_TC-83, DMSO passage 3C VEEV_TC-83, and SRI-34329 passage 3C VEEV_TC-83 by Illumina deep sequencing. Comparative analysis of the sequence reads from all three samples revealed 5 mutations present in SRI-34329 passage 3C VEEV_TC-83 in more than 5% of reads (>20,000 reads per site) (Table 3). One site, an adenine (A) at nucleotide position 1954, was mutated to either a guanine (84.4% of reads) or a cytidine (13.1% of reads) in 97.5% of sequence reads, resulting in an amino acid substitution from tyrosine (Y) to cysteine (C) or serine (S) at residue 102 in the nsP2 gene (Y102C or Y102S, respectively) (Table 3). The next most common mutation detected was a synonymous nucleotide change at position 5450 in the nsP3 gene (Table 3). To evaluate the role of the most frequent coding change detected in conferring resistance to SRI-34329, we introduced a Y102C mutation into nsP2 by site-directed mutagenesis into the VEEV_TC-83 genome and evaluated for effects on VEEV_TC-83 susceptibility to quinolinone compounds. As shown in Fig. 6A, VEEV_TC-83 nsP2 Y102C replication was completely resistant to 10 μM SRI-33394 or SRI-34329. In the absence of drug, we did not detect any difference in viral yields over time from NHDFs infected with WT or nsP2 Y1012C VEEV TC-83 (Fig. 6A), suggesting that this mutation does not substantially alter VEEV_TC-83 replication in cell culture. A dose-response assay revealed that the IC₉₀ of SRI-34329 is >50-fold higher against VEEV_TC-83 nsP2 Y102C compared with WT VEEV_TC-83 (Fig. 6B), confirming nsP2 as the target for SRI-34329. Moreover, VEEV_TC-83 nsP2 Y102C also displayed resistance to a number of additional analogs (see Fig. S1A to D in the supplemental material).

TABLE 3.

Mutations associated with viral resistance to SRI-34329

Nucleotide position	Mutation	Gene	Amino acida	% of reads
1218	G→T	nsP1	A392S	12.4
1954	A→G	nsP2	Y102C	84.4
1954	A→C	nsP2	Y102S	13.1
5450	T→C	nsP3	NA	76.0
10755	A→G	E1	K251R	12.5

^aNA, not applicable.

FIG 6.

VEEV_TC-83 nsP2 Y102C mutation confers resistance to SRI-33394 and SRI-34329. (A) Following a 1 h pretreatment with DMSO, 10 μM SRI-33394, or 10 μM SRI-34329, normal human dermal fibroblasts (NHDFs) were infected with wild-type (WT) or nsP2 Y102C VEEV_TC-83 at an MOI of 1 PFU/cell. At the indicated times postinfection, infectious virus in the supernatant was quantified by plaque assay. P values were determined by two-way ANOVA followed by Bonferroni’s multiple-comparison test. ***, P < 0.001; ****, P < 0.0001. (B) NHDFs were pretreated for 1 h with DMSO or with 2-fold serial dilutions of SRI-34329 ranging from 100 μM to 0.2 μM. Treated cells were then infected with WT or nsP2 Y102C VEEV_TC-83 at an MOI of 1 PFU/cell. At 48 hpi, infectious virus in the supernatant was quantified by plaque assay, and the percent infection relative to DMSO control was quantified.

The amino acid at nsP2 position 102 accounts for differences in alphavirus susceptibility to quinolinones.

Sequence alignment of the region within the N-terminal region of VEEV nsP2 predicted to be responsible for resistance to the quinolinone compounds revealed a Y at position 102 for VEEV and other encephalitic alphaviruses. In contrast, arthritogenic alphaviruses, including CHIKV, ONNV, and RRV, encode a lysine (K) residue at this position (Fig. 7A). To further evaluate the importance of the amino acid at position 102 for virus susceptibility to the quinolinones, we introduced a K102Y mutation into the nsP2 gene of CHIKV_181/25. Following inoculation of NHDFs, CHIKV nsP2 K102Y did not display growth differences compared with WT CHIKV (Fig. 7B). In dose-response assays, the CHIKV nsP2 K102Y mutant virus was more susceptible to SRI-34329 (IC₉₀ = 2.45 μM) than WT CHIKV (IC₉₀ = 7.7 μM) (Fig. 7C), suggesting that a Y at nsP2 position 102 renders alphaviruses more sensitive to this class of quinolinone compounds.

FIG 7.

Mutation of CHIKV nsP2 (K102Y) confers sensitivity to quinolinones. (A) Amino acid sequence alignment of alphavirus nsP2 residues 90 to 118 identifies a natural single-nucleotide polymorphism (SNP) at amino acid position 102 (red box). (B) An nsP2-K102Y mutation was introduced into the cDNA clone for CHIKV_181/25. NHDF cells were infected with WT and CHIKV_181/25 nsP2-K102Y virus at an MOI of 1 PFU/cell. Culture supernatants were collected at the indicated times, and titer was determined by plaque assay. (C) Triplicate wells of NHDFs were pretreated for 1 h with 2-fold serial dilutions of SRI-34329, ranging from 140 μM to 0.078 μM. Treated cells were then infected with WT CHIKV_181/25 or CHIKV_181/25 nsP2-K102Y at an MOI of 1 PFU/cell. At 24 hpi, infectious virus was quantified by plaque assay, and IC₉₀ values were determined by nonlinear regression analysis. Data are representative of two independent experiments.

Binding mode of quinolinones elucidated using computational docking studies.

Since position 102 of nsP2 was identified as a molecular target for SRI-34329, computational docking studies were performed to elucidate a possible binding mode of the quinolinone scaffold. Although there is no published structure of full-length VEEV nsP2, the residues between VEEV and CHIKV around position 102 are highly conserved, as illustrated in Fig. 7A. Thus, a homology model of VEEV nsP2 for compound docking studies was built using the CHIKV nsP2 crystal structure (32) as the template. A reasonable docked pose of SRI-34329 was found in the VEEV nsP2 homology model with four major interactions at a binding pocket around Y102 (Fig. 8A). Specifically, the quinolinone ring forms pi-pi stacking with Y102; the furan ring forms pi-pi stacking with F98; the basic nitrogen is protonated and forms a salt bridge with D116; and, last, the two methyl groups on the quinolinone ring can effectively reach to a hydrophobic region centered around L103 to form hydrophobic contacts. Not surprisingly, when SRI-34329 was docked to drug-resistant mutants nsP2 Y102C and Y102S, without the pi-pi stacking with Y102 as an anchor, it bound to random positions without the high binding specificity shown in the WT VEEV nsP2 docked pose (Fig. 8B). Docking studies also captured this trend in the case of CHIKV nsP2. A reasonable binding mode was found when SRI-34329 was docked to the drug-sensitive CHIKV nsP2 K102Y mutant, which resembled that of WT VEEV nsP2 (Fig. 8C). In contrast, the drug-resistant WT CHIKV nsP2 did not favor the binding of SRI-34329 according to docking. The K102 in WT CHIKV nsP2 competes with the basic nitrogen of SRI-34329 for binding with D116. Moreover, the formation of the salt bridge between K102 and D116 blocks SRI-34329 from binding to this position (Fig. 8D).

FIG 8.

Docked poses of lead SRI-34329 at VEEV and CHIKV nsP2 WT and position 102 mutants. Docked ligand is colored in green. Salt bridges, pi-pi stacking, and hydrophobic contacts are indicated by dashed cyan, green, and orange lines, respectively. (A) Pose at VEEV nsP2 WT. (B) Superposition of the ligands from docked poses of VEEV nsP2 Y102C (cyan) and Y102S (orange) onto the docked pose of VEEV nsP2 WT. (C) Pose at CHIKV nsP2 K102Y. (D) Superposition of the ligand from docked pose of CHIKV nsP2 K102Y (turquoise) onto the docked pose of CHIKV nsP2 WT.

Analogs were also docked to the same binding pocket of SRI-34329 at VEEV nsP2 to provide a molecular explanation of the SAR studies. As summarized in Fig. 9A, the docking results and anti-VEEV potencies are highly correlated. Specifically, deleting the two methyl groups (R¹ and R²), or the basic nitrogen (R⁴), or the furan ring (R⁵) causes loss of the favorable hydrophobic contacts, the salt bridge with D116, or the pi-pi stacking with F98 and, accordingly, results in the loss of potency. Consistently, analogs retained antiviral potency with R⁴ substitutions that still possess the basic nitrogen at the proper position to form the salt bridge (e.g., 3i and 3j) as shown in Fig. S1E. Notably, compound 3p is not potent, even though the replacement of furan with benzene should maintain pi-pi stacking. However, according to docking, the bare benzene ring is too hydrophobic to stay at the furan binding region, which is solvent accessible and has a mixture of polar and nonpolar residues. On the other hand, the nitrogen on the quinolinone ring (R³) did not form any critical interaction in the docked poses. Thus, substitutions at this position are tolerable. Besides the five R groups, another intriguing SAR finding is that analogs lost potency when the sulfur was replaced by an oxygen, namely, thiourea to urea. As shown in Fig. 9B, when the sulfur is replaced with oxygen, it forms an intramolecular hydrogen bond with the basic nitrogen, and thus reduces the latter’s chance to form the favorable salt bridge with D116. Therefore, the reduced specificity and affinity of urea analogs to VEEV nsP2 may explain their loss of potency. In retrospect, the in silico binding mode of the quinolinone scaffold at VEEV nsP2 is consistent with mutagenesis studies as well as the anti-VEEV SAR and therefore can be used to guide the rational design of this scaffold for further antiviral drug development.

FIG 9.

Docking results of SRI-34329 analogs. (A) Docked pose of compound 3d (green) at VEEV nsP2 WT. Hydrogen bonds, pi-pi stacking, and hydrophobic contacts are indicated by dashed black, green, and orange lines, respectively. (B) Summary of the SAR in line with docking results. Analogs belonging to each case are listed in either black (active and retention of certain interaction in docked poses) or red (inactive and loss of certain interaction in docked poses).

DISCUSSION

In this study, we report that several quinolinones potently restrict the replication of attenuated and pathogenic VEEV strains in multiple cultured cell types with limited cytotoxicity. The original quinolinone compound, SRI-33394, was identified from a library of 348,140 compounds (26). By performing thorough SAR studies, we developed a more potent quinolinone compound, SRI-34329, which blocked synthesis of both viral genomic and subgenomic RNAs at a postentry step.

To identify the molecular target of these small molecules, we selected drug-resistant viral strains and then used deep sequencing and reverse genetics approaches to define sequence polymorphisms that confer drug resistance. These efforts identified two unique resistance mutations in the nsP2 gene (Y102S/C) that endowed VEEV with resistance to these small molecules. Remarkably, a prior study found that a nsP2 Y102C mutation also conferred VEEV resistance against a quinazolinone compound termed CID15997213 (26). Similarly to the quinolinones identified in this study, CID15997213 potently restricted VEEV replication at a postentry step. Residue 102 of the alphavirus nsP2 protein is located in the stalk α-helix domain of the nsP2 N-terminal helicase, which forms part of a top cover for the single-stranded RNA (ssRNA) binding groove (32). Thus, the binding of small molecules at this site may interfere with RNA binding and/or unwinding activities of the nsP2 helicase, which are essential for viral RNA replication (32). Based on our docking results with the VEEV nsP2 helicase domain (nsP2h), SRI-34329 binds near the RNA binding groove but distal to the ATPase active site (Fig. 10). Although SRI-34329 does not directly occupy the RNA binding space, it forms multiple interactions with residues (e.g., D116 and L156) from region 1B (amino acids [aa] 110 to 175), which is known to be an important region for stabilizing the conformational rearrangements of nsP2h upon RNA binding (32, 33). Therefore, rather than inhibiting the ATPase activity of nsP2h, SRI-34329 is more likely to disrupt the RNA binding as a mechanism for inhibiting VEEV RNA replication. Future studies using purified nsP2 could be used to confirm the impact of quinolinones on not only RNA binding and RNA helicase activity but also on the RNA triphosphatase and nucleotide triphosphatase activities performed by the N-terminal domain of nsP2 (34, 35).

FIG 10.

Zoomed-out view of the docked pose of SRI-34329 at the VEEV nsP2 helicase. The helicase domain is represented in ribbons. Region 1B is colored in yellow. The single-strand RNA 14-mer and ADP from the CHIKV nsP2 helicase crystal structure (PDB identifier 6JIM) were superimposed to the analogous positions at the VEEV nsP2 helicase in silico homolog to indicate the RNA binding groove and ATPase active site.

In antiviral breadth studies, we found that the quinolinones identified in this study had inhibitory activity against other alphaviruses, including MAYV, ONNV, and RRV, although the potency was greatly reduced, and little to no inhibitory activity was found against CHIKV and UNAV. These findings are similar to antiviral breadth studies with CID15997213, which also displayed no detectable inhibitory activity against CHIKV replication (26). In contrast to VEEV, the arthritogenic alphaviruses encode a Lys residue at nsP2 position 102, suggesting that these viruses may be naturally resistant to these quinolinones and indicating that pi-pi stacking with Y102 is critical for stable compound docking and antiviral activity. To test this idea, we introduced a K102Y mutation into the nsP2 gene of CHIKV and found that the mutant virus had enhanced sensitivity to SRI-34329. Moreover, our molecular docking studies revealed that WT CHIKV nsP2 did not support SRI-34329 binding. However, a binding mode for SRI-34329 with CHIKV nsP2 harboring an nsP2 K102Y mutation was readily observed. These studies provide strong confirmation that the quinolinones identified in this study target the N-terminal helicase domain of nsP2.

In summary, our studies identified a new class of compounds with antiviral activity against VEEV and other alphaviruses and provide further evidence that therapeutics targeting the nsP2 N-terminal helicase domain, in addition to the C-terminal protease domain (36), may be useful against alphavirus infection. In addition, we have developed a path forward for the development of compounds targeting alphaviruses containing non-Tyr residues at nsP2 position aa 102 and for modifying quinolinones with improved binding characteristics.

MATERIALS AND METHODS

Viruses.

VEEV_TC-83 was derived from the pVEEV-TC-83 plasmid encoding the viral genome, kindly provided by Ilya Frolov (University of Alabama—Birmingham) (37). This plasmid was linearized and used as a template for in vitro transcription with SP6 DNA-dependent RNA polymerase (Ambion). RNA was electroporated into BHK-21 cells, and at 24 h postelectroporation, cell culture supernatant was collected and clarified by centrifugation at 1,721 × g. Clarified supernatants were aliquoted and stored at −80°C. To generate VEEV_TC-83 nsP2 Y102C, the pVEEV-TC-83 plasmid was mutated by site-directed mutagenesis (Agilent). Restriction digestion was performed with BssHII and EcoRI to release a 580-bp fragment containing the desired mutation, and this fragment was ligated in BssHII- and EcoRI-digested, unmutagenized plasmid. The ligated fragment was sequenced to confirm that only the desired mutation was present. Virus encoding nsP2 Y102C was generated using the same methods described above, and the presence of the mutation in the encapsulated RNA genome was confirmed by sequence analysis of a reverse transcription-PCR (RT-PCR)-generated amplicon spanning the mutated region. CHIKV, ONNV, MAYV, RRV, and UNAV were obtained from Robert Tesh (University of Texas Medical Branch—Galveston). Virus stocks were propagated in Aedes albopictus C6/36 mosquito cells by infecting with a low multiplicity of infection. At 48 to 72 h postinoculation (hpi), viral particles were pelleted through a 15% sucrose cushion by ultracentrifugation (76,618 × g), and pellets were resuspended in phosphate-buffered saline (PBS). Stock virus titers were quantified by plaque assay on BHK-21 or Vero cells as previously described (38, 39). To generate CHIKV harboring a nsP2 K102Y mutation, a plasmid containing the CHIKV_181/25 infectious clone genome sequence was mutagenized at position nsP2-K102 (AAG) to contain a tyrosine (TAC) using the Q5 site-directed mutagenesis kit (catalog no. E0554; New England BioLabs) and the forward and reverse primers 5′-CGTAAACAGATACTTTACACCACATTGC and 5′-AACTCTCTTTCGTTGTAC, respectively. Following PCR amplification and selection, the mutagenized plasmid was sequence confirmed, and viral RNA (vRNA) was generated by in vitro transcription. Infectious CHIKV particles were recovered 2 days following transfection of vRNA into fibroblasts.

All studies with the ZPC738 strain of VEEV were conducted under biosafety level 3 (BSL3) conditions by trained personnel wearing approved biosafety equipment. All protocols and standard operating procedures associated with the handling, storage, and disposal of this virus were approved by the Institutional Biosafety Committee at the University of North Carolina.

Cells.

Normal human dermal fibroblasts (NHDFs) (ATCC PCS-201-012), telomerized human foreskin fibroblasts (THF-ΔIRF3, from Victor DeFilippis), and Vero cells (ATCC CCL-81) were grown at 37°C in complete Dulbecco’s modified eagle medium (DMEM; Corning) containing 10% fetal bovine serum (FBS; Thermo Scientific) and supplemented with 1× penicillin-streptomycin-glutamine (Life Technologies). BHK-21 cells (ATCC CCL10) were grown in α-minimal essential medium (Life Technologies) supplemented with 10% bovine calf serum (HyClone), 10% tryptose phosphate broth, penicillin, streptomycin, and 0.29 mg/ml l-glutamine. Aedes albopictus C6/36 cells (ATCC CRL-1660) were grown at 28°C in complete DMEM containing 10% FBS.

VEEV high-throughput screen.

A high-throughput screen of compounds for antiviral activity against VEEV_TC83 has been previously described (26). Briefly, 348,140 compounds were plated in 384-well black-walled plates seeded with 4,500 Vero 76 cells/well in a single dose of 20 μM in Eagle’s minimum essential medium with 5% heat-inactivated FBS, 1% penicillin/streptomycin/l-glutamine, 1% HEPES, and 0.2% DMSO. Twenty-five microliters of 175 50% tissue culture infective dose (TCID₅₀) of VEEV_TC83 were added to each well using a Matrix WellMate. The plates were incubated for 3 days in an actively humidified incubator with 5.0% CO₂ at 37°C and 95% humidity. The ability of the compounds to block cytopathic effect was determined by luminescence using the CellTiter-Glo assay kit (Promega). The Z factor values were calculated as follows: 1 − [(3 × standard deviation of cell control [σc]) + (3 × standard deviation of the virus control [σv])]/[mean cell control signal (μc) − mean virus control signal (μv)].

Cellular cytotoxicity.

Compound cytotoxicity was measured at concentrations ranging from 100 to 0.41 μM, following the CellTiter-Glo luminescent cell viability assay protocol (Promega). Briefly, starting at a concentration of 100 μM, compounds were diluted 1:3 in a 96-well plate with DMEM supplemented with penicillin-streptomycin-glutamine (PSG) and 5% FBS. A total of 50 μl of diluted compound was added to the wells of a 96-well black-walled plate (Corning) seeded with 1 × 10⁴ NHDF cells/well at 1 day prior to the assay. At 24 h after compound treatment, 50 μl of CellTiter-Glo substrate was added to each well, followed by 2 min on an orbital rocker and a 10-min incubation. The luminescence of each well was measured using a Synergy HTX microplate reader (BioTek, Winooski, VT). Well luminescence, indicative of the number of living cells per well, was converted to percent cell viability in Microsoft Excel by dividing luminescence values in experimental wells by the value in control wells containing untreated cells and multiplying by 100. These values were then used to calculate the CC₅₀ values by nonlinear regression analysis of graphs with log compound concentration plotted versus cell viability, using Prism v8 (GraphPad Software, Inc.).

Validation screen.

Confluent 48-well plates of NHDFs were treated 1 h prior to infection with DMSO or 10 μM compound for viral load reduction assays (VLR) or a with a 10-point curve of compound ranging from 100 μM to 0.19 μM for dose-response assays. Compound dilutions were generated by diluting compounds 1:1 with DMEM supplemented with 5% FBS and PSG. Treated NHDFs were inoculated with either VEEV_TC83, CHIKV, CHIKV K102Y, MAYV, ONNV, RRV, or UNAV at a multiplicity of infection (MOI) of 1 or 3 PFU/cell. At 2 hpi, infection medium was removed, and cells were washed twice with PBS. Fresh medium containing compound was then added to each well. At 24 or 48 hpi, 25 μl medium from each well was collected into 96-well plates and frozen at −80°C. Yields of progeny virus were determined by limiting dilution plaque assays. The frozen 96-well plates were thawed, a 10-fold dilution series was made in the 96-well plate, and 100 μl of each dilution was transferred to confluent monolayers of Vero cells in 48-well plates. At 2 hpi, the cells were overlaid with 250 μl of medium containing 0.3% high-viscosity and 0.3% low-viscosity carboxy-methylcellulose. At 48 hpi, the plates were fixed with 3.7% formalin for 20 min, washed, and stained with 0.2% methylene blue dye for 10 min. Plaques were counted using a stereomicroscope. Data were tabulated in Excel and compared with control wells to determine fold change in release of infectious virus.

Time-of-addition studies.

For time-of-addition analysis, 12-well plates containing NHDFs were inoculated with VEEV_TC83 at an MOI of 3 PFU/cell. At times before, during, or after virus inoculation, culture medium containing 10 μM SRI-34329 was added. At 48 hpi, the amount of infectious virus in culture supernatants was quantified by plaque assay.

Northern blot analysis.

NHDF cells were pretreated with 10 μM compound for 1 h prior to infection with VEEV_TC83. At 2 hpi, cells were washed with PBS and fresh medium containing compound (10 μM) was added. At 12 hpi, total RNA was isolated from cells using TRIzol reagent (Life Technologies) according to the manufacturer’s protocol. Isolated RNA was quantified using a NanoDrop spectrophotometer. A 10-μg sample of total RNA was applied to a 1% formaldehyde-agarose gel and separated by gel electrophoresis. RNA was transferred from the gel to a Hybond-N+ membrane (Amersham Biosciences) using the capillary transfer method. After transfer, the RNA was fixed to the membrane via UV cross-linking. Digoxigenin (DIG)-labeled probes were generated by PCR amplification of a 1-kb fragment spanning the E2-6K-E1 coding region of the VEEV genome using the forward primer 5′-C TCACTACACGCACGAGCTCATATC and the reverse primer 5′-C TGAGCTTGAGACTGTTCTGGATTG. A human β-actin DIG-labeled probe was generated by PCR amplification of DNA encoding the human β-actin gene using the forward primer 5′-ACCCTGAAGTACCCCATCGA and the reverse primer 5′-CGGACTCGTCATACTCCTGC. Probes were hybridized overnight at 68°C in hybridization buffer. Blots were washed twice for 5 min with a low stringency buffer and twice with 1× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate) containing 0.1% SDS at room temperature, and then washed twice for 15 min with 0.1× SSC containing 0.1% SDS at 68°C. The washed blots were blocked for 3 h in blocking solution (Roche) at room temperature and then incubated with anti-DIG-alkaline phosphatase antibody for 30 min at room temperature. The blots were washed in buffer containing 0.1 M maleic acid, 0.5 M NaCl, and 0.3% Tween 20, followed by equilibration in detection buffer (0.1 M Tris-HCl and 0.1 M NaCl [pH 9.5]). Antibodies were visualized using a CSPD alkaline phosphatase substrate and visualized on CL-XPosure film (Thermo Scientific).

Selection for drug resistance and viral genome sequencing.

VEEV_TC83 was passaged in triplicate wells of NHDFs in the presence of DMSO or 10 μM SRI-34329. At 48 hpi, cell culture supernatants were collected, and yields of infectious virus were quantified by plaque assay. To confirm the acquisition of drug resistance, passaged virus showing evidence of resistance (i.e., increased viral titers in the presence of SRI-34329) was rescreened against SRI-34329. Viral RNA was purified from untreated, drug-resistant, and DMSO control virus stocks using TRIzol reagent according to the manufacturer’s protocol (Life Technologies) and used in Illumina-based sequencing as previously described (40). Isolated viral RNA (500 ng) was reverse transcribed using 500 ng of random primer 9 (NEB) and SuperScript II reverse transcriptase (Thermo Fisher Scientific). The reaction mixture was then equilibrated at 25°C for 2 min before the addition of 200 U of SuperScript II reverse transcriptase. The reverse transcription reaction mixture was incubated at 25°C for 10 min and at 42°C for 180 min and then inactivated at 70°C for 15 min. Double-stranded cDNA was prepared by using the NEBNext mRNA second strand synthesis module (NEB). Sequencing libraries were prepared from the double-stranded cDNA using the Nextera XT DNA library preparation kit (Illumina). Residual nucleotides were removed by using Agencourt AMPure XP beads (Beckman Coulter) at a DNA-to-bead ratio of 0.6:1. Library size and quality were measured by using a 2100 Bioanalyzer (Agilent Technologies) and quantified with a Qubit fluorometer by using the Qubit double-stranded DNA (dsDNA) high-sensitivity (HS) assay kit (Thermo Fisher Scientific). Sequencing reactions were performed on a MiSeq desktop sequencer (Illumina). Reads were sorted and aligned, and mutations were characterized using Bowtie 2 (41) and SAMtools (42).

Statistical analysis.

All data were analyzed using GraphPad Prism 8 software. Data were evaluated for statistically significant differences using a two-tailed, unpaired t test, one-way analysis of variance (ANOVA) followed by Tukey’s multiple-comparison test, or two-way analysis of ANOVA followed by Bonferroni’s multiple-comparison test. A P value of <0.05 was considered statistically significant.

Computational docking studies.

All modeling work was performed using different modules in Schrödinger Small Molecule Drug Discovery Suite (43). Five protein structures were prepared for docking studies. CHIKV nsP2 was prepared from its crystal structure (PDB identifier 6JIM) following the standard protocol of Protein Preparation Wizard (44). A homology model of VEEV nsP2 was built using the CHIKV nsP2 crystal structure as the template via the energy-based structure prediction approach (45) implemented in Schrödinger. To prepare the CHIKV nsP2 K102Y mutant, residue K102 of CHIKV nsP2 was mutated in silico to Y102. Subsequently, a geometry minimization was performed on Y102 and residues within 5 Å using the Prime module (46). Similar procedures were used to prepared the VEEV nsP2 Y102C and Y102S mutants from the VEEV nsP2 homology model. The three-dimensional (3D) conformations of the compounds were generated using the LigPrep module. Lead SRI-34329 was then docked to a pocket around residue 102 of all five protein structures using the induced fit docking protocol (flexible ligand and receptor) (47, 48) implemented in Schrödinger. All other compounds were docked to the same pocket of VEEV nsP2 following the same protocol.

Chemical synthesis.

The high-resolution (HR) mass spectral data were obtained on an Agilent liquid chromatography-mass spectrometry time-of-flight (LC/MS-TOF) or Bruker BioTOF II instrument by electrospray ionization (ESI). ¹H nuclear magnetic resonance (NMR) spectra were recorded at 400 MHz on an Agilent-Varian MR-400 spectrometer in CDCl₃, CD₃OD, or DMSO-d₆ as a solvent. Chemical shifts (δ) are given in ppm downfield from standard tetramethylsilane (TMS). High-performance liquid chromatography (HPLC) analyses of final compounds were run on an Agilent 1100 liquid chromatograph equipped with a diode array UV detector and were monitored at 254 and 280 nm using the following methods.

(i) Method A.

Phenomenex Kinetex phenyl-hexyl column (2.6 μm, 100 Å, 50 mm × 4.6 mm) using solvent A, 95:5 H₂O:acetonitrile (MeCN) with 1% HCO₂H, and solvent B, MeCN with 0.1% HCO₂H; flow rate, 2.0 ml/min; 4-min linear gradient from 5 to 95% B.

(ii) Method B.

Ascentis phenyl-hexyl column (2.7 μm, 3 cm × 4.6 mm) using solvent A, 0.1% HCO₂H in 95:5 H₂O:MeCN, and solvent B, 0.1% HCO₂H in MeCN; flow rate, 1.0 ml/min; 4-min linear gradient from 10 to 90% B.

(iii) Method C.

Sunfire C₁₈ column (5 μm, 4.6 mm × 150 mm) using solvent A, 0.1% HCO₂H in H₂O, and solvent B, 0.1% HCO₂H in MeCN; flow rate, 2.0 ml/min; 20-min linear gradient from 5 to 95% B, hold for 2 min, then back to 95% A over 3 min.

The HPLC purity of compounds 3a to 3h, 3k, 3n, 3o, and 3p was determined by method A; that of compounds 3j, 3l, and 3m by method B; and that of compound 3i by method C. The HPLC purity of all final compounds obtained was >96%.

General synthetic scheme and procedures.

The general synthetic scheme is shown in Table 1 and is further described as follows.

(i) Step 1 (N-alkylation).

Following the procedure described in the literature (28), alkyl iodide (1.5 equivalent) and potassium carbonate (1.5 equivalent) were added to quinolinone 1 (1 mmol) in dimethylformamide (DMF) (5 ml), and the reaction mixture was stirred for 3 to 4 h at room temperature. After completion (checked by thin-layer chromatography [TLC]), the reaction mixture was poured into ice-cold water (10 ml), whereupon a product precipitated out; the product was filtered, washed with water, and dried. The product was purified by recrystallization from 70% aqueous ethanol.

(ii) Step 2 (reductive amination).

Similarly to a procedure reported in the literature (29), a solution of quinolinone 1 (0.65 mmol) and appropriate corresponding amine (1.1 equivalent, 0.71 mmol) in dry methanol (MeOH; 5 ml) was stirred for 4 h at room temperature. Then, sodium borohydride (1 equivalent, 0.71 mmol) was added in portions to the reaction mixture and stirred for additional 30 min. Methanol was removed by evaporation, and the crude residue was washed with water and extracted with dichloromethane (DCM; 2 × 10 ml). The organic layer was dried over Na₂SO₄ and concentrated under vacuum to afford the desired amine. It was used in the next step without purification.

(iii) Step 3 (urea formation).

A mixture of quinolinone amine (1 equivalent) from step 2 and the desired isocyanate (1.1 equivalent) in dry DCM (5 ml) was stirred for 18 h at room temperature. Progress of the reaction was monitored by TLC, which showed that all starting material was consumed. Silica gel was added to the reaction mixture and evaporated to dryness. This preadsorbed mixture on silica was subjected to chromatography on silica gel via column chromatography (0 to 100% ethyl acetate [EtOAc] in hexanes) to afford the desired final compound.

Compounds.

Structures for the compounds described below are shown in Table 1.

(i) 1-((6,7-Dimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-1-(2-(dimethylamino)ethyl)-3-(furan-2-ylmethyl)urea (compound 3a).

The compound was obtained as a white solid (71% yield). Proton nuclear magnetic resonance ¹H NMR (400 MHz, DMSO-d₆) δ 11.80 (s, 1H), 7.52 (d, J = 17.5 Hz, 3H), 7.36 (s, 1H), 7.07 (s, 1H), 6.34 (t, J = 2.4 Hz, 1H), 6.15 (d, J = 3.1 Hz, 1H), 4.29 to 4.11 (m, 4H), 2.25 (d, J = 12.5 Hz, 13H). High-resolution electrospray ionization mass spectrometry (HR-ESIMS): calculated (calcd.) for C₂₂H₂₈N₄O₃ (M+H)⁺ as 397.2234, found 397.2233.

(ii) 1-(2-(Dimethylamino)ethyl)-3-(furan-2-ylmethyl)-1-((2-oxo-1,2-dihydroquinolin-3-yl)methyl)urea (compound3b).

The compound was obtained as a light-yellow solid (62% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 11.91 (s, 1H), 7.62 (d, J = 7.9 Hz, 2H), 7.51 to 7.41 (m, 3H), 7.30 (d, J = 8.2 Hz, 1H), 7.24 to 7.09 (m, 1H), 6.34 (dd, J = 3.2, 1.9 Hz, 1H), 6.15 (d, J = 3.1 Hz, 1H), 4.28 (s, 2H), 4.20 (d, J = 5.4 Hz, 2H), 2.38 (t, J = 6.3 Hz, 2H), 2.14 (s, 6×H). HR-ESIMS: calcd. for C₂₀H₂₄N₄O₃ (M+H)⁺ as 369.1921, found 369.1919.

(iii) 1-((6,7-Dimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-3-(furan-2-ylmethyl)-1-((5-methylfuran-2-yl)methyl)urea (compound 3c).

The compound was obtained as a yellow solid (67% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 11.79 (s, 1H), 7.51 (ddd, J = 15.2, 1.9, 0.8 Hz, 1H), 7.39 (d, J = 7.2 Hz, 2H), 7.05 (s, 1H), 6.41 to 6.26 (m, 2H), 6.20 to 6.06 (m, 2H), 5.90 (dd, J = 3.0, 1.2 Hz, 1H), 4.42 (s, 2H), 4.30 to 4.10 (m, 4H), 2.25 (d, J = 11.2 Hz, 6×H), 2.11 (d, J = 1.0 Hz, 3H). HR-ESIMS: calcd. for C₂₄H₂₅N₃O₄ (M+H)⁺ as 420.1918, found 420.1920.

(iv) 1-(2-(Dimethylamino)ethyl)-3-(furan-2-ylmethyl)-1-((1,6,7-trimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)urea (compound 3d).

The compound was obtained as a yellow solid (64% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 7.68 (dd, J = 7.8, 1.4 Hz, 1H), 7.62 to 7.47 (m, 4H), 7.36 (s, 1H), 7.32 to 7.21 (m, 1H), 6.39 to 6.24 (m, 1H), 6.13 (d, J = 3.2 Hz, 1H), 4.31 (s, 2H), 4.19 (d, J = 5.4 Hz, 2H), 3.64 (d, J = 0.9 Hz, 3H), 2.38 (t, J = 6.3 Hz, 2H), 2.14 (s, 6H). HR-ESIMS: calcd. for C₂₁H₂₆N₄O₃ (M+H)⁺ as 383.2078, found 383.2081.

(v) 1-(2-(Dimethylamino)ethyl)-3-(furan-2-ylmethyl)-1-((1-methyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)urea (compound 3e).

The compound was obtained as a white solid (70% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 7.55 to 7.45 (m, 2H), 7.41 (s, 2H), 7.33 (s, 1H), 6.34 (dd, J = 3.3, 1.8 Hz, 1H), 6.14 (d, J = 3.1 Hz, 1H), 4.28 (s, 2H), 4.19 (d, J = 5.4 Hz, 2H), 3.61 (s, 3H), 2.36 (s, 4H), 2.26 (s, 3H), 2.13 (s, 6×H). HR-ESIMS: calcd. for C₂₃H₃₀N₄O₃ (M+H)⁺ as 411.2391, found 411.2389.

(vi) 1-(2-(Dimethylamino)ethyl)-3-(furan-2-ylmethyl)-1-((2-oxo-1,2-dihydroquinolin-3-yl)methyl)thiourea (compound 3f).

The compound was obtained as an off-white solid (58% yield); ¹H NMR (400 MHz, DMSO-d₆) δ 11.96 (s, 1H), 7.63 (dd, J = 8.0, 1.4 Hz, 1H), 7.56 (d, J = 3.3 Hz, 2H), 7.46 (ddd, J = 8.5, 7.1, 1.4 Hz, 1H), 7.32 to 7.28 (m, 1H), 7.17 (td, J = 7.5, 1.1 Hz, 1H), 6.38 (dd, J = 3.2, 1.9 Hz, 1H), 6.29 (s, 1H), 4.74 (s, 2H), 4.66 (d, J = 4.6 Hz, 2H), 3.68 (s, 2H), 2.45 (s, 1H), 2.09 (s, 6H). HR-ESIMS: calcd. for C₂₀H₂₄N₄O₂S (M+H)⁺ as 385.1693, found 385.1693.

(vii) 1-((6,7-Dimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-3-(furan-2-ylmethyl)-1-((tetrahydrofuran-2-yl)methyl)thiourea (compound 3g).

The compound was obtained as an off-white solid (70% yield); ¹H NMR (400 MHz, DMSO-d₆) δ 11.86 (s, 1H), 8.50 (s, 1H), 7.51 (s, 1H), 7.42 (s, 1H), 7.37 (s, 1H), 7.07 (s, 1H), 6.35 (t, J = 2.5 Hz, 1H), 6.23 (d, J = 3.2 Hz, 1H), 4.91 to 4.55 (m, 4H), 4.17 (s, 1H), 3.74 (dt, J = 8.7, 6.9 Hz, 1H), 3.61 (td, J = 7.8, 5.8 Hz, 1H), 3.45 (dd, J = 14.7, 8.1 Hz, 1H), 2.25 (d, J = 13.7 Hz, 6×H), 2.09 to 2.05 (m, 0H), 1.94 to 1.69 (m, 2H), 1.47 (dt, J = 11.8, 7.9 Hz, 1H). HR-ESIMS: calcd. for C₂₃H₂₇N₃O₃S (M+H)⁺ as 426.1846, found 426.1845.

(viii) 1-((6,7-Dimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-3-(furan-2-ylmethyl)-1-((5-methylfuran-2-yl)methyl)thiourea (compound 3h).

Compound was obtained as an off-white solid (64% yield); ¹H NMR (400 MHz, DMSO-d₆) δ 11.89 (s, 1H), 8.73 (s, 1H), 7.50 (dd, J = 1.9, 0.9 Hz, 1H), 7.37 (s, 1H), 7.32 (s, 1H), 7.07 (s, 1H), 6.35 (dd, J = 3.2, 1.8 Hz, 1H), 6.22 (dd, J = 3.2, 0.8 Hz, 2H), 5.92 (dd, J = 3.0, 1.2 Hz, 1H), 5.01 (s, 2H), 4.74 (d, J = 5.1 Hz, 2H), 4.59 (s, 2H), 2.25 (d, J = 11.9 Hz, 6×H), 2.09 (d, J = 1.0 Hz, 3H). HR-ESIMS: calcd. for C₂₄H₂₅N₃O₃S (M+H)⁺ as 436.1689, found 436.1689.

(ix) 3-(Furan-2-ylmethyl)-1-((6-methyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-1-(1-methylpyrrolidin-3-yl)thiourea (compound 3i).

Compound was obtained as an off-white solid (55% yield). ¹H NMR (400 MHz, CD₃OD) δ 7.70 (s, 1H), 7.42 (dt, J = 3.0, 1.5 Hz, 1H), 7.41 (s, 1H), 7.39 to 7.31 (m, 1H), 7.25 (dd, J = 8.5, 1.6 Hz, 1H), 6.39 to 6.28 (m, 2H), 5.14 (d, J = 17.4 Hz, 1H), 5.01 (d, J = 17.6 Hz, 1H), 4.79 (d, J = 1.6 Hz, 2H), 2.91 (d, J = 9.3 Hz, 2H), 2.47 to 2.13 (m, 8H), 2.13 to 1.94 (m, 2H). HR-ESIMS: calcd. for C₂₂H₂₇N₄O₂S (M+H)⁺ as 411.1850, found 411.1844.

(x) 3-(Furan-2-ylmethyl)-1-((6-methyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-1-(pyrrolidin-3-yl)thiourea (compound 3j).

Compound was obtained as an off-white solid (45% yield). ¹H NMR (400 MHz, CD₃OD) δ 7.55 (s, 1H), 7.38 (dq, J = 6.1, 1.9 Hz, 2H), 7.26 (d, J = 8.9 Hz, 1H), 7.21 (dd, J = 1.9, 0.9 Hz, 1H), 6.25 (dd, J = 3.3, 1.8 Hz, 1H), 6.22 (dd, J = 3.2, 0.9 Hz, 1H), 6.04 (p, J = 8.8 Hz, 1H), 4.86 to 4.70 (m, 2H), 4.59 (t, J = 1.9 Hz, 2H), 3.62 (dd, J = 12.1, 8.9 Hz, 1H), 3.53 (ddd, J = 11.7, 8.6, 2.9 Hz, 1H), 3.32 to 3.18 (m, 2H), 2.40 (s, 3H), 2.31 (dtd, J = 15.7, 7.7, 2.8 Hz, 1H), 2.18 (dtd, J = 13.1, 10.3, 8.8 Hz, 1H). HR-ESIMS: calcd. for C₂₁H₂₅N₄O₂S (M+H)⁺ as 397.1693, found 397.1685.

(xi) 1-(2-(Dimethylamino)ethyl)-3-(furan-2-ylmethyl)-1-((1,6,7-trimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)thiourea (compound 3k).

Compound was obtained as a yellow solid (59% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 7.55 (s, 1H), 7.43 (s, 2H), 7.35 (s, 1H), 6.38 (t, J = 2.5 Hz, 1H), 6.27 (s, 1H), 4.72 (s, 2H), 4.66 (d, J = 4.7 Hz, 2H), 3.70 (s, 2H), 3.62 (s, 3H), 2.51 (s, 2H), 2.37 (s, 3H), 2.27 (s, 3H), 2.13 (s, 7H). HR-ESIMS: calcd. for C₂₃H₃₀N₄O₂S (M+H)⁺ as 427.2162, found 427.2160.

(xii) 1-(2-(Dimethylamino)ethyl)-1-((1-ethyl-6,7-dimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)-3-(furan-2-ylmethyl)thiourea (compound 3l).

Compound was obtained as an off-white solid (91% yield). ¹H NMR (400 MHz, CDCl₃) δ 7.80 (d, J = 2.5 Hz, 1H), 7.36 (dt, J = 2.0, 0.9 Hz, 2H), 7.13 (s, 1H), 6.33 (dd, J = 3.4, 1.8 Hz, 1H), 6.29 (dt, J = 3.1, 0.8 Hz, 1H), 5.02 (s, 2H), 4.77 (d, J = 4.1 Hz, 2H), 4.33 (q, J = 7.1 Hz, 2H), 3.58 (s, 2H), 2.52 (s, 2H), 2.40 (s, 3H), 2.30 (s, 3H), 2.11 (s, 6×H), 1.39 to 1.29 (m, 3H). HR-ESIMS: calcd. for C₂₄H₃₃N₄O₂S (M+H)⁺ as 441.2319, found 441.2386.

(xiii) 1-((6,7-Dimethyl-2-oxo-1-propyl-1,2-dihydroquinolin-3-yl)methyl)-1-(2-(dimethylamino)ethyl)-3-(furan-2-ylmethyl)thiourea (compound 3m).

Compound was obtained as an off-white solid (84% yield). ¹H NMR (400 MHz, CDCl₃) δ 7.85 (s, 1H), 7.41 to 7.32 (m, 2H), 7.08 (s, 1H), 6.33 (dd, J = 3.2, 1.9 Hz, 1H), 6.29 (d, J = 3.2 Hz, 1H), 5.01 (s, 2H), 4.77 (d, J = 4.1 Hz, 2H), 4.26 to 4.16 (m, 2H), 3.74 to 3.46 (m, 2H), 2.54 (s, 2H), 2.40 (s, 3H), 2.30 (s, 3H), 2.12 (s, 6×H), 1.75 (h, J = 7.4 Hz, 2H), 1.04 (t, J = 7.4 Hz, 3H). HR-ESIMS: calcd. for C₂₅H₃₅N₄O₂S (M+H)⁺ as 455.2475, found 455.2523.

(xiv) 3-Cyclopropyl-1-(2-(dimethylamino)ethyl)-1-((1,6,7-trimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)thiourea (compound 3n).

Compound was obtained as an off-white solid (62% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 9.02 (s, 1H), 7.47 (d, J = 3.8 Hz, 2H), 7.36 (s, 1H), 4.68 (s, 2H), 3.62 (d, J = 0.7 Hz, 5H), 3.01 (dq, J = 7.2, 3.6 Hz, 1H), 2.46 (s, 1H), 2.37 (s, 3H), 2.26 (s, 3H), 2.17 (s, 6×H), 0.67 (td, J = 7.0, 4.8 Hz, 2H), 0.51 to 0.43 (m, 2H). HR-ESIMS: calcd. for C₂₁H₃₀N₄OS (M+H)⁺ as 387.2213, found 387.2212.

(xv) 1-(2-(Dimethylamino)ethyl)-3-methyl-1-((1,6,7-trimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)thiourea (compound 3o).

Compound was obtained as a white solid (68% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 8.16 (s, 1H), 7.46 (s, 1H), 7.39 (s, 1H), 7.36 (s, 1H), 4.65 (s, 2H), 3.79 (d, J = 8.4 Hz, 2H), 3.63 (s, 3H), 2.87 (d, J = 4.1 Hz, 3H), 2.56 (s, 2H), 2.36 (s, 3H), 2.24 (d, J = 12.4 Hz, 9H). HR-ESIMS: calcd. for C₁₉H₂₈N₄OS as 361.2057, found 361.2060.

(xvi) 3-Benzyl-1-(2-(dimethylamino)ethyl)-1-((1,6,7-trimethyl-2-oxo-1,2-dihydroquinolin-3-yl)methyl)thiourea (compound 3p).

Compound was obtained as a white solid (68% yield). ¹H NMR (400 MHz, DMSO-d₆) δ 7.43 (s, 2H), 7.35 (s, 1H), 7.30 to 7.14 (m, 4H), 4.70 (d, J = 5.3 Hz, 3H), 3.80 (d, J = 9.8 Hz, 2H), 3.62 (s, 3H), 2.60 (s, 2H), 2.37 (s, 3H), 2.28 (s, 3H), 2.16 (s, 6×H). HR-ESIMS: calcd. for C₂₅H₃₂N₄OS (M+H)⁺ as 437.2370, found 437.2371.

Data availability.

Newly determined sequences have been deposited in GenBank under accession numbers MZ399798 and MZ399799.