Novel Indole-2-Carboxamide Compounds Are Potent Broad-Spectrum Antivirals Active against Western Equine Encephalitis Virus In Vivo

ABSTRACT

Neurotropic alphaviruses, including western, eastern, and Venezuelan equine encephalitis viruses, cause serious and potentially fatal central nervous system infections in humans for which no currently approved therapies exist. We previously identified a series of thieno[3,2-b]pyrrole derivatives as novel inhibitors of neurotropic alphavirus replication, using a cell-based phenotypic assay (W. Peng et al., J. Infect. Dis. 199:950–957, 2009, doi:http://dx.doi.org/10.1086/597275), and subsequently developed second- and third-generation indole-2-carboxamide derivatives with improved potency, solubility, and metabolic stability (J. A. Sindac et al., J. Med. Chem. 55:3535–3545, 2012, doi:http://dx.doi.org/10.1021/jm300214e; J. A. Sindac et al., J. Med. Chem. 56:9222–9241, 2013, http://dx.doi.org/10.1021/jm401330r). In this report, we describe the antiviral activity of the most promising third-generation lead compound, CCG205432, and closely related analogs CCG206381 and CCG209023. These compounds have half-maximal inhibitory concentrations of ∼1 μM and selectivity indices of >100 in cell-based assays using western equine encephalitis virus replicons. Furthermore, CCG205432 retains similar potency against fully infectious virus in cultured human neuronal cells. These compounds show broad inhibitory activity against a range of RNA viruses in culture, including members of the Togaviridae, Bunyaviridae, Picornaviridae, and Paramyxoviridae families. Although their exact molecular target remains unknown, mechanism-of-action studies reveal that these novel indole-based compounds target a host factor that modulates cap-dependent translation. Finally, we demonstrate that both CCG205432 and CCG209023 dampen clinical disease severity and enhance survival of mice given a lethal western equine encephalitis virus challenge. These studies demonstrate that indole-2-carboxamide compounds are viable candidates for continued preclinical development as inhibitors of neurotropic alphaviruses and, potentially, of other RNA viruses.

IMPORTANCE There are currently no approved drugs to treat infections with alphaviruses. We previously identified a novel series of compounds with activity against these potentially devastating pathogens (J. A. Sindac et al., J. Med. Chem. 55:3535–3545, 2012, doi:http://dx.doi.org/10.1021/jm300214e; W. Peng et al., J. Infect. Dis. 199:950–957, 2009, doi:http://dx.doi.org/10.1086/597275; J. A. Sindac et al., J. Med. Chem. 56:9222–9241, 2013, http://dx.doi.org/10.1021/jm401330r). We have now produced third-generation compounds with enhanced potency, and this manuscript provides detailed information on the antiviral activity of these advanced-generation compounds, including activity in an animal model. The results of this study represent a notable achievement in the continued development of this novel class of antiviral inhibitors.

INTRODUCTION

Infections caused by arthropod-borne viruses (arboviruses) represent dramatic examples of disease reemergence (1), due in part to significant urban growth and the ease of worldwide travel, which produce conditions that facilitate arbovirus epidemics (2, 3). Furthermore, the threat posed by the intentional exposure of a population center to a virulent arbovirus has prompted the United States federal government to designate numerous arboviruses high-priority biodefense pathogens, particularly those that infect the central nervous system (CNS), causing encephalitis. The Alphavirus genus within the Togaviridae family contains some 30 mosquito-borne, enveloped, single-stranded, positive-sense RNA viruses, one-third of which cause significant diseases in humans and animals worldwide (4). The encephalitic alphaviruses, including western, eastern, and Venezuelan equine encephalitis viruses (WEEV, EEEV, and VEEV), directly infect neurons, resulting in CNS inflammation and neuronal destruction (5–8). These highly virulent pathogens can cause severe disease in humans, with case fatality rates of up to 70% and long-term neurological sequelae in most survivors (9, 10).

There are currently no licensed vaccines or antiviral drugs against alphaviruses. Formalin-inactivated vaccines for WEEV or EEEV and a live attenuated vaccine against VEEV (TC-83 strain) are available on an investigational drug basis, the use of which is limited primarily to laboratory personnel working with these infectious agents. The development of alternative live attenuated, chimeric, and DNA-based alphavirus vaccines is being actively pursued, but the broad clinical application of these next-generation vaccines is likely years away (11). Furthermore, the combination of active vaccination plus antiviral therapy may be a more effective response in the setting of an outbreak caused either by natural transmission or by intentional exposure to one of these viral pathogens (12). Although numerous compounds have been reported to inhibit alphavirus replication in cultured cells, only a select few have shown any activity in animal models (13–17). Thus, there is a pressing need to identify new antiviral compounds and drug targets as part of a comprehensive medical-countermeasure strategy to prevent or mitigate illness, suffering, and death resulting from infections caused by these virulent pathogens (18).

We previously identified a novel class of thieno[3,2-b]pyrrole-based alphavirus inhibitors via high-throughput screening using a cell-based WEEV replicon assay and a defined small-molecule diversity library (19). Through extensive structure-activity relationship (SAR) analyses and targeted medicinal-chemistry efforts, we produced second-generation indole-2-carboxamide derivatives that display increased potency and striking enantiospecific activity against WEEV and related alphaviruses in cultured cells and that also improve survival in mice infected with a neuroadapted strain of Sindbis virus (SINV) (13). We have now produced third-generation analogues through continued medicinal-chemistry-directed efforts that have increased antiviral potency, where we have achieved submicromolar half-maximal inhibitory concentrations (IC₅₀) against WEEV replicons in cultured cells (20). In this report, we detail the antiviral activity of three closely related third-generation compounds, CCG205432, CCG206381, and CCG209023. These inhibitors display potent and broad-spectrum antiviral activity in cultured cells against alphaviruses, picornaviruses, bunyaviruses, and paramyxoviruses. We also provide evidence for their mechanism of action, which involves, at least in part, suppression of cap-dependent translation. Furthermore, we demonstrate that CCG205432 and CCG209023 improve both the clinical disease course and survival of mice given a lethal WEEV challenge.

MATERIALS AND METHODS

Mice.

Female C57BL/6 mice (5 to 6 weeks of age) were purchased from the Jackson Laboratory (Bar Harbor, ME). All animals were housed and used on site under specific-pathogen-free conditions in a federally certified animal biosafety level 3 (ABSL3) facility in strict accordance with guidelines set by the National Institutes of Health and Centers for Disease Control and protocols approved by the University of Michigan Committee on the Use and Care of Animals and the Institutional Biosafety Committee. Mice were housed with a 10/14 h light/dark cycle in ventilated cages containing no more than 5 animals per cage, and food and water were available ad libitum. For WEEV infections, mice were inoculated subcutaneously with 10³ PFU WEEV suspended in 100 μl phosphate-buffered saline (PBS). CCG205432 and CCG209023 were solubilized in dimethyl sulfoxide (DMSO) as a stock solution at a concentration of 100 mM and then diluted in PBS to generate working solutions for intraperitoneal injections into infected mice on a twice-daily dosing schedule for 7 days following virus challenge. Mice were monitored for an additional 7 days after treatment was completed, and all surviving animals were euthanized at day 14. All mice were weighed daily and assigned a clinical score using a 1-to-5 scale as previously described (13).

Plasmids.

The WEEV replicon plasmids, pWR-LUC and pWR-ΔLUC, have been previously described (19). To generate the WEEV nonstructural protein (nsP) expression plasmid, pW-nsP, we inserted the nsP1- through nsP4-coding region from pWR-LUC into pCITE-4a(+) (EMD Millipore, Darmstadt, Germany). The helper plasmids expressing the SINV capsid protein or envelope glycoproteins, pSINV-HC or pSINV-HGP, respectively (21), were obtained from Ilya Frolov (University of Alabama at Birmingham, Birmingham, AL). The control simian virus 40 (SV40) promoter-driven firefly luciferase (fLUC) expression plasmid, pGL3-Promoter, which we renamed pSV40-LUC, was purchased from Promega (Madison, WI). To generate the T7 promoter-driven fLUC expression plasmid pT7/CITE-LUC, we inserted the fLUC cassette from pTRE2hyg-LUC (Clontech, Mountain View, CA) into pCITE-4a(+), subsequently removed the region encoding the polyadenylation tract to produce pT7/CITE-LUC(-A), and finally removed the cap-independent translation element (CITE) from each to produce pT7-LUC and pT7-LUC(-A). Cloning details are available upon request.

Inhibitors and antibodies.

Synthesis of compounds CCG205432, CCG206381, and CCG209023 has been previously described (13, 20). Ribavirin, mycophenolic acid (MPA), antimycin A, phosphonoacetic acid (PAA), and cycloheximide were all purchased from Sigma-Aldrich (St. Louis, MO). Hippuristanol was obtained from Junichi Tanaka (University of Ryukyus, Nishihara, Japan). Generation of mouse monoclonal antibodies against SINV E2 glycoprotein has been previously described (22).

Cell culture.

BE(2)-C, BHK-21, Vero, CHO, SH-SY5Y, U87, and C6/36 cells were purchased from the American Type Culture Collection (Manassas, VA). Differentiated BE(2)-C/m cells were generated as previously described (23). HEK293 cells were obtained from David Markovitz (University of Michigan, Ann Arbor, MI), Huh-7 cells were obtained from Raymond Chung (Massachusetts General Hospital, Boston, MA), BHK-21 cells stably expressing bacteriophage T7 RNA polymerase (BSR-T7 cells) were obtained from Klaus Conzelman (Max von Pettenkofer-Institut, Munich, Germany) and Sonja Gerrard (University of Michigan, Ann Arbor, MI), and BHK-21 cells stably expressing dengue virus (DENV) replicons encoding a Renilla LUC (rLUC) reporter gene were obtained from Andrew Tai (University of Michigan, Ann Arbor, MI). BSR-T7 cells were cultured as previously described (13), and all other cells were cultured in complete Dulbecco's modified Eagle medium (cDMEM) containing 5% bovine growth serum, 1% sodium pyruvate, 0.1 mM nonessential amino acids, 10 U/ml penicillin, and 10 μg/ml streptomycin.

Viruses.

The Cba-87 strain of WEEV was initially generated from the full-length cDNA clone pWE2000 as previously described (23). A low-passage isolate of infectious virus produced in Vero cells was expanded twice in C6/36 mosquito cells to obtain viral stocks that were stored in single-use aliquots at −80°C. All experiments with infectious WEEV were performed under biosafety level 3 (BSL3) containment conditions. Encephalomyocarditis virus (EMCV), the BFS-283 strain of California encephalitis virus (CEV), and the CM4-146 strain of Fort Morgan virus (FMV) were purchased from the American Type Culture Collection. The TC-83 vaccine strain of VEEV was obtained from Robert Tesh (University of Texas Medical Branch, Galveston, TX), and the New Guinea C and H241 strains of DENV-2 and DENV-4, respectively, were obtained from Sonja Gerrard (University of Michigan, Ann Arbor, MI). Green fluorescent protein (GFP)-tagged Sendai virus (GFP-SeV) was obtained from Valery Grdzelishvili (University of North Carolina at Charlotte, Charlotte, NC) and has been previously described (24).

We generated chimeric WEEV-SINV pseudoinfectious particles (PIPs) in BHK-21 cells using a three-component system with RNAs derived from the plasmids pWR-LUC, pSINV-HC, and pSINV-HGP. Plasmids were linearized, and capped mRNAs were generated with T7 (pWR-LUC) or SP6 (pSINV-HC and pSINV-HGP) RNA polymerase using the Ribomax system (Promega) according to the manufacturer's instructions. Full-length transcript production was checked by agarose gel electrophoresis, and transcription reaction mixtures were used directly for electroporation into BHK-21 cells using a Bio-Rad Gene Pulser, 0.2-cm cuvettes, and the preset BHK-21 protocol according to the instructions of the manufacturer (Bio-Rad, Hercules, CA). After electroporation, cells were transferred to tissue culture plates in cDMEM, incubated for 24 h, and subjected to one freeze-thaw cycle at −80°C to facilitate PIP release. Samples were centrifuged at 2,000 × g for 5 min to pellet cellular debris. Clarified supernatants containing chimeric WEEV-SINV PIPs were dispensed into single-use aliquots and stored at −80°C, and optimal target cell-PIP combinations were determined empirically to maximize signal-to-noise ratios.

Virus replication assays.

Infectious titers for all viruses were determined by plaque assay on Vero cell monolayers as previously described (17). WEEV replicon assays used BSR-T7 cells and pWR-LUC, pW-nsP, or pWR-ΔLUC and measured fLUC accumulation as a surrogate marker for viral RNA replication as previously described (13, 19). GFP-SeV replication was determined by fluorescence accumulation as previously described (24). Quantitative reverse-transcription (RT)-PCR was done as previously described (19, 24, 25), where threshold cycle (ΔΔC_T) values were calculated to determine relative RNA levels. The sequences for the 18S rRNA and WEEV envelope glycoprotein 1 primers have been previously published (25). The sequences for the forward and reverse DENV primers were 5′-TCAATATGCTGAAACGCGCGAGAAACCG-3′ and 5′-TTGCACCAACAGTCAATGTCTTCAGGTTC-3′, respectively, and corresponded to a capsid region conserved between serotypes. The sequences for the forward and reverse FMV envelope glycoprotein 1 primers were 5′-GCGACCACTGTGCCAAATGT-3′ and 5′-CTCCTGGTTTCATAGCTCCA-3′, respectively.

For PIP assays, tissue culture plates were precoated with 0.01% poly-l-lysine and seeded with HEK293 cells 18 h prior to compound or PIP addition, and fLUC activity was routinely assayed at 9 h postinfection (hpi). For pretreatment groups, optimized PIP dilutions in binding buffer (PBS with 1% bovine serum albumin [BSA]) were incubated with CCG205432, MPA, and control anti-SINV or anti-hemagglutinin (HA) monoclonal antibodies for 30 min at 4°C. Samples were subsequently added to tissue culture-plated cells, incubated for an additional 60 min at 4°C with gentle rocking to allow PIP binding, washed twice with ice-cold binding buffer, and incubated in cDMEM at 37°C for 9 h prior to fLUC assays. For posttreatment groups, untreated PIPs were used and compounds (CCG205432 or MPA) were added to cells after the binding and washing steps described above.

For in vitro viral RNA-dependent RNA polymerase (RdRp) assays, we isolated membrane fractions from mock- and WEEV-infected BHK-21 cells as previously described (26). Reaction mixtures contained 10-μl membrane fractions with 50 mM Tris (pH 8.0), 50 mM potassium acetate, 15 mM magnesium acetate, 1,600 U/ml RNAseOUT (Invitrogen), 10 μg/ml actinomycin D, 2 mM (each) ATP, CTP, and GTP, 100 μM unlabeled UTP, and 10 μCi [³²P]UTP in a 25-μl volume and were incubated for 30 min at 30°C. Samples were extracted once with phenol-chloroform and desalted with Bio-Rad P-30 columns, and the reaction products were separated in 0.8% non-denaturing agarose gels. After electrophoresis, gels were dried under vacuum conditions, and ³²P-labeled RNA was detected by autoradiography.

Viability and translation assays.

Cell viability after viral infection or drug treatment was routinely determined by 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazolium bromide (MTT) assay as previously described (23). For select additional cell viability measurements, we used CellTiter-Glo and CytoTox-Glo assays (Promega) according to the manufacturer's instructions. Messenger RNAs for cell transfection and in vitro translation experiments were generated with T7 RNA polymerase using the Ribomax in vitro transcription system (Promega) and capped with the ScriptCap m⁷G capping system (Epicentre Biotechnologies, Madison, WI) according to the manufacturers' instructions. Translation assays were done with nuclease-treated rabbit reticulocyte lysates (RRLs) or wheat germ extracts, and posttranslation lysates were analyzed for fLUC expression using Steady-Glo according to the instructions of the manufacturer for all three systems (Promega). Messenger mRNAs were transfected into BSR-T7 cells using a TransIT-mRNA transfection kit (Mirus Bio LLC, Madison, WI) according to the manufacturer's instructions.

Microarray and pathway analyses.

Transcriptome analyses were performed on independent sets of cultures using Affymetrix human U133 plus 2.0 microarray chips as previously described (24). The Genomatix ChipInspector software package (Genomatix Software Inc., Ann Arbor, MI) was used for primary microarray data analysis. The following parameters were chosen to identify sets of differentially regulated transcripts: (i) a false-discovery rate of 5%; (ii) three-probe minimum coverage; and (iii) an expression level log₂ change of ≥1 (2-fold) from that of the control. The list of genes preferentially up- or downregulated in BE(2)-C cells treated with CCG205432 or CCG206381 was analyzed using Ingenuity Pathway Analysis software (Ingenuity Systems, Redwood, CA). Significance was measured by determining the ratio of the number of genes from the data set that map to a particular canonical pathway to the total number of genes for that pathway and calculating a subsequent P value using a Fisher exact test. Association with a particular canonical pathway was considered significant at a P value of <0.01.

Statistical analyses.

We used a two-tailed Student's t test assuming equal variances for routine comparative analyses, and we performed virus titer statistical analyses on log₁₀-transformed data. Microarray and pathway statistical analyses are described above. Differences in survival among cohorts of WEEV-infected mice were measured using a log-rank (Mantel-Cox) test. Differences at a P value of <0.05 were considered significant, except for the pathway analysis described above. Unless otherwise indicated, results are representative of at least three independent experiments, where quantitative data represent the mean ± standard error of the mean.

Microarray data accession number.

Complete original data files have been deposited in the Gene Expression Omnibus database (www.ncbi.nlm.nih.gov/geo) under accession number GSE51909.

RESULTS

CCG205432 and CCG206381 have antiviral activity against infectious WEEV in cultured cells.

The initial series of thieno[3,2-b]pyrrole derivatives that we identified by high-throughput screening and validation in a cell-based assay had modest potency (19), where the lead compound, CCG32091 (Fig. 1A, left), had an IC₅₀ of 24.4 μM against WEEV replicons (13). Initial SAR development based on medicinal-chemistry principles and guided by antiviral activity led to the production of second-generation indole derivatives with improved potency, where the lead compound, CCG203926 (Fig. 1A, right), had an IC₅₀ of 6.5 μM against WEEV replicons and was also active against neuroadapted SINV in vivo (13). We directed subsequent SAR development toward continued improvement in potency, but we also sought to enhance crucial chemical parameters predictive of in vivo activity, such as solubility, metabolic stability, and reduced efflux by P glycoprotein, in efforts to advance preclinical development. We produced a series of third-generation compounds with submicromolar IC₅₀ values against WEEV replicons (20), and we initiated detailed antiviral studies on two potent derivatives, CCG205432 and CCG206381. Structures for these compounds, plus IC₅₀ values, half-maximal cytotoxicity (CC₅₀) values, and selectivity indices (CC₅₀/IC₅₀) for the replicon assay in BSR-T7 cells are shown in Fig. 1B and C. Both compounds contain a terminal 4-pyridylethyl carboxamide, which proved to be a key functional determinant for potency (13, 20). The only molecular difference between CCG205432 and CCG206381 is the core ring structure, which was maintained as a bicyclic indole in CCG205432 (Fig. 1B) but was replaced by a monocyclic pyrrole in CCG206381 (Fig. 1C) to reduce molecular weight. Compared to the lead second-generation compound, CCG203926, the third-generation compound, CCG205432, had 10-fold-increased potency, 2-fold-increased aqueous solubility, and 6-fold-increased metabolic stability measured by mouse liver microsome assay (20). For CCG206381, potency and aqueous solubility increased by 10-fold and 15-fold, respectively, and susceptibility to efflux by P glycoprotein was reduced, whereas there was little change in metabolic stability compared to that for CCG203926 (20).

FIG 1.

Structures of novel antiviral compounds CCG32091 and CCG203926 (A), CCG205432 (B), and CCG206381 (C). The IC₅₀, CC₅₀, and selectivity indices (SI) for each compound in the WEEV replicon assay are shown for reference (13, 19, 20). The terminal carboxamide R2 group, the main modification site for production of third-generation compounds, is circled in the structure for the second-generation compound, CCG203926.

Since the primary target cells of neurotropic alphaviruses in the CNS are neurons (5–8), we performed the initial assessment of CCG205432 and CCG206381 antiviral activity against WEEV in BE(2)-C human neuronal cells (Fig. 2A to C), an in vitro model used extensively to study WEEV replication and pathogenesis (23–25, 27). We initially examined cytotoxicity and found that CCG205432 showed mild toxicity in BE(2)-C cells at high concentrations, with CC₅₀ values of 60 to 80 μM as determined by MTT, CellTiter-Glo, and CytoTox-Glo assays (data not shown), similar to results observed in BSR-T7 cells (Fig. 1B). CCG206381 showed no measurable toxicity in BE(2)-C cells up to 100 μM (data not shown). In direct antiviral assays, CCG205432 potently suppressed infectious WEEV production after a low-inoculum (multiplicity of infection [MOI] = 0.1) infection, with a maximal titer suppression of >100-fold at 50 μM and an IC₅₀ of 1.8 μM (Fig. 2A). Although the IC₅₀ in virus production assays was 3-fold higher than in replicon assays (Fig. 1B), CCG205432 was still >10-fold more potent than ribavirin in suppressing WEEV production. In contrast, the IC₅₀ of CCG206381 in virus production assays was 13.9 μM, similar to that for ribavirin (Fig. 2A), but was 20-fold higher than its IC₅₀ in replicon assays (Fig. 1B). We obtained similar IC₅₀ values for CCG205432 and CCG206381 when we infected BE(2)-C cells with FMV, a lower-virulence WEEV serogroup alphavirus (28) that can be safely handled under BSL2 conditions (data not shown). Treatment of BE(2)-C cells with CCG205432 or CCG206381 at 25 μM or 2.5 μM also suppressed WEEV-induced cytopathic effect (CPE) and increased cell viability by approximately 2-fold at 24 hpi (Fig. 2B). Finally, we examined CPE reduction in differentiated BE(2)-C/m human neuronal cells, where WEEV-induced cell death is delayed despite similar levels of virus production (23), and where the rapid shutdown of host cell translation seen with alphavirus infections in most mammalian cells is significantly delayed (25). The delayed virus-induced CPE and translational suppression in differentiated BE(2)-C/m cells did not significantly alter the protective activity of CCG205432 or CCG206381 when we examined viability up to 72 hpi (Fig. 2C). However, consistent with the results of the virus production assays (Fig. 2A), CCG205432 was more active than CCG206381 in reducing WEEV-induced CPE (Fig. 2B and C).

FIG 2.

CCG205432 and CCG206381 are active against WEEV in cultured cells. (A) BE(2)-C human neuronal cells were infected with WEEV at an MOI of 0.1 and simultaneously treated with decreasing concentrations of the indicated compound. Virus titers in cell culture supernatants were determined at 24 hpi by plaque assay, and IC₅₀ values for each compound are shown on the graph. (B) BE(2)-C cells were infected as for panel A and treated with vehicle control (DMSO), CCG205432, or CCG206381 at the indicated concentrations. Cell viability was measured at 24 hpi by MTT assay. Results are presented as percent viability of that for compound-treated but uninfected controls. **, P < 0.0001. (C) Differentiated BE(2)-C/m neuronal cells were infected as for panel A and simultaneously treated with the indicated compound. Residual cell viability was measured at 24, 48, and 72 hpi by MTT assay. (D) Infectious virion production in single-step growth assays. BE(2)-C cells were infected with WEEV at an MOI of 10 and treated with DMSO or the indicated compound at a concentration of 25 μM. Virus titers in tissue culture supernatants were determined by plaque assay at 6, 12, 24, and 48 hpi. Plaque assay sensitivity was 10² PFU/ml. *, P < 0.05 for ribavirin and CCG205432 at 12 hpi, for all compounds at 24 hpi, and for CCG205432 and CCG206381 at 48 hpi. (E) Quantitative RT-PCR analysis of WEEV RNA accumulation in single-step growth assays. Cells were infected and treated as for panel D, total RNA was harvested at the indicated time points, and primers corresponding to either the nsP1 or E1 WEEV genome were used to amplify and quantify either genomic (nsP1) or genomic plus subgenomic (E1) RNA accumulation. Results are presented as WEEV RNA levels relative to those for infected DMSO-treated control cells. *, P < 0.05; **, P < 0.005. (F) Virion production in various mammalian cells. Individual cell lines were infected with WEEV at an MOI of 0.1 and simultaneously treated with DMSO or the indicated compound at a concentration of 25 μM. Virus titers in tissue culture supernatants were determined by plaque assay at 24 hpi for BE(2)-C, BE(2)-C/m, BHK-21, and Vero cells, 48 hpi for CHO, HEK293, and SH-SY5Y cells, and 72 hpi for Huh-7 and U87 cells, which represented predetermined optimal time periods for WEEV virion production in individual cell lines (17). *, P < 0.05; **, P < 0.005 compared to results for DMSO-treated controls.

We confirmed the antiviral activity of CCG205432 and CCG206381 against WEEV in single-step growth assays using BE(2)-C cells infected with virus at an MOI of 10 (Fig. 2D and E). We used CCG205432, CCG206381, and the positive control ribavirin all at a concentration of 25 μM, and we measured virus titers at 6, 12, 24, and 48 hpi (Fig. 2D). All three compounds reduced WEEV production by 12 hpi, and both CCG205432 and CCG206381 caused stable 10- to 20-fold reductions in virus titers at 24 to 48 hpi, where CCG205432 produced the largest decline in titers (Fig. 2D, open triangles). We also measured viral RNA accumulation by reverse transcription-quantitative PCR (qRT-PCR) up to 24 hpi and found that both CCG205432 and CCG206381 suppressed accumulation of viral genomic and subgenomic RNA at all time points examined after infection, whereas ribavirin was less active in these assays (Fig. 2E).

To further characterize the antiviral properties of CCG205432 and CCG206381, we examined their activity against WEEV in several cell lines derived from various mammalian species and tissues (Fig. 2F and Table 1). We infected cells at a low inoculum (MOI = 0.1) and used predetermined optimal time points to harvest supernatants and analyze WEEV titers (17). Both compounds suppressed infectious WEEV production in all cell lines tested, although CCG205432 was generally more active than CCG206381 when used at a concentration of 25 μM (Fig. 2F). The magnitude of suppression varied between cell lines from approximately 7- to 159-fold for CCG205432 and 4- to 93-fold for CCG206381 (Table 1). This variance was not due to differential compound toxicities (data not shown) or intrinsic characteristics of the cell lines, such as the expression of drug efflux pumps, as the pattern of suppression varied significantly between CCG compounds and the positive control, ribavirin. For example, while all three compounds suppressed WEEV production in Huh-7 cells, with <5-fold differences in titer reduction, their activities in BHK-21, SH-SY5Y, and U87 cells were quite divergent (Table 1). Taken together, these results indicate that the third-generation compounds, CCG205432 and CCG206381, have demonstrable antiviral activity against infectious WEEV in cultured cells, where CCG205432 has greater potency.

TABLE 1.

WEEV titer reduction mediated by CCG205432 or CCG206381 in various mammalian cell lines

Cell line	Species	Tissue/cell type	WEEV titer (fold reductiona) with:
			Ribavirin	205432	206381
BE(2)-C	Human	Neuron	4	78	6
BE(2)-C/m	Human	Differentiated neuron	ND	144	19
BHK-21	Hamster	Kidney fibroblast	263	7	4
HEK293	Human	Kidney epithelium	27	114	11
Vero	Primate	Kidney epithelium	1	13	11
CHO	Hamster	Ovarian epithelium	68	14	12
Huh-7	Human	Hepatocyte	18	63	93
SH-SY5Y	Human	Neuron	4	159	37
U87	Human	Astrocyte	3	70	28

^aValues represent the average decrease in infectious WEEV production in cells treated with 25 μM ribavirin, CCG205432, or CCG206381. Tissue culture supernatants were harvested at 24 hpi for BE(2)-C, BE(2)-C/m, BHK-21, and Vero cells, 48 hpi for CHO, HEK293, and SH-SY5Y cells, and 72 hpi for Huh-7 and U87 cells (17). ND, not determined.

CCG205432 and CCG206381 are active against VEEV, CEV, EMCV, and SeV, but not DENV.

The first- and second-generation compounds, CCG32091 and CCG203926, respectively, have activity against WEEV and the related alphaviruses FMV and SINV (13, 19, 20). To explore the breadth of third-generation compound antiviral activity, we examined the ability of CCG205432 and CCG206381 to suppress replication of RNA viruses from the Togaviridae (VEEV), Bunyaviridae (CEV), Picornaviridae (EMCV), Paramyxoviridae (SeV), and Flaviviridae (DENV) families (Fig. 3). For VEEV, CEV, and EMCV, we used infectious nonrecombinant virus and measured the effects of CCG205432 or CCG206381 (25 μM and 2.5 μM concentrations) on virus production in BE(2)-C cells (Fig. 3A). In initial time course experiments, all three viruses showed peak production by 24 hpi after a low-inoculum (MOI = 0.1) infection, although the titer magnitude varied by almost 100,000-fold, from 10⁵ PFU/ml for EMCV to 10¹⁰ PFU/ml for VEEV (17). Both third-generation compounds suppressed the production of infectious VEEV, CEV, and EMCV, although CCG205432 was more potent than CCG206381 (Fig. 3A), similar to results with WEEV (see Fig. 2). Both compounds also suppressed recombinant GFP-SeV replication at multiple time points after infection (Fig. 3B). In contrast, neither CCG205432 nor CCG206381 suppressed DENV replication of two different serotypes, DENV-2 and DENV-4, whereas MPA potently inhibited DENV replication (Fig. 3C), as expected (29, 30). Surprisingly, both CCG205432 and CCG206381 enhanced DENV RNA accumulation in BE(2)-C cells (Fig. 3C), which was confirmed by RT-PCR analysis of viral RNA accumulation in infected Vero cells and reporter gene expression in DENV replicon-bearing BHK-21 cells (data not shown). Taken together, these results indicate that CCG205432 and CCG206381 have broad, but not unlimited, activity against a range of RNA viruses.

FIG 3.

CCG205432 and CCG206381 have antiviral activity against VEEV, CEV, EMCV, and SeV, but not DENV. (A) Activity against VEEV, CEV, and EMCV. BE(2)-C cells were infected with the indicated virus at an MOI of 0.1 and simultaneously treated with the indicated compound. Virus titers in tissue culture supernatants were determined by plaque assay at 24 hpi, which was determined to be an optimal period for peak WEEV production (17). *, P < 0.05; **, P < 0.005 compared to results for DMSO-treated controls. (B) Activity against SeV. BE(2)-C cells were infected with GFP-SeV at an MOI of 1 and treated simultaneously with the indicated compound. GFP fluorescence was measured at 6, 12, 24, 36, 48, 72, and 96 hpi. Similar results were obtained with higher (MOI = 10) and lower (MOI = 0.1) inocula concentrations (data not shown). (C) Activity against DENV. BE(2)-C cells were infected with either DENV-2 or DENV-4 and treated with DMSO, MPA at a concentration of 25 μM, or the indicated CCG compound. Total RNA was harvested at 96 hpi, and DENV RNA accumulation was analyzed by qRT-PCR. Preliminary experiments indicated the absence of detectable DENV replication in BE(2)-C cells with either serotype prior to 72 to 96 hpi (data not shown). Results are presented as DENV RNA levels relative to those for infected DMSO-treated control cells, where the dashed line is provided for baseline reference. Viral RNA levels in MPA-treated cells were <0.1% and 0.3% those of DMSO controls for DENV-2 and DENV-4, respectively.

Time-of-addition and chimeric WEEV-SINV PIP assays confirm CCG205432 antiviral activity correlates with viral RNA replication.

We used WEEV replicons for the majority of the small-molecule drug discovery and development efforts leading to the production of the third-generation compounds, CCG205432 and CCG206381 (13, 19, 20). Thus, suppression of viral RNA replication is the likely mechanism whereby these antivirals inhibit WEEV and related alphaviruses. To support this conclusion, we conducted time-of-addition studies with infectious FMV in BE(2)-C cells (Fig. 4A and B). We first examined the temporal appearance of viral RNA by qRT-PCR in single-cycle growth assays in the absence of inhibitor and detected FMV RNA accumulation beginning at 6 hpi with a subsequent rapid increase after 10 to 12 hpi (Fig. 4A). We then treated cells with CCG205432 (25 μM) at various intervals from 2 h before to 18 h after FMV infection and measured virus titers in tissue culture supernatants at 24 hpi (Fig. 4B). We found equivalent inhibitory activities with early or delayed treatment up to 10 to 12 hpi, after which time the antiviral activity of CCG205432 diminished, corresponding to the rapid increase in viral RNA accumulation (Fig. 4A).

FIG 4.

Efficacy of CCG205432 inhibition correlates with WEEV RNA replication. (A) Time course of viral RNA accumulation. BE(2)-C cells were infected with FMV at an MOI of 10, total RNA was harvested at the indicated times postinfection, and FMV RNA accumulation was quantified by qRT-PCR using primers corresponding to an E1 genome region. Results are presented as the fold change (ΔΔC_T) from results for the 2-hpi time point, which represented input virion RNA. (B) Virus titers after delayed addition of CCG205432. Cells were infected as described above, CCG205432 was added at the indicated time points (−2 to 18 hpi) to a final concentration of 25 μM, and FMV titers in tissue culture supernatants were determined by plaque assay at 24 hpi. The value at 24 hpi represents cells treated with control DMSO at the time of infection. (C) Activity against chimeric WEEV-SINV PIPs. HEK293 cells were infected with predetermined optimal concentrations of chimeric WEEV-SINV PIPs and simultaneously treated with decreasing concentrations of the indicated compound. fLUC activity was measured at 9 hpi. Results represent percent fLUC activity of that of the DMSO-treated controls. IC₅₀ values for each compound are shown on the graph. (D) Control prebinding treatment PIP assay. PIPs were preincubated with optimized dilutions of antibodies against HA (negative control) or SINV glycoprotein E2, and fLUC activity was measured at 9 hpi. (E) PIP binding assays with CCG205432 and MPA. The indicated compounds were preincubated with PIPs prior to cell binding (Pre Rx) or were added to cultures after PIPs were allowed to bind to cells at 4°C (Post Rx), and fLUC activity was measured at 9 hpi. CCG205432 and MPA were used at concentrations of 1.5 and 1.0 μM, respectively, which represented concentrations 2- to 3-fold above IC₅₀ values (see Fig. 6C).

We also developed and validated a WEEV PIP replication system to evaluate the activity of candidate inhibitors on the processes of viral binding, entry, and RNA replication in the absence of infectious virion production, similar to systems developed for SINV (31) and VEEV (32). We used a three-component system with the SINV capsid and envelope proteins encoded on separate plasmids to produce chimeric WEEV-SINV PIPs when mRNAs derived from these plasmids were coelectroporated with mRNA derived from the plasmid encoding the WEEV-fLUC replicon. We chose to produce chimeric PIPs, as initial studies indicated low-efficiency production with homotypic helper vectors (data not shown), and because WEEV is a natural recombinant virus whose envelope genes were originally derived from a SINV-like virus (33). Furthermore, the presence of SINV envelope proteins facilitated the use of well-described monoclonal antibodies against defined envelope regions (22) as a control for prebinding neutralization assays (see below). We used HEK293 cells as PIP assay targets, as preliminary experiments indicated robust fLUC activity in this human epithelial cell line after chimeric WEEV PIP infection, whereas fLUC activity in BE(2)-C cells was significantly lower under similar experimental conditions (data not shown). The WEEV PIP IC₅₀ values for CCG205432 and MPA in HEK293 cells (Fig. 4C) were similar to WEEV replicon IC₅₀ values in transfected BSR-T7 cells and infectious WEEV or FMV IC₅₀ values in BE(2)-C cells for both compounds (Fig. 1 and 2) (20).

We subsequently used HEK293 cells and chimeric WEEV PIPs to complete binding-competition experiments with CCG205432 and MPA (Fig. 4D and E). As an additional control for these experiments, neutralizing antibodies against SINV glycoprotein E2 were used, where preincubation with PIPs prior to cell addition resulted in measurable suppression of fLUC activity compared to that for control HA antibodies (Fig. 4D). In contrast, when we preincubated PIPs with CCG205432 or MPA prior to addition to cells, no suppression of fLUC activity was seen (Fig. 4E, closed bars), whereas addition of either compound after PIPs were allowed to bind cells resulted in significant fLUC suppression (Fig. 4E, open bars). Taken together with the time-of-addition studies described above, these results support the conclusion that CCG205432 functions at a step in the WEEV replication cycle downstream of virus binding and entry and that its activity correlates with suppression of viral RNA replication.

CCG205432 does not directly target WEEV RdRp or other viral enzyme activities readily amenable to drug resistance selection.

We used a phenotypic cell-based assay rather than a targeted assay against a specific viral or cellular protein to identify, validate, and develop novel compounds with inhibitory activity against WEEV and related viruses in an effort to broaden the potential targets and mechanisms of action for candidate antivirals (13, 19, 20). The use of phenotypic cell-based antiviral assays also potentially increases the probability of selecting compounds having a cellular rather than a viral target (34). However, both thienopyrrole- and indole-based compounds have been developed as allosteric inhibitors of hepatitis C virus (HCV) NS5B RNA polymerase activity (35, 36), leading us to examine whether CCG205432 directly inhibited WEEV RNA polymerase activity using an in vitro RdRp assay (Fig. 5). Alphaviruses assemble their RNA replication complexes on intracellular membranes that can be readily isolated from infected cells by differential centrifugation (26). In vitro alphavirus RdRp activity can subsequently be measured in a cell-free system by analyzing the incorporation of radiolabeled ribonucleotides in the presence of actinomycin D, which inhibits cellular DNA-dependent RNA polymerases but does not disrupt alphavirus RdRp activity (37). We found that WEEV RdRp activity was suppressed by both EDTA (Fig. 5, lane 3) and phosphonoacetic acid (Fig. 5, lane 4), a known calicivirus RdRp inhibitor (38), whereas CCG205432 had no notable inhibitory activity even at concentrations of >300-fold above its WEEV replicon IC₅₀ (Fig. 5, lanes 5 to 7). Similar results were obtained with CCG206381 (data not shown). The high mutation rate of viral RdRps facilitates the rapid selection of drug-resistant mutant viruses in cell culture, which can be used in a forward genetics approach to identify candidate targets for antiviral compounds. This approach has been used successfully to help identify the mechanism of action for several alphavirus inhibitors (39–41). Despite repeated attempts, we were unable to generate WEEV or FMV mutants resistant to CCG205432 by daily serial passage for 20 days in cultured Vero or BE(2)-C cells (data not shown). These results indicate that CCG205432 does not directly target WEEV RdRp or other viral enzyme activities that would promote the development of drug-resistant viral mutants.

FIG 5.

CCG205432 does not inhibit WEEV RdRp activity in vitro. Membrane fractions from mock-infected control BHK-21 cells (lane 1) or WEEV-infected BHK-21 cells (lanes 2 to 7) were incubated with ³²P-labeled UTP and unlabeled ribonucleotides with vehicle control DMSO (lane 2), positive control 1 mM EDTA (lane 3) or 200 μM phosphonoacetic acid (PAA; lane 4), or decreasing concentrations of CCG205432 (lanes 5 to 7). Radiolabeled reaction products were separated by non-denaturing agarose gel electrophoresis and detected by autoradiography. Results from one representative experiment of three are shown.

CCG205432 and CCG206381 target a host factor that modulates cellular cap-dependent translation.

We initially explored a focused approach to examine potential cellular targets for CCG205432 and CCG206381 using combination treatments to determine possible synergy or antagonism between inhibitors with known mechanisms of action against cellular pathways. We used WEEV replicons in BSR-T7 cells and pairwise combination treatments with CCG205432, CCG206381, the purine biosynthesis inhibitor, MPA (30, 42), or the mitochondrial electron transport chain and pyrimidine biosynthesis inhibitor, antimycin A (17), to calculate Chou-Talalay parameters and combination index values (43). We found additive effects between CCG205432 and CCG206381 but moderate synergy between CCG205432 and MPA or antimycin A (Table 2). These results suggested that CCG205432 and CCG206381 target a similar pathway not associated with nucleotide biosynthesis. We confirmed this conclusion by complementation studies, where neither purine nor pyrimidine supplementation rescued CCG205432-mediated inhibition of WEEV replicon activity (data not shown).

TABLE 2.

Synergy antagonism assays of CCG205432, CCG206381, antimycin A, and MPA

Compound/combinationa	Molar ratiob	Chou-Talalay parameterc			Combination indexd	P valuee
		Potency (Dm)	Shape (m)	Conformity (r)
CCG205432	NA	0.65 ± 0.09 (μM)	−0.81 ± 0.03	−0.99 ± 0.01	NA	NA
CCG206381	NA	1.06 ± 0.27 (μM)	−1.03 ± 0.07	−0.96 ± 0.01	NA	NA
Antimycin A	NA	4.2 ± 0.9 (nM)	−1.28 ± 0.20	−0.94 ± 0.02	NA	NA
MPA	NA	0.29 ± 0.05 (μM)	−1.37 ± 0.33	−0.97 ± 0.02	NA	NA
CCG205432 + CCG206381	1:1	NA	−0.77 ± 0.04	−0.96 ± 0.01	1.01 ± 0.12	NS
CCG205432 + antimycin A	233:1	NA	−0.96 ± 0.05	−0.96 ± 0.01	0.66 ± 0.07	0.017
CCG205432 + MPA	3:1	NA	−1.37 ± 0.05	−0.98 ± 0.01	0.51 ± 0.09	0.012

^aBSR-T7 cells transfected with pWR-LUC were treated with the indicated compounds at a range of concentrations around their individual IC₅₀ concentrations or at a fixed molar ratio around the CCG205432 IC₅₀ concentration.

^bFixed molar ratios were determined based on IC₅₀ values to obtain equipotency concentrations. NA, not applicable.

^cWe used the CompuSyn program (www.combosyn.com) to calculate Chou-Talalay parameters and combination index values.

^dThe combination index is a quantitative measure of synergistic, additive, or antagonistic effects between compounds. This parameter indicates synergism with values of <1, nearly additive effects with values of ∼1, and antagonism with values of >1. The range of values is 0 to 1 for synergy and 1 to ∞ for antagonism.

^eP values were calculated using the results of a one-sample Student's t test compared to a hypothetical mean value of 1. NS, not significant.

To further identify potential cellular targets for the antiviral activity of CCG205432 and CCG206381, we used an unbiased genome-wide transcriptional microarray analysis. We identified 108 upregulated and 5 downregulated genes in BE(2)-C cells treated with CCG205432 and 85 upregulated and 4 downregulated genes in the same cells treated under identical conditions with CCG206381 (see Table S1 in the supplemental material). Furthermore, there was a 40% concordance in the gene sets between the two treatment groups when all genes regulated ≥2-fold were compared, and this increased to 90% concordance when only genes regulated ≥2.5-fold were analyzed. There were 57 genes coregulated in BE(2)-C cells treated with CCG205432 or CCG206381, and there was a significant correlation (R = 0.97; P < 10⁻³²) in the magnitude of transcriptional changes of individual genes in cells treated with either compound. We subsequently conducted in silico analyses with differentially regulated genes that were assigned to known cellular pathways, using Ingenuity Pathway Analysis software. We identified 22 canonical pathways preferentially modulated after treatment with CCG205432 or CCG206381, where far and away the most significantly associated pathway was eukaryotic initiation factor 2 (eIF2) signaling (Table 3). The eIF2 signaling pathway is an essential component of gene expression and translational control in eukaryotes, it plays a critical role in cellular stress responses to multiple stimuli, including viral infections, and it is targeted for subversion by numerous viruses (44).

TABLE 3.

Ingenuity canonical pathways preferentially modulated in BE(2)-C neuronal cells treated with CCG205432 or CCG206381

Canonical pathwaya	No. of genes	P value
eIF2 signaling	17	3.98 × 10−14
Mitochondrial dysfunction	9	0.000002
Breast cancer regulation by stathmin 1	9	0.000015
RhoGDI signaling	8	0.000049
mTOR signaling	8	0.000100
Signaling by rho family GTPases	8	0.000355
Regulation of eIF4 and p70S6K signaling	6	0.000891
fMLP signaling in neutrophils	5	0.001380
Androgen signaling	5	0.001585
Remodeling of epithelial adherens junctions	4	0.001698
CCR3 signaling in eosinophils	5	0.001778
CCR5 signaling in macrophages	4	0.001905
Ephrin B signaling	4	0.002344
IL-8 signaling	6	0.002951
α-Adrenergic signaling	4	0.004365
Epithelial adherens junction signaling	5	0.004898
Gαq signaling	5	0.005248
Gap junction signaling	5	0.006457
Tec kinase signaling	5	0.006457
Protein kinase A signaling	8	0.008128
Phospholipase C signaling	6	0.008511
Role of NFAT in regulation of immune response	5	0.009550

^aThe listed pathways were identified as significantly modulated by CCG205432 or CCG206381 if P was <0.01 and there were at least three genes that were up- or downregulated. GDI, guanine nucleotide dissociation inhibitor; fMLP, N-formyl-Met-Leu-Phe; IL-8, interleukin-8; Gα_q, G protein alpha-q; NFAT, nuclear factor of activated T cells.

Inhibition of cellular or viral gene expression through translational suppression has been identified as a candidate mechanism of action for a recently described novel antiviral compound active against some RNA viruses (45). Although transcriptome analyses did not reveal global suppression of host transcription, we examined whether CCG205432 or CCG206381 could directly modulate reporter gene expression in the absence of virus (Fig. 6). To facilitate these experiments, we constructed a series of fLUC expression plasmids to generate either nuclear- or cytoplasmic-transcribed mRNAs (Fig. 6A). For nucleus-derived transcription of capped and polyadenylated mRNAs, we used a plasmid containing the SV40 promoter (pSV40-LUC). For cytoplasm-derived transcription, we used plasmids containing the bacteriophage T7 RNA polymerase promoter with or without a polyadenylation signal (pT7/CITE-LUC and pT7/CITE-LUC(-A), respectively) and BSR-T7 cells for transfection. Transcripts produced by T7 RNA polymerase in cells are largely uncapped (46), and thus we included a CITE segment, which contains the internal ribosome entry site (IRES) from EMCV (47), to improve translation efficiency. As controls for these experiments, we used hippuristanol, which is an eIF4A helicase inhibitor that suppresses both cap-dependent and EMCV IRES-mediated translation (48), and we also measured cell viability by MTT assay. Both CCG205432 (Fig. 6B, right) and CCG206381 (data not shown) potently suppressed pSV40-LUC expression in the absence of significant cellular toxicity but had minimal impact on the expression of pT7/CITE-LUC or pT7/CITE-LUC(-A). In contrast, hippuristanol suppressed fLUC expression from all three vectors and also displayed more cellular toxicity (Fig. 6B, left).

FIG 6.

Novel thienopyrrole- and indole-based antiviral compounds target a host factor that modulates cellular gene expression. (A) Schematic of fLUC-encoding reporter plasmids used for mRNA expression in cells and in vitro. Plasmid designations are shown on the left, and the large block arrows indicate the promoter. CITE, cap-independent translation element; A_n, polyadenylation sequence. (B) fLUC reporter gene expression in plasmid-transfected cells. BSR-T7 cells were transfected with pSV40-LUC, pT7/CITE-LUC, or pT7/CITE-LUC(-A) and treated with decreasing concentrations of hippuristanol (left) or CCG205432 (right). Both viability via MTT assay and fLUC activity were measured 20 h later. (C) Thienopyrrole- and indole-based antiviral compounds inhibit pSV40-LUC reporter gene expression. Cells were transfected with pSV40-LUC and treated with decreasing concentrations of select compounds from the indicated chemical generation series. fLUC activity was measured 20 h later. For reference, the terminal carboxamide R2 group (see Fig. 1A) for each tested compound is shown on the right, along with the IC₅₀ calculated from the graphs on the left. (D) Correlation between WEEV replicon and pSV40-LUC vector IC₅₀ values for thienopyrrole- and indole-based antiviral compounds. (E) fLUC reporter gene expression in mRNA-transfected cells. BSR-T7 cells were transfected with reporter fLUC mRNA containing either a 5′ m⁷G cap or CITE and treated with decreasing concentrations of CCG205432. Both viability via MTT assay and fLUC activity were measured 20 h later. Transcripts were generated with T7 RNA polymerase using pT7-LUC or pT7/CITE-LUC (see Fig. 6A) for m⁷G-capped and uncapped mRNAs, respectively. (F) fLUC reporter gene in vitro translation. Nuclease-treated RRLs were incubated with the indicated mRNAs and 10 μM cycloheximide (Cx), 10 μM hippuristanol (Hip), or decreasing concentrations of CCG205432. fLUC activity was measured after 90 min of incubation. Transcripts for in vitro translation were generated as described above for panel E. IVT, in vitro translation.

We have generated thienopyrrole- and indole-based antiviral compounds with a wide range of activity, including striking enantiospecificity within the terminal carboxamide R2 group (see Fig. 1A) in the second-generation series (13). We utilized this diversity of antiviral activity to examine whether suppression of WEEV replicon expression correlated with pSV40-LUC inhibition (Fig. 6C and D). First-, second-, and third-generation compounds all suppressed pSV40-LUC expression (Fig. 6C), and there was a significant correlation between the IC₅₀ values for WEEV replicon and pSV40-LUC inhibition (Fig. 6D). Of particular note, the (R) isomer in the second generation series, CCG203926, actively suppressed pSV40-LUC expression, whereas the corresponding (S) isomer, CCG203927, was inactive in this assay (Fig. 6C, middle), identical to results obtained using these enantiomers in assays that examined inhibition of WEEV replicon activity, infectious WEEV or SINV production and CPE reduction in cultured cells, and efficacy in SINV-infected mice (13). Furthermore, the (R) isomer in the third-generation series, CCG212090, was more active in suppressing pSV40-LUC expression than the corresponding racemate, CCG206329 (Fig. 6C, right), similar to results with WEEV replicon antiviral activity (20).

Based on the plasmid transfection experiments described above, we could not differentiate the activities of CCG205432 and related compounds on reporter gene nuclear transcription; mRNA processing, export, or stability; or cap-dependent translation. To further investigate the mechanism(s) whereby these antiviral compounds suppressed gene expression, we directly transfected mRNA encoding fLUC into BSR-T7 cells and measured reporter gene expression. Both CCG205432 (Fig. 6E) and CCG206381 (data not shown) suppressed fLUC expression in cells transfected with a capped mRNA, but not in cells transfected with a CITE-containing mRNA, similar to results obtained with plasmid-transfection experiments (Fig. 6B). Finally, we investigated whether CCG205432 or CCG206381 suppressed translation in vitro using RRLs and transcribed mRNAs. Neither CCG205432 (Fig. 6F) nor CCG206381 (data not shown) notably altered translation from capped or uncapped mRNA templates, even at concentrations of >150-fold above their IC₅₀ values for WEEV replicon inhibition or suppression of pSV40-LUC expression. In contrast, both cycloheximide and hippuristanol, which directly block translation elongation and eIF4A-dependent translation initiation, respectively (48, 49), suppressed RRL-mediated translation from all RNA templates in vitro (Fig. 6F) (data not shown). Similar results were obtained when we used mRNA transcripts without polyadenylated tails, in wheat germ extracts, or when we used a coupled in vitro transcription-translation system with plasmids rather than purified mRNA transcripts (data not shown). Taken together, these results suggest that CCG205432 and related first-, second-, and third-generation thienopyrrole- and indole-based antiviral compounds target a host factor that modulates cellular or viral gene expression. Furthermore, they suggest that these novel antiviral compounds function, in part, through suppression of cap-dependent translation without directly inhibiting ribosome-mediated initiation or elongation.

CCG205432 does not inhibit WEEV RNA replication in a cell-based trans-replicon system.

The observation that CCG205432 and CCG206381 did not disrupt CITE-mediated cap-independent translation in cells (Fig. 6) presented the opportunity to directly examine whether these novel antiviral compounds inhibited WEEV nsP enzymatic activity using a trans-replicon system (Fig. 7). For some positive-sense RNA viruses, nsP translation and viral RNA replication, including subgenomic RNA synthesis, can be unlinked by providing a replication-competent template that does not include functional nsP-coding sequences and a separate replication-incompetent mRNA encoding nsPs that are active in trans. Spuul et al. recently described a novel trans-replication system for alphaviruses using Semliki Forest virus (50), and we developed a similar system for WEEV (Fig. 7A). We removed the majority of the nsP1-to-nsP4-coding sequences from the cis-replicon plasmid, pWR-LUC, to produce a template, pWR-ΔLUC. This template still contains the authentic viral 5′ and 3′ termini with corresponding untranslated regions and the subgenomic promoter-driven fLUC reporter gene, but it does not encode functional nsPs and therefore cannot initiate viral RNA replication. To provide functional nsPs, we cloned the entire WEEV nsP1-to-nsP4-coding sequence behind a T7 promoter and CITE to produce pW-nsP. This plasmid allowed cytoplasmic transcription and cap-independent translation in BSR-T7 cells, thereby avoiding the inhibitory activity of CCG205432 and related compounds on cellular cap-dependent translation. We initially validated the trans-replication system in the absence of inhibitors (Fig. 7B). BSR-T7 cells transfected with either pW-nsP or pWR-ΔLUC alone showed minimal fLUC expression over baseline, whereas cells cotransfected with both plasmids showed an approximate 100-fold increase in fLUC expression, albeit an order of magnitude lower than the increase seen with the cis-replicon pWR-LUC.

FIG 7.

CCG205432 does not inhibit WEEV RNA replication using a trans-replicon system. (A) Schematic of plasmids encoding cis- or trans-WEEV replicons. The cis replicon-encoding plasmid pWR-LUC is shown for reference (19). Plasmid designations are shown on the left, and the large block arrows indicate the promoter. U, untranslated region; δRz, hepatitis δ ribozyme; A_n, polyadenylation sequence; SG_P, subgenomic promoter; CITE, cap-independent translation element. (B) fLUC reporter gene activity in BSR-T7 cells transfected with equivalent amounts of empty vector, pWR-LUC (LUC), pW-nsP (nsP), pWR-ΔLUC (ΔLUC), or both pW-nsP and pWR-ΔLUC. Results are expressed as relative luciferase units (RLU). (C) Viability and reporter gene activity in cells expressing WEEV trans-replicons treated with either CCG205432 or mycophenolic acid (MPA). BSR-T7 cells were transfected with optimized ratios of pW-nsP and pWR-ΔLUC and treated with decreasing concentrations of the indicated compound. Both viability and fLUC activity were measured 18 to 20 h later.

We subsequently transfected BSR-T7 cells with optimized ratios of pW-nsP and pWR-ΔLUC, treated cells with inhibitors for 18 to 20 h, and measured fLUC activity (Fig. 7C). As a control for these experiments, we used the purine biosynthesis inhibitor MPA (30, 42), which should be active in both cis- and trans-replication systems. Indeed, MPA potently suppressed fLUC expression in cells cotransfected with pW-nsP and pWR-ΔLUC with an IC₅₀ of approximately 100 nM (Fig. 7C, open triangles), similar to its antiviral activity in infectious virus, PIP, and cis-replication systems (17) (Fig. 4). In contrast, neither CCG205432 (Fig. 7C, open circles) nor CCG206381 (data not shown) had notable inhibitory activity in the WEEV trans-replication system. These results are consistent with the RdRp results (Fig. 5) and indicate that CCG205432 and related compounds do not directly inhibit WEEV nsP enzymatic activities, but rather target a host factor.

Third-generation indole-2-carboxamide inhibitors are active in vivo in mice infected with WEEV.

Candidate antiviral compounds that demonstrate activity in vitro frequently do not retain efficacy in vivo. For example, several recently developed pyrimidine biosynthesis inhibitors with potent antiviral activity against a range of viruses in cultured cells were inactive in corresponding virus-infected mice (51–53). We previously demonstrated that the second-generation derivative, CCG203926, improves survival in mice inoculated intracerebrally with a neuroadapted strain of SINV (13). To examine whether third-generation analogues were viable candidates for further development as antiviral compounds against virulent neurotropic alphaviruses that cause CNS disease in humans, we examined the antiviral activity of CCG205432 in mice infected with WEEV (Fig. 8A to C). The recombinant Cba-87 strain of WEEV is highly pathogenic in mice (54), and initial dose-titration experiments showed that subcutaneous inoculation of 10³ PFU routinely produced high mortality in C57BL/6 mice, with a mean day of death (MDD) between 10 and 11 days (data not shown). Pilot experiments also showed that 30 mg/kg CCG205432 delivered via intraperitoneal injection was well tolerated, and therefore we used this dose twice daily for 7 days starting on the day of infection. Vehicle-treated control mice showed progressive loss of body weight from day 4 after infection, overt clinical signs of disease by day 5 to 6, and 90% mortality by day 14, with a MDD of 10.6 ± 0.2 days. In mice treated with CCG205432, weight loss (Fig. 8A), clinical disease severity (Fig. 8B), and overall survival (Fig. 8C), were all improved compared to those for vehicle-treated control animals, and the MDD in CCG205432-treated mice was prolonged to 11.4 ± 0.3 days (P < 0.05). We also examined the impact of CCG205432 treatment on CNS virus replication in infected mice. Pilot experiments showed that CNS titers in multiple brain regions peaked 5 to 7 days after WEEV challenge (data not shown). Although mice treated with CCG205432 had decreased WEEV titers in the cerebrum at days 2, 4, and 6 after virus infection, this reduction did not reach statistical significance (P = 0.0515).

FIG 8.

Third-generation indole-based antiviral compounds reduce clinical disease severity and enhance survival in mice infected with WEEV. (A through C) Clinical disease severity and survival in WEEV-infected mice treated with CCG205432. C57BL/6 mice were infected subcutaneously with 10³ PFU WEEV and treated twice daily with DMSO or 30 mg/kg CCG205432 via intraperitoneal injection. Body weight (A), clinical disease severity scores (B), and mortality (C) were monitored daily for 14 days postinfection. Results shown are a composite of the results of four independent experiments, where n = 4 to 10 mice per group per experiment. (D) Structure of compound CCG209023. The circled R1 benzyl group is the only difference in structure from that of CCG205432 (see Fig. 1B). (E and F) Clinical disease severity (E) and survival (F) in WEEV-infected mice. C57BL/6 mice were infected as described above and treated twice daily with DMSO or 30 mg/kg CCG209023 via intraperitoneal injections. Clinical disease severity scores and mortality were monitored daily for 14 days postinfection. Results shown are a composite of the results of two independent experiments, where n = 5 to 10 mice per group per experiment. *, P < 0.05; **, P < 0.005 compared to results for DMSO-treated controls.

To further assess the in vivo activity of indole-2-carboxamide inhibitors, we examined clinical disease and survival in WEEV-infected mice treated with CCG209023, an additional third-generation compound closely related to CCG205432 (Fig. 8D to F). The chlorine on the R1 benzyl group of CCG205432 was removed in CCG209023 (Fig. 8D), which resulted in a 3-fold increase in replicon IC₅₀ values, but also a notable 7-fold increase in metabolic stability (20). Treatment with CCG209023 both reduced clinical disease severity (Fig. 8E) and improved overall survival (Fig. 8F) compared to those for vehicle-treated control mice given a lethal challenge of WEEV. Taken together, these results indicate that third-generation indole-2-carboxamide inhibitors decrease clinical disease severity and prolong survival in mice with WEEV encephalitis.

DISCUSSION

The lack of approved antiviral drugs for infections caused by neurotropic alphaviruses, if not most arboviruses, highlights the need for a more comprehensive strategy to prevent or treat infections caused by these virulent pathogens. In this report, we describe the antiviral characteristics of a novel class of indole-2-carboxamide inhibitors. We draw four main conclusions from our studies: (i) compounds CCG205432 and CCG206381 display potent antiviral activity against WEEV RNA replication and virion production in cultured cells; (ii) these compounds are active against several alphaviruses, but also inhibit the replication of bunyaviruses, picornaviruses, and paramyxoviruses in cultured cells; (iii) these compounds target a host factor that modulates cellular cap-dependent translation; and (iv) CCG205432 and CCG209023 have in vivo activity in WEEV-infected mice. These results support the continued preclinical development of indole-2-carboxamide inhibitors as novel antiviral agents against neurotropic alphaviruses and potentially against other RNA viruses as well.

We used a cell-based phenotypic assay and a WEEV replicon for the initial discovery (19) and SAR-directed development (13, 20) of the indole-2-carboxamide inhibitors. This approach does not exclude the identification of compounds that directly target viral proteins, as a potent HCV NS5A inhibitor in advanced clinical development, daclatasivir (BMS-790052), was identified using a similar replicon cell-based assay (55). However, RNA viruses utilize a number of known host factors for their replication (56), and this list will undoubtedly grow as we gain further understanding of the complex host-pathogen interactions that control virus replication. Thus, the number of cellular processes that serve as potential targets for antiviral agents likely exceeds the number of viral targets with pathogens such as alphaviruses, whose genome encodes a limited number of proteins with known enzymatic activities (4). Several recently published cell-based antiviral screens have identified a number of compounds that target host factors, from pyrimidine biosynthesis (17, 51, 57) to protein translation (45, 58). Targeting a cellular process also has the potential advantage of presenting a higher barrier for the development of resistance, a concept with substantial experimental and clinical support. For example, alisporivir (Debio 025) is another HCV inhibitor in advanced clinical development that targets host cyclophilin A, where the emergence of resistance is low and requires multiple viral mutations (59–61). However, cell-targeted antiviral compounds also have potential disadvantages, including toxicity and the potential for limited in vivo activity due to the redundancy of physiologic systems, such as exogenous dietary pyrimidine uptake mitigating the antiviral activity of dihydroorotate dehydrogenase inhibitors in mice (51). We found no evidence that CCG205432 or CCG206381 inhibited WEEV RdRp activity in vitro or viral RNA replication in cells when nsPs were provided in trans, suggesting that these compounds do not target alphavirus-encoded enzymatic activities, such as nsP1 methyltransferase (62), nsP2 helicase (63) or protease (64), or nsP4 terminal nucleotide transferase or polymerase (65) activities. Although we cannot definitively exclude a viral target for the indole-2-carboxamide inhibitors based on negative results, the preponderance of evidence supports our conclusion that these compounds target a host factor that modulates virus replication.

The precise molecular target and mechanism of action for CCG205432 and related indole-2-carboxamide compounds remain unknown, but the enantiospecific activity of these compounds indicates that the interaction between the target and inhibitor is precise. Furthermore, there was excellent correlation between potency in antiviral assays and pSV40-LUC reporter gene assays with first-, second-, and third-generation compounds, suggesting that the molecular target for these inhibitors modulates a cellular process readily amenable to experimental investigation. The identification of crucial cellular processes such as transcription or translation as antiviral targets presents obvious potential complications related to host toxicity. However, we saw no evidence for global transcriptional or translation suppression with either CCG205432 or CCG206381, suggesting a more selective inhibitory activity. For example, these compounds did not suppress expression from IRES-containing mRNA transcripts in cells or inhibit translation in vitro, thus differentiating their inhibitory activity from the general protein synthesis inhibitors cycloheximide and hippuristanol (48, 49). Furthermore, the selectivity indices of CCG205432 and CCG206381 are also both >100, indicating the existence of a therapeutic window that can potentially be exploited in the continued development of these compounds.

The observation that indole-2-carboxamide inhibitors disrupt translation of capped mRNAs within cells but not in vitro using cell extracts was intriguing but not without precedent. For example, rocaglamides are naturally occurring herbal compounds with potent anticancer activity that were originally identified as protein synthesis inhibitors (66), where these compounds suppress protein translation within cells but not using cell extracts in vitro (67). Detailed studies have revealed that rocaglamides target the regulatory cellular proteins prohibitin 1 and 2, thereby inhibiting the Raf-MEK-ERK signal transduction pathway that controls protein synthesis (68). In addition, rapamycin indirectly inhibits cap-dependent translation in cells by disrupting target-of-rapamycin (TOR) signaling (69). One intriguing hypothesis is that indole-2-carboxamides function through a similar indirect mechanism, which is consistent with our observation that treatment with CCG205432 or CCG206381 modulates eIF2 signaling pathway component expression. The results with the trans-replication system suggest that one activity of this hypothetical cell-signaling pathway is to modulate translation of authentic capped viral genomic, but not subgenomic, mRNAs, as neither CCG205432 nor CCG206381 inhibited viral RNA replication and subgenomic reporter gene expression in the presence of CITE-driven nsP translation. Experiments are currently in progress to further examine the precise molecular target(s) of indole-2-carboxamide compounds, and we have generated a series of active CCG205432 analogues with cross-linking chemical groups to facilitate a chemoproteomics approach for definitive target identification (70).

The indole-2-carboxamide inhibitors described herein share interesting parallels and contrasts with two recently described novel antivirals. Kaur et al. used an immunofluorescence-based assay to screen a library of highly purified natural products against the Old World alphavirus chikungunya virus and identified harringtonine, an alkaloid derived from the plant Cephalotaxus harringtonia, as a novel antiviral that blocks protein translation (58). Detailed mechanism-of-action studies, breadth of antiviral activity outside the alphavirus genus, and in vivo activity were not explored with this natural product-derived compound, so a more detailed comparison with CCG205432 and related compounds is not yet possible. However, Wang et al. recently described a novel benzomorphan compound (NITD-451) with in vitro and in vivo antiviral activity (45) that shares intriguing contrasts with our indole-2-carboxamide compounds. The benzomorphan series of compounds were first developed as antivirals against DENV, and NITD-451 shows potent activity against several flaviviruses but not WEEV, whereas CCG205432 is active against WEEV and related alphaviruses but not DENV. Furthermore, although both series of compounds have enantiospecific activity and modulate cellular gene expression, NITD-451 suppresses translation in cells and in in vitro assays (45), whereas CCG205432 was active only in cells and did not inhibit expression from mRNA templates in vitro, suggesting different mechanisms of action. However, Wang et al. were also unable to generate NITD-451-resistant DENV, similar to our failed attempts to generate CCG205432-resistant WEEV and FMV, consistent with both compounds having host factors as targets. Further molecular mechanism-of-action studies with NITD-451 and CCG205432 may reveal additional details that shed light on the complex interactions between RNA viruses and cells that ultimately control virus replication and pathogenesis.

Numerous compounds have been shown to suppress alphavirus replication in cultured cells, but only a few have shown any activity in animal models (13–17). We demonstrate that CCG205432 and CCG209023 reduce clinical disease and improve survival in mice given a lethal challenge of WEEV. Although the cumulative in vivo activity is modest, it represents a notable achievement in the preclinical development of this novel class of antiviral compounds. We did not design the in vivo efficacy trials to optimize dosing regimens or to emulate potential outbreak scenarios such as pre- or postexposure treatment. Our current indole-2-carboxamide compounds are still in development and have not been fully optimized from the standpoint of chemical parameters such as solubility, metabolic stability, and CNS penetration. Rather, we set relatively stringent criteria with initial animal trials to identify evidence for antiviral activity in vivo, thereby advancing only compounds with the highest probability for productive future development. For example, we limited drug injections to twice daily during the first 7 days after infection, in part due to the practical considerations of performing treatment trials under ABSL3 conditions. However, preliminary pharmacokinetic data on CCG205432 suggests that more frequent or extended dosing may be needed to achieve optimal drug concentrations and adequate CNS penetration (20). We are currently optimizing pharmacokinetic and pharmacodynamic parameters as we produce fourth-generation indole-2-carboxamide compounds.