Identification of a Broad-Spectrum Arenavirus Entry Inhibitor

Abstract

Several arenaviruses, including Lassa virus (LASV), are causative agents of hemorrhagic fever, for which effective therapeutic options are lacking. The LASV envelope glycoprotein (GP) gene was used to generate lentiviral pseudotypes to identify small-molecule inhibitors of viral entry. A benzimidazole derivative with potent antiviral activity was identified from a high-throughput screen utilizing this strategy. Subsequent lead optimization for antiviral activity identified a modified structure, ST-193, with a 50% inhibitory concentration (IC₅₀) of 1.6 nM against LASV pseudotypes. ST-193 inhibited pseudotypes generated with other arenavirus envelopes as well, including the remaining four commonly associated with hemorrhagic fever (IC₅₀s for Junín, Machupo, Guanarito, and Sabiá were in the 0.2 to 12 nM range) but exhibited no antiviral activity against pseudotypes incorporating either the GP from the LASV-related arenavirus lymphocytic choriomeningitis virus (LCMV) or the unrelated G protein from vesicular stomatitis virus, at concentrations of up to 10 μM. Determinants of ST-193 sensitivity were mapped through a combination of LASV-LCMV domain-swapping experiments, genetic selection of viral variants, and site-directed mutagenesis. Taken together, these studies demonstrate that sensitivity to ST-193 is dictated by a segment of about 30 amino acids within the GP2 subunit. This region includes the carboxy-terminal region of the ectodomain and the predicted transmembrane domain of the envelope protein, revealing a novel antiviral target within the arenavirus envelope GP.

Arenaviruses are a diverse family of small, enveloped, single-stranded RNA viruses which are generally propagated through asymptomatic, chronic infection of specific rodent hosts. They are phylogenetically grouped into Old World and New World lineages (9). Several arenaviruses are significant human pathogens, including five distinct hemorrhagic fever viruses designated category A by the CDC and NIAID, which is indicative of the level of highest threat to civilian populations (40). The most prevalent of these is Lassa virus (LASV), an Old World arenavirus endemic in West Africa, with several hundred thousand cases estimated annually (38). The mortality rate among hospitalized Lassa fever patients is 15 to 20% (36), but it has been reported to be higher than 50% for some outbreaks (23). Treatment or prevention options for arenavirus infections are limited. Intravenous ribavirin has shown efficacy against Lassa fever in high-risk patients (37), although its use can be associated with dose-limiting anemia (47). Ribavirin has also been used to treat isolated infections with the New World arenaviruses Junín virus (JUNV) (19), Machupo virus (MACV) (30), and Sabiá virus (6), although extensive clinical data are lacking. A vaccine candidate for Argentine hemorrhagic fever (JUNV) has demonstrated good efficacy among agricultural workers in South America (34), and several Lassa fever vaccine candidates have demonstrated efficacy in nonhuman primate models (21, 22, 27). The paucity of effective treatment options in the clinic, however, prompted the Working Group on Civilian Biodefense to recommend the pursuit of new antiviral therapies for these pathogens (8).

Anti-infective drug discovery for LASV presents significant logistical and safety challenges due to the requirement for maximum laboratory containment (biosafety level 4 [BSL-4]). Therefore, a surrogate assay, in which the LASV envelope glycoprotein (GP) was incorporated into lentiviral pseudotypes, was used as a high-throughput screening platform. Arenavirus entry is mediated by this single virally encoded protein, categorized as a class I viral fusion protein (20, 26, 53), facilitating the effective use of pseudotypes for antiviral screening. Inhibitors of LASV GP-mediated viral entry could thus be identified from a library of small-molecule compounds. As an essential component of the viral life cycle, the entry process is an attractive target for the development of antiviral pharmaceuticals. For example, two distinct classes of viral entry inhibitor, enfuvirtide (35) and maraviroc (15), have recently been approved for human immunodeficiency virus (HIV) treatment.

A benzimidazole derivative identified through high-throughput screening and subsequent lead optimization, ST-193, was found to be a potent LASV inhibitor in vitro and demonstrated protection superior to that of ribavirin against a lethal LASV challenge in a small-animal model (K. Cashman and M. Guttieri, unpublished data). Similarly potent in vitro activity was shown against viral entry mediated by other arenavirus envelopes, including the category A pathogens MACV, JUNV, Sabiá virus, and Guanarito virus. However, lymphocytic choriomeningitis virus (LCMV), an Old World arenavirus, was much less sensitive to ST-193 and thus served as a useful tool to map antiviral sensitivity determinants. Sensitivity to the benzimidazole derivative overlaps partially, but not completely, with that to ST-294, a previously described, chemically distinct inhibitor of New World arenaviruses such as JUNV (7). The convergence of sensitivity to diverse small-molecule inhibitors thus identifies a robust new target for arenavirus antiviral discovery within the viral entry phase.

MATERIALS AND METHODS

Virus, cells, and compounds.

Viruses and Vero cells were described previously (7). 293T/17 cells (ATCC CRL-11268) were maintained in Dulbecco's modified Eagle medium with 10% heat-inactivated fetal bovine serum (FBS) (Invitrogen) at 37°C with 5% CO₂. Initial compound lots were purchased from commercial suppliers (ST-37 from Asinex and ST-193 from InterBioScreen), while subsequent batches have been custom synthesized by a number of different vendors. Compound stock solutions were made at 10 mM in dimethyl sulfoxide (DMSO).

Viral GP cloning.

Viral RNA isolated with a QIAamp viral RNA kit (Qiagen) served as a template for cDNA synthesis using SuperScript one-step reverse transcription-PCR (Invitrogen) and GP-specific primers. GP inserts, except for JUNV GP (see below), were subsequently subcloned into the mammalian expression vector pCAGGS (41). Sabiá virus RNA was a kind gift from Ricardo Carrion, Jr. JUNV GP cDNA, a kind gift from Jack Nunberg, was subcloned into pCI (Promega) via PCR with engineered, flanking KpnI-BamHI restriction sites. GenBank accession numbers (and citations where appropriate) are provided in the Fig. 5 legend, with the following deviations relative to the deposited sequences (all numbering is relative to the relevant GP open reading frame): MACV GP had one synonymous nucleotide substitution (T₅₅₂C); JUNV GP had one synonymous nucleotide substitution (C₁₄₄₆T); Pichinde virus (PICV) GP had one coding substitution (G₃₉₅A nucleotide change, coding for an S₁₃₂N amino acid change); and Tacaribe virus (TCRV) GP had three nucleotide substitutions, one synonymous (C₂₉₇T) and two (GA to AG at nucleotides [nt] 1336 and -7) coding for an E₄₄₆R amino acid change. The LASV-LCMV chimeras were created by overlapping PCR, fusing the 5′ 1,245 (LASV) or 1,263 (LCMV) nt with the 3′ 231 (LASV) or 234 (LCMV) nt of the GP open reading frame (numbering includes a termination codon); the protein fusion junction is thus C terminal of a common TEML (single-amino-acid code) sequence. Arenavirus GP point mutations were introduced with the QuikChange site-directed mutagenesis kit (Stratagene). LASV V₄₃₁M (LASV #1 in Fig. 6) was generated by replacement of 2 nt (G₁₂₉₁A T₁₂₉₃G) within codon 431, LASV V₄₃₅M (LASV #2 in Fig. 6) contained a single-nucleotide substitution (G₁₃₀₃A), and LASV dbl contained both amino acid changes. LCMV M₄₃₇V (LCMV #1 in Fig. 6) was generated by a A₁₃₀₉G nucleotide substitution, and LCMV dbl (M₄₃₇V M₄₄₁V) also contained an A₁₃₂₁G nucleotide change. PICV T₄₄₅V (PICV #1 in Fig. 6) was made by substituting GT for AC at nt 1333 and -4.

FIG. 5.

Amino acid alignment of a portion of the arenavirus GP2 subunit. Amino acid numbering is as in Fig. 4. Residues identified as ST-193 sensitivity determinants (Fig. 4) are in bold, and the predicted TMD is indicated. The highlighted residues at positions 421 and 425 are sensitivity determinants for which some sequence divergence is observed across the Arenaviridae family. Abbreviations (GenBank accession numbers) are as follows: LASV, LASV strain Josiah (J04324) (5); LCMV, LCMV strain Armstrong 53b (AY847350) (29); MACV, MACV strain Carvallo (AY619643); JUNV, JUNV strain MC2 (D10072); TCRV, TCRV strain TRVL 11598 (P31840) (3); GTOV, Guanarito virus (AY129247) (13); SABV, Sabiá virus (U41071) (28); Chapare, Chapare virus (EU260463) (14); LATV, Latino virus (AF485259) (4); PICV (U77602); PIRV, Pirital virus (AF485262) (4); and WWAV, Whitewater Arroyo virus (AF228063) (12). LASV and LCMV are Old World arenaviruses, while New World arenaviruses are represented by PICV and PIRV (clade A); MACV, JUNV, TCRV, GTOV, SABV, and Chapare virus (clade B); and LATV (clade C). WWAV is likely a recombinant of clades A and B.

FIG. 6.

Site-directed mutagenesis of arenavirus GPs identifies important ST-193 sensitivity determinants. Pseudotype infectivity was measured as in Fig. 1 (representative experiments are shown in panels B to D). (A) List of GP constructs and their sensitivity to ST-193 (*, numbering based on alignment with the corresponding TCRV GP amino acid). Data shown are averages from at least four experiments. (B) LASV GP is much less sensitive when either site 1 or 2 is converted to the corresponding LCMV amino acid, and all arenavirus-specific ST-193 activity is lost when both sites are changed. (C) LCMV GP does not gain ST-193 sensitivity when the reciprocal substitutions are introduced (sites 1 and 2 converted to LASV amino acids). (D) PICV GP becomes significantly more sensitive to ST-193 when valine is substituted at site 1 (to match LASV GP). WT, wild type.

Generation of ST-193-resistant TCRV variants.

Vero cells were initially infected with TCRV at a multiplicity of infection of 0.1 in minimal essential media with 2% FBS in the presence of 1.2 μM ST-193. Virus was harvested when cytopathic effect became apparent (∼7 to 10 days) and was passaged again at a higher ST-193 concentration (1.8 μM); this process was repeated at 2.4 and 3 μM ST-193. RNA was isolated from the last virus harvest and cDNA prepared as described above, followed by cloning into TOPO TA vector pCR2.1 (Invitrogen). Multiple clones were then sequenced and subcloned into pCAGGS using EcoRI-XhoI restriction sites.

Pseudotyped virus production.

Pseudovirions were generated with a three-plasmid, HIV-based expression system (39). 293T/17 cells were transfected (CalPhos mammalian transfection kit; BD Biosciences) with a 1:1:1.7 ratio (1:1:0.6 for the G protein from vesicular stomatitis virus [VSVg]) of pΔR8.2, pHR′-Luc, and GP expression construct (10, 10, and 17 μg per 10-cm plate) and induced with 10 mM sodium butyrate for 6 h at 20 to 26 h posttransfection (applied only to non-VSVg envelopes). Supernatants were harvested at 48 h posttransfection, clarified by low-speed centrifugation (100 × g for 4 min.), filtered with a 0.45 μm syringe filter, and stored in aliquots at −80°C.

Infections and inhibition assays.

293T/17 cells seeded in poly-d-lysine-coated 96-well plates were infected with pseudovirions in Dulbecco's modified Eagle medium with 7.5% FBS and 0.5% DMSO and assayed for firefly luciferase activity (luciferase assay system from Promega) at 72 h postinfection. Luminescense was quantitated on a Wallac EnVision 2102 multilabel plate reader (Perkin-Elmer) using a 1-s read time per well. The luciferase signal was directly proportional to the inoculum size over several orders of magnitude. To test antiviral activity, serial compound dilutions in DMSO were added in triplicate to cells immediately prior to virus addition, maintaining a final DMSO concentration of 0.5% in all wells; the compound remained on cells throughout the infection. Each 96-well plate had a minimum of four replicates of a negative control (no virus) and eight replicates of a positive control (virus without compound). Luciferase activity was converted to % of the positive control, and 50% inhibitory concentrations (IC₅₀s) were calculated using XLfit (IDBS) for Microsoft Excel with a one-site dose-response curve fit.

RESULTS

Identification and activity of ST-193.

A chemically diverse, random library of about 400,000 small molecules was screened with lentivirus-based pseudotypes incorporating the LASV GP. Hit compounds (>75% inhibition at a test concentration of 5 μM) were filtered through a battery of follow-up tests, including specificity assays (inhibition of pseudotypes with unrelated GPs and inhibition of a panel of RNA and DNA viruses), cytotoxicity assays (42), confirmation assays against authentic LASV, and validation of antiviral activity with resynthesized compound. The initial hit rate from the primary screen was about 1.2%, although 90 to 95% of these initial hits were found not to be specific for LASV GP and/or cytotoxic. ST-37 (359.4 Da), a benzimidazole derivative exhibiting an average IC₅₀ against LASV GP pseudotypes of 16 nM (Fig. 1A), was identified from this process. Although the panel of specific hits included some clusters of chemically related structures, ST-37 was unique. In order to characterize the relationship between chemical structure and biological activity (structure-activity relationship [SAR]), analogs of ST-37 were assayed for antiviral activity. The initial SAR study was designed to explore the maximum chemical space with compounds obtained from commercial sources, to test the effect of electronic properties, and to examine the effect of the sizes of substituents of the benzene rings. In an iterative process, additional compounds were obtained based on the nascent SAR, details of which will be described elsewhere. One of these analogs, ST-193 {1-(4-methoxyphenyl)-N-[(4-propan-2-ylphenyl)methyl]benzimidazol-5-amine} (371.5 Da), has a substitution of an isopropyl for a methoxy group (Fig. 1A), resulting in significantly greater antiviral potency (Fig. 1B). A nonarenavirus envelope, VSVg, was used as a specificity control. As a practical consideration, compounds were tested at concentrations of no higher than 50 μM, a level at which ST-193 exhibits visual evidence of insolubility (crystals observed by light microscopy); this is also near the 50% cytotoxicity concentration (48 μM for ST-193 on 293T/17 cells), and thus the VSVg inhibition seen at the highest concentration is considered to be nonspecific activity (IC₅₀ of around 30 μM [Fig. 1B]).

FIG. 1.

Structure and antiviral activity of ST-37 and ST-193. (A) Benzimidazole derivative ST-37 (left) was modified to generate ST-193 (right). (B) Inhibition of LASV GP- or VSVg-pseudotyped HIV infection with ST-37 or ST-193. Infectivity was measured with a luciferase reporter relative to controls with no compound. Each point is an average of three replicates, with error bars designating standard deviations.

The activity and specificity of ST-193 have been assessed by a variety of assays, including virus yield reduction, cytopathic effect, plaque reduction, and pseudotype infectivity. The pseudotype platform was the most reproducible and amenable to direct comparison between diverse viruses, particularly given that several viruses of interest are restricted to use under BSL-4 containment. As shown in Table 1, ST-193 potently inhibits envelopes derived from clade B New World arenaviruses, the phylogenetic cluster containing all four South American hemorrhagic fever viruses currently designated category A (JUNV, MACV, Guanarito virus, and Sabiá virus). ST-193 also potently inhibits viral entry mediated by the GP from the prototypic New World arenavirus, TCRV. TCRV, another clade B member, is not known to be a significant human pathogen. Surprisingly, ST-193 exhibits only a nonspecific level of activity against the LASV-related LCMV GP, with an IC₅₀ some 4 orders of magnitude greater than that against LASV (Table 1). Antiviral activity against authentic TCRV is shown in Fig. 2. Inhibition of live LASV, JUNV, MACV, and Guanarito virus has been confirmed by plaque reduction assays under BSL-4 conditions in collaboration with a research team at USAMRIID (Mary Guttieri, Amy Shurtleff, Philip Ferro, Kathleen Cashman, and Anna Honko). Generally, effective IC₅₀s appear to be higher against live virus (0.02 to 0.4 μM) (Fig. 2 and unpublished data) than against pseudotyped viruses. Although the reason for the different values is unclear, they do not appear to be due solely to differences in assay platform. For example, ST-336, a previously described inhibitor of New World arenavirus entry (7), generates similar IC₅₀s against both pseudovirions and live virus (∼0.1 to 0.3 μM). Possible causes for the divergence in values include GP density, differential GP modification (e.g., producer cell type-specifc glycosylation patterns), target cell differences (human cells are used for pseudovirions, while arenavirus plaque assays are generally performed on Vero cells from the African green monkey), and involvement of other arenavirus proteins (e.g., the Z protein).

TABLE 1.

ST-193 antiviral activity against lentivirus-based pseudotypes incorporating heterologous envelope GPs

GP	Phylogeny of GP source	Mean IC50 (μM) ± SEM
LASV	Old World arenavirus, Josiah strain	0.0016 ± 0.0003
LCMV	Old World arenavirus, Armstrong 53b	31 ± 4
PICV	New World arenavirus, clade A	2.6 ± 0.5
JUNV	New World arenavirus, clade B1	0.0002 ± 0.00003
MACV	New World arenavirus, clade B1	0.0023 ± 0.0013
TCRV	New World arenavirus, clade B1	0.004 ± 0.002
Guanarito virus	New World arenavirus, clade B2	0.00034 ± 0.00007
Sabiá virus	New World arenavirus, clade B3	0.012 ± 0.003
VSV	Rhabdovirus	29 ± 2.5

FIG. 2.

ST-193 inhibition of TCRV infection. TCRV was diluted and used to infect Vero cell monolayers in 35-mm wells in the presence of 0.1% DMSO with or without ST-193. Bars represent averages of six replicates (error bars designate standard errors of the means).

Mapping of determinants of ST-193 sensitivity.

To gain insight into the mechanism of ST-193 inhibition and potential binding sites, three strategies were employed to identify determinants of ST-193 sensitivity within the arenavirus envelope GP. First, domain-swapping experiments were performed using two closely related arenavirus GPs, one sensitive (LASV) and one insensitive (LCMV) to ST-193. Second, a nonpathogenic surrogate virus (TCRV) was passaged in the presence of ST-193 to select for less-sensitive variants. Finally, site-directed mutagenesis, informed by the genetic pattern of arenavirus sensitivity, identified two significant amino acids located within the predicted transmembrane domain (TMD) of the GP2 subunit.

The C-terminal portion of the LASV GP2 subunit confers ST-193 sensitivity.

Lentiviral pseudotypes incorporating chimeric arenavirus envelopes were constructed to identify the region dictating ST-193 sensitivity. Constructs that exchanged the C-terminal one-third of the GP2 subunit (76 amino acids [aa] of LASV and 77 aa of LCMV) were found to retain viral entry function. These chimeras partition the entire endodomain and predicted TMD from the signal peptide, the GP1 subunit, and most of the GP2 subunit ectodomain (Fig. 3A), including the N-terminal (HR1) and most of the C-terminal (HR2) heptad repeats characteristic of class I viral fusion proteins (20, 26, 53). ST-193 sensitivity was found to be conferred by the C-terminal portion of GP2 (Fig. 3B).

FIG. 3.

Domain swapping demonstrates that the C-terminal third of GP2 determines sensitivity to ST-193. (A) Schematic representation of LASV-LCMV chimeric GPs. Predicted TMD and heptad repeat domains (HR1 and HR2) are indicated. The two GPs are spliced 12 aa N terminal of the predicted TMD. (B) Pseudotype infectivity as in Fig. 1, using the arenavirus GPs shown (from a representative experiment). Average IC₅₀s ± standard errors of the means for these constructs (across at least four experiments) were as follows: LASV, 0.0016 ± 0.0003 μM; LCMV, 31 ± 4 μM; LASV_N-LCMV_C, 13.6 ± 2.2 μM; and LCMV_N-LASV_C, 0.0005 ± 0.0003 μM. WT, wild type.

Generation and characterization of ST-193-resistant TCRV variants.

In order to genetically map sensitivity determinants, TCRV was serially passaged in the presence of escalating concentrations of ST-193 to select for variants with decreased sensitivity. A selected population (193R) was thus generated that yielded equivalent viral titers when cultured in the presence or absence of 3 μM ST-193; in contrast, the unselected TCRV titer is reduced ∼1,000-fold under corresponding conditions (data not shown). RNA from the selected population was used to synthesize cDNA, and multiple GP sequences were determined from clones derived from this cDNA. Coding changes found within any GP sequence were subsequently introduced into a TCRV GP expression plasmid for viral pseudotype production in order to directly evaluate the contribution of a given mutation to ST-193 sensitivity. Although this approach gives little consideration to viral fitness, it allows for simple and rapid identification of multiple sensitivity determinants.

Individual 193R mutations that reduce ST-193 sensitivity were found in or immediately N terminal of the predicted TMD of the GP2 subunit (Fig. 4). The 193R variations resulted in a range of resistance, as single-amino-acid changes increased the ST-193 IC₅₀ by 30-fold to more than 1,000-fold. At position 413, located within the predicted ectodomain near the TMD, two distinct mutations (Q₄₁₃H and Q₄₁₃R) resulted in greatly reduced ST-193 sensitivity.

FIG. 4.

ST-193 sensitivity determinants. (A) Schematic representation of an arenavirus GP2 subunit and the sites of 193R variations. The TCRV amino acid sequence is shown for the region indicated, with sites identified as determinants of ST-193 sensitivity indicated with raised letters and numbering (TCRV GP amino acid numbering). Predicted TMD and heptad repeat domains (HR1 and HR2) are indicated (the TMD is underlined in the sequence). (B) Single-amino-acid variations that reduce ST-193 sensitivity in TCRV GP. Resistance is the fold change in IC₅₀ as measured with TCRV GP-pseudotyped HIV, relative to wild type, and averaged across at least four independent experiments. “Allele” refers to the common name for the mutant GP.

Interestingly, resistance to a previously described inhibitor of New World arenavirus entry, ST-294, has been mapped to a similar location within the TCRV GP2 subunit (7). Three of the four recognized ST-294-resistant TCRV variants (DR1 to -4) also exhibited reduced sensitivity to ST-193 (Fig. 4), while the fourth (DR3, or S₄₃₃I) retained full sensitivity. The ST-294 DR4 mutation (F₄₃₆I) was also identified from within the 193R population. Conversely, the other 193R variants retain ST-294 sensitivity (data not shown). A mutant identified from a third distinct TCRV selection, R₄₁₂T, was also found to display ST-193 resistance (Fig. 4). This third screen was designed to identify resistance to ST-761, another New World arenavirus inhibitor that targets the GP protein (unpublished data). As with the ST-294 comparison, 193R variants retain sensitivity to ST-761, with the exception of the F₄₃₆I variant. ST-193, ST-294, and ST-761 are chemically diverse small molecules with no obvious common characteristic.

Two TMD residues regulate ST-193 sensitivity.

An alignment of arenavirus GP protein sequences reveals conservation of the majority of ST-193 sensitivity determinants identified above (Fig. 5). Positions 421 and 425 show some diversity, however, which could in part explain the relative insensitivity of LCMV and PICV GPs to ST-193. Position 421 in particular is striking in that arenavirus GPs that are sensitive to ST-193 in the nanomolar range contain valine, while those that are not contain other residues (methionine or threonine). To directly test the functional significance of these residues with respect to ST-193 sensitivity, the LASV GP was engineered to contain the corresponding amino acid from ST-193-insensitive LCMV at one or both of these sites (Fig. 6A). Additionally, some of the reciprocal changes were made in the LCMV GP, and the Val₄₂₁ present in ST-193-sensitive GPs was introduced into the PICV GP. Viral pseudotypes were made to evaluate ST-193 sensitivity for each of these genotypes.

Introduction of a methionine into either position 421 or 425 dramatically reduces the sensitivity of LASV to ST-193, and replacement of both residues further reduces the sensitivity such that LASV dbl GP (V₄₂₁M V₄₂₅M) and LCMV GP display nearly equivalent ST-193 sensitivity (Fig. 6B). The reciprocal substitutions (LCMV dbl and LCMV #1), however, do not convert LCMV GP to greater sensitivity, indicating that other residues play a role in LCMV resistance to ST-193 inhibition (Fig. 6C). The significance of position 421 is further highlighted by the increased ST-193 sensitivity of PICV GP when valine is exchanged for the native threonine residue (Fig. 6D). The correlation between genotype and ST-193 sensitivity suggest that other New World arenaviruses linked to hemorrhagic fever, including Whitewater Arroyo virus in the United States (11) and the recently identified Chapare virus in Bolivia (14), may well be inhibited by this compound class as well (Fig. 5).

Mechanism of ST-193 activity.

The stage of viral entry at which ST-193 acts has not yet been elucidated. Binding studies have shown no effect of ST-193 on pseudovirion binding to target cells (data not shown), but these studies have been hampered by high nonspecific virus binding (not mediated by GP). ST-193 has been shown to inhibit pH-dependent cell-cell fusion mediated by LASV (IC₅₀ of <50 nM) or JUNV GP (IC₅₀ of 0.7 μM) and also inhibits pH-induced GP1 shedding (57), a marker of envelope activation. An inhibitor of entry of New World arenavirus, ST-336/-294, has previously been shown to tightly bind or inactivate native virus particles (7). This was based in part on the ability of ST-336 to inhibit virus infection in a compound-virus mix even after dilution to normally ineffective concentrations (7) (Fig. 7B). ST-193, however, does not exhibit this effect (Fig. 7A), indicating either weaker binding or a distinct mechanism of action.

FIG. 7.

Dilution of antiviral activity. LASV GP (A) or TCRV GP (B) pseudotype infectivity was measured after addition of ST-193 to 80 nM (A) or of ST-294 to 2 μM (B) and following two serial 20-fold (A) or 10-fold (B) dilutions (shown as open squares, with error bars designating standard deviations). The line in each graph indicates the standard inhibition curve assembled from 10 independent experiments (dots are averages of three replicates from one experiment) using the XLfit dose-response one-site model. ST-193 activity is diluted as predicted, while ST-294 inhibitory activity is retained.

DISCUSSION

The benzimidazole derivative described here, ST-193, has been demonstrated to be a potent inhibitor of arenavirus entry. Several lines of circumstantial evidence suggest that the primary mechanism of ST-193 action is likely not direct blocking of virus attachment. First, domain-swapping experiments demonstrate that the difference in sensitivity between LASV and LCMV resides not within the receptor-binding GP1 domain, but in the GP2 domain. Second, no sensitivity determinants were identified within GP1 following selection for ST-193-resistant arenavirus. Finally, arenavirus entry is mediated by a diversity of host cell surface receptors that do not appear to correspond to ST-193 sensitivity. LASV and LCMV utilize α-dystroglycan (10), while the category A hemorrhagic New World arenaviruses use transferrin receptor 1 (44). Clade C New World arenaviruses, like the Old World arenaviruses, use α-dystroglycan (51). Some other New World arenaviruses, including ST-193-sensitive TCRV, likely use a distinct, yet-unidentified receptor (24, 43, 45, 46). These observations, combined with the presence of multiple ST-193 sensitivity determinants within the fusogenic GP2 subunit, suggest that ST-193 blocks GP-mediated entry at a postattachment stage.

ST-193 sensitivity is modulated by a 28-aa span of the GP2 subunit of the envelope protein, encompassing nearly the entire predicted TMD. Although these results implicate this region in ST-193 activity, the location and spacing of the identified sensitivity determinants suggest that at least some of these sites may not be directly involved in inhibitor binding. Confirmation of the ST-193 binding site awaits direct biochemical data. Notably, the two residues implicated in the phylogeny of ST-193 sensitivity, positions 421 and 425, are located 4 aa apart. Within an α-helical TMD, these two residues would be predicted to be adjacent to one another; position 418, another important sensitivity determinant (Fig. 4B), would be predicted to lie on the same helical face as well. The predicted physical property of ST-193 (average calculated partition coefficient of 5.5) is consistent with binding a hydrophobic pocket. ST-193 might prevent or perturb conformational rearrangement of GP or might inhibit lipid mixing, thereby preventing fusion with the host cell. Alternatively, ST-193 could interact with the membrane-proximal domain, in the 409 to 413 region identified following genetic selection; although this region is more hydrophilic, the membrane-proximal ectodomain of HIV gp41 (the class I viral fusion protein equivalent of arenavirus GP2) is a target of broadly neutralizing monoclonal antibodies (52). Another possibility is that ST-193 might disrupt protein-protein interactions within or near the TMD. Interactions between the TMDs of alphavirus E1 and E2 proteins are important for fusion (50). Several recent studies have investigated the unusual role of the arenavirus GP signal peptide (1, 2, 17, 18, 25, 32, 48, 49, 54-56), which remains associated with GP and is required for efficient GP1-GP2 processing, transport to the plasma membrane, membrane fusion, and virus infectivity. This cleaved, 58-aa peptide has two conserved hydrophobic domains and a hydrophilic amino terminus. The interaction between the stable signal peptide and the processed GP1-GP2 complex, then, might provide a vulnerable antiviral target specific to arenaviruses. Sensitivity to at least three dissimilar antiviral compounds (ST-193, ST-294, and ST-761) has been mapped to the TMD region of GP2, with overlapping yet quite distinct patterns of genetic resistance. By way of analogy, several different classes of small-molecule HIV entry inhibitor, including maraviroc, are thought to bind within a pocket created by four TMDs of CCR5, an important HIV coreceptor (31). Similarly, structurally distinct small-molecule fusion inhibitors of respiratory syncytial virus have demonstrated cross-resistance patterns (16). Recently, several novel inhibitors of arenavirus entry with activity against both Old and New World arenaviruses have been described (33). Although none of these small molecules is structurally related to ST-193, it will be of interest to examine cross-resistance patterns.

Arenaviruses constitute an important class of hemorrhagic fever pathogens, with five viruses designated category A. The potency and spectrum of ST-193 activity make it a strong candidate antiviral for the prevention and treatment of disease caused by these viruses. Preclinical activities have been initiated to support that development. ST-193 pharmacokinetics, oral bioavailability, toxicity, and in vivo efficacy have been evaluated in multiple animal models, with promising preliminary results (to be shown elsewhere), thus validating the drug discovery approach of identifying viral entry inhibitors through a surrogate (pseudotype) platform.