Identification and Characterization of Mefloquine Efficacy against JC Virus In Vitro^{▿ †}

Source and identity of potential therapeutic agents.

The Spectrum collection (MicroSource Discovery Inc., Groton, CT) consists of ∼1,000 bioactive compounds and natural products plus ∼1,000 Food and Drug Administration-approved drugs that are defined according to the name designations set forth in the USP Dictionary of USAN and International Drug Names (57). An alphabetical listing of the compounds is available at http://www.msdiscovery.com/spectrum.html. The compounds are supplied as 10 mM solutions in dimethyl sulfoxide (DMSO). In our follow-up experiments, mefloquine and all other drugs were purchased from Sigma-Aldrich (St. Louis, MO), and 100 mM stock solutions were prepared in DMSO. Two mefloquine enantiomers were separated from a commercial mefloquine sample by Chiral Technologies (West Chester, PA) by chiral high-pressure liquid chromatography on a Chiralpak IA column to a purity of >98.6% for each enantiomer. Mefloquine analogs were purchased from BioBlocks, Inc. (San Diego, CA.). Cidofovir (Vistide) was obtained from Gilead as a 0.238 M aqueous stock solution.

Propagation and identity of cultured cells.

SVG-A cells (a gift from Walter Atwood, Brown University), established by transforming human fetal glial cells with an origin-defective simian virus 40 (SV40) mutant (38), were cultured in 1× Eagle minimal essential medium supplemented with 10% heat-inactivated fetal bovine serum (FBS) and 4 mM l-glutamine (Mediatech, Inc.). Viral infection was performed in the same medium supplemented with 2% heat-inactivated FBS.

Human astrocytes were isolated from a fetal cerebral cortex (ScienCell Research Laboratories, Carlsbad, CA) and cultured in proprietary astrocyte growth medium (ScienCell Research Laboratories). Viral infection of the astrocytes was performed in this medium.

Cultures of primary human fetal glial (PHFG) cells were prepared from fetal brain tissue by using a modification of an earlier protocol (44). Briefly, tissue was received in glial cell medium (Dulbecco modified Eagle's minimal essential medium [DMEM] supplemented with 3% FBS, 7% bovine calf serum, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μg/ml gentamicin, and 2.5 μg/ml amphotericin B [Fungizone]). After removal of the meninges and blood vessels, the tissue was washed twice in 10 to 30 ml glial cell medium without gentamicin containing 200 μg/ml amikacin (Sigma) and was then pelleted in a clinical centrifuge (3 min, 1,120 rpm). The tissue was transferred to a sterile petri dish containing 10 to 20 ml medium, and tissue fragments were reduced in size by expressing them through a 10-ml syringe (without a needle) and a 40-mesh screen in a tissue dissociation device (Sigma). Cells derived by this procedure were seeded on 100-ml tissue culture plates in glial cell medium by using a volume of 0.5 ml of cells per plate. After 1 week, if the astrocyte layer was confluent, the cultures were placed in maintenance medium (DMEM supplemented with 3% FBS, penicillin, and streptomycin). If the astrocyte layer was not confluent, the cultures were placed in glial cell medium without amikacin. At 2 to 3 weeks after they were seeded, the cells were prepared for freezing. Briefly, cell cultures were incubated overnight in DMEM containing 10% FBS. On the next day, the heterogeneous population of cells containing astrocytes and small round cells clustered on top of the astrocyte layer was washed with saline A (trypsin buffer, 0.8% NaCl, 0.4% KCl, 0.1% glucose, 0.035% NaHCO₃, 0.2% EDTA, pH 7.0). After aspiration, saline A was added again to each plate, and the cells were incubated for 3 min at 37°C. Trypsin (in saline A) was added, and cells were incubated for 4 min at 37°C. DMEM containing 10% FBS was added, and the cells were gently collected in 50-ml conical tubes. The cell suspension was centrifuged, the pellet was washed once with DMEM containing 10% FBS, and the cells were then pelleted again. The cells were suspended in freezing medium (DMEM, 10% FBS, 10% DMSO) and were stored in liquid nitrogen until they were used for infectivity assays with JCV(Mad1).

JCV isolates.

The hybrid virus JCV(M1/SVEΔ) (a gift from Walter Atwood) was constructed by first inserting SV40 regulatory sequences into the JCV(Mad1) transcriptional control region to create the hybrid genome, JCV(M1/SVE) (58). Transfection of SVG-A cells with this DNA yielded JCV(M1/SVEΔ), which comprised JCV-SV40 enhancer signals linked to the JCV(Mad1) genome. To produce purified preparations of virus, SVG-A cells were plated at 50% confluence and were infected with a 1:50 dilution of JCV(M1/SVEΔ) for 1 h at 37°C (25, 36). The cells were cultured for 3 weeks and were then scraped from the flasks, pooled (along with cells that had detached during prior medium changes), and pelleted. The cell pellet was resuspended in 20 ml of supernatant and disrupted in a microfluidizer (Microfluidics, Inc.). Deoxycholate was added to the cell lysates at a final concentration of 0.25%, and the mixture was incubated at 37°C for 30 min. The virus-containing supernatant was centrifuged at 10,000 rpm for 30 min in an SA600 rotor and aliquoted and stored at −80°C. The JCV(Mad4) isolate (46) was obtained from the American Type Tissue Collection (Manassas, VA), and the prototype JCV(Mad1) isolate (24, 45) was a gift from Duard Walker (Madison, WI).

Detection of antibodies.

PAb597 (a gift from Walter Atwood [19]), a mouse monoclonal antibody directed against SV40 major capsid protein VP1, cross-reacts with JCV VP1 (4) and was used with an Alexa-Fluor 488-labeled goat anti-mouse secondary antibody (Invitrogen, Carlsbad, CA) to visualize JCV-infected cells by indirect immunofluorescence. Cell nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) (Invitrogen). Neutralizing anti-JCV rabbit antiserum was a gift from Walter Atwood (3).

JCV infectivity assay.

SVG-A cells were seeded at 2,000 cells/well/0.075 ml of culture medium in flat-bottom 96-well plates (Corning). The compounds to be tested for antiviral activity were prepared the next day in assay medium (1× Eagle minimal essential medium with 2% heat-inactivated FBS and 4 mM l-glutamine). A master viral plate was prepared by mixing equal volumes of 2× compound and virus diluted two times to yield the final working concentrations of compound and virus. The plate containing cells was gently inverted and shaken to remove the medium. From the master plate, 0.035 ml of compound plus virus was added to designated wells. The cells were incubated with the compound-virus mixture for 60 min in a humidified, 37°C incubator containing 5% CO₂. Medium containing the final concentration of the drug was added to designated wells to bring the final volume to 0.1 ml/well. After incubation of the plates for 3 days, the cells were washed once with 1× phosphate-buffered saline (PBS) and fixed in 2% paraformaldehyde-1× PBS for 30 min at room temperature. The fixative was removed, and the cells were permeabilized with 0.5% Triton X-100 in PBS for 30 min. Cells infected with JCV(M1/SVEΔ) were visualized by staining with 0.05 ml of PAb597 (2 μg/ml in 1× PBS) for 60 min at 37°C. Following a wash step with 1× PBS, an Alexa-Fluor 488-labeled goat anti-mouse secondary antibody (1:100 dilution in 1× PBS) and DAPI (1 μg/ml, 0.05 ml/well) were added to the cells for 30 min at 37°C. The cells were washed with 1× PBS, and field images of each well were acquired and analyzed with a Cellomics ArrayScan imager (Thermo Scientific Inc.) and Target Activation software.

The infectivity assay was also performed with human astrocytes or PHFG cells but with the following modifications. Human astrocytes were seeded at 4,000 cells/well/0.075 ml of culture medium in flat-bottom 96-well plates. The culture medium was used as the virus diluent, and the duration of the infection was 6 to 10 days. PHFG cells (∼1.2 × 10⁵) were seeded onto 35-mm plates containing 2.0 ml DMEM supplemented with 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine. The cells were exposed 48 h later to 5 hemagglutinating units of JCV(Mad1) diluted in DMEM; 0.5% heat-inactivated FBS; and 2.5 μM, 5 μM, 7.5 μM, or 10 μM of mefloquine in 0.01% DMSO. Three independent sets of experiments were performed, each of which was performed with two samples at each time point. Cells not exposed to mefloquine were infected with 1 or 5 hemagglutinating units of JCV(Mad1) and served as controls to determine mefloquine's inhibitory activity and to confirm that the infected cell numbers increased proportionally with a fivefold increase in the virus inoculum. Virus was allowed to adsorb to the cells for 3 h at 37°C, and then DMEM supplemented with 3% heat-inactivated FBS and the appropriate amount of mefloquine was added. The medium was replaced 5 days postinfection, and DNA was extracted from the cells at days 7 and 10 postinfection.

DNA extraction and sample preparation.

DNA was extracted from cells with a QIAamp 96 blood kit (Qiagen, Inc.) and were treated with RNase A, which is optional. The DNA was quantitated with a Quant-iT double-stranded DNA high-sensitivity assay, according to the manufacturer's recommendations (Molecular Probes Inc., Eugene, OR). The purified DNA was stored at −20°C until use.

Real-time quantitative PCR (qPCR) assay.

TaqMan forward and reverse primers and MGB probes were designed to recognize conserved JCV early coding sequences and were designed with Primer Express (version 3.0) software (Applied Biosystems, Foster City, CA). The sequence of the forward primer was AGGCAGCAAGCAATGAATCC, that of the reverse primer was ATGGCAATGCTGTTTTAGAGCAA, and that of the 6-carboxyfluorescein-labeled probe was CCACCCCAGCCATAT. To create a copy number standard curve for absolute quantification, we linearized plasmid pUC19 containing the JCV genome with SmaI. Quadruplicate PCRs were run in a 384-well optical plate (Applied Biosystems). Real-time reactions were performed in a 7900HT thermal cycler (Applied Biosystems) under the following conditions: 50°C for 2 min, 95°C for 10 min, and 40 cycles of 95°C for 15 s and 60°C for 60 s with 900 nM forward and reverse primers, 200 nM TaqMan probe, and 1× Universal master mix (Applied Biosystems). The JCV copy number was determined for each experimental sample by comparison to the JCV plasmid standard curve by using Sequence Detection software (Applied Biosystems) and normalization to the total amount of DNA extracted from a sample. Zero copies of JCV DNA were detected in a noninfected negative control. P values were calculated by a Student t test.

Determination of rates of viral inhibition.

We determined the percentage of cells infected following exposure to JCV by dividing the number of immunofluorescent cells by the total number of nucleated cells and then multiplying that value by 100, that is, percent JCV-positive cells = (total number of VP1-positive cells)/(total number of DAPI-positive cells)·100. The percent viral inhibition by a compound was calculated on the basis of the percentage of JCV-positive cells rather than the number of VP1-positive cells: percent JCV inhibition = [1 − (percent JCV-positive cells with a compound − percent JCV-positive cells in the negative control)/(percent JCV-positive cells in the positive control − percent JCV-positive cells in the negative control)]·100, where the positive control represents cells infected with virus in the absence of any compound and the negative control denotes cells not infected with virus. The number of JCV-positive cells in the negative control samples quantitated with the Cellomics ArrayScan imager was always <1% of the number of JCV-positive cells in the positive control. The percentage of JCV DNA replication inhibition was calculated as [1 − (JCV DNA copy number in cells treated with a compound/JCV DNA copy number in the positive control)]·100. Zero copies of the JCV genome were detected in negative control samples. For high-throughput screening of the compounds, we calculated the Z factor, which was equal to 1 − [3·(σ_p + σ_n)/|μ_p − μ_n|], where μ is the mean, σ is the standard deviation, p is the positive control, and n is the negative control (61). The intraplate intragroup coefficient of variance was always below 20%, and Z′ was always >0.5. EC₅₀s were calculated with Prism software (GraphPad Software Inc., La Jolla, CA). The EC₅₀ was calculated by fitting of a sigmoid dose-response curve, Y = bottom + {(top − bottom)/[1 + 10^(X − log EC₅₀)]}, where X is the logarithm of the inhibitor concentration, Y is the response (i.e., percent inhibition), top is the highest observed value, and bottom is the lowest observed value.

Molecular modeling and in silico screening.

The ROCS program (version 2.3.1; Openeye Scientific Software, Santa Fe, NM) was used to compare the three-dimensional shapes of diverse compound classes and to perform virtual screening against clinical compounds extracted from the MDL Drug Data Report database (Symyx Technologies, Inc.). The program OMEGA2 (version 2.3.0; Openeye Scientific Software) was used to generate multiple low-energy conformations of the compounds. The three-dimensional enumerated conformers were compared on the basis of the implicit Mills-Dean atom chemical force field scheme in conjunction with the shape-base matching algorithm of the ROCS program (version 2.3.1; Openeye Scientific Software). The overlays with the best scores were visualized in the PyMOL program (version 1.0.0b15; DeLano Scientific LLC, Palo Alto, CA).

Primary screen: JCV infectivity assay.

To identify drugs with anti-JCV activity, we screened a commercially available collection of approximately 2,000 approved drugs and bioactive compounds, called the Spectrum collection, for their anti-JCV activities in an in vitro viral infectivity assay (48). As a primary screen, we monitored inhibition of the viral infection rate in a human fetal astroglial cell line (SVG-A) infected with JCV(M1/SVEΔ), a modified form of JCV. SVG-A cells (38) were chosen for use in the primary screening assay, as they represent one of the few available cell lines permissive for JCV replication, and JCV(M1/SVEΔ) was chosen because its infection of SVG-A cells results in an accelerated rate of viral replication that permits the earlier detection of infected cells relative to that observed with other cell types and JCV isolates (3 versus 6 to 10 days postinfection). The JCV(M1/SVEΔ) genome includes the coding sequences of prototype strain JCV(Mad1) isolated from a patient with PML (24, 45) linked to hybrid JCV-SV40 noncoding regulatory sequences (58). This rearrangement of transcriptional signals extends the species and cell type host range of the virus.

To facilitate screening of the Spectrum collection, we adapted the JCV infectivity assay (48) to a 96-well format and employed a Cellomix ArrayScan high-content imager to measure JCV replication. In this system, infected cells can be detected by immunofluorescent staining with antibodies that recognize the JCV capsid protein, VP1. The total numbers of cells in the culture were visualized by staining with DAPI, which stains DNA (Fig. 1A). Use of the Cellomics ArrayScan imager allowed us to simultaneously identify and count each DAPI- and VP1-positive event in the assay well and routinely counted 300 to 800 VP1-positive and 8,000 to 16,000 DAPI-positive events per well, thus minimizing variability due to a nonuniform cell growth pattern and/or intrawell viral spread. By using this format, 4 to 7% of all cells were found to express JCV VP1 72 h postinfection, and the number of infected cells (i.e., VP1-positive cells) at the end of the culture period was proportional to the number of infectious viral particles used to infect the cell culture (Fig. 1B). In each experiment, the highest viral dilution that yielded maximum infectivity was used. Using neutralizing rabbit anti-JCV serum as a positive control for the inhibition of infectivity, we showed that the assay responds to viral inhibition in a predictive fashion (Fig. 1C).

We initially tested the compounds for their antiviral activities at a single dose (10 μM) and noticed that some of the compounds that inhibited the number of virus-infected cells (i.e., VP1-positive cells) to a great extent also dramatically reduced the total number of cells (i.e., DAPI-positive events), suggesting the occurrence of cytotoxic or cytostatic effects. To determine whether a particular compound had decreased the number of virus-infected cells because of its antiviral effect, we calculated the percent viral inhibition using the infection rate (i.e., the percentage of JCV-positive cells) rather than the total number of JCV-infected cells (i.e., the number of VP1-positive cells). With this system, we observed that treatment with cidofovir, a drug that has been tested clinically for its efficacy in patients with PML (21, 27, 39), inhibited the number of infected cells and the total number of cells in culture to the same degree, indicating that the percent inhibition of the infection rate (i.e., the percentage of JCV-positive cells) was not significant (Fig. 1D). Similar effects were noted for other drugs reported to have cytotoxic effects, e.g., mitomycin C and cytarabine (data not shown). Therefore, we determined the antiviral activity of each compound by calculating viral inhibition on the basis of the percentage of JCV-positive cells rather than the absolute number of JCV-positive cells per group.

Drug screening and selection.

A number of drugs and compounds that inhibited the JCV infection rate by ≥20% without causing significant cell toxicity (<20% total cell number inhibition) were identified (Fig. 2). We chose 20% as a cutoff for the first-pass screening because the coefficient of variance value of our assay was consistently <20%; 67 compounds fulfilled this criterion (see the table in the supplemental material). These compounds were subsequently tested in the same assay by using several different concentrations to further evaluate their therapeutic potential. On the basis of the dose-response data, 14 drugs proved to be effective and demonstrated ≥50% inhibition of virus-infected cells (EC₅₀, <20 μM) without inducing significant cell toxicity (<20% total cell number inhibition) (Table 1). The compounds that had reduced the total cell number by ≥80% were retested at lower concentrations, but none that demonstrated a clear anti-JCV effect without a concomitant cytotoxic or cytostatic effect were identified (Fig. 2). We chose to proceed with testing only with the drugs that did not reduce the total cell numbers to diminish the chance of confounding an antiviral effect with a cytotoxic or cytostatic effect.

JCV actively replicates and destroys oligodendrocytes in the CNS of PML patients. Therefore, it is crucial to the success of any potential PML therapy that the candidate drug be capable of achieving an efficacious concentration in the brain. Unfortunately, many drugs are incapable of penetrating the BBB. On the basis of a review of the literature (Table 1), only 1 of the 14 compounds with anti-JCV activity, mefloquine, has been shown to accumulate in the brains of treated patients at the concentration at which it is efficacious in vitro (EC₅₀, 3.9 ± 2.1 μM) (Fig. 3A). A postmortem brain analysis of people taking mefloquine prior to their deaths measured 35 to 50 nmol of drug per gram of brain tissue, or approximately 35 to 50 μM (33, 47). Since potentially efficacious doses of mefloquine could be achieved in the brains of patients receiving approved doses of the drug, subsequent experiments focused upon characterization of the anti-JCV activity of mefloquine.

Characterization of mefloquine activity in primary cell culture.

To further characterize the effect of mefloquine on the activity of JCV, experiments were performed to evaluate whether the JCV-inhibitory effect of mefloquine was dependent on the cell line used in the primary screen. The SVG-A cell line has been propagated in vitro for many generations and was transformed by an SV40 large T antigen that can enhance JCV replication. To evaluate whether mefloquine is capable of inhibiting viral replication in cells more closely resembling the JCV target cell in the human brain, experiments were performed with primary glial cells. In vitro infection of homogeneous cultures of primary oligodendrocytes, a primary JCV target in the brains of patients with PML, has not been established; therefore, we employed human fetal astrocytes to test the ability of mefloquine to inhibit viral infection in a primary cell culture. In these cells, mefloquine inhibits JCV infection with essentially the same efficacy as it inhibits viral infection in SVG-A cells (Fig. 3B). These data suggest that the anti-JCV effect of mefloquine occurs in a more relevant cellular background.

Characterization of mefloquine effects on natural JCV isolates.

Experiments were performed to demonstrate that the inhibitory effect of mefloquine is not limited to the JCV(M1/SVEΔ) construct used in the primary screening assay. JCV(M1/SVEΔ) was used because of its faster growth kinetics and ease of preparation relative to those of in vivo isolates of JCV; however, the transcription of its JCV genes is not regulated by an authentic JCV enhancer element. To ensure that mefloquine's antiviral activity is not limited to this modified virus, we tested mefloquine's ability to inhibit JCV(Mad4), an in vivo isolate from a PML patient (46). On the basis of the results of five independent experiments, mefloquine inhibited JCV(Mad4) infection of SVG-A cells with the same efficacy as it did JCV(M1/SVEΔ) infection (Fig. 3C), demonstrating that mefloquine has an effect against a known pathogenic form of JCV.

Characterization of mefloquine's effect on JCV DNA replication.

To better understand mefloquine's mechanism of action and to address whether this drug inhibits viral DNA replication as opposed to a later step in the viral life cycle, a qPCR assay was employed to measure viral DNA levels in treated and untreated cells. The results presented in Fig. 3D indicate that the percentage of JCV DNA replication inhibited by mefloquine closely paralleled the percent inhibition of the infection rate, suggesting that mefloquine inhibits the infection rate at one of the steps involved in viral DNA replication and not VP1 protein expression. We sought to extend this observation to a JCV(Mad1) infection of the naïve cells most commonly used to propagate the virus in culture, PHFG cells. At days 7 and 10 postinfection, mefloquine inhibited JCV(Mad1) DNA replication in these cells with nearly the same efficacy (EC₅₀, ∼5 μM) that it inhibited JCV(M1/SVEΔ) DNA replication in SVG-A cells (Fig. 3E).

Effect of mefloquine on established JCV infection.

Our experiments indicated that mefloquine inhibited JCV infection when it was added to cells at the same time that the virus was added, but it was not clear from the results of those experiments whether mefloquine inhibited viral entry of the cells or a later step in the viral life cycle. Once PML is diagnosed, many cells have already been infected, and a preferred drug candidate for PML treatment should demonstrate an ability to inhibit an ongoing viral replication cycle. To investigate this possibility, mefloquine was added to cells at the time of infection or 3 or 24 h postinfection and the antiviral activity was measured. Mefloquine effectively inhibited JCV infection under all three conditions (Fig. 3F). Since most of the virus enters the cells within 1 h and all of the virus entered the cells by 24 h postinfection in our assay (the addition of JCV-neutralizing antiserum 24 h after virus addition was completely ineffective at blocking viral infection), our results suggest that mefloquine would effectively inhibit viral replication in cells already infected with JCV.

Lack of inhibitory effect of CSF on anti-JCV effect of mefloquine.

While 98% of the mefloquine in plasma is reported to be protein bound, high ratios of the concentration in tissue to the concentration in plasma have been described (41). It is not clear how mefloquine's high level of protein binding might affect its anti-JCV activity; therefore, we tested the drug's efficacy in the presence of increasing concentrations of human cerebrospinal fluid (CSF) to mimic the conditions encountered by the drug in the brain. To this end, we added CSF at a final concentration of 2%, 10%, or 20% over a range of mefloquine concentrations and calculated the EC₅₀ for viral inhibition. As shown in Fig. 4, the EC₅₀s were consistent for all levels of CSF added to the culture (up to 20%).

Characterization of antiviral effects of mefloquine enantiomers and analogs.

Mefloquine is a racemic mixture of (11R,12S) and (11S,12R) enantiomers of (2,8-bis-trifluoromethyl-quinolin-4-yl)-piperidin-2-yl-methanol hydrochloride (Fig. 5A and B). While there is a minimal difference between the activities of the enantiomers against malaria (5, 13), the (R,S) enantiomer has a much more potent (∼1,000-fold) antagonistic activity against the A2a adenosine receptor (59). The two enantiomers also display different pharmacokinetics and brain penetration properties (6, 11). To better understand the anti-JCV effects of the two components of the marketed mefloquine racemate, each enantiomer was separated from the racemate by chiral chromatography and tested in the JCV inhibition assay. The two enantiomers were found to have similar efficacies in inhibiting JCV. Although the difference in activity was small, it was statistically significant, with (R,S)-mefloquine being less than twofold more active (2.7 versus 4.5 μM; P = 0.02) than (S,R)-mefloquine or a racemate (2.7 versus 4.0 μM; P = 0.036) (Fig. 5A and B). On the basis of this result and on the reported brain concentrations for mefloquine enantiomers (6), we conclude that an efficacious concentration of mefloquine would be achieved in the brains of PML patients taking the drug. Furthermore, the very small difference in the anti-JCV activities between the enantiomers suggests that this effect is not mediated by the A2a adenosine receptor.

To further explore the structure-activity relationship for mefloquine, we acquired and tested other mefloquine analogs. Specifically, we tested a racemic mixture of (11S,12S) and (11R,12R) enantiomers of mefloquine (Fig. 5C) and of the pyridine analog of mefloquine, (2,8-bis-trifluoromethyl-quinolin-4-yl)-pyridin-2-yl-methanol (Fig. 5D). The activity of threo-(R*,R*)-mefloquine (Fig. 5C) was almost the same as the activity of erythro-(R*,S*)-mefloquine (4.6 versus 4.5 μM), indicating that some degree of flexibility exists in the chiral centers at positions 11 and 12 of the mefloquine molecule required for the inhibition of JCV replication. On the other hand, when the piperidine moiety was replaced with the pyridine, as in (2,8-bis-trifluoromethyl-quinolin-4-yl)-pyridin-2-yl-methanol (Fig. 5D), the anti-JCV activity was dramatically reduced. These results indicate that the presence of a saturated heterocycle at this position is crucial for target inhibition.

Modeling studies suggest shape similarities among chemically diverse JCV inhibitors.

To gain insights into the potential mechanism of action or molecular target of mefloquine in the inhibition of JCV replication, we analyzed the chemical classes of drugs with inhibitory activity (see the table in the supplemental material) and found that arylanthranilic and arylalkanoic acid nonsteroidal anti-inflammatory drugs (NSAIDs) (Table 1; Fig. 6) represented the most frequent class of inhibitors. This observation suggests that a common structural motif is shared by at least some JCV inhibitors. To explore the structural relationships among chemically diverse JCV inhibitors, we compared their three-dimensional shapes while assuming that molecules have similar shapes if their volumes overlay well. Conversely, any volume mismatch would represent a measure of dissimilarity. Such shape- and chemical feature-based comparisons (52) indicate that mefloquine, mefenamic acid, and indomethacin can occupy the same conformational space (Fig. 7). We then extended this analysis to compounds in clinical testing reported in the MDL Drug Data Report database and discovered that mefloquine overlaid well with several nucleoside analogues, such as 8-chloroadenosine 3′,5′-monophosphate (Fig. 7) and 3-deazaadenosine (data not shown). The last two compounds inhibited the rate of infection of JCV(M1/SVEΔ) in SVG-A cells with efficacies similar to the efficacy observed with mefloquine (Fig. 7 and data not shown). Taken together, these data suggest that chemically diverse JCV inhibitors may have a common mechanism of viral inhibition and act, in part, as mimetics of nucleoside analogues.