Screening and identification of CMLDBU6128.
We generated single- and multiple-reporter vaccinia viruses with fluorescent proteins under the control of temporally regulated early (C11R
), intermediate (G8R
), and/or late (F17R
) viral promoters () that can be used to monitor stages of vaccinia virus gene expression (7
). For screening, we used a dual reporter virus that contained late mCherry (red) and early Venus (LREV virus). Compounds were screened for the ability to reduce viral reporter gene expression, determined by well fluorescence, following high-MOI infection of A549 cell monolayers (A). The compound library was developed at CMLDBU and consisted of 2,070 compounds with high stereochemical and positional variation to provide broad coverage of chemical space in a limited screen.
Replication-competent vaccinia virus reporter viruses (based on strain Western Reserve)
Fig 1 (A) Screen summary. Confluent A549 cells were infected with a fluorescence reporter vaccinia virus at a high MOI. A diversity-oriented synthesis library was screened for compounds that reduced reporter expression at 12 h postinfection. (B) Structure and (more ...)
None of the library compounds reduced early Venus expression at 12 h postinfection, and the majority also did not prevent late mCherry expression. However, three compounds reduced late mCherry expression to 29 to 56%. Of these, one did not show activity upon resynthesis, and another was antiviral but had a cell-type-specific cytopathic effect. The third, CMLDBU6128, was chosen for further characterization. The structure of the pyridopyrimidone compound CMLDBU6128 and the optimized synthesis scheme that was developed are depicted in B.
CMLDBU6128 is an antiviral compound.
The effect of CMLDBU6128 on a single-reporter virus containing late Venus (LV virus) is shown in C. The 50% inhibitory concentration (IC50
) for LV reporter expression at 12 h postinfection was ~5.3 μM, and the IC90
was ~10.4 μM, with similar results obtained in both A549 and HeLa cells. Little to no cell cytotoxicity was observed at 24 h for up to 80 μM CMLDBU6128 in either cell type. We therefore chose a compound concentration of 20 μM for subsequent experiments. The noncytotoxic, antiviral properties of CMLDBU6128 are further illustrated in D. A549 cells were infected with LV virus at a low MOI, and the infection was allowed to spread for 48 h. With DMSO but not CMLDBU6128, this resulted in a monolayer of Venus-expressing cells from complete virus transmission. Cell monolayers were then subjected to medium exchange to drug-free conditions and to secondary, high-MOI infection with a single-reporter virus containing late mCherry (late red [LR] virus). Twenty-four hours following secondary infection, mCherry-expressing cells were observed only with the CMLDBU6128-limited primary infection, as vaccinia virus-infected cells do not support superinfection (3
). The results show that CMLDBU6128 limits virus spread and that viable, infection-supporting cells remain in this limited infection.
To confirm the antiviral effect more directly, low- and high-MOI viral growth assays using wild-type vaccinia virus strain Western Reserve were performed (). Two reference compounds were included: ST-246, which inhibits the egress of infectious viral progeny, and AraC, a nucleoside analog and inhibitor of DNA replication (31
). In high-MOI assays with A549 cells (A), the virus yield was reduced by 3.6 log (CMLDBU6128), 0.3 log (ST-246), and 3.5 log (AraC) at 24 h. In low-MOI assays with A549 cells (B), the virus yield was reduced by 3.3 log (CMLDBU6128), 1.0 log (ST-246), and 4.5 log (AraC) at 72 h. Similar results were obtained in low-MOI assays with Vero cells, indicating that the inhibitory effect was not cell or species specific (C) (1.9-, 2.0-, and 4.9-log reductions with CMLDBU6128, ST-246, and AraC, respectively). Microscopy of the 72-h multicycle endpoints showed intact cell monolayers with the inhibition of low-MOI virus spread by CMLDBU6128 (D). Consistent with inhibition of late viral reporter gene expression, CMLDBU6128 was antiviral in viral proliferation assays. This was further confirmed in a multicycle growth assay, where it reduced infectious progeny formation 24, 48, and 72 h following low-MOI infection of A549 cells (E).
Fig 2 (A) Single-cycle growth yields on A549 cells. (B and C) Multicycle growth yields on A549 and Vero cells, respectively. (D) Phase-contrast microscopy of cell monolayers of A549 and Vero cells at the endpoint of multistep growth for panels B and C (10× (more ...) CMLDBU6128 interferes with intermediate and late viral gene expression.
We next tested the effect of CMLDBU6128 on a panel of single-reporter viruses harboring early Venus (EV virus), intermediate Venus (IV virus), or late Venus (LV virus) (). A fluorescence plate reader was used for automated hourly monitoring of Venus expression after high-MOI infection of A549 cells (). Reporter expression by the canonical promoter elements occurred with the anticipated timing and magnitude, and there was no detectable Venus expression with a virus that contained Venus with no preceding promoter element (promoterless Venus [PLV] virus) (A to C and data not shown). EV reporter expression was unaffected by CMLDBU6128, consistent with the lack of effect on early reporter expression in our initial screen (A). In contrast, IV reporter expression initiated correctly but reached only half-maximal levels (B). A similar but more pronounced defect was observed with LV reporter expression, with maximum fluorescence reduced by 85% (C). Virtually identical results were observed when the compound was added 4 h before virus, as opposed to simultaneously, indicating that these differences were not due to kinetic limitations of drug uptake or action (data not shown).
Fig 3 (A) A549 cells were infected with early Venus (EV; A), intermediate Venus (IV; B), and late Venus (LV; C) reporter viruses at a high MOI in the presence of DMSO or CMLDBU6128. Viral reporter fluorescence was measured hourly for 12 h. (D) Promoter-dependent (more ...)
We analyzed reporter mRNA accumulation to determine if this was the underlying defect in reporter expression. This procedure utilizes quantitative RT-PCR and normalization to infection with the promoterless Venus virus to determine promoter-dependent Venus mRNA accumulation (7
). Over an 8-h time course, peak Venus mRNA accumulation for EV, IV, and LV virus infections occurred at 1, 4, and 8 h postinfection, respectively (D). EV reporter mRNA accumulation was unaffected by CMLDBU6128; however, IV and LV reporter mRNA accumulation initiated correctly but failed to reach maximal levels. Peak IV reporter mRNA was reduced by 41%, and peak LV reporter mRNA was reduced by 73%, mirroring the reporter expression defects described above.
The results obtained using these single-reporter viruses were validated by microscopy using a multireporter triple virus (TrpV) which contains early Venus, intermediate mCherry, and late TagBFP (7
). A549 cells were infected with TrpV at a high MOI and visualized at 12 h postinfection (E). Consistent with the results described above, early Venus expression was unaffected and intermediate mCherry and late TagBFP expression was reduced uniformly in the infected cell population. Phase-contrast imaging following this high-MOI infection showed that despite this, CMLDBU6128 was unable to prevent the cell morphological changes associated with infection-induced cytopathic events.
CMLDBU6128 does not prevent viral DNA replication or factory formation.
Viral DNA replication resets the transcriptional landscape to postreplicative gene transcription, and consequently, DNA replication inhibitors block intermediate and late viral gene expression (16
). To monitor viral DNA replication, we used quantitative PCR to track Venus DNA copy numbers after infection of A549 cells with LV virus. A is a log-scale plot of Venus copies per cell equivalent (c.e.) showing that viral DNA replication occurred normally in the presence of CMLDBU6128. The onset of DNA replication was between 2 and 3 h postinfection, and the number of Venus DNA copies/c.e. increased from 1 h postinfection to 8 h postinfection, from 1 to 2 copies to 574 and 529 copies with DMSO and CMLDBU6128, respectively.
Fig 4 (A) Venus DNA copy number per A549 cell equivalent (c.e.) after high-MOI infection with LV reporter virus. Venus copy numbers were calculated from a standard curve of Venus-containing plasmid. (B) High-magnification microscopy after infection of A549 (more ...)
We next used high-magnification microscopy to determine if viral factory formation occurred normally in the presence of CMLDBU6128. B depicts images obtained 1, 8, and 24 h after infection of A549 cells with a vaccinia virus harboring a Venus fusion of the late-expressed viral core protein A4L (Venus-A4L virus). A DAPI (4′,6-diamidino-2-phenylindole)-staining perinuclear viral factory was evident by 8 h and persisted to 24 h postinfection in both DMSO- and CMLDBU6128-treated infections. However, only with DMSO treatment was Venus-A4L visible as a punctate signal in and around viral factories at 8 h postinfection. By 24 h, this signal had spread throughout the cell as Venus-A4L-containing progeny viruses trafficked toward the cell membrane (11
). CMLDBU6128-induced defects in intermediate and late gene expression were therefore uncoupled from any gross defects in viral DNA replication or factory formation.
A CMLDBU6128-arrested infection has a general loss of protein synthesis.
To investigate more globally the effects of CMLDBU6128 on viral gene expression, we used [35S]methionine-cysteine pulse labeling at various times postinfection to monitor active protein translation (A). Cellular protein synthesis in mock-infected cells was unaffected following 12 h of treatment with CMLDBU6128. In control vaccinia virus-infected cells, a characteristic shutoff of host gene expression and transition to viral protein synthesis were observed. CMLDBU6128 did not prevent this host cell shutoff; however, viral protein synthesis was generally reduced. The appearance of viral proteins was diminished slightly at 6 h postinfection and more profoundly at late time points. Despite the virtual absence of active protein synthesis, these infection-arrested cells continued to be viable in metabolic activity assays 12 and 24 h after infection (data not shown).
Fig 5 (A) [35S]methionine labeling of cells to visualize active translation. Densitometry values are provided normalized to 12-h mock-infected cells (−). A representative slice of the Coomassie blue-stained gel is also shown. (B) Time course analysis (more ...)
Because vaccinia virus gene expression occurs in a cascade mechanism, with a requirement of preceding gene expression products for subsequent gene expression, we tested whether delayed drug addition could inhibit established late gene expression. We also tested whether drug withdrawal from an initially inhibited infection would alleviate inhibition. The LV reporter virus was used to allow for quantitative analysis (B). Cells received an initial 0-h treatment followed by medium removal and a second treatment at 10 h, and LV reporter fluorescence was then measured hourly for 12 h (from 10 h to 22 h). Unimpeded infections showed high Venus expression at 10 h, which continued to increase over the time course. CMLDBU6128 addition to an unimpeded infection at 10 h resulted in a flattening of Venus expression within 2 h of addition, showing that the compound could inhibit late gene expression following diffusion into infected cells.
In the complementary experiment, a CMLDBU6128-arrested infection had significantly reduced Venus expression at 10 h, as seen above. However, drug removal at this time did not result in any detectable rise in Venus expression. Indeed, the level of Venus expression following drug removal was indistinguishable from that with sustained drug inclusion. For comparison, infection of A549 cells with LV virus at the 10-h time point showed a clear rise in reporter expression over the course of this analysis. The results indicate that CMLDBU6128 can inhibit late gene expression even after it is established. They also suggest that a CMLDBU6128-arrested infection may be irretrievably defunct.
Mutations in the large subunit of the viral RNA polymerase bypass inhibition.
We next determined whether drug-resistant virus could be obtained through serial passage of LV virus in the presence of CMLDBU6128. After six serial passages, this resulted in clear drug resistance, as seen in low-MOI viral spread assays of two independent selection pools (A, drug-resistant pools DR1 and DR2). Two clones from each of these pools were randomly chosen and plaque purified to yield clones DR1a, DR1c, DR2b, and DR2c. These viruses were completely resistant to CMLDBU6128 in single-cycle growth assays (B).
Fig 6 (A) Fluorescent foci from low-MOI infection of A549 cell monolayers with parental LV virus or two selection pools (DR1 and DR2) in the absence or presence of CMLDBU6128. Images were taken using a 2.5× objective. (B) Single-cycle growth yields (more ...)
A stock of parental LV virus and all four clones were sequenced by Illumina whole-genome sequencing. Compared to the reference GenBank sequence for vaccinia virus strain Western Reserve, the parental LV virus had four potential differences: G1091A, A1092G, T194371C, and G194372T. These fell within terminal repeat regions of the viral genome and were likely mapping errors. In addition, all four drug-resistant viruses contained a silent mutation (G185627T; VACWR208 gene, amino acid P140P), which we presume was acquired by LV virus in an intermediate passage. Each drug-resistant pool yielded a single, different coding change in J6R, the gene encoding the large subunit of the viral RNA polymerase. DR1a and DR1c had a T85842G mutation, which encodes a V576G mutation, and DR2b and DR2c had a C86976T mutation, which encodes an A954V mutation (C).
We confirmed that these unique DR1 and DR2 mutations could confer drug resistance in a marker rescue experiment (D). A549 cells were infected with wild-type virus and subsequently transfected with an ~1-kb PCR fragment surrounding either the J6R V576 or J6R A954 region generated from the corresponding drug-resistant isolate or wild-type control virus. CMLDBU6128 was added to these infections, and viral harvest was performed 2 days later. Viral yield assays showed that rescue with the DR1 fragment resulted in a 40-fold increase in virus yield. Rescue with the DR2 fragment resulted in a 10-fold increase in yield. Taken together, these data indicate that either a V576G mutation or an A954V mutation in J6R can confer resistance to CMLDBU6128.
CMLDBU6128 has activity against multiple orthopoxviruses.
J6R is highly conserved among orthopoxviruses, with >96% identity between vaccinia, monkeypox, cowpox, and variola viruses. This suggested that CMLDBU6128 might act as a broad-spectrum inhibitor. We tested the effect of CMLDBU6128 on orthopoxviruses other than vaccinia virus strain Western Reserve. The results of single-cycle growth assays of monkeypox virus Zaire 1979 (MPOX), cowpox virus Brighton red (CPX), and vaccinia virus strain IHDJ (VACV IHDJ), using A549 cells, are shown in . CMLDBU6128 inhibited the replication of all of these viruses, reducing viral yields by 2.5 log (MPOX), 4.0 log (CPX), and 2.4 log (VACV IHDJ). The replication of these viruses was also inhibited on Vero cells (data not shown). These findings are consistent with drug targeting of a conserved orthopoxvirus function and demonstrate that CMLDBU6128 has broad-spectrum efficacy against orthopoxviruses.
Single-cycle growth yields for A549 cells infected with monkeypox virus Zaire 1979 (MPOX), wild-type cowpox virus (CPX), or vaccinia virus (VACV) IHDJ in the absence or presence of CMLDBU6128.