Results
In reporting the results of our experiments, we apply the term “infection” to measurements of virus cytopathic effect in permissive cells, and the term “replication” to the measurement of genome copies or protein expression.
We chose to use GA C15:1 as the main GA type in our experiments because it is the one most used experimentally to date. However, we also tested GA C13:0 and GA C17:1 in some of our experiments. Unless stated otherwise, GA C15:1 was used.

GA IC50 vs. EC50
The toxicity EC50 of GA on HEp-2 cells in MEM supplemented with 1% fetal bovine serum (FBS) was 27.63 ± 2.21 μM (Fig. S1A) or 63.5 ± 3.2 μM (Fig. S1B) on cells supplemented with 5% FBS. In cells infected with HCMV strain CH19 and supplemented with 1% FBS, the 50% inhibitory concentration (IC50) was 6.83 ± 1.08 μM, and with 10% bovine serum, the IC50 of GA was 41 μM (Fig. 1A). The toxicity EC50 of GA on HFF cells was 14.00 μM ± 0.1 in MEM medium supplemented with 1% FBS (Fig. S1C), and 137.04 μM ± 10.44 (Fig. S1D) in MEM medium supplemented with 10% FBS. The results indicated that the activity and toxicity of GA is affected by the serum concentration in the medium (see discussion).
Figure 1 Ginkgolic Acids (C13:0, C15:1, C17:1) inhibit HCMV in a dose-dependent manner and prevents plaque formation. (A) Monolayers of human foreskin fibroblasts (HFF) were inoculated with 50–100 cells infected with one of two cell-associated clinical isolates of HCMV (CH19 and BI-6) in a total of 4 wells per drug concentration followed by treatment with medium containing 0–10 µM GA. Viral infection was allowed to progress for 7 days when the mean number of HCMV plaques per concentration was counted, and the IC50 was determined. The effects of different concentrations of FBS and time delay on the IC50 are also shown. (B) Dose-dependent inhibitory effect of GA on infection by the cell-free virus strain PT30CMV-GFP. (C) Inhibitory effect of 10 µM GA C15:1 compared to 16 µM GCV on HCMV-GFP cell-free infection. (D) Inhibitory effect of 10 µM GA C13:0 on HCMV-GFP infection, and (E) Inhibitory effect of 10 µM GA C17:1 on cell-free HCMV-GFP infection. (B–E) representing average of 20 scanned fields.

GA inhibits HCMV infection in human foreskin fibroblasts (HFF)
To assess the effect of GA on HCMV infection, dilutions of 1 µM to 20 µM were made (Fig. 1A). Monolayers of human foreskin fibroblasts (HFF) were incubated for 1 hour with GA and then infected with two cell-associated clinical isolates of HCMV: CH1914 and BI-615. Because neither virus produces extracellular virus, plates for plaque assays were inoculated with 50–100 infected cells per well, and then treated with medium containing 1–20 µM GA, or vehicle control. Viral infection was allowed to progress for 7 days, and then quantitated by plaque reduction assay14. GA inhibited infection of BI-6 and CH19 with an IC50 of 7.26 ± 0.92 μM and 6.83 ± 1.08 μM, respectively.
To further assess the effect of GA on plaque formation, we used the CMVPT30-GFP strain, which produces cell-free virions. HFF monolayers in 24-well plates pretreated with GA (1–20 µM) for 1 h were inoculated with 50–100 plaque-forming units of CMVPT30-GFP per well.16, (Fig. 1B). The number of plaques were counted after 7 days and the IC50 was determined. As predicted by the previously-determined IC50, at 5 and at 10 µM the antiviral effect was apparent. Furthermore, at 10 µM, only single cells were infected, indicating that there was no cell-to-cell virus transmission. These results may be explained as inhibition of secondary infections or that viral entry may be a target of GA. To verify these results, we compared the inhibition of infection by GA using a plaque reduction assay (PRA) similar to the PRA for ganciclovir (GCV), the preferred antiviral HCMV drug of choice. The results showed that although there was strong inhibition of the virus at 16 µM GCV, there were still visible plaques (>5 adjacent cells showing HCMV cytopathic effect) indicating cell-to-cell transmission of the virions. On the contrary, CMVPT30-GFP infected monolayers that were treated with 10 µM GA only displayed individual infected cells in intact HFF monolayers, (Fig. 1C), thus indicating no cell-cell transmission. We also tested GA C17:1 and GA C13:0. Both GA compounds had a strong inhibitory effect on HCMV infection by PRA, similar to GA C15:1 (Fig. 1D,E). None of the GA structures showed cytotoxicity at their active concentrations, respectively.

GA inhibits HSV-1 and ZIKV infection
We tested GA on HSV-1, a rapidly replicating and lytic DNA virus, and on ZIKV, an RNA virus which has neither glycoprotein conservation nor common receptors with HCMV or HSV-1. To test whether GA inhibits HSV-1, we designed an experiment to test the direct effect of GA on the virus. For this experiment, we used HEp-2 and 293T cells. 1 × 107 PFU HSV-1 strain F was treated for 1 hour with GA (50 µM) or with vehicle in serum free DMEM and then each of these were used to infect HEp2 or 293T cells at an MOI of 0.5 in 199V medium with a final concentration of 2.5 µM GA. As a control, the virus that was originally treated with the vehicle was then supplemented also with 2.5 µM GA. To test for successful viral infection of HEp2 and 293T cells, production of HSV-1 immediate early (ICP27), early (ICP8), and late (US11) proteins was analyzed by Western Blot (Fig. 2A,B). The results showed that in the treated HSV-1 F stock there was complete inhibition of viral replication, as indicated by lack of protein synthesis. This implies that viral entry may be a target of GA, because direct treatment of HSV-1 with GA blocks downstream HSV-1 protein production. To test the effect of GA on HSV-1 replication, HEp2 cell monolayers were grown to 90% confluency in a 6-well plate, treated with 10 µM GA C15:1 or vehicle for 1 hour in 199V medium, and then inoculated with HSV-1 strain F for protein analysis. HSV-1 ICP27, ICP8, and US11 proteins were detected by Western Blot (Fig. 2C). The results showed profound inhibition of immediate early, early, and late viral proteins. The inhibition of the full temporal range of HSV-1 proteins implies that inhibition of viral replication occurs by blocking the entry of virions into the target cell. To evaluate the effect of GA on progeny virions, the supernatant of GFP-HSV-1 strain 17+ infected Vero cells at an MOI of 1 was collected after 20 h and titered by plaque reduction assay (Fig. 2D). The results showed a significant decrease of approximately 2 log in the GA-treated cell titers.
Figure 2 GA inhibits HSV-1 infection. (A) 1 × 107 PFU HSV-1 strain F were treated for 1 hour with 50 µM GA (Lanes 1–3) or with vehicle (Lanes 4–6) in serum free DMEM and then were used to infect HEp2 or 293T cells at an MOI of 0.5 in 199V medium with final concentration of 2.5 µM GA. 6-well cultures of HEp-2 (A) or 293T (B) were then infected, and at 5, 10 and 29 hours post-inoculation, cells were collected and a fraction of the total cell lysate was subject to Western Blot analysis using antibodies directed against ICP8, ICP27, US11 and β-Actin. (C) HEp2 cell monolayers were treated with 10 µM GA C15:1 or vehicle for 1 hour in 199V medium and then infected with HSV-1 strain F at 4, 8, 12, 24 and 32 hours post-inoculation cells were collected and a fraction of the total cell lysate was subject to Western Blot analysis using antibodies directed against ICP8, ICP27, US11 and β-Actin. Normalized ratios of protein expression are in the bar graph. (D) Titration by plaque assay of Vero cells pretreated with 10 µM GA C15:1 or vehicle and then infected with GFP-HSV-1 strain 17+.
To test the effect of GA on ZIKV infection, NHA were grown to 90% confluency in a 24-well plate, treated with GA or DMSO for 3 hours without serum and then infected with ZIKV strain PRVABC59 at an MOI of 0.3. The next day, supernatant was replaced with fresh medium containing GA or DMSO and cells were incubated at 37 °C, 5% CO2. On day 7, viability was determined with the MTS Cell Proliferation Assay (Promega), and then these cells were harvested to extract total RNA for quantification of ZIKV RNA by Taqman based qRT-PCR. The results showed 70% to 80% viability at 5 µM to 20 µM GA, compared to less than 40% viability for the vehicle-treated cells. Furthermore, an 80–90% decrease in ZIKV RNA was observed at 5 µM to 20 µM GA (Fig. 3A,B). We concluded that GA inhibits entry of Zika virions and prevents NHA cell death. This suggests that the mechanism of inhibition that appears to be targeted in HSV and HCMV by GA is conserved among enveloped viruses that have no homologous glycoproteins and use different cell receptors for entry. To test the direct effect of GA on ZIKV, 5 × 105 PFU ZIKV were treated for 1 hour with GA (10 µM/ml) or with vehicle in 199V medium, and then were used to infect Vero cells at an MOI of 0.5. Supernatant was collected at 4, 24 and 48 hours post-inoculation for cell free viral RNA copy number determination (Fig. 3C). The results indicated that the GA-treated ZIKV was completely inhibited, as indicated by cell free ZIKV RNA copy number. This suggests that viral entry of ZIKV, may be a target of GA because direct treatment of ZIKV with GA blocks downstream ZIKV RNA production.
Figure 3 GA inhibits ZIKV infection. (A) Viability of NHA infected with ZIKV and treated with GA. NHA were grown in 24-well plates to >80% confluency and treated with GA (0–20 µM) or DMSO for 3 hours and infected with ZIKV strain PRVABC59 at an MOI of 0.3. At 12 hpi, supernatants were carefully removed and replaced with fresh AGM. (B) 7 days post infection, samples were analyzed for cell viability by the MTS assay and then live cells were harvested for RNA and processed with Taqman based real-time PCR to quantify ZIKV RNA. Experiments were performed in duplicate three independent times and the data were analyzed by t-test (* indicates p ≤ 0.05). (C) Cell free ZIKV in 199V medium was pretreated with 10 µM GA or with DMSO for 1 hour at 37 °C. The pretreated ZIKV was used to infect Vero cells grown in 6 well plates >80% confluency with 0.5 MOI for 1 hour. The infected cells were washed twice with warm DPBS and supplemented with DMEM supplemented with 5% FBS. Supernatant was collected at 4, 24 and 48 hours post-inoculation for cell free viral RNA copy number determination. Experiments were performed in triplicates.

GA inhibits a non-enveloped adenovirus
To assess the effect of GA on non-enveloped virus, we also tested its antiviral activity on adenovirus, which is internalized by endocytosis. Monolayers of Vero cells were incubated for 1 hour with medium containing 10 µM GA C15:1 or DMSO vehicle. The cells then were inoculated with a replication-defective human adenovirus type 5 (dE1/E3) containing GFP (Ad-GFP) or GFP-HSV-1 virus, at an MOI of 0.5 in a medium containing 10 µM GA C15:1 for 24 hours. The inoculated Vero cells were evaluated by fluorescence microscopy (Fig. 4). The results showed significant HSV-1 inhibition, as well as inhibition of entry of the adenovirus.
Figure 4 GA inhibits non-enveloped human adenovirus. (A) Monolayers of Vero cells were incubated for 1 hour with medium containing 10 µM GA C15:1 or DMSO vehicle. The cells were infected with GFP-HSV-1 virus at an MOI of 0.5 in medium containing 10 µM GA C15:1 for 24 hours. Infection was evaluated by light (lower panels) and fluorescence (upper panels) microscopy. (B) Monolayers of Vero cells were incubated for 1 hour with medium containing 10 µM of GA C15:1 or DMSO vehicle. The cells were inoculated with a replication-defective human adenovirus type 5 (dE1/E3) containing GFP (Ad-GFP) at an MOI of 0.5 in a medium containing 10 µM GA C15:1 for 24 hours. Inhibition of entry was evaluated by light (lower panels) and fluorescence (upper panels) microscopy.

GA inhibits cell fusion induced by all three classes of viral fusion proteins
Our data suggested that GA targets virus-cell fusion. To assess the activity of GA C15:1 on virus-cell fusion we modeled virus-cell fusion by the fusion of cells expressing viral fusion proteins to target cells. Cell-cell fusion was monitored by the spread of fluorescent dyes. Based on crystallographically identified three-dimensional structures, the folds of all virus fusion proteins fall into one of three classes. Effector COS7 cells were transfected to express all three classes of fusion proteins (Table S2), including ZIKV, HIV, EBOV, IAV, SFV, VEEV, VSV and EBV.
The COS7 cells were loaded with calcein AM and bound to 293T target cells that were either unlabeled or, for purposes of microscopic identification, loaded with the aqueous dye CMAC, as illustrated for EBOV GP-induced fusion (Fig. 5A). We tested the effect of adding GA on cell-cell fusion mediated by four class I, two class II, and two class III fusion proteins. For all eight proteins, 10 µM GA C15:1 completely blocked fusion (Fig. 5B). Furthermore, when EBOV GP was treated with 5 µM or with 10 µM GA C13:0, GA C15:1, or GA C17:1 (Fig. 5C), 10 µM of all GA derivatives completely blocked fusion.
Figure 5 GA inhibits cell fusion induced by all three classes of fusion proteins. (A) In the two top panels, typical images used to visualize the extent of fusion in the absence (control, left) and presence of GA (right) are shown. Fused cells (light blue, due to mixing of the green calcein in effector cells and dark blue CMAC in target cells) are marked by arrows in the control. Cells did not fuse in the presence of 10 µM GA. The extent of fusion is quantitated in the bar graph. The experiments used EBOV GP as the fusion protein. (n = 3). For all bars of experiments using cell-cell fusion, means ± SEM are shown. (B) Extent of cell-cell fusion induced by fusion proteins from three classes of viral fusion proteins. The proteins for class I are: ZIKV (Zika E), HIV Env, EBOV (Ebola GP), IAV (Influenza HA); for class II: SFV (Semliki forest virus) E1/E2, VEEV (Venezuelan Equine Encephalitis Virus) E; for class III: VSV (Vesicular Stomatitis Virus) G, EBV (Epstein Barr Virus) gB. Effector cells were transfected (except for HIV Env which was stably expressed in a cell line) with the indicated viral proteins. Each was bound and allowed to fuse with target cells. The reduction in fusion caused by adding 10 µM GA is shown as low grey bars labeled 10 µM GA. For every fusion protein, the presence of GA abolished fusion. n = 3 for each fusion protein. (C) Effect of two different concentrations of GA with different side chains on cell-cell fusion.
When GA was removed, fusion was restored (Fig. 6A, Bar 4), indicating that GA interferes with fusion in a reversible manner. Viral fusion, regardless of fusion protein, proceeds through the creation of hemifusion, an intermediate state in which proximal lipid monolayer leaflets of membranes, in contact with one another, have merged, but the distal monolayers remain distinct. Because inhibition of fusion was universal, independent of viral protein, it is extremely unlikely that GA targeted the fusion proteins themselves. This is consistent with GA inhibiting the creation of the hemifusion intermediate, and universal inhibition would be expected. It is well known that agents conferring positive spontaneous curvature, such as lysophosphatidylcholine (LPC), inhibit fusion induced by viral and non-viral fusion proteins17. Likewise, the cone-shaped lipid oleic acid (OA) has a negative spontaneous membrane curvature, which favors hemifusion when present in the outer bilayer, and its presence relieves the inhibition of fusion18. The addition of OA together with GA abolished the inhibitory effect of GA (Fig. 6B), indicating that GA induces positive membrane curvature.
Figure 6 OA together with GA abolished the inhibitory effect of GA. (A) Dissecting the stages of fusion affected by the presence of GA. Bar 1, control: GA was not present. Bar 2: GA was maintained throughout the experiments. Bar 3: Effector and target cells were incubated for 30 min. in the presence of 10 µM GA. The neutral pH bathing solution was then replaced by a pH 5.7 solution, which did not contain GA, and this was maintained for 10 min. The low pH bathing solution was replaced by one at neutral pH, also without GA, and fusion was measured. Bar 4: The same protocol was followed as for experiments of bar 3, but prior to acidification, the GA-containing neutral pH solution was replaced by a GA-free neutral pH solution that was maintained for 10 min to allow GA in the membranes to dissociate into the aqueous solution. This almost restored the extent of fusion to that of control. Bar 5: The same protocol as for experiments of bar 3 was used, but GA was maintained through acidification. The subsequent neutral pH wash solution did not contain GA. EBOV GP was the fusion protein for these experiments. (n = 3). (B) The inhibition of cell-cell fusion by GA is reversed by the presence of oleic acid (OA). Upper panels: Massive dye mixing was observed for control (left image), but none was observed in the presence of 10 µM GA (middle image). Addition of 250 µM OA along with GA (right image), resulted in the same degree of fusion as for the control. Extent of fusion is quantified in the bar graphs. (n = 3).

GA inhibition of enveloped viruses correlates with inhibition of viral DNA and protein synthesis
To validate the PRA observations, we tested whether there is an effect on downstream viral DNA synthesis in HCMV infected cells when treated with GA. HFF were inoculated with HCMV for 3 hours, then washed and incubated with different concentrations of GA or vehicle. DNA was extracted from 4 pooled wells of HFF at 7 days post inoculation, and then analyzed by qPCR targeting the viral polymerase gene. DNA copies were reduced in a dose dependent manner by the addition of GA to culture medium (Fig. 7A). Although HCMV DNA copy numbers were reduced by GA, this could be secondary to inhibition of virus entry rather than direct inhibition of DNA replication.
Figure 7 GA inhibits HSV-1 replication by inhibition of viral protein synthesis and HCMV by DNA synthesis. (A) Effect of GA on HCMV DNA synthesis. Viral replication is inhibited in the presence of increasing concentration of GA, as determined by decreasing viral DNA copies. DNA was extracted from 4 pooled wells of HFF at 7 dpi, then analyzed by qPCR targeting the viral polymerase gene. DNA copies were reduced in a dose dependent manner by the addition of GA to culture medium as determined by ANOVA analysis with pairwise post-hoc t-test analysis in SPSS. Data are presented as mean percent of control. (*5 µM p = 0.014, 8 µM p = 0.003, 10 µM p = 0.002). 6 well cultures of HEp-2 (B) or 293T (C) were infected with HSV-1 strain F at an MOI of 1 for 2 hours, then washed with 199V and supplemented with 10 µM GA (Lanes 1–4) or vehicle (Lanes 5–8). Mock (M), 5, 11 and 24 hours after inoculation, cells were collected and a fraction of the total cell lysate was subject to Western Blot analysis using antibodies directed against ICP8, ICP27, β-Actin and US11. Normalized ratios of protein expression are in the bar graphs.
Viral protein synthesis was determined in HSV-infected cells following GA treatment. Untreated HEp-2 and 293T cells were infected with HSV-1 F’ at an MOI of 1 for 2 hours, allowing the virions to internalize into the cells, then washed with 199V medium and supplemented with 10 µM GA or vehicle. The viral replication was evaluated by Western Blot detection of HSV-1 proteins. Although the infection of HEp-2 and 293T cells had been initiated, the addition of 10 µM GA to the infected HEp-2 and 293T cells inhibited the virus replication from the point of exposure to GA (Fig. 7B,C). The observed effect could be the result of inhibition of secondary infections. However, the early down regulation of HSV proteins, occurring prior to a complete viral replication cycle, suggests that there could be a secondary inhibitory mechanism targeting viral protein synthesis.