Evaluation of fendiline treatment in VP40 system with nucleation-elongation process: a computational model of Ebola virus matrix protein assembly

ABSTRACT Ebola virus (EBOV) infection is threatening human health, especially in Central and West Africa. Limited clinical trials and the requirement of biosafety level-4 laboratories hinder experimental work to advance our understanding of EBOV and the evaluation of treatment. In this work, we use a computational model to study the assembly and budding process of EBOV and evaluate the effect of fendiline on these processes in the context of fluctuating host membrane lipid levels. Our results demonstrate for the first time that the assembly of VP40 filaments may follow the nucleation-elongation theory, as this mechanism is critical to maintaining a pool of VP40 dimers for the maturation and production of virus-like particles (VLPs). We further find that this nucleation-elongation process is likely influenced by fluctuating phosphatidylserine (PS), which can complicate the efficacy of lipid-targeted therapies like fendiline, a drug that lowers cellular PS levels. Our results indicate that fendiline-induced PS reduction may actually increase VLP production at earlier time points (24 h) and under low fendiline concentrations (≤2 µM). However, this effect is transient and does not change the conclusion that fendiline generally decreases VLP production. In the context of fluctuating PS levels, we also conclude that fendiline can be more efficient at the late stage of VLP budding relative to earlier phases. Combination therapy with a VLP budding step-targeted drug may therefore further increase the treatment efficiency of fendiline. Finally, we also show that fendiline-induced PS reduction more effectively lowers VLP production when VP40 expression is high. Taken together, our results provide critical quantitative information on how fluctuating lipid levels (PS) affect EBOV assembly and egress and how this mechanism can be disrupted by lipid-targeting molecules like fendiline. IMPORTANCE Ebola virus (EBOV) infection can cause deadly hemorrhagic fever, which has a mortality rate of ~50%–90% without treatment. The recent outbreaks in Uganda and the Democratic Republic of the Congo illustrate its threat to human health. Though two antibody-based treatments were approved, mortality rates in the last outbreak were still higher than 30%. This can partly be due to the requirement of advanced medical facilities for current treatments. As a result, it is very important to develop and evaluate new therapies for EBOV infection, especially those that can be easily applied in the developing world. The significance of our research is that we evaluate the potential of lipid-targeted treatments in reducing EBOV assembly and egress. We achieved this goal using the VP40 system combined with a computational approach, which both saves time and lowers cost compared to traditional experimental studies and provides innovative new tools to study viral protein dynamics.

Difference in VLP production between 2 µM and 10 µM fendiline application decreases with later application time.The difference in VLP measurement is not significant at 24 h when fendiline is applied at 0 or 12 hours, and at 48 h when fendiline is applied at 24 or 36 hours (Table S10).fendiline Error bars indicate the SEM.

Figure S1 .
Figure S1.Simulation from the 'As2' model.(A) Oligomer ratio at 24 h.(B) VLP production at 24 h.(C) VP40 budding ratio at 48 h.(D) VP40 plasma membrane localization.(E) Relative VLP production.(F) Relative oligomer frequency.Both the deceasing trend of relative frequency from membrane VP40 dimer to 42mer, and the increasing trend in higher oligomers under higher PS level are reproduced.The three bars in each of the sub-column are 14.39%, 16,52%, 20% PS from left to right separately.Error bars indicate SEM from top 5 fits.

Figure S2 .
Figure S2.Simulation from the 'As3' model.(A) Oligomer ratio at 24 h.(B) VLP production at 24 h.(C) VP40 budding ratio at 48 h.(D) VP40 plasma membrane localization.(E) Relative VLP production.(F) Relative oligomer frequency.Both the deceasing trend of relative frequency from membrane VP40 dimer to 42mer, and the increasing trend in higher oligomers under higher PS level are reproduced.The three bars in each of the sub-column are 14.39%, 16,52%, 20% PS from left to right separately.Error bars indicate SEM from top 5 fits.

Figure S3 .
Figure S3.Parameter distribution of groups with fendiline treatment induced VLP production in late and early budding dynamic groups.(A) Fendiline-induced VLP increase groups in late VLP production type have typical low k3 and k3,1'.(B) Other Fendiline-induced VLP increase groups are more related to high k4.Values are normalized to the sampling range of parameters.The Y axis range shows the relative value of each parameter in their LHS.0 indicates lower bound and 1 indicates upper bound.

Figure S6 .
Figure S6.Representative confocal images of HEK293 cells expressing EGFP-VP40 with vehicle (DMSO) or fendiline treatment.HEK293 cells post-VP40 transfection were treated with either vehicle (DMSO) or fendiline (at varying concentrations for analysis at different time points 24 or 48 hours).Confocal imaging was performed (48 hours as in Fig. S5) and image analysis (plasma membrane localization pre-VLP formation) was performed by counting pre-VLPs at the plasma membrane per cell slice by moving the Z plane of the image of and down.The number of pre-VLPs were counted per imaging frame for an equal number of VP40 expressing cells over the course of three independent experiments and normalized to WT VP40 production with vehicle treatment.Scale bar = 10 mm.

Figure S5 .
Figure S5.Difference in VLP production between 2 µM and 10 µM applied at different time experimentally.Difference in VLP production between 2 µM and 10 µM fendiline application decreases with later application time.The difference in VLP measurement is not significant at 24 h when fendiline is applied at 0 or 12 hours, and at 48 h when fendiline is applied at 24 or 36 hours (TableS10).fendiline Error bars indicate the SEM.

Table S4 . Statistical analysis for impact of fendiline application time on VLP production (Simulation). Table S3. Efficiency of fendiline application at different time: Experiment.
*Two-tailed unpaired t test was performed.

Table S2 . Statistical analysis for parameter distribution in fendiline-induced VLP production simulation. Table S8. Fendiline concentration experimental data Table S6. Statistical analysis for impact of fendiline concentration on VLP production at different application time (Experiment).
Two-tailed unpaired t test was performed. *

Table S7 . PS inhibition experimental data Table S5. Statistical analysis for impact of fendiline application time on treatment efficiency per hour (Simulation). Table S9. Weight of cost calculation Table S10. Statistical analysis for membrane dimer concentration
*Paired one-way ANOVA multiple comparasion was performed.