Ebola Virus Requires Phosphatidylserine Scrambling Activity for Efficient Budding and Optimal Infectivity

Abstract

Ebola virus (EBOV) attaches to target cells using two categories of cell surface receptors: C-type lectins and phosphatidylserine (PS) receptors. PS receptors typically bind to apoptotic cell membrane PS and orchestrate the uptake and clearance of apoptotic debris. Many enveloped viruses also contain exposed PS and can therefore exploit these receptors for cell entry. Viral infection can induce PS externalization in host cells, resulting in increased outer PS levels on budding virions. Scramblase enzymes carry out cellular PS externalization; thus, we targeted these proteins in order to manipulate viral envelope PS levels. We investigated two scramblases previously identified to be involved in EBOV PS levels, transmembrane protein 16F and Xk-related protein 8 (XKR8), as possible mediators of cellular and viral envelope surface PS levels during the replication of recombinant vesicular stomatitis virus containing its native glycoprotein (rVSV/G) or the EBOV glycoprotein (rVSV/EBOV-GP). We found that rVSV/G and rVSV/EBOV-GP virions produced in XKR8 knockout cells contain decreased levels of PS on their surfaces, and the PS-deficient rVSV/EBOV-GP virions are 70% less efficient at infecting cells through PS receptors. We also observed reduced rVSV and EBOV virus-like particle (VLP) budding in ΔXKR8 cells. Deletion of XKR8 in HAP1 cells reduced rVSV/G and rVSV/EBOV-GP budding by 60 and 65%, respectively, and reduced Ebola VLP budding more than 60%. We further demonstrated that caspase cleavage of XKR8 is required to promote budding. This suggests that XKR8, in addition to mediating virion PS levels, may also be critical for enveloped virus budding at the plasma membrane. IMPORTANCE Within the last decade, countries in western and central Africa have experienced the most widespread and deadly Ebola outbreaks since Ebola virus was identified in 1976. While outbreaks are primarily attributed to zoonotic transfer events, new evidence is emerging outbreaks may be caused by a combination of spillover events and viral latency or persistence in survivors. The possibility that Ebola virus can remain dormant and then reemerge in survivors highlights the critical need to prevent the virus from entering and establishing infection in human cells. Thus far, host cell scramblases TMEM16F and XKR8 have been implicated in Ebola envelope surface phosphatidylserine (PS) and cell entry using PS receptors. We assessed the contributions of these proteins using CRISPR knockout cells and two EBOV models: rVSV/EBOV-GP and EBOV VLPs. We observed that XKR8 is required for optimal EBOV envelope PS levels and infectivity and particle budding across all viral models.

Keywords: Ebola; XKR8; budding; phosphatidylserine.

FIG 1

PS scrambling abilities of WT and knockout (KO) HAP1 cells. HAP1, ΔTMEM16F, and ΔXKR8 cells were infected with rVSV/G for 16 h or treated with a calcium ionophore A23187 (5 μM) for 5 min, double stained with AnV-PacBlue and PI, and analyzed using a BD LSRII flow cytometer. Anv-PacBlue fluorescence of live, single, nonnecrotic cells is shown as (A) histograms and (B) average mean fluorescence intensities (MFIs). (C) Immunoblot probing for β-actin and TMEM16F confirming the absence of TMEM16F in ΔTMEM16F cells. The values shown are averages from at least three independent experiments ± SEM. *, P < 0.05, **, P < 0.01, and ***, P < 0.001, by Student's t test.

FIG 2

rVSV/EBOV-GP replication and cell-cell spread are delayed in cells lacking XKR8. HAP1, ΔTMEM16F, and ΔXKR8 cells were infected with rVSV/EBOV-GP (A), rVSV/G (B), or MeV (C). Cell supernatants (A and B) or cell-associated virus (C) was collected at the indicated time points and titrated on Vero cells. The values shown are the average log₁₀ TCID₅₀/ml from at least three independent experiments ± SEM. To observe cell-cell viral spread, cells were infected with rVSV/EBOV-GP (D) or rVSV/G (E), and viral spread was monitored by GFP fluorescence over time. Three representative images were taken at the indicated time points (magnification, ×20).

FIG 3

XKR8 is not required for rVSV/EBOV-GP entry or replication. (A) HAP1 and ΔXKR8 cells were infected with rVSV/G or rVSV/EBOV-GP. After 2 h, early viral entry was determined by quantifying the number of viral genomes within the cell lysate using qRT-PCR. (B) rVSV/G and rVSV/EBOV-GP entry was assessed by counting the number of cells producing the virally encoded GFP reporter by flow cytometry. Cells were infected with rVSV/G or rVSV/EBOV-GP for 12 h, after which cells were harvested, fixed, and analyzed for GFP fluorescence. To assess viral replication, cells were infected with rVSV/G or rVSV/EBOV-GP, and viral RNA was quantified in the cell lysates 12 h (rVSV/G) or 24 h (rVSV/EBOV-GP) postinfection. The values shown are averages from at least three independent experiments ± SEM. *, P < 0.05 by Student's t test.

FIG 4

XKR8 is required for efficient rVSV budding. (A) Cells were infected with rVSV-MnLuc/G or rVSV-MnLuc/EBOV-GP for 12 h, after which cell lysate and supernatant nano-luciferase (nLuc) activity was measured. Budding efficiency was calculated by dividing supernatant nLuc activity by lysate nLuc activity, and values were normalized to HAP1 budding efficiency. The values shown are averages from five independent experiments ± SEM. *, P < 0.05, and **, P < 0.01, by Student's t test. (B) Immunoblot probing for VSV-M and VSV-G or EBOV-GP in rVSV/G- or rVSV/EBOV-GP-infected HAP1 and ΔXKR8 cell lysates and supernatants after one round of replication. GAPDH levels are shown as a cell lysate loading control. (C to F) Transmission electron microscopy analysis of rVSV/EBOV-GP particles budding from the surfaces of HAP1 cells (C) or ΔXKR8 cells (E) and rVSV/G particles budding from the surfaces of HAP1 cells (D) or ΔXKR8 cells (F). Images in panels C to F were captured at the following magnifications: 6,000× (a), 5,000× (b), 12,000× (c), 15,000× (d), 6,000× (e), and 15,000× (f).

FIG 5

Knocking out XKR8 reduces rVSV/EBOV-GP infectivity and PS levels. (A) rVSV/G and rVSV/EBOV-GP were propagated in HAP1 and ΔXKR8 cells for 24 h. Supernatants were collected, concentrated, and titrated via plaque assay, and viral genomes were quantified by qRT-PCR. (B) Viral genome/PFU ratios were determined by calculating the ratio of viral genome copy numbers to infectious particle numbers (PFU/ml) for each sample. (C to E) Equal amounts of rVSV/G and rVSV/EBOV-GP particles produced in HAP1 and ΔXKR8 cells were bound to aldehyde-sulfate latex beads and immunostained for VSV-G (C) or EBOV-GP (D). E) Viral particles were stained for surface PS levels with AnV-PacBlue, and MFIs were quantified. (F) Equivalent genome copies of rVSV/EBOV-GP particles produced by HAP1 and ΔXKR8 cells were TCA precipitated and analyzed for VSV-M and EBOV-GP protein content by immunoblot. Ratios of EBOV-GP to VSV-M were quantified using densitometry analysis. The values shown are averages from at least three independent experiments ± SEM. *, P < 0.05, and **, P < 0.01, by Student's t test.

FIG 6

rVSV/EBOV-GP infectivity is restored following one passage in Vero cells. HAP1-made or ΔXKR8-made rVSV/G or rVSV/EBOV-GP virus examined in Fig. 5A and B was passaged once in Vero cells. (A) After 24 h, supernatants were collected and titrated via plaque assay. Viral genomes were quantified by qRT-PCR. (B) Viral genome/PFU ratios were determined by calculating the ratio of viral genome copy numbers to infectious particle numbers (PFU/ml) for each sample. The values shown are averages from at least three independent experiments ± SEM.

FIG 7

XKR8 enhances VLP budding. (A) HAP1 and ΔXKR8 cells were transfected with plasmids encoding EBOV VLP components NP, GP, and VP40 and plasmids encoding either tetherin or empty vector. After 48 h, lysates and pelleted supernatants were subjected to immunoblot analysis for vinculin, VP40, and GP. Asterisks (*) indicate nonspecific protein bands. (B) Supernatants from HAP1 and ΔXKR8 cells producing nLuc-VLPs were collected and pelleted 24 h following transfection. nLuc-VLPs were normalized using luciferase levels, denatured, and subjected to immunoblot analysis for VP40 and GP content. EBOV-GP/VP40 ratios were quantified using densitometry analysis. (C and D) HAP1 and ΔXKR8 cells were transfected with plasmids encoding NP, GP, and VP40-nLuc, and either tetherin (C), XKR8_FLAG (D), or empty vector. After 24 h, VLP budding efficiencies were quantified by measuring luciferase levels in lysates and pelleted supernatants. (E) HAP1 and ΔXKR8 cells were transfected with plasmids encoding VP40-nLuc and empty vector, GP, NP, GP plus NP, or SARS-CoV-2 S_Met1Δ21. nLuc-VLP budding efficiencies were determined by measuring luciferase levels in cell lysates and pelleted supernatants 24 h following transfection. The values shown are averages from at least three independent experiments ± SEM. Additional statistics are summarized in Table 1.

FIG 8

Caspase-cleaved XKR8 is required for VLP budding enhancement. (A) nLuc-VLPs were produced in HAP1 cells treated with DMSO or 20 μM pan-caspase inhibitor Z-VAD-FMK. nLuc-VLP budding efficiencies were determined by measuring luciferase levels in cell lysates and pelleted supernatants 24 h following transfection. (B) ΔXKR8 cells were transfected with plasmids encoding EBOV-NP, -GP, and -VP40-nLuc, and plasmids encoding either XKR8_FLAG, caspase-resistant XKR8(2DA)_FLAG, or empty vector. After 24 h, cell lysates and pelleted supernatants were subjected to immunoblot analysis to detect VP40-nLuc and full-length (∼48 kDa) or cleaved (∼42 kDa) XKR8_FLAG. Vinculin levels are shown as a cell lysate loading control. (C) ΔXKR8 cells were transfected with plasmids encoding VP40-nLuc, GP, NP, and either XKR8_FLAG, XKR8(2DA)_FLAG, or empty vector for 24 h. VLP budding efficiency was quantified by measuring luciferase levels in lysates and pelleted supernatants. The values shown are averages from at least three independent experiments ± SEM. *, P < 0.05, and **, P < 0.01, by Student's t test.

FIG 9

Pharmacological activation of PS scrambling enhances VLP budding. HAP1 and ΔXKR8 cells were transfected with plasmids encoding VP40-nLuc, GP, and NP. Twelve hours following transfection, cells were treated with 0.5 μM calcium ionophore A23187 or DMSO. After an additional 36 h, VLP budding efficiencies were quantified by measuring luciferase levels in lysates and pelleted supernatants. The values shown are averages from at least three independent experiments ± SEM. * P < 0.05, and **, P < 0.01, by Student's t test.