Jamaican fruit bats' competence for Ebola but not Marburg virus is driven by intrinsic differences

Abstract

Ebola virus (EBOV) and Marburg virus (MARV) are zoonotic filoviruses that cause hemorrhagic fever in humans. Correlative data implicate bats as natural EBOV hosts, but neither a full-length genome nor an EBOV isolate has been found in any bats sampled. Here, we model filovirus infection in the Jamaican fruit bat (JFB), Artibeus jamaicensis, by inoculation with either EBOV or MARV through a combination of oral, intranasal, and subcutaneous routes. Infection with EBOV results in systemic virus replication and oral shedding of infectious virus. MARV replication is transient and does not shed. In vitro, JFB cells replicate EBOV more efficiently than MARV, and MARV infection induces innate antiviral responses that EBOV efficiently suppresses. Experiments using VSV pseudoparticles or replicating VSV expressing the EBOV or MARV glycoprotein demonstrate an advantage for EBOV entry and replication early, respectively, in JFB cells. Overall, this study describes filovirus species-specific phenotypes for both JFB and their cells.

Fig. 1. Infection of JFB with EBOV or MARV does not cause overt clinical disease.

A Depiction of challenge study of Jamaican fruit bats with EBOV-Mayinga (magenta) or MARV-Ozolin (purple) via subcutaneous (SC), intranasal (IN), and oral routes. Image created in BioRender [https://BioRender.com/a66b266]. B Change in body temperature of bats (n = 4) infected with EBOV followed through day 28. Two-way ANOVA with Dunnett’s multiple test comparison. Data plotted as mean ± S.D. P values for comparisons to baseline weight are indicated. *** <0.001, ** <0.01, * <0.05. C Percent body weight of bats (n = 4) infected with EBOV followed through day 28. Two-way ANOVA with Dunnett’s multiple test comparison. D Change in body temperature of bats (n = 4) infected with MARV followed through day 28. Two-way ANOVA with Dunnett’s multiple test comparison. Data plotted as mean ± S.D. E Percent body weight of bats (n = 4) infected with MARV followed through day 28. Two-way ANOVA with Dunnett’s multiple test comparison. F Number of circulating white blood cells (WBC), lymphocytes, neutrophils, and monocytes in EBOV-infected bats (n = 4) at baseline and at necropsies. A mixed-effects model with matching and Dunnett’s multiple comparison test was applied to evaluate change in each cell population at necropsy compared to the matched baseline value. G Number of circulating WBC, lymphocytes, neutrophils, and monocytes in MARV challenged bats (n = 4) at baseline and at necropsies. A mixed-effects model with matching and Dunnett’s multiple comparison test was applied to evaluate change in each cell population at necropsies compared to the matched baseline value.

Fig. 2. Infected JFBs shed infectious EBOV orally.

RT-qPCR for EBOV (A, B) or MARV (C, D) RNA in the oral (A, C) or rectal (B, D) swabs. Limit of detection 1.63 log₁₀ copies/mL. E Infectious titer of EBOV in the oral swabs. Limit of detection 0.5 log₁₀ TCID₅₀/mL. F Infectious titer of EBOV in the rectal swabs in span of days where viral RNA was detected in swabs. Limit of detection 0.5 log₁₀ TCID₅₀/mL. G Infectious titer of MARV in the oral swabs at baseline and on day 6 post-challenge. Limit of detection 0.5 log₁₀ TCID₅₀/mL.

Fig. 3. EBOV but not MARV infection is disseminated in JFBs.

A RT-qPCR for EBOV in tissue samples collected at 3-, 7-, and 28 days post-infection (DPI) (n = 4). Limit of detection 2.2 log₁₀ copies/g. B Infectious titer of EBOV in tissues collected at D3, 7, and 28 (n = 4). Limit of detection 3.0 log₁₀ copies/g. C RT-qPCR for MARV in tissue samples collected at necropsy day (DPI) 3, 7, and 28 (n = 4). Limit of detection 2.2 log₁₀ copies/g. D Infectious titer of MARV in tissues collected at D3, 7, and 28 (n = 4). Limit of detection 3.0 log₁₀ copies/g.RT-qPCR. EBOV (E) or MARV (F) in serum samples collected at baseline, 3-, 7-, 14-, 21-, and 28 days post-infection (DPI). Limit of detection 1.67 log₁₀ copies/mL.

Fig. 4. EBOV infection induces robust innate antiviral and adaptive immune responses.

RT-qPCR of interferon stimulated gene (ISG) (A, C, E) or pro-inflammatory cytokine (PIC) (B, D, F) mRNA in the skin at inoculation site (A, B), liver (C, D), or spleen (E, F) collected from healthy control bats (n = 4) and at day (D) 3 or 7 necropsy of EBOV or MARV-infected bats (n = 4). A–F ΔC_T values were normalized to the average ΔC_T value of the healthy control bats to calculate ΔΔC_T. Fold change of each gene at necropsy day was compared to the healthy controls using a two-way ANOVA with Dunnett’s multiple test comparison. G Serum binding antibodies and (H) neutralizing titers were determined via ELISA assay with a standard curve of EBOV glycoprotein (GP) or neutralization assay, respectively. Neutralizing titer is presented as the lowest dilution that protected 50% of wells in a TCID₅₀ assay. I Serum binding antibodies and (J) neutralizing titers were determined via ELISA assay with a standard curve of MARV GP or neutralization assay, respectively. Neutralizing titer is presented as the lowest dilution that protected 50% of wells in a TCID₅₀ assay. K Cryopreserved splenocytes from naïve controls (n = 8) and D28 EBOV (n = 4) infected bats were analyzed for intracellular expression of CD3ε and CD79a to identify the proportion of T cells and B cells from a live lymphocytes gate. L CD79a+ B cells from cryopreserved splenocytes, naïve controls (n = 8), and D28 EBOV (n = 4) infected bats, were analyzed for interaction with Alexa-fluor 598 conjugated recombinant EBOV-GP receptor binding domain. The percentage of live CD79a⁺ B cells with a positive EBOV-GP staining signal after subtracting the fluorescence minus one control signal. Data plotted as mean ± S.D. A two-sided Mann-Whitney statistical test was used to assess statistical significance. A–F, K Box defines the upper (75th percentile) and lower (25th percentile) quartiles with whiskers extending from minimum to maximum with all values shown, and the line as the median. P values adjusted for multiple comparisons <0.05 for comparisons versus healthy controls are indicated. **** <0.0001, *** <0.001, ** <0.01, * <0.05.

Fig. 5. EBOV enters and replicates in JFB cells more efficiently than MARV.

A Infectious titers of three EBOV strains (Kikwit, Makona C05, and Mayinga) and MARV strains (Angola, Musoke, and Ozolin) stocks (n = 3) measured on Vero E6 and JFB uropatagium-derived fibroblasts (AjUFi_RML6). Three independent stock vials were titrated on different days. The titer of each virus was compared on the two cell lines using a two-way ANOVA with Dunnett’s multiple test comparison. B Replication incompetent vesicular stomatitis virus (VSV) with glycoprotein gene replaced with green fluorescent protein pseudotyped with either EBOV or MARV glycoprotein (GP) were titrated on Vero E6, Jamaican fruit bat kidney (AjKi_RML1 and AjKi_RML2), or Egyptian rousette bat kidney (RoNi) cells. To determine the relative entry ratio, we normalized the ratio of EBOV/MARV infectious units for each cell line and divided by the ratio on Vero E6 cells. Three independent experiments were performed with four technical replicates each experiment. One-way ANOVA with Dunnett’s multiple test comparison. C Infectious titers of recombinant VSV expressing GFP (rVSVwt-GFP) (n = 2) or GFP plus EBOV-Mayinga (n = 3) or MARV-Ozolin (n = 3) GP were measured on Vero E6 and AjKi_RML2. Two (VSVwt) or three (EBOV and MARV GP) independent stock vials were titrated on two different passages of cells. The titer of each virus was compared on the two cell lines using a two-way ANOVA with Dunnett’s multiple test comparison. D Vero E6 and AjKi_RML2 cells were infected at multiplicity of infection (MOI) 0.005 for 1 h and supernatants were collected at 16-, 48-, and 72 h post-infection. The ratio of infectious titers for rVSV-, EBOV-GFP, or rVSV-MARV-GFP was calculated for both cell lines and normalized to the ratio on Vero E6 cells (Replication kinetics shown in Supplementary Fig. 4). Two independent experiments were performed with three technical replicates each experiment. The infection ratio at each time point was compared using a two-way ANOVA with Dunnett’s multiple test comparison. A–D Data plotted as mean ± S.D. P values for comparisons to healthy controls are indicated. **** <0.0001, ** <0.01, * <0.05.

Fig. 6. EBOV antagonizes JFB’s type I interferon pathway more efficiently than MARV.

A Infectious titers of EBOV-Mayinga and MARV-Ozolin on Jamaican fruit bat kidney cells (AjKi_RML2) infected with multiplicity of infection (MOI) 0.1. Data presented as log₁₀ transformed values. Data are from two independent experiments with three technical replicates each. Two-way ANOVA with Sidak’s multiple test comparison to evaluate the difference between EBOV and MARV titer at each time point. B Immunoblot of AjKi_RML2 cells infected with EBOV or MARV at MOI 0.1. The presented panel is representative of three independent immunoblots. RT-qPCR of IFNb1 (C), interferon stimulated gene (D), or pro-inflammatory cytokine (E) mRNA in AjKi_RML2 cells infected with EBOV or MARV at MOI 0.1. F Infectious titers of three EBOV and MARV strains on AjKi_RML2 cells infected with MOI 0.1. Data presented as log₁₀ transformed values. Data are from one experiment conducted in triplicate. RT-qPCR of Infb1 (G), interferon stimulated gene (H), or pro-inflammatory cytokine (I) mRNA in AjKi_RML2 cells infected with three different EBOV or MARV strains at MOI 0.1. J Infectious titers of three EBOV and MARV strains on Jamaican fruit bat immortalized kidney (AJi), primary kidney (AjKi_RML1), primary lung (AjLu_RML3) infected with MOI 0.1. Data presented as log₁₀ transformed values. Data are from one experiment conducted in triplicate. RT-qPCR of Ifnb1 (K), interferon stimulated gene (L), or Tnfa (M) mRNA in Jamaican fruit bat cells infected with three EBOV Mayinga or MARV Ozolin strains at MOI 0.1. C–E; G–I; K–M ΔC_T values were normalized to the average ΔC_T value of mock infected cells to calculate ΔΔC_T. Data from three biological replicates. Fold change of each gene at each time point for EBOV and MARV-infected cells was compared to mock using a two-way ANOVA with Dunnett’s multiple test comparison. A, C–M Data plotted as mean ± S.D. A, C–E, G–I, K, L P values for comparisons to mock cells are indicated. **** <0.0001, *** <0.001, ** <0.01, * <0.05.

Fig. 7. Inability of MARV to antagonize IFN-I signaling partially contributes to attenuated replication.

A Schematic mechanisms of EBOV VP24 and MARV VP40 antagonism of the type-I interferon signaling pathway. Image created in BioRender [https://BioRender.com/b428467]. B Immunoblot of cytoplasmic and nuclear fractions of Jamaican fruit bat kidney cells, AjKi_RML2, infected with MOI 1.5 of EBOV-Mayinga or MARV-Ozolin for 24 h prior to the addition of recombinant Chiroptera IFN-β. The presented panel is representative of three independent immunoblots. C Quantification of phosphorylated STAT1 in the cytoplasmic and nuclear fractions. Three independent experiments were performed. Percentage of pSTAT1 in each of the fractions for EBOV and MARV-infected cells was compared mock using a two-way ANOVA with Dunnett’s multiple test comparison. Data plotted as mean ± S.D. P values for comparisons to mock are indicated. **** < 0.001, * < 0.05. D AjKi_RML2 cells were infected with MOI 0.1 EBOV-Mayinga or MARV-Ozolin for with vehicle control DMSO or itacitinib immediately after infection. The experiment was performed in triplicate. The effect of itacitinb on EBOV and MARV replication was determined using a one-way ANOVA with Dunnett’s multiple test comparison. Data plotted as mean ± S.D. P values for comparisons to mock are indicated. * < 0.05. E Immunoblot showing the effect of itacitinib treatment on the activation of the immune response and VP40 expression. F Quantification of VP40 for the immunoblot presented in (D). The effect of itacitinb on EBOV and MARV VP40 expression was determined using a one-way ANOVA with Dunnett’s multiple test comparison. The data is from three independent western blots. Data plotted as mean ± S.D. G Infectious titers of EBOV-Mayinga or MARV-Ozolin at 48 h post-infection. AjKi_RML2 cells were treated with recombinant Chiroptera IFN-β in 10-fold serial dilution for 24 h prior to infection with MOI 0.1. Data presented as log₁₀ transformed values. The experiment was performed in triplicate. To test a change in viral titer at each dose compared to mock treated cells, we performed a two-way ANOVA with Dunnett’s multiple test comparison. Data plotted as mean ± S.D. P values for comparisons to mock are indicated. **** <0.001, * <0.05.