VSV-EBOV Induces Temporal and Dose-Dependent Transcriptional Responses in Non-human Primates

Zaire Ebola virus (EBOV), the causative agent of Ebola virus disease (EVD), is a member of the Filoviridae family. EVD is characterized by innate and adaptive immune dysregulation that leads to excessive inflammation, coagulopathy, lymphopenia, and multi-organ failure. Recurrent outbreaks of EBOV emphasize the critical need for effective and deployable anti-EBOV vaccines. The FDA-approved VSV-EBOV vaccine protects non-human primates (NHPs) and humans from EBOV when given at a 10–20 million PFU dose. We recently demonstrated that a dose as small as 10 PFU protected NHPs from lethal EBOV infection. Furthermore, 1 PFU of VSV-EBOV protected 75% of vaccinated NHPs. In this study, we performed a comparative transcriptional analysis of the whole blood transcriptome in NHPs vaccinated with doses of VSV-EBOV associated with complete protection (10M PFU), protection with mild EVD (10 PFU), and break-through protection (1 PFU) before and after challenge with a lethal dose of EBOV Makona. Transcriptional findings demonstrated that, regardless of dose, vaccination significantly attenuated the upregulation of genes associated with fatal EVD. Genes involved in T- and B-cell activation were more highly expressed in groups receiving 10 or 10M PFU than in 1 PFU–vaccinated animals. Furthermore, the singular vaccinated (1 PFU) non-survivor exhibited a transcriptional signature distinct from both surviving vaccinated animals and controls that received an irrelevant vaccine. These findings provide additional insight into mechanisms of vaccine-mediated protection and informing public policy on vaccine distribution during outbreaks.

The 2013-2016 West Africa EBOV Makona epidemic and the recent EBOV outbreaks in the Democratic Republic of the Congo (DRC) and Guinea highlight the urgent need to understand mechanisms of anti-EBOV immunity and vaccine-mediated protection (13,16,17).
The dual function of VSV-EBOV as a prophylactic and rapidacting, therapeutic vaccine represents a critical public health strategy for outbreak control. Administering lower, yet equally effective, doses of VSV-EBOV can facilitate vaccine distribution during outbreaks and evade adverse side effects associated with high doses (33,43,45,46). We recently showed that NHPs vaccinated with as little as dose as 10 4 PFU were completely protected from EVD (19). Furthermore, animals receiving a dose as small as 10 PFU-∼1 millionth the FDA-approved dose-experienced mild EVD but were completely protected from fatal disease. Moreover, 75% of NHPs vaccinated with 1 PFU succumbed to infection. The molecular mechanisms surrounding this dose-dependent efficacy remain unknown, but critical for understanding VSV-EBOV-mediated protection and anti-EBOV immunity.
In this study, we performed a transcriptional analysis of historical whole blood samples collected from NHPs vaccinated with either the fully protective 10M PFU dose of VSV-EBOV or doses associated with full protection against fatal disease (10 PFU), or break-through protection (1 PFU) (19). Expression of genes associated with EVD (e.g., inflammation, apoptosis, myeloid cell mobilization) were downregulated in vaccinated NHPs in a dose-dependent manner. Canonical genes involved in B-and T-cell activation were most highly expressed in animals receiving 10 or 10M PFU of VSV-EBOV. The animal vaccinated with 1 PFU and succumbing to challenge exhibited transcriptional signatures intermediate of vaccinated counterpart and controls. Collectively, these findings suggest a critical role for dosage in VSV-EBOV-mediated immunity and protection during EBOV infection.

Study Design
Whole blood-derived RNA samples were leveraged from a previous study (19). All infectious work with EBOV was performed in the maximum containment laboratory at the Rocky Mountain Laboratories (RML), Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health. RML is an AAALAC-accredited institution. All procedures followed standard operating procedures approved by the Institutional Biosafety Committee. Animal work was performed in strict accordance with the recommendations described in the Guide for the Care and Use of Laboratory Animals of the National Institute of Health, the Office of Animal Welfare and the Animal Welfare Act, United States Department of Agriculture. The study was approved by the Institutional Animal Care and Use Committee (IACUC). Procedures were conducted in animals anesthetized by trained personnel under the supervision of veterinary staff. The humane endpoint criteria for euthanasia were specified and approved by the IACUC. All efforts were made to ameliorate animal welfare and minimize animal suffering in accordance with the Weather all report on the use of non-human primates in research (https://royalsociety. org/policy/publications/2006/weatherall-report/).
In brief, RNA was extracted from whole blood of animals vaccinated with either VSV-EBOV (1 PFU, 10 PFU, or 10M PFU) or VSV-MARV (10M PFU) before vaccination (d28), following vaccination (d14) and following challenge (d6, d28, d42) as previously described. Animals were challenged with a lethal dose (1,000 PFU) of EBOV-Makona in the caudal thighs. Animals were monitored daily for signs of disease and behavioral changes.

cDNA Library Construction
Quality and quantity of RNA were determined using an Agilent 2100 Bioanalyzer. cDNA libraries were constructed using the NEB Next Ultra II Direction RNA Library Prep Kit (Thermo Fisher). RNA was treated with RNase H and DNase I following depletion of ribosomal RNA (rRNA). Adapters were ligated to cDNA products and the subsequent ∼300-400 base pair (bp) amplicons were PCR amplified and selected by size exclusion. All cDNA libraries were assessed for quality and quantity prior to 100 bp single-end sequencing using the Illumina NovaSeq platform.

Bioinformatics Analysis
Preliminary data analysis was performed with RNA-Seq workflow module of systemPipeR, developed by Backman and Girke (47). RNA-Seq reads were demultiplexed, qualityfiltered, and trimmed using Trim Galore (average Phred score cut-off of 30, minimum length of 50 bp). FastQC was used to generate quality reports. Alignment was performed with hisat2 using reference genome Macaca fascicularis (Macaca_fascicularis.Macaca_fascicularis_5.0.dna.toplevel.fa) and annotation file Macaca_fascicularis.Macaca_ fascicularis_5.0.94.gtf for annotation. Raw expression values (gene-level read counts) were generated using the summarizeOverlaps function and normalized (read per kilobase of transcript per million mapped reads, rpkm) using the edgeR package. Statistical analysis with edgeR was used to determine differentially expressed genes (DEGs) meeting the following criteria: genes with average rpkm of ≥1, an FDR corrected p ≤ 0.05, and a |log 2 fold change| ≥1 compared with either untreated animals (d28) or control-vaccinated animals at 6 days postchallenge. Two-way forward regression analysis was performed with MaSigPro (48). DESeq2 analysis without biological replicates was used to compare the dose 1 non-survivor (n = 1) with the control animals (n = 2) at d6. The fold change (dose 1 non-survivor relative to average of controls) was calculated using the exactTest function. A |log 2 fold change| ≥3 cut-off was used to identify significant protein-coding genes (SGs) with notably higher or lower expression in the dose 1 non-survivor. Genes exhibiting dose-dependent regulation were determined using Short Time-Series Expression Miner (STEM) (49).
Functional enrichment of DEGs was performed using Metascape to identify relevant gene ontology (GO) biological process terms and immunological signatures (50). Heatmaps, bubbleplots, Venn diagrams, and beanplots were generated using R packages ggplot2 and VennDiagrams. GO network plots were generated in Cytoscape (version 3.5.1). Bar graphs and statistics were generated using GraphPad Prism software (version 8).
Digital cell quantification was performed using ImmQuant with the IRIS database, which includes transcriptional profiles from 22 immune cell types in normal human blood (51). This algorithm predicts changes in immune cell frequencies within heterogeneous populations, including whole blood and lymphoid tissue, using bulk transcriptional data. Human references were used given the close genetic homology between NHPs and humans as well the fact that NHP recapitulate clinical hallmarks of Ebola virus disease (52,53).
To understand the biological relevance of DEGs, we performed functional enrichment of DEGs induced at d6 in each group (Supplementary Figure 1C, Figure 1C). DEGs detected d6 enriched to GO terms reflecting hallmarks of EVD: antiviral defense (e.g., "response to virus"), inflammation (e.g., "inflammatory response"), cell death (e.g., "apoptotic signaling pathway"), and innate immunity (e.g., "myeloid leukocyte activation") ( Figure 1C). The number of DEGs and the significance of the enrichment were significantly larger in the control group compared with the vaccinated groups ( Figure 1C). Gene expression changes associated with inflammation, like "cellular response to IL-1" and "regulation of ERK/2 cascade" GO terms, were only observed in controls and dose 1. GO terms associated with antigen presentation (e.g., "autophagy, " "antigen processing and presentation;" MR1, PSMB8, TAP1) were unique to dose 10. DEGs enriching to GO terms associated with T-cell activation were only seen in the dose 10M group (Figure 1C).
We next examined the magnitude of expression for DEGs enriching to GO terms illustrating EVD (Figures 1D-G). DEGs enriching to GO term "response to virus" and shared by all groups included canonical components of type I/II interferon (IFN) signaling (e.g., IFNGR2, STAT1, STAT2), nucleic acid sensors (e.g., DDX58, DHX58), and interferon stimulated genes (ISGs) (e.g., EIF2AK2, RNASE6) ( Figure 1D). These DEGs were significantly more highly upregulated in controls compared with vaccinated animals. Additional DEGs detected only in controls that enriched to "response to virus" included genes that play roles in viral RNA editing (e.g., ADAR, APOBEC3G) and inflammation (e.g., FGR, RIPK2) ( Figure 1E). Similarly, DEGs that mapped to "myeloid leukocyte activation" GO term and detected in all groups were also more highly upregulated in the controls. Notable genes within this set included IL18RAP, S100A8, and S100A12, which are important for neutrophil activation and inflammation ( Figure 1F). Some DEGs enriching to "myeloid cell activation" were only detected in VSV-EBOV-vaccinated animals ( Figure 1G). These included CD59 and MMP8 that play a role in complement-mediated immunity (dose 1), SERPINB6 important for (dose 10), and BATF3 and JAK2 important for myeloid cell differentiation (dose 10M) (Figure 1G). The 20 upregulated DEGs detected in dose 10M and enriching to "T cell activation" primarily regulated T-cell activity (e.g., ARG2, IDO1, CD274, TNFAIP8L2).

Transcriptional Indicators Reflecting Hallmarks of EVD Are Attenuated in Protected Animals
To better understand how VSV-EBOV dose modulates responses to EBOV, we directly compared VSV-EBOV-vaccinated NHPs with control-vaccinated NHPs at peak infection (d6) (Figure 2A). We detected similar numbers of DEGs for dose 1 (n = 629) and dose 10 (n = 699) while a smaller number of DEGs (n = 499) were detected in dose 10M (Figure 2A). The majority of DEGs were downregulated and shared among all VSV-EBOV-vaccinated groups, indicating that these genes were upregulated in fatal disease (Figure 2B). DEGs from each group enriched to similar GO terms notably "myeloid leukocyte activation, " "neutrophil degranulation, " "response to virus, " and "leukocyte migration" (Figure 2C). DEGs detected only in the dose 1 group additionally mapped to cell death pathways ("necrotic cell death") and oxidative stress (e.g., "reactive oxygen species metabolic process") ( Figure 2C). DEGs detected only in higher vaccine doses groups (10 and 10M PFU) enriched to "response to IFN-γ".
Since whole blood samples were not available for flow cytometry, we used our transcriptomic data to perform digital cell quantification with the IRIS database to predict changes in immune cell frequencies (Supplementary Figure 1G) (51, 54). B-and T-cell lymphopenia was predicted for all groups at d6, including vaccinated animals. Reduced frequencies of memory B cells were more significant in the dose 1 group vaccinated nonsurvivor, and less pronounced in the dose 10 and 10M vaccine groups. While activated monocyte populations (monocyte d1/7) were more significantly increased in the control at d6, less dramatic increases were predicted for dose 10 and dose 10M. Dendritic cell and neutrophil populations were also predicted to increase at d6, but not significantly for neutrophils. Cell frequencies were predicted to return to baseline at d28 for all surviving animals.

Genes Involved in B-and T-cell Activation Are Upregulated in Dose-Dependent Manner in Vaccinated Animals at Peak Infection
We next identified clusters of genes that exhibited dosedependent regulation among vaccinated groups at d6 using STEM with dose as a variable instead of time (Figure 3) (49). We determined three clusters of genes ( Figure 3A). Expression of the genes in clusters 1 and 2 increased in a dose-dependent manner. In contrast, genes in cluster 3 were most highly expressed in dose 1-vaccinated animals. Genes in cluster 1 enriched to GO terms associated with adaptive immunity (e.g., "lymphocyte activation, " "T cell differentiation") ( Figure 3B). Genes in cluster 1 included canonical genes involved in T-cell activation (e.g., CD2, CD28, CD3D, CD8A, ICOS, IL7R) and B-cell-mediated immunity (e.g., LCK, IGHG2, SLAMF7), as well as indicators of cellular cytotoxicity (e.g., KLRD1, KLRK1) ( Figure 3C). Genes playing roles in cellular cytotoxicity such as PRF1 and CD69, and inflammation mediated by monocytes/macrophages like CCR5 and CD36, were part of cluster 2 ( Figure 3D). In contrast, genes in cluster 3 enriched to GO terms associated with innate immunity and inflammation (e.g., "myeloid leukocyte activation, " "coagulation") ( Figure 3B). Components of the complement pathway (e.g., C1QB), neutrophil-mediated immunity (e.g., CD99, S100A8), and acute inflammatory response (e.g., CD163, IL18R1, IL1R2) belonged to cluster 3 ( Figure 3E).

Dose 1 PFU VSV-EBOV-Vaccinated Non-survivor Exhibits a Distinct Transcriptional Signature From Survivors and Negative Controls
In the dose 1 group, one animal succumbed to infection while three survived. Therefore, we performed a two-way forward regression analysis to identify clusters of genes that were similarly and differently regulated following EBOV challenge between the non-survivors and survivors (Figures 4A-D). We identified two gene clusters (Figure 4A). Genes in cluster 1 were highly upregulated in the non-survivor at d6 compared with survivors. Functional enrichment of cluster 1 genes mapped to GO terms associated with antiviral defense, cellular stress, leukocyte activation, and intracellular vesicle mobilization (Figure 4B). Of the genes enriching to "response to virus" and "regulation of defense response, " many were ISGs and viral restriction factors such as HERC5, MX1, and SAMHD1 (Figures 4B,C). Genes that enriched to GO term "myeloid leukocyte activation" played a role in neutrophil-mediated immunity such as activation marker CD177 and NETosis factor S100A12. This term also encompassed mediators of the complement pathway (e.g., CD59 and CD93) and cell adhesion (e.g., CAPZA1, OLFM4) (Figures 4B,C). The 16 genes in cluster 2 were downregulated by the non-survivor but upregulated in the survivors from d6 to d28 (Figures 4A,C). These few genes included CDC34 and PSMD9 involved in protein degradation, HYAL3 in maintenance of epithelial integrity, and ICAM4 in cell migration (Figures 4A,C).

DISCUSSION
VSV-EBOV (rVSV-ZEBOV, Ervebo by Merck) is currently the only FDA-approved vaccine to protect against EVD. Human clinical trials initiated during the 2013-2016 West Africa epidemic and the 2018-2020 outbreak in the DRC demonstrated robust safety and efficacy, although vaccination was frequently associated with transient muscle pain, arthralgia, fever, and fatigue (43,44). This is most likely due to high dose of the vaccine used in these vaccination campaigns. Recurrent EBOV outbreaks in endemic regions highlight the need to improve the availability of vaccines and resolve undesirable side effects in target populations without the loss of protective efficacy. Our previous study determined the dose-dependent efficacy of VSV-EBOV in cynomolgus macaques vaccinated 28 days before lethal EBOV challenge (19). We showed that complete protection is achieved with a dose of 10 6 PFU while doses ranging from 10  PFU to 10 4 PFU are sufficient to prevent severe EVD, but not viremia. On the other hand, 25% of animals receiving 1 PFU of VSV-EBOV succumbed to challenge and exhibited hallmarks of EVD. Here, we investigated the molecular mechanisms surrounding dose-dependent efficacy using transcriptomics with historical samples. Interestingly, we observed functionally similar transcriptional response in vaccinated animals and control-vaccinated animals at d6. The transcriptional changes contained hallmarks of antiviral responses including an induction of ISGs, myeloid cell activation factors, and apoptosis despite the absence of viremia in 10M PFU-vaccinated animals, suggesting that small amounts of antigen can trigger a robust response (4,12,55). However, the magnitude of induction was significantly higher in animals that succumbed to disease, as was previously reported for both NHP and humans, indicative of a dysregulated inflammatory response (4-6, 24, 56-60). Indeed, we previously reported lower plasma levels of IFNγ, IL-6, and TNFα, and stable levels of platelets and AST in vaccinated animals throughout challenge as compared with controls (19). Analysis of dose-dependent expression of genes also revealed a cluster of pro-inflammatory genes highly expressed in only dose 1-vaccinated animals.
We detected induction of genes involved in humoral immunity in dose 1-vaccinated animals during challenge, suggesting that infection may serve as an antigen "boost" since these animals initially presented with low antibodies. A similar transcriptional response including B-cell activation genes was previously seen in animals vaccinated with VSV-EBOV only 3 days before EBOV challenge (27,61). The significant upregulation of genes important for regulatory T-cell function in dose 10M-vaccinated animals compared with controls suggests a controlled response to infection not seen in fatal NHP and human cases (46,59,(62)(63)(64). Aberrant T-cell responses are associated with fatal EVD and immunological impairment during other viral diseases such as EBV reactivation in humans (59,65,66). Furthermore, we identified a cluster of genes that were significantly upregulated in dose 10-and 10M-vaccinated animals compared with dose 1 animals that included canonical members of T-and B-cell signaling pathways like CD28 and LCK, as well as NK-and T-cell-mediated cytotoxicity. The role of humoral immunity is well-established for VSV-EBOV while the roles of NK-and T-cell-mediated immunity are less clear (12,28,(67)(68)(69). Nonetheless, these findings indicate that VSV-EBOV dosage likely plays a key role in shaping nature and durability of long-term immunity, although future studies are needed to test this hypothesis.
Digital cell quantification demonstrated significant decreases in T-and B-cell subsets for all groups. This lymphopenia could indicate two different processes in controls vs. vaccinated animals. Specifically, the reduced frequency of B and CD8 T cell could indicate (1) lymphocyte apoptosis in controls and (2) recruitment into tissues of vaccinated animals. This hypothesis is supported by the more significant upregulation of apoptotic genes in the control group. Monocytosis was predicted to be significant in the controls and animals that received lower vaccine doses. The downregulation of canonical neutrophil chemoattractant CXCL8 in all vaccinated animals further suggests a lack of EVD-associated neutrophilia in these animals even though we did detect a greater expression of inflammatory and neutrophil-associated genes in dose 1, but not doses 10 and 10M, animals. Since lymphoid tissues serve as key sites of EBOV pathogenesis, future studies should address the impact of vaccine dose within these sites (1,58).
Given that a single animal in the dose 1-vaccinated group succumbed to EBOV challenge, we had the opportunity to explore transcriptional signatures differentiating nonsurvivors from survivors within the same vaccine group as well as non-survivors from controls. Several hundred genes that play a critical role in inflammation and apoptosis were only upregulated by the non-survivor. These transcriptional changes may provide critical diagnostic markers. However, induction of canonical genes influencing myeloid and lymphocyte activation pathways/differentiation (e.g., IFNB) was less significant in the non-survivor compared with control-vaccinated animals. Therefore, vaccinated non-survivors are associated with intermediate transcriptional responses of protected NHPs and unvaccinated NHPs.
In summary, we determined a downregulation of EVD-related genes in vaccinated survivors and dose-related expression of key components of T-and B-cell-mediated immune responses. This supports the proposal for a reduction in vaccine dose to prevent adverse side effects and facilitate greater vaccine distribution in endemic regions, which represents a critical strategy for controlling ongoing outbreaks. However, there are several caveats to this study. First, the transcriptional findings reported here are restricted to whole blood and not cell specific. Second, we were unable to confirm the up-and downregulation of genes at the protein level, as well as changes in immune cell frequencies with flow cytometry. Third, the impact of dosage on durability of anti-EBOV immunity remains to be investigated. Future studies will focus on determining (1) the contributions of both innate and adaptive cells to dose-dependent immunity, (2) the durability of dose-dependent immunity, and (3) the therapeutic efficacy of lower doses of VSV-EBOV.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found at: https://www.ncbi.nlm.nih. gov/, PRJNA747033.

ETHICS STATEMENT
The animal study was reviewed and approved by Rocky Mountain Laboratories (RML), Division of Intramural Research, National Institute of Allergy and Infectious Diseases, National Institutes of Health Institutional Animal Care and Use Committee.