Filovirus infection induces an anti-inflammatory state in Rousettus bats

The filoviruses Ebola (EBOV) and Marburg (MARV) cause severe disease in humans. In contrast, the Egyptian rousette bat (Rousettus aegyptiacus), a natural reservoir of MARV, exhibits a subclinical phenotype with limited MARV replication and nearly undetectable EBOV replication. Rousettus cell lines support replication of filoviruses, however. To understand the bat-filovirus interaction, transcriptomes of tissues from EBOVand MARV-infected R. aegyptiacus bats were analyzed. While viral transcripts were only detected in liver, a systemic response was observed involving other tissues as well. By focusing on evolutionarily divergent (from human homologues) protein-coding genes, we identified novel transcriptional pathways that suggest infected bats exhibit impaired coagulation, vasodilation, aberrant iron regulation, and impaired complement system leading to muted antibody responses. Furthermore, a robust T-cell response and an antiinflammatory state driven by M2 macrophages were identified. These processes likely control infection and limit pathology. All data can be freely explored and downloaded through our tools (http://katahdin.girihlet.com/shiny/bat/). bat-filovirus-jayaprakash


Introduction
Ebola virus (EBOV) and Marburg virus (MARV) are members of the family Filoviridae, which is comprised of filamentous, enveloped viruses with non-segmented, negative-sense RNA genomes 1 . EBOV and MARV cause outbreaks of severe, often fatal, disease in humans 2 . The second largest filovirus outbreak, caused by EBOV, is ongoing in the Democratic Republic of Congo and has resulted in 3,444 infections and 2,264 deaths as of March 2020 4 . The largest human outbreak of MARV occurred in Angola during 2004-2005, causing a reported 252 infections and 227 deaths 5 . Despite the aggressive use of a recently approved vaccine, control of the ongoing Congolese outbreak has been difficult, demonstrating the need for continued exploration of the pathobiology of these high-impact viruses.
There is substantial evidence that bats serve as hosts for filoviruses. MARV has been isolated from Egyptian rousette bats (Rousettus aegyptiacus) from sites in Uganda and Sierra Leone 6-8 . Ecological and experimental studies have demonstrated that these bats serve as a natural reservoir for MARV 7,9 . Circumstantial evidence suggests an association of bats with EBOV outbreaks 10,11 . Surveillance studies identified the presence of EBOV antibodies and RNA in several species of bats, implicating them as potential reservoirs of EBOV and other filoviruses 12 . However, infectious EBOV has never been isolated from a bat 12 . Additional evidence of association of filoviruses with bats has been obtained via detection of diverse filoviruses in bat tissues. These include Bombali virus (a novel species in the genus Ebolavirus identified in Chaerephon pumilus and Mops condylurus bats), Lloviu virus (the sole member of the genus Cuevavirus identified via RNA detected in Miniopterus schreibersii bats in Spain and Hungary), and Mengla virus (the sole member of the proposed genus Dianlovirus detected in the liver of a bat from the genus Rousettus) [13][14][15][16][17] .
Experimental infections with MARV in Egyptian rousette bats have demonstrated virus replication in various tissues, but minimal clinical signs of disease. The virus is shed in saliva, urine, and feces [18][19][20][21][22] . Co-housed bats can transmit MARV from one individual to another 9 . A serial sacrifice study following subcutaneous inoculation of MARV demonstrated mild pathology as evidenced by transient elevations of alanine aminotransferase and, lymphocyte/monocyte counts as well as by modest levels of inflammatory infiltrates in livers 20,22 . Animals were able to clear MARV, and develop adaptive immune responses, including MARVspecific neutralizing IgG 23 . Serological data suggests the possibility that EBOV may infect bats of the Rousettus genus in nature [24][25][26] although it is unknown whether it was a systemic or abortive peripheral infection (often referred to as "exposure") with bats becoming seropositive as a result. EBOV can replicate in R. aeqyptiacus cell lines 27 . However, experimental inoculation of EBOV in these bats has not previously been associated with disease and has not produced definitive evidence of significant virus replication 27 .
Hypothesizing that this difference arises from altered function of evolutionarily divergent genes, we identified differentially expressed genes in R. aegyptiacus bats that also had significant differences in amino acid sequence relative to their human homologues. The pathways impacted by these genes suggest that they may be involved in the remarkable ability of bats to avoid clinical illness during filovirus replication. Our transcriptomics data suggest that unlike humans, infected bats 1) are in a state of impaired coagulation and increased vasodilation (which may have the effect of lowering blood pressure), 2) exhibit anti-inflammatory signatures including an early transition to an M2 macrophage phenotype and tissue regeneration in the liver, 3) exhibit downregulation of critical components of the complement system that facilitate antibody activity suggesting a muted antibody response (to MARV 7 ), and 4) appear to mount a robust T cell response, which is a component of successful viral clearance in humans. This is the first comprehensive in vivo study of the transcriptomic changes induced by filovirus infection in bats. Our results highlight key parts of the systemic response that facilitate the ability of bats to survive filovirus infections and suggest potential host-targeted therapeutic strategies with utility in human infection.

Experimental methods
Viruses. Recombinant wild-type EBOV, strain Mayinga, was recovered from the full-length clone and support plasmids in HEK 293T cells and passaged twice in Vero E6 cells for amplification, as described previously 29 . Recombinant wild-type MARV, strain Uganda, was recovered similarly in BHK-21 cells as described previously 30 and was also passaged twice in Vero E6 cells for amplification.
Bat experimental protocol. Adult Egyptian rousettes were obtained from a commercial source and quarantined for 30 days under ABSL-2 conditions. Animals were implanted with microchip transponders for animal ID and temperature data collection. For studies with EBOV and MARV, animals were transferred to the Galveston National Laboratory ABSL-4 facility. Animals were segregated into groups of three. Except for one MARVinfected male, all bats were female. Each group was housed in a separate cage for inoculation with the same virus. After acclimation to the facility, animals were anesthetized with isoflurane and infected subcutaneously in the scapular region with 10 5 focus forming units (FFU; titrated on Vero E6 cells) of EBOV or MARV. Every other day, animals were anesthetized by isoflurane, weighed, temperature was determined via transponder, and 100-150 µL of blood was collected from the propatagial vein. Nucleases in blood were inactivated in 1 mL of TRIzol reagent (Thermo-Fisher Scientific). Samples were then removed from ABSL-4 containment, and RNA was extracted. Droplet-digital RT-PCR (ddRT-PCR) with primers specific to the nucleoprotein (NP) gene was used to detect viremia. If fewer than 10 6 MARV RNA copies/mL viremia were detected in a MARV-inoculated bat, the animal was observed for additional 2 days to allow the animal to reach a higher viral RNA load. In 48-96 hours after first observation of viremia, the animal was euthanized under deep isoflurane sedation via cardiac exsanguination confirmed by bilateral open chest. All EBOV-inoculated bats were euthanized 48 hours after the first detection of viremia, independent of viral RNA load. Tissues were collected (listed in Table S1) and immediately homogenized in an appropriate volume of TRIzol reagent and stored at -80°C. Tissue sections were also homogenized in minimal essential media (MEM) supplemented with 10% fetal bovine serum and stored at -80°C. Additional tissue sections were fixed in 10% neutral buffered formalin for histopathology.

bat-filovirus-jayaprakash
Tissues and PBMCs were also collected from three uninfected control animals. Given that the course of infection appears to be relatively short in these animals 21 , we sacrificed them shortly after onset of viremia in the infected animals to ensure adequate capture of changes in transcriptional dynamics. As such, animals were bled every other day, and viral loads were assessed via ddRT-PCR. In addition, animals were weighed, and temperature was determined with each bleed.
All animal procedures were performed in compliance with protocols approved by the Institutional Animal Care and Use Committee at the University of Texas Medical Branch at Galveston. Leukocyte isolation. Leukocyte isolation was performed using ACK lysis buffer (Gibco). Ten volumes of lysis buffer were added to whole blood, incubated for 2-3 minutes, and then neutralized with complete DMEM media containing 10% FBS. Following neutralization, samples were centrifuged at 250 g for 5 minutes at 4°C, after which the supernatant was decanted from the pellet. This process was repeated several times per sample until a white pellet of cells free of contaminating red blood cells remained. Density gradient purification was not performed on these samples prior to or after red blood cell lysis; therefore, these leukocyte populations were assumed to contain granulocytes in addition to PBMCs. mRNA sequencing. Total RNA was isolated from bat tissues using Ambion's RNA isolation and purification kit. For most samples, polyA-tailed mRNA was selected using beads with oligo-deoxythymidine and then fragmented. A few samples with poor (RNA Integrity Number) scores were treated with Ribominus (targeting human ribosomal RNA) to enrich for polyA-tailed mRNA before fragmentation. cDNA was synthesized using random hexamers and ligated with bar-coded adaptors compatible with Illumina's NextSeq 500 sequencer. A total of 88 samples were sequenced on the NextSeq 500, as 75 base pair single reads.

Expression Analyses.
Kallisto was used to determine transcript expression levels from reads. Kallisto uses pseudo-alignments and is relatively more accepting of errors/variants in reads 31 . This is the appropriate tool to quantify transcripts in the samples, as the mRNA sequences from different species can have mismatches to error-free reads. Kallisto uses a parameter "k" while indexing the database to specify how sensitive it is to matches with smaller k values leading to more spurious hits. We empirically determined k=23 to be an appropriate parameter value with which to index the reference mRNA dataset. We used the transcripts-permillion (tpm) value as the transcript expression levels to determine changes in expression across samples.
We used the presence of viral transcripts to confirm proper assignment of samples. This approach has previously helped us to identify and correct mistakes of annotation in some of the cell line data and also identified a problem with a published dataset 32 , where all the naïve (uninfected) samples showed signs of viral infection. Furthermore, to ensure there was no mislabeling of samples from individual bats, we identified single nucleotide variants to ensure that all tissue samples from an individual had the same set of variants Tools for data exploration and interrogation. We developed a web browser-based tool using Shiny in R, which is accessible at http://katahdin.girihlet.com/shiny/bat/. This allows exploration of the data across various samples on a gene-by-gene basis, as well as analysis of viral expression in the samples. Samples can also be compared using scatter plots and hierarchical clustering. Data from the outlier excluded from our analyses is available in the online tool.
Statistics. Datasets obtained from in vivo studies are inherently noisy, for a variety of biological reasons: viral replication and infection of cells is variable across samples/tissues and the samples consist of a heterogenous mixture of cell types that can vary from sample to sample, even from the same tissue type. Bats can also be infected by other pathogens in addition to the filovirus. Large changes in expression profiles were readily detected by comparing averages across replicates, since such changes are less affected by noise; however, subtle changes (less than 2-fold) were difficult to reliably detect due to lack of power in the sample size and variability between samples. To accommodate this most effects noted in our study are greater than 2-fold up-or downregulated.

Pathway analyses.
A basic assumption underlying our study is that bats are mammals that possess innate and adaptive responses to infections that roughly mirror those seen in humans. The data from comparative filovirus infections in human and bat cell lines supports this. 29 Despite similarities in the basic architecture of the networks, effects seen at the organismal level are likely to arise from subtle differences in the systemic responses. We were, therefore, able to guide our analysis by what is known from human and mouse studies of the relevant pathways and genetic networks.
We identified pathways of interest, based on divergent genes that were differentially expressed in bat liver upon filovirus infection, as explained in the results (Fig. 3).

Clinical, virological, and pathological findings in inoculated bats
Inoculated bats showed no apparent clinical signs or changes in behavior, and body weights and temperatures remained relatively constant ( Fig. 1-I,II). Viral RNA was detected (using ddRT-PCR) in the blood of MARVinoculated animals earlier (and higher) than in the EBOV-inoculated animals ( Fig. 1-III), particularly in the liver, spleen, and kidney samples. We examined the potential for excretion via salivary glands and urine in bat-filovirus-jayaprakash MARV-inoculated bats. Two of three animals had virus detectable by plaque assay in the salivary glands, while one of three animals had detectable virus in the kidneys (Fig 1-IV). The virus was not detected by plaque assay (limit of detection is 10 Pfu/cc) in most or all tissues collected from EBOV-inoculated animals (Fig 1-V).
In MARV-infected animals, histopathological observations were largely consistent with prior publications 18,21 . For MARV-infected animals, cytoplasmic immunostaining was performed using a pan-filovirus antibody. In these bats, we observed diffuse cytoplasmic immunostaining with moderate frequency in the absence of histopathological lesions in the mammary glands and testes, suggesting presence of virus in these organs. Two of the three EBOV-inoculated animals presented with noteworthy histopathological lesions in the liver, consisting of pigmented and unpigmented infiltrates of aggregated mononuclear cells compressing adjacent tissue structures, and eosinophilic nuclear and cytoplasmic inclusions. Focal EBOV immunostaining was observed in the liver of one animal, using both pan-filovirus and EBOV-VP40 antibodies, but very few foci were found, suggesting low viral replication (Fig 1-VI). Since the animals did not come from a colony known to be pathogen-free and viral loads were determined to be extremely low, any connections between the histopathology and EBOV infection are unclear.

Filovirus infection of bats results in a significant and consistent response in the transcriptome
mRNA from tissues (liver, spleen, kidney, lungs, salivary glands, large and small intestine, and testes) collected from infected and uninfected bats was deep sequenced. Multi-dimensional scaling (MDS, Fig. 2A) established that one non-infected control bat liver sample (labeled cb1 in the shiny tool) seemed to be an outlier compared to the other two (cb2 and cb3 in the shiny tool); many inflammatory genes were upregulated in this sample , suggesting that cb1 may have had an unexpected injury or infection. Although we left cb1 out of the analysis, cb1 data are available for exploration in the shiny tools.
Consistent with the fact that liver is one of the main targets of MARV 33 and abundant viral transcripts were present only in the liver samples, we expected to detect robust transcriptional response in the liver tissue. However, samples from other tissues also clustered separately based on the type of infection (MARV, EBOV and mock, Fig. 2B, S1). This suggests that even though the liver was the focus of replication, the response to infection was system wide.
We summarize the effect of filoviral infection on the liver transcriptome using an upset plot (Fig. 3), which is just another way of drawing a Venn diagram, showing membership of genes in six sets (genes up/down regulated in EBOV-infected samples compared to mock-infected samples, genes up/down regulated in MARVinfected samples compared to mock-infected samples, and genes up/down regulated in EBOV-infected samples compared to MARV-infected samples). The various intersections between the sets show members unique to that intersection. A large set of genes was found to respond to infection with either virus, indicating the response is broader than a simple perturbation of the naïve state. We also found that more genes responded to MARV infection than to EBOV infection, concordant with more robust replication of MARV in these animals (Fig. 3).
As such, most of our analyses concentrated on liver RNA transcripts. For some genes, we also analyzed transcriptional response in kidneys and/or spleens in order to understand the regulation of certain pathways (e.g., Renin is expressed in kidney and regulates the blood pressure system).

Responsive, evolutionarily divergent bat genes guide pathway analysis
bat-filovirus-jayaprakash Genomic and transcriptomic datasets provide a rich field for developing theories, but they can also be minefields when the analysis is carried out without guidance from the relevant biology. Routine pathway analyses using standard tools usually result in lists of pathways/functions replacing gene lists, often without offering much additional insight. If we start with the full list of genes responsive to filovirus infections, the list will be dominated by the interferon response genes, obscuring the pathways responsible for the systemic response. The multiple testing problem becomes apparent at this level because, with the large set of responsive genes it is not clear which genes are important and if some are highlighted due to random chance To guide our exploration of the datasets, we identified a set of bat genes (2439 genes) with significant differences from their human/mouse homologues, defined as genes whose homologues could not be identified using BLASTn with default settings. This is based on our hypothesis that these divergent genes must be the foundation of the differences in the systemic responses to filoviruses between bats and humans systemic. Considering only genes with reasonable expression (tpm > 20) in at least one class of liver samples (MARV-, EBOV-or mock-infected) we refined the list down to 264 genes. Of these, 151 genes were responsive in at least one class of bat livers, defined as up-or down-regulated at least 2-fold upon infection with EBOV or MARV. This process of narrowing down the list of genes of interest is depicted in Fig. 4.
Tables S1-S8 show the 151 genes split into various classes, upregulated (the log2 ratio to the mock-infected sample is greater than 0.6) in both, downregulated (log2 ratio to the mock-infected samples is less than -0.6) and various combinations thereof. The tables have annotated various pathways/processes that the genes participate in. The major themes identified by this preliminary analysis and the connections between them are outlined in Fig. 5. We found that the following pathways/systems were impacted by filovirus infection: acute phase proteins (Table 1), interferon responsive genes (Fig. 6,7), macrophage polarization ( Fig. 8,9,10), the complement system (Fig. 11), the adaptive immune system (Fig. 12) and the vascular system ( Fig. 13,14,15).

Infected bats have transcriptional profiles indicative of a robust innate immune response to filovirus infection
We detected a robust innate immune response in vivo not only to MARV, but also to EBOV, despite limited viral replication. Innate immune genes are seen to be responsive to MARV and EBOV infection (Fig. 5,6,7). Our previous in vitro transcriptomic studies demonstrated that the innate immune response in bat cells is broadly similar to that in human cells 29 . Additionally, most innate immune bat genes are not divergent from their human homologs, as we have defined it. Thus, we believe that the innate immune response likely cannot explain, on its own, the drastic difference in infection phenotype observed in bats relative to humans.

Infected bats exhibit an acute phase response, involving multiple acute phase proteins (APP)
Inflammation and injury 34,35 trigger inflammatory cytokines (e.g., Interleukin-1(IL-1), IL-6, and TNFα). These cytokines subsequently trigger transcriptional events that lead to an increase in serum levels of some acute phase protein (APP) 36 and a decrease of others (e.g., transferrin, albumin 37 ). APPs are produced by hepatocytes in the liver, and are an important part of the innate immune response. 38 Depending upon the combination of cytokines, the specific reaction can vary in response to different inflammatory conditions. [39][40][41] The IL-6 response is often not directly detected by mRNA-seq data due to low expression; however, the APPs respond strongly.
Our data indicates that MARV infection, and to a lesser extent EBOV infection, elicits a strong APP response in bats ( Table 1). SAA expression, for example, increased 1,000-fold in response to MARV infection. Curiously, bat-filovirus-jayaprakash c-reactive protein (CRP), known as a marker for inflammation/acute-phase-response in humans, was not expressed in bat livers. Potentially, CRP may not be present in an active form in bats at all ( Table 1).
Persistent expression of IL-6, known to cause chronic inflammation, is tightly regulated via negative feedback loops. We found that several of these negative feedback loops were strongly upregulated in liver tissues in response to MARV infection and to a lesser degree in response to EBOV infection. These include 1) GP130 (IL-6ST), which acts to initiate the Janus kinase (JAK)-STAT3 pathway, 2) STAT3, which induces various IL-6 responsive genes such as APPs and the SOCS (suppressor of cytokine signaling) genes, and 3) the SOCS genes themselves, which bind to JAK and GP130 to stop IL-6 signaling (Fig. 6,7).
Inflammation is also mediated by leukotrienes (LTC4) and prostaglandin E, which are produced by microsomal glutathione S-transferases (MGST1 and MGST2) 42 . Under filoviral infection in bats, MGST1 and MGST2 are highly upregulated.

Infected bats exhibit a transcriptional profile suggestive of an early transition from an M1 dominated to an M2 dominated macrophage population
Macrophages recognize and phagocytize foreign organisms and damaged host cells. Macrophages, an important early target for filoviruses 43 , play a major role in the immune response in the liver, one of the primary sites of filovirus replication. Macrophages can be either in an M1 state (inflammatory, assisting innate immunity) or in an M2 state (anti-inflammatory, assisting tissue regeneration) (Fig. 8,9,10) and can polarize or shift from one state to the other. Key markers of M1 macrophage activation include IRF5, NF-κB, AP-1, and STAT1 ( Fig. 8), which subsequently lead to the secretion of pro-inflammatory cytokines such as IFN-γ, IL-1, IL-6, IL-12, IL-23 and TNFα.
M2 macrophages have additional anti-inflammatory subclasses (M2a, M2b and M2c) that share some markers and are distinguished by others 44 . M2a macrophages enhance tissue regeneration and inhibit inflammatory responses. They are activated by IL-4 and IL-13, which, in turn, upregulate arginase-1(ARG1), IL-10 and TGF- 45 . The M2b macrophages have anti-inflammatory activity by producing IL-1, IL-6, IL-10, TNF-. M2c macrophages suppress inflammatory response. They are activated by IL-10, transforming growth factor beta (TGF-), and glucocorticoids, and they produce IL-10 and TGF-β. In the anti-inflammatory state, mitochondrial activity is increased and primarily involved in fatty acid metabolism.
Prolonged M1 activity can be harmful and is modulated by the negative feedback that transforms macrophages from M1 to M2 state 46,47 , thereby controlling inflammation during infections and the transition to tissue repair 48,49 (Fig. 8). M1 macrophages rely upon glycolysis and M2 macrophages utilize fatty acid oxidation as an energy source. The switch between states is achieved by simply disrupting cellular energy metabolism. M2 macrophage polarization is accompanied by mitochondrial biogenesis. Hypoxia-inducible factor 1 (HIF1) is a key regulator inducing M2 polarization through hypoxia 50 . HIF1A [51][52][53] , promotes mitophagy and is required by M1 macrophages, whereas M2 macrophages depend on the mitochondrial oxidative metabolism. Inactivating HIF1A also promotes M2 polarization.
We found that filovirus-infected bats exhibit upregulation of key markers of M1 macrophages, including IRF5, NF-κB, AP1G1 (a subunit of the AP-1 complex), and STAT1 (Fig. 6,7). These lead to the secretion of proinflammatory cytokines such as IFN-γ, IL-1, IL-6, IL-12, IL-23 and TNFα, all of which were upregulated in filovirus-infected bats, which we infer through either direct or indirect evidence. Expression of these markers bat-filovirus-jayaprakash was stronger in MARV-infected animals, corresponding to greater replication of the virus, but it is difficult to clearly delineate the cause/effect relationship.
Although we did not detect expression of IL-4 and IL-13 in any bat tissues, genes regulated by them, such as MRC1, TGFB1 and ARG1 were found to be highly expressed in livers of bats infected with both viruses. CSF1 is a cytokine that controls the production, differentiation, and function of macrophages. The CSF1 receptor (CSF1R) that mediates the biological functions of CSF1 was also upregulated in filovirus infected bat livers, the upregulation was greater in EBOV infected animals (Fig. 8,9). Several genes related to fatty acid oxidation were found to be upregulated by filovirus infection (Tables 1-3). Infected bats also upregulated multiple markers of mitochondrial abundance, such as TFAM, OPA1, MFN1/2, and DNM1L, more so in MARV than in EBOV. The pyruvate dehydrogenase, PDK1, involved in the response to hypoxia was also upregulated (more so in MARV than EBOV) 54 . HGF-MET and PPARGC1A involved in mitochondrial biogenesis 55 , are upregulated upon MARV infection.
In MARV-infected bats, SOCS3, which promotes M1 polarization, was upregulated. Several M2 markers, including TGFB1, ARG1, and MRC1 were also upregulated ( Fig. 8,9). Additional evidence for M2 polarization is provided by the non-anemic state of EBOV-infected bats, inferred from the presence of abundant iron, which enhances macrophage M1 to M2 polarization. 56 PKM, HIF1AN and HGF, which play an important role in inactivating HIF1A to promote the M1 to M2 switch, are all upregulated in filovirus infected bats (Fig. 8,9).
Finally, the gene GPD2, the mitochondrial glycerol-3-phosphate dehydrogenase which contributes to the shift in core metabolism in macrophages associated with the M1 to M2 transition during infection aiding tissue repair 57 was found to be upregulated by filovirus infection (Fig. 9).
These findings indicate that bats may transition from an M1-dominated macrophage response to an M2dominated response relatively early in infection compared to humans. This switch may be an important component of the ability of the animals to control infection and avoid clinical disease by suppressing inflammation and promoting tissue regeneration. This would reduce or prevent immunopathology and allow the adaptive immune system to effectively control the virus.

Expression of key components of the classical complement pathway is inhibited by filoviral infection
The complement pathway has three branches, the classical pathway, the mannose-binding lectin pathway and the alternative pathway 58 . The classical pathway recognizes antigens bound to antibodies; the lectin pathway binds to sugar molecules on the virus particles and the alternative pathway is a component of the innate defense against infections.
Both the classical and lectin complement pathways were activated by filovirus infection (Fig. 11). However, several elements of the classical pathway were downregulated or even not expressed in filovirus infected livers, including C1R, C3, C8G, and MASP2. Downregulation or suppression of expression of C1R, C3, and MASP2 would compromise the classical pathway as they are key to the antibody effector activity. This likely reduces the efficiency of the antibody activity in infected bats, consistent with the finding that antibody-mediated virus neutralization is not the dominant mode for filovirus clearance in R. aegyptiacus bats. 59

Infected bats exhibit transcriptional signatures of robust T cell activity bat-filovirus-jayaprakash
We found that multiple markers of CD8+ T cell activity, including CCL3, ANAX1, TIMD4 and MAGT1 were upregulated by filovirus infection (Fig. 12), indicating that bats may mount a strong cellular response despite the apparent weakening of the humoral response. Overall, this suggests that control and clearance of filovirus infection in bats may largely depend upon a robust cellular response, similar to what has been observed in humans, where individuals who recover tend to mount robust cellular responses. [60][61][62] Infected bats exhibit dramatic signatures in the vascular system, with low coagulation, vasodilation and non-anemic states despite HAMP upregulation Three major interconnected pathways of the vascular system are, a) iron metabolism (Fig. 13), b) blood pressure (Fig. 14) and c) coagulation (Fig. 15). The interplay between the three is complicated (outlined in Fig.  5). We highlight genes involved in the response affecting the three pathways and the connections between them. To better present our results in context, give a brief overview of iron metabolism.
Iron, an essential component of heme needed for oxygen transport, is tightly regulated, 63 mostly by hepcidin (HAMP) 64 (Fig. 13). HAMP controls the internal absorption of iron 65 , by binding Ferroportin (SLC40A1/FPN1), to block export of iron across membranes and cause Ferroportin degradation. HAMP is upregulated by iron in serum and pro-inflammatory stimuli (IL-6), such as infection. 66 In blood, iron forms a complex with Transferrin (TF), which is enabled by ceruloplasmin (CP) and MOXD1, involved in processing copper. 67 In the cytosol, iron is bound to ferritin (comprised of a heavy chain, FTH1 and a light chain FTL), synthesized by cells in response to increased iron. Thus, ferritin is a marker for iron levels in serum. 68 In mitochondria, iron is bound to FTMT, the mitochondrial ferritin. 69 PCBP2, an iron chaperone is also needed for iron transport within the cytosol 70 . STEAP3 helps transport iron from Transferrin to the cytosol of erythrocytes 71 .
Most iron is in hemoglobin (66%), the remainder is stored mostly in macrophages in the liver, which take up iron through the CD163 receptor. The bone marrow can suppress HAMP synthesis, in response to anemia, leading to export of iron from macrophages, and increased uptake of iron from diet.
In EBOV-and MARV-infected bat tissues, HAMP was upregulated (Fig. 13). In MARV infection, macrophage expansion/infiltration (as was observed in histology sections of infected tissues) and lowered hemoglobin expression suggest that red blood cell production might be impaired, which is potentially a sign of anemia. Further, CD164, which suppresses hematopoietic cell proliferation, was also upregulated by MARV infection (Fig. 13).
In EBOV-infected bats, FTH1 and FTMT were both upregulated (Fig. 13), reflecting increased iron abundance in serum. In contrast, FTH1 and FTMT, along with PCBP2 were downregulated in MARV-infected bats. Upregulation of HBB in EBOV-infected samples was also observed. This suggests that hematopoiesis was impaired in MARV-infected bats, but not in EBOV-infected bats. It is likely that this was a result of the divergent pathobiology of EBOV and MARV infections in Egyptian rousettes. The early control of EBOV by these bats suggest that the iron levels in bats that eventually clear MARV infections may resemble the iron levels in EBOV-infected animals.

Filovirus-infected bats exhibited transcriptional signatures of vasodilation, which may indicate a low blood pressure state
Blood pressure (Fig. 14) can be controlled by either vasoconstriction (e.g., AGT2 activity and the renin/angiotensin system), or by changing the concentration of salts in blood (e.g., regulation of aldosterone by bat-filovirus-jayaprakash CYP11B2). The primary means of blood pressure regulation is renal expression of renin, which converts Angiotensinogen to Angiotensin I. Angiotensin converting enzyme (ACE) converts Angiotensin I to Angiotensin II, which constricts blood vessels to increase pressure. ACE expression is upregulated by a feedback loop triggered by low blood pressure. Angiotensin II also enhances production of active plasmin increasing coagulation, connecting the pressure and coagulation pathways 72 . Inflammation upregulates the SERPIN genes, several complement genes, and HAMP, which connects the iron, blood pressure and coagulation pathways. Prostaglandin I2 synthase (PTGIS), which inhibits platelet aggregation and reduces blood pressure, CYP11B1 and CYP11B2 (which reduce blood pressure and inflammation) all connect blood pressure, inflammation and coagulation 73 .
During filovirus infection in bats, we found that ACE was upregulated, while angiotensin and AGT were downregulated (Fig. 14). Additionally, we found that PTGIS was upregulated. In EBOV-infected bats, CYP11B2 (which regulates blood pressure by synthesizing aldosterone) was upregulated (Fig. 14) The vascular response might be another key to the response of bats to filovirus infection. Humans infected with EBOV or MARV in many cases eventually exhibit excessive bleeding, low blood pressure, and excessive and dysregulated coagulation in the form of disseminated intravascular coagulation 74 . Our data suggest that bats use multiple strategies to protect their vascular systems during filovirus infection, an important mechanism for limiting the pathology.

DISCUSSION
The ability of bats to serve as hosts for a variety of diverse viruses has been a topic of considerable interest and scientific attention, with several theories being proposed to explain this phenomenon.
One theory posits that bats have constitutively expressed interferons or permanently active innate immune system, ready and waiting for pathogens to appear 75 , although this has not been a universal observation in all bat species 76,77 . The reported levels of constitutive expression reported 75 are extremely low (at least 5 orders of magnitude lower than ribosomal RNA), which makes them undetectable in mRNAseq, but also raises questions about nature of the constitutive expression. Further, in an mRNA-seq study on PBMCs from EBOV-infected humans, individuals who succumbed to disease showed stronger upregulation of interferon signaling and acute phase response-related genes compared to survivors during the acute phase of infection 82 . Therefore, the differences in responses between human and bats goes beyond any upregulation or constitutive expression of interferons.
Another theory suggested that components of the innate immune response (e.g., STING(TMEM173)) could be mutated to become less effective in bats 78 . It is unlikely that a single gene is the "magic bullet" that explains the profound differences observed between human and bat responses to filovirus infection. Instead, our data together with the extant literature strongly suggest that modifications of entire systems is required to produce the observed divergence in the response to infection.
The innate response of human and bat cell lines to filovirus infections is almost identical, but in vivo, the clinical course and outcomes in humans and bats are different. Such differential responses likely involve a variety of tissue/cell types and the interactions between them are driven by numerous genes. Identifying divergent genes in this large set and using them to identify the systemic differences between bats and humans provides a rational basis for the analysis of this data. By limiting ourselves to genes that fit this requirement, the multiple testing problem was ameliorated, by drastically reducing the number of genes being considered.

bat-filovirus-jayaprakash
A study in humans infected with EBOV 83 analyzed 55 biomarkers in blood, showing viremia to be associated with elevated levels of tissue factor and tissue plasminogen activator, which is consistent with coagulopathy. Nonfatal cases had higher levels of sCD40L expression, a marker for T cell activity, consistent with our data that suggest that T cells are highly active during filovirus infection of bats while antibody-mediated virus neutralization is potentially less important for filovirus clearance 59 .
The state of the bat under filoviral infection and the fact that they do survive these infections suggests potential approaches to helping human patients.
We believe the anti-inflammatory state induced in bats upon filovirus infection is a natural application of this strategy, especially the early switch to M2 macrophage polarization. This allows the adaptive defenses of bats to clear the virus and avoid damage from immunopathology.
Thus, an attempt could be made to reduce the human hyperinflammatory response 88 to filovirus infections by modulating the innate response to prevent damage and allow other processes to clear the infection and allow for wound healing. For example, Anti-inflammatory agents could also be used to emulate the protective physiological conditions observed in bats (e.g., through the inhibition of IL-6). One approach would be to target the IL-6 receptor through the use of tocilizumab (Actemra), an antibody directed against the IL-6-receptor 95 . Alternatively, IL-6 could be targeted directly with agents such as siltuximab (Sylvant) 96 . Another class of antiinflammatory agents are LTC4 inhibitors, used to treat asthma, may be of benefit in filovirus infection which in bats upregulate MGST1 and MGST2, in turn inducing leukotrienes (LTC4) and prostaglandin E, which are mediators of inflammation 42 .
Our evidence further suggests that bats, upon infection by filovirus, may naturally vasodilate and reduce their blood pressure (mimicking the action of ACE inhibitors). They further make the endothelial system antithrombotic. Surprisingly, use of ACE inhibitors and statins has already been tried in field studies which have suggested they might help humans infected with EBOV 90 . Along these lines, another potentially useful drug is Prostaglandin I2 (PGI2, or epoprostenol, its drug name), a powerful vasodilator and anti-coagulant that acts by binding to the prostacyclin receptor. This has potential for use in human filovirus infections to emulate the physiological conditions (low blood pressure and coagulation) in bats that we believe have protective effects 91 .
In Infected bats, high HAMP expression seems decoupled from the levels of iron, which should normally be depressed by HAMP. This suggests HAMP inhibitors, used to treat anemia, might prove useful in filoviral infections. Two Hamp inhibitors, Heparin 92 and erythropoietin (EPO) 93,94 , have additional effects, anticoagulation and RBC synthesis respectively, which might make them particularly useful. Vitamin D is also a HAMP inhibitor which could be used with minimal side-effects.
A limitation of our study is our inability to conduct genetic manipulations that would help tease out details of interactions that we have uncovered here. Another limitation is our inability to pursue potential therapeutic agents we have identified. Investigating these effects further requires either reconstituting systems in vitro or experimenting on live animal models, both of which are beyond the scope of this work.

CONCLUSIONS
Bats are an ideal model system for research into the pathobiology of filovirus infection. The resistance of bats to clinical illness provides a useful basis for comparison to human infection. Based on transcriptional analyses, we have composed a framework for understanding a filovirus-infected bat's remarkable resilience to serious bat-filovirus-jayaprakash disease, with induction of anti-inflammatory state to be one of the most striking observations. Our study identifies several ways in which the systemic responses in bats and humans to filoviruses differ. These studies have the potential to aid in the development of new strategies to effectively treat filovirus infections in humans.

Data
All data underlying the balloon plots is available as csv files on the shiny tool website (http://katahdin.girihlet.com/shiny/bat/). Additionally, a fasta file containing all the mRNA sequences used in our analysis is available. The raw sequencing reads will be deposited with GEO, and the shiny site has several tools for analysis and exploration of data.        54 , which leads to vascular effects 80 . During filovirus infection, the system is in an anti-inflammatory state (markers for M2 are up), while during MARV infection, the inflammatory M1 state is also seen, consistent with the acute phase response involving down regulation of albumin and upregulation of SAA1/2. The M1-M2 switch is anti-inflammatory and promotes wound healing, low-angiotensin, hematopoiesis etc. Different SOCS family molecules are molecular switches that control M1/M2 macrophage polarization. High expression of SOCS3 in MARV infection promotes M1 polarization. M2 markers are also upregulated. The M2 state is probably the key to the resilience of bats during filovirus infection, allowing bats to fight off filoviruses without significant adverse effects. Under MARV infection, there is limited viral replication and disease symptoms before it is cleared, consistent with this, the anti-inflammatory state is not as pronounced as in the case of EBOV-infected bats. The anti-inflammatory state is also characterized by tissue regeneration, which is facilitated by an increase in mitochondrial numbers and fatty acid oxidation activity. There is also a connection between iron metabolism and macrophage M1/M2 polarization with increasing iron favoring M2 54    Filovirus infections in bat eventually triggers tissue regeneration, reflected in the M2 macrophages, which are antiinflammatory. The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6. There is more activity under MARV infection, consistent with higher viral loads.  Most are upregulated by filovirus infection, reflecting enhanced CD8 T cell activity, which probably plays a major role in controlling the infections. The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.
bat-filovirus-jayaprakash  Genes whose activation cannot be measured by mRNAseq are shown in white. Catalytic activity is represented by a red dot. The serine protease plasmin opposes this process by attacking the fibrin mesh to dissolve clots. Urokinase-type plasminogen activator (uPA, PLAU gene) activates plasminogen to generate plasmin. SERPINE1 (PAI-1) inhibits the activity of uPA, blocking the creation of Plasmin, thereby stabilizing clots. SERPINF2 also stabilizes clots by directly inhibiting plasmin. PTGIS creates prostacyclin, which prevents platelet aggregation, yet another path for inhibiting clot formation and coagulation. Prostacyclin is also a potent vasodilator. Failure in the control of plasmin (e.g. inactivation of SERPINF2 which inhibits plasmin) can lead to hemorrhagic diathesis, while blocking plasmin activity can lead to excessive clotting. Filovirus infection in bats leads to low F2 (which lowers fibrin) and raises levels of PTGIS which suggest that they are in a low coagulation state. The colored bands on either side of the gene names depict the effect of filovirus infection on expression (MARV-left and EBOV-right). The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.  Table 1 Acute Phase Proteins respond strongly to inflammatory cytokines (IL-6, TNFα etc.). SAA1/2 is highly up regulated by the filovirus (MARV more than EBOV), as is ORM2 . CP, involved in copper/iron metabolism is also massively up regulated by MARV infection, while exhibiting a smaller effect in EBOV. CRP, used as a marker for acute phase response in humans, does not appear to be expressed in these bats.TF is highly expressed in all samples, but does not react to filoviral infection, while TTR is not expressed in any of the samples. "UP/up" stands for > 5/2-fold increase, "Neutral" is when expression did not change and "down" represents >2-fold decrease. The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.   inhibits production of inflammatory cytokines by macrophages S100A 12 paralog of S100A8 IL-10 induced monocytes macrophages proinflammatory mast cell chemoattractant innate immune response PLAC8 expressed by macrophage phospholipid metabolic process ISG15 regulation of IFNG production Mito function in macrophages CCL7 regulates macrophages attracts monocytes eosinophils but not neutrophils cellular response to interferon-gamma innate immunity CLEC4Fendocytosis pathogen detector CLECL1 expressed by dendritic and B cells enhances IL-4 production regulates immune response CXCL3 chemoattractant for neutrophils immune response   positive regulation of sodium ion export UNDEF425 UNDEF464 bat-filovirus-jayaprakash gene process macrophages BPI negative regulation of IL-6 production expressed by macrophages Bactericidal permeability-increasing protein Complement  inflammation, mitochondria, lipid metabolism, tissue regeneration, Macrophages, T cell function and the complement are common themes running through these lists. The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.   inflammation, mitochondria, lipid metabolism, tissue regeneration, T cell function and the complement are common themes running through these lists. The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.  Table S7 Divergent genes downregulated by EBOV infection. Only 3 known genes, from innate immunity, vascular(coagulation) and Digestion, which seems to occur occasionally in these list of genes (probably related to liver function) . The balloon plot compares responses of genes to EBOV, MARV and mock against each other. The radius of circle is proportional to log2(ratio), gray is used when absolute values of log2(ratio) < 0.6.