The IFN Response in Bats Displays Distinctive IFN-Stimulated Gene Expression Kinetics with Atypical RNASEL Induction

Bats are recognized as important viral reservoirs, and many highly pathogenic viruses, including Nipah virus (1), Hendra virus (2), Marburg virus (3), severe acute respiratory syndrome–like coronaviruses (4), and Ebola virus (5), have been detected in various bat species. In experimental infection studies, certain bats can be productively infected with pathogenic viruses without obvious disease symptoms (6–9). A recent study identified bats as hosts for a greater proportion of zoonotic viruses than all other mammalian orders tested, with the highest viral richness found in flaviviruses, bunyaviruses, and rhabdoviruses (10). Although the mechanisms underlying disease resistance are not known, it has been speculated that bats possess unique immune features that confer innate antiviral protection (9, 11). In vertebrates, one of the first lines of defense against viral pathogens is the IFN response. Upon viral infection, pattern-recognition receptors sense viral components and initiate a signaling cascade that results in the secretion of IFNs. These IFNs bind their cognate receptors to activate the JAK–STAT signaling pathway, leading to the transcriptional induction of hundreds of IFN-stimulated genes (ISGs), many of which have antiviral activity (12).

The black flying fox (Pteropus alecto) is an asymptomatic natural reservoir for the highly lethal Hendra virus (2, 13, 14). Studies of the recently sequenced black flying fox genome revealed that genes for key components of antiviral immunity are conserved in bats, such as pathogen sensors, including TLRs (15), RIG-I–like helicases, and NOD-like receptors, IFNs and their receptors, and ISGs (11, 16). Previous efforts to study bat–virus interactions have mainly focused on the host response to viral infection (17–20), and global transcriptional responses to type I IFN remain uncharacterized.

Because the black flying fox harbors pathogenic human viruses and has an annotated genome, we sought to characterize the IFN-induced transcriptional response in this species. Gene-expression analyses revealed that bat cells induce a pool of common ISGs. However, they also induced a small number of apparent novel ISGs, including 2-5A–dependent endoribonuclease (RNASEL). Kinetic analyses revealed that bat ISGs can be categorized into distinct groups, depending on their temporal gene-expression patterns. Additionally, maintenance of ISG expression over time differed between bat and human cells, suggesting distinct mechanisms of gene regulation.

Black flying fox–derived PaBr (brain), PaLu (lung), and PaKi (kidney) immortalized cell lines (21) were grown at 37°C and 5% CO₂ and passaged in DMEM/F-12 (catalog number 11320033; Life Technologies) supplemented with 10% FBS. Human A549 lung adenocarcinoma and HEK293 cells were grown at 37°C and 5% CO₂ and passaged in DMEM (catalog number 11995065) supplemented with 10% FBS and 1× nonessential amino acids (NEAA) (catalog number 11140076; both from Life Technologies).

Yellow fever virus (YFV)-Venus (derived from YF17D-5C25Venus2AUbi) stocks were generated by electroporation of in vitro–transcribed RNA into STAT1^−/− fibroblasts, as previously described (12). Vesicular stomatitis virus (VSV) encoding GFP, Indiana serotype (provided by J. Rose) was generated by passage in BHK-J cells. For both viruses, virus-containing supernatant was centrifuged to remove cellular debris and stored at −80°C until use.

Cells were seeded into 24-well plates at a density of 1 × 10⁵ cells per well. Viral stocks were diluted into DMEM/F-12 media supplemented with 1% FBS to make infection media. Media were aspirated and replaced with 200 μl of infection media. Infections were performed at 37°C for 1 h and then 800 μl of DMEM/F-12 media supplemented with 10% FBS was added back to each well. After approximately one viral life cycle (5 h for VSV, 25 h for YFV), cells were harvested for flow cytometry.

Cells were detached using Accumax, fixed in 1% PFA for 10 min at room temperature, and pelleted by centrifugation at 800 × g. Fixed cell pellets were resuspended in 200 μl of FACS buffer (1× PBS supplemented with 3% FBS). Samples were run on a Stratedigm S1000 instrument using CellCapTure software and gated based on GFP fluorescence. Data analysis was done using FlowJo software (v9.7.6).

Cells were seeded into six-well plates at 3 × 10⁵ cells per well. The following day, cells were treated with 2 ml of DMEM/F-12 media supplemented with 10% FBS and 50 U/ml Universal Type I IFN-α (catalog number 11200; PBL Assay Science). The cells were harvested by aspirating the media, washing twice with 2 ml of sterile 1× PBS, and lysing with 350 μl of Buffer RLT from the RNeasy Mini Kit (QIAGEN). The cell lysate was stored at −80°C until RNA isolation was completed using an RNeasy kit, following the manufacturer’s protocol.

Total RNA samples for each time point were prepared in three independent experiments, as described above. An Agilent 2100 Bioanalyzer was used to determine RNA quality; all samples had an RNA Integrity Number > 9. A Qubit fluorometer was used to determine RNA concentration. Libraries were prepared using the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina), following the manufacturer’s instructions, summarized as follows. Four micrograms of DNase-treated total RNA was used as input. Poly-A RNA was purified and fragmented before strand-specific cDNA synthesis. cDNA was then A-tailed and ligated with indexed adapters. Samples were PCR amplified using the following parameters: 5 μl of PCR Primer Cocktail was added to 20 μl of adapter ligated library and then 25 μl of PCR Master Mix was added to each sample. Samples were mixed by pipette, and the plate was sealed and cycled on a thermal cycler with a 100°C heated lid under the following conditions: initial denaturing at 98°C for 30 s; 15 cycles of 98°C for 10 s, 60°C for 30 s, and 72°C for 30 s; and final extension at 72°C for 5 min, hold at 4°C. Samples were then purified with AMPure XP beads and revalidated on an Agilent 2100 Bioanalyzer and Qubit. Normalized and pooled samples were run on an Illumina HiSEquation 2500 using SBS v3 reagents.

Paired-end 100-bp read length FASTQ files were checked for quality using FastQC (http://www.bioinformatics.babraham.ac.uk/projects/fastqc) and FastQ Screen (http://www.bioinformatics.babraham.ac.uk/projects/fastq_screen/) and were quality trimmed using fastq-mcf (https://github.com/ExpressionAnalysis/ea-utils/blob/wiki/FastqMcf.md). Trimmed fastq files were mapped to black flying fox assembly ASM32557v1 (https://ftp.ncbi.nih.gov/genomes/Pteropus_alecto) using TopHat (22). Duplicates were marked using Picard tools (https://broadinstitute.github.io/picard/). Reference annotation–based transcript assembly was done using cufflinks (23), and read counts were generated using featureCounts (24). Pairwise differential expression analysis was performed using edgeR (25), and longitudinal analysis was performed using time course (26) after data transformation by voom (27). Differentially expressed unannotated genes were manually annotated using a nucleotide Basic Local Alignment Search Tool (28) search for homologous genes.

Data were deposited in the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo/) under accession number GSE102296.

Heat maps were constructed using Morpheus (https://software.broadinstitute.org/morpheus/).

A customized panel targeting black flying fox and human genes (Schoggins_1_C4066; NanoString Technologies, Seattle, WA) was used to measure the expression of 50 genes per species. Probes were designed to target as many known transcript variants as possible (see Supplemental Table II for accession numbers of targeted genes). Total RNA was isolated from IFN-treated samples, as described above. One hundred nanograms of total RNA in 5 μl was used as input in each reaction for NanoString assay. RNA samples were processed according to the NanoString nCounter XT CodeSet Gene Expression Assay manufacturer’s protocol. Following hybridization, transcripts were quantitated using an nCounter Digital Analyzer.

Total RNA was prepared as described above. Reactions were prepared with a QuantiFast SYBR Green RT-PCR kit (catalog number 204154; QIAGEN), as recommended by the manufacturer, using 50 ng of total RNA per reaction. Samples were run on an Applied Biosystems 7500 Fast Real-Time PCR System using 7500 Software v2.0.6. The program consisted of a reverse-transcription step at 50°C for 10 min, a polymerase-activation step at 95°C for 5 min, and cycling steps alternating between 95°C for 10 s and 60°C for 30 s (40 cycles). Melt curves were generated for all experiments by ramping up the temperature 1°C/min from 60 to 95°C. Only a single melt curve peak was observed for all primer sets. Primers used to amplify RNASEL (XM_006907762.2) are 5′-CCACCCTGGGGAAAATGTGA-3′ and 5′-GGAGGATCCTGCTTGCTTGT-3′. Primers used to amplify reference gene RPS11 (XM_006905029) are 5′-ATCCGCCGAGACTATCTCCA-3′ and 5′-GGACATCTCTGAAGCAGGGT-3′. Relative expression in IFN-treated samples versus untreated samples was calculated using the comparative C_T method using RPS11 as the reference gene.

lentiCRISPR v2 was a gift from F. Zhang (plasmid number 52961; Addgene). RNASEL targeting guides were generated by cloning annealed complementary 20-bp oligonucleotides with Esp3I-compatible overhangs (5′-caccgAGACCCACACCCTCCAGCAG-3′ and 5′-aaacCTGCTGGAGGGTGTGGGTCTc-3′) targeting the black flying fox RNASEL gene into the lentiCRISPR v2 backbone, as described (29). Clustered regularly interspaced short palindromic repeats (CRISPR) guide oligonucleotides were designed using CRISPRdirect (30). Proper assembly was confirmed using Sanger sequencing.

Lentiviral pseudoparticles were made as described (31), with some modifications. Briefly, 2 × 10⁶ HEK293T cells were seeded on a poly-lysine–coated 10-cm plate. The following day, media were changed to 7.5 ml of DMEM with 3% FBS and 1× NEAA. Cells were cotransfected with 5 μg of lentiCRIPSR v2, 2.5 μg of pCMV-VSVg, and 3.5 μg of pGag-pol. Four hours posttransfection, media were changed to 7.5 ml of fresh DMEM with 3% FBS and 1× NEAA. Supernatant was collected 48 h posttransfection, passed through a 0.45-μm filter to remove debris, and stored at −80°C until use.

A total of 3 × 10⁵ PaKi cells was seeded on six-well plates. The following day, media were changed to DMEM/F-12 supplemented with 3% FBS, 4 μg/ml polybrene, and 20 mM HEPES. lentiCRISPR v2 lentiviral pseudoparticles were added, and cells were spinoculated at 800 × g for 45 min at 37°C. Media were changed to DMEM/F-12 with 10% FBS immediately following spinoculation. Forty-eight hours after transduction, cells were pooled into a 10-cm dish and selected in DMEM/F-12 with 10% FBS and 5 μg/ml puromycin.

Genomic DNA was isolated from wild-type (WT) or RNASEL-knockout (KO) bulk populations using a DNeasy Blood and Tissue Kit (QIAGEN). The region flanking the CRISPR-targeted sequence was amplified via PCR using primers 5′-ATGGAGACCAAGAGCCACAACA-3′ and 5′-CGTCCTCGTCCTGGAAATTGA-3′. PCR products were gel purified using a QIAquick Gel Extraction Kit (QIAGEN) and subsequently used in TOPO cloning reactions with a TOPO TA Cloning Kit (Thermo Fisher). Several colonies were selected for each cell background, and colony PCR was used to amplify the CRISPR-targeted region. Samples were analyzed using Sanger sequencing.

A total of 3 × 10⁵ PaKi cells was seeded on six-well plates. The following day, universal IFN (100 U/ml) was added to the media. After 24 h, cells were transfected with 100 ng/ml polyinosinic-polycytidylic acid [poly(I:C)] in Opti-MEM using Lipofectamine 3000. After 4 h, RNA was harvested using an RNeasy Mini Kit (QIAGEN), and RNA integrity was measured on a Total RNA Nano Chip using an Agilent 2100 Bioanalyzer.

Previous studies have shown that exogenous IFN-α and IFN-γ treatment of cells from the black flying fox can suppress Pteropine orthoreovirus Pulau virus (32) and Hendra virus (33), respectively. We treated immortalized black flying fox kidney-derived (PaKi) cells for 24 h with universal IFN-α, which is designed for activity across multiple species. We then infected the cells with two model GFP-expressing reporter viruses: a negative-sense RNA rhabdovirus (VSV) and a positive-sense RNA flavivirus (YFV). We observed a dose-dependent inhibition of both viruses with IFN treatment (Fig. 1A, 1B). VSV infection was maximally inhibited by only 50%, whereas YFV infection was suppressed completely at the highest IFN dose. In addition, we confirmed IFN-mediated dose-dependent and time-dependent induction of the canonical ISG MX1 (Fig. 1C, 1D) (34). These data confirm that PaKi cells are capable of mounting an antiviral response and highlight virus-specific sensitivities to IFN in this cellular background.

We next used total RNA sequencing (RNA-Seq) to profile the global transcriptional response of PaKi cells treated with IFN-α over time (Fig. 2A). Transcriptome assembly analysis across all experimental conditions returned ∼30,000 genes, of which 11,559 met a minimal read count threshold (mean log₂CPM ≥ 0). Differential gene expression analysis revealed that IFN induced 93 genes at 4 h, 104 genes at 8 h, 103 genes at 12 h, and 103 genes at 24 h (fold change [FC] ≥1.5, false discovery rate [FDR] ≤ 0.05) (Fig. 2B). There were no downregulated genes at 4 h, 2 downregulated genes at 8 h, 105 downregulated genes at 12 h, and 279 downregulated genes at 24 h. However, statistical significance of upregulated genes was more robust than the statistical significance of the downregulated genes. Overall, IFN treatment of PaKi cells produced a positive gene-induction signature.

Next, heat maps were generated to assess individual gene induction, using FDR ≤ 0.05 and FC ≥ 4 for at least two time points (Fig. 2C). Many genes in this list have known roles in innate immunity, including well-known ISGs (IFIT1, MX1, OAS2, RSAD2/viperin, USP18), members of the JAK–STAT signaling cascade (STAT1, STAT2, IRF9), pattern recognition receptors (DDX58/RIG-I, IFIH1/MDA5, ZBP1), and transcription factors (ETV7, IRF7, SP110) (35). Notably, we detected induction of GVIN1 (IFN-induced very large GTPase), which is predicted to encode a protein in bats but is annotated as a pseudogene in humans. This list also included transcripts predicted to encode an endogenous retrovirus (ERVK25) and several transcripts predicted to be long noncoding RNAs.

Differentially expressed genes were cross-referenced to the INTERFEROME v2.0 database (36) to determine whether they had previously been reported as IFN-induced genes. At the 4- and 8-h time points, >80% of the genes in our data set overlapped with INTERFEROME v2.0. Because the INTEFEROME database consists predominantly of human and mouse data sets, this result suggests that antiviral responses in bat cells include a conserved repertoire of IFN-inducible genes commonly found in other mammalian species.

We next used a clustering algorithm to group genes in the RNA-Seq data set based on induction kinetics, without imposing FC or FDR cutoffs (26). This analysis revealed that genes were organized into eight subclusters (SC) based on changes in expression levels throughout the IFN time course (Fig. 3A, Supplemental Table I). Genes in SC1 and SC2 increased in expression after 8 h. Genes in SC3 and SC4 were induced by 4 h and peaked at 4–8 h, with genes in SC3 returning close to baseline by 12 h and SC4 genes remaining elevated for ≥24 h. These two SCs were highly enriched for genes found in the INTERFEROME v2.0 database. Genes in SC5 increased slightly and peaked at 8 h, returning to baseline levels by 12 h. Levels of SC6 genes either increased or decreased by 4 h, followed by decreased levels at 12–24 h. SC7 gene expression levels decreased sharply by 8 h, followed by a partial recovery by 12 h and a further decrease in expression by 24 h. Finally, genes in SC8 exhibited a sustained decrease in expression levels starting at 8–12 h. These data demonstrate that IFN induces distinct subsets of genes that are characterized by differing temporal expression patterns.

To validate our RNA-Seq results, we used NanoString nCounter technology, which uses colorimetrically barcoded DNA probes for direct detection of mRNA without nucleic acid amplification. A customized nCounter CodeSet was designed containing 50 genes from several temporal SCs, with a focus on the ISG-rich SC3 and SC4 (Supplemental Fig. 1, Supplemental Table II). Temporal expression profiles using NanoString were generally similar to the RNA-Seq data (Fig. 3B). SC1 and SC2 contained genes with increased expression levels at the later time points. SC3 and SC4 had genes with peak expression levels at 8 h, with genes in SC4 exhibiting expression levels that remained elevated at 24 h. Genes in SC8 decreased in expression over time, with the lowest expression levels observed at 24 h.

We next used NanoString to compare temporal regulation of ISGs between bat and human cell lines. We first compared gene expression between PaKi and HEK293 cells, because both cell lines are kidney derived, but HEK293 cells responded poorly to type I IFN (Supplemental Fig. 1). After screening for robust IFN responses in other human cell lines, we chose human A549 cells for comparative studies (Fig. 4A). A striking difference was observed between expression profiles of SC1 and SC2 when comparing PaKi and A549 cells. Of the five selected genes in SC1 and SC2, none was significantly upregulated in IFN-treated A549 cells. Similarly, the decreased expression levels observed for bat genes in SC8 were not observed in A549 cells.

For further analysis, we chose to focus on previously reported ISGs, particularly those found in SC3 and SC4. When comparing changes in gene expression in SC3, the median FC of all genes in SC3 followed similar trends between bat and human cell lines (Fig. 4B). However, in SC4, genes from A549 cells exhibited higher fold induction and remained elevated at later time points compared with PaKi cells.

We next determined the time point at which we observed maximum gene induction. For SC3 and SC4, ∼80% of IFN-induced PaKi genes peaked at 8 h (Fig. 4C). However, A549 genes had a bimodal pattern within both SCs, with most genes peaking at 4–8 h and a smaller subset peaking at ≥12 h.

To identify potential differences in induction kinetics, we calculated the percentage of genes that were induced to ≥50% of maximum expression for each time point. More than 80% of PaKi genes in both SCs were induced by 4 h (Fig. 4D). A549 genes in SC3 behaved similarly, although a small subset of genes was induced by 2 h. However, ∼40% of A549 genes in SC4 were induced by 2 h, indicating faster induction of this subset of genes. Genes in this list include HERC5, IFI6, IFIH1, MX1, NLRC5, OAS2, OASL, and PARP12. Notably, PaKi cells express higher baseline levels of ISGs, such that by full induction at 8 h, they express similar or greater mRNA counts compared with A549 cells, despite faster induction in A549 cells (Fig. 4D, 4G, Supplemental Fig. 1).

Next, we calculated the percentage of genes that had recovered to <50% of maximum expression for each time point. In both SCs, recovery occurred earlier in PaKi cells, with most genes recovering by 16 h (Fig. 4E). In contrast, the expression levels for most A549 genes remained elevated throughout the time course. Together, these data suggest that the regulatory mechanisms governing IFN-mediated gene induction and maintenance of gene expression differ in each cell type, particularly with genes in SC4.

It was recently reported that black flying fox tissues have constitutive and ubiquitous expression of IFN-α, suggesting that cells from this species may be primed to inhibit viral infection as a result of constitutive expression of certain ISGs (32). Indeed, we observed higher overall ISG mRNA levels in unstimulated PaKi cells, particularly for SC4 genes (Fig. 4F). In addition, we found that more than half of SC4 PaKi ISGs were induced to higher maximum counts than corresponding A549 ISGs (Fig. 4G).

Our gene-expression profiling revealed several genes not previously reported to be ISGs. These included EMC2, FILIP1, IL17RC, OTOGL, SLC10A2, and SLC24A1 (Fig. 2A). It is unclear whether the induction of these genes is cell type or species specific. In addition, we observed IFN-mediated induction of RNASEL, which encodes a 2′-5′-oligoadenylate synthetase–dependent RNase. This protein exerts its antiviral effect by degrading viral RNA in response to 2′-5′-linked oligoadenylates, which are generated by the IFN-inducible oligoadenylate synthase (OAS) family of enzymes upon stimulation by dsRNA in the cytosol (37). In human cells, RNASEL is a constitutively expressed latent enzyme and is not IFN inducible (Fig. 4A, Supplemental Fig. 1) (38). Interestingly, we observed a dose-dependent induction of RNASEL in IFN-treated PaKi cells (Fig. 5A). Of note, similar mRNA expression levels of RNASEL were observed in unstimulated human and bat cell lines (Fig. 4F, Supplemental Fig. 1). In addition, we observed IFN-mediated RNASEL induction in brain-derived (PaBr) and lung-derived (PaLu) black flying fox cell lines (Fig. 5B), suggesting that IFN-mediated induction of RNASEL in the black flying fox is not cell type specific.

Next, we constructed RNASEL-deficient PaKi cells using CRISPR (39). Because of the lack of a bat-specific RNASEL Ab, we were not able to monitor RNASEL protein levels. However, we detected modifications to the RNASEL gene in the PaKi-KO bulk population compared with the parental WT population (Supplemental Table III).

To test whether RNASEL protein was functional, we activated RNASEL with poly(I:C) transfection after IFN or mock treatment and monitored RNA integrity (Fig. 5C) (40, 41). We observed rRNA degradation when cells were treated with poly(I:C) but not with IFN alone. Treating cells with IFN for 24 h to induce RNASEL expression before poly(I:C) transfection resulted in increased RNA degradation and accumulation of smaller products, suggesting increased RNASEL activity. The RNA degradation observed in WT cells was reduced in the bulk KO cells. Together, these data indicate that RNASEL is a functional RNase that, unlike its human ortholog, is IFN inducible.

To test whether the presence of RNASEL is important in the context of viral infection, we infected WT and bulk KO cells with YFV17D-Venus and quantified infectivity after one viral life cycle (Fig. 5D). RNASEL-KO cells were more permissive to infection at all doses used, suggesting that RNASEL is important for suppression of initial viral infection. To test whether RNASEL induction played a role in the protective effect of the IFN response, we treated PaKi WT and KO bulk cells with increasing doses of IFN-α for 24 h, followed by infection with YFV17D-Venus (Fig. 5E). Consistent with previous results, KO bulk cells were more permissive to infection than WT cells. In addition, KO bulk cells were resistant to the protective activity of IFN. IFN-α (100 U/ml) pretreatment resulted in 80% reduction of infection in WT cells but only 50% reduction in bulk KO cells. Together, these data suggest RNASEL plays a significant role in the inhibition of viral infection, particularly in the context of the IFN response.

This study aimed to identify IFN-stimulated transcripts in a cell line from the black flying fox. Transcriptional analysis revealed >100 genes induced in response to IFN-α. Most of these genes have been previously described as ISGs, suggesting strong evolutionary conservation of the ISG pool, as would be predicted by previous genomic studies of immune genes in the black flying fox (11, 16).

We have provided a framework of black flying fox ISGs, organized by early, mid, and late responses to type I IFN. Temporal expression profiling delineated two ISG pools based on unique temporal induction profiles. Although both SCs are characterized by similar peak mRNA levels and a subsequent decline by 12–24 h, SC4 contained some genes that remained elevated. These genes may offer residual antiviral protection, even when IFN signaling has returned to basal levels. In addition, many genes in SC4 have higher baseline and higher maximal induction levels in bat PaKi cells compared with human A549 cells, which could result in species-specific differences in susceptibility to viral infection. It remains unclear why only particular ISGs are differentially expressed in this context, but it would be interesting to determine whether there were unique features or functions of these ISGs that could explain their differing expression patterns. Compared with human A549 cells, bat PaKi cells have a more rapid decline in ISG levels, suggesting tightly regulated expression kinetics. The reason for this strict transcriptional regulation remains unclear, but such a mechanism may exist to prevent excessive inflammation in a highly metabolically active host (11).

We also identified several previously unrecognized or noncanonical ISGs, including RNASEL. Notably, while our manuscript was under review, another transcriptome study of IFN-treated black flying fox cells also uncovered RNASEL as an ISG (42). In addition, a transcriptomic study done in Jamaican fruit bats (Artibeus jamaicensis) reported that RNASEL levels are increased in the spleen following experimental infection with Tacaribe virus, indicating that RNASEL induction can be observed in certain bat species in vivo (43). The OAS/RNASEL pathway, through which OAS proteins use viral dsRNA to create short oligonucleotides that act as second messengers to activate the constitutively expressed latent enzyme RNASEL, is used to clear viral genetic material from the cell. As the result of cleavage of viral and cellular RNA, RNASEL activation can also lead to apoptosis of infected cells (44). In addition, the short RNA fragments created by RNASEL can potentiate the IFN response by activating the cytosolic RNA sensor RIG-I (45). The induction of RNASEL in response to IFN in bats may provide an additional layer of antiviral protection. Indeed, KO of RNASEL increased viral susceptibility of black flying fox–derived cells. Although induction itself does not result in significant nuclease activity, stimulation with poly(I:C) is sufficient to cause degradation of total RNA in the cell. Unlike in humans, in whom only the upstream OAS proteins are induced by IFN, bat cells and tissues induce both components of the OAS/RNASEL pathway, likely creating a more rapid and potent effect that would inhibit viral replication before extensive viral spread could occur. Induction of RNASEL could also be a way of attenuating the effect of viruses that affect RNASEL activity either by direct inhibition, as seen with the L* protein of murine Theiler’s virus (46), or via increased expression of an RNASEL inhibitor, as seen with HIV (47) and encephalomyocarditis virus (48).

There is some evidence of modest (∼2-fold) RNASEL induction in certain mouse cell lines (49, 50). However, one mouse cell line with low endogenous levels of RNASEL been shown to increase RNASEL levels ∼10-fold in response to high doses of IFN (51). Rodents are also important viral reservoirs (10), but additional studies are needed to determine whether IFN-mediated RNASEL induction plays a role in host–virus interactions in mice.

We acknowledge several limitations of this study. First, temporal kinetics analysis was performed between one bat kidney cell line and one human lung cell line. It is possible that some differences observed may be due to intrinsic differences between cell lines and not differences in species. Studies in multiple cell backgrounds from both species may help to determine whether kinetic differences between bat and human ISGs can be generalized to the whole species. Second, in using universal IFN in place of black flying fox IFN, we made the assumption that the IFN response would be comparable between both types of IFN. Although we observed a transcriptional profile that is expected for IFN-treated cells, we cannot rule out possible differences in downstream signaling kinetics or magnitude of the response between universal and bat-derived IFN. Nonetheless, a recent transcriptomic study done in PaKi cells using black flying fox IFN showed ISG profiles that were comparable to those observed in our study (42).

Overall, this work lays the foundation for future investigation into the potential unique features of the bat IFN response. Although this study focused on ISG induction, we know little about whether bat ISG-encoded effectors possess unique antiviral properties. Uncovering mechanisms of bat ISGs will provide insight into the innate immune responses of an important viral reservoir and may inform research and development of antiviral therapies.

We thank Jeanine Baisch and Cynthia Silverman (Genomics Core at Baylor Research Institute) for assistance with NanoString samples and the McDermott Center Sequencing Core and Bioinformatics Core for sequencing and analysis.

The authors have no financial conflicts of interest.