Functional comparisons of the virus sensor RIG-I from humans, the microbat Myotis daubentonii, and the megabat Rousettus aegyptiacus, and their response to SARS-CoV-2 infection

ABSTRACT Bats (order Chiroptera) are a major reservoir for emerging and re-emerging zoonotic viruses. Their tolerance toward highly pathogenic human viruses led to the hypothesis that bats may possess an especially active antiviral interferon (IFN) system. Here, we cloned and functionally characterized the virus RNA sensor, retinoic acid-inducible gene-I (RIG-I), from the “microbat” Myotis daubentonii (suborder Yangochiroptera) and the “megabat” Rousettus aegyptiacus (suborder Yinpterochiroptera) and compared them to the human ortholog. Our data show that the overall sequence and domain organization are highly conserved and that all three RIG-I orthologs can mediate a similar IFN induction in response to viral RNA at 37° and 39°C but not at 30°C. Like human RIG-I, bat RIG-Is were optimally activated by double stranded RNA containing a 5’-triphosphate end and required mitochondrial antiviral-signaling protein (MAVS) for antiviral signaling. Moreover, the RIG-I orthologs of humans and of R. aegyptiacus, but not of M. daubentonii, enable innate immune sensing of SARS-CoV-2 infection. Our results thus show that microbats and megabats express a RIG-I that is not substantially different from the human counterpart with respect to function, temperature dependency, antiviral signaling, and RNA ligand properties, and that human and megabat RIG-I are able to sense SARS-CoV-2 infection. IMPORTANCE A common hypothesis holds that bats (order Chiroptera) are outstanding reservoirs for zoonotic viruses because of a special antiviral interferon (IFN) system. However, functional studies about key components of the bat IFN system are rare. RIG-I is a cellular sensor for viral RNA signatures that activates the antiviral signaling chain to induce IFN. We cloned and functionally characterized RIG-I genes from two species of the suborders Yangochiroptera and Yinpterochiroptera. The bat RIG-Is were conserved in their sequence and domain organization, and similar to human RIG-I in (i) mediating virus- and IFN-activated gene expression, (ii) antiviral signaling, (iii) temperature dependence, and (iv) recognition of RNA ligands. Moreover, RIG-I of Rousettus aegyptiacus (suborder Yinpterochiroptera) and of humans were found to recognize SARS-CoV-2 infection. Thus, members of both bat suborders encode RIG-Is that are comparable to their human counterpart. The ability of bats to harbor zoonotic viruses therefore seems due to other features.

viruses) can cause severe disease in humans, e.g., SARS coronaviruses 1 and 2 (5)(6)(7) or Marburg viruses (8).The taxonomic order Chiroptera was recently divided into the two suborders Yangochiroptera and Yinpterochiroptera (9), which largely, but not entirely, overlap with the previous division into microbats (having the ability of echolocation) and megabats (large fruit eating bats), respectively.Although members of both suborders can host highly pathogenic viruses, they rarely show clinical signs of disease, indicat ing an ability to tolerate and resist infection to an unprecedented level (10).Infection tolerance is proposed to be mediated by a dampening of pro-inflammatory responses that would otherwise lead to tissue damage (11)(12)(13)(14)(15). Infection resistance is supposed to be due to special features of the antiviral type I interferon (IFN) system, which is in bats under strong positive selection (16)(17)(18)(19)(20).In line with this, some Yinpterochiroptera megabats exhibit elevated base levels of type I IFNs (21,22) or IFN-stimulated genes (ISGs) (23), whereas for others this was not the case but an expanded tissue distribution of the master IFN regulator IRF7, diversified IFN gene loci, or a set of nonstandard ISGs were detected (24)(25)(26)(27).Myotis "microbats" (Yangochiroptera) express uniquely high paralog numbers of the broadly antiviral IFN effectors Tetherin (BST2) and PKR (28)(29)(30).
Type I IFNs (IFN-α/β) are cytokines that constitute the first line of defense against viral infection (31).Several intra-and extracellular pattern recognition receptors (PRRs) are able to sense viral hallmark structures called pathogen-associated molecular patterns (PAMPs), e.g., genomic RNA (32), and initiate a signal transduction chain that leads to the upregulation of IFN genes, first of all IFN-β.The RNA sensor retinoic acid inducible gene I (RIG-I) is one of the most important PRRs for virus infection (33,34).It has three major domains: an N-terminal domain encompassing two caspase recruitment domains (CARDs), a central DExD/H box RNA helicase domain, and a C-terminal domain involved in ligand binding and regulation (35,36).RIG-I recognizes double-stranded RNA (dsRNA) structures-especially when containing a 5'-triphosphate moiety-as it is present in many RNA virus genomes (37)(38)(39)(40).RNA ligand-bound RIG-I then undergoes a conforma tional change to expose the N-terminal CARD domains, thus enabling interaction with the signal adaptor mitochondrial antiviral signaling protein (MAVS) and the eventual activation of the IFN transcription factor, interferon regulatory factor 3 (IRF3) (34).Newly produced and secreted IFN then binds to its cognate receptor on the cell surface to activate the expression of antivirally active ISGs, thus establishing a virus-resistant cellular state (31).
However, despite the importance of bats as major hosts for zoonotic viruses and the central role of RIG in antiviral defense, functional data on bat RIG-I proteins are entirely lacking.To fill this gap, we cloned and expressed RIG-I sequences from two bat species from "microbats" and "megabats", namely Myotis daubentonii (belonging to the Yangochiroptera) and Rousettus aegyptiacus (belonging to the Yinpterochiroptera), investigated their function in virus recognition, temperature dependence, PAMP binding and IFN induction, and compared them to human RIG-I.Moreover, we tested the involvement of the bat RIG-Is in recognition of SARS-CoV-2 infection.

Bat cells induce an innate immune response upon virus infection
We have previously shown that cell lines derived from M. daubentonii (MyDaNi) and R. aegyptiacus (Ro6E-J) are capable of responding to exogenously added pan-species IFN (29,49).Here, we investigated whether these cells could also produce endogenous antiviral IFN in response to virus infection.We employed a pair of genetically matched viruses that are either inducing IFN (La Crosse virus lacking the IFN suppressor NSs; LACVΔNSs), or blocking IFN induction (La Crosse virus expressing the IFN suppressor NSs; wt LACV) (50).Replication of the NSs-deleted virus mutant LACVΔNSs is severely reduced in IFN competent cells and animals (51).Besides the two bat cell lines, we employed simian Vero E6 cells, which lack type I IFN expression (52) as a negative control, and the IFN competent human A549 cells as a positive control.Each of the cell lines was infected with either of the two viruses at a low MOI (0.01) to ensure multistep growth, and incubated for 72 hours.Figure 1A shows that supernatants harvested from Vero E6 cells contained similar amounts of infectious viruses, indicating comparable replication.In the cell lines A549, MyDauNi and Ro6E-J, by contrast, yields of the LACV ΔNSs were reduced by approximately two orders of magnitude compared to the wt LACV.Thus, just as in the human positive control line A549, the two bat cell lines are restricting replication of the IFN-sensitive virus mutant, strongly suggesting that the bat cell lines have mounted an antiviral IFN response.
To directly compare how the bat cells activate genes for endogenous IFN and ISGs (including RIG-I), we performed RT-qPCR analyses for RIG-I (DDX58) as well as for a series of other antiviral marker genes.IFN-β thereby represents exclusively virus-dependent genes (29,53), similar to the chemokine CXCL10 (54) which is however also inducible by .Mx1 (MxA in humans) is established as an ISG that only reacts to IFN, but not directly to infection (56), whereas OAS1 can be induced by both virus infection and IFN (53).The A549, MyDaNi, and Ro6E-J cells were either infected with the strong IFNinducing Rift Valley fever virus (RVFV) mutant clone 13 (54) or treated with 1,000 U/mL pan-species IFN-α.Total RNA samples were taken 6 and 24 hours later and tested for mRNA levels of the mentioned marker genes using RT-qPCR.Upon virus infection, all three cell lines induced IFN-β as expected but with different kinetics (Fig. 1B).A549 and MyDauNi cells showed an initial peak at 6 hours post infection (p.i.), which then decreased to the later 24 hours p.i. time point while Ro6E-J cells had a delayed IFN-β induction which peaked 24 hours p.i. (Fig. 1B).Levels of viral RNA were however similar between the two bat cell lines, indicating true differences in IFN induction kinetics.RIG-I was upregulated in all three cell lines by clone 13 infection and by IFN-α treatment, but again the Ro6E-J cells reacted not as quickly to infection as the other two cell lines (see Fig. 1B).A similar Ro6E-J-specific pattern was observed for CXCL10, whereas all cells upregulated Mx1 and OAS1 in a similar manner (Fig. S1).These results indicate that the applied micro-and megabat cells are fully competent in launching an antiviral IFN response to viral infection or type I IFN treatment.

Amino acid sequence comparison of bat RIG-Is
As a first step towards functional characterization of bat RIG-I orthologs, we assembled the full-length RIG-I sequence from M. daubentonii, using the data from our transcrip tome studies (29,49).Figure 2 shows the amino acid sequences alignment of the human, M. daubentonii and R. aegyptiacus RIG-I.Overall the sequences are well conserved, with 93.2% to 94.6% similarity between the different species.However, both bat orthologs exhibit three apparent differences to human RIG-I (red asterisks in the figure), namely an insertion of one amino acid and a deletion of two amino acids in the hinge region between the CARDs and the helicase domain, and an insertion of five amino acids in the Hel-2i helicase subdomain which is responsible for auto-inhibition of RIG-I in the absence of an RNA ligand (35).Moreover, M. daubentonii RIG-I has a specific one-amino acid deletion in the hinge region (black asterisks in the figure).An extended multiple alignment that includes RIG-I amino acid sequences from 15 additional bat species confirmed the deletions and insertions identified with M. daubentonii and R. aegyptiacus RIG-I for the other members of their respective suborders (including the Yangochiropteraspecific one-amino acid deletion in the hinge region), with the only exception that Pipistrellus kuhlii species (Yangochiroptera) lacks the five amino acids insertion in the Hel-2i helicase subdomain (Fig. S2).
Using the SMART database (http://smart.embl-heidelberg.de/),we identified the two N-terminal CARDs followed by a DEXD-like helicase domain, a helicase domain and the C-terminal regulatory domain as being conserved between all three RIG-Is (Fig. S3).Thus, while the overall sequence and domain organization are highly conserved, the bat RIG-Is contain several specific indels.

Bat RIG-Is are functional
Attempts to knock down endogenous RIG-I in the bat cells or to find antibodies with sufficient cross-reactivity were unsuccessful (data not shown).We, therefore, constructed cDNA plasmids to express the three RIG-I orthologs, each equipped with an N-terminal FIG 2 Amino acid sequence alignment for human, micro-and megabat RIG-I.Identical amino acid residues are marked with a small asterisk (*), replacements with conserved (:), semi-conserved residues with dots (.), gaps with a dash (-).Above the alignment is the different domains of human RIG-I depicted based on Kowalinski et al. (35).RIG-I sequences derived from human, M. daubentonii, and R. aegyptiacus are designated as huRIG-I, myRIG-I, and roRIG-I, respectively.The indels distinguishing human and bat RIG-I sequences are marked by big red asterisks, and the Myotis-specific deletion is marked by a big black asterisk.

Full-Length Text
Journal of Virology 3×Flag epitope, in RIG-I-deficient cells.Then, we used reporter assays to test whether human HEK293 ΔRIG-I cells (39) could be transcomplemented with the cloned bat orthologs.The cells were transfected with the different RIG-I constructs [or a 3×Flag-con trol (CTRL) protein] together with plasmids encoding firefly luciferase under the control of the inducible human ISG54-promoter and a transfection control plasmids encoding Renilla luciferase under the control of the SV40-promoter.Since the ISG54 promoter is inducible by both virus infection and IFN, we also employed the exclusively virus-respon sive IFN-β-promoters of mouse and of R. aegyptiacus in parallel.At 24 hours after plasmid transfections, the inducible promoters were stimulated by supertransfection of the cells with the RIG-activating genome RNA of vesicular stomatitis virus (VSV).Following a further 16 hours of incubation, firefly (inducible promoter) and Renilla (transfection baseline for normalization) luciferase activities were measured.The normalized data in Fig. 3A show an absence of inducible firefly luciferase activity in the CTRL-transfected samples, as expected, whereas expression of the human RIG-I alone already resulted in a 30-400 fold induction, depending on the promoter.As the CTRL-transfected cells did also not respond after VSV RNA transfection, any specific activity detected when RIG-Is are overexpressed is due to transcomplementation.When the cells expressing human RIG-I had been stimulated with VSV RNA, overall induction levels were 1.5-to 2.5-fold higher than in the unstimulated counterpart (see Fig. 3A).Unlike human RIG-I, the bat RIG-I orthologs did not exhibit background promoter induction, but in presence of VSV RNA they stimulated the promoters by 20-to 300-fold for Myotis and 20-to 400-fold for Rousettus (see Fig. 3A).
To make sure that the differences in induction levels were not due to differences in RIG-I expression, we analyzed the cell lysates by immunoblotting.All three RIG-Is from human, Myotis and Rousettus showed comparable levels and the expected apparent molecular weight of approximately 100 kDa (Fig. 3B).Moreover, we could also success fully transcomplement RIG-I -/-mouse embryo fibroblasts with all three RIG-I orthologs, indicating that the RIG-I of humans and bats can rescue RIG-I deficiency irrespective of the species background (Fig. S4).
Although the promoter activations exhibited a wide variation which is most likely owed to the fact that the HEK293 ΔRIG-I cells (which are also quite sensitive) had to be transfected twice, these results demonstrate that the RIG-I orthologs cloned from microand megabat cells can be stimulated by viral RNA and are able to initiate an antiviral signaling that results in the transactivation of virus-responsive promoters.

Bat and human RIG-I show a similar induction pattern under different temperatures
The body temperature of bats is remarkably variable, from down to 11°C during sleep or torpor (e.g., in hibernation) to up to 41°C during flight (57).As the IFN system is known to be influenced by temperature (58), we wondered whether temperature might affect the activity of the bat RIG-I orthologs.To test this, we transcomplemented the HEK293 ΔRIG-I cells and allowed RIG-I expression for 24 hours at 37°C, but then placed the cells in incubators set to 30°C, 37°C, or 39°C at 1 hour before stimulation with VSV RNA, and kept them for another 16 hours at the different temperatures (the cells did not tolerate 41°C, data not shown).As shown in Fig. 4A, at 30°C none of the RIG-I orthologs could be specifically stimulated by VSV RNA.At 37°C, there was a robust activation as shown above, which was maintained at the temperature of 39°C.RIG-I expression levels were comparable at all temperatures, with an additional smaller band observed for Myotis RIG-I at 37°C and 39°C (Fig. 4B).The appearance of the band might indicate RIG-I degradation, enforced at higher temperatures.Taken together, these results suggest that the human and bat RIG-Is are functional at normal body temperature or higher but not at 30°C.

Bat RIG-I signaling via MAVS
Upon recognition of 5′-triphosphorylated dsRNA, RIG-I undergoes a conformational change to expose the CARDs for interaction with MAVS which, in turn, initiates the antiviral signaling.To investigate whether the bat RIG-I orthologs are also signaling via MAVS, we compared their ability to activate the ISG54-promoter reporter in HEK293 ΔRIG-I and HEK293 ΔMAVS cells.The cells with the respective genotype were transfected with the expression constructs and stimulated with VSV RNA as described.RIG-Idependent ISG54-promoter activation was only detected in the presence of MAVS (Fig. 5A), with comparable expression levels of the Flag-tagged RIG-I orthologs (Fig. 5B).Thus, the bat RIG-Is, like the human ortholog, are signaling via MAVS.

RNA ligand interaction
We investigated the interaction of the bat RIG-I orthologs with RNA.Firstly, we per formed pulldowns with the biotin-labeled dsRNA analog polyI:C (HMW) that is bound to streptavidin-beads.Cell lysates from HEK293 ΔRIG-I cells expressing either Flag-tagged control or RIG-I proteins (Fig. 6A, left panel) were mixed with the polyI:C-coated beads.To make sure that the binding was specific, we also incubated the lysates with either empty beads or added free VSV RNA as a competitor.After incubation, washing, and elution, the eluates were subjected to SDS-PAGE and immunoblot detection of the Flag epitope tag (Fig. 6A, right panel).None of the RIG-I orthologs bound unspecifically to the empty beads, but all exhibited a binding to the poly I:C-coated beads which could be efficiently outcompeted by free VSV RNA.
Next, we investigated whether the activation of the RIG-I orthologs depends on the 5′-triphosphate moiety and the double-strandedness of a viral RNA.To this aim, genomic RNA isolated from RVFV clone 13 particles were either mock, Calf Intestinal Alkaline Phosphatase (CIAP, 5′-phosphatase), RNaseIII (double-stranded RNA (dsRNA) specific endoribonuclease), or RNaseA treated.The RNAs were then transfected into HEK293 ΔRIG-I cells that had been transfected with the RIG-I expressing plasmids and the ISG54reporter plasmid.After 24 hours of incubation, ISG54 promoter activity was measured.Pretreatment of the RNA with CIAP more than halved the induction by all RIG-Is com pared to mock treated RNA and pretreatment with either of the two RNases reduced the induction level to background, irrespective of the particular RIG-I (Fig. 6B and C).Together with the polyI:C pulldown, the above results indicate that bat RIG-I, like human Representative data from six independent experiments are shown.n.s., non-significant; *P < 0.05.
and mouse RIG-I (32,34), are specifically binding dsRNA and that full activation by viral genome RNA requires both a 5′-triphosphate and double stranded RNA.

IFN induction by SARS-CoV-2 can be enabled by RIG-I orthologs
Rhinolophus megabats (suborder Yinpterochiroptera) are proposed to be reservoirs of SARS-coronaviruses including the pandemic SARS-CoV-2 (5).Infection of human lung epithelial cells with SARS-CoV-2 induces a certain level of type I IFN (59), which was mostly found to be mediated by MDA5, a PRR that is structurally and functionally related to RIG-I (60-62).However, also RIG-I was found to be involved in IFN and cytokine induction by SARS-CoV-2, either directly via antiviral signaling (63,64), or indirectly by controlling viral RNA synthesis down to non-inducing levels (65).We tested the involvement of our bat RIG-I orthologs for their ability to sense SARS-CoV-2 infection.
Transcomplementation experiments in human ACE2-HEK293 ΔRIG-I cells demonstrated that the RIG-I orthologs of humans and of R. aegyptiacus are indeed capable of inducing IFN-β mRNA synthesis in response to SARS-CoV-2 (Fig. 7A, top left panel).For the RIG-I of M. daubentonii, by contrast, the reaction to SARS-CoV-2 was not statistically significant.R. aegyptiacus RIG-I also raised mRNA levels of the chemokine CXCL10, but just by a factor below 2, whereas for the RIG-Is from the two other species CXCL10 induction was not statistically significant (Fig. 7A, top right panel).For both human and R. aegyptiacus RIG-I orthologs, the cytokine mRNA induction levels by SARS-CoV-2 were approximately five to more than tenfold lower than for the positive control virus clone 13.This is expected since clone 13 is a RVFV mutant devoid of any IFN antagonistic factor (66, 67), whereas SARS-CoV-2 is a wild-type virus expressing several proteins that counteract IFN induction (68,69).No differences in SARS or RVFV RNA (Fig. 7A, bottom panels) or protein levels (Fig. 7B) were observed between the negative control (3×Flag-ΔMx) and the RIG-I expressing samples, and expression of all three RIG-I orthologs was confirmed (Fig. 7B; Fig. S5).Thus, the RIG-I orthologs of both humans and the megabat R. aegyptiacus enable innate immune sensing of SARS-CoV-2, whereas for the RIG-I of the microbat M. daubentonii we could not measure a statistically significant effect.However, expres sion of the various RIG-I orthologs also induced MDA5 (Fig. 7C).Although the levels of MDA5 induction were comparatively low, overexpressing RIG-I orthologs could not entirely clarify whether the IFN response to SARS-CoV-2 was owed to the respective RIG-I orthologs, or rather to the endogenous MDA5 they are inducing.

MDA5 is dispensable for SARS-CoV-2 innate immune sensing
To investigate whether the RIG-I-dependent IFN induction by SARS-CoV-2 could be indirectly mediated, we abrogated MDA5 expression by siRNA treatment before transfection of ACE2-HEK293 ΔRIG-I cells with the two SARS-CoV-2-reactive RIG-I orthologs (human and R. aegyptiacus).As shown in Fig. 8A, the IFN induction in response to SARS-CoV-2 occurred irrespective of whether cells were treated with the control siRNA or with the MDA5 siRNA, as long as human or R. aegyptiacus RIG-I were present.The previously seen (low) effect of SARS-CoV-2 on CXCL10 induction could not be repro duced in the siRNA-transfected cells (Fig. S6A).Again, even in the RIG-I expressing cells the MDA5 levels were comparatively low and suppressed by the specific siRNA as expected (Fig. 8B; Fig. S6B through D).Moreover, viral RNA levels including those of SARS-CoV-2 were not influenced by the overexpressed RIG-I orthologs (Fig. S6E and F).Thus, in our system MDA5 seems not to contribute to the RIG-I-mediated IFN induction by SARS-CoV-2.
Overall, these experiments demonstrate that the RIG-I orthologs of R. aegyptiacus and humans, but not of M. daubentonii, are indeed capable of sensing infection by SARS-CoV-2.

DISCUSSION
RIG-I is responsible for one of the earliest steps of the innate immune response, the recognition of intruding viral RNA and the induction of antiviral IFNs (34,40).Since bats are a major reservoir for zoonotic viruses and suspected to have a special IFN system (1, 3, 5), we set out to deep-characterize RIG-I orthologs from one species each of the microbats (M.daubentonii, suborder Yangochiroptera) and the megabats (R. aegyptiacus, suborder Yinpterochiroptera), and compare them to human RIG-I.
In terms of gene expression, we could stimulate mRNA synthesis of the bat RIG-I orthologs in parental cells by type I IFN treatment as well as by virus infection, results that are in line with previous findings (29,(41)(42)(43)(44)(45)(46)48).The amino acid sequences derived from cloned RIG-I cDNAs were more than 90% similar to each other and compared to human RIG-I (with the notable exception of small indels) and the predicted domain structure (34) was obviously conserved.
Our functional characterization of the M. daubentonii and R. aegyptiacus RIG-I orthologs suggest that bat RIG-Is are not substantially different from the human counterpart.All three RIG-I orthologs were capable of binding dsRNA, are in their function partially dependent on the 5'-triphosphate RNA end, are signaling via MAVS, and can trigger the induction of IFN and other cytokines in response to viral RNA or virus infection.In the overexpression experiments, neither protein levels nor dsRNA binding or 5′-triphosphate end dependency were substantially different.Nonetheless, the RIG-I of M. daubentonii, exhibited in most settings a somewhat reduced activity, whereas R. aegyptiacus RIG-I tended more often to perform comparably to human RIG-I.Importantly, the reduced activity by M. daubentonii RIG-I occurred in both a human and a mouse cell background.When parental bat cells were stimulated with a RIG-I-inducing virus, by contrast, neither the antiviral response in total nor IFN induction per se was much different between the M. daubentonii and R. aegyptiacus RIG-I.Thus, Myotis RIG-I may not possess a systematically lower antiviral signaling capability, but rather need a cofactor that is only present or only fitting in parental cells and regulates a signaling step downstream of RNA recognition.The most obvious possibility would be MAVS as the prime interactor and downstream antiviral signal transmitter.Further studies are however necessary to clarify this.
As RIG-I is a key PRR for the very first step of the IFN response (34), it is often targeted by viral counterstrategies (68)(69)(70)(71).Some highly pathogenic viruses modify their 5′ genome ends to avoid recognition by RIG-I (72, 73), whereas others encode IFN antagonists that directly interact (74)(75)(76)(77)(78)(79) or induce its degradation (80,81).Also the bat-borne Nipah virus as well as SARS-CoV-2 express proteins with such activities (78,82,83).Moreover, it is known that species-specific differences in the matching of viral IFN antagonists to their cellular targets correlate with host range or virulence (84,85).It will therefore be interesting to see whether and to what extent the Myotis and Rousettus RIG-Is we characterized as well as human RIG-I can be inhibited by antagonists from bat viruses, and whether such interactions take place in conserved domains that are also present on human RIG-I.Testing viral RIG-I antagonists for potential broad-species IFN antagonism may help to assess the zoonosis risk of newly identified bat viruses.
The involvement of RIG-I in IFN induction by SARS-CoV-2 is still not entirely clarified.While most reports concluded that MDA5 rather than RIG-I is the relevant PRR (60)(61)(62), two studies reported it is additionally RIG-I-dependent (63,64).Moreover, a direct antiviral effect of RIG-I against SARS-CoV-2 was shown (65), and the RIG-I antagonistic activity of SARS-CoV-2 proteins also indicates a certain relevance for the virus (82,83).In our hands, both human and R. aegyptiacus RIG-I enabled HEK293 ΔRIG-I cells to induce IFN in response to SARS-CoV-2 infection and in an MDA5-independent manner.RIG-I, therefore, seems to be another PRR sensing SARS-CoV-2 under certain conditions, which is in line with the results by Arai et al. who identified RIG-I-activating small virus RNAs produced excessively by SARS-CoV-2, but not by the common cold coronaviruses OC43 and 229E (63).Besides this, it is interesting that the RIG-I of R. aegyptiacus, but not of M. daubentonii, is reacting to SARS-CoV-2 as Rousettus belongs to the same suborder Yinpterochiroptera as Rhinolophus, the major host of SARS-coronaviruses (5).
Unlike humans, bats have to endure extreme changes in body temperature when they switch between torpor, sleep, and flight (57,86).Torpor is a state of reduced metabolic activity, applied, e.g., during hibernation (87,88).Despite the principal differences between humans and bats regarding temperature regulation, we did not observe any species-specific activity of the respective RIG-Is that would have been temperature dependent.The two bat and the human RIG-I orthologs showed RNAresponsive activity at 37°C and at 39°C, while at 30°C there was only background activity (which was highest for the human ortholog).We cannot exclude a dominant influence of the human cell background, i.e., that cellular factors of the antiviral signaling chain other than RIG-I are shutting down at lower temperatures.Our observations, however, are in agreement with studies showing a reduced and delayed IFN system activity at lower temperatures in human, primate, and mouse systems (58) and the lack of any transcriptional response in virus-infected Myotis myotis cells at conditions simulating torpor (89).Our data showing the absence of RIG-I activity at low temperature are also compatible with the report that hibernating bats elicit very little immune responses to a fungal pathogen (86).Taken together, there appears to be a relative immune dormancy at low body temperature, which is expected to render the animals more susceptible to virus infections.Therefore, torpor, which also occurs in tropical bats as a response to environmental heat (87), may contribute to the role of bats as virus reservoir.
The antiviral IFN system of bats has attracted great attention as it might be the key to understand the role of these animals as virus reservoirs.The results presented are corroborating previous studies and predictions on the high similarity of bat RIG-I orthologs with those of other mammals regarding sequence, domain structure, and regulation.Our functional data on dsRNA binding and the dependency on 5′-triphos phate ends, MAVS, and temperature, as well as on the recognition of SARS-CoV-2 infection do not indicate substantial differences between the RIG-Is of humans and bats.This insight complements recent observations that the antiviral IFN effectors Tetherin and PKR of bats are basically functioning the same way as their human counterparts (28,30).On the other hand, however, compared to humans some bats encode multiple copies of the genes for Tetherin and PKR as well as for IFNs or express non-standard ISGs (24)(25)(26)(28)(29)(30).Thus, it could be speculated that the IFN system of bats has functionally conserved key factors but is probably enforced by increased copy numbers of antiviral genes or novel ISGs.
In any case, it is hoped that the data and tools we generated will enable further insights into the role of the antiviral IFN system for the unique capacity of bats to harbor a huge range of different human pathogenic viruses without falling prey to them.

Virus titration
Supernatants from cells that had been infected with LACV wt or LACV ΔNS were collected and cleared by centrifugation at 800 × g for 5 minutes.The supernatants were titrated on Vero E6 with an overlay medium containing MEM (Gibco, 21935028) supplemented with 10% fetal bovine serum, 2 mM glutamine, 120 U/mL penicillin, 100 g/mL streptomycin (Gibco, 10378016), and 1,5% Avicel [FMC BioPolymer, (91)].After 48 hours of incubation, the medium was removed and the cells washed with PBS before fixing with PBS-4% paraformaldehyde (Roth, 0335-4) for 24 hours at 4°C.Then, the fixed cells were again washed with PBS and permeabilized with PBS-0,1% Triton X-100 (Sigma, T9284-500ML) for 20 minutes.The cells were again washed and then incubated for 16 hours at 4°C with anti-LACV N (1:1,000, kind gift from Georg Kochs, University or Freiburg, Germany) in TBS-T buffer containing 1% non-fat milk powder.After washing the cells with PBS, they were incubated with secondary antibody IRDye800 conjugated anti-rabbit (1:10,000, Rockland, 611-132-122) and DRAQ5 (1:10,000, eBioscience, 65-0880-92) in TBS-T-1% milk powder at room temperature.Finally, the wells were washed first with PBS and then with H 2 O, and the fluorescence signals were detected and the foci counted using the Odyssey instrument (LI-COR).

cDNA cloning
Primers for cloning human, M. daubentonii and R. aegyptiacus RIG-I, were gener ated using the In-Fusion cloning tool (https://www.takarabio.com/learning-centers/cloning/primer-design-and-other-tools) and ordered from Eurofins Genomics.The full-length sequences of RIG-I (DDX58) from M. daubentonii was assembled from previous data (29) while the human and R. aegyptiacus RIG-I sequences were available as GenBank entries (NM_014314.4 and XM_016130339.2,respectively).Primers were designed for cloning the cDNAs into the vector pI.18 with a 5′ 3×Flag-tag sequence added to the forward primer.Human RIG-I fwd: 5′ TGA CAC GAT CGG ATC CAT GGA CTA CAA AGA CCA TGA CGG TGA TTA TAA AGA TCA TGA TAT CGA TTA CAA GGA TGA CGA TGA CAA GAC CAC CGA GCA GCG ACG C 3′ and rev 5′ TCT AGA ATT CCT CGA GTC ATT TGG ACA TTT CTG CTG GA 3′; M. daubentonii and R. aegyptiacus RIG-I: fwd 5′ TGA CAC GAT CGG ATC CAT GGA CTA CAA AGA CCA TGA CGG TGA TTA TAA AGA TCA TGA TAT CGA TTA CAA GGA TGA CGA TGA CAA GAC GGC CGA GGA GCG GCG G 3′, M. daubentonii rev 5′ TCT AGA ATT CCT CGA GTC ATT TGG ACA TTT CTG CTG GAT C 3′, R. aegyptiacus rev 5′ TCT AGA ATT CCT CGA GTC ATT TGG GCA TTT CTG CAA CAT CG 3′.cDNAs were generated from RNAs isolated from A549, MyDauNi, and Ro6E-J cells that were treated for 16 hours with 1,000 U/mL IFN-α (B/D) (PBL Assay Science).The RNAs were isolated using RNeasy Mini Kit (Qiagen, Cat No./ID: 74106), and 1 µg was used for cDNA synthesis with the PrimeScript High Fidelity RT-PCR Kit (Takara, R022B).Of the cDNA, 2 µL were used as PCR template for amplification with the KOD Polymerase (Calbiochem, 71086-3).The PCR products were cloned into the pI.18 vector, digested with BamHI (NEB, R3136S) and KpnI (NEB, R3142S) using the In-Fusion Kit (Takara, 638911).The ligated product was transformed into Stellar competent cells accompanying the In-Fusion Kit, and spread onto agar plates at 37°C for 16 hours before colonies were picked for DNA isolation.Correctness of the inserts was confirmed by sequencing.

Isolation of genomic VSV and RVFV clone 13 RNA
Genomic RNAs were isolated from VSV or RVFV clone 13 particles by PEG800-precipita tion and phenol-chloroform extraction as described (72).Briefly, the supernatant from a T175 flask infected with VSV or RVFV clone 13 (MOI 0.001) was collected and cleared by centrifugation at 800 × g for 5 minutes, after 48 hours (VSV), or 72 hours (RVFV) of incubation.The cleared supernatant was then mixed with 10.8 mL PEG8000 buffer (30% (wt/vol) PEG 8000, 10 mM Tris-HCl (pH 7.5), 1 mM EDTA, 100 mM NaCl). and 1.5 mL 5M NaCl followed by incubated for 1 hour at 4°C with head-over-tail rotation.The solution was then centrifugated at 4,600 × g for 1 hur at 4°C and the resulting pellet dissolved in 1.5 mL peqGOLD TriFast (peqlab, 30-2030).After vortexing and incubation at room temperature for 5 minutes, a phase-separation was performed by centrifugation at 4,600 × g for 11 minutes.The upper aqueous phase was collected and mixed with 700 µL chloroform (Roth, 7331) and glycogen (Roche, 10 901 393 001).The RNA was then precipitated at −20°C for 16 hours and pelleted by centrifugation at 4,600 × g for 33 minutes at 4°C.The pellet was washed twice with 70% Ethanol (Roth, 9065.4) and then dissolved in H 2 O to a final concentration of 250 ng/µL.

Poly I:C pull down
A total of 2 × 10 5 HEK293 ΔRIG-I cells per well were seeded in 6-well plates.After 24 hours, the cells were transfected with 100 ng/well of either pI.18-3×Flag-ΔMx (CTRL), pI.18-3×Flag-huRIG-I, pI.18-3×Flag-myRIG-I, or pI.18-3×Flag-roRIG-I using GeneJammer (Agilent, 204130) transfection reagent.After 16 hours of incubation, the cells were scraped off in PBS and pelleted by centrifugation at 800 × g for 5 minutes.The cells were then lysed in lysis buffer [0.5% Triton X-100 (Sigma, T9284-500ML), 1× Protease inhibitor (c0mplete, Roche, 4693116001) in PBS] for 10 minutes at 4°C. Cell debris were removed by centrifugation at 15000 × g, 10 minutes, 4°C and the supernatants used for pull down.As input from each reaction, 5% of the volume was put aside.Dynabeads M-270 Streptavidin (Invitrogen, 65305) (250 µg/IP) was coupled to 100 ng/IP Poly(I:C) (HMW) Biotin (Invivogen) according to manufacturer's instructions.After washing, the beads were re-suspended in the cell lysates and incubated at 4°C with head over tail rotation for 16 hours.To compete for RIG-I binding, 10 µg of VSV RNA was added to the lysate before incubation.After incubation the beads were washed three times in lysis buffer and the bound fraction eluted in 0.5% Triton X-100 in PBS mixed 4:1 in 4 × sample buffer supplemented with phosphatase and protease inhibitors.The samples were then boiled and the supernatant used for immunoblot analysis.

Statistical analysis
Two-tailed paired t tests were done using GraphPad Prism (version 7.05, GraphPad Software Boston, USA) based on the logarithmic values for viral growth assays while all other comparisons were made on fold-induction over control values.

FIG 3
FIG 3 Transcomplementation of RIG-I-deficient human cells by bat RIG-I orthologs.(A) HEK293 ΔRIG-I cells were transfected with plasmids encoding 3xFlag-ΔMx control protein (-), or 3xFlag-tagged hu-, my-, or roRIG-I together with firefly-luciferase expressing plasmids under the control of the indicated innate immune promoters and Renilla-luciferase under SV40-control promoter.After 24 hours, the cells were stimulated with VSV genomic RNA for 16 hours.Upon harvesting the cells, the firefly/Renilla luciferase activities were measured.The promoter activation by human RIG-I in the absence of VSV RNA is most likely due to the fact that cells were incubated for 40 hours after transfection of the RIG-I expression construct.Graphs show data points for fold induction over the untreated negative control (column 1), with mean values and standard deviations from seven independent replicates.(B) Immunoblot analysis was performed with antibodies against the indicated antigens.Representative data from five independent experiments are shown.*P < 0.05.

FIG 4
FIG 4 Temperature dependent IFN induction by exogenously expressed human and bat RIG-I orthologs.(A) HEK293 ΔRIG-I cells were transfected, stimulated and assayed as described for Fig. 3, but incubated at different temperatures as indicated.Each graph shows the results from six independent replicates.(B) Immunoblot analysis was performed with antibodies against the indicated antigens.Representative data from three independent experiments are shown.*P < 0.05.

FIG 5 9 FIG 6
FIG 5 MAVS-dependent antiviral signaling by human and bat RIG-I orthologs.(A) HEK293 ΔRIG-I or ΔMAVS cells were transfected, stimulated, and assayed as described in Fig. 3.The graph shows the results from six independent replicates.(B) Immunoblot analysis was performed with antibodies against the indicated antigens.The expected size of the 3xFlag-ΔMx control (−) and RIG-I proteins are indicated with arrows.Representative data from five independent experiments are shown.

FIG 7 12 FIG 8
FIG 7 Innate immune response to SARS-CoV-2 in cells overexpressing RIG-I orthologs.(A) ACE2-HEK293 ΔRIG-I cells were transfected with the indicated plasmids for 24 hours, followed by infection with either SARS-CoV-2 or RVFV clone 13 (MOI 1).After 16 hours, total RNA was isolated and used for detection of IFN-β and CXCL10 mRNA, SARS E RNA and RVFV L RNA by RT-qPCR.The graph shows the results from six independent replicates.(B) Immunoblot analysis with antibodies against the indicated antigens.Representative data from six independent experiments are shown.(C) RT-qPCR analysis for MDA5 mRNA with the samples used in (A).ns, non-significant; *P < 0.05.