Transcriptomics of primary cells derived from three Myotis yumanensis individuals after immune stimulation unveil GBP5 as a top ISG

To better characterize cell-autonomous innate immunity in bats, we performed a transcriptomic approach in primary cells from three individuals of Myotis yumanensis under type I interferon (IFN) stimulation (Fig 1A–1C). We choose this species because it has genome-wide evidence of adaptations in response to viral infections and exceptional genome sequence and annotation [13]. Briefly, primary fibroblast-like cells were derived from wing punches of three M. yumanensis bat individuals in the US Southwest (see Materials and methods). Following previous studies on the IFN stimulation of bat cell lines [22–24], we stimulated the primary cells with, or without, universal type I IFN for 6 hours. We then isolated total cellular RNA for sequencing (RNAseq) and subsequent bioinformatic analyses. We used Salmon [25] to quantify M. yumanensis gene expression. The genome of M. yumanensis is one of the best annotated genomes of all mammals, with BUSCO scores of 94.5% and around 19,000 annotated protein-coding genes, leaving few unannotated regions [13]. This allowed extremely high quality of our RNAseq with one-to-one (human and Myotis) ortholog analysis. Through pairwise comparisons with DESeq2 [26], we identified genes that were significantly upregulated (n = 1,036) and downregulated (n = 859) by a log2 fold change (FC) ≥ 2 and with an adjusted p value < 0.05, and which we defined as significantly responding to type I-IFN (Fig 1D). Amongst the significantly upregulated genes, we found several innate immune factors, indicating that I-IFN treatment led to an efficient activation of the innate immune responses in these primary cells. Specifically, among the most differentially expressed genes, we identified some involved in the innate immune signaling pathways (e.g., IRF7, IFIH7), cytokines (e.g., CXCL10) and antiviral ISGs, such as ISG15, ISG20, OAS1, OASL, Mx2, or RSAD2/Viperin (Fig 1D, 1E; Fig 1E highlights counts per gene in each condition and per M. yumanensis individuals). Notably, GBP5 was the most differentially expressed ISG in M. yumanensis primary cells compared to the unstimulated control in which GBP5 is not expressed (Fig 1D, 1E; of note, the lower significance for GBP5 in Fig 1D was solely due to more variations in differential expression between cells derived from different Myotis individuals, which do not express any GBP5 at basal level, similarly to CXCL10 or RSAD2, as shown in Fig 1E), an interesting finding in light of the broad antiviral activities recently ascribed to this factor.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. GBP5 is the most differently upregulated ISG in Myotis yumanensis primary cells upon type I interferon immune stimulation.

A, Photo of a Myotis yumanensis bat at a sampling site (Credit: Elise Lauterbur). B, Primary cells derived from a biopsy wing punch. C, Schematic of the workflow from bat sampling to RNA sequencing analyses. Bat primary cell lines were similarly derived from three M. yumanensis individuals and treated with, or without, universal type I IFN to trigger ISG expression. Total RNA was extracted and sequenced, followed by analyses to identify differentially expressed genes between stimulated cells compared to control cells. D, Volcano plot representing the differential gene expression between untreated and IFN-stimulated cells from three M. yumanensis individuals. Genes significantly (p < 0.05) differentially expressed (DE) are colored in red. Several known ISGs are highlighted on the graphic. GBP5 is the most upregulated ISG. E, Expression of ISGs from panel D in M. yumanensis primary cells derived from three individuals in control (light gray) and I-IFN (dark gray) conditions. TPM, Transcripts per million counts (log10 scale; complete table in S1 Dataset). F, Comparative analysis of differential gene expression (Log2 Fold Change) of GBP5 upon universal type I IFN treatment M. yumanensis cells as compared to six other mammalian species cells in similar experimental setup ([22]; details in Materials and methods). Pictos from https://www.phylopic.org/ and Illustrations from NIAID NIH BioArt Source (https://www.bioart.niaid.nih.gov/). The data underlying this Figure can be found in S1 and S2 Dataset. See Data availability for access to code.

https://doi.org/10.1371/journal.pbio.3003760.g001

To decipher whether this was typical of mammalian GBP5 expression upon interferon stimulation, we performed a comparative analysis. We used publicly available results from a transcriptomic study defining the interferome of cells among mammalian species treated in similar conditions [22]. We found that GBP5 was also an ISG in six other mammalian species (Fig 1F). Yet, the upregulation of GBP5 transcripts in Myotis yumanensis primary cells seemed remarkably high compared to the others, including its closest related species with transcriptomic information, Myotis lucifugus (1.5-fold higher; Fig 1F).

GBP5 has evolved under episodic positive selection during mammalian evolution

Antiviral factors that are important in vivo have often evolved under recurrent positive selection, as a consequence of their engagement in molecular arms races with pathogenic viruses [2,27]. We therefore characterized the evolutionary history of GBP5 in mammals, using phylogenetic and positive selection analyses. First, we retrieved GBP5 homologous sequences in mammalian genomes from public databases using human GBP5 as a query (details in Materials and methods). We further obtained de novo sequences of bat GBP5s from closely related Myotis species that we sampled in the wild and from a cell line of Eptesicus fuscus (see Materials and methods). In total, we analyzed 348 nucleotide sequences of mammalian orthologous GBP5s from the main orders: Chiroptera (n = 55), Rodentia (n = 80), Primates (n = 37), Perissodactyla (n = 9), Artiodactyla (n = 97), Lagomorpha (n = 10), and Carnivora (n = 51), as well as other mammalian sequences that were used as outgroups: Proboscidea (n = 3), Pilosa (n = 1), Cingulata (n = 4), and Xenarthra (n = 1). We then codon-aligned the GBP5 coding sequences with PRANK [28] and performed phylogenetic analyses with IQ-TREE [29]. The GBP5 gene tree seemed to follow the accepted mammalian species tree (Fig 2A), with some exceptions due to gene duplications/losses. For example, we confirmed an ancient duplication event in Lagomorpha [30] and the loss of GBP5 in primate Old World Monkeys [31,32]. We also detected duplication events, indels and high variability in sequences during rodent and bat evolution (Fig 2A, See Data Availability for alignments and phylogenetic details). To identify whether, and when, episodic positive selection of GBP5 has occurred during mammalian evolution, we used a branch-specific method, adaptive Branch-Site Random Effects Likelihood (aBSREL) [33]. aBSREL tests whether a given lineage has undergone significant positive selection and estimates the ω ratio, which is the ratio of the nonsynonymous to synonymous substitution rates (dN/dS). We found evidence of episodic positive selection in GBP5 during mammalian evolution, in particular in rodents, primates, lagomorphs, and bats (Fig 2A). We further analyzed each mammalian order independently, using BUSTED [34] from HyPhy [35]. We confirmed that mammalian GBP5 has evolved under positive selection in several orders (Fig 2B), particularly during Rodentia and Chiroptera evolution, as well as in Artiodactyla and Lagomorpha.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Mammalian GBP5 has evolved under episodic positive selection, with strong diversification in rodents and bats.

A, Phylogenetic analysis of GBP5 across mammals. The phylogeny was built from a PRANK [28] codon alignment of 348 GBP5 homologous sequences. Maximum likelihood phylogenetic tree was built with IQ-TREE [29] with 1000-bootstrap replicates for statistical support (see Data availability). Branches under significant positive selection (p-value < 0.05) assigned by aBSREL are thickened and in red. The scale bar indicates the number of substitutions per site. B, Evidence of positive selection in GBP5 from several mammalian orders. Positive selection analyses were performed with BUSTED [34] using a PRANK codon alignment for each mammalian order. OWM, Old world monkeys. Species silhouettes are from https://www.phylopic.org.

https://doi.org/10.1371/journal.pbio.3003760.g002

The data underlying this Figure can be found in S2 Dataset, the alignments for phylogenetic analyses are all openly available in FigShare Dataset (https://doi.org/10.6084/m9.figshare.26180785.v1, https://doi.org/10.6084/m9.figshare.26180764.v1, https://doi.org/10.6084/m9.figshare.26180803.v1, https://doi.org/10.6084/m9.figshare.26180761.v1, https://doi.org/10.6084/m9.figshare.26180767.v1, https://doi.org/10.6084/m9.figshare.26180776.v1, https://doi.org/10.6084/m9.figshare.26180794.v1, https://doi.org/10.6084/m9.figshare.26180797.v1).

GBP5 is under strong diversifying evolution in bats at the genomic and genetic levels

Because we found evidence of positive selection in bat GBP5 at the gene-wide level, as well as pairwise-amino acid sequence divergence that could reach 20% within Chiroptera (S1 Fig), we performed comprehensive evolutionary analyses in this order. To identify when positive selection has occurred during bat evolution, we used aBSREL [33] from the Chiroptera-specific GBP5 alignment. We found that several bat lineages were the targets of intensive episodic positive selection, including the branch at the origin of the Microchiroptera suborder, as well as other internal and terminal lineages spread-out in the Yangochiroptera and Yinpterochiroptera clades (Fig 3A, 3B). This indicates differential selective pressure during bat evolution and may represent lineage-specific adaptations driven by past viral epidemics.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. GBP5 has evolved under strong positive selection in bats.

A, Phylogenetic and positive selection analyses of bat GBP5. Maximum likelihood phylogenetic tree of bat GBP5 was built with IQ-TREE and statistical support is from 1,000 bootstrap replicates (values are shown below branches). Branches under significant positive selection (p-value < 0.05) assigned by aBSREL are in red and the corresponding estimated values of ω are reported in panel B. The scale bar indicates the number of substitutions per site. The black arrows identify species with GBP5 genes that were functionally tested in this study. 1.1 and 1.2 identify GBP5 duplicates within a given bat species. B, Evidence of lineage-specific positive selection during bat GBP5 evolution. aBSREL identified at least six branches under significant positive selection. ω1 and ω2, estimation of ω in the rate class not allowing positive selection, and allowing positive selection (ω > 1), with % of sites in this class in parentheses, respectively. C, Sites under positive selection (PS) in bat GBP5. Site-specific positive selection analyses were performed using FUBAR [36], MEME [37], and FEL [38] from HYPHY/Datamonkey [35,39]. Only the sites above the indicated “statistically significant cut-off” are shown (PP, posterior probabilities for FUBAR; p-value for MEME and FEL). In bold are the sites identified by several methods. Nb of PSS, number of positively selected sites. “PS sites” numbering is according to the codon numbering in the PRANK codon alignment. D, Schematic representation of GBP5 with its functional domains and the herein identified sites under positive selection (red arrows at the top). LG domain, large GTPase domain. MD, Middle domain. GED, GTPase effector domain. The size of the domains is not to scale. Residues identified by at least two positive selection methods are highlighted in bold. In blue, sites or motifs involved in known functions in human GBP5. The bat phylogenetic tree from Fig 3A can be found in https://doi.org/10.6084/m9.figshare.30924512 and all results from positive selection analyses in https://doi.org/10.6084/m9.figshare.30924551.

https://doi.org/10.1371/journal.pbio.3003760.g003

Beyond evidence of lineage-specific positive selection, we identified multiple, independent duplication events in bat GBP5, specifically in Phyllostomus hastatus, Antrozous pallidus, Corynorhinus townsendii, and Ia io (Fig 3A), which diverged a long time ago (Fig 3A). Despite all the duplication events being recent (i.e., only at terminal branches), we found important divergence between the copies (Figs 3A and S1).

Furthermore, we carried out several analyses to specifically identify the sites that have evolved under positive selection. Using three maximum likelihood methods from HyPhy (see Materials and methods), we found many positively selected residues in bat GBP5 (Fig 3C). Conservatively, considering only residues identified by at least two methods, we identified 16 positively selected sites—of note, if we consider sites identified by at least one method, we have 30 sites total. The rapidly evolving residues are found in all GBP5 protein domains, except for the hinge domain, which contains no positively selected sites (Fig 3D). Overall, we identified a very high concentration of sites under positive selection, as well as numerous indels (insertion/deletion) and lineage-specific early stop codons in the C-terminal part of bat GBP5.

Taken together, our transcriptomic and phylogenomic approaches (Figs 1–3) indicate that GBP5 could be an important innate immune factor engaged in a molecular conflict in bats. This suggests that GBP5 could play important and lineage-specific roles against various pathogens and that past epidemics may have shaped bat innate immunity.

Natural variation in subcellular localization of bat GBP5

To test how past genetic diversification may have impacted GBP5 functions, we focused on 10 bat species (indicated by the black arrows in Fig 3A) that represent: (i) a large panel of orthologous GBP5 sequences encoding for maximum amino acid variations at the sites identified under positive selection (S2 Fig) and (ii) covering various bat families of different clades—some of which are very distant from each other while others are very close (e.g., Myotis spp.). This approach allowed us to functionally cover the evolution of bat GBP5 over a period of around 60 million years, at both ancient and more recent times.

In humans, the localization of GBP5 at the TGN is absolutely required for its antiviral activities [16–18]. Its localization is driven by a prenylation sequence present at the C-terminal end of the protein (CaaX motif) that anchors GBP5 at the TGN membrane. Synthetic mutation in this domain prevents both TGN localization and antiviral activities (Homo sapiens GBP5-C583A) [16,40].

To study the subcellular localization of bat GBP5, we cloned the 10 selected orthologous bat GBP5s (Fig 3A) with an HA tag in a eukaryotic expression vector, and compared them to human GBP5 (wt), a prenylation-defective mutant (human GBP5-C583A) and a control vector (i.e., not expressing any GBP5). We ectopically expressed GBP5 in TZM-bl cells and imaged the cells 48 hours later by confocal immunofluorescence microscopy. We found colocalization of GBP5 with the TGN46 marker for 8 out of 10 bat species (Fig 4), showing GBP5 localization to the TGN similarly as human GBP5. Surprisingly, two closely related bat species Eptesicus fuscus and Pipistrellus kuhlii exhibited a diffused subcellular distribution in the cell, similarly to what we observed for the Homo sapiens-C583A mutant (Fig 4). Thus, bat GBP5s present natural variations in their subcellular localization, from TGN to cellular diffusion.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 4. Natural variation in subcellular localization of bat GBP5s.

TZM-bl cells were transfected with a plasmid coding for indicated HA-GBP5 species proteins: 10 bat orthologs, and 2 human GBP5s: wt and mutant C583A. Two days post-transfection, GBP5 localization was analyzed by confocal fluorescence microscopy with the indicated marker. Nuclei and trans-Golgi-network (TGN) were stained with DAPI and anti-TGN46, respectively. A, All the channels and a zoom are shown for Homo sapiens, the mutant Homo sapiens-C583A, Myotis yumanensis and Eptesicus fuscus. B, Only the merge is shown for the remaining bat species. The complete panel is shown in S3 Fig. The pictures present representative results observed in 3 independent experiments. Scale bar indicates 15 μm.

https://doi.org/10.1371/journal.pbio.3003760.g004

Bat GBP5s restrict the retrovirus HIV-1 in a species-specific manner

The antiviral functions of human GBP5 have been particularly studied in the context of HIV-1 infection [16,18], where it was shown to target HIV-1 envelope glycoproteins, reducing the intrinsic infectivity of neosynthesized virions. Specifically, GBP5 alters the cleavage of the envelope precursor gp160 in the TGN, preventing maturation into gp120 and gp41 proteins, thereby leading to the production of immature viral particles. However, the precise mechanisms and whether additional inhibitory effects exist remain unclear. Recent reports showed that human GBP5 specifically alters HIV-1 gp120 glycosylation, which can be experimentally witnessed by a typical electrophoretic mobility shift of viral glycoproteins [16,18]. Nonetheless, the anti-retroviral effect of human GBP5 is strictly dependent on its TGN localization. We therefore assessed the antiviral activity of bat GBP5s, using HIV-1 replication.

HEK-293T cells were cotransfected with increasing doses of GBP5 or control vector (Empty vector, EV, encoding E2-Crimson only), as well as a plasmid encoding the HIV-1 envelope (Env NL4.3) and an HIV-1 genome plasmid expressing all other HIV-1 proteins along a luciferase reporter (Fig 5A for the experimental setup). Total DNA transfected was help constant across conditions. Two days later, we quantified viral particles in the supernatant, by measuring HIV-1 reverse transcriptase (RT) activity with RT-qPCR (S4A Fig). We noted that only high doses of GBP5 affected HIV-1 RT activity in the supernatant (up to 10-fold reduction for some species, S4A Fig and S1 Table). To determine if, as human GBP5, bat GBP5s influenced the intrinsic infectivity of the neosynthesized viral particles, we infected HeLa P4P5 cells (i.e., Hela cells stably expressing the HIV-1 receptors and co-receptors) with the same amount of HIV-1 virions: equivalent to 30 mU RT. In this experimental setup, we measure the effect of GBP5 on HIV infectivity independently of the possible effect on the amount of produced viral particles (Fig 5A). Three days post-infection, we found that several bat GBP5s significantly inhibited the virion intrinsic infectivity of HIV-1 in a dose-dependent manner, similarly to human GBP5 (Fig 5B and S1 Table). As examples, GBP5 from Pteropus giganteus (pteGig), Miniopterus natalentis (minNat), Phyllostomus discolor (phyDis), and Myotis spp. (myoXxx) significantly restricted HIV-1 intrinsic infectivity, with up to 1-log restriction compared to the control condition (EV). Interestingly, the GBP5s from three bat species, which were well expressed in the cells (Fig 5C), exhibited a distinct behavior: Rhinolophus ferrumequinum (rhiFer) GBP5 showed only mild HIV-1 restriction, and Pipistrellus kuhlii (pipKuh) and Eptesicus fuscus (eptFus) GBP5s did not restrict intrinsic infectivity, similarly to the human GBP5-C583A mutant (Fig 5B).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 5. Species-specificity in bat GBP5 restriction of retrovirus (HIV-1).

A, Experimental setup. Briefly, HEK-293T cells were transfected with plasmids coding for HA-GBP5 or the control (EV “empty vector” expressing E2-Crimson), and for HIV-1 LAI genome and Luciferase reporter (Bru∂EnvLuc2 vector), NL4.3 Envelope and Rev. 48 hour post-transfection, cells were harvested for western blot analysis, and supernatant for western blot analysis after ultracentrifugation, titration of virus by RT-qPCR (RT activity), and infection of the RT-normalized viruses in HelaP4P5 cells. Infection was quantified 72 hour later by luminescence from the viral-encoded Luciferase reporter. B, HIV-1 intrinsic infectivity (RLU, Relative light units) normalized to RT activity in the supernatant, in the context of control (EV, encoding solely E2-crimson) or a dose of HA-GBP5 (1, 2, or 4 µg; total DNA identical across conditions). The corresponding species of GBP5 is shown (name follows the UCSC nomenclature, three letters from genus followed by three letters from species). Results from three independent experiments, bars correspond to SD. Statistics vs. the corresponding control EV condition, by one-way ANOVA, Dunnett’s test: *, p value < 0.05, **, p value < 0.01. C and D, Corresponding western blot analyses of HA-GBP5, HIV-1 Env, and HIV-1 Gag, and Tubulin from the total cell lysates of the HIV-1 producer cells in the context of a GBP5 dose (constant total transfected DNA). The cladogram at the top shows the phylogenetic relationships of the tested species. The RT activity titers and a western blot of purified supernatants are in S4 Fig. Panel C shows one representative experiment and panel D the quantification of western blots from three independent experiments (Tubulin is in S5A Fig). Illustrations from NIAID NIH BioArt Source (https://www.bioart.niaid.nih.gov/). The data underlying this Figure can be found in S2 Dataset.

https://doi.org/10.1371/journal.pbio.3003760.g005

To further characterize viral proteins in the presence of GBP5s, we analyzed and quantified their expressions by western blot in cell lysates from the three independent experiments (Figs 5C, 5D, and S5A). Overall, we did not find any statistically-significant differences in viral protein expression levels; yet, in the producer cells, there was a trend for Gag p41 and p24 decreased expression by some GBP5s, as compared to Gag p55 (Figs 5C, 5D, and S4B). Of note, we did not observe any variation in Tubulin expression in these assays, in which we had collected total cell lysates and run the same volumes (S5A Fig). We also probed virus supernatants from two experiments (S4B Fig shows one), which suggested possible lower expression of the mature gp120 and increased expression of the immature gp160 precursor in restrictive GBP5s. In the cell, this seemed concomitant with a change in the electrophoretic mobility of Env, as well as mild, albeit not statistically-significant, changes in the gp160:gp120 ratios (Fig 5C, 5D), as previously reported for human GBP5 [19]. Yet, we found a case in which GBP5 from Pteropus giganteus (pteGig) strongly restricted HIV-1 intrinsic infectivity (Fig 5B) without any evidence of electrophoretic mobility of Env (Fig 5C–5E). It is therefore possible that, beyond previously described phenotypes on glycoproteins and independently of its prenylation, GBP5 may have additional antiviral functions, which remain to be investigated. Finally, as control, we performed cell viability assays in the context of HIV and GBP5 exogenous expressions and found no evidence of general cytotoxicity (S5B Fig).

Overall, we revealed that bat GBP5s are antiviral factors, restricting HIV-1 replication by inhibiting viral intrinsic infectivity, but that this is species-specific, suggesting that bat GBP5 diversification is the result of virus-host arms races.

Virus-specificity and increased antiviral breadth in bat GBP5s

To determine the breadth of the bat GBP5 antiviral restriction, we tested their effect on the infectivity of viral particles bearing glycoproteins from two other RNA viruses: VSV and EBLV-1, which are independent of furin cleavage. Previous studies showed that human GBP5 inhibits the infectivity of retrovirus (HIV) pseudotyped with the glycoprotein of EBVL-1 (isolated from Eptesicus serotinus bats) but not with that of VSV [18]. Yet, a recent report by Veler and colleagues 2024 found that human GBP5 does restrict HIV:VSVg infectivity, strongly suggesting that GBP5 has an antiviral activity independent of glycoprotein cleavage by the furin [19].

Here, HEK-293T cells were cotransfected with 2 µg GBP5 or the control vector, and plasmids encoding HIVΔenv luciferase reporter viruses pseudotyped with EBLV-1g or VSVg (HIV:EBLV-1g or HIV:VSVg, respectively). We quantified RT activity from the supernatant, as previously described, and measured the virions’ infectivity in new HEK-293T target cells (Fig 6A for the experimental setup). First, for both pseudoviruses, we similarly observed a decrease in RT activity in the supernatant for all GBP5s (S6 Fig). Second, and after normalization of virion amounts (RT activity), we found that bat GBP5s reduced the infectivity of HIV:EBLV-1g, but in a species-specific manner in the magnitude of restriction (Fig 6B). GBP5 from Pteropus giganteus, Phyllostomus discolor, Miniopterus natalensis, and Myotis spp significantly reduced HIV:EBLV-1g infectivity by 70%–80%. Interestingly, the same inhibitory effect was observed with the human GBP5 and GBP5-C583A mutant, suggesting a restriction mechanism independent of GBP5 isoprenylation at the CaaX motif. In Rhinolophus ferrumequinum, Pipistrellus kuhlii, and Eptesicus fuscus, the restriction was milder in the range of 35−45%. By comparison with HIV:HIV-1-Env restriction, these are the same species with the least impacting antiviral effect. This suggests that genetic variations in these bat GBP5s are probably located at common interfaces required for restriction of EBLV-1g and HIV Env mediated infectivity (i.e., common determinants). Third, we found that most bat GBP5s, as human GBP5, had a mild inhibitory effect on HIV:VSVg infectivity. Yet, GBP5 from two Vespertilionidae bat species, Myotis occultus, and Pipistrellus kuhlii, significantly restricted HIV:VSVg infectivity (up to 80% inhibition) (Fig 6B). In the latter case, it was particularly unexpected given the lack of HIV infectivity restriction (Fig 5) by Pipistrellus Kuhlii GBP5. It therefore supports that GBP5 antiviral functions can also be virus-specific. It further suggests that specific genetic innovations in GBP5 from some Vespertilionidae enabled a gain in antiviral activity, contributing to specificities unique to them. Altogether, these results show that the antiviral restriction of bat GBP5 is species- and virus-specific.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 6. Bat GBP5 restriction is both species- and virus-specific.

A, Experimental setup similar to Fig 5A but with viral pseudotyping with VSVg (VSV condition) or EBLV-1 envelope (EBLV-1 condition). B, Infectivity of RT-normalized pseudotyped-viruses produced in the presence of GBP5, normalized to the condition without GBP5 (EV, Empty vector E2-Crimson control) at 100%. The cladogram on the left shows the phylogenetic relationships of the tested GBP5 species. RLU, Relative light units. Results from three independent experiments, bars correspond to SD. Statistics vs. the corresponding control condition, One-way ANOVA, Dunnett’s test: *, p value < 0.05, **, p value < 0.01, ***, p value < 0,001. Illustrations from NIAID NIH BioArt Source (https://www.bioart.niaid.nih.gov/). The data underlying this Figure can be found in S2 Dataset.

https://doi.org/10.1371/journal.pbio.3003760.g006

The CaaX prenylation motif of GBP5 was lost through early stop codon in one lineage of Vespertilionidae bats

To identify molecular determinants of the species-specific subcellular distribution and antiviral activity of bat GBP5s, we performed comparative phenotypic and phylogenetic analyses. Remarkably, we uncovered that several species (n = 6) from the same taxon in the Vespertilionidae family share an early, fixed stop codon in their genetic sequence (Figs 7A and S2, complete alignment: see Data availability). This includes Pipistrellus kuhlii and Eptesicus fuscus, which have remarkably different subcellular localization and reduced antiviral functions compared to others (Figs 4–6). Importantly, beyond data from publicly available sequences, we obtained and treated Eptesicus fuscus cells with universal type I interferon, extracted total RNA and performed RT-PCR and Sanger sequencing to de novo sequence the IFN-induced GBP5 from mRNA transcripts (see Materials and methods). This confirmed the presence of an early stop codon. The acquisition of the latter certainly occurred through a single nucleotide mutation (C to T) in the common ancestor of this bat clade, approximately 22.6 million years ago (MYA), leading to a nonsynonymous change: “CGA” coding for Arginine (R) to “TGA” Stop codon (Fig 7A). Of note, the nucleotide sequence of the corresponding GBP5 mRNA transcripts, after the early stop codon, did not greatly diverge (Figs 7A and S8, complete alignment available), confirming a relatively recent emergence of this mutation. Interestingly, this natural change led to the complete loss of the prenylation CaaX motif in the GBP5 C-terminal region (Figs 7A, S2, and S8, complete alignment available) and this correlated with different subcellular localization and antiviral functions.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 7. Ancestral reconstruction of the prenylation motif in Eptesicus fuscus relocalizes GBP5 to the TGN, but is not sufficient to retrieve anti-pseudoretroviral functions.

A, Ancestral state sequence reconstruction upstream of the Eptesicus fuscus-CaaX prenylation motif. C-terminal end of the protein alignment of the 10 bat GBP5s tested in functional assays (asterisk, stop codon). Phylogenetic tree was used to infer the ancestral sequence of the C-terminal region, the branch where the prenylation motif was lost by a premature stop codon is annotated on the tree. The site of mutagenesis for reconstruction is indicated by the blue arrow. B, Reconstruction of the C-ter relocalizes Eptesicus fuscus GBP5-CaaX to the trans-Golgi network (TGN). Briefly, TZM-bl cells were transfected with plasmids encoding HA-GBP5s and, 48 hour later, were analyzed by confocal fluorescence microscopy. GBP5, nuclei and TGN were stained with anti-HA, DAPI and anti-TGN46 antibodies, respectively. Scale bar indicates 15 μm. C, GBP5 mean intensity at the Golgi vs. the total cell was quantified for the wild-type eptFus and the mutant eptFus-CaaX. Each dot corresponds to one cell. Two independent replicates are identified by different dot colors. D, Pearson coefficient correlation per cell calculated between GBP5 and TGN signals for the wild-type eptFus and the mutant eptFus-CaaX. Data are represented as a mean ± SD. Statistics vs. the corresponding control condition, Nested t test: **, p-value < 0.01 (n = 2). E–G, Ancestral reconstruction of the prenylation CaaX did not increase Eptesicus fuscus GBP5 restriction of intrinsic viral infectivity. E, Infectivity of RT-normalized HIV-1 pseudotyped-viruses in the presence of GBP5, normalized to the condition without GBP5 (EV control) at 100%. Dose of GBP5 plasmids: 1, 2, and 4 µg with constant total DNA transfected across conditions. Experimental setup as in Fig 5A. RLU, Relative light units. Viral titers (RT activity) are shown in S6 Fig. F, Corresponding western blot showing the expression of HA-GBP5, HIV-1 Env and Gag in the viral producer Tzm-bl cells with beta-actin as loading control (kDa, on the right). Quantification of three independent experiments is shown in S7 Fig. G, Intrinsic infectivity of (RT-normalized) VSVg or EBLV-1g pseudotyped retroviruses in the presence of GBP5s, normalized to EV control at 100%. Experimental setup as in Fig 6A.**, p-value < 0.01 (versus control). The data underlying this Figure can be found in S2 Dataset.

https://doi.org/10.1371/journal.pbio.3003760.g007

Ancestral loss of GBP5 prenylation motif in a lineage of bats led to its subcellular relocalization in Eptesicus fuscus bat cells and increased restriction against rhabdovirus.

To test Eptesicus fuscus bat GBP5 functions in the context of bat cells, we successfully exogenously expressed myoYum GBP5, eptFus GBP5, and the CaaX-reconstructed mutant in the Eptesicus fuscus bat cell lines by DNA transfection (of note, in the absence of IFN stimulation, the endogenous GBP5 mRNA transcript is not detectable in these cells). First, through confocal immunofluorescence imaging, we found that eptFus GBP5 expression was diffused in its own species cells and that it re-localized to a specific compartment upon CaaX reconstruction similar to the myoYum GBP5 (of note, the TGN antibody did not cross-react in Eptesicus bat cells) (Fig 8A), suggesting a similar subcellular phenotype in these bat cells than in the human ones.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 8. Ancestral reconstruction of the prenylation motif in eptFus GBP5 leads to its subcellular relocalization in Eptesicus fuscus cells and to a loss of the capacity to restrict VSV-GFP infections.

A, Eptesicus fuscus cells were transfected with plasmids encoding HA-GBP5s and, 48 hours later, were analyzed by confocal fluorescence microscopy with anti-HA antibody. Nuclei were stained with DAPI. Of note, anti-TGN46 antibody did not cross-react in bat cells. MyoYum GBP5 was also transfected as a control of TGN subcellular localization. B and C, VSV-GFP infections of eptFus bat cells expressing or not GBP5s: total % of cell death (B) and % of VSV-GFP infected live cells as measured by flow-cytometry. Each point corresponds to an independent replicate. D, 3D protein structure prediction (AlphaFold) of the reconstructed Eptesicus fuscus-CaaX GBP5 dimer. Colored and gray chains each correspond to a monomer. Blue, GTPase domain. Green, hinge domain. Yellow, middle domain. Orange, catalytic domain. Red, residues different from Myotis yumanensis. Credit: https://www.phylopic.org/. The data underlying this Figure can be found in S2 Dataset.

https://doi.org/10.1371/journal.pbio.3003760.g008

Second, we tested the production of infectious pseudoretroviral particles in these Eptesicus fuscus, as well as Myotis yumanensis, bat cells in +/− GBP5 conditions. However, the production of infectious virions was intrinsically restricted in these cells, as noted by the absence of Gag maturation (only p55 is expressed) and the detection of Env expression in cell lysates (S9 Fig), as well as absence of detected RT activity in supernatants. Therefore, we finally tested another viral infection in the Eptesicus fuscus bat cells, using full-length infectious rhabdovirus VSV-GFP. Briefly, bat cells were seeded and transfected 48 hours later with 2 µg of EV or GBP5 plasmids in 6-well plates. Thirty-two hours later, cells were infected with equal amounts of VSV-GFP. 16 hours post-infection, cells were fixed for flow-cytometry analyses and we quantified cell death (Fig 8B) and GFP+ infected living cells (Fig 8C). We found that human GBP5 protected against VSV-induced cell death and VSV-GFP infections, confirming its antiviral activity against VSV (Veler and colleagues 2025) even in bat cells. Moreover, we showed that eptFus GBP5 also had antiviral activity against VSV-GFP. Finally, we found that the eptFus GBP5 with the CaaX reconstruction did not protect against VSV-GFP-induced cell death and did not restrict VSV-GFP infection. This shows that the loss of this motif is necessary for eptFus GBP5 antiviral function against VSV.

Altogether, our study supports an evolutionary model in which the loss of the C-terminal CaaX motif in GBP5 in the common ancestor of Pipistrellus and Eptesicus bats, ~22 MYA, led to changes in GBP5 subcellular localization and may have been beneficial (i.e., selected) in the context of ancient Rhabdoviral-like infections, thanks to its increased antiviral activity.

Ethics statement, bat sampling, and generation of cell lines

Myotis spp. (yumanensis, occultus, auriculus, and thysanodes) bats sampled for this study were wild caught under scientific collection permits from California (California Department of Fish and Wildlife) and Arizona (Arizona Game and Fish Department): AZ SP407155, AZ SP407113, AZ SP403977, SCP-2672, and UA IACUC 20-615. Protocols were validated by Animal Care and Use Committees at UC Berkeley and University of Arizona: IACUC AUP-2020-11-13795, AUP-2023-04-16265, and UA IACUC 20-615. Bats were sampled using standard mist-netting procedures, including taking standard body measurements, following USGS recommendations for White-Nose Syndrome and COVID-19 prevention. Two 3-mm wing punch biopsies were taken from the left and right plagiopatagium of each donor individual and placed in a live cell collection media consisting of DMEM/F12 (Gibco) supplemented with 15mM HEPES (Gibco), 20% FBS (Gibco), and 0.2% puromycin (Invivogen) [49–51]. Wing punches were then brought back to a cell culture facility in Berkeley, where they were used to generate cell lines as previously described [49–51].

Cell culture

Human embryonic kidney 293T (ATCC, cat. CRL-3216), HelaP4P5 (Hela cells expressing CD4^hiCCR5^hiCXCR4^hi), and TZM-bl (NIH AIDS Research and Reference Reagent Program, Cat. 8129; Hela cells expressing CD4^hiCCR5^hiCXCR4^hi and the luciferase reporter under an LTR promoter) cells were grown in DMEM supplemented with 5% fetal bovine serum (FCS, Sigma cat. F7524) and 100 U/ml of penicillin/streptomycin (Sigma–Aldrich). Bat Myotis yumanensis primary cells were maintained in DMEM supplemented with 10% fetal bovine serum (FCS, Sigma cat. F7524) and 100 U/ml of penicillin/streptomycin (Sigma–Aldrich). Bat Eptesicus fuscus immortalized skin fibroblasts (a gift from Cédric Feschotte and Rachel Cosby [52]) were maintained in DMEM supplemented with 7.5% fetal bovine serum and 100 U/ml of penicillin/streptomycin.

RNA extraction for next-generation sequencing and quality assessment

Myotis yumanensis primary cells from the three individuals were plated in 12-well plates. Cells from each individual constituted a replicate. Twenty-four hours later, cells were either treated with universal type I interferon treatment (PBL assay science) at 1,000 U/mL or the medium was replaced with IFN-free DMEM for the mock condition. After6 hours, cells were collected and pelleted, and then stored at −20 °C. RNA was extracted from frozen cell pellets according to RNA isolation Macherey–Nagel kit protocol. To maximize RNA extraction quality, DNAse incubation time was increased to 30 min. Total RNA concentration was quantified using Qubit RNA high sensibility kit and Qubit4 machine (Thermofisher). RNA integrity was assessed by capillary electrophoresis on Tapestation 4,150 (Agilent). All samples reached an RNA integrity number superior to 9 and were used for library preparation.

Library preparation and RNAseq

Samples were enriched in mRNA using NEBNext Poly(A) mRNA Magnetic Isolation Module (New England Biolabs, E7490- lot n°10151189) following the manufacturer’s protocol. Enriched mRNA samples were stored at −80 °C prior to library concentration. RNA libraries were generated using CORALL Total RNA-Seq Library Prep kit (UDI 12 nt set B1, Lexogen). Libraries’ control assessments were performed using Qubit dsDNA HS Assay Kit on Qubit4 and capillary electrophoresis Tapestation 4,150. Libraries were then mixed in equimolar concentration. The sequencing was performed using three successive runs on Illumina NextSeq 500 platform in paired-end 78-pb with dual indexing at IGFL sequencing platform.

Bioinformatic analyses of RNAseq data

Raw data quality was assessed using FASTQC (version 0.12.0) and results were compiled using MultiQC (version v1.14). The data were trimmed to remove adapters/UMI index and low-quality reads using FASTP (version 0.23.1, with the parameter -min_len 40 and -q 20).

De novo sequencing of GBP5 genes

Total genomic RNA was extracted from bat primary fibroblast cells of Myotis auriculus, Myotis occultus, Myotis thysanodes, and Myotis yumanensis, as well as from the Eptesicus fuscus immortalized skin fibroblasts, using Macherey–Nagel NucleoSpin RNA kit following the manufacturer’s protocol. Total RNA was reverse transcribed into complementary DNA (cDNA) with random primers and oligo(dT), using the SuperScript III One-Step reverse transcription polymerase chain reaction (PCR) kit (Thermo Fisher Scientific, Poland). GBP5 mRNA was then amplified from each species using 10 ng of cDNA and different sets of primers (S2 Table) that were specifically designed using an alignment of publicly available GBP5 sequences. The PCR reactions were performed using the New England Biolabs (NEB) Q5 High-Fidelity DNA Polymerase, following the manufacturer’s protocol, including a final volume of 50 µl, a 0.5 µM primer concentration, and an annealing temperature of 61 °C. PCR products with multiple bands were excised and purified from gel using the NucleoSpin Gel and PCR Clean-up Kit from Macherey–Nagel. Sanger sequencing of GBP5 was performed by a commercial company (Genewiz, Azenta Life Sciences, Germany).

Collection of GBP5 orthologous sequences

Homologs of all six human GBPs were searched in 536 mammalian reference genomes (downloaded from NCBI in June 2023) by sequence similarity using the miniprot software [53]. Homologous matches that were at least 50% of the length of the human GBP5 were selected. Because genome sequences can include indel errors that destroy the reading frame, a first round of multiple sequence alignment of the retrieved GBP homologs was run with MACSE v2 [54]. MACSE uses a statistical model that takes frameshifts into account in order to properly align homologous codons. The resulting coding sequences were then processed with PREQUAL [55] to eliminate segments with insufficient homology and likely to be misaligned. Then, the remaining coding sequences were aligned with MACSE again. The GBP5 orthologs to the human GBP5 were then identified by generating a phylogenetic tree with IQ-TREE2 (under a GTR + R6 model as identified using ModelFinder) [29].

In total, GBP5 coding sequences from Chiroptera (n = 55; including 5 newly generated sequences), Primates (n = 37), Rodentia (n = 80), Artiodactyla (n = 97), Carnivora (n = 51), Lagomorpha (n = 10), Perissodactyla (n = 9), Proboscidea (n = 3), Pilosa (n = 1), Cingulata (n = 4), and Xenarthra (n = 1) were retrieved.

Phylogenetic and positive selection analyses of GBP5 orthologous sequences

GBP5 orthologous codon sequences were aligned for each mammalian group separately using PRANK [28] or Muscle [56], and were manually curated. A phylogenetic tree was built using the maximum likelihood method implemented in IQ-TREE using ModelFinder to identify the best substitution model [29]. Node statistical support was computed through 1,000 bootstrap replicates. The gene codon alignments were then submitted to positive selection analyses using the HYPHY package (either through command line or the DataMonkey webserver [35,39]). First, Branch-site Unrestricted Statistical Test for Episodic Diversification (BUSTED, [34]) was used to determine whether the gene has experienced episodic positive selection (p value < 0.05). Second, an adaptive branch-site REL test for episodic diversification (aBSREL) [33] was used to identify the branches under positive selection with a cutoff at p value < 0.1 as well as to quantify the dN/dS ratio for each branch independently (the tree was newly generated in aBSREL). Lastly, we used three statistical analyses to identify the specific sites under positive selection: MEME, which detects individual sites subject to episodic diversifying selection ([37], cutoff p value < 0.01), FUBAR (A Fast, Unconstrained Bayesian AppRoximation for Inferring Selection, [36]; cut off posterior probabilities PP > 0.90) and FEL (Fixed Effects Likelihood, [38]; cutoff p value < 0.05).

Plasmids

GBP5s were amplified from bat cDNA (species: Myotis auriculus, Myotis occultus, Myotis thysanodes, Myotis yumanensis, and Eptesicus fuscus) of fibroblast cells (details above) or synthesized by a commercial company (Genewiz, Azenta Life Sciences, Germany) (species: Rhinolophus ferrumequinum, Pteropus giganteus, Pipistrellus kuhlii, Miniopterus natalensis, and Phyllostomus discolor). They were cloned (via BstBI and XhoI restriction sites) into the vector MT06 (RRL.sin.cPPT.CMV/-BamHI-HA-BstBI-E2-crimson-XhoI.IRES-puro.WPRE, shared by Caroline Goujon: Addgene plasmid # 139448 [57]), where an HA tag was added between BamHI and XhoI restriction sites. Ancestral reconstruction of the Eptesicus fuscus isoprenylation motif expression was generated from wild-type Eptesicus fuscus pMT06-GBP5 by replacing the Stop codon with corresponding ancestral residue, using the Quik Change Lightning Site-Directed Mutagenesis Kit (Agilent) following the manufacturer’s instructions (set primers available in S2 Table). For pseudovirus production, a plasmid coding for LAI-∆Env-Luc2 genome (Bru-∆Env-Luc2 that has no Env and encodes a firefly luciferase in place of the nef gene, shared by Michael Emerman) was used and pseudotyped with HIV-1 NL4.3 Env (along an HIV-1 Rev plasmid to facilitate nuclear export of Env, [58]), VSVg (pMD2.G, Didier Trono: Addgene plasmid # 12259) or European bat lyssavirus EBLV-1 Env (pLTR-EBLV1env, a gift from Daniel Sauter [18]).

HIV-1 production and infection assay

A total of 4 × 10⁵ HEK-293T cells were seeded in 6-well plates and transfected the next day with plasmids coding for LAI-∆Env:Luc2 genome (1.2 µg), NL4.3 envelope (120 ng), HIV-1 Rev (75 ng), as well as indicated doses of HA-GBP5 (0, 1, 2, and 4 µg) or empty vector EV control (expressing E2-crimson) using HBS/CaCl2 to produce infectious HIV-1 single-round viruses. Forty-eight hours later, cells were harvested for western blot and virus-containing supernatants were clarified for virus titration, by measuring HIV-1 RT activity (see below), western blot (see below) and infections. For infection, 1 × 10⁴ HeLaP4P5 were seeded into white-clear bottom 96-well plates. Twenty-four hours later, cells were infected with 30 mU RT of virus supernatant in the presence of DEAE-Dextran solution (20 µg/ml). Luciferase activity from the reporter HIV-1: Luc2 virus was measured 72 hours post-infection by Tecan Spark Luminometer.

VSV and EBLV-1 pseudovirus productions and infection assays

A total of 4 × 10⁵ HEK-293T cells were seeded in 6-well plates and transfected the next day with Bru-∆Env-Luc2 (shared by Michael Emerman) (1.2 µg), VSV-g or EBLV-1 envelope, as well as indicated doses of HA-GBP5 or empty vector control using HBS/CaCl2 to produce infectious pseudotyped virus. Virus-containing supernatant was harvested, clarified for viral titration, by measuring RT activity and infection. For infection, 1 × 10⁴ HEK-293T were seeded into white-clear bottom 96-well plates and infected with 50 mU RT of virus supernatant. Luciferase activity from the reporter pseudovirus was measured 72 hours post-infection by Tecan Spark Luminometer.

Virus titration by RT activity

Titration of HIV-1, VSVg and EBLV-1 Env pseudoviruses from supernatant were quantified by measuring reverse transcriptase (RT) activity as in [59]. Briefly, virions were lysed in a homemade lysis buffer (0.25% Triton X-100, 50 mM KCl, 100 mM Tris−HCl pH 7.4, 40% glycerol) and 0.4 U/μL RiboLock RNase inhibitor (Thermo Scientific). The viral lysates were then diluted in H2O and added to the RT-qPCR mix: 35 nM bacteriophage MS2 RNA (Roche) as a template for RT, 100 nM of each primer (5′-TCCTGCTCAACTTCCTGTCGAG-3′ and 5′-CACAGGTCAAACCTCCTAGGAATG-3′) and Takyon No ROX SYBR 2X MasterMix blue dTTP buffer (Eurogentec) in a total reaction volume of 20 μL. Purified RT enzyme (Abnova) was used to make the standard curve. The RT–qPCR reaction was carried out in a Bio-Rad CFX96 cycler with the following parameters: 42 °C for 20 min, 95 °C for 5 min and 40 cycles (95 °C for 5 s, 60 °C for 30 s and 72 °C for 15 s).

Viability measurement

A total of 1 × 10⁴ HEK-293T cells were seeded in white-clear bottom 96-well plates and transfected the following day with plasmids encoding the LAI-∆Env:Luc2 genome (30 ng), NL4.3 envelope (3 ng), HIV-1 Rev (1.9 ng), as well as a plasmid encoding HA-GBP5 (100 ng) or an empty vector control using TransIT-LT1 (Mirus). Of note, the quantities were calculated to maintain the same ratio used for HIV production with the higher dose (4 µg) of GBP5 described above. Forty-eight hours later, cell viability was assessed by measuring adenosine triphosphate (ATP) levels using the CellTiter-Glo Luminescent Cell Viability Assay (Promega). Treatment with etoposide (100 µM) for 24 hour was used as a positive control (Sigma). The luminescence of ATP activity was quantified using a Tecan Spark luminometer.

Western blot and antibodies

Cells were lysed by ice-cold RIPA buffer (50 mM Tris pH8, 150 mM NaCl, 2 mM EDTA, 0.5% NP40) supplemented with protease inhibitor cocktail (Roche) and by sonication. Cell-free virions were concentrated and purified from the supernatant by centrifugation (2 hour, 30,000 rpm, 4 °C) through a 20% sucrose cushion and resuspended for lysis in Exo-RT lysis buffer. Proteins from cell or virus lysates were mixed in Laemmli buffer, heated at 95 °C for 5 min and separated by SDS-PAGE on 4%–20% Bis-Tris gels (Invitrogen) before being transferred to PVDF membrane by an overnight wet transfer at 4 °C. Membranes were blocked in a TBS-T 1X solution (« Tris Buffer Saline », Tris HCl 50 mM pH8, NaCl 30 mM, 0.05% of Tween 20) with 5% powder milk. Proteins were stained with a mouse anti-p24 HIV-1 CA (1:1000, NIH HIV Reagent Program, cat.183-H12-5C), a mouse anti-HIV-1 Env (16H3) (1:1000, NIH-ARP, cat. 12559), a rabbit anti-HA (1:5000, Sigma, cat. H6908), a mouse anti-tubulin (1:5000, Sigma, cat. T5168) or a mouse anti-β-actin (1:5000, Sigma, cat. A2228), and then with a donkey anti-rabbit-HRP (1:5000, Sigma, cat.AP188P), or a donkey anti-mouse-HRP secondary antibody (1:5000, Sigma, cat.AP16017). Detection was made with SuperSignal West Pico Chemiluminescent Substrate (ThermoFisher Scientific) using the Chemidoc Imagina System (Biorad).

Immunofluorescence imaging of human and eptFus bat cells

A total of 3 × 10⁴ TZM-bl cells or 2.5 × 10⁴ eptFus bat cells were seeded on glass coverslips and were transfected with 1 µg of GBP5 plasmid with jetPRIME (Polyplus) or Lipofectamine 3,000, respectively, according to the manufacturers’ instructions. Forty-eight hours later, cells were fixed with 4% paraformaldehyde (PFA) (Sigma) for 15 min at room temperature and permeabilized in 0.25% Triton-TX100 (Sigma) for 5 min. A blocking step was carried out for 1 hour at room temperature with 3% BSA (Sigma) and 0.1% Triton TX100 in PBS. Primary antibody incubation was carried out for 1 hour at RT with rabbit anti-HA (1:500, Sigma, cat. H6908) and sheep anti-TGN46 (1:500, Biorad, cat.AHP500GT; not cross-reacting in bat cells) to label respectively HA-GBP5 proteins and the TGN. Primary antibodies were detected with secondary donkey anti-rabbit AlexaFluor-594 (Invitrogen, cat.A212027) and donkey anti-sheep AlexaFluor-647 (Thermofisher, cat.A-21448), for 1 hour at RT. All cells were labeled with DAPI (4′,6-diamidino-2-phenylindole) (Thermo Scientific) - containing solution (1:10,000 dilution in PBS). Images were acquired using Zeiss LSM 980 AiryScan (human cells) or LSM 800 (bat cells) confocal microscopes and analyzed with Imaris software (human cells). For quantification analyses, segmentation of the cells was first performed using Cellpose algorithm (version 3.0.11) [60]. Proportion of GBP5 at the TGN and colocalization analysis were then performed on ImageJ software using an automatized macro or the BIOP JACoP plugin, respectively.

Infection by VSV-GFP in bat cells exogenously expressing GBP5s

Bat Eptesicus fuscus immortalized cells were all seeded at 5 × 10⁴ cells/ml in 6-well plates (final volume of 2 ml) at day 1. 48 hours later, they were transfected by lipofectamine 3,000 in three independent experiments with 2 µg DNA of EV or GBP5 plasmids per condition. Thirty-two hours later, cells were infected with equal amounts (6 µl) of VSV-GFP, whose stock was previously made from 293T cells. 16 hours post-infection, eptFus cells were fixed with Fix and Perm Cell Permeabilization kit (Invitrogen, Thermo Fisher Scientific) for flow-cytometry analyses using MacsQuant cytometer (SFR BioSciences) and we quantified cell death and GFP+ infected living cells.

Other software and statistical analyses

Differences between conditions were statistically analyzed using GraphPad Prism 9 software or R. A Nested t test was used for immunofluorescence analysis (Fig 8). One-way ANOVA with Dunnett’s test for all other datasets. For each of these tests, the p-value was considered significant when inferior to 0.05: *, p value < 0.05, **, p value < 0.01, ***, p value < 0,001. Error bars in graphics are SD from three independent experiments.