key: cord-1026642-uw4wfpwl
authors: Bondet, Vincent; Le Baut, Maxime; Le Poder, Sophie; Lécu, Alexis; Petit, Thierry; Wedlarski, Rudy; Duffy, Darragh; Le Roux, Delphine
title: Constitutive IFNα protein production in bats
date: 2021-06-22
journal: bioRxiv
DOI: 10.1101/2021.06.21.449208
sha: 29aae824e8b0996f0169d1043fca5fb031bd0fca
doc_id: 1026642
cord_uid: uw4wfpwl

Bats are the only mammals with self-powered flight and account for 20% of all extant mammalian diversity. In addition, they harbor many emerging and reemerging viruses, including multiple coronaviruses, several of which are highly pathogenic in other mammals, but cause no disease in bats. How this relationship between bats and viruses exists is not yet fully understood. Existing evidence supports a specific role for the innate immune system, in particular type I interferon (IFN) responses, a major component of antiviral immunity. Previous studies in bats have shown that components of the IFN pathway are constitutively activated at the transcriptional level. In this study, we tested the hypothesis that the type I IFN response in bats is also constitutively activated at the protein level. For this we utilized highly sensitive Single Molecule (Simoa) digital ELISA assays, previously developed for humans that we adapted to bat samples. We prospectively sampled four non-native chiroptera species from French zoos. We identified a constitutive expression of IFNα protein in the circulation of healthy bats, and concentrations that are physiologically active in humans. Expression levels differed according to the species examined, but was not associated with age, sex, or health status suggesting constitutive IFNα protein expression independent of disease. These results confirm a unique IFN response in bat species that may explain their ability to coexist with multiple viruses in the absence of pathology. These results may help to manage potential zoonotic viral reservoirs and potentially identify new anti-viral strategies.

and Eppendorf tubes to obtain plasma. PAXGene tubes were stored at -20°C until extraction, while Eppendorf tubes were centrifuged at 2500rpm for 10min. Plasma was then removed and stored at -80°C until Simoa analysis. 

Bat epithelial cells from Tb 1 Lu cell line (ATCC, United States) were cultured at 37°C, 5% CO 2 , in complete culture medium composed of MEM Eagle medium with 2 mM Glutamine supplemented with 50 U/mL penicillin, 50 μg/mL streptomycin (all from Lonza, Belgium) and with 10% of decomplemented fetal calf serum (Gibco, Thermo Fisher Scientific, France). Before stimulation, cells were plated in 1mL of complete medium per well of 24 well plates and maintained at 37°C, 5% CO 2 until they reached 2x10 6 cells/well. Supernatants were removed and 1mL of complete medium including 500HAU/mL mouse influenza virus (Strain H1N1 A/PR81934) was added to the cells or not (unstimulated control). Before stimulation and 1 hour, 3.5 hours, and 23 hours after stimulation, supernatants were sampled and frozen at -80°C for IFNα protein quantification.

The Rousettus aegyptiacus IFNα DNA sequence was obtained from a previously published study (32) . Nucleotide bases that correspond to the signal peptide were removed, a start codon, spacers, and codons for a 6His tag and a TEV cleavage site were added in the 5' termination. The cDNA coding for the recombinant protein was chemically-synthesized with optimization for expression in Escherichia coli. The recombinant gene was then introduced in a pT7 expression plasmid under the control of a Lac operator and harboring kanamycin resistance. E. coli strains were transformed and kanamycin-resistant clones were selected. After optimization, protein production was done, culturing the selected clones in a Luria-Bertani kanamycin (LBkan) medium at 16°C during 16 hours after induction with 1mM isopropyl β -D-1-thiogalactopyranoside (IPTG). Bacteria were then harvested by centrifugation. The pellet was lysed and the soluble extract was obtained after a second centrifugation. This soluble extract was directly used at different dilutions as a calibrator for the digital ELISA assay. It was aliquoted and stored at -80°C before use.

All plasma samples were first thawed and centrifuged at 10.000g, +4°C for 10 minutes to remove debris. Because bats can harbor many viruses, supernatants were treated in a P2 laboratory for viral inactivation using a standard solvent/detergent protocol used for human blood plasma products (33, 34) and described in (35) and in (36). Briefly, samples were treated with Tri-n-Butyl Phosphate (TnBP) 0.3% (v/v) and Triton X100 (TX100) 1% (v/v) for 2 hours at room temperature. After treatment, TnBP was removed by passing the samples through a C18 column (Discovery DSC-18 SPE from Supelco). For digital ELISA assays, inactivated samples and stimulated cell supernatants were diluted in the Detector / Sample Diluent (Quanterix) added with NP40 0,5% (v/v). They were then incubated for one hour at room temperature before analysis. Global dilution factor was generally 1/6 for plasma samples and 1/3 for stimulated cell supernatants depending on the amount of material available and to allow the optimal protein detection.

The Simoa IFNα2 assay was developed using the Quanterix Homebrew kit and described in (36).

The BMS216C (eBioscience) antibody clone was used as a capture antibody after coating on paramagnetic beads (0.3mg/mL), and the BMS216BK biotinylated antibody clone was used as the detector at a concentration of 0.3ug/mL. The SBG revelation enzyme concentration was 150pM.

The assay follows a 2-step ELISA configuration. Two calibrators were used; recombinant human IFNα2c (hIFNα2c) purchased from eBioscience and Rousettus aegyptiacus IFNα (bIFNα) produced in Escherichia coli for this study. The limit of detection (LOD) was calculated by the mean value of all blank runs + 2SD after log conversion.

Whole blood RNA was extracted manually from PAXGene tubes, following manufacturer's instructions (Blood RNA extraction kit, Qiagen, France). After extraction, samples were inactivated at 65°C for 5min then stored at -80°C until RT-qPCR. RT-qPCR was done using the qScript XLT One-Step RT-qPCR mix following manufacturer's instructions (Quanta BioSciences, Inc., United States). Taqman probes (Applied Biosystems, ThermoFisher Scientific, France) and primers (Eurofins, France) for IFN-α1, IFNα2 and IFNα3 were described previously for P. alecto bat species in (30) and used here. Probes and primers for the bat GAPDH housekeeping gene were designed using Primer-BLAST from NCBI (https://www.ncbi.nlm.nih.gov/tools/primer-blast/) and are presented in table S2. All data were normalized relative to the housekeeping gene (GAPDH) as indicated. The expression level of the target genes was calculated using the standard curve method and expressed as copy numbers relative to the housekeeping gene.

RT-qPCR for potential coronaviruses in bat whole blood RNA was performed from the mRNA extracted previously and following the protocol previously published in (37).

GraphPad Prism 8 was used for statistical analysis. Mann-Whitney tests were used to compare two groups such as female and male, or healthy and disease. ANOVA tests (Kruskal-Wallis) with

Dunn's post testing for multiple comparisons were used to test for differences between multiple bat species. For all analyses, p values less than 0.05 were considered statistically significant, with *p< 0.05; **p<0.01; ***p<0.001; ****p<0.0001. Median values were reported on figures. Spearman correlations are used to compare continuous variables such as mRNA level or age and protein production.

All available data from the bat cohort are shown in Supplementary Table S1 .

Anti-bat IFNα antibodies are not commercially available for the development of bat-specific IFNα ELISA. However, given the ultra-sensitivity of human IFNα digital ELISA which detects protein at attomolar concentrations (31), and potential cross-species reactivity, we hypothesized that our existing human assay could also detect bat IFNα. As a first proof of concept, we stimulated a bat lung epithelial cell line with influenza virus (Strain H1NI A/PR81934) and tested the recovered supernatant with a human IFNα2 digital ELISA. We observed a significant induction of IFNα2 protein at 1hr and 3.5 hrs as compared to the unstimulated control (Fig 1a) .

These initial results were extrapolated from a standard curve of a human recombinant IFNα2 protein. To better adapt our assay to bat species, we produced a recombinant Rousettus aegyptiacus IFNα protein (bIFNα) from Escherichia coli competent bacteria. SDS-Page analysis of the soluble and insoluble extracts obtained from the bacteria pellet showed that bIFNα was mainly produced as an insoluble form even at low induction temperature ( Fig S1a) . Comparing profiles before and after induction of the protein expression, SDS-Page analysis showed that the unique bIFNα band appeared alone at this mass ( Fig S1a) . Western-Blot analysis of the soluble fraction after induction at 16°C using the IFNα2 assay detection antibody revealed that the protein was expressed in a single band at the expected molecular weight (Fig S1a) . The purification from the soluble extract failed: the bIFNα protein was not selected at the expected molecular weight ( Fig   S1b) and the western-Blot analysis revealed no affinity at the purified molecular weight (Fig S1a) .

The purification from the insoluble extract succeeded, but the renaturation of the protein failed ( Fig   S1c) . So we used the soluble extract itself as a calibrator after quantification of bIFNα. Global protein quantification of the soluble extract was done using the BCA assay, and the bIFNα protein concentration in the soluble extract was estimated after gel densitometry for potential use as a digital ELISA calibrator.

To explore the ability of the IFNα2 digital ELISA assay designed for the quantification of human interferons, to quantify bat IFNα species, we compared the responses of the assay to bIFNα protein and all 13 human IFNα subtypes (Fig. 1b) . As expected the assay revealed a weaker response for bIFNα in comparison with hIFNα2c. However, the affinity of the human mAb for the bat protein was comparable to the human subtypes, with bIFNα and hIFNα21 showing very similar affinities, and two human species showing weaker responses (Fig. 1b) . Using the bIFNα protein as the calibrator, we re-calculated the cellular response after in vitro influenza stimulation and observed similar results with the highest concentrations present after 1hr of influenza stimulation (Fig 1c) . The only difference observed was related to the scale of these results, due to the lower affinity of the mAb for the bIFNa2 calibrator.

To better understand the cross-species reactivity we compared the sensitivities of the 13 IFNα subtypes as previously described (36) S2a) . The LMNED sequence also appears in human IFNα16 and IFNα17, two species for which the IFNα2 assay shows a positive response. The M→L substitution concerns two apolar amino-acids. The K→D substitution changes a positive with a negative charged amino-acid, but these two amino-acids are then hydrophilic and so do not produce a detrimental α-helix coil in the structure. This in silico analysis provides support for how the IFNα2 antibody assay may recognize Rousettus and Pteropus IFNα protein.

Having validated the assay for its ability to detect bat IFNα, we analyzed plasma samples from 4 bat species sampled from French zoos (Table 1) with the digital ELISA assay. Results are presented using the two calibrators; hIFNα2c (Fig. 2a) and bIFNα (Fig. 2b) . A greater number of samples (Table S1 ) were associated with IFNα protein levels in all species but found no significant associations (Fig S3) . We also tested for presence of corona viruses in the blood but found no evidence (data not shown). These results support the hypothesis that certain bat species have physiological levels of circulating IFNα protein in healthy conditions.

The constitutive mRNA expression of bat IFNα genes has been previously described for Rousettus IFNα3, and between Pteropus rodricensis and Pteropus lylei for IFNα3. This also indicates that high levels of variation could be observed within a same genus, and also within the same species.

While Eidolon helvum showed lower mRNA and protein levels as compared to the other species,

and Pteropus rodricensis medium mRNA levels and a high protein level, Pteropus lylei showed the highest RNA and the lowest protein levels (Fig. 2a, b and Fig. 3a) . Therefore, we observed no overall direct link between IFNα mRNA expression and protein plasma concentrations across all species examined. To directly explore this hypothesis, we tested the correlations for each gene and protein for Pteropus rodricensis samples where sufficient detectable measurements were available for both parameters. In this relatively small sample size (n=8 for paired samples) we observed a strong statistically significant correlation (Rs=0.952, p=0.001) between IFNα2 RNA and protein levels (Fig. 3b) . No correlation was observed with IFNα1 and IFNα3 genes (Fig. 3b) . This result may reflect that the ELISA assay is designed for the IFNα2 subtype.

Type I interferons trigger the downstream activation of hundreds of critical genes as part of the anti- Our study contains some inherent weaknesses. IFNα protein concentrations were calculated using human IFNα2c or Rousettus aegyptiacus IFNα protein (bIFNα) produced in E. coli as calibrators, with results respectively in the fg/mL and the ng/mL ranges. Such a difference could be due to different antibody specificities, potentially due to incorrect folding of the bIFNα protein produced in E. coli strains. This is supported by the observation that the bIFNα protein is mainly expressed in the insoluble fraction even at 16°C (Fig. S1c) , that the protein failed to renature after urea purification (Fig. S1b) , and that the viral inactivation solvent/detergent protocol had a greater effect on the IFNα2 assay response to bIFNα (>1Log) than hIFNα2c (Fig. S4) , suggesting greater insolubility of the bat protein. Additional improvements of the assay could be envisioned such as production of a purer and better folded recombinant protein in mammalian cells, and eventually the production of a bat specific monoclonal antibody against this protein.

Despite these technical limitations, we were able to show elevated levels of plasma IFNα protein in certain bat species, which also correlated with expression levels of the IFNα2 gene. While additional confirmatory experiments will be required, the inter-species differences in plasma IFNα protein is an interesting observation. It would also be interesting in future studies to assess whether these IFNα protein differences have an impact on viral levels and diversity within the different bat species. Lastly, our results raise additional new questions on the nature of bat physiology, in particular how the constitutively activated type I IFN response is maintained in bats without resulting in pathological conditions such as those observed in human autoimmune disease.

Chronic IFN activation, in particular during growing and development phases, can have significant neurological effects as observed in interferonopathies such as STING mutation patients (42) . Correlation plots between the Eq. bIFNα protein concentration obtained using the IFNα2 digital ELISA assay and the GAPDH-normalized number of IFNα1, IFNα2 and IFNα3 mRNA copies.

Spearman method is used for correlation analysis with Spearman's Rank Correlation Coefficient R (Rs) and p values reported (n=8). IFNα2 assay response (AEB) as a function of IFNα concentrations for hIFNα2c and bIFNα untreated or after viral inactivation.

Tables Legends Table 1 .

Number of individuals, gender, species and origins for the bat cohort. Data are shown as the n (%). 

A molecular phylogeny for bats illuminates biogeography and the 39

Does type-I interferon drive systemic autoimmunity?

Neuropsychiatric side effects of HCV therapy and their treatment: focus on IFN alpha-induced depression

Type I interferon-mediated autoinflammation due to DNase II deficiency

STING-Mediated Lung Inflammation and Beyond

Autoantibodies against type I IFNs in patients with life-threatening COVID-19

We thank the ProteoGenix company (Schiltigheim, France) for the production of the bIFNα ELISA