Attenuation of clinical and immunological outcomes during SARS‐CoV‐2 infection by ivermectin

In order to study the effects of IVM chemical therapy on clinical outcome, we assessed body weight, clinical score, and olfactory performance daily for 4 days post‐infection. The golden hamster model reproduces a moderate‐to‐severe COVID‐19 in humans, with a clinical phase lasting 5–6 days post‐infection followed by a complete recovery by 2 weeks post‐infection, with no death occurring. The viral titers and viral RNA loads in the airways of infected animals are elevated after 2–4 days post‐infection, but they drastically drop by 7 days post‐infection (Chan et al, 2020; Sia et al, 2020; de Melo et al, 2021).

The IVM‐treated infected animals, of both sexes, showed a decrease in body weight similar to that observed in saline‐treated infected hamsters (Fig EV1A). However, IVM‐treated infected animals exhibited a significant reduction in the severity of the clinical score, in a sex‐dependent manner: Infected males presented a significantly reduced clinical score, and it fully returned to normal in infected females (Fig 1A). Remarkably, IVM treatment reduced the olfactory deficit in infected animals (Figs 1B and EV1B–D): 66.7% (12/18) of the saline‐treated infected hamsters presented with hyposmia/anosmia, whereas only 22.2% (4/18) of IVM‐treated infected hamsters presented with signs of olfactory dysfunction (Figs 1B and EV1B–D). The olfactory performance was also influenced by sex: 83.3% (10/12) of the saline‐treated infected males presented with hyposmia/anosmia, against only 33.3% (4/12) of IVM‐treated infected males (Figs 1B and EV1B–D). In females, while 33.3% (2/6) of saline‐treated infected animals presented with hyposmia/anosmia, no olfactory deficit was observed in IVM‐treated infected females (0/6; Figs 1B and EV1B–D).

Since males presented a higher incidence of anosmia/hyposmia, we subsequently performed a dose‐response curve to test the effect of IVM on the clinical presentation and olfactory functions of infected males: Lower doses of IVM (100 or 200 µg/kg) elicited similar clinical outcomes as the anti‐parasitic dose of 400 µg/kg (Fig EV2). As expected, no signs of olfactory deficit were observed in mock‐infected hamsters (Fig EV1B–D).

To evaluate the effect of IVM treatment on the viral load in the respiratory tract, we tested the nasal turbinates and lungs of infected hamsters using both classical RT–qPCR (Fig 1C) and the highly sensitive technique of digital droplet PCR (Suo et al, 2020; Fig 1D, Appendix Fig S1). Surprisingly, the viral RNA load in the respiratory tract remained unaffected by IVM treatment in both samples in both sexes. Furthermore, IVM treatment did not influence viral replication rate, as evaluated by the ratio between structural and non‐structural gene transcription (Fig 1D, Appendix Fig S1). Finally, IVM treatment did not alter infectious viral titers in the lungs (Fig 1E). These results illustrate that no antiviral activity of IVM is detected in vivo at the standard anti‐parasitic dose of 400 µg/kg, in contrast to a previous report suggesting that IVM inhibits the replication of SARS‐CoV‐2 in vitro, albeit used at far higher concentrations (Caly et al, 2020). Therefore, the action of IVM on COVID‐19 clinical signs of infected golden hamsters does not result from a decrease in viral replication.

Anosmia is a typical symptom of COVID‐19 in humans, with some sex‐dependent differences (Han et al, 2020; Qiu et al, 2020; Xydakis et al, 2020). Inflammation in the nasal cavity, following olfactory sensory neurons infection and deciliation, has been shown to be an underlying factor for smell loss during SARS‐CoV‐2 infection (de Melo et al, 2021), and the chemokine Cxcl10 could be directly implicated due to its neurotoxic potential (Oliviero et al, 2020). We therefore tested a possible modulation by IVM of the local inflammatory response in hamsters and in particularly in the nasal turbinates, the primary target tissue of SARS‐CoV‐2 infection (de Melo et al, 2021), that could correlate its effect on the olfactory score. To this aim, a panel of cytokines (Il‐6, Il‐10, Il‐1β, Tnf‐α, Ifn‐β, Ifn‐γ, and Ifn‐λ) and chemokines (Cxcl10 and Ccl5) were used to assess the impact of IVM treatment on immune responses in the nasal turbinates of SARS‐CoV‐2‐infected hamsters at 4 dpi. Upon treatment with IVM, females presented a significant downregulation of Il‐6, Il‐10, and Tnf‐α, which are key inflammatory mediators of prognostic value in COVID‐19 patients (McElvaney et al, 2020a), and of Cxcl10 (Fig 1F), in line with their better olfactory performance observed in the food finding tests (Fig 1B). The differences between sex groups are illustrated by the increase in three pro‐inflammatory mediators (Ifn‐γ, Ifn‐λ, and Ccl5) only in males (Fig 1F). No difference for the Il‐6/Il‐10 ratio was observed in the nasal turbinates.

All animal experiments were performed according to the French legislation and in compliance with the European Communities Council Directives (2010/63/UE, French Law 2013–118, February 6, 2013) and according to the regulations of Pasteur Institute Animal Care Committees. The Animal Experimentation Ethics Committee (CETEA 89) of the Institut Pasteur approved this study (200023; APAFIS#25326‐2020050617114340 v2) before experiments were initiated. Hamsters were housed by groups of 4 animals in isolators and manipulated in class III safety cabinets in the Pasteur Institute animal facilities accredited by the French Ministry of Agriculture for performing experiments on live rodents. All animals were handled in strict accordance with good animal practice.

The isolate BetaCoV/France/IDF00372/2020 (EVAg collection, Ref‐SKU: 014V‐03890) was kindly provided by Sylvie Van der Werf. Viral stocks were produced on Vero‐E6 cells infected at a multiplicity of infection of 1 × 10⁻⁴ plaque‐forming units (PFU). The virus was harvested 3 days post‐infection, clarified, and then aliquoted before storage at −80°C. Viral stocks were titrated on Vero‐E6 cells by classical plaque assays using semisolid overlays (Avicel, RC581‐NFDR080I, DuPont) (Baer & Kehn‐Hall, 2014).

Male and female Syrian hamsters (Mesocricetus auratus; RjHan:AURA) of 5–6 weeks of age (average weight 60–80 g) were purchased from Janvier Laboratories and handled under specific pathogen‐free conditions. The animals were housed and manipulated in isolators in a Biosafety level‐3 facility, with ad libitum access to water and food. Before manipulation, animals underwent an acclimation period of 1 week.

Animals were anesthetized with an intraperitoneal injection of 200 mg/kg ketamine (Imalgène 1000, Merial) and 10 mg/kg xylazine (Rompun, Bayer) and received one single subcutaneous injection of 200 µl of freshly diluted ivermectin (I8898, Sigma‐Aldrich) at the classical anti‐parasitic dose of 400 µg/kg (Beco et al, 2001) (or at 100–200 µg/kg for the dose–response experiment). Non‐treated animals received one single subcutaneous injection of 200 µl of physiological solution. 100 µl of physiological solution containing 6 × 10⁴ PFU of SARS‐CoV‐2 was then administered intranasally to each animal (50 µl/nostril). Mock‐infected animals received the physiological solution only.

Infected and mock‐infected animals were housed in separate isolators, and all hamsters were followed up daily during 4 days at which the body weight and the clinical score were noted. The clinical score was based on a cumulative 0–4 scale: ruffled fur, slow movements, apathy, and absence of exploration activity.

At day 3 post‐infection (dpi), animals underwent a food finding test to assess olfaction as previously described (Lazarini et al, 2012; de Melo et al, 2021). Briefly, 24 h before testing, hamsters were fasted and then individually placed into a fresh cage (37 × 29 × 18 cm) with clean standard bedding for 10 min. Subsequently, hamsters were placed in another similar cage for 2 min when about five pieces of cereals were hidden in 1.5 cm bedding in a corner of the test cage. The tested hamsters were then placed in the opposite corner, and the latency to find the food (defined as the time to locate cereals and start digging) was recorded using a chronometer. The test was carried out during a 15‐min period. As soon as food was uncovered, hamsters were removed from the cage. One minute later, hamsters performed the same test but with visible chocolate cereals, positioned upon the bedding. The tests were realized in isolators in a Biosafety level‐3 facility that were specially equipped for that.

At 4 dpi, animals were euthanized with an excess of anesthetics (ketamine and xylazine) and exsanguination (AVMA, 2020), and samples of nasal turbinates and lungs were collected and immediately frozen at −80°C. Fragments of lungs were also collected and fixed in 10% neutral buffered formalin.

Frozen tissues were homogenized with TRIzol (15596026, Invitrogen) in Lysing Matrix D 2‐ml tubes (116913100, MP Biomedicals) using the FastPrep‐24™ system (MP Biomedicals) at the speed of 6.5 m/s during 1 min. Total RNA was extracted using the Direct‐zol RNA MicroPrep Kit (R2062, Zymo Research: nasal turbinates) or MiniPrep Kit (R2052, Zymo Research: lung) and reverse‐transcribed to first‐strand cDNA using the SuperScript™ IV VILO™ Master Mix (11766050, Invitrogen). qPCR was performed in a final volume of 10 μl per reaction in 384‐well PCR plates using a thermocycler (QuantStudio 6 Flex, Applied Biosystems). Briefly, 2.5 μl of cDNA (12.5 ng) was added to 7.5 μl of a master mix containing 5 μl of Power SYBR Green Mix (4367659, Applied Biosystems) and 2.5 μl of nuclease‐free water with nCoV_IP2 primers (nCoV_IP2‐12669Fw: 5′‐ATGAGCTTAGTCCTGTTG‐3′; nCoV_IP2‐12759Rv: 5′‐CTCCCTTTGTTGTGTTGT‐3′) at a final concentration of 1 μM (WHO, 2020). The amplification conditions were as follows: 95°C for 10 min, 45 cycles of 95°C for 15 s and 60°C for 1 min; followed by a melt curve, from 60 to 95°C. Viral load quantification of hamster tissues was assessed by linear regression using a standard curve of eight known quantities of plasmids containing the RdRp sequence (ranging from 10⁷ to 10⁰ copies). The threshold of detection was established as 200 viral copies/µg of RNA. The Golden hamster gene targets were selected for quantifying host inflammatory mediator transcripts in the tissues using the Hprt (hypoxanthine phosphoribosyltransferase), the γ‐actin, and/or the actinB genes as reference (Appendix Table S1). Variations in gene expression were calculated as the n‐fold change in expression in the tissues from the infected hamsters compared with the tissues of the uninfected ones using the $2^{‐\mathrm{\Delta }\mathrm{\Delta }{\mathrm{C}}_{\mathrm{t}}}$ method (Pfaffl, 2001).

Frozen lung fragments were weighted and homogenized with 1 ml of ice‐cold DMEM supplemented with 1% penicillin/streptomycin (15140148, Thermo Fisher) in Lysing Matrix M 2‐ml tubes (116923050‐CF, MP Biomedicals) using the FastPrep‐24™ system (MP Biomedicals) and the following scheme: homogenization at 4.0 m/s during 20 s, incubation at 4°C during 2 min, and new homogenization at 4.0 m/s during 20 s. The tubes were centrifuged at 10,000 g during 1 min at 4°C, and the supernatants were titrated on Vero‐E6 cells by classical plaque assays using semisolid overlays (Avicel, RC581‐NFDR080I, DuPont) (Baer & Kehn‐Hall, 2014).

RNA preparation was used to construct strand‐specific single‐end cDNA libraries according to the manufacturers' instructions (TruSeq Stranded mRNA sample prep kit, Illumina). Illumina NextSeq 500 sequencer was used to sequence libraries. The RNA‐seq analysis was performed with the Sequana framework (Cokelaer et al, 2017). We used the RNA‐seq pipeline (v0.9.16), which is available online (https://github.com/sequana/sequana_rnaseq). It is built on top of Snakemake 5.8.1 (Köster & Rahmann, 2012). Reads were trimmed from adapters using Cutadapt 2.10 (Martin, 2011) and then mapped to the golden hamster MesAur1.0 genome assembly from Ensembl using STAR 2.7.3a (Dobin et al, 2012). FeatureCounts 2.0.0 (Liao et al, 2014) was used to produce the count matrix, assigning reads to features using annotation MesAur1.0.100 with strand‐specificity information. Quality control statistics were summarized using MultiQC 1.8 (Ewels et al, 2016). Statistical analysis on the count matrix was performed to identify differentially regulated genes, comparing infected versus non‐infected samples considering all samples and separating by sex. Clustering of transcriptomic profiles was assessed using a principal component analysis (PCA). Differential expression testing was conducted using DESeq2 library 1.24.0 (Love et al, 2014) scripts based on SARTools 1.7.0 (Varet et al, 2016) indicating the significance (Benjamini–Hochberg‐adjusted P‐values, false discovery rate FDR < 0.05) and the effect size (fold change) for each comparison. Finally, enrichment analysis was performed using modules from Sequana, first by converting golden hamster ensembl ids to gene names and then using human annotations for GO terms and KEGG pathways. The GO enrichment module uses PantherDB (Mi et al, 2019) and QuickGO (Huntley et al, 2014) services; the KEGG pathways enrichment uses gseapy (https://github.com/zqfang/GSEApy/), EnrichR (Chen et al, 2013), KEGG (Kanehisa & Goto, 2000), and BioMart services. All programmatic accesses to the online web services were performed via BioServices (Cokelaer et al, 2013).

Lung fragments fixed in 10% neutral buffered formalin were embedded in paraffin. Four‐µm‐thick sections were cut and stained with hematoxylin and eosin staining. The slides were then scanned using Axioscan Z1 Zeiss slide scanner, using the Zen 2 blue edition software.

Lung fragments fixed in 10% neutral buffered formalin were washed in PBS and then embedded in O.C.T compound (4583, Tissue‐Tek), frozen on dry ice, and cryostat‐sectioned into 20‐µm‐thick sections. Sections were rinsed in PBS, and epitope retrieval was performed by incubating sections for 20min in citrate buffer pH 6.0 (C‐9999, Sigma‐Aldrich) at 96°C for 20 min. Sections were then blocked in PBS supplemented with 10% goat serum, 4% fetal calf serum, and 0.4% Triton X‐100 for 2 h at room temperature, followed by overnight incubation at 4°C with primary antibodies: rat anti‐Ly6G (1/100, 551459, BD‐Biosciences), chicken anti‐Iba1 (1/500, 234006, Synaptic Systems), rabbit anti‐Arg1 (1/250, PA5‐29645, Invitrogen), and rabbit anti‐SARS‐CoV nucleoprotein (1/500, provided by Dr Nicolas Escriou, Institut Pasteur, Paris). After rinsing, slides were incubated with the appropriate secondary antibodies (1/500: goat anti‐rat Alexa Fluor 546, A11081, Invitrogen; goat anti‐rabbit Alexa Fluor 488, A11034, Invitrogen; goat anti‐chicken Alexa Fluor 647, A32933, Invitrogen) for 2 h at room temperature. All sections were then counterstained with Hoechst (H3570, Invitrogen), rinsed thoroughly in PBS, and mounted in Fluoromount‐G (15586276, Invitrogen) before observation with a Zeiss LM 710 inverted confocal microscope through a Plan Apochromat 20x/0.8 Ph2 M27 lens. Cell quantification was performed in an automated manner using ImageJ. Single‐channel images were extracted, thresholded, and converted to binary images. Cells were then counted using the Particles Analyzer ImageJ plug‐in.

Statistical analysis was performed using Prism software (GraphPad, version 9.0.0, San Diego, USA), with P < 0.05 considered significant. Quantitative data were compared across groups using log‐rank test or two‐tailed Mann–Whitney test. Randomization and blinding were not possible due to pre‐defined housing conditions (separated isolators between infected and non‐infected animals). Ex vivo analysis was blinded (coded samples). All animals were included, and data were provided from 2 replications, except food finding in males, that were replicated 3 times.