Hamster lung and brainstem ex vivo cultures are viable and susceptible to for SARS-CoV-2 infection.
Since SARS-CoV-2 mainly targets lungs, and may also infect the brainstem, we have developed new ex vivo models of these organs from naive suckling hamsters, based on our previous experience organotypic cerebellum cultures 23. Lung and brainstems were isolated and sliced, at 500 µm and 350 µm thickness respectively, based on the stability of each structure in the slicing process. The 3-dimensional cultures were then maintained on a polytetrafluoroethylene (PTFE) membrane in order to maintain an air-liquid interface for up to 4 days (Fig. 1a). The metabolic activity, the main parameter reflecting viability of the cultures, did not decrease over time as quantified by Alamar blue assay (Fig. 1B). To evaluate the permissiveness of ex vivo cultures to SARS-CoV2 infection, we quantified the activity of the entry receptor and the expression level of the proteases known to cleave and mature the virus surface Spike glycoprotein (S) as well as of the virus receptor ACE2. The ACE2 activity/µg of total protein lysate from hamster organs on the day of slice preparation was assessed in the lung and brainstem cultures and compared to cerebellum cultures used as a reference. ACE2 inhibitor was used to confirm that the quantified activity was real. The three cultures were found to exhibit a similar ACE2 activity, with the lung displaying a slightly higher activity (Fig. 1c). The transcription level of TMPRSS2 and Cathepsin B mRNA, two proteases known to cleave and activate SARS-CoV-2 S, were quantified in cultures by RT-qPCR. As in human tissues, TMPRSS2 was highly expressed in hamster lung but at very low level in brainstem and cerebellum. On the contrary, Cathepsin B mRNA expression level was significantly higher (≈2*106 mRNA copies/µg of RNA) in the brain cultures compared to the lung (3.6*105 mRNA copies/µg of RNA) (Fig. 1d). These results suggest that the lung organotypic model harbor all main components required for SARS-CoV-2 infection while the brain derived models express fewer TMPRSS2 and might be less permissive to SARS-CoV-2 infection.
Organotypic cultures are susceptible to SARS-CoV-2 infection
In order to evaluate the infectability of these new models, hamster lungs, brainstem (and cerebellum) organotypic cultures were infected with recombinant SARS-CoV-2_neon green (icSARS-CoV-2-mNG). In parallel, to compare the permissiveness of the organotypic cultures with other encephalitic respiratory viruses, the cultures were also infected with recombinant Nipah virus (NiV) and measles virus (MeV), both expressing enhanced green fluorescent protein (EGFP) (referred to as rNiV-EGFP and MeV IC323-EGFP-F L454W, a CNS adapted MeV that can infect in the absence of know receptors25, respectively). Viral entry (Fig. 2a,b,c) and dissemination (Extended Data Fig. 1) were followed by microscopy at day 1 and 4 days post infection (dpi). Interestingly, icSARS-CoV-2-mNG and NiV-EGFP entered and disseminated in both lung and brain cultures (Fig. 2a,b and Extended Data Fig. 1) whereas MeV IC323-EGFP-F L454W only infected the brainstem and the cerebellum (Fig. 2c and Extended Data Fig. 1). Even at 4 dpi the hyperfusogenic MeV did not infect any cells in the lung cultures, suggesting that MeV entry in lung epithelial cells may require additional factors in this model (Fig. 2c and Extended Data Fig. 1). SARS-CoV2 infection was then followed by genome quantification by RT-qPCR. The viral load was higher in lung than in brainstem and cerebellum (Fig. 2d). This difference was already observed on the day of infection (Fig. 2d). The greater susceptibility of lung cultures might arise either from the thickness of the slice (thicker than that of the brainstem and cerebellum cultures) or perhaps from a greater number of ACE2 expressing cells (as supported by the slightly higher ACE2 activity (Fig. 1c). In the lung, the viral replication was very fast and the number of SARS-CoV-2 genomes/µg of RNA almost reached the plateau after 1 dpi (1.2*108 genomes/µg of RNA). In the brainstem and cerebellum, the replication was delayed, with an increase in genome copies being observed only after 2 days, and reached values lower than those in the lung (Fig. 2d). Altogether, these results show that all analyzed ex vivo cultures are susceptible to the SARS-CoV-2 infection, although with the slightly different kinetics.
SARS-CoV-2 infection can be blocked by Remdesivir in organotypic lung and brainstem cultures
Remdesivir is in clinical use for COVID-19 26–28 and we used it to validate our models for drug evaluation. The slices were treated daily at two different concentrations of Remdesivir starting two hours after infection (Fig. 2e). After 4 days of treatment the metabolic activity of both lung and brainstem remained very close to 100% and to that of non-treated slices, suggesting the very low, not to say null, effect on metabolic activity of the drug at these doses (Fig. 2f). The lower dose (2 µM) of Remdesivir did not have a significant inhibitory effect on infection as assessed by RT-qPCR. However, after treatment with 10µM of Remdesivir, infection was reduced by almost 100% at 4 dpi in both lung and brainstem. These data confirm utility of these models for assessing drugs prior to in vivo experiments.
SARS-CoV-2 preferentially targets neurons in brain, and ciliated cells, type 1 and type 2 pneumocytes in lungs
To evaluate which cells are the main to be targeted, viral tropism was evaluated by transmission electron microscopy (TEM) in lung and brainstem organotypic cultures, followed by immunofluorescent staining analyzed by confocal microscopy (Fig. 3). Based on the kinetics of virus replication, all the organotypic cultures shown in figures.3 and 4 were collected at 1 dpi for the lung and at 2 dpi for the brainstem and cerebellum. In lung cultures infection was observed in type 1 and type 2 pneumocytes, as well as in the ciliated cells from the general area of the bronchioles (Fig. 3a,b,c). The immunofluorescence analysis confirmed these observations by showing the presence of SARS-CoV-2_S staining in cells positive for surfactant protein C (SP-C), Aquaporin 5 (AQP5) and α acetylated Tubulin (Tub) staining that are specific for type 2, type 1 pneumocytes and ciliated cells, respectively (Fig. 3d,e,f). Most of the cells display microvillae as expected in young animals (Fig. 3c and Extended Data Fig. 2a). Infected cells harbored a large number of vacuoles and showed multiple signs of cell degradation: cytoplasmic material degradation, membrane coiling (blue star), large empty vacuoles (green arrow) (Fig. 3b1,b2). We observed autophagosomal vacuoles containing virions or degraded viral particles in all types of infected cells (Fig. 3a,b,c,g and Extended Data Fig. 2c). Virions were also found attached on the microvillae outside the cells (Extended Data Fig. 2a) and several cells showed disorganization of the smooth endoplasmic reticulum (SER), as well as accumulation of lipids and mitochondria that were undergoing degradation (Extended Data Fig. 2b).
In the brainstem and in the cerebellum, the TEM analysis showed viruses in granular neurons (Fig. 3g and Extended Data Fig. 2d) with a developed Golgi apparatus. Moreover, the viral particles are often localized in double-membraned vacuoles (red arrows) inside the cells where other Coronavirus are generally observed during their cell cycle (Fig. 3g1,g2)29,30. In the immunofluorescence analysis, SARS-CoV-2_S staining colocalized with NeuN positive cells (Fig. 3h), confirming the infection of granular neurons. Myelin Basic Protein (MBP) staining surrounds the SARS-CoV-2_S without colocalization, suggesting that the cells positive for the infection could also be myelinated neurons and not oligodendrocytes (Fig. 3i). The cultures were also stained for microglia marker (Iba1), astrocytes (GFAP) and Olig2 (used as a second marker for oligodendrocytes) and SARS-CoV-2_S staining was not found in these cells (Fig. 3i,j,k). In the cerebellum, the TEM analysis showed the infection of Golgi neurons with viral particles in autophagosomes. We did not observe infection of Purkinje neurons (Extended Data Fig. 2d1,d2), suggesting that under these conditions there is a selective infection of specific neuronal subtypes by SARS-CoV-2.
SARS-CoV-2 in these models infects almost all epithelial lung cells, but is selective for neuronal subtypes in the CNS. In both lungs and brain organotypic cultures, the infection led to a marked cell degeneration that could conceivably affect organ functions.
Apoptotic, necroptotic and pyroptotic signatures are detected in both organotypic cultures
Since unbalanced inflammatory responses can provoke organ failure, we evaluated cell death signatures in the SARS-CoV-2-infected organotypic cultures. First, we performed TEM analysis which highlighted the presence of apoptotic and necrotic cellular disorders in both infected lungs and brainstem (Fig. 4a,b). The involvement of apoptosis was verified by TUNEL assays in both organotypic cultures (Fig. 4c,d). TUNEL staining was observed similarly in both non infected and infected cultures, potentially due to the experimental procedure (data not shown). However, most of the cells positive for SARS-CoV-2 staining were not positive for TUNEL, confirming that apoptotic cell death observed by TEM might not be the direct consequence of viral infection. Alternatively, the cell death related to viral infection may be caspase-independent (Fig. 4c,d). Using RT-qPCR we corroborated necroptotic events in line with microscopic observations despite an erratic expression of Tumor Necrosis Factor α (TNFα) throughout the 4 days of infection (Fig. 4f). Indeed, we observed in both cultures a sharp increase in Mixed Lineage Kinase Domain Like Pseudokinase (MLKL) mRNA levels (Fig. 4e) which is known to be associated with Caspase 8 deficiency and Inflammatory Bowel Disease commonly observed in patients. Moreover, we showed that pyroptosis also occurs during viral infection as inferred from the increase in Gasdermin D mRNA levels in both infected lungs and brainstem cultures (Fig. 4g). Gasdermin D is also known to be substrate of inflammation-related caspases, thus imbalancing inflammatory response potentially leading to organ failure 31,32. Interestingly, whereas Gasdermin D levels decreased at day 4 post-infection in the lungs, its expression kept increasing in the brainstem, possibly due to a difference in infection kinetics between both tissues. Furthermore, while the levels of Interleukin 18 (IL-18) mRNA remained low in both organotypic cultures (Fig. 4h), we documented a difference in the expression of Interleukin 1 β (IL-1β) mRNA that increased in the brainstem while it remained low in the lungs (Fig. 4i). These data reveal that distinct cellular mechanisms lead to pyroptosis in the two organs.
Innate and inflammatory responses are increased in both organotypic cultures
To characterize the recapitulation of the responses of these models to SARS-CoV-2 infection, we transcriptomically profiled infected and uninfected organotypic cultures of both hamster brainstem and lungs (Fig. 5a,b,c,d). The transcriptomic first level of analysis pointed out the strong stimulation of the immune response with 19 and 20 out of 20 mainly altered Gene Ontology (GO) categories related to immunity in lung and brainstem respectively (Fig. 5a,b). Alternatively, eight GO categories related to lymphocyte responses were altered in brainstem versus two in lung. To go further, the dichotomy in cellular responses occurring in both tissues at day 4 post-infection was confirmed by the gene expression patterns, highlighting organ-dependent specificities in the host response to the infection (Fig. 5c,d). Indeed, the most significantly differentially expressed genes (DEG) in the lung include a plethora of upregulated interferon-stimulated genes (Fig. 5c), while these significant DEG brainstem conversely contained many downregulated neuronal markers (Fig. 5d). This is consistent with the observation that the percentage of polyadenylated transcripts aligning to the SARS-CoV-2 genome is 6.74 fold higher in lung than brainstem (2.90% vs 0.43%) (Fig. 5c,d).
In parallel, specific immunological markers including Myxovirus Resistance 1 (MX1), Interferon Stimulated Exonuclease Gene 20 (ISG20), C-X-C Motif Chemokine Ligand 10 (CXCL10) and C-C Motif Chemokine Ligand 5 (CCL5) mRNA were quantified by RT-qPCR (Fig. 5e,f,g,h). However, specific genes encoding interferon-stimulated genes (ISG) or chemokines were similarly upregulated in both organotypic cultures, pointing to these as potentially relevant to the innate immune response. Indeed, RT-qPCR profiles showed that the expression of MX1 and ISG20 ISGs and of CCL5 and CXCL10 chemokines were increased within the 4 days kinetics following SARS-CoV-2 infection in both cultures. Interestingly, ISG20 mRNA amounts decreased rapidly to lower levels determining that its reducing antiviral exonuclease activity may be compensated at later time (Fig. 5f). Moreover, while exhibiting a similar trend, we noticed that all those responses were delayed in brain cultures compared to lung ones (Fig. 5e,f,g,h).