Wide distribution and replication of SARS-CoV-2 in organs of rhesus macaques. SARS-CoV-2 is a positive-sense single-stranded RNA virus, and its replication is thought to occur in double-membrane vesicles, where negative-sense RNAs are produced. Based on this viral information, green and red fluorescence-labeled probes were designed to recognize sense and antisense viral RNAs, respectively. Combined with a RNAscope assay, the cellular replication of SARS-CoV-2 was well detected. Here, we used SARS-CoV-2 to infect rhesus macaque, and performed RNAscope to investigate both the in situ distribution and replication of SARS-CoV-2 in different tissues, including lung, heart, cerebral cortex, cerebellum, liver, kidney microvascular vessel and testis (Fig. 1 and Extended Data Fig. 1). The rhesus macaques were infected with SARS CoV-2 virus at 106 TCID50, and the tissues were collected at 7 days post infection (dpi) in the ABSL3 laboratory11. Upon SARS-CoV-2 infection, a severe lung interstitial inflammation with extrudate of proteins in alveoli was evidenced in all animals examined (Fig. 1a), implying a well-established rhesus macaque model for COVID-19. All RNAscope analyses for expression (Green fluorescence, FITC-labelled) and replication (Red fluorescence, Cy3-labelled) of SARS-CoV-2 were performed on formalin-fixed paraffin-embedded (FFPE) sections. The fluorescence signals were acquired and analyzed using TissueGnostics. We found that in addition to the severely infected lung epithelial cells showing a focal signal for both SARS-CoV-2 expressions and replications (Fig. 1a), myocardiac cells in left and right atria and ventricles also displayed expression of SARS-CoV-2, mainly located at both side of enlarged nuclei (Fig. 1b). Likewise, abundant SARS-CoV-2 can be detected in testicular Sertoli cells (Extended Data Fig. 1a), probably due to their high expression of angiotensin-converting enzyme 2 (ACE2)12. However, only a small portion of hepatocytes showed invasion of SARS-CoV-2 (Extended Data Fig. 1b). In the kidney cortical regions, sparse expression and replication of SARS-CoV-2 was observed in proximal, distal convoluted tubules, and collecting tubules (Extended Data Fig. 1c), as well as endothelial cells of the interstitial micro-vessels (Extended Data Fig. 1d). No SARS-CoV-2 signal was observed in Leydig and spermatocytes. Notably, we found a sparse signal of SARS-CoV-2 expression and replication in various cerebral cortex regions, including cerebral cortical neuronal cells (Fig. 1c) and cerebellar Purkinje cells (Fig. 1d). Collectively, these data showed a wide distribution and replication of SARS-CoV-2 in multiple organs of rhesus macaques, with rich levels in the lung, heart and testis (Supplementary Table 1), suggesting its higher infectivity and potential pathogenicity.
Distinct transcriptomic features of rhesus macaque organs. To investigate the organ-specific transcriptomes of the healthy control samples, we obtained the transcriptomic profiles by RNA-seq in each of 14 organs from non-infected control rhesus macaque (Fig. 2a). The tissue markers specifically for different organs were selected from Human Cell Landscape13 and used to validate the data quality. The markers with high expression levels in their corresponding organs were demonstrated (Fig. 2b, c, Extended Data Fig. 2a and Supplementary Table 2), justifying our collected samples suitable for further analysis. We then performed principle components analysis (PCA) and hierarchical clustering for different organs and found that functionally closed organs were well clustered, such as tissues of cerebral cortex, brainstem, cerebellum and medulla of the brain organ, intestine and stomach of digestive system, and left ventricle, right ventricle and muscle related cardiovascular system (Fig. 2d and Extended Data Fig. 2b).
To further identify the organ-specific genes and their functions in rhesus macaque, we performed K-means clustering analysis for all tissues, and 8 clusters were categorized by their whole transcriptomic profiling (Fig. 2e,g, Extended Data Fig. 2c-h and Supplementary Table 3). The genes from different clusters displayed highly conserved enrichments in both their tissue-specific expressions and organ-related functions. For example, genes from Cluster 1 were highly expressed in the brain organ, especially medulla (Fig. 2e), and showed enrichment in nervous system related pathways, including locomotory behavior, positive regulation of axonogenesis, learning and so on (Fig. 2f), while the cardiovascular system related genes were mainly identified in Cluster 2 and shown high expression levels in heart and muscle organs (Fig. 2g,h). A similar enrichment mode was also observed in genes from Cluster 3 to 7, such as Cluster 3 in cerebellum with functional enrichment in synapse assembly, Cluster 4 in lung and kidney with smoothened signaling and hypoxia, Cluster 5 in liver with cholesterol metabolism, Cluster 6 in intestine with carbohydrate metabolic processing and Cluster 7 in spleen with innate immune and inflammatory responses (Extended Data Fig. 2c-g; 2i-m). Moreover, genes from Cluster 8 were ubiquitously expressed in all organs (Extended Data Fig. 2h) with enrichment of fundamental cellular functions, like translation and RNA processing (Extended Data Fig. 2n). These results illustrated tissue/organ-specific features of whole transcriptome in gene expression and functional enrichment.
Transcriptomic landscape of multiple tissues in SARS-CoV-2 infected rhesus macaques. To determine the SARS-CoV-2-induced transcriptional changes in 14 tissues, the tissues were collected from three rhesus macaques infected with the virus for 7 days and subjected to RNA sequencing. Based on the whole transcriptome profiles, lung samples from control and infected rhesus macaques were distinctly separated, while other control and infected tissues from the same organ were clustered together, suggesting that lung tissue underwent serious injury at the early stage of infection 14 (Extended Data Fig. 3a and Supplementary Table 4). Intriguingly, post virus infection, the transcriptomic profiles from right ventricle was clustered with brain tissues except cerebral cortex which showed a similar expression pattern with uninfected right ventricle (Extended Data Fig. 3a). This distinct clustering result was further validated by t-distributed stochastic neighbor embedding (t-SNE) analysis (Extended Data Fig. 3b), suggesting a differential yet closely related responses to SARS-CoV-2 infection in right ventricle and cerebral cortex.
To further evaluate the differentially expressed genes (DEGs) in multiple organs post SARS-CoV-2 infection, we compared all the infected organ samples with their corresponding controls. Since the infected rhesus macaques include 2 male and 1 female, the DEG analysis for testis was only based on 2 male infected samples. We totally identified 18,730 dysregulated genes (fold change ≥ 2 and P value < 0.05) among all 14 tissues and most of them were located in cerebral cortex, cerebellum and right ventricle (Fig. 3a and Supplementary Table 5). We further compared the expression level of ACE2, the first identified receptor for SARS-CoV-215, in multi-tissues between control and infected groups (Fig. 3b). Intriguingly, among all the tissues tested, ACE2 was highly expressed in intestine, which is consistent with previous reports16,17, and remarkably increased in lung15,18 and testis12 post infection (Supplementary Tables 4–5). Moreover, we performed both PCA and hierarchical clustering analysis for expression changes among all dysregulated genes in 14 organs, and found that right ventricle was more closer to cerebellum and brainstem but apart from cerebral cortex (Fig. 3c,d), suggesting that differential signaling pathways are involved in the individual tissue/organ response to SARS-CoV-2 infection.
Spatial response of cerebral cortex, cerebellum and right ventricle. We then performed Gene Ontology enrichment analysis to determine the up- or down-regulated genes and their related functions in each organ upon SARS-CoV-2 exposure (Fig. 3e and Extended Data Fig. 3c). The transcriptomic changes in each type of organ post SARS-CoV-2 infection were involved in diversely distinct functions. Interestingly, cerebellum and right ventricle shared similar transcriptomic changes in both gene dysregulation and annotated functions (Fig. 3c,d), such as up-regulated genes related to with skeletal system development, actin cytoskeleton organization and signal transduction pathways (Fig. 3e) and down-regulated ones involving cilium assembly and telomere maintenance pathways (Extended Data Fig. 3c). Beyond that, the other up-regulated genes of right ventricle were significantly annotated in chemical synaptic transmission, locomotory behavior, neurotransmitter secretion, nervous system development and learning pathways (Fig. 3e), suggesting that the homeostatic genes for nerve cells were significantly affected in right ventricle post SARS-CoV-2 infection, resulting in an enhanced secretion of neurotransmitter at early stage of viral infection. Moreover, specific transcriptomic changes in cerebral cortex post infection (Fig. 3c,d and Extended Data Fig. 3a,b) showed that the down-regulated genes were mainly related to synapse, locomotory behavior and learning terms (Extended Data Fig. 3c), while up-regulated ones participated in muscle contraction, transcription, angiogenesis and Nuclear factor-kappa B (NF-κB) signaling pathways (Fig. 3e). This data suggests that genes related to immune response were specifically induced in the cerebral cortex post SARS-CoV-2 infection, whereas genes functioning in the original learning and motor behavior functions were inhibited.
Relative to cerebellum and right ventricle, cerebral cortex showed an opposite tendency in dysregulated genes in terms of differential expression and functional annotation post SARS-CoV-2 infection (Fig. 3c-e and Extended Data Fig. 3b-d). Moreover, the same opposite changes were also observed on those significantly dysregulated genes in cerebral cortex relative to cerebellum and right ventricle (Extended Data Fig. 3e). A relatively higher proportion of intersection, especially between cerebral cortex and right ventricle, was also observed (Extended Data Fig. 3f). Specifically, some metabolic process related genes were up-regulated in cerebral cortex but down-regulated in cerebellum/right ventricle (Extended Data Fig. 3g), while nervous system development related genes exhibited a reverse pattern (Extended Data Fig. 3h). These results suggest that brain and heart undergo drastic transcriptomic reprogramming at the early stage of SARS-CoV-2 infection.
Elevated innate immune response of cerebral cortex. Since innate immune response is the first line of defense against SARS-CoV-2 infection, which triggers the production of interferons (IFN), pro-inflammatory cytokines and chemokines19, we further investigated the expression of interferon-related genes in each organ (Fig. 4a). Intriguingly, the dysregulated genes were enriched in cerebral cortex, cerebellum and right ventricle, and moreover, interferon gamma (IFN-γ) receptor 1 (IFNGR1) and interferon lamda (IFN-λ) receptor 1 (IFNLR1) expressions were up in cerebral cortex but down in cerebellum and right ventricle (Fig. 4b), a similar opposite pattern as the cluster analyses on whole transcriptome and dysregulated genes (Fig. 3d, Extended Data Fig. 3d,e). We then focused on interferon-stimulated genes (ISGs)20 (Fig. 4c, Extended Data Fig. 4a,c and Supplementary Table 6), and observed that in cerebral cortex, a large proportion of ISGs showed an elevated expression with enrichment in the response pathways to IFN- γ and cytokine (Fig. 4d), while reversely the down-regulated ISGs were predominant in right ventricle and cerebellum and enriched in defense responses to viral and metabolic process, respectively (Extended Data Fig. 4b,d). Moreover, most of the up-regulated ISGs (112/259) in cerebral cortex showed down-regulation in both cerebellum and right ventricle (Fig. 4e) and mainly enriched in type I interferon production, response to cytokine and metabolic process pathways (Fig. 4f). These findings suggest that IFN- γ related genes were specifically up-regulated in the SARS-CoV-2 infected cerebral cortex.
IFN- γ, as the first macrophage-activating factor (MAF) identified so far, facilitates immune response by promoting proinflammatory cytokine synthesis, phagocytosis and antigen-presenting capacity in activated M1 macrophages21,22. To elucidate the immune response of cerebral cortex caused by IFN- γ, we evaluated the expression of inflammatory factors from polarized macrophages, and found that IL-23, the key factor in Th1-antigen-specific responses, as well as CD80 and CD86, the co-stimulatory factors for pro-immune responses, had a significantly higher expression in cerebral cortex post infection (Fig. 4g). In addition, anti-inflammation factors of M2 macrophages also showed high expression in the cortex probably for balancing the inflammatory response (Extended Data Fig. 4e). However, those genes were preferentially down-regulated in the infected cerebellum and right ventricle tissues (Fig. 4g and Extended Data Fig. 4e). To further determine the proportion of different stages of macrophages (M0: inactive macrophages, M1: pro-inflammatory macrophages, M2: anti-inflammatory macrophages), we employed CIBERSORT23 to evaluate the amounts of immune cell types based on transcriptomic data of cerebral cortex, cerebellum and right ventricle tissues (Extended Data Fig. 4f). The cerebral cortex exhibited an increased proportion of macrophages within active M1 and M2 phases upon infection, while most macrophages in cerebellum and right ventricle were in inactive M0 (Fig. 4h), suggesting that cerebral cortex harbors a specifically elevated immune response in the early stage of SARS-CoV-2 infection.
Transcriptional regulatory network of hyperinflammation in cerebral cortex. We then analyzed the signaling network regulated by transcription factors (TFs), which might associate with the susceptibility or severity of SARS-CoV-2 infection24. Among 745 TFs identified in rhesus macaque based on human transcriptional regulatory interactions of TRRUST database25, 567 showed dysregulation in the infected tissues, especially in cerebral cortex, cerebellum and right ventricle (Extended Data Fig. 5a and Supplementary Table 7). By matching the TF-target pairs and the dysregulated genes among these three infected tissues, we built the regulatory interaction network about the up-regulated TFs and their targets by Cytoscape26 (Extended Data Fig. 5b-d). To further determine the infection-elicited TF network, four up-regulated TFs with their most interacting targets were retrieved to build a relatively simple regulatory interaction network for cerebral cortex (Fig. 5a), cerebellum (Fig. 5c) and right ventricle (Fig. 5e). The up-regulated targets in cerebral cortex post infection were significantly annotated in response pathways to inflammation and hypoxia (Fig. 5b), while in cerebellum and right ventricle the up-regulated ones were mainly annotated in transcription pathways (Fig. 5d,f). These findings suggest that SARS-CoV-2 might hijack the host transcriptional regulatory mechanism to induce hyper-inflammatory state via TFs in cerebral cortex, but not in cerebellum and right ventricle.
Hypercytokinemia, thrombosis, angiogenesis and fibrotic factors elevated in cerebral cortex. The hallmark of COVID-19 pathogenesis is the cytokine storm with higher levels of proinflammatory cytokines, such as interleukin-1β (IL-1 β), interleukin-2 (IL-2), interleukin-6 (IL-6), tumor necrosis factor (TNF-α), granulocyte-macrophage colony-stimulating factor (GM-CSF), chemokines (C-C-motif chemokine ligand (CCL)-2, CCL-3 and CCL-5), also interleukin-2 (IL-2), interleukin-7 (IL-7) and, interleukin-10 (IL-10)27. The severe cases with hyperinflammatory syndrome can include coagulation impairment28, abnormal angiogenesis29, cystic fibrosis30 and even death27,31. We then examined the expression of those factors using the transcriptomic data from various tissues of rhesus macaques, and observed their dysregulation in the early stage of infection, especially in cerebral cortex, cerebellum and right ventricle (Extended Data Fig. 6a and Supplementary Table 8).
The cytokines retrieved from M9809 gene set of Molecular Signatures Database in Gene Set Enrichment Analysis (GSEA)32, such as IL-7, CCL2, CXCL10 and IFNs, were significantly up-regulated in cerebral cortex, but markedly down-regulated in cerebellum and right ventricle post infection (Fig. 6a). Moreover, these cytokines were enriched in the pathways of cytokine-mediated signaling pathway and acute inflammatory response (Extended Data Fig. 6b), suggesting a state of hypercytokinemia or cytokine storm in the cerebral cortex. In addition, among the coagulation factors retrieved from gene set of GSEA, over half of them were dysregulated, and showed the most significant changes in three aforementioned tissues. Specifically, the coagulation factors with significant upregulation in cerebral cortex include coagulation factors prothrombin XIIIa (F13A1) 33,34, glycoprotein VWF binding to factor VIII (F8)33,34 and plasminogen activator inhibitor 1 (SERPINE1)35 (Fig. 6b). These up-regulated cytokines were annotated in blood coagulation pathway (Extended Data Fig. 6c), indicating an impaired coagulation in cerebral cortex. Similarly, most angiogenesis factors, annotated in nCounter PanCancer Progression Panel (NanoString Technologies), were aberrantly regulated with significant changes in cerebral cortex relative to cerebellum and right ventricle (Fig. 6c), and enriched in angiogenesis and some signaling pathways (Extended Data Fig. 6d). suggesting an abnormal angiogenetic state in the virus infected cerebral cortex.
Fibrosis is usually divided into four stages: initiation is triggered by a cascade of immune response and stress, inflammation with activated inflammatory signaling pathways, proliferation with differentiation and proliferation of fibroblasts, and modification with restructured extracellular matrix (ECM) composed of immune cells and fibroblasts. We have quantified the expression of 771 fibrosis-related genes from mCounter Fibrosis Panel (NanoString Technologies), and found that 567 genes were dysregulated in at least one tissue post viral infection. We built an interaction network from String database36 using these altered genes by Cytoscape26 (Fig. 6d and Extended Data Fig. 6e). Dysregulated fibrosis markers were dominant in cerebral cortex, cerebellum and right ventricle, while inflammatory fibrosis markers were significantly dysregulated in cerebral cortex (Fig. 6d), revealing an abnormal fibrotic response in cerebral cortex at the early stage of SARS-CoV-2 infection.
Thus, SARS-CoV-2 infection rapidly induces the abnormal state of hypercytokinemia, thrombosis, angiogenesis and fibrosis, in particular the cerebral cortex. As these imbalanced activities could facilitate the formation of microthrombi37, SARS-CoV-2 patients have higher risk to develop microthrombi in brain38,39 and other systems as well4.
SARS-CoV-2 receptors might mediate the differential transcriptomic changes of cerebral cortex. We next examined the expression of reported receptors18,40−44 for SARS-CoV-2 in multi-tissues post infection and observed that Neuropilin-1 (NRP1) and Niemann-Pick disease type C intracellular cholesterol transporter 1 (NPC1) were significantly up-regulated in cerebral, but down-regulated in right ventricle (Fig. 7a,b). It has been reported that NRP1 expression is higher in nasal epithelium rendering susceptible to SARS-CoV-2 infection40,41. It is likely that SARS-CoV-2 may enter the cerebral tissues by crossing the blood-brain barrier via an upper nasal transcribrial route45. Thus, the higher expression of NRP1 in the cerebral tissue might serve as a potential receptor to promote SARS-CoV-2 infection in brain.
To elucidate the potential interactions among organs post infection, we used the ligand-receptor pairs from CellTalkDB46 to examine their transcriptomic changes in the infected organs, and observed a close communication between cerebral cortex and right ventricle/cerebellum (Fig. 7c) with higher expression of ligands in right ventricle and cerebellum and its corresponding receptors in cerebral cortex. Those genes were mainly involved in the regulation of cell migration and axon guidance pathways (Fig. 7d), which is consistent with the findings of significantly up-regulated genes in these two organs (Fig. 3e). The results suggest a plausible mechanism through which the signaling molecules secreted by right ventricle and cerebellum could transmit to cerebral cortex through the circulatory system to target the receptors and induce the responses to SARS-CoV-2 infection (Fig. 7e).