SARS-CoV-2 Infects Endothelial Cells In Vivo and In Vitro

Abstract

SARS-CoV-2 infection can cause fatal inflammatory lung pathology, including thrombosis and increased pulmonary vascular permeability leading to edema and hemorrhage. In addition to the lung, cytokine storm-induced inflammatory cascade also affects other organs. SARS-CoV-2 infection-related vascular inflammation is characterized by endotheliopathy in the lung and other organs. Whether SARS-CoV-2 causes endotheliopathy by directly infecting endothelial cells is not known and is the focus of the present study. We observed 1) the co-localization of SARS-CoV-2 with the endothelial cell marker CD31 in the lungs of SARS-CoV-2-infected mice expressing hACE2 in the lung by intranasal delivery of adenovirus 5-hACE2 (Ad5-hACE2 mice) and non-human primates at both the protein and RNA levels, and 2) SARS-CoV-2 proteins in endothelial cells by immunogold labeling and electron microscopic analysis. We also detected the co-localization of SARS-CoV-2 with CD31 in autopsied lung tissue obtained from patients who died from severe COVID-19. Comparative analysis of RNA sequencing data of the lungs of infected Ad5-hACE2 and Ad5-empty (control) mice revealed upregulated KRAS signaling pathway, a well-known pathway for cellular activation and dysfunction. Further, we showed that SARS-CoV-2 directly infects mature mouse aortic endothelial cells (AoECs) that were activated by performing an aortic sprouting assay prior to exposure to SARS-CoV-2. This was demonstrated by co-localization of SARS-CoV-2 and CD34 by immunostaining and detection of viral particles in electron microscopic studies. Moreover, the activated AoECs became positive for ACE-2 but not quiescent AoECs. Together, our results indicate that in addition to pneumocytes, SARS-CoV-2 also directly infects mature vascular endothelial cells in vivo and ex vivo, which may contribute to cardiovascular complications in SARS-CoV-2 infection, including multipleorgan failure.

Keywords: SARS-CoV-2; animal models; aorta ring; endothelial cell infection; hACE2.

Figure 1

Perivascular inflammation and infection of pulmonary ECs by SARS-CoV-2 in Ad5-hACE2 mice. (A) Perivascular inflammation in both Ad5-empty (top images) and Ad5-hACE2 mice (bottom images) 3 days post infection (dpi). Control Ad5-empty mice have lower numbers of nucleated cells compared to Ad5-hACE2 mice. The mask of the infiltration analysis (top right and bottom right) shows the 100μm interface surrounding the tunica adventitia where the cell density was quantified: 5,461 cells/μm (top right) and 8,775 cells/μm (bottom right). (B) Quantification of perivascular inflammation of SARS-CoV-2 (CoV-2) infected mouse lungs on 3, 6, and 12 dpi. Perivascular inflammation was quantified by the density of nucleated cells within 100μm of the tunica adventitia of small arterioles, n = 4-6 for each time point per group. * indicates P<0.05 using two-tailed, unpaired t-test. (C) Representative immunostaining images showing co-localization (arrowhead) of SARS-CoV-2 proteins with CD31 in the lung of infected Ad5-hACE2 mice on 3 dpi. Autofluorescence (blue) differentiates red blood cells which fluoresce in the 488 (green), 568 (red), and 647 (blue) channels. Green: SARS-CoV-2; white: DAPI; red: CD31; blue: autofluorescence.

Figure 2

RNA of SARS-CoV-2 spike protein can be found inside endothelial cells. (A) Representative RNAscope images showed co-localization of RNA of spike protein with CD31 RNA in SARS-CoV-2 infected Ad5-hACE2 mouse lungs (right panel, n = 2), but not in lungs of control mice transduced by DMEM (left panel, n = 3) or Ad5-empty (middle panel, n = 3). White arrow: Spike RNA signals showing light green; black arrow: CD31 RNA signals showing red; yellow arrows: co-localization of two RNAs showing purple. (B) Immunoelectron microscope showed SARS-CoV-2 proteins (black particles indicated by red arrows, n = 1) inside endothelial cells of infected Ad5-hACE2 mice lungs. Pneu, pneumocyte; Er, erythrocyte.

Figure 3

The infection of pulmonary ECs by SARS-CoV-2 in SARS-CoV-2 infected non-human primates and patients who died from severe COVID-19. (A) Representative image (left panel, n = 3) shows colocalization (yellow, arrows) of SARS-CoV-2 (green) and CD31 (red) in the lung of an African green monkey (AGM). Channel separation (right in inset) demonstrates that the SARS-CoV-2 and CD31 signal form an identical pattern, and that there is no autofluorescence (blue) in these areas. Quantitative analysis of SARS-CoV-2-infected endothelial cells in the lung of Naïve and infected AGM (right panel). Multilabel immunofluorescence histochemistry was used to quantify the proportion of SARS-CoV-2 positive CD31+ endothelial cells in total infected cells in the lung of AGMs (PA16, PA20 and PA24). White=DAPI, green=SARS-CoV-2, red=CD31, and blue=autofluorescence. *P < 0.05 vs naïve monkeys by one tail student T test (B) Representative RNAscope image of SARS-CoV-2 and CD31 RNA in the lung of infected AGM (n = 3). Yellow arrows: co-localization of Spike (shown in red) with CD31 (shown in light blue) RNA signal (shown in purple). (C) Colocalization of SARS-CoV-2 with CD31 in two cells (arrows) of a patient that died from COVID-19 (Case 3). Channel separation distinguishes true signal from red blood cells (arrowheads) that lack nuclei (DAPI channel) and can be seen in the autofluorescent/empty channel (blue). (D) RNAscope of SARS-CoV-2 RNA (red, black arrow) in the lung of the decreased COVID-19 patient (Case 3) (E) RNAscope of SARS-CoV-2 RNA (red) and CD31 RNA (green) in the lung of Case 2 and Case 1. Co-localization of SARS-CoV-2 RNA with CD31 was found in the lung of Case 2 patient (left panel) but not Case 1 patient (right panel). Yellow arrows: co-localization of Spike (showing in red) with CD31 (shown in light blue) RNA signal (shown in purple).

Figure 4

(A–M) Immunostaining for SARS-CoV-2 and double immunostainings for SARS-CoV-2 plus ACE-2 or SARS-CoV-2 plus CD34. (A) SARS-CoV-2 staining of the nasal mucosa (used as positive control) and single epithelial cells show the dark signals (arrow) indicating SARS-CoV-2 infection. (B–D) Tissue sections of mouse aorta after ARA and SARS-CoV-2 infection. Dark signals in the innermost layer called intima (I). Single cells covering the collagen gel area inside the aortic lumen (cl-i) also stained positive for SARS-CoV-2 (arrows) while others are negative (arrowhead). (E) ARA section exposed to SARS-CoV-2 infection but treated by secondary antibody only in the immunostaining as control. (F–G) Tissue sections of ARA that were not exposed to SARS-CoV-2 infection but stained with both primary and secondary antibodies in the immunostaining procedure. Cells of the aortic intima that sprouted into the aortic lumen are also visible (arrow). Figure panels (A–G) Dark staining: SARS-CoV-2 staining, red staining: counterstaining with Calcium red, cl-i: collagen gel inside the aortic lumen, cl-a: collagen gel around the aortic wall, I: intima, M: media, A: adventitia. Figure panels (H–M) Double immunostainings for SARS-CoV-2 plus ACE-2 or SARS-CoV-2 plus CD34 on tissue sections of ARA and FIA. (H): Co-localization of SARS-CoV-2 (dark signals) and ACE-2 (red) in intimal ECs and adventitial sprouting cells (arrowhead: intimal ECs; arrows: sprouting adventitial cells). Inset: higher magnification demonstrates the clear co-localization of both dark and red staining signals in the aforementioned cells. Of note, some SARS-CoV-2 positive stained cells (white arrows, dark staining) remained negative for ACE-2. (I) only ACE-2 staining (red signals) are detectable in intimal ECs and adventitial sprouting cells (arrows) in ARA without exposure to SARS-CoV-2. (J) No specific staining signals are detectable in the tissue section of FIA. Note the autofluorescence signals (red) of elastic laminae of the aortic wall. (K) Co-localization of SARS-CoV-2 (dark staining signals) and CD34 (green) in intimal ECs and sprouting adventitial cells (arrowhead: intimal ECs; arrow: sprouting adventitial cells). Inset: higher magnification demonstrates the clear co-localization of both dark and green staining signals in the aforementioned cells. Note that tissue sections in (H) and (K) were prepared as serial sections from the same ARA, demonstrating a very similar pattern of cellular localization. (L) Only CD34 staining (green) is detectable in intimal ECs and sprouting adventitial cells (arrows) in ARA without exposition to SARS-CoV-2. (M) Only CD34 staining (green) is detectable in the wall of FIA, particularly strong in cells within the aortic adventitia as expected. I: intima, M: media, A: adventitia.

Figure 5

(A) Light and electron microscopic analyses of SARS-Cov-2-infected ARA. (A) toluidine blue-stained aortic wall section displaying the layers intima (I), media (M) and avdentitia (A). (B) Tissue section of SARS-CoV-2-infected ARA in EM with low and (C) higher magnification displaying the aortic wall layers as in (A). (D) Higher magnification shows the aortic intima with EC and inner elastic lamina (iel) (E–I) SARS-CoV-2 viral particles (arrows) in a EC process into the aortic lumen (E), in the intimal layer (arrows) (F), inside of an EC process within the collagen matrix (arrowhead) (G), attached to the inner elastic lamina (iel) (H), in a cell that was migrated into the collagen gel (cl-a) around the aortic wall (I). I: intima, M: media, A: adventitia; EC: endothelial cell; cl-a: collagen gel around the aortic wall; iel: inner elastic lamina of the aortic wall. (B) Light and electron microscopic analyses of non-infected ARA. (A) Toluidine blue-stained aortic wall section displaying the layers intima (I), media (M) and adventitia (A). (B, C) tissue section of non-infected ARA in EM with low and higher magnification displaying the aortic wall layers as in A. (D–F): No SARS-CoV-2 viral particles were detected in the cells of the aortic wall, neither in ECs of the intima (D, E) nor in those sprouted into the collagen matrix (F). I: intima, M: media, A: adventitia; cl-a: collagen gel around the aortic wall; iel: inner elastic lamina of the aortic wall.

Figure 6

Endothelial cell infection occurs with the presence of hACE2 in the SARS-CoV-2 infected Ad5-hACE2 mice. (A) Representative image shows co-localization (arrowhead) of SARS-CoV-2 proteins (green) with hACE2 (blue) and CD31 (red) in the lungs of infected Ad5-hACE2 mice at 3 DPI (n = 5). (B) The ratio of infected hACE2⁺ endothelial cells (hACE2⁺, SARS-CoV-2⁺ and CD31⁺) to total infected cells. (C) Representative image shows co-localization (arrowhead) of SARS-CoV-2 proteins (green) with hACE2 (blue) and Pan-CK (red) in the lungs of infected Ad5-hACE2 mice at 3 DPI (n = 5). (D) The ratio of the infected hACE2⁺ pneumocytes (hACE2⁺, SARS-CoV-2⁺ and Pan-CK⁺) to total infected cells. Multilabel immunofluorescence histochemistry was used to quantify the proportion of SARS-CoV-2 positive cells expressing the markers PanCK and CD31 with or without ACE2, in the infected Ad5-hACE2 mice lung at 3DPI (Mean ± SEM, n = 5, quantitative analysis was performed from 96 and 106 microscopic regions from 5 fluorescent immunohistrochemsitry staining slides of the 5 infected Ad5-hACE2 mice respectively (1 slide per mouse). WT: SARS-CoV-2-infected B6 wild type mice; and Ad5-hACE2: SARS-CoV-2-infected Ad5-hACE2 mice. ***P < 0.001 vs WT, and ****P < 0.0001 vs WT.