The most common clinical feature reported in patients with COVID-19 is respiratory symptoms (1, 2). In addition to primarily causing pulmonary symptoms, COVID-19 disease is accompanied by vascular pathology, endothelial damage, and vascular coagulopathy (3–5). Reports emerged around the world confirming a disproportionate prevalence of abnormal thrombotic events and vascular pathology in patients with COVID-19, even in those not in intensive care units (6–14). Theories regarding the mechanism of vascular disease observed in patients with COVID-19 have been proposed, including direct infection of endothelial cells and systemic inflammatory responses (15–22). However, these hypotheses remain largely untested, and the cellular basis of vascular pathology remains controversial. In this study, we used a set of new mouse genetic tools (23) to rigorously test endothelial contribution to COVID-19-associated vascular pathology.

Cellular expression of ACE2 is indispensable for SARS-CoV-2 infection in pneumocytes (25, 26), but SARS-CoV-2 is unable to bind mouse ACE2. To determine if endothelial cells directly contribute to lethal infection, we generated animals that express human ACE2 (hACE2) from the mouse Ace2 locus in a manner that enables cell-specific loss of hACE2 using Cre recombinase (hACE2^fl/y mice) (23). We crossed hACE2^fl/y mice onto a Tie2-Cre transgenic mouse line that drives Cre expression in endothelial cells (ECs) to generate mice that express hACE2 in all cells except vascular ECs. hACE2^fl/y; Tie2-Cre⁺ mice and control littermates were exposed to 10⁵ PFU of SARS-CoV-2 virus via nasal inhalation. hACE2^fl/y; Tie2-Cre⁺ mice showed no significant difference in survival after exposure to SARS-CoV-2 compared with the littermate controls (Figure 1A). Histological analysis revealed the presence of alveolar infiltrates and pulmonary vascular thrombi in the lungs of infected hACE2^fl/y; Tie2-Cre⁺ mice that were indistinguishable from findings observed in control hACE2^fl/y mice (Figure 1B). Histological analysis using hematoxylin-eosin staining of tissue sections from the small intestine, kidney, liver, and heart also failed to identify any vascular pathology (Supplementary Figure S1). Next, we evaluated the expression of inflammation-induced protein intracellular adhesion marker 1 (ICAM1) and the pro-coagulant, inflammation-induced protein Von Willebrand factor (vWF) in the mice following the SARS-CoV-2 infection, given both ICAM1 and vWF have been closely associated with COVID-19 induced vascular damage (27, 28). Expression of ICAM1 and vWF were also similar in the lung capillary endothelial cells of SARS-CoV-2-infected hACE2^fl/y and hACE2^fl/y; Tie2-Cre⁺ mice (Figures 1C,D).

The studies described above suggested that endothelial cell infection is not required for vascular COVID-19 pathology when hACE2 is expressed at endogenous levels. In fact, immunostaining of lung sections using anti-ACE2 antibodies was able to detect ACE2 expression in epithelial but not endothelial cells (23) (Figure 1E). To more rigorously test the role of endothelial hACE2, we next crossed Tie2-Cre onto a recently described Cre-activated gain of function hACE2 allele (loxP-stop-loxP-hACE2 or LSL-hACE2^+/0) (23) to over-express hACE2 in vascular endothelial cells. Tie2-Cre;LSL-hACE2^+/0 animals exhibited very high endothelial-specific expression of hACE2, assessed by immunostaining of tissue sections compared with hACE2^fl/y mice (Figure 1F). To ensure that Tie2-Cre;LSL-hACE2^+/0 animals would be productively infected following SARS-CoV-2 exposure, we generated Tie2-Cre;LSL-hACE2^+/0;hACE2^fl/y animals that support robust infection of the nasal and respiratory epithelium (23) (Figure 1F). Despite high levels of endothelial hACE2 expression, we failed to detect nucleocapsid protein that colocalized with PECAM⁺ endothelial cells following nasal SARS-CoV-2 infection (Figure 1F). In contrast, we have previously shown that this gain of function allele is sufficient to drive hACE2 expression and support SARS-CoV-2 infection in both neuronal cells and lung epithelial cells (23). These studies support the conclusion that SARS-CoV-2 does not confer endothelial cell damage and vascular thrombosis through direct viral infection of those cells. They further demonstrate that the levels of circulating virus are too low to infect even endothelial cells that express very high levels of hACE2, and therefore that most COVID-19 pathology arises due to aerosol infection of the nasal and pulmonary epithelium.

It has been debated whether direct viral infection of endothelial cells or indirect damage from systematic inflammation underlie COVID-19-associated vascular pathology (3). Our murine vascular endothelial loss and gain of function studies reported here provide strong in vivo evidence that endothelial ACE2 and direct infection of vascular endothelial cells do not contribute significantly to the diverse vascular pathology associated with COVID-19. These findings are consistent with previously reported in vitro studies that showed human endothelial cells are not readily infected by SARS-CoV-2 (21). Together with our recently reported studies, these findings strongly support a mechanism in which SARS-CoV-2 infection of nasal epithelial and neuronal cells stimulates a powerful inflammatory response that is the cause of COVID-19 vascular pathology.

In the present study, we utilized both loss of function and gain of function hACE2 mouse lines and demonstrated that direct endothelial viral infection does not contribute to COVID-19-associated vascular pathology. Future studies are needed to define the cytokines that likely drive secondary vascular inflammation and thrombosis and to understand the molecular mechanism by which systemic inflammation damages endothelium following SARS-CoV-2 infection. We used the original isolate SARS-CoV-2 USA-WA1/2020 strain in our study because that isolate is the best characterized regarding vascular complication. Omicron BA.1 variant failed to confer lethal disease and associated vascular phenotypes in our mouse models (23). Future studies testing the impact of other variants on the vascular system will be needed. We performed our studies on male mice due to the Ace2 allele being located on the X chromosome, enabling a straightforward comparison. Mouse models are not humans, and our mouse model hACE2^fl/y expresses a higher level of hACE2 as previously reported (23). Thus there are likely to be differences in pathogenic mechanisms identified using our model compared to human studies. However, this difference should bias toward rather than against a direct endothelial infection mechanism and it does not weaken our negative conclusions. Future studies looking at longer-term vascular events in mice with lower levels of hACE2 expression will be needed to address non-acute mechanisms of COVID-19-related cardiovascular disease.