Stimulation of vascular organoids with SARS-CoV-2 antigens increases endothelial permeability and regulates vasculopathy

Brief report

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-associated vasculopathy and endotheliitis are common in severe cases of COVID-19¹. Binding of the viral Spike glycoprotein to the host protein angiotensin-converting enzyme 2 (ACE2) is required for cellular entry, with different proteolytic reactions and co-receptors required for efficient virion infection^2,3. Mechanisms of endotheliitis are poorly understood with conflicting evidence available as to whether SARS-CoV-2 infection damages 2D endothelial monolayers in vitro. Indeed, ACE2 expression in human endothelium is not sufficient for endothelial activation^4,5 and infection of primary lung endothelial cells with SARS-CoV-2 results in limited infection^6,7. Nevertheless, ACE-2-positive endothelial cells and disruption of intercellular junctions are observed in COVID-19 post-mortem lung sections⁸. Single cell RNAseq studies of healthy and fibrotic lungs have shown ACE2 expression on type II alveolar cells in healthy lungs and significant RNA expression on arterial vascular cells in fibrotic lungs⁹, a clinical complication observed in cohorts known to be vulnerable to COVID-19 infection¹⁰. In this letter, we show that SARS-CoV-2 antigens, namely the Spike glycoprotein and Nucleocapsid protein, induce pericyte activation and increase endothelial permeability, and this is associated with increased thrombotic complications in severe COVID-19 patients.

In the absence of direct evidence of endothelial viral uptake and subsequent damage, we reason that altered crosstalk between mural and endothelial cells (EC) promotes endotheliitis and regulates thrombosis. Pericytes are known to highly express ACE2 and regulate vascular integrity, permeability and blood flow^2,6,11. We first assessed the presence of the Spike glycoprotein in the pericytes of lung tissues obtained post mortem from eight patients who died from COVID-19. All patients had comorbidities (Supplementary information). Spike glycoprotein was detected in the pericytes (NG2⁺ cells) of 4 patients, which also expressed high levels of the activation marker ICAM-1 (Figure 1A). As pericyte activation regulates endothelial permeability, we assessed the expression of VE-cadherin (CD144), a key regulator of vascular permeability, in non-ventilated COVID-19 patients. Compared to control lung sections from age-matched non-COVID-19 patients, CD144 expression was decreased in SARS-CoV-2 infected lungs, most markedly in patients with confirmed thrombosis (Figure 1B, C). These results suggest pericyte activation and compromised endothelial barrier function underlie the pathophysiology of severe SARS-CoV-2 infection.

• Download figure
• Open in new tab

Figure 1: Increase in endothelial permeability and pericyte activation in COVID-19 lung sections and SARS-CoV-2 antigens-treated vascular organoids.

(A) Immunofluorescence imaging of NG2⁺, Spike glycoprotein and ICAM-1 in formalin fixed and paraffin embedded lung from patient who died from COVID-19 and control COVID-19^- lung section. (B, C) Representative immunofluorescence imaging of CD144 (VE-cadherin) in lung sections (2 patients who died from non-respiratory-associated diseases (P1, P2), 2 COVID-19 patients without detectable thrombi (P3, P4 Thr^-) and 2 patients with detectable microthrombi (P5, P6 Thr⁺)) as assessed by post-mortem histology. (C) Quantification of CD144 levels using ImageJ shows a significant reduction in CD144 expression in COVID-19⁺ patients, which is more notable in those with thrombosis. (D) SARS-CoV-2 England 2 virus (Wuhan) (M.O.I = 0.5) were added to the organoids for 48h. SARS-CoV2 trimeric Spike (S) and nucleocapsid (N) proteins (100nM) were added to vascular organoids model for 72h. (E) Vascular organoids were generated as reported by Wimmer et al. using a step wise differentiation of human induced pluripotent stem cells. First cells are committed to a mesodermal lineage as non-adherent aggregates, before embedding in a mixed matrix hydrogel which induces vascular sprouting. (F) On maturation, these samples are large 3D networks of branched endothelium (CD31⁺ CD144⁺) supported by pericytes (NG2⁺ CD140b⁺) and fibroblasts (NG2^- CD140b⁺). (G) Immunofluorescence imaging of ACE-2 and CD144⁺ positive cells in organoids (12µm organoid sections). (H) Immunofluorescence imaging of CD144 and nuclei (DAPI) in whole vascular organoids treated with SARS-CoV-2 (M.O.I= 0.5) for 48h or spike glycoprotein (S)(100nM) for 72h. (I) Activation of endothelial, pericytes and fibroblasts in the vascular organoids treated with Spike glycoprotein (S) and nucleocapsid (N), or a combination of both proteins for 72h in the presence or absence of IL1-β (20ng/ml) compared to control was measured by flow cytometry (Dako Cyan cytometer) after collagenase-digestion of vascular organoids. Median of fluorescence (MFI) of CD144 and ICAM-1 on CD31⁺ cells, ICAM-1 and podoplanin (PDPN) on CD140⁺NG2⁺ cells and CD140⁺NG2^- cells were measured by flow cytometry (n=3). (J) Detection of the levels of soluble IL-6 and IL-8 in the supernatant of organoids treated for 72h with viral antigens in the presence or absence of IL-1β by ELISA (n=4-5). One-Way ANOVA with multiple comparisons performed for each statistical test with significance at (* p = < 0.05, ** p = < 0.001).

In order to assess the direct role of SARS-CoV-2 on pericytes and endothelium, we generated 3D human vascular organoids ^12,13 (Fig 1D, E) comprised of vascular endothelium (CD31⁺ CD144⁺), pericytes (CD140b⁺ NG2⁺), and fibroblast (CD140b⁺ NG2^-) (Figure 1D,E, F) (Supplementary information). This model has been recently used to show that ACE2 antibody blockade limits viral replication in vitro¹⁴. ACE2 was primarily observed in pericytes adjacent to CD144⁺ cells, consistent with recent evidence of poor expression of ACE2 on endothelium⁴ (Figure 1G). Organoids were then exposed to i) live SARS-CoV-2 virus; ii) recombinant active trimeric spike glycoprotein (S) ^15,16; iii) nucleocapsid protein (N) ¹⁷ or iv) a combination of both (S + N). Endothelial survival, permeability and activation, as well as pericyte and fibroblast survival and activation were assessed by flow cytometry and immunofluorescence (Figure 1D,E).

Treatment of vascular organoids with SARS-CoV-2 decreases CD144 expression on EC as observed by immunofluorescence, and trimeric S mimics this effect (Figure 1H). We next assessed the effect of S, N or S+N on endothelial, pericyte and fibroblast activation by flow cytometry. The decrease of CD144 was also observed following treatment with N and S+N Figure 1H, I) without significant changes in ICAM-1 (Figure 1I), CD62P or VCAM-1 expression on EC (data not shown for both), indicating that viral antigens do not directly activate the endothelium. Similarly, no activation of pericytes as measured by an increase in ICAM-1 expression was observed (Figure 1I). To assess whether endothelial activation alter EC response to S, N or S+N, we investigated the effect of SARS-CoV-2 antigens on cell activation in the presence of IL-1β, a chemokine known to activate endothelium and a critical component of the COVID-19 cytokine storm¹⁸. The SARS-CoV-2 antigen-dependent decrease in CD144 was not altered by the addition of IL-1β. Endothelial and pericyte cell activation by IL1-β, as assessed by the upregulation of ICAM-1, was not altered by the addition of SARS-CoV-2 proteins (Figure 1I). Neither recombinant SARS-CoV-2 antigens or IL-1β induced activation of fibroblasts in the vascular organoids, as measured by podoplanin expression. These results indicate that the SARS-CoV-2 antigens alter pericyte and endothelial survival and decrease CD144 on endothelial cells, increasing endothelial permeability.

Activated EC and pericytes release inflammatory cytokines such as IL-6 and IL-8, and higher levels are observed in severe COVID-19 patients ¹⁹. The levels of soluble IL-8 and IL-6 in the supernatant of treated organoids were modestly increased with SARS-CoV-2 antigens (Figure 1I). IL-1β increased IL-8 levels without further change by addition of SARS-CoV-2 antigens, showing that S and N have no additional effect on EC activation.

Pericyte activation and dysfunction increase endothelial cell procoagulant activity and endothelial permeability, as constitutive hypoplasia of pericytes increases VWF release, tissue factor expression and platelet adhesion⁶. Moreover, CD144 is crucial for endothelial stability and blockade of CD144 contributes to coagulopathy, particularly in the lung microvasculature^6,20,21. We therefore assessed whether there is a link between CD144 expression and thrombosis in the lungs of fatal SARS-CoV-2 infected patients (n=8) (Figure 2A, B). Diffuse alveolar damage with extensive hyaline membrane rich in fibrin was observed in all patients (Figure 2A,B). Thrombi in the microvasculature was observed in 4 of eight lung autopsies of COVID-19 patients (Figure 2A). CD144 expression was decreased in all sections, and was almost undetectable in patients with diagnosed thrombosis (Figure 1B, 2B, C). Tissue factor was upregulated in all lung sections, independently of thrombosis (n=9), suggesting no direct correlation between thrombosis and upregulation of tissue factor. Thrombosis was associated with increased platelet recruitment and accumulation in 2 patients as single platelets or in microaggregates, without evidence of fibrin deposition on these thrombi in the alveolar capillaries while fibrin rich thrombi did not contain platelets, which were localized on the endothelium (Figure 2B). It is possibly that the presence of procoagulant phosphatidylserine-positive platelets observed in severe COVID-19 patients support the generation of fibrin-rich thrombi ²². VWF and platelets were observed in some but not all patients diagnosed with thrombosis, suggesting distinct thrombus composition (P6). Moreover, megakaryocytes were also observed in 3 patients, with enrichment around hyaline membrane^23,24. We further compared the expression of CD144 in Middle East respiratory syndrome coronavirus (MERS-CoV)- and rhinovirus-to SARS-CoV-2-infected lung specimen. A decrease in CD144, increased fibrin deposition and platelet recruitment in the microvasculature is observed in a MERS-infected lung associated with thrombosis (Figure 2D) whereas CD144 expression was more evident in rhinovirus-infected lung negative for thrombosis, fibrin deposition and platelet recruitment (Figure 2E). These results suggest that a decrease in CD144 is associated with higher risk of thrombosis in the lung microvasculature during SARS-CoV-2 and MERS-CoV infections. Further studies with larger cohorts of viral infections are necessary to confirm these observations.

• Download figure
• Open in new tab

Figure 2: Endothelial permeability and thrombosis in COVID-19 lungs.

(A) H&E staining of lung sections from patients who died from COVID-19 with or without evidence of thrombosis. (B) Immunofluorescence imaging of CD144, Von Willebrand Factor (VWF), tissue factor (TF), platelets (CD42b) and fibrin in formalin fixed and paraffin embedded lung sections from control, COVID-19, Rhinovirus and MERS-CoV patients. Nuclei/DNA were stained with Dapi. Images was captured using Epi fluorescence microscope and slide scanner Axio Scan. Z1.

In summary, using a vascular organoid model, we show that SARS-CoV-2 antigens directly alter endothelial permeability by decreasing CD144 expression and have a detrimental effect on endothelial and pericyte populations. The presence of comorbidities might account for increased endothelial permeability and favour the passage of the virus across the endothelial barrier and pericyte infection altering endothelial permeability, exacerbating endotheliitis, and ultimately contributing to the thrombotic complications which are fatal in severe infections. This effect appears independent of classical endothelial activation, however the elevated inflammatory cytokines such as IL-1 β could contribute to the endotheliitis. Dual therapies regulating endothelial activation and permeability is likely to be beneficial in severe patients to limit endotheliitis and thrombosis. However, as thrombus composition is not homogenous among patients, biomarker-guided strategies might be required to target thrombosis in COVID-19 patients.

Materials and Methods

Authorship

Contribution: AOK and JR designed research, performed research, analyzed data and wrote the paper; JSR, JHB and MC performed research; MLN, JDA and MC contributed vital new reagents; EY, PGM, GT, AGR contributes to clinical information and samples collection; ZS and MP contributed vital new reagents, performed experiments and analysed data; AF contributed vital new reagents and contributed to data analysis. All authors read and approved the paper.

Conflict-of-interest disclosure

The authors declare no competing financial interests.