Introduction
Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) causes the coronavirus disease 2019 (COVID-19) that in the last year gave rise to a global pandemic. SARS-CoV-2 primarily invades alveolar epithelial cells and causes acute respiratory distress syndrome. However, increasing evidence indicates that endothelial cell dysfunction and vascular events are major complications of the disease. Indeed, vascular inflammation, barrier defects leading to tissue edema, activation of disseminated intravascular coagulation and microthrombi were reported in moderate to severe COVID-19 cases [
3,
8,
14,
16,
27,
28,
30]. In addition, pre-existing impaired endothelial function, i.e., in diabetes mellitus patients and underlying vascular pathologies were shown to worsen clinical outcome of COVID-19 [
3,
8,
14,
16,
27,
28,
30]. Whether vascular complications can be attributed to a systemic inflammatory response or are a direct consequence of the viral infection of endothelial cells is currently under debate. To-date, SARS-CoV-2 has been reported to directly infect vascular organoids in vitro [
19] and first case studies reported endothelial infection in glomerular capillary loops, skin lesions [
8,
10,
30], as well as provided evidences for endotheliitis in COVID-19 patients [
17,
30]. However, the expression of the putative SARS-CoV-2 receptor, i.e., angiotensin-converting enzyme 2 (ACE2) is low in endothelial cells compared to mural cells and recent studies suggest that endothelial cells may not be the primary target of SARS-CoV-2 in the vascular wall [
11]. This would imply that the endothelium might be affected independently of direct viral action during the course of the disease, whereby the cytokine storm syndrome associated with elevated levels of pro-inflammatory cytokines such as IL-1β, IL-6, and TNFα may cause the loss of anti-thrombotic and anti-inflammatory functions of endothelial cells [
23,
28].
To gain insight into whether endothelial cells are a primary target of SARS-CoV-2, we studied human endothelial cells derived from several different vascular beds in vitro.
Materials and methods
Cells and cardiac tissues
Human umbilical vein endothelial cells (HUVEC; CC-2935), human coronary artery endothelial cells (HCAEC; CC-2585), human cardiac microvascular endothelial cells (HCMVEC; CC-7030), and human lung microvascular endothelial cells (HLMVEC; CC-2527), and human pulmonary arterial cells isolated from diabetics (D-HPAEC; CC-2924) were purchased from Lonza and cultured in endothelial basal medium (EBM; CC-3156, Lonza) supplemented with 10% fetal calf serum (FCS; 4133, Invitrogen), amphotericin-B, ascorbic acid, bovine brain extract, endothelial growth factor, gentamycin sulfate, and hydrocortisone (EGM-Bullet Kit; CC-3124, Lonza) or EBM-2 supplemented with 10% FCS, hydrocortisone, FGF, VEGF, R3-IGF, ascorbic acid, EGF and GA-1000 (EGM-2-Bullet kit; CC-3162, Lonza) at humidified atmosphere, at 37 °C/5% CO2.
Primary human umbilical vein endothelial cells were isolated and purified using CD144 antibody-coated magnetic beads (Dynal Biotech, Hamburg, Germany) and cultured as reported previously [
6]. The human umbilical cords were obtained from local hospitals in Frankfurt am Main and the use of human material in this study conforms to the principles outlined in the Declaration of Helsinki. The isolation of human cells was approved by the ethics committee at the Goethe University in Frankfurt am Main. HUVEC were cultured in endothelial basal medium (EBM; CC-3156, Lonza) as mentioned above for the commercial HUVEC.
Living human heart slices were generated and cultured as recently described [
9]. Samples of left ventricular myocardium were obtained from failing hearts during transplantation in the Clinic of Thoracic and Cardiovascular Surgery, Heart and Diabetes Center, Bad Oeynhausen, Germany. The procedure has been approved by the institutional ethics board, and patients have provided informed consent to the scientific use of the explanted tissue. In brief, heart slices were generated from the explanted failing human myocardium by cutting 300-µm-thick vibratome sections. Slices were mounted and cultured in biomimetic culture chambers in a standard incubator (37 °C, 5% CO
2, 20% O
2, 80% humidity) [
9]. Pacing was performed at 0.5 Hz with bipolar 50 mA pulses comprised of 1 ms charging and discharging pulses separated by a 1 ms interval. Slices were cultured in Medium 199 supplemented with penicillin/streptomycin, insulin/transferrin/selenite and 50 µM 2-Mercaptoethanol. Medium was exchanged in part (1.6 ml of 2.4 ml total volume in each biomimetic cultivation chamber) at 36–48 h intervals.
Infection
SARS-CoV-2 (strains: D614, G614, B.1.1.7, B.1.351, and P.2) were isolated and propagated in CaCo2 cells as previously described [
4,
12]. For infection of endothelial cells, the viral stock was diluted to the desired MOI (multiplicity of infection) in the respective medium supplemented with 1% FCS and incubated for 2 h. Then, the medium was changed to the respective culture media (see above). Five days after infection, endothelial cells were fixed in 4% paraformaldehyde (PFA) for 10 min or lysed for RNA isolation. CaCo2 cells were fixed after 24 h. Living human heart slices were incubated with 200 µl of viral stock (1.10
7 TCID 50/ml) for 3–5 days using the above-mentioned medium without 2-Mercaptoethanol.
For experiments that mimic pro-inflammatory conditions, endothelial cells were treated for 24 h with 30 ng/ml TNFα (210-TA-005, R&D Systems), 10 ng/ml IFNβ (8499-IF-010, R&D Systems) or 10 ng/ml IFNγ (285-IF-100, R&D Systems) prior to viral infection.
To block proteasomal degradation, cells were incubated with the virus together with 10 µM MG132 (M7449, Sigma-Aldrich; solved in DMSO) for 24 h.
To inhibit viral infection, HCAECs pre-incubated with 10 µM chloroquine (PHR1258, Sigma-Aldrich), 1 µM (in DMSO) cathepsin inhibitor N-Acetyl-l-leucyl-l-leucyl-l-methional (ALLM; 0384, Tocris) or 20 µM Furin Inhibitor I (344390-AMG, Merck). 5 µg/ml human recombinant ACE2 (933-ZN, R&D Systems) was mixed with the virus and incubated for 30 min, prior to infection. Cells were fixed 3 days after infection.
Immunofluorescence labeling
Cells were fixed with 4% PFA and were permeabilized with 0.1% Triton X-100/PBS for 10 min. Cells were blocked with blocking solution (5% donkey serum in PBS) for 1 h at RT. Primary antibodies were incubated in blocking solution overnight at 4 °C. After four 5-min washes with PBS, secondary antibodies and DAPI were incubated for 1 h in blocking solution. Finally, two 5-min washes of PBS were performed before cell observation.
Heart slices were fixed in 4% HistoFix (P087.4, Carl Roth GmbH) for 24 h at 4 °C. Slices were transferred to a series of ascending sucrose concentration: 4% sucrose in PBS (1 h, 4 °C), 15% sucrose in PBS (4 h, 4 °C) and finally 30% sucrose in PBS (overnight, 4 °C). The day after, slices were washed twice for 30 min with 100 mM glycine at 4 °C and once with PBS for 30 min at 4 °C. Cardiac slices were permeabilized with 1% Triton X-100 in PBS overnight at 4 °C. Slices were then washed three times with PBS for 30 min at 4 °C and blocked with blocking solution (3% BSA in 0.3% Triton X-100) overnight at 4 °C. Slices were washed three times in 0.3% Triton X-100 and incubated with primary antibodies that were diluted in blocking solution (overnight, 4 °C). Slices were again washed trice with 0.3% Triton X-100/PBS (30 min each) and incubated with the secondary antibodies that were diluted in PBS (overnight, 4 °C). Finally, slices were again washed trice with 0.3% Triton X-100/PBS before mounting (see Suppl. Tab. 1 for detailed antibody information).
RT-qPCR and gel electrophoresis
Total RNA was isolated using RNeasy Mini Kit (217004, Qiagen) according to the manufacturer’s instructions including an on-column DNase I digestion step (79254, Qiagen). Reverse transcription was performed using 500 ng RNA, random hexamers and MuLV reverse transcriptase (N8080018, Thermo Fisher). Fast SYBR Green qPCR were carried out by StepOnePlus real-time PCR systems (4385617, Thermo Fisher). RPLP0 amplification was used for data normalization. Relative expression levels were calculated by 2‒ΔCt. Primer sequences are provided in Suppl. Table 2. PCR products were visualized on a 1.5% agarose gel in 1xTAE buffer and visualized with Midori Green Advance (617004, Biozym Scientific GmbH).
Quantification of virus titer in cell culture supernatants
Supernatants from infected endothelial cells were collected 5 days post-infection. Confluent layers of CaCo2 cells in 96-well plates were infected with serially diluted supernatants. Cytopathogenic effect (CPE) was assessed visually 48 h after infection. The infectious titer was determined as TCID50/ml.
Single-nuclei RNA-sequencing analysis
Single-nuclei RNA-sequencing data from human cardiac samples were derived from a previously published data set [
21]. Data were analyzed with the Seurat (v3) package.
Statistical analysis
Data are represented as mean and error bars indicate standard error of the mean (SEM). Data were statistically assessed for Gaussian distribution using Kolmogorov–Smirnov and Shapiro–Wilk test. For comparison of two groups, statistical power was determined using two-tailed, unpaired t test. For multiple comparisons, ordinary one-way ANOVA with a post hoc Dunnett’s or Turkey’s multiple comparison was used.
Discussion
The results of the current investigation suggest that SARS-CoV-2 does not permissively infect microvascular or venous endothelial cells from different sources. However, HCAEC appear to take up the virus and show positive spike protein in the endosomal compartment. Since no permissive infection was detected in endothelial cells, SARS-CoV-2 might indirectly induce endothelial cell dysfunction via the systemic inflammatory response that causes the observed striking vascular effects in patients with COVID-19.
Interestingly, of all the endothelial cells studied only HCAECs were positive for spike protein after SARS-CoV-2 infection, suggesting different responses of endothelial cells derived from different vascular beds. Endothelial cell heterogeneity and specificity is crucial for the homeostasis of the different organs [
2]. In our study, HCAECs show a higher expression of the SARS-CoV-2 receptor ACE2, which may be responsible for mediating the uptake of the virus. HCMVEC, HLMVEC, HUVEC and HPAEC did not express ACE2 and were not permissive for SARS-CoV-2. However, transducing endothelial cells with recombinant ACE2 may enable SARS-CoV-2 infection as shown recently [
20], suggesting that the lack of sufficient ACE2 expression might be a limiting factor. Subsequent steps in the viral life cycle appear to be blocked in HCAECs. Since we did not detect double-strand viral RNA in HCAECs upon infection, likely early steps in uncoating of the incoming virus, or endosome-virus membrane fusion, or viral RNA synthesis are haltered in HCAECs. It would be of interest to understand the mechanisms that allow this endothelial cell type to block subsequent virus replication [
29]. Interestingly, we observed an early induction of EDEM1, which is known to be involved in clearance of misfolded proteins in the ER [
18]. One may speculate that the induction of this gene may be involved in the removal of spike protein limiting further viral activities.
The observation that endothelial cells take up SARS-CoV-2 without propagating the virus was also made previously [
22,
32]. However, these studies were performed with stem cell-derived endothelial cells, which may not fully resemble primary cultured lung or cardiac endothelial cells, as used in our study. In addition, one report found no evidence of direct viral infection of vascular endothelial cells in an ex vivo lung culture of one COVID-19 patient [
13], which was supported by others [
24] and by our own ex vivo heart slice model, where we could not detect spike positive endothelial cells. To our knowledge, this is currently the first study to compare the effect of SARS-CoV-2 variants on endothelial cells. We demonstrate a higher uptake of the three tested virus variants in HCAECs, however, B.1.1.7 was the only variant that affected cell number.
One limitation of the present study and previous reports is that we cannot exclude that endothelial cells in humans might react differently and that the in vitro culture and expansion of endothelial cells change their gene expression and responses. We also tested whether a pro-inflammatory environment would have facilitated virus infection. However, stimulation of endothelial cells with pro-inflammatory cytokines or using endothelial cells cultured from diabetic patients did not augment the SARS-CoV-2 infection in the present in vitro study. Recent findings suggested that inflammatory cytokines contribute to a complicated course of COVID-19 [
15]. Therefore, we speculate that pre-treatment of endothelial cells with pro-inflammatory cytokines might sensitize endothelial cells for viral infection. However, we did observe reduced levels of viral spike protein upon TNFα stimulation, whereas IFNβ and IFNγ had no influence. It is unclear why TNFα did reduce viral spike protein load in endothelial cells and limited information is available regarding the effects of TNFα on SARS-CoV-2 virus entry and infection rate. One study reports a link between TNFα and ACE2 expression, which was reduced by systemic anti-TNFα treatment in intestinal cells [
26]. However, we did not find a transcriptional regulation of ACE2 in endothelial cells (Suppl. Fig. 6). Thus, one may speculate that TNFα might interfere with other mechanisms resulting in a reduction of viral uptake or destabilization of spike protein.
In conclusion, the lack of a direct cytotoxic or pro-inflammatory effect by SARS-CoV-2 infection of endothelial cells may suggest that the massive endothelial dysfunction and microvascular thrombotic complications observed in patients suffering from COVID-19 is mainly secondarily caused by the inflammatory cascades mediated by the cytokine release syndrome. However, due to reduction of cell number observed after infection of HCAECs with the variant B.1.1.7, patients infected with these SARS-Co-V-2 variants should be closely monitored. Overall, therapeutic interventions aiming at endothelial protection may be warranted to protect against primary or secondary effects to maintain organ integrity during later development of the disease.
Acknowledgements
We thank Lisa-Maria Kettenhausen, Lena Stegmann and Kerstin Euler for experimental assistance and Isabel Winter for isolating HUVEC from umbilical veins. J.W. and D.B. have performed the experiments, J.W. has performed statistical analysis of the data. J.K. has performed qRT-PCR quantification of SARS-CoV-2 variants, G.A., M.S. and G.L. have performed stainings; L.N. has analysed single cell sequencing data; H.M. and A.D. provided human heart slices, A.H., T.J. and C.D. have provided tools and advice regarding ER stress responses, as well as bulk RNA sequencing; I.F. and I.B. has provided primary endothelial cells and contributed to drafting the manuscript; A.M. Zeiher and S. Ciesek have provided conceptual advice, J.C. and S.D. have designed and supervised the study, J.W. and S.D. wrote the manuscript.