COVID-19 PLASMA INDUCES SUBCELLULAR REMODELLING WITHIN THE PULMONARY MICROVASCULAR ENDOTHELIUM

Abstract

Background:

COVID-19 caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can affect multiple organ systems, including the pulmonary vasculature. Endothelial cells (ECs) are thought to play a key role in the propagation of COVID-19, however, our understanding of the exact scale of dysregulation sustained by the pulmonary microvasculature (pMV) remains incomplete. Here we aim to identify transcriptional, phenotypic, and functional changes within the pMV induced by COVID-19

Methods and Results:

Human pulmonary microvascular endothelial cells (HPMVEC) treated with plasma acquired from patients hospitalised with severe COVID-19 were compared to HPMVEC treated with plasma from patients hospitalised without COVID-19 but with other severe illnesses. Exposure to COVID-19 plasma caused a significant functional decline in HPMVECs as seen by a decrease in both cell viability via the WST-1 cell-proliferation assay and cell-to-cell barrier function as measured by electric cell-substrate impedance sensing. High-content imaging using a Cell Painting image-based assay further quantified morphological variations within sub-cellular organelles to show phenotypic changes in the whole endothelial cell, nucleus, mitochondria, plasma membrane and nucleolus morphology. RNA-sequencing of HPMVECs treated with COVID-19 plasma suggests the observed phenotype may, in part, be regulated by genes such as SMAD7, BCOR, SFMBT1, IFIT5 and ZNF566 which are involved in transcriptional regulation, protein monoubiquitination and TGF-β signalling.

Conclusion and Impact:

During COVID-19, the pMV undergoes significant remodelling, which is evident based on the functional, phenotypic, and transcriptional changes seen following exposure to COVID-19 plasma. The observed morphological variation may be responsible for downstream complications, such as a decline in overall cellular function and cell-to-cell barrier integrity. Moreover, genes identified through bulk RNA sequencing may contribute to our understanding of the observed phenotype and assist in developing strategies that can inform the rescue of the dysregulated endothelium.

Graphical Abstract;

1. INTRODUCTION

The COVID-19 pandemic has been a severe threat and burden to public health in recent years (1). It has been well documented that endothelial cells (ECs), a monolayer of specialised cells that line the lumen of the systemic vasculature, play an essential role in the propagation of COVID-19 and contribute to a variety of vascular pathologies seen during severe disease states (2). A continued discussion has revolved around whether endothelial cell dysfunction, as a result of COVID-19, results from direct viral infection of endothelial cells or as an indirect effect of systemic immune responses mediated by distally infected cells (3, 4). Mounting evidence supports an indirect mechanism of damage that causes sustained injury to the endothelium inhibiting clearance and causing progression to severe COVID-19 and mortality (5). Studies by McCracken et al. and Muhl et al., have shown that human endothelial cells lack the required angiotensin-converting enzyme 2 (ACE-2) receptors needed to internalise severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) (6, 7). This proof has been further supported by emergent data in mouse models that express human ACE-2 (8) and in vitro studies using human ECs (9) to show COVID-19-associated pathologies independent of ACE-2 mediated infection (10). Amongst the functional characteristics of healthy ECs, cellular viability and intercellular barrier function were consistently adversely affected in cases of severe COVID-19 (11-13). This underscores their potential as key hallmarks of endothelial cell dysregulation. Studying EC dysfunction and the establishment of a pro-inflammatory and pro-coagulative state can provide valuable insight into the nature of this dysregulation and identify targets to initiate repair (14, 15).

It is vital to understand the nature and extent of damage sustained to the endothelial cell compartment during severe COVID-19. Through the phenotypic and transcriptional profiling of pulmonary microvascular endothelial cells (HPMVEC), a mechanistic insight into the nature of early EC dysfunction can be observed and highlight new potential therapeutic targets for the treatment of chronic COVID-19. There is still limited information regarding the specific burden placed on the subcellular organelles that are damaged within ECs, which may manifest in phenotypic and eventual genotypic alterations that promote aberrant function. Quantification of subcellular variations is now possible through high-content imaging, a powerful tool for morphological analysis. (16) Mirabelli et al. have previously used high-content screening techniques to repurpose drugs to identify rapidly deployable treatments for COVID-19 (17). Here we leverage this technology to determine the impact of COVID-19 plasma exposure on the endothelium to identify organelle-specific variations within the endothelial cell compartment in an agnostic, pathway-independent manner. To explain the observed phenotype, we have paired these findings with RNA sequencing to provide genotypic annotation to the observed phenotype.

2. METHODS

2.1. Donor Patients and Plasma Acquisition

Fresh frozen plasma samples from 150 patients hospitalised between April and June 2021, were obtained through Bioresource NHS Labs, UK, under Study Approval #SR1491 AM01 220421. Plasma samples from patients that received a positive COVID-19 real-time quantitative polymerase chain reaction (RT-qPCR) test result during hospitalisation were classified as ‘COVID-19 plasma’ while plasma from donor patients that were hospitalised with alternative acute conditions and received a negative RT-qPCR test during hospitalisation was classified as ‘control plasma’, (detailed in Table 1).

Table 1.

Baseline Characteristics, Medical History and Clinical Chemistry of Study Population

Variable	All Patients	Hospitalised Patients (non-SARS-CoV-2)	Hospitalised Patients (SARS-CoV-2)
Size of cohort, n	150	75	75
Sex - Male, n (%)	73 (48.7)	34 (45.3)	39 (52.0)
Death, n (%)	50 (33.4)	24 (32.0)	26 (34.7)
Age, median (IQR)	71.0 (56.3, 82)	71.0 [56.5, 81.5]	71.0 [56.0, 82.5]
Medical History, n (%)
Chronic Obstructive Pulmonary Disease	30 (20)	23 (30.7)	7 (9.3)
Cancer	26 (17.3)	16 (21.3)	10 (13.3)
Heart Failure	<20 (13.4)	15 (20)	< 5 *(<6.7%)
Pulmonary Embolism	14 (9.4)	6 (8.0)	8 (10.7)
Angina	<10 (6.7)	< 5 *(<6.7%)	< 5 *(<6.7%)
Asthma	20 (13.3)	10 (13.3)	10 (13.3)
Obesity	12 (8)	4 (5.3)	8 (10.7)
Diabetes	32 (21.4)	14 (18.7)	18 (24.0)
Stroke	16 (10.7)	8 (10.7)	8 (10.7)
Dementia	19 (12.7)	6 (8.0)	13 (17.3)
Haematology and Clinical Chemistry, median (interquartile range)
Prothrombin Time	13.5 (12.0, 15.0)	13.0 [12.0, 15.0]	14.0 [12.0, 15.0]
C Reactive Protein	58.6 (28.0, 106.8)	25.5 [9.8, 77.0]	92.0 [46.2, 136.5]
Creatinine	79 (67.0, 111.3)	74.0 [66.5, 99.5]	84.0 [67.5, 123.0]
Urea	6.4 (4.5, 10.3)	5.7 [4.3, 8.4]	7.1 [4.7, 12.1]
Haemoglobin	130.5 (114,147.6)	129.0 [112.0, 144.2]	132.0 [116.0, 151.0]
White Cell Count	9.2 (6.5, 12.0)	11.3 [8.0, 14.9]	7.0 [4.9, 9.1]
Platelet Count	224.5 (176.5, 279.9)	248.0 [203.0, 304.2]	201.0 [150.0, 255.5]

2.2. Cell Culture and Viability Assays

Human pulmonary microvascular endothelial cells (HPMVEC) [Lonza, Switzerland] were cultured in endothelial microvascular growth media Microvascular Growth Medium 2 (EGM-2MV) + Bullet Kit [Lonza, Switzerland]). HPMVECs were then subjected to treatment with patient plasma, which was diluted at a ratio of 1:5 in a solution composed of 2 parts endothelial growth medium (EGM) and 3 parts phosphate-buffered saline (PBS).

Cell viability was tested using water-soluble tetrazolium salt (WST-1) [Roche, Switzerland] on HPMVECs treated with COVID-19 and control plasma according to the protocol described by Dupont et al (11). An additional cell viability assay, Cell Titre Blue [Promega, USA], and alternative cell line- human coronary artery endothelial cells (HCAEC) [Lonza, Switzerland] were also utilised to confirm findings from the WST-1/HPMVEC assay using the described protocol above. Measurements were taken on a Tecan Sunrise Microplate Reader and all viability readings were standardised against untreated controls.

2.3. Electric Cell Impedance Sensing (ECIS)

Electric Cell substrate impedance testing (ECIS) was performed using an 8-well 8W10E+ array (Applied Biophysics, USA) on an ECIS Z-Theta station (Applied Biophysics, USA) at 4kHz. Arrays were prepared by treatment with 10 mM L-cysteine (Sigma-Aldrich, USA) in 0.2% gelatin and 10ug/mL fibronectin (Sigma-Aldrich, USA). HPMVEC cells (6x10⁵) were seeded directly into array wells and cultured overnight under standard conditions. Plasma, diluted 1:5, was administered directly into the wells after 24 hours and impedance was measured at 7.5-second intervals during the 20-hour incubation. Measurements were expressed as electrical resistance (Ohms) and normal distribution was assumed based on previous studies (18).

2.4. Cell Painting and High Content Imaging (HCI)

HPMVECs were plated at a density of 1x10⁴ cells per well in a 96-well, flat bottom optically clear PhenoPlates (PerkinElmer, USA) pre-coated with 0.2% gelatin (Sigma-Aldrich, USA) and incubated for 48 hours. Following incubation, cells were treated with diluted patient plasma, for 1 hour. Plates were then washed once with DPBS and stained using a six-dye Cell Painting Kit across six subcellular organelles [nucleus, endoplasmic reticulum, mitochondria, plasma membrane and Golgi apparatus] and actin filaments. (PerkinElmer, USA). Imaging was performed using a six-channel, confocal Opera Phenix Plus High Content Screening System (PerkinElmer, USA).

Cell paint basic analysis was conducted using Harmony High Content Imaging and Analysis Software (PerkinElmer, USA). and yielded 1,500 morphological features based on size, shape, texture, and intensity (16). Features were then ranked and analysed using simple z-scores to determine statistical relevance and standardised against untreated controls using work packages on Spotfire (Tibco, USA) for 2D analysis and StratoMineR (Core Life Analytics, Netherlands) for 3D analysis. Images were annotated using Signals Image Artist (Perkin Elmer, USA).

2.5. Bulk RNA sequencing and differential gene expression analysis

RNA was isolated from HPMVEC cells treated with COVID-19 plasma (n=5) and control plasma (n=5) following a 1-hour incubation, using a RNeasy Kit (Qiagen, Germany). Library preparation was performed using an Illumina polyA selection library followed by total RNA sequencing on an Illumina NovaSeq sequencer at a depth of 20 million paired end reads per sample (Genewiz (New Jersey USA)).

Raw reads from each sample were mapped to a primary human assembly transcriptome (GENECODE version 38) and quantified using RSEM (version 1.3.0, bowtie2). Downstream analysis was carried out using R (version 4.2.1). Genes that on average expressed fragments per kilobase of transcript per million mapped reads (FPKM) values higher than 1 within a sample were retained for further analysis. Tximport (version 1.24.0) was used to aggregate isoform level counts to provide gene-level counts and downstream analysis. Principle component analysis (PCA) was used to map percentage variance between conditions using DESeq2-normalised reads (v1.36). Differentially expressed genes (DEGs) were mined using DESeq2. Genes with an absolute fold change value >1.5 and adjusted p-value of <0.05 were selected as genes of interest (GOI).

Gene-gene interaction network analysis was conducted using Gene Mania (version 3.5.2)(19). GeneMania finds functionally similar genes from a provided query list and reports weighted genes (graphically shown as different-sized circles) that indicate the importance and association from the query list of genes. Functionally similar genes were identified through GeneMania’s algorithm based on physical interactions, co-expression, shared protein domains and co-localisations. By linking external datasets including GEO, BioGRID, I2D, and Pathway Commons associated pathways were identified (20). In this analysis, only Homo sapiens-specific interactions were used along with default settings used to provide an expanded list of genes and functional interactions from the provided gene list.

2.6. Statistical Analyses

Statistical analyses were performed using GraphPad Prism (version 9.5.1) (GraphPad Software, USA). Normality (Normal Gaussian distribution) was determined using GraphPrism’s Normality and Lognormality tests. Comparative testing between two experimental conditions was performed using an unpaired two-tailed t-test when data was normally distributed or a Mann-Whitney U test when data were not normally distributed.

3. RESULTS

3.1. COVID-19 plasma causes a functional decrease in cellular viability and intercellular barrier capabilities of HPMVECs.

To accurately document the effects of patient plasma on the microvascular endothelium we treated HPMVECs with plasma from patients hospitalised with COVID-19 or other severe illnesses. Functional, transcriptomic and phenotypic changes were then analysed (Figure 1A). Patient demographic data, co-morbidities and clinical chemistry for all plasma-donating patients were tabulated in Table 1.

Figure 1: Exposure to SARS-CoV-2 plasma causes functional changes to human pulmonary microvascular endothelial cells.

(A) The experimental pipeline of the in vitro (blue) and in silico (yellow) analyses. (Bi-ii) Cell viability assay using WST-1 in HPMVEC cells (n =75, p = 0.02) (Bi) or HCAEC (n = 3, p = 0.003) (Bii). (Biii) Viability assay using the Promega cell titre blue assay in HPMVEC cells (n = 5, p = 0.03). (Ci) Electric cell-substrate impendence testing (ECIS) of COVID-19 plasma (blue), control plasma (red), and untreated cells (grey) (n=3). (Cii) Net change in barrier function versus untreated condition measured for 20 hours (n = 3, p = 0.0325). Normality was determined using the Shapiro-Wilk Test followed by p-value determination using Welch’s t-test. Data points and error bars represent the mean ± standard error of the mean. *p<0.05, **p<0.01

A decrease in HPMVEC viability was observed in the COVID-19 plasma-treated cohort showing a 0.85-fold change between groups based on a WST-1 cell viability assay [Figure1 Bi] (n=75, p = 0.01). This decrease in viability was confirmed in a different primary endothelial cell line – human coronary artery endothelial cells (HCAEC) [Bii] (n=3, p = 0.003) to show the effect of plasma was not only observed in the pulmonary-derived endothelium. An alternative cell viability assay, Cell Titre Blue [Biii] (n=5, p=0.03), also showed a decline in endothelial cell viability upon COVID-19 plasma treatment compared to controls, demonstrating the effect of the plasma was not spurious.

Given the vascular leakage observed in severe COVID-19, (13) and our data showing decreased EC viability upon treatment with COVID-19 plasma, we next tested the effect of plasma incubation on EC intercellular barrier function using electric cell-substrate impedance sensing (ECIS). Assessment of endothelial barrier integrity in HPMVEC showed a sharp decline upon COVID-19 plasma treatment (0.51 Ohms) compared to those treated with control patient plasma (0.96 Ohms) over 20 hours [Fig1 Ci]. The change in resistance from the time of plasma introduction to post-20 hours showed a 32% drop in intercellular resistance in COVID-19 plasma-treated cells and a 4.7% drop in resistance in control plasma-treated cells when compared to untreated control samples [Cii] (n=3, p = 0.03).

This collectively demonstrated a functional decline in the pulmonary microvascular ECs in vitro upon exposure to COVID-19 patient plasma, similar to observed vascular changes in cases of severe COVID-19 (11).

3.2. Subcellular organelles within the pulmonary microvascular endothelium undergo significant phenotypic remodelling upon exposure to COVID-19 plasma.

We were next interested in identifying if any associated phenotypic changes supported the observed functional decline in our plasma-treated ECs. Plasma-treated HPMVEC were morphologically profiled using a cell painting high-content imaging assay [Figure 2A, n=4], which showed differentially modified features across five subcellular organelles (the nucleus, endoplasmic reticulum, plasma membrane, mitochondria, nucleolus) and actin filaments.

Figure 2: Morphological and phenotypic variation is observed in COVID-19 plasma-treated cells.

(A) Representative images of the cell painting imaging assay, which stained six-subcellular organelles for untreated, control plasma and COVID-19 plasma-treated HPMVECs. (B) Whole-cell level analysis revealed variation in whole cell area (i), perimeter (ii), and cell roundness (iii) in the COVID-19 plasma-treated cohort. (C) Nuclear features include perimeter (i), width: length ratio (ii), and area (iii), and 3D features include cross-section area (iv), footprint area (v), and sphericity (vi). (D) Mitochondrial features include radial mean (i), axial length (ii), and SER spots (iii). (E) Nucleolar features showing changes in spot intensity (i) and number of nucleoli (ii). (F) Plasma membrane features show changes in region length (i), membrane region area (ii), and region roundness (iii). Normality was determined using the Shapiro-Wilk Test followed by p-value determination using Welch’s t-test (normal distribution) or *Mann Whitney Test (non-normal distribution). Data points and error bars represent the mean ± standard error of the mean. *p<0.05, **p<0.01, ***p<0.001

COVID-19 plasma-treated cells displayed a reduction relative to controls in their size, significantly within the cell region area (Figure 2Bi) (p=0.0009), cell perimeter (2Bii) (p=0.0002) and an increase in cell roundness (2Biii) (p=0.006). Within the literature, shrinkage in the whole cell can be an indicator of severe dysfunction and early signs of apoptosis. (21) We additionally noted a shrinkage of the plasma membranes in our COVID-19 treated conditions characterised by a decrease in (Fi) membrane length (p=0.002), (Fii) and membrane region area (p=0.0002), and (Fiii) an increase in membrane roundness (p=0.006). Which has previously been linked to extracellular leakage and reduced barrier function during cases of severe infection (22).

At a subcellular level, significant differences in phenotype were observed in the nucleus (Figure 2C), mitochondria (Figure 2D), nucleolus (Figure 2E) and plasma membrane (Figure 2F) between COVID-19 plasma-treated samples and controls. The nuclei demonstrated a pronounced decrease in size across 2-dimensional features: (Ci) perimeter (p=<0.0001), (Cii) width: length ratio (p=0.01)], (Ciii) area (p=<0.0001) and 3-dimensional features: (Civ) cross-section area (p = 0.02), footprint area (Cv)(p=0.02) and an increase in (Cvi) nuclear sphericity (p=0.006). Previous reports correlate nuclear condensation with a functional reduction in transcription and chromatin organization, which in turn could also contribute to a decrease in whole-cell viability (23, 24). Similarly, the number of nucleoli (p=0.01) were decreased in the COVID-19 experimental condition (Figure 2Ei) and a reduction in nucleolus spot intensity (p=0.007) - a feature that correlates to the nucleolar size and composition – was observed (Figure 2Eii). A reduction in the number of nucleoli directly corresponds to a decrease in protein packaging via the nucleolus (25).

The mitochondria also showed a decrease in overall size upon exposure to COVID-19 plasma, quantified by a shrinkage in the overall mitochondria. A decrease in size was observed by a reduction in the (Dii) radial mean (p=0.0004) and (Diii) axial length (p=0.0014)] and changed textural properties such as small endoplasmic reticulum-derived spots (SER Spots) (Div) (p=0.02) compared to control plasma.

Taken together, the phenotypic profiling of the HPMVEC compartment revealed alterations in various subcellular organelles, as demonstrated by morphological deviation from control samples, to show significant remodelling to accommodate the shifts in vascular pathology occurring during severe COVID-19.

3.3. RNA sequencing reveals transcriptional changes complementary to the observed phenotype.

Next, we performed bulk RNA sequencing (n=5, COVID-19 plasma-treated HPMVECs and n=5, control plasma-treated HPMVECs) to identify genes that were differentially regulated in COVID-19 compared to similarly diseased controls. The five COVID-19 plasma-treated conditions clustered separately compared to control plasma-treated samples, as seen through principal component analysis [Figure 3A]. In the COVID-19 cohort, we found 12 genes to be significantly down-regulated and 5 up-regulated based on a fold change of 1.5 and p<0.05 compared to the controls[3B].

Figure 3: Bulk RNA sequencing analysis reveals distinct transcriptional signatures in COVID-19-treated microvascular endothelial cells.

(A) PCA plot of COVID-19 and controls plasma-treated HPMVECs. (B) Heatmap of differentially regulated genes between COVID-19 and control plasma-treated HPMVECs. LogFC (1.5), p-value < 0.05. (C, D) Gene-gene interaction network for upregulated (C) and downregulated (D) genes in COVID-19 plasma-treated cohort (GeneMania). Each node represents one gene with the size of the circle corresponding to the strength of the interaction with the DEG. Coloured segments on genes correspond to GO term co-annotations that represent molecular functions, biological processes, or cellular component hierarchies active within the gene network.

As this analysis revealed only a few differentially regulated genes, we further investigated with which genes the DEG genes interact and to which pathways the gene interactions could be linked, using GeneMania (20).

Genes that were upregulated in COVID-19 were not specifically linked to one pathway but rather connected with other genes that are involved in several pathways (Figure 3C). SMAD family 7 (SMAD7), found in our bulk seq, physically and genetically interacted with pathway-specific genes such as SMAD Specific E3 Ubiquitin Protein Ligase 1 and 2 (SMURF1 and 2), SMAD family 6 (SMAD6) and TGF-beta receptor R1 (TGFBR1), amongst others, to cumulatively regulate signalling cascades associated for aberrant signalling (regulation of transforming growth factor beta receptor signalling pathway, negative regulation of transmembrane receptor protein serine/threonine kinase signalling pathway, negative regulation of cellular response to growth factor stimulus).

Whereas other genes upregulated in our COVID-19 cohort such as BCL corepressor (BCOR) and interferon-induced protein with tetratricopeptide repeats 5 (IFIT5) were associated with pathways responsible for subcellular maintenance such as histone monoubiquitination, histone H2A ubiquitination, protein monoubiquitination and protein kinase complex activity.

Upregulated genes that did not show a functional annotation via GeneMania, namely Scm-like with 4 MBT Domains 1 (SFMBT1) and zinc finger protein 566 (ZNF566) were manually curated using GO terms available via Genecards (Supplementary Table 1)(26). Annotations of these two genes showed that ZNF566 is involved in the negative regulation of transcription by RNA polymerase II while SFMBT1 showed involvement in chromatin organisation and the negative regulation of DNA-templated transcription.

Genes downregulated in COVID-19 were associated with canonical pathways responsible for inflammation and immune regulation such as cytokine binding, leukocyte chemotaxis, cellular response to biotic stimulus, acute inflammatory response, regulation of pri-mRNA transcription by RNA polymerase II and RNA polymerase II transcription regulator complex (Figure 3D).

Together, this data shows a small number of genes are upregulated during COVID-19 independent of canonical inflammatory markers present in most types of severe pro-inflammatory diseases.

4. DISCUSSION

In this study, we observed changes in both the phenotypic and functional characteristics of ECs when treated with COVID-19 plasma. Additionally, we mapped out the transcriptional changes occurring in ECs treated with COVID-19 patient plasma, comparing them to control samples. We first found that hallmarks of endothelial cell health such as cellular viability and intercellular barriers were compromised upon treatment with COVID-19 plasma, indicating endothelial dysfunction (27). Evaluation of subcellular organelles within endothelial cells revealed morphological variations and organelle shrinkage within the nucleus, nucleolus, mitochondria, and plasma membrane. Transcriptional profiling of plasma treated HMVECs showed distinct transcriptional profiles compared to cells treated with hospitalised control plasma. Our analysis also yielded several pathways that may provide mechanistic insight into dysregulation and signalling cascades activated during severe COVID-19 with therapeutic advantage.

While SARS-CoV-2 disseminates to multiple organs, (28) the lungs are primarily affected with reported angiopathic, pro-thrombotic and pro-inflammatory manifestations seen in severe cases, making it the primary organ system in which to model COVID-19 (29-31). We previously reported that the crosstalk between the heart and lung axis plays a crucial role in the dissemination of severe COVID-19 and hence chose to study pulmonary and coronary microvascular endothelial cells (3). From previous studies, it is evident that changes in cellular tone and intercellular leakage are important side effects of endothelial dysfunction and may be early predictors of cell death (32, 33). Following from the work of Karamapini et al. (34) and Dupont et al. (11), we were also able to show these changes in our WST-1 cell viability assay across a larger sample size of 75 biological replicates, compared to previously reported experiments with n=39 and n=10 samples, respectively. Additionally, we compared our functional readouts from COVID-19 plasma-treated cells with plasma from other hospitalised patients without COVID-19 infection instead of against healthy patients, highlighting lower cell viability and poor intercellular barrier function may be more pronounced in cases of severe COVID-19.

We were then interested in identifying a microvascular phenotype that may be specially acquired during severe COVID-19. Traditionally, during the onset of disease, the endothelium elicits a damage response, causing the activation of immune cascades and triggering dysregulated ECs to undergo apoptosis, allowing for early clearance and rapid repair of the endothelial monolayer (35). However, during severe COVID-19, endothelial cells detach more rapidly and do not regenerate as effectively causing a cascade of downstream complications such as the formation of microthrombi, endotheliopathy and disseminated intravascular coagulation which contribute to chronic disease progression (36-39). Hence, we employed a high-content imaging approach to analyse and quantify changes within six subcellular compartments within plasma-treated ECs (16).

At a cellular level, an overall decrease in cell size and an increase in cellular roundness have been associated with poor cellular health and are early predictors of apoptosis (40, 41). In the context of past coronaviruses such as avian coronavirus infection bronchitis virus (IBV), a decrease in nuclear size was also observed in Vero and BHK-21 cells (42). This decrease contributed to perturbations in the cell cycle machinery, and cell cycle arrest, which may also be relevant in the case of COVID-19. Notably, a decrease in nuclear size may directly impact transcriptional regulation, (43) which is further in line with the upregulation of genes such as BCOR, ZNF566 and SFMBT1 that influence histone ubiquitination and the negative regulation of transcription as evidenced by their pathway analysis and Genecards annotations,. Moreover, genes downregulated in COVID-19 such as FOS proto-oncogene (FOS) and early growth response 1 (EGR1), which are responsible for pathways modulating RNA Polymerase II activity, may further shed light on aberrant transcriptional activity observed during severe COVID-19 that downstream contributes to EC dysfunction.

We also observed changes in the number of nucleoli and a shrinkage of the mitochondria. A decrease in the number of nucleoli may correspond to fewer ribosomes and the generation of fewer proteins downstream (44, 45). Our pathway analysis identified an upregulation in the gene IFIT5, which contributes to protein kinase complex activity, defence response against viruses and nucleic acid binding activity (26, 46). Moreover, there is a direct link between mitochondrial DNA maintenance and nuclear protein synthesis that regulates cell growth and health (47). Hence decreased nuclear capabilities may also affect mitochondrial activity and global ATP synthesis within HPMVECs.

We finally conducted a bulk RNA sequencing analysis paired with a gene interaction network analysis to find several genetic interactions that provide possible explanations for the observed phenotype and functional variation. Several signalling pathways are associated with the maintenance of cellular homeostasis, two of which were identified in our RNA sequencing: the transforming growth factor beta (TGF-β) signalling pathway, and negative regulation of the transmembrane receptor protein serine/threonine kinase (TRPSTK) signalling pathway. The role of TGF-B has been previously described in the context of COVID-19 (48). In a clinical setting, activation of the TGF-β pathway has been identified in the sera of critical COVID-19 patients and has correlated to higher levels of platelet counts and immune responses (49). In a microvascular context, the TGF-β pathway may promote vascular leakage and barrier disruption that stimulates endothelial dysfunction in the lung and distal organs causing progression to a chronic disease state (50, 51). On the other hand, the TRPSTK pathway has not been linked to COVID-19 in the current literature. The canonical function of this pathway is to regulate cellular processes, more specifically cellular signalling and intercellular communication (52). The GeneMania analysis suggested the negative regulation of this pathway is associated with our COVID-19 condition and is further functionally supported by the observed reduction in barrier function in vitro. Genes present in the TRPSTK pathway also belong to the TGF-β superfamily highlighting a shared role (53).

Furthermore, genes involved in promoting the TGF-β pathway in the microvasculature during COVID-19 have been underlined in several other studies and have also been identified in our RNA sequencing (49, 51). SMAD7, an inhibitory gene that is a key regulator and agonist of the TGF-β and bone morphogenetic protein (BMP) pathways and associated genes SMURF 1 and SMURF2 from the gene interaction network analysis were identified to play a role in the propagation of the TGF-β pathway. While our study only shows an upregulation of a small portion of genes regulating the cascade, other genes that were downregulated in our analysis due to our similarly diseased controls, may also have a pronounced role in the propagation of the TGF-β response or amplify other pathways in conjunction that cumulatively lead to an amplified immune response.

Interestingly, previous studies have also suggested that inhibiting E3 ligases like SMURF1 and SMURF2 could serve as potential therapies for COVID-19 (54). In the context of the Middle Eastern respiratory syndrome coronavirus (MERS-CoV), inhibiting SMAD7 in renal and pulmonary epithelial cells led to the inhibition of MERS-CoV replication and protected cells from virus-induced cytopathic effects (55). Hence, SMAD7 and other reported genes could hold promise as a treatment for severe COVID-19 and may be valuable targets in repairing the damaged endothelium.

Together our findings show that there is a COVID-19 plasma-induced phenotype in ECs which directly correlates to a decrease in the functional output of ECs and alters the EC transcriptome. The COVID-19-associated phenotype identified from the high-content imaging aligns with several aspects of the transcriptional profiles identified through the bulk RNA sequencing. While we have been able to directly investigate the effect of patient plasma, a limitation is that we only identified a small population of genes from largely variable samples. Hence, there are likely many other genes and pathways also regulating the progression of COVID-19. However, the differentially regulated genes identified have high value, as they are consistently upregulated across the samples regardless of additional co-morbidities and circulating factors.

In conclusion, this study has employed a comprehensive in vitro approach to assess endothelial dysfunction during COVID-19. By combining functional analyses with morphological and transcriptional assessment, we identify genes and subcellular variations that may contribute to a unique endothelial signature associated with exposure to COVID-19 plasma. Genes like SMAD7, SMURF1, SMURF2, SFMBT1, ZNF566, BCOR and IFIT5 are implicated in the observed phenotype and might hold therapeutic potential following further functional investigation to understand their involvement in COVID-19-related EC alterations. Crucially, our data are supportive of the concept that plasma derived from patients with severe COVID-19 contributes to indirect mechanisms of endothelial cell damage.

ABBREVIATIONS

ACE-2

angiotensin-converting enzyme 2

BCOR

BCL6 corepressor

BMP

bone morphogenic protein

COVID-19

coronavirus disease 2019

DEG

differentially expressed genes

EC

endothelial cells

ECIS

electric cell-substrate impedance sensing

EGM2-MV

endothelial microvascular growth media

EGR1

early growth response 1

FPKM

fragments per kilobase of transcript per million mapped reads

GOI

genes of interest

HCAEC

human coronary artery endothelial cells

HCI

high content imaging

HPMVEC

human pulmonary microvascular endothelial cells

IBV

avian coronavirus infection bronchitis virus

IFIT5

interferon-induced protein with tetratricopeptide repeats 5

MERS-CoV

Middle Eastern respiratory syndrome coronavirus

PBS

phosphate-buffered saline

PCA

principal component analysis

RT-qPCR

real-time quantitative polymerase chain reaction

SARS-CoV-2

severe acute respiratory syndrome coronavirus 2

SER

small endoplasmic reticulum

SFMBT1

Scm like with 4 MBT Domains 1

SMAD6

SMAD family 6

SMAD7

SMAD family 7

SMURF 1 and 2

SMAD Specific E3 Ubiquitin Protein Ligase 1 and 2

TGF-β

transforming growth factor beta

TGFBR1

TGF-beta receptor R1

TRPSTK

transmembrane receptor protein serine/threonine kinase

WST-1

water soluble tetrazolium salt

ZNF566

zinc finger protein 566