Relationship: 2311: Binding to ACE2 leads to ACE2 dysregulation

AOPs Referencing Relationship

Key Event Relationship Description

This KER summarises the evidence for dysregulation of ACE2 (KE 1854) as a result of its interaction with the viral spike (S) protein of SARS-CoV and SARS-CoV2 (KE1739). This interaction is likely an important aspect of COVID-19 pathogenesis. As the initial step of SARS-CoV2 infection, it is a potential target for intervention at the point of viral entry on cellular level, but also latter for treatment of the disease during viral replication at organism level. In addition,  it is important to summarise exiting evidence for this KER and evaluate the WoE for the prevailing hypotheses that COVID-19 pathogenesis (and hence relevant treatments) is largely governed by the binding of the S protein to ACE2 leading to down-regulation of its physiological function as a protease that converts Angotensin II to Angiotensin 1-7 and hence disturbs the balance within the RAS system. Evidence summary and evaluation should also help identify key data/evidence gaps and inform critical tests and approaches to fill them.

Evidence Supporting this KER

Angiotensin-converting enzyme 2 (ACE2) is a transmembrne protein which under normal physiological conditions acts as a carboxypeptidase with broad substrate specificity. Angiotensin II is recognised as the main substrate of ACE2, but ACE2 hydrolyses a number of other biologically active peptides including DABK, apelin-13, neurotensin(1–11), dynorphin A-1–13), β-casomorphin-(1–7), and ghrelin (Vickers 2002, Humming, 2007). Hence, ACE2 is one of the major regulating enzymes in the local tissues and systemic plasma RAS and KKs systems, mainly balancing the action of its homologue ACE.

In the gut, ACE2 interacts with the Broad Neutral (0) Amino Acid Transporter 1 (B⁰AT1) (Slc6a19) and chaperons its membrane expression. As B⁰AT1 mediates the uptake of neutral dietary amino acids, such as tryptophan and glutamine, into intestinal cells (Camargo, 2009), ACE2 is indirectly involved in regulating intestinal amino acid homeostasis, expression of antimicrobial peptides, and the gut microbiome (Camargo, 2020, Hashimoto, 2012).  

Considering the above, it is plausible that interference with any aspect of the ACE2 expression, it’s enzymatic or B⁰AT1s-shaperone function represents a potential target for dysregulation with a wide range of downstream effects.

ACE2 is not known to act as a classical receptor that would itself initiate intracellular signalling cascades after binding of ligands or substrates and very few selective inhibitors or stimulators of its activity are available to study the effects of potential stressors and modulators of ACE2.  However, binding of SARS-CoV and SARS-CoV2 spike protein to ACE2 has been demonstrated (refs from Event: 1739) and this interaction represents an essential step for viral entry into the cells (refs from Event: 1738).

How interaction of ACE2 with the spike protein (KE upstream) perturbs ACE2 function (KEdownstream) is evaluated in the empirical evidence section of this KER.

At a molecular level, it is plausible that binding of the spike protein to the extracellular domain of ACE2 may perturb the catalytic activity of both, the cell membrane mACE2 or the cleaved catalytically active ectodomain sACE2, by allosteric interaction with the catalytic site (Liu & Nussinov 2016; Dutta 2022; Wang DS 2022; Trozzi 2022).

Interaction with spike protein may also perturb the shedding of ACE2 catalytic ectodomain. ACE2 is constitutively cleaved (shed) mainly by another membrane bound metalloprotease ADAM17, also known as TACE, TNFalpha Converting Enzyme (Zunke 2017) in a number of cells and tissues (Guy 2008 in human cardiac myofibroblasts; Peng Jia, 2009 in human lung epithelial cells and BALF; Werner, 2005 in canine epithelial polarised kidney MDCKII cells stabilly expressing ACE2; Iwata, 2009 in Chinese Hamster Ovary Cells). The exact role of constitutive ACE2 shedding is not well understood, but proteolytc ectodomain shedding of membrane proteins is a fundamental post-translational regulatory mechanism of the activity/function of a wide variety of proteins, including growth factors, cytokines, receptors and cell adhesion proteins (Lichtenthaler et al., 2018).

Binding of the spike (S) protein from both SARS-CoV has been associated with cleavage of the ACE2 ecodomain by TACE/ ADAM17 (Haga 2008, Haga 2010, Jocher, 2022), but also by another host protease TMPRASS2 (Heurich, 2014; Shula 2011) which also cleaves the Spike protein itself thus mediating viral entry by fusion of the cellular and viral membranes.  Spike protein mediated ADAM17 shedding of ACE2 has not been demonstrated for SARS-CoV2 (to my knowledge). However, the interplay between spike and these proteases has been suggested as critical and linked events based on analogy and also some structure-function considerations (Schreiber 2020; Heasly 2022; Zipeto 2020).  It is plausible then, that the spike induced shedding of ACE2 (via ADAM17 or other proteases) perturbs the distribution of the catalytically active mACE2/sACE2 and consequently differentially dysregulates ACE2 function/activity locally and systematically (Wysocki 2010). This in turn would have significant implications for the development of therapies that would target ACE2 activity and the RAS system appropriately in terms of delivery and timing.

Additionally, it is plausible that membrane bound ACE2 is down-regulated due to endocytosis of viral particles as suggested by earlier studies with SARS-CoV spike protein (Inoue, 2007; Wang H., 2008; Wang S., 2008) but also with SARS-CoV2 spike protein (Raghavan 2021, Gorshkov 2022). The relative contribution of the membrane fusion versus endocyitic route varies for the two SARS-CoV viruses (CoV1 and CoV2). This difference has been suggested as an important aspect of the differential infectivity and transmissibility of the two viruses (Zhu, 2021). It may also affect the relative magnitude and direction of mACE2 and sACE2 dyregulation.

Finally, transcriptional dysregulation of ACE2 mRNA synthesis is also plausible, but the process is not likely to be an immediate consequence of spike binding to ACE2, but rather requires some downstream signalling (potentially other intermediate KEs)and time for the biosynthesis of the mRNA and protein. Consistent with this, there is evidence that Ace2 (human but not the mouse), is an Interferon Stimulated Gene (ISG) activated also by some other viruses, inflammation, hypoxia and Synthetic dsRNA, poly(I:C) (Ziegler, 2020, Smith, 2020; Salka 2021, Zhuang 2020). However, it emerges that interferons and viruses stimulate a particular truncated form of ACE2, the dACE2 and not the full length ACE2 (Onabajo 2020; Scagnolari, 2021; Blume, 2021; Oliveto, 2022). The significance of this (post?)transcriptional modulation of Ace2 mRNA is not clear, and the enzymatic activity of the product of the alternatively spliced or, Ace2 mRNA initiated from a different promotor (which contains the catalytic domain but not the spike protein binding site) has not been examined. In addition, there is some evidence that S-protein binding can also lead to (direct?) suppression of ACE2 and Type I Interferon synthesis. To facilitate the analysis of the WoE for relevant perturbation of ACE2 as related to mRNA synthesis following the interaction with SARS-CoV and SARS-CoV2 spike protein, existing empirical evidence is outlined in the KER highlighting considerations for stressor type and time concordance between the two KEs.

This KER summarises the evidence for dysregulation of ACE2 (KE 1854) as a result of binding of the viral spike (S) protein of SARS-COV and SARS-COV2 (KE1739). This is likely an important aspect of COVID19 pathogenesis as an initial step it is a potential target for intervention at (re)infection but also for treatment of the disease. In addition, it is important to summarise the exiting evidence for this KER as the majority of hypotheses on the pathogenesis of COVID19 and potential treatments consider S protein binding leads to down-regulation of ACE2.

The evidence included in this KER was identified in the period between October 2020 and March 2021 with a Pubmed search syntax……, and additional targeted searches for specific aspects to ACE2 regulation. The searches were aimed to identify the evidence for the effect of SARS-COV or SARS-COV2 on any aspect ofACE2 regulation i.e. effect on expression of ACE2 at mRNA or protein level, ACE2 shedding and enzymatic activity, all described in KE 1854.

The studies generally use SARS-COV or SARS-COV2 Virus like Particles (pseudovirus), recombinant or transiently expressed S protein from SARS-COV and SARS-COV2 (in Table 1 designated S protein (1) and S-protein (2), respectively) in various cell lines e.g. Vero E6 and Huh7-that express endogenous ACE2 or HEK293T transfected with ACE2 expressing vectors. Concurrent increase of the soluble form of ACE2 (sACE2) and/or enzymatic (peptidase) activity is monitored in some of the studies (Table).

The evidence mostly relates to evaluation of direct binding of S protein to ACE2, but some evidence is indirect and can potentially relate to dysregulation resulting from infection progression/viral replication and reinfection of new cells in the context of the parallel inflammatory processes.

Details of the evidence for dysregulation of ACE2 at mRNA and protein level (including enzymatic activity) as a result of interaction with SARS-COV 1 and 2 S proteins:  : https://aopwiki.org/system/dragonfly/production/2022/02/12/8oiauj0w7m_Empirical_evidence_ACE2_dysregulation_120222.pdf

GI tract specific evidence for dysregulation at mRNA level

Evidence Collection Strategy

The literature search is not comprehensive or systematic and needs continuous improvement.

In all of the studies reviewed here, ACE2 dysregulation following exposure to recombinant S-protein and replicating virus is considered as equivalent for the prpose of evaluationg ACE2 dysregulation. However, it should be noted that the literature screening step showed that inflammation and exposure to other stressors (nanomaterials and other viruses, like influenza) were also associated with ACE2 dysregulation, indicating that aspects of the stressor multiplication and/or persistency, additional to the S-protein binding, may be responsible for the observed ACE2 dyregulation. These references were tagged but not further analysed and not included in this KER that is focused on S-protein binding. They can be informative if this KER is eventually modified and potentially split into two: (i) direct/adjacent KER (studies examining ACE2 dysregulation after direct and short term exposure to S-protein binding or non-replicating pseudoviruses) and (ii) indirect/nonadjacent KER (studies with replication virus and persistent infection).

Focus on GI tract relevant evidence (published in https://doi.org/10.3390/jcm11185400)

Both ACE2 down and up regulation are observed in the gut as in the other test systems (see gut unreated evidence table). The apparent inconsistences regarding the direction and magnitude of ACE2 dysregulation in the different studies may reflect the dynamic, temporal components of the dysregulation driven not only by the interaction of ACE2 with the surface viral components i.e. S-protein binding, but also by the interaction of the replicating viral components with the innate immunity response elements, particularly in the test systems with replicating viruses used with the GI-tract-derived organoids.

ACE2 mRNA down-regulation in SARS-COV2 treated GI-derived organoids is reported in one study using scRANseq analysis (Triana et al., (2021). The down-regulation was specific to enterocytes actively replication the virus. Another study Nataf & Pays (2021) also reports profound but transient ACE mRNA down regulation.

Evidence for SARS-COV2 mediated up-regulation of ACE2 mRNA in GI-tract derived organoids is also available (Lamers et al., 2020; Lee at all (2020); Heuberger et al., 2021) and consistent with the similar studies in many other tissue/organ systems (see evidence table for other organs above). This is also consistent with the finding that ace2 is an Interferon Stimulated Gene (IFG) in airway epithelial cells (Ziegler, 2020) and also in colon enterocytes (Heuberger et al., 2021). In fact, all there studies also demonstrate a time concordance of ACE2 mRNA up-regulation with stimulation of ISG response in the infected organoids (Lamers et al., 2020; Lee at all (2020); Heuberger et al., 2021). Interestingly, even the scRNAseq study by Triana et al., 2021 which zoomed on specific cells within the organoid found that SARS-COV2 treatment induced distinct proinflammatory and ISG expression profiles in infected and bystander cells in the organoid. Namely, expression of interferon-stimulated genes was pronounced in bystander cells, while the infected cells showed strong NFkB/TNF-mediated pro-inflammatory response but a limited production of ISGs, suggesting that while SARS-CoV-2 may activate ISG by paracrine signaling, it may suppress the autocrine action of interferon i.e inducton of IGS including ACE2 in infected cells. This would be consistent with down-regulation of ACE2 in the infected cells observed in this study. In addition, this may explain why in some studies down-regulation of ACE2 mRNA can be observed under certain conditions (e.g. bulk mRNA measurements) and at some (earlier) time points of replication. Furthermore, the relationship of an observed increase of ACE2 mRNA to dysregulation at protein and enzymatic level remains to be elucidated. Indeed, most recently Harnik et al. 2021 (10.1038/s42255-021-00504-6) examined the spatial discordances between mRNAs and proteins in the intestinal epithelium and their significance for interpretation of transcriptomic data. Such apparent discordances have also been reported in the heart and lung tissue in mice and human (KE1854).

Identification of alternative forms of ACE2 mRNA and protein, an N-terminus truncated dACE2, which appears to have distinct transcriptional regulation profile compared to flACE2 (Onabajo et al, 2020, Janakowski et al (2021) - 0.1016/ j.isci.2021.102928) may also account for some of the observed inconsistences. However, a detailed analysis of experimental conditions in past, and careful design of probes and primers in future studies is necessary. Interestingly, concomitant down and up regulation of 97kD and 80kD anti-ACE2 polyclonal Ab-reacting proteins has been detected in differentiating human colon adenocarcinoma cell line HT29 (Bártová et al. (2020) 10.18632/aging.202221). Considering only one form of ACE2 relevant (97KD the only form detected in HEK293 and A549), the authors conclude that ACE2 is down-regulated in mature differntiatiatiated enterocytes compared to undifferentiated ones. This is in contrast to all mRNA and even cytoimmunostaining studies described above which demonstrate that a highest level of ACE2, both mRNA and protein is detected in the mature enterocytes and at the brush borders of the intestine and 3D organoids (eg. Lamers et all., 2020, Lee et al., 2020; Hauberge et al., 2021; Triana et al., 2021). Notably, the initial analysis by Onabajo et al, 2020 found that dACE2 mRNA is enriched in in squamous tumors of the respiratory, gastrointestinal and urogenital tracts.

The uncertainties and uncosnstances disused above illustrate clearly the need for careful characterization of the test systems to facilitate robust interpretation of the results.

In addition, we note that to date (to our knowledge), the majority of studies related to SARS-COV2-mediated ACE2 dysregulation, focus on ACE2 mRNA expression while structural/protein and functional studies are lacking, particularly in the gastro intestinal system. The novel gut-derived organoid systems can help address this gap by monitoring level and cell distribution of ACE2 protein as well as its function as B⁰AT1chaperon, via monitoring the membrane expression and/or the transporter function of B⁰AT1 itself. In addition, treatment with S-protein and/or non-replicating SARS-COV2 pseudo-viruses [Minghai & Zhang (2021) 10.7150/ijbs.59184], may help address better any potential direct effect of S-binding on ACE2 dysregulation. Finally, a development of more complex organoid systems that would also include microbiota or elements of the immune and/or vascular system are needed to better examine ACE2 dysregulation by SARS-COV2 but also the effects of such dysreulation at higher organizational level and in conjunction with the other elements of the RAS system.

Finally, evidence on up- or down-regulation of ACE2 in the GI tract of SARS-COV infected patients is not available to our knowledge. Examining the GI specific transcriptomic, proteomic and biomarker databases of COVID19 patents may help address some of these uncertainties.

Quantitative Understanding of the Linkage

Relationship: 2312: Binding to ACE2 leads to Downregulation of ACE2

AOPs Referencing Relationship

Key Event Relationship Description

The current KER relates to evidence for downregulation of the ACE2 following the binding of the viral spike S protein of SARS-COV (1 or 2).

According to the evidence (mostly from the study of SARS-CoV-1 in the years preceding the SARS-CoV-2 emergence and the COVID-19 pandemic) the infection of cells by SARS-CoV leads to downregulation of ACE2, either:

- in the form of shedding of the ACE2 (which has other repercussions from a RAAS point of view)

- internalization / endocytosis of the ACE2

- downregulation of mRNA / protein expression.

The literature focusing on SARS-CoV-2 is quite scarce, especially in relation to animal work (only 5 studies have been identified) up to November 2021.

Relationship: 2831: ACE2 dysregulation leads to (Micro)vascular dysfunction

AOPs Referencing Relationship

Key Event Relationship Description

ACE2, is a carboxypeptidase with broad substrate specificity that deactivates or generates active biological peptides such as Angiotensin II (Ang II), Angiotensin 1-7 (Ang1-7), des-Arg9-bradykinin (DABK) but also others (10.1042/BJ20040634; 10.1074/jbc.M200581200; 10.1002/path.2162). These bioactive peptides and their receptors constitute the Renin Angiotensin (RAS) andKinin Kallikrein Sytem  (KKS) which represent balancing regulatory networks of the vascular tone, inflammation, cell activation/proliferation, tissue remodelling and pain (10.1152/ajpheart.00723.2018; 10.3389/fnins.2020.586314).

This KER aims to summarise the evidence that addresses the relationship between ACE2 dysregulation and microvascular dysfunction as the vasculature represents one of the main targets of the RAS system (10.1152/ajpheart.00723.2018).

This KER was developed as part of the CIAO project (www.ciao-covid.net) and is therefore it focused mainly on one stressor of ACE2 dysregulation (SARS-COV2 and its spike protein which is the viral protein that binds to ACE2). Additional relevant evidence on the same or other stressors are welcomed.

Evidence Supporting this KER

Microvascular dysfunction manifesting as vasculitis of the small blood vessels in different organs has been identified in COVID19 patients in a number of studies (e.g. 10.1016/j.ebiom.2020.10318 10.1056/NEJMoa2015432; 10.1016/j.anndiagpath.2020.151645).

Angiotensin Converting Enzyme 2 (ACE2) is recognised as the main receptor for the COVID19 causing SARS-CoV2 virus. In particular the viral spike (S) protein binds to ACE2 mediating viral entry (10.1016/j.cell.2020.02.052; 10.1128/JVI.00127-20) although other receptors or less specific (recognizing glycans on the virion surface of many viruses rather than specific viral protein) to mediators of viral entry have also been suggested (10.1002/rmv.2207; 10.3390/v14112535).

ACE2 is also known to be a cell receptor for SARS-CoV (10.1038/nature02145) although its spike protein shows ~10- to 20-fold lower binding affinity for ACE2 (10.1126/science.abb2507). Evidence from SARS patients is nowhere near that for COVID19, but some evidence of multiorgan injury, including to systemic vacuities has been reported (10.1002/path.1440).

ACE2, is a carboxypeptidase with broad substrate specificity that deactivates or generates active biological peptides such as Angiotensin II (Ang II), Angiotensin 1-7 (Ang1-7), des-Arg9-bradykinin (DABK) but also others (10.1042/BJ20040634; 10.1074/jbc.M200581200; 10.1002/path.2162). These bioactive peptides are involved in regulating vascular tone, inflammation, cell activation/proliferation and tissue remodelling via their receptors Angiotensin II receptor 1 (AT1R), Angiotensin II receptor 2 (AT2R), Mas-Related G Protein-Coupled Receptor (Mas) (10.1152/ajpheart.00723.2018; 10.3389/fnins.2020.586314).

Therefore it is plausible to explore the evidence examining spike protein-mediated dysregulation of ACE2 on aspects of (micro)vascular dysfunction.

The protein profile of blood plasma represents a good indicator of (micro)vascular dysfunction at tissue level.

Increase of soluble ACE2 protein (sACE2) in plasma of SARS-CoV-2 positive patients was correlated with increase of the level of von Willebrand factor (vWF) and coagulation factor VIII comparing COVID19 to healthy controls by ELISA assay (10.1002/jmv.27144). Another study using proteomic/OLING approach and comparing viremic and aviremic COVD19 patients (10.1172/JCI148635) found increase of a larger panel of angiogenesis and endothelial damage related markers, including the vWF, and also coagulation/fibrinolysis related factors: tissue factor III and SERPINE1/Plasminogen activator inhibitor-1 (PAI-1), as well as complement related pathway proteins (PTX3, C1QA). In this study PROC was decreased in the viremic versus aviremic SARS-CoV-2 patients. Importantly, the proteomic study (10.1172/JCI148635) find that the dysregulation of specific endovascular damage markers (ANGPT2, ESM1), coagulation pathway markers (F3/tissue factor, vWF and PROC) and also the complement related pathway proteins (PTX3, C1QA) was specifically associated with SARS-CoV-2 viremia and not with generalised inflammation comparing SARS-CoV2 negative participants with respiratory dysfunction and an inflammatory profile (CRP >10 mg/dL). However, it is not clear whether ACE2 level is also significantly increased specifically in the SARS-CoV2 viremic patients compared to non COVID-19 controls (the supplementary data is not detailed -see sup fig 7). ACE2 mRNA is known to be upregulated by other infections, inflammation, hypoxia and synthetic dsRNA, poly(I:C) (Ziegler, 2020, Smith, 2020; Salka 2021, Zhuang 2020).

Unlike the proteomic study, the (10.1002/jmv.27144) did not find evidence for dysregulation of SERPINE1/PAI-1 in COVID-19 vs healthy controls. In fact, SERPINE1/PAI-1 protein levels were unaffected as were the level of ACE in plasma in COVID-19 patients compared to healthy controls.

Macroscopic indicators of dysfunction of the vasculature have also been reported in hACE2 transgenic mice infected with SARS-CoV (PMID 17974127) and SARS-CoV2 (10.1101/2020.08.19.251249 – bioRxiv; 10.1172/jci.insight.142032).  (PMID 17974127) examined different organs and found: in LUNG: oedema, focal haemorrhage, degenerated blood vessels surrounded by monocytic and lymphocytic infiltrates; KIDNY: glomerular capillaries dilated markedly and engorged with lymphocytic infiltrate present in the renal interstitial; BRAIN - vasculitis, and haemorrhage. With SARS-CoV2 multiple intra-alveolar and intra-arteriolar microthrombi adherent to the endovasculature were observed that stained positive for the platelet marker CD61 (glycoprotein IIIa), as well as intra-bronchiolar blood clots and interstitial hemorrhagic patches (10.1101/2020.08.19.251249). Another study with SARS-CoV2 (10.1172/jci.insight.142032) report that vasculitis was the most prominent finding in the lungs of infected transgenic mice.

Wild type mice are resistant to SARS-CoV-2 infection as mouse ACE2 is not binding efficiently to the viral S protein, supporting the role of S-binding to ACE2 (direct or by the virtue of amplification of the stressor) in mediating the observed vascular damage and dysfunction. The findings are also consistent with autopsy findings in COVID-19 patients that showed widespread microthrombi and microangiopathy (/10.1016/S2352-3026(20)30216-7, 10.1007/s00428-020-02886-6, 10.1038/s41379-021-00793-y 10.1161/CIRCRESAHA.120.317447 and references therein).

There is also supporting evidence in lung tissue from autopsies of SARS-CoV2 infected humans compared to controls that died from unrelated causes including car accidents, that indicate increase of ACE2 protein in the lings (by immunostaining) correlated with increase of clotting factor VIII (FVIII) by immunostaining (10.3390/diagnostics10080575; 10.1002/path.5549). Another study identifies gross changes to the vascular component in both the interstitial capillaries and the mid-size vessels, with capillary fibrin micro-thrombi, as well as organized thrombi even in medium-sized arteries, in most cases not related to sources of embolism.

Similarly, increased ACE2 expression associated with multiple fibrin trombi in medium size pulmonary vessels without signs of inflammation is reported in COVID-19 autopsies compared to control [10.1002/path.5549].  Based on CD34 staining authors observe preserved but desquamation of the endothelial cells from the vascular walls, indicating endothelial damage and perturbed intercellular communication 10.1002/path.5549]. These case studies are based on very small number of patients and are not quantitative, but appear to converge on the finding that ACE2 immunohystochemical staining is increased in the lungs of patients with terminal COVID-19, particularly in the regions of small vasculature and alveolar damage. In lung autopsies from COVID-19 patients [10.1093/cvr/cvac097] S-protein collocalised (immunostaining) with the pirocyte marker Neural-glial antigen 2(NG2)/ Chondroitin sulfate proteoglycan 4 (CSP4). Collocalisaton staining was more pronounced in patients with thrombotic complications. Endothelial marker CD144/Cadhirin was also decreased in COVID-19 autopsies compared to age matched controls, and pronounced in patients with thrombotic complications. At the same time WWF factor was increased. Authors note that NG2+ Spike+ cells (actiated pirocytes phenotype) were not lining the endothelium, in particular, cells with high spike staining was associated with ICAM-1 staining suggesting pericyte activation and detachment in terminal cases of COVID-19. Together with the decrease of CD144 these finding strongly suggest dysregulation of the pirocyte-endothelial communication. ACE2 was not evaluated in vovo, but the same study included analisis in human iPCS derived 3D vascular organoids (see section on in vitro data).

10.1016/j.anndiagpath.2020.151645 examined biopsies form different tissues of patients with severe COVID19 (24 out of 26 with fatal outcome). Different degrees of (micro)vascular damage was observed in the different tissues, also without signs of observable inflammation. Small (micro) and bigger thrombi were observed in all tissues in the different size vasculature examined. Endothelial and sub-endothelial microvascular deposition of C3d, C4d and/or C5b-9 was detected by Immunohystochemical staining in the lung, liver, heart and brain autopsies. However, viral RNA was not detected (by in situ hybridisation) in endothelial cells but in cells with the morphology of macrophages. Furthermore, endothelial cells of the microvasculature showed no, or low (~10% of cells), immuno staining for ACE2 in most tissues, except the subcutaneous fat and in the brain where it collocalised with spike protein. Authors suggest spike-containing abortive pseudovirions can be released from infected cells and they can dock on ACE2 positive endothelial cells most prevalent in the skin/subcutaneous fat and brain, leading to activation of the complement pathway/coagulation cascade resulting in thrombogenic vascular dysfunction. In an earlier study focused on skin biopsies of COVID19 patients, [10.1016/j.humpath.2020.10.002] found evidence of endothelial cell injury also largely unaccompanied by significant infiltration of inflammatory cells, where the affected microvascularure in the skin showed significant immunohistochemical staining for C5b-9, C3d, and C4d. C5b-9 colocalized with ACE2 and spike protein.

A number of in vitro studies are available that use particularly recombinant spike protein (or its S1 or S2 subdomains) to explore the mechanistic drivers of the relationship between ACE2 dysregulation elicited by the binding of the spike protein and any aspect of the observed (micro)vascular disfunction at tissue/organ level. Mainly, different endothelial cell cultures (primary and immortalised) are used, but also 2D and 3D co-cultures.

Different types of dysregulation are explored and reported depending on the focus on the researchers, methods used for detection of specific markers of dysregulation and cell systems available:

•Loss of the barrier integrity based on: Electric Cell-Substrate Impedance sensing (ECIS), trans-endothelial electrical resistance (TEER), FITC-dextran or or FITC-BSA permeability assays. [10.1016/j.nbd.2020.105131; 10.1152/ajplung.00223.2021; 10.3389/fcvm.2021.687783;  10.4081/monaldi.2022.2213].

•Cell death and apoptosis preferentially of pericytes and to a lesser extent of ECs [10.1093/cvr/cvac097] – 72hpt

•Surface expression of adhesion proteins (↑ICAM-1, ↑VCAM-1) based on FACS-SCAN assay [10.1016/j.nbd.2020.105131 – significant outcome detected as early as 4hpt; (↓JAM-A, ↓Connexin-43, ↓PECAM based on Western blotting, time not specified) - 10.3389/fcvm.2021.687783]. No effect on mRNA for VCAM and E-selection  have also been  reported monitored 24hpt [10.4081/monaldi.2022.2213].

• Up-regulation of pro-inflammatory cytokines (IL6, IL1, CCL5, CXCL10) (qRT-PCR) [10.1016/j.nbd.2020.105131 significant outcome detected as early as 4hpt]. IL8, CXCL10, TGFbeta, MCP-1/CCL2 (qRTPCR) [10.4081/monaldi.2022.2213 measured only at 24hpt, CCL5 and IL6 not significantly changed in this study].

• Up-regulation of matrix metalloproteinases (e.g.MMP2, MMP3, MMP9, MMP1) and their regulators (e.g. MMP inhibitor TIMP1 (qRT-PCR) [10.1016/j.nbd.2020.105131 significant outcome detected as early as 4hpt];

• Up-regulation of anti-inflammatory cytokines (e.g. IL4, IL11, IL10RB) and down-regulation of inflammasome (NLRP3) and apoptotic pathways (CCL17) (Nanostring nCounter Assay and qRT-PCR for selected mRNAs to confirm) [10.1161/STROKEAHA.120.032764 – study monitored outcomes at 24hpt]

• Production of coagulation and/or fibrinolytic factors such as the plasminogen activator inhibitor 1(PAI-1), the inhibitor of the pro-fibrinolytic factors plasminogen activator (tPA) and urokinase (uPA). [10.1165/rcmb.2020-0544OC – study monitored outcome 24hpt at protein level by western blotting of cell lysates with anti PAI-1 antibody). Time course study in a co-culture test system showed increase of synthesis and secretion of coagulation factors: tissue factor (TF), factor (F)-V, thrombin, and fibrinogen with decrease in tissue factor pathway inhibitor (TFPI). mRNA, but also protein synthesis and secretion was measured) [10.3390/ijms231810436 – effect seen as early as 6hpt with two variants of S-protein].

The above findings have been obtained with different cell systems:

10.1016/j.nbd.2020.105131 - 2D&3D primary culture models with human brain microvascular endothelial cells (hBMVECs/D2) and immortalised hCMEC (telomerase-immortalized human brain endothelial cell line ) 3D perfused cultures. Assays done 4hpt.

10.1093/cvr/cvac097 – vascular organoids derived from human-induced pluripotent stem cells consisting of endothelial cells and pirocytes

10.1152/ajplung.00223.2021; 10.3390/ijms231810436 - human lung microvascular endothelial cells (HLMEC), alone or in co culture with neutrophils

10.1161/STROKEAHA.120.032764 -human brain microvascular endothelial cells (HBMECs) & human umbilical vascular endothelial cells HUVECs. This study also identified some differences between the activation profiles of HBMECs and HUVECs. Characteristically different in HBMEC was the observed ↑ in C3, CCL8, IL15, IL9 that was not observed in HUVECs.

10.4081/monaldi.2022.2213 - human umbilical vascular endothelial cells HUVEC

10.1165/rcmb.2020-0544OC - Human pulmonary microvascular endothelial cell (HPMECs)

In vitro 3D vascular organoid system [10.1093/cvr/cvac097] also supported studies with replicating virus. Organoids infected with SARS-CoV2 (0.5 MOI for 48h), spike protein immunofluoresecence (imaging) co-localised 2x more with pyrocyte markers (CD140b(PFGFR)+/NG2+ cells)  than with UAE1-positive ECs. Pirocyte apploptosis was also significantly higher (TUNEL assay). Presumably, UAE1 rather than CD144 was chosen as a “housekeeping” EC marker as ACE2 (spike receptor) co-localised 2x less with EC (CD144+) cells than with perycites (CD140b(PFGFR)+/NG2+ cells) in uninfected organoids. Taken together, the results strongly suggest that that SARS-COV2 is preferentially taken up by pirocytes via the ACE2.  Viral entry was associated with increased cell apoptosis (TUNEL assay) which was significantly higher in pericytes (markers) compared to ECs (markers).

Given the preferencial infection and death of pericytes, the relatively low EC death (TUNEL) and the observed CD144 decrease (immunostaining) suggest that ACE2-dependent virus entry into pyrocites could lead to cell death and further to vascular dysfunction due to loss of functional pirocyte-EC communication.

In the same organoid test system [10.1093/cvr/cvac097], treatment with recombinant viral S or N proteins (72h) decreased the percentage of live pirocytes and ECs (FACS scan for live/total CD140b(PFGFR)/NG2+ and CD144+/CD31+ cells, respectively). In the case of ECs the % decrease of live cells appears comparable after treatment with S, N and combination of S+N protein. In the case of perycites, effect of S alone and in combination with N appears much more pronounced.  For both cell types, treatment with proinflmatory IL1-beta togather with S, N or both antigens, did not have additional cytotoxic effect. IL1beta alone, had greater cytotoxic effect on ECs compared to pirocytes.   However, neither EC nor pirocytes appeared to be activated by the S or N protein treatment as judged by the lack of significant change of CD31 or CD140b(PFGFR)/NG2+/ICAM-1+  cells, respectively.  Similarly expression of WVF, IL6 or IL8 (ELISA of supernatants) was not affected by S or N treatment of ECs. In contrast, IL1-beta alone activated both cell types (as judged by increase of ICAM-1+ cells of both types) and also induced of WVF, IL6 or IL8 production by ECs.There was no difference between the teratemnt with IL1-beta alone or with the recombinant viral proteins, suggesting that viral replication rather than interaction with the viral spike (or N) protein may be driver of cell activation observed in the patients lung authopsies . At mRNA level, joined S + N treatment induced ACE2 expression (qRT-PCR) as early as 4hpt, peak at 24hpt, maintained up to 72hpt. Individual antigen treatment is not tested (or shown?).

Taken together thehe lines of evidence from testing systems of different complexity [10.1093/cvr/cvac097]  suggests that in vivo the most likely primary target for viral infection or uptake of S-containing abortive pseudoviral particles in he (micro)vasculature are the pericytes. Infection/antigen uptake induce pericyte damage which dysregulates the intercellular crosstalk with ECs which in turn dysregulates overall vascular permeability (one potential aspect by interference with CD144 expression/function). In a parallel process EC may be also be activated by the inflammatory response in vivo, leading to significant increase in vascular permeability and microvasuclar dysfunction. Neutrophils are not present in this test system where apparently different results were obtained [10.3390/ijms231810436], where on the other hand pericytes are not present.

Uncertainties:

It is difficult to decipher or generalise which if any particular cell type in the (micro)vasculature represent a key factor in the putative spike-protein mediated (micro)vasculature dysfunction, particularly because the expression of ACE2 varies significantly in the cell systems used, (see Table below). Closer analysis is needed regarding the reagents/antibodies used in these studies and comparative characterisation of the experimental systems, as it appears that ACE2 expression may depend on the dimensionality (2D/3D, co-cultures) but also the dynamics (medium flow or stasis) of the test system.

Notably, ACE2 expression in HBMEC & HUVEC perfusion culture is stimulated by flow (HBMEC < HUVEC) (qRT-PCR); also it is increased by flow intensity and vessel shape in the MCA 3D model of stenosis (immunostaining cells) [10.1161/STROKEAHA.120.032764].

Single­cell RNA­seq analysis of quiescent mouse endothelial cells isolated from different tissues by flow cytometry without the cell culture step, does not identify ace2 enrichment in the ECs from any of the mouse tissues while ace is one of the top-50 marker genes in ECs from arteries in the brain, testes and small intestine. scRNAseq of primary cell cultures from EC from human lung identified similar cell subpopulations and marker genes in mice and human, but interestingly, scRNAseq of commercially available primary lung ECs demonstrated a loss of their native lung phenotype in culture . (Schupp 2021 - 10.1161/CIRCULATIONAHA.120.052318). 10.1038/s41419-020-03252-9

There are also some apparently conflicting results. For example although [10.1016/j.nbd.2020.105131] shows S-mediated increase of many markers of MV dysfunction, without evidence for any change in ACE2 levels as measured by Western blot of cell culture lysates. This suggests that although the effect is spike protein mediated it may not necessarily be ACE2 mediated and that other receptors for S protein may be involved in MV.

Additional uncertainty in this KER evidence analysis is that most of the studies that use various cultured endothelial cell models, do not specifically measure/identfy ACE2 dysregulation aspects in the same time and binding of S-protein to ACE2 is often assumed. In fact, infection of endothelial cells by SARS-CoV2 is still debated. [10.1021/acscentsci.0c01537] show that ACE2 is expressed at very low level in HUVEC-TERT, immortalised umbilical endothelial cell culture. In contrast two C-type lectin receptors, CD209L (also known as L-SIGN) and CD209 (also known as DC-SIGN), are highly expressed. CD209L and CD209 bind spike RBD domain and mediate SARS-CoV-2 entry into HUVEC-TERT (demonstrated by interference with CD209L). Thus, while endothelial cells may be permissive to SARS-CoV-2 entry/replication, the involvement of ACE2 in these undifferentiated cells in monolayer culture, is uncertain. The same study demonstrates that, when expressed, ACE2 mediates pseudovirus entry more efficiently and in vitro binds S-protein RBD with higher affinity than CD209L and CD209 [10.1021/acscentsci.0c01537].

Sumilarly, [10.3390/ijms231810436 ] that demonstrates effect of spike treatment on a number of markers of endothelial dysfunction in HLMEC alone, or in co-culture with neutrophils also does not characterise the test system for ACE2 status or time concordance with any potential ACE2 dysregulation in either of the cell types in the co-culture. Such characterisation would be helpful in understanding the primary target of the viral or abortive spike-containing particles, as well as the intercellular paracrine interactions driving adversity at tissue level.

Therefore, although in some studies the dependence of aspects of (micro)vacular dysfunction on ACE2 is demonstrated [10.1128/mBio.03185-20], the type of ACE2 dysregulation/effect (upregulation/downregulation) that could be related to the observed aspect of (micro)vascular dysregulation is not clear.

Other plausible mechanisms:

Viral replication and inflammation-related oxidative stress are also explored as significant factors in coagulation and DIC following SARS-CoV-2 infection/stress and may be more direct drivers of the (micro)vascular dysfunction compared to ACE2 dysregulation. Consistent with this, [10.1128/mBio.03185-20] find that primary cultures of scattered endothelial cells derived from microvascular tissue from different organs, express very low levels of ACE2 (if any) and cannot be infected by SARS-CoV2. However, when transduced to overexpress hACE2 they could be infected and the interaction with the replicating virus led to stimulation of mRNA synthesis for coagulation and pro-inflammatory mediators: tissue factor (TF), thrombomodulin (TM), tumour necrosis factor alpha (TNFa), interleukin 6 (IL-6), IL-1b, and E-selectin (qRT-PCR assay following 6h interval time course from 0-24hpi showing significant stimulation for TF and TNFa, as early as 6hpi and for the rest of the mediators  at latter time points). E selectin was also stimulated early with pick at 12hpi. In human heart tissue of COVID-19 infected patients ACE2 immuno-histological staining showed increasing expression towards the small vessels:  capillaries >> arterioles/venules. The main coronary arteries were virtually devoid of ACE2 receptor [10.1016/j.ebiom.2020.103182]. In the same study, staining of harts from influenza infected patients showed much lower expression of ACE2 even in capillaries, and in small vessels, practically the same in coronary artery.

ACE2 is known to be upregulated by other inflammation, hypoxia and synthetic dsRNA, poly(I:C) (Ziegler, 2020, Smith, 2020; Salka 2021, Zhuang 2020, indicating that inflammation, low oxygen levels and viral replication ma precede or in parallel potentiate ACE2 mediated MV dysfunction by increasing the ACE2 levels in the MV cells which then maintain viral infection cycle but also govern MV dysfunction via spike-protein binding mechanism in the surviving cells.

ACE2 independent (though spike-mediated) mechanisms leading to microvasuclar disfunction (TEER disruption), have also been suggested [10.1038/s41467-022-34910-5]. In this study human pulmonary microvascular endothelial cells (HPMEC) were used as a representative endothelial cell relevant to COVID-19 pathology that do not endogenously express ACE2 and are not permissive to SARS-CoV-2 infection (as demonstrated by the authors in their study). That allowed them to separate viral replication from S-mediated dysfunction and compare them to transfected ACE2 overexpressing line of HPMEC and also HPMEC in which ace2 gene was disrupted by CRISPR-Cas9. The alternate mechanism appears to involve perturbation of surface levels of key endothelial glycocalyx layer (EGL) components, including sialic acid (SIA), heparan sulfate (HS), hyaluronic acid (HA), and chondroitin sulfate (CS) (immunofluorescence staining) and EGL-degrading enzymes such as hyaluronidase and neuraminidase. Integrins and TGFbeta sugnaling also appar essencial for tis mechanism. Intradermal and intranasal administration of recombinant spike protein also appeared to disrupt endothelial barrier function in vivo in leak models in WT C57BL/6 mice which do not express human ACE2 and are not permissive to infection by most SARS-CoV-2 variants. Binding of spike protein to alternative (co)receptors including integrins and heparin sulfate has been implicated as well [10.3390/v13040645; 10.1016/j.cell.2020.09.033]. It should be noted that in some test systems ACE2 mRNA was up-regulated as a result of spike treatment in vitro (eg. 80% confluent HUVEC cell [10.1016/j.jbc.2022.101695s]) even though the initial spike interaction was likely to an alternative receptors leading to activation of the quiescent endothelium. Resulting ACE2 up-regulation and/or activation of quiescent endothelium could initiate additional and parallel ACE2 dependent perturbations leading to vascular dysfunction [10.1021/acscentsci.0c01537].Contribution of the effects of the activation of the adaptive immune response to spike and other viral components or endogenous host prpteins relevant to (micro)vascular dysfunction  is possible but not explored sufficiently by this evidence collection strategy, although it has been indicated [10.3390/pathophysiology29020021; 10.3389/fimmu.2022.906063]

Gaps:

Overall, the studies with more differentiated 2D/3D cultures and recombinant S protein as a stressor, strongly suggest that spike-mediated ACE2 dysregulation could also, in part lead to aspects of (micro)vascular dysfunction. The initial target cells still remain to be elucidated. This may vary in different organs. Notably, a very recent cell atlas of adult human myocardium made by single nuclear transcriptome analysis found that the highest level of ACE2 mRNA in the heart was found in the pericytes a type of perivascular mural cells [10.1093/cvr/cvaa078]. Therefore, more complex and better characterised in vitro, optimally microfluidic dynamic systems, are needed to better understand molecular and cellular aspects of SARS-CoV-2, spike protein-mediated or more general toxicity (including general inflammation) -mediated (micro)vascular dysfunction. One potentially very useful system has already been described. [10.1016/j.cell.2020.04.004] developed human capillary organoids from induced pluripotent stem cells (iPSCs). They resembled human capillaries with a lumen, CD31+ endothelial lining, PDGFR+ pericyte coverage, as well as formation of a basal membrane. Consistent with [10.1021/acscentsci.0c01537} these organoids were permissive to SARS-CoV-2 infection and replication that could be significantly but only partially blocked by soluble human recombinant ACE2 in a dose dependent manner. However, the target cell and the ACE2 status of the different cells in the organoids are not reported and authors discuss the possibility that there might be other co-receptors/auxiliary proteins or even other mechanisms by which viruses can enter cell.

However, in another iPCS derived organoid system with characterised ACE2 expression, authors show that perycites express significantly higher levels of ACE2 also show preferential uptake of viral particles. In addition productive infection of perycites has been found in primary cultures of cardiac perycites in vitro and in hart tissue from COVID-19 patents [10.1016/j.jacbts.2022.09.001].

In addition, the role of microparticles (MPs) and microvesicles (MV) in mediating intercellular communication or its perturbation need to be examined in more detail.  [10.1152/ajpheart.00409.2022] show that plasma MV from platelet and white blood cell origin contain high amounts of ACE2 that binds the spike protein of SARS-CoV-2. MVs also contain other adhesion moolecules (eg. Tissue Factor- CD152 ) mediating EC adhesion and alternative entry pathway by endocytosis for viral or abortive spike-containing particles with affinity for ACE2, ultimately leading to activation and dysfunction of the EC and micro(vsculature). It is interesting that [10.1152/ajpheart.00409.2022] found difference in the ACE2 containing MVs between normal and diabetic patients which have increased incidence of severe and fulminant COVID-19.

Extracellular vesicles (EVs) including microvesicles (MVs) and exosomes can also mediate cell–cell signalling in a paracrine and/or even endocrine mechanism via their protein and RNA cargo (10.1093/nar/gky985; 10.1016/j.cell.2016.01.043). Non coding RNAs such as microRNAs (miRNA or miR) have been implicated in various biological processes including angiogenesis and vascular dysfunction (10.1530/VB-19-0009). A number of miRs (eg miR122, miR143, miR155, miR181a and miR1246, miR421) have also been implicated in regulation of ACE2 in different tissues and plasma (10.1016/j.ncrna.2020.09.001; 10.1016/j.yjmcc.2020.08.017; 10.1016/j.omtn.2022.06.006; 10.1042/CS20130420). Therefore interference with the expression of relevant miRs by exogenous stressors, including RNA viruses, LPS or dysregulated endogenous signalling molecules (eg. vasoactive peptides) can potentially lead to ACE2 dysregulation and (micro)vascular dysfunction and need to be examined in greater detail.

Finally, the distribution and role in (micro)vascular dysfunction, of the receptors for the substrates and products of ACE2 need to be investigated in these systems to better elucidate a potential link with up- or down-regulation of ACE2 function locally and/or systemically. Test systems that could monitor activation of the complement system proteins would also be valuable.