Observational studies have demonstrated that patients who experience sCOVID-19 are also more susceptible to IS. A recent meta-analysis indicated that the risk of stroke increased by 500% for those who were confined to the ICU due to sCOVID-19 (OR 5.1, 95% CI 2.72–9.54) [13]. Observational research, however, is not well-suited to distinguishing between causal associations and those resulting from confounding factors or reverse causation. Consistent with past studies [5, 8–9, 13], the present MR approach revealed that COVID-19 hospitalization and sCOVID-19 were causally associated with an elevated IS risk level. No comparable causal relationship was observed regarding genetic susceptibility to SARS-CoV-2 infection. This report presents genetic evidence linking COVID-19 severity to the risk of ischemic stroke through a comprehensive MR approach. Furthermore, ten genes (KCNJ2, CFL2, TPM1, LHFPL6, PELI1, FAM20C, SPAG9, C15orf39, TNFSF8, and HLA-L) have been identified as potentially causal in the incidence of IS among patients with severe COVID-19. This finding offers significant insight into the underlying mechanisms of this association, suggesting these genes may serve as promising targets for therapeutic interventions in cases of severe COVID-19 complicated by IS.

Factors suggested to play a role in IS incidence among COVID-19 patients include alternative RAAS pathway activity and immune-mediated thrombosis or hypercoagulopathy [14–15]. This report identifies shared causal genes primarily associated with the immune response and the activation of the RAAS pathway. Hemostasis is intricately linked to the activity of the immune response and the process of inflammation. In instances of COVID-19, both the innate and adaptive immune systems are activated, leading to significant activation of inflammatory cells, including macrophages and neutrophils, alongside complement activation and the release of various pro-inflammatory cytokines, such as Interleukin (IL)-1, IL-2, IL-6, IL-8, IL-10, and IL-17, resulting in a phenomenon known as a cytokine storm [15–16]. Activated immune cells are capable of causing damage to endothelial cells through the exposure of tissue factor (TF) and the induction of microvascular thrombosis. In a hyperinflammatory condition, locally activated platelets can induce the release of tissue factor-coated neutrophil extracellular traps (NETs), initiating the extrinsic coagulation cascade and resulting in thrombin production [17]. Inappropriate host immune responses in this context lead to interactions among platelets, endothelial cells, and various immune system components, resulting in hypercoagulability and excessive microvascular immune-mediated thrombosis. Angiotensin-converting enzyme 2 (ACE2) facilitates the entry of SARS-CoV-2 into cells, which is essential for the viral replication process [18]. It also serves a necessary role as a suppressor of angiotensin II (Ang II) in the RAAS pathway [19]. In healthy individuals, angiotensin II is converted into Ang (1–7) via ACE2 [15, 19]. ACE2 dysfunction may result from SARS-CoV-2 binding in COVID-19 patients, potentially impacting Ang II conversion into Ang (1–7) [20], with Ang II accumulating in affected individuals. When Ang II binds to its receptor (AT1R), it functions as a potent vasoconstrictor that can adversely affect various tissues by promoting oxidative stress, inflammation, fibrosis, and vascular remodeling [20]. Further research on IS in animals has shown that Ang (1–7) can activate the Mas receptor and AT2R to locally produce anti-inflammatory, antioxidant, and vasodilatory effects within the grain [21]. The ability of SARS-CoV-2 to disrupt these neuroprotective functions of ACE2 may thus culminate in the occurrence of strokes.

The CFL2 gene encodes cofilin-2, an actin-binding protein expressed in a range of eukaryotic cells. Cofilins are attributed with numerous functions, playing a critical role in cell stress responses, locomotion, and cytokinesis, and are pertinent in various pathological contexts [22]. In mouse model systems, myocardial infarction results in cofilin-2 upregulation and cardiac NLRP3 inflammasome upregulation [23]. Inflammatory cytokines can promote the upregulation of CFL2 in cells treated with lipopolysaccharide (LPS). In the context of infection, cofilin-2 can exert damaging pro-apoptotic effects through its ability to promote mitochondrial cytochrome c release [25–26]. L-cofilin-2 also reportedly plays a role in LPS-induced immunosuppressive responses [23]. The Golgi casein kinase Fam20C phosphorylates the SxE/pS motifs of proteins that are secreted [24]. Multiple Fam20C substrates associated with coagulation were found using the phosphoproteomic analysis of serum and plasma samples. Upon vascular and tissue injury, thrombin cleaves fibrinogen to generate fibrin peptides, resulting in blood clot formation and cessation of bleeding [24]. The ability of phosphorylated fibrinogen to bind to thrombin has been reported to be enhanced, allowing for the release of a more significant number of fibrin peptides, resulting in more rapid coagulation [25]. Fam20C is reportedly capable of the direct phosphorylation of the gamma and alpha chains of fibrinogen in vitro, in addition to phosphorylating two SxE sites within the von Willebrand factor (vWF) A2 domain (pSer1517 and pSer1613). The modifications facilitate platelet adhesion at the site of vascular injury, thereby contributing to coagulation. Fam20C can phosphorylate the C3 and C4 complement proteins. Liu et al. identified a positive correlation between the infiltration of macrophages, neutrophils, and dendritic cells and the expression of Fam20C. They also proposed that Fam20C may be involved in Treg activation and the induction of T cell exhaustion [26]. Inflammatory and coagulation crosstalk may thus explain the roles that CFL2 and Fam20C play in sCOVID-19 and IS development.

The E3 ubiquitin ligase pellino1 (Peli1) exhibits a high degree of conservation and functions through its ability to mediate the ubiquitin modification of target proteins [27]. Peli1 has been demonstrated to play a crucial role in regulating various inflammatory signaling pathways involving Toll-like receptors (TLRs), IL-1 receptors, MAPKs, PI3K/AKT, and NF-kB [27]. Peli1 is proposed to contribute to infections, coagulation, and immunological responses by regulating glycolysis, DNA damage, autophagy, necrosis, pyroptosis, and apoptosis [27]. Peli1 can promote inflammatory activity by affecting the IL-1R and TLR pathways, activating NF-κB. In contrast, it can also interact with the TGF-β-induced Smad6 protein to exert anti-inflammatory effects [28]. Yang et al. explored transcriptomic datasets from patients with COVID-19. They determined that PELI1 upregulation was only evident in cases of moderate COVID-19 but not sCOVID-19, suggesting a link between this gene and a reduction in disease severity [29]. PELI1 also serves as a vital regulator of stroke pathogenesis. In large-artery atherothrombotic stroke patients, PELI1 expression was downregulated, consistent with the inhibition of IS [30].

KCNJ2, located on chromosome 17 in humans, encodes the Kir2.1 K + channel, expressed by mononuclear cells in peripheral blood. Its expression levels are positively connected with ventricular expression and inversely correlated with IL-1 and CRP levels in patients with acute infections [31]. Kir2.1 modulates the plasma membrane potential (Vm) of macrophages, and this regulatory function is crucial for food intake and subsequent pro-inflammatory metabolic reprogramming. In the absence of Kir2.1 activity, the depolarization of macrophage Vm induces a condition of caloric restriction, leading to the depletion of epigenetic substrates and altering the histone methylation status of metabolism-responsive inflammatory gene clusters, thereby inhibiting their transcriptional activation [32]. Pharmacological efforts to target Kir2.1 can protect against LPS- or bacteria-induced inflammation in sepsis model systems while protecting against sterile inflammation in human samples [32]. As Kir2.1 plays a selective anti-inflammatory role, it can reportedly decrease the size of infarcts in MCAO animal models [33]. TNFRSF8, a type I transmembrane glycoprotein belonging to the TNRSF superfamily, possesses a cytosolic domain that contains a TNFR-associated factor (TRAF) binding domain. This domain can interact with TRAF1 and TRAF2, increasing the activation of NF-κB and thereby positively influencing T cell activity [34]. TNFRSF8 is an essential regulator of T cell-mediated immune responses directed against intracellular bacteria. TNFRSF8-deficient mice with M.avium infections present with a reduction in IFN-γ-producing cell numbers, a more significant bacterial burden, and abnormal inflammation [35]. There is a potential relationship between TPM1 and long COVID-19. According to Gui et al., when the influenza A virus is present, SPAG9 can stimulate necroptotic, pyroptotic, and apoptotic cell death by interacting with DAI via the c-Jun N-terminal kinase pathway [36]. The expression of LHFPL6 was positively correlated with M2 macrophage abundance, with these cells serving as robust immunosuppressive agents [37]. C15orf39 has been established as a novel substrate of MAPKs, which can regulate hemorrhagic and ischemic vascular diseases [38]. However, relatively little is known about the link between these variables and the pathophysiology of sCOVID-19 or IS.

This study has some limitations. Firstly, the generalizability of these findings is unknown, as they were conducted using data from a European population. Furthermore, the sample size of the GWAS data was somewhat constrained, limiting the number of SNPs available as instrumental variables. These GWAS studies also neglected to account for all risk variables associated with sCOVID-19, such as obesity and diabetes, potentially influencing the study outcomes. Furthermore, the control groups were not screened, raising the possibility that some healthy controls had moderate or asymptomatic infections. The outcomes of eQTL-based analyses are less likely to have been impacted by confounding variables. Thirdly, while efforts to reduce bias, including additional sensitivity analyses, were implemented, it is impossible to exclude the possibility that these results stem from horizontal pleiotropy. Therefore, these findings and candidate genes should only be considered “potentially causal,” highlighting the necessity of future validation research in larger patient cohorts.