Type-I interferons in atherosclerosis

Cardiovascular diseases (CVDs) are the leading cause of death worldwide (Hess et al., 2017; Lozano et al., 2012; Wang et al., 2016). Atherosclerosis, the main underlying causal factor of CVDs, is a chronic inflammatory disease driven by lipid accumulation in the arterial intima where modified low-density lipoprotein (mLDL) deposition provokes the recruitment of blood-derived immune cells and triggers inflammatory cascades (Rafieian-Kopaei et al., 2014; Tabas et al., 2007). Pro-atherogenic lipids are taken up mainly by smooth muscle cells (SMCs) and monocytes/macrophages, which subsequently secrete pro-inflammatory cytokines and chemokines (den Brok et al., 2018;,Owsiany et al., 2019). In lipid-laden macrophage-derived foam cells, endoplasmic reticulum stress-associated apoptosis can be induced by high cholesterol, mLDL-triggered pattern recognition receptor (PRR) signaling, and increased inflammatory cytokines in the plaques (Seimon and Tabas, 2009). As such, atherosclerotic lesion development is characterized by the recruitment of monocytes and macrophages, accumulating pro-apoptotic foam cells and SMCs, infiltrating leukocytes, plaque-stabilizing collagen deposition, and phagocytes responding to engulfed cellular debris (Hansson, 2005; Moore and Tabas, 2011; Williams et al., 2019). As the activation hierarchy progresses to a chronic process, the spatiotemporal homeostasis between inflammation and disease-suppressing resolution pathways is disrupted. Unresolved inflammation, together with subsequently impaired efferocytosis, leads to cell necrosis, microvessel formation, fibrous cap thinning, and destabilization of the advanced atherosclerotic plaques (Kojima et al., 2017; Rafieian-Kopaei et al., 2014). Various cell types with high heterogeneity are involved in this pathogenic process. Notably important are differentially activated monocytes and macrophages, dendritic cells, neutrophils, T and B lymphocytes, endothelial cells (ECs), and SMCs (Döring et al., 2017; Ketelhuth and Hansson, 2016; Stöger et al., 2012).

The unstable advanced plaques are prone to rupture, increasing the risk of thrombosis and consequent ischemic heart diseases and stroke. Traditional pharmacological strategies for atherosclerosis prevention and treatment focus mainly on reducing plasma low-density lipoprotein (LDL) levels. Given the success of cholesterol-lowering therapy, mainly by statins for secondary prevention of CVDs, new therapeutic approaches usually develop on top of the widespread statin treatment. However, emerging evidence from both clinical and experimental studies reinforced the beneficial effect of dampening inflammation in atherothrombosis where the Canakinumab Anti-Inflammatory Thrombosis Outcomes Study, applying anti–IL-1β antibody therapy, exhibited a significantly reduction of the risk of recurrent cardiovascular events, independent of lipid-levels (Bäck and Hansson, 2015; Hansson, 2005; Libby et al., 2009; Ridker et al., 2017, 2018; Zhang and Reilly, 2018).

IFNs are a group of cytokines named by their characteristics of viral interference (Isaacs et al., 1957;Pestka, 2007). IFNs can be classified into three families (type-I, -II, and -III) according to the protein structure and the receptors they signal through. Type-I IFNs are important immune modulator altering both innate and adaptive immunity (González-Navajas et al., 2012;Kopitar-Jerala, 2017; Trinchieri, 2010). Accumulating evidence from both human and murine studies supports their role in atherogenesis and linked clinical manifestations. Experimental data show that systemic or intraplaque type-I IFNs deteriorate atherogenesis by activating endothelium and immune cells, promoting foam cell formation, altering progenitor cell function, and enhancing pro-inflammatory leukocyte recruitment to arteries. Further, individuals suffering from autoimmune diseases with elevated type-I IFN signatures, such as systemic lupus erythematosus (SLE), are predisposed to accelerated atherosclerosis leading to increased risk of CVDs and cardiovascular (CV)-associated mortality. The syndromic concurrences of interferonopathy and cardiovascular manifestations may result from shared pathogenic processes (Ganguly, 2018).

In this review, we focus on the role of type-I IFNs in atherogenesis and discuss the potential opportunities to dampen inflammation for prevention and therapeutic intervention of atherosclerosis. In particular, we specify the effect of type-I IFNs on various atherogenic cell types (Table 1, provided at the end of the review), highlight their involvement in accelerated atherosclerosis in patients with IFN-associated autoimmune diseases, and discuss how anti–type-I IFN treatments could serve as multi-target therapy in disease.

IFNs have been known for their antiviral activity for decades. Studies have broadened the range of their biological roles in metabolic rewiring, viral and nonviral infections, cancer, and autoimmunity (Ganguly, 2018; Ivashkiv, 2018; Trinchieri, 2010). Each of the three IFN family members signals through distinct receptors to exert their function. The type-I IFN receptor (IFNAR) is a heterodimer of IFNAR1 and IFNAR2 (Mogensen et al., 1999; Fig. 1); the type-II IFN receptor (IFNGR) consists of subunits IFNGR1 and IFNGR2 (Walter et al., 1995); and the type-III IFN receptor (IFNLR) is formed by IFNLR1 and IL-10 receptor β. The last is also a subunit of IL-10 receptor (Kotenko et al., 2003;Sheppard et al., 2003). The type-II IFN family has one single member, IFN-γ, with no structural similarity to type-I and -III IFNs (Epstein, 1982). IFN-γ is mainly produced by innate lymphoid cells, T cells, and natural killer cells coordinating cell-mediated and innate immune responses against viruses, bacteria, and tumors (Boehm et al., 1997; Ivashkiv, 2018; Schroder et al., 2004). In contrast, type-I and -III IFNs include various subtypes with some protein sequence homology (∼15–19% amino acid identity), and part of the antiviral activities of these two families can be compensated by either one (Crotta et al., 2013; Sheppard et al., 2003). While the three IFN families share some similar properties and signaling cascades (Raza et al., 2014; Waddell et al., 2010), how different cell types regulate the expression and decipher distinct IFN signals is not completely clear and has been shown to be pathogen dependent (Cheng et al., 2019; Lazear et al., 2019; Mesev et al., 2019;Singhania et al., 2019). In this review, we will focus on type-I IFNs. An overview of the three types of IFNs has been extensively reviewed elsewhere (Borden et al., 2007; González-Navajas et al., 2012; Kim et al., 2016; Lazear et al., 2019;Mesev et al., 2019; Wack et al., 2015).

In humans, type-I IFNs consist of 13 IFN-α isoforms (14 in mice), IFN-β, IFN-ε, IFN-κ, IFN-δ, and IFN-ω, which all signal through IFNAR (Kopitar-Jerala, 2017; Ng et al., 2016). As IFN-α and IFN-β are the most characterized and the most predominantly expressed type-I IFNs, we mainly discuss these two subtypes in this review. Most nucleated cells in the body are able to produce type-I IFNs in response to viral pathogens and the bacterial cell wall component LPS. This process is driven by activation of PRRs that induce cGAMP synthase (cGAS), stimulator of IFN genes (STING), retinoic acid–inducible gene I (RIG-I), MDA5, and IPS-1, which subsequently provoke nuclear translocation of IFN regulatory factor 3 (IRF3) and IRF7 (Fig. 1; Honda et al., 2006; Kawai and Akira, 2010; Kawai et al., 2005; Li and Chen, 2018; Zhao et al., 2015). Innate immune cells, including monocytes and plasmacytoid dendritic cells (pDCs), can secrete type-I IFNs by triggering TLR3, TLR4, and TLR7–9, which activate TRIF, MyD88, IRF1, 3, 5, 7, and 9, and myeloid-specific IRF8 (Blasius and Beutler, 2010; Borden, 2019; Li et al., 2011b; Muskardin and Niewold, 2018). In addition, TNF induces IFN-β expression by activating TNF receptors (TNFRs) and IRF1 in macrophages and ECs (Yarilina et al., 2008). Notably, pDCs are considered to be the predominant producers of IFN-α, while other cell types, including lymphocytes, dendritic cells, macrophages, ECs, and fibroblasts, mainly produce IFN-β (Gilliet et al., 2008; Izaguirre et al., 2003; Kerkmann et al., 2003; Takauji et al., 2002).

Upon type-I IFN stimulation, IFNAR1 and IFNAR2 employ tyrosine kinase 2 (TYK2) and JAK1, respectively, which results in the dimerization and activation of STAT; Hervas-Stubbs et al., 2011; Fig. 1). The most important STAT is the STAT1–STAT2 heterodimer, which assembles with IRF9 to form the ISGF3 complex (Stark and Darnell, 2012) provoking the transcription of IFN-stimulated genes (ISGs; Budhwani et al., 2018). In addition to STAT1 and STAT2, type-I IFNs also activate STAT3–6 that can form different homo- or heterodimers with distinct and sometimes even opposing biological functions. The detailed roles and regulation of different IFN signaling pathways are reviewed elsewhere (Ivashkiv and Donlin, 2014;Mesev et al., 2019; Schreiber, 2017). The varied and pleiotropic IFNAR signaling pathways suggest that the effects of type-I IFNs are cell type– and stimulus-specific, highlighting their complex crosstalk in coordinating the immune system.

Plaque-residing monocytes and macrophages are highly heterogeneous populations with different functions during atherosclerosis development. Accumulated lipids in the arterial intima are mainly taken up by these cells. Type-I IFN treatment facilitates lipid uptake in both human and mouse monocytes/macrophages through up-regulation of scavenger receptors class A, leading to increased foam cell formation (Table 1, Fig. 2, andFig. 3; Ahmed et al., 2018; Boshuizen et al., 2016; Li et al., 2011a; Pulliam et al., 2014). Elevated scavenger receptors class A expression can also be found in human peripheral blood mononuclear cells (PBMCs) from subjects with elevated IFN signatures, such as SLE patients and HIV-infected individuals (Table 1 and Fig. 3; Li et al., 2011a; Pulliam et al., 2014). While SLE and HIV infection are associated with accelerated atherogenesis (Kearns et al., 2017; McMahon and Skaggs, 2014), these findings hint on a causal link between type-I IFN–accelerated foam cell formation and the increased CVD risk. Consistently, mice receiving IFN-β treatment showed increased atherosclerotic macrophage area and larger plaque size (Table 1; Goossens et al., 2010).

Another role of monocytes and macrophages in atherogenesis is their pro- and anti-inflammatory function. Single-cell transcriptomic studies deconvoluted the complex spectrum of blood-derived monocytes and macrophages in the murine plaque where distinct subpopulations were distinguished. A Ly-6C⁻ monocyte subset expressing highIfna1 and Il12b and low anti-inflammatory Il10 mRNA was identified by single-cell RNA sequencing (scRNA-seq) of murine atherosclerotic plaques (Table 1 and Fig. 3; Winkels et al., 2018). Whereas some monocyte and macrophage subsets might serve as a source of type-I IFNs in the murine plaques, cytokine production and IFN-responsive genes are actually suppressed in mLDL-treated or cholesterol-laden peritoneal macrophages (Table 1; Jongstra-Bilen et al., 2017; Qin et al., 2017; Spann et al., 2012). Consistently, lipid-probe based analysis suggests that intimal nonfoamy macrophages, rather than foamy cells, exhibit more pro-inflammatory characters in the murine atherosclerotic lesions (Table 1; Kim et al., 2018).

In addition to clusters of cells exhibiting pro-inflammatory NF-κB signaling signatures (M1-like) or expressing genes associated with type 2 cytokine activation (M2-like), macrophages showing up-regulated ISGs were also identified in mouse plaques, suggesting a type-I IFN responsive subset (Table 1; Kim et al., 2018; Lin et al., 2019). This IFN-associated subpopulation was enriched in progressing lesions compared with the regressing ones (Lin et al., 2019), implying the involvement of local type-I IFN signaling in the complex cytokine crosstalk during atherosclerosis development. Of note, type-I IFN–stimulated macrophages can exert both pro- and anti-inflammatory phenotypes (Cheng et al., 2019). Type-I IFNs have been shown to suppress both mouse and human macrophages’ inflammatory cytokine production and inflammasome activity through anti-inflammatory cytokine induction, such as IL-10, epigenetic regulation affecting chromatin accessibility, and metabolic rewiring, for instance, oxysterol 25-hydroxycholesterol production (Table 1 and Fig. 3; Ahmed et al., 2018; Cheng et al., 2019; Guarda et al., 2011; Reboldi et al., 2014). By contrast, upon type-I IFN stimulation, human peripheral monocyte-derived macrophages (MDMs) restore their inflammatory function silenced by TNF-induced LPS tolerance through reprogramming the accessibility of NF-κB–associated chromatin regions (Table 1; Park et al., 2017). These in vitro data corroborate the observation that isolated myeloid dendritic cells (mDCs) and macrophages from IFN-α–treated human plaque tissue ex vivo show markedly enhanced LPS-triggered TNF secretion as well as other pro-atherogenic cytokines such as IL-12, IL-23, and MMP-9 production (Table 1, Fig. 2, and Fig. 3; Niessner et al., 2007).

Neutrophils are infiltrating leukocytes present in both mouse and human plaques (van Dijk et al., 2016;,van Leeuwen et al., 2008;,Zernecke et al., 2008). During human atherosclerosis development, their number significantly increases in the vulnerable lesions (van Dijk et al., 2016). Particularly, neutrophil distribution positively correlates with lipid core size, macrophage numbers, and micro-vessels and negatively with collagen content and SMC numbers in human plaques (Ionita et al., 2010; Maracle et al., 2018). In murine atherosclerosis models, intraplaque neutrophils distribute mainly in the shoulder regions (Rotzius et al., 2010) and accumulate in the early high-fat diet period; then the number gradually reduces over time (Drechsler et al., 2010;,Rotzius et al., 2010;,van Leeuwen et al., 2008). In advanced murine atherosclerotic lesions, neutrophils correlate with necrosis, plaque size, and vulnerability, and inversely correlate with SMCs and fibrous cap thickness, while no correlation was found to the collagen content, macrophage, and EC activation, which is in contrast to human data (Maracle et al., 2018; Silvestre-Roig et al., 2019).

Neutrophil recruitment to the arterial wall has been shown to be driven by CCL5 in mice (Drechsler et al., 2010), which is induced in type-I IFN–stimulated macrophages and ECs (Table 1, Fig. 2, and Fig. 3; Goossens et al., 2010; Nakano et al., 2012). While the function of neutrophils in atherosclerosis is not completely clear (Drechsler et al., 2010), over recent years, emerging evidence has revealed the contributory role of neutrophils in atherogenesis and plaque instability (Teixeira and Tam, 2018). As a phagocytic granulocyte, neutrophils are capable of discharging granule proteins, proteases, and reactive oxygen species (ROS) and producing neutrophil extracellular traps (NETs; Döring et al., 2017). NETs consist of web-like chromatin with cytosolic and granule proteins that can trap pathogenic microbes and have been shown to be involved in many (chronic) inflammatory diseases (Papayannopoulos, 2018), including rheumatoid arthritis (RA), vasculitis, and SLE.

The presence of NET structures has been shown in atherosclerotic plaques and arterial thrombi in humans and mice, especially in cholesterol-rich or erosion-prone areas (Fig. 2; Franck et al., 2018; Quillard et al., 2015; Warnatsch et al., 2015). Consistently, cholesterol crystals can induce NET formation (NETosis) in vitro (Warnatsch et al., 2015). Studies proposing the mechanistic relation of NETs and atherogenesis show the ability to induce endothelial cytotoxicity, cause high-density lipoprotein (HDL) oxidative modifications, promote macrophage accumulation and cytokine release, activate T helper (Th) 17 cells, and trigger type-I IFN production from pDCs (Table 1,Fig. 2, and Fig. 3; Döring et al., 2012; Knight et al., 2014; Smith et al., 2014; Villanueva et al., 2011; Warnatsch et al., 2015). Of note, several studies suggest an imbalance between NETosis and NET degradation in autoimmune diseases, which may in part account for the increased CVD risk in these patients (Carmona-Rivera et al., 2015; Kessenbrock et al., 2009; Khandpur et al., 2013). While a recent study shows inhibiting NETosis via Pad4 deficiency inLdlr^−/− mice does not impede atherogenesis but prevents EC injury and thrombus formation (Franck et al., 2018), applying myeloid-specific Pad4-deficient model onApoe^−/− mice exerts reduced atherosclerosis burden (Liu et al., 2018). Similarly, inhibiting NETosis by pan-protein-arginine deiminase (PAD)–inhibitor chloramidine treatment before starting the high-fat diet feeding decreased atherosclerosis and mitigatedIfna expression inApoe^−/− mice, suggesting an initiating role of NETs in atherogenesis. Furthermore, the atheroprotective effects of chloramidine could not be reproduced inApoe^−/−Ifnar1^−/− mice, indicating a causal role of type-I IFN signaling in NET-driven atherogenesis (Knight et al., 2014).

Low-density granulocytes (LDGs), an immature, pro-inflammatory subtype of neutrophils identified in SLE individuals, display increased expression ofIFNA and are associated with IFN signature enrichment in these patients (Denny et al., 2010; Kegerreis et al., 2019). Further, LDGs are capable of inducing an IFN signature in ECs, thus disrupting their differentiation from the progenitors (Table 1 and Fig. 2; Denny et al., 2010). In addition to affecting endothelial maturation directly, LDGs are also capable of inducing mitochondrial ROS and NETs upon stimulation (Table 1; Kaplan, 2011; Lood et al., 2016; Villanueva et al., 2011), leading to increased plaque vulnerability. Several studies have shown that LDG NETs are the main inducer of type-I IFN secretion in pDCs via TLR9 signaling in SLE patients (Table 1; Garcia-Romo et al., 2011; Lande et al., 2011; Villanueva et al., 2011). Moreover, after adjustment for traditional risk factors, noncalcified plaque burden in SLE patients is directly associated with LDG levels (Carlucci et al., 2018). In addition to this particular neutrophil subset LDG, atherosclerotic mouse models applying myeloid-specific deletion of Ifnar1, which mainly affects neutrophils and monocytes/macrophages, revealed smaller lesions, reduced neutrophil and macrophage recruitment, and less necrosis (Fig. 3; Goossens et al., 2010). Of note, while the plasma CCL5 was increased in mice treated with IFN-β, these animals did not show a difference in neutrophil area in the plaque when compared with the untreated controls (Table 1 andFig. 3; Goossens et al., 2010). These data reinforce the concept of positive feedback between NET formation and type-I IFNs in neutrophils and other cell types during atherosclerosis progression.

Though the direct role of eosinophils in atherogenesis remains unclear, clinical studies reveal that absolute eosinophil count, intraplaque eosinophil markers, and eosinophil-specific chemokines are associated with increased risk of CVDs (Haley et al., 2000; Kitano et al., 2017; Lee et al., 2001;Niccoli et al., 2010; Tanaka et al., 2012). While EG2⁺ eosinophils are considered to be absent during the atherosclerosis development according to immunohistochemistry (IHC) staining of human plaques (van Dijk et al., 2016), the occurrence of eosinophil extracellular trap was identified in human thrombosed atherosclerotic specimen by double staining of eosinophil major basic protein and citrullinated histone H3 (Pertiwi et al., 2019), which might trigger IFN-α production in pDCs. Circulating eosinophils can affect atherogenesis through their cytokine production. In vitro data revealed that oxidized LDL (oxLDL) triggers the expression of type-I IFNs in cultured mouse eosinophils via CD36 and suppresses Il4 expression, and subsequently polarizes macrophages to a pro-inflammatory phenotype (Table 1 and Fig. 2; Qin et al., 2017).

pDCs are known to have the ability to rapidly produce massive amounts of type-I IFNs and are present in both murine and human atherosclerotic lesions as well as intact aorta (Chistiakov et al., 2014; Cole et al., 2018; Döring et al., 2012; Yun et al., 2016). Single-cell data revealed an increased amount of pDCs in aortas from Apoe^−/− mice fed a high-fat diet compared with mice fed normal chow (Cole et al., 2018). In human plaques, CD123⁺ pDCs localize closely to T cells and mDCs in the unstable shoulder region (Chistiakov et al., 2014; Niessner et al., 2006, 2007) and accumulate during atherogenesis (Fig. 2; Li et al., 2017).

The role of pDCs in atherosclerosis is still a subject of discussion. Both whole human plaque tissue and isolated pDCs respond to TLR9 agonists with increased IFN-α transcription and secretion, amplifying pro-inflammatory responses by mDCs and cytolytic T cells (Table 1 and Fig. 2; Niessner et al., 2006, 2007; Yun et al., 2016). Also, sorted mouse pDCs treated with oxLDL show an enhanced capacity to phagocytose and to stimulate antigen-specific T cell responses without altering IFN-α/β secretion (Döring et al., 2012). However, pDC-selective deletion in different atherosclerosis murine models showed conflicting results. Some studies found that antibody-mediated cell ablation or inducible pDC depletion in Ldlr^−/− mouse deteriorated atherogenesis, suggesting an atheroprotective character of pDCs. This observation was specifically linked to the pDCs’ ability of tuning T cell activation and proliferation in murine models (Table 1; Daissormont et al., 2011; Yun et al., 2016). On the contrary, studies applying similar techniques showed a more favorable phenotype in pDC-depletedApoe^−/− mice with reduced lesion size, more stable plaques, and a reduction in IFN-α serum levels, supporting the pro-atherogenic role for pDCs in murine atherosclerosis (Table 1; Döring et al., 2012; Macritchie et al., 2012).

Of note, without CpG stimulation, IFN-α protein levels in the intact aorta or atherosclerotic plaques in mice are low or undetectable (Macritchie et al., 2012; Yun et al., 2016). This implies that without interferogenic stimuli, the effect of IFN-α–mediated atherogenesis might be minor or localized to restricted regions in these murine model settings. On the other side, NETs have been demonstrated to be present in human plaques and may serve as stimuli triggering pDC type-I IFN production via TLR signaling, which subsequently primes neutrophils for NET formation (Fig. 2). This positive feedback between NETs and type-I IFNs produced by pDCs might be the causal factor exacerbating atherosclerosis in autoimmune diseases with elevated IFN signatures. Moreover, mice stimulated with type A CpGs showed an advanced atherosclerotic plaque phenotype in a pDC-dependent manner (Döring et al., 2012), implying a central role of pDCs in autoimmune-associated atherogenesis. Taken together, the responsiveness of pDCs to TLR7/TLR9 activation and type-I IFN induction corroborates the link between accelerated atherosclerosis and chronic inflammation, and thus serves as additional pathogenic mechanism for increased cardiovascular events in interferonopathies.

EC dysfunction is an important player in the initiation of atherosclerosis, where the permeation and accumulation of lipoprotein particles take place at the arterial lesion-prone areas (Davignon and Ganz, 2004; Gimbrone and García-Cardeña, 2016). During plaque progression, ECs activated by oxLDL produce chemokines that attract circulating monocytes into the arterial intima. Inflammatory factors expressed by recruited immune cells further increase plaque instability by modulating EC phenotype and extracellular matrix (Gimbrone and García-Cardeña, 2016).

Type-I IFN–treated HUVECs produce more adhesion molecules and chemokines, such as VCAM1 and MCP1 (Fig. 2), possibly through the suppression of NOS3 (encoding, endothelial nitric oxide synthase) expression and insulin-mediated NO production (Table 1, Fig. 2, and Fig. 3; Buie et al., 2017; Jia et al., 2018). Though the effect on ICAM1 is not clear, studies showed that more human neutrophils attach to the type-I IFN–treated HUVECs (Table 1 and Fig. 2; Buie et al., 2017; Giorelli et al., 2002; Jia et al., 2018; Kobayashi et al., 2008; Shen et al., 1997). However, some studies revealed conflicting results. In vitro IFNα2b exposure of human aortic ECs (HAEC) was shown not to alter the proliferation, tubule formation, or NO production while IFN-signature genes were induced (Table 1; Reynolds et al., 2014). Also, both IFN-α and IFN-β, but not IFN-γ, show a protective effect on growth factor deprivation-induced apoptosis of HAECs in vitro (Table 1; Sano et al., 2012).

In murine atherosclerosis models, IFN-β enhances leukocyte arrest via EC-macrophage CCL5-CCR5 interaction, thus promoting plaque destabilization (Fig. 2; Goossens et al., 2010). The reciprocal interaction between type-I IFNs and other inflammatory mediators exacerbates this process. TNF triggers IFN-β production via TNFR2-IRF1–dependent signaling in ECs (Fig. 1). The secreted IFN-β serves as an autocrine signal triggering HUVECs to generate chemotactic factors, such as VCAM1, that promote monocyte recruitment (Table 1, Fig. 1, andFig. 2; Venkatesh et al., 2013). IFN-α exposure was shown to disrupt the endothelium, and impair endothelial vasorelaxation and endothelial progenitor cell (EPC) function, resulting in accelerated thrombosis and platelet activation in lupus-prone and non–lupus-prone mice (Table 1 and Fig. 3; Diao et al., 2016; Thacker et al., 2012). In line with this, lupus-prone mice showed impaired function and increased type-I IFN signatures in the EPC compartments already without additional IFN-α application (Thacker et al., 2010, 2012). Corroborating the in vitro and animal data, type-I IFN signatures or high serum IFN-α are associated with impaired EPC and EC function, reduced flow-mediated dilation, and elevated endothelial activation markers, such as sVCAM1 and endothelial microparticles, in individuals with autoimmune diseases including SLE, antiphospholipid syndrome (APS), and RA (Table 1, Fig. 2, and Fig. 3; Denny et al., 2007;Somers et al., 2012; Grenn et al., 2017; Lee et al., 2007; Rodríguez-Carrio et al., 2014; Tydén et al., 2017). Many of the characteristics of EPC dysfunction in these patients could be overcome by neutralization of type-I IFNs or their receptor ex vivo (Table 1 and Fig. 3; Denny et al., 2007; Grenn et al., 2017). Comparable results were found in animal models. Ifnar1 deficiency in lupus-prone mice exhibit an improved EPC function and endothelial vasorelaxation whereasApoe^−/−Ifnar1^−/− mice fed with high-fat diet showed a more preferable atherosclerosis phenotype (Table 1 and Fig. 3; Thacker et al., 2012).

T lymphocytes are adaptive immune cells present in both human and murine atherosclerotic plaques (Witztum and Lichtman, 2014). During plaque development, accumulated innate immune cells activate their pro-inflammatory response, exerting deleterious effects. The majority of plaque-residing T cells are CD4⁺, whereas the remaining are mainly CD8⁺ (Jonasson et al., 1986). Among the CD4⁺ T cell subsets, the role of Th17 and Th2 cells in atherosclerosis are still debated, while Th1 and regulatory T (T reg) cells are considered to be pro- and anti-atherogenic, respectively. Details of different subsets and functions are reviewed elsewhere (Ketelhuth and Hansson, 2016; Lahoute et al., 2011). Of note, T cell subset differentiation is not irreversible, as some T cells can acquire different properties under certain (chronic) inflammatory conditions (Geginat et al., 2014; Sakaguchi et al., 2013).

Many animal studies have verified the T reg to Th1 cell plasticity driven by atherosclerotic and hypercholesterolemic environments, in which T reg cells lost the immunosuppressive characteristics and acquired a pro-inflammatory phenotype (Tabas and Lichtman, 2017). Interestingly, the instability of T reg cells can also be found in many autoimmune diseases, implying a possible shared pathophysiology driving both atherosclerosis and autoimmunity (Dominguez-Villar and Hafler, 2018). Plasma from SLE patients with type-I IFN activity impaired T reg cell activation and function in vitro (Golding et al., 2010). Moreover, limited T reg cell generation and increased IFN-γ⁺ T cell proliferation were observed in activated SLE PBMCs (Table 1;Golding et al., 2010). Consistent with this observation, in a murine colitis model,Trex1 deficiency induced type-I IFN expression and attenuated T reg cell activation and proliferation while promoting Foxp3^low/neg T cell expansion and pro-inflammatory function (Table 1; Srivastava et al., 2014a). IFN-α/β directly inhibits both human and murine T reg cell proliferation in vitro, while the murine Ifnar1-deficient counterpart is unaffected. This type-I IFN signaling-mediated inhibition of T reg cells can be observed in murine lymphocytic choriomeningitis virus infection and tumor models (Gangaplara et al., 2018;Srivastava et al., 2014b). However, studies also showed IFNAR signaling to be important for T reg cells exerting their immunomodulation function as IFNAR deficient T reg cells failed to dampen the effector T cell activation in a mouse model of inflammation (Metidji et al., 2015). These studies suggest that type-I IFNs can alter the pro- versus anti-inflammatory balance of T cell subsets.

Besides the effect of phenotype switching on T reg cells, IFN-α may directly modulate human CD4⁺ T cells’ cytotoxic function by increasing the expression of TNF-related apoptosis-inducing ligand (TRAIL) on the T cell surface (Table 1 and Fig. 2; Kayagaki et al., 1999; Niessner et al., 2006). Further, experiments applying human plaque-derived or peripheral blood-isolated CD4⁺ T cells to coronary SMC monolayers showed TRAIL-dependent apoptosis of SMCs, which could be amplified by IFN-α pretreatment of CD4⁺ T cells (Fig. 2 and Fig. 3). Of note, IFN-α alone did not induce SMC apoptosis in this study (Niessner et al., 2006).

As mentioned in a previous section, IFN-α–producing pDCs colocalize with T cells in human atherosclerotic plaques (Fig. 2; Niessner et al., 2006). The pro-atherogenic effect of type-I IFNs might be through paracrine signaling and exacerbate in autoimmune patients with elevated systemic IFN levels.

B cell subsets affect the atherosclerotic process through both cellular and humoral immunity functions. While whole blood gene expression profile implies an association between B cells and coronary heart diseases (Huan et al., 2013), the functional characteristics of B cell subtypes, the antibody targets, and different immunoglobulin classes in atherosclerosis are still elusive (Tsiantoulas et al., 2014). Two major subfamilies of B cells, B1 and B2 cells, process distinct machineries in plaque formation (Domeier et al., 2018; Tsiantoulas et al., 2015). In general, B1 cells are considered to be atheroprotective, as they can secrete natural IgM against oxidation-specific epitopes, thus limiting EC dysfunction and immune cell activation (Fig. 2; Tsiantoulas et al., 2014). B2 subsets, the majority of B cells in the body, show pro-atherogenic properties in some animal models, though the underlying mechanism remains under debate (Srikakulapu and McNamara, 2017;Tsiantoulas et al., 2015).

Atherosclerotic plaques contain apoptotic and necrotic cell debris, endogenous nucleic acids, and oxidized lipid species, which all can serve as a source of modified self-antigen and TLR ligands (Miller et al., 2011) triggering autoantibody production and adaptive immune responses. Murine atherosclerosis studies have also revealed an increase of anti–double-stranded DNA (dsDNA) antibody titers in high-fat diet–fedApoe^−/− mice (Table 1 and Fig. 2; Döring et al., 2012), again suggesting a shared pathophysiological continuum between autoimmunity and atherosclerosis. Detailed information on the role of B cells and immunoglobulins in atherogenesis is reviewed elsewhere (Sage et al., 2019; Tsiantoulas et al., 2015).

Besides the production of various antibodies, stimulated and early transitional stage B cells can secret type-I IFNs, which are important for B cell development and which might contribute to the pathogenesis of some autoimmune diseases (Table 1; Bénard et al., 2018; Green et al., 2009; Hamilton et al., 2017; Ward et al., 2016). This type-I IFN–IFNAR autocrine loop promotes autoreactive B cell development in the germinal center and strengthens follicular B cells’ ability to produce antibodies in a T cell–independent manner (Table 1; Domeier et al., 2018; Swanson et al., 2010). On the other hand, self-reactive anti-dsDNA IgE is associated with SLE disease activity and potentiates IFN-α production in human pDCs in vitro through promoting phagocytosis of DNA and TLR9 signaling (Table 1 and Fig. 2; Henault et al., 2016). As both IgE and pDCs are present in human atherosclerotic plaques (Wang et al., 2011), the aforementioned mechanisms might exacerbate atherogenesis and plaque instability, especially in individuals with systemically elevated type-I IFN levels or signaling signatures. Since B cells are capable of producing and responding to type-I IFNs, the complex nature of B cells in atherosclerosis and how type-I IFNs shape this process appear to be valuable subjects for further investigation.

While type-I IFN administration in murine models exacerbates atherosclerosis, the effects of type-I IFNs in humans, both direct and indirect to the plaque-residing cells, could result in clinical manifestation, especially in individuals with elevated IFN signatures. Therapeutic IFN treatment has been applied in cancer and chronic viral infection patients. In addition to the antivirus and antitumor effects, several clinical studies observed treatment-associated hypocholesterolemia, together with a decreased level of HDL, in these patients (Robertson and Ghazal, 2016). Further, the use of pegylated IFN and ribavirin as an anti–hepatitis C virus therapy is associated with risk factors of atherosclerosis such as insulin resistance (Brandman et al., 2012).

Compared with atheroma-free carotid wall samples, human atherosclerotic plaques show higher IFNA expression, and the level is even higher in inflamed and thrombosed lesions (Table 1 and Fig. 3; Niessner et al., 2006). Similarly, elevated IFN levels and/or its signatures are also observed in some autoimmune diseases, such as RA and SLE. The association between antiviral IFN-α therapy and the presence of autoantibodies indicated the casual role of type-I IFNs in systemic autoimmune pathogenesis (Fabris et al., 1992; Gota and Calabrese, 2003; Mavrogiannis et al., 2001; Mayet et al., 1989). Many clinical and epidemiological studies showed the concordance of accelerated atherosclerosis or increased risk of cardiovascular events in these interferonopathy-associated autoimmune diseases (Parker and Bruce, 2013; Pereira et al., 2009). A significant increase in aortic wall inflammation with mild to moderate disease activity is observed in SLE patients (Carlucci et al., 2018). In addition, type-I IFN–regulated proteins, such as IFITM1 and RPKRA, are increased in platelets isolated from SLE patients, and are associated with previous CV events including myocardial infarction and venous and arterial thrombosis (Table 1; Lood et al., 2010). Similarly, increased platelet activation was also observed in lupus-prone or normal mice receivingIfna-expressing virus and reduced inIfnar^−/− mice (Table 1 and Fig. 3; Thacker et al., 2012). Gene expression differential analysis of PBMC or blood leukocytes from SLE patients also revealed a strong enrichment of IFN signaling signatures (Baechler et al., 2003; Bennett et al., 2003; Carlucci et al., 2018).

Of note, type-I IFNs play distinct roles in different autoimmune diseases. The response to type-I IFNs and the signature gene expression are usually lower in mononuclear cells from untreated multiple sclerosis (MS) patients compared with heathy controls (Feng et al., 2002). Opposite SLE and RA, many symptoms of MS benefit from recombinant IFN-β as a disease-modifying treatment (Reder and Feng, 2013). MS patients receiving IFN-β usually show a decrease in total cholesterol and an increase in triglyceride plasma level (Fig. 3; Morra et al., 2004; Robertson and Ghazal, 2016; Sena et al., 2000; Uher et al., 2017), similar to the effect of type-I IFN therapies in individuals with cancers or viral infections. In line with this, IFN-β treatment is associated with increased CV risk factors including lower HDL level and higher diastolic blood pressure (Sternberg et al., 2014). However, the increased prevalence of cardiovascular events such as myocardial infarction in MS patients cannot be fully explained by traditional risk factors, and its correlation with disease-modifying treatments was not clear (Marrie et al., 2019). Notably, applying type-I IFNs in murine atherosclerosis models showed increased or unchanged plasma cholesterol or triglyceride levels (Goossens et al., 2010; Levy et al., 2003).

Similarly, traditional risk factors for cardiovascular events including age, gender, smoking, hypertension, and dyslipidemia have been correlated with but cannot fully explain the exacerbated atherogenesis in SLE and RA patients (Bruce et al., 2003; Castañeda et al., 2015; Esdaile et al., 2001). While the serum type-I IFN activity is independently associated with biomarkers of atherosclerosis development in lupus patients (Somers et al., 2012), some studies reported that preventive therapies for traditional risk factors, such as hypercholesterolemia and hypertension, failed to significantly reduce the incidence of CVDs in lupus patients (Petri et al., 2011; Schanberg et al., 2012; Tselios et al., 2016; Wigren et al., 2015). Notably, increased subclinical atherosclerosis prevalence is observed in individuals with primary (i.e., antineutrophil cytoplasmic antibody–associated vasculitis) or secondary (i.e., associated with SLE or RA) vasculitis (Argyropoulou et al., 2018; Chironi et al., 2007; Guillevin and Dörner, 2007). Whether vasculitis and atherosclerosis share common pathogenic mechanisms and whether IFNs play a causal role in this disease continuum warrant further investigation.

Although the role for type-I IFN in the pathogenesis and instability of atherosclerosis is supported by studies in both human and mouse models (Table 1; Goossens et al., 2010; Levy et al., 2003; Niessner et al., 2006), there is no statistically significant association between the risk of coronary artery events and IFN-α production–correlated single nucleotide polymorphisms (Nelson et al., 2015). Alternatively, mutations in some components of the IFN regulation and signaling, such as IRF8 and JAK2, are associated with the increased risk of cardiovascular events in different populations (Jaiswal et al., 2017; Leonard et al., 2013).

Immunomodulating therapies that dampen IFN signatures have been applied in many autoimmune diseases. Whereas the relation between immunosuppressive steroids and atherosclerosis remains controversial (Wu et al., 2016), hydroxychloroquine (HCQ), an alkalinizing lysosomatropic drug for SLE treatment, has been attested to be atheroprotective and is effective in dampening disease activity and mortality, reducing required steroids dosages, and preventing organ damage accrual and thrombovascular events (Fig. 3; Fasano et al., 2017; Ponticelli and Moroni, 2017; Yang et al., 2019). Mechanistically, the atheroprotective effects of HCQ may be due to interfering with IFN-α and TNF production by TLR7/TLR9-activated pDCs in SLE individuals (Table 1 and Fig. 3; Sacre et al., 2012). In addition to immune cell modulation and the prevention of EC dysfunction, HCQ offers beneficial effects on other traditional CV risk factors, such as dyslipidemia and diabetes (Floris et al., 2018), which are also shown to be associated with systemic type-I IFNs. Similarly, chloroquine, a closely related anti-malarial compound that may also be used in SLE treatment, is shown to be able to inhibit NETosis and the subsequent HDL oxidation in human neutrophils and LDGs from patients in vitro (Smith et al., 2014).

Inhibitors directly targeting downstream pathways of IFN signaling, such as JAK–STAT pathways and IRFs, have been proposed as a potential treatment strategy in CVD (Schwartz et al., 2017; Szelag et al., 2016). RA patients receiving tofacitinib, one of the first-generation JAK inhibitors, showed an increase in plasma total cholesterol, LDL, and HDL without changes in atherogenic index (Kang et al., 2018). In agreement with this observation, in vitro experiments and in vivo animal models revealed that tofacitinib up-regulated ABCA1 expression, promoted anti-inflammatory macrophage polarization, improved EC function, and attenuated atherosclerosis (Table 1 and Fig. 3; De Vries et al., 2019; Furumoto et al., 2017; Pérez-Baos et al., 2017; Wang et al., 2017). However, recently the Oral Surveillance study revealed a statistically significant and clinically relevant increase in pulmonary embolism and an increased mortality in patients older than 50 yr with an increased CV risk when treated with a tofacitinib dose of 10 mg twice daily (Fig. 3; FDA, 2019; Pfizer, 2019). In this respect, more selective JAK1 inhibitors may demonstrate a more favorable safety profile. Another anti-IFN approach is to inhibit type-I IFNs and IFNAR signaling directly. Blocking IFNAR1 in mice has been shown to be cardioprotective in MI (King et al., 2017) and could stimulate mouse arteriogenesis without affecting atherosclerosis burden after 4-wk treatment (Table 1 and Fig. 3; Teunissen et al., 2015). The potential beneficial effects of inhibiting type-I IFNs and their signaling have been applied to clinic trials. Anifrolumab, a fully human IgG1κ monoclonal antibody blocking IFNAR1 signaling, decreased disease activity, reduced NETosis, dampened chemokine serum level elevation, and improved cholesterol efflux capacity in SLE patients (Fig. 3; Casey et al., 2018; Furie et al., 2017; White et al., 2018). Further, sifalimumab, an anti–IFN-α monoclonal antibody, also yielded promising results in SLE patients (Khamashta et al., 2016), though the effect on atherosclerosis and cardiovascular outcome has not been investigated yet.

Given the positive outcomes of these anti-IFN and/or immunomodulating medicines dampening IFN signaling, it is of great interest to further investigate their effects on atherogenesis in autoimmune diseases, from premature, subclinical atherosclerosis to severe CV comorbidities.

Dyslipidemia and inflammation are central in the development of atherosclerosis. In addition to the great breakthrough of lipid-lowering therapy, there has been progress in understanding the role of inflammation that leads to plaque development and clinical complications. The causal role of chronic inflammation accelerating atherosclerosis in several systemic autoimmune diseases is well established. Studies applying elegant single-cell profiling approaches and functional studies have advanced our understanding on how type-I IFNs and the IFNAR signaling are regulated in different cell types involved in the pathogenesis, and helped to guide therapy development. As a pleiotropic cytokine modulating pro- and anti-inflammatory phenotypes in various cell types, the putative role of type-I IFNs in atherogenesis and autoimmunity has been verified, though still much remains to be investigated about the complex crosstalk of IFN signaling in the immune system and metabolism (Fig. 3). Future studies should continue to investigate the delicate homeostasis between IFNs’ antitumor and anti-pathogen effects and the pro-inflammatory and systemic lipid-altering effects.

We envision that type-I IFNs and their signaling pathways will be shared therapeutic targets in both atherosclerosis and rheumatic diseases. However, because type-I IFNs mediate important immune responses against infections and cancers, there is reason for caution for applying long-term anti-IFN therapy for CVD prevention. Advanced understanding of the genetic and epigenetic regulation and the biological underpinnings of IFN signaling in cardiovascular and autoimmune diseases might enable us to intervene in these diseases via fine-tuned modulation of type-I IFN signaling using targeted pharmacological technologies, which might serve as a promising and safe treatment strategy for atherosclerosis in patients with autoimmune diseases, and might be generalizable to the nonautoimmune population as well.

This work was supported by the Netherlands Heart Foundation (CVON 2011/B019, CVON 2017-20), Spark-Holding BV (2015B002), the European Union (Innovative Training Networks grant EPIMAC SEP-210163258), Leducq Foundation (Transatlantic Network Grant, 16CVD01), REPROGRAM (EU Horizon 2020, 667837), and the Amsterdam University Medical Centers fellowship.

The authors declare no competing financial interests.

Author contributions: H.J. Chen and M.P.J. de Winther decided on the topics included in the manuscript. H.J. Chen conceptualized and wrote the manuscript and prepared the figures and the table. S.W. Tas and M.P.J. de Winther conceptualized the article and edited the manuscript. All authors contributed to manuscript editing, and read and approved the final version.