Breaking tolerance: the autoimmune aspect of atherosclerosis

Abstract

Atherosclerotic cardiovascular disease (ASCVD) is a chronic inflammatory disease of the arterial walls and is characterized by the accumulation of lipoproteins that are insufficiently cleared by phagocytes. Following the initiation of atherosclerosis, the pathological progression is accelerated by engagement of the adaptive immune system. Atherosclerosis triggers the breakdown of tolerance to self-components. This loss of tolerance is reflected in defective expression of immune checkpoint molecules, dysfunctional antigen presentation, and aberrations in T cell populations — most notably in regulatory T (T_reg) cells — and in the production of autoantibodies. The breakdown of tolerance to self-proteins that is observed in ASCVD may be linked to the conversion of T_reg cells to ‘exT_reg’ cells because many T_reg cells in ASCVD express T cell receptors that are specific for self-epitopes. Alternatively, or in addition, breakdown of tolerance may trigger the activation of naive T cells, resulting in the clonal expansion of T cell populations with pro-inflammatory and cytotoxic effector phenotypes. In this Perspective, we review the evidence that atherosclerosis is associated with a breakdown of tolerance to self-antigens, discuss possible immunological mechanisms and identify knowledge gaps to map out future research. Rational approaches aimed at re-establishing immune tolerance may become game changers in treating ASCVD and in preventing its downstream sequelae, which include heart attacks and strokes.

Introduction

Atherosclerosis is an inflammatory disease affecting the arterial walls of large and medium-sized arteries and is characterized by the accumulation of low-density oxidized lipoproteins and immune cell-rich plaques¹. Rupture or erosion of these plaques causes catastrophic thrombosis, which triggers immediate ischaemia in the heart, brain or extremities, subsequently culminating in tissue infarction¹. According to the latest report of the American Heart Association, the global age-standardized mortality rate of cardiovascular disease (CVD) is 240 per 100,000 (20 million cases per year) whereas morbidity rate is 7,354 per 100,000 (608 million cases per year)². The incidence of atherosclerotic cardiovascular disease (ASCVD) can be reduced by regulating low-density lipoprotein (LDL) cholesterol through statins and PCSK9 (proprotein convertase subtilisin/kexin type 9) inhibitors³. Despite the success of statins and PCSK9 inhibitors, there is a significant residual inflammatory risk even when LDL cholesterol is treated to target⁴; the risk of major adverse cardiovascular events (MACEs) remains at 47% to 53% (refs. 4,5). Anti-inflammatory therapies such as the IL-1β inhibitor canakinumab have shown promise in reducing MACEs⁶. However, such therapies impair host defence mechanisms, resulting in other complications including lethal infections⁶. Other anti-inflammatory drugs, such as colchicine, have also shown promise in lowering MACEs in patients with stable coronary disease^7,8.

Although it has been known for more than 20 years that patients with atherosclerosis have circulating autoantibodies against apolipoprotein B (ApoB)-containing lipoproteins, including LDL^9,10, it was only recently that this observation was linked with a potential systemic loss of tolerance to self-components in atherosclerosis^11,12. In mouse models of atherosclerosis, self-reactive CD4⁺ T cells⁴ enable cytotoxic CD8⁺ T cell-mediated killing of host cells¹³. Other self-reactive CD4⁺ T cells enter germinal centres, which promotes B cell affinity maturation, isotype switching and the development of plasma cells secreting high-affinity antibodies⁹. In some studies, high levels of IgG antibodies to LDL predict poor outcomes of ASCVD¹⁰, suggesting that the autoimmune response in atherosclerosis is maladaptive. However, the correlation of IgG and IgM levels to oxidized LDL and to ApoB peptides varies among studies. In most studies, no significant correlation was shown between the presence of IgG antibodies specific for oxidized low-density lipoprotein (oxLDL) and the occurrence of ASCVD. One study¹⁴ has actually reported that oxLDL-specific IgG antibody titres negatively correlated with Gensini scores (used to evaluate the severity of coronary artery disease) in patients undergoing coronary angiography. However, in another study¹⁵, oxLDL-specific IgG titres positively correlated with coronary artery stenosis (that is, narrowing of the blood vessels) in patients with myocardial infarction. In all studies reporting IgM against oxLDL, the levels of these antibodies were negatively correlated with disease severity in patients^16–19, which supports findings made in mouse models of disease²⁰. Therefore, there is data suggesting that atherosclerosis is associated with a loss of tolerance, but how this contributes to ASCVD outcomes is still unclear. In this Perspective, we review the emerging evidence that atherosclerosis is associated with a break of tolerance to self-antigens, discuss possible immunological mechanisms and identify knowledge gaps to map out future research.

Antigen presentation in atherosclerosis

Self-antigens recognized by T cells

In atherosclerosis lesions, LDL particles entrapped within the sub-endothelial space undergo oxidation. This process leads to the formation of malondialdehyde and 4-hydroxynonenal. This oxidative transformation correlates with significant structural alterations in LDL, involving fragmentation of ApoB and the creation of diverse aldehyde and phospholipid adducts bound to ApoB-derived peptides. Modified LDL can be taken up by scavenger receptors such as CD36, but CD4⁺ T cell responses to native ApoB sequences have also been demonstrated^21,22. It is improbable that ApoB is the only relevant autoantigen; for example, heat shock proteins have also been proposed to be autoantigens in atherosclerosis²³. In atherosclerosis, ApoB-specific CD4⁺ T cells show clonal expansion²⁴. In a single-cell RNA sequencing (scRNA-seq) study with T cell receptor (TCR) sequencing, the ApoB-specific CD4⁺ T cells were identified by using tetramers to an ApoB epitope. Uniform Manifold Approximation and Projection for Dimension Reduction (UMAP) showed that the transcriptomes of most ApoB-specific cells were more similar to those of memory T cells than regulatory T cells²⁴. Other atherosclerosis-associated autoantigens include oxLDL, malondialdehyde-modified LDL, β₂-glycoprotein I and heat shock protein 60 (HSP60)^21,23,25. However, the epitopes of these autoantigens that are targeted by T cells remain to be identified. It is also probable that more atherosclerosis-associated autoantigens remain to be discovered.

Antigen presentation

Atherosclerosis in humans and model organisms is always accompanied by autoantibody production. These atherosclerosis-associated autoantibodies are commonly measured by reactivity to oxLDL⁹, which contains many epitopes. Isotype-switched IgG antibodies specific for oxLDL require the assistance of T cells, specifically CD4⁺ follicular helper T (T_FH) cells, which recognize antigenic epitopes displayed by MHC class II molecules. Circumstantial evidence suggests that antigen presentation occurs in the lymph nodes draining arteries affected by atherosclerosis (Fig. 1). In mice, the axillar and cervical lymph nodes draining the aortic and carotid arches become very large in response to atherosclerosis, suggesting fulminant antigen presentation¹. Later, antigen-experienced T cells can be re-activated in a recall response in the atherosclerotic plaque proper²⁶. The arterial adventitia is endowed with lymphatics, which have been shown to be involved in reverse cholesterol transport and provide a plausible pathway for the migration of dendritic cells (DCs) from atherosclerotic plaques to draining lymph nodes²⁷. However, DCs carrying atherosclerosis-associated antigens to draining lymph nodes have not been shown experimentally. It should be noted that it is technically very challenging to trace these DCs. First, the vascular lymphatics are only incompletely described²⁷. From what we know, the afferent lymphatics are very small and, therefore, hard to cannulate. Second, the kinetics of DC recruitment from the artery wall are not known.

Fig. 1 |. Antigen presentation in atherosclerosis.

The adventitia of large arteries has afferent lymphatics that drain into local lymph nodes, but in the setting of atherosclerosis, how immune cells can migrate from the plaque to the adventitia is unknown. Antigen-presenting cells (APCs) may carry antigen from the artery to the draining lymph node, wherein they enter germinal centres (indicated by dashed lines) that are populated by B cells (blue) and some T cells (red). High endothelial venules (HEVs) are the gateways enabling entry of naive T cells from the blood into lymph nodes. In atherosclerosis, plaque forms a neointima (yellow) and low-density lipoprotein (LDL) accumulates and becomes oxidized. In the draining lymph node, some T cells undergo clonal expansion in response to arterial antigens including ApoB. Some T cells leave via efferent lymphatics and traverse the thoracic duct (dark blue-shaded area) to enter the blood circulation. Clonally expanded T cells home back to the plaque (arrows), some via CC-chemokine receptor 5 (CCR5)-mediated trafficking. Over time, arterial tertiary lymphatic organs form in the adventitia, wherein APCs encounter naive and antigen-experienced B cells and T cells. T cell clonal expansion is more extensive in arterial tertiary lymphoid organs (ATLOs) than in lymph nodes, suggesting that ATLOs may provide a ‘shortcut’ that accelerates autoimmunity in atherosclerosis. Some germinal centre B cells become plasma cells and secrete antibodies to atherosclerosis-associated antigens including oxidized LDL (not shown).

Failure of self-tolerance in atherosclerosis

Normally, autoimmune T cell responses are kept in check owing to central and peripheral tolerance mechanisms (see Boxes 1 and 2). Failure of peripheral T cell tolerance in atherosclerosis was first formally shown in studies by Wang et al.¹¹ and Depuydt et al.¹², both published in 2023. The main findings of these studies were that several factors were associated with this failure of peripheral T cell tolerance: (1) defective immune checkpoint expression, (2) clonal expansion of CD4⁺ T cell, CD8⁺ T cell and regulatory T (T_reg) cell populations, (3) T cell exhaustion, (4) the probable conversion of T_reg cells to T helper 17 (T_H17) cells, and (5) dysfunctional antigen presentation^11,12. Both of these recent studies have used scRNA-seq with TCR-seq, in which TCRα and TCRβ sequences are matched with transcriptomes in the same cells. In a published mouse study¹¹, the highest degree of plaque-specific clonal T cell expansion was seen in effector CD8⁺ T cells, and these clonally expanded T cells expressed genes such as Cd69, Fos and Fosb, indicative of recent TCR engagement¹¹. Similar observations were made in a very recent study of human CD8⁺ T cells in subjects with coronary artery disease²⁸. In the mouse study¹¹, T cell transcriptomes with paired TCRα and TCRβ sequences were assessed by scRNA-seq in aorta-draining renal lymph nodes (RLNs) of healthy or atherosclerotic mice, arterial tertiary lymphoid organs (ATLOs) and atherosclerotic plaques¹¹. scRNA-Seq with TCR-Seq also revealed transcriptomes of ApoB-tetramer-specific CD4⁺ T cells in humans²⁴. There were progressive changes from wild-type RLNs to Apoe^−/− RLNs, ATLOs and atherosclerotic plaques, reflected in a notable reduction in the proportions of T_reg cells and naive CD4⁺ T cells, coupled with a significant increase in the proportions of effector memory (T_EM) and central memory (T_CM) CD8⁺ T cells. The majority of TCRα and TCRβ CDR3 sequences observed in CD8⁺ T cells within atherosclerotic plaques shared patterns with those seen in CD8⁺ T cell populations expanded in the aorta-draining RLNs of Apoe^−/− mice, but not in the RLNs of wild-type mice. Single-cell TCRα and TCRβ CDR3 sequences showed clonal expansion, which progressively increased from wild-type RLNs to Apoe^−/− RLNs to ATLOs and finally to plaques.

Box 1. T cell development and central tolerance.

T cell development in the thymus

T cells develop from bone marrow-derived progenitors that home to the thymus and undergo a complex and irreversible differentiation process⁶⁷. Following T cell lineage commitment and population expansion, T cell receptor (TCR) gene rearrangement occurs, leading to the development of either γδ or αβ T cell progenitors. At this stage, all T cell precursors are double-negative (DN) for the expression of CD4 and CD8 (ref. 67). The somatic recombination of TCR V, D and J genes forms a highly diverse array of unique TCRs with stochastic specificities. The fate of αβ T cell progenitors is determined by the affinity of their TCR for self-peptides presented by MHC class I or MHC class II molecules⁶⁷. Most αβ TCR⁺ DN thymocytes fail positive selection and die by apoptosis. Other αβ TCR⁺ DN thymocytes pass positive selection⁶⁷, proliferate and generate a large population of CD4⁺CD8⁺ double-positive (DP) T cells⁶⁷. During this maturation, many T cells are eliminated by negative selection. Negative selection eliminates cells whose TCR has high affinity for self-peptide–MHC complexes. Cells with lower-affinity TCRs pass negative selection, which is the mechanism for central tolerance^67,68.

Central tolerance

One of the most important mechanisms of central tolerance in the thymus is the negative selection, also called clonal deletion, of immature T cells that bear high-affinity TCRs against self-antigens⁶⁹. However, negative selection is incomplete^70–75. A second layer of central tolerance, known as clonal diversion, ensures that self-reactive T cells that survive past negative selection are imprinted with a regulatory phenotype^64,65. These thymic-derived regulatory T (tT_reg) cells suppress T cell activation and effector function in the periphery⁷⁶. tT_reg cells are thought to express TCRs with intermediate affinity for self-epitopes. Both negative selection and clonal diversion depend upon the recognition of self-peptide–MHC complexes displayed on thymic antigen-presenting cells, particularly the medullary thymic epithelial cells that specialize in the ectopic expression of many antigens in the thymus^77,78. Insufficient expression or presentation of self-antigens in the thymus and engagement of TCRs with low affinity and avidity allow the developing T cells to escape central tolerance.

Box 2. Mechanisms of peripheral tolerance.

Mature autoreactive T cells that escape thymic negative selection⁷⁹ are restrained by peripheral tolerance mechanisms^37,67, namely, ignorance⁸⁰, anergy⁸¹, expression of inhibitory molecules⁸² and active suppression by regulatory T (T_reg) cells⁸³.

Ignorance

When self-antigens are exclusively expressed in immune-privileged sites, such as the brain or the eye, they are effectively hidden from circulating immune cells by anatomical barriers^72,80,84. When the levels of expression or presentation of these self-antigens in peripheral tissue and lymphoid organs are too low to trigger T cell activation, autoreactive T cells against these self-antigens are maintained in a quiescent state. Hidden and low-expressed self-antigens are ignored by the immune system^72,80,84. However, antigen-inexperienced naive T cells against such self-antigens can be activated when sufficient quantities of cognate antigen are presented to them by activated antigen-presenting cells (APCs). This can happen under conditions of tissue damage and chronic inflammation^85–87, which is relevant to atherosclerosis.

Anergy

T cell anergy occurs when naive T cells encounter self-antigens in the absence of proper co-stimulation⁸⁸. They enter a state of hypo-responsiveness characterized by low IL-2 production and poor proliferation⁸⁹. This anergic state results from improper regulation of T cell receptor (TCR) proximal signalling events (for example, excessive calcium–NFAT signalling and defective mTOR and RAS–MAPK activation) and involves active gene silencing, for example, at the IL-2 locus, as well as transcriptional upregulation of anergy-specific genes (for example, GRAIL and Cbl-b)⁸¹. T cell anergy can be reversed by cytokines such as IL-2 and IL-15 that induce proliferation and cell survival^86,90.

Inhibitory molecules

Activation-dependent upregulation of co-inhibitory checkpoint molecules such as CTLA4 and PD1 help to control excessive autoreactive T cell responses⁹¹. CTLA4 predominantly controls naive T cell activation in lymphoid organs^92,93, whereas PD1 primarily exerts its regulatory effect during recall responses in peripheral tissues⁹⁴. CTLA4 interacts with CD80 and CD86 on APCs and blocks co-stimulation from CD28 (ref. 95). CTLA4 inhibits T cell activation by triggering T cell anergy⁹⁶. PD1 ligands PDL1 and PDL2 are constitutively expressed on APCs and are induced in non-immune cells in inflamed tissues⁹⁷. PD1 limits ongoing effector T cell activation and proliferation by inducing a dysfunctional state known as T cell exhaustion⁹⁸. Blocking PD1 or CTLA4 can overcome this, a mechanism used to enhance cancer immunotherapy. In the context of atherosclerosis, such as in autoimmune diseases, PD1 and CTLA4 are probably beneficial and dampen disease progression^99,100.

Regulatory T cells

The principal mechanism of peripheral tolerance is based on suppression of effector T cell responses by Treg cells. Treg cells express the transcription factor FOXP3 and the high-affinity IL-2 receptor CD25 but do not express the IL-7 receptor CD127 (refs. 101,102). They exert their suppressive effects through various mechanisms, including the secretion of immunosuppressive cytokines such as IL-10 and transforming growth factor-β (TGFβ), direct cell–cell contact inhibition, and modulation of APCs^103,104. These mechanisms collectively work to maintain immune tolerance and prevent autoimmunity. Treg cells exhibit considerable heterogeneity and have tissue-specific, characteristic transcriptomes^105,106. Most T_reg cells originate in the thymus (tT_reg cells) during T cell development¹⁰⁷. Peripherally induced regulatory T cells (pT_reg cells) develop from conventional CD4⁺ T cells in the periphery in response to cytokine signals, specifically TGFβ and retinoic acid^103,108. Both tT_reg cells and pT_reg cells display a broad range of antigen specificities including many self-antigens.

T_reg cells are not naive. Their cell surface phenotype and transcriptome suggest that they repeatedly engage their TCRs. At steady state, T_reg cells proliferate in vivo^29,109. T_reg cells increase in number during inflammation, a mechanism that avoids excessive inflammation associated with disease. However, T_regcells are not stable. In the presence of repeated TCR engagement and inflammatory stimuli, T_reg cells become ‘exT_reg’ cells⁴⁰. Such exT_reg cells can express either T-bet and acquire a T helper 1 (T_H1)-like transcriptome^110,111, express RORγt and acquire a T_H17-like transcriptome⁴⁰, express GATA3 and acquire a T_H2-like transcriptome^112,113, express BCL-6 and acquire a follicular helper T (T_FH) cell-like phenotype³², or express CD16 and CD56 on the cell surface and become cytotoxic exT_reg cells²⁹.

Several defects in the expression of immune checkpoint molecules and other immune receptors were previously reported¹¹, but none have been validated so far. There was indirect evidence for T_reg cells in ATLOs acquiring a T_H17-like phenotype, a finding that needs to be confirmed by T_reg cell lineage tracking¹¹. Breakdown of tolerance was also seen in scRNA-Seq data from human coronary and carotid artery plaques¹¹, which was associated with a strong cytotoxic gene signature including CST1, GZMB, GZMK, GZMM, NKG7 and PRF1 (ref. 12).

In a recent human carotid plaque study¹², clonal expansion of effector CD4⁺ and CD8⁺ T cells was reported in the plaque, with evidence for recent TCR engagement supported by expression of CD69, FOS and FOSB (ref. 12). Of particular interest was the expansion of TEMRA (terminally differentiated effector memory cells re-expressing CD45RA) clusters, which are a type of T cells that express CD45RA and are thought to be terminally differentiated. These cells expressed GZMB, PRF1 and NKG7 and lacked CD27 and CD28 (ref. 12). Another TEMRA cluster expressed KLRD1, KLRG1 and FCGR3A (ref. 12), the gene encoding CD16. We recently identified this signature as indicative of terminally differentiated exT_reg cells²⁹. The percentage of large (1 to 10%) and medium (0.1 to 1%) clones was higher in plaques than in peripheral blood mononuclear cells (PBMCs). One CD8⁺ T cell clone that was very highly expanded in PBMCs (>10%) had a TCRα sequence known to be specific for cytomegalovirus and showed no evidence of recent TCR engagement¹². On the basis on these findings, the authors proposed atherosclerosis as an ‘autoimmune-like disease’¹². In 2018, we suggested that atherosclerosis is “a chronic inflammatory disease with an autoimmune component”⁴. The idea that atherosclerosis is an autoimmune disease has previously been discussed²³. Previous evidence was limited to the presence of autoantibodies to (modified) LDL⁹. A recent study has provided evidence of clonal T cell expansion in plaque and ATLOs¹¹.

Mechanisms for tolerance breakdown in atherosclerosis

The mechanisms involved in breaking tolerance to self-antigens in the setting of atherosclerosis are unknown. However, potential candidate mechanisms include T_reg cell instability and dysfunction or the activation of self-reactive naive T cells.

T_reg cell instability in atherosclerosis

T_reg cell instability has been observed in the context of atherosclerosis, whereby T_reg cells convert to pro-inflammatory exT_reg cell phenotypes^21,24. Such exT_reg cells have been described in many mouse studies of atherosclerosis^30,31. Three early studies have provided evidence that T_reg cells acquire a T_H1-like gene expression signature^30,31 that includes expression of the chemokine receptor CCR5, a receptor that was shown to promote CD4⁺ T cell homing to atherosclerotic arteries in vivo³¹. In another mouse atherosclerosis study, T_reg cells were shown to switch to a T_FH cell-like phenotype³². In a study of patients with atherosclerosis, ApoB-specific T cells were shown to express FOXP3 and RORγt or FOXP3 and T-bet, assessed at the protein level by flow cytometry³³.

In a recent study, we mapped exT_reg cell transcriptomes from the Apoe^−/− mouse model of atherosclerosis — obtained from sorting CD4⁺ T cells from T_reg lineage tracker mice onto human scRNA-seq data — and showed that human exT_reg cells express CD16, CD56 and an array of cytotoxic genes²⁹. These exT_reg cells cannot regulate effector T cell proliferation. Instead, they kill target cells with remarkable efficiency, comparable to killing mediated by CD8⁺ cytotoxic T lymphocytes (CTLs)²⁹. Thus, it appears that exT_reg cells in atherosclerosis exist in multiple versions. Some still express FOXP3 and acquire additional T_H1-like, T_H17-like or T_FH-like gene signatures³⁴. The near-complete loss of FOXP3 in exT_reg cells reported previously²⁹ suggests that these cells may be the most terminally differentiated exT_reg cells. The breakdown of tolerance to self may be related to the conversion of T_reg cells to exT_reg cells because many T_reg cells express TCRs specific for self-epitopes (Fig. 2).

Fig. 2 |. Conversion of regulatory T cells in atherosclerosis.

FOXP3⁺CD25⁺CD127⁻ regulatory T (T_reg) cells (blue) can convert into at least four types of ‘exT_reg’ cells in atherosclerosis. In studies of the Apoe^−/− mouse model, T helper 1 (T_H1)-likeT_reg cells (yellow) have been described with a FOXP3^lowCD25⁻ T-bet⁺ CCR5⁺ phenotype^30,31. Another study has shown that T_reg cells become increasingly T_H17-like (orange) in response to a western diet²¹. A similar T cell phenotype (FOXP3⁺RORγt⁺ T_H17-like) was shown by flow cytometry in patients with atherosclerosis³³. In Apoe^−/− mice, some exT_reg cells can acquire a T_FH-like phenotype (pink) and express CXCR5, PD1, BCL-6 and ICOS³². In humans and mice with atherosclerosis, terminally differentiated cytotoxic exT_reg cells (red) express very low levels of FOXP3 and no CD25, but they acquire expression of CCL3, CCL4 and CCL5, TBX21, and a cytotoxic signature including NKG7 (ref. 29). The mechanisms that drive the conversion of T_reg cells to exT_reg cells are still unknown.

The mechanisms leading to T_reg cell instability in the context of atherosclerosis remain to be defined, but previous studies have shown that chronic antigen exposure, pro-inflammatory signalling and metabolic reprogramming can destabilize T_reg cells (see Box 3).

Box 3. What drives regulatory T cell instability?

Several lines of evidence suggest that chronic antigen exposure, pro-inflammatory signalling and metabolic reprogramming at inflamed tissue sites work in concert to cause regulatory (T_reg)cell instability^47,114,115.

Dysregulation of the FOXP3 locus

A stable Treg cell programme is maintained both by Foxp3 expression and regulation of its target genes, as well as by preserving a hypomethylated or demethylated epigenetic landscape at specific chromatin sites¹¹⁶. T_reg cell-specific deletion of TET2 and TET3 proteins that normally help to maintain demethylated chromatin states in the Foxp3 upstream enhancer region triggers loss of Foxp3 expression and T_reg cell instability^117,118. Treg cells activated in the absence of the chromatin-modifying enzyme EZH2 were unable to resolve inflammation at tissue sites, thereby leading to multi-organ autoimmunity¹¹⁹.

Prolonged or unregulated TCR signalling

Human and mouse Treg cells proliferate in vivo²⁹. T_reg cell maintenance in the periphery requires a basal level of continuous activation through TCR-mediated, IL-2R-mediated and CD28-mediated co-stimulation that induces a T_reg cell-specific suppressive gene programme and a permissive epigenetic landscape¹²⁰. However, prolonged engagement of TCR–CD28 signalling complexes^121–123 can cause hyperactivation of the PI3K–AKT–mTOR pathways¹²². This can trigger chromatin modifications at the Foxp3 locus^110,122. T_reg cells that are deficient in NRP1 (ref. 124) or PTEN¹²⁵ are more susceptible to pathogenic conversion. NRP1 deficiency makes T_reg cells lose their suppressive capacity, although FOXP3 expression remains intact⁶⁰. These cells express T_H1 cell-associated markers such as IFNγ, T-bet, CXCR3 and Eomes⁶⁰. PTEN deletion in T_reg cells also drives instability and loss of function via CD25 and FOXP3 downregulation and increased production of IFNγ, resulting in systemic lupus-like autoimmunity¹²⁶. Downregulation of the transcription factor Eos reprogrammes T_reg cells into FOXP3⁺ helper-like T cells, called Eos-labile T_reg cells¹²⁷.

Toll-like receptor signalling

Tol-like receptor (TLR) signalling can promote autoimmune conditions⁸⁵ and TLR engagement has been shown to block the suppressive action of CD4⁺CD25⁺ T_reg cells¹²⁸. TCR engagement and TLR signalling can mediate T_reg cell instability by activating common downstream pathways. One example is autoantigen-mediated TCR stimulation combined with TLR activation by LPS, which can destabilize T_reg cells by strongly activating the MEK–ERK pathway¹²⁹. Increased glucose uptake and glycolytic activity impairs the FOXP3-dependent suppressive programme of T_reg cells¹³⁰.

Pro-inflammatory cytokine signalling

Pro-inflammatory cytokines may also contribute to destabilizing T_reg cells. In the inflamed synovial joints of patients with rheumatoid arthritis, tumour necrosis factor (TNF) upregulated the expression and activity of protein phosphatase 1 (PP1), which specifically dephosphorylated the Ser418 residue on FOXP3 in T_reg cells¹³¹. This compromised the transcriptional activity of FOXP3 and converted the T_reg cell-suppressive programme into an effector programme, characterized by IL-17 and IFNγ expression¹³¹. FOXP3 phosphorylation and T_reg cell suppressive function were restored by treating patients with anti-TNF therapy (infliximab)¹³¹. IL-12 was shown to promote T_H1-like T_reg cells^44,45 (Fig. 2). Mechanistically, this was mediated by enhanced phosphorylation of AKT and its downstream targets FOXO1 and FOXO3, which triggered IFNγ production and loss of suppressive ability¹³². IL-12, together with a T_reg cell-intrinsic hyperactive Notch signalling pathway, led to STAT4 phosphorylation and IFNγ production¹³³. IL-1β can also reprogramme T_reg cells into IL-17-producing effector cells^134,135, and treatment of mouse and human T_reg cells with IFNγ resulted in a loss of their suppressor programme⁶⁰. T_reg cells from inflamed tissues, such as the tumour microenvironment, were shown to be more susceptible to loss of suppressive functions owing to their increased expression of IFNGR1 (ref. 60), which is directly relevant to T_reg cell instability induced by cytokine signalling.

TGFβ, retinoic acid and IDO protect T_reg cell stability

The destabilizing effects of prolonged TCR activation and pro-inflammatory cytokines can be counterbalanced by T_reg cell-promoting factors such as TGFβ¹²¹ and all-trans acid¹³⁶. Indoleamine-pyrrole 2,3-dioxygenase (IDO), produced by a tolerogenic subset of activated plasmacytoid DCs, can also block T_reg cell plasticity¹²⁷. In IDO-deficient mice, ligation of TLR9 by CpG, particularly on plasmacytoid DCs, triggered IL-6 production that caused reprogramming of T_reg cells into T_H17-like exT_reg cells¹³⁷. Immunization of mice in the presence of IDO blockade converted T_reg cells into IL-17-producing and IL-22-producing non-suppressive effector T cells¹³⁸. Most of them also co-expressed IL-2, CD40L and TNF^127,138,139.

Regulatory T cell dysfunction

Although T_reg cells are known to be stable under homoeostatic conditions³⁵, several recent reports have shown T_reg cell dysfunction³⁶ and instability in many autoimmune diseases^37,38. Fate-mapping experiments in mouse models have confirmed that T_reg cells downregulate Foxp3 during severe lymphopenia³⁹, in experimentally induced autoimmune encephalomyelitis⁴⁰, in collagen-induced autoimmune arthritis⁴¹, in type 1 diabetes⁴² and in atherosclerosis²⁹. These exT_reg cells produce effector cytokines such as IL-2 (ref. 39), IFNγ (refs. 40,42) and IL-17 (ref. 41), and when adoptively transferred to naive mice, they trigger tissue damage and promote disease^39–42. IL-17A⁺ FOXP3⁺ human T_reg cells are present in the skin lesions of patients with psoriasis⁴³ and in the synovium of patients with rheumatoid arthritis⁴¹. Significantly higher frequencies of IFNγ⁺ FOXP3⁺ T_reg cells were also reported in patients with relapsing remitting multiple sclerosis⁴⁴ or with type 1 diabetes⁴⁵ than in healthy controls. Chronic tissue inflammation and antigen-driven T cell activation, which are hallmarks of these diseases⁴⁶, rewire T_reg cells and drive the conversion of their suppressive programme to a pathogenic one^47,48. In summary, T_reg cells appear to become dysfunctional in almost all autoimmune disease models that have been studied.

In mouse models of atherosclerosis, fate-mapping studies demonstrated the accumulation of T-bet⁺ T_H1-like^30,31 and BCL-6⁺ T_FH-like³² exT_reg cells³⁴, particularly in the inflamed aortic wall and in organ-draining lymph nodes. Transcriptomic analysis of T_reg and exT_reg cells from the spleen and lymph nodes of 20-week-old atherosclerotic mice revealed an upregulation of cytotoxic molecules, such as granzymes and perforin, in the exT_reg cells, and similar results were observed in exT_reg cells from patients with atherosclerosis²⁹. In all these studies^29–32, exT_reg cells were reported to have compromised suppressive capacity. Instead, they acquired effector programmes such as cytotoxicity²⁹ and secretion of inflammatory cytokines such as TNF and IFNγ^24,29–31. Consistent with previous observations that downregulation of CD25 has an important role in T_reg cell instability⁴⁹, exT_reg cells in atherosclerosis displayed low levels of CD25 expression^29–32. In most cases, this was accompanied by a concomitant loss in Foxp3 expression^29,31,32. Both human²⁹ and mouse^30,31 exT_reg cells expressed pro-inflammatory chemokine receptors such as CXCR3 and CCR5. Adoptive transfer of the reprogrammed T_reg cells promoted lesion formation and systemic IFNγ production in atherosclerotic mice^24,31, suggesting that pathogenic conversion of T_reg cells to exT_reg cells in atherosclerosis may have a causal association with disease progression.

Antigen-mediated activation of T cells has been suggested to be a driver of T_reg cell instability in mouse models of multiple sclerosis⁴⁰ and autoimmune arthritis⁴¹. TCR activation seems to be necessary but not sufficient for T_reg cells to covert to exT_reg cells. An inflammatory microenvironment also has a pivotal role as >90% of adoptively transferred T_reg cells remained stable in the absence of an inflammatory setting in a mouse model of graft-versus-host disease⁵⁰. Therefore, it is probable that both TCR-mediated antigen-specific T cell activation and a local inflammatory milieu promote autoreactivity in atherosclerosis⁵¹.

RNA sequencing of mouse CD4⁺ T cells specific for the atherosclerosis-associated autoantigen ApoB revealed lower expression of T_reg cell-associated genes — such as Foxp3, Il2ra, Ctla4 and Il10 — in ApoB-specific CD4⁺ T cells than in non-ApoB-specific bulk CD4⁺ T cells²¹. Instead, ApoB-specific T cells were enriched in genes encoding transcription factors, cytokines and chemokine receptors typically associated with T_H17, T_H1 and T_FH cells including Rora, Tbx21, Bcl6, Il17a, Il21, Ifng, Cxcr3 and Cxcr5 (ref. 21). Overall, the transcriptomes of mouse ApoB-tetramer⁺ cells²¹ resembled those of the CCR5⁺FOXP3⁺T-bet⁺ reprogrammed T_reg cells that were previously described³¹. Thus, atherosclerosis-associated antigen-specific T_reg cells may be more plastic than other T_reg cells. Within the ApoB-tetramer⁺ subset, T cells that expressed FOXP3, but not other transcription factors associated with effector T cell responses, were no longer detectable by flow cytometry in old (more than 20 weeks old) atherosclerotic Apoe-deficient mice²¹. Human ApoB-reactive CD4⁺ T cells from patients with cardiovascular disease also exhibited heightened co-expression of the T_H1 and T_H17 lineage-defining transcription factors T-bet and RORγt in FOXP3⁺ CD4⁺ T cells³³. Their transcriptomes displayed signatures of effector and memory T cells, rather than of T_reg cells²⁴. Indeed, single-cell sequencing of atherosclerotic tissue samples have shown that clonally expanded populations of pro-inflammatory effector and memory T cell subsets accumulate in mouse^11,52 and human^12,53,54 lesions. In humans^12,53 and mice¹¹, T cells found in plaques expressed the genes encoding the pro-inflammatory chemokine receptors CCR5, CXCR6 and CX₃CR1, which have an important role in T cell infiltration into inflamed aortas^31,55. All three chemokine receptors were shown to be expressed at the protein level²⁹. All three receptors are known to be causally involved in atherosclerosis, based on knockout studies^31,55,56. CCR5 is involved in CD4⁺ T cell homing to atherosclerotic aorta³¹ and CXCR6 was also shown to be involved in T cell homing⁵⁵. CCR5, CXCR6 and CX3CR1 are also expressed on human and mouse exT_reg cells^29–31.

MHC class II-dependent interaction between T_reg cells and activated APCs can trigger T_reg cell reprogramming, requiring TCR engagement and an inflammatory environment⁵⁷. Pro-inflammatory APCs, such as dendritic cells, that accumulate in progressing atherosclerotic plaque can drive T_reg cell activation and tune their phenotypes by expressing a diverse array of co-stimulatory molecules, cytokines and chemokines⁵⁸. CCL17 secreted by a subset of dendritic cell has a vital part in progression of atherosclerosis by recruiting inflammatory T cells to lesion sites and hindering T_reg cell maintenance and expansion in lymphatic tissue⁵⁹.

RNA-seq analysis of mouse ApoB⁺ CD4⁺ T cells from young (8 weeks) versus old (20 weeks) Apoe^−/− mice showed activation of inflammatory signalling pathways in autoreactive T cells from old atherosclerotic mice²¹. Specific enrichment of CCR5 (refs. 30,31) and IL-6Rα³² on the surface of exT_reg cells in mice suggests that these cells have acquired increased responsiveness to pro-inflammatory ligands such as CCL5 and IL-6. Mouse¹¹ and human⁵³ plaque CD4⁺ T cells were also enriched in the receptor for the IFNγ receptor IFNGR1, whose expression has been shown to drive loss of suppressive function of T_reg cells in a tumour microenvironment⁶⁰.

A recent study has reported that inflammatory DCs are the predominant DC type found in in atherosclerotic plaque tissue in mouse models of the disease, whereas tolerogenic DCs dominated in lymph nodes¹¹. Human carotid artery plaque macrophages from patients with symptomatic disease also exhibit higher expression of CCL5 than those from the asymptomatic group⁵³. These and other studies^52,61–63 have shown that DC and macrophage subsets in atherosclerotic plaques express inflammatory molecules such as TLR2, IL1B, CCL3, CCL4 and TNF, all of which could contribute to the loss of T cell tolerance within the plaque microenvironment⁴⁹. As such, these are candidate pathways for mediating the loss of tolerance in atherosclerosis and warrant further investigation. It is also possible that breakdown of tolerance may be achieved by aberrant activation of naive T cells, resulting in clonal expansion and a possibly aggressive, pro-inflammatory effector phenotype (Fig. 3).

Fig. 3 |. Activation and expansion of naive T cells.

Dendritic cells (DCs) derived from plaques may enter draining lymph nodes through afferent lymphatics and present self-antigens on MHC class I (MHC-I) or MHC-II molecules to T cells in the context of co-stimulation. This can initiate the activation and clonal expansion of naive CD8⁺ or CD4⁺ T cells. These cells may acquire effector phenotypes, leave the lymph node through efferent lymphatics, and enter the circulation, homing back to atherosclerotic arteries^26,31. Whether such newly activated T cells acquire their effector programme at the site of initial antigen encounter (that is, in the draining lymph node) or in the peripheral (plaque) tissue is not known⁶⁶. HEV, high endothelial venule; TCR, T cell receptor.

However, there is currently no evidence for this mechanism being operative in atherosclerosis. As such, whether the autoimmune component of atherosclerosis is based on exT_reg cells or on self-reactive effector T cells, or even both, remains to be clarified.

Future directions

Currently, it is not known how the different subpopulations of exT_reg cells (namely T_H1-like, T_H17-like, T_FH-like and cytotoxic exT_reg cells) are related to each other. The trigger responsible for converting T_reg cells that are specific for atherosclerosis-associated autoantigens into exT_reg cells is also unknown. It is not even clear whether there is one triggering pathway or whether multiple pathways interact. On the APC side, it is possible that the migration or function of tolerogenic DCs may be defective in atherosclerosis⁶⁴, and there may be atherosclerosis-associated factors that result in fewer to lerogenic APCs and more pro-inflammatory APCs. Discovering the factors that promote the shift of T_reg cells to exT_reg cells and tolerogenic APCs to inflammatory APCs will be of paramount importance for designing possible immune-based therapeutic or preventative strategies for atherosclerosis. Capitalizing on these insights may enable the development of tolerogenic vaccines to restore tolerance to self, an approach that has shown promise in mouse models of atherosclerosis⁶⁵.