Role of the adaptive immune system in atherosclerosis

Abstract

Atherosclerosis, the pathology underlying heart attacks, strokes and peripheral artery disease, is a chronic inflammatory disease of the artery wall initiated by elevated LDL cholesterol levels. LDL accumulates in the artery wall, where it can become oxidized to oxLDL. T cell responses to ApoB, a core protein found in LDL and other lipoproteins, are detectable in healthy mice and people. Most of the ApoB-specific CD4 T cells are FoxP3+ regulatory T cells (Treg). In the course of atherosclerosis development, the number of ApoB-reactive T cells expands. At the same time, their phenotype changes, showing cell surface markers, transcription factors and transcriptomes resembling other T-helper lineages like Th17, Th1 and follicular helper (TFH) cells. TFH cells enter germinal centers and provide T cell help to B cells, enabling antibody isotype switch from IgM to IgG and supporting affinity maturation. In people and mice with atherosclerosis, IgG and IgM antibodies to oxLDL are detectable. Higher IgM antibody titers to oxLDL are associated with less, IgG antibodies with more atherosclerosis. Thus, both T and B cells play critical roles in atherosclerosis. Modifying the adaptive immune response to ApoB holds promise for preventing atherosclerosis and reducing disease burden.

Introduction

Atherosclerosis is a disease of the arterial wall. Predilection sites are characterized by oscillatory or disturbed flow (1). In humans, atherosclerosis of the coronary arteries leads to coronary artery disease (CAD), angina, unstable angina, and ultimately myocardial infarction. Cerebrovascular disease (CVD, typically affects the carotid arteries and the arteria cerebri media) leads to transient ischemic attacks and ultimately stroke. Peripheral artery disease (PAD), commonly occurring in the lower limbs, leads to intermittent claudication, gangrene and ultimately amputation. All forms of atherosclerotic disease are positively correlated with LDL cholesterol levels, age, smoking, hypertension and other classical risk factors, which also correlate with biomarkers of inflammation (2).

One factor in determining where atherosclerotic lesions occur is endothelial cell activation, either by cytokines (3) or flow patterns (1). There is evidence that pro-inflammatory and thus pro-atherogenic flow patterns exist in arteries. Such observations can explain where in the arterial tree (branch points, bends) atherosclerosis starts. However, they do not explain why atherosclerosis affects arteries but not veins, in which oscillatory shear stress is common. In a healthy artery, the intimal layer consists only of the endothelium and its basement membrane. Vascular macrophages are constitutively present in the adventitia of large arteries (4). Initially, vascular macrophages were thought to be a homogeneous population, seeded early in embryonic development and maintained by local proliferation (4). Recent studies using single cell RNA sequencing (scRNA-Seq) show at least two resident macrophage subsets (5), and at least 3 more macrophage subsets that appear in atherosclerosis. Macrophages have long been known to be important in atherosclerosis (6–9). However, the role of the constitutively present vascular macrophages is not known yet, since models of specific depletion or alteration of vascular macrophages are not available. During atherosclerosis, monocytes infiltrate the arterial wall. Some differentiate to macrophages, others to foam cells and dendritic cells. Together with T and B cells, smooth muscle cells and extracellular matrix, cholesterol crystals, necrotic core areas and often cartilage- and bone-like structures, these immune cells form the neointima (figure 1).

Figure. Adaptive Immunity in Atherosclerosis.

Mature T cells (naive CD4: light blue, CD8: dark blue, Tregs: green) leave the thymus and migrate to lymph nodes, spleen and the artery wall (blue arrows). B cell precursors leave the bone marrow and seed lymph nodes, spleen and the artery wall (thin red arrows) with B cells (red). Naive T and B cells are mostly in the adventitia of the artery. In the plaque, T cells interact with macrophages (blue and brown to indicate subsets) and dendritic cells (DCs, peach) in a recall response. They differentiate to Th1 and Th17 (pink). T cells from the artery wall leave and home to lymph nodes (orange arrow). There, naive T cells differentiate into Th1, Th17, Treg and TFH (purple). In germinal centers of the spleen and lymph nodes, TFH provide critical help to B cells (red), allowing them to isotype switch and home to the bone marrow (green arrows). In the bone marrow, they differentiate to plasma cells (large red cells) and secrete IgG and IgM specific for atherosclerosis antigens. The small dark red arrow in the capsule of the lymph node indicates where soluble antigens and antigen-loaded dendritic cells from tissue enter. I wish to acknowledge Andreas Schober, LMU Munich, Germany for the histological image of a human carotid artery with severe atherosclerosis and Servier http://smart.servier.com/ for digital artwork I used for the immune cells.

The Adaptive Immune System

This review is focused on adaptive immunity, comprising B and T cells. The enormous diversity of antibodies, B cell receptor (BCRs) and T cell receptors (TCRs) is caused by unique recombination events of genomic sequence elements. Adaptive immunity evolved in sharks (10) when the recombinase-activating genes (RAG)-1 and 2 appeared. All cells of the adaptive immune system require both the RAG-1 and RAG-2 gene products to undergo complicated recombination events (11, 12). The V, D and J segment genes of TCRs and BCRs recombine at the single cell DNA level. These recombined genes form most of the variable part of each TCR and BCR, called CDR3, flanked by constant regions and signaling subunits.

B cells develop in the bone marrow and reside in secondary lymphoid organs including lymph nodes and spleen (13). B cells mature in germinal centers with the help of TFH cells (14). During this process, additional mutations are introduced in the sequence of the antibodies under the influence of the enzyme AID, and antibodies switch isotypes from IgM to various IgG isotypes, IgE or IgA. The variable portions of the light and heavy chains of these antibodies share sequence identity with the BCR of the B cell that produces them. Mature B cells can form memory, meaning that they persist for years and can be reactivated when the same antigen is experienced again (15). Other B cells become antibody-secreting plasma cells (15).

T cells develop in the thymus, which is seeded by precursor cells from the bone marrow that undergo complex maturation and selection steps (16). Positive selection requires that each TCR recognizes an endogenous self antigen peptide with some (low) affinity (17). Negative selection means that T cells that recognize such self antigens with high affinity are eliminated by induced apoptosis (17). Once they are mature, T cells leave the thymus to reside in secondary lymphoid organs. Some specialized tissue-resident T cells reside in tissues, especially in epithelia (18). The TCR requires a multimeric protein complex called cluster of differentiation-3 (CD3) and many downstream signaling molecules to be functional (19, 20). Some T cells express γδ TCRs and are not considered further, because their role in atherosclerosis is unclear. Most T cells express αβ TCRs, which associate with CD3γ, δ, ε and ζ. Most αβ T cells express one of two co-receptors, CD4 or CD8, but usually not both.

CD4 T cells recognize antigenic peptide sequences called epitopes. These peptide epitopes must be presented by major histocompatibility complex (MHC)-II, which is expressed on dendritic cells, B cells and other antigen-presenting cells. MHC-II has an “open” groove; thus, it can bind peptides of 9 amino acids (minimum binding sequence) or longer. CD8 T cells recognize peptide epitopes bound to MHC-I that are typically 8-10 amino acids long. MHC-I has a “closed” groove, which limits the peptide length. MHC-I is expressed on all nucleated cells.

CD4 T cells can be regulatory and curb inflammation. FoxP3+ regulatory T cells (Tregs) inhibit effector T cell proliferation and cytokine production by secreting TGF-β and IL-10, and by contact-dependent mechanisms (19, 20). FoxP3-negative regulatory T cells are called Tr1 and mostly secrete IL-10 (21). CD4 helper (h) T cells include Th1, which produce interferon-γ, Th2, which produce IL-4, IL-5 and IL-13, Th9, which produce IL-9, Th17, which produce IL-17A and IL17F, and follicular helper T cells (TFH), which produce IL-21. Th1 (22, 23) and TFH (24) cells clearly promote atherosclerosis, Tregs dampen atherosclerosis (25). The role of the other subsets is unclear (20).

In humans, the genes encoding both MHC-I and II are highly polymorphic. Several genes are expressed (HLA-A, B, C for MHC-I, HLA-DR, DQ, DP and others for MHC-II), and these genes are among the most highly variable genes in the human genome. Thus, no two persons except identical twins share the same MHC-I and II alleles. This becomes important for the design of vaccines and cell-based therapies (see Perspectives).

B cells are not MHC-restricted. Hence, B cells not only recognize linear peptides, but can also recognize non-linear epitopes in proteins, lipoproteins, glycoproteins, glycolipids and modified self, for example, haptens bound to proteins, or chemically modified amino acids. When a B cell first encounters a cognate antigen, its BCR binds the antigen, which results in signaling events that lead to secretion of IgM antibody specific to that antigen (26). Physical contact with a TFH cell in a germinal center (GC) will initiate isotype switching from IgM to IgG and other isotypes (14). Importantly, the antigen specificity of the IgG initially is the same as that of the IgM. However, the enzyme AID allows for hypermutation in variable segments of the BCR sequence that, through natural selection in multiple rounds of expansion in GCs, ultimately leads to higher affinity antibodies (27).

Atherosclerosis Antigens

Low density lipoprotein (LDL) is the best-known and best-characterized atherosclerosis antigen (12). LDL contains a core of cholesterol esters and triglycerides, a shell of phospholipids and cholesterol, and the apolipoprotein ApoB. Controlling LDL cholesterol by statins or PCSK9 inhibitors has been highly successful in reducing the atherosclerosis burden and major adverse cardiovascular events (MACE) such as myocardial infarction, stroke or cardiovascular death. Lowering LDL cholesterol to targets in the range recommended by heart associations can lower MACE rates by up to 35% (28). LDL binds LDL receptor (LDLR), mainly expressed in hepatocytes, where LDL can be targeted for degradation and excretion through the bile and other pathways (29). The LDLR normally recycles, meaning that each LDLR molecule can take up many LDL molecules. However, the soluble enzyme PCSK9 can prevent LDLR recycling and promote LDLR degradation (30). Therefore, people with elevated PCSK9 are at higher risk for MACE. Monoclonal antibodies to block PCSK9 are approved for clinical use and have also been shown to lower MACE rates by about 35% (31). It is not clear whether statins and PCSK9 inhibitors have additive effects (32). As far as we know, lowering LDL cholesterol by either method shows similar benefits.

Residual Inflammatory Risk

As we have seen, lowering LDL cholesterol by statins or PCSK9 inhibitors to “optimal” levels reduces MACE by 35%. In patients treated to target, the “cholesterol risk” is considered to be under control. However, the reduction in MACE is incomplete and leaves many patients at risk. This remaining risk is called “residual inflammatory risk” (32). In several major clinical trials, the relevance of this residual inflammatory risk was proven. Specifically, the CANTOS trial showed that blocking the inflammatory cytokine IL-1β by canakinumab reduced MACE in patients with elevated inflammatory markers (33). The COLCOT trial showed that low-dose colchicine reduced MACE (34, 35), whereas methotrexate or TNF blockade did not. Thus, there is specificity to anti-inflammatory interventions that provide benefit to people with atherosclerosis who are at risk of MACE. Currently, we do not know why some anti-inflammatory inventions work and others don’t. Some of the residual inflammatory risk is genetic (36). Predictive genetic risk scores can be calculated even from low resolution single nucleotide polymorphism (SNP) maps (37). However, the function of most of the SNPs used to calculate genetic risk scores is not known. Most SNPs are in introns or intergenic regions, with no obvious target gene. Thus, genetic atherosclerosis risk scores are correlative, not causative.

Adaptive Immunity and Atherosclerosis

Much of the inflammation in atherosclerosis is controlled by the adaptive immune system. Atherosclerosis is not an autoimmune disease, because it is not initiated by a faulty immune reaction to self antigens. However, an autoimmune response is always detectable in patients with atherosclerosis and in mouse models. The main mechanisms by which adaptive immunity controls inflammation is through antibodies and cytokines. The Fc portion of antibodies binds Fc receptors (38) on myeloid and other cells. Fc receptor ligation usually results in enhanced inflammation (39, 40). Antibodies also opsonize by binding and activating complement (41). Many complement factors including C3a and C5a are pro-inflammatory. Most cytokines of the adaptive immune system are produced by T cells. CD8 T cells mainly produce interferon-γ, which activates macrophages and is strongly pro-inflammatory and pro-atherogenic. CD4+ Th1 cells also produce interferon-γ. Th17 cells produce IL-17A and F, both of which are pro-inflammatory. Th2 cells produce IL-4, IL-5 and IL-13. The role of the Th2 cytokines in atherosclerosis is unclear. Some effects that were previously attributed to Th2 cells may be mediated by innate lymphocyte-like cells (ILC)-2 (42), which produce the same cytokines as Th2 cells. IL-9, the product of Th9 cells, is also pro-atherogenic. The two types of regulatory T cells, Tregs and Tr1 cells, secrete TGF-β and IL-10, both of which are atheroprotective.

Understanding adaptive immunity in atherosclerosis can inform possible preventive and therapeutic strategies. The adaptive immune system can be manipulated by vaccination, tolerization, cell-based therapies, or specific cytokine manipulations. Vaccination is a process inducing an immune response to an antigen that is present in the vaccine and similar or identical to the intended target. Most vaccines require an adjuvant, which is an additive that alerts the pattern recognition receptors of the immune system and allows antigen-presenting cells to initiate an immune response (43, 44). Tolerization is achieved by offering an offending antigen at low dose, repeatedly and without adjuvant. A tolerogenic vaccine is designed to induce Tregs with specificity to the offending antigen. Cell-based therapies are commonly used in cancers, especially leukemias and lymphomas. CD19-targeted CAR-T cells are very successful at controlling CD19+ lymphomas (45). Currently, there is no cell-based approach for atherosclerosis treatment or prevention. Specific cytokine manipulations can be used to induce desired immune cell populations. For example, injecting IL-2 complexed by IL-2 antibodies has been shown to induce Tregs and reduce atherosclerosis in mouse models (46).

Towards a Tolerogenic Atherosclerosis Vaccine

My lab is interested in finding atherosclerosis-relevant epitopes to build a tolerogenic vaccine. To achieve this goal, the epitopes must be discovered, validated and then tested for the kind of immune response each induces. ApoB contains hundreds of MHC-II restricted epitopes and at least as many, if not more, MHC-I restricted epitopes. My lab recently published 30 human ApoB peptide sequences that bind common alleles of MHC-II expressed in humans with high affinity (47). Similarly, we published 27 MHC-II-restricted epitopes in mouse ApoB (48). Some MHC-I restricted mouse ApoB epitopes have also been published (49). However, finding epitopes does not predict how strong the immune response to these peptides will be. Strongly immunogenic peptides are called immunodominant epitopes. My lab is actively working on finding immunodominant epitopes in ApoB. Knowing the peptide epitope and the MHC allele usually cannot predict the TCR sequence (50). However, TCRα and β sequences can be determined experimentally by single cell RNA sequencing (scRNA-Seq) followed by reconstruction using algorithms like TRACER (51). TRACER generates a “combinatorial recombinomes” of all possible recombination events between polymorphic V and J segments to allow mapping of the demultiplexed reads from each single cell and assemble them into contigs. Thus, the entire CDR3 region, responsible for antigen specificity, can be sequenced. The experimental tools to discover relevant epitopes and the T cells recognizing them are now available.

To find ApoB-specific T cells, two fundamentally different approaches can be used. In one approach, recombinant MHC-I or MHC-II molecules are biotinylated, loaded with peptide epitopes and tetramerized using fluorescently tagged streptavidin (45). MHC-I tetramers are commercially available and, in general, more robust than MHC-II tetramers. Each tetramer is specific for the MHC allele from which it was constructed and hence will only work in subjects expressing that MHC allele. Each tetramer reagent must be validated for specificity of binding and against background binding to myeloid cells (47). In our hands, many MHC-II tetramers fail validation, usually due to excessive non-specific binding. The other methods are restimulation assays (45, 52, 53). In these assays, peripheral blood mononuclear cells (PBMCs) are restimulated with the peptide epitopes of choice. Short-term restimulation assays are also called antigen-induced molecule (AIM) assays. The readout is expression activation markers such as CD40L or CD69, measured by flow cytometry (54). Longer-term (up to 14 days) restimulation assays expand the cells recognizing peptide epitopes in the PBMC culture in the presence of IL-2. The limitation of restimulation assays is that the cell culture conditions bias for some types of T cells and against others. Cell culture also influences the phenotype and polarization of T cells (table 1).

Table 1.

Assays for detecting epitope-specific T cells.

Method > Criteria V	Tetramers	AIM assays	Expansion and Restimulation
Specificity	Excellent	Excellent	Excellent
Number of cells	Low (~100)	Low	High
Phenotype	Preserved	Somewhat biased	Highly biased
Validation	Difficult	Unproblematic	Unproblematic
Positive control	Vaccinated mice	Polyclonal stimulation by anti-CD3 and CD28
Negative control	MHC mismatch	No peptides, irrelevant peptides

Using MHC-II tetramers, my lab showed that most CD4 T cells specific for the interrogated ApoB epitope are Tregs in individuals without detectable CVD (47). This study was very limited in that only 21 MHC-matched PBMC samples were tested. No samples from men were included. Most samples were from women living with HIV (HIV+). In this cross-sectional study, the T cells were only interrogated at one time point. Thus, the dynamics of T cells numbers and phenotype are unknown. However, the validated APOB p18:DRB1*07:01 tetramer can now be used for all individuals who express this allele. DRB1*07:01 is relatively common (about 8% of the general population in the US).

In building a tolerogenic vaccine, the instability of Tregs (55, 56) must be considered. Mouse studies provide evidence that Tregs change their phenotype with the onset and progression of atherosclerosis (57, 58). Tregs express the transcription factor FoxP3 and the high affinity IL-2 receptor CD25. Using the Apoe^−/− model of atherosclerosis, two studies (57, 58) showed CD25- FoxP3^low cells that had acquired T-bet, the hallmark transcription factor for Th1 cells, and expressed the Th1-typic cytokine IFN-γ. Lineage tracking studies showed that such cells are likely ex-Tregs: regulatory T cells that once expressed FoxP3 but no longer do (24, 59). Some of these cells switch to a follicular helper (TFH) phenotype and are pro-atherogenic (24). In all these studies, the antigen specificity of these “switched” CD4 T cells is unknown.

Longitudinal studies of human CD4 and CD8 T cells specific for ApoB or other atherosclerosis antigens (12) are sorely needed. Of particular interest are samples from “converters”, defined as subjects who acquired cardiovascular disease over time and for whom PBMC samples are available before and after the onset of CVD. Both tetramer-based and restimulation assays work in frozen PBMC samples (47), a material that is available from several cohorts (60). Deep phenotyping can be achieved by mass cytometry and single cell RNA sequencing (scRNA-Seq) (8, 9). scRNA-Seq can be combined with oligonucleotide-tagged antibodies to define the cell surface phenotype coupled to the transcriptome in the same cell (7). This type of data is necessary to understand how the immune system responds to and modulates atherosclerosis.

Developing a tolerogenic vaccine is attractive, because, if successful, it would be a “home run”, benefitting millions or billions of at-risk women and men throughout the world. Another vaccination approach is to target PCSK9. PCSK9 vaccines are designed to induce neutralizing antibodies to PCSK9 (43). The challenge facing PCSK9 vaccines include the limitation that PCSK9 is a self-antigen, and thus the antibody titers achieved by vaccination remain low and may not be neutralizing. Targeting cytokines like IL-1β (33) or innate immune cells by low-dose colchicine have shown evidence of clinical benefit (35). Ultimately, it remains to be seen which of these strategies will be safe, effective and successful in the healthcare marketplace.

Perspectives

Importance of the field.

Blocking IL-1β by canakinumab has shown reduced MACE rates in subjects with high levels of the inflammation marker CRP (33). Low-dose colchicine also showed clinical benefit (35), as have other interventions (61–63). This has directed the focus of atherosclerosis research to the immune system.

Current thinking.

Vaccinations are the most successful intervention in medicine and have greatly reduced the burden of infectious diseases. Conceptually, vaccination can work for atherosclerosis (19, 64). Tolerogenic vaccination for atherosclerosis was proposed more than 25 years ago: rabbits immunized with oxLDL showed less atherosclerosis (65). Mice vaccinated with MHC-II-restricted ApoB peptide epitopes are protected from atherosclerosis (19, 48, 66) by producing a tolerogenic response.

Future directions and challenges.

A major challenge in designing tolerogenic vaccines is that ApoB-specific T cells can switch their phenotype away from Tregs and towards effector T cells in autoimmune and inflammatory settings (55, 56). This is well documented in mice and humans, where natural ApoB-specific CD4 T cells are Tregs, but in humans with documented CAD or CVD, may switch to Th17, Th1, TFH and intermediate phenotypes (47). Cell-based immunotherapy could conceivably be used to target atherosclerosis using engineered Tregs (67). In such schemes, ApoB-specific Tregs would be constructed or expanded in vitro and infused back into the patient, which is costly (68).