T Cell Development and Function

Abstract

T cells play critical roles in the immune system, including in responses to cancer, autoimmunity, and tissue regeneration. T cells arise from common lymphoid progenitors (CLPs) that differentiate from hematopoietic stem cells in the bone marrow. CLPs then traffic to the thymus, where they undergo thymopoiesis through a number of selection steps, resulting in mature single positive naive CD4 helper or CD8 cytotoxic T cells. Naive T cells are home to secondary lymphoid organs like lymph nodes and are primed by antigen-presenting cells, which scavenge for both foreign and self-antigens. Effector T cell function is multifaceted, including direct target cell lysis and secretion of cytokines, which regulate the functions of other immune cells (refer to “Graphical Abstract”). This review will discuss T cell development and function, from the development of lymphoid progenitors in the bone marrow to principles that govern T cell effector function and dysfunction, specifically within the context of cancer.

Introduction

Tcells are a subset of lymphoid-derived immune cells, which are able to specifically recognize a diverse array of antigens presented on the major histocompatibility complex (MHC) of cells through their T cell receptor (TCR). This recognition is the product of multiple TCR rearrangements that occur in the thymus during T cell development, resulting in ∼10⁸ unique T cell clones with different antigen specificities being present in the body at any given time.¹ Thymic T cell development results in two phenotypically and functionally distinct naive or antigen-inexperienced αβ T cell subtypes: CD4+ helper and CD8+ cytotoxic T cells,² in addition to γδ T cells³ and NKT cells.⁴

T cells are activated when they are primed by antigen-presenting cells (APCs), a process that involves the processing of antigen by APCs and their subsequent presentation to T cells.⁵ Upon activation, T cells migrate into tissues where CD8+ and CD4+ T cells exercise effector functions by directly lysing target cells through the action of granzymes and perforins, and secreting both pro inflammatory and anti-inflammatory cytokines.^6,7 Activated T cells differentiate into different T cell subsets, including long-lived memory cells that allow rapid responses upon re-encountering antigen.⁸

T cells have been demonstrated to be crucial to the immune system's response to cancer, which has led to significant interest in T cell-based cancer immunotherapies.⁹ A typical example is adoptive T cell transfer, where T cells are harvested from patients and expanded before being reintroduced.⁹ These T cells can be either tumor-infiltrating lymphocytes or chimeric antigen receptor (CAR) T cells, which are genetically modified polyclonal T cells that can specifically recognize tumor cells.¹⁰

To maximize the therapeutic potential of T cells, it is important to understand the basic principles that govern the development, function, and dysfunction of these cells. Consequently, we present a review of T cell development and function, starting with a description of the development of lymphoid progenitors in the bone marrow and thymopoiesis. The mechanism of T cell priming is outlined, after which T cell differentiation, effector function, and dysfunction are discussed. The review ends by discussing the role of T cells in antitumor immunity, and ways by which tumors evade T cell activity.

T Cell Development

Development of lymphoid progenitors in the bone marrow

Hematopoietic stem cells (HSCs), identifiable by the expression of c-Kit^hiLineage⁻Sca-1⁺ in the bone marrow fraction,¹¹ are multipotent cells whose daughter cells differentiate to continually replenish all classes of blood cells. HSC differentiation in the bone marrow, in response to differentiation signals, is thought to be multistaged with successive loss of capacity for self-renewal. Specifically, at least three multipotent stages have been identified: long-term HSC and short-term HSC, which have long-lasting and limited ability for self-renewal, respectively, as well as multipotent progenitors (MPPs), which have no capacity for self-renewal.^11,12

HSCs differentiate into common lymphoid progenitors (CLPs) and common myeloid progenitors (CMPs), which specify the lymphoid and myeloid lineages, respectively. The lymphoid lineage comprises T cells, B cells, and NKs, while the myeloid lineage is made up of megakaryocytes, erythrocytes, granulocytes, and monocytes/macrophages.^11,13 Dendritic cell lineage classification is unclear, as dendritic cells can arise from both CLPs and CMPs.¹¹ The traditional understanding of the differentiation of the lymphoid and myeloid lineages from HSCs is based on the “classical” model of hematopoiesis, where the lymphoid and myeloid lineages symmetrically diverge from MPPs in an irreversible lineage commitment from HSCs.

Recent findings, however, suggest a different model of differentiation based on a step-wise loss of myeloid differentiative potential during early lymphoid differentiation. This asymmetric, stepwise loss of CMP potential was shown to depend on the relative expressions of Flt3 and VCAM on MPPs, with Flt3^lowVCAM^hi being the most primitive of the MPPs and having the highest potential to differentiate into CMPs, while Flt3^hiVCAM^low MPPs are most restricted and predominantly differentiate into CLPs in vivo.^11,13

T cells are generally accepted to arise from CLPs; however, it is unclear when and how the earliest thymic progenitors (ETPs), which seed the thymus for subsequent T cell differentiation, diverge from the other lymphoid lineages. Recent studies have cast doubt on the traditionally accepted paradigm of T cells being generated solely from CLPs, suggesting that T cell lineage progenitors could occur upstream of the CLPs in the bone marrow.¹⁴ This is mainly due to the observation that, while CLPs can intrinsically differentiate into T cells, the ETPs, which home the thymus, are unresponsive to the cytokine interleukin 7 (IL7; at the ETP stage), and do not express the IL7Rα receptor: a defining marker for CLPs.^14,15

Thus, IL7Rα-expressing CLPs are more likely to home regions in the bone marrow that express IL7, rather than traffic to the thymus, and so would not necessarily be the sole contributor to thymopoiesis. There is, however, consensus that the divergence of T cell progenitors must occur after the major lymphoid-myeloid branch point in hematopoiesis, even though the exact point of divergence is yet to be elucidated.¹³

CD4+ CD8+ thymocyte generation in the thymus

Early thymic progenitor cells home the thymus from the bone marrow, where they undergo T cell lineage commitment and begin the process of maturation, identifiable by the expression of a specific combination of cell surface markers and TCR rearrangements. In the thymus, T cell precursors lose the potential for B cell differentiation by engaging the Notch-1 ligand expressed by thymic stromal cells at the cortex-medulla junction.^16,17

The immature thymocytes, currently lacking the expression of CD4/CD8 co-receptors, as well as surface TCRs are characterized as double negative (DN) populations. DN thymocytes undergo four DN stages of sequential differentiation (DN1 to DN4) distinguishable by their expression of CD44 and CD25: CD44⁺CD25⁻, CD44⁺CD25⁺, CD44⁻CD25⁺, and CD44⁻CD25⁻, respectively.² The DN thymocyte populations can proceed to differentiate into two main classes of T cells: γδ or the more common αβ TCR-expressing cells. The path to αβ T cell differentiation first involves the expression of the pre-TCR-α by DN3 thymocytes, which do not undergo genetic rearrangement.²

Pre-TCR-α forms stable interactions with the TCR β-chain, which undergoes somatic DNA rearrangement: V (variable), D (diversity), J (junction) recombination, which allows T cells to eventually recognize the myriad of antigens present (both self-antigens and foreign antigens).¹⁸ Thymocytes that undergo a productive TCR β-chain rearrangement survive and proliferate, while those that do not undergo productive TCR β-chain rearrangement do not receive the requisite survival signal, in a process known as β-selection.^17,19 Post β-selection, thymocytes lose the expression of the pre-TCR-α, and express the TCR-α, which is a product of genetic V and J recombination. Thymocytes also gain the expression of both CD4+ and CD8+ expression and are referred to as double-positive (DP) thymocytes.²

Thymic selection

DP αβ T cells proceed to undergo positive selection by engaging MHC complexed self-peptides expressed on thymic epithelial cells in the cortex and negative selection by engaging peptides on both thymic dendritic cells (DCs) and epithelial cells, mainly in the medulla.¹⁹ In positive selection, thymocytes that fail to engage self-peptide-MHC complexes die by neglect, while those that do receive survival signals promote positive selection.¹⁹ Over 90% of DP thymocytes die at this stage.

Studies have shown that peptides involved in positive selection are often low-affinity antagonists or weak agonists that would not ordinarily lead to mature T cell activation.^19,20 In fact, it is believed that mature T cells achieve homeostasis through tonic TCR signaling by engaging the positively selecting peptides they re-encounter outside the thymus.^19,20 Cortical thymic epithelial cells (cTECs) are integral to positive selection and work by forming 3D scaffold niches that provide intimate interactions with thymocytes.¹⁹ cTECs present peptides on both MHC I and MHC II to facilitate CD8+ and CD4+ T cell positive selection, respectively.²⁰

In contrast to positive selection, thymocytes that engage too strongly with self-peptides die by negative selection through apoptosis, or are tolerized into regulatory T cells.¹⁹ Peptides involved in negative selection are often high-affinity agonists, which can lead to mature T cell activation.²⁰ Thus, conditions that disrupt thymic negative selection lead to spontaneous autoimmunity.¹⁹ In negative selection, medullary thymic epithelial cells (mTECs) are responsible for generating tissue-restricted antigens, which are presented to thymocytes either directly by the mTECs or handed over to APCs like thymic dendritic cells for presentation.^19,21 Positive and negative selection and subsequent downregulation of co-receptors ultimately result in thymocyte differentiation into mature single-positive CD4+ or CD8+ T cells that egress the thymus to the periphery.²

T Cell Priming

Structure and function of secondary lymphoid organs

While the bone marrow and the thymus, which are the main sites for hematopoiesis and thymopoiesis, are classified as primary lymphoid organs, secondary lymphoid organs (SLOs) like lymph nodes provide the microenvironment for lymphocyte activation and effector differentiation.²² The development of SLOs begins during embryogenesis and proceeds during development. Lymph node development begins with endothelial cells budding from larger veins and forming lymph sacs during embryogenesis.²³ Nerve fibers produce retinaldehyde dehydrogenase 2, an enzyme that is known to convert retinal to retinoic acid (RA), and which in turn results in mesenchymal cells (lymphoid tissue organizer [LTo] cells) producing CXCL13 in response to RA.^22,24

The lymph node mesenchymal cells are thought to differentiate from adipocyte progenitor cells by signaling of the lymphotoxin beta receptor (LTβR).²² CXCL13 expression leads to the recruitment of lymphoid tissue inducer (LTi) cells, which cluster with the stromal cells to form lymph node anlage.²⁴ LTi cells develop from lymphoid cell precursors and express LTα1β2, which binds to LTβR expressed on the LTo cells and leads to a positive feedback loop resulting from the expression of high levels of ICAM-1, VCAM-1, and MAdCAM-1, as well as the increased expression of CXCL13, and leads to the accumulation of more LTi cells to the developing lymph node.^22,24 LTo cells ultimately differentiate into lymph node stromal populations like the fibroblastic reticular cells (FRCs) or follicular dendritic cells, which are critical for lymph node organization and lymphocyte activation.²²

The developed lymph node is a highly specialized organ, which sits at the convergence of major blood vessels, and whose role is to filter incoming lymph and facilitate antigen-specific activation of lymphocytes. The lymph node architecture is stratified into the cortex, paracortex, and medulla.²⁵ The cortex is B cell rich, and mainly comprises the primary and secondary lymphoid follicles, which house naive B cells and B cells undergoing antigenic stimulation, respectively. Secondary lymphoid follicles comprise the mantel, germinal center, and marginal zone compartments.²⁵ While B cells undergoing active immune response reside in the germinal centers, memory and plasma B cells are enriched in the marginal zone and medulla, respectively.²⁵

In contrast to the cortex, the paracortex is T cell rich, and houses both naive T cells and T cells undergoing antigenic stimulation.^25,26 T cell entry and retention in the paracortex are highly dependent on the expression and presentation of the chemokines CCL19 and CCL21 on lymph node stromal cells such as FRCs. FRCs have also been shown to produce the cytokine IL7, which is important for the survival of naive T cells.²⁶ Lymph enters lymph nodes through the afferent vessels and flows into a series of sinuses, including the subcapsular and cortical sinuses where the lymph is filtered for antigen. The medulla, which resides at the innermost compartment of lymph nodes, is made up of sinuses that collect lymph that has been filtered through the lymph node and drains to the efferent lymphatic vessels to exit the node.²⁷

Mechanism of T cell priming in lymph nodes

T cells enter lymph nodes through high endothelial venules (HEVs) and traffic to the paracortex where they interact with APCs to initiate the process of T cell priming.²⁶ It is generally believed that T cell migration within the paracortex of the lymph node is largely random.²⁸ However, some studies have shown that the apparent random walk behavior is due to the structure of the stromal cell networks in the paracortex, where FRCs make up a 3D network of randomly branching collagen-rich reticular fibers, providing T cells with tracks on which to migrate.²⁸

Other reports have speculated that the random walk behavior could be due to the interactions between the densely packed cells, as it has been shown that T cells sequentially interact with multiple DCs before T cell activation is initiated.²⁸ T cells that find their cognate antigen establish long-term contact with DCs, while those that do not egress lymph nodes through the cortical and medullary sinuses to the efferent lymph.^26,28

Contact between T cells and APCs during T cell priming is dynamic, and occurs in three successive stages: transient serial encounters, stable contacts, and high T cell motility with rapid proliferation²⁹ (Fig. 1). Intravital imaging showed that during the first phase of activation, T cells and APCs formed short successive contacts lasting about 4 minutes within the first 2 hours, increasing to about 7 minutes at 8 hours.²⁹ However, these contacts become stable after 12 hours, with most conjugates lasting longer than 30 minutes with concomitant production of cytokines. Brief contacts resumed at about 48 hours, where T cells showed high motility and started undergoing rapid proliferation.²⁹

FIG. 1.

T cell priming in the lymph node. Priming of naive T cells in lymph nodes involves three broad phases: Phase 1: T cells and APCs form short, successive contacts where T cells scan for the presence of cognate antigen. Phase 2: In the presence of cognate antigen, T cells arrest and undergo productive interactions with APCs. The engagement of co-stimulatory ligands amplifies T cell activation and avoids anergy. This phase is also characterized by cytokine secretion, which further enhances T cell activation and instructs T cell effector differentiation. Phase 3: T cell activation is followed by rapid proliferation and motility, and subsequent egress from lymph nodes into target tissues. APCs, antigen-presenting cells.

Productive interaction of T cells with their cognate antigen on APCs requires successful processing and presentation of antigen by APCs on their MHCs. While MHC class I presentation happens in all cell types, MHC class II presentation mostly occurs in professional APCs.⁵ MHC class I peptides are generated endogenously from the cell's own translational machinery, while MHC class II molecules bind to peptides derived from proteins that access the endosomal–lysosomal antigen-processing compartments through autophagy, and various endocytic pathways such as phagocytosis and pinocytosis, or by receptor-mediated endocytosis.^5,30 In addition, while MHC class I has a closed peptide-binding groove that can only accept peptides that are 8–10 amino acids long for presentation to CD8+ T cells, MHC class II processed peptides are generally about 13–25 amino acids long and must fit within the MHC class II open peptide-binding groove for recognition by CD4+ T cells.⁵

T cells interact with peptide-MHC complexes by engaging their TCR, which determines T cell antigen specificity and provides downstream signaling for T cell activation. The TCR is a transmembrane complex composed of an extracellular region, a transmembrane region, and a shorter cytoplasmic tail.³¹ The extracellular domain comprises a variable region and a constant region. T cell antigen specificity is dictated by the TCR variable (V) domain, which is generated by somatic genetic recombination in the thymus, and comprised the V alpha and V beta TCR polypeptides.³²

Both components of the V segment have three hypervariable regions: CDR1, CDR2, and CDR3. The CDR3 hypervariable region is the largest and determines TCR antigen specificity, and undergoes conformational changes that allow it to bind the peptide-MHC complex.³² The TCR constant domain interacts with the CD3 complex: δ, γ, ɛ, and ζ chains, and facilitates downstream TCR signaling through ITAM phosphorylation.³¹ In vitro approaches to polyclonal T cell activation and expansion bypass TCR engagement to directly target the CD3 complex.

Two signals are required, in addition to TCR-peptide MHC interactions, to allow productive T cell activation: engagement of co-stimulatory ligands and cytokine instruction. T cell anergy, a state of T cell unresponsiveness to re-stimulation, can occur in the presence of weak TCR signaling.³¹ Co-stimulatory molecules, like CD28, engage B7.1 (CD80) and B7.2 (CD86) on APCs to strongly amplify weak TCR signals, avoiding T cell anergy and resulting in T cell proliferation and differentiation.³¹ Other co-stimulation signals include engagement of LFA-1/ICAM-1 and ICAM-2, and CD2/LFA-3.³² In addition to co-stimulation, cytokines like IL2, IL15, and interferon gamma (IFNγ) are secreted by both T cells and APCs during T cell activation, and serve to enhance T cell activation and proliferation, as well as provide instructions for T cell function and effector differentiation.

In particular, IL2 and IL15 both facilitate T cell proliferation, cytotoxic function and Th-1 differentiation, but while IL2 results in terminal effector T cell differentiation and activation-induced cell death, IL15 enriches for long-lived memory T cells.³³ In addition, IL2, and not IL15, promotes regulatory T cell maintenance.³³ Importantly, IL2 is known to be produced in soluble form primarily by T cells during activation, while IL15 mainly functions through the trans-presentation of membrane-bound IL15–IL15Rα in a cell contact-dependent manner, and can be secreted by a variety of cell types, including APCs, T cells, B cells, and stromal cells.³⁴ In vitro, T cell activation generally includes CD28 engagement by antibodies and supplementation of cytokines such as IL2 and IL15 for T cell maintenance and proliferation.

The dynamics of lymph node architecture evolve during the course of T cell priming to accommodate the increased cellular activity resulting from T cell activation and clonal expansion. For instance, it has been reported that the FRCs, along which T cells, migrate, form collagen-based reticular networks, which contract collagen under resting conditions, and relax during inflammation.^35,36 A relaxed collagen architecture results in altered lymph node stiffness and an expanded reticular network, which allows lymph nodes to increase substantially in size and permits T cell expansion.³⁵

T Cell Effector Function and Dysfunction

Stages of T cell effector/memory differentiation

After priming, naive T cells undergo massive clonal expansion and assume phenotypically and functionally distinct differentiation states with different capacities for effector function, accompanied by significant alterations in gene expression profiles. The major stages of T cell differentiation broadly include naive, memory, and effectors^37–39 (Fig. 2). Naive T cells are antigen inexperienced, have low effector function, and can be identified by the expression of the following cell surface markers in humans: CCR7⁺CD45RA⁺CD127⁺CD62L⁺³⁹.

FIG. 2.

Stages of T cell differentiation. Naive T cells assume various differentiation states upon activation and after antigen clearance. These differentiation states are accompanied by characteristic phenotypic profiles, which can be used for their identification.

Memory T cells can be categorized into central memory, effector memory, and effector memory re-expressing CD45RA (TEMRA).⁴⁰ Central memory T cells are primarily found in circulation or in SLOs, have low effector function, although higher than those of naive T cells, and can be identified in humans as CCR7⁺CD45RA⁻CD45RO⁺CD127⁺CD62L⁺³⁹.

Effector memory, TEMRA, and effector T cells have the highest capacities for effector function and lowest proliferative potentials, and are mostly found in circulation or in tissues. Effector memory T cells are typically identified by the expression of CCR7⁻CD45RA⁻CD45RO⁺CD127⁻CD62L⁻, while TEMRAs are mostly CCR7⁻CD45RA⁺CD45RO⁻CD127⁻CD62L⁻³⁹. Recent reports have identified two additional T cell differentiation states: stem cell memory and tissue-resident memory T cells. Stem cell memory T cells are antigen-experienced T cells with characteristics of naive and memory T cells, which have the ability to self-renew and differentiate into effector phenotypes.⁴¹

Stem cell memory T cells are identified by the following surface markers in humans: CD45RA⁺CD62L⁺CCR7⁺CD95⁺CD27⁺CD122⁺, and are characterized by rapid response to antigens.⁴¹ Tissue-resident memory T cells are a distinct T cell subset that occupies tissues without re-circulating. These T cells express high levels of effector molecules upon antigen exposure, and are seen as pivotal to providing first response against re-encountered antigen.⁴² Recent studies also suggest that T cell phenotype is inherently linked to their location, and that nonlymphoid tissues could play a role in directing T cell fate.⁴³

Several models for memory T cell differentiation from naive T cells have been proposed (Fig. 3). The first is the linear model where memory T cells arise from effector T cells after antigen clearance (i.e., Naive→Effector→Memory T cell differentiation).³⁹ The second is the divergent model where naive T cells can differentiate directly into either effector or memory T cells, bypassing the effector cell stage.³⁹ The third model is a variation of the linear model, where the strength and duration of antigenic stimulation dictate the memory potential of the T cell: short antigen stimulation favors central memory, while longer stimulation favors effector memory and effector T cells.³⁹

FIG. 3.

Models of memory T cell differentiation. Linear model: memory T cells are generated from effector T cells after antigen clearance. Divergent model: memory T cells can be generated directly from naive T cells without having to go through the effector stage. Variation model: strength and duration of antigen stimulation dictate memory potential. Short antigen stimulation favors central memory T cell generation, while a longer stimulation favors effector memory and effector T cells. Decreasing potential hypothesis model: persistent antigen exposure progressively results in T cell populations with decreased capacity for both effector functions and memory formation.

The fourth model is the decreasing potential hypothesis, where persistent and/or cumulative antigen exposure leads to a decreased effector T cell function and concomitant decreased ability to form memory T cells.³⁹ As there exists notable evidence for each model, T cell differentiation may follow a combination of these models, perhaps dependent on the context.

Several changes in transcription factor profiles occur during T cell effector differentiation, which result in distinct functional states. For instance, TCF1, LEF1, and FoxO1 are highly expressed in naive T cells, but are downregulated during effector differentiation and re-expressed during memory T cell development.³⁷ These transcription factors have also been shown to be expressed in stem cell memory T cells. The transcription factors T-bet, Blimp-1, and IRF4 have been shown to be important for the effector T cell program, while Bcl6 and Eomes are memory-related transcription factors.³⁷ The tissue-resident memory T cell program has been shown to require Runx3, Blimp-1, and Hobit, as well as concomitant repression of KLF2.³⁷ These differences in transcription factor profiles and changes in chromatin accessibility lead to differential expression of effector molecules such as granzyme B, IFNγ, IL2, and tumor necrosis factor alpha (TNFα).³⁷

Functional roles of CD4+ and CD8+ T cells

CD4+ and CD8+ T cells have unique and complementary roles in establishing protective immunity. While CD4+ T cells are helper cells that provide signals to induce CD8+ T cell and B cell proliferation in the presence of antigen, CD8+ T cells are characterized as killer T cells that secrete lytic enzymes to destroy infected cells.⁶ After activation, CD4+ T cells can differentiate into multiple T helper (T_H) subsets, including T_H1 and T_H2, based on whether they produce IFNγ or IL4, respectively. While T_H1 CD4+ T cells are important for controlling intracellular pathogens, T_H2 T cells function primarily to control extracellular pathogens. CD8+ T cells also secrete IL2, but to a lesser extent. They mainly mediate effector function either by secreting cytolytic granzymes and perforins or TNFα.

CD4+ and CD8+ T cells differ in their response to the strength and duration of antigenic stimulation. For instance, different co-stimulatory requirements seem to be important for CD4+ and CD8+ T cells: in mice, CD8+ T cells seem to depend more on CD40L, CD28, or OX-40 for co-stimulation, while CD4+ T cells are more affected by the presence of 41BB.⁴⁴

In addition, the threshold for CD8+ T cell activation is thought to be lower than that for CD4+ T cells^39,44 and is less dependent on co-stimulation after initial activation.³⁹ CD8+ cells accumulate activation signals faster than CD4+ T cells,³⁹ and have a faster rate of expansion after stimulation than CD4+ T cells.^39,44 CD8+ T cells also more quickly develop into effector T cells than do CD4+ T cells.⁴⁴ CD4+ T cells, and to a lesser extent CD8+ T cells, also have the ability to form regulatory T cells, which are generated either in the thymus during negative selection (central tolerance), or in the periphery in response to various stimuli. CD4+ regulatory T cells are CD25+FoxP3+ and function by suppressing the immune system to promote self-tolerance and avoid autoimmune complications.^45–47

T cell dysfunction

T cells can undergo dysfunction in broad categories, including anergy, senescence, and exhaustion, each of which has distinct molecular programs⁴⁸ (Fig. 4). T cell anergy results from suboptimal stimulation, usually without the presence of co-stimulation.^48,49 Anergic T cells do not produce IL2 and are unresponsive to antigen re-stimulation.⁴⁹ In vivo, it is thought that T cell anergy is important in inducing peripheral tolerance and avoiding autoimmunity, as the presentation of self-antigen is mostly done by immature APCs without co-stimulation.⁴⁸ The transcription factor, early growth response gene 2, has been shown to be an important regulator of T cell anergy.^48,49 T cell senescence is generally characterized by cell cycle arrest, shortened telomeres, and downregulation of the co-stimulation marker CD28, and is pronounced in aged individuals.^49,50 Senescence in T cells can be caused by repetitive stimulation, stress signals, and DNA damage agents.

FIG. 4.

Modes of T cell dysfunction. Anergy: driven by suboptimal stimulation, usually in the absence of co-stimulation. Senescence: caused by repetitive stimulation, stress signals, and DNA damage. Exhaustion: results from chronic antigen stimulation and accompanied by expression of various inhibitory markers.

Senescent T cells generally have decreased cytotoxic activity, and acquire senescence-associated secretory phenotype, which comprises increased secretion of proinflammatory molecules, including IL6, IL8, and TNF, as well as the suppressive cytokines IL10 and transforming growth factor β (TGF-β).^48,50 T cell exhaustion occurs in the presence of chronic antigen stimulation and inflammation. Exhausted T cells gradually lose proliferative capacity, IL2 production, and expression of TNFα, IFNγ, and granzyme B.^48,50 The gradual functional impairment is accompanied by the expression of inhibitory/exhaustion markers including PD1, LAG3, TIM3, CTLA4, and BTLA.⁴⁸ It should be noted that most of these markers are transiently expressed during T cell activation, but remain stably expressed in exhausted T cells.⁴⁸

The expression of the inhibitory markers also contributes to T cell exhaustion by engaging their respective ligands and facilitating downstream signaling. For instance, the interaction between CTLA4 and CD80/CD86 can outcompete CD28-CD80/CD86 co-stimulation.⁵¹ In addition, PD1 binds to PDL1, which represses TCR signaling and diminishes T cell expansion and effector function.⁴⁸ Approaches to rescue T cell exhaustion have included the use of checkpoint blockade antibodies such as PD1 and CTLA4 antibody blockade alone or in combination, as well as TIM3 or LAG3 blockade.^48,51

T Cells and Cancer

Cancer development

Cancer results when cells become unresponsive to cellular checkpoints that govern normal cell behavior in homeostasis and proliferate uncontrollably. Such cells can invade normal tissues and eventually spread to other parts of the body (metastasis).⁵² Benign tumors remain confined to their location, and do not invade the surrounding tissues or metastasize, while malignant tumors are capable of invading into normal surrounding tissues or metastasizing.⁵² Benign tumors can transform into malignant tumors over time.⁵² Cancers can be grouped into one of three groups depending on the cells of origin: carcinomas are derived from epithelial cells, sarcomas are derived from connective and fibrous tissue, and leukemias or lymphomas are blood cancers that arise from cells in the hematopoietic lineage.⁵²

Cancer development is both clonal and multistep. Clonality is a feature in cancer that describes its single-cell origin.⁵² It is thought that cancers begin from single cells that proliferate uncontrollably, and studies analyzing X chromosome inactivation have been used to demonstrate this phenomenon.⁵² Cancer cells, however, accrue mutations over time in a multistep process during development.⁵² Tumor initiation, where a single cell acquires mutations that result in abnormal proliferation, is followed by tumor progression, where additional mutations confer selective advantages to some cells, causing those cell populations to become dominant within the tumor population. This iterative process, called clonal selection, results in further tumor progression and eventual metastasis.⁵²

Several factors are known to cause cancer, including genetic predisposition, carcinogens, hormones, and viruses. Genetic predispositions to cancer result from germline mutations that are thought to be responsible for certain types of cancers.⁵³ Multiple cancer types, including familial adenomatous polyposis, breast cancer, and ovarian cancer, have been linked to some level of genetic predisposition.⁵³ Carcinogens can either be chemical or radiation based, and mostly act by causing DNA damage, which can lead to eventual mutations during DNA repair.⁵² Other carcinogens, like phorbol esters, cause cancers by stimulating proliferation through the activation of protein kinase C.⁵²

Hormones, like estrogen, are also thought to result in cancers like endometrial cancer, where the proliferation of endometrial cells can be stimulated by the hormone.⁵² Viruses can cause cancer either directly, by the expression of viral oncogenes that lead to cancer formation, or indirectly as a result of prolonged inflammation that can result in mutations in host cells.⁵⁴ Several virus-linked cancers have been reported, including cervical and penile cancers caused by high-risk human papillomaviruses, Kaposi's sarcomas caused by the Kaposi's sarcoma herpesvirus, and Burkitt's lymphoma caused by the Epstein–Barr virus.⁵⁴

The tumor microenvironment

In solid tumors, the tumor microenvironment is a complex, heterogenous and dynamic milieu consisting of cancer cells, infiltrating immune cells, stromal cells, extracellular matrices (ECMs), and secreted factors, which can act to either suppress, or more often promote tumor growth and metastasis.^55,56 Immune cells, for instance, are critical to antitumor response. Macrophages and neutrophils, in their M1 and N1 states, respectively, have been shown to phagocytose and kill tumor cells.⁵⁵ Similarly, dendritic cells in the tumor are important for capturing and presenting tumor antigens to T cells in SLOs.⁵⁵ However, immune cells can play protumorigenic functions, as seen in the hypoxia and soluble factor-induced skewing of macrophages and neutrophils to the M2 and N2 phenotypes, as well as the tolerization of dendritic cells that can lead to their active support of tumor growth.^55,56

Stromal cells are a heterogenous group of mesenchymal cells that are capable of differentiating into fibroblasts, endothelial cells, and adipocytes.⁵⁷ While cancer-associated stromal cells have been traditionally linked to tumor progression due to their role in tumor angiogenesis, neovascularization, malignant transformation, metastasis, maintenance of cancer cell stemness, and resistance of tumor cells to chemotherapy treatment, there is evidence that stromal cells could be important for tumor control.⁵⁷ Treatment of both rats and humans with bone marrow-derived mesenchymal stromal cells was shown to control the growth of colorectal cancers, mainly due to the increased infiltration of T cells and NK cells, and enhancement of macrophage phagocytic activity.⁵⁷

Cancer-associated fibroblasts, in particular, produce most ECM components, including collagens, which are the dominant forms of ECMs in most tumors.⁵⁷ ECM mainly specifies the mechanical properties of the tumors, with increased ECM deposition and cross-linking, altering tumor stiffness and viscoelasticity, respectively. Stiffness is a property that describes a tissue's resistance to deformation in response to applied stresses.⁵⁸ All tissues, including tumors, are also viscoelastic, a distinct feature from stiffness that confers a time-dependent response to deformation^58,59; the stress needed for a particular deformation decreases with time for a viscoelastic substrate, but remains the same for a perfectly elastic material.

The viscoelasticity of tissues, while previously underappreciated, is gaining increasing attention as growing evidence suggests it can independently influence cellular behavior.⁵⁹ Significant data suggest that the ECM may be important for both tumor control and progression. While some therapies that have sought to deplete the ECM have resulted in increased tumor aggressiveness,^60–62 changes in tumor mechanics have also been shown to result in epithelial to mesenchymal transition,^63,64 promote integrin and PI3K signaling,⁶⁵ regulate multiple transcription factors, including YAP/TAZ,⁶⁶ and influence cytoskeletal and nuclear dynamics,^67,68 resulting overall in cells with enhanced mobility, invasion, and metastatic potential.

T cell-mediated antitumor immunity

Tumor-reactive T cells are elicited when dead or dying cancer cells release tumor antigen, which are then picked up and processed by APCs for presentation and T cell priming in SLOs. Tumor antigens can be in the form of tumor-associated antigens (TAAs) or tumor-specific antigens (TSAs).⁶⁹ TAAs are expressed by both normal and malignant cells, but are overexpressed on the latter. TAA-based immune responses are limited by the possibility of central and peripheral T cell tolerance, since TAAs are expressed in normal cells.⁶⁹ TSAs, on the other hand, are neoantigens that arise from tumor-associated mutations and elicit T cells that are specific to the tumor.⁶⁹ However, TSAs tend to be patient specific and may be heterogeneous within the same tumor; as a result, a generalized approach for eliciting TSA-based T cell responses could be complex.

After being activated, tumor-reactive T cells exit SLOs; they migrate to the tumor site where they engage tumor cells in a variety of ways. CD4+ T cells mediate antitumor immunity by providing help to CD8+ T cells directly by expression of cytokines such as IL2, or indirectly by supporting APCs, including cross presenting dendritic cells.⁷⁰ CD4+ T cells also function by facilitating antibody production by B cells and producing effector molecules such as IFNγ and TNFα.⁷⁰

Although to a lesser extent, granzyme B+ CD4+ T cells have also been shown to directly kill MHC II-expressing tumor cells.⁷⁰ CD8+ T cells elicit antitumor responses mainly by directly killing tumor cells by granzyme and perforin secretion.⁷¹ Perforin causes pores to form on cell membranes, while granzyme B transports through those pores and elicits apoptosis in target cells.⁷¹ CD8+ T cells also secrete other effector molecules such as IFNγ and TNFα, and engage in Fas/FasL interactions with tumor cells that instigate cell death⁷¹ (Fig. 5).

FIG. 5.

T cells and the tumor microenvironment. The functional tumor microenvironment comprises antitumorigenic myeloid and lymphoid cells and normal stroma. In this study, T cells engage with and lyse tumor cells through the secretion of granzyme B, IFNγ, or TNF, as well as by Fas-fas ligand interactions. Dysfunctional tumor microenvironments are characterized by dysfunctional or protumorigenic immune cells and abnormal stroma. Tumors evade T cell action by secreting protumorigenic soluble factors, downregulating expression of MHC I and upregulating inhibitory ligands such as PDL1.

A number of studies have reported the presence of tertiary lymphoid organs (TLOs) in some tumors, including colorectal, breast cancer, and melanoma.⁷² TLOs are lymphoid organ-like structures that form under chronic inflammation and are characterized by the presence of stromal cells, B cells, T cells, and HEVs.⁷² Some studies have reported distinct T cell and B cell zones in TLOs. They are thought to potentially provide a local niche for immune response inside the tumor, similar to the immune responses that occur in SLOs.⁷² TLOs provide a localized niche for T cell priming and activation with an abundant source of antigen, and have been seen as a good prognostic indicator in many tumor settings.⁷²

T cell dysfunction in cancer

Tumors evade T cell control through multiple mechanisms, including T cell exhaustion, active immunosuppression, and altered ECMs (Fig. 5). T cell exhaustion due to chronic tumor antigen stimulation results in gradual loss of T cell function and concomitant expression of inhibitory receptors (Fig. 5). Tumor cells evade T cell recognition by loss of tumor antigens, downregulating the expression of or shedding MHC I, secreting immunosuppressive molecules such as IL10 and TGF-β, upregulating antiapoptotic molecules and overexpressing inhibitory ligands such as PDL1, which engage inhibitory receptors on T cells to shut down effector function.⁵⁶

In addition to tumor cells, myeloid cells can play active roles in T cell dysfunction. For instance, immature dendritic cells in the tumor microenvironment can present antigen to T cells without the appropriate co-stimulation, and as a result generate anergic or tolerant T cells.⁵⁶ Myeloid-derived suppressor cells, a heterogenous population of immature myeloid cells, have been shown to suppress T cell function by secreting immunosuppressive molecules such as Arg1, inducible nitric oxide synthase, indoleamine 2, 3-dioxygenase (IDO), and reactive oxygen species,^73–75 which can result in the inhibition of T cell function and even T cell apoptosis. Regulatory T cells also contribute to intratumoral T cell immunosuppression.

Conventional T cells are induced to become regulatory T cells in the presence of cytokines such as IDO, IL10, and TGF-β, which form part of the tumor microenvironment cytokine milieu during tumor progression. Regulatory T cells exert their effects, in part, by becoming a sink for IL2, an important cytokine for T cell maintenance and expansion, due to their high expression of the IL2 receptor CD25, as well as their ability to shed their CD25 receptor.^56,76 Regulatory T cells also secrete TGF-β and IL10, which induce more regulatory T cells and suppress dendritic cells, macrophages, and conventional T cell activity.⁷⁷

Regulatory T cells have also been shown to actively remove peptide-MHC class II complexes from dendritic cells, debilitating dendritic cell capacity as APCs.⁷⁸ Finally, alterations in ECM deposition and cross-linking is thought to impact T cell activity by serving as physical barriers to T cell infiltration, as well as promoting tumor cell invasion and metastatic potential. With changes in tumor mechanical environment being widely accepted as pivotal to both cancer progression and control, how ECM-mediated mechanical changes directly affect the phenotype and ultimate function of tumor-infiltrating T cells is still unknown.

Summary and Outlook

This review summarizes the process of T cell development in the bone marrow and thymus, T cell priming in SLOs, T cell activation, and effector function, as well as modes of T cell dysfunction. The specific roles of T cells in antitumor immunity, and strategies of tumor evasion were also discussed. To elicit durable antitumor immunity, recent work has focused on minimizing the effects of T cell exhaustion and the suppressive tumor environment by using checkpoint blockade inhibitors,⁵¹ eliciting host T cell responses by therapeutic cancer vaccination⁷⁹ and adoptively transferring in vitro engineered CAR T cells or TCR T cells.⁸⁰

To increase the safety of T cell therapies, there have been significant efforts to genetically engineer complex circuit systems that make T cells more responsive to their microenvironment,⁸¹ including synthetic notch systems, which enhance CAR T cell specificity by employing combinatorial gating strategies that require the presence of two antigens to elicit T cell responses.⁸² In developing these systems, it is important to strike a balance between increased efficacy and safety.

Biomaterial-based approaches to enhance T cell efficacy within the context of cancer have also gained significant interest. For instance, synthetic APCs have been developed to enhance in vitro T cell activation and proliferation by providing both the needed T cell activation ligands and paracrine signaling of bioinstructive soluble factors.⁸³ Macroscale biomaterials have also been developed to locally deliver T cells in vivo, while eliciting robust host T cell response by therapeutic cancer vaccination.⁸⁴ Nanoparticle-based biomaterial strategies have been developed to locally conjugate bioactive factors to T cells, which may otherwise be cytotoxic when administered systemically.⁸⁵

It is becoming increasingly clear that interdisciplinary approaches will be needed to develop the next generation of T cell therapies. For example, more advanced computational modeling approaches could lead to better prediction of MHC peptide-TCR interactions and lead to a more rational selection of candidate TCR clonotypes. Employing both biomaterial and genetic engineering approaches could lead to safe and efficient generation of CAR T cells in vivo. Importantly, basic T cell immunology research could benefit from advanced computational and bioinformatics approaches, which could in turn inform the design of engineering solutions.

Authors' Contributions

K.A.-B.: conceptualization, writing—original draft, and visualization. F.O.O.: writing—original draft and visualization. D.J.M.: supervision and writing—review and editing.

Disclaimer

The contents are those of the author(s) and do not necessarily represent the official views of, or an endorsement by, the funding agencies.

Author Disclosure Statement

No competing financial interests that directly relate to this work.

Funding Information

The authors acknowledge funding from the National Institutes of Health (R01 CA276459), the Food and Drug Administration (R01FD006589), and Wellcome Leap HOPE Program.