T follicular helper cell biology: A decade of discovery and diseases

SUMMARY

Helping B cells and antibody responses is a major function of CD4⁺ T cells. It has been 10 years since the publication of Bcl6 as the lineage defining transcription factor for T follicular helper (Tfh) differentiation and the requirement of Tfh cells as the specialized subset of CD4⁺ T cells needed for germinal centers (the microanatomical sites of B cell mutation and antibody affinity maturation) and related B cell responses. A great deal has been learned about Tfh cells in the past 10 years, particularly regarding their roles in a surprising range of diseases. Advances in the understanding of Tfh cells differentiation and function are discussed, as are the understanding of Tfh cells in infectious diseases, vaccines, autoimmune diseases, allergies, atherosclerosis, organ transplants, and cancer. This includes discussion of Tfh cells in the human immune system. Based on the discoveries to date, the next decade of Tfh research surely holds many more surprises.

eTOC blurb

Tfh cells are the CD4⁺ T cells specialized for helping B cells. Bcl6 was identified as the Tfh lineage defining transcription factor 10 years ago. Crotty reviews the extensive progress made in understanding Tfh cells since then, including surprising roles of Tfh cells in a range of diseases.

Introduction

While T cell help to B cells was one of the very first defined functions of T cells (Crotty, 2015), the identification of the CD4⁺ T cell subset responsible for T cell-dependent humoral immune responses developed much later. The term T follicular helper (Tfh) cell, was coined in a series of studies on human tonsillar germinal center (GC) CD4⁺ T cells and CXCR5⁺ CD4⁺ T cells in blood (Crotty, 2011). Early converts rightly point to those papers as seminal for establishing the field, but it was not until Bcl6 was determined to be a lineage defining transcription factor (TF) of Tfh cells that Tfh cells were broadly accepted among immunologists as a distinct lineage of helper CD4⁺ T cells and required for GCs. That trifecta of papers was published 10 years ago (Johnston et al., 2009; Nurieva et al., 2009; Yu et al., 2009), including the demonstration that the TFs Blimp1 and Bcl6 are potent reciprocal antagonists (Johnston et al., 2009) (Figure 1). Now is a good time to review what has been accomplished over the past decade of Tfh research. By any metric, the field of Tfh research has exploded in that time. Not all of the discoveries can be covered here. This review will highlight key findings in major areas of research that provide insights into what Tfh cells do, how they do it, and how proper and improper Tfh biology impacts diseases.

Figure 1. Features of Tfh cells.

Tfh cell differentiation is induced by DCs in the presence of antigen and signals from the microenvironment. Tfh cell versus non-Tfh cell differentiation bifurcates early, based on the balance of inductive and inhibitory signals. Some intermediate cell fates are also possible (grey). Early Tfh cells then subsequently differentiate to GC-Tfh cells.

Molecular and cellular biology of Tfh cells

Tfh differentiation usually starts with interaction of a naive CD4⁺ T cell with a myeloid professional antigen-presenting cell (APC) such as a dendritic cell (DC). Tfh differentiation can largely be modelled as an early cell fate decision between becoming a Tfh cell and a non-Tfh effector cell, e.g. Th1, Th2, or Th17 cell. The Tfh versus non-Tfh cell fate decision is predominantly made within the first two cell divisions (Choi et al., 2011; 2013; DiToro et al., 2018). Tfh cells stay resident in lymph nodes (LNs) and spleen because their purpose is to help B cells, while non-Tfh effector cells such as Th1, Th2, Th17, cytotoxic CD4 T cells, or Th9 cells are destined to predominantly leave lymphoid tissue and traffic to sites of infection or inflammation, as actions at those sites are the primary purposes of most non-Tfh cells. Interleukin-6 (IL-6) and the costimulatory molecule ICOSL are early factors involved in Tfh differentiation, in mice (Crotty, 2014; Vinuesa et al., 2016). IL-2 is the most potent inhibitor of Tfh differentiation, and IL2R signaling can be dose limiting for Tfh differentiation (Ballesteros-Tato et al., 2012; DiToro et al., 2018; Johnston et al., 2012). This is due to IL-2 induction of Blimp-1 expression and signal transducer and activator of transcription-5 (STAT5) activity (Johnston et al., 2012) and is associated with metabolic differences between Tfh and non-Tfh cells (Ray et al., 2015). T cell receptor (TCR) signal strength plays an important role in Tfh differentiation (DiToro et al., 2018; Fazilleau et al., 2009; Tubo et al., 2013), but the relationship appears to be non-linear (Tubo et al., 2013) and likely depends on which alternative differentiation pathway is available (e.g. Th1 or Th2), the cytokine environment, and the strength of available costimulatory receptor signaling (Figure 1).

Tfh differentiation can be primed by several classes of conventional DCs (Kato et al., 2015; Krishnaswamy et al., 2017) (Figure 2, 3). Murine monocytes can affect Tfh differentiation via cytokine production (IL-6) but not as APCs (Chakarov and Fazilleau, 2014). The diversity of DCs capable of priming Tfh differentiation likely reflects the fact that antibody (Ab) responses are of value for protection against almost all viral, bacterial, fungal, and parasite pathogens (Figure 4), and are of value at all body surfaces and in most body organs.

Figure 2. Signals to and from Tfh cells.

Tfh cells receive signals from a number of cell types that can regulate Tfh differentiation and/or function. Tfh cells function by providing help to B cells at a series of stages of antigen-specific B cell differentiation. B_Act, activated B cells. B_GC, GC B cells. B_PC, PC. B_pb, plasmablasts. B_Mem, memory B cells.

Figure 3. Tfh and B cell response kinetics and interactions.

A. A time series of Tfh differentiation and function. Details of the steps are summarized in the main text. For simplicity, only one TF is indicated in each cell. TCF1 expression is retained in Tfh cells and not non-Tfh effector cells. An animated versions is included. B. Tfh functions and signals that regulate B_GC cycling, B_GC➔B_PC differentiation, and B_GC➔B_Mem differentiation. “Int” = intermediate.

Figure 4. Relationships of Tfh cells to diseases.

Over the past decade Tfh cells have been associated with a wide range of diseases, with the Tfh cells contributing protective or pathogenic roles in different contexts, discussed in the main text.

DC priming is generally required for Tfh differentiation but is not sufficient. Tfh differentiation is a multistage, multifactorial process. Normal Tfh cell differentiation usually requires two APCs: DCs and B cells (Crotty, 2014). DCs are generally required for early Tfh differentiation, while B cells are required for later events and full GC-Tfh maturation (Figure 3A, Movie). Similar signals are provided by both DCs and B cells (Figure 2), but B cells are usually incapable of priming CD4⁺ T cell responses because of the rarity of antigen-specific (and thus antigen-presenting) B cells early in an immune response, in contrast to DCs. Additionally, DCs predominantly localize to T cell zones and thus have the greatest opportunity to prime CD4⁺ T cells. Nevertheless, B cells can prime Tfh cell responses under certain conditions (Hong et al., 2018). B cells are excellent producers of a wide range of cytokines, including IL-6 which can facilitate murine Tfh cell differentiation (Karnowski et al., 2012).

B cells are outstanding APCs for Tfh cells downstream of the initiating event for Tfh differentiation. Antigen-specific B cells undergo exponential expansion and thus become more abundant APCs; activated B cells also upregulate major histocompatibility complex class II proteins (MHCII) and costimulatory molecules, such as CD80 and CD86, which make the activated B cells more potent APCs for Tfh cells. Additionally, activated B cells migrate to the T cell:B cell (T-B) border to enhance their likelihood of interaction with Tfh cells (Figure 3A).

Upon priming by DCs, CD4⁺ T cells receiving Tfh cell-inductive signals upregulate Bcl6 (Choi et al., 2011; DiToro et al., 2018). This facilitates expression of the chemokine receptor CXCR5 and repression of CCR7 among other molecules, allowing for migration to the T-B border. These differentiating Tfh cells migrate and interact with B cells in an ICOS:ICOSL and peptide:MHC-dependent manner. The additional signals from cognate B cells drive further GC-Tfh differentiation and the formation of GCs. The migration of Tfh cell into GCs is facilitated by repression of the migration-associated receptor PSGL1 and the chemoattractant receptor Ebi2, and changes in the expression of S1P receptors, SLAM family receptors, and integrins (Meli et al., 2016; Vinuesa et al., 2016). ICOS is particularly fascinating, as it functions as both a costimulatory molecule and a migration receptor for Tfh cells (Pedros et al., 2016; Xu et al., 2013). Signaling through ICOS is complex (Leavenworth et al., 2015; Weber et al., 2015) and is a major factor for Tfh differentiation in many contexts, including an elegant study elucidating transformational progression of Tfh cells to angioimmunoblastic T cell lymphoma (AITL) (Cortes et al., 2018).

The TF Bcl6 is essential for Tfh differentiation, and constitutive ectopic expression of Bcl6 enhances Tfh differentiation (Hollister et al., 2013; Johnston et al., 2009; Nurieva et al., 2009; Yu et al., 2009). Bcl6 acts on multiple aspects of Tfh differentiation and function (Hatzi et al., 2015). Most available data indicate that Bcl6 is an obligate repressor of gene expression (Béguelin et al., 2016; Hatzi et al., 2015; 2013). Multiple TFs in addition to Bcl6 play important roles in Tfh differentiation (Vinuesa et al., 2016). TCF1 and LEF1 are involved in induction of Bcl6, as are STAT3 and STAT1. Inhibition of Prdm1 (Blimp1) is required for Tfh differentiation and is accomplished by Bcl6 (Figure 1) in coordination with TCF1 and additional factors (Choi et al., 2015; Wu et al., 2015; Xu et al., 2015). CXCR5 expression, as one example Tfh cell gene, is dependent on E protein TFs such as E2A and Ascl2. E protein activity is inhibited by Id2, and Id2 is inhibited by Bcl6 (Shaw et al., 2016). E2A is present in resting and early-activated CD4⁺ T cells, with Ascl2 becoming highly expressed in GC-Tfh cells (Liu et al., 2014; Shaw et al., 2016). More encyclopedic reviews of TFs involved in Tfh differentiation are available (Crotty, 2011; Vinuesa et al., 2016). While major advances have been made in our understanding of the transcriptional and epigenetic regulation of Tfh cell differentiation and function, much remains to be learned.

Most GC-Tfh cells are quite similar in their gene expression profile, indicating that ‘vanilla’ Tfh cells predominate in vivo (Crotty, 2018); nevertheless, substantial phenotypic and functional heterogeneity exists among Tfh cells in disease states, which may be important causes or effects of those diseases (Crotty, 2014) (see below). It was clear from early on that Tfh cells can express molecules associated with other CD4⁺ T cell lineages, such as the TF Tbet and IFNγ (Johnston et al., 2009); however, transcription of those genes is normally potently repressed by Bcl6 in fully differentiated GC-Tfh cells (Crotty, 2014; Hatzi et al., 2015). Additionally, it is unclear what productive biological function those molecules have in GC-Tfh cells, if any (see Infectious Diseases section below).

Human and mouse GC-Tfh cells are similar in surface protein marker phenotype, gene expression profile, and BCL6 expression, indicating robust evolutionary conservation of many aspects of Tfh cell biology; however, cytokines involved in induction of Tfh differentiation appear to be substantially different between the two species. IL-6 is a potent inducer of murine Tfh cell differentiation, and is the most potent inducer of IL-21 expression by newly activated naive murine CD4⁺ T cells (Crotty, 2014); however, the role of IL-6 in human Tfh differentiation is far less clear. IL-6 has little impact on IL-21 induction in human CD4⁺ T cells in vitro. IL-12 is the most potent human cytokine at inducing IL-21-expressing CD4⁺ T cells, and IL-12 signals through different STATs than IL-6 (Schmitt et al., 2014). Curiously, there remains no reproducible in vitro system for murine Tfh cell differentiation. In contrast, in vitro human Tfh cell differentiation can be accomplished via coculture with TGFβ + IL-12, or the cytokines activin A + IL-12, with essentially no impact of IL-6 under equivalent conditions (Locci et al., 2016; Schmitt et al., 2014). TGFβ + IL-12 or activin A + IL-12 do not differentiate murine Tfh cells in vitro (Locci et al., 2016). Human in vitro Tfh cell differentiation is most optimal when IL-2 is blocked (Locci et al., 2016), consistent with IL-2’s potent inhibitory activity repressing murine Tfh differentiation. While substantial differences between human and murine Tfh cells are apparent, conclusions about the magnitude of the differences are clouded by the fact that there are multiple pathways to Tfh cell differentiation. In mice, IL-27 can largely substitute for IL-6 (when expressed) (Harker et al., 2013; Vijayan et al., 2016) and the IL-12 family member IL-23 can largely substitute for IL-12 to induce IL-21 and human Tfh cell differentiation in vitro (Schmitt et al., 2014). This is consistent with the concept that in vivo pathogen detection mechanisms need to induce Tfh cell differentiation to diverse pathogens.

Human genetics has provided valuable insights into Tfh cell biology. While robust conclusions are frequently limited by the rarity of donors, there is substantial evidence of Tfh cell-intrinsic defects in a number of human primary immunodeficiencies, including ICOSL, ICOS, SH2D1A, STAT3, IL21, IL21R, IL12RB1, and PI3KD (Ma et al., 2015; Preite et al., 2018).

Interactions Between Tfh and B cells

Tfh cells depend on B cells in most contexts, and GC B (B_GC) and plasma cells (PCs) depend on Tfh cells. These feedback loops are central aspects of proper regulation of humoral immunity. Early in Tfh cell differentiation, the Tfh cells migrate to the T-B border and interact with antigen-specific B cells (Figure 3A). Most B cell responses cannot progress beyond this point without Tfh cell help; the B cells most likely die within 24h of antigen recognition in the absence of T cell help (Akkaya et al., 2018). Tfh cells provide IL-21 and CD40L signals required for B cell proliferation and differentiation towards both GC and extrafollicular fates (Lee et al., 2011) (Figure 3A). IL-4 expression by Tfh cells is only acquired later, upon full differentiation to GC-Tfh cells (Weinstein et al., 2016; Yusuf et al., 2010). B cells compete for Tfh help at the T-B border based on the amount of p:MHCII presented by the B cell to the Tfh cell (Schwickert et al., 2011; Yeh et al., 2018). The success of B cells competing for early Tfh help depends on the frequency of the B cells and their antigen affinity (Abbott et al., 2018; Schwickert et al., 2011). Rare B cells with low affinity may be excluded from immune responses at this early Tfh cell checkpoint, which has implications for vaccines because neutralizing epitope-specific B cells against some pathogens may be quite rare and low affinity (Havenar-Daughton et al., 2017).

GCs are the microanatomical sites of B cell mutation and antibody affinity maturation. GC responses are evolution in miniature. All evolutionary processes require generations (GC cell division), mutation (SHM), and limiting resources that drive evolutionary selection. Tfh help is a key limiting resource in GCs. B_GC cells compete for GC-Tfh help. GCs consistent of a light zone (LZ) containing GC-Tfh cells, follicular dendritic cells (FDCs), and B_gc cells, and a dark zone (DZ) of proliferating B_GC cells (Figure 3A). Selection occurs in the LZ. FDCs organize the microanatomic structure of GCs and retain and provide antigens to the B_GC cells (Heesters et al., 2013; 2016). Deletion of FDCs rapidly impairs GCs (Wang et al., 2011). B_GC cells obtain intact antigen from the surface of FDCs by B cell receptor (BCR)-mediated endocytosis and process the antigen into peptides for surface presentation in p:MHCII complexes. GC-Tfh cells preferentially provide help to B_GC cells with higher expression of p:MHCII, which serves as an indirect measurement of higher BCR antigen affinity. B_GC cells that do not receive GC-Tfh help die. B_gc cells helped by GC-Tfh cells then migrate to the DZ and undergo rounds of division and division-linked SHM commensurate with the magnitude of the GC-Tfh cell help each B_GC cell receives (Ersching et al., 2017; Gitlin et al., 2014) (Figure 3A). One model is that GC-Tfh cell help is the only B_GC cell selecting event in GCs. If that is the case, changes in MHCII expression by B cells should alter GC selection. However, B cells with MHCII haploinsufficiency compete in GCs equivalently to WT cells under conditions of physiological antigen concentrations (Yeh et al., 2018), indicating that B cell competition for Tfh cell help predominantly occurs at the T-B border, not in the GCs. B_gc cells with alterations in MHCII degradation also have similar GC competitive fitness (Bannard et al., 2016). Thus, considering experimental systems avoiding supraphysiological stimuli, both BCR signaling and Tfh cell help signals appear to be integrated in B_GC cells to determine survival and proliferation in a GC (Bannard et al., 2016; Kräutler et al., 2017; Yeh et al., 2018). That model is further supported by findings that B_GC cells have rewired BCR and CD40 signaling. Naive B cells can induce Myc TF expression via CD40 or BCR signaling due to cross-regulation of NFκB and PI3K pathways; however, in B_GC cells those signaling pathways become siloed, such that CD40L:CD40 engagement triggers NFκB and BCR antigen signaling engages PI3K signaling (Luo et al., 2018) (Fig 3B). In part this is due to synaptic differences in B_GC cells, which provides more stringent affinity discrimination by B_gc cell BCR engagement (Nowosad et al., 2016).

The help provided by GC-Tfh cells to B_GC cells classically consists of IL-21, IL-4, CD40L, and CXCL13, in addition to production of more widely expressed cytokines such as IL-2 and TNF. The combination of IL-21 and IL-4 expression by GC-Tfh cells maximally supports B_GC cells (Weinstein et al., 2016), in addition to CD40L. While GC-Tfh cells express substantial amounts of IL-21 and IL-4 RNA, GC-Tfh cells secrete infinitesimal quanta of IL-21 and IL-4 protein, consistent with GC-Tfh cells constraining GCs as a resource limited environment (Dan et al., 2016; Havenar-Daughton et al., 2016c). GC-Tfh cells may also directly kill B_GC cells via FasL-Fas (Butt et al., 2015); however, GC-Tfh cells express little FasL (Bentebibel et al., 2011) and a role of GC-Tfh FasL has not been directly demonstrated in GC selection.

One substantial difference in Tfh cell help between mice and humans is CXCL13 expression (Figure 3B), discussed later. A second substantial difference is the surprising expression of the neurotransmitter dopamine by human GC-Tfh cells (Papa et al., 2017). Dopamine secretion by human GC-Tfh cells enhances rapid ICOSL surface expression by human B_GC cells, resulting in a feedforward help loop between the interacting GC-Tfh and B_GC cells (Figure 3B). No equivalent dopamine-related process is known in mice (Papa et al., 2017). IL-10 also appears to have species-specific functions. IL-10 is made by some human GC-Tfh cells, and IL-10 has long been known as a PC differentiation factor for human B cells (Bryant et al., 2007). It is unclear if this biology is conserved in mice. IL-10 is almost exclusively an immunosuppressive cytokine in mice, but there are at least certain cases of IL-10⁺ murine Tfh cells providing help to B cells (Xin et al., 2018).

GC-Tfh cells not only control selection of high affinity B_GC cells, GC-Tfh cells also are critical regulators of the development of long-term humoral immunity via B_GC cell differentiation to memory B cells and PCs (Figure 3). Early GC-derived plasmablasts (PBs, differentiating PCs still expressing Ki67) are induced by IL-21 from GC-Tfh cells, in synergy with IL-6 provided by stromal cells (Zhang et al., 2018). GC-Tfh cell induction of long-lived PCs requires IL-21. Differentiation of PCs from B_GC cells is a multi-step transcriptional reprogramming process requiring reduction of Bcl6, increase of the TF Irf4, and subsequent expression of Blimp1 (Ise et al., 2018; Kallies et al., 2007). Most GC-Tfh cell interactions with B_GC cells are brief (minutes) (Liu et al., 2015a; Shulman et al., 2014). Regulation of CD40L availability to B_GC is sophisticated in Tfh cells. Strong CD40L signaling from GC-Tfh cells to B_gc cells can transiently induce enhanced SLAM (CD150), ICAM-1, and CD40 expression by a LZ B_GC cell (Ise et al., 2018) (Figure 3B). Those B_GC cell surface changes can cause extended GC-Tfh:B_GC cell synaptic engagement, which can result in elevated Irf4 and reduced Bcl6 expression. This may be facilitated by LFA-1 adhesion enhancing GC-Tfh gene expression (Meli et al., 2016), or feedforward T:B engagement by ICOS enhancing GC-Tfh cell surface CD40L (Liu et al., 2015a). The process may also involve direct BCR signaling (Kräutler et al., 2017). GC-Tfh cell IL-21 is then likely required to induce Blimp1 expression in Irf4^hiBcl6^lo early PCs to complete cell fate commitment (Ise et al., 2018; Kräutler et al., 2017) (Figure 3B). Selection of memory B cells appears to involve converse B_GC cell characteristics. A fraction of LZ B_GC cells with lower affinity are prone to progress to resting memory B cells in a Bach2 TF-dependent manner, associated with CCR6 expression (Shinnakasu et al., 2016; Suan et al., 2017; Wang et al., 2017). GC➔memory B cell differentiation appears to occur when antigen-specific B_GC cells receive moderate help signals from GC-Tfh cells (Figure 3B), which are stochastically more likely to occur to B_gc cells with low or intermediate affinity in a GC population (Shinnakasu et al., 2016; Suan et al., 2017). While impressive advances have been made in understanding how GC-Tfh cells control GCs—even when focusing on work in the past two years—much remains to be learned, in part due to the sophisticated reciprocal signaling between the cells and the inherent challenges of studying the molecular biology of rapid and repetitive interactions of GC-Tfh and B_GC cells in vivo. While current mechanistic models are appealing, and are based on sophisticated experimental approaches, it is unclear if they are general principles or if they selectively apply to particular types of antigens and adjuvants. For example, while most experimental systems have been chosen in the past based on their ease of use and robust GC and Ab responses, the outcome of an infection or immunization can change dramatically depending on the precursor frequency and affinity of the antigen-specific B cells (Abbott et al., 2018; Tarlinton, 2019).

The study of memory recall responses have more limited data, but data suffice to conclude that memory Tfh cells and memory B cells rapidly interact upon re-exposure to antigen (Ise et al., 2014; Suan et al., 2015). IL-9 appears to have a role in memory B cell recall, but there are conflicting accounts (Takatsuka et al., 2018; Wang et al., 2017).

There exist Tfh cell functions outside the GC. Some of these functions are to help B cells or memory B cells at the T-B border early in an immune response. However, those same functions may occur at other times, and GC-Tfh cells can migrate in and out of GCs (Shulman et al., 2013). Tfh cells outside of GCs express PSGL1 and may have heterogenous phenotypes (Bentebibel et al., 2011; Kim et al., 2018b; Suan et al., 2015). In addition to Tfh cells at the T-B border, there can also be Tfh cells that downregulate CXCR5 and become extrafollicular Tfh cells, distant from the follicle or T-B border. In such cases one can distinguish fTfh (follicle, T-B border, and interfollicular zone Tfh cells) and eTfh cells (extrafollicular), if histological data support the conclusion. This nomenclature highlights the ontological relatedness of the cells (both depend on Bcl6 and express IL-21). A detailed gene expression profiling and histological examination of eTfh cells is lacking and would provide additional insights into their relationship to GC-Tfh cells and Tfh-like cells found outside of lymphoid tissues. Memory Tfh cells are an additional population of Tfh cells. Memory Tfh cells appear to be derived from GC-Tfh cells or mTfh cells (Crotty, 2014; Tubo et al., 2016) and constitute a major fraction of human memory CD₄⁺ T cells.

Tfh Cells in Infectious Diseases

The primary function of Tfh cells is to provide protection from infectious diseases. Tfh cells facilitate Ab responses to viral, bacterial, parasite, and fungal infections (Figure 4). As one example, the IgG response to vaccinia virus infection is reduced 98% in the absence of Tfh cells (Xiao et al., 2014). The importance of Tfh cells for Ab responses to infections has been demonstrated in mice, non-human primates (NHPs), and humans (Crotty, 2014). Tfh cells are required for GC responses to infections, as well as at least some pre-GC extrafollicular Ab responses (Lee et al., 2011). Tfh cells have been shown to be important for control of chronic lymphocytic choriomeningitis (LCMV) infection via support of GCs and the slow development of neutralizing Abs (nAbs) over time (Greczmiel et al., 2017; Harker et al., 2011). Similar findings on the importance of Tfh cells for protective Abs have been made for chronic malarial infections (Butler et al., 2012; Ryg-Cornejo et al., 2016).

Antigen-specific GC-Tfh cells are difficult to study because they are exclusively located in lymphoid tissue and they secrete small quantities of cytokines (Dan et al., 2016; Havenar-Daughton et al., 2016c). The highly constrained cytokine secretion may be a direct effect of Bcl6 expression (Hatzi et al., 2015) and is consistent with the primary function of GC-Tfh cells, which is to preferentially help B_GC cells during cognate interactions. Extensive secretion of cytokines throughout a GC would likely be counterproductive to the competitive evolution of B_gc cell composition. As a result of the challenge of identifying antigen-specific GC-Tfh cells, most studies have focused on bulk GC-Tfh cells. Antigen-specific GC-Tfh cells have been studied in HIV, SIV, and Streptococcus pyogenes infections. S. pyogenes is the cause of recurrent tonsillitis (RT), a classic childhood disease. It was a longstanding mystery as to why some children are susceptible to RT. Predisposition to RT disease is associated with particular HLA class II alleles and significantly fewer GC-Tfh cells (Dan et al., 2019). An aberrant GC-Tfh cell population is present in RT that expresses granzyme B, and the granzyme B⁺ GC-Tfh cells are capable of killing B cells (Dan et al., 2019). S. pyogenes targeting of human GC-Tfh cells is an unexpected immune evasion strategy.

Cytotoxic gene expression and Tfh cell-associated gene expression were thought to be exclusive to different differentiation pathways. However, CXCR5⁺ CD8⁺ T cells (‘follicular CD8⁺ T cells’) co-opt Tfh cell-associated gene expression, including Bcl6, in a TCF1-dependent manner, while maintaining some cytotoxic CD8⁺ T cell functionality (Im et al., 2016; Leong et al., 2016). TCF1 is a key regulator of Tfh differentiation. Id2 inhibits CXCR5⁺ CD8 T cell development (He et al., 2016b; Leong et al., 2016), and Id2 inhibits Tfh differentiation (Shaw et al., 2016). The CXCR5⁺ follicular CD8⁺ T cells may be critical for controlling pathogens located in B cell follicles or infecting Tfh cells, such as EBV and HIV.

HIV- and SIV-specific GC-Tfh cells have been studied both because HIV can infect Tfh cells (Banga et al., 2016) and because Tfh cells are positively associated with broad HIV nAb responses (Locci et al., 2013; Moody et al., 2016). Env-specific GC-Tfh cells develop in SIV infection and are associated with Env-specific IgG responses (Yamamoto et al., 2015). A substantial fraction of GC-Tfh cells in SIV-infected monkeys are CXCR3⁺ (Velu et al., 2016). CXCR3 expression is primarily regulated by Tbet and is a signature gene of Th1 cells. Many GC-Tfh cells in SIV⁺ animals are Tbet⁺, and IFNγ⁺ by histology, and thus the cells co-express Tfh and Th1 cell-associated gene expression programs and can be considered GC-Tfh1 cells (Velu et al., 2016). The consequences of Tbet⁺ IFNγ⁺ GC-Tfh cells is unclear. While IFNγ regulates IgG2 class switch recombination (CSR) in mice, IFNγ is not an important regulator of class switch in humans (Crotty, 2014), and thus the mouse model may be largely uninformative about Tbet biology in human Tfh cells. Tbet does not appear to block human Tfh differentiation (Schmitt et al., 2016), and the impact of Tbet and IFNγ on Tfh cell help to B cells appears to vary depending on the in vitro culture conditions (Morita et al., 2011; Schmitt et al., 2016; Velu et al., 2016). CXCR3 expression by Tfh cells may drive localization away from GCs and towards the T-B border, where CXCR3 ligands are common during inflammatory conditions. Such a location may bias B cell responses towards non-GC destinies, such as PCs. CXCR3⁺ Tfh have been found to be poor helpers of B cells in a mouse malarial model, and the impairment was reversed by blocking IFNγ, resulting in improved Tfh cell localization, GCs, and Abs (Ryg-Cornejo et al., 2016). The inhibitory receptor PD1 inhibits expression of CXCR3 by Tfh cells, which is important for proper localization to follicles (Shi et al., 2018). Much remains to be learned about the biology of antigen-specific GC-Tfh cells to pathogens, particularly in humans.

T follicular regulatory cells (Tfr. Foxp3⁺Bcl6⁺CXCR5⁺CD4⁺) are a subset of natural Treg (nTreg) cells that express Tfh cell-associated genes (Sage and Sharpe, 2016). Absence of Foxp3⁺ Tfr cells results in a surprisingly mild impact on GC and Ab responses to infections. Anti-influenza Abs and B_GC cells are essentially identical in the presence or absence of Tfr cells (Botta et al., 2017; Fu et al., 2018). Similarly mild effects have been observed in Tfr-deficient mice administered immunizations (Wu et al., 2016). In contrast, dramatic effects and rapid lethality are observed in Foxp3-deficient mice when total Treg cells are absent. The data from studies of infections indicate that the primary physiological role of Tfr cells is to limit the development of autoreactive B cells (Botta et al., 2017). While Tfr cells are uncommon early (d10) in influenza or LCMV infection, they accumulate substantially over time (d30) and prevent the development of autoreactive (anti-histone) B cells late in GC responses (Botta et al., 2017).

In addition to facilitating protective immunity against pathogens, Tfh cells are important regulators of the microbiota (Figure 4), via the importance of Tfh cells for generation of high affinity IgA responses in GC-rich Peyer’s patches against pathobionts (Bunker et al., 2015; Kubinak et al., 2015; Proietti et al., 2014; Wei et al., 2011) (though not IgA CSR itself (Reboldi et al., 2016)). Vice versa, gut microbiota also affect Tfh cell biology (Macpherson, 2018; Preite et al., 2018). Gut microbiota such as segmented filamentous bacteria (SFB) that reduce IL-2 responses enhance Tfh cell responses (Teng et al., 2016). There are likely multiple mechanisms by which variation in gut microbiota can modify Tfh differentiation or function, and this is an area ripe for future study, given the broad relevance of microbiota in immune system physiology.

Tfh Cells and Vaccines

The important roles of Tfh cells in immune responses to vaccines parallels their roles in development of immune responses to pathogen infections. Tfh cells are important for GC, Ab, and long-lived PC responses to protein immunizations, including the process of affinity maturation (Crotty, 2014; Victora and Nussenzweig, 2012). Understanding the roles of Tfh cells in the development of highly affinity matured Ab responses to vaccines, and in the development of long-lived memory to vaccines, has substantial promise for improving future vaccines.

The role of Tfh cells in human vaccines has been most extensively studied in the context of annual influenza vaccine immunizations. GC-Tfh cells are not present in blood, but a population of recently activated Tfh cells (CXCR5⁺ICOS⁺, also Ki67⁺PD1^hi) is present in blood. Tfh cells in blood (CXCR5⁺) are commonly termed circulating Tfh (cTfh) cells. Resting memory Tfh cells recirculate regularly through blood, like other central memory CD4⁺ T cells, and therefore it is important to distinguish activated cTfh cells (CXCR5⁺ICOS⁺ or Ki67⁺PD1^hi) from resting memory Tfh cells (T_CM-Tfh resting CXCR5⁺CD45RO⁺ cells) in human blood cTfh cell studies (Heit et al., 2017). Resting memory Tfh cells represent the majority of cTfh cells in blood (Locci et al., 2013). Activated cTfh cells are most likely derived from GC-Tfh cells (Heit et al., 2017; Locci et al., 2013) or fTfh cells (He et al., 2013; Tsai and Yu, 2014). Flu vaccination recalls an oligoclonal memory Tfh cell response and flu-specific activated cTfh cell frequencies increase sharply at d7 (Bentebibel et al., 2013; Herati et al., 2017). Antigen-specific activated cTfh cells in human blood are also observed in response to an adenoviral vector (Heit et al., 2017) and correlated with anamnestic IgG responses. The large majority of flu-specific activated cTfh cells are CXCR3⁺ and CXCR3⁺ cTfh cells correlate with acute anti-influenza IgG (Bentebibel et al., 2013); however, CXCR3⁺ memory cTfh cells are particularly poor providers of B cell help in vitro (Locci et al., 2013; Morita et al., 2011). The observation that most flu-specific activated cTfh cells are CXCR3⁺ may be connected with the poor efficacy of the flu vaccine. The annual flu vaccine is frequently only ~30% protective and is associated with poor quality Ab responses. CXCR3⁺ cTfh cells may preferentially provide help for short-lived PB differentiation. It is possible that flu vaccines would be more efficacious if they successfully elicited Tfh cells with other properties.

Adjuvants are key components of vaccines. Live attenuated vaccines are self-adjuvanted, but other vaccines generally depend on adjuvants. Aluminum salts (alum) are the most common adjuvant in human vaccines and have been used for almost 100 years (McKee and Marrack, 2017; Plotkin et al., 2018). The mechanisms of adjuvanticity of alum remains controversial. Alum does stimulate Tfh cell responses, but relatively poorly (Preite et al., 2018; Riteau et al., 2016). Oil-in-water adjuvants, such as MF59, are also used in human vaccines and induce Tfh cell-dominated CD4⁺ T cell responses via IL-6 stimulation of Bcl6 expression, at least in mice (Mastelic Gavillet et al., 2015; Riteau et al., 2016). The direct mechanism(s) of action of oil-in-water adjuvants remains unclear, but likely involves endogenous danger signals or stress responses (Riteau et al., 2016; Vono et al., 2013). NKT adjuvants are promising and can enhance Tfh cell responses (Polonskaya et al., 2017). The primary mechanism of action of NKT adjuvants may be the provision of early cytokines such as IL-4 or IL-21 by NKT cells to B cells prior to the availability of IL-4 and IL-21 by Tfh cells (Chang et al., 2012; Gaya et al., 2018; King et al., 2012). There is extensive interest in the use of toll-like receptor (TLR) agonists as vaccine adjuvants. TLR4, TLR6, TLR7, TLR8, and TLR9 agonists can all enhance Tfh cell and Ab responses to protein antigens compared to alum (Kasturi et al., 2011; Leroux-Roels et al., 2016; Moon et al., 2012), and human vaccines containing TLR4 and TLR9 agonists are now available. However, it largely remains unclear if TLR agonist adjuvants preferentially induce Tfh cells or work via other mechanisms, such as direct effects on B cells and overall increases in CD4⁺ T cell response (Liang et al., 2017; Rookhuizen and DeFranco, 2014). In addition, it is unclear, on a case-by-case basis, if the mechanisms of action of adjuvants in mice are conserved in humans for Tfh cell and GC B cell biology. In the case of TLR8, ‘vita-PAMP’ agonists have a direct effect on human Tfh differentiation (Ugolini et al., 2018).

An additional aspect of adjuvants is antigen delivery. GC-Tfh cells require continuous antigen presentation for their maintenance (Baumjohann et al., 2013; Deenick et al., 2010). The magnitude of the Tfh cell response is also dependent on the antigen quantities presented by APCs to prime the CD4⁺ T cells. Thus, adjuvants and immunization methods that enhance CD4⁺ T cell priming, or sustain GC-Tfh cells, can substantially improve humoral immune responses (Cirelli et al., 2018; Tam et al., 2016).

Much remains to be learned about Tfh cell responses to vaccines and candidate vaccines and how they could be improved. A major challenge is the limitation of exclusively studying blood in most human and NHP studies. A fundamental attribute of GCs is that they do not exist in blood, they exist in lymphoid tissues. To understand GC-Tfh and B_gc cell biology it is essential to study lymphoid tissue. This is obviously challenging, and efforts to identify surrogate cell types and biomarkers of GCs and GC-Tfh cells in blood are worthwhile (Bentebibel et al., 2013; Havenar-Daughton et al., 2016b; Herati et al., 2017), but it is clear from many areas of immunological research that the cell types in tissue can be dramatically different than what is measured in blood. LN fine needle aspirates (FNA) provide one avenue for minimally invasive direct assessment of GC-Tfh cells and B_GC cells in human and NHP LNs (Havenar-Daughton et al., 2016a). Using this approach, it has been determined in a monkey immunization study that nAb development correlates with Tfh quality and the magnitude of the GCs in LNs (Havenar-Daughton et al., 2016a). LN GC magnitude predicted the magnitude of the nAb responses in a subsequent study (Pauthner et al., 2017). A major advantage of LN FNAs is that they can be performed longitudinally on the same LN, allowing for kinetic measurements of GC-Tfh and GC biology in individuals. A first-of-its-kind longitudinal study of antigen-specific B_GC cells and GC-Tfh cells in NHPs has revealed that B cell immunodominance is a major factor shaping the GC and Ab response to a candidate vaccine immunogen (Cirelli et al., 2018). LN FNAs are performed daily in most major hospitals for cancer pathology surveillance, and thus infrastructure is in place for utilizing LN FNAs for human GC investigation. This will hopefully result in improved understanding of human GC-Tfh cells and B_GC cells in diverse contexts in the coming years.

Tfh Cells in Autoimmune Diseases

Tfh cells are associated with a wide range of autoimmune diseases, both in mice and humans. These fall into three categories: autoantibody-mediated autoimmune diseases, autoantibody-associated autoimmune diseases for which there is no clear causal role for autoantibodies, and autoimmune diseases not associated with autoantibodies. While associations of Tfh cells with autoantibody-mediated autoimmune diseases are expected, Tfh cell associations with other autoimmune diseases have been more surprising (Figure 4). Type 1 diabetes (T1D) is an example of the second category, where disease can be transferred by Tfh cells in a mouse model (Kenefeck et al., 2015). This may be relevant to humans, as increased T1D antigen-specific cTfh (Serr et al., 2016) and activated cTfh (PD1^hi) cells are seen in children with recent onset T1D or at risk for T1D (Viisanen et al., 2017). Separately, Tfh cells have also been associated with multiple sclerosis in an EAE mouse model (Quinn et al., 2018). Due to space constraints, I will focus on two major autoimmune diseases here with Tfh cell involvement, SLE and RA.

In humans, perhaps the clearest connection between Tfh cells and an autoantibody-mediated autoimmune disease is systemic lupus erythematosus (SLE). A clear causal role for Tfh cells in disease progression in mice has been shown for autoantibody-associated autoimmunity in the sanroque SLE model (Linterman et al., 2009) and related models (Gensous et al., 2018). SLE can be considered an autoimmune disease for which the autoantibodies have an active role. In humans, the majority of autoantibody-producing B cells are somatically mutated (Tiller et al., 2007), indicating that they are likely GC-derived. Frequencies of activated cTfh cells in blood are increased in SLE patients (Gensous et al., 2018; He et al., 2013). Low dose IL-2 is being considered as a therapeutic treatment for SLE, on the basis that IL-2 is a potent inhibitor of Tfh cells (He et al., 2016a). Lupus nephritis is one of the most severe manifestations of human SLE, frequently requiring kidney transplantation, and ectopic GCs and Tfh cells are frequently found in lupus nephritis patient kidneys (Chang et al., 2011). Thus, it appears that Tfh cells have a pathogenic role in SLE. Interestingly, mice with Ets1-deficient Tfh cells develop spontaneous SLE (Kim et al., 2018a). The Ets1-deficient Tfh cells co-express a Th2 cell-associated gene expression program (Bcl6⁺GATA3^hi) and are thus Tfh2 cells (Kim et al., 2018a). In contrast, Tfh2 cells were not observed in wild-type mice, implying that Tfh2 may be specifically pathological. IgE CSR is dependent on Tfh cells, not Th2 cells or Tfh2 cells, in helminth infections, as shown via deletion of Gata3 or Il4 in Tfh cells (Liang et al., 2012; Meli et al., 2017) (see Allergies, below). Ets1-deficient mice with SLE have greatly exacerbated IgE synthesis (Kim et al., 2018a), and anti-dsDNA IgE is associated with human SLE (Henault et al., 2016). Ets1 simultaneously inhibits signature Tfh- and Th2-associated genes, providing a potential mechanism for Tfh2 cell development. IL-4⁺IL-13⁺CXCR5⁺ Tfh2 cells are present in SLE patients, but are far less abundant in healthy individuals (Kim et al., 2018a). Thus, Tfh cells may be part of the pathogenesis of human SLE via aberrant Tfh2 cell differentiation resulting in unusual anti-DNA IgE. Additionally, human SLE is associated with ectopic lymphoid structures (ELS), which may directly contribute to the disease. Relationships between Tfh cells and ELS are discussed more below.

IL-21 and CD40L are both key Tfh help molecules, and both proteins have been associated with SLE disease. IL-21 or Tfh cells have been shown to be involved in SLE disease pathogenesis in several animal models with relatively diverse disease presentations (Choi et al., 2017; Kim et al., 2017), consistent with Tfh cells being important for SLE development in a range of contexts. In a pristane SLE mouse model, the nucleotide receptor P2rx7 (which is highly expressed on Tfh cells) regulates pathogenic Tfh cells that drive autoantibody responses, consistent with disrupted responsiveness to P2RX7 by cTfh cells from SLE patients (Faliti et al., 2019). Lastly, engagement of OX40 on human CD4⁺ T cells enhances BCL6 expression and IL-21 expression, and OX40L⁺ myeloid cells positively correlated with cTfh cell frequencies in SLE patients (Jacquemin et al., 2015).

Rheumatoid arthritis (RA) is strongly associated with autoantibodies, particularly against modified-self citrullinated proteins (Smolen et al., 2018). RA pathogenesis is multifactorial, involving many cell types and multiple organs. ELS in joints are associated with disease in a substantial subset of RA patients (Bombardieri et al., 2017) (Figure 4). These structures contain B cells, CD4⁺ T cells, and GCs, among other cell types. Other RA patients have ‘diffuse infiltrates’, of which a plurality of the cells are CD4⁺ T cells. B cells present in the lesions are autoreactive and have SHM (Amara et al., 2013; Humby et al., 2009), indicative of a requirement for Tfh cells in RA progression. The ELSs in joints are particularly associated with CXCL13-expressing CD4⁺ T cells (Manzo et al., 2008). This is notable, because ectopic CXCL13 expression induces ELS formation and recruits B cells to nonlymphoid tissues in mice (Luther et al., 2000). In human lymphoid tissue, GC-Tfh cells are the primary producers of CXCL13 (Gu-Trantien et al., 2017; Havenar-Daughton et al., 2016b). However, the CXCL13-expressing CD4⁺ T cells in RA joints are predominantly CXCR5 and BCL6 negative (Kobayashi et al., 2013; Manzo et al., 2008; Rao et al., 2017). Those synovial T cells expressed CD40L and, in technically challenging experiments, have been demonstrated to help synovial B cells in vitro in an IL-21 and SLAMF5 dependent manner, similar to GC-Tfh cells (Rao et al., 2017). Both Tfh cells and CXCR5^neg CXCL13⁺ CD4 T cells co-localize with B cells in synovial ELSs, while in ‘diffuse infiltrate’ RA synovial tissue it is almost exclusively CXCR5^neg CXCL13⁺ CD4⁺ T cells observed to colocalize with B cells. Thus, given that the CD4⁺ T cells co-localize with B cells in vivo, express Tfh-associated B cell help molecules, and provide help to B cells in vitro, these cells are clearly B cell helpers in vivo.

These CXCL13⁺CXCR5^negPD1⁺IL-21⁺ CD4⁺ T cells (Rao et al., 2017) appear to be related to the surprising CXCL13⁺CXCR5^neg CD4⁺ T cells found in ELS of breast cancer (see below). It is unclear whether the CXCL13-expressing CD4⁺ T cells in breast cancer and RA are Tfh cells with impaired CXCR5 expression, or a more distantly related cell type. Based on the facts that the breast cancer-associated CXCL13⁺ cells express multiple Tfh cell-associated proteins (including BCL6), tumor extract could inhibit CXCR5 expression by GC-Tfh cells in vitro, and TGFβ could induce CXCR5^negCXCL13⁺ CD4 T cells in vitro (Gu-Trantien et al., 2017; Kobayashi et al., 2016), the breast cancer researchers concluded that the CXCL13⁺CXCR5^neg CD4 T cells were most likely Tfh cells with CXCR5 inhibited by the tumor microenvironment and termed the cells TfhX13 (Gu-Trantien et al., 2017). The connection with TGFβ is consistent with the finding that major components of human Tfh differentiation can be induced by TGFβ or activin A (see above) (Locci et al., 2016; Schmitt et al., 2014), and TGFβ is common in tumor microenvironments. It is plausible that a similar explanation applies to the RA synovial CD4⁺ T cells, but direct comparisons have not been done. In either case, the non-lymphoid-tissue CD4⁺ T cells in these disparate diseases possess central attributes of Tfh cells, and thus a clearer understanding of the genesis of these cells will help better understand Tfh cell biology and the immunobiology of these diseases.

The connection between Tfh and Tfh-like cells and ELS structures is intriguing, but poorly understood. The simplest hypotheses for the function of Tfh cells in RA ELSs, and autoimmune ELS biology more generally, are that (1) Tfh cells may help develop or support ELSs that are a recruitment site for other cells that actively engage in autoimmune pathology, or (2) Tfh cells may support autoantibody responses by B cells (Figure 4). In favor of the first model, Tfh cells are robust producers of CXCL13, and CXCL13 and B cells are usually required for ELS development and maintenance (Pitzalis et al., 2014). The role of Tfh cells in ELSs is challenging to assess in animal models. In mice, CXCL13 is predominantly made by stromal cells (including FDCs), not GC-Tfh cells. Additionally, ELSs are associated with chronic disease processes, which are more difficult to model and study than acute events. Nevertheless, ELSs develop in mice and are associated with Tfh cells in some cases (Pitzalis et al., 2014), and the biology of Tfh cell-ELS interactions is certainly worthy of more extensive exploration.

Tfh cells or IL-21 have been shown to be important in multiple RA animal models (Gensous et al., 2018). IL-6 is a signature cytokine of human RA, and blockade of IL-6 signaling can provide therapeutic benefit (Smolen et al., 2018); however, the role of IL-6 in human Tfh differentiation is unclear, as discussed above. Overall, there is substantial evidence for Tfh cell and Tfh-like cell contributions to RA, but mysteries remain regarding the cells themselves, their relationship to RA ELS structures, and the mechanisms by which they regulate autoantibody development.

Tfr Cells in the Regulation of Autoimmunity

The main function of Tfr cells appears to be to prevent or eliminate autoreactive B cells generated by SHM during an antigen-specific immune response (Botta et al., 2017; Wu et al., 2016). This is supported by data on the specificity of Tfr cells. While rare foreign antigen-specific Tfr cells can be detected as a type of induced Treg (iTreg), >99% of Tfr cells have been found to be natural Treg cells (nTreg) (Aloulou et al., 2016; Maceiras et al., 2017), consistent with a role primarily limiting autoreactivity. It is unclear if Tfr cells directly regulate Tfh cells in vivo under physiological conditions. Tfr cells in GCs are rare, so it is difficult to see how Tfr function could be primarily on GC-Tfh cells (Sayin et al., 2018; Wing et al., 2017). However, since most GC-Tfh cell interactions with B_gc cells are brief, it is possible that rare Tfr cells in GCs may suffice to influence GC biology due to the large number of cellular interactions that can occur in a GC in a single day. Murine Tfh cells do not express MHC class II, so there cannot be direct cognate recognition of Tfh cells by Tfr cells. IL-2 consumption by nTreg cells is one important mechanism of action by which Treg cells control CD4⁺ T cell responses (Liu et al., 2015b). However, IL-2 consumption by Treg cells generally has a positive impact on Tfh cells (León et al., 2014), as Tfh cells are inhibited by IL-2 and generally express low or no IL-2Rα (Ballesteros-Tato et al., 2012; Johnston et al., 2012). Thus, the main actions of Tfr cells in mice may be by direct influence on B cells. In contrast, human CD4⁺ T cells express MHC class II, and therefore Tfr cells could directly regulate Tfh cells in humans but not mice. Regarding direct regulation of B cells by Tfr cells, one study has found that CTLA4 is the major mechanism of action of Tfr cells, via directly regulating CD80 and/or CD86 on B cells (Wing et al., 2014), though a parallel study suggested other, undetermined, CTLA4 mechanisms were more important (Sage et al., 2014). Of note, there remain differences in defining Tfr cells, between species and between scientific studies, which may underlie some of the differences observed between studies (Ritvo et al., 2017). A major challenge to understanding the importance of Tfr cells in human autoimmune diseases is the limitation of studying blood. As discussed above, GCs and GC-Tfh cells are not present in blood, and thus studies of lymphoid tissue are required to understand the biology of human Tfh cells, Tfr cells, and GCs.

Tfh Cells in Allergies

Tfh cells are required for Ab-mediated allergies. It is still commonly assumed that IgE responses are driven by Th2 cells (Gould and Ramadani, 2018). That is incorrect. It was first shown in 2012, using a combination of experimental approaches in murine helminth infections models, that Tfh cell-derived IL-4, not Th2 cell -derived IL-4, was the source for IgE induction (Liang et al., 2012). More recently, the requirement for Tfh cell-derived IL-4 to develop IgE responses was elegantly shown in a different helminth model using mice engineered to have a specific Il4 deficiency in Tfh cells but not Th2 cells (Meli et al., 2017). Bcl6⁺ Tfh cells are distinct from GATA3^hi Th2 cells (Liang et al., 2012). The majority of IL-4 producing CD4⁺ T cells in LNs and spleen are Tfh cells, while Th2 cells are found in peripheral tissues (Reinhardt et al., 2009). IL-4 expression by Tfh cells is regulated by distinct signaling pathways and enhancer loci that excludes expression of IL-5 and IL-13 (Vijayanand et al., 2012; Yusuf et al., 2010). Il5 and Il13 are in the same chromosomal locus as Il4 and are specifically expressed in Th2 cells due to Th2 cell-specific enhancers.

While those studies indicate that allergic IgE responses to environmental antigens are likely to also depend on Tfh cells, they did not directly examine allergens. In a mouse allergic airway inflammation model, allergen-induced Tfh cells (CXCR5⁺Bcl6⁺IL-4⁺IL-21⁺ IL-5⁻IL-13⁻GATA3^l°) and Th2 cells (GATA3^hiIL-4⁺IL-5⁺IL-13⁺ CXCR5⁻Bcl6⁻) are again distinct (Kobayashi et al., 2017). IgE responses to allergen are completely lost in the absence of Tfh cells (Bcl6^fl/flCre^cd4) or ICOS (Kobayashi et al., 2017). Th2 cell responses and lung eosinophil inflammation are essentially unchanged. Those Tfh versus Th2 findings are consistent with a large human GWAS study that found genetic polymorphism associations with serum IgE and asthma are distinct, suggesting distinct pathways (Moffatt et al., 2010). A subsequent, analogous study used the authentic allergen peanut flour and found essentially identical results; IgE responses are dependent on Tfh cells, which were distinct from Th2 cells, and Tfh-deficient mice are protected from peanut exposure IgE anaphylaxis (Dolence et al., 2018).

Tfh cells express both IL-21 and IL-4. Tfh cells most likely induce IgE CSR under conditions of low IL-21 expression by Tfh cells or low IL21R signaling by B cells (Suto et al., 2002; Tangye, 2015). IgE CSR may be negatively regulated by Tfr cells (Yao et al., 2019).

While Tfh and Th2 are distinct cell types, nevertheless most aspects of CD4⁺ T cell biology can become complex under differing in vivo conditions, potentially due to a requirement for ‘fuzzy logic’ in T cell responses to prevent immune evasion by pathogens (Crotty, 2018). The relationship between Tfh and Th2 cell-associated transcriptional programs is no exception to exceptions. Tfh cell conversion into Th2 cells can be a primary source of Th2 cell effectors, shown in an elegant lung allergic asthma model study (Ballesteros-Tato et al., 2016). However, Tfh cells are not required for Th2 cell responses in similar models (Dullaers et al., 2017; Hondowicz et al., 2016), apparently due to differences in antigen burden (Dullaers et al., 2017).

Tfh Cells in Heart Disease

Atherosclerosis, the leading cause of death in the United States, is an inflammatory disease of coronary arteries. The causes of atherosclerotic inflammation are multifactorial, and recent evidence points to a functional role for Tfh cells. Abs against oxidized low-density lipoprotein (oxLDL) have a role in atherosclerosis, with most human data indicating that IgM is protective and IgG is potentially pathogenic (van den Berg et al., 2018). In a Notch-signaling-deficient B cell mouse model of atherosclerosis, rapid onset atherosclerosis has been observed and is associated with elevated frequencies of splenic GC-Tfh cells (Nus et al., 2017). The arterial disease can be prevented by blocking ICOSL, which prevented the splenic GC-Tfh cell accumulation. In a second atherosclerosis mouse model, a pro-atherosclerotic role for Tfh cells has also been demonstrated (Gaddis et al., 2018). Unexpectedly, nTreg cells de-differentiated (exTreg) and preferentially converted to autoreactive Tfh cells. Atherosclerosis was ameliorated in Tfh cell-deficient mice (Bcl6^fl/flcre^Cd4). OxLDL appears to drive nTreg cell conversion into Tfh cells via reduction of IL2Rα and enhancement of IL6R expression (Gaddis et al., 2018). LDL can also increase IL-6 and IL-27 expression by DCs of atherogenic mice (Ryu et al., 2018).

Interestingly, no obvious role for Abs have been observed in the atherosclerosis mouse models (Nus et al., 2017), suggesting that the pro-atherosclerotic activities of Tfh cells are not due to increased Abs. This is similar to the situation for multiple autoimmune diseases that are not thought to be primarily autoantibody-mediated, wherein Tfh cells are associated with disease development but not via providing help for pathogenic autoantibody responses. Alternative functions for Tfh cells in such situations are largely hypothetical, and this is an area of Tfh cell research that requires rigorous examination in the future. ELSs with Tfh cells are observed in mouse models of atherosclerosis and in human atherosclerosis (Clement et al., 2015), and may have a role in atherosclerosis disease, as has been hypothesized for SLE, multiple sclerosis, and RA.

Tfh Cells in Organ Transplant Diseases

Alloantibodies have long been associated with organ transplant or skin graft rejection. GCs are associated with organ rejection in a murine heart transplant chronic graft-versus-host disease (cGVHD) model (Ali et a l., 2016), and those responses can be driven by alloantigen-specific Tfh cells (Conlon et al., 2012). Tfh cells and GCs in transplanted kidneys have been associated with acute and chronic human kidney transplant rejection (Chenouard et al., 2017; de Graav et al., 2015). cGVHD is a major complication of human allogeneic HSC bone marrow transplantation. In an allogeneic HSC transplant murine model, cGVHD disease is dependent on Tfh cells and lung disease and alloantibody responses can be efficiently blocked by elimination of Cxcr5, or neutralization of IL-21, CD40L, or ICOS (Flynn et al., 2014). In a mouse skin GVHD disease model, disease development was associated with skin Tfh cells (CXCR5⁺PD1⁺ICOS⁺ CD4⁺) and blocking IL-21 prevents disease (Taylor et al., 2018).

Tfh Cells in Cancer

Perhaps the most surprising connections found between Tfh cells and human diseases in the past decade have come from the cancer field. Tfh-related cells have been positively associated with long term survival of humans with breast cancer or colorectal cancer (Bindea et al., 2013; Gu-Trantien et al., 2013). The strongest correlation with immunoprotective function is CXCL13 expression. In the context of breast cancer, the CXCL13-expressing CD4⁺ T cells are strongly correlated with ELSs (Gu-Trantien et al., 2017). In human lymphoid organs, GC-Tfh cells are the primary producers of CXCL13 (Gu-Trantien et al., 2017; Havenar-Daughton et al., 2016b); however, while breast cancer tissue-associated CXCL13-expressing CD4⁺ T cells have multiple attributes of GC-Tfh cells (including BCL6, CD200, PD1, and ICOS), they are predominantly CXCR5^neg (Gu-Trantien et al., 2017). Similar cells have been observed in melanoma tumors (Li et al., 2019). Notably, while many breast cancer tumors exhibit ELSs containing B cells, tumors with GCs are less common, and those tumors do contain fully differentiated GC-Tfh cells in addition to CXCR5^neg cells (Gu-Trantien et al., 2017). The Tfh-relatedness of the CXCR5^neg cells was discussed above.

The simplest hypotheses for the value of Tfh cells in immune responses against tumors parallel the concepts for autoimmunity: (1) Tfh may help develop or support ELSs, which are a site of recruitment for CD8⁺ T cells, NK cells, and macrophages that engage in anti-tumor immunity. Alternatively, (2) Tfh cells may support anti-tumor Ab responses by B cells. There is little data available regarding Tfh cells supporting anti-tumor Ab responses (Garaud et al., 2018), and more research is warranted at the granular level of the antigen-specificity (or lack thereof) of the B cells co-localizing with Tfh-like cells in tumors. In addition to breast cancer, survival with colorectal cancer has been positively associated with the presence of ELSs and CXCL13 (Becht et al., 2016; Coppola et al., 2011). Finally, given that there are over 100 types of cancers, it is not surprising that there appear to be differences in Tfh cell associations depending on the cancer. Tfh cells are negatively associated with survival in a mouse model of hepatocellular carcinoma with a prominent role of IgA⁺ PBs (Shalapour et al., 2017). There are clearly interesting complexities to Tfh-associated biology in the context of cancer, and the available data show that much more needs to be learned.

Perhaps the most exciting connection between Tfh cells and cancer biology are the potential implications of PD1 and PDL1 immunotherapies. As discussed above, it has been thought that Tfh cells and cytotoxic transcriptional programs are exclusive, but it has become clear that CD8⁺ T cells can appropriate Tfh cell-associated gene expression in a TCF1- and Id2-dependent manner to express CXCR5 and other Tfh cell-associated genes, and it is those CXCR5⁺ CD8⁺ T cells that respond most robustly to PD1 immunotherapy (Im et al., 2016). While CD8⁺ T cells are the primary target of PD1 mAbs in cancer immunotherapy, CD4⁺ T cells will, of course, also be affected. Given that GC-Tfh cells express very high levels of PD1 and are present in lymphoid organs throughout the human body, and there are Tfh-associated cells in multiple tumor types, it is surprising that the phenotypic and functional impact of PD1 and PDL1 immunotherapy on Tfh cells in humans has not been described.

Perspective: Future Directions

It is clear that both broad and deep advances have been made in the scientific understanding of Tfh cells during the past decade of research. Select advances were highlighted here, due to space constraints, and many other interesting Tfh-related findings have been made. Looking at this summary of the past decade of research, the obvious question is, ‘What does the future hold?’ Future directions and open questions have been discussed in numerous places throughout this article. Broadly, emphasis will need to continue to be put on understanding how antigen-specific Tfh cells differentiate and function, and increasing emphasis will surely be placed on how to apply that knowledge to Tfh-related therapeutics. GCs and numerous aspects of Tfh cell-associated biology are quite complex and frequently elude simple universal answers, but the high biomedical and scientific value of understanding these processes is clear and thus worth the effort. There is great interest in the use of Tfh cell-associated biology for enhancement for vaccines. There is also great interest in targeting Tfh cells for therapeutic interventions in human autoimmune diseases, allergies, atherosclerosis, organ transplants, and cancers. Therefore, well-defined causal roles for Tfh cells in disease pathologies remain a high priority research area. Key aspects of these human diseases are difficult to model in mice, such as ELS biology, the importance of CXCL13 production by Tfh cells, Ab isotype biology and CSR, and the cytokines that preferentially regulate human versus mouse Tfh differentiation. These challenges require creative solutions and should not be ignored.

Highlights.

• Tfh cells are the CD4⁺ T cells specialized for helping B cells.

• Bcl6 was identified as the Tfh lineage defining transcription factor 10 years ago.

• Tfh cell differentiation and functions have now been elucidated in great detail.

• Tfh cells have been found to be involved in a wide range of disease.