T follicular regulatory cells and antibody responses in transplantation

Abstract

De novo donor specific antibody (DSA) formation is a major problem in transplantation, and associated with long-term graft decline and loss as well as sensitisation, limiting future transplant options. Forming high-affinity, long-lived antibody responses involves a process called the germinal center (GC) reaction, and requires interaction between several cell types, including GC B cells, T follicular helper (Tfh) and T follicular regulatory (Tfr) cells. Tfr cells are an essential component of the GC reaction, limiting its size and reducing nonspecific or self-reactive responses.

An imbalance between helper function and regulatory function can lead to excessive antibody production. High proportions of Tfh cells have been associated with DSA formation in transplantation; therefore Tfr cells are likely to play an important role in limiting DSA production. Understanding the signals that govern Tfr cell development and the balance between helper and regulatory function within the GC is key to understanding how these cells might be manipulated to reduce the risk of DSA development.

This review discusses the development and function of Tfr cells and their relevance to transplantation. In particular how current and future immunosuppressive strategies might allow us to skew the ratio between Tfr and Tfh cells to increase or decrease the risk of de novo DSA formation.

Introduction

In the current era of transplantation short-term outcomes are excellent¹. However, long-term graft attrition has remained relatively unchanged over the past few decades^2–4, with a steady decline and rate of loss after the first year posttransplant¹ that has not improved despite improvements in organ retrieval, organ allocation and immunosuppressive regimens^1,5–7. There is increasing evidence that chronic rejection, associated with and potentially mediated by antibodies^8–10, is a major cause of long term graft loss^11–13. In kidney transplantation 8-10% of recipients develop de novo donor-specific anti-HLA antibodies (DSA) within the first year^14,15, and between 15-30% within 10 years^10,16,17. These antibodies are associated with an increased risk of graft failure^8,10,18,19 and therefore the cells that interact to produce alloantibody are becoming increasingly recognised as important targets in transplantation to try to improve long-term outcomes^20,21.

Many researchers have looked at the cells involved in the development of antibody responses against transplanted tissue, in particular T follicular helper (Tfh) and germinal center (GC) B cells. These studies have previously been comprehensively reviewed^22,23 however control of the GC reaction is provided by a specialised subset of regulatory T cells (Tregs) known as T follicular regulatory (Tfr) cells. This review summarises the current literature on Tfr cells and their relevance to transplantation, including how we might manipulate them to alter antibody responses to transplanted tissue.

The Germinal Center Reaction

In order to produce long-lived, high affinity antibody responses, mature naïve B cells must enter the B cell follicle of secondary lymphoid organs (SLOs) and interact with T cells in a process called the GC reaction²⁴.

Anti-HLA antibodies, and particularly posttransplant DSA, are predominately class switched^25–29, and persist in the circulation for many years both pre and posttransplant, suggesting that they have been produced by plasma cells that have come from the GC^30–32. HLA-specific B cells with memory markers (CD27, CD28) have also been identified in the circulation^33–35, and HLA-specific plasma cells in the bone marrow³⁰ of transplant patients, suggesting a requirement for GC formation, and hence for T cell help³⁶, in the development of a response to the transplanted organ³⁴. Animal models have supported this, showing that antibody mediated rejection is T cell dependent^37–41. Studies in humans have been more limited because of the difficulty of obtaining secondary lymphoid tissue, but kidneys removed following rejection have shown evidence of somatic hypermutation in intragraft B cell aggregates⁴². It is therefore likely that the GC reaction is necessary for development of anti-HLA antibodies and particularly DSAs.

The GC reaction is a process that allows generation of a broad spectrum of highly specific, high affinity antibodies to provide protection against the multiple pathogens that are encountered over the lifetime of an individual²⁴. Over the course of an antibody response, for example to vaccination, the affinity of antibodies for antigen increases in a process known as affinity maturation^43,44. In order to increase affinity, proliferating GC B cells undergo somatic hypermutation^45,46 (SHM) of their B cell receptor (BCR) genes. However, random mutation may generate BCRs with both lower and higher affinity for antigen, as well as potentially self-reactive BCRs⁴⁷. In order to ensure only higher affinity B cells go on to produce antibodies, which are soluble forms of the BCR, a selection process takes place within the GC that is dependent on T cells^22,23. T cell help comes from Tfh cells, CD4 T cells that have downregulated CCR7, the chemokine receptor that directs them to the T cell zone of SLOs, and upregulated CXCR5, the chemokine receptor that traffics cells to the B cell zone. After entering the follicle, Tfh cells are key players in the maintenance of the GC response, and selection of GC B cells, with GCs collapsing in the absence of Tfh cells⁴⁸. Tfh cells contain preformed CD40L that can be rapidly expressed on the cell surface to provide CD40 signalling to GC B cells during cognate T:B interactions^49,50. Tfh can also provide help in the form of cytokines. IL-21 is the classical cytokine associated with Tfh cells, maintaining Bcl6 expression in GC B cells and stimulating plasmablast development^51–54. It is thought that B cells with higher affinity BCRs can take up more antigen from follicular dendritic cells (FDCs) and out-compete lower affinity B cells for T cell help⁵⁵, hence lower affinity B cells either do not receive CD40L signalling and undergo apoptosis, or receive signals to induce reentry into cell cycle and further rounds of SHM⁵⁶. This would allow only the higher affinity cells to differentiate into long-lived memory B cells or plasma cells that then migrate to bone marrow or gut niches to form long-lived plasma cells⁵⁷. However, in the presence of excessive Tfh cells, lower affinity or self-reactive B cells can receive help^58–60. Thus control of the GC reaction is essential and is thought to both reduce the likelihood of autoantibodies and potentially ensure higher affinity antibody production. Tfr cells are a subset of Tregs that have co-opted the Tfh pathway to upregulate CXCR5 and enter the GC (figure 1) to provide regulation of GC responses.

Figure 1. Dynamics of the germinal centre and development of Tfh and Tfr cells.

A. Development of Tfh and GC B cells. Priming of naïve CD4 T cells occurs in the T cell zone by dendritic cells. Following this interaction, Tfh form a fate decision and can exit the SLO and enter the circulation as cTfh, or migrate towards the T-B border and undergo cognate interactions with B cells, allowing exchange of survival and proliferation signals. This induces Bcl6 in pre-Tfh, allowing further upregulation of CXCR5 and entry into the B cell follicle.

B. Initiation of the GC and development of Tfr. After priming by pre-Tfh at the T-B border, B cells move into the follicle and begin to proliferate, forming a germinal centre that is supported by the Tfh that have upregulated CXCR5 and entered the follicle. Within 3-4 days a GC has been established and both cell types are proliferating, supported by follicular dendritic cells (FDCs). Tregs are similarly primed by DCs and can either exit the SLO and enter the circulation as cTfr or, following interactions with B cells at the T-B border, enter into the B cell follicle to become bona fide Tfr cells and regulate the GC response.

C. Collapse of the GC response. By 7 days the GC response has reached its peak. High affinity B cells receive help to become long lived plasma or memory B cells and exit the SLO to enter the circulation, where the plasma cells will traffic to bone marrow niches. Tfr are present in the B cell follicle and begin to suppress the GC reaction, but it is not know if they exert effects on Tfh, GC B cells, the interactions between them, or a combination. By day 10 Tfh numbers are falling, Tfr numbers still rising, and the GC begins to collapse.

Cells with regulatory properties have long been of interest in transplantation because it was hoped they might be the key to immunological tolerance to the transplanted organ without compromising protective immunity. Much research has been done looking at CD4⁺ Tregs in particular and their potential use in tolerogenic immunosuppressive strategies, summarised by Wood and Sakaguchi⁶¹. Both animal models and human studies have lead to clinical trials in autoimmune disease and now transplantation. The recently completed ONEstudy - http://www.onestudy.org/ - has used several regulatory cell types in kidney transplantation in an attempt to induce tolerance to a transplanted organ^62,63 and detailed results are awaited.

However, Tregs are a diverse group of cells containing many subsets, designed to regulate a range of immune responses. Tregs act through a number of mechanisms, including anti-inflammatory cytokines, such as IL-10 and TGFβ, as well as cell-contact dependent inhibition of T cell activation by inhibitory receptors, such as CTLA-4 (CD152). In addition, Tregs can express transcription factors traditionally associated with other Th subsets alongside Foxp3 and tailor their suppression to the response generated^64–66. It is not surprising therefore that Tregs can also co-opt the Tfh pathway, expressing Bcl6 alongside Foxp3, and upregulating surface markers associated with Tfh, such as CXCR5, ICOS and PD-1^67–69 to enter the GC as Tfr cells.

T follicular regulatory cell development

While the presence of regulatory cells in SLOs and in the GC has long been established^68,70 the first detailed description of Tfr cells, Tregs that could enter the B cell follicle and control the production of antibody, occurred in 2011. Two groups simultaneously described Tfr cells^67,71, and their findings were confirmed by a third group⁶⁹. Under these experimental conditions, Tfr cells derived from thymic precursors, and expressed Foxp3 alongside Bcl6 and Blimp1^67,69,71. The latter was of particular interest, as Bcl6 and Blimp1 are mutually antagonistic, but Linterman et al showed that the dynamic expression of these transcription factors was required, with Bcl6 required for the development of the follicular phenotype in Tregs, and Blimp1 required to limit the size of the Tfr population⁶⁷.

In a similar way to Tfh cells, Tfr cells seem to require initial priming by DCs^72,73, however unlike Tfh cells, which require achaete-scute homologue-2 (Ascl2) for expression of CXCR5 and inhibition of Th1 and Th17 pathways⁷⁴, Tfr cells seem to be more dependent on nuclear factor of activated T cells (NFAT)-2 for initiation of CXCR5 expression⁷⁵. Both Tfh and Tfr cells are highly dependent on NFAT signaling^75–77 which is of interest in transplantation as calcineurin activation dephosphorylates NFAT and allows it to translate to the nucleus⁷⁸, meaning calcineurin inhibitors (CNIs) may inhibit Tfh and Tfr cell differentiation to a greater extent than other T cell subsets.

Upregulation of CXCR5 following priming by DCs allows both pre-Tfh and pre-Tfr to migrate towards the T-B border^56,73. In a similar way to Tfh, Tfr cells can then either exit the SLO into the circulation as memory-like circulating Tfr (cTfr) cells with lower expression of ICOS and PD-1 (discussed below), or move into the B cell follicle to regulate the GC response⁷³. Much like Tfh cells, Tfr cells are dependent on signalling via the TCR as well as costimulation^67,79,80, and both CD28 and ICOS are essential costimulatory molecules for Tfr development^67,80,81, as cd28-/- and icos-/- mice lack Tfr cells.

Interestingly, in the absence of B cells, cTfr cells can be found but Tfr cells in SLOs are absent⁷³, suggesting that interactions with DCs are sufficient for initiation of the Tfr pathway, but interactions with B cells, whether at the T-B border prior to entry into the follicle or in the follicle itself, are required for pre-Tfr to differentiate into Tfr cells and gain full suppressive capacity⁷³. There is debate as to the location of B cell interaction in part because it is not clear if Tfr cells are specific for the immunising antigen, and could in fact develop from naïve T cells via interaction with antigen-specific B cells under certain circumstances⁸² or have a TCR repertoire more similar to Tregs, developing in a non-antigen-specific way⁸³.

This distinction is likely to be extremely important in the context of transplantation, particularly in the era of regulatory cell therapies. If antigen-specific Tfr cells could reduce the development of antigen-specific antibodies, then altering the ratio of antigen-specific Tfh and Tfr may help to prevent the development of DSAs without limiting the development of antibodies against other antigens, such as bacterial antigens. If this could be managed, it may fulfil the elusive goal of transplant tolerance without compromising protective immunity. Aloulou et al showed that Tfr cells in the LN could develop from induced Tregs and be specific for the antigen as long as the stimulus was one that would normally generate induced Tregs⁸². However, if the antigen was foreign rather than self, they demonstrated that the balance was shifted towards antigen specific Tfh cells, rather than Tfr cells, which predominated when immunisation was with self-antigen⁸².

In the context of transplantation, the inflammatory insult of surgery and ischaemia-reperfusion injury^5,84 around transplantation and the presence of nonself antigen would be likely to promote antibody formation a high Tfh to Tfr ratio among antigen-specific cells⁸², however there is the potential for manipulation of this system, either by depleting or blocking Tfh cells^85,86 or possibly by promoting the development of Tfr cells.

The discovery that induced Tregs are capable of developing into antigen-specific Tfr cells⁸² could potentially allow ex vivo generation of antigen-specific Tregs, rather than polyclonal Tregs, which are currently being generated by the ONE Study⁸⁷. If these cells could be induced to form donor-antigen-specific Tfr cells, perhaps through PD-L1 stimulation as shown by this group⁸², this could potentially allow reduction in the development of de novo DSAs. If this were possible, it would presumably limit therapy to those with a live donor, as the generation of polyclonal Tregs currently takes around 14 days⁸⁷. Given that generation of antigen specific Tfh and Tfr cells begins within 48 hours of an immunising event^74,82, allospecific Tfr cells would need to be introduced very early post transplant, meaning only those with planned surgical dates could have allospecific cells generated ex vivo to avoid prolonged cold storage. However, this data is exciting and may support currently proposed strategies to move from polyclonal Treg therapy to allospecific Treg therapy^88,89, allowing not just allospecific Treg, but also allospecific Tfr to be generated.

Further work is required to determine if these antigen-specific Tfr cells are short lived or capable of forming long lived memory responses. It would be of interest to know if antigen-specific Tfr cells can be seen in the circulation following immunisation, as circulating Tfr cells are thought to represent a memory population⁷³. The requirement for lifelong acceptance of the transplanted organ in the presence of other inflammatory insults such as cellular rejection episodes; infections and future surgical insults would mean that a short-lived response would be of limited benefit to transplant recipients.

It may be of more use to understand the factors that may drive development of antigen-specific Tfr. Manipulation of antigen presenting cells (APCs) to shift the balance of responding cells from a pro-inflammatory to a tolerogenic profile is currently being trialled in transplantation^90,91. As Tfr cells are also dependent on the signals from DCs in order to develop, using antigen-specific DCs expressing high levels of PD-L1 may be able to drive antigen-specific Tfr development to reduce GC responses to a particular antigen⁸². Manipulation of these pathways is currently being explored with the aim of improving vaccination responses (multiple animal studies, reviewed by Linterman and Hill⁹²). However considerable work is required to refine our understanding of the pathways that both enhance and inhibit Tfr cell development and how we might safely manipulate them to improve or inhibit GC responses.

At present, several molecules are known to inhibit Tfr cell development, including CTLA4 (discussed below), PD-1 and IL-2. PD-1 is an inhibitory receptor of great interest in the field of autoimmunity^93–96, cancer⁹⁷ and vaccination⁹⁸ and is a key marker of Tfh and Tfr cells^99,100. PD-1 and its ligand PD-L1 are important for the induction and maintenance of induced Tregs and in the absence of PD-L1 signalling, induction of Tregs from naïve CD4 T cells is much lower¹⁰¹. In contrast, while it is not entirely clear what the function of PD-1 on Tfh and Tfr cells is, it appears to have an inhibitory role. In pdcd1-/- mice, Tfr cells are seen in greater proportion in the LN and blood, while Tfh cells are seen in greater proportion in the blood, but reduced in the LN, consistent with the greater suppressive capacity of LN Tfr cells compared to cTfr cells⁸⁰. It is not clear how PD-1 signalling curtails Tfh or Tfr function, or why it is expressed so highly on Tfh and Tfr cells. However the authors of this paper speculate that, as strong TCR signalling slows T cell movement, PD-1 may reduce the strength of the signal and allow Tfh cells to scan more B cells, giving help only to the highest affinity cells which would elicit the strongest TCR signal.

Interestingly Aloulou et al showed that antigen-specific Tfr cells required PD-L1 signalling in order to develop⁸², this may be due to the fact that these cells develop from induced Tregs, rather than natural Tregs, and thus are highly dependent on PD-L1 signalling for precursor development¹⁰¹. Further signals that promote conversion of induced Tregs to Tfr cells are yet to be elucidated, and it is not clear if they would follow the same pathway of development, and require the same DC and B cell interactions as Tfr cells derived from thymic precursors.

It is also of interest that Tregs are highly dependent on IL-2 signalling^102,103 and yet this cytokine suppresses Tfh and Tfr cell responses^104,105. Botta et al showed that at the peak of influenza infection, the high levels of IL-2 promoted Blimp-1 expression in Tregs, which prevented Bcl6 upregulation and hence prevented them developing into Tfr cells. As the infection cleared and IL-2 levels fell, a proportion of Tregs downregulated CD25, upregulated Bcl6 and took on a Tfr phenotype¹⁰⁶. This allows early initiation of an antibody response, as high Treg consumption of IL-2 is thought to permit Tfh cell development¹⁰⁷, but then curtailing of this response after the infection is cleared, to prevent uncontrolled GC responses leading to autoantibody production.

The dynamic regulation of Tfr cell activation and maturation is a key factor in developing an effective humoral immune response. At a resting state, animal models suggest that, while Tfr cells are a rare population (approximately 1% of the total CD4 T cell population in mice) depending on the tissue studied, in the B cell follicle they can be at equal proportion to Tfh cells^73,80,108. During the course of an immune response, both Tfh and Tfr cells begin to proliferate, however Tfh cells proliferate faster and skew the proportion in favour of helper capacity. By day 7, the peak Tfh response and when GCs begin to form^56,109, Tfr cells represent less than 20% of the follicular CD4 T cell population⁸⁰, but by day 10 Tfh numbers have started to fall while Tfr cell proliferation continues, and so ratio returns to the resting state⁸⁰. In tissues exposed to constant antigen stimulation the ratio of Tfr to Tfh cells is lower, for example Peyer’s patches where constant IgA production is required¹¹⁰, and the spleen, where rapid responses to blood borne antigens are essential¹¹¹. The ratio of Tfr to Tfh cells is not only important for the control of normal immune responses, but to limit autoimmunity, as uncontrolled Tfh responses lead to spontaneous autoimmunity^58,60. Understanding more about how this process is controlled and the interplay between different cell types and signalling pathways is important to identify targets that might allow us to manipulate this process.

Biomarkers of GC responses

Many studies of human disease have used circulating Tfh-like cells (cTfh) and circulating Tfr-like cells (cTfr) as biomarkers looking at both normal antigen responses in vaccination^112,113 and abnormal responses in autoimmune disease and chronic viral infection. Studies following influenza vaccination demonstrate the potential use of cTfh cells as a peripheral biomarker of GC activity. Following vaccination, subjects showed increased ICOS⁺PD-1⁺CXCR3⁺CXCR5⁺CD4⁺ Tfh cells at 7 days, which correlated both with plasmablast appearance in the circulation and development of strain-specific influenza antibody¹¹³. However, these studies did not look at Tfr markers to separate helper from regulatory cells. Animal models have suggested that cTfr cells, circulating cells resembling lymph node Tfr cells, also represent a memory population⁷³ and are reflective of an ongoing GC response. As secondary lymphoid tissue is difficult to obtain in humans, it is important to elucidate how closely the ratio of circulating cells reflects the tissue resident populations. This has been most closely investigated in autoimmune disease and chronic viral infection.

In human autoimmune disease, circulating Tfh and Tfr cells have been reported in many different conditions. It was originally suggested that high proportions of cTfh cells correlated with autoantibody levels in many diseases^114–118, however many of these early studies did not differentiate between Foxp3⁺ and Foxp3^- cells in the circulation, describing whole populations of CD4⁺CXCR5⁺ cells with the addition of PD-1 or ICOS as markers of Tfh-like cells. Further studies have looked specifically at cTfr levels and ratios, but reports have been variable, with some studies showing higher percentages of cTfr cells in certain autoimmune diseases¹¹⁹, while others demonstrated lower levels¹²⁰ with a skew towards high cTfh cells. In all studies higher proportions of circulating cells of both types seem to correlate with active inflammation and antibody production. However, extensive animal work suggests that uncontrolled Tfh cell responses lead to autoimmunity^58–60, combining this with the human studies suggesting high cTfh cells correlate with active disease and high autoantibody levels^118,121, it would seem likely that uncontrolled Tfh responses in humans are also associated with autoimmunity. It is not yet clear whether these are due to ineffective Tfr responses, or whether the breakdown in self tolerance is the key factor that allows a persistent antibody response, driving the immune system towards Tfh-mediated antibody production and suppressing Tfr responses to a mistaken ‘nonself’ antigen⁸².

In the absence of a breakdown in self-tolerance, the ratio of cTfh to cTfr seems to be important in driving antibody formation. In chronic viral infection higher levels of cTfr correlated positively with higher levels of hepatitis B or C viral DNA¹²², suggesting that in the presence of chronic antigen stimulation, having higher levels of Tfr cells prevents the formation of antibodies that might help clear the infection. In HIV, ineffective Tfh responses and excessive Tfr responses are thought to prevent the development of bnAbs^123–126, allowing persistence of the virus due to inadequate GC responses. Indeed HIV+ patients show inadequate responses to vaccination, suggesting a generalised defect in GC responses, not specific to the virus^125,127,128. The balance between Tfh and Tfr responses is therefore finely tuned and likely to be a key factor in the development of antibodies to persistent antigens.

In transplantation, this balance may also be important, as, like a chronic virus, the transplanted tissue provides a persistent antigenic stimulation that cannot easily be cleared. However, studies of cTfh and cTfr cells have been limited. Some human observational studies have suggested that high levels of cTfh correlate with DSA production, and low levels of cTfr cells correlate with chronic rejection ^86,129–131. This is in keeping with animal models showing that Tfh cells are required for DSA formation³⁸ and blocking pathways important for GC development reduces DSA levels¹³². The data on Tfr cells in transplantation is even more limited. In graft-versus-host disease (GVHD), a complication of haematopoietic stem cell transplantation where repopulating cells see the recipient tissues as nonself and therefore mount a response, low levels of Tregs are associated with worse disease¹³³. Additionally, Tfh cells seem to be important in driving chronic GVHD, as transplantation of CXCR5-deficient T cells showed attenuated disease, as did blocking IL-21, ICOS and CD40¹³⁴. McDonald-Hyman et al showed that infusion of Tregs prior to haematopoietic stem cell transplantation or after established GVHD led to attenuation of disease, and this was thought, in part, to be mediated by an increase in Tfr cells limiting Tfh cells and antibody production¹³⁵. Whether this finding also holds true in solid organ transplant is yet to be elucidated. However, animal models have shown that Tregs increase in protocols designed to induce transplant tolerance¹³⁶ and Tregs are higher in patients who are operationally tolerant of their grafts^137,138, cTfh cells have also been shown to be lower in operationally tolerant renal transplant recipients¹³⁹ suggesting that the balance between follicular effector and regulatory cells is important for tolerance to transplanted antigens.

Impact of Immunosuppression on Tfr cells

While an antibody response to donor antigen is a normal response to nonself, and therefore a skew in favour of Tfh cells would be expected early post transplant, many of the immunosuppressive agents are T cell targeted and may interfere with the normal process of dynamic regulation.

Commonly used induction strategies include blockade of CD25, the alpha chain of the IL-2 receptor, with agents such as basiliximab, and lymphocyte depletion with agents such as ATG and alemtuzumab. Blockade of CD25 is designed to prevent activation of T cells, in part by blocking autocrine production of IL-2, limiting the levels of this cytokine. Lower levels of IL-2 would prevent IL-2 mediated upregulation of STAT5, thus preventing STAT5-dependent blockade of Bcl6 upregulation¹⁰⁴. This would be permissive of Tfh cell development, but the lack of IL-2 signalling would also allow Tregs to upregulate Bcl6 and develop into Tfr cells and potentially promote Tfr cell development over Tfh cells¹⁰⁶. However, given Treg cells are highly dependent on IL-2 signalling and Tfr are thought to develop from nTregs, it may also reduce Tfr numbers by reducing precursor cells. Therefore it is not clear whether induction therapy with basiliximab or other CD25 blocking mAbs would lead to a balanced increase in both Tfh and Tfr cells, or whether there is a skew towards regulatory or helper function. Although in wide use, it is also not clear from epidemiological studies if anti-CD25 mAbs alter the risk of de novo DSA formation compared to alternative induction methods¹⁴⁰. Further observational studies with immunophenotyping of recipients would be useful to establish the effect of both induction and maintenance therapy on the ratio of Tfh to Tfr cells, and further elucidate the relative risk of de novo DSA formation.

The impact of lymphocyte depleting agents is another unknown, in part due to the complexity of the recovering cell population dynamics. Alemtuzumab, another widely used induction agent, leads to widespread lymphocyte depletion with variable recovery, however from phenotyping studies in autoimmunity, it appears to increase the proportion of Tregs in the recovering CD4 T cell population^141,142. Given the evidence suggesting Tfr cells can develop from induced Tregs, alemtuzumab induction may skew the balance in favour of Tfr responses when the pro-inflammatory milieu (and hence IL-2 levels) has reduced. However, autoimmunity following alemtuzumab, which occurs in up to a third of multiple sclerosis patients treated with the drug, has been shown to be IL-21 dependent¹⁴³ therefore suggesting that the increase in the proportion of Tregs, which will consume IL-2 and therefore allow Tfh cell development, may skew the balance towards helper function rather than regulation. More work is required to understand the impact of these agents on the cells that interact to form antibodies.

Current maintenance immunosuppressive agents tend to target T cell pathways, and hence may affect both Tfh and Tfr cells. Tfh and Tfr cells are both highly dependent on NFAT signalling^75,77, which is itself dependent on calcineurin dephosphorylation, hence disruption by CNIs may affect these cells more than other cell subsets. Additionally, MTORC1 has been found to promote the generation of Tfr cells and be important for suppressive function^76,144, so use of MTOR inhibitors may adversely affect Tfr cell function, skewing the balance towards Tfh cells. This is supported by animal models suggesting that the use of MTOR inhibitor rapamycin (sirolimus) after alemtuzumab induction increases the proportion of Tfh cells and consequently of DSAs¹⁴⁵. Given the complexity of the pathways that interact to generate a GC response, it would be prudent to understand these pathways and the influence of drugs prior to targeting the GC response to either promote or reduce antibody responses.

An example of unintentional manipulation of the GC response has recently been demonstrated. Costimulatory blockade using belatacept, a CTLA-4/IgG1 fusion protein, has been approved for use in transplantation as an alternative maintenance agent for those unable to tolerate CNIs or where CNI toxicity has been demonstrated. CTLA-4 is an inhibitory receptor with a much higher affinity for CD80/86 than CD28, its stimulatory counterpart, and is an important mechanism of Treg mediated suppression through disruption of CD28-CD80/86 signalling^146,147.

While this is a key mechanism of Treg mediated suppression, it is not clear how Tfr suppress GC responses. For example, IL-10, a key anti-inflammatory cytokine produced by Tregs¹⁴⁸ and Tfr (and possibly Tfh⁷³), appears to have a role in supporting GC B cell survival¹⁴⁹ and maintaining normal light zone/dark zone differentiation¹⁵⁰. The function of CTLA-4 on Tfr cells is also unclear. While germline deletion of CTLA-4 in mice is fatal in early life due to overwhelming autoimmunity and autoantibody formation^151,152, and in humans mutations in CTLA-4 leads to profound immune dysregulation¹⁵³, blockade or deletion of CTLA-4 later in life does not cause these issues and may even be protective against certain forms of autoimmunity by increasing IL-10 production¹⁵⁴. In animal models, Treg specific deletion of CTLA-4 leads to a significant increase in Tfr cell numbers in SLOs and of cTfr cells in the circulation^147,155, suggesting a directly inhibitory effect on Tfr cells. However, while Tfr numbers may be increased, these cells show reduced suppressive capacity in vitro^147,155, suggesting it is a key molecule for Tfr function. Interestingly in one study, while mice with CTLA-4 deficient Tfr cells had higher numbers of Tfh cells and higher antibody level, this was of lower affinity, with the authors postulating that suppression by Tfr cells may reduce the amount, but increase the affinity of antibody produced by ensuring only B cells with the highest affinity for antigen go on to receive T cell help and develop into plasma cells¹⁵⁵.

In humans, CTLA-4 Ig has been widely used as a treatment for autoimmune disease and as maintenance therapy in transplantation in an attempt to mimic its function on Tregs^{132,156–158}. It has good efficacy compared to CNIs, with an increased rate of acute rejection early post transplant but improved long term outcomes¹⁵⁹. Interestingly, belatacept has been linked to smaller GC responses, lower levels of Tfh in LN and a lower rate of DSA formation in a non-human primate model¹³². Additionally, mouse models show CTLA-4Ig can suppress both de novo and memory DSA responses by reducing Tfh numbers¹⁶⁰. This is supported by data from the BENEFIT-EXT trial showing a markedly reduced rate of DSA formation with belatacept treatment compared to ciclosporin¹⁶¹. These studies support the idea that inhibition through CTLA-4 is one of the key factors in Tfr-mediated suppression of GC responses, however exactly how this is mediated is unclear. Interestingly in vitro work by this same group suggests that this is not due to a direct effect of CTLA-4Ig on Tfh cells or their interactions with B cells¹⁶². These apparently conflicting results require further investigation, and it would be of interest to see long-term phenotyping studies on samples from patients receiving belatacept maintenance therapy to see whether the ratio of cTfr to cTfh cells, or the function of these cells changes over time, as the improved outcomes from belatacept compared to CNIs tend to manifest later post transplant.

Another trial of great interest will be the ONEstudy. With the advent of cell therapy, multiple regulatory cell populations have been generated ex vivo and infused into patients undergoing living donor transplantation. Multiple sites have been set up, with different sites investigating the safety and feasibility of using different cell subsets, ranging from regulatory T cells to tolerogenic DCs and suppressive macrophage subsets. Standardised immunophenotyping has been undertaken for all patient groups in order to examine the effects of different cell therapies on the circulating immune profile¹⁶³. Unfortunately these standardised panels do not include cTfh or cTfr markers, so it is unclear if there will be any impact of cell therapy on GC activity. It will be of interest to see whether therapy with polyclonal induced Tregs will lead to any alterations in the rate of de novo DSA formation. Given the animal data suggesting that Tfr can develop both from thymic and induced Tregs, it is possible that infused Tregs could also take on a Tfr phenotype and provide suppression of GC responses. However, the excess consumption of IL-2 from an infusion of Tregs may allow differentiation of Tfh cells and thus increase the likelihood of DSA development, although it is unclear how much impact CNI maintenance therapy, which is given alongside Tregs, may affect the cell populations. Additionally, there is concern that the infused Tregs may not be stable, and in the context of inflammatory stimuli such as infection, operative intervention, or indeed acute rejection, could potentially convert to other helper cell phenotypes¹⁶⁴. These early studies will hopefully provide some answers and guide future investigations.

Future directions

Although it is clear that Tfh cells must play a role in alloantibody formation^38,86,131 it is not yet clear what the role of Tfr cells is in modulating alloantibody formation. Inhibition of Tfr cell function through conditional deletion of CTLA-4 leads to more but lower affinity antibodies¹⁵⁵. One of the postulated roles of Tfr cells is to prevent antibody responses in the context of low levels of antigen⁷³ or where B cells have only generated low-affinity receptors through somatic hypermutation^67,71. Paradoxically, it may therefore be that a skew towards Tfr responses leads to fewer, but higher affinity DSAs that are more likely to lead to rejection. Alternatively, given the persistence of antigen, the potential need for manipulation to increase the ratio of Tfr to Tfh cells long-term may limit antibody formation in all areas, both to donor antigens and to pathogens, thus increasing the risk of infection.

Until a mechanism can be found to selectively enhance and maintain the development of antigen-specific Tfr cells to reduce the production of clinically undesirable antibodies such as DSAs without reducing production of helpful antibodies, altering the ratio of Tfr to Tfh cells where there is long term exposure to and persistence of nonself antigen may not be a feasible approach to reduce the risk of DSA formation.

It seems we may inadvertently have stumbled upon one such manipulation strategy with the use of belatacept, although further work is required to establish the exact mechanism of action and if this is mediated through GC responses or other pathways.

In summary, the GC response and roles of Tfh and Tfr cells are at the forefront of immunological research, and understanding the regulation of the GC and antibody production is essential as the field of transplantation medicine moves forward. Tackling de novo DSA formation is one of the key targets for improving transplant longevity and reducing waiting times for those who are sensitised. Understanding how the GC response changes posttransplant using the biomarkers of cTfh and cTfr is important in understanding and predicting those at risk of de novo DSA formation. Additionally, understanding how different immunosuppressive regimens and strategies influence humoral immunity will be key to finding treatment options for these patients. There are many questions still to be answered, but with new scientific tools and comprehensive phenotyping panels, marrying high quality bench science with clinical trial protocols can provide answers to at least some of these questions.

Abbreviations

(c)Tfh

(circulating) T follicular helper cell

(c)Tfr

(circulating) T follicular regulatory cell

APC

antigen presenting cell

Ascl-2

achaete-scute homologue-2

BCR

B cell receptor

CNI

calcineurin inhibitor

DC

dendritic cell

DNA

deoxyribonucleic acid

DSA

donor specific antibody

FDC

follicular dendritic cell

GC

germinal center

HIV

human immunodeficiency virus

HLA

human leucocyte antigen

LN

lymph node

MTOR

mammalian target of rapamycin

NFAT

nuclear factor of activated T cells

SAP

SLAM-associated protein

SHM

somatic hypermutation

SLO

secondary lymphoid organ

TCR

T cell receptor

TLO

tertiary lymphoid organ

Treg

regulatory T cell