“Germinal Center B cell initiation, GC maturation, and the coevolution of its stromal cell niches”

Summary

Throughout the developing GC response, B cell survival and fate choices made at the single cell level are dependent on signals received largely through interactions with other cells, often with cognate T cells. The type of signals that a given B cell can encounter is dictated by its location within tissue microarchitecture. The focus of this review is on the initiation and evolution of the GC response at the earliest time points. Here we review the key factors influencing the progression of GC B cell differentiation that are both stage and context dependent. Finally, we describe the coevolution of niches within and surrounding the GC that influence the outcome of the GC response.

1. Introduction

Effective immune responses to pathogens and vaccines critically depend on the generation of high affinity antibodies. The steady increase in antibody affinity during adaptive immune responses, known as affinity maturation, requires the formation of an organized cluster of proliferating B and T lymphocytes in germinal centers (GC). Germinal center responses are initiated within secondary lymphoid tissues, that are inherently organized, where antigen-inexperienced T and B cells are separated into T cell zones and B cell follicles. CD4+ T cells expressing T cell receptors (TCR) specific for newly arriving antigen are activated to both proliferate and differentiate through interactions with dendritic cells (DCs) that have internalized antigen and presented proteolytically degraded fragments on surface MHC II molecules. At the same time, antigen-specific B cells become activated after encountering antigen in its intact, unprocessed form and experiencing B cell receptor (BCR) ligation. B cells that have internalized antigen bound to its specific BCR can then also present processed components in the context of surface MHC II. A series of coordinate interactions between activated T and B cells specific for the same (cognate) antigen are the foundation for B cell activation throughout the GC response, both during the initial pre-GC phase and within the mature GC.

The GCs that emerge within B cell follicles of secondary lymphoid tissue are highly organized structures, within which GC B cells compete to acquire additional antigen and engage follicular helper T cells. With contemporary thinking, GC B cells expressing BCR with higher affinities are selected to further proliferate, while less fit B cells expressing BCRs with weaker affinities may not receive sufficient survival or instructive signals and undergo apoptotic cell death. In addition to survival, T/B interactions also drive differentiation of GC B cells into one of two lineages that are fundamental to durable immunity; long-lived high-affinity plasma cells and memory B cells^1–4. In this way, GCs mold, amplify and add persistence to the most effective B cells of the immune response.

Throughout the developing GC response, B cell survival and fate choices made at the single cell level are dependent on signals received largely through interactions with other cells, often with cognate T cells. The type of signals that a given B cell can encounter is dictated by its location within tissue microarchitecture. A major component of B cell activation and differentiation in the GC response involves sequential changesin migration patterns that move the cell from one specialized zone to the next. The focus of this review is on the initiation and evolution of the GC response at the earliest time points. We’ll review the key factors influencing the progression of GC B cell differentiation that are both stage and context dependent. The review will finish with a discussion of the niches within and surrounding the GC that may influence the outcome of the GC response.

2. Landmarks and unique features of established GCs

2.1. Tissue compartments of mature GCs

To better appreciate the similarities and disparities of the initial GC response, we begin with an overview of the fundamentals of fully mature GCs. As the newly established GC expands, non-responding B cells are peripherally displaced and surround the emerging GC, forming what is known as the follicular mantle. Mature GCs have a complex architecture that is comprised of two prominent subdomains, the light zone (LZ) and dark zone (DZ), originally named for their relative appearance in hematoxylin/eosin-stained tissue sections^5–9. The light zone (LZ) is distinguished by the presence of the fine processes of a stromal cell subset, follicular dendritic cells (FDC), which entirely comprise the stromal cell matrix in this region and form a reticular network that GC B cells infiltrate⁸. Within primary follicles FDCs are a key source of CXCL13, a chemokine that is requisite for the correct organization of GCs into compartmentalized DZ and LZs¹⁰. FDCs retain deposits of activated complement components and antigen–antibody complexes via the low-affinity Fc receptor for IgG (FcγRIIb) and the complement receptors CD21 and CD35^11–15. FDCs can serve as long-term depots of undegraded Ag 11, 13, 15–17, and for this reason they have classically been considered an important site of antigen presentation to B cells. Intrafollicular CD4+ T helper cells known as T follicular helpers (Tfh) are predominantly found within the LZ, sometimes in great numbers, but are also present throughout the follicle mantle^{5, 18, 19}. Tfh are critical to the formation and maintenance of a productive GC, largely due to their provision of IL-21, IL-4 and CD40^20–26. At the peak of the response, Tfh can be distinguished from other cell subsets by a high level of expression of the molecules PD-1, ICOS, and the chemokine receptor CXCR5^27–30.

The DZ contains a higher density of proliferating cells and harbors another stromal cell type that produces CXCL12, the ligand for CXCR4, and lacks the distinctive CD35 expression of FDCs^{31, 32}. The majority of DZ resident B cells, also referred to as centroblasts, express lower levels of a variety of surface markers, including BCRs, CD40, and CD86, but a higher level of Bcl6 and CXCR4 compared to their LZ counterparts^32–34. Many LZ resident GC B cells, aka centrocytes, are smaller in size and are typically CD86hi, CD83hi, and CXCR4lo³³. Although these phenotypic differences have come to be viewed as surrogate markers of GC B cell location, it is important to note that B cells residing within each zone are far from homogeneous. GCs are consistently oriented within B cell follicles (Fig 1). The LZ is reproducibly more proximal to the subcapsular sinus of lymph nodes or the marginal sinus enveloping the white pulp of spleens. The dark zone balloons beneath the FDC network, extending toward the border of the follicle with the T cell zone.

Figure 1.

Lymph node tissue compartments influencing the developmental stages of germinal center B cells.

2.2. Selection of higher affinity variants, inter-zonal migration and cyclic reprogramming

During their extensive clonal expansion, B cells with higher affinity B-cell receptor (BCR) variants are selectively enriched^35–38. GC B cells express high levels of activation induced cytidine deaminase (AID) that drives a unique process of somatic hypermutation (SHM) of immunoglobulin (Ig) genes that can introduce amino acid substitutions in the antibodies produced^{39, 40}. Because SHM is introduced during DNA replication, it was thought to occur in the DZ where mitotic cells are more evident, especially in tonsils^{32, 33, 41, 42}. Selection for higher affinity B cells has been proposed to result from competition for the acquisition of antigen^{6, 43, 44}. As FDCs are the most evident site of antigen accumulation within follicles, and have the ability to promote the survival of GC B cells that are prone to apoptosis, the selection process is thought to occur within the LZ compartment^45–47. Early thinking on GC B cell selection envisioned that a key element involved the crosslinking of BCRs while in contact with antigen-laden FDCs^48–50. However, a recent study by Luo et al indicates that some aspects of canonical BCR signaling are rewired in Bcl6hi GC B cells, such that BCR crosslinking per se cannot fully instruct the extensive self-renewal observed within mature GCs^{51, 52}. Other studies strongly support a model in which positive selection requires GC B cells to form immunological synapses with Tfh cells^53–55 and reviewed in^{56, 57}. B cells with higher affinity acquire more protein antigen, resulting in increased surface presentation of MHC II-antigen complexes, and a greater capacity to stimulate Tfh cells with a cognate specificity^{33, 58–62}. It is important to note that BCR signaling during antigen acquisition and Tfh cell engagement need not be mutually exclusive and could be synergistic. In support of this, the combination of BCR and CD40 agonism more robustly invokes cMYC expression and further cell division then either alone⁵². Other models for affinity maturation postulated forms of negative selection, which could operate in concert with positive selection^{48, 63–66}.

To reconcile a perceived need for sequential events to occur in spatially separated regions of the GC, the “cyclic re-entry” model was proposed^{67, 68}. With a contemporary view of this model, DZ resident B cells that have exited cell cycle lose the transcriptional profile typical of centroblasts and with lower CXCR4 levels subsequently migrate to the LZ^{10, 32, 33, 69}. Intravital imaging studies of GC B cell migration patterns are consistent with infrequent movement between compartments^{59, 70–72}. Within the LZ, the recent immigrants compete for access to antigen, potentially experience BCR crosslinking during antigen acquisition, and then compete for contacts with Tfh cells that selectively provide factors such as CD40L, IL-21 and IL-4^{52, 73, 74}. FoxO1, cMyc and AP4 are among the transcription factors known to play a critical role in the transition of such positively selected centrocytes to the centroblast transcriptional program, although expression of these transcription factors (TF) can be short-lived^{34, 75–78}. Thus, the LZ is currently seen as the compartment in which both self-renewal and differentiation toward long-lived effector lineages are initiated, the latter evidenced by the expression of antibody secreting cell (ASC) associated TFs by B cells within the LZ harboring the majority of Tfh cells^{55, 79–82}. Self-renewing GC B cells undergo a centrocyte to centroblast transition that is characterized by a reciprocal increase in CXCR4, a decrease in costimulatory molecules, initiation of cell division and their return to the DZ to complete cell division there³³. Within the DZ, centroblasts reduce their surface BCR and recycle existing MHC II complexes, and in this way have an opportunity to better express their newly mutated BCR that can be more fully tested when antigen is next encountered^{83, 84}.

The bulk of research on GCs has interrogated mature GCs. However recent studies have implications on its initiation and development. The focus of the remainder of this review is on the progression of the GC response at the earliest time points, both of the responding B cells and within the evolving stromal compartments that coordinate the context of their cellular interactions.

3. T-B cell contacts during GC initiation

Exploration of the earliest stages of GC development had long been hampered by the seemingly inscrutable developmental steps that preceded the appearance of the classical surface markers of GC B cells. Markers typically ascribed to GC B cells include high levels of CD95 (Fas), the GL7 epitope, a loss of IgD and an increased ability to bind peanut agglutinin (PNA)^85–87. B cells within established murine GCs have uniformly low levels of CD38 in mice, while curiously exactly the opposite pattern is seen with human GC B cells⁸⁸. However, it is important to note that during the early phases of an immune response, some of these markers are also expressed by developing plasmablasts^{89, 90}. A further challenge to the study of early B cell activation events is the downregulation of surface BCR after crosslinking, making antigen specificity hard to confirm in some cases. A superior indicator of this lineage commitment emerged when the requisite role of Bcl6 was demonstrated, empowering investigations into the early steps in this pathway that were previously enigmatic. Expression of the transcriptional repressor Bcl6 is required for both Tfh and GC B cell differentiation, a process that involves shifts in the transcription of other influential TFs^91–95. In order to more easily identify and track the behavior of antigen specific T and B cells, we and others have developed strains of mice that ubiquitously express fluorescent proteins such as GFP or RFP and harbor transgenic or knockin BCR or TCR alleles. With the transfer of such fluorescently tagged cells, specific cells can be interrogated while remaining agnostic on all other phenotypic aspects during their activation and as they undergo significant change during subsequent differentiation.

The creation of a productive GC involves step-wise changes to the transcriptional profiles and phenotype of responding lymphocytes, often including shifts in their chemotactic tendencies. Prior to activation, antigen-naïve B and T cells are segregated into B cell follicles and T cell zones (Fig 1). This is the result of migration patterns not dictated by structural or anatomical limits, but rather by localized expression of chemotactic agents drawing in B and T cells to their respective zones based on their expression of receptors for those agents. By virtue of B cell expression of CXCR5, they are attracted to its ligand chemokine CXCL13 that is produced by FDCs centrally positioned in follicles^{96, 97} and by Marginal Reticular Cells (MRC) that line the subcaspular sinus (SCS) of LNs and splenic marginal sinus (SMS)^98–100. T cell localization to the T cell zone is mediated by their expression of CCR7, the receptor for CCL19/21^{101, 102}. An additional chemoattractant within lymphoid tissue includes the compound EBI2L that is enzymatically modified by stroma lining the SCS and SMS. EBI2L is also found at the cortical sinuses descending through the interfollicular (IF) regions, at the border between T cell zones and follicles (T-B border), as well as at the bridging channels positioned between follicles in spleens (Fig 1 and^103–109). The combinatorial expression of even just these 3 chemoattractants by stromal cell populations establishes multiple distinct architectural domains within which lymphocytes with similar tropisms can aggregate (Fig 1).

With an influx of inflammatory material via lymph to lymph nodes, or via blood in the spleen, B and T cells specific for any newly arriving antigens become activated and adjust their migratory propensities. Within lymph nodes, naïve B cells within follicles first encounter their cognate antigen delivered via lymphatic vessels, cortical sinuses or at the surface of SCS lining macrophages^110–116. Exposure to pathogen associated molecular motifs and BCR crosslinking result in increased levels of CCR7, CD86 and MHC II, as well as receptivity to CD40 signaling^{117, 118}. With this increase in CCR7, activated B cells rapidly relocate to the periphery of the follicle^119–121 and are retained in part by further expression of EBI2^{103, 104}. Activated CD4+ T cells also adjust their chemotactic propensities by an increase of their expression of CXCR5 and EBI2^{96, 122}. As a result, recently activated CD4+ T cells and B cells are similarly CXCR5+ CCR7+ and EBI2+ and both migrate to and congregate at the T-B border and interfollicular regions within LNs or the analogous bridging channels within spleens. After engaging T cells that share their cognate specificity, activated B cells commit to one of several potential fates; they either re-enter the follicle and commit to the formation of a new GC, or they differentiate outside of GCs into low-affinity short-lived antibody secreting cells (ASC), early memory B cells, or other effector B cell lineages^123–125. It is at these locations that elevated Bcl6 protein levels are first apparent in both responding B and T cells^{126, 127}. Once Bcl6 mediated transcriptional repression is fully enforced in nascent GC B cells and Tfh cells, they lose expression of chemokine receptors such as EBI2 and CCR7 that retained them at the follicle periphery, resulting in revised chemotactic patterns that instead drive them to the follicle interior shortly thereafter^{108, 127–130}.

Importantly, the requirements for commitment to the intra-follicular GC pathway differ for T and B cells. It is clear that T cell differentiation to the perifollicular Tfh state can be driven initially by their interactions with DCs 131. Bcl6 expression by a subset of responding T cells is detectable very early, within one-two days of antigen encounter^{126, 132, 133}. Interestingly, typical Tfh markers CXCR5 and PD-1 are quickly upregulated on many responding T cells, but after three days of interactions with B cells at the follicle periphery are then only maintained on a population that ultimately gains entry to the follicle and continues to interact with B cells^{126, 132–135}. In contrast, full activation of B cells to commit to the GC pathway depends on those B cells interacting with cognate T cells.

Over the course of several days, cognate B and T cells form stable immunological synapses, but repeatedly engage new partners during this protracted and formative time period, all the while remaining at the follicle periphery. Bcl6 expression by B cells is not detectable until after the emergence of Bcl6+ Th cells. Thus B cell commitment to the GC pathway, reentry to the follicle and the coalescence to form a proliferative foci there requires several days of ongoing serial interactions with perifollicular Tfh. Interestingly, Tfh entry into the follicle proceeds that of GC B cells¹²⁶, suggesting that the availability of Tfh-support within the follicle is a prerequisite for the survival of newly arriving GC B cells. Early, preGC cognate interactions that occur at the follicle periphery are therefore critical to establishing the GC and determining immune outcome.

4. Progressive GC B cell differentiation has distinctly regulated stages

4.1. Germinal center B cells are developmentally delayed

Shortly after BCR signaling in vitro and in vivo, activated B cells are well poised to engage Th cells and respond to CD40 ligation¹¹⁷. However, many cell divisions are required before Bcl6 protein expression becomes evident in responding B cells²⁶. This is true even under conditions in which activation and cell contacts are occurring with rapidity. The delay in the appearance of GC B cells had been a long-standing paradox. Multiple models have proposed mechanisms that may influence the differentiation of activated B cells to either the GC or ASC lineages. In vitro studies have suggested that B cell fate decisions could be stochastic and B cell intrinsic^136–138. Alternatively, the initial BCR signal strength^{139, 140} or competition for antigen and its presentation to cognate Th cells¹⁴¹ may be deciding factors. However, the delay in GC B cell development could also be influenced by other potential constraints. For example, Tfh cells require multiple days to complete their effector differentiation, but once complete might be capable of instructing GC B cell differentiation expeditiously. A teleologic explanation for the delay in GC B cell formation might instead relate to the need for multiple lineage fates to be created from every antigen-specific B cell after its initial activation. Perhaps multiple rounds of proliferation are key to ensuring that an investment in the germinal center response only occurs after the production of early memory and short-term ASC B cells. Here we review the results of recent studies that suggest additional mechanisms and provide some insights and into these aspects of B cell differentiation.

4.2. Impact of CD40 signaling

B cell lineage commitment is governed by opposing transcriptional networks that can be mutually antagonistic. Bcl6 acts as a transcriptional repressor of the TFs IRF4 and BLIMP1, and whereas it is essential for GC B cell formation^{93, 127, 142–144}, when over-expressed it can prevent ASC formation^145–147. Reciprocally, high levels of IRF4 or induction of BLIMP1 represses transcription of Bcl6 but facilitates ASC differentiation^{148, 149}. To further confound our understanding of transcriptional regulation, a transient and low expression of IRF4 promotes the transcription of Bcl6 and is required for proper GC B cell progression¹⁵⁰, despite the fact that it directly binds to and represses the Bcl6 promoter when at higher levels^{92, 151}. Thus, IRF4 is required for both ASC and GC development, although ASC associated genes are promoted when IRF4 is at sustained or high levels of expression^{79, 150, 152–155}.

Tfh cells express the surface co-stimulatory molecule CD40L and supply the cytokines IL-4 and IL-21 to GC B cells, all of which are critical to the formation of fully mature and productive GCs^20–26. The combined expression of IL-4 and IL-21 is unique to Tfh cells, however all Th cell populations defined to date are thought to express CD40L, and very early in an immune response^{156, 157}. Signaling via CD40 ligation invokes the non-canonical NFkB pathway and uniquely the nuclear translocation of heterodimeric RelB/p52^158–160. Although CD40 signaling is required for GC B cell development^{1, 161}, their formation can also be antagonized by chronic CD40 signaling that instead promotes ASC development^{151, 162–165}. ASC formation during chronic CD40 signaling is in part mediated by IRF4, a TF whose nuclear translocation is promoted by CD40 signaling^{151, 166}. These seemingly conflicting roles of CD40 pose a conundrum: CD40 and CD40L are required for GC B cell formation, but continued CD40 signaling discourages it.

Some insight into the paradoxical roles of CD40 was garnered by our recent work that examined the temporal relationship between CD40 signaling and TF shifts during GC B cell differentiation²⁶. In this study, B cells were first observed to express intermediate levels of Bcl6 within the interfollicular regions of LNs. Intriguingly, these B cells also were found at their first emergence to co-express RelB and IRF4 at intermediate levels, factors known to be mutually antagonistic to Bcl6^{92, 151}. Consistent with the presence of nuclear RelB and IRF4, CD40 signaling was required for this very first step in Bcl6 protein production. At this early stage, Bcl6int GC precursors retained some plasticity - the infusion of either anti-CD40 or additional antigen in vivo at this incomplete stage of development could still divert them towards Bcl6lo IRF4hi ASCs²⁶. However it is important to note that Bcl6lo IRF4hi ASC can be seen prior to the detection of Bcl6int pre-GC B cells²⁶, an observation that is consistent with the co-existence of an alternative pathway for extra-follicular ASC formation that is CD40 dependent but does not involve Bcl6.

Although CD40 signaling may establish a transcriptional landscape that is initially compatible with intermediate expression levels of Bcl6, the presence of RelB and the antagonistic TF it promotes, IRF4, is incompatible with a complete repression of Bcl6 target genes within preGC B cells²⁶. Consistent with this, preGC B cells display a partial GC B cell phenotype (CD38int, PNAint)²⁶. Bcl6 is a member of the BTB-POZ zinc finger family of TFs that forms a homodimer capable of binding the corepressors SMRT, NCOR and BCOR^{167, 168}. Bcl6 controls GC B cell differentiation by regulating cell cycle genes, repressing terminal differentiation factors, and suppressing some signaling pathways, including some aspects of B cell receptor signaling^{91, 169}. Repressed target genes in dark zone GC B cells in mice include cd38, irf4 and prdm1 (encoding BLIMP1)^{92, 143, 144}. Notably, the extent of repression of potential Bcl6 target genes is heavily influenced by co-repressors that are themselves independently regulated^{91, 147, 170, 171}. Therefore, Bcl6 expression per se does not ensure target repression. It is unclear whether the ineffective repression of Bcl6 target genes in GC B cell precursors is the result of subthreshold levels of Bcl6, or alternatively the absence of other co-repressors that would otherwise enable effective transcriptional repression^{170, 172, 173}.

Although CD40 signaling is required for this first step in preGC B cell development, CD40 signaling must cease at the immediately subsequent stage to allow for that incomplete stage of differentiation to further progress²⁶. The intermediate differentiative stage of preGC B cells is characterized by mutually antagonistic TFs (IRF4int Bcl6int) and is followed shortly thereafter by the appearance of mature GC B cells that are Bcl6hi with low levels of IRF4 and RelB and a more classical phenotype (CD38lo PNAhi)²⁶. Interestingly, the shift from the immature preGC B cell state to the more mature Bcl6hi state involves both transcriptional and post-transcriptional regulation²⁶, the latter a form of regulation that can be mediated by microRNAs^{162, 174, 175}. Importantly, disrupting T-B interactions or inhibiting CD40 signaling in vivo during this early transitional stage encourages GC B cell maturation. However, when CD40 agonism is prolonged or additional soluble antigen is introduced, GC precursors are instead expanded and shunted away from the intra-follicular GC pathway^{26, 165}. Based on these results, a model of initial GC B cell differentiation has been proposed to be a multi-staged process completed by a transient diminution in the receipt of T cell help within its immediate microenvironment (Fig 2). Here it is envisioned that GC B cell differentiation is a two-stage process involving first a T-cell contact and CD40-dependent phase that generates RelB+ IRF4+ Bcl6int GC precursors and second, a maturation phase immediately following that leads to the Bcl6hi intrafollicular state that is prevented by further CD40 signaling. Thus the impact of CD40 signaling may be stage specific in that it promotes GC B cell precursor formation, but diverts precursor progression once formed. It is intriguing that the shifting transcriptional profiles of GC B cell precursors are staged over the course of multiple cell divisions. More research is required to understand how the sequential introduction and subsequent removal of TFs is advantageous to the initial formation of GC B cells.

Figure 2.

Temporal and context dependent shifts in transcription factor composition during GC B cell development.

It is unclear whether a transient abstinence from CD40L stimulation plays a similar role in established GCs. The transition between the classical centrocyte phenotype to that of an idealized centroblast may involve progressive changes as well. LZ B cells experiencing CD40L stimulation during engagement with Tfh cells gain cMyc expression, which subsequently promotes expression of AP4, a TF that is maintained within B cells by IL-21 signaling⁷⁸. Subsequent loss of AP4 requires further cell division(s) with the DZ, a process that is needed to eventually exit cell cycle⁷⁸. The potential impact of additional CD40 signaling by DZ resident B cells, and whether it could occur in the absence of IL-21, is unclear. Since there are far fewer Tfh cells within the DZ than the contiguous LZ, this compartment offers a relative dearth of T cell derived CD40L. The sheltered spaces within the DZ compartment appear to provide other advantages as well to newly transitioning centroblasts to GC dark zones. A study by Bannard et al indicates that GC B cells unable to escape the Tfh-rich environment of the LZ have a competitive disadvantage over time³², suggesting that in the long run the cyclic movement between zones ultimately empowers a selective advantage in a way that is not yet understood. Although some analogies can be made between the initial development of GC B cells at the follicle periphery and the cyclic reprogramming of mature GCs, there are two major differences – 1) newly arriving Bcl6hi centrocytes that are seeking antigen and Tfh help are already experiencing some level of transcriptional repression by Bcl6, and 2) a reservoir of antigen is accessible within the FDC network. More research is needed to untangle this complex series of events.

How might preGC B cells transiently avoid CD40 signaling and resolve their mixed state of mutually antagonistic TFs? One means by which RelB+ pre-GC B cells could lose that status is by an asymmetric cell division (ACD) that results in one daughter receiving less RelB. Consistent with this, we have observed an additional cell division of preGC B cells correlates with reduced RelB levels and a rise in Bcl6²⁶. In this scenario, preGC B cells could undergo an ACD in which Bcl6 is unequally partitioned to one of the emerging daughters and RelB to the other, resulting in two distinct cell fates. ACDs involving an unequal distribution of Bcl6 has been reported among GC B cells, although its role in GC maintenance has been questioned^{176, 177}. Alternatively, engagement with perifollicular cognate T cells could be attenuated by a sub-threshold level of antigen presentation. Multiple cell divisions could lead to a reduced capacity to present antigen, particularly under conditions when antigen is limiting within that local microenvironment. This would be consistent with previous reports of progressively shorter T-B interaction lengths with non-infectious immunogens^{59, 126, 127}. In further support of this, conditions that instead create persistent pathogen-derived antigen promote extrafollicular ASCs and impair GC responses¹⁷⁸. Moreover, antigen that has been internalized after BCR capping has been reported to be asymmetrically distributed between daughters after B cell division, resulting in two daughters that differ in their ability to engage cognate T cells^{179, 180}. However, once GC B cells have migrated into the follicle interior, they again have access to a rich source of antigen within the FDC network.

4.3. Role of IL-21 and IL-4 in GC B cell maturation and intrafollicular development.

Similar to CD40L, the Tfh-derived cytokines IL-4 and IL-21 play distinct roles at discreet phases of GC B cell development. Although many Tfh cells found within mature GCs can produce both IL-4 and IL-21, the cytokine expression profiles of Tfh cells can shift during their development and can differ at distinct locations within lymphoid tissue^{21–23, 181, 182}. Earlier in adaptive immune responses, peri-follicular Tfh predominantly express IL-21 and less IL-4^{23, 182}, while the reverse is true within the distal LZ of fully mature GCs where Tfh cells are more likely to produce IL-4 than IL-21²³. CXCR5+ iNK T cells positioned at follicular boundaries are also capable of providing IL-21 and/or IL-4 and promoting the GC lineage development of B cells that are presenting glycolipid antigens, but the resulting GCs are short-lived and unproductive without the added participation of intrafollicular Tfh cells responding to protein antigens^183–186. Importantly, the delivery of IL-4 and/or IL-21 by iNK T cells is insufficient for the formation of sustainable GCs when CD40 signaling is absent¹⁸⁵. Other innate cell types including eosinophils and basophils are also known to produce IL-4 and can function as antigen presenting cells to Th2 cells^187–190, but they have not been observed within intrafollicular GCs¹⁸⁸ and it is unclear whether such innate cell types have the ability to provide IL-4 in a directed fashion discreetly to antigen-specific B cells.

Both IL-4 and IL-21 are critical to GC B cell formation. Although B cells that are unable to signal via IL-21R or IL-4Rs can still develop GCs in vivo, the stunted GCs that do form are significantly reduced in size^{182, 191–194}. In vitro, both IL-21 and IL-4 can enhance the transcription and/or translation of Bcl6 in B cells^{195, 196}. Both cytokines are also known to promote ASC formation^{22, 197–201}. In the case of IL-21, an inability to signal via its receptor has a direct effect on B cells, resulting in lower Bcl6 levels among the more limited number of GC B cells that form^{182, 191, 192}.

IL-4R and IL-21R complexes activate distinct signaling pathways. Like other cytokine receptors employing the common gamma chain, both IL-21R and IL-4R signal via the Janus kinase molecules JAK1 and JAK3^202–205. However, these cytokine receptors activate different signal transducer and activator of transcription (STAT) molecules. Whereas IL-21 predominantly activates STAT3 and STAT1, and to a lesser extent STAT5,^206–209, IL-4 and IL-13 uniquely and requisitely activate STAT6,^{210, 211}.

The results of a recent study we performed suggest that B cell signaling by these cytokines may influence GC B cells at different stages¹⁸². These experiments employed the cell transfer of antigen-specific B cells that were either deficient in IL-21R, STAT6, or both in order to assess the impact of IL-21 and IL-4/IL-13 to B cells intrinsically. We found that the initial transition from the peri-follicular preGC B cell state to the intra follicular stage is reduced in IL-21R−/− B cells, but unaffected in STAT6−/− B cells, suggesting that IL-21 has a greater influence on differentiation at the time of preGC B cell re-entry to the follicle interior¹⁸². Importantly, in the absence of both IL-21 and IL-4 signaling pathways, Bcl6+ B cells can still form, however they fail to establish intrafollicular GCs¹⁸². These findings support the idea that GC B cell development undergoes a step-wise progression in which the initial steps toward this lineage can occur in the absence of IL-4 and IL-21, but is dependent upon CD40L. The Bcl6-expressing B cells that form without IL-21R or STAT6 signaling can persist transiently in vivo, but their expansion is limited, indicating that GC B cell development can be arrested at this stage. This could functionally provide a key checkpoint in this lineage pathway that would allow for the initial creation of GC precursors, without committing to their furthered proliferation. In this way, a diverse repertoire could be potentiated without a concomitant clonal expansion of antigen reactive B cells for which there are not yet cognate Tfh cells capable of sustaining a mature GC.

The results of this study further suggest that proper GC form and function within follicles is highly reliant on both cytokines¹⁸². IL-21 and IL-4 play non-redundant roles in the subsequent maturation and self-renewal of GC B cells within follicles. Smaller numbers of intra-follicular GC B cells form in the absence of either of these signaling pathways, but with aberrant phenotypic features¹⁸². When GCs mature in the absence of either IL-21 or STAT6 signaling, the efficiency in the transition between the centroblast and centrocyte states is compromised and self-renewal is reduced^{182, 197}. IL-21R−/− B cells can form CXCR4hi centroblasts but they overexpress some features that are more typical of centrocytes such as CD40 and CD86¹⁸². Moreover, IL-21R−/− B cells are predominantly located within the FDC network, further suggesting that IL-21 signaling is required to properly instruct the centocyte to centroblast transition. By contrast, IL-4 dependent signaling appears to have a greater contribution to the proper formation of the centrocyte profile. Although STAT6−/− GC B cells have a broadly normal spatial distribution within their smaller GCs, albeit with a smaller percentage of centrocytes, the CXCR4lo centrocytes that form aberrantly underexpress CD86. Together, these results suggest that newly generated centroblasts formed without prior IL-4 exposure might be compromised in their ability to recapitulate the complete centrocyte program. This is consistent with a study by Turqueti-Neves and co-workers that examined the response of IL-4 deficient mice to helminth infection and observed a preponderance of DZ phenotype GC B cells that could be corrected by the introduction of IL-4 sufficient cells in BM chimeras¹⁹⁴. Together, these intriguing studies raise more questions than they answer. Can these signals be received independently during sequential cell contacts with Tfh cells that differ in their cytokine profile? Are DZ centroblasts engaging Th cells within that compartment? Although more work is needed to address these and other questions, it is clear that the programing of GC B cells capable of responding to both cytokines is different from that obtained with one but not the other.

5. Establishment of intra-follicular GCs and the coevolution of follicular stromal cell niches.

The cellular composition of tissue compartments effectively creates micro-climates that can be quite different. The cell types residing in each compartment, together with the cytokines and chemokines they secrete, likely shape the nature of lymphocyte activation, proliferation and differentiation^{10, 110}. Differences in local accessory cells ensures that B cell experiences within the molecular milieu of one compartment will be different from those experienced in another compartment. For example, the composition of the stromal cell population, their activation status, and any molecular products could potentially create a microenvironment that fosters modes of B cell growth or effector lineages that are distinct from those occurring elsewhere. Although stromal communities may delineate the boundaries of neighborhoods defined by characteristic chemokine compositions, they are themselves capable of change under inflammatory conditions²¹². Here we review the progression of events that occur within the follicle as Tfh and GC B cells arrive, stromal cell populations evolve, and the unique compartments of the mature GC are established.

5.1. Early intra-follicular events and FDC maturation

Follicular dendritic cells are derived from a stromal cell subset that uniquely differentiates within B cell follicles and are the primary source of CXCL13 in this zone of primary lymphoid tissues^{97, 213, 214}. Within lymphoid tissue of naïve mice, FDCs can be identified as uniquely expressing the long variant of the complement receptor CR1 (CD35) within follicles¹⁴. They are centrally and sparsely distributed within primary B cell follicles and are substantially less dendritic then is typically observed within mature GCs (Fig 3a,b). Immunization promotes the furthered development of FDCs from perivascular associated stromal cells, which then protrude long dendritic extensions forming an extensive network of dendrites⁹. Immunization can also elicit the formation of FDC-like cells from the marginal reticular cells (MRC) that descend from the SCS of LNs or marginal sinus lining cells of spleens, and hence contribute to the CD35+ CXCL13-producing dendritic network closer to the outer follicle sinus^{98, 99}. Both a numerical increase and furthered differentiation occurs with these MRC-derived FDCs, known to uniquely produce RANK-L within the follicle⁹⁹.

Figure 3.

Stages of GC maturation and associated tissue niches. (A) Schematic representation of the stages of germinal center (GC) maturation. (B - E) Immunofluorescence staining of popliteal lymph node (B) and spleen (C-E) from C57BL/6 mice that received an adoptive transfer of OVA-specific T cells and antigen-specific B cells enriched by immunomagnetic purification (StemCell Technologies). T cells specific for OVA were obtained from spleens of OT-II mice while either B1-8+ Jκ-/- or MD4 mice served as donors for anti-NP or anti-HEL specific B cells respectively. Recipients were immunized after transfer with 50 μg of either NP-OVA/CFA in the footpad (lymph node, panels B), NP-OVA/alum injected i.p. (spleen, panels C, E), or HEL-OVA/alum (spleen, panel D). (B) Popliteal lymph node histology time-course demonstrating GC maturation after immunization. Stages of maturation represented include the unimmunized follicle (d0/naïve), maturation of the FDC network and light zone expansion (day 4), centroblast increase, zonal segregation and follicle swelling (day 5), and a mature GC (day 10). Sections were stained to highlight CD4+ T cells (blue), IgD+ non-responding B cells within the follicular mantel (green), and CD35+ FDC network (red). (C) Histology showing a maturing GC (day 6) stained to highlight IgD+ naïve B cells (grayscale), PNA+ GC cells (green), and the distribution of sonic hedgehog (SHH, red) in relation to the FDC network stained with CD35 (blue). (D) Representative histology of splenic GCsdemonstrating initial differentiative events shortly after the arrival of Ag-specific B cells to the follicle interior(day 3). Sections were stained to highlight GFP+ Ag-specific B cells (green), CD35 FDCs (blue), Bcl6 (red) toidentify GC committed B cells, and IgMa to assess the level of cytosolic immunoglobulin. Arrows point to Bcells within the FDC network that are cIg hi (consistent with ASC effector lineage), Bcl6hi (GC lineage), andBcl6neg cIg lo (of an undefined alternative fate). (E) Histology showing GFP+ Ag-specific B cell interactionwith dendritic cells (DC) at the T-B border (day 6). Section were stained for CD4+ T cells (blue), B220+ Bcells (red), GFP+ Ag-specific B cells (green) and CD11c+ DC cells (grayscale). Higher magnification imagesof the T-B border demonstrate the ability of GC B cells, DCs and T cells at the T zone border to directly engage each other.

After Tfh and preGC B cells have further differentiated at the interfollicular zone and T-B border, they are observed to migrate directly from that region toward the follicle interior, with the migration of disengaged cognate T cells occurring independently and preceding that of GC B cells¹²⁶. Responsiveness to CXCL13, downregulation of EBI2 and CCR7 all promote this initial intra-follicular movement^{10, 103, 104, 129}. Once within the follicle, Bcl6hi B cells are primarily found in the outer follicle, the upper follicle more distal to the T cell zone, and appear there typically 3–4 days post immunization in non-infectious experimental systems^{126, 127}.

FDCs undergo a form of activation after immunization that is influenced by TLRs and NF-kB signaling pathways, as well as immune complex binding^215–218. This activation results in an increase in the expression of numerous proteins that can play a role in GC B biology, including Fc receptor FcγRIIb, complement receptors CD21/35, ICAM1 and VCAM1^{215, 217–219}. FcγRIIb together with the complement receptors CD21/CD35 mediates the bulk of the antigen retention by the binding to proteins in immune complexes or tagged with activated complement components²²⁰. After immunization, FDCs also increase their production of Mfge8 that will promote the phagocytosis of emerging apoptotic B cells by tingible body macrophages^{7, 221}. FDC activation and dendrite extension is similar in timing to the arrival of Tfh and GC B cells to the follicle interior. Notably, Tfh cells when first arriving to the follicle interior within LNs can first be found primarily in contact with the centrally located and underdeveloped CD35+ FDCs, but are also found in the follicular space above nearer the SCS (Fig 3b and^{126, 127}). Although coincident, there is only in vitro evidence to suggest that Tfh cells can influence FDC activation²²²; there is no evidence to date in vivo. However, the TLR dependent activation of FDCs suggests that some aspects of their furthered differentiation can take place after exposure to adjuvant alone²¹⁶.

As the GC matures, FDC network differences become more apparent. ICAM, FCγRIIb expression²²³ and sonic hedgehog (SHH) expression (Fig 3c and²²⁴) by FDCs is more evident within the GC LZ, and in particular at the LZ-DZ interface, than is observed with the FDCs within the FM more proximal to the SCS. Given the positioning of MRC-derived FDCs closer to the SCS of LNs, it may be that their contribution to the FDC network found within the FM is larger⁹⁹. The functional differences between these FDC zones suggest that the nature of the cellular interactions with lymphocytes within these distinct domains could differ. For example, SHH found within GC LZs is known to influence self-renewal and cell-fate determination in other tissue compartments, including as hematopoietic niches, hair follicles and thymocyte selection^225–228. In support of the idea that SHH could influence LZ centrocytes, it has been reported that a subset of GC B cells express both of the required components of the Hedgehog (Hh) receptor, Ptc and Smo²²⁴. Interestingly, expression of Smo can be increased with CD40 or BCR signaling²²⁴, suggesting that successful engagement with stimulatory antigen and/or Tfh cells would empower hedgehog signaling. Hedgehog signaling in GC B cells promotes their viability in vitro and reduces Fas-induced apoptosis²²⁴, providing further evidence that within the GC LZ both FDC-derived and Tfh-derived factors can promote protection from the apoptosis prone state of GC B cells^{215, 229}.

5.2. B cell differentiation within the newly formed LZ

Studies that have examined the formation of BM-resident long-lived plasma cells have observed that the plasma cells that persist are predominantly generated after the peak of the GC response^{125, 230}. However, there is evidence of B cell differentiation toward this effector lineage within the FDC network, even at this very early stage. In addition to Bcl6+ GC B cells, antigen specific B cells with large amounts of cytosolic immunoglobulin (cIg) can be seen centrally placed in follicles and adjacent to FDCs (Fig 3d,¹²⁶). It is unclear whether the interactions with accessory cells within the still evolving LZ microenvironment are sufficient to instruct the transcriptional program that is needed for longevity in plasma cells²³¹.

At this same place and time, within the emerging GC LZ, responding B cells lacking both Bcl6 and cIg can also be observed, suggestive of an alternative lineage fate such as a GC-derived memory B cell. Although the lineage fate of such Bcl6neg antigen-specific B cells remains unclear, and whether they have the capacity to persist long-term is unknown, these observations nevertheless suggest that not all GC B cell progeny are committed to Bcl6 expression and self-renewal in this arena even during this stage of LZ expansion. Although Tfh cells are already in residence at this point in time, it is intriguing that the first events are not entirely devoted to centroblast formation. Given that the production of long-lived memory B cells is the predominant effector cell output of younger GCs that persists^{125, 232}, it is tempting to speculate that the unique conditions of this stage may contribute to that lineage-fate preference. There are multiple differences within newly formed GCs (days 3–5 with most immunization regimes): 1) FDCs have more recently undergone activation, 2) antigen deposits may be less likely initially to be in complexes with high-affinity antibody, possibly enabling more processes reliant on activated complement^{220, 233}, 3) newly arrived Tfh cells are more prone to produce IL-21 than IL-4 than at later stages in GC maturation²³, and 4) LZ resident B cells have just transitioned from the peri-follicular preGC program and could therefore theoretically not have yet obtained the centrocyte program or higher affinities that are typical of a late time point^{55, 234}.

5.3. B cell expansion and effector differentiation at peripheral sinuses

When colonization of the FDC network by GC B cells has just begun, B cell differentiation toward other non-GC lineages is also apparent by histology at other peripheral arenas. Strikingly, at this very early stage of the immune response, expanded populations of antigen-specific B cells form immediately under the SCS of LNs^{126, 127} and at splenic marginal sinuses^{235, 236}. This phenomena is short lived and quickly diminishes as the GC response progresses. During this early phase, responding B cells at the SCS were seen to exit follicles via the sub-capsular lumen of LNs during intravital imaging, but importantly, the reverse was not true – antigen-specific B cells near the SCS were not observed to enter the follicle via that route, strongly suggesting that proliferative B cells at the SCS do not contribute to the GC reaction¹²⁶. In this study, Bcl6 protein was not apparent within B cells at that location as assessed histologically. This is in contrast to the results of another histologic study that discerned a temporal relationship of first a predominance of Ki67+ GL7+ B cells adjacent to the marginal sinus, followed by an appearance of aggregated GL7+ B cells in follicle interiors. Here GL7 expression was presumed to be a marker uniquely expressed by GC destined B cells. Although GL7 expression appears restricted to B cells within GCs at mature timepoints⁸⁷, it is important to note that it is also known to be expressed by LPS stimulated B cells in vitro, including by B cells forming ASC⁸⁹ as well as by IgM+ early memory B cells²³⁷. In the Coffey et al study, Bcl6 levels were assessed by qRT PCR of mRNA²³⁵. However, Bcl6 can be transcribed with little translation to protein in some cell types¹⁶². Thus the difference in these studies vis a vis GC B cell formation might therefore relate to the well known post-transcriptional regulation of Bcl6^238–240

There is some evidence of ASC development among activated B cells at LN SCS. Some SCS-adjacent B cells have copious amounts of clg light chain, consistent with commitment to a ASC lineage¹²⁶. However in another study, B cells adjacent to splenic marginal sinus lacked the CD138 expression that is characteristic of late stage plasma(blast) cell formation and had not undergone class switch recombination to IgG²³⁵. Together the results of these studies remain consistent with the idea that a subset of sinus lining B cells are in the process of differentiating to IgM ASCs. In addition to the SCS, ASC cell formation can also be seen at the T-B border, and among the remaining cognate B and T cells within the interfollicular region^{126, 241}. Consistent with this, prdm1-expressing ASC B cells have also been observed to move from deeper peri-follicular locations toward medullary sinuses⁹⁰. The lineage fates of the non-ASC B cells at these alternative locations remains unclear, but given the absence of Bcl6 protein and prior reports of very early IgM+ memory B cells that are GC-independent²³⁷, it is likely that a subset is destined to become early memory B cells or alternative B cell effectors^{123, 125}.

5.4. Zonal segregation and DZ evolution

Depending on the immunogen and mode of immunization, the emergence of a discernable DZ may not occur for another day or two. This zonal segregation requires, by definition, the presence of GC B cells outside of the FDC network, but as described below there are several other unique features of the DZ compartment that evolve during the course of the response. One key element of the centroblast program is the increased expression of CXCR4, which enables centroblast migration to CXL12 producing stroma initially found most proximal to the basement of the follicle, near the T-B border^{10, 33, 242}. These CXCL12-producing reticular cells (CRC) also have a reticular morphology, although the dendritic extensions lack the high levels of CD35 expression and complement binding that are distinctive for FDCs^{31, 32, 242}. Interestingly, in some but not all GCs, there are DZ areas that are not filled by a CRC network²⁴². The DZ stromal cells without CRC features are as yet uncharacterized, but are found closer to the DZ/LZ interface and at this position could potentially shelter GC B cells from the influences of both CRC and FDCs. In this regard, the distribution of freshly minted centroblasts is intriguing. The emerging centroblasts typically form both a collection immediately under the FDC network as well as a loose column that stretches toward the T-zone border (Fig 3b,^{141, 182}). This is consistent with chemotaxis toward CXCL12 as the primary drive for movement out of the FDC zone¹⁰.

CRC can differ phenotypically and morphologically within the DZ, perhaps relating to differences in either their derivation or mode of activation. Although fibroblastic reticular cells (FRC) similar to T zone FRC are only peripherally present at the edge of the B cell follicle in naïve LNs, after immunization the expanding GC swells the follicle, causing both IgD+ and GC B cells to shift past the large cortical lymphatic sinuses present at the naïve border and into the T zone FRC rich area^{100, 105, 106}. Results presented by Mionnet et al are consistent with the idea that at least some DZ stroma may derive from T zone FRC cells¹⁰⁰. In this way, new microenvironments are created by evolving stromal cells thereby forming a novel venue for events that might not have otherwise been possible, in this case an avenue bringing DZ GC B cells in close contact with the T zone border. Like other T zone FRC, the stromal network newly infiltrated by the expanding follicle also exhibits a collagen-dense core typical of FRC-ensheathed lymph conduits^{243, 244}, a feature that is less apparent among the more centrally positioned CRC within mature DZs^{32, 242}. Given that CRCs are also found within primary non-reactive B cell follicles²⁴², it seems possible that CXCL12-expressing stroma within DZ compartments can derive from more than one cell type.

In addition to the blasting centroblasts, GC DZs also harbor follicular T cells as well as DCs (Fig 3b). Not all follicular T cells are located in the LZ. They can also be found within the center of DZs as well as at the outer edges of DZs adjacent to the FM, albeit in lower densities (Fig 3b,^{21, 245}). Although little is known about how these CD4+ T cells and DC subsets might differ from their counterparts at the T-B border and GC LZ, there is evidence to suggest that Tfh cells producing IFNγ or IL-21 can localize to the DZ outer edge within the follicle^{21, 23}. Follicular regulatory T cells (Tfr) have also been observed within DZs²⁴⁵. Such DZ resident T cells are in a unique position to influence centroblast behavior, either favorably as they might engage Tfh cells within the LZ, or in the case of DZ Tfr cells influencing centroblast self-renewal or differentiation to long-lived memory or plasma cells^246–248.

5.5. Follicular Mantle

As the GC expands, GC B cells form a large egg-shaped aggregate that displaces non-responding B cells, causing them to form the follicular mantle. Intravital imaging has demonstrated that non-responding B cells are not excluded by a physical barrier and can move freely through GCs^{70, 72}. Although early stages of GC development show extensive admixing with IgD+ non-cognate B cells (Fig 3b), this is uncommon within mature GCs. The follicular mantle harbors more than just non-responding B cells. In addition to the sinus-proximal FDCs described above, a distinct subset of Tfh cells can also be found within FMs. The follicular mantle surrounding mature GCs harbors Tfh cells that are phenotypically and functionally distinct, with a lower expression of SAP and IL-4^{22, 122}. After the resolution of a primary response, both memory B cells and memory Tfh cells can be found within FMs, a positioning that is influenced by their EBI2 expression^{122, 249, 250}.

Theoretically, this compartment could influence the differentiative state of antigen-specific B cells that are leaving the GC via this route. As cells within this arena are less likely to be in active cell cycle, this environment may be more conducive to gaining a quiescent state, sheltered from the cellular interactions that promote a centroblast program and their self-renewal. It is important to note that this is not the predominant route of exit for the prdm1-expresing ASCs leaving GCs, which largely leave via the GC outer zone through the DZ base due to their down-regulation of CXCR5 and increased responsiveness to CXCL12^{82, 90, 251–253}. However, the FM could be the preferred exit route of GC-derived memory B cells, as has been shown for Tfh memory cells during secondary responses¹²². CCR6 expression has been identified as a distinctive feature of memory B cells^{254, 255}, including those observed to be completing their differentiation within GC LZs²⁵⁶. A ligand for CCR6, CCL20, is known to be produced by some epithelial cell types²⁵⁷ and importantly by lymphatic endothelial cells lining the subcapsular sinus of LNs near B cell follicles, the latter shown to be key to the localization of IL7Ra+ CCR6hi innate-like lymphocytes²⁵⁸. FM non-responding B cells are also CCR6hi, in sharp contrast to CCR6lo GC B cells. However naive B cells lack responsiveness to CCL20 despite expressing the receptor for this chemokine, unlike CCR6hi memory B cells which chemotax well to CCL20²⁵⁵. Although memory B cells have been reported to be positioned within lymphoid tissues at marginal zones, perifollicular zones and some follicles after the resolution of response^{255, 259, 260}, two studies have reported a predominant FM location for Ki67-memory B cells in spleens and LNs^{249, 250}. Although the above studies support the idea that memory B cells primarily exit via follicle adjacent sinuses, it is worth noting that CCR6hi memory B cells are also responsive to CXCL12²⁵⁵.

5.6. The unique niche of the DZ-T zone border in mature GCs

The follicular mantle is thin to non-existent at the true base of the DZ in mature GCs. In cross sections of mature GCs that are central and en face, IgD+ B cells can be infrequent or absent at the base of the DZ. This DZ-T zone interface is commonly observed in splenic and LN GCs when canvased in their entirety via serial sections (Fig 3e, Haberman lab unpublished data and^{32, 82, 242, 261}). The DZ-T zone interface can be quite extensive in larger GCs and could theoretically allow DZ GC B cells to interact with cell types that are not present in other GC compartments.

This arena of mature GCs is poorly understood but clearly has the potential to be an influential venue for self-renewing centroblasts as well as GC-derived long-term plasma cells. Maturing plasmablasts can be found at this location^{82, 251, 262}, and in a recent study found to be in association with CD157hi podoplanin+ fibroblastic reticular cells as early as 6 days post immunization²⁴¹. Similar to the CRC cells within the DZ, the reported CD157hi FRC at the DZ base of young GCs also express CXCL12, but appear to lack dendritic extensions into the DZ^{241, 242}. An intriguing analysis of LN stromal cells by single-cell RNAseq further suggests that FRC at the T zone interface with B cell follicles are heterogenous, and may include both CCL19lo and CCL19hi subsets, as well as activated subsets responding to inflammation¹⁰⁹.

There are several features to this unique niche that have the potential to influence centroblasts and other DZ resident cell types at this DZ-T zone border. Here there are opportunities for direct contact with peri-follicular Th cell and DC subsets (Fig 3e), as well as Tregs²⁴⁵. There is also the potential for fresh antigen delivery via incoming local lymph within adjacent sinuses^{106, 243, 263}. Scavenging APCs could theoretically participate in antigen acquisition from local sinuses, similar to DCs probing lymphatic sinuses or FRC junctures^{244, 264}, and provide antigen processing and presentation to peri-follicular Tfh cell present at the DZ-T border. In this regard it is interesting to note that some DCs have a specialized capacity to retain and present antigens in native form without degradation, theoretically permitting BCR crosslinking of centroblasts in direct contact with DCs presenting incoming lymph-borne antigen^{115, 265}. Thus at this niche, DZ GC B cells may have access to antigen to process and present, as well as APCs and T cell help, which in toto could be sufficient for any of the proposed mechanisms for GC B cell selection^{33, 43, 45, 59}. Consistent with this niche playing a role in DZ B cell dynamics, movement to and from the DZ-T zone border has been observed by intravital imaging⁷⁰. Future work in this arena is very much needed to gain insight into this poorly understood niche.

Conclusion

GC development is a complex series of events that requires that cognate B and T cells periodically engage during discreet stages in their differentiation and at distinct locations within tissue. While GC B cell development begins at the periphery of B cell follicles, this line of differentiation is progressive and is characterized by step-wise changes in transcription and phenotype over the course of multiple cell divisions. GC B cell development can be paused at an intermediate stage of differentiation when conditions are not quite right, permitting the refinement of the immune response when antigen is in excess, or when the differentiation of regional activated T cells has not sufficiently progressed. Under conditions when recently activated CD40L+ CD4+ T cells are present, but IL-21 and IL-4 are lacking, preGC B cells will transiently persist at perifollicular locations. This is in part regulated by a stage specific effect of CD40 signaling; while an early phase in GC B cell development is dependent on CD40 signaling, the subsequent phase is refractory to CD40 agonism and likely involves a transient abstention from T cell help at the periphery of B cell follicles. When IL-21 or IL-4 signaling also occurs, then GC B cell maturation is allowed to progress to the next stage that empowers follicular entry, but only if CD40 agonism is transiently avoided. The migration of Tfh cells into the follicle interior before GC B cell arrival might therefore help at multiple stages and locations by: 1) vacating the perifollicular sites and thereby reducing the chance of CD40L stimulation, and 2) entering into the follicle in advance, immediately ensuring that centrocyte and centroblast instruction occurs properly and sustainably, events that are reliant on both IL-21 and IL-4. Thus early B and T cognate interactions at the follicle periphery change in quality as they co-evolve, critically influencing when and how intrafollicular GCs are established.

All these early events are temporally and spatially choreographed by the serial movements of cognate B and T cells to different and discreet tissue locales. The stroma at these tissue destinations are themselves influenced by the evolving immune response in their arena. Together these interactions help to ensure that activated B cells receive the correct signals for their stage of GC B cell differentiation and coordinate the timing and extent of that differentiation. Finally, we conclude that it is advantageous for GC B cell differentiation to be staged and slowly progressive in that this creates checkpoints that allow for the initial creation of GC precursors, without committing to an energy intensive, and potentially misdirected and unproductive germinal center.