A synchronous IRF4-dependent gene regulatory network in B and Th cellsorchestrating the antibody response

Abstract

Control of diverse pathogens requires an adaptive antibody response, dependent on cellular division of labor to allocate antigen dependent B and CD4⁺ T cell fates that collaborate to control the quantity and quality of antibody. This is orchestrated by the dynamic action of key transcriptional regulators mediating gene expression programs in response to pathogen-specific environmental inputs. We describe a conserved, likely ancient, gene regulatory network that intriguingly operates contemporaneously in B and CD4⁺ T cells to control their cell fate dynamics and thus, the character of the antibody response. The remarkable output of this network derives from graded expression, designated by antigen receptor signal strength, of a pivotal transcription factor that regulates alternate cell fate choices.

Diverse antibody production requires both B and CD4⁺ T cells in vertebrates

Upon antigen-dependent activation of the B cell receptor (BCR, see glossary) and supportive signals from CD4⁺ T cells, B cells differentiate into cells capable of BCR diversification by somatic hypermutation (SHM) and/or class switch recombination (CSR), antibody secretion, or long-term memory [1] (Figure 1). Each of these possible cell fates are governed by state-specific gene programs and associated transcriptional regulators. For example, germinal center B cells (GCBs) are dependent on the transcription factor (TF) Bcl6 for differentiation, and on the activity of the enzyme Activation-Induced Deaminase (AID) for SHM and CSR [2]; by contrast, antibody secreting plasma cells (PCs) depend on the activation of TF Blimp-1 for their terminal differentiation [3], [4]. Thus, distinct cell types execute antibody quantity (secretion) and quality (a product of diversification).

Figure 1. Naïve B and CD4⁺ T cell progenitors activate and differentiate into progeny that can act as either effectors, or cells inducing antibody diversification in vertebrates.

Upon antigen specific activation and co-stimulation, B cells can terminally differentiate into effector, antibody secreting plasma cells (PC). Conversely, the B cell can participate in the germinal center and undergo affinity maturation - a process of mutation and selection that modifies the antigen specific BCR repertoire. Likewise, after antigen specific activation and co-stimulation, CD4⁺T cells can terminally differentiate into effector (Teff), pro-inflammatory cytokine secreting cells. Alternatively, the CD4⁺ T cell differentiates into a T follicular helper cell (Tfh), and participates in the germinal center reaction to select high affinity B cells (germinal center B cells: GCB). Not discussed here are progenitor memory cells - long lived cells for rapid reactivation upon a second antigen encounter. However, where memory cells arise in this process remains to be determined.

Antigen engagement of the T cell receptor (TCR) drives the differentiation of CD4⁺ T cells that orchestrate an appropriate immune response [5]. Similar to B cells, CD4⁺ T cell fate choices are dictated by distinctive gene programs and transcriptional regulators [6]. CD4⁺ T follicular helper (Tfh) cells, dependent on the TF Bcl6, provide help to B cells in the form of survival, activation, and regulatory signals during the process of affinity maturation [7]–[9]. In contrast, CD4⁺ T effector (Teff) cells, dependent on the transcription factor Blimp-1, migrate to peripheral tissues to participate in local immune processes [10]–[13]. Thus, the antibody response is dependent on the proportional allocation of Tfh cells, which is in turn regulated by the relative expression of Bcl6 and Blimp-1 [14].

Bcl6 and Blimp-1 act as counter-antagonists to mediate cell fate allocation. Findings that the TF IRF4 was both essential and upstream of the Bcl6 and Blimp-1 cell fate determinants in both B and CD4⁺ T cells, prompted the formation of a novel model for how cellular concentrations of IRF4 quantitatively and qualitatively might tune the antibody response [15]–[18]. Herein, we describe the genetic and signaling perturbations that led to the description of a primary mechanism of B and CD4⁺ T cell alternate cell fate choice. We provide a molecular view for how B and CD4⁺ T cells integrate IRF4 cell concentrations and assemble a core gene regulatory network (Box 1) that explains the dynamics of the network (Box 2) in orchestrating alternate cell fate choice and ultimately, the nature of the antibody response (Figure 2, Key Figure). Altogether, a larger model emerges whereby IRF4, operating contemporaneously in B and CD4⁺ T cells, may enable the collaboration of specialized antigen specific B and CD4⁺ T cells to regulate the makeup of the mammalian antibody response [15], [16]. The information in this review pertains to experiments conducted either on mouse and/or human cell lines, and/or primary mouse and/or human cells.

Box 1. State-specific gene programs and Gene Regulatory Networks (GRNs).

Upon antigen recognition, B and T cells integrate the second messenger signals of cell activation to drive cellular differentiation. Many key TFs and epigenetic mechanisms, as well as miRNAs and lncRNAs, have been identified that orchestrate necessary changes in gene expression [1]. Intense research over the past 15 years has illuminated the key role of GRNs in specification and commitment of diverse effector cell fates [109]. GRNs are assemblies of both positively and negatively acting TFs that operate to drive gene programs important for specific cell fate trajectories. As such, distinct GRNs are deployed that depend on a unique set of transcriptional regulators for each cell fate. Because a given naïve cell and its corresponding antigen receptor is thought to be capable of differentiating into any given effector cell, it is hypothesized that cell state-specific GRNs are coordinated with each other in a dynamic manner to allocate the correct proportion of each cell type appropriate for a given immune response [124]. The dynamics of a GRN are governed by both positive and negative inputs as well as by biophysical features that include time, abundance and affinity. Thus, the architecture of a GRN is key to the precise and timely allocation of cell fate [134]. Coherent networks function in a feed-forward manner to commit to a specified cell fate. Conversely, incoherent or paradoxical networks enact both positive and negative regulation on the same target that enable temporal or fold change detection capabilities in response to graded inputs [109]. For example, signal strength of the antigen receptor’s response to antigen is a graded input. This review focuses on a conserved, incoherent GRN, linked to the antigen receptor of both B and CD4⁺ T cells to shape the antibody response by cell type allocation. Specifically, the incoherent GRN is based on a single activator, IRF4, which regulates the expression of two counter-antagonistic cell fate determinants, Bcl6 and Blimp-1.

Box 2. Modeling the IRF4-based Gene Regulatory Network (GRN).

The IRF4 GRN is an incoherent network, where a single activator (IRF4) functions to promote the expression of two mutually antagonistic cell fate determinants (Bcl6/Blimp-1). A multi-scale mathematical simulation has been used to elucidate these dynamics over time [40]. Ordinary differential equations estimated the synthesis and degradation rates of multiple molecular species into the strength of a given regulatory interaction and were applied to the GRN as a function of varying IRF4 amounts. When the choice to undergo CSR or PC differentiation was analyzed, a dynamic emerged termed “kinetic control”. The kinetic control model states that activated B cells pass through an obligate state that promotes CSR prior to becoming a PC[40]. This obligate state is unstable when IRF4 accumulates to high amounts and antibody diversification is aborted in favor of antibody secretion. In contrast to a bifurcating cell fate decision, where B cells either diversify their antibody or secrete their antibody, kinetic control suggests that all cells gain the ability to diversify but this opportunity is terminated if a given cell accumulates high amounts of IRF4[40]. Kinetic control also likely coordinates the timing of CSR in vivo [95]. Like CSR, the GCB state also involves antibody diversification and exhibits low amounts of IRF4 expression; however, GCBs are stable unless the BCR and CD40 are appropriately engaged during selection. To accommodate the new findings of IRF4’s role in GCBs [16], [18], the original kinetic control model would need an additional negatively regulatory node to prevent further IRF4 accumulation and to maintain the GCB state. Potential negative regulatory candidates include Bcl6, which binds the IRF4 locus [91], [96], and Cbl-b and c-Cbl, which degrade IRF4 protein [97].

Given the conservation of the IRF4 GRN in B and CD4⁺ cells, we speculate that the kinetic control model might also function in Tfh –Teff cell fate choices, where the Tfh state represents the obligate state through which all CD4⁺ T cells pass. However, despite the similarities, the maintenance of IRF4 expression in differentiated CD4⁺ T and B cells is unique to each lineage. Specifically, PCs require and maintain high amounts of IRF4. In contrast, in protein-adjuvant immunization settings, both Teff and Tfh cells express comparable IRF4 amounts [15], suggesting that CD4⁺ T cells might retain memory of antigen encounters by adopting different fates, but not by the amount of IRF4 they express. As Blimp-1 is often associated with irreversible terminal differentiation, dampening IRF4 expression in Teff cells might be one of many mechanisms that maintain plasticity of CD4⁺ T cell responses. A fuller interpretation of these results awaits analysis of IRF4 expression dynamics in infectious systems where antigen availability is actively replenished.

Figure 2. (Key Figure). Variations in the antibody response by independent control of IRF4 concentrations in B and CD4⁺ T cells.

A) Topology of the IRF4-dependent GRN orchestrating alternative cell fate choice. The network combines an activating connection between the antigen receptor and IRF4 with an incoherent type 1 feedforward edge topology (from IRF4 to Blimp-1/Bcl6). B) Left, Intensity of antigen receptor signaling scales with IRF4 concentrations. Right, dynamics of the type 1 incoherent feedforward motif in regulating alternate cell fate choice as a function of antigen receptor signal intensity / IRF4 concentrations. The architecture of the incoherent motif is predicted to compute fold change detection of IRF4 concentrations to execute alternate cell fate choice by regulation of Blimp-1 or Bcl6. C) During T-dependent antibody responses, independent control of IRF4 concentrations by antigen receptor signal intensity in T and B cells is predicted to result in different outcomes on the quantity and quality of the antibody response. PC: plasma cells; GCB: germinal center B cells; Teff: effector T cell; Tfh: T follicular helper cell.

Core components of the B and CD4⁺ T cell IRF4 Gene Regulatory Network

The antigen receptor: BCR and TCR

Upon antigen binding by antigen receptors, a “signalosome” is assembled that recruits a variety of intermediates which branch to activate multiple second messenger signaling cascades [19]. The Irf4 locus in mammals is an immediate early gene induced by antigen receptor signaling in both B and CD4⁺ T cells [20], specifically through the downstream effectors NF-kB, NFAT, and mTOR [21]–[23]. Thus, antigen receptor signaling activates multiple second messenger cascades that lead to broad cellular activation, and which also converge on the Irf4 locus. Although IRF4 cellular concentrations are important for B and CD4⁺ T cell differentiation, the basis for, and whether the converging pathways are additive, synergistic, or subject to Boolean logic, remains to be explored.

CD4⁺ T and B cell differentiation is affected by both the biophysical properties of antigen recognition and subsequent signaling intensity [24], [25]. High signal strength in B cells enhances the proportions of effector, antibody secreting PCs, whereas low signal strength favors antibody diversification through CSR in the germinal center (GC) [26]–[31]. In CD4⁺ T cells, greater antigen receptor signal intensity favors the formation of inflammatory effector CD4⁺ T cells (Teff) cells, whereas lower signal strength favors differentiated cells that function to help B cells [15], [32]–[34]. While the “signalsome” is common to both strong and weak antigen receptor signaling, it is a matter of current research to identify the nature of the receptor signal strength that regulates cell fate allocation during the immune response.

Transcription factor IRF4

The Interferon Regulatory Family (IRF) includes a group of TFs involved in signaling pathways essential for multiple cellular processes, including cell growth and immune regulation [35]. Nine human IRF genes have been identified that all share significant homology in their DNA-binding domains, comprised of five tryptophan repeats, three of which contact at DNA sequences termed Interferon-Sensitive Response Elements (ISRE) [36], [37]. The carboxy terminus includes the IRF associated domain (IAD), which displays greater sequence diversity and has been linked to binding partner specificity and autoregulation of DNA binding [35]. Unlike all other IRFs, IRF4 is not regulated by Type I or Type II interferons (IFN) [20], but is instead activated by antigen receptor signaling and restricted classes of Toll-like receptors (TLR) (TLR4 or TLR9) and Tumor Necrosis Factor Receptor (CD40), suggesting a unique role for IRF4 [17], [38]–[40].

. IRF4 DNA binding has been shown to occur at multiple sites, which depend on distinct partner requirements and composite sequence elements (Table 1). For example, biochemical analysis of the immunoglobulin light chain enhancer region showed that IRF4 is in a ternary complex with either PU.1 or SpiB (both of the Ets family) and Ets Irf Composite Elements (EICE) [41]. Biochemical analysis of the ternary complex revealed that the Ets proteins recruit IRF4 to enable high affinity IRF4 binding, whereas binding by IRF4 alone was extremely inefficient [42]. In addition, IRF4 partners in a quaternary complex with BATF/c-jun or BATF/JunB heterodimers (AP-1 family members) and AP-1 IRF Composite Elements (AICE) [16], [43]–[46]. Indeed, biochemical experiments using the AICE probe sequence have revealed a similar partner dependency for high affinity binding, as described for EICE[41]. Further, two variants of AICE sequences display slightly different dependencies on IRF4 concentrations for efficient binding suggesting even greater complexity to the generalizations in this review [43], [47]. Finally, chromatin immunoprecipitation sequencing (ChIP-seq) studies in activated B cells have highlighted IRF4 bound regions, including the Prdm1 (Blimp-1) locus, that contain multiple copies of the ISRE sequence [16], [48]. However, ISRE binding by IRF4 has been shown to be of comparatively lower affinity to that of either EICE or AICE [15], [16], [49]. Thus, a general biochemical scenario emerges whereby IRF4 displays relatively higher affinity DNA binding when it is recruited by Ets or AP-1 members to EICE or AICE composite elements, respectively, but comparatively poorer DNA binding efficiency to ISRE multimers. As we discuss below, these properties of differential DNA affinity likely enable B and CD4⁺ T cells to decode graded IRF4 cell concentrations set by antigen receptor signal strength to execute alternate cell fate decisions.

Table 1.

IRF4 DNA binding sites and partners driving B and CD4⁺ T cell fate decisions.

DNA recognition site	Binding Partner	Relative DNA affinity	Aligned with Gene Program	Sequence Logo
EICE Ets-IRF Composite Element	PU.1 or SpiB	high	B cell GCB
AICE AP1-IRF Composite Element	BATF/JunB or BATF/cJun	high	B cell GCB / T cell Tfh
ZICE Ikaros-IRF Composite Element	Ikaros	high	B cell GCB
ISRE Interferon Stimulated Response Element	IRF4** (homodimer)	low	B cell PC / T cell Teff

^*Currently, a comprehensive understanding of each complex’s relative DNA binding efficiency is not known as only pairwise comparisons have been made. These have shown that generally, ISRE binding efficiency is substantially weaker compared to either EICE or AICE sequences[15], [16].

^**The logo for AICE2 is shown. AICE1 lacks the 4 nucleotide spacer between the AP1 and IRF sites[47]. The biological basis for AICE1 and AICE2 consensus sequences has yet to be determined. Two additional variants of AICE2 have been described (not depicted). The first is composed of a partial AP-1 motif and exhibits a greater dependency on IRF4 amounts for binding [43]. The second, which is preceded by a thymidine residue 4 nucleotides upstream from the IRF site, exhibits greater IRF4 binding than a guanine variant [47]. Together, these observations raise the possibility that variations in DNA sequence binding affinities coordinated with distinct partner interactions can drive IRF4 genome localization dynamics to control cell function and alternate cell fate decisions.

^***Heterodimerization with other IRF family members is formally possible and remains to be tested. PC: plasma cells; GCB: germinal center B cells; Teff: effector T cell; Tfh: T follicular helper cell.

A number of other composite sequences and binding partners have also been identified in B and T cells, including NFAT, E2A and Ikaros [50]–[52]. In relation to NFAT and Stat3, these TFs are often found to co-bind with IRF4 to cis-regulatory elements in genome-wide experiments; however, whether this reflects cooperative assembly on composite elements remains to be determined[53], [54]. Thus, the regulation of IRF4 genome binding is highly complex, suggesting that the dynamics of IRF4 DNA binding and partner interactions may underlie the differential activity of IRF4.

The phenotype of Irf4 deficient mice was first described in 1996, where the formation of the bone marrow and splenic naïve B and T cell compartments appeared normal; however, all tested antigen dependent responses were compromised [38]. Since then, seminal work from many laboratories demonstrates that IRF4 plays a broad role in B and T cell responses, and furthermore, suggests that, unlike lineage-specifying TFs (e.g. Tbet, Rorc), IRF4 plays a novel and fundamental role in the activation and differentiation of multiple subsets of B and T cells [18], [23], [60]–[69], [38], [70], [71], [52], [53], [55]–[59]. Cumulatively, these important findings prompted us to pursue a distinct line of reasoning regarding the role of IRF4 in T cell differentiation. We describe this below in the context of the T-dependent antibody response.

Transcription factor Bcl6

B cell lymphoma 6 (Bcl6) was first identified as a potential proto-oncogene from a chromosome translocation frequently seen in diffuse large B-cell lymphoma [72] and was later shown in cell lines to be involved in apoptosis regulation [73]–[75]. Bcl6 is a zinc-finger TF that also bears a POZ (or BTB or ZIN) domain in the amino terminus [72], [76]; this domain is primarily responsible for transcriptional repression via recruitment of corepressor proteins that include histone deacetylases[76]–[78]. Bcl6 is highly expressed in GCBs; Bcl6^−/−mice are impaired in GCB differentiation, and mice which constitutively express Bcl6 in B cells have enhanced GCB responses relative to wild type (WT) mice [79]–[81]. Furthermore, Bcl6 has also been shown to be both necessary and sufficient for Tfh differentiation in mice [7]–[9]. Therefore, transcriptional repression by Bcl6 is key to adopting GCB and Tfh cell fates, in large part through the suppression of alternative cell fate determinants (including Blimp-1) [14], [82].

Transcription factor Blimp-1

B lymphocyte induced maturation protein-1 (Blimp-1; encoded by Prdm1) is a zinc finger TF that functions as both a repressor and activator of gene expression [83]–[85]. Blimp-1 contains five Krüppel-type zinc finger motifs, a proline-rich domain implicated in gene repression, and a SET domain that functions to control antibody secretion [84], [86], [87]. Blimp-1 was first described in murine B cells as a gene highly induced during B-cell differentiation into PCs, and was later shown to be both necessary and sufficient for PC differentiation [86], [88]. In PCs, Blimp-1 regulates a large gene program important for high titer antibody secretion –operating at both the post-transcriptional and post-translational levels [83]–[85]. Subsequently, Blimp-1 was shown to play an essential role in effector T cell (Teff) differentiation and the acquisition of migratory, cytolytic, and inflammatory gene programs [12], [13], [61]. In addition to regulating a gene program important for PC and Teff cell function, Blimp-1 can also repress the expression of Bcl6 –thus forming a counter-antagonistic transcriptional regulatory loop [7], [82], [84]. Thus, reciprocal repression of Blimp-1 and Bcl6 is believed to be at the core of the mechanism enabling divergent cell fate choices for B and CD4⁺ T cells.

The B and CD4⁺ T cell IRF4 Gene Regulatory Network

IRF4 induces both Blimp-1 and Bcl6 expression

The essential role of IRF4 in the antibody response has been known for over 20 years, when immunized Irf4^−/− mice failed to produce both antigen-specific serum antibody, and GCBs [38]. Subsequent ex vivo experiments using these same Irf4^−/−B cells demonstrated that PC differentiation depended on IRF4 induction of Blimp-1 in a cell autonomous manner [17]. These studies also revealed that IRF4 was essential for the expression of AID, which placed IRF4 in a network upstream of essential regulators of B cell differentiation. IRF4 was later shown to be essential for cell autonomous upregulation of Bcl6 and the generation of GCBs, but timing was important as this was only observed when IRF4 was deleted prior to antigen activation and not post antigen activation (using CD19- or Cγ1- Cre drivers, respectively) [16], [18], [39]. The timing of deletion exhibited by the different Cre drivers suggested that IRF4-dependent GCB cell fate decisions occured in the first few days following antigen encounter in mice.

Simultaneous with B cell studies, Tfh cells required the expression of Bcl6 and were antagonized by Blimp-1 in gain- and loss-of-function experiments [7]. However, the factor(s) responsible for activating Bcl6 and/or Blimp-1 cell fate determinants in CD4⁺ T cells were unknown. Later, it was demonstrated that Irf4^−/− mice failed to generate Tfh cells [53]. A cell autonomous role of IRF4 in generating Tfh cells was confirmed using bone marrow chimeras; and this study also established that Irf4^−/− CD4⁺ T cells failed to generate Blimp-1-expressing Teff cells that co-expressed Tbx21 (T-bet) and secreted proinflammatory IFNγ[ 15]. Together, these results demonstrated that IRF4 was central to both diversification and effector cell fates in both B and CD4⁺ T cells, positioned upstream of the key cell fate determinants Blimp-1 and Bcl6 (Figures 3 and 4).

Figure 3. Antigen dependent alternate B cell fate decisions.

Antigen recognition induces the expression of IRF4 in naïve B cells. Intensity of BCR and CD40 signaling scale with induced IRF4 concentrations. Higher IRF4 amounts favors the induction of Blimp-1 expression and preferential PC cell fates. Alternatively, lower IRF4 amounts favor the induction of Bcl6 expression and preferential germinal center B cells (GCB) cell fates. Varying IRF4 concentrations affect its partitioning into distinct assemblies of DNA binding partners and DNA recognition elements that are aligned with components of distinct gene programs to effect cell fate choice, including that of Blimp-1 and Bcl6[16]. PC: plasma cells.

Figure 4. Antigen dependent alternate CD4⁺ T cell fate decisions.

Antigen recognition induces the expression of IRF4. Intensity of TCR signaling and costimulation scale with induced IRF4 concentrations. Higher IRF4 amounts favors the induction of Blimp-1 expression and preferential effector T cell (Teff) cell fates. Alternatively, lower IRF4 amounts favor the induction of Bcl6 expression and preferential follicular helper T (Tfh) cell fates. Varying IRF4 concentrations can affect its partitioning into distinct assemblies of DNA binding partners and DNA recognition elements that are aligned with components of distinct gene programs to effect cell fate choice, including that of Blimp-1 and Bcl6[15].

IRF4 Expression dictates Blimp-1 or Bcl6 gene induction

The data described above raised the question, “how can IRF4 simultaneously activate the expression of two counter-antagonistic transcriptional regulators”? Insight arose with the observation that PCs were restricted to high IRF4 expression, whereas GCBs, to low IRF4 expression [17], [89]. This raised the possibility that IRF4 cell concentrations could coordinate the cell fate decision between Blimp-1-expressing PC and Bcl6-expressing GCB. Genetic perturbation experiments were consistent with this idea [17], [40], [90], but strong evidence came from the analysis of a mouse germline tetracycline-inducible allele ofIrf4, which enabled orthogonal control of IRF4 expression in vitro and in vivo in antigen specific B cells (uncoupled from BCR/TCR) [16]. This system showed that low, transient amounts of IRF4 could rescue GCB differentiation but were insufficient for PC differentiation. Conversely, when IRF4 expression was enforced above WT levels, PC differentiation was enhanced at the expense of the GCB response. To formally test whether IRF4 abundance directed CD4⁺ T cell fate determination, the same mouse germline encoded Irf4-inducible system in antigen specific TCR transgenic cells was used in vivo [15]. Enforced induction of IRF4 above WT, resulted in a substantial redirection of Tfh to Teff cell fates. This tetracycline-inducible IRF4 system demonstrated that despite the multiple pathways induced by antigen receptor signaling, the pathway controlling IRF4 induction appeared to be key for B and CD4⁺ T cell fate choices [15].

IRF4 expression scales with antigen receptor signaling intensity

Due to Irf4 being an immediate early gene downstream of the BCR, it was reasoned that BCR signal strength might regulate IRF4 expression amounts. Indeed, stimulating antigen-specific B cells in vitro withincreased antigen abundance or avidity raised IRF4 cell concentrations in a manner that scaled with signal intensity [40]. Moreover, using the SWHEL transgenic BCR in vivo system, in which low affinity HEL antigens predominantly generate a GCB response and high affinity HEL antigens promotes a PC response[26], IRF4 expression in activated SWHEL B cells was reported to correlate with BCR signal strength [16]. These systems directly linked antigen affinity, and thus, BCR signaling intensity, to PC and GCB cell fate choice via modulation of IRF4 expression. Furthermore, CD40 signaling also induced IRF4 expression, suggesting that B cell engagement with T cells through CD40:CD40L binding could control IRF4 abundance [17], [39], [91]. Because of the temporal delay of CD40-induced versus BCR-induced IRF4 expression, it remains to be determined whether the overall magnitude and/or the timing of IRF4 induction functions as an additional layer of regulation in alternative cell fate choices and thus, in the quantity and quality of the antibody response.

As seen in B cells, the relationship between the antigen receptor and IRF4 expression also occurs in CD8⁺ T cells [23], [58], [59]. To examine this possibility in CD4⁺ T cells, an approach was adopted in which a series of altered peptide ligands (APLs) were used that displayed a range of potency for the 5c.c7 TCR peptide:MHC system in mice[25]; the study showed that Irf4 transcript and protein abundance scaled with increased TCR signal strength both in vitro and in vivo [15]. These results were consistent with in vitro anti-CD3 experiments from another study, which showed that IRF4 protein was induced at higher concentrations with greater TCR stimulation [47]. Together, these observations establish an important link between antigen receptor signal strength and IRF4 abundance.

The in vivo relationship between the TCR and IRF4 cell concentrations led to testing whether TCR signal strength influenced Teff and Tfh cell fate decisions. Indeed, upon immunization with low potency APL, responding mouse 5c.c7 TCR transgenic cells of the lymph node exhibited higher frequencies of Tfh cells, coincident with diminished Teff responses relative to high potency APL [15]. This was true at both early and late time points, corresponding to specification and commitment of cell fates, respectively [92]. Conversely, immunization with higher potency APL reversed these trends and favored Teff cell fates at the expense of Tfh [15]. These results confirmed that increased TCR signaling could induce Teff cell fate determination at the expense of Tfh.

The role of TCR signal strength in CD4⁺ T cell fate choice is well established; however, in relation to Tfh cell fate, whether strong or weak signaling is important remains controversial. Although earlier reports suggested that increased TCR signal strength favored Tfh over Teff cell allocation [93], [94], recent reports in different TCR systems suggested that increased TCR signal strength in CD4⁺ T cells favored Teff over Tfh cell fates [32]–[34], supporting the data described above. The basis for the divergent conclusions is presently unclear because of the different TCR systems used, raising the possibility that the range of signal strength along the curve to which a cell is confined to may lead to different interpretations. For example, when interpreting cell fate outcomes stimulated from two weak APLs; the stronger of the weak APLs may lead to comparatively more Tfh because neither APL elicits sufficient TCR signal strength to cross the crucial threshold important for Teff cell choice. We expect future research to identify the biophysical and/or dynamical parameter of TCR signal strength that corresponds to quantitative changes in IRF4 abundance and CD4⁺ T cell fate outcomes. Lastly, it is clear that IRF4 expression scales with increasing TCR signal strength, and that orthogonally increasing IRF4 is sufficient to derail a given regime of TCR signal strength and its allocation to Tfh and Teff cell fates [15] suggesting that quantifying IRF4 expression in these different systems may resolve the debate.

IRF4 abundance can dictate DNA binding site choice

Previously, the manner in which B cells were thought to detect IRF4 concentrations remained ambiguous, until IRF4 genome localization was measured using Chromatin Immunoprecipitation-deep sequencing (ChIP-seq) [16]. The majority of IRF4 binding events in activated B cells occurred at regions harboring EICE sequences and were precipitated with PU.1. The remaining third of binding events were enriched for either AICE sequences prior to differentiation, or enriched for tandem ISRE sequences upon PC differentiation, demonstrating the complexity in IRF4 binding site choice [16]. In contrast to the EICE, which contains a single ISRE sequence, the tandem ISRE sequences identified in the IRF4-bound regions were found in multiple copies, which raised the possibility that IRF4 could efficiently bind to those sites, perhaps as a dimer or multimer. In fact, biochemical experiments demonstrated specific binding of IRF4 to a tandem ISRE site found in the Prdm1 (Blimp-1) locus and binding was shown to be considerably weaker when compared to IRF4 binding to an EICE probe in the presence of PU.1 [16]. Thus, at the genome level, IRF4 localized to multiple regions that harbored distinct binding sites, which seemed to differ in their ability to recruit IRF4[16]. To investigate the nature of IRF4 binding with gene regulation and cell fate determination, ChIP-seq datasets were integrated with genome wide expression analysis. IRF4 binding site choice correlated significantly with the binding of genes expressed in cells undergoing different cell fates. Generally, IRF4-bound EICE --but not ISRE sequences--aligned with genes of the GCB program (including Bcl6) and complementary to this, IRF4-bound ISRE-- but not EICE sequences--, localized to genes affiliated with the PC program (including Prdm1)[16]. Given that ISRE sequences displayed lower IRF4 binding affinity than EICE sequences in the presence of PU.1, this finding suggested that higher amounts of IRF4 were required in PCs to enable efficient occupancy of ISRE-containing regulatory regions of PC genes [16]. Of note, not all binding events in PCs are restricted to low affinity binding sites, as observed for a high affinity IRF4-Ikaros DNA binding complex (ZICE [Zinc finger-IRF Composite Elements]) at select B cell lineage genes, including Ebf1, known to antagonize PC differentiation [51]. Together, a model emerges whereby B cell fate choice, and therefore the character of the antibody response, are shaped by a given cell’s IRF4 abundance, in turn dictating IRF4 binding partner interactions, and the ability of this TF to occupy low affinity DNA binding sites.

To better understand how IRF4 abundance molecularly controls CD4⁺ T cell fate choices, the relationships between accessible chromatin (putative regulatory regions) have been analyzed in CD4⁺ T cells with differential gene expression by ATAC-seq, along with the presence of low or high affinity IRF4 DNA binding sites [15]. Cells depleted of IRF4 did not express genes of either the Tfh or Teff program, cells with WT amounts of IRF4 expressed genes of either the Tfh or Teff programs and these cells had greater accessibility to Tfh program genes, including Bcl6. Moreover, cells with abundant IRF4 predominantly expressed, and had greater accessibility to genes of the Teff program, including Prdm1, while exhibiting a collapse of the Tfh program. Low affinity ISRE sequences were significantly enriched in regions belonging to the Teff program and depleted in regions aligned with the Tfh program. Conversely, high affinity AICE sites were enriched in regulatory regions of both the Tfh and Teff programs [15]. Hence, high affinity AICE sequences might also regulate Teff genes, which suggests that as IRF4 accumulates, transient binding to AICE sequences might “prime” Teff-associated loci for later expression during Teff cell commitment. Thus, as in B cells, the dynamics of IRF4 accumulation enable differential occupancy of high and low affinity DNA binding sites of genes belonging to the Tfh and Teff programs to control cell fate allocation. Nevertheless, further analyses are warranted to fully validate these models, and more could be done by using mathematical interpretations (Box 3).

Box 3. A Conserved IRF4 Gene Regulatory Network in the immune system?

The striking conservation of the IRF4 GRN in both B and CD4⁺ T cells suggests that the GRN may have preceded the evolutionary bifurcation of T and B cells.In line with this, the family of innate immune receptors “Agnathan paired receptors resembling antigen receptors” (APAR) may have served as the ancestral vertebrate antigen receptor that also possessed signaling capacity [98]–[100]. Given the signaling propensity of the APAR, we speculate that the IRF4 GRN may have been fixed at that point of evolutionary time. In fact, IRF4, Bcl6, and Blimp-1 homologs have been identified in the ancient genomes of sea urchins and agnathans [101]–[104]. Given the multiple interfaces for interaction (e.g. protein-protein, protein-DNA elements) orchestrated to execute function, it is plausible that the IRF4 GRN arose from an ancestral apomere (cell type-specific cellular module) [105] already present in agnathans and has been evolutionarily constrained. As such, the IRF4 GRN could have been inherited during the emergence of adaptive immunity and the diversification of the variable component of the ancestral receptor in gnathostomes [106]–[108]. Although it is unclear what linked the IRF4 activator to Blimp-1 and Bcl6 fate determinants, a common theme in development is the pairing of counter-antagonistic regulators to enable alternate cell fates [109]. In contrast to all other members of the IRF family, IRF4 is not induced by either Type I or II IFN signaling [20]; instead, its regulation is largely governed by NFkB signaling [21], [71], [91], [110]. However, the nature of the selective forces enabling this shift during the course of apomere inheritance from agnathans is speculative. Despite this divergence, IRF4 has maintained the wide-ranging roles of IRFs in general cell biology, as evidenced by that of IRF4 in lymphocyte activation, metabolism, and differentiation. Notably, IRF4 loss-of-function variants are extremely rare in most human genome-wide association studies, leading to the hypothesis that IRF4 is subject to purifying selection, consistent with its essential role in immunity [113]. This finding makes it unlikely that the IRF4 GRN is only deployed for the initiation of naïve B and CD4⁺ T cell responses. GCBs undergoing cell selection during affinity maturation “test” their SHM-mutated antibody, resulting in BCR and CD40 signaling, which resembles the initial steps of Tdependent B cell responses [29], [111], [112]. The IRF4 GRN also plays a role in CD8⁺ T cytolytic responses, where IRF4 expression scales with TCR signal strength and IRF4 gene dosage affects the proportions and downstream cell states of CD8⁺ T cells [114] [58]. Further, IRF4 is expressed in monocytes and dendritic cells where it also plays an essential role in cell fate decisions [115]–[118]; however, we note that it is unclear whether those decisions are dependent on Blimp-1 or Bcl6 [116], [119], [120].

Concluding Remarks

There is genetic evidence for the IRF4 gene regulatory network (GRN) in specifying antigen dependent B and CD4⁺ T cell fate trajectories that orchestrate the antibody response. However, it is important to note that the IRF4 GRN is a core network and thus, it is expected that signal-induced pathways activate other GRNs that interface directly or indirectly with it to modulate the overall dynamics depending on the output of those GRNs [121]-[122]. In fact, recent computational modeling and wet lab experimental validation with B cells suggest that a reciprocally antagonistic NFκB GRN composed of RelA and c-Rel can interface with the IRF4 GRN on the Blimp-1 node in an antagonistic manner, dampening PC differentiation [123]. In contrast to these deterministic GRN models, an expanding literature of modeling and single cell visualization experiments suggest that B and CD4⁺ T cell fate trajectories are subject to stochasticity [124], [125]. In light of the inherent “noise” of transcription and translation rates that have been documented using high throughput methodologies [126], it is not surprising that randomness is observed. Thus, a fascinating scenario emerges where deterministic and stochastic events combine to regulate clonal selection and fate allocation to enable the micro-evolution of the antigen-induced repertoire of diversified B and CD4⁺ T cells [127].

Given that the IRF4 GRN is rooted in TF biology, it is subject to regulation by the nuclear chromatin environment that includes nuclear position and looping as well as epigenetic modifications that act at specific loci. Of note, PCs undergo dynamic de- and pro-DNA methylation events that may enable PC differentiation [128]. How this is regulated and impacts the dynamics of the IRF4 GRN will be exciting to unravel. Another interesting finding is that IRF4 activity has been linked to the regulation of genes important for metabolism [58], [117]. Metabolic changes are key to cell activation and division, but different cell states and their transitions exhibit distinct metabolic requirements –including GCB, PC, Tfh, and Teff [129]-[133]. The role of IRF4 in metabolic processes and how metabolites may feedback on the IRF4 GRN remains for future studies but is certainly timely. Overall, the core IRF4 GRN both influences and interfaces with a multitude of regulatory processes comprised of gene and signaling regulatory modules to control the outcome of B and CD4⁺ T cell fate allocation.

In sum, the essential components of a dynamic IRF4 GRN that regulate alternate B and CD4⁺ T cell fates have been identified, but further work is required to map the ancillary factors, GRNs, and metabolic effectors that integrate the quantity and quality of antibodies in the immune response. Given the large numbers of nodes involved, it is clear that modeling will play an important role in clarifying the nature of these interactions and how they affect the dynamics of the overall process (see Outstanding Questions). The existing and future endeavors in this area promise to advance the fundamental understanding of cell fate determination and provide novel inroads for potential therapeutic targeting

Figure 5. Antigen-Dependent Alternate CD4⁺ T-Cell Fate Decisions.

Antigen recognition induces the expression of IRF4. Intensity of TCR signaling and costimulation scale with induced IRF4 concentrations. Higher IRF4 amounts favor the induction of Blimp-1 expression and preferential effector T-cell (Teff) fates. Alternatively, lower IRF4 amounts favor the induction of Bcl6 expression and preferential follicular helper T (Tfh) cell fates. Varying IRF4 concentrations can affect its partitioning into distinct assemblies of DNA-binding partners and DNA recognition elements that are aligned with components of distinct gene programs to effect cell fate choice, including that of Blimp-1 and Bcl6 (15]. Abbreviations: Bcl6, B-cell lymphoma 6; Blimp-1, B lymphocyte-induced maturation protein-1; IRF, interferon regulatory family.

Outstanding Questions.

• Does the IRF4 GRN play a role in GCBs during affinity maturation? Since BCR and CD40 signals play a role in selection of affinity matured variants and BCR and CD40 signals regulate IRF4 amounts, this raises the possibility that the IRF4-dependent GRN is used in affinity maturation.

• Where is the bifurcation of memory B and CD4⁺ T cells in the progenitor (naïve) to progeny (effector) map of the antibody response? Given that cell fate trajectory is responsive to BCR signaling and CD4⁺ T cell help, this raises the possibility that IRF4 abundance plays a role in influencing the antibody response.

• How is the IRF4 GRN layered in the epigenetic and nuclear landscape of transitioning cells? As cells transition from progenitors to differentiated cell states, they undergo extensive epigenetic changes that may intersect with the IRF4 GRN.

• Does IRF4 control the metabolic changes associated with B and CD4⁺ Tcell fate transitions and do the metabolic products feedback to shape cell fate? Metabolic products have emerged as critical mediators of cell state dynamics, including epigenetics, indicating that metabolic re-programming may influence the outcome of the IRF4-dependent GRN.

• Do post-translational mechanisms play a role in the dynamics of the IRF4 GRN? Post-translational regulation is a relatively understudied, yet important, feature of cell state.

• As IRF4 appears to be a central node controlling cell state and differentiation in these immune cell subsets, could future therapies targeting IRF4 improve specific disease outcomes? IRF4 has already been implicated in modulating allograft and cancer survival for certain cases, suggesting that manipulation of the IRF4 GRN might be beneficial for specific diseases, pending robust investigation.

Highlights.

• The antibody response is orchestrated by the development of distinct B and CD4⁺ T cell fates with specialized function.

• Distinct B and CD4⁺ T cell fates are controlled by an antigen receptor signal strength-responsive, incoherent gene regulatory network, determined in mice

• The transcription factor IRF4 is a central and essential node of this network, whose expression scales with the intensity of antigen receptor signal strength

• Graded expression and differential DNA affinity of IRF4 can function to control the outcome of a counter-antagonistic gene regulatory interactions governing alternate cell fates

Glossary

5c.c7 system

A well characterized T cell receptor that recognizes a peptide derived from moth cytochrome c in the context of MHC II I-E^k. Variants of this peptide, called Altered Peptide Ligands (APL), have been identified that display weak and strong agonistic as well as antagonistic activities. These variants reflect TCR contact residues and not MHC contact residues that would attenuate general peptide:MHC density. A genetically modified mouse model based on this TCR enables antigen specific T cells to be tracked in response to the APLs in vivo

B cell receptor (BCR) / T cell receptor (TCR)

The antigen receptor expressed by B and T cells formed by assembling Variable and Joining segments by DNA rearrangements during their development. The receptor recognizes antigen

Germinal Center B cell (GCB)

A B cell state within a germinal center in a lymphoid follicle; it is acquired upon antigen encounter and T cell-derived signals with a genetic program for the diversification of the antigen specific repertoire by somatic mutation of its immunoglobulin variable genes. Selected variants differentiate into plasma and memory B cells

Plasma cell (PC)

A B cell state acquired upon antigen encounter and T cell-derived signals with a genetic program for antibody secretion thus maintaining serum immunoglobulin titers

SWHEL system

A genetically modified mouse model that enables antigen specific B cells to be tracked in response to an allelic series of antigens that vary in affinity to the antigen receptor

T follicular helper cell (Tfh)

A T cell state acquired upon antigen encounter and B cell derived signals with a genetic program for the coordination of B cell activation and differentiation

T effector cell (Teff)

A T cell state acquired upon antigen encounter with a genetic program for the coordination of inflammatory responses at peripheral tissue sites

T-dependent antibody responses

Antigen-specific antibody derived by obligate interactions between antigen specific B and CD4⁺ T cells. The two-way interactions lead to B and T cell differentiation into the GCB, PC, and Tfh cell states described in this review

Toll Like Receptor (TLR)

A system of innate immune receptors that signal inflammatory responses upon recognition of common microbial patterns. Signaling is often through the NFκB system