IKZF1 (Ikaros Zinc Finger Protein 1), encoding for Ikaros, is a zinc finger transcription factor that is essential for immune cell development, homeostasis, and function (1). It regulates transcriptional programs through the coordination of six highly conserved C2H2 zinc finger (ZF) domains; the first four at the N-terminus are essential for regulating gene transcription through DNA binding, and the last two at the C-terminus facilitate multimer formation as both a homodimer and as a heterodimer with other family members: IKZF2 (Helios), IKZF3 (Aiolos), IKZF4 (Eos), and IKZF5 (Pegasus) (2). IKZF1 is able to regulate gene expression through direct binding and activating of promoter regions (2–4), but predominantly negatively regulates gene transcription through repression of chromatin and recruitment of co-repressors (4–11). The expression of IKZF1 is important in guiding hematopoietic cell fate decisions in the early development of uncommitted lymphocytes as well as controlling transcriptional programs essential for later B and T cell development. IKZF1 expression can be detected from the earliest stage of hematopoietic development in the fetal liver and adult bone marrow and its expression is continued through until mature B and T cells (12–15).

Lymphocyte development first begins with pluripotent and self-renewing hematopoietic stem cells (HSCs) upregulating Flt3 to become multipotent progenitors (MPPs) (16). This upregulation of Flt3 removes the ability of MPPs to self-renew and restricts them to either a lymphocyte or myeloid fate (17). In mice, the subsequent upregulation of interleukin-7 receptor (IL7Rα or CD127) will commit a cell to the lymphoid lineage, defining them as a common lymphoid progenitor (CLP) (18), thereby facilitating a dependency on IL7 that is secreted by stromal cells in the bone marrow (19–21). This however, is true only in mice as human B cells do not require IL7 for normal development (22). From here, CLPs can then differentiate into either B, T, dendritic cells, or natural killer cells (18, 23, 24). IKZF1 is an important regulator that guides lymphoid commitment. In the complete absence of IKZF1 expression, a failure to establish the lymphoid program in the fetal liver results in the absence of B and T cell development (25, 26) that can be attributed to the loss of Il7r and Flt3 upregulation (27). The expression of IKZF1 however, is not limited to early hematopoietic progenitors as IKZF1 also has important roles both early in B cell development (28) and in the regulation of mature B cells (29).

Multiple isoforms of IKZF1 have been described in both humans and mice, all products of alternative splicing. These isoforms differ in the number and combination of zinc finger domains and can be broadly split into two groups: DNA binding and non-DNA binding, based on whether the N-terminal zinc finger domains are present or not (2, 3, 30, 31). As the information available describing both murine and human IKZF1 isoforms are not free of discrepancies and scientific controversies, we will focus on the most generally accepted data to avoid overcomplicating this review. Dimers containing IKZF1 isoforms that lack a DNA-binding domain are transcriptionally inactive and are not able to activate gene transcription in-vitro (2). This generally suggests a dominant negative effect of the shorter isoforms that are non-DNA binding, yet later studies demonstrated that this is only observed when there is a drastic reduction in the amount of longer DNA binding isoforms available (7). Over six isoforms have been discovered in mice and humans with slight differences between murine and human isoforms (Figure 1). In mice, all isoforms maintain the C-terminal dimerization domain consisting of ZF5 and ZF6. Surprisingly, only two of the six murine isoforms contain a DNA binding domain with at least three N-terminal zinc fingers, compared with approximately half of the human isoforms. This would suggest that in mice, only the canonical isoform VI and potentially isoform V are functionally relevant as transcriptional activators as they are the only isoforms able to bind DNA with high affinity. A larger variety of isoforms have been discovered in humans, the majority of which are DNA binding isoforms that maintain at least three N-terminal zinc finger domains.

IKZF1 binds to pericentromeric repeats through direct DNA binding at pericentromeric-heterochromatin (PC-HC) (5) and is able to recruit repressed genes to PC-HC by binding directly to their promoter regions (7). It was originally hypothesized that PC-HC localization and promoter binding are exclusive events, as IKZF1 binding to pericentromeric repeats would obscure the DNA binding domain and prevent binding to target genes (5). Yet the ability of IKZF1 to both directly bind target genes and simultaneously bind to pericentromeric repeats was later shown to be facilitated through the formation of IKZF1 multimers consisting of individual IKZF1 monomers bound to either target genes or PC-HC (7). In endogenous systems, IKZF1 can be alternatively spliced to produce multiple isoforms containing different combinations of each zinc finger (Figure 1) that are expressed during different stages of lymphoid development (3, 8, 30). These isoforms can interact with each other in different configurations, either as a homo or heterodimers, the outcome of which produces different functional activities (2, 31–34). Whilst the longest isoform mainly associates with PC-HC, some evidence suggests that other isoforms can direct transcriptional regulation outside of PC-HC (35). Furthermore, it was originally believed that shorter isoforms that do not contain the DNA binding domain disrupted the DNA binding ability of longer isoforms that do contain the N-terminal DNA binding domain (2, 5). Yet experiments by Trihn et al. (7) demonstrated that this in fact is not the case. Co-expressing murine isoforms VI and I in NIH3T3 cells and observing their nuclear location using immunofluorescence, showed localization of both isoforms at PC-HC. Yet over-expression of the shorter isoform I alone showed diffuse staining indicating an inability to localize to centromeric foci consistent with its lack of DNA binding domain (7), thus demonstrating that isoforms that lack the DNA binding domain are still able to associate with PC-HC when co-recruited by an isoform that contains a DNA binding domain. Furthermore, the expression of the smaller isoform I had no effect on DNA binding by endogenous IKZF1 isoforms (7). The authors therefore concluded that smaller isoforms that are unable to bind DNA are able to associate with larger isoforms in multimeric structures at PC-HC and that dominant negative effects of shorter isoforms are only apparent when the expression of the longer isoforms is lost.

Our understanding of IKZF1 biology has been greatly enhanced through the analysis of many Ikzf1 mouse models [summarized in Heizmann et al. (36)]. The first reported Ikzf1 mutant mice have a dominant negative mutation that removes ZF1, ZF2, and ZF3 in the DNA binding domain, rendering the protein unable to bind DNA (2, 25). Mice homozygous for this mutation (Ikzf1^DN/DN) have a complete absence of all lymphoid progenitors in the fetal liver and postnatal bone marrow and have no mature B, T, or NK cells, thus defining IKZF1's role in the establishment of the lymphoid fate (25). Additionally, Ikzf1^DN/DN mice have only a very rudimentary thymus and no lymph nodes, and die at a very young age due to increased opportunistic infections and septicemia (25). Surprisingly, a few potential CD4 progenitors could be found in the rudimentary thymus in postnatal mice, potentially suggesting differential requirements of IKZF1 in T cell establishment compared with B cell establishment (25). Mice heterozygous for this mutation (Ikzf1^+/DN) show no obvious B cell defects but spontaneously develop thymic derived leukemia as a result of the loss of the wildtype allele in clonally expanded populations (37). Ikzf1^DN/DN mice highlighted the crucial role IKZF1 plays in hematopoietic stem cell commitment to the lymphoid lineage, but Ikzf1^+/DN mice demonstrated the effects that mutant Ikzf1 alleles can have on the functioning of wildtype Ikaros family members. By definition, the dominant negative nature of the Ikzf1^DN mutation meant that in a heterozygous state, the mutant IKZF1^DN protein could interfere with DNA binding of the wildtype IKZF1 protein, thus reducing the functionality of the wildtype protein (2, 37). This opened the door for exploring the effects that this type of mutation would have on the functioning of other Ikaros family members. It was plausible to assume that in a homozygous state, the IKZF1^DN protein could similarly be interfering with IKZF3 DNA binding (13), thus obscuring the sole effect of IKZF1 during lymphoid development.

In attempt to address this, the same group created an Ikzf1-null mouse (Ikzf1^−/−) (26), where deleting exon 7 at the C-terminal dimerization domain resulted in the production of unstable null proteins that are not functional and are transcriptionally inactive (2). Wildtype IKZF1 isoforms VI and V (described as Ik-1 and Ik-2) were unable to be detected by western blot in thymocytes from Ikzf1^−/− mice (26). No lymphocyte progenitors could be found in the fetal liver (26) due to a reduction in HSC numbers and activity (38) leading to a subsequent absence of B, T, and NK cells. B cell precursors could not be found in the bone marrow of adult mice and subsequently, no mature B cells ever developed in Ikzf1^−/− mice (26). Conversely, low frequencies of T cell precursors were identified in the thymus of postnatal Ikzf1^−/− mice, despite there being no evidence of lymphocyte commitment in the fetal liver. Aberrant TCR signaling in the thymus (39) resulted in a skew toward CD4 single positive thymocytes at the expense of double positive thymocytes. As Ikzf1^−/− mice aged, T cells could be found at almost normal frequencies in the spleen, suggesting seeding from the thymus in adult mice was not affected (26). In support of this, γδ T cells that are normally found in the skin and epithelium that are seeded from the thymus during fetal development (40–45), were not present in Ikzf1^−/− mice, yet γδ T cells in the spleen and lymph nodes that are seeded from the thymus postnatally could be identified (26).

Unlike Ikzf1^DN/DN mice, Ikzf1^−/− mice are born at the expected frequency and survive into adulthood (26). Whilst Ikzf1^−/− mice were intended to overcome the issue of dominant negative effects of the Ikzf1^DN/DN mutation, the result was unfortunately not as clear-cut as first expected. It later became evident that removing IKZF1 expression entirely results in compensation from other Ikaros family members (31, 46, 47). Through both homodimerization and heterodimerization, IKZF1 creates macromolecular complexes that serve functionally different roles in the hematopoietic system (13, 14, 31, 48, 49). The complete removal of IKZF1 expression or function such as in Ikzf1^−/− and Ikzf1^DN/DN mice creates a hole in the repertoire that is potentially able to be filled by other Ikaros family members, thereby obscuring the direct role of IKZF1 in the hematopoietic system. This was demonstrated through the discovery of an ENU-induced point mutation in ZF3, H191R, that specifically disrupts DNA binding and localization to PC-HC but maintains protein scaffold structure and expression (46). When in homozygosity, the H191R mutation (Ikzf1^Plstc/Plstc) creates a much more severe phenotype than the Ikzf1^−/− and Ikzf1^DN/DN mutations, with embryonic lethality at day E15.5 and a complete loss of B and T cell development in the fetal liver. Reconstitution experiments using either 100% Ikzf1^Plstc/Plstc, or 50:50 Ikzf1^+/+ and Ikzf1^Plstc/Plstc fetal liver cells failed to produce any macrophages, myeloid, or lymphoid cell subsets that were derived from the mutant cells. Heterozygous mice (Ikzf1^+/Plstc) survive into adulthood but develop T cell lymphoma. A partial blockage of B cell development at the pro-B cell stage in the adult bone marrow also causes a reduction in mature B cell numbers in Ikzf1^+/Plstc mice. As all three mutations (Ikzf1^−/−, Ikzf1^DN/DN, and Ikzf1^Plstc/Plstc) affected DNA binding and PC-HC localization, the difference in severity between the three mice models could not simply be accounted for by the absence of IKZF1 binding DNA. By preserving protein expression in Ikzf1^Plstc/Plstc mice, the niche that IKZF1 fills does not have to be maintained by other Ikaros family members and therefore the effects of the sole loss of IKZF1 functionality could account for the severity of the Ikzf1^Plstc/Plstc phenotype. Paradoxically, it was noted in Schjerven et al. (50) that when Ikzf1^−/− mice were attempted to be bred onto a pure C57BL/6 background, no homozygotes were born, likely due to embryonic lethality. This would suggest that the reported difference in severity between Ikzf1^−/− mice originally bred on a mixed 129SV background, and Ikzf1^Plstc/Plstc mice bred on a C57BL/6 background, could simply be due to genetic differences among background strains. Despite this, several other authors have reported differences in phenotypes between mice with the DN or null mutation when bred on the same mixed 129SV × C57BL/6 background (39, 51) suggesting differences in the underlying mechanism between the two mutations.

It has become increasingly clear that IKZF1 is absolutely required for both T and B cell development, but exactly how IKZF1 discerns between these two transcriptional programs is still under debate. The absence of lymphopoiesis in the fetal liver of Ikfz1^DN/DN, Ikzf1^−/−, and Ikzf1^Plstc/Plstc mice, but presence of some lymphoid development in the thymus of surviving adult mice suggests differential requirements of IKZF1 in T and B cell commitment between the fetal liver and the bone marrow. Additionally, the thymus of Ikzf1^Plstc/Plstc embryos contains a decent proportion of B220⁺CD19⁻ precursor B cells in lieu of any Thy-1⁺ precursor T cells (46). This shift in T and B frequencies in the thymus affirmed previous suggestions that IKZF1 has a role in the regulation of T and B cell fate decisions.

The role of IKZF1 in B cell development was originally investigated by inserting a β-galactosidase reporter in-frame into the second exon of Ikzf1, which is present in all IKZF1 isoforms, just upstream of the N-terminal zinc finger domains (28). Mice homozygous for this mutation (Ikzf1^L/L) lowly express a truncated form of IKZF1 and no full-length protein. Interestingly, this causes a complete block of B cell development in the fetal liver, and a partial blockage in the adult bone marrow between the pro- and pre-B cell stages. This was later shown to be because IKZF1 ensures B cell lineage commitment by positively regulating V_H rearrangements by directly binding to and activating Rag1 and Rag2 expression, as well as facilitating accessibility and compaction of the Igh locus (52). The few B cells that were able to successfully rearrange their BCR in Ikzf1^L/L mice surprisingly had a lower threshold for BCR activation in-vitro, and fewer germinal centers in response to bovine serum albumin (BSA) immunization, as well as reduced basal IgG3 secretion (28). These mice demonstrated that the low expression of a smaller IKZF1 protein can still establish B cell commitment in the postnatal bone marrow, compared with a complete loss of IKZF1 expression and/or function where B cell development is completely abolished.

To further this work, selectively removing either zinc finger 1 or zinc finger 4 demonstrated differential roles that each zinc finger plays in directing B and T cell development (50). Removing zinc finger 1 (Ikzf1^ZF1/ZF1) induces a B cell specific defect with a partial block of development at the pro-B cell stage in the adult bone marrow, whilst deleting zinc finger 4 (Ikzf1^ZF4/ZF4) prevents T cell development in the thymus and attenuation of development in large pre-B cells in the bone marrow. This would suggest that ZF4 is largely responsible for T cell development, but also has roles in B cell development, whilst ZF1 is exclusively responsible for B cell development. This work could also suggest that ZF1 and ZF4 are responsible for site-directed DNA binding but thus far, no DNA motifs that are specifically recognized by either ZF1 or ZF4 have been identified. Furthermore, the deletion of the entire zinc finger in these models has also removed the regulatory regions in the upstream flanking sequences (50), potentially impacting the interpretations of the independent roles each zinc finger has in regulating lineage-specific genes.

Recently, our group has described a mouse with a novel mutation in ZF1, L132P, that similarly to the Plastic mice have altered DNA binding capabilities but maintain full-length protein expression (53). Mice homozygous for this mutation, Ikzf1^L132P/L132P, have reduced B cell development and an impaired T dependent humoral immune response, yet show no defects in T cell development. As the mutation lies within the first zinc finger, this provides further evidence that each zinc finger plays a different role in regulating lineage specific events.

With early lymphoid development in both the fetal liver and the adult bone marrow being dependent on IKZF1 expression, it has been impossible to determine the role of IKZF1 in mature B cell development and function. To achieve this, Schwickert et al. (29) created a conditional knock-out where Ikzf1 is deleted in mature B cells only using Cd23-Cre. Ikzf1^B− mice have normal early B cell development, but develop splenomegaly and autoimmunity with elegant experiments demonstrating how IKZF1 maintains anergy induction in follicular B cells by directly regulating the anergic signature and also by restraining MyD88-dependent TLR signaling (29). Additionally, by conditionally knocking-out Ikzf1 in the germinal center using Aicda-Cre, they demonstrated the crucial and intrinsic role of IKZF1 in the humoral immune response as Ikzf1^GCB−, mice failed to elicit a strong germinal center and long-lived plasma cell response after T-dependent immunization (29).

Whilst collectively these mice demonstrate how even small amino acid substitutions to larger deletions of entire regions can cause widespread failure of the immune system, the differences between each mouse model highlight how complex IKZF1's involvement is from the earliest common lymphoid progenitor through until mature B cells. The mechanism of how IKZF1 aids different functions across the immune system by interacting not only with itself in different isoforms, but also with its family members, adds another layer of complexity that is yet to be fully explored. It is currently unknown how IKZF1 regulates such vastly different transcriptomes and how this shapes infection and immunity.

IKZF1 acts in a complex and dynamic network in-vivo, yet current attempts to understand the molecular mechanism of IKZF1 mutations are occurring in isolated in-vitro systems. The exclusive analysis of overexpression systems, whereby, specific point mutations that have been discovered in human patients are artificially expressed in-vitro, does not necessarily or completely pinpoint the underlying molecular defect causing the full spectrum of the clinical phenotype of the patient. To understand these biological networks in the context of mammalian disease, in-vivo studies demonstrating cause and effect of known IKZF1 mutations in in-vivo systems could be required. The majority of existing murine models of Ikzf1 mutations take a sledge-hammer approach, whereby, removal of the entire gene or segments of the gene, causes significant disruptions to the expression of the protein. This effect of the loss of protein structure and/or expression is however, not reflected in the majority of human IKZF1 mutations (54, 55). In general, the loss or reduction of IKZF1 expression in mice causes severe phenotypes. Ikzf1^−/− mice have no B cells and very few T cells (26), whilst Ikzf1^DN/DN mice have a complete absence of lymphocytes and die prematurely (2). Selectively removing sections of Ikzf1 to either reduce its expression (Ikzf1^L/L) (27), or reduce the specific activity of IKZF1 (Ikzf1^ZF1/ZF1 and Ikzf1^ZF4/ZF4) (50) also have broader and more severe defects than those observed in human patients with heterozygous IKZF1 mutations. Conversely, a single point mutation in ZF3 that still maintains correct protein expression leads to a severe phenotype with embryonic lethality when in homozygosity as well as complete absence of lymphoid cell development (46). Phenotypes to this extreme are not commonly observed in human patients with IKZF1 mutations, and whilst there are some phenotypes that are reproduced in both humans and mice expressing mutant IKZF1 alleles, the majority of more subtle clinical features arising from human IKZF1 mutations have not been replicated in murine models (Table 1). An exception to this however, is the Ikzf1^+/L132P mouse, where a heterozygous missense mutation in ZF1 leads to a decrease in mature B cells and progressive decline in immunoglobulin titers as well as a failure to elicit a robust humoral immune response (53). These features are commonly observed in patients with IKZF1 mutations and demonstrates how specific point mutations, rather than entire gene deletions, are more equipped to study the molecular effects of IKZF1 mutant alleles.

Through the use of CRISPR-Cas9, the study of specific point mutations originally discovered in human patients that have been redeveloped into the murine germline has been increasing in popularity (65, 66). This allows for the exact replication of specific point mutations in mice and facilitates the study of how these mutant alleles function in an in-vivo system, that has as of yet been unexplored in the context of IKZF1-related diseases. Additionally, maintaining the endogenous repertoire of Ikaros family members preserves these interactions in a biologically relevant way. Analysis of the changes these mutations make to the RNA and chromatin landscape, combined with analysis of site-directed DNA binding outside of heterochromatin and downstream phenotypic effects will allow for an in-depth characterization of the effect of individual point mutations in IKZF1 causing immune defects in human patients.