AIOLOS, encoded by the IKZF3 gene, is a member of the IKAROS zinc finger (IKZF) transcription factor family. AIOLOS was first discovered as a binding partner of the pioneer molecule of the IKZF family IKAROS (1). The early studies of Aiolos-deficient mice revealed that Aiolos regulates B cell activation and maturation, Th17 differentiation, and natural killer (NK) cell maturation (2–4). Ikaros and Aiolos cooperatively regulate the expression of λ5, a pre-B cell receptor component, in pre-B cells (5). They also suppress c-Myc expression in pre-B cells, inhibiting pre-B cell proliferation (6). Aiolos, together with Ikaros, is also involved in T cell development. They regulate CD8α expression in thymocytes and regulate Bcl-6 expression in follicular helper T cells (7, 8).

IKZF family transcription factors share the N-terminal C2H2 zinc finger (ZF) domain that is required for DNA binding and the C-terminal C2H2 ZF domain that mediates homo- and heterodimerization (Figure 1A) (9). IKZF proteins comprise the nucleosome remodeling, histone deacetylase complex and polycomb repressive complex 2 in the form of homo- and heterodimers (10, 11). They modulate the chromatin status of the target genes and mediate the activating as well as the repressive effects. Moreover, these proteins play pivotal roles in maintaining and developing various hematopoietic cells, but are especially important for the development of lymphocytes.

IKAROS was the first member of IKZF family protein to be discovered as the causative gene in the patients with inborn errors of immunity (IEI) (12–14). Heterozygous loss-of-function variants of IKZF1 result in IKAROS deficiency, which is characterized by a reduction in the levels of B cells and hypogammaglobulinemia often associated with recurrent sinopulmonary infections and other infections (9). Autoimmunity and hematologic malignancies are frequently observed to occur. In particular, IKAROS^N159S/T variants exert dominant-negative effect against wild-type IKAROS (14–16). Even though the wide range of clinical phenotypes is observed in the patients with IKAROS loss-of-function variants, the patients with IKAROS N159 missense variants present with a profound combined immunodeficiency phenotype (9, 17). All patients developed Pneumocystis jirovecii pneumonia (PjP) in the first years of life. Various bacterial, viral, fungal, and parasitic infections were also observed. The patients with IKAROS N159 missense variants manifest profound B-lymphopenia associated with severe hypogammaglobulinemia. T cells were present but manifested predominantly naive phenotype, and Th polarization defect was observed. The heterozygous and homozygous germline variants of HELIOS have been recently identified in the patients with recurrent sinopulmonary infections, chronic Candida infection, hypogammaglobulinemia, and lymphoma (18, 19). Heterozygous HELIOS variants have also recently been reported in patients with autoimmune diseases such as systemic lupus erythematosus, and Epstein–Barr virus (EBV)-associated hemophagocytic lymphohistiocytosis (20). In addition, heterozygous PEGASUS variants have been identified in a cohort with hereditary thrombocytopenia (21).

Somatic mutations of AIOLOS have been reportedly found in hematologic malignancies, including B-cell acute lymphoblastic leukemia (ALL) and chronic lymphocytic leukemia (CLL) (22–25). Two recent studies revealed the germline variants of AIOLOS in human diseases (Figure 1B) (26, 27). Therefore, we reviewed the clinical characteristics and murine models of the AIOLOS variants in these studies.

IKZF family of transcription factors is evolutionally well-conserved among species (28), with human and murine Aiolos sharing 87% of the amino acid sequences. In the studies of AIOLOS variants in IEI, knock-in mouse models were produced using CRISPR/Cas9 gene-editing technology to further investigate detailed immunophenotypes and elucidate the pathogenic mechanisms caused by these AIOLOS variants. AIOLOS^G159R and AIOLOS^N160S in humans correspond to Aiolos^G158R and Aiolos^N159S in mice, respectively.

The heterozygous AIOLOS^G159R/Aiolos^G158R and AIOLOS^N160S/Aiolos^N159S variants caused adaptive immune defects in humans and mice (Figure 2). Both of G159R and N160S variants reside in the ZF2 of the AIOLOS protein, one of four N-terminal ZF that mainly mediates DNA binding. The roles of each ZF are extensively studied in IKAROS. ZF2 and ZF3 of IKAROS are essential in binding to the core IKZF motif, whereas ZF1 and ZF4 determine the specificity to the flanking sequences (31, 32). Within ZF2 and ZF3 of IKAROS, substitutions of the amino acids at -1, 2, 3, and 6 positions in reference to the beginning of the α-helix result in abrogated DNA binding (31, 33, 34). Regarding the positions of substituted amino acids in the AIOLOS variants, G159 and N160 correspond to 2 and 3 position of helical positions, respectively. Thus, missense variants of these amino acids would alter the DNA binding ability of AIOLOS. In vitro and in silico analyses conducted in the studies of AIOLOS^G159R and AIOLOS^N160S supported this hypothesis. The fundamental question lies in understanding the clinical and immunological phenotypes of the patients and mouse models; the mutated amino acid is next to each other, but why were there so many differences? For example, the AIOLOS^G159R variant is associated with a profound decrease in B cells and possible EBV susceptibility, but two of the three patients were not hypogammaglobulinemic. AIOLOS^N160S variant is associated with P. jirovecii infection and severe hypogammaglobulinemia, but B cells were not decreased in the patients’ peripheral blood. The phenotypes of the patients were well recapitulated in the mouse models with the homologous variants. Several hypotheses can explain the difference between AIOLOS^G159R/Aiolos^G158R and AIOLOS^N160S/Aiolos^N159S, such as 1) the severities of functional defect are different among variants; 2) neomorphic attributes of the variants affect the phenotype; or 3) protein interactions differ among the variants. Both variants were loss-of-function in terms of binding to the canonical consensus sequence and formation of PC-HC and acted in negative-dominance against wild-type AIOLOS. AIOLOS^G159R exerted the dominant-negative effect against IKAROS as demonstrated by PC-HC formation and luciferase reporter gene assays but not AIOLOS^N160S. Thus, the degree of functional damage may be greater in the G159R variant. However, patients with AIOLOS^N160S did not necessarily exhibit milder phenotypes. The neomorphic aspects of the mutations may also explain the phenotypic differences. ChIP-seq analyses of AIOLOS^G159R and Aiolos^G158R detected the novel binding motifs preferentially bound by these variants. In contrast, AIOLOS ChIP-seq in the T cells of patients with AIOLOS^N160S did not alter the binding motifs but might have been affected by the bindings of wild-type AIOLOS as half of the AIOLOS protein in the cells would have been the wild-type. In this regard, it is important to compare genome-wide bindings of Aiolos and Ikaros between Aiolos^G158R/G158R and Aiolos^N159S/N159S mice to evaluate how differential bindings by the different AIOLOS variants affect the phenotypes of the patients and mouse models. The current evidence suggests that the newly acquired bindings by AIOLOS^G159R and Aiolos^G158R do not have cis-regulatory effects on surrounding genes but rather work as sequestration of IKAROS to non-physiological binding sites. Although the N-terminal ZFs (ZF1-4) of AIOLOS is known to mediate DNA binding, there is a possibility that the variants in these ZFs also affect protein–protein interaction. The alteration of protein–protein interactions by each variant will be the subject of the future research.

Alternative mRNA splicing of IKZF molecules results in multiple isoforms with differential patterns of domains present in each isoform. The isoforms lacking the whole DNA binding domain or core binding ZFs (ZF2 and ZF3 for Ikaros) are designated as dominant-negative isoforms. Dominant-negative isoforms lack DNA binding ability while maintaining dimerizing capacity, and they inhibit the DNA binding of dimerizing partners. The dominant-negative isoforms of IKZF proteins can be detected in normal lymphocytes (35–37). However, the specific roles of these dominant-negative isoforms in the lymphocyte development are still not known. The expressions of dominant-negative isoforms of IKZF molecules are also observed in various hematologic malignancies (37–41). IKAROS^N159S somatic mutation is also found in ALL (42). Contribution of dominant-negative IKAROS to the leukemogenesis is further supported by the observations of spontaneous development of T-cell leukemia and lymphoma in the heterozygous Ikaros dominant-negative (ΔZF1-3) mice and the mice harboring heterozygous Ikaros^H191R mutation (43, 44). Conditional deletion of Ikaros exon 5 (coding ZF2 and ZF3) in common lymphoid progenitors (Cd2-Cre; Ikaros E5^fl/fl ) also developed T-lymphoid malignancies (45). Pre-B cells of Ikaros Cd2-Cre; Ikaros E5^fl/fl mice underwent leukemic transformation when transplanted to immunodeficient NOD SCID gamma (NSG) mouse. Ikaros-deficient mice lack B cells, NK cell, and fetal T cell, while T cells develop postnatally (46). On the other hand, Ikaros dominant-negative (ΔZF1-3) mice and mice with ENU-mediated missense mutation in ZF3 (H191R) demonstrate severer lymphocyte developmental defect characterized by absent T, B, and NK cells (44, 47). Ikaros^H191R/H191R fetal liver cells demonstrated defect in terminal myelopoiesis and erythropoiesis. Moreover, Ikaros^H191R/H191R mice were fatally lethal (44). The severer phenotypes observed in Ikaros dominant-negative mice when compared to Ikaros-deficient mice have been hypothesized as the results of dominant-negative effect against other IKZF molecules by the mutant Ikaros. The studies of AIOLOS variants provided counterpart of this hypothesis. Heterozygous and homozygous Aiolos^G158R and Aiolos^N159S variants resulted in severer lymphocyte developmental defects when compared to Aiolos-deficient mice. These observations indicate dominant-negative effect of Aiolos^G158R and Aiolos^N159S variants against other IKZF molecules. Precisely, it is not clear whether the phenotypes of these mutant mice solely resulted from the dominant-negative effects against other molecules or partially due to neomorphic function by the variants. Also, the roles of each ZF of Aiolos in lymphocyte development have not been studied. In this regard, it is essential to compare the phenotypes of Aiolos dominant-negative (ΔZF1-4) mice, mice lacking individual Aiolos ZFs, and the missense Aiolos^G158R and Aiolos^N159S mutant mice.

One interesting feature of the AIOLOS^N160S variant is its susceptibility to PjP. The homologous variant in IKAROS (N159S) is associated with similar clinical and immunological phenotypes, including PjP susceptibility (14, 15). B cell levels in patients with AIOLOS^N160S were normal to high, whereas class-switched memory B cell and plasmablast levels were low. In addition, CD23 and CD21 expressions were decreased on the B cells. In contrast, B cells in IKAROS^N159S patients were almost absent. Both variants were associated with severe hypogammaglobulinemia. T cell abnormalities were characterized by increased naive CD4⁺ T cells and impaired Th polarization (decreased Th1 in AIOLOS^N160S and decreased Th1, Th2, and Th17 in IKAROS^N159S patients) in both AIOLOS^N160S and IKAROS^N159S patients. Impaired CD40/CD40L signals were the cause of PjP susceptibility in patients with AIOLOS^N160S but no CD40L signal defects were reported in patients with IKAROS^N159S. The precise mechanism underlying PjP susceptibility in patients with AIOLOS^N160S and IKAROS^N159S should be investigated in future studies.

The development of hematologic malignancies is a common clinical characteristic of IEI caused by AIOLOS variants. EBV-associated lymphoma was observed in two of the three patients with AIOLOS^G159R, and CLL was observed in one of the four patients with AIOLOS^N160S. The tendency to develop hematologic malignancy in patients with the AIOLOS variants is plausible as they were observed in patients with IKAROS and HELIOS deficiencies (17, 19). Additionally, germline variants of IKAROS were also detected in cohorts of patients with B-ALL (48). Although, the precise mechanism underlying tumorigenesis in patients with the AIOLOS variants remains unknown. Two patients with AIOLOS^G159R developed EBV-associated lymphoma, whereas another patient presented with chronic and recurrent EBV infections. Thus, the underlying cause of lymphomagenesis is possibly EBV susceptibility. On contrast, somatic loss-of-function AIOLOS mutations were reported in B-ALL (22). Moreover, aged Aiolos-null mice develop B-cell lymphoma (2). These studies and reports of IEI caused by AIOLOS variants indicate a tumor-suppressive role of AIOLOS and the development of hematologic malignancies following the functional deterioration of AIOLOS. Alternatively, altered interaction with other IKZF molecules and heteromerizing partners or neomorphic features of the AIOLOS variants may have led to cancer development in the patients. The somatic AIOLOS^L162R mutation was repeatedly reported in CLL (25). AIOLOS^L162R variant was not proven to be loss-of-function in DNA binding and PC-HC targeting abilities (27, 30). Studies on the murine homologous variant Aiolos^L161R suggested that this variant enhanced BCR and NF-κB signaling, contributing to the CLL-like disease development in aged mice. It is not clear if CLL development in patients with AIOLOS^N160S shared the same molecular mechanism as in AIOLOS^L162R/Aiolos^L161R.

Studies on AIOLOS^G159R/Aiolos^G158R demonstrated that these variants can cause pathogenicity through heterodimerization with Ikaros, interfering with the physiological function of Ikaros. Seeking therapeutic potential for the patients with IEI caused by AIOLOS variants, we performed gene therapy for Aiolos^G158R mice. A small indel was introduced before the dimerizing ZF (ZF5) of the mutant Aiolos allele, designated as Aiolos^G158R:Δc-ZF , using CRISPR/Cas9 gene-editing technology (26). This mutant theoretically had a DNA-binding domain, but it could not form a dimer with wild-type Aiolos or Ikaros and other dimerizing partners. The Aiolos^{+/G158R:Δc-ZF} mice recovered the B-cell levels in the peripheral lymphoid tissues and early B-cell development in the bone marrow was normalized. T cell abnormalities that were observed in Aiolos^+/G158R mice were also normalized. These findings suggest the therapeutic potential of gene therapy and small molecules that interfere with the AIOLOS–IKAROS interaction. However, applying this strategy to the patients will be challenging. Germinal center B cells increased in mice with biallelic C-terminal deletion in the Aiolos, as previously reported in Aiolos-deficient mice. Considering autoimmunity and spontaneous lymphoma development in Aiolos-knockout mice, selective inhibition of mutant AIOLOS allele/protein would be ideal strategy.

Increasing number of novel IEIs is being reported with the advancement of genetic diagnostic technologies. However, most newly described IEIs are sporadic and are often confined to a handful of patients. For example, the first reports of IEI caused by AIOLOS variants include only seven patients in two families. Although these studies address extremely rare disease, detailed phenotyping of the patients and thorough understanding of underlying mechanism using animal models can provide insights into the unknown but essential aspects of Aiolos in the development and function of lymphocytes. Future studies should address the differences in the phenotypes of immunodeficiency caused by AIOLOS variants and novel therapeutic approaches that specifically inhibit mutant AIOLOS proteins without interfering with the unaffected allele and other heterodimerizing partners should be developed.

MY and TM wrote the manuscript. All authors contributed to the article and approved the submitted version.

This work was supported by JSPS Grant-in-Aid for Young Scientists JP20K16884 (MY) and Scholarship grant 91AA191397 (TM).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The handling editor declared a past collaboration with the authors MY and TM.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.