FOXI3 haploinsufficiency contributes to low T cell receptor excision circles and T cell lymphopenia

Abstract

Background:

Newborn screening can identify neonatal T cell lymphopenia through the detection of low copies of T cell receptor excision circles in dried blood spots collected at birth. After a positive screen, further diagnostic testing is required to determine if the subject has severe combined immunodeficiency or other causes of T cell lymphopenia. Even after thorough evaluation, approximately 15% of children with a positive T cell receptor excision circle newborn screening remain genetically undiagnosed. Identifying the underlying genetic etiology is necessary to guide subsequent clinical management and family planning.

Objective:

We sought to elucidate the genetic basis of patients with T cell lymphopenia without an apparent genetic diagnosis.

Methods:

We used clinical genomic testing, functional and immunological assays to identify and elucidate the genetic and mechanistic basis of T cell lymphopenia. For detailed methods, please see the methods section in this article’s Online Repository at www.jacionline.org

Results:

We report two unrelated individuals with non-severe T cell lymphopenia and abnormal T cell receptor excision circles, who harbor heterozygous loss of function variants in forkhead box I3 transcription factor, FOXI3.

Conclusion:

Our findings support the notion that haploinsufficiency of FOXI3 results in T cell lymphopenia with variable expressivity and that FOXI3 may be a key modulator of thymus development.

Capsule summary:

Our data suggest that previously undescribed loss of function variants leading to haploinsufficiency of FOXI3 results in T cell lymphopenia and abnormal T cell receptor excision circle new-born screening results.

Introduction

Despite a large increase in the number of genetically defined inborn errors of immunity¹, a significant minority of individuals identified by newborn screening (NBS) with low number of T-cell receptor excision circles (TRECs) and T cell lymphopenia (TCL) remain without a molecular diagnosis to inform prognosis and guide treatment². We identified a subject (Figure 1A, AII.4) who had an abnormal TREC NBS, followed by persistent TCL. Genomic workup identified a previously unreported heterozygous c.280C>T (p.Gln94Ter) nonsense variant in exon 1 of FOXI3 (NM_001135649.2, Figure 1A). Through ClinVar³, we ascertained a second subject (Figure 1B, BII.2) with TCL and a heterozygous c.74_75dup (p.Ala26Profs*116) frameshift variant in exon 1 of FOXI3.

Figure 1:

Pedigrees of subject 1(A) and subject 2(B). The variants are indicated in text and the Sanger traces covering the variant region are shown for the corresponding individuals. The dark shaded boxes depict the probands in respective families. The probands are indicated by arrows.

Results and discussion

Subject 1 (Figure 1A, AII.4) is an eight-month-old dizygotic twin male, born at 36 2/7 weeks estimated gestational age, who had critical NBS positive for low TRECs (repeated two times, Ct values = 36.58 and 36.81). Immunological laboratory studies indicated a persistent deficiency of both CD4+ and CD8+ T cells with otherwise normal immunological profile (Figure 2A, Supplemental Table 1). The absolute concentration of total B cells was modestly decreased at the most recent examination (1.2 years). Approximately 12% of the B cells had a CD27⁺ memory phenotype; with approximately 4% exhibiting a CD27⁺IgM⁻IgD⁻ switched memory phenotype. Antibody responses were largely normal (Supplemental table 2). He developed transient hyperbilirubinemia after birth that resolved without clinical intervention. His calcium levels were normal (Supplementary Table 1). He has been growing well and meeting developmental milestones and has no cardiac anomalies. He had no dysmorphic features noted after birth but developed a disproportionate increase in head circumference (90^th percentile) compared to other growth parameters (25^th percentile). He passed his hearing test and did not have skin, hair or fingernail abnormalities. He has had two upper respiratory tract infections that did not require antimicrobial medications. At 2 months of age, he received hexavalent vaccines without any adverse effects. His unaffected mother (Figure 1A, AI.2) was not taking any immunosuppressant medication during pregnancy and did not have a history of infections or autoimmunity. His father is also clinically unaffected and does not have a history of recurrent infections or autoimmunity. The subject’s twin sister (Figure 1A, AII.5) passed her NBS, has normal T cell numbers and is clinically well (Figure 2A, Supplemental Table 1). His three older sisters have no history of recurrent or severe infection and no evidence of TCL (Supplemental Table 1). The results of the TREC NBS of these three older siblings were not available. However, the family does not recall being notified of abnormal NBS results. Because oligoclonal T-cell expansions may occur in patients with T-cell immunodeficiencies and may result in normal T-cell counts, we also characterized the TCRVβ repertoire for subject 1 (Figure 1A, AII.4) and his siblings (AII.1,AII.2 and AII.3) and did not observe any significant skewing in any of the subjects (Supplemental Figure 1). In addition, proliferation to TCR agonist (anti-CD3/CD28) was similar in S1 and in control, suggesting that S1 had a normal response to TCR stimulation (Supplemental Figure 2).

Figure 2:

A) Laboratory values of different cell types of the indicated individuals from Figures 1A and 1B. Affected subjects are indicated by arrows in the legend. Persistent T cell lymphopenia (CD3+, CD4+ and CD8+) in subjects 1 (AII.4, green) and 2 (BII.2, brown). The B cells (CD19+) and NK cell counts (CD16+CD56+) were normal in these subjects. The apparently unaffected twin sister of subject 1 (II.5 in Figure 1A) is depicted in purple (AII.5). The corresponding laboratory values for the other siblings in family A are also depicted (AII.1, AII.2, AII.3) in red, blue and orange respectively. The shaded area represents the normal range of the corresponding cells at the respective ages shown along x axis. The y axis represents counts of the corresponding cells in × 10³ cells/mcL.

B) Imaging: Frontal and lateral Chest X-ray images at one month of age of subject 2 demonstrate absence of thymic shadow (arrows).

Subject 2 (Figure 1B, BII.2) is an eight-year-old male who had abnormal TREC NBS and persistent TCL (both CD4+ and CD8+ T cells) ever since (Figure 2A, Supplemental Table 1). He had an otherwise normal immunological profile (Figure 2A, Supplemental Table 1), normal growth and development and was nondysmorphic. A chest X-ray within the first month of life demonstrated an absence of thymic shadow (Figure 2B). He has seasonal allergies but no history of infections. He has adequate antibody responses and lymphocyte proliferation in response to mitogens (Supplemental tables 2 and 3). The calcium levels of S2 were not determined. However, S2 has no features consistent with hypocalcemia/DiGeorge across 7 years of life. He has two apparently healthy siblings (BII.1 and BII.3, Figure 1) with normal new-born screening for TRECs. His parents were clinically unaffected with an unremarkable family medical history, although paternal family history is limited because the father was adopted.

Both subjects underwent genetic testing to identify the genetic basis of TCL. For subject 1, genome sequencing identified a previously unreported heterozygous c.280C>T (p.Gln94Ter) nonsense variant in exon 1 of FOXI3 (Figure 1A). Segregation studies using targeted Sanger sequencing (Figure 1A) showed that this variant was inherited from the apparently unaffected father (Figure 1A, AI.1, Supplemental Table 1) and inherited by two apparently unaffected elder sisters (Figure 1A, AII.2 and AII.3, Supplemental Table 1). The unaffected mother (Figure 1A, AI.2), twin sister (Figure 1A, AII.5, Supplemental Table 1), and the oldest sister (Figure 1A, AII.1, Supplemental Table 1) were negative for this variant. The c.280C>T (p.Gln94Ter) variant has not been observed in gnomAD⁴. A chromosomal microarray analysis (CMA) was unremarkable. There were no other pathogenic or likely pathogenic variants in known genes associated with SCID or in genes associated with TCL⁵ (Supplemental Table 4). Incidentally, we detected a heterozygous c.226C>T (p.Leu76Phe) variant of uncertain significance in UNC119 in subject 1, his father (A1.1), and all siblings except AII.2. The c.226C>T (p.Leu76Phe) variant has not been reported in the literature, however a different variant in UNC119 (p.Gly22Val) has been reported in association with diminished T cell responses to TCR stimulation, CD4⁺ T cell lymphopenia, and infections of viral, bacterial, and fungal origin in a single patient⁶. Furthermore, UNC119 has also been associated with cone-rod dystrophy⁷ which we did not appreciate in any of the family members harboring the UNC119 variant. Based on our data, we do not anticipate a synergistic effect of UNC119 and FOXI3 on TCL in S1 (AII.4).

For subject 2 (Figure 1B, BII.2), genetic testing using a clinical primary immunodeficiency gene panel (consisting of 407 genes that included all known SCID genes and candidate genes) detected a heterozygous c.74_75dup (p.Ala26Profs*116) frameshift variant in exon 1 of FOXI3. This variant was inherited from the apparently unaffected father (Figure 1B, BI.1, Supplemental Table 1). Subject 1’s unaffected siblings (Figure 1B, BII.1 and BII.3), who had normal newborn TREC screening results, did not harbor the FOXI3 variant. The c.74_75dup (p.Ala26Profs*116) variant is rare in the general population and has been observed in heterozygous state in gnomAD (v3.1) in two individuals with a minor allele frequency of 1.38 × 10⁻⁵. This variant has not been reported in the literature. Furthermore, CMA analysis was unremarkable. In addition, a heterozygous c.293C>T (p.Pro98Leu) variant of uncertain clinical significance in BCL11B was also detected in subject 2 (Supplemental Table 3). Although variants in BCL11B have been associated with TCL, the phenotypic spectrum is broad and includes neurological, skeletal and dysmorphic features, among others, all of which were absent in BII.2. Additionally, this variant in BCL11B is reported in 218 alleles in gnomAD v3 and 86 alleles in gnomAD v2, rendering it unlikely to be pathogenic. Whether this variant may be modifying the TCL seen in subject 2 (BII.2) remains unclear at this time. Furthermore, the commercial panel did not include the gene UNC119. We were unable to determine any changes in UNC119 in subject 2. We note that subject 2 did not have any clinical features consistent with cone-rod dystrophy -the primary disease association of UNC119. FOXI3 is located on human chromosome 2 and consists of two exons which encode a 420 amino acid transcription factor with a forkhead DNA binding domain⁸. Homozygous Foxi3 knockout mice have impaired thymus, craniofacial and ear development, whereas heterozygous Foxi3 mutant mice exhibit thymic hypoplasia^8,9,10. Previously, five microdeletions at 2p11.2 have been identified in five families⁹, all with selective TCL and varying degrees of thymic hypoplasia. A common 647 Kb region was deleted in these patients that encompassed nine genes including FOXI3 (Figure 3A). Furthermore, the levels of T cells in our subjects with truncating FOXI3 variants were similar to that observed in patients with deletions encompassing FOXI3⁹ (Figure 3B), suggesting that FOXI3 haploinsufficiency may be driving T cell lymphopenia. The intrafamilial variability in T cell lymphopenia suggests incomplete penetrance.

Figure 3:

A) The previously published deletions encompassing FOXI3 are indicated by black bars. UCSC genes are displayed below the tracks (UCSC Genome Browser, GRCh37/hg19). The position of the variants on the FOXI3 protein are shown. Purple and cyan regions depict the forkhead and DNA binding domains respectively.

B) Comparison of CD3+, CD4+ and CD8+ cells (in × 10³ cells/mcL, y axis) among subjects harboring FOXI3 nonsense, frameshift or deletion variants encompassing FOXI3. The size of the deletion or the variants are indicated next to the respective subjects. The shaded area represents the normal range of the corresponding cells at the respective ages shown along x axis.

C) CADD scores of variants seen in gnomAD in FOXI3 as well as the novel nonsense variant seen in subject 1 are plotted against the log10 of the minor allele frequency in gnomAD.

D) Dual luciferase assays in HEK-293T cells show that the variants detected in subjects 1 and 2 (hFOXI3 FL Gln94Ter and hFOXI3 FL 74_75 dup respectively) have no transcriptional activity compared to the wild type human FOXI3 (hFOXI3 WT).

The variants detected in our two families reside in the first exon of FOXI3 and are predicted to result in nonsense mediated mRNA decay of the FOXI3 transcript, thereby resulting in FOXI3 haploinsufficiency (Figure 3A). These variants have high combined annotation–dependent depletion (CADD) scores (c.280C>T CADD= 35; c.74_75dup CADD= 22.6), consistent with their predicted loss of function effect on FOXI3 (Figure 3A, C). Consistent with incomplete penetrance, three predicted rare loss of function variants are present in gnomAD (v3.1) in this gene, raising the possibility of a wider population of individuals with subclinical TCL (Figure 3C). To determine whether the truncating variants resulted in nonsense mediated mRNA decay of FOXI3 transcripts, we performed a dual luciferase assay using the FOXI3-responsive AE4 promoter¹¹. Both variants showed the same low level of transcriptional activation as the control mock vector (Figure 3D). Our results suggest that these variants are likely functional null alleles and support haploinsufficiency as the mechanism of pathogenesis.

To assess whether FOXI3 variants impair the ability of hematopoietic stem cell to differentiate into mature T cells, we used an artificial thymic organoid (ATO) system¹². CD34⁺ cells, isolated from fresh peripheral blood samples of subject 1 (S1, Figure 1A II.4) and a healthy control (Ctrl), were aggregated with MS5-DLL4 stromal cells to generate ATOs and then cultured in the presence of appropriate growth factors. ATOs were harvested and stained with markers specific of early and late stages of T cell maturation (see Methods for details). Cells isolated from the ATOs generated both from the Ctrl and S1 showed high viability (around 88% and 90%, respectively, Figure 4A (first row) and efficiently differentiated into CD4⁺CD8β⁺ double positive cells (about 72% of L/D⁻CD45⁺CD56⁻ cells in Ctrl and 79% of L/D⁻CD45⁺CD56⁻ cells in S1) and TCRαβ⁺CD3⁺ cells (about 59% in Ctrl and 54% in S1). The absolute counts per ATO of cells at various stages of differentiation (Figure 4B) confirmed that CD34⁺ cells isolated from S1 retained the ability to differentiate into CD4⁺CD8β⁺ double positive and TCRαβ⁺CD3⁺ cells, with absolute count/ATO similar to those obtained in the ATOs generated with Ctrl CD34⁺ cells. These results indicate that the FOXI3 variant carried by S1 is not affecting the T cell differentiation ability of hematopoietic stem and progenitor cells, suggesting that the lymphopenia observed in this subject is likely caused by defects in thymic tissue functionality and/or development caused by the lack of FOXI3.

Figure 4:

In vitro T cell differentiation of positively-selected peripheral blood CD34⁺ cells obtained from a healthy control (Ctrl) and a subject carrying the FOXI3 variant (S1) after 6 weeks of culture in the ATO system. A) FACS plots showing expression of early and late T cell differentiation markers CD7, CD5, CD1a, CD4, CD8β, TCRαβ and CD3 upon gating on LIVE/DEAD (L/D)⁻ CD45⁺ CD56⁻ cells. B) Bar graphs showing the absolute counts/ATO of the different stages of T cell differentiation in the cells isolated from the ATOs.

Our data suggests that FOXI3 haploinsufficiency may contribute to thymus hypoplasia and dysfunction, resulting in lymphopenia. These phenotypes are similar to those caused by heterozygous defects in TBX1^13,14 and FOXN1^15,16, shown to act within the same pathway as FOXI3 in mice⁸. The intrafamilial variability in individuals with FOXI3 haploinsufficiency or with microdeletions encompassing FOXI3⁷ is reminiscent of TCL due to heterozygous disruption of FOXN1. The severity of TCL in individuals with heterozygous pathogenic variants in FOXN1 was less pronounced beyond the first two years of age and normal in most adults. It is also possible that the premature birth of Subject 1 contributed to his TCL, but not in his genotypically affected siblings (AII.2 and AII.3)¹⁷. However, in most cases, TCL associated with prematurity tends to resolve over time, whereas in Subject 1 it remains persistent. The genetic basis of incomplete penetrance remains largely unclear. Although we cannot exclude the possibility that the variants in UNC119 (in Subject 1) and BCL11B (in Subject 2) contribute to T cell lymphopenia in our subjects, further studies are needed to elucidate this possibility.

Here, we report two patients with loss of function variants in FOXI3 that results in abnormal TRECs and T cell lymphopenia and establishes FOXI3 as one of the genes contributing to severe TCL that may lead to abnormal TRECs on NBS. When a child is identified as having low T cells, a defined genetic etiology can help guide management plans and family expectations. Our findings suggest that children who have TCL secondary to pathogenic variants in FOXI3 should be followed closely, but will likely not require definitive treatment. T-cell development may not be impaired significantly enough to cause an increased risk for infection, and it may improve over time. Furthermore, lack of phenotype in relatives with FOXI3 haploinsufficiency, indicates that this condition should be considered in subjects with T-cell lymphopenia, but may not be diagnostic on its own. Clinical care, including frequency of laboratory testing, administration of live vaccines, ability to continue breastfeeding, travel restrictions and isolation requirements, should be individualized and managed in accordance with T-cell numbers and overall clinical history.

Key messages:

• We establish that haploinsufficiency of FOXI3 leads to T cell lymphopenia and abnormal T cell receptor excision circles in newborn screening.

• The T cell lymphopenia in patients with FOXI3 haploinsufficiency is likely due to thymus defects.

Abbreviations:

TCL

T cell lymphopenia

TREC

T cell receptor excision circle

NBS

newborn screening

ATO

artificial thymic organoid