Recurrent microdeletions at chromosome 2p11.2 are associated with thymic hypoplasia and features resembling DiGeorge Syndrome

Abstract

Background:

Thymic hypoplasia/aplasia occurs as a part of DiGeorge Syndrome, which has several known genetic causes, and with loss-of-function mutations in FOXN1.

Objective:

We sought to determine the cause of selective T cell lymphopenia with inverted kappa/lambda ratio in several kindreds.

Methods:

Patients were identified through newborn screening for severe combined immunodeficiency using the T cell Receptor Excision Circles (TREC) assay. Those found to have selective T cell lymphopenia underwent testing by chromosomal microarray analysis (CMA). Three-week-old mice heterozygous for a loss-of-function mutation in FOXI3, a candidate gene within the common deleted region found in patients, were compared to wild-type littermates. Assessments included body and organ weights, flow cytometric analysis of thymocytes and splenocytes, and histologic/transcriptomic analyses of thymic tissue.

Results:

Five kindreds with similar immunophenotypes that included selective T cell lymphopenia had overlapping microdeletions at chromosome 2p11.2 that spanned FOXI3 and, in most cases, the immunoglobulin kappa light chain locus. Studies in a mouse knockout strain for FOXI3 revealed smaller body weights and relatively lower thymus weights in heterozygous, compared to wild-type animals. Histology and flow cytometry on spleen and thymus from three-week-old pups for T and B cell subsets and epithelial cells did not show any significant qualitative or quantitative differences. Transcriptome analysis of thymic RNA revealed divergence in global transcriptomic signatures, and Ingenuity Pathway Analysis revealed predicted dysfunction in epithelial adherens junctions.

Conclusions:

Microdeletions at chromosome 2p11.2 are associated with T cell lymphopenia and probable thymic hypoplasia in humans, and haploinsufficiency for FOXI3, a candidate gene within the deleted region, is the likely underlying cause.

Capsule summary:

A new microdeletion syndrome at chromosome 2p11.2 describes a likely new cause of thymic hypoplasia due to haploinsufficiency of FOXI3.

INTRODUCTION

Thymic hypoplasia/aplasia can occur as a part of the chromosome 22q11.2 deletion syndrome, the most common pathogenic copy number variation (CNV) with an estimated prevalence of 1:2000-4000 children.(1) It often occurs in association with parathyroid hypoplasia/aplasia and conotruncal heart malformations, a triad referred to as DiGeorge syndrome. Other genetic defects described as including features of DiGeorge syndrome include heterozygous loss-of-function (LOF) mutations in CHD7(2), TBX1(3) (which resides within the chromosome 22q11.2 deletion region), TBX2(4), and chromosome 10p deletions.(5) Homozygous or compound heterozygous LOF mutations in FOXN1 cause thymic hypoplasia/aplasia with epithelial defects and are responsible for the similar nude phenotype in mice.(6)

DiGeorge syndrome/thymic hypoplasia is the most common immunologic disorder identified in newborn screening for severe combined immune deficiency (SCID).(7) However, mutations in genes that cause a selective intrinsic defect in T cell development (e.g. IL7R, CD3D, CD3E, CD247, and PTPRC) can mimic thymic hypoplasia in laboratory testing.(8) Both types of defects will cause selective decreases in T cell receptor excision circles (TRECs) and decreased or absent numbers of circulating T cells with normal B and NK cell populations. The size and appearance of the thymus by imaging studies is unreliable since severe intrinsic defects in T cell development often lead to greatly reduced thymic volume.(9) However, in severe cases, the difference is crucial since correction of thymic aplasia requires a thymus transplant while genetic defects severely impairing T cell development in the presence of a normal thymus must be corrected with allogeneic hematopoietic stem cell transplant or, more recently, specific gene therapy in hematopoietic stem cells.(10)

In this report, we describe infants from five kindreds, four of which were identified through newborn screening for SCID and found to have overlapping microdeletions at chromosome 2p11.2 by chromosomal microarray analysis (CMA). The proband from the fifth kindred had previously been diagnosed with a variant of DiGeorge syndrome and was also found to have a 2p11.2 microdeletion by CMA. We noticed that this region spans a FOX gene family member, FOXI3, and that a single previous publication had documented haploinsufficiency for this gene in a patient with aural atresia and agenesis of the internal carotid artery, features suggestive of a defect in pharyngeal apparatus embryogenesis that might extend to thymic development.(11) Although previous studies in mice with a FOXI3 LOF mutation had focused on ear development, we examined heterozygous pups and found a previously unrecognized effect on thymus development, strengthening this as a candidate gene for thymic hypoplasia in our patients.

METHODS

Patients and animals.

Parents of all affected children gave their consent to the research studies and publication of this report in a protocol approved by the University of Alabama at Birmingham Institutional Review Board. Mice heterozygous for FOXI3 LOF were maintained by Dr. Groves in his colony approved by the Baylor Institutional Animal Care and Use Committee (IACUC). For some histology and flow cytometry studies and for tissue processing for RNA/DNA, three week old animals were transferred to UAB and euthanized under approval of the UAB IACUC. The FOXI3 deletion allele (Foxi3-del) was maintained by breeding heterozygous mice from animals derived as previously described.(12) Primers used to genotype embryos were f3G1 (5’-GGC CTT GTC TCA ACC AAC AG-3’), f3G2 (5’-GTT TCC TGT ATC CCT GGC TG-3’) and f3G3 (5’-CTT GGA ATG GGT TGA CTG AG-3’). f3G1 and f3G2 produce a 350bp band corresponding to the wild-type allele, and f3G1 and f3G3 yield a 600bp band corresponding to the Foxi3-del allele. Three-week-old mice were compared to wild-type littermates with respect to body weight, thymus and spleen weights, flow cytometry on thymocytes and splenocytes, histology of spleen and thymus, and thymic transcriptome. Animals were euthanized by isoflurane inhalation followed by cervical dislocation.

Chromosomal microarray analysis (CMA):

Array comparative genomic hybridization (aCGH) was performed using the Agilent 4x180k aCGH+SNP oligonucleotide array (Agilent Technologies, Santa Clara, CA) as part of the patients’ routine laboratory workup. The 2p11.2 microdeletion was confirmed by metaphase fluorescence in situ hybridization (FISH) using the RP11-39F20 probe within the deleted region. Parental and sibling FISH analyses were performed using the same probe to determine whether the deletion was de novo or inherited.

Tissue Digestion and Organ Preparation.

For analysis of cells by flow cytometry, organs were harvested from mice and prepared for single cell suspensions as previously described.(13) In brief, thymus and spleens were isolated from animals and placed in storage buffer on ice. Suspensions of splenocytes from minced organs were stored in RPMI 1640 buffer (Thermofisher Scientific) + 0.5% heat-inactivated fetal calf serum (Thermofisher Scientific) on ice until analysis by flow cytometry. Thymi were harvested and digested for single cell suspensions as described.(13) Cell pellets were resuspended in 10 mL of phosphate buffered saline (PBS, Invitrogen) with 0.5% BSA (0.2 μM filtered prior to use), filtered through 100 μm cell filters (BD Biosciences) and stored on ice until time of flow cytometric analysis. Cell counts were carried out prior to analysis for absolute cells/mL with enumeration by trypan blue exclusion and hemocytometer count.

Flow cytometry

Analysis of single-cell suspensions was carried out on a NovoCyte Flow Cytometer (ACEA Biosciences) with the following antibodies: Alexa Fluor 647 anti-mouse Cluster of Differentiation 326 (CD326, EpCAM-1, Biolegend, Cat 118211); APC-CY7 anti-mouse CD4 (BD Biosciences, Cat B552051); Alexa Fluor 488 anti-mouse CD8α (BD Biosciences, Cat B557668); PE-CY7 anti-mouse CD3ε (BB Biosciences, Cat B552774) and PerCP-Cy5.5 anti-mouse CD45 (Biolegend, Cat 103131).

DNA extraction and sjTREC qPCR analysis.

For DNA extraction, 0.1-0.3 g of minced thymus or spleen was placed in a microcentrifuge tube and combined with lysis buffer and Proteinase K. Organs were digested and DNA extracted as per manufacturer’s recommendations with the DNeasy Blood and Tissue Kit (Qiagen). DNA was eluted in TE buffer and stored at −20°C until time of qPCR analysis. Remaining tissue was flash frozen and stored at −80°C for later analyses.

At time of qPCR analysis, DNA was thawed and prepared for analysis on a LightCycler 480 instrument (Roche) for sjTRECs as previously described.(14) In brief, SYBR green master mix (Roche) was combined with sjTREC primers (Table I, Invitrogen, 1 μM final concentration), and T cell receptor alpha chain (Table I, TCRα, Invitrogen, 1 μM final concentration) and then cycled as per previous specification: 95°C for 5 mins for polymerase activation, followed by 45 cycles of 95°C for 5 seconds, 50°C for 15s, and 72°C for 10s, and then melt curve analysis from 45 to 95°C, before 40°C cooldown for 30s. All samples were analyzed in 20 μl volume reactions. All resultant Cycle threshold (Ct) values were calculated for total sjTREC concentrations as per ΔCt comparison method comparing sjTREC and TCRα Ct values.

Table I.

Sequence of qPCR Primers.

Gene of Interest	Directionality	Primer Seq (5’-3’)
sjTREC	Forward	CAAGCTGACAGGGCAGGTTT
	Reverse	TGAGCATGGCAAGCAGTACC
mTCRα	Forward	TGACTCCCAAATCAATGTG
	Reverse	GCAGGTGAAGCTTGTCTG

Thymic transcriptome analysis.

RNA extraction from tissue.

Prior to RNA extraction, thymus or spleen samples were sterilely dissected from euthanized mice. Tissue was stored in RNAlater (Invitrogen) at −20°C until time of nucleic acid extraction to prevent sample degradation. At time of extraction, tissue was decanted from RNAlater solution and placed in sterile tissue grinder apparatus (Thermofisher Scientific) and ground in Trizol solution (Invitrogen). Samples were then spun at 10,000 x g for 10 min at 4°C to pellet non-homogenized tissue. Resultant supernatant was then processed as per manufacturer’s specifications for RNA. RNA was resuspended in sterile nuclease-free H₂O (Thermofisher Scientific) and then stored at −80°C until used for downstream analyses.

Microarray hybridization.

RNA collected per sample was prepared for hybridization to the Affymetrix Clariom S Mouse GeneChip using protocols recommended by the GeneChip manufacturer (Affymetrix). Quality and quantity of the RNA was ensured beforehand using the Bioanalyzer (Agilent) and NanoDrop (Thermofisher Scientific) respectively. Post hybridization, GeneChips were washed using the Affymetrix Fluidics Station, stained with streptavidin phycoerythrin solution (Molecular Probes), then enhanced by using an antibody solution containing 0.5 mg/mL of biotinylated anti-streptavidin (Vector Laboratories). Post wash and stain, the Affymetrix Gene Chip Scanner 3000 was used to scan the GeneChips and the Affymetrix AGCC software used to generate one CEL file per hybridized RNA.

Microarray data analysis.

CEL files were imported into the Transcriptome Analysis Console (Thermofisher Scientific) and the Console used to generate 22,206 gene-level SST-RMA expression measurements per file. These measurements were then imported into R (www.cran.r-project.org) and analyzed using supported functions therein. Quality of the data and absence of sample-level outliers was assured by covariance-based Principal Component Analysis (PCA) scatter plot (princomp, cor=F) and correlation-based heat map (cor). Noise modeling of the data by sample type (Coefficient of Variation ~ Mean) via locally weighted scatterplot smoothing (lowess) suggested expression measurements with value < 4 to be noise-biased. Measurements that were less than this value were then floored to this value. Measurements for a gene were discarded altogether if there were no measurements greater than the floored-to-value. The Welch-modified t-test (t.test, paired=F, var.equal=F) was then applied to measurements for surviving genes under Benjamini-Hochberg (BH) correction condition and no correction condition. A volcano plot was used to summarize the testing results and the legend therein used to describe the number of genes selected as having dysregulated expression between the two sample types respectively (uncorrected P < 0.05 and absolute difference of means ≥ 1.5X). To evaluate how selected genes collectively separate samples by type, both a 3D PCA scatterplot (plot3d) and clustered heatmap (heatmap.2) were generated. Enriched functions and pathways for the selected genes were tested for and identified using the Ingenuity Pathway Analysis tool (Qiagen).

Statistical analyses.

Testing for Gaussian residuals was tested by Anderson-Darling and Shapiro-Wilk tests to determine parametric / non-parametric data distribution. For parametric data, two-tailed unpaired T tests were utilized for all data analyses. For non-parametric data, Mann-Whitney test was used for all data analyses. Data were graphed and analyzed statistically using GraphPad Prism (v. 8.0, GraphPad). Unless otherwise noted, data are represented by group mean ± Std Dev. For statistical analyses, p < 0.05 was considered statistically significant, and the following abbreviations were used for graphical representation: * for p < 0.05, ** for p < 0.01, and *** for p < 0.001.

RESULTS

Clinical features and laboratory test and imaging results (Table 2, Figure 1).

Table II.

Patient Immunophenotyping

Subject Number:	A.I.1	A.II.1	A.II.2	A.II.3	A.II.4	A.II.5	B.II.2	C.II.2	D.I.2	D.II.3	E.I.1	E.II.2
TRECs/mm3	ND	ND	ND	< 1	5.2	ND	< 5	14.2	ND	9.7	ND	ND
Initial Flow Cytometry (age)	31 yr	6 yr 10 mo	45 mo	3 mo	3 mo	3 mo	1 mo	2 wk	23 yr	2 wk	28 yr	4.5 mo
% CD3	59	58*	51*	6*	21*	36*	3*	31*	78	49*	79	35*
% CD4	32	32	34	5*	18*	24*	3*	22*	30	28*	31	21*
%CD8	25	18	11*	0.3*	2*	9*	0.1*	9*	33	19	37	11*
CD4/CD8 (1.5-2.5)	1.28	1.75	3.12**	15.37**	7.48**	2.5	34.52**	2.37	0.93*	1.48	0.85	1.89
Abs CD3/mm3	1300	1413	1164*	308*	916*	1928*	91*	1655*	1423	1377*	1132	1307*
Abs CD4/mm3	700	774	772	273*	796	1299*	91*	1158*	544	734*	444	794*
Abs CD8/mm3	547	435	250*	16*	88	487*	3*	474*	599	480*	530	416*
% CD19	13	22	20	78**	56**	51**	51*	8*	10	34**	13	54**
Abs CD19/mm3	275	534	455	4272**	2498**	2775**	1431	405	191	971	182	2039
kappa/lambda (1.4-1.9)‡	0.66*	0.68*	0.74*	0.9*	0.99*	1.08*	0.47*	0.42*	1.15*	0.54*	1.84	1.4
% CD3−/CD16+/CD56+	24	17	25**	15	19**	9	46**	58**	4.6	11	8	9
Abs CD3−/CD16+/CD56+	525	411	568**	820	840	487	1291**	3055**	83	311	115	340
Immunoglobulins				2 yr	2 yr	2 yr	2 yr	7 mo
IgG	ND	ND	ND	730***	463	551	453	431	ND	ND	ND	662
IgA	ND	ND	ND	33	17	35	40	33	ND	ND	ND	36
IgM	ND	ND	ND	83	73	64	28*	54	ND	ND	ND	93
IgE	ND	ND	ND	23	3	5	57	ND	ND	ND	ND	ND
Vaccine responses	ND	ND	ND	28 mo	7 mo	7 mo	26 mo	7 mo	ND	ND	ND	11 mo
Tetanus#				ND	3.89	3.26	0.52	1.12				0.2
Pneumococcal§ (Number protective of 14)				13 of 14	10 of 14	10 of 14	8 of 14	8 of 14				10 of 14

^*Low for age (Reference: Shearer et al 2003. Lymphocyte subsets in healthy children to 18 years)

^**High for age

^***on IgG replacement

^‡Deneys et al, 2001. J Immunol Meth 253:23-36.

^#Protective ≥ 0.15 μg/Ml

^§Protective ≥ 1.3 μg//mL

ND: Not Done

Figure 1:

Pedigrees of Five Kindreds.

Kindred A:

Two patients from a set of non-identical triplets screened positive for possible SCID and were referred for further evaluation. Flow cytometry testing revealed that all three of the patients as well as two older siblings and their father had a kappa/lambda ratio inversion and variable T cell lymphopenia (normal in father and one older sibling, mildly to moderately low in all three triplets and one older sibling). CMA and/or FISH analyses demonstrated that all five patients had a 2p11.2 microdeletion (~1.38 Mb in size) with breakpoints at linear genomic positions 87,771,588 and 89,160,192 bp (GRCh37/hg19). Affected patients with lower absolute T cell numbers also had lower relative CD8 T cell numbers and elevation of the CD4/CD8 ratio. One of the triplets (A.II.4) had an asymmetric crying face and another (A.II.3) had feeding difficulties due to dysphagia and primary aspiration. A.II.3 also had transient hypocalcemia in the neonatal period and was started on calcium supplements in the Neonatal Intensive Care Unit, which were subsequently weaned off. An older sibling (A.II.1) had suffered from chronic otitis media, recurrent right cholesteatoma and had mild to moderate conductive hearing loss in his right ear. Magnetic resonance imaging of the temporal bone revealed a markedly abnormal appearance of the right temporal bone with opacified, hypoplastic mastoid air cells and apparent bony destruction of the portions of the ossicles and scutum.

Kindred B:

One patient from a non-identical set of twins screened positive for possible SCID and was referred for evaluation. Flow cytometry showed that circulating T cells were < 100/mm³ with virtual absence of CD8+ T cells. The kappa/lambda ratio was strongly inverted. Because of his very low TRECs and peripheral blood T cells, he was referred for an allogeneic bone marrow transplantation (BMT). CMA demonstrated a 2p11.2 microdeletion (~4.11 Mb in size) with breakpoints at linear genomic positions 85,990,640 and 90,105,896 bp (GRCh37/hg19). This deletion was shown by parental FISH analyses to be de novo in origin. Additional genetic testing included whole exome sequence (WES), which did not reveal pathogenic variants in any relevant genes. He underwent an unconditioned BMT at two months of age from his unaffected HLA-identical female twin sibling. His immunologic reconstitution has been somewhat delayed, specifically with regard to CD8+ T cell development, though he weaned off intravenous immunoglobulin. Twenty-four months after BMT his absolute CD3+ and CD8+ T cell numbers were 491/mm³ and 80/mm³ respectively.

Kindred C:

The proband, a full-term vaginal delivery born without complications, was referred for flow cytometry testing because of abnormal SCID screen. He has been healthy since delivery. Flow cytometry performed three months apart has shown persistent T cell lymphopenia and inversion of the kappa/lambda ratio. CMA demonstrated a 2p11.2 microdeletion (~2.9 Mb in size) with breakpoints at linear genomic positions 87,325,301 and 90,234,023 bp (GRCh37/hg19). This deletion was shown by parental FISH analyses to be inherited from his asymptomatic mother. He has a healthy older brother. The family declined further immunological testing for the mother or brother given the normal clinical course of the proband.

Kindred D:

The proband (a full-term vaginal delivery without forceps) was transferred from his birth hospital to an outside Neonatal Intensive Care Unit at 24 hours post-delivery. He was noted to have an asymmetric crying face. His initial serum calcium on transfer was 7.5 (normal range 8.5-10.2 mg/dL), and this worsened to 6.4 on day 4 (ionized Ca 0.64 μmol/L, normal range 1.12-1.30). PTH was in the low normal range at 23 (normal range 10-65 ng/L), and phosphorus was elevated at 8.9 (normal 4.3-6.8 mg/dL). His calcium and phosphorus levels normalized over the first two months of life, and he was weaned off supplemental calcium and Vitamin D. Genetic studies undertaken because of his positive SCID screen included chromosome analysis of peripheral blood leukocytes, which demonstrated a mosaic 45,X[10]/46,XY[20] karyotype. FISH analysis for the 22q11.2 deletion was negative, but CMA demonstrated a 2p11.2 microdeletion (~2.9 Mb in size) with breakpoints at linear genomic positions 87,325,301 and 90,234,023 bp (GRCh37/hg19). This deletion was shown by parental FISH analyses to be inherited from his asymptomatic mother. Family history is negative for related medical problems. He has one full sibling and one maternal half-sibling who are at risk for having inherited the deletion but have not undergone testing.

Kindred E:

The proband was born prior to the institution of state-wide newborn screening for SCID and was noted to have multiple congenital malformations. She was diagnosed antenatally with hydrocephalus and a head ultrasound/MRI after delivery at 38.2 weeks gestation showed marked lateral and mild 3rd ventriculomegaly, absent septum pellucidum, and holoprosencephaly. She also has cleft palate, cup-shaped ears, micrognathia and low-set, widely spaced nipples. Skeletal survey demonstrated a sickle-shaped sacrum, T11 butterfly vertebrae, hypoplastic 12th rib, and hip dysplasia with bent femurs. Chromosome analysis of peripheral blood leukocytes demonstrated a normal female 46,XX karyotype but CMA revealed two deletions, the first at 2p11.2 (~647 kb in size) with breakpoints at linear genomic positions 88,369,124 and 89,016,224 bp, and the second at 4q33q34.1 (~1.04 Mb in size) with breakpoints at linear genomic positions 171,516,361 and 172,553,132 bp (GRCh37/hg19). The second deletion spanned two long non-protein coding RNA genes (LOC100506107 and LOC100506122) and a microRNA gene (MIR6082), and was interpreted as a variant of uncertain clinical significance. A ventriculo-peritoneal shunt was placed successfully, but she required tracheostomy temporarily because of her micrognathia and airway problems. Echocardiogram showed small/moderate apical muscular and several tiny midmuscular ventricular septal defects with a mildly dilated left atrium and a patent foramen ovale. She was hypocalcemic for the first week of life with total calcium levels in the 6-7 mg/dL range (normal 8.4-11.9), and phosphorus was elevated at 8.2 mg/dL (normal 3.9-6.5). Flow cytometry at about 4.5 months of age showed selective T cell lymphopenia (absolute T cells 1307/mm³, normal 2500-5600)(15) with normal subset distribution. She has moderate motor and intellectual disability as well as severe short stature with hip dysplasia and a possible tethered spinal cord. Whole genome sequencing (WGS) to exclude additional genetic diagnoses is pending.

Hearing screens were being undertaken in all patients at the time of manuscript submission because of the involvement of FOXI3 in inner ear development. To date, the proband in kindred E has been tentatively diagnosed with hearing loss that is at least moderate in one ear, and an older sibling in kindred A has mild to moderate conductive hearing loss in one ear with imaging abnormalities described above.

CMA reveals overlapping microdeletions at 2p11.2 in five kindreds:

Five overlapping 2p11.2 microdeletions were detected by CMA in the five unrelated kindreds that range in size from 647 kb to 4.11 Mb (Figure 2A). The smallest region of overlap (SRO), noted in kindred E, is 647 kb in size and spans 8 annotated RefSeq genes, namely SMYD1, FABP1, THNSL2, FOXI3, TEX37, EIF2AK3, BC046476, and RPIA, as well as a microRNA gene (MIR4780) (Figure 2B). EIF2AK3 and RPIA are autosomal recessive Online Mendelian Inheritance in Man (OMIM) morbid genes. No benign CNVs spanning this genomic region have been reported in the Database of Genomic variants. The 2p11.2 microdeletions noted in four kindreds (A, B, C, and D) also span the immunoglobulin kappa light chain locus, which maps immediately proximal to the SRO (Figure 2B). With the exception of the 4.11 Mb microdeletion, the genomic breakpoints of the other four deletions map within large flanking segmental duplication clusters that contain modules >10 kb in size and with >97% sequence homology, which are shown in Figure 2B under the track named ‘Duplications of >1000 Bases of Non-Repeat Masked Sequence’ and identified by the dashed line boxes. This genomic organization is consistent with non-allelic homologous recombination (NAHR)-mediated loss and explains the recurrent nature of these microdeletions (Figure 2B). The 4.11 Mb microdeletion, noted in kindred B, is notably larger than the other four deletions and has one breakpoint (proximal) mapping within the proximal segmental duplication cluster whereas the distal breakpoint does not map within any segmental duplications, which suggests a different molecular mechanism causing this deletion. This large 4.11 Mb deletion spans 25 additional annotated RefSeq genes, including 4 OMIM morbid genes, in the distal portion of 2p11.2 that includes the genomic region between the distal breakpoints of kindred B deletion and kindreds C and D deletions (Figure 2B). Among these gene is CD8A, which is relevant to the patient’s initial immunophenotype at one month of age (complete absence of CD8+ T cells) because homozygous mutations in this gene have been reported to cause CD8 T-lymphocyte deficiency (OMIM 608957). However, WES performed on this patient’s genomic DNA did not reveal any pathogenic variants in the other CD8A allele. WGS in the proband of kindred E is pending as noted above.

Figure 2:

Mapping of the recurrent 2p11.2 microdeletion (A) Array CGH plots from five UAB kindreds (A to E) with 2p11.2 microdeletions, which are flagged in red. The smallest region of overlap (SRO) is represented by the dashed line box. (B) Genomic map of the recurrent 2p11.2 microdeletion region generated using the University of California at Santa Cruz (UCSC) genome browser. Tracks shown from top to bottom include genomic ruler, cytogenetic band, OMIM genes, UCSC and RefSeq genes, UAB 2p11.2 microdeletion patients (red bars), DECIPHER patients with either 2p11.2 microdeletions or microduplications (red bars represent deletions, blue bars represent duplications, and gray bars represent either deletions or duplications), and segmental duplication clusters (named ‘Duplications of >1000 Bases of Non-Repeat Masked Sequence’ and identified by the dashed line boxes). The SRO is represented by the blue shaded area.

Thymic hypoplasia identified in mice heterozygous for loss-of-function mutations in FOXI3.

In order to explore the role of FOXI3 haploinsufficiency in the selective T cell lymphopenia observed in the majority of patients bearing 2p11.2 microdeletions, we made use of a mutant strain of mice in which a LOF mutation has been introduced in this gene.(12) Studies in three week old heterozygous mice and littermate controls revealed a statistically smaller body weight and statistically smaller absolute and normalized thymic weights in the mutant animals (Fig. 3, Suppl. Fig. 1). However, there were no differences noted in the histology of thymus and spleen from the mutant animals and littermate controls (Suppl. Fig. 2). Flow cytometry for CD3, CD4, CD8 positive cells in thymus and spleen and for thymic epithelial cells did not reveal any significant differences nor were there significant differences in normalized sjTRECS in thymus or spleen.

Figure 3:

Comparison of wild type (WT) and FOXI3 mutant (MT) mouse body and thymic organ weights from three-week-old animals. (A) Body weights (g) of WT and MT mice. (B) Gross thymus weights, and (C) normalized weight of thymus / body weight in WT and MT mice. Bars represent mean ± SD.

Thymic RNA expression differences between FOXI3 mutant (MT) and wildtype (WT) mice revealed discrete differences in transcriptomic profiles as evidenced by principal component analysis (PCA) (Figure 4A). Volcano plot of the genes tested for dysregulated expression between samples representing FOXI3 MT and WT revealed 97 genes with a fold difference of ≥ (+)1.5x (p<0.05) and 28 genes ≤ (−)1.5x (p<0.05) respectively (Figure 4B). A heat map describing the relationship between FOXI3 MT and WT samples for the same 125 genes further highlights differential expression (Figure 4C). To enumerate and compare the differences in significantly enriched biological pathways Ingenuity Pathway Analysis (IPA^®) was employed. Signatures suggestive of perturbations in epithelial adherens junctions in FOXI3 MT mice as compared to WT controls were ultimately identified; top ranked enriched pathways were “remodelling of epithelial adherens junctions” p=0.000363 and “epithelial adherens junction signaling” p=0.000933.

Figure 4.

RNA Expression Differences between FOXI3 MT and WT. A) Covariance-based Principal Component Analysis (PCA) scatter plot describing the relationships between FOXI3 MT (yellow) and WT (blue) samples using expression for 125 gene probes (Affymetrix Clariom_S_Mouse) identified to be dysregulated between them (Welch-modified T-test P < 0.05 & absolute fold-difference of means ≥ 1.5x). B) Volcano plot summary of the gene probes tested for dysregulated expression between samples representing FOXI3 MT and WT using the Welch-modified T-test. Blue-colored circles indicate gene probes having both a Welch-modified T-test P < 0.05 and fold-difference of means ≥ (+)1.5x between FOXI3 MT and WT samples (n=97). Yellow-colored circles indicate gene probes having both a Welch-modified T-test P < 0.05 and fold-difference of means ≤ (−)1.5x between FOXI3 MT and WT samples (n=28). Gray-colored circles indicate gene probes not observed to have either a Welch-modified T-test P < 0.05 or an absolute fold-difference of means ≥ 1.5x between FOXI3 MT and WT samples (n=19,792). C) Heat map describing the relationship between FOXI3 MT and WT samples by Euclidean-based dendrogram in columns (agglomeration method = complete) using expression for the same 125 gene probes described in A). Relationships across these probes is similarly described by dendrogram in rows. Expression observed per probe is summarized using a two-color coding with 64-levels of blue (low expression) to yellow (high expression).

DISCUSSION

This report details the description of a new genomic region at 2p11.2 associated with features of DiGeorge syndrome including apparent thymic hypoplasia, indicated by selective T cell lymphopenia. The association of this 2p11.2 microdeletion with selective T cell lymphopenia in the four kindreds detected by the newborn screen for SCID and, in particular, in the three affected non-identical triplets, indicates that this genomic region contains at least one gene of importance in T cell differentiation/development. The effect of haploinsufficiency appears to be generally milder than that seen in chromosome 22q11.2 deletion syndrome, and there is no distinctive facies or severe congenital heart disease in any of our affected patients. Transient hypocalcemia and asymmetric crying face are seen in some patients. Although the degree of T cell lymphopenia was generally mild to moderate in infants, two of our patients had severely reduced circulating T cells, and one of these underwent a successful allogeneic unconditioned BMT from a matched related donor at two months of age. This patient is now 24 months post-transplant, and despite persistent CD8 lymphopenia (absolute CD8 = 80/mm³), he no longer requires intravenous immunoglobulin and has normal post vaccination serologic titers (see Table II, patient B.II.2).

The five overlapping 2p11.2 microdeletions described here have a 647 kb SRO that spans 8 RefSeq genes. Two of these genes are autosomal recessive OMIM morbid genes (EIF2AK3 and RPIA) that are implicated in phenotypes that do not include T cell lymphopenia. Five other genes have loss-intolerance scores (pLI) between 0.00 and 0.01 suggesting that they are not haploinsufficient. The one remaining gene, namely FOXI3, does not have a pLI score available; however, it is the most likely candidate gene in this region. The detrimental effect of FOXI3 haploinsufficiency on T cell development is likely related to its impact on thymic development. Evidence to support this includes the following: 1) previous clinical report of a patient with a LOF mutation in FOXI3 (11) and congenital malformations suggestive of a defect in the development of the pharyngeal apparatus, 2) studies in mutant mice demonstrating the critical importance of FOXI3 in the development of the pharyngeal apparatus, particularly structure derived from the anterior branchial arches, (12) 3) our studies in FOXI3 knockout mice (Fig. 3, Suppl. Fig. 1) demonstrating that the heterozygous animals have a proportionately smaller thymus than their wild type litter mates, and 4) our thymic transcriptomic data from mice demonstrated clustering on the basis of FOXI3 haploinsufficiency (Fig. 4). Of note, four of the 2p11.2 microdeletions described in this report span the immunoglobulin kappa light chain locus, which explains the inverted kappa/lambda ratio noted in these patients.

FOXI3 is a member of the large FOX family of transcription factors and is known to be required for branchial arch derived structures.(16) Complete LOF is embryonic lethal in mice, but studies in embryonic knockout mice and normal controls have demonstrated that FOXI3 is required for development of otic placodes.(17) Mutant animals also fail to develop the first branchial arch, and the mandibular branch of the trigeminal ganglion is absent.(12) FOXI3 has also been demonstrated to regulate multiple aspects of hair follicle development and homeostasis, and a seven base pair duplication in exon 1 underlies the hairless phenotype and dental abnormalities in three breeds of dogs.(18–21) Examination of the thymus in at least one of these breeds, the Mexican hairless, revealed that the organ underwent atrophy by two months of age with a lymphocyte population that was too sparse to demarcate the cortex and medulla.(22) Thus, the effects of FOXI3 deficiency appear to bear similarities to TBX1 haploinsufficiency, which causes disruption in development of the pharyngeal apparatus, and loss of function in FOXN1, which likely acts to impair thymic epithelial cell development.(6, 23–27) Whether either or both of these potential mechanisms is at play in the effect of FOXI3 deficiency on thymus development remains to be fully delineated. However, in a recently published study, the authors demonstrate that TBX1 and FOXI3 are both important in segmentation of the pharyngeal apparatus, and loss of function, either individually or in combination gives rise to anatomic defects seen in DiGeorge Syndrome including thymic hypoplasia/aplasia.(28) In another very recent publication, Bosticardo and colleagues demonstrate that heterozygous loss of function variants in FOXN1 cause thymic hypoplasia and T cell lymphopenia, most likely due to a gene dosage effect on thymic epithelial development.(29) Given the number of genes observed to be dysregulated in FOXI3 MT mice and that the top two enriched pathways identified for these genes are adherens junction-related, further exploration of the role of FOXI3 in development of the epithelium and its effect on the architecture of the thymus during development would seem warranted. Specifically, dysregulation of genes involved in the formation of epithelial adherens junctions are known to have consequences for epithelial integrity during development, homeostasis, and remodeling.(30)

The 2p11.2 microdeletion region spanning the FOXI3 gene reported here is flanked by segmental duplications that contain genomic modules with more than 97% sequence homology (shown in Figure 2B under the track named ‘Duplications of >1000 Bases of Non-Repeat Masked Sequence’ and identified by the dashed line boxes). This genomic organization is consistent with NAHR-mediated loss and explains the recurrent nature of these microdeletions. However, due to the large size of the flanking segmental duplications, these recurrent microdeletions can vary in size from 650 kb to 2.9 Mb depending on which flanking homologous genomic modules are involved in the NAHR-mediated loss. Similar overlapping deletions have been reported in the DECIPHER and ClinGen Databases to be associated with variable phenotypes and have been interpreted as uncertain; however, it is not certain that these patients underwent an immunologic examination.

CONCLUSIONS:

We report five unrelated kindreds with overlapping microdeletions at 2p11.2 that are associated with low TRECs and T cell lymphopenia. We present data indicating that the T cell lymphopenia is probably due to thymic hypoplasia. Based on the genomic and biological data present here, we suggest that the FOXI3 gene, which maps within the SRO, is likely haploinsufficient and is the likely candidate gene for the T cell lymphopenia in these patients. Based on breakpoint mapping and the genomic architecture of this region, these 2p11.2 microdeletions are recurrent and are most likely mediated by NAHR. We therefore suggest that this 2p11.2 microdeletion defines a previously unrecognized microdeletion syndrome. Patients with this microdeletion may be further identified by inversion of the kappa/lambda ratio, which is due to heterozygous deletion of the immunoglobulin kappa locus, in addition to low T lymphocyte numbers and TRECs. We hypothesize that the phenotype can vary in severity based on other unknown genetic modifiers. More clinical data from other patients with deletions in this region are needed to better define the phenotypic range, and more studies in animal models will be needed to further explore the genetic and molecular mechanisms involved.

Supplementary Material

1

Suppl Figure 1: Mouse thymus whole organ photos

Suppl Figure 2: Histology of mouse thymus and spleen (H&E stained FFPE sections of thymus and spleen as indicated)

Clinical Implications:

We describe a new syndrome of congenital thymic hypoplasia likely due to heterozygous loss-of-function of FOXI3 associated with microdeletions at chromosome 2p11.2. Patients with such microdeletions may often be identified by inversion of the kappa/lambda ratio in addition to low T cell numbers and TRECs.

Abbreviations:

aCGH

Array comparative genomic hybridization

BMT

Bone marrow transplantation

CMA

Chromosomal microarray analysis

CNV

Copy number variation

FISH

Fluorescence in situ hybridization

FOXI3

Forkhead Box I3

pLI

Loss-intolerance scores

LOF

Loss-of-function

MT

Mutant

NAHR

Non-allelic homologous recombination

OMIM

Online Mendelian Inheritance in Man

SCID

Severe combined immunodeficiency

SRO

Smallest region of overlap

TCRα

T cell receptor alpha chain

TRECs

T cell receptor excision circles

WES

Whole exome sequence

WGS

Whole genome sequence

WT

Wild type