Exome Sequencing Diagnoses X-Linked Moesin-Associated Immunodeficiency in a Primary Immunodeficiency Case

Background We investigated the molecular etiology of a young male proband with confirmed immunodeficiency of unknown cause, presenting with recurrent bacterial and Varicella zoster viral infections in childhood and persistent lymphopenia into early adulthood. Aim To identify causative functional genetic variants related to an undiagnosed primary immunodeficiency. Method Whole genome microarray copy number variant (CNV) analysis was performed on the proband followed by whole exome sequencing (WES) and trio analysis of the proband and family members. A >4 kbp deletion identified by repeated CNV analysis of exome sequencing data along with three damaging missense single nucleotide variants were validated by Sanger sequencing in all family members. Confirmation of the causative role of the candidate gene was performed by qPCR and Western Blot analyses on the proband, family members and a healthy control. Results CNV identified our previously reported interleukin 25 amplification in the proband; however, the variant was not validated to be a candidate gene for immunodeficiency. WES trio analysis, data filtering and in silico prediction identified a novel, damaging (SIFT: 0; Polyphen 1; Grantham score: 101) and disease-causing (MutationTaster) single base mutation in the X chromosome (c.511C > T p.Arg171Trp) MSN gene not identified in the UCSC Genome Browser database. The mutation was validated by Sanger sequencing, confirming the proband was hemizygous X-linked recessive (–/T) at this locus and inherited the affected T allele from his non-symptomatic carrier mother (C/T), with other family members (father, sister) confirmed to be wild type (C/C). Western Blot analysis demonstrated an absence of moesin protein in lymphocytes derived from the proband, compared with normal expression in lymphocytes derived from the healthy control, father and mother. qPCR identified significantly lower MSN mRNA transcript expression in the proband compared to an age- and sex-matched healthy control subject in whole blood (p = 0.02), and lymphocytes (p = 0.01). These results confirmed moesin deficiency in the proband, directly causative of his immunodeficient phenotype. Conclusion These findings confirm X-linked moesin-associated immunodeficiency in a proband previously undiagnosed up to 24 years of age. This study also highlights the utility of WES for the diagnosis of rare or novel forms of primary immunodeficiency disease.

Method: Whole genome microarray copy number variant (CNV) analysis was performed on the proband followed by whole exome sequencing (WES) and trio analysis of the proband and family members. A >4 kbp deletion identified by repeated CNV analysis of exome sequencing data along with three damaging missense single nucleotide variants were validated by Sanger sequencing in all family members. Confirmation of the causative role of the candidate gene was performed by qPCR and Western Blot analyses on the proband, family members and a healthy control.
Results: CNV identified our previously reported interleukin 25 amplification in the proband; however, the variant was not validated to be a candidate gene for immunodeficiency. WES trio analysis, data filtering and in silico prediction identified a novel, damaging (SIFT: 0; Polyphen 1; Grantham score: 101) and disease-causing (MutationTaster) single base mutation in the X chromosome (c.511C > T p.Arg171Trp) MSN gene not identified in the UCSC Genome Browser database. The mutation was validated by Sanger sequencing, confirming the proband was hemizygous X-linked recessive (-/T) at this locus and inherited the affected T allele from his non-symptomatic carrier mother (C/T), with other family members (father, sister) confirmed to be wild type (C/C). Western Blot analysis demonstrated an absence of moesin protein in lymphocytes derived from the proband, compared with normal expression in lymphocytes derived from the healthy control, father and mother. qPCR identified significantly lower MSN mRNA transcript expression in the proband compared to an age-and sex-matched healthy control subject in whole blood (p = 0.02), and lymphocytes (p = 0.01). These results confirmed moesin deficiency in the proband, directly causative of his immunodeficient phenotype.

INtRoDUCtIoN
Primary immunodeficiency disorders (PIDs) encompass a diverse group of mostly inherited genetic disorders that compromise immune system function and predispose individuals to recur rent infections and other immune disorders such as auto immunity, hematological disorders and lymphoid malignancies (1)(2)(3). As defined by the International Union of Immunological Societies, there are almost 300 single gene defects described which are causative for the wide range of phenotypes (4), making PIDs a rapidly expanding field of medicine. Loss of function mutations in multiple genes coding for proteins that regulate the actin cytoskeleton are known to cause PID, the most well characterized form being WiskottAldrich (WAS) syndrome (5). This is a rare X chromosomelinked PID where expression and function of the WAS protein results in failure of Arp2/3mediated actin polymerization causing a combined defect of innate and adaptive immunity associated with micro thrombocytopenia, eczema, blood cancers, and increased risk of autoimmunity (5).
Nextgeneration DNA sequencing technologies such as whole genome sequencing (WGS) and whole exome sequenc ing (WES) have revolutionized the field of genomics by enabling a highthroughput and increasingly costeffective method for complex and rare disease diagnosis (6). While WGS is the most comprehensive technique able to identify around 4 million variants within the entire genome, including coding and noncoding regions, WES provides more focus on only proteincoding exons, which harbor around 85% of pathogenic mutations (7,8). Currently, WES permits lower operating costs (US$800 per sample compared to US$1,500 per sample for WGS) with faster data generation and less complex analysis, however, it is unable to identify structural variants (6,9). Here, we used WES to elucidate possible diseasecausing variants in a case diagnosed previously as immunodeficient of unknown cause.

BaCKGRoUND
The proband is a 24year-old male. At age 7 weeks, he had boils in his groin and axillae. At 18 months of age, he had a prolonged episode of bronchitis for which he received antibiotics but was not hospitalized. He suffered repeated upper respiratory tract infections and ear infections for the next few years. He had two episodes of chicken pox (Varicella) at around 18 months and 3 years. The first episode was severe, with widespread skin lesions, but no secondary infection or hemorrhagic lesions. At about 3 years of age, he had a persistent eye infection, which progressed to periorbital cellulitis, secondary to Pseudomonas aeruginosa infection. Neutropenia was first noted during that episode. Bone marrow aspirate showed normal myeloid matu ration, consistent with an immune etiology. No treatment was started at that stage. At 4 years of age, he had an episode of thrombocytopenia, with platelets falling to <5. A further bone marrow aspirate was consistent with an immune mediated thrombocytopenia, and he responded rapidly to intravenous immunoglobulin. He had some further skin infections and paronychiae at this time, and continued to suffer repeated respiratory infections, although not requiring hospitalization. At 6 years 6 months of age, he had his first episode of pneumonia and had several hospitalizations.
At this stage, a diagnosis of combined immunodeficiency was made: he had persistent absolute lymphopenia, normal Tcell receptor Vβ distribution, low numbers of Tcell receptor excision circles (TRECs), no mutation in the common γ chain gene, low IgG, IgA, and IgM. He was negative for HIV by PCR. He started immunoglobulin replacement (IVIg) and GCSF treatment in early 2000 and has very much improved symptomatically since then, however, lymphocyte counts have remained severely decreased (Tables 1and 2).
We previously investigated a functional molecular cause for this undiagnosed immunodeficiency through whole genome arrays. Briefly, analysis of Affymetrix 250K SNP microarray data of the case and a matched healthy control subject via a copy num ber analysis tool identified hyperploidy of a region centromeric to chromosome 14q11.2, mapped over the interleukin 25 (IL25) gene (10). IL25 (a member of the IL17 family of cytokines) induction in mice has been previously associated with a Thelper (Th) 2like pathological immune response (11) and shown to regulate the development of autoimmune inflammation medi ated by IL17producing cells (12,13) and was deemed a plausible candidate for further analyses.
Green et al. (10) paired genetic analysis with transcrip tional profiling and found a large number of genes associ ated with Th1/Th2 profiles to be differentially expressed in peripheral blood lymphocytes of the proband, indicative of a Th2 bias confirmed by flow cytometric analysis. Tcells from the proband and control subject were cultured and activated in vitro with qPCR analysis demonstrating higher IL25 expression in the proband when compared to the control. The skewed Th2 immunity hypothesized in the proband was in line with his susceptibility to infections normally cleared by Th1 responses, such as Varicella and Pseudomonas. Further in vitro studies were performed in Bcell line models to measure the effect of IL25 treatment on cellular proliferation and viability, however when proband and control lymphocytes were treated with exogenous IL25, no differences were observed (data not shown).

Wes analysis
Raw sequences from each library were aligned to the GRCh37/ Hg19 reference genome via the Ion Torrent Server TMAP alignment algorithm. Trio analysis was performed on the Ion Reporter Suite V.5.0 (Life Technologies) where variant annotation identified included singlenucleotide polymorphisms (SNPs) and indels (insertions and deletions) for each exome library.
At a total readdepth of 20×, the target base coverage was 94.63%. A combined total of 51,997 variants were identified with 482 of these unique to the proband. Variant filtering then focused on deleterious variants (frameshift insertion and deletion, stoploss, missense, and nonsense) followed by screening using the OMIM public database as a filter for genes related to immunological disorders. 13 identified variants were then further filtered using in silico missense variant effect prediction tools with the following parameters designated as damaging: SIFT < 0.05, PolyPhen > 0.8, and Grantham Score > 100. SIFT and PolyPhen predict both the structural and functional impact of a variant based on properties such as accessible surface area, ligand contacts, solvent acces sible area, and change in residue side change volume (14). The Grantham score focuses on the difference between amino acid atomic composition, polarity, and volume where a radical amino acid substitution gives a score above 100 (15). MutationTaster predicts whether a variant is tolerable, polymorphic, or damaging and diseasecausing to identify potential diseasecausing variants (16

Western Blotting
Total protein was extracted from cell lysates using Runx protein lysis buffer (containing protease and phosphatase inhibitors) as previously described (18). Protein concentration was measured using the Qubit™ Protein Assay Kit (Invitrogen). A Western Blot using 30 µg protein per sample was performed using an antiMSN primary antibody (ab52490, Abcam) and an HRPconjugated sec ondary antibody (antiRabbit IgG #7074, Cell Signaling). Sample loading was normalized using HRPconjugated antiBetaactin (#5125S, Cell Signaling). Detection of target protein was carried out with ECL (Clarity™ ECL, BioRad) using the Fusion Spectra chemiluminescent system (Vilber Lourmat, Fisher Biotec) and optical density quantitation assessed using Bio1D software.

ethics and Cell Line Validation
This study was performed in accordance with the recommen dations of the Queensland University of Technology Human Research Ethics Committee (Approval number 1400000125). All subjects gave written informed consent in accordance with the Declaration of Helsinki. Written informed consent was also obtained from the proband and family members for the publica tion of this case report. Cell line validations were performed for commercial Toledo nonHodgkin lymphoma (NHL) Bcell and MCF7 breast cancer cell lines (data not shown).

Identification and Validation of a sIRpβ1 Deletion
Copy number variant analysis of the WES data identified a large >4 kbp deletion in the SIRPβ1 gene in the proband (chr20:1, 576,140-1,580,375 Hg38) not present in other family members. Further inspection suggested the deletion in the proband encom passed exon 2 of the SIRPβ1 gene. Validation of this deletion by touchdown PCR and Sanger sequencing confirmed a deletion of exon 2, which has high sequence homology with another exon in the same gene (data not shown). Further Sanger validation to determine the size of the deletion was discontinued due to high sequence homology in that region. Furthermore, this exon 2 deletion has also been found in a number of healthy individuals (19); and therefore, this variant was considered to be an unlikely pathogenic candidate for the immunodeficiency.

Identification and Validation of a Hemizygous X-Linked MsN Mutation
Filtering of the WES data on the Ion Reporter Suite yielded three variants in three immune function genes TET2, NLRP8 and MSN with damaging SIFT and Polyphen scores predicted to be diseasecausing/polymorphic by MutationTaster ( Table 3) The proband is hemizygous Xlinked for the mutation (/T), inheriting the affected allele from his heterozygous (C/T) mother. The TET2 SNP and NLRP8 rare variant occur in healthy individuals with no known disease associations, therefore, they were considered unlikely candidates and not analyzed further. WES coverage for the MSN mutation was 145× for the proband, 119× for the father, 150× for the mother, with an average total coverage of 138×. The MSN mutation was validated by Sanger sequencing ( Figure 1A) and confirmed using IGV for whole genome and exome data in all four family members ( Figure 1B) and as such, considered to be the most likely candidate.

protein Detection and Quantification
MSN encodes the moesin protein (MSN) which is a member of the ezrinradixinmoesin (ERM) family of cell structurerelated proteins. Western Blot analysis confirmed the presence of the 68 kDa MSN protein in the Toledo NHL Bcell line (positive control) and absence of MSN in the MCF7 breast cancer cell line (negative control). MSN was shown to be present in the healthy control, father, mother, and is absent in the proband (Figure 2A). Relative MSN concentrations were normalized against betaactin concentrations where the MCF7 cells and the proband show very low MSN concentrations (<0.5 μg; Figure 2B). These data confirm the proband has a MSN protein deficiency, not present in the healthy control or parents, and is the most likely sole genetic cause for the persistent lymphopenia.

MsN mRNa transcript expression
MSN gene expression was measured in all samples (healthy con trol, father, mother and proband). In whole blood the proband  had significantly lower moesin mRNA expression than the control (p = 0.02), father (p = 0.002) and mother (p = 0.01) and the mother had significantly lower expression than the father (p = 0.01) and control (p = 0.04) ( Figure 2C). In isolated lym phocytes, the proband (p = 0.01) and mother (p = 0.004) had sig nificantly lower expression than the control (Figure 2D). Overall expression levels in all participants were higher in lymphocytes than in whole blood, with MSN mainly expressed in lymphocytes, however, it is also expressed in monocytes, neutrophils and plate lets. The lower levels observed in the mother compared to the father and control and higher levels than the proband are likely due to the heterozygous C > T mutation in the MSN gene. High levels were observed in the healthy control who was age and sexmatched to the proband.

DIsCUssIoN
The variant identified here in the MSN gene is a missense, hemizy gous, singlebase mutation in the proband. This mutation was not identified in current online databases (UCSC Genome Browser, gnomAD) with no reference number and no disease association reported in OMIM at the time of query (March 2016). A genome wide association study (GWAS) search in the GWAS Catalog (20) produced no results for diseasetraitassociated variants in the MSN gene, or other polymorphisms within the MSN gene.
More recently an OMIM entry for Immunodeficiency50 was identified describing a novel Xlinked moesinassociated immu nodeficiency (XMAID) (21). This was the first study to docu ment a moesinassociated disease in humans and describes six male cases from four unrelated families with the same R171W MSN missense mutation as identified in the proband. A seventh case had a different mutation, p.R553X, which caused a frameshift and truncation of MSN in the Factin binding domain. The authors show an impact on lymphocyte function in vitro where moesin deficiency impaired proliferation following activation with mito gen, increased adhesion, and reduced migration (21).
Moesin is a member of the ERM family of cell structurerelated proteins that regulate the cell's actin cytoskeleton (5,22). The mutation identified results in a radical amino acid substitution from basic arginine to nonpolar tryptophan within the func tional fourpointone, ezrin, radixin, moesin domain. However, interestingly both our study and that of LagreslePeyrou et al. suggest that rather than interfering with protein function, the R171W MSN mutation triggers degradation of moesin mRNA in lymphocytes with a loss of expression at the transcript and protein level (21). Moesin has been shown to have a crucial and nonredundant role in lymphocyte homeostasis in mice (23). As such, a lack of moesin protein could prevent efficient lymphocyte migration and egress from lymphoid organs causing persistent absolute lymphopenia in the peripheral blood, as observed in the case. The low TRECs exhibited by the proband at a young age, are an indicator of recent thymic emigrants (24). We compared the LagreslePeyrou et al. cases' (P1P7) and proband phenotypes in more detail and we surmise the P3 case would be the most similar to the proband in terms of clinical outcomes due to both indi viduals presenting with no eczema, no Molluscum contagiosum, and presentation of an autoimmune reaction in the form of TTP LG provided funding support and finalized the manuscript for submission.

aCKNoWLeDGMeNts
We would like to acknowledge Dr. Rachel Okolicsanyi and Chieh (Jade) Yu for sharing their cell culture expertise and K. M. Taufiql Arif for sharing his Western Blot expertise. We would like to sincerely thank the case and family members, as well as healthy control subjects, who agreed to donate blood and share their medical information for this study.

FUNDING
Funding for this study has been generously provided by the Herbert family. Funding has also been provided by the GRC Genomics Lymphoma Project fund.
(P3) and ITP (proband). Considering we cannot confirm whether the P3 case has an IL25 or NLRP8 variant which could play a role in the absence of eczema or M. contagiosum, it would be difficult to determine whether these variants are playing a role in the proband. In addition, as there are no outstanding differences between the proband and all of the other cases, we can be more confident that the R171 MSN mutation is the sole molecular cause for this disease in the proband ( Table 4).
Infants with genetic defects of the immune system that cause severe combined immunodeficiency, or SCID, are effectively identified by populationwide screening practices in the United States (25). Newborns are routinely screened for SCID by using qPCR to quantitate TRECs from DNA extracted from dried blood spots (24); however, this is a laborious assay with a number of limitations. Recently, a newborn case with an R171W MSN mutation was identified by WES (26) when undergoing screening for SCID. Together with our report of an Australian case with XMAID due to the R171W MSN mutation, this data suggests that although rare, it is a site of recurrent mutation.

CoNCLUDING ReMaRKs
Together with our case report, the findings by LagreslePeyrou et al. and Delmonte et al. confirm the effectiveness and validity of WES as a diagnostic method for PIDs and SCID, as well as an investigative method for the identification of novel or rare variants causing novel forms of PID not yet been documented. Identification of these rare and novel variants early in the patient's life will not only aid with swift intervention prior to lifethreatening infections, but may also provide possibility for application of gene therapies and personalized treatment options for these patients.

etHICs stateMeNt
This study was performed in accordance with the recommen dations of the Queensland University of Technology Human