Unusual clinical manifestations and predominant stopgain ATM gene variants in a single centre cohort of ataxia telangiectasia from North India

Germline ATM gene variations result in phenotypic heterogeneity characterized by a variable degree of disease severity. We retrospectively collected clinical, genetic, and immunological data of 26 cases with A-T. Clinical manifestations included oculocutaneous telangiectasia (100%), ataxia (100%), fever, loose stools or infection (67%), cerebellar atrophy (50%), nystagmus (8%), dysarthria (15.38%), and visual impairment (8%). Genetic analysis confirmed ATM gene variations in 16 unrelated cases. The most common type of variation was stopgain variants (56%). Immunoglobulin profile indicated reduced IgA, IgG, and IgM in 94%, 50%, and 20% cases, respectively. T cell lymphopenia was observed in 80% of cases among those investigated. Unusual presentations included an EBV-associated smooth muscle tumour located in the liver in one case and Hyper IgM syndrome-like presentation in two cases. Increased immunosenescence was observed in T-cell subsets (CD4+CD57+ and CD8+CD57+). T-cell receptor excision circles (TRECs) were reduced in 3/8 (37.50%) cases.


Genetic investigations. Genomic DNA was isolated from EDTA blood samples using QIAmp DNA Blood
Mini kit (Qiagen, Germany) as per the manufacturer's instructions. Genetic investigations could be carried out in 16 unrelated cases. Genetic analysis in 5 A-T cases was performed at the National Defense Medical College, Japan using a targeted NGS gene panel containing 29 genes implicated in severe combined immunodeficiency and all coding exons of the ATM gene with ion semiconductor sequencing and multiplex PCR amplicons. In the remaining cases, next-generation sequencing was performed using a custom-made panel comprising 44 common PID genes including all coding exons of the ATM gene at the Allergy Immunology Laboratory, Department of Pediatrics, PGIMER, Chandigarh. Briefly, the targeted gene panel consisting of the ATM gene was designed using Ion AmpliSeq Designer (ThermoFisher Scientific, USA). Genomic DNA was quantified using QubitTM Fluorometer (ThermoFisher Scientific, USA) followed by target amplification using PID 2 × 2-primer pool panel. The amplified product was partially digested followed by adapter ligation, barcoding, library purification, and amplification. Further, DNA fragments were immobilized on an ion sphere particle and clonally amplified using Ion One TouchTM 2 instrument. The emulsion PCR results in beads containing amplified and cloned DNA fragments. Elimination of empty beads was done using a robotic enrichment system. Sequencing was performed using Ion S5 instrument and simultaneously processed on an Ion torrent server for assembly and further analysis. Variant calling and analysis of results were made using Ion reporter software (ThermoFisher Scientific, USA). As large deletions and duplications are not detected by this method, BAM files of the patients were also analysed on Integrative Genomics Viewer (IGV) software to check for large deletion/duplication; Supplementary Fig. 1. The identified variants were validated using Sanger sequencing. We have described the ATM variants based on transcript NM_000051.4. Novelty of variants was determined using Human Gene mutation database (HGMD), 1000 genomes database, Genome Aggregation Database (gnomAD) browser. The effect of pathogenic variants on the protein structure and functions was predicted using in-silico tools (Mutation Taster 2.0, PROVEAN, SIFT, and PolyPhen2). SPLICE AI, MaxEnt, and dbscSNV scores were obtained to predict the effect of splicing variants through ensembl variant effect predictor (https:// www. ensem bl. org/ Tools/ VEP). MoBiDiC Prioritization Algorithm (MPA) scores were also obtained to predict the splicing defect.
Statistical analysis. Data

Results
Clinical presentation. Twenty six patients were clinically diagnosed with A-T during the study period.
The mean age at onset of symptoms was 2 years (SD 0.88; range 9 months to 3 years). However, the mean age at diagnosis was 7 years (SD 3.01; range 2-12). A family history of sibling death because of recurrent infections was noted in 3 patients. Parental consanguinity was reported in 5 patients. Two of the patients were related. Ocular or oculocutaneous telangiectasia was present in all patients; Fig. 1A. Some patients presented with unusual but previously reported clinical manifestations. One of the cases (P16) presented with a smooth muscle tumor (SMT) located in the liver, which was confirmed to be an EBV-associated SMT (by EBER ISH). This tumor was initially misdiagnosed as a leiomyoma on routine histopathological examination ( Fig. 1B1-3). Median alpha-fetoprotein (AFP) level was 178.60 ng/ml (range 52.68-611.9 ng/ml). In addition, 2 patients (P8, P15) presented with recurrent infections and an immunological phenotype mimicking Hyper IgM syndrome. The detailed clinical profile has been provided in Table 2.
Immunodeficiency in A-T. A history suggestive of immune defect i.e., recurrent infections, fevers, and loose stool were noted in 14/26 (53.8%) patients. Lymphopenia was noted in all cases (100%) with a mean abso-   . IgG subclasses were analyzed in five cases. IgG1 was found to be reduced (P7 and P8) in two cases and markedly enhanced in one case (P3). IgG2 was reduced in three cases (P1, P7 and P9). Two patients (P8, P15) presented with elevated IgM and decreased level of IgG and A, conforming to a phenotype of Hyper IgM syndrome (HIGM). One of these two patients also had severe T cell lymphopenia. Both these children had recurrent infections since early infancy. Both the patients were initially diagnosed and managed as Hyper IgM syndrome. They developed ataxia and ocular telangiectasia later in life, following which genetic testing confirmed the diagnosis of A-T. Retinitis pigmentosa was observed in one case who also had recurrent infections and respiratory tract infections. Genetic analysis. ATM gene variants were found in 16 out of 25 clinically diagnosed unrelated A-T cases (including a sibling pair). Among these 68.75% (11/16) had homozygous, whereas 31.25% (5/16) had compound heterozygous ATM gene variations. Representative chromatograms revealing compound heterozygous variants in P11 describing trans locations are provided in Fig. 4. In nine cases, genetic investigations could not be performed. These cases were clinically diagnosed when facilities for genetic diagnosis were not available. Karyotyping to detect 7, 14 translocations could not be performed in these patients and most of these patients were lost to follow-up. However, clinical features of patients without a genetic diagnosis were not different from those with a confirmed molecular diagnosis. All diagnosed patients either based on clinical criteria or with a confirmed molecular diagnosis had characteristic features of ataxia, ocular telangiectasia, and increased levels of alpha fetoprotein.
Among 16 unrelated cases (32 alleles) 14 different variants were found. The details of variants have been provided in the Table 3. The most frequent variant was c.67C>T [5/32 (15. Fig. 5A. Premature termination (Stopgain) was found to be the most common genetic variation, present in 56% of cases followed by frameshift (22%), Missense (13%) and splicing variants (9%). Two missense  5 12 9  3  3.5 2  6 10  3  7  9  4  8  8  6  2  9  9  8  7  12  7  3  10  8  11 Age of onset (years) 3 2 1.8 0.9 0.9 0.9 2 2.5 1.   www.nature.com/scientificreports/ variants were found that probably disrupt the splicing site according to the in silico tools provided in Supplementary Table 2. Unfortunately, these predictions have not been confirmed by published transcriptional studies. Figure 5B depicts the distribution of various types of variants among A-T cases. Seven of the 14 variants were found to be affecting the FAT domain, one affecting PI3K domain, and six variants were affecting the N-terminal domains. Position of one variant was observed to be between the FAT and PI3K domain. The most frequent variant c.67C>T affects the N-terminal Tel1/ATM N-Terminal Motif (TAN) of ATM protein. Figure 6 depicts the domain-wise location of ATM gene variants. Out of 8 A-T cases that harbored variants affecting FAT domain, 6 had homozygous variation and 2 had compound heterozygous biallelic variations. In the remaining 2 cases with bi-allelic variant (compound heterozygous state) at-least one allele affected the FAT domain. The details have been provided in Table 3.

Discussion
Ataxia telangiectasia has remained an orphan disease in India. Due to variability in clinical manifestations, differential diagnosis becomes difficult and definite diagnosis is delayed. Recently, Mahadevappa et al. reported clinical features of 100 A-T cases and reported early-onset cerebellar ataxia with mucocutaneous telangiectasia as cardinal manifestations along with recurrent sinopulmonary infections, oculomotor abnormalities, and extrapyramidal involvement as common manifestations 10 . Unusual manifestations have previously been reported in the cases with A-T. Hyper IgM phenotype (elevated IgM with decreased IgG and IgA along with switch defect of B lymphocytes) has been observed in approximately 10% cases with A-T 11 . HIGM is also associated with severe immunological manifestations including autoimmunity, lymphoproliferation and poor outcomes 12 . Noordzij et al. studied the A-T cases with HIGM and proposed addition of DNA repair disorders as a possible cause of HIGM 11 . Various reports indicate that often A-T patients erroneously diagnosed as HIGM 11,[13][14][15][16] . However, marked reduction of absolute T cell numbers along with elevated IgM and AFP could differentiate HIGM 11 . Minto et al. have also reported presence of cutaneous granulomas along with HIGM 17 . The presence of retinitis pigmentosa in the A-T cases has rarely been reported. Lymphoproliferations have previously been reported with A-T, however, its association with EBVs are rarely reported. Reyes et al. reported EBV associated laryngeal leiomyosarcoma and jejunal cellular leiomyoma in patient with A-T 18 . Majority of these cases also manifest immunodeficiency and may be related to immunosuppressive consequences. It would be critical to detect EBVs in A-T for better clinical management. Moreover, ATM protein could be implicated in the life cycle of EBV by using damage repair functions of ATM to facilitate replication 19 .    www.nature.com/scientificreports/ We herein report the spectrum of genetic variants in a North Indian A-T cohort. The majority of AT cases in our cohort harboured non-sense variants (56%) and small frameshift deletions or insertions (22%). The combined outcome from predominant (78%) stop-gain and frameshift variants in A-T results in the severe outcomes. Null mutations contribute to the morbidity and mortality of the A-T disease. This finding is consistent with previous findings that predominantly truncating variants contribute to loss of ATM protein in A-T 20-24 . Sandoval et al. reported nonsense and small frameshift deletions or insertions in more than half of the A-T cases investigated in their cohort 20 . The C → T transition was found to be the most common alteration in our cohort. The C → T transition is associated with methylation of cytosine residues in DNA 25,26 . These epigenetics mediated alterations are associated with CGA → TGA transitions leading to creation of stop codon 27,28 . The CGA (Arg) → TGA transitions are considered to be the most frequent nonsense mutations in the genome followed by CAG(Gln) → TAG transitions 29 . Therefore, GC rich regions have preponderance for hypermutability as described previously in model organisms 30 and various other diseases 31,32 . CpG-rich loci have been associated with germline single nucleotide variations [33][34][35] .
Micol et al. reported that malignancies and respiratory tract infections were the major cause for mortality in the null and hypomorphic variants, respectively 36 . The c.67C>T variant was present in three unrelated patients among which two had homozygous and one had heterozygous variants (Five allelic variants). The germline variant at this position has also been associated with mantle cell lymphoma 37 , gastric cancer 38 and A-T 39 (Clin-Var Accession: VCV000232248.14; Invitae: SCV000748724.5). The c.8473C>T variant was also found in two unrelated A-T cases. All other variants were private variants. Six variants are also being reported as a result of collaborative centralized rapid genetic diagnosis for severe combined immunodeficiency (SCID) in Japan (Manuscript in submission).
Initial studies, after decoding the sequence of ATM gene, identified many pathological variants 40 . Populationspecific ATM variant spectrum has also been reported. Byrd et al. described six mutations affecting the N terminal part of ATM protein in the Irish population 41 . One of these variants was associated with a haplotype and was found to be common in four unrelated families. Similarly, Baumer et al. reported four novel variations in the cases with four different origins. One of the variants was found in the Turkish case that harbored homoallelic 174-bp deletion in the ATM gene due to consanguineous marriage 42 . Previous studies have also reported that the majority of variations in the ATM genes are heteroallelic 20,22,23 . However, similar to our cohort where 68.75% cases had homozygous variants, other recent studies have also reported homozygous variants in the patients with A-T 43,44 . Homoallelic variants in ATM gene develop in a population reflects founder effect. Kuznetsova et al. detected two novel variations in a Caucasian family, a mononucleotide deletion and a mononucleotide insertion, in children as well as prenatal testing in these families 45 .
Though population-based variability in the type of spectrum has been observed, the presence of nonsense and/or premature truncation has been a consistent finding. A recent study conducted in the Russian population reported founder mutations in the patients of Slavic origin with three recurrent mutations (c.5932G>T, c.450_453delTTCT, and c.1564_1565delGA) constituting 57% of total identified alleles 46 . The c.5932G>T (p.Glu1978*) allele was found to be the most frequent in the Russian population. Unlike the Russian population, missense (c.8147T>C) and splice site (c.3576G>A) variants were prominently identified to cause A-T in Dutch, Italian, German, and French cohort 47 . Moreover, the phenotype for these variants manifested as clinically milder A-T and reported longer survival and lesser susceptibility to having cancer, respiratory disease, and immunodeficiency in comparison to retrospectively identified Dutch classical AT cases. There were four variants (two splice site and two missense) in our cohort affecting the splicing event (c.6198+1G>T; c.3077+1G>T; c.7307G>A; c.7788G>C). The c.7307G>A; c.7788G>C alterations occur in the last base pair of coding exon 49 and exon 52, respectively. These alterations are predicted to affect the mRNA splicing event (Supplementary   55 . Various other studies have reported cases with atypical, delayed and/or mild clinic-biological presentations necessitating confirmation of pathogenicity of ATM variants [56][57][58] . Leuzzi et al. reported an atypical case of A-T who recovered from ataxia at the age of 12. She harboured compound heterozygous variation in ATM gene (Inframe deletion: p.Glu709_Lys-750del42; Missense mutation: p.Asp2708Gln). The transcriptional profile of this case matched 90% to the classic A-T profile, however, 10% matched to healthy controls. Amongst the transcripts matched to healthy controls, two critical genes were SCRB1 and THOC3. These genes are involved in wound repair, and genome stability, respectively. Another gene THOC3 is known to be involved in transcriptional elongations, nuclear RNA export, and genome stability 59  However, this study does not vindicate that the differences in the DNA methylation profile between typical and atypical A-T is independent of the genotype. Variability of the phenotype may be associated with preserved function of ATM in the cases harbouring hypomorphic variants rather than independently due to epigenetic alterations.
In the absence of DNA damage, FAT and PI3K domains interact to form a dimer and inhibit the activity of ATM protein. In response to DNA damage, autophosphorylation of serine residue at 1981 location results in the activation of ATM protein due to dissociation of dimer 63 . The majority of mutant alleles in our cohort were confined to the region affecting FAT domain (at C-terminal) and a few affecting the HEAT sequences (N-terminal). HEAT sequences facilitate protein-protein interactions. Our cohort may have a severe outcome due to the involvement of FAT/PI3K domains and the presence of homozygous alterations. Three cases in our cohort with c.6547G>T ATM gene variation affecting FAT domain (position Glu2183), had onset at nine months of age. Another case with onset at 13 months had biallelic variant one of which (c.6100C>T) affected FAT domain at position Arg2034. Truncating variants have a huge impact on the structure, quantity and function of the protein and may be attributed to the severity in A-T. Four among five cases of our cohort who reported recurrent fever, loose stool, and infections harboured variants in the distal regions of the ATM gene and affected FAT or PI3K domains (p.Arg2436Lys; p.Glu2596Asp; p.Arg2486Ter; p.Gln2825Ter). However, the genetic analysis could not be established in one case.
ATM protein acts as a master regulator of double-stranded break response. The impairment in the response to DSBs leads to altered V(D)J recombination process which further results in cellular anomalies due to reduced end tethering; altered cell cycle checkpoints; programmed cell death 64 67 . We also reported enhanced T cell immunosenescence especially in the T cytotoxic cells.
We could not study the role of signalling cascade of MRE11/RAD50/NBN (MRN) complex in our A-T cohort. The molecular expression and interaction of ATM protein with MRN complex influence the clinical severity. The MRN complex recruits ATM protein upon sensing the DSBs and facilitates repair. Moreover, MRN complex facilitates telomere stability [68][69][70] and may involve in the immunosenescence observed in the A-T cases. Epigenetic modifying factors may also govern the A-T phenotype and provide resilience to some cases. Murine RAD50 s model has shown alleviation of classic A-T manifestations including senescence, radiosensitivity, and malignancy 71 .
Since, this was a retrospective analysis, some of the patients in the cohort are now non-ambulatory and not coming for follow-up. We could not firmly establish in the case of compound heterozygous variants in patients whether they were in cis or trans. We could only predict the functional impact of splice site variants using insilico prediction tools but could not perform transcript analysis in four cases with splice site variants. Moreover, large genomic rearrangements including deletions/duplications could not be performed by alternative methods such as MLPA. Oncologic risk management could not be established in consanguineous parents.
We report the data of 26 cases clinically correlated with Ataxia-Telangiectasia phenotype according to the hallmark clinical and biochemical markers along with genetic analysis in sixteen unrelated cases. To the best of our knowledge, this is the first study reporting genetic spectrum and T-cell receptor excision circles data of Indian cases with Ataxia Telangiectasia. Current retrospective analysis limits us to deduce the temporal change in www.nature.com/scientificreports/ severity due to limited data. The available data reveal higher null and truncating variants, predominantly reduced IgA, as well as hyper IgM-like phenotype in two cases, altered TRECs and sjKRECs, indicating an underlying immune defect. However, a larger cohort with classical and atypical A-T phenotypes would confirm these findings or give a clearer picture of the spectrum of ATM mutations in North India.

Data availability
Data will be made available on request.