Next Generation Sequencing for Detecting Somatic FAS Mutations in Patients With Autoimmune Lymphoproliferative Syndrome

Autoimmune lymphoproliferative syndrome (ALPS) is a primary immune regulatory disorder clinically defined by chronic and benign lymphoproliferation, autoimmunity and an increased risk of lymphoma due to a genetic defect in the FAS-FASL apoptotic pathway. Genetic defects associated with ALPS are germinal and somatic mutations in FAS gene, in addition to germinal mutations in FASLG, FADD, CASP8 and CASP10 genes. The accumulation of CD3+TCRαβ+CD4-CD8- double negative T-cells (DNT) is a hallmark of the disease and 20-25% of ALPS patients show heterozygous somatic mutations restricted to DNT in the FAS gene (ALPS-sFAS patients). Nowadays, somatic mutations in the FAS gene are detected through Sanger sequencing in isolated DNT. In this study, we report an ALPS-sFAS patient fulfilling clinical and laboratory ALPS criteria, who was diagnosed through NGS with a targeted gene panel using DNA from whole blood. Data analysis was carried out with Torrent Suite Software and variant detection was performed by both germinal and somatic variant caller plugin. The somatic variant caller correctly detected other six ALPS-sFAS patients previously diagnosed in the authors’ laboratories. In summary, this approach allows the detection of both germline and somatic mutations related to ALPS by NGS, avoiding the isolation of DNT as the first step. The reads of the somatic variants could be detected even in patients with DNT in the cut off limit. Thus, custom-designed NGS panel testing may be a faster and more reliable method for the diagnosis of new ALPS patients, including those with somatic FAS mutations (ALPS-sFAS).

Heterozygous acquired or somatic mutations in the FAS gene are found in 20-25% of ALPS patients (10,13,17). These mutations are mainly restricted to the exons 7, 8 and 9 of FAS gene, mostly detectable in DNT-cell population, but also in approximately 10-20% of single-positive mononuclear cells (18). Phenotypically, DNT derives from both CD4 and CD8 effector memory T-cells with CD45RA expression, representing a subset of terminally differentiated T-cells that should have been deleted by apoptosis (19). Although elevated DNT is a required criterion of the disease, its pathologic role in ALPS remains unclear, but as ALPS-FAS and ALPS-sFAS are clinically indistinguishable, these cells could contribute to disease pathogenesis (10,13).
Decisional trees for the diagnosis of ALPS have been proposed (5,14,20). It is established that if the clinical features of ALPS are present, DNT immunophenotyping and ALPS-biomarkers assessment are the first step of the algorithm (14). If DNT and any other ALPS-biomarkers are elevated, ALPS suspicion is confirmed and sequencing of the ALPS-related genes is mandatory. If no germinal mutation is found in these genes, searching for a somatic FAS mutation should be performed through Sanger sequencing in isolated DNT. Sanger sequencing can be a useful and fast tool for sequencing of specific genomic regions. However, Next Generation Sequencing (NGS) technologies such as targeted gene panels and whole exome sequencing have become powerful tools that allow simultaneous screening of heterogeneous genetic diseases as ALPS and ALPS-like disorders (14).
Generally, DNT represent less than 3% of total leukocytes (17). The finding of somatic mutations in the FAS gene requires the isolation of DNT population, a costly and time-consuming technique that may not be feasible in some routine laboratories. Based on the classical NGS application for detecting a relatively low mutational burden (21)(22)(23), we propose the possibility of detecting somatic mutations by DNA sequencing from whole blood using a targeted gene panel. This approach could allow the simultaneous screening of ALPS patients with both germline mutations in the ALPS related genes and somatic mutations in the FAS gene.
In this study we report a new case of ALPS-sFAS diagnosed by NGS through a targeted gene panel. The application of the somatic variant caller was validated with the detection of somatic mutations in other ALPS-sFAS patients previously diagnosed in the authors' laboratories (13,24,25).

DNT Immunophenotyping
For direct immunofluorescence, peripheral whole blood was incubated using the corresponding monoclonal antibodies as previously described (14). Cellular acquisition was performed using a Beckman Coulter Navios cytometer (Beckman Coulter, Miami, FL, USA) and data were analyzed with Kaluza 2.1 software (Beckman Coulter).
Eight ALPS-sFAS patients were included in this study showing different DNT percentages (from 4 to 30.7%) ( Table 1).

Isolation of DNT
DNT were isolated by immunomagnetic cell separation (MACS) using a double-negative T-cell isolation Kit (Miltenyi Biotec GmbH, Germany) according to the manufacturer's instructions as previously described (14).

FAS-Induced Apoptosis
Functional studies for FAS-mediated apoptosis was performed as previously described (14) for screening of ALPS patients.

Variant Caller, Annotation and In Silico Prediction
Data analysis was carried out with Torrent Suite Software (version 5.12). Data produced were aligned and mapped to the human GRCh37 (hg19) genome assembly by Torrent Mapping Alignment Program (TMAP). For germline variants, the variant caller plugin with default setting of "Generic PGM Germ Line Low Stringency" (Thermo Fisher Scientific) was performed.
Somatic variants were detected by customization of the default setting of somatic variant caller plugin from Torrent Suite Software (TVC, v5.12. Thermo Fisher Scientific) considering only the exons 7, 8 and 9 of FAS gene. The following parameters were adapted from the somatic variant caller of Torrent Suite server: minimum allele frequency permitted was 0.01, minimum allele coverage was defined as 3 reads, no minimum coverage on either strand was defined but those variants with strand bias above 0.85 and strand bias pvalue below 0.05 was filtered out.
Variant annotation was performed with Ion Reporter Software (version 5.10). The software Integrative Genomics Viewer (IGV) (26) was used to inspect the aligned reads (bam files) of each sample in order to check the consistency of nucleotide calls. VarSome (27) was used for in silico pathogenicity prediction and for a comprehensive interpretation of the variants.
The somatic FAS mutations were detected with this approach in seven out eight patients. It is important to consider that in the patient 8 (P8) which we could not be able to detect the somatic mutation, he had the DNT in the limit for our consideration as an ALPS patient (required criterium). In P1 and P4, the mutations were detected even with slightly high DNT.

Sanger Sequencing
Germline and somatic variants were confirmed by Sanger sequencing using the appropriate specific primers.

Statistics
Sensitivity and specificity of the assay was calculated considering sensitivity as true positive rate (87.5% of ALPS-sFAS patients were correctly identified) and specificity as true negative rate (93.1% of controls were correctly identified).

Patients
Patient 1 (P1) was admitted to the hospital with fever and exanthema at ten months of life. One year later he suffered from another febrile episode with exanthema and ALPSsuspected phenotype: splenomegaly, lymphadenopathies and bicytopenia (thrombocytopenia and neutropenia) and positive anti-smooth muscle antibodies. DNT, vitamin B12, IL-10, sFASL and immunoglobulins G and A were elevated ( Table 1) confirming the ALPS suspicion. Patients 2-8 (P2-P8) are ALPS-sFAS patients previously diagnosed in the authors' laboratories ( Table 1) (13,24,25). The studies involving human participants were reviewed and approved by ComiteÉ́t ico de Investigación Clıńica, University Hospital 12 de Octubre and by ethic committee of CPP Ile-de-France (CPP IDF2 DC-2014-22722015-03-03 AF).

Identification of Germinal and Somatic Mutations Through Targeted Next Generation Sequencing
NGS using DNA extracted from whole blood of P1 ( Figure 1) revealed two germline heterozygous mutations in PRF1 (c.1179C>A; p.Cys393Ter) and UNC13D (c.2411A>G; p.Tyr804Cys) genes ( Figures 1A, B), both considered as ALPS-phenotype-modifiers but not ALPS-causing genes (11,12,15). Functionally, intracellular perforin staining and degranulation assay in NK cells were similar to the healthy control tested (data not shown).  Figure 1C). For somatic variant confirmation, DNT from P1 were isolated through MACS separation, DNA was extracted and somatic variant in intron 7 of FAS gene was detected by Sanger sequencing ( Figure 1D).

Customized Somatic Variant Caller Validation
After detecting the somatic variant in P1 (c.651+2T>C) by NGS using DNA from whole blood, it was necessary to validate the somatic variant caller analyzing samples from seven other ALPS-sFAS patients previously diagnosed in the authors' laboratories by Sanger sequencing in isolated DNT (P2-P8) (13,24,25).
To ascertain the quality of the somatic variants identified by our method, forty-four samples from non-ALPS patients were also included (C1-C44). In total, nine different variants were called in the 52 samples analyzed ( Table 2). Strand bias, as the significantly different genotype inferred from the positive and negative strand, was around 0.5 in all cases, ensuring a good reads balance in both strands thus ruling out possible NGS errors. Somatic mutations of P1, P2, P3, P4, P5, P6 and P7 were correctly detected with a VAF of 4.2%, 7.6%, 11.5%, 2.7%, 7.9%, 8.7% and 1.9%, respectively (Table 2). P8, with three base-pair deletion (c.812_814del; p.Ala271del) (25) was not detected by the somatic variant caller plugin, although the deletion was observed   Table  2). All somatic mutations and variants calls were visually checked with the IGV (Figure 2). Based on the experimental data, the minimum depth of coverage of 101 reads allows to detect a VAF of 7.9% (case of P5). When the depth reaches 212 reads, we can detect a minimum VAF of 1.9% (case of P7). Variant annotation and pathologic interpretations for new variants were performed with Ion Reporter Software and VarSome, respectively (Table 3). Somatic mutations previously described in P2-P7 (13,24,25) and germline variants were excluded from variant annotation and in silico analyses. Somatic variant of P1 (c.651+2T>C) was predicted as pathogenic and was previously reported as disease causing in ALPS-FAS patients (10,28). The somatic variant detected in both P1 and P4 (c. 587A>G) was not found in Clinvar and was classified as benign (twenty-one benign versus no pathogenic scores), but it was not detected in Sanger sequencing from their isolated DNT (data not shown). The somatic variant found in C5 (c.675T>C) is probably a benign synonymous variant that not predicted an aberrant splicing (27,29) (Table 3). Although we cannot exclude that these two somatic variants are present in another cell population in the three cases (c. 587A>G in P1 and P4 and c.675T>C in C5), they have been considered as false positive variants. Data show that sensitivity of the assay is 87.5% and the specificity is 93.1%. These results are obtained without annotation and interpretation of the variants. If we consider the no

FAS-Mediated Apoptosis
Nowadays, FAS-mediated apoptosis is not considered essential for the diagnosis of ALPS (accessory criterion) (5), as some ALPS patients (ALPS-sFAS or ALPS-FASLG) reported normal values of FAS-mediated apoptosis (14). In the case of ALPS-sFAS patients a small percentage of PBMCs (DNT cells) carry the FAS mutation, while in the case of ALPS-FASLG, FAS-mediated apoptosis is overlooked, therefore it's necessary to demonstrate an impaired activation-induced cell death (AICD) (14). Apoptosis testing is an intensive, costly labor, available only in a few specialized centers and not really helpful in routine evaluation of patients. According to the proposed ALPS diagnostic algorithms, the presence of a pathogenic mutation in ALPS-related genes establishes a diagnosis of ALPS. Although in some circumstances, FAS-mediated apoptosis assay can be useful to clarify the pathogenicity of variants of uncertain significance (VUS). Interestingly, FAS-mediated apoptosis induced with anti-FAS  mAb in T-cell blasts was partially reduced in P1 (Figure 3) as previously described in other ALPS-sFAS patients (13).

DISCUSSION
ALPS is mostly caused by FAS mutations either germline or somatic, being restricted mainly to DNT in the latter case. Due to both ALPS-FAS and ALPS-sFAS patients cannot be clinically differentiated, sequencing of FAS gene in DNA from isolated DNT is recommended for ALPS-suspected patients without molecular diagnosis (30). However, the time, cost and technical requirements makes the DNT sorting a technique not accessible to all laboratories. NGS has been demonstrated to be a powerful tool that enables to detect very low allele frequency variants. Detecting somatic mutations by NGS is a critical part of cancer research and diagnostics (31)(32)(33). Sequencing depth is required since mutations need to be observed on a sufficient number of reads to pass predetermined variant calling threshold (21). Thus, targeted gene panel sequencing is the optimal option to allow very deep coverage in a cost-effective way. Increasingly, new approaches are being developed for the exploration of somatic disease-causing mutations, being more frequent than expected. High-throughput NGS is presented as an important tool for the discovery of new diseases caused by postzygotic mosaicism (34)(35)(36)(37)(38). Hence, there is a role for customized targeted NGS approach in some circumstances looking for somatic variants when germline variant calls remain noncontributory and might help correlate clinical and cellular phenotypes with genomic datasets.
In an attempt to facilitate the molecular diagnosis of ALPS-sFAS patients, we describe clinical, immunologic and molecular findings of a new ALPS-sFAS patient diagnosed through NGS targeted gene panel. The somatic c.651+2T>C mutation affecting the intracellular domain of the FAS gene was detected with a somatic variant caller plugin as well as six other ALPS-sFAS previously diagnosed (13,24). The mutation identified in this domain has the highest clinical penetrance (>80%) because it exerts dominant-negative effect on wild-type FAS protein (14). This strategy can be very useful allowing a first screening of possible somatic variants for their subsequent confirmation by Sanger sequencing in isolated DNT. This approximation could guide the next diagnostic steps and help assess whether DNT isolation is necessary, especially in the case of pediatric patients, as P1, where the limitation of the significant amount of blood needed, together with the technical requirements for the isolation of DNT, can lead in many cases to patient mismanagement (39).
As limitations of this study, we could consider that in the case of T-cell lymphopenia of the ALPS patients, the detection of the somatic FAS mutation would be more difficult in the whole blood (P1-P8 did not show T-lymphopenia). This fact must be considered in the case of ALPS patients under treatment when the diagnosis has been delayed for several circumstances. By the contrary, at diagnosis ALPS patients show normal/high numbers of T-lymphocytes.
Other limitation of the study is in case of ALPS patients with DNT percentages in the limit of the normal values ( Table 1). In the case of P8, an ALPS-sFAS patients with borderline DNT ( Table 1) and presenting a complete codon deletion (c.812_814del; p.Ala271del) (25), the mutation was not detected with the variant caller plugin used. However, the mutations of P1 and P4, also presented with borderline DNT (Table 1), were correctly detected. Probably a combination of the lowest DNT percentage and the nature of the P8 mutation are the main limitations for detecting the mutation by NGS.
Somatic mutation in FAS or in combination with germline mutation in genes other than FAS as XIAP, PRF1, UNC13D or CASP10 have been reported (11-13, 15, 40). The second mutation is not an ALPS-causative hit and the codified proteins are not involved directly in FAS-FASL pathway but have been proposed as ALPS-phenotype modifiers (11,12,15).
In this work, in addition to the somatic FAS mutation causing ALPS-sFAS detected in P1, a germline variant caller plugin revealed two heterozygous mutations in PRF1 and UNC13D genes predicted as possibly pathogenic, but not sufficient to cause hemophagocytic lymphohistiocytosis (41,42) due to the autosomal recessive form of the disease. Furthermore, the expression of perforin and degranulation in NK cells were preserved in the patient, and probably these variants at best contribute marginally to clinical phenotype.
In summary, we report the first ALPS patient with a somatic FAS mutation detected through NGS targeted gene panel. Identification of somatic FAS mutations with this novel strategy, even in the cases of borderline DNT, is a cost and time-effective option to expedite the diagnosis and treatment of these patients avoiding further complications due to a delayed diagnosis.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The name of the repository and accession numbers can be found below: National Center for Biotechnology Information (NCBI) GenBank, https://www.ncbi.nlm.nih.gov/ genbank/, MW674786, MW674787, MW674788 and MW674789.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by ComitéÉtico de Investigacioń Clıńica, University Hospital 12 de Octubre and by ethic committee of CPP Ile-de-France (CPP IDF2 DC-2014-22722015-03-03 AF). Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

AUTHOR CONTRIBUTIONS
ML-N designed the pipeline, analyzed the results and drafted the manuscript, tables, and figures. JD-C designed the bioinformatics algorithm. JR, RR-P, CG-L, SS-R, JGH, AM and FR-L conducted the clinical and immunological follow-up of the patient. MJD-M, MAD-M, PMP, EP-A, AM and FR-L did the molecular (NGS and Sanger sequencing) and functional studies (immunophenotyping, biomarkers and functional studies of apoptosis). LA conceived the idea for the manuscript and drafted the manuscript, tables, and figures. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by grants from Fondo de Investigación Sanitaria (FIS-PI16/2053) to LA. The project has been co-financed with FEDER funds. ML-N was co-financed by "Fondo Social Europeo, Programa Operativo de empleo juvenil (YEI)". The study was also supported by the Institut National de la Santé et de la Recherche Médicale (INSERM) and by government grants managed by the Agence National de la Recherche as part of the "Investment for the Future" program (ANR-10-IAHU-01 and ANR-18-RHUS-0010), the Ligue Contre le Cancer-Comité de Paris, Fondation ARC pour la recherche sur le CANCER, the Centre de Référence Déficits Immunitaires Héréditaires (CEREDIH), the Agence National de la Recherche (ANR-14-CE14-0026-01 "Lumugene"; ANR-18-CE17-0001 "Action").