Potential Role of STAG1 Mutations in Genetic Predisposition to Childhood Hemato-Oncological Diseases

Background. Genetic predisposition has been recently assessed in about 5-10% of pediatric tumors, including leukemias. Among genetic syndromes associated with an increased risk of developing leukemia, a possible link has been proposed for Cohesinopathies, like Cornelia de Lange Syndrome (CdLS), caused by mutations in Cohesin family genes. Somatic mutation in these genes, known for their fundamental role in cell cycle and DNA repair mechanisms, are described in myeloid malignancies (10-20% of AML, 50% of DS-AMKL, 5-15% of MDS and 10% of MPN) and solid tumours. We previously described the first germline variant of NIPBL in a CdL case with acute lymphoblastic leukemia (ALL). The identification of germline variants can provide new insights in the pathogenesis of hemato-oncological diseases. Objectives. This study aims to define the association of germline variants of the Cohesin family genes with predisposition to pediatric hemato-oncological diseases, including ALL. Functional assays have been performed on a STAG1 unique variant. Methods. We set up an NGS capture-base DNA panel, including 40 genes associated with 8 pathways associated to ALL predisposition, including Cohesin genes. Patients' bone marrow (BM) diagnostic samples have been analyzed together with a germline tissue (BM at remission or buccal brush). Bioinformatic analysis has been carried out by Sophia DDM software (Variant Fraction >5%; Coverage at least 500X; Variant Allele Frequency 1%) and variants were annotated as certainly pathogenic, potentially pathogenic or VUS. A Lymphoblastoid Cell Line (LCL) was obtained from PB of a MDS patient, then used for functional studies. Genomic stability and X-ray induced DNA-damage repair were evaluated in the STAG1-mutated LCL by Sister Chromatids Exchange (SCE) assay, cell cycle analysis (Propidium Iodide) and pH2AX staining. Results. Overall, 117 consecutive pediatric ALL patients and sporadic cases with familial recurrency of cancer or other hematological disorders have been analyzed through the NGS panel. A total of 225 mutations were identified in ALL, distributed as it follows: 77 variants in Transcription factors, 68 in Signaling-associated genes, 56 variants in RAS pathway, 44 in chromatin modifier genes, 38 variants in B-cell differentiation genes, 30 in DNA repair genes and 16 variants in genes associated to predisposition. Overall, 12/117 unique variants (5%) were found in Cohesins family genes; 11/12 were germline, while a single somatic variant was found in STAG2 in the same position as a germline variant. In details, 3 mutations were found in BRD2 gene while 2 variants were identified in STAG1, STAG2 and NIPBL genes, respectively. Only 1 variant was detected in SMC1A, SMC3 and BRD2. No mutations were detected in HDAC8 and RAD21 genes. Interestingly, the Arg1167Gln variant in STAG1 is located in a highly conserved region involved in solid tumors. A STAG1 variant (Arg1187Gln) was identified in a pediatric MDS patient, which was then further characterized in the LCL. First, we evaluated the effects on chromosomal stability, and we observed a higher number of abnormal chromatids exchanges in mutated LCL (STAG1 patient mean = 4.8 exchanges/metaphase compared to four control LCLs mean = 3.05; p <0.0001), as shown in the attached figure. Differences in cell cycle phases between STAG1-mutated LCL and controls, even after X-ray irradiation (3Gy - 6Gy), are not significant. Mutated cells showed a significantly lower capability to repair DNA after an ionizing radiation. γH2AX phosphorylation status of STAG1 mutated LCL appears higher also in basal condition (T0: 1.7X, p <0.01; T24: 2.2X, p <0.0001; T48: 2.4X, p <0.0001; MFI STAG1 over MFI control LCLs) and it remains at higher levels at 48 hours after the irradiation (T48: 3.7X, p<0.001 [3Gy]; 4.1X, p<0.0001 [6Gy]; MFI STAG1 over MFI control LCLs). These data are confirmed by the ratio of the STAG1-mutated LCL median MFI over controls (T48/T0: 3.42 vs 0.92, p <0.0001 [3Gy]; T48/T0: 2.33 vs 0.64, p <0.01 [6Gy]). Conclusion. In pediatric hematological diseases, NGS screening showed several variants among Cohesin genes, with either a potentially pathogenic or unknown significance. Our study confirms a considerable effect of a STAG1 germline variant on genetic instability that leads to oncogenesis, opening new scenarios for Cohesins' contribution to genetic predisposition to leukemias. Figure 1 Figure 1. Disclosures Biondi: Colmmune: Honoraria; Incyte: Consultancy, Other: Advisory Board; Amgen: Honoraria; Bluebird: Other: Advisory Board; Novartis: Honoraria.

Recurrent Germline Variant in the Cohesin Complex Gene RAD21 Predisposes Children to Lymphoblastic Leukemia and Lymphoma

Introduction: Cohesin complex genes are commonly mutated in cancer particularly in myeloid malignancies. Yet patients with germline mutations in cohesin genes, leading to cohesinopathies like Cornelia-de-Lange syndrome (CdLS) are generally not known to be tumor-prone. The complex plays a major role in chromosome alignment and segregation (Uhlmann, Nature Reviews Molecular Cell Biology, 2016), homologous recombination-driven DNA repair (Ström et al., Molecular Cell, 2004) and regulation of gene expression (Busslinger et al., Nature, 2017). To deepen the understanding of cohesin variants in cancer predisposition, we performed TRIO Sequencing in two independent pediatric cancer cohorts. Thereby, we identified a novel recurrent heterozygous germline variant in the cohesin gene RAD21 not described in CdLS patients , located in the binding domain of the cofactors WAPL and PDS5B . Methods: Whole exome sequencing (WES) in a TRIO (child-parent datasets) setting was carried out in two independent, unselected cancer cohorts (TRIO-D, n=158 (Wagener et al., European Journal of Human Genetics, 2021) and TRIO-DD, n=60). To investigate the oncogenic potential of the novel RAD21 variant molecular and functional assessment was performed focusing on potential implications on the complex. Results: The newly identified RAD21 variant at amino acid position 298 resulting in a Proline to Serine (p.P298S) and a Proline to Alanine exchange, respectively, (p.P298A) is only rarely mutated in the general population (gnomAD database n=118,479; RAD21 p.P298S MAF <10 -6 and RAD21 p.P298A MAF <10 -5). While both patients did not show any signs of CdLS, they both have a remarkable family history of cancer. Patient 1 (13y) was diagnosed with T-cell acute lymphoblastic leukemia (T-ALL) whose father had died from breast cancer (41y), while patient 2 (2y) presented with precursor B-cell lymphoblastic lymphoma (pB-LBL) whose uncle had died from pediatric cancer of unknown subtype (8y). To assess the influence of RAD21 p.P298S/A on the binding capacity of the complex, RAD21 variants and the wildtype (WT) were cloned and transfected into HEK293T cells, respectively. Immunoprecipitation analysis of RAD21 with the cofactors WAPL and PDS5B showed no differential binding between the WT and the variants, suggesting that RAD21 p.P298S/A does not impact the formation of the complex. Nevertheless, on a transcriptional level 83 genes were significantly differentially expressed in RAD21 p.P298S and p.P298A compared to the wildtype (fc>1.5, adj. p-value <0.05) with enrichment of genes in p53 signaling pathways. We further observed an increased number of γH2AX and 53BP1 co-localized foci compared to the WT (p≤0.01; Student's t-test). In line, following ionizing radiation, primary patients' samples showed increased cell cycle arrest at G2/M cell-cycle stage compared to a healthy control (p.P298S: p=0.0049 [6Gy]; p=0.0026 [10Gy]; p.P298A: p=0.0054 [6Gy]; p=0.0006 [10Gy]; Student's t-test). For cross-validation of the germline variant RAD21 p.P298S/A and its potential role in pediatric lymphoblastic malignancies, we analysed a third cohort of 150 children with relapsed ALL (IntReALL) for RAD21 p.P298S/A. We again identified RAD21 p.P298A in a boy (12y) with B-cell precursor acute lymphoblastic leukemia. To compare our data to a non-pediatric cancer setting, a cohort of 2300 young adults (<51 years) with cancer was mined (MASTER program). Here, one patient carrying RAD21 p.P298A with a solid tumor was identified. Therefore, amongst all cohorts, RAD21 p.P298S/A was found to be enriched in pediatric vs. adult cancers (3/479 vs. 1/2299; Fisher's exact test; p=0.018). Conclusion: Taken together, we present for the first time the potential role of RAD21 germline variants in pediatric lymphoblastic malignancies. This may shed new light on the many roles of the cohesin complex and its implication outside the typical syndromal presentation. Disclosures No relevant conflicts of interest to declare.

Investigating Germline Predisposition to Clonal Hematopoiesis through Perturbation of a Variant-Harboring Enhancer of TET2

Clonal hematopoiesis (CH) is the age-related expansion of hematopoietic stem and progenitor cells (HSPCs) due to acquired genetic changes and is associated with increased blood cancer risk. Despite considerable progress in understanding how specific acquired somatic mutations result in clonal expansion, we have a limited understanding of the role of germline mutations that also predispose to clonal expansion. Recent work has revealed a low frequency germline variant found exclusively in individuals of African diasporic descent that confers a 2.4-fold increased risk for CH (Bick et al., Nature, 2020). Remarkably, this variant is found within a putative enhancer of the TET2 gene, which encodes a key epigenetic modifier that catalyzes conversion of methylated cytosines to 5-hydroxymethylcytosine, thereby facilitating DNA demethylation. However, the precise role of this enhancer variant in altering TET2 activity and human hematopoiesis is poorly understood. We specifically hypothesize that this germline variant may alter TET2 activity subtly in hematopoietic stem cells (HSCs) to modify DNA methylation at a number of HSC regulatory elements and subsequently gene expression, which are likely mediated through selective changes in transcription factor (TF) activity. To explore this hypothesis, we performed deletions of the putative TET2 enhancer in adult CD34 +HSPCs using CRISPR/Cas9 editing through the introduction of Cas9 ribonucleoproteins. Following deletion of this enhancer, we observed a moderate increase in the total number of phenotypic long-term reconstituting hematopoietic stem cells (LT-HSCs; as marked by CD34 +CD45RA -CD90 +CD201 +CD133 +CD49c +) and in primary colony formation unit (CFU) assays compared to control editing (AAVS1 edited). Interestingly, the enhancer deletion did not cause an increase in the number of phenotypic short-term HSCs (CD34 +CD45RA -CD90 +CD201 +CD133 +CD49c -), HSPC proliferation, or secondary CFU plating. Assessment of the deletion stability showed selection against enhancer deleted cells during myeloid differentiation, however some cells could still be identified after more than 30 days following editing. This result suggests that the enhancer likely functions in a selective manner within HSCs. The overall phenotypes we observe suggest some overlap, but distinct presentations in comparison to concomitant TET2 coding disruption that we have also performed. Ongoing studies seek to use these promising phenotypic results to define changes in accessible chromatin and DNA methylation states in this isogenic perturbed model. This will enable insights into the specific enhancers altered through this perturbation and we plan to examine TF motif alterations at these regulatory elements. While the perturbation performed may represent a larger perturbation than is seen through the naturally-occurring variant, this provides a platform for defining how this CH predisposition arises with a larger impact perturbation. Use of base editors in a similar manner should enable more selective perturbations, as well. Together, these results will further reveal potential germline genetic and molecular origins of CH and further explain broader mechanisms of TET2 function and the importance of proper DNA methylation during human hematopoiesis that may provide clinical relevance for the potential prevention of blood cancers. Disclosures Sankaran: Cellarity: Consultancy; Forma: Consultancy; Novartis: Consultancy; Ensoma: Consultancy; Branch Biosciences: Consultancy.

Whole Genome Sequencing Uncovers a Rare Germline Variant in ATM Associated with Familial Myeloproliferative Neoplasms

Introduction: Myeloproliferative neoplasms (MPN) are sporadic diseases characterized by a somatic driver mutation in the JAK2, CALR or MPL gene. Although it is generally considered a sporadic disease, approximately 10% of MPN cases display familial clustering, and there is a 5 to 7-fold increased risk of developing an MPN among first degree relatives of MPN patients. In contrast to other myeloid malignancies, investigation of large pedigrees with familial clustering of MPN has failed to identify high-risk predisposition genes relevant to the general MPN population. Genome wide association studies (GWAS) have identified common, low penetrance risk alleles for MPN predisposition in multiple genes including JAK2, TERT, TET2, ATM and SH2B3. In order to identify novel germline predisposition variants in MPN, an unbiased whole genome sequencing (WGS) approach was utilized to examine genomic structure and germline variations in a cohort of individuals with familial MPN. Methods: The study cohort was comprised of 67 individuals with familial MPN enrolled in a prospective research registry at Johns Hopkins Hospital. Familial MPN was defined as a diagnosis of MPN in an individual with a family history of MPN or related myeloid malignancy (myelodysplastic syndrome and chronic myelomonocytic leukemia) in a first or second degree relative. Neutrophil genomic DNA was subjected to WGS using Illumina HiSeq platform and sequenced to 60x depth. We performed germline variant calling using HaplotypeCaller and following the GATK best practices. The variants detected were further enriched for germline by allele frequency 40-60% or >90% and presence in the gnomAD database. Non-synonymous coding variants that occurred in the study cohort at a statistically higher frequency that in the general population (gnomAD) were selected for further analysis. Prediction of variant deleteriousness was assessed by 4 algorithms (Provean, SIFT, Polyphen-2, CADD). Results: Filtering of 32,788 non-synonymous, likely germline variants produced 148 that occurred at a higher frequency in our cohort than in the general population (p < 0.01, Fisher's exact test). Of these, 29 were predicted to be pathogenic in 3 out of 4 algorithms. Five unrelated individuals were found to harbor a heterozygous p.Leu2307Phe variant in the ATM gene (chr11:108326169 C>T). The clinical characteristics of these individuals are presented in Table 1. The structure prediction of ATM indicates that Ser2306 is a potential phosphorylation site of protein kinase A, suggesting that the methionine-aromatic bond between M2026 and the mutated F2307 may block the phosphorylation of S2306 (Figure 1). Conclusions: We identified a rare ATM germline variant (chr11:108326169 C>T; p.Leu2307Phe) present in 5 individuals with familial MPN. ATM is involved in DNA damage repair and important in the maintenance of genomic integrity. Heterozygous germline variants in ATM are known to predispose to multiple cancer types, including breast, prostate, pancreatic and melanoma. Further, common polymorphisms in ATM have been found to be associated with the MPN phenotype via GWAS. Our data suggests that this variant may impact ATM activation and its function in DNA damage repair, and functional studies are in progress. These data implicate a rare germline ATM variant as a novel risk factor for development of MPN. Figure 1 Figure 1. Disclosures Hourigan: Sellas: Research Funding.

ClinGen CDH1 specifications for the ACMG/AMP guidelines: improvement of germline variant clinical assertions and updated curation guidelines

Purpose: The Clinical Genome Resource (ClinGen) CDH1 Variant Curation Expert Panel (VCEP) developed specifications for CDH1 variant curation with a goal to resolve variants of uncertain significance (VUS) and with ClinVar conflicting interpretations for effective medical care. In addition, the CDH1 VCEP continues to update these specifications in keeping with evolving clinical practice and variant interpretation guidelines. Methods: CDH1 variant classification specifications were modified based on updated genetic testing clinical criteria, new recommendations from ClinGen, and expert knowledge from ongoing CDH1 variant curations. Trained biocurators curated 273 variants using updated CDH1 interpretation guidelines and incorporated published and unpublished data provided by diagnostic laboratories. All variants were reviewed by the ClinGen VCEP and classifications submitted to ClinVar. Results: Updated CDH1-specific variant interpretation guidelines include eleven major modifications since the initial specifications from 2018. Using the refined guidelines, 97% (36/37) of variants with ClinVar conflicting interpretations were resolved into benign, likely benign, likely pathogenic, or pathogenic, and 35% (15/43) of VUS were resolved into benign or likely benign. Overall, 88% (239/273) of curated variants had non-VUS classifications. Conclusion: The development and evolution of CDH1-specific criteria by the expert panel results in decreased uncertain and conflicting interpretations of variants in this clinically actionable gene.

Performance Comparison of Computational Prediction Methods for the Function and Pathogenicity of Non-coding Variants

Non-coding variants in the human genome greatly influence some traits and complex diseases by their own regulation and modification effects. Hence, an increasing number of computational methods are developed to predict the effects of variants in the human non-coding sequences. However, it is difficult for users with insufficient knowledge about the performances of computational methods to select appropriate computational methods from dozens of methods. In order to solve this problem, we assessed 12 performance measures of 24 methods on four independent non-coding variant benchmark datasets: (Ⅰ) rare germline variant from ClinVar, (Ⅱ) rare somatic variant from COSMIC, (Ⅲ) common regulatory variant dataset, and (Ⅳ) disease associated common variant dataset. All 24 tested methods performed differently under various conditions, indicating that these methods have varying strengths and weaknesses under different scenarios. Importantly, the performance of existing methods was acceptable in the rare germline variant from ClinVar with area under curves (AUCs) of 0.4481 - 0.8033 and poor in the rare somatic variant from COSMIC (AUCs: 0.4984 - 0.7131), common regulatory variant dataset (AUCs: 0.4837 - 0.6472), and disease associated common variant dataset (AUCs: 0.4766 -0.5188). We also compared the prediction performance among 24 methods for non-coding de novo mutations in autism spectrum disorder and found that the CADD and CDTS methods showed better performance. Summarily, we assessed the performances of 24 computational methods under diverse scenarios, providing preliminary advice for proper tool selection and new method development in interpreting non-coding variants.