ALBERT

Description: There is growing evidence supporting an inherited basis for susceptibility to acute lymphoblastic leukemia (ALL) in children. In particular, we and others reported recurrent germline ETV6 variants linked to ALL risk, which collectively represent a novel leukemia predisposition syndrome. To understand the influence of ETV6 variation on ALL pathogenesis, we comprehensively characterized a cohort of 32 childhood leukemia cases arising from this rare syndrome. Of 34 nonsynonymous germline ETV6 variants in ALL, we identified 22 variants with impaired transcription repressor activity, loss of DNA binding, and altered nuclear localization. Missense variants retained dimerization with WT ETV6 with potentially dominate negative effects. Whole transcriptome and whole genome sequencing of this cohort of leukemia cases revealed a profound influence of germline ETV6 variants on leukemia transcriptional landscape, with distinct ALL subsets invoking unique patterns of somatic cooperating mutations. 70% of ALL cases with damaging germline ETV6 variants exhibited hyperdiploid karyotype with characteristic recurrent mutations in NRAS, KRAS, and PTPN11. In contrast, the remaining 30% cases had a diploid leukemia genome and an exceedingly high frequency of somatic copy number loss of PAX5 and ETV6, with a gene expression pattern that strikingly mirrored that of ALL with somatic ETV6-RUNX1 fusion. Two ETV6 germline variants gave rise to both AML and ALL, with lineage-specific genetic lesions in the leukemia genomes. ETV6 variants compromise its tumor suppressor activity in vitro with specific molecular targets identified by ATAC-seq profiling. ETV6-mediated ALL predisposition exemplifies the intricate interactions between inherited and acquired genomic variations in leukemia pathogenesis.

Description: Acute lymphoblastic leukemia (ALL) in children is a prototype of cancer that can be cured by chemotherapy alone. However, the molecular mechanisms for anti-leukemic drug sensitivity and genetic basis of inter-patient variability in treatment response are not fully understood. Taking a genome-wide approach, we recently identified genetic variants in the ARID5B gene that strongly predispose children to developing ALL and also a high risk of relapse following therapy (J Clin Oncol 2012 30:751, Nat Genet 2009 41:1001). To understand the mechanisms by which ARID5B is linked to treatment outcome in childhood ALL, we sought to 1) characterize ARID5B expression in different genetic subtypes of ALL, 2) determine the effects of ARID5B expression on cytotoxicity of chemotherapeutic agents commonly used in ALL therapy, and 3) describe molecular pathways linking ARID5B to anti-leukemic drug sensitivity. In 567 children with newly diagnosed ALL treated at St. Jude Children’s Research Hospital (GSE33315), ARID5B expression was highest in cases with hyperdiploid karyotype (〉50 chromosomes) and lowest in T-cell ALL and cases with MLL rearrangements. This pattern was validated in an independent cohort of 106 children from the Dutch Childhood Oncology Group (GSE13351). In 59 patients treated on the Children’s Oncology Group (COG) CCG1961 trial, lower ARID5B expression was associated with higher rates of relapse (P=0.01, GSE7440). Importantly, when we compared matched newly-diagnosed vs. relapsed ALL blasts from a cohort of 60 patients enrolled in COG trials (GSE28460), ARID5B expression was further downregulated at disease recurrence (P=0.0009). shRNA-mediated ARID5B knockdown in 3 ALL cell lines (Nalm6, SEM, and UOCB-1) substantially increased resistance to antimetabolites (an average of 5.16 and 35.3-fold increase in IC50 for methotrexate [MTX] and 6-mercaptopurine [6MP], respectively), with minimal effects on glucocorticoids, vincristine, asparaginase, and daunorubicin. Because cytotoxic effects of MTX and 6MP are highly dependent on the rate of cell proliferation, we postulate that ARID5B directly influences cell cycle entry. In all 3 cell lines, ARID5B knockdown led to significant blockade at the G1/S checkpoint, increasing the percent of cells in G0/G1 phase. At the molecular level, downregulation of ARID5B resulted in higher levels of p21 and reduction in phosphorylated Rb, consistent with the retention at G0/G1 phase. ARID5B expression was restricted to nucleus but affected both nuclear and cytoplasmic p21 expression in a time-dependent fashion. Interestingly, there was a highly significant negative correlation between p21 expression and MTX- and 6MP-induced apoptosis in all 3 ALL cell lines. Taken together, we hypothesize that lower expression of ARID5B impairs ALL cell cycling by upregulating p21, contributing to resistance to MTX and 6MP and eventually leukemia relapse. Finally, we compared global gene expression in ARID5B knockdown vs. control ALL cells, and via the Connectivity Map analysis we identified histone deacetylase (HDAC) inhibitors as promising agents for overcoming ARID5B-related drug resistance. Indeed, ARID5B knockdown cells were significantly more sensitive to panobinostat than controls, suggesting HDAC inhibitors as potential therapeutic options for patients with ARID5B-deficient and drug resistant ALL. Disclosures No relevant conflicts of interest to declare.

Description: Key Points AGGF1-PDGFRB is a novel oncogenic fusion gene in Ph-like ALL. Genomic profiling and functional studies identified a novel PDGFRB mutation directly related to TKI resistance.

Description: Background There is increasing evidence for an inherited predisposition to pediatric acute lymphoblastic leukemia (ALL). We and others have previously reported rare and highly penetrant variants in hematopoietic transcription factors (PAX5 and ETV6) and tumor suppressor genes (TP53) in both sporadic and familial ALL. IKZF1encodes the founding member of the Ikaros family of zinc finger transcription factors, and is a critical regulator of lymphoid development. IKZF1 is frequently targeted by somatic deletions and mutations in high-risk B-ALL, particularly Ph+ and Ph-like ALL, and is associated with poor outcome. IKZF1 alterations have previously been shown to result in the acquisition of stem cell-like features, overexpression of adhesion molecules causing aberrant cell-cell and cell-stroma interaction, and decreased sensitivity to tyrosine kinase inhibitors. Genome-wide association studies have also identified an association between common polymorphisms at the IKZF1locus and risk of developing ALL, however the nature and effects of germline IKZF1variation in the pathogenesis of ALL are poorly understood. In this study, we sought to comprehensively characterize germline IKZF1 genetic variation and to determine the extent to which they contribute to predisposition to ALL. Methods We recently identified a germline frameshift IKZF1 variant (D186fs) in the proband of a family with BCR-ABL1 ALL with incompletely penetrant autosomal dominant inheritance, and carriers of this variant showed varying degree of B cell deficiency. We sequenced IKZF1in germline DNA from 5,008 children with ALL (4902 B-ALL and 106 T-ALL) enrolled on ChildrenÕs Oncology Group and St. Jude ChildrenÕs Res. Hosp. frontline ALL trials. We examined each variant for its effects on transcriptional repression, DNA-binding, cellular localization, homodimerization, and leukemic cell adhesion in mouse BCR-ABL1Arfnull B-ALL cells and/or in HEK 293T cells. All variants were assayed for their effects on cell viability and proliferation, cell-cell adhesion, and IKZF1 protein expression and localization in BCR-ABL1 Arfnull pre-B cells. Representative variants, including M31V (N-term), H163Y (DNA-binding domain), D186Tfs (familial index), M306* (truncation of C-terminus), and A434G (C-terminus) were also assayed in detail for their ability to dimerize with wild type IKZF1, bind to DNA, or dominant negative effects on transcription repressor activity in HEK293T cells. IKZF1 variants were also evaluated for inducing perturbations in cell adhesion and THY1, ITGA5, SELL expression in the mouse PreB cells, and adhesion within the bone marrow niche by ex vivo imaging of calvaria. Finally, the effects of variants on dasatinib sensitivity were assessedin vitro and in vivo. Results We identified 28 germline IKZF1variants in children with ALL, mostly in B-ALL (Figure 1). Among these variants, 3 were frameshift or nonsense resulting in truncated IKZF1 proteins. Of the remaining missense variants, 2 were located within the N-terminal DNA-binding domain, 1 in the C-terminal dimerization domain, and 22 in other parts of IKZF1 protein with clustering proximal to the C-terminal zinc fingers. In mouse BCR-ABL1 Arfnull pre-B cells, all but 4 variants (P18T, P420Q, H432Q, and M518K) variably perturbed IKZF1 function. In contrast to expression of wild-type IKZF1, which caused growth arrest, 24 of the ALL variants were tolerated; 18 caused cellular aggregation; 15 displayed cytoplasmic mislocalization; and 14 out of 20 variants analyzed had significant upregulation of the adhesion molecules THY1, ITGA5 and/or SELL that are normally repressed by IKZF1. In HEK293T cells, 3 IKZF1 truncating variants showed dramatic loss of transcription repressor activity and no longer dimerized with wildtype IKZF1. DNA-binding domain variants (R162P, H163Y) failed to repress target promoter transcription but also altered wildtype IKZF1 function in a dominant negative fashion. In comprehensively characterization of representative variants (M31V, H163Y, D186Tfs, M306X, and A434G), these variations caused cell-stroma adherence in the bone marrow niche in vivo, and significantly reduced sensitivity of leukemic cells to dasatinib in vitro and in vivo. Conclusions These results identify IKZF1 as a new ALL predisposition gene, and suggest that these germline risk variants have roles in both leukemia pathogenesis and treatment responsiveness. Disclosures No relevant conflicts of interest to declare.

Description: While acute lymphoblastic leukemia (ALL) is a prototype of cancer that can be cured by chemotherapy alone, current ALL treatment regimens rely primarily on conventional cytotoxic agents with significant acute and long-term side effects. Better understanding of genomic landscape of ALL is critical for developing molecularly targeted therapy and implementing genomics-based precision medicine in this cancer. In particularly, sentinel chromosomal translocations are common in ALL and often involve key transcription factors important for hematopoiesis. Epigenetic regulator genes are also frequently targeted by somatic genomic alterations such as sequence mutations (e.g., CREBBP) and gene fusions (e.g., MLL, EP300). To comprehensively define transcriptomic abnormalities in childhood ALL, we performed RNA-seq of an unselected cohort of 231 children enrolled on the MaSpore frontline ALL protocols in Singapore or Malaysia. In total, we identified 58 putatively functional and predominant fusion genes in 125 patients (54.1%), the majority of which have not been reported previously. In particular, we described a distinct ALL subtype with a characteristic gene expression signature driven by chromosomal rearrangements of the ZNF384 gene with different partners (i.e., histone acetyl-transferases EP300 and CREBBP, TAF15, and TCF3). In 9 of 11 ALL cases with ZNF384 rearrangements, the breakpoint in this gene was invariably between exon 2 and exon 3, resulting in deletion of the 5'-UTR and then in-frame fusion of the entire ZNF384 coding sequence with the partner genes. The top two most significantly up-regulated genes in the ZNF384-rearranged group were CLCF1 and BTLA, whose expression levels were 15.5- and 15.0-fold higher than in ALL cases with wildtype ZNF384, respectively. In fact, ZNF384 binding was identified within the CLCF1 and BTLA loci (particularly the promoter regions) by chromatin immunoprecipitation sequencing in B lymphoblasoid cells. Using luciferase transcription driven by CLCF1 promoter in HEK293T cells as a model system, we observed significantly greater transcription activity with EP300-ZNF384 fusion compared to cells expressing wildtype ZNF384, suggesting that this chimeric gene resulted in gain of ZNF384 function. Similar results were obtained with luciferase transcription assay driven by the BTLA promoter. In human ALL cells, CLCF1 and BTLA promoter activities were consistently and significantly higher in ZNF384-rearranged ALL than in ALL cell line with wildtype ZNF384. To examine the effects of ZNF384 fusion on hematopoietic stem and progenitor cell (HSPCs) function, we also evaluated colony forming potential of HSPC in vitro upon ectopic expression of ZNF384 fusions. While there was marked suppression of colonies from myeloid and erythoid lineages, expression of EP300-ZNF384 or CREBBP-ZNF384 significantly stimulated preB cell colony formation. However, neither EP300- nor CREBBP-ZNF384 fusion was able to transform mouse hematopoietic precursor cell Ba/f3 in vitro, but instead increased the transforming potential of other oncogenic mutations (NRASG12D). EP300-ZNF384 and CREBBP-ZNF384 fusion proteins lacked the histone acetyltransferase (HAT) domain, and showed only 25% and 10% of HAT activity of full-length EP300 and CREBBP, respectively, with dominant-negative effects. Also, expression of EP300-ZNF384 led to significant decrease in global H3 acetylation in Ba/f3 cells in vitro. Finally, in NRASG12D-transformed Ba/f3 cells, co-expression of EP300-ZNF384 or CREBBP-ZNF384 substantially potentiated cytotoxic effects of histone deacetylase inhibitor vorinostat. Similarly, in a panel of human ALL cell lines, ZNF384-rearrangement was also associated with increased sensitivity to vorinostat, suggesting that some ZNF384-rearranged ALL may benefit from therapeutic agents targeting histone acetylation regulation. In conclusion, our results indicate that gene fusion is the major class of genomic abnormalities in childhood ALL and chromosomal rearrangements involving EP300 and CREBBP may cause global epigenetic deregulation in ALL with potentials for therapeutic targeting. Disclosures No relevant conflicts of interest to declare.

Description: Previous studies by us and others have linked germline genetic variants to the familial predisposition to childhood B-cell acute lymphoblastic leukemia (B-ALL), with pathogenic variants discovered in TP53, PAX5, ETV6, and IKZF1 (J. Clin. Oncol 2018, Nature Genet 2014, Lancet Oncol 2015, Cancer Cell 2018). However, genetic predisposition to T-ALL is much less understood. Rare care reports of T-ALL pedigrees with germline RUNX1 point to its potential role in ALL susceptibility. RUNX1 plays significant roles in definitive hematopoiesis and primarily functions as a transcription factor. RUNX1 germline variants are associated with familial platelet disorder, with a significant proportion of patients also developing myeloid malignancies. To comprehensively examine the pattern and prevalence of RUNX1 germline variation in T-ALL, we performed targeted germline sequencing of 1,231 cases enrolled on the Children's Oncology Group AALL0434 trial. In this largely unbiased T-ALL cohort, we identified 13 germline RUNX1 variants in 16 cases (Figure 1), including six missense (46.2%), two nonsense (15.4%), three frameshift (23.1%), and two indel variants (15.4%). These variants are divided into three groups: Group I, truncating both the DNA-binding RHD domain and the transcriptional activation AD domain (p.K117* and p.S141fs); Group II, truncating the AD domain only (p.Q213fs, p.R232fs, and p.Y287*); and Group III, missense and indel variants. To comprehensively characterize the function of these T-ALL-related RUNX1 variants, we performed a variety of biochemical and cellular assays in different model systems. Using reporter gene assays, we first directly evaluated the transcriptional activity of RUNX1 variants in Hela cells and identified both loss-of-function (e.g., Group I variants) and dominant-negative effects (e.g., p.G365R in Group III variants). Group I variants also showed dramatic subcellular mislocalization in the cytoplasm, with concomitant loss of CBFβ binding, both of which were significantly subtler for Groups II and III variants. Focusing on representative variants in these three groups (p.S141fs, p.R232fs, Y287*, and p.G365R), we next examined their effects on hematopoietic phenotypes in vitro. Ectopic expression of Group II and III variants in human CD34+ cells significantly increased CFU-M/GM colony formation and long-term proliferation, while repressing BFU-E colonies. Variant RUNX1 cells also showed defects in megakaryocyte and pre-T cell differentiation, with decreased apoptosis compared to cells expressing wild-type RUNX1. Expression of Group I variant led to phenotypes similar to that of empty vector, suggesting a complete loss of RUNX1 function. In parallel, we engineered isogenic T-ALL single clones with epitope-tagged RUNX1 variant introduced at the endogenous locus via CRISPR-Cas9 mediated homology recombination. Chromatin immunoprecipitation (ChIP)-seq profiling of these cells suggested a varying degree of changes in RUNX1 binding sites across the genome as a result of the RUNX1 genetic variation. On the other hand, RNA-seq profiling identified down-regulation of genes that were activated by wild-type RUNX1, again confirming the loss-of-function effects of these variants. Finally, we performed whole-genome seq of matched leukemia and germline samples and RNA-seq of leukemia cells in 7 T-ALL cases with RUNX1 predisposition variants. In this analysis, we observed a significant enrichment of JAK3 mutations (5 of 7 cases, 71.4%) compared to a cohort of 264 T-ALL with wild-type RUNX1 in the germline (P=3.39×10-7). By comparison, only 27.3% (3 of 7) of T-ALL with a somatic mutation in RUNX1 had concurrent JAK3 mutations in this cohort. Unsupervised clustering based on RNA-seq derived gene expression profile showed that RUNX1-mutated cases, either germline or somatic, clustered tightly with early T precursor (ETP) and near-ETP immunophenotypes. In conclusion, we comprehensively characterized 13 RUNX1 germline variants in T-ALL, ~40% of which are frameshift or nonsense. These variants result in a loss of function, by disrupting DNA binding or deleting the transcriptional activation domain, and in some cases in a dominant-negative fashion. RUNX1 genetic variation also results in significant defects in hematopoietic cell differentiation and functions in vitro, but additional somatic lesions are most likely required for overt leukemogenesis. Disclosures Gastier Foster: Incyte Corporation: Other: Commercial Research; Bristol Myers Squibb (BMS): Other: Commercial Research. Raetz:Pfizer: Research Funding. Zweidler-McKay:ImmunoGen: Employment. Mullighan:Illumina: Honoraria, Membership on an entity's Board of Directors or advisory committees, Other: sponsored travel; Pfizer: Honoraria, Other: speaker, sponsored travel, Research Funding; AbbVie: Research Funding; Loxo Oncology: Research Funding; Amgen: Honoraria, Other: speaker, sponsored travel. Hunger:Amgen: Consultancy, Equity Ownership; Bristol Myers Squibb: Consultancy; Novartis: Consultancy; Jazz: Honoraria. Relling:Servier Pharmaceuticals: Research Funding. Loh:Medisix Therapeutics, Inc.: Membership on an entity's Board of Directors or advisory committees.

Description: There is increasing evidence for an inherited genetic basis of susceptibility to acute lymphoblastic leukaemia (ALL) in children. In particular, genome-wide association studies (GWAS) have identified nine genomic regions with common polymorphisms associated with ALL risk. However, these loci cumulatively only explain a minority of the genetic risk of childhood ALL, and differences in the etiology amongst molelcular subtypes of ALL remain poorly understood. In this study, we performed one of the largest ALL susceptibility GWAS, totalling 5,321 cases and 16,666 controls of European descent, to comprehensively map ALL risk loci. With this large sample size, we also sought to identify risk genes for two major subtypes of ALL, namely those with ETV6-RUNX1 fusion and hyperdiploid ALL. This meta-analysis included four previously published GWAS datasets with European cases from the UK ALL cohorts and the German Berlin-Frankfurt-Münster ALL trials, and the US cases enrolled on ALL trials in Children's Oncology Group and at St. Jude Children's Research Hospital (Nat Genet2009, Blood 2012, Nat Genet 2013, Blood 2013, Nat Commun 2018). The replication series included 2,237 ALL cases of non-European ancestry from the COG and St. Jude GWAS cohorts. For each GWAS dataset, the genotypes of ~10 million SNPs in each study were imputed. After filtering out SNPs on the basis of minor allele frequency and imputation quality, we assessed associations between ALL status and SNP genotype in each study using logistic regression. Risk estimates were combined through an inverse variance weighted fixed-effects meta-analysis. The meta-analysis identified 16 risk loci above genome-wide significance (P 〈 5 x 10-8), of which ten confirmed previously published associations. Of the six new candidate risk loci, four were validated in the replication series (P 〈 0.05): for all B-ALL at 9q21.3 (nearest gene TLE1), for hyperdiploid ALL at 5q31.1 (C5orf56) and 6p21.31 (BAK1), and for ETV6-RUNX1 positive ALL at 17q21.32 (IGF2BP1). As well as providing further evidence for the 21q22.3 (ERG) association for all B-ALL, we also identified novel risk variants with the hyperdiploid subtype-specific association. To gain insight into the biological basis of association signals, we examined the epigenetic landscape of risk regions in B-cells. Of particular note, hyperdiploid ALL-specific risk variant at 6p21 is located within an intron 1kb downstream of the BAK1 transcription start site and possess histone marks characteristic of active promoter activity and open chromatin accessibility. The top SNP at this locus falls within a transcription factor binding cluster, and the C-risk allele is associated with reduced BAK1 expression (P

Description: Acute lymphoblastic leukemia (ALL) is the most common cancer in children, and the etiology of this aggressive cancer is not fully understood. Common germline polymorphisms in lymphoid development genes and tumor suppressor genes have been associated with ALL susceptibility, although most have modest effects. Only a small fraction of ALL cases are thought to be related to congenital genetic disorders and consequently hereditary predisposition is rarely considered in clinical practice. However, a growing number of rare germline genetic mutations have been discovered in familial ALL (e.g., PAX5, TP53), raising the possibility that the proportion of ALL attributable to inherited predisposition may be higher than currently proposed. In particular, germline ETV6 variations were recently reported in families with hereditary thrombocytopenia and dramatically increased susceptibility to hematologic malignancies (Nat Genet 2015 47: 180 and 535). ETV6 is a transcriptional repressor essential for hematopoiesis and is frequently targeted by somatic genomic aberrations in childhood ALL (e.g., the ETV6-RUNX1 fusion). Therefore, we sought to comprehensively identify ALL predisposition variants in ETV6 and to determine the extent to which these variants contribute to childhood ALL risk in general. We first identified a family with three cases of childhood ALL at St. Jude Children's Research Hospital. Whole exome sequencing of this family (mother and 2 daughters with ALL, the unaffected father and 1 unaffected daughter) identified a single variant in ETV6 (p.R359X) in the 3 cases with ALL and also in the healthy daughter. This nonsense variant is predicted to create a stop codon within the ETS domain of ETV6, resulting in a truncated protein without DNA-binding function. This highly damaging variant is likely to be responsible for the ALL predisposition in this family with a high albeit incomplete penetrance. To comprehensively determine the prevalence of ALL-predisposing alleles in ETV6, we performed targeted sequencing of this gene in 4,405 children with newly-diagnosed ALL enrolled on the Children's Oncology Group (COG) AALL0232, P9904, P9905 and P9906 protocols and St. Jude Total Therapy XIIIA, XIIIB and XV studies. We identified a total of 43 germline variants in the exonic regions of ETV6. Thirty-one of the 43 ETV6 variants were defined as "ALL-related" because they were not found or extremely rare in non-ALL populations (N=60,706). These ALL risk variants included 4 nonsense, 21 missense, 1 splice site, and 5 frameshift variants occurring in 35 children (0.79% of ALL cases studied). Fifteen of the 31 ALL-relatedvariants (48.4%) were clustered in the ETS DNA-binding domain of ETV6. We used the combined annotation dependent depletion algorithm (CADD) to predict deleterious effects of each variant. ALL-related ETV6 variants were significantly more likely to be damaging compared to germline variants observed in the non-ALL population (mean CADD phred-like score of 25.6 vs 15.2, respectively, p