Whole Kinome Profiling In Pediatric Acute Leukemia.

Abstract 1029

With almost 130.000 entries in PubMed by today, the kinase family represents one of the gene families with the best documented history in malignant diseases.

Acute myeloid leukemias account for 4.7% of all childhood cancers and 13.7% of all childhood leukemias. Recent german statistics (Kinderkrebsregister Mainz) indicate a 5-year survival rate of only 66%, emphasizing the need to discover novel therapeutic targets and prognostic markers. Here we present an adaptable workflow to detect mutations in kinase family members on genomic and expression level as well as in silico assessment of their functional consequences.

To undertake the venture of resequencing the human kinome of leukemia patients, we constructed a custom SequenceCapture microarray (NimbleGen) based on all exons of all 518 members of the kinase family previously reported by Greenman and colleagues (Greenman et al, Nature, 2007) as well as a set of TaqMan® low density arrays (Life Technologies) to confirm the expression of potential candidates in tumor material. Genomic DNA from the initial tumor sample and germline material from a 16 year old female patient diagnosed with a AML FAB1 with Auer bodies, 93% blasts, CD33⁺ CD117⁺ CD38⁺ and partial CD15⁺ and from a 16 year old female patient with a bilinear CD22⁺ CD33⁺ leukemia was enriched on SequenceCapture arrays and sequenced on the Roche GS FLX Titanium platform. Acquired sequences were mapped against the current human reference (hg19/GRCh37). RNA derived from the initial leukemic blasts was used for kinome-wide expression profiling on TaqMan low density arrays.

From sequencing the enriched DNA, we recieved between 85 and 129 megabases of tumor and germline sequence, of which 97.7–99.4% mapped to the human reference. About 90% of the targeted regions were covered at least once. Mapping and comparison to the germline material led to 28 and 46 tumor specific non-synonymous single nucleotide variants sequenced at a depth of ≥10x that were not previously reported as SNPs (dbSNP130). Analyzing these candidates with UniProt revealed that 6 and 12 were located in or very close to important functional domains of their respective protein. Structural protein models of one patients remaining mutations underscore the potential influence of detected mutations on the protein function. TaqMan® expression profiling of all kinases present on the SequenceCapture Array identified 3 and 4 of the remaining genes to be expressed in the tumor on mRNA level. In conclusion, starting from 518 candidates, we present a powerful workflow for high throughput analysis of the whole kinome in leukemic samples as well as CDK4, DYRK1B, GSG2 (pt1) and MASTL, MARK3, RP6KA5, CLK4 (pt2) as potential targets for further functional analysis.