Identification of Somatic Mutations in LGL Leukemia through Whole Genome Sequencing and Correlation of STAT3 Y640F Mutation with Treatment Response to Methotrexate.

Open Access
Olson, Thomas Lynn
Graduate Program:
Molecular Medicine
Doctor of Philosophy
Document Type:
Date of Defense:
June 18, 2013
Committee Members:
  • "Thomas P Loughran, Jr", Dissertation Advisor
  • Rosalyn Bryson Irby, Committee Chair
  • Thomas E Spratt, Committee Member
  • Richard James Courtney, Committee Member
  • Bruce A Stanley, Special Member
  • large granular lymphocytes
  • leukemia
  • STAT3
  • TBET
  • genomics
The rare disorder, large granular lymphocyte (LGL) leukemia is characterized by a clonal expansion of cytotoxic cells. Normal LGL are critical to the removal of virally-infected or tumor cells from the body. These cells normally expand when there is an immune challenge and then rapidly undergo apoptosis when the antigen is cleared, but leukemic LGLs are resistant to apoptosis. Serological evidence indicates that LGL leukemia may be driven by chronic infection of an unknown retrovirus. Patients with LGL leukemia frequently experience autoimmune conditions such as rheumatoid arthritis and also experience cytopenias requiring medical intervention. The study of LGL leukemia is therefore important both from a patient standpoint and as a model of normal LGL. We have recently discovered and characterized multiple somatic mutations in exon 20 and 21 of STAT3 from exome sequencing of an LGL leukemia patient and confirmed this mutation to be present in both Natural Killer and T cell leukemias. A small percentage was also found to have mutations in STAT5B, with some associated with an aggressive phenotype. Concurrently, we have undertaken whole genome sequencing to determine what other mutations may contribute to LGL leukemia and to expand our knowledge of STAT3 activation. Given the antigen driven nature of LGL leukemia, it is unclear what mutations are necessary for their lack of apoptosis and whether these will resemble those seen in other leukemias. The current treatment of choice for LGL leukemia is immunosuppressive methotrexate. Firstly, we present the first large prospective trial of immunosuppressive therapy in LGL leukemia. In this trial we identify a correlation between a particular Y640F mutation in STAT3 and a favorable response to first line treatment with methotrexate. We measure and report 27 serum cytokines collected for this trial, none of which were a priori predictive of treatment response. Through microarray analysis we identified a gene expression signature indicative of response. Interestingly, this gene signature does not appear to be driven by STAT3 mutation in all samples. We propose that there are other methods of STAT activation in these samples and advance a model of how the Y640F mutation enforces response. Secondly, we collaboratively sequenced 6 paired, leukemic/saliva genomes from three LGL patients at a high rate of coverage. Two patient genomes contain a direct STAT3 mutation. In the third patient we observe a mutation in the 3' UTR of the IL6R that could activate JAK-STAT signaling if it is shown to regulate translation. Mutations in the histone modifiers, MLL2, MLL3 and NCOR1 are supported in the sequence reads of all three genomes. Alterations in this pathway would be a new observation for LGL leukemia. We identify and catalog numerous other mutations which may give insight into LGL biology that could allow us to better understand how symptoms arise and differ between patients and why current treatments may fail. Lastly, we present data identifying important cis-regulatory modules in LGL leukemia gathered through ChIP-Seq of genomic regions occupied by the master regulator TBET. We report evidence of binding in the promoter region of numerous constituents of lytic granules which may indicate these are important downstream effectors of TBET. Overlap between the patients used for genome work above and ChIP-Seq is allowing us to determine whether mutations can have an effect on TBET binding and subsequent transcription of nearby genes. As we collect the footprints of additional factors, this holds the promise of elucidating the potential impact of the tens of thousands of non-coding mutations observed in these samples. In summary, we present the results of large collaborative efforts to sequence mutations in LGL leukemia and determine their effect on LGL biology. We advance the knowledge of STAT3 mutations in LGL biology while identifying and stratifying other mutations for the future study of their potential impacts. Bioinformatic methods of analysis were focused on publicly available resources such as Galaxy for which new tools, file formats, and workflows were developed. This allows easy reproduction of methods by other groups and facilitates visualization and future analysis by our own.