Kocher, Jean‐Pierre

Abstract Background: Traditionally, mutational burden and mutational signatures have been assessed by tumor-normal pair DNA sequencing. The requirement of having both normal and tumor samples is not always feasible from a clinical perspective, and led us to investigate the efficacy of using RNA sequencing of only the tumor sample to determine the mutational burden and signatures, and subsequently molecular cause of the cancer. The potential advantages include reducing the cost of testing, and simultaneously providing information on the gene expression profile and gene fusions present in the tumor.Results: In this study, we devised supervised and unsupervised learning methods to determine mutational signatures from tumor RNA-seq data. As applications, we applied the methods to a training set of 587 TCGA uterine cancer RNAseq samples, and examined in an independent testing set of xxx TCGA colorectal cancer RNAseq samples. Both diseases are known associated with microsatellite instable high (MSI-H) and driver defects in DNA polymerase ɛ (POLɛ). From RNAseq called variants, we found majority (>95%) are likely germline variants, leading to C>T enriched germline variants (dbSNP) widely applicable in tumor and normal RNAseq samples. We found significant associations between RNA-derived mutational burdens and MSI/POLɛ status, and insignificant relationship between RNAseq total coverage and derived mutational burdens. Additionally we found that over 80% of variants could be explained by using the COSMIC mutational signature-5, -6 and -10, which are implicated in natural aging, MSI-H, and POLɛ, respectively. For classifying tumor type, within UCEC we achieved a recall of 0.56 and 0.78, and specificity of 0.66 and 0.99 for MSI and POLɛ respectively. By applying learnt RNA signatures from UCEC to COAD, we were able to improve our classification of both MSI and POLɛ. Conclusions: Taken together, our work provides a novel method to detect RNAseq derived mutational signatures with effective procedures to remove likely germline variants. It can leads to accurate classification of underlying driving mechanisms of DNA damage deficiency.

Abstract Motivation DNA methylation can be measured at the single CpG level using sodium bisulfite conversion of genomic DNA followed by sequencing or array hybridization. Many analytic tools have been developed, yet there is still a high demand for a comprehensive and multifaceted tool suite to analyze, annotate, QC and visualize the DNA methylation data. Results We developed the CpGtools package to analyze DNA methylation data generated from bisulfite sequencing or Illumina methylation arrays. The CpGtools package consists of three types of modules: (i) ‘CpG position modules’ focus on analyzing the genomic positions of CpGs, including associating other genomic and epigenomic features to a given list of CpGs and generating the DNA motif logo enriched in the genomic contexts of a given list of CpGs; (ii) ‘CpG signal modules’ are designed to analyze DNA methylation values, such as performing the PCA or t-SNE analyses, using Bayesian Gaussian mixture modeling to classify CpG sites into fully methylated, partially methylated and unmethylated groups, profiling the average DNA methylation level over user-specified genomics regions and generating the bean/violin plots and (iii) ‘differential CpG analysis modules’ focus on identifying differentially methylated CpGs between groups using different statistical methods including Fisher’s Exact Test, Student’s t-test, ANOVA, non-parametric tests, linear regression, logistic regression, beta-binomial regression and Bayesian estimation. Availability and implementation CpGtools is written in Python under the open-source GPL license. The source code and documentation are freely available at https://github.com/liguowang/cpgtools. Supplementary information Supplementary data are available at Bioinformatics online.

Abstract Massively parallel sequencing analyses have revealed a common mutation within the MYD88 gene (MYD88L265P) occurring at high frequencies in many non-Hodgkin lymphomas (NHL) including the rare lymphoplasmacytic lymphoma, Waldenström’s macroglobulinemia (WM). Using whole exome sequencing, Sanger sequencing and allele-specific PCR, we validate the initial studies and detect the MYD88L265P mutation in the tumor genome of 97% of WM patients analyzed (n=39). MYD88L265P was detected at lower frequencies in other indolent lymphomas including LPL (0%), MALT (4%), nodal MZL (5%) and splenic MZL (8%); all but one MYD88L265P was heterozygous. Due to the high frequency of MYD88 mutation in WM and other NHL, and its known effects on malignant B cell survival, therapeutic targeting of MYD88 signaling pathways may be useful clinically. However, while the effects of MYD88L265P on the activity of IRAK1/4 and NF-κB are have been studied previously, we are lacking a thorough characterization of the role of intermediary signaling proteins such as TRAF6 and TAK1 on the biology of MYD88L265P-expressing B cells. A better understanding of the proteins involved in MYD88L265P signaling may lead to the development of more targeted and effective therapeutic approaches. In an attempt to identify MYD88L265P –specific therapeutic targets we first wanted to characterize the role of intermediary signaling proteins that facilitate the downstream activation of NF-κB. Upon activation of TLRs or IL-1b receptors, MYD88 forms a homodimer and recruits IRAK1/4 and TRAF6 into a complex resulting in association and phosphorylation of TAK1 followed by activation of NF-κB. We monitored the formation of a complex comprised of MYD88, IRAK1, IRAK4 and TRAF6 and immunoprecipitation of either endogenous IRAK4 or IRAK1 revealed constitutive association of IRAK with TRAF6 and MYD88L265P. To assess if the formation of a MYD88L265P/IRAK/TRAF6 complex results in downstream activation of TAK1, constitutive TAK1 phosphorylation was measured and detected in all three cell lines that express MYD88L265P. An association between TAK1 and TRAF6, another measure of TAK1 activation, was also detectable. When a similar analysis of TAK1 was performed in DLBCL cells expressing wild-type MYD88, no phosphorylation of TAK1 was detected, nor was TAK1 associated with TRAF6. IRAK1, IRAK4, TAK1, TRAF6, and MYD88 were expressed at similar levels in all cell lines studied and therefore did not contribute the differences in MYD88 complex formation observed between cell lines. These studies were further confirmed using HEK 293T cells that were transduced with either a vector control plasmid or HA-tagged MYD88WT or MYD88L265P expression plasmids. Together, these studies suggest that MYD88L265P forms a complex with IRAK and TRAF6 resulting in constitutive activation of TAK1 and NF-κB. To confirm the significance of TAK1-mediated MYD88L265P signaling on lymphoma cell growth, the effect of the selective TAK1 inhibitor, (5Z)-7-Oxozeaenol, on cell proliferation was tested. All MYD88L265P-expressing cell lines were sensitive to TAK1 inhibition in a dose-dependent manner (0-10 μM). In contrast, NHL cells expressing MYD88WT were found to be insensitive to TAK1 inhibition. We next tested the impact of the TAK1 inhibitor on a MYD88L265P positive WM patient sample. Similar to what was seen in the WM cell lines, the TAK1 inhibitor inhibited WM cell growth and survival in a dose dependent manner. Additionally, the TAK1 inhibitor significantly reduced the level of IL-10 secreted by each of the cell lines. Together, these data suggest that MYD88L265P drives cell proliferation and cytokine secretion through a TAK1-dependent mechanism. In conclusion, we are the first to validate by NGS in a large patient cohort the high prevalence and specificity of MYD88L265P in WM. Cells harboring the L265P mutation but not wild-type MYD88 exhibit constitutive signaling leading to the hyperactivation of NF-κB. We have established the role of TAK1 as an integral component of MYD88L265P signaling in both WM and DLBCL cell. Our data suggest that targeting TAK1 clinically may be an effective strategy for the treatment of WM and other lymphomas driven by MYD88L265P signaling. Disclosures: Fonseca: millennium: Consultancy; amgen: Consultancy; Binding site: Consultancy; onyx: Consultancy; medtronic: Consultancy; Genzyme: Consultancy; Otsuka: Consultancy; Celgene: Consultancy; lilly: Consultancy; Onyx: Research Funding; cylene: Research Funding.

Abstract Background: Peripheral T-cell lymphomas (PTCLs) comprise a heterogeneous group of relatively uncommon and generally aggressive non-Hodgkin lymphomas. Recent studies have begun to reveal their mutational landscape, such as the recurrent mutations affecting RHOA, IDH2, and TET2 in angioimmunoblastic T-cell lymphomas (AITLs) and related PTCLs. Here, we studied the exomes of 63 PTCLs to address several unanswered questions: (1) How frequently do PTCLs have mutations with a strong rationale for being actionable using clinically available targeted therapies? (2) What pathways are enriched in PTCLs that could point to additional targeted therapies? (3) What additional recurrent mutations exist that may provide insight into PTCL pathogenesis and classification? Methods: We performed exome capture and sequencing in 63 frozen PTCL samples (with matched germline controls from 22), including 13 PTCLs, not otherwise specified; 6 AITLs; 37 anaplastic large cell lymphomas (ALCLs: 12 ALK+, 19 ALK-, and 6 cutaneous); 4 extranodal NK/T-cell lymphomas; 2 cutaneous T-cell lymphomas, and 1 enteropathy associated T-cell lymphoma. Following primary variant calling, data were filtered by removing variants that were: outside the capture region, polymorphic (0.1% in ExAC), seen in paired germline or 50 healthy controls, within repetitive elements, or synonymous. Potentially actionable variants required published evidence for altering protein function and demonstrated in vivo or in vitro sensitivity to an available targeted agent (including open clinical trials). Mutated cancer genes (COSMIC) were examined for enrichment of KEGG canonical signaling pathways. A recurrent mutation encoding MSC E116K was validated in tumor DNA and MSC protein was examined by immunohistochemistry. Results: Of 25 unique variants with published evidence suggesting actionability (including ErbB, JAK/STAT, and RAS genes), at least one variant was present in 16/63 patients (25%). Variants in any COSMIC gene were seen in 55/63 samples (87%). Top genes were STAT3, RHOA, ARID1A, MLL2 (KMT2D), PTPRB, TP53, and TET2. The KEGG ErbB pathway was most significantly enriched for mutated genes in our dataset (EGFR, ERBB2, ERBB3, ERBB4, and others; q=3.88x10-3), followed by neurotrophin and mitogen-activated protein kinase signaling pathways. Among 25 recurrent non-synonymous variants, we identified MSC E116K with predicted gain-of-function exclusively in 3 ALK-ALCLs and 1 cutaneous ALCL. All mutated cases strongly expressed MSC protein. Among 96 cases, 13/18 ALK-ALCLs (72%) were positive for MSC, compared to 9/78 (12%) from other PTCL subtypes. Conclusions: Mutations that were potentially actionable based on published variant-specific data were identified in 25% of PTCLs, indicating promise for the role of sequencing in individualizing therapy. Additionally, remaining variants were enriched for several targetable pathways, most significantly ErbB. A novel recurrent mutation, MSC E116K, was seen exclusively in ALK-ALCLs (systemic and cutaneous). MSC protein, a repressor of E2A-mediated transcription of targets such as MYC, was expressed in most ALK- ALCLs and generally was absent in other PTCL subtypes. This finding extends evidence for a genetic relationship between systemic and cutaneous ALCLs and suggests MSC may help distinguish ALK- ALCLs from other PTCLs. Disclosures Link: Genentech: Consultancy, Research Funding; Kite Pharma: Research Funding.

Abstract DNA methylation plays an import role in both tumorigenesis and therapy resistance. Demethylating drug regimens have shown efficacy in hematological malignancies, and are actively studied for other cancer types. High variability in patients’ responses to these drugs has been observed and highlights the demand to identify predictive biomarkers for these demethylating drugs. In particular, decitabine (latest FDA-approved demethylating drug) demonstrated as the most potent agent and has the best response on hematological cell lines. Two public datasets were collected to identify methylation biomarkers of decitabine. The first one is from Genomics of Drug Sensitivity in Cancer, with methylation data from Illumina 450K array and drug response from cell lines. The other one is a publicly available patient dataset (methylations from RRBS sequencing and decitabine response from clinic). 64900 CpG sites common to the two methylation platforms were used to train machine learning model (Elastic Net), based on the methylation data and decitabine response (AUC) from 108 hematological cell lines in GDSC. 73 CpGs were identified as predictive biomarkers by Elastic Net. The trained model was then used to make predictions on 39 patients (20 responders and 19 non-responders) with their methylation profiles of the same set of CpGs. The predicted response of responder group is indeed better than non-responder group (p-value = 0.0501), indicating the validity of identified methylation biomarkers. The top one CpG identified by Elastic Net resides in the promoter region of gene C20orf85, and is significantly hypermethylated in non-responder group (p-value = 0.0458), suggesting its potential involvement in the resistance to decitabine. Citation Format: Xuewei Wang, Jean-Pierre Kocher, Zhifu Sun. Methylation biomarkers of cellular response to demethylating drug [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 681.

AbstractAnaplastic large cell lymphomas (ALCLs) represent a relatively common group of T-cell non-Hodgkin lymphomas (T-NHLs) that are unified by similar pathologic features but demonstrate marked genetic heterogeneity. ALCLs are broadly classified as being anaplastic lymphoma kinase (ALK)+ or ALK−, based on the presence or absence of ALK rearrangements. Exome sequencing of 62 T-NHLs identified a previously unreported recurrent mutation in the musculin gene, MSCE116K, exclusively in ALK− ALCLs. Additional sequencing for a total of 238 T-NHLs confirmed the specificity of MSCE116K for ALK− ALCL and further demonstrated that 14 of 15 mutated cases (93%) had coexisting DUSP22 rearrangements. Musculin is a basic helix-loop-helix (bHLH) transcription factor that heterodimerizes with other bHLH proteins to regulate lymphocyte development. The E116K mutation localized to the DNA binding domain of musculin and permitted formation of musculin–bHLH heterodimers but prevented their binding to authentic target sequence. Functional analysis showed MSCE116K acted in a dominant-negative fashion, reversing wild-type musculin-induced repression of MYC and cell cycle inhibition. Chromatin immunoprecipitation–sequencing and transcriptome analysis identified the cell cycle regulatory gene E2F2 as a direct transcriptional target of musculin. MSCE116K reversed E2F2-induced cell cycle arrest and promoted expression of the CD30–IRF4–MYC axis, whereas its expression was reciprocally induced by binding of IRF4 to the MSC promoter. Finally, ALCL cells expressing MSCE116K were preferentially targeted by the BET inhibitor JQ1. These findings identify a novel recurrent MSC mutation as a key driver of the CD30–IRF4–MYC axis and cell cycle progression in a unique subset of ALCLs.