Search

The main objective of our project is to obtain a comprehensive insight into oncogenic signaling in order to develop novel diagnostic tools for molecular subtypes of breast cancer (BRCA). We have provided about 100 BRCA samples along with full pathology and clinical annotations to the TCGA program. These cancer samples have been extensively characterized using all available genomics platforms and the obtained molecular data contributed to the molecular characterization of BRCA. This integrated effort identified four major molecular subtypes of breast cancer that can be identified across five profiling platform including genomics, transcriptomics and proteomics (Nature, 490 (7418):61-70). Building upon these datasets we aim to translate these genomics profiles into diagnostic tools that can be used in every day medical practice. In the first step, we have generated tissue microarrays (TMAs) from FFPE BRCA samples obtained from the patients enrolled into the TCGA project. Next, the TMA slides were used for immunohistochemistry (IHC) using about 100 antibodies specific for major cancer markers, including oncogenic kinases frequently activated or overexpressed in BRCA. The obtained IHC results have been scored and verified independently for each marker by two board-certified pathologists. To identify markers that correlate with or are specific to the molecular subtypes of BRCA, the TMA/IHC read-outs from each tumor sample were correlated with all TCGA genomic data. In addition, we integrate the pathology and genomics results with clinical data obtained from the TCGA-enrolled patients, including up to 4 years follow-up. Finally, we apply system biology tools to integrate the genomic data from TCGA with the proteomic analysis to understand causalities between changes in DNA, transcriptome and signal transduction pathways. Our long-term goal is to identify novel diagnostic biomarkers that will precisely identify each molecular subtype of BRCA and which may be predictive of patient response to therapy thus paving the way for novel personalized therapies for cancer. Citation Format: Jannik Andersen, Parantu Shah, Konstanty Korski, Matthew Ibbs, Violetta Filas, Michal Kosiedowski, Juliusz Pukacki, Cezary Mazurek, Yuanqing Wu, Edward Chang, Carlo Toniatti, Giulio Draetta, Maciej Wiznerowicz. Applying TCGA data for breast cancer diagnostics and pathway analysis. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 4272. doi:10.1158/1538-7445.AM2014-4272

To study the influence of subcellular localization as a determinant of signal transduction specificity, we assessed the effects of wild-type transmembrane and cytoplasmic protein tyrosine phosphatase (PTP) epsilon on tyrosine kinase signalling in baby hamster kidney (BHK) cells overexpressing the insulin receptor (BHK-IR). The efficiency by which differently localized PTPepsilon and PTPalpha variants attenuated insulin-induced cell rounding and detachment was determined in a functional clonal-selection assay and in stable cell lines. Compared with the corresponding receptor-type PTPs, the cytoplasmic PTPs (cytPTPs) were considerably less efficient in generating insulin-resistant clones, and exceptionally high compensatory expression levels were required to counteract phosphotyrosine-based signal transduction. Targeting of cytPTPepsilon to the plasma membrane via the Lck-tyrosine kinase dual acylation motif restored high rescue efficiency and abolished the need for high cytPTPepsilon levels. Consistent with these results, expression levels and subcellular localization of PTPepsilon were also found to determine the phosphorylation level of cellular proteins including focal adhesion kinase (FAK). Furthermore, PTPepsilon stabilized binding of phosphorylated FAK to Src, suggesting this complex as a possible mediator of the PTPepsilon inhibitory response to insulin-induced cell rounding and detachment in BHK-IR cells. Taken together, the present localization-function study indicates that transcriptional control of the subcellular localization of PTPepsilon may provide a molecular mechanism that determines PTPepsilon substrate selectivity and isoform-specific function.

Focal Adhesion Kinase (FAK) is a key regulator of cancer cell migration and invasion and overexpression of FAK and Src has been associated with resistance to current anticancer therapies. Thus, inhibition of FAK may represent a promising therapeutic approach to anti-cancer treatment. We investigated the anti-cancer activity of the FAK inhibitor (GSK2256098) in a panel of 26 human glioblastoma (GBM) cell lines. GBM is one of the most aggressive and lethal forms of cancer, characterized by exponential growth and diffuse invasiveness. Herein, we report that 8/26 GBM cell lines displayed sensitivity to FAK inhibition in both migration and Matrigel invasion assays. GSK2256098 had minimal effect on 2D cell proliferation. Integrated genomic and proteomic analyses revealed up-regulation of the PI3K and MAPK pathways in the FAK inhibitor-resistant GBM cell lines, suggesting possible drug combination strategies. To test this hypothesis, we assessed the effects of combining GSK2256098 with a PI3K inhibitor and a MEK inhibitor in the FAK inhibitor-resistant GBM cell lines. Our results suggest that multi-agent inhibition of FAK and MEK or PI3K may provide an attractive therapeutic strategy for GBM. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 3714. doi:1538-7445.AM2012-3714

Pharmacodynamic (PD) biomarkers are an increasingly valuable tool for decision-making and prioritization of lead compounds during preclinical and clinical studies as they link drug-target inhibition in cells with biological activity. They are of particular importance for novel, first-in-class mechanisms, where the ability of a targeted therapeutic to impact disease outcome is often unknown. By definition, proximal PD biomarkers aim to measure the interaction of a drug with its biological target. For kinase drug discovery, protein substrate phosphorylation sites represent candidate PD biomarkers. However, substrate phosphorylation is often controlled by input from multiple converging pathways complicating assessment of how potently a small molecule drug hits its target based on substrate phoshorylation measurements alone. Here, we report the use of quantitative, differential mass-spectrometry to identify and monitor novel drug-regulated phosphorylation sites on target kinases. Autophosphorylation sites constitute clinically validated biomarkers for select protein tyrosine kinase inhibitors. The present study extends this principle to phosphorylation sites in serine/threonine kinases looking beyond the T-loop autophosphorylation site. Specifically, for the 3'-phosphoinositide-dependent protein kinase 1 (PDK1), two phospho-residues p-PDK1(Ser410) and p-PDK1(Thr513) are modulated by small-molecule PDK1 inhibitors, and their degree of dephosphorylation correlates with inhibitor potency. We note that classical, ATP-competitive PDK1 inhibitors do not modulate PDK1 T-loop phosphorylation (p-PDK1(Ser241)), highlighting the value of an unbiased approach to identify drug target-regulated phosphorylation sites as these are complementary to pathway PD biomarkers. Finally, we extend our analysis to another protein Ser/Thr kinase, highlighting a broader utility of our approach for identification of kinase drug-target engagement biomarkers.

The human genome encodes some 500 protein kinases and an estimated twenty percent of drugs currently in development target this enzyme class. Because kinase inhibitors frequently bind to the conserved ATP-binding pocket, in vitro kinase selectivity profiling of drug candidates are an integral part of the small-molecule drug discovery process. Likewise, the co-development of pharmacodynamic (PD) biomarkers is critical to drug development as they link drug-target inhibition in cells with biological activity. As such PD biomarkers are essential for the interpretation of in vivo PK/PD/efficacy studies, which ultimately inform medicinal chemistry. Here, we present a universal mass-spectrometry-based approached to discover kinase drug-target engagement biomarkers using the phosphoinositide dependent protein kinase 1 (PDK1) as an example. Western blot analysis using a commercial available phosphosite-specific PDK1 antibody (Ser241) revealed that classical, ATP-competitive PDK1 inhibitors do not dynamically impact PDK1 T-loop phosphorylation in cells, despite cell biochemical inhibition of phospho-AKT(Thr308), a bona fide PDK1 substrate phosphorylation site. However, because receptor tyrosine kinases (RTKs), lipid kinases and other pathways often signal via AKT, the monitoring of phospho-AKT as a PD biomarker for PDK1 is not sufficient for delineating the specificity of small-molecule inhibitors and ascertain the degree of PDK1 target engagement in cells. Here, we used stable isotope labeling by amino acids (SILAC) in cell culture combined with immunoprecipitation of PDK1 protein from cells for quantitative mass-spectrometry-based identification of phosphorylation sites modulated by PDK1 inhibitor treatment. We mapped 10 Serine/Threonine phosphorylation sites on PDK1 including several novel phosphorylation sites. Of these, two PDK1 phosphorylation sites (Ser410 and Thr513) were modulated by pharmacological PDK1 inhibition and the degree of dephosphorylation correlated with inhibitor potency in PDK1 enzymatic assays. By contrast, pharmacological inhibition of PDK1 did not quantitatively modulate PDK1 T-loop phosphorylation (Ser241) consistent with the initial western blot analysis. Altogether, our study provides proof-of-concept for identifying novel kinase target engagement PD biomarkers for PDK1, complementing pathway biomarkers such p-AKT and p-S6RP, which are modulated broadly by PI3K-pathway targeting agents including inhibitors of PI3K/mTOR and RTKs. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4953.

Protein phosphorylation plays a key role in cell signaling and pathophysiological alterations in protein kinases and phosphatases contribute to human diseases. We used stable isotope labeling by amino acids (SILAC) in cell culture combined with phospho-antibody based enrichment for quantitative mass-spectrometry-based identification of differentially phosphorylated proteins in response to small-molecule PI3K-pathway inhibition. We quantified over 500 non-redundant serine/threonine phosphopeptides (NSTPs) containing either the AKT-, MAPK-substrate, or PDK1-docking motif. Of these NSTPs, 71 phosphoproteins were modulated by inhibitors targeting either AKT, PDK1 or PI3K/mTOR and a common set of 11 were modulated by all three drugs. Bioinformatics analysis of the regulated phosphoproteins identified core components of the canonical PI3K pathway and showed enrichment in adaptor and scaffolding molecules involved in cell polarity (PAK4), cytoskeletal reorganization (Filamin, Stathmin, Dynamin) vesicle transport (LARP1, VPS13D, SLC20A1), protein translation (S6RP and PRAS40), and transcription (EIF4BP-1). These results have guided the rational selection of antibodies for biomarker and patient stratification assays and a newly developed phosphor-specific antibody against PRAS40T246 shows specificity for oncogenic PI3K-pathway activation in PTEN-deficient mouse prostate tumor tissue by immunohistochemistry. In addition, the phospho-PRAS40T246 biomarker was evaluated across a panel of 67 breast and 96 lung cancer cell lines and in triple negative human breast tumor tissue. In these datasets, phopspho-PRAS40T246 positively correlates with phospho-AKTS473, but not PTEN protein expression. As such, we have positioned PRAS40T246 as a clinically relevant biomarker for the identification of PI3K-pathway activated tumors to enable individualized cancer therapy. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 5560.

Among the many types of molecular mechanisms that are currently understood, none is more important for the control of cellular faith than the network of protein kinases and phosphatases. The importance of understanding these signaling networks is furthermore highlighted by the fact that virtually all major pharmaceutical companies have significant projects focused on targeted interference against oncogenic kinases as potential anti-tumor therapies. Yet, a system's wide analysis of signaling phosphorylations by these inhibitors in in vivo (i.e. tissue or xenograft models) is cost prohibitive and rarely performed. The use of mass spectrometry is a powerful approach to study complex mixtures, and some techniques have used either chemical or metabolical isotopic labeling strategies to quantify protein differences between cohorts. While these techniques have been shown to be useful and are routinely applied for cell culture systems, they do not extend readily to in vivo studies as these require either combining of samples or are dependent on the duty cycle of the mass spectrometer. Here we report a generally applicable proteomic approach to identify global protein phosphorylation changes in in-vivo systems. This label-free discovery platform, named phospho-differential mass spectrometry (phospho-dMS) does not require complex mixing or pooling strategies or isotope labeling, and instead identifies statistically significant changes in full scan mass spectrometry data. Hence, in-vivo samples can be analyzed individually, allowing the use of longitudinal designs and within-subject data analysis. We demonstrate phospho-dMS on 27 breast tumor tissues samples (3 concentrations, 3 time points, 3 mice each) excised from a PTEN-LOF mouse model. We quantified in excess of 2000 high confidence phosphorylation events. The most prominent protein modules inhibited by compound treatment (P Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 101st Annual Meeting of the American Association for Cancer Research; 2010 Apr 17-21; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2010;70(8 Suppl):Abstract nr 4561.

Activity of enzymes, such as protein tyrosine phosphatases (PTPs), is often associated with structural changes in the enzyme, resulting in selective and stereospecific reactions with the substrate. To investigate the effect of a substrate on the motions occurring in PTPs, we have performed molecular dynamics simulations of PTP1B and PTP1B complexed with a high-affinity peptide DADEpYL, where pY stands for phosphorylated tyrosine. The peptide sequence is derived from the epidermal growth factor receptor (EGFR 988–993 ). Simulations were performed in water for 1ns, and the concerted motions in the protein were analyzed using the essential dynamics technique. Our results indicate that the predominately internal motions in PTP1B occur in a subspace of only a few degrees of freedom. Upon substrate binding, the flexibility of the protein is reduced by ∼10%. The largest effect is found in the protein region, where the N-terminal of the substrate is located, and in the loop region Val 198 -Gly 209 . Displacements in the latter loop are associated with the motions in the WPD loop, which contains a catalytically important aspartic acid. Estimation of the pK a of the active-site cysteine along the trajectory indicates that structural inhomogeneity causes the pK a to vary by approximately ±1 pK a unit. In agreement with experimental observations, the active-site cysteine is negatively charged at physiological pH.