Yuhong Wei

Integrated Omic analysis of lung cancer reveals metabolism proteome signatures with prognostic impact

Cancer results from processes prone to selective pressure and dysregulation acting along the sequence-to-phenotype continuum DNA → RNA → protein → disease. However, the extent to which cancer is a manifestation of the proteome is unknown. Here we present an integrated omic map representing non-small cell lung carcinoma. Dysregulated proteins not previously implicated as cancer drivers are encoded throughout the genome including, but not limited to, regions of recurrent DNA amplification/deletion. Clustering reveals signatures composed of metabolism proteins particularly highly recapitulated between patient-matched primary and xenograft tumours. Interrogation of The Cancer Genome Atlas reveals cohorts of patients with lung and other cancers that have DNA alterations in genes encoding the signatures, and this was accompanied by differences in survival. The recognition of genome and proteome alterations as related products of selective pressure driving the disease phenotype may be a general approach to uncover and group together cryptic, polygenic disease drivers.

Primary tumor xenografts of human lung adeno and squamous cell carcinoma express distinct proteomic signatures.

Nonsmall cell lung carcinoma (NSCLC) accounts for 80% of lung cancers. The most prevalent subtypes of NSCLC are adenocarcinoma (ADC) and squamous cell carcinoma (SCC), which combined account for approximately 90%. Ten resected NSCLC patient tumors (5 ADC and 5 SCC) were directly introduced into severely immune deficient (NOD-SCID) mice, and the resulting xenograft tumors were analyzed by standard histology and immunohistochemistry (IHC) and by proteomics profiling. Mass spectrometry (MS) methods involving 1- and 2-dimensional LC-MS/MS, and multiplexed selective reaction monitoring (SRM, or MRM), were applied to identify and quantify the xenograft proteomes. Hierarchical clustering of protein profiles distinguished between the ADC and SCC subtypes. The differential expression of 178 proteins, including a comprehensive panel of intermediate filament keratin proteins, was found to constitute a distinctive proteomic signature associated with the NSCLC subtypes. Epidermal growth factor receptor (EGFR) was expressed in ADC and SCC xenografts, and EGFR network activation was assessed by phosphotyrosine profiling by Western blot analysis and SRM measurement of EGFR levels, and mutation analysis. A multiplexed SRM/MRM method provided relative quantification of several keratin proteins, EGFR and plakophilin-1 in single LC-MS/MS runs. The protein quantifications by SRM and MS/MS spectral counting were associated with superior dynamic range and reproducibility but were otherwise consistent with orthogonal methods including IHC and Western immuno blotting. These findings illustrate the potential to develop a comprehensive MS-based platform in oncologic pathology for better classification and potentially treatment of NSCLC patients.

Proteomic profiles of human lung adeno and squamous cell carcinoma using super-SILAC and label-free quantification approaches.

Nonsmall cell lung cancer (NSCLC) accounts for 85% of lung cancers, and is subdivided into two major histological subtypes: adenocarcinoma (ADC) and squamous cell carcinoma (SCC). There is an unmet need to further subdivide NSCLC according to distinctive molecular features that may be associated with responsiveness to therapies. Four primary tumor‐derived xenograft proteomes (two‐each ADC and SCC) were quantitatively compared by using a super‐SILAC labeling approach together with ultrahigh‐resolution MS. Proteins highly differentially expressed in the two subtypes were identified, including 30 that were validated in an independent cohort of 12 NSCLC primary tumor‐derived xenograft tumors whose proteomes were quantified by an alternative, label‐free shotgun MS methodology. The 30‐protein signature contains metabolism enzymes including phosphoglycerate dehydrogenase, which is more highly expressed in SCC, as well as a comprehensive set of cytokeratins and other components of the epithelial barrier, which is therefore distinctly different between ADC and SCC. These results demonstrate the utility of the super‐SILAC method for the characterization of primary tissues, and compatibility with datasets derived from different MS‐based platforms. The validation of proteome signatures of NSCLC subtypes supports the further development and application of MS‐based quantitative proteomics as a basis for precision classifications and treatments of tumors. All MS data have been deposited in the ProteomeXchange with identifier PXD000438 (http://proteomecentral.proteomexchange.org/dataset/PXD000438).

CHCHD2 Is Coamplified with EGFR in NSCLC and Regulates Mitochondrial Function and Cell Migration

Coiled-coil-helix-coiled-coil-helix domain-containing 2, a mitochondrial protein, encoded by CHCHD2 is located at chromosome 7p11.2 and proximal to the EGFR gene. Here, bioinformatic analyses revealed that CHCHD2 is consistently coamplified with EGFR in non–small cell lung carcinoma (NSCLC). In addition, CHCHD2 and EGFR protein expression levels were positively correlated and upregulated relative to normal lung in NSCLC tumor-derived xenografts. Knockdown of CHCHD2 expression in NSCLC cells attenuated cell proliferation, migration, and mitochondrial respiration. CHCHD2 protein–protein interactions were assessed by the complementary approaches of affinity purification mass spectrometry and in vivo proximity ligation. The CHCHD2 interactome includes the apparent hub proteins C1QBP (a mitochondrial protein) and YBX1 (an oncogenic transcription factor), and an overlapping set of hub-associated proteins implicated in cell regulation. Implications: CHCHD2 influences mitochondrial and nuclear functions and contributes to the cancer phenotype associated with 7p11.2 amplification in NSCLC. Mol Cancer Res; 13(7); 1119–29. ©2015 AACR.

Molecular heterogeneity of non-small cell lung carcinoma patient-derived xenografts closely reflect their primary tumors.

Availability of lung cancer models that closely mimic human tumors remains a significant gap in cancer research, as tumor cell lines and mouse models may not recapitulate the spectrum of lung cancer heterogeneity seen in patients. We aimed to establish a patient‐derived tumor xenograft (PDX) resource from surgically resected non‐small cell lung cancer (NSCLC). Fresh tumor tissue from surgical resection was implanted and grown in the subcutaneous pocket of non‐obese severe combined immune deficient (NOD SCID) gamma mice. Subsequent passages were in NOD SCID mice. A subset of matched patient and PDX tumors and non‐neoplastic lung tissues were profiled by whole exome sequencing, single nucleotide polymorphism (SNP) and methylation arrays, and phosphotyrosine (pY)‐proteome by mass spectrometry. The data were compared to published NSCLC datasets of NSCLC primary and cell lines. 127 stable PDXs were established from 441 lung carcinomas representing all major histological subtypes: 52 adenocarcinomas, 62 squamous cell carcinomas, one adeno‐squamous carcinoma, five sarcomatoid carcinomas, five large cell neuroendocrine carcinomas, and two small cell lung cancers. Somatic mutations, gene copy number and expression profiles, and pY‐proteome landscape of 36 PDXs showed greater similarity with patient tumors than with established cell lines. Novel somatic mutations on cancer associated genes were identified but only in PDXs, likely due to selective clonal growth in the PDXs that allows detection of these low allelic frequency mutations. The results provide the strongest evidence yet that PDXs established from lung cancers closely mimic the characteristics of patient primary tumors.

Abstract 1822: Proteome signatures distinguish lung cancer subtypes, define metabolism states, and have prognostic impact

Proceedings: AACR 106th Annual Meeting 2015; April 18-22, 2015; Philadelphia, PA Lung cancer is one of the most common cancers worldwide, and is the number one cause of cancer death in both men and women. Non-small cell lung cancer (NSCLC) accounts for 85% of lung cancers and is subdivided into two major histological subtypes: adenocarcinoma (ADC) and squamous cell carcinoma (SCC). There is an unmet need to better understand and stratify NSCLC according to distinctive molecular features that may help develop diagnostic and therapeutic strategies to improve patient outcomes. The proteome is expected to have a pronounced and direct effect on cancer phenotypes. Therefore, proteomic approaches hold promise as superior methods to characterize cancers, with the ultimate goal to translate research results into clinical utilities. Mass spectrometry (MS)-based quantitative comprehensive proteome analysis resolved the proteomes of human lung ADC and SCC primary tumor-derived xenografts. A multi-protein signature able to distinguish between ADC and SCC was identified and validated in an independent cohort of samples. The signature is comprised of various components of the epithelial barrier and metabolism enzymes. Signatures composed of metabolism proteins were found to be highly recapitulated between primary and matched xenograft tumors, and when extrapolated to DNA alterations in the encoding genes, to have prognostic impact for overall patient survival. The ability of NSCLC primary tumors to engraft in severely immune deficient mice is an independent predictor of shorter disease-free survival in early-stage NSCLC patients. Therefore, we sought to identify proteome signatures of engraftment, which would then be tested for prognostic impact. The proteomes of a series of >50 NSCLC primary tumors that engrafted or not were quantitatively compared by using the so-called super-SILAC method, in which a mixture of metabolically-labeled, stable-isotope-encoded NSCLC-derived cell lines were used as an internal standard. ADC and SCC tumors were analyzed, and a signature of proteins including enzymes involved in central carbon metabolism were identified as differentially expressed between engrafting and non-engrafting tumors. These results highlight the significance of metabolic remodeling as a feature that might be a determinant of more aggressive cancer phenotypes. These results support the further development of proteome signatures to diagnose, stratify, and precisely treat NSCLC. Citation Format: Wen Zhang, Paul Taylor, Lei Li, Yuhong Wei, Jiefei Tong, Vladimir Ignatchenko, Nhu-An Pham, Thomas Kislinger, Ming-sound Tsao, Michael Moran. Proteome signatures distinguish lung cancer subtypes, define metabolism states, and have prognostic impact. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 1822. doi:10.1158/1538-7445.AM2015-1822

Abstract SY33-04: Integrated omic analysis of lung cancer reveals metabolism-proteome signatures with prognostic impact

Proceedings: AACR 106th Annual Meeting 2015; April 18-22, 2015; Philadelphia, PA Cancer results from processes prone to selective pressure and dysregulation acting along the sequence-to- phenotype continuum DNA→RNA→Protein→Disease. An Omics Array integrating DNA gene copy number, mRNA transcriptome, and quantified proteome was assembled into a genetic map representing non-small cell lung carcinoma (NSCLC). Data were collected from patient-matched normal lung, primary tumors, and patient tumor-derived xenograft (PDX) tumors. Dysregulated proteins not previously implicated as cancer drivers were found encoded throughout the genome including but not limited to regions of recurrent DNA amplification/deletion in NSCLC. Unsupervised clustering revealed signatures comprising metabolism proteins particularly highly recapitulated between matched primary and PDX tumors, and which distinguished between the major NSCLC histological subtypes adenocarcinoma (ADC) and squamous cell carcinoma (SCC). Interrogation of The Cancer Genome Atlas (TCGA) revealed sizeable cohorts of NSCLC patients with DNA alterations in genes encoding the metabolism proteome signatures, and accompanied by differences in survival. Similar to the proteome signatures from which they were extrapolated, the gene mutation signatures with prognostic impact discriminated between the lung ADC and SCC subtypes. Serine hydroxymethyltransferase 2 (SHMT2), a key enzyme in serine/glycine and folate- dependent one-carbon metabolism, is upregulated in the proteomes of NSCLC primary and PDX tumours, and is implicated as a driver of recurrent chromosome 12q14.1 amplification in NSCLC. SHMT2, along with other enzymes implicated as anti-folate targets, is also part of a metabolism proteome signature associated with poor outcome in lung ADC. The interrogation of cancer genomes and proteomes for alterations that are related products of selective pressures driving the cancer phenotype may be a general approach to uncover and group together cryptic, polygenic cancer drivers, which might represent new anti-cancer therapeutic targets. Citation Format: Michael F. Moran, Lei Li, Yuhong Wei, Paul Taylor, Christine To, Jiefei Tong, Vladimir Ignatchenko, Melania Pintilie, Nhu-An Pham, Wen Zhang, Lakshmi Muthuswamy, Frances A. Shepherd, Thomas Kislinger, Ming S. Tsao. Integrated omic analysis of lung cancer reveals metabolism-proteome signatures with prognostic impact. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr SY33-04. doi:10.1158/1538-7445.AM2015-SY33-04

Abstract 5127: Characterization of lung cancers by integrated genomic and proteomic analysis

Non-small cell lung carcinoma (NSCLC) represents 80% of lung cancers, the deadliest cancer worldwide. The genomic profiling of DNA and mRNA, and characterization of proteomes have begun to address the objective to stratify and treat tumors according to their molecular features. However, these data sets have largely been used independently and typically have not been integrated. Hence most cancers including NSCLC continue to be classified largely based on histology. Our aim for this study was to integrate a set of comprehensive functional genomics data sets in order to stratify a set of NSCLC primary tumors and establish that primary tumor xenografts mirror closely the primary tumors, and hence may serve as validated pre-clinical models. Our preliminary analyses indicated that engraftment is prognostic of poor clinical outcome (John et al., 2011, Clin Cancer Res, 17:134-41), and that the major NSCLC subtypes adenocarcinoma and squamous cell carcinoma are readily resolved according to their distinctive proteomes (Wei et al., 2011, J Proteome Res 10:161-74). Herein we characterized a collection of 12 each primary tumor (T), primary tumor xenograft (X), and patient-matched normal lung (N) by using mass spectrometry for proteome analysis, Illumina 1M Omni-Quad for somatic copy number alterations (SCNAs), and Illumina Omni-1 Quad HT-12 v4 for mRNA expression. Unsupervised hierarchical clustering of protein abundances and SCNAs independently revealed that primary tumor and xenografts are highly correlated with each other. This correlation was significantly enhanced in the proteome data when a small number of highly abundant blood-associated proteins were systematically identified and subtracted. We identified tumor-specific dysregulated proteins and SCNAs in T and X using N as a reference. Two thirds of T and X matched pairs could be identified based on Pearson Correlation Coefficients of the dysregulated proteins. This clearly demonstrates that the xenografts accurately recapitulated tumor proteomes. Proteins upregulated in tumors were expressed to a significant extent from regions of SCNA gain, and we found a high degree of concordance between mRNA expression levels and SCNAs. Some primary tumors had very highly correlated proteomic profiles, suggesting they may be effectively stratified according to their proteome signatures. In conclusion, our integrated analysis has validated the primary xenograft model, provided an initial systems level perspective on the central dogma in cancer, and reinforces the proteome as a distinctive molecular feature for lung tumor stratification. 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 5127. doi:1538-7445.AM2012-5127

Abstract 4805: Distinctive phospho-proteome signatures in non-small cell lung carcinoma tumors

Non-small cell lung carcinoma (NSCLC) is a major, lethal cancer worldwide. Various molecular and cellular abnormalities are found in NSCLC, and an established theme is the alteration of protein phosphotyrosine (pY), as exemplified by elevated epidermal growth factor receptor (EGFR) activity recognized in a subset of NSCLC. In order to gain a broad perspective of pY-mediated signaling networks in NSCLC, mass spectrometry was used to analyze pY-containing peptides affinity purified from protease-digested xenograft tumors. Seventeen tumors representing the major NSCLC subtypes adenocarcinoma (ADC) and squamous cell carcinoma (SCC) were characterized. The major classes of pY proteins detected were cytoskeletal, protein kinases, and adhesion proteins. Pathways and cellular functions enriched in the data set included focal adhesion, adherens junction, and ErbB signaling. Five receptor tyrosine kinases (RTKs) were detected, with EGFR and EPHA2 most prevalent. Src-family kinases (YES, LYN, FRK and LCK) and focal adhesion kinase (FAK) comprised the majority of non-receptor tyrosine kinases measured. Analysis of individual pY sites on 14 tyrosine kinases indicates their activation by kinase domain activation loop phosphorylation, but also revealed additional pY sites suggestive of regulation by trafficking/endocytosis and inter- and intra-molecular protein-protein interactions involving, for example, SH2 and SH3 domains, and extra-kinase domain regions. Analysis of tumors based on protein-pY suggested subtypes distinguished by their activated tyrosine kinase networks. Genetic analysis of 3 tumors that displayed EGFR-associated phospho-proteome signatures confirmed that they carried either EGFR activating mutations (in two ADCs) or increased gene copy number (in one SCC). Western blot analysis of protein pY confirmed different patterns of cellular protein pY among the tumor phospho-types in agreement with the phospho-proteomic analysis. Quantification of kinase and substrate pY sites, according to their integrated extracted ion currents, revealed a set of phosphorylations that constituted a molecular signature that differentiated between the ADC and SCC subtypes. In summary, quantitative phospho-proteomic profiling of NSCLC tumors is feasible and may have utility in the categorization of individual tumors according to their networks of activated tyrosine kinases and substrates. 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 4805. doi:1538-7445.AM2012-4805

Abstract 4919: Multiplexed EGFR signaling pathway analysis in FFPE tissue using quantitative mass spectrometry

Proceedings: AACR 102nd Annual Meeting 2011‐‐ Apr 2‐6, 2011; Orlando, FL The epidermal growth factor receptor (EGFR) is a drug target for both small molecule and antibody therapeutics and has been approved in non small-cell lung carcinoma (NSCLC) and colorectal carcinoma (CRC) among other indications. These drugs block receptor signaling though blockade of the tyrosine kinase domain, or through inhibition of ligand binding. Current genomic tests measure receptor amplification, RNA levels, the mutation status of receptor or pathway molecules (EGFR or kRAS mutations) but no current assay can directly assess the activation state of the EGFR or its downstream signaling pathway components. Indeed, the EGFR mutation positive NSCLC tumors (thought to be constitutively active) show a high response rate to TKI therapy, but the many non responders (50% or more) demonstrate the limitation of genomic analysis. Since activation of EGFR is necessary for the response to these targeted agents, it is critical to measure what levels of receptor activation and downstream signaling determines tumor responsiveness to EGFR targeted therapies in these patients. For this reason, we have developed a panel of new diagnostic assays which measure the activation of the EGFR and key downstream signaling proteins through quantitation of the phosphorylation state of these proteins. These assays are based on the Liquid Tissue®-SRM technology platform. This approach enables relative and absolute quantification of proteins and their phosphorylation status directly in formalin fixed paraffin embedded (FFPE) tissue. We preclinically validated the multiplexed Liquid Tissue® phospho-SRM assay on formalin fixed EGF stimulated A431 cells. We followed up these in vitro studies with phospho-SRM analysis of FFPE NSCLC xenograft explants where extensive independent histopathologic and molecular characterization had been performed, allowing us to benchmark our phospho-SRM analysis with standard diagnostic analyses. We have now extended these quantitation studies by measuring the expression of EGFR and phospho-EGFR in FFPE tissues obtained from relevant human clinical trial cohorts – Gefitinib treated NSCLC and Cetuximab treated CRC. It is hoped that we will be able to correlate EGFR expression, activation and signaling in these tumors with responsiveness to EGFR targeted therapy, and to validate this assay for use as a companion diagnostic to guide therapy in both NSCLC and CRC. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4919. doi:10.1158/1538-7445.AM2011-4919