WO2012170614A2 - Dna methylation markers in non-small cell lung cancer and methods of using the same - Google Patents

Abstract

Provided herein are unique sets of genes that are regulated by methylation in non-small cell lung cancer ("NSCLC") that are useful as a biomarker or in an assay for determining, diagnosing or subtyping lung cancer and/or the outcome of certain drug therapies.

Description

DNA Methylation Markers in Non-Small Cell Lung Cancer

and Methods of Using the Same

FIELD OF THE INVENTION

This invention relates to DNA methylation biomarkers and more particularly relates to such markers in non-small cell lung cancer.

CROSS REFERENCE TO RELATED INVENTIONS

This application claims the benefit of US provisional patent application number

61/494,142 filed June 7, 2011 which is incorporated herein by reference.

STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT

This invention was made with government support under SPORES P50CA070907 awarded by the National Institute of Health. The government has certain rights in the invention.

THE NAMES OF THE PARTIES TO A JOINT RESEARCH AGREEMENT

None.

REFERENCE TO SEQUENCE LISTING

None.

BACKGROUND OF THE INVENTION

DNA methylation is a key epigenetic mechanism for transcriptional regulation. DNA methylation is a stable marker that allows interrogation of the signal from multiple sources (tumor samples, blood products, circulating tumor cells). Using existing amplification technologies, this type of marker can be derived as a signal from single cells or from blood in order to help risk stratify patients. Further, since this type of marker is stable and not subject to degradation (unlike RNA-based technologies that look at mRNA expression changes), it can be robust. Notwithstanding the advantages of a DNA methylation biomarker, a comprehensive analysis of the genes that may be regulated by methylation in non-small cell lung cancer ("NSCLC") has not been reported. SUMMARY OF THE INVENTION

Provided herein are unique sets of genes (also referred herein as "GRMs") regulated by methylation in non-small cell lung cancer ("NSCLC") that are useful as a biomarker (also referred to herein as "a marker") or as an assay for determining, diagnosing or subtyping lung cancer and/or the outcome of certain drug therapies. An integrative analysis of genome-wide methylation profiling and gene expression was performed to decipher which genes are regulated by methylation in NSCLC. As a result, we discovered that DNA methylation of certain genes, or expression of these genes at the protein or mRNA level, can be used for a number of purposes. First, these GRMs may be useful for diagnosing or subtyping lung cancer, either by looking at DNA methylation or expression of GRMs in tissue or other materials (e.g. blood or sputum). Second, the methylation or expression of these GRMs may be useful for predicting clinical outcome or drug response in lung cancer patients. Third, a subset of these genes can distinguish epithelial cells from the more mesenchymal-like cell types. In this context, we refer to the genes as the "EMT-GRMs" or "EMT-GRMs genes." The EMT-GRMs are useful for predicting response to EGFR inhibitors such as erlotinib and other anticancer drugs such as docetaxel in NSCLC cells, and the methylation status or expression of these genes are both useful in as biomarkers for drug sensitivity in NSCLC patients. Fourth, a number of GRMs may be therapeutic targets in NSCLC including but not limited to SPINT-1,2 and Axl. Fifth, we have further discovered that CBLC is a marker of response to erlotinib.

BRIEF DESCRIPTION OF THE SEVERAL VIEWS OF THE DRAWINGS

Figures 1 A through ID show that unsupervised clustering of NSCLC cell lines based on NSCLC GRM predicts epithelial and mesenchymal cells in cell lines and tumors.

Figures. 2A and 2B show that GRM EMT signature predicts for drug response.

Figures 3A through 3F show that CBLC is methylated in NSCLC cell lines and is associated with EMT.

Figures 4A through 4D show that CBLC methylation is associated with erlotinib sensitivity in the context of KRAS mutation and MET amplification.

Figures 5A and 5B show that the association of GRM signature and CBLC expression is a predictor for erlotinib, vandetinib, or sorafenib 8wkDC in the BATTLE trial.

Figures 6A and 6B collectively represent the EMT-GRM genes.

Figures 7 A and 7B collectively represent the erlotonib-GRM genes. Figures 8A through 8U collectively represent of the GRM genes.

DETAILED DESCRIPTION OF THE INVENTION

We have identified a set of GRMs in NSCLC, and associated these GRMs with the response to different therapies (e.g. EGFR inhibitors, chemotherapy) and cellular phenotypes (e.g. epithelial or mesenchymal phenotypes). DNA methylation status at CpG sites around the GRMs (referred to as "GRM methylation status"), or expression of the GRMs at the gene expression or protein level, can be assessed in clinical samples including but not limited to tumor samples, blood products, sputum, circulating tumor cells, biopsy specimens from airways. Methylation status can be used to diagnosis cancer or risk of cancer, as a marker for the presence of cancer or preneoplasia, and as a response to anticancer treatment and/or clinical outcome. One such gene is CBLC, a marker of response to EGFR inhibitors such as erlotinib. Methylation status can also be useful status to identify epithelial or mesenchymal tumors, and as predictors of drug response (e.g. to erlotinib) or clinical outcome (e.g. risk of metastatic spread).

We have identified the GRMs by profiling DNA methylation in NSCLC cell lines in vitro and then correlating these methylation sites with expression of the corresponding genes using mRNA gene expression profiles. Certain GRMs were defined as genes whose expression correlated inversely with the degree of DNA methylation within a defined distance from the gene. Many of these GRMs were validated by methylation specific PCR and other methods. As defined herein, we further investigated these GRMs by examining their regulation after treatment with 5-azacytidine in a published data set. We identified a large number of genes not known to be regulated by methylation in lung cancer (See Figures 6, 7 and 8). A large percentage of GRMs were associated with EMT status. We then correlated GRMs with cell line sensitivity to different anticancer drugs including docetaxel and erlotinib and identified GRMs that are markers of drug response. One of these was CBLC methylation and gene expression which was found to be a marker of erlotinib response.

As such, we provide diagnostic assays useful in diagnosing cancer or cancer risk; assessing prognosis or likelihood of response to different drugs (i.e., a predictive marker) including erlotinib and docetaxel, or for assessing EMT status comprising GRMs as defined herein.

As acknowledged in the art, "epigenetic changes such as DNA methylation act to regulate gene expression in normal mammalian development. However, promoter hypermethylation also plays a major role in cancer through transcriptional silencing of critical growth regulators such as tumor suppressor genes. Other chromatin modifications, such as histone deacetylation and chromatin-binding proteins, affect local chromatin structure and, in concert with DNA methylation, regulate gene transcription. The DNA methylation inhibitors azacitidine and decitabine can induce functional re-expression of aberrantly silenced genes in cancer, causing growth arrest and apoptosis in tumor cells. These agents, along with inhibitors of histone deacetylation, have shown clinical activity in the treatment of certain hematologic malignancies where gene hypermethylation occurs. Alterations in DNA methylation in cancer, effects on gene expression, and implications for the use of hypomethylating agents in the treatment of cancer." Baylin, S. Nature Clinical Practice Oncology (2005) 2, S4-S11, incorporated in its entirety herein by reference.

"Epigenetic events play a significant role in the development and progression of cancer. Mutations occurring in oncogenes frequently result in a gain of function, while mutations or deletions associated with tumor suppressor genes cause a loss or inactivation of negative regulators. Loss of function, however, can also occur through epigenetic changes such as DNA methylation. 'Epigenetics' refers to heritable changes in gene expression that do not result from alterations in the gene nucleotide sequence. When DNA is methylated in the promoter region of genes, where transcription is initiated, genes are inactivated and silenced. This process is often dysregulated in tumor cells. In cancer, epigentic silencing through methylation occurs at least as frequently as mutations or deletions and leads to aberrant silencing of normal tumor-suppressor function." Id.

"DNA methylation occurs by covalent addition of a methyl group at the 5' carbon of the cytosine ring, resulting in 5-methylcytosine. These methyl groups project into the major groove of DNA and effectively inhibit transcription. In mammalian DNA, 5-methylcytosine is found in approximately 4% of genomic DNA, primarily at cytosine-guanosine dinucleotides ("CpGs"). Such CpG sites occur at lower than expected frequencies throughout the human genome but are found more frequently at small stretches of DNA called CpG islands. These islands are typically found in or near promoter regions of genes, where transcription is initiated. In contrast to the bulk of genomic DNA, in which most CpG sites are heavily methylated, CpG islands in germ- line tissue and promoters of normal somatic cells remain unmethylated, allowing gene expression to occur." Id. Epidermal growth factor receptor (EGFR)-tyrosine kinase inhibitor (TKI) resistance in EGFR wild-type (WT)/KRAS WT non-small cell lung cancers (NSCLCs) is caused by epithelial-to-mesenchymal transition (EMT) by the downregulation of CBLC, a gene involved in EGFR turnover, through promoter DNA methylation. EGFR wild-type NSCLC cells are resistant to EGFR-TKIs. Further, EMT has been attributed to resistance to erlotinib (an EGFR- TKI). We have discovered that epigenetic silencing of CBLC can be an important contributor of erlotinib resistance in EMT and that a CBLC-based predictive biomarker can be useful in providing different therapeutic strategies for reversing drug resistance.

Patients who have activating EGFR mutations (-10%) benefit greatly from targeted agents such as erlotinib, which, unlike traditional chemotherapy, have a low therapeutic index; however, for the majority of patients without EGFR mutations, the drug's benefit is difficult to predict. Identifying the subset of patients who may benefit from this drug will enable personalized therapies for patients with NSCLC.

EMT confers resistance to certain therapies yet may be modulated for preventive and therapeutic purposes. For example, epigenetic silencing of CBLC, as well as CBLC's association with mesenchymal-type cells, can be responsible for EGFR-TKI resistance in EGFR WT NSCLC. Further, CBLC is involved in the process of EMT and EGFR-TKI resistance. As such, CBLC is a useful as a predictive biomarker for EGFR-TKI resistance. Several mechanisms are involved in acquired drug resistance to EGFR-TKI and EMT has been implicated in both intrinsic and acquired drug resistance. Yauch, R.L., et al, Epithelial Versus Mesenchymal Phenotype Determines in Vitro Sensitivity and Predicts Clinical Activity of Erlotinib in Lung Cancer Patients, Clinical Cancer Research 11, 8686-8698 (2005); Uramoto, FL, et al, Epithelial- Mesenchymal Transition in EGFR-TKI Acquired Resistant Lung Adenocarcinoma, Anticancer Research 30, 2513-2517 (2010).

Two major mechanisms are involved in EGFR-TKI resistance, acquired and intrinsic. Acquired resistance occurs in tumors that were initially sensitive (most of which are EGFR mutants) but become resistant after a course of EGFR-TKI therapy. Engelman, J. A., et al., Mechanisms of Acquired Resistance to Epidermal Growth Factor Receptor Tyrosine Kinase Inhibitors in Non-Small Cell Lung Cancer, Clinical Cancer Research 14, 2895-2899 (2008). Intrinsic resistance results from genetic alterations intrinsic to the tumors themselves that confer resistance to therapy. EGFR T790M mutation and MET amplification are the two most common acquired-resistance mechanisms after TKI therapy, accounting for -50% and 5-10% of cases, respectively, KRAS mutations account for -30% of intrinsically resistant tumors. Onitsuka, T., et al., Acquired Resistance to Gefitinib: The Contribution of Mechanisms Other than the T790M, MET, and HGF Status, Lung Cancer 68, 198-203 (2010); Linardou, H.,et al, Assessment of Somatic K-RAS Mutations as a Mechanism Associated with Resistance to EGFR-Targeted Agents: A Systematic Review and Meta-Analysis of Studies in Advanced Non-Small-Cell Lung Cancer and Metastatic Colorectal Cancer, The Lancet Oncology 9, 962-972 (2008). EMT has been implicated as one such mechanism and can function as both an intrinsic- and acquired- resistance mechanism. Yauch, R.L., et al, Epithelial Versus Mesenchymal Phenotype Determines in Vitro Sensitivity and Predicts Clinical Activity of Erlotinib in Lung Cancer Patients, Clinical Cancer Research 11, 8686-8698 (2005), Uramoto, H., et al, Epithelial- Mesenchymal Transition in EGFR-TKI Acquired Resistant Lung Adenocarcinoma, Anticancer Research 30, 2513-2517 (2010).

Casitas B-lineage lymphoma (Cbl) proteins in EGFR regulation and TKI resistance.

CBLC belongs to a class of three E3 ubiquitin ligases (Cbl-1 (or c-Cbl), Cbl-2 (or Cbl-b), and Cbl-3 (or "CBLC")) that negatively regulate RTK signaling, namely TCR in thymocytes and EGFR in non-hematopoietic tissues, by targeting activated tyrosine kinases for degradation¹⁵. Although all three family members can attenuate EGFR signaling, in vivo gene targeting studies have demonstrated that Cbl-1 regulates T-cell signaling in thymocytes and that Cbl-2 is required for regulation of mature T-cell signaling. Swaminathan, G., et al, The Cbl Family Proteins: Ring Leaders in Regulation of Cell Signaling, Journal of Cellular Physiology 209, 21-43 (2006). CBLC is the most recently discovered family member, existing only in mammals. Keane, M.M., et al, Cbl-3: A New Mammalian Cbl Family Protein, Oncogene 18, 3365-3375 (1999); Kim, M., et al., Molecular Cloning and Characterization of a Novel Cbl-Family Gene, cbl-c, Gene 239, 145-154 (1999). Its expression is completely restricted to the epithelial lining of organs, such as the bronchi and bronchioles of the lung (but not in the surrounding alveoli), intestines, esophagus, stomach, endometrium, vagina, and skin (gene atlas and wvvw.proteinatlas.org). Griffiths, E.K., et al, Cbl-3-Deficient Mice Exhibit Normal Epithelial Development, Molecular and Cellular Biology 23, 7708-7718 (2003). In the lung, Cbl-1 expression is confined to macrophages, and Cbl-2 is ubiquitously expressed in both epithelial and non-epithelial cells (www.proteinatlas.org). This may be meaningful biologically, in that there is temporal interplay between Cbl-1 and Cbl-2; furthermore, Cbl-1 accomplishes rapid ubiquitination after EGFR activation, followed by dissociation of Cbl-1 and reassociation of Cbl-2 in internalized receptors for a second peak of ubiquitination, leading to receptor degradation. Pennock, S., et al, A Tale of Two Cbls: Interplay of C-Cbl and Cbl-B in Epidermal Growth Factor Receptor Downregulation, Molecular and Cellular Biology 28, 3020-3037 (2008). Knockdown of both of these genes in embryonic kidney cells (293T) results in complete abrogation of receptor downregulation but not receptor endocytosis. Pennock, S., et al, A Tale of Two Cbls: Interplay of C-Cbl and Cbl-B in Epidermal Growth Factor Receptor Downregulation, Molecular and Cellular Biology 28, 3020-3037 (2008). Hence, CBLC, along with Cbl-2, can play an important role in EGFR regulation in normal lung physiology as well as in cancer cells, whereas Cbl-1 and Cbl-2 interaction may play a more important role in hematologic tissues.

Integrative Analysis of DNA Methylation and Gene Expression in NSCLC Cells Reveals Epigenetic Regulation of EMT

We have performed genome-wide Illumina promoter methylation and Illumina gene expression microarrays on 70 NSCLC cell lines. By integrating the promoter methylation status of genes with that of gene expression for these cell lines, we sought to identify genes whose expression was highly inversely correlated with DNA methylation. By using the Spearman rho cutoff of -0.6, we identified 342 probes, corresponding to 273 genes that we termed genes regulated by methylation (GRM). GeneGo pathway analysis for this GRM panel revealed that the top two biological processes are related to the EMT process. Indeed, when we characterized these cells with EMT markers (CDHl, VIM, TWIST, ZEBl, CDH2, and FN), we found that by unsupervised clustering, the cell lines were able to segregate into epithelial and mesenchymal- like groups (Figure 1A). By using correlation matrix analysis, we found a group of 36 genes whose methylation status was co-methylated in the mesenchymal-like cells (E-cadherin low cells). We termed this group of genes the "EMT Core GRM" (Figure IB). Gene expression of the GRM also clusters the cell lines into epithelial and mesenchymal types (Figure 1C), which is also seen in tumors (Figure ID). Hence, EMT can be an epigenetically-regulated process through coordinated methylation of genes involved in the EMT process. GRM Signature Predicts for Cellular Response to Erlotinib and Docetaxel Treatment.

EMT has been implicated in therapeutic resistance. We tested the relationship between the NSCLC GRM and cell response to several clinical drugs and radiation. Using beta uniform mixture modeling, we segregated the cells into resistant and sensitive types based on IC50 (drug) or SF2 (radiation) values. We then conducted unsupervised clustering of these cell lines and found that the EMT core GRM was able to predict response to erlotinib and docetaxel. The methylation signature clustered resistant and sensitive cell lines based on the degree of methylation of each of the genes and independently clustered these according to mesenchymal and epithelial phenotypes, respectively (Figure 2). These data show that the EMT core GRM is a robust signature to identify phenotypic and functional epithelial and mesenchymal cell types in NSCLC.

CBLC Methylation is Associated with EMT and Erlotinib Resistance in EGFR WT Cells in the Context of KRAS Mutation and EGFR and MET Amplification.

Among the 36 genes co-methylated and regulated in the EMT core GRM, we found CBLC methylation to be strongly associated with the mesenchymal cell type and closely related to E-cadherin expression (Figure 3 A, 3B). CBLC is located on chrl9ql3.2 and has a CpG island located at the promoter region (UCSC Genome Browser). Cbl-1 is located on chrl lq23.3 and also contains a CpG island, whereas Cbl-2 is located on chr3ql3.11 but lacks a CpG island. In the 70 NSCLC cell lines, we found no methylation for either Cbl-1 or Cbl-2; but for CBLC, 61% of the cells (n=43) were fully methylated, 19% (n=13) were hemimethylated, and 20% (n=14) were unmethylated. We found CBLC methylation to be strongly inversely proportional to gene expression (Figure 3C). We validated the array data by bisulfite pyrosequencing and found good correlation between the array and the pyrosequencing results (Figure 3D). Furthermore, treating mesenchymal cells (HI 299) with 1 micromolar 5'aza-2-deoxycytidine (DAC) reversed the degree of promoter methylation and gene re-expression (Figure 3E, 3F). These results suggest that CBLC expression is regulated by DNA methylation in lung cancer cells.

Since Cbl proteins control EGFR signaling and degradation, we next determined whether CBLC methylation itself can predict for erlotinib response after excluding EGFR mutants (n=6), which are exceptionally sensitive to erlotinib (data not shown). Indeed, for all of the EGFR WT cell lines (n=54), we found that CBLC methylation significantly predicted erlotinib resistance, and cells with unmethylated CBLC exhibited better erlotinib sensitivity (p=0.018) (Figure 4A). However, if we restricted our analysis to just the KRAS mutant (n=20) or the MET amplified (n=10) cell lines, genetic alterations known to be predictive for resistance to EGFR-TKI, the association of erlotinib sensitivity with CBLC methylation status was no longer significant (Figure 4B, 4C).

Biomarkers-Integrated Approaches of Targeted Therapy for Lung Cancer Elimination (BATTLE) Program.

In this program, which included phase II trials of biomarkers that may predict sensitivity to erlotinib, sorafenib, and vandetanib, we enrolled NSCLC patients with advanced chemotherapy-refractory disease. Baseline biopsies and global gene expression profiling were performed for each patient. A total of 139 core biopsies were processed for gene expression profiling by using the Affymetrix HG 1.0 ST array. The primary end point of the study was 8- week disease control (8wkDC). Using principal component analysis (PCA) and a logistical regression model, we tested the entire set of the GRM signature (342 probes) and found that it trended toward predicting poorer 8wkDC in the erlotinib arm (p=0.079) but not in the vandetinib or sorafenib arms (p=0.215 and p=0.756, respectively) (Figure 5). When we confined the analysis to individual genes, we found CBLC also trended toward predicting 8wkDC only in the erlotinib arm (p=0.149) but not in the vandetinib or sorafenib arms (p=0.647 and p=0.782, respectively) (Figure 5).

EXAMPLE I

Integrative Analysis of DNA Methylation and Gene Expression Profiles

Identifies Epigenetic Regulation of Genes

Associated With Epithelial-To-Mesenchymal Transition (EMT)

Background

Abnormal expression of genes in cancer comes from the aberrant regulation of gene transcription through various mechanisms. We sought to determine the genes whose expressions are largely regulated by DNA methylation in a large panel of non-small cell lung cancer cell lines.

Methods

We performed genome-wide DNA methylation and gene expression profiling in 74 non- small cell lung cancer (NSCLC) cell lines using the Infinium HumanMethylation27 beadchip and the Affymatrix U133 gene chip, respectively. To identify genes that are potentially regulated by DNA methylation, we performed integrative analysis between the two datasets to find genes whose expression is strongly inversely correlated to the degree of DNA methylation.

Results

Using a FDR cutoff of 0.5% (p<0.0001) and a Spearman correlation cutoff of -0.65, we identified 241 CpG site methylation probes, corresponding to 147 unique genes, that are strongly inversely correlated with gene expression. Unsupervised cluster analysis of these genes segregated these cell lines into two distinct groups. By pathway analysis, a subset of genes that separate the two groups is associated with Epithelial-to-Mesenchymal Transition (EMT). Indeed we find that the DNA methylation of these genes is predictive of the mesenchymal phenotype. We independently verified these findings on a separate gene expression platform using the Illumina WG6-v2 microarray platform.

Conclusions

By identifying genes that are regulated through promoter CpG island methylation, we discovered that these genes define cells with epithelial or mesenchymal phenotypes. This shows that DNA methylation can be an important regulator of EMT. Given the putative role of EMT in metastatic spread and drug resistance, these results show that DNA methylation can be a useful biomarker for clinical outcome and therapeutic response in NSCLC and that epigenetic therapies could be used as targeted reversal of EMT and drug resistance.

EXAMPLE II

Integrative Analysis of DNA Methylation and Gene Expression Identifies a DNA

Methylation Signature Associated With Erlotinib Resistance in EGFR Wild Type Non-Small Cell Lung Cancer Cells

Background

While EGFR mutation status dictates sensitivity to EGFR Tyrosine Kinase Inhibitors (TKI) in non-small cell lung cancers (NSCLC), the molecular pathways responsible for sensitivity or resistance to TKIs in EGFR wild type cells are relatively unknown. We sought to determine if there is a DNA methylation profile that can help predict for TKI response in a panel of NSCLC cell lines.

Methods

As a screen to find genes whose gene expression is strongly regulated by DNA methylation, we performed integrative analysis between DNA methylation profiles of 74 NSCLC cell lines using the Infmium HumanMethylation27 beadchip and gene expression changes in a subset of these cells using both Affymatrix U133 and the Illumina Expression Beadchip (WG6- v2) microarray platforms. To find the gene signature that correlates with responsiveness to TKIs, we performed two sample t-test for each gene and applied Beta Uniform Mixture modeling for adjusting multiple comparison. Using appropriate False Discovery Rate (FDR) cutoffs, we identified gene signature that are associated with erlotinib responsiveness in EGFR wild type cells.

Results

Using FDR= 0.5 % (corresponding p < 0.0001) and Spearman correlation (rho) cutoff of -0.65, we identified 147 unique genes that are significantly inversely correlated with the degree of methylation. We then correlated these genes with Erlotinib resistance. We found the DNA methylation of a subset of these genes to be highly related to the Epithelial-to-Mesenchymal Transition (EMT) phenotype of these cells based on the expression of E-Cadherin and β-catenin proteins using Reverse-Phase Protein Arrays. We independently confirmed these results using the Illumina WG6-v2 microarray platform.

Conclusions

We have found a gene panel whose expression is largely regulated by DNA methylation in a panel of NSCLC cell lines. By analyzing the degree of methylation of these genes with drug responsiveness to erlotinib in EGFR wild type cells, we have discovered a DNA methylation signature for EMT that predicts for TKI responsiveness which can serve as a useful biomarker to identify TKI resistance in lung cancer patients.

EXAMPLE III

Prophetic

CBLC Effects of Knockdown and Overexpression on EGFR Turnover and EGFR-TKI Sensitivity in the Context of CBLC Methylation

Based on our preliminary results, we found that CBLC is a methylated gene whose degree of methylation predicts EGFR-TKI in cell lines in the context of known pathways of resistance. Our working hypothesis is that epigenetic regulation of CBLC affects EGFR turnover and signaling, thereby conferring EGFR-TKI resistance. To test this, the impact of CBLC knockdown or overexpression on EGFR turnover and TKI sensitivity in NSCLC cell lines is evaluated as follows: Rationale

The coordinated downregulation of epithelial-related genes involved in EGFR signaling confers EGFR resistance to TKIs. Since CBLC, an E3 ubiquitin ligase involved in receptor downregulation after receptor activation, is often undetectable in more than 60% of lung cancers (www.proteinatlas.org), CBLC downregulation by DNA methylation results in EGFR-TKI resistance.

Research

We will examine EGFR ubiquitination, turnover rates, and erlotinib sensitivity in EGFR WT/KRAS WT/MET unamplified NSCLC cells with or without CBLC. We will use three EGFR WT/KRAS WT/MET unamplified cell lines that are CBLC unmethylated and erlotinib sensitive (US) and three that are CBLC methylated and erlotinib insensitive (MIS) for these experiments. We will evaluate EGFR ubiquination, turnover rates, and erlotinib sensitivity by performing long term knockdown of CBLC using shRNA in the US cells and CBLC overexpression in the MIS cells with retroviral transduction. Our preliminary data showed that CBLC methylation status and erlotinib sensitivity are no longer associated in the background of either KRAS mutation or MET amplification. To study this further, the same cell lines will be further transduced with vectors containing KRAS mutation (G34T) or with expression vectors for MET. We will determine whether CBLC's effect on erlotinib sensitivity or receptor turnover is altered with the introduction of KRAS mutation or MET overexpression. We will study EGFR phosphorylation and CBLC interaction by these alterations.

Expected Outcomes

CBLC is a major player in EGFR activity. Also, erlotinib sensitivity is associated with the methylation status of CBLC. The role of KRAS mutation in the context of CBLC and EGFR-TKI resistance will be determined and can be used for further exploration of potential mechanisms underlying this process, leading to identification of additional therapeutic targets to modulate EGFR WT/KRAS mutants in EGFR-TKI resistance. Note, the CBLC effect on EGFR turnover could also be dependent on other unknown factors that are also turned off during the EMT process, along with CBLC. If there is no CBLC effect on EGFR turnover when CBLC is overexpressed in the mesenchymal cells, MET should be induced in the mesenchymal NSCLC cells by expressing mir200a, a known inducer of MET, in conjunction with CBLC overexpression. Gregory, P.A., et al, The Mir-200 Family and Mir-205 Regulate Epithelial to

Mesenchymal Transition by Targeting ZEB1 and SIP1, Nature Cell Biology 10, 593-601 (2008).

EXAMPLE IV

Prophetic

The Effect of CBLC on EMT

To confirm whether the association between CBLC and the EMT process is in NSCLC tumors, we will first examine whether CBLC is associated with the EMT phenotype using resected NSCLC tumors. Then, to confirm whether CBLC actually contributes to the epithelial or mesenchymal phenotype, we will show that CBLC knockdown can be a driver of EMT and that overexpression in CBLC null mesenchymal cells can cause mesenchymal-to-epithelial transition (MET) to occur.

Rationale

CBLC is 1 of 36 genes found to be co-methylated in the EMT core GRM. Its methylation status is strongly associated with mesenchymal-type cells. Since CBLC is a multivalent adaptor protein that is capable of interacting with a plethora of proteins involved with multiple receptor signaling pathways, Swaminathan, G., et al, The Cbl Family Proteins: Ring Leaders in Regulation of Cell Signaling, Journal of Cellular Physiology 209, 21-43 (2006), CBLC could be a critical link in the pathways leading to EMT or MET. So far, CBLC's involvement in EMT has not been described. CBLC expression can modulate processes involved in EMT.

Research Determine the relationship of CBLC methylation, immunohistochemical staining, and EMT markers in lung tumors.

In our preliminary data, we found that low CBLC gene expression is associated with mesenchymal cells in cell lines and tumors. To prove that CBLC is related to EMT in NSCLC, we will determine histologically the relationship between CBLC expression and EMT markers in archived tumor specimens. We have identified 200 resected NSCLC from the MDACC tissue bank, stage I-IIB, half adenocarcinoma (n=100), and half squamous cell carcinoma (n=100), with clinical annotation for tumor recurrence, adjuvant therapy, smoking status, and overall survival. IHC analysis will be performed for five markers (CBLC, E-Cad, vimentin, EGFR, and MET) using tissue microarrays (TMA) of these tumors. To prove that CBLC methylation may be responsible for the 64% of human lung cancers not expressing CBLC (wvvw.protematlas.org), DNA will be extracted from the formalin-fixed, paraffin-embedded (FFPE) specimen to perform bisulfite pyrosequencing. An IRB-approved protocol is currently in place (LAB09-0841) to allow for the study of these samples.

Determine Whether CBLC can be a Driver of EMT in NSCLC Cell Lines.

In the cellular background of low to no CBLC expression in mesenchymal-like cells, we will determine whether overexpression of CBLC can induce MET by examining morphologic changes, protein expression changes of EMT markers (E-Cad, vimentin, fibronectin), and invasion/migration alterations. To determine whether CBLC downregulation can induce EMT, we will stably knock down CBLC using shRNA in epithelial cell lines. We will also determine whether CBLC acts upstream or downstream of known drivers of EMT (transcription factors such as TWIST, or growth factors such as TGF-beta), by overexpression of CBLC, the addition of TGF beta, or overexpression of TWIST.

Outcomes

There is a relationship between EMT in tumors and CBLC downregulation. We expect to determine that CBLC downregulation by DNA methylation in tumors is in line with complete gene silencing in tumors and to determine the degree that DNA methylation controls gene expression by TMA. Since we already see an association of expression of CBLC with EMT markers in gene expression microarrays, we believe that this experiment will show the association of EMT with CBLC expression. In our cell culture experiments, we will determine whether CBLC downregulation is a mere consequence of EMT or whether it can be a driver of EMT.

We propose that CBLC drives EMT and is not just a "passenger." However, since the 36-gene EMT core GRM signature is co-repressed by DNA methylation, some other master regulatory switch may control EMT and regulate the expression of these genes. However, the evidence we have suggests that this is not necessarily the case. CDH1 is part of the 36-gene signature, which, along with other genes such as SPINT1, are known inducers of EMT. CBLC, being a gene with a possible widespread role in the cell, could possibly regulate EMT as well. EXAMPLE V

Prophetic

The Predictive Value of CBLC Expression

We have demonstrated that CBLC methylation predicts erlotinib resistance in EGFR/KRAS WT cell lines and that CBLC gene expression can very likely predict erlotinib sensitivity in pts. We will develop a high-throughput and sensitive method to detect CBLC gene expression and protein expression in an initial cohort of NSCLC patients that can be adapted as a biomarker for EGFR-TKI resistance using specimens from a completed randomized phase III study of erlotinib versus vandetanib. We will clinically validate the predictive value of CBLC expression for progression-free survival (PFS) in patients included in the phase III Zactima Efficacy when Studied versus Tarceva (ZEST) trial.

Rationale

We have preliminary evidence that CBLC expression is a predictive of EGFR-TKI response in EGFR WT/KRAS WT cell lines and in tumors from patients in the BATTLE trial. However, to develop CBLC as a robust biomarker to help select EGFR/KRAS WT patients to receive EGFR-TKI, a necessary step is to develop and validate robust Clinical Laboratory Improvement Amendments (CLIA)-certifiable assays that can be performed with use of FFPE tissue. For this purpose, we will assess the performance of quantifying CBLC expression with use of a quantitative nuclease protection assay (qNPA) along with the gold standard IHC assay in a large panel of resected NSCLC tissues and NSCLC cell lines. Part of this work can be linked to the objectives as set forth in Example IV. If validated, we hope to apply this technology to samples from a large phase III study comparing erlotinib and vandetinib in previously treated pts with advanced NSCLC.

Develop CBLC as a biomarker with use of a CLIA-certifiable assay of FFPE tissue from a large collection of specimens from cell lines and resected NSCLC.

We will use High Throughput Genomics (HTG) 96-well ArrayPlate™ assay with qNPA to assess CBLC expression. This new technology was designed for detecting small changes in gene expression levels, in both frozen and FFPE samples, without the need for RNA extraction, labeling, or amplification. Roberts, R.A., et al, Quantitative Nuclease Protection Assay in Paraffin-Embedded Tissue Replicates Prognostic Microarray Gene Expression in Diffuse Large- B-Cell Lymphoma, Laboratory Investigation 87, 979-997 (2007); Rimsza, L.M., et al, Gene Expression Predicts Overall Survival in Paraffin-Embedded Tissues of Diffuse Large B-Cell Lymphoma Treated with R-CHOP, Blood 112, 3425-3433 (2008) Only 1/4 of a 3- to 5-micron tissue slice is required per sample. Two sets of samples will be applied to the HTG ArrayPlate platform: resected NSCLC tissues and NSCLC cell lines. The same 200 surgically resected NSCLC specimens from Example IV will be used. FFPE pellets from 5- x 5-micron sections will be measured in triplicate with use of the HTG ArrayPlate assay. Normalization will be done with use of five housekeeping genes and a negative control. These data will be compared with protein expression by IHC analysis, which will be performed with use of an automated microscope-based Ariol® system to assess express scores for membrane, cytoplasm, and nucleus ranging from 0 to 100. The sum of the membrane, cytoplasmic, and nuclear scores will be used to generate a total composite score.

Along with these two assays, we will conduct methylation analyses of CBLC in parallel. As shown in the preliminary data, our pyrosequencing results showed a high level of correlation between the methylation percentage found by pryosequencing and the array results. We will, however, adopt the pyrosequencing assay for a small amount of clinical samples. The purpose here is fundamentally different from the objectives of Example IV, for which we will prepare DNA from a larger piece of tissue. We will minimize sample loss by using the Zymo EZ- methylation Direct Kit, which obviates the need for DNA extraction by performing tissue digestion and bisulfite treatment in one step. Since this minimizes sample loss, it is adapted for use with small amounts of FFPE tissue, and the signal can be detected from as few as 10 cells. Sensitivity and specificity will be determined by comparing the results to larger amounts of sample, which will reveal the limit of detection, the effect of bisulfite treatment on small samples, and the need for whole genome amplification (WGA) techniques. We will compare the methylation status of CBLC with IHC and HTG results, with the expectation of an inverse relationship of CBLC gene/protein expression and methylation.

Validate the predictive value of CBLC as a biomarker for PFS in the ZEST trial.

This completed phase III trial compared the efficacy of vandetanib (an oral inhibitor of VEGFR, EGFR, and RET) with erlotinib in pts with advanced, previously treated NSCLC. The primary objective was to show superiority in PFS for vandetanib vs. erlotinib. A total of 1240 pts were randomized to receive vandetanib or erlotinib. There was no difference in PFS between the 2 arms. Natale, R.B., et al., Vandetanib Versus Erlotinib in Patients with Previously Treated Advanced NSCLC: A Randomized, Double-Blind Phase III Trial (ZEST), Journal of Thoracic Oncology 4, S358-S358 (2009). We hypothesize that CBLC expression will be associated with PFS in EGFR/KRAS WT patients in the erlotinib but not in the vandetanib arm, as we found in the BATTLE patients, but no conclusions can be reached by the limited patient numbers. We will use the HTG ArrayPlate assay and IHC analysis as described above.

Statistical analysis

From the 1240 patients included in the trial, we will obtain FFPE samples in 310 (25%) with known EGFR and KRAS status. Based on the overall rate of mutation in the trial, we expect 70% of patients (217 of 310) to be EGFR/KRAS WT, 50% (n=108) of whom should be in the erlotinib arm. Although we would not be able to account for MET amplification, we expect less than 10% to be MET amplified. In our preliminary results, cells with MET amplification mixed in were not sufficient to degrade the predictive value of CBLC for erlotinib sensitivity in the overall group of cell lines. The median PFS for this patient population is 2.5 months. With use of a median gene expression score of CBLC to dichotomize the patients into low (group I) and high (group II) groups with 54 patients each, a log rank test will achieve 82% power at a 2- sided 0.05 significance level to detect a 20% PFS difference between group I (40%>) and group II (60%) at 2.5 months. This difference corresponds to a hazard ratio of 1.8. Association among various continuous and discrete clinic-pathologic characteristics will be assessed by exploratory data analysis. Comparison among subgroups will be performed by using chi-square or Fisher's exact test for discrete variables and by t-test, Wilcoxon rank sum, or Kruskal-Wallis test for continuous variables. PFS will be estimated by the Kaplan-Meier method. A Cox model will be fitted to estimate the effect of the CBLC gene signature score and other covariates on time-to- event end points with continuous and discrete biomarker expressions as appropriate. The CBLC gene score will also be explored for the vandetanib group.

Outcome

CBLC is a validated biomarker for testing with use of the HTG ArrayPlate assay and IHC analysis with validated, commercially available antibodies. We will determine whether the readout for the expression of CBLC by HTG is predictive of IC50 in the cell lines. If so, we will adopt the technology. However, if no correlation is seen, we will use IHC analysis. In the event that correlation for gene expression and protein expression by IHC analysis is also not adequate, we will assay for methylation of CBLC by pyrosequencing, provided that we are able to develop the methylation assay as proposed above. Since this is not an established high- throughput method, developing the predictive assay for CBLC gene expression with use of established and validated methods, such as the HTG platform and automated IHC, is our first choice.

Claims

CLAIMS We claim:

1. A biomarker comprising the gene CBLC.

2. An assay comprising the EMT-GRMs genes.

3. An assay comprising the EMT Core GRM of Figure IB.

4. A diagnostic assay for assessing prognosis or likelihood of response to erlotinib docetaxel comprising the methylation of CBLC.

5. An assay comprising any of the DNA methylation biomarkers shown or described Figures 1 through Figures 8.