Gene Array and Fluorescence in Situ Hybridization Biomarkers of Activity of Saracatinib (AZD0530), a Src Inhibitor, in a Preclinical Model of Colorectal Cancer

Published OnlineFirst August 3, 2010; DOI: 10.1158/1078-0432.CCR-10-0066 Published OnlineFirst on August 3, 2010 as 10.1158/1078-0432.CCR-10-0066

Cancer Therapy: Preclinical Clinical Cancer Research Gene Array and Fluorescence In situ Hybridization Biomarkers of Activity of Saracatinib (AZD0530), a Src Inhibitor, in a Preclinical Model of Colorectal Cancer

John J. Arcaroli1,2, Basel M. Touban1,2, Aik Choon Tan1,2, Marileila Varella-Garcia1, Rebecca W. Powell1,2, S. Gail Eckhardt1,2, Paul Elvin3, Dexiang Gao3, and Wells A. Messersmith1,2

Abstract Purpose: To evaluate the efficacy of saracatinib (AZD0530), an oral Src inhibitor, in colorectal cancer (CRC) and to identify biomarkers that predict antitumor activity. Experimental Design: Twenty-three CRC cell lines were exposed to saracatinib, and baseline gene expression profiles of three sensitive and eight resistant cell lines in vitro and in vivo were used to predict saracatinib sensitivity in an independent group of 10 human CRC explant tumors using the gene array K-Top Scoring Pairs (K-TSP) method. In addition, fluorescence in situ hybridization (FISH) and immunoblotting determined both Src gene copy number and activation of Src, respectively. Results: Two of 10 explant tumors were determined to be sensitive to saracatinib. The K-TSP classifier (TOX>GLIS2, TSPAN7>BCAS4, and PARD6G>NXN) achieved 70% (7 of 10) accuracy on the test set. Evaluation of Src gene copy number by FISH showed a trend toward significance (P = 0.066) with respect to an increase in Src gene copy and resistance to saracatinib. Tumors sensitive to saracatinib showed an increase in the activation of Src and FAK when compared with resistant tumors. Conclusions: Saracatinib significantly decreased tumor growth in a subset of CRC cell lines and explants. A K-TSP classifier (TOX>GLIS2, TSPAN7>BCAS4, and PARD6G>NXN) was predictive for sensitivity to saracatinib. In addition, increased activation of the Src pathway was associated with sensitivity to saracatinib. These results suggest that FISH, a K-TSP classifier, and activation of the Src pathway have potential in identifying CRC patients that would potentially benefit from treatment with saracatinib. Clin Cancer Res; 16(16); 4165–77. ©2010 AACR.

Src is a pleiotropic nonreceptor tyrosine kinase that believed that Src affects growth during earlier stages of plays a major role in the development and progression tumorigenesis, but alters other processes such as adhesion, of many human cancers (1–4). It has been well estab- invasion, and motility during later stages of tumor lished that one of the main functions of Src is regulating progression. cellular proliferation through its interaction with many Many studies on Src have revealed the complex molec- different growth factor receptors such as the epidermal ular mechanisms and interactions with many different growth factor receptor (EGFR), fibroblast growth factor, substrates that are important in altering tumor cell behav- hepatocyte growth factor receptor, and insulin growth fac- ior. In particular, Src binds and activates focal adhesion tor receptor (5–8). In addition to proliferation, elevated kinase (FAK), a kinase that facilitates changes in the cyto- levels and activation of Src lead to the disruption of focal skeleton by disrupting cell-cell and cell-matrix interactions adhesions and adherens junctions resulting in an increase (9, 11). Both Src and FAK have been reported to be simul- in motility and invasiveness of cancer cells (2, 9, 10). It is taneously increased in many human malignancies (12). In addition, Src activates Stat-3, resulting in a subsequent increase in the transcriptional activation of vascular endo- Authors' Affiliations: 1Division of Medical Oncology, University of thelial growth factor, an important regulator of tumor Colorado Denver, and University of Colorado Cancer Center, Anschutz Medical Campus; and 2Gastrointestinal Cancer Program and angiogenesis (13, 14). Recently, Src has been shown to ac- Experimental Therapeutics Program, University of Colorado Cancer tivate Trask in many human epithelial cancers and is be- Center, Aurora, Colorado; and 3AstraZeneca, Alderley Park, lieved to be activated during detachment and shedding of Macclesfield, United Kingdom, and University of Colorado Denver; Research Institute, The Children's Hospital, Aurora, Colorado tumor cells (15). Several selective Src kinase inhibitors Corresponding Author: Wells Messersmith, 12801 East 17th Avenue, have been developed and have shown potential in inhibit- MS 8117, Aurora, CO 80045. Phone: 303-724-3808; Fax: 303-724- ing the growth of tumor cells as well as altering the meta- 3892; E-mail: [email protected]. static phenotype evidence by a reduction in migration and doi: 10.1158/1078-0432.CCR-10-0066 invasiveness of cancer cells. A study by Huang et al. (16) ©2010 American Association for Cancer Research. tested dasatinib (Src/Abl inhibitor) against 23 breast

www.aacrjournals.org 4165

Arcaroli et al.

gested to play a central role in the induction and progres- Translational Relevance sion of tumors, especially in tumors with high malignant potential (22). When compared with the normal mucosa In colorectal cancer, studies have shown that the ex- of the colon, Src activity is elevated in adenomatous pression of Src is elevated in premalignant lesions and polyps (23). In addition, Talamonti et al. (24) showed subsequently increases as tumors progress to meta- Src activity to be the greatest in liver metastatic lesions static lesions. The objective of this study was to evalu- showing a role of Src in CRC progression. Finally, Src is ate the efficacy of saracatinib, a Src inhibitor, and to an independent prognostic indicator at all stages of CRC identify predictive biomarker(s) of sensitivity or resis- (25). Taken together, increased levels of Src expression tance to saracatinib in a colorectal cancer human and activation have been shown to play a very important explant preclinical model. These data show that saraca- role in the sequential progression from premalignant le- tinib had antitumor activity in a small subset of tu- sions of the colon to metastasis. mors. Additionally, fluorescence in situ hybridization There have been numerous developments in the CRC and a K-Top Scoring Pairs gene classifier derived from field over the past year, the most important being studies baseline gene array data had predictive power for sar- showing that patients with codon 12/13 KRAS mutations acatinib sensitivity. Because clinical trials of Src inhibi- do not benefit from the EGFR inhibiting antibodies, cetux- tors with unselected patients have been disappointing imab and panitumumab (26, 27). In addition, dual use of thus far, these predictive biomarkers may prove to be EGFR- and vascular endothelial growth factor–targeting valuable in the clinical setting by preselecting patients antibodies has been found to harm patients (27, 28). who may benefit from saracatinib therapy. These findings show that it is extremely important to have an understanding of the specific genetic/signaling defects of a particular tumor. To this end and in conjunction with the observations that Src may potentially be a significant cancer cell lines, and a genomic signature was developed contributor in CRC, we set out to identify possible that predicted dasatinib susceptibility. The 161-gene pre- mechanisms and potential predictive biomarkers of sensi- dictor was refined to a 6-gene predictor consisting of tivity and resistance to saracatinib (a Src kinase inhibitor ANXA1, CAV1, CAV2, EPHA2, PTRF, and IGFBP2, and this currently in phase II clinical development). These studies was associated with a “triple-negative” pattern of estrogen show that saracatinib has activity in a CRC subset, which receptor (ER), progesterone receptor (PR), and human may be identified by the use of fluorescence in situ hybrid- EGFR 2 breast cancer when tested against archival breast ization (FISH), a K-Top Scoring Pairs (K-TSP) classifier, cancer specimens. and Src activation status. The Src gene is located on chromosome 20q12-q13. It is widely accepted in the field that the Src gene is not ampli- Materials and Methods fied. However, a review of the Sanger database (http:// www.sanger.ac.uk/) does suggest that there may be an in- Cell lines and culture crease in Src gene copy in some colorectal cancer (CRC) Human CRC cell lines were obtained from the American cell lines. Type Culture Collection. Cells were cultured in RPMI sup- Previously, a human pancreas cancer explant model was plemented with 10% fetal bovine serum, 1% nonessential developed whereby excess nondiagnostic tissue is ob- amino acids, and 1% penicillin/streptomycin, and were tained from the operating room and then placed directly maintained at 37°C under an atmosphere containing into mice (17). The advantage of this model is that there is 5% CO2. The cells were routinely screened for the presence a continuous supply of human tumor tissue to develop of Mycoplasma (MycoAlert; Cambrex BioScience) and and to validate novel correlative studies with new drugs were exposed to saracatinib (Astra Zeneca) when they and drug combinations. Most importantly, the tumor cells reached ∼70% confluence. All cell lines were tested and are not placed in culture and appear to retain most of authenticated in the University of Colorado Cancer Center the genetic features of the original tumor (17). This model DNA Sequencing and Analysis Core. CRC cell line DNA has been used to identify predictive biomarkers of sensitivity was tested using the Profiler Plus kit (Applied Biosystems). to two different Src inhibitors in pancreatic cancer (18, 19). The data obtained were compared with American Type We have applied the same concept in CRC to identify predic- Culture Collection data to ensure the cell lines have not tive biomarkers of sensitivity and resistance to saracatinib. changed. The cell lines were last tested in September 2009. CRC is one of the most common malignancies in West- ern countries, with an annual incidence of ∼300,000 in Evaluation of cytotoxicity the United States and Europe (20). About 50% of patients Cytotoxic effects were determined using the sulforhoda- with locally advanced or metastatic disease will ultimately mine B assay (13). Briefly, cells in logarithmic growth die of their disease. There is strong evidence that aberrant phase were transferred to 96-well flat-bottomed plates Src kinase activity is a critical component in all stages of with lids. Cell suspensions (100 μL) containing 500 to CRC (21). The Src pathway is activated in 80% of all 3,000 viable cells were plated into each well and incubated human colon tumors, and its activation has been sug- overnight before exposure to saracatinib for 72 hours.

4166 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

After drug treatment, medium was removed and cells then analyzed by high-resolution melting with 25 data ac- were fixed with cold 10% TCA for 30 minutes at 4°C. Cells quisitions per degree of temperature increase, from 40°C were then washed with water and stained with 0.4% sul- to 90°C. The Lightcycler 480 Gene Scanning software forhodamine B (Fisher Scientific) for 30 minutes at room using the known WT control samples for baseline calcula- temperature. The plates were washed with 1% acetic acid tion were used for these analyses. Src and BRAF mutations followed by stain solubilization with 10 mmol/L Tris. The were analyzed by PCR amplification and direct sequencing plate was then read on a plate reader (Biotek Synergy 2) of the products as previously described (29, 30). set at an absorbance wavelength of 565 nm. Cell proliferation curves were derived from the raw absorbance data. Development of Src probe for FISH A Src FISH probe was developed to evaluate Src gene CRC cell line xenografts copy number. The PAC clone RP4-681F2 (http://www. SW620, SW480, LS174T, and LS180 were injected into ensembl.org), which encompasses the SRC gene was the left and right flanks of 4- to 6-week-old female athymic selected for this study. The clone was obtained from (nu+/nu+) mice (Harlan Laboratories). Mice were random- CHORI and streaked on agar plates. Single-cell colonies ized into the treatment group (saracatinib) or vehicle were selected, and the PAC clone was validated using group when tumor volumes reached ∼200 mm3.Mice PCR (forward, 5′-AGTGTCTGACTTCGACAACGCCAA-3′; were treated daily with saracatinib 50 mg/kg or vehicle reverse, 5′-GAAGGAGACAAGGAAAGGAGGGAA-3′). After by oral gavage. Mice were monitored daily for signs of validation, the clone was expanded and purified using the toxicity, and tumor size was evaluated twice per week by Qiagen QIAamp DNA Mini kit. Repli-G kit (Qiagen) was caliper measurements using the following formula: tumor used as per manufacturer's instructions for whole genome volume = (length × width2) × 0.52. amplification. The DNA inserts were then labeled by nick translation (Vysis Nick Translation kit) with SpectrumRed- Flow cytometric analysis of cell cycle distribution conjugated dUTPs. Labeled DNA was ethanol precipitated Cells (2 × 105 per well) were seeded in six-well plates with herring sperm and human Cot-1, and each DNA and allowed to attach for 24 hours. The cells were then pellet was resuspended in 20 μL of c-DenHyb (Insitus Bio- treated with saracatinib (1 μmol/L) for 24 and 48 hours, technologies). The SRC probe was combined with the washed in PBS, and resuspended in Krishanaposs stain. Aquarius LPE 20G probe (Cytocell) as a control for the The cells were stained for 24 hours at 4°C before analysis centromere on chromosome 20. by flow cytometry at the University of Colorado Cancer Center Flow Cytometry Core Facility. Evaluation of Src gene copy in CRC cell lines and explants Induction of apoptosis Slides were incubated at 56°C for 4 hours, soaked in Cells (1,000/well) were plated in 96-well, white-walled CitriSolv, and washed thrice for 10 minutes each. After plates and allowed to attach for 24 hours. The cells were drying, the slides were incubated in 2xSSC at 75°C for then treated with 1 μmol/L saracatinib and apoptosis was 20 minutes and then in 0.25 mg/mL proteinase K at 45°C determined by the measurement of caspase-3 and caspase- for 20 to 25 minutes. After washing and dehydration, 7 activity using a luminometric Caspase-Glo-3/7 assay the specimens were sequentially placed in 70% ethanol, (Promega) according to the manufacturer's protocol and 85% ethanol, and 100% ethanol for 2 minutes, and al- read using a 96-well plate reader (Biotek). lowed to air dry. The appropriate volume of the SRC/ CEP20 probe was applied followed by DNA denaturation Sequencing of Src, KRAS, and BRAF (85°C for 15 min) and hybridization (37°C for 36-48 h). We performed sequencing to examine the mutational Posthybridization washes were done in 2xSSC/0.3% NP40 status of Src, KRAS, and BRAF in CRC cell lines and ex- at 74°C for 2 minutes and then in 2xSSC for 2 minutes plants. DNA was isolated using the Qiagen DNA extrac- each. After dehydration, 14 μLof4′,6-diamidino-2- tion kit (Qiagen). KRAS mutations were analyzed by the phenylindole/anti-fade (0.3 μg/mL in Vectashield mount- University of Colorado Cancer Center Pathology Core using medium) was added to the slide. Analysis was done on ing a high-resolution melting temperature method with an epifluorescence microscope using single interference custom primers and the Roche LC480 real-time PCR ma- filter sets for green (FITC), red (Texas red), and blue chine. Briefly, template DNA was tested by high-resolution (4′,6-diamidino-2-phenylindole) as well as dual (red/ melting analysis using a Lightcycler 480 real-time PCR in- green) and triple (blue, red, green) band pass filters. At strument (Roche Applied Science). Approximately 60 ng least 20 metaphase spreads (cell lines) and 100 interphase of template DNA, wild-type (WT) control DNA, and nuclei (cell lines and explants) were analyzed for Src gene mutant control DNA were amplified on the Lightcycler copy number and CEP 20 signals. Continuous variables 480 instrument using a high-resolution melting master for Src gene copy number were categorized, and a con- mix (Roche Diagnostics), with the RASO1 and RASA2 pri- servative approach was used for cell lines and explants mers and 1.75 mmol/L MgCl2 in 10 μL on a 96-well plate, in determining a gain in Src gene copy number. A Src copy using a two-step cycling program (95° melting, 72° an- number of ≥3 was considered a gain, and <3 was consid- nealing and extension) for 45 cycles. PCR products were ered no gain.

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4167

Arcaroli et al.

CRC explant xenograft model K-TSP classifier Patient-derived colorectal adenocarcinoma tumor We used the K-TSP algorithm (33) to construct a discrim- specimens were obtained from consenting patients at the inative classifier in predicting tumors sensitive to University of Colorado Hospital in accordance with pro- saracatinib. In brief, the algorithm exploits the information tocols approved by the Colorado Multiple Institutional contained in the rank-based matrix by focusing on “marker Review Board (COMIRB # 07-0570). Tumor material not gene pairs” (i, j) for which there is a significant difference in required for histopathologic analysis was collected and the probability of the event (Ri < Rj) across the N samples placed in medium consisting of RPMI supplemented with from class Y = 1 (saracatinib sensitive) to Y = −1 (saracatinib 10 μmol/L HEPES, 4.5 g/L glucose, 1 μmol/L pyruvate so- resistant), in which the event (Ri < Rj) is equivalent to the dium, 200 U/mL penicillin, and 200 μg/mL streptomycin. rank of gene i is less than the rank of gene j if and only if After washing several times in media, tumors were cut into gene i is expressed less then gene j (relative expression). 3 2- to 3-mm pieces in antibiotic-containing medium and Here, the quantities of interest are pij(m) = Prob(Ri < Rj | + coated in Matrigel. Four to 6-wk-old female athymic (nu / Y = m), m =(1, −1), i.e., the probabilities of observing Ri < + nu ) mice were obtained from Harlan Laboratories under Rj in each class. These probabilities were estimated by the an approved research protocol by the Institutional Animal relative frequencies of occurrences of Ri < Rj within profiles Care and Use Committee [IACUC # 51402007(09) 2E]. and over samples. Let Δij denote the “score” of gene pair The tumor pieces were then implanted in mice, and expan- (i, j), in which Δij =|pij(1) − pij(−1)|. A score Δij is computed sion of the F1-F3 generations was carried out as previously for every pair of genes i, j ∈{1,…,P}, i ≠ j. Gene pairs with described (17, 31). Both mutational analysis of common high scores were viewed as most informative for classifica- genes mutated in CRC (Ras, Raf, PI3K, PTEN, and Src) as tion. Using an internal leave-one-out cross-validation, the well as gene expression patterns were evaluated in F0 and final K-TSP classifier uses the k disjoint pairs of genes, which F3 generations to ensure genetic stability of tumors in achieve the k best scores from the training set. In this study, subsequent generations. Tumors were expanded in the left maximum number of pairs (kmax) was fixed as 10. and right flanks of five to six mice (10 evaluable tumors per group). Mice were randomized into the treatment group Immunoblotting (saracatinib) or vehicle group when tumor volumes reached Cells were seeded into six-well plates 24 hours before ∼200 mm3. Mice were treated daily with saracatinib 50 mg/ treatment with 500 nmol/L saracatinib. After treatment, kg or vehicle by oral gavage for 28 days. Mice were moni- 100 μL of 1× prechilled cell lysis buffer + 1 mmol/L phe- tored daily for signs of toxicity, and tumor size was evaluated nylmethylsulfonyl fluoride (Cell Signaling Technology) twice per week by caliper measurements using the following were added to the well, and cells were scraped and placed formula: tumor volume = (length × width2) × 0.52. in a microfuge tube. Tumor tissues (50-75 mg/mouse) were minced on ice and homogenized using a Dounce Determination of sensitive and resistant homogenizer, and centrifuged at 16,000 g at 4°C for molecular biomarkers 10 minutes. The total protein in samples was determined Total RNA (RNeasy kit) from all CRC explants and cell using the Bio-Rad Dc Protein Assay. Forty micrograms of lines were profiled using the Affymetrix Human Gene 1.0 sample were electrophoresed on 4% to 12% Bis-Tris pre- ST Array. This gene array has ∼764,885 probes comprising cast gels (Bio-Rad Laboratories, Inc.). After electrotransfer ∼28,869 genes. Sample preparation and processing proce- to Immobilon-P membranes (Millipore), membranes dure were done as described in the Affymetrix GeneChip were blocked at room temperature with TBS [10 mmol/L Expression Analysis Manual (Affymetrix, Inc.). The gene ex- Tris-HCl (pH 7.5), 0.5 mol/L NaCl, and 0.1% (v/v) Tween pression levels were converted to a rank-based matrix and 20] containing 5% nonfat milk (Bio-Rad) for 1 hour. Pri- standardized (mean, 0; SD, 1) for each microarray. Using this mary antibodies for Src, p-Src family (Tyr416), FAK, Stat-3, preprocessing method, we can cluster the same explants from p-Stat-3 (Tyr705) and Actin (Cell Signaling Technology), different data sets based on their gene expression profiles. and p-FAK Tyr861 (BioSource International) were diluted Data analyses were done on this rank-based matrix. at 1:1,000 in TBST containing 5% protease-free bovine serum albumin (Sigma-Aldrich), and the membranes were Differentially expressed genes analysis incubated with primary antibodies overnight at 4°C with We used two different methods to identify the differen- rocking. After washing thrice with TBST, the membranes were tially expressed genes in this study. When comparing two incubated for 1 h at room temperature with anti-rabbit/ phenotypes (e.g., saracatinib sensitive versus resistant), mouse IgG horseradish peroxidase–conjugated antibody first, we used t statistics to compute the P values for each at a final dilution of 1:50,000 in TBST. After washing thrice gene and only call differentially expressed genes when P < with TBST, bound antibodies were detected by enhanced 0.0001 (DEG list 1). We also performed Statistically Anal- chemiluminescence (Millipore). Immunoblot experiments ysis of Microarray (32) with stringent thresholds (fold were done in triplicate for each antibody. change, >1.5; false discovery rate, <0.0001) to determine the differentially expressed genes (DEG list 2). Only genes Immunohistochemistry found to be common in both lists were considered differ- Tumor tissues from control and saracatinib-treated mice entially expressed genes. were fixed in formalin immediately after surgical excision.

4168 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

Fig. 1. The effects of saracatinib on proliferation of CRC cell lines. A, 23 CRC cell lines were treated with saracatinib (0.0625 μmol/L → 2 μmol/L) and assayed by SRB 72 h after exposure to drug. Cell lines (H508, LS180, and LS174T) with an IC50 of <1 μmol/L were considered sensitive to saracatinib. Cell lines are listed in order from the most resistant to the most sensitive; (R), resistant; (S), sensitive. B, two sensitive and two resistant cell lines (identified from the SRB assay) were injected in a xenograft model to confirm the antiproliferative effects of saracatinib in vivo. When tumor volumes reached ∼200 mm3, mice were randomized and treated with saracatinib 50 mg/kg/d by oral gavage for 7 to 28 d. Tumor size was evaluated twice per week by caliper measurements using the formula: tumor volume = (length × width2) × 0.52. Treatment with saracatinib resulted in a significant decrease in growth in the LS180 (*, P < 0.05) and LS174T (***, P < 0.001). C, cell cycle was analyzed in two sensitive (H508 and LS174T) and two resistant (SW620 and

SW480) cell lines by flow cytometry. Treatment with saracatinib resulted in a G1 cell cycle arrest (*, P < 0.05) in the sensitive cell lines. There were no effects on cell cycle in the resistant cell lines.

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4169

Arcaroli et al.

Table 1. Src mutational status and Src gene copy (FISH) in CRC cell lines and explants

Specimen ID Saracatinib Src SRC Centromere SRC/ SRC copy effects (531) 20 centromere 20 number gain Mean SD Mean SD

Cell lines SW620 Resistant WT 3.23 0.85 3.03 0.39 1.07 Yes SW480 Resistant WT 4.40 1.21 4.40 1.21 1.00 Yes LS123 Resistant WT 4.09 0.70 4.04 0.74 1.01 Yes LS1034 Resistant WT 4.93 0.56 4.94 0.58 1.00 Yes COLO205 Resistant WT 5.10 1.38 4.95 0.96 1.03 Yes HCT116 Resistant WT 2.02 0.38 2.08 0.51 0.97 No HCT15 Resistant WT 2.10 0.48 2.02 0.48 1.04 No SW837 Resistant WT 2.99 0.98 1.64 0.69 1.82 No SW1417 Resistant WT 6.94 1.33 4.93 0.90 1.41 Yes HT29 Resistant WT 4.99 0.61 4.19 0.63 1.19 Yes GEO Resistant WT 4.00 0.64 2.95 0.56 1.36 Yes SW403 Resistant WT 4.76 0.65 4.76 0.65 1.00 Yes SW48 Resistant WT 2.02 0.14 2.00 0.14 1.01 No LS513 Resistant WT 2.00 0.25 2.02 0.20 0.99 No T84 Resistant WT 3.52 0.75 3.56 0.76 1.0 Yes LoVo Resistant WT 2.10 0.44 2.66 0.59 0.79 No SW948 Resistant WT 4.30 0.63 4.25 0.61 1.01 Yes RKO Resistant WT 3.40 1.03 3.22 0.97 1.06 Yes CaCo2 Resistant WT 7.76 1.75 7.56 1.70 1.03 Yes SW1463 Resistant WT 5.91 0.59 4.83 0.47 1.22 Yes LS180 Sensitive WT 1.99 0.27 1.98 0.32 1.01 No NCI-H508 Sensitive WT 6.19 1.02 3.91 0.90 1.58 Yes LS174T Sensitive WT 2.10 0.46 2.20 0.51 0.95 No Explants CRC-001 Resistant WT 2.24 1.02 2.12 0.8 1.06 No CRC-006 Resistant WT 2.44 0.97 2.24 0.8 1.09 No CRC-007 Sensitive WT 1.84 0.65 1.86 0.64 0.99 No CRC-010 Resistant WT 3.14 1.16 2.78 0.89 1.13 Yes CRC-012 Resistant WT 4.12 3.04 2.38 2.02 1.73 Yes CRC-021 Resistant WT 2.04 0.83 1.42 0.5 1.44 No CRC-026 Resistant WT 3.38 1.1 3.04 1.34 1.11 Yes CRC-027 Resistant WT 5.3 1.56 5.66 2.26 0.94 Yes CRC-034 Resistant WT 3.98 1.39 2.96 1.26 1.34 Yes CRC-040 Sensitive WT 2.44 0.86 2.4 0.83 1.02 No

NOTE: A Src copy number of ≥3 was considered a gain, and <3 was considered no gain.

Sections were deparaffinized using standard histologic an SRB assay. We considered an IC50 of <1 μmol/L as sen- procedures to characterize a suite of immunohistochemi- sitive to saracatinib. As shown in Fig. 1A, LS180, H508, cal biomarkers across several biological effect areas. Im- and LS174T were sensitive to saracatinib with an IC50 of munohistochemical staining was scored by eye (from a 0.5 μmol/L. All other cell lines evaluated were shown to minimum of 1,000 cells for the % markers) by pathology be resistant to saracatinib. To confirm the in vitro responses concordance panel (blinded). to saracatinib, we treated two of the most resistant cell lines (SW620 and SW480) and two sensitive cell lines Results (LS180 and LS174T) with saracatinib (50 mg/kg/d) in vivo. The two resistant cell lines proved resistant to saracatinib, Saracatinib inhibits growth in a subset of colorectal whereas the two sensitive cell lines were shown to be sen- cell lines in vitro and in vivo sitive to saracatinib in the xenograft model (Fig. 1B). This We tested the antiproliferative effects of saracatinib shows that the in vitro responses to saracatinib were the (dose range, 0.0625-2 μmol/L) on 23 CRC cell lines using same in vivo with these cell lines.

4170 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

To investigate the mechanisms of inhibition of cellular 83.3-100%), and 91% (95% confidence interval, 85.3- proliferation, the two resistant cell lines (SW620 and 97.3%) for accuracy, sensitivity, and specificity, respectively SW480) and two sensitive cell lines (H508 and LS174T) (34). The three gene pairs are thymocyte selection-associated were treated with 1 μmol/L of saracatinib for 48 hours, high mobility group box (TOX), GLIS family zinc finger and cell cycle progression was determined by flow cytome- 2 (GLIS2), tetraspanin 7 (TSPAN7), breast carcinoma am- try. H508 and LS174T cell lines treated with saracatinib plified sequence 4, (BCAS4), par-6 partitioning defective showed an increase in G1 cell cycle arrest at 48 hours 6 homologue γ (PARD6G), and nucleoredoxin (NXN). (Fig. 1C). In contrast, the resistant cell lines (SW620 and The interpretation for the top gene pair is as follows: If SW480) did not show any differences in cell cycle arrest the expression of TOX is higher than GLIS2, then predict after treatment. In addition, both resistant and sensitive sensitive to saracatinib, or else predict resistant to saraca- cell lines did not show any increases in apoptosis eval- tinib. Similar interpretation applies to the second and uated by a caspase-3/7 assay (data not shown). These re- third gene pairs. Figure 3A depicted the heat map of the sults suggest that the antiproliferative effects of saracatinib K-TSP gene pair classifier. on sensitive cell lines are through cell cycle arrest and not apoptosis. K-TSP achieves high accuracy in predicting saracatinib sensitivity in CRC explants Heterogeneity in Src gene copy in CRC cell lines We next tested the predictive power of the K-TSP classifier and explants from the CRC cell line training set in 10 patient-derived We developed a new FISH assay to assess Src gene copy human CRC explants. Before treatment, baseline gene number. Table 1 shows that a gain in Src gene copy num- array was analyzed for the six markers. Patients CRC007, ber was common among CRC cell lines and explants. In CRC001, and CRC006 (TOX>GLIS2, TSPAN7>BCAS4, addition, the use of a probe for centromere 20 shows that and PARD6G>NXN) were predicted to be sensitive, and the gain in Src gene copy number is mainly from an in- patients CRC010, CRC012, CRC021, CRC026, CRC027, crease in ploidy of chromosome 20, showing that the CRC034, and CRC040 were predicted to be resistant Src gene was not amplified. We next examined the rela- (Fig. 3B). The 10 cases were then treated with saracatinib tionship between Src gene copy and sensitivity or resistance to saracatinib in CRC cell lines and explants, and showed a (Fisher's exact) trend toward significance (P = 0.066) with respect to an increase in Src gene copy number and resistance to saracatinib (Table 1). A Src gene copy number of ≥3 was considered a gain, and <3 was considered no gain. A representative figure shows that CRC007, a sensitive tumor, had normal Src gene copy number, whereas CRC027, a resistant tumor, has an increase in Src gene copy (Fig. 2A and B).

Mutational analysis of Src It has been suggested that Src (codon 531) has an activating mutation in 12% of patients with advanced CRC (29), although this finding has not been confirmed in other laboratories. Evaluation of this mutation in our CRC cell lines and explants did not reveal any mutations (Table 1).

Identification of sensitive and resistant biomarkers by K-TSP scoring in CRC We next sought to develop a K-TSP gene classifier for predicting saracatinib sensitivity in CRC. Baseline gene expression profiles of three sensitive cell lines and a subset of 8 of 20 resistant cell lines in vitro (cell lines cultured) and in vivo (cell lines in xenograft model) were used as the training set. We used both in vitro and in vivo baseline gene array for a given cell line in the training set so that differences relating purely to the tumor microenvironment would be minimized. The K-TSP algorithm determined Fig. 2. Evaluation of Src gene copy number by FISH. A, representative three gene pairs from the training set with an estimated graphs of interphase nuclei showing normal copies (∼2 genes per cell) leave-one-out cross-validation of 91% (95% confidence of the Src gene (CRC007). B, gain (∼5.3 gene copies per cell) in Src gene interval, 86.6-96.2%), 92% (95% confidence interval, copy (CRC027).

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4171

Arcaroli et al.

Fig. 3. Heat map of the K-TSP gene pairs for predicting saracatinib sensitivity. A, relative gene expression profiles in the training set of the three gene pairs in (three sensitive and eight resistant) CRC cell lines in vitro and in vivo. The three gene pairs that predict sensitivity to saracatinib include TOX > GLIS2, TSPAN7 > BCAS4, and PARD6G > NXN. High (red) and low (green) gene expression. B, heat map of expression in the validation set of the three gene pairs. CRC006 and CRC007 were predicted to be sensitive.

(50 mg/kg/d, oral gavage), and tumor growth inhibition for 1 hour resulted in a decrease in the phosphorylation (TGI) was evaluated 28 days after treatment. Recognizing of Src (Y416), FAK (Y861), and Stat-3 (Y705). Further- that the explant model has a limited dynamic range, in more, the sensitive explants CRC007 and CRC040 had in- this validation set, we considered a TGI of ≤50% as sensi- creased activation of Src, but not FAK and Stat-3 when tive and a TGI of >50% as resistant. CRC007 (TGI of 23%) compared with the resistant CRC explants (Fig. 5B). Eval- and CRC040 (TGI of 37%) were sensitive, whereas the uation of tumors from CRC007 and CRC027 28 days after other eight explants were resistant to saracatinib (Fig. 4). saracatinib treatment showed a decrease in Src activation The gene classifier predicts 7 of 10 correct (70% accuracy), (Fig. 5C). Finally, immunohistochemistry shows that Src yielding 50% sensitivity (correctly predicting 1 of 2 in the activity was increased in the sensitive tumor (CRC007) sensitive cases) and 75% specificity (correctly predicting 6 versus the resistant tumor (CRC027; Fig. 5D). of 8 in the resistant cases). Discussion Sensitive CRC cell lines and explants have increased levels and activity of Src and downstream mediators Significant advances in the molecular characterization of Next, we assessed protein levels and activation of Src, aberrant pathways in cancer have led to the development FAK, and Stat-3 in two of the most resistant cell lines of targeted agents. However, there is considerable hete- (SW620 and SW480), two of three sensitive cell lines rogeneity with respect to dysregulated pathways and (H508 and LS174T), and 10 CRC explants by Western mutations in genes involved in cell cycle control, prolifer- blotting. We hypothesized that cell lines and explants that ation, angiogenesis, and metastasis. Consequently, many were more sensitive to saracatinib would have higher le- targeted agents lack efficacy because they usually benefit vels of Src pathway activity when compared with resistant a small subset of patients, and clinical trials are not suffi- cell lines and explants. As shown in Fig. 5A, both protein ciently powered to detect these differences. Therefore, levels and activity of Src, FAK, and Stat-3 were significantly identifying this patient population through predictive elevated in the sensitive cell lines when compared with the biomarkers before treatment may lead to more effective resistant cell lines. Treatment with saracatinib (0.5 μmol/L) therapy and ultimately better outcomes. In this study, we

4172 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

explored the efficacy of saracatinib (a Src inhibitor) in pharmacologically relevant concentrations. Although sara- CRC and used a K-TSP classifier, FISH, and pathway acti- catinib did not inhibit the growth of LoVo, Colo205, or vation to predict sensitivity to saracatinib in our patient- HT29 tumors in vivo, Src kinase activity was inhibited as derived CRC explant model. evidenced by reduced phosphorylation of Src substrates Saracatinib, a potent and selective inhibitor of Src/Abl FAK and paxillin. Other Src inhibitors such as bosutinib kinase activity currently in phase II clinical development, and dasatinib have also shown antiproliferative effects in has shown a range of antiproliferative effects in vitro in se- CRC (36, 37). In particular, dasatinib showed antiproli- lected lung, breast, ovarian, prostate, pancreas, and colon ferative activity in the CRC cell lines DLD-1, HT29, WiDr, cell lines (35). Whereas LoVo and SW403 CRC cell lines Caco-2, and SW48 (37). Additionally, bosutinib was were particularly sensitive to in vitro growth inhibition shown to inhibit tumor growth in HT29, Colo205, by saracatinib, HT29 and HCT116 were not inhibited at HCT116, and DLD-1 in a mouse xenograft model (36).

Fig. 4. Three gene pairs identified by K-TSP predicts saracatinib sensitivity in an independent test set. A, prospective validation of efficacy in 10 xenografts predicted by the K-TSP classifier. The 10 xenografts predicted for saracatinib sensitivity were treated with saracatinib 50 mg/kg/d by oral gavage for 28 d. Tumor size was evaluated twice per week by caliper measurements using the formula: tumor volume = (length × width2) × 0.52. TGI was calculated by relative tumor growth of treated mice divided by relative tumor growth of control mice × 100. Cases with a TGI of <50% were considered sensitive, and TGI of >50% were considered resistant to saracatinib. Two xenografts (CRC006 and CRC040) were sensitive to saracatinib (TGI ≤ 50%), and eight xenografts were resistant to saracatinib (TGI > 50%). CRC001, CRC006 (predicted sensitive), and CRC040 (predicted resistant) by the K-TSP classifier were misclassified. Columns, mean (n = 8-0 tumors per group); bars, SEM. *, significance (P < 0.05) compared with vehicle-treated tumors. B, representative graph of a sensitive tumor (CRC007) and (C) resistant tumor (CRC027) treated with saracatinib.

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4173

Arcaroli et al.

Fig. 5. Saracatinib-sensitive CRC cell lines and explant have increased activation of Src and downstream mediators. A, the sensitive cell lines H508 and LS174T had greater activation of Src, FAK, and Stat-3 when compared with the resistant cell lines SW620 and SW480. Treatment with saracatinib (500 nmol/L) for 1 h resulted in a decrease in the phosphorylation of Src, FAK, and Stat-3 (n = 3). B, evaluation of Src, FAK, and Stat-3 activation in CRC explants by immunoblotting showed that CRC007 and CRC040, sensitive tumors, had the highest activation of Src and FAK but not Stat-3. C, representative graph of two tumors (CRC007 and CRC027) treated with saracatinib (50 mg/kg/d) showed a decrease in the activation of Src. D, immunohistochemical staining of p-Src (Y416) shows positive staining in the sensitive tumor (CRC007) and negative staining in the resistant tumor (CRC027).

Although Src substrates such as FAK and paxillin may be such as FAK and paxillin, or cell cycle perturbation (37– suitable biomarkers of Src kinase activity, clearly there is a 39), the pattern of gene expression in response to Src ki- need for biomarkers that are predictive of tumor growth nase inhibition seems to be pharmacophore related. inhibition in vivo. Whereas saracatinib, dasatinib, and bosiutinib are all po- Here, we investigated the efficacy of saracatinib on 23 tent inhibitors of Src (and Src family kinases), their off- CRC cell lines and identified 20 resistant and 3 sensitive target kinase inhibition profiles are likely to differentially cell lines. Importantly, treatment of resistant and sensitive modulate gene expression. For example, K-TSP analysis of cell lines in a xenograft model confirmed the in vitro re- pancreatic tumors treated with bosutinib or saracatinib sults; SW620 and SW480 were resistant, and LS180 and identified different pairs of genes indicative of sensitive LS174T were sensitive to saracatinib. Moreover, we show tumors. We have previously described the development in our human CRC explant model that 2 out 10 explants of K-TSP classifiers for Src kinase inhibitors in pancreas responded to saracatinib. Additionally, we show that the cancer. The K-TSP derived for bosutinib in pancreatic ex- antiproliferative effects seen with saracatinib in cell lines plant models included the gene pairs GPR116>LQK1, are through G1 cell cycle arrest and not apoptosis. Taken CEACAM7>SPFH2, and ALPK3>GUF1 (18), whereas sara- together, these findings suggest that inhibition of Src with catinib sensitivity was defined by LRRC19>IGFBP2 (19). a selective inhibitor may prove to be beneficial in a subset Some pancreatic explant models showed an equivalent of CRC patients. The identification of patient stratification response to saracatinib or bosutinib, whereas other tumors markers for targeted therapies such as Src kinase inhibitors were resistant to one agent but sensitive to the other. The is a key challenge for drug development. kinase inhibitor profile, together with the pattern of Whereas several Src kinase inhibitors have consistently pathway activation and tumor specific differences, are inhibited Src-mediated events such as cell adhesion and likely determinants of sensitivity and successful predictors cell motility in vitro, the phosphorylation of key substrates of response.

4174 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

The advantage of the K-TSP approach is that it is easy to previously. We have not seen Src gene amplification in interpret as the classifier involves only a few informative either cell lines or human explants. Additionally, we genes. In this study, we tested the LLRC19>IGFBP2 gene show that cell lines and tumors that have a gain in Src pair in our CRC model; however, this gene pair was not gene copy number are associated with resistance to sara- predictive, suggesting that the mechanisms of sensitivity catinib. As Src mutations were not found in our study, and resistant are different between pancreatic and colon these results indicate that an increase in Src gene copy cancer. These results may also be related to enzyme inhi- rather than mutations are more prevalent in CRC and bition profiles of the Src kinase inhibitors and the relative may possibly be a predictor of resistance to saracatinib. potency versus key enzymes in different tissues. The data Characterization of Src gene copy number in annotated suggest that a gene signature defining sensitivity to par- premalignant and malignant human colorectal tumor clin- ticular kinase inhibitors may not be universally appli- ical samples, as well as clinical trial samples in patients cable and will need to be established for each target treated with Src inhibitors, may elucidate the relevance tumor type. of these findings. Consequently, we evaluated other gene sets to predict Levels and activation of Src are elevated at all stages of sensitivity to saracatinib in CRC and identified three gene CRC in which it is postulated to enhance tumor growth pairs that predict sensitivity: TOX>GLIS2, TSPAN7>BCAS4, and progression by activating many downstream targets and PARD6G>NXN. Four (TOX, TSPAN7, GLIS2, and such as FAK and Stat-3. We found that the activation of NXN) of the K-TSP genes were among the top 100 differen- Src, FAK, and Stat-3 are increased in sensitive cell lines tially expressed genes. Using the training set (CRC cell lines when compared with resistant cell lines. Moreover, the ac- in vitro and in vivo), the gene pairs were able to predict with tivation of Src was also the highest in the sensitive CRC 70% accuracy in the CRC explants. explants (CRC007 and CRC040) when compared with re- Many of the genes identified by the K-TSP classifier sistant explants. This suggests that Src pathway activation have been implicated in cancer. Tetraspanins such as may be a marker of sensitivity to Src inhibitors such as sar- TSPAN7 are transmembrane proteins that form com- acatinib in CRC. plexes with one another; in cancer, higher expression For this study, we developed a patient-derived human levels of tetraspanins are associated with metastasis CRC explant model to evaluate efficacy and identify bio- (40). GLIS family zinc finger 2 is a transcriptional re- markers of sensitivity to saracatinib. Although there are pressor that has been shown in conjunction with Src to several advantages of this preclinical model compared enhance the nuclear translocation of p120 catenin, a with cell lines, there are also several weaknesses (47). protein important in facilitating cell-cell interactions First, it is unknown whether this model will be able (41). The BCAS4 gene was one of the most amplified to predict activity in the clinic. Future clinical trials genes in breast cancer cell lines (42); however, its func- where these biomarkers are prospectively evaluated in tion and role in cancer is unknown. TOX is a chroma- patients will likely resolve this issue. A second limita- tin-associated protein that specifically binds to DNA. tion of this model is that tumor cells in this xenograft Genome-wide studies in breast cancer have revealed model do not metastasize. Therefore, we could not eval- TOX as a susceptibility gene (43). NXN is a member uate the effects of saracatinib on metastasis, one of the of the thioredoxin family of proteins, which possess main functions of Src in altering tumor cell behavior. A thiol-oxidoreductase activity. Recently, it has been third challenge is the limited dynamic range of the shown to regulate the Wnt/β-catenin pathway, an im- model, in which small differences in TGI are difficult portant pathway regulating cellular differentiation and to interpret and may be irrelevant. A final issue with proliferation (44). Finally, PARD6G is a cell polarity the model is the inability to assess immunologic me- component that plays a role in cell division (45). Be- chanisms because the model uses immunocompro- cause Src has hundreds of client proteins with which it mised mice to prevent graft rejection. interacts, it is not surprising that many proteins may Like many targeted agents, the clinical activity of sara- contribute to drug sensitivity. Further functional studies catinib as a monotherapy has been modest. There were are warranted to specifically address the importance no responses in a small phase II trial of 10 patients in and mechanism of these genes in altering sensitivity third or fourth line CRC (48). It is likely that the future to this drug. of this class of agents lies in proper selection of sensitive Studies have shown that the expression of Src is elevated patient subsets, as well as combinational strategies. Initial in premalignant lesions and subsequently increases as results of a combination trial of FOLFOX, cetuximab, and tumors progress with some of the highest levels in meta- dasatinib showed an impressive response rate with fur- static lesions (24). It has also been shown that a gain in ther results pending (49). It has been shown that oxali- chromosome 20 commonly occurs in CRC (46). Given platin could lead to an increase in SRC expression and that Src levels are elevated in CRC, we investigated Src activition in HT29 CRC cell line (50). Furthermore, treat- gene copy number by FISH. Here, we show that there is ment with oxaliplatin and dasatinib resulted in signifi- heterogeneity in Src gene copy number, with most of the cant reduction in tumor volume when compared with CRC cell lines and explants having a gain in Src gene nontreated mice in a liver metastatic preclinical model, number. To our knowledge, this has not been reported whereas there were no differences when single agent

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4175

Arcaroli et al.

oxaliplatin or dasatinib were administered (50). Although Disclosure of Potential Conflicts of Interest the clinical implications of human explant models re- mains unknown, our results suggest that patient subsets J.J. Arcoli: advisory board for saracatinib, AstraZeneca. The other authors disclosed no potential conflicts of interest. sensitive to Src inhibitors may be identified. Follow-up biomarker-driven and/or combinatorial clinical trials may Received 01/11/2010; revised 05/05/2010; accepted 05/27/2010; be warranted. published OnlineFirst 08/03/2010.

References 1. Bromann PA, Korkaya H, Courtneidge SA. The interplay between Src 20. Jemal A, Siegel R, Ward E, Hao Y, Xu J, Thun MJ. Cancer statistics, family kinases and receptor tyrosine kinases. Oncogene 2004;23: 2009. CA Cancer J Clin 2009;59:225–49. 7957–68. 21. Yeatman T. A renaissance for src. Nat Rev Cancer 2004;4:470–80. 2. Frame MC. Src in cancer: deregulation and consequences for cell 22. Malek RL, Irby RB, Guo QM, et al. Identification of Src transformation behaviour. Biochim Biophys Acta 2002;1602:114–30. fingerprint in human colon cancer. Oncogene 2002;21:7256–65. 3. Schlessinger J. New roles for Src kinases in control of cell survival 23. Cartwright C, Meisler A, Eckhart W. Activation of the pp60c-src and angiogenesis. Cell 2000;100:293–6. protein kinase is an early event in colonic carcinogenesis. Proc Natl 4. Summy J, Gallick G. Src family kinases in tumor progression and Acad Sci U S A 1990;87:558–62. metastasis. Cancer Metastasis Rev 2003;22:337–58. 24. Talamonti MS, Roh MS, Curley SA, Gallick GE. Increase in activity 5. LaVallee T, Prudovsky I, McMahon G, Hu X, Maciag T. Activation of and level of pp60c-src in progressive stages of human colorectal the MAP kinase pathway by FGF-1 correlates with cell proliferation cancer. J Clin Invest 1993;91:53–60. induction while activation of the Src pathway correlates with 25. Park J, Meisler A, Cartwright C. c-Yes tyrosine kinase activity in migration. J Cell Biol 1998;141:1647–58. human colon carcinoma. Oncogene 1993;8:2627–35. 6. Peterson J, Kulik G, Jelinek T, Reuter C, Shannon J, Weber M. Src 26. Karapetis CS, Khambata-Ford S, Jonker DJ, et al. K-ras mutations phosphorylates the insulin-like growth factor type I receptor on and benefit from cetuximab in advanced colorectal cancer. N Engl J the autophosphorylation sites. Requirement for transformation by Med 2008;359:1757–65. Src. J Biol Chem 1996;271:31562–71. 27. Hecht JR, Mitchell E, Chidiac T, et al. A randomized phase IIIB trial 7. Rahimi N, Hung W, Tremblay E, Saulnier R, Elliott B. c-Src kinase of chemotherapy, bevacizumab, and panitumumab compared activity is required for hepatocyte growth factor-induced motility with chemotherapy and bevacizumab alone for metastatic colorectal and anchorage-independent growth of mammary carcinoma cells. cancer. J Clin Oncol 2009;27:672–80. J Biol Chem 1998;273:33714–21. 28. Tol J, Koopman M, Cats A, et al. Chemotherapy, bevacizumab, and 8. Tice D, Biscardi J, Nickles A, Parsons S. Mechanism of biological cetuximab in metastatic colorectal cancer. N Engl J Med 2009;360: synergy between cellular Src and epidermal growth factor receptor. 563–72. Proc Natl Acad Sci U S A 1999;96:1415–20. 29. Irby R, Mao W, Coppola D, et al. Activating SRC mutation in a subset 9. Hauck C, Hsia D, Schlaepfer D. The focal adhesion kinase: a of advanced human colon cancers. Nat Genet 1999;21:187–90. regulator of cell migration and invasion. IUBMB Life 2002;53:115–9. 30. Jhawer M, Goel S, Wilson AJ, et al. PIK3CA mutation/PTEN expres- 10. Jones RJ, Brunton VG, Frame MC. Adhesion-linked kinases in sion status predicts response of colon cancer cells to the epidermal cancer;emphasis on src, focal adhesion kinase and PI 3-kinase. growth factor receptor inhibitor cetuximab. Cancer Res 2008;68: Eur J Cancer 2000;36:1595–606. 1953–61. 11. Schaller M. Biochemical signals and biological responses elicited 31. Dangles-Marie V, Pocard M, Richon S, et al. Establishment of by the focal adhesion kinase. Biochimica et Biophysica Acta 2001; human colon cancer cell lines from fresh tumors versus xenografts: 1540:1–21. comparison of success rate and cell line features. Cancer Res 2007; 12. Owens L, Xu L, Craven R, et al. Overexpression of the focal adhesion 67:398–407. kinase (p125FAK) in invasive human tumors. Cancer Res 1995;55: 32. Tusher VG, Tibshirani R, Chu G. Significance analysis of microarrays 2752–5. applied to the ionizing radiation response. Proc Natl Acad Sci U S A 13. Niu G, Wright K, Huang M, et al. Constitutive Stat3 activity up- 2001;98:5116–21. regulates VEGF expression and tumor angiogenesis. Oncogene 33. Tan AC, Naiman DQ, Xu L, Winslow RL, Geman D. Simple decision 2002;21:2000–8. rules for classifying human cancers from gene expression profiles. 14. YuCL,MeyerDJ,CampbellGS,etal.EnhancedDNA-binding Bioinformatics 2005;21:3896–904. activity of a Stat3-related protein in cells transformed by the Src 34. Dreyfus GMG. Withdrawing an example from the training set: an oncoprotein. Science 1995;269:81–3. analytic estimation of its effect on a non-linear parameterised model. 15. Wong CH, Baehner FL, Spassov DS, et al. Phosphorylation of the Neurocomputing 2000;35:195–201. SRC epithelial substrate Trask is tightly regulated in normal epithelia 35. Green TP, Fennell M, Whittaker R, et al. Preclinical anticancer ac- but widespread in many human epithelial cancers. Clin Cancer Res tivity of the potent, oral Src inhibitor AZD0530. Mol Oncol 2009;3: 2009;15:2311–22. 248–61. 16. Huang F, Reeves K, Han X, et al. Identification of candidate 36. Golas JM, Lucas J, Etienne C, et al. SKI-606, a Src/Abl inhibitor molecular markers predicting sensitivity in solid tumors to dasatinib: with in vivo activity in colon tumor xenograft models. Cancer Res rationale for patient selection. Cancer Res 2007;67:2226–38. 2005;65:5358–64. 17. Rubio-Viqueira B, Jimeno A, Cusatis G, et al. An in vivo platform for 37. Serrels A, Macpherson IR, Evans TR, et al. Identification of potential translational drug development in pancreatic cancer. Clin Cancer biomarkers for measuring inhibition of Src kinase activity in colon Res 2006;12:4652–61. cancer cells following treatment with dasatinib. Mol Cancer Ther 18. Messersmith WA, Rajeshkumar NV, Tan AC, et al. Efficacy and 2006;5:3014–22. pharmacodynamic effects of bosutinib (SKI-606), a Src/Abl inhibitor, 38. Nam S, Kim D, Cheng JQ, et al. Action of the Src family kinase in freshly generated human pancreas cancer xenografts. Mol Cancer inhibitor, dasatinib (BMS-354825), on human prostate cancer cells. Ther 2009;8:1484–93. Cancer Res 2005;65:9185–9. 19. Rajeshkumar NV, Tan AC, De Oliveira E, et al. Antitumor effects 39. Chang YM, Bai L, Liu S, Yang JC, Kung HJ, Evans CP. Src family and biomarkers of activity of AZD0530, a Src inhibitor, in pancreatic kinase oncogenic potential and pathways in prostate cancer as cancer. Clin Cancer Res 2009;15:4138–46. revealed by AZD0530. Oncogene 2008;27:6365–75.

4176 Clin Cancer Res; 16(16) August 15, 2010 Clinical Cancer Research

Saracatinib Activity Biomarkers and Colorectal Cancer

40. Le Naour F, Andre M, Greco C, et al. Profiling of the tetraspanin ity as a non-canonical tumor suppressor. J Cell Sci 2008;121: web of human colon cancer cells. Mol Cell Proteomics 2006;5: 1141–50. 845–57. 46. Nanashima A, Yamaguchi H, Yasutake T, et al. Gain of chromosome 41. Hosking CR, Ulloa F, Hogan C, et al. The transcriptional repressor 20 is a frequent aberration in liver metastasis of colorectal cancers. Glis2 is a novel binding partner for p120 catenin. Mol Biol Cell Dig Dis Sci 1997;42:1388–93. 2007;18:1918–27. 47. Jimeno AAS, Karikari C, Garrido-Laguna I, et al. A prospective 42. Barlund M, Monni O, Weaver JD, et al. Cloning of BCAS3 validation of a direct tumor xenograft model in pancreatic ductal (17q23) and BCAS4 (20q13) genes that undergo amplification, adenocarcinoma (PDA). J Clin Oncol 2008;26:4500. overexpression, and fusion in breast cancer. Genes Chromosomes 48. Fogelman DR, Kopetz S, Eng C. Emerging drugs for colorectal Cancer 2002;35:311–7. cancer. Expert Opin Emerg Drugs 2008;13:629–42. 43. Easton DF, Pooley KA, Dunning AM, et al. Genome-wide association 49. KopetzSRAW,GloverK,HenryL,etal.PhaseIstudyofSrc study identifies novel breast cancer susceptibility loci. Nature 2007; inhibition with dasatinib in combination with 5-fluoruracil, leucovorin, 447:1087–93. oxaliplatin (FOLFOX) and cetuximab in metastatic colorectal cancer. 44. Funato Y, Miki H. Nucleoredoxin, a novel thioredoxin family member ASCO-2008 Gastrointestinal Cancers Symposium 2008;325. involved in cell growth and differentiation. Antioxid Redox Signal 50. Kopetz S, Lesslie DP, Dallas NA, et al. Synergistic activity of the 2007;9:1035–57. SRC family kinase inhibitor dasatinib and oxaliplatin in colon carcinoma 45. Lee M, Vasioukhin V. Cell polarity and cancer-cell and tissue polar- cells is mediated by oxidative stress. Cancer Res 2009;69:3842–9.

www.aacrjournals.org Clin Cancer Res; 16(16) August 15, 2010 4177

Gene Array and Fluorescence In situ Hybridization Biomarkers of Activity of Saracatinib (AZD0530), a Src Inhibitor, in a Preclinical Model of Colorectal Cancer

John J. Arcaroli, Basel M. Touban, Aik Choon Tan, et al.

Clin Cancer Res Published OnlineFirst August 3, 2010.

Updated version Access the most recent version of this article at: doi:10.1158/1078-0432.CCR-10-0066

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/early/2010/07/30/1078-0432.CCR-10-0066. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.