Systematic Interrogation of 3Q26 Identifies TLOC1 and SKIL As Cancer Drivers

Published OnlineFirst June 13, 2013; DOI: 10.1158/2159-8290.CD-12-0592

RESEARCH ARTICLE

Systematic Interrogation of 3q26 Identiﬁ es TLOC1 and SKIL as Cancer Drivers

Daniel Hagerstrand 1 , Alexander Tong 1 , Steven E. Schumacher 2 , 5, Nina Ilic 1 , Rhine R. Shen 1 , Hiu Wing Cheung 1 , 5, Francisca Vazquez 1 , 5, Yashaswi Shrestha 1 , 5, So Young Kim 1 , 6, Andrew O. Giacomelli 1 , Joseph Rosenbluh 1 , 5, Anna C. Schinzel 1 , Nicole A. Spardy 1 , David A. Barbie 1 , 5, Craig H. Mermel 1 , 5, Barbara A. Weir 5 , Levi A. Garraway 1 , 3 , 5, Pablo Tamayo 5 , Jill P. Mesirov 5 , Rameen Beroukhim 1 , 2 , 3 , 5, and William C. Hahn 1 , 3 , 4 , 5

ABSTRACT 3q26 is frequently amplifi ed in several cancer types with a common amplifi ed region containing 20 genes. To identify cancer driver genes in this region, we interrogated the function of each of these genes by loss- and gain-of-function genetic screens. Specifi cally, we found that TLOC1 (SEC62 ) was selectively required for the proliferation of cell lines with 3q26 amplifi cation. Increased TLOC1 expression induced anchorage-independent growth, and a second 3q26 gene, SKIL (SNON ), facilitated cell invasion in immortalized human mammary epithelial cells. Expres- sion of both TLOC1 and SKIL induced subcutaneous tumor growth. Proteomic studies showed that TLOC1 binds to DDX3X, which is essential for TLOC1-induced transformation and affected protein translation. SKIL induced invasion through upregulation of SLUG (SNAI2 ) expression. Together, these studies identify TLOC1 and SKIL as driver genes at 3q26 and more broadly suggest that cooperating genes may be coamplifi ed in other regions with somatic copy number gain.

SIGNIFICANCE: These studies identify TLOC1 and SKIL as driver genes in 3q26. These observations provide evidence that regions of somatic copy number gain may harbor cooperating genes of different but complementary functions. Cancer Discov; 3(9); 1044–57. ©2013 AACR.

INTRODUCTION RESULTS The most common alterations found in cancer genomes are recurrent somatic copy number alterations (SCNA; refs. 1, 3q26 Is Frequently Amplifi ed in Ovarian, Breast, 2 ). Although some of these SCNAs harbor known oncogenes and Non–Small Cell Lung Cancers or tumor suppressor genes, the gene(s) targeted by most of We previously used the Genomic Identifi cation of Signifi - these SCNAs remain unclear. For example, a recent study cant Targets in Cancer (GISTIC) analytic approach to identify of more than 3,000 cancer samples identifi ed 158 recurrent recurrent regions of SCNA in a set of 3,131 tumor samples SCNAs in several cancer types, of which 122 did not harbor a ( 1 ). Many of the most frequently amplifi ed regions harbored known oncogene or tumor suppressor gene ( 1 ). known oncogenes, but the identity of specifi c driver genes was 3q26 has been reported to be amplifi ed in several cancer unknown for several recurrently amplifi ed regions. Specifi - types including breast, prostate, ovarian, non–small cell lung, cally, 3q26 was amplifi ed in 22% of these tumor samples, and and head and neck squamous carcinomas. The distal arm 8.4% of tumors harbored focal amplifi cations, as defi ned as a of 3q also contains other oncogene candidates including region less than half a chromosome arm long. The minimal PIK3CA, SOX2 , and TP63 (3–5 ). However, the analysis of a common amplifi ed region contained 20 protein-coding genes. large number of human cancers showed that the minimal When we investigated the specifi c cancer types in Tumorscape common amplifi ed region at 3q26 contains 20 genes, which ( 6 ) that harbor this amplicon, we found that 3q26 is ampli- frequently does not include the neighboring genes PIK3CA , fi ed in 43.7% of ovarian, 31.7% of breast, and 31.2% of non– SOX2 , or TP63 . small cell lung carcinomas (NSCLC; ref. 1 ; Fig. 1A ). When we Here, we applied both gain- and loss-of-function approaches interrogated the current Cancer Genome Atlas (TCGA; 5,547 to interrogate the 20 genes resident in the minimal common samples; refs. 2 , 7 ), we found that 3q26 is amplifi ed in lung amplifi ed region of 3q26 for effects on proliferation, anchor- squamous cell (31.5%), serous ovarian (19.3%), cervical squa- age-independent growth, and invasion. We found two genes mous cell (11.8%), head and neck (11.5%), pancreatic (7.1%), that cooperated to confer a tumorigenic phenotype: TLOC1 uterine corpus endometrial (7.1%), and stomach adenocarci- and SKIL . noma cancers (6.1%). Amplifi cation of 3q26 often extends to include larger regions of 3q. Because PIK3CA , SOX2 , and TP63 also reside on Authors’ Affi liations: 1 Departments of Medical Oncology and 2 Cancer 3q, we investigated at what frequency these genes are ampli- Biology; 3Center for Cancer Genome Discovery, Dana-Farber Cancer Insti- fi ed in conjunction with 3q26. Of more than 3,000 cancer 4 tute; Department of Medicine, Brigham and Women’s Hospital, Harvard samples in Tumorscape, 718 (22%) displayed amplifi cation Medical School, Boston; 5 Broad Institute of Harvard and MIT, Cambridge, Massachusetts; and 6 Department of Molecular Genetics and Microbiology, of 3q26. Of these 718 samples, 59 did not include PIK3CA, Duke University Medical Center, Durham, North Carolina SOX2, or TP63 (Fig. 1B ). To investigate whether PIK3CA is Note: Supplementary data for this article are available at Cancer Discovery mutated at a higher frequency in samples that lack amplifi ca- Online (http://cancerdiscovery.aacrjournals.org/). tion of the minimal common amplifi ed region, we analyzed Corresponding Author: William C. Hahn, Dana-Farber Cancer Institute, 450 data from 3,953 samples from the TCGA (2 ) for which both Brookline Avenue, Boston, MA 02215. Phone: 617-632-2641; Fax: 617- copy number and mutation information were available. We 632-4005; E-mail: [email protected] found no signifi cant enrichment of PIK3CA mutations in doi: 10.1158/2159-8290.CD-12-0592 samples that either harbor or lack amplifi cation of 3q26 ©2013 American Association for Cancer Research. ( P = 0.37, χ 2 test). Together these observations confi rm that

SEPTEMBER 2013CANCER DISCOVERY | 1045

RESEARCH ARTICLE Hagerstrand et al.

A Samples Ovarian Breast NSCLC EVI1 MDS1 ARPM1

m MYNN

ar LRRC34 p

3 LRRC31 SAMD7

e TLOC1 GPR160 PHC3 PRKCI

entromer SKIL C CLDN11 SLC7A14 RPL22L1

arm EIF5A2 q

3 SLC2A2 3q26 TNIK PIK3CA PLD1 SOX2 FNDC3B TP63 43.7% 31.7% 31.2% 3q26 Amplification

B 3p arm Centromere 3q arm Chromosome 3 170 mb 1801 mb 190 mb 3q region

3q26 PIK3CA SOX2 TP63 Samples

Figure 1. 3q26 is frequently amplifi ed in ovarian cancer, breast cancer, and NSCLC. A, copy number plots of the samples in Tumorscape for the q arm of chromosome 3. A vertical line represents each sample, where red represents a high chromosomal copy number ratio, blue represents low, and white represents neutral. Chromosomal bands are shown to the left. Two horizontal blue lines indicate the minimal common amplifi ed region, and the genes are listed to the right. B, illustration of samples with 3q26 amplifi cation. Samples that did not exhibit copy number gain of the PIK3CA , SOX2 , or TP63 locus are shown at the top. The 3q distal regions have been magnifi ed to show the position of these in relation to each other. ( continued on following page)

3q26 is frequently amplifi ed in several cancer types in a in 11 cell lines that do or fi ve cell lines that do not harbor manner that is independent of copy number alterations of amplifi cation of 3q26 ( Fig. 1D ). We used RNA interference PIK3CA, SOX2 , or TP63 . (RNAi) Gene Enrichment Ranking (RIGER) analysis (8 ) to identify genes that were selectively essential for proliferation Systematic Interrogation of 3q26 Identifi es of cell lines with 3q26 amplifi cation. This method takes the TLOC1 and SKIL as Transforming Genes effect of all shRNAs for one gene into account and compares To identify genes resident in 3q26 that contribute to the score for each gene to other genes. Specifi cally, we sum- malignant transformation, we interrogated the function of marized the effects of multiple shRNAs (fi ve on average) the genes in the region by conducting systematic gain- and targeting each gene into a single fi nal score called normalized loss-of-function studies (Fig. 1C ). Specifi cally, we suppressed enrichment score (NES), which considers the relative ranking or overexpressed each of the 20 genes present in the minimal of and the magnitude of gene-specifi c suppression effects of common amplifi ed region and assessed the effects on prolif- the shRNAs (Supplementary Table S1). The enrichment score eration, anchorage-independent growth, and invasion. refl ects the degree to which the shRNAs targeting each gene To identify genes whose expression was necessary for the are overrepresented at the top or bottom of the ranked list. proliferation of cell lines that harbored the 3q26 amplicon, The scores are further normalized to account for the size of we conducted an arrayed short hairpin RNA (shRNA) screen each set of shRNAs against each gene to yield a NES. Using

1046 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

C Loss-of-function Gain-of-function

Phenotype Proliferation Transformation Invasion

Assay shRNA screen Anchorage-independent Matrigel invasion growth

Experimental Cancer cell lines with or HMLE–MEKDD cells model without 3q26 amplification

D Centromere 3q26

3q26 Cell Tissue amp line origin Yes T47D Breast Yes HCC1937 Breast Yes H3255 Lung Yes HCC95 Lung Yes H1819 Lung Yes H28 Lung Yes TE6 Esophageal Yes RPMI8226 Myeloid Yes Fu-Ov-O1 Ovarian Yes COLO320 Colorectal Yes MCF7 Breast No MDA-MB-231 Breast No ZR75-1 Breast No HMEC Breast No HCC364 Lung No DLD1 Colorectal Chromosome 3

Figure 1. (Continued) C, a summary of the screens conducted to identify cancer driver genes residing in the minimal common amplifi ed region of 3q26. D, cell lines with normal or amplifi ed levels of 3q26 used for the proliferation screen. The amplifi cation data of each cell line are illustrated as above for A. An arrow marks the 3q26 region.

this analysis, we found that suppression of GPR160 , TLOC1 , age-independent growth and Matrigel invasion, that measure TNIK, and PHC3 selectively inhibited the proliferation of cells phenotypes associated with malignant transformation. harboring the 3q26 amplifi cation, with a false discovery rate In the anchorage-independent growth assay, TLOC1 was (FDR) of less than 0.25 ( Fig. 2A ). We confi rmed that the degree the only gene that induced increased anchorage-independent of gene suppression induced by TLOC1-specifi c shRNAs growth in HMLE–MEKDD cells in a manner comparable with correlated with the effects on cell proliferation in cells that that observed when we expressed a myristoylated version of harbor 3q26 copy number gain (Supplementary Fig. S1). AKT1 (>2 SD above the median; Fig. 2B ). We note that three In parallel to these studies, we overexpressed each of the genes, PLD1 , PRKCI , and MYNN , scored >1 SD over the median. genes in immortalized human mammary epithelial cells In addition, we carried out proliferation experiments on a set (HMLE) expressing an activated allele of MAP2K1 (MEKDD). of cell lines that do or do not harbor the 3q26 amplifi cation. By In prior studies, we showed that these cells (HMLE–MEKDD) stably expressing shRNAs, we found that cell lines with 3q26 do not grow in an anchorage-independent manner nor do amplifi cation required the expression of TLOC1 for prolifera- they form tumors in animals, but the expression of oncogenes tion as compared with cells that lack this amplifi cation ( Fig. 2C ). such as HRAS , AKT1 , and IKBKE rendered these cells tumori- We concluded that TLOC1 expression was required for the genic (9 ). As such, these cells serve as an experimental model proliferation of cell lines that harbor the 3q26 amplifi cation. for mammary epithelial transformation. We then assessed the In contrast, when we assessed whether expression of each consequences of expressing these genes in two assays, anchor- of these genes affected the capacity to induce invasion in a

SEPTEMBER 2013CANCER DISCOVERY | 1047

RESEARCH ARTICLE Hagerstrand et al.

AB100 90 80 SKIL EVI1 PRKCI EIF5A2 FNDC3B SAMD7 shRNAs genes LRRC34 LRRC31 MDS1 PHC3 GPR160 CLDN11 MYNN ARPM1 SLC7A14 SLC2A2 TLOC1 TNIK PLD1 70 2 2 60 Median + 2 SD 50 40 30

1 Colony number 20 Median 10 0 1 0 Survival score SKIL PLD1 AKT1 PHC3 MDS1 TERC PRKCI MYNN TLOC1 ARPM1 SAMD7 LRRC31

1 GPR160 CLDN11 FNDC3B SLC7A14 myr-AKT1 Normalized enrichment score FDR 0.21 0.21 0.24 0.75 0.75 0.07 0.02 0.01 0.02 0.16 0.30 0.35 0.43 0.43 0.65 0.71 0.80 0.35 0.16

TLOC1 TLOC1 shGFP sh 1159 sh 1267 TLOC1 PLD1 SLC7A14 myr-AKT1 C 3q26 amplified

5 HCC1937 6 H1819 6 H28 4 5 5 D 4 P 4 P 3 < 0.05 < 0.05 180 P < 0.01 P < 0.05 3 3 160 2 P < 0.01 2 2 140

Cell doublings 1 1 120 1 Median + 2 SD 0 0 0 100 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 80

Cell number 60 3q26 amplified Normal copy number of 3q26 40 Median 14 8 COLO320 MDA-MB-231 35 HCC364 20 12 30 0 6 10 P < 0.05 25 YFP

8 SKIL

20 PLD1 PHC3 SOX2 TERC MDS1

4 MYNN

P PRKCI shGFP

= n.s. TLOC1 EIF5A2 shEcad SAMD7 ARPM1

6 PIK3CA

15 SLC2A2 LRRC31 CLDN11 GPR160 FNDC3B 4 2 10 SLC7A14 Cell doublings 2 5 0 0 0 0 1 2 3 4 5 6 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 Days Days Days

Figure 2. Identifi cation of TLOC1 and SKIL as tumor driver genes in 3q26. A, RIGER analysis. Differential proliferation scores for each shRNA as compared between cell lines with normal or amplifi ed 3q26 levels are represented by blue lines. Survival scores less than 1 indicate a relative reduced proliferation in cells with 3q26 amplifi cation. The NES is represented by a red line and is calculated for each gene based on the proliferative effect of all shRNAs against that gene. FDRs are listed below. All the proliferation effects were normalized against 10 different GFP-targeting shRNAs. B, anchorage- independent growth for HMLE–MEKDD cells expressing each gene from the minimal common region. Each bar shows average number of colonies. Myris- toylated AKT1 (Myr-AKT1) was used a positive control. The bottom and top dotted horizontal lines indicate the median and median + SD colony number for the tested genes. Representative images of formed colonies are shown below the graph. C, effect of TLOC1-specifi c shRNAs on the proliferation of four cell lines with 3q26 amplifi cation and two cell lines with diploid 3q26 level. D, Matrigel-induced invasion in HMLE–MEKDD cells overexpressing genes from the common amplifi ed region. The bars indicate the number of invaded cells. The lower dotted horizontal line represents median number of invaded cells and the upper dotted line the median + 2 SD. ( continued on following page)

Transwell Matrigel invasion assay, we found that SKIL induced to tumor initiation and implicate both TLOC1 and SKIL as signifi cant invasion ( Fig. 2D and E ; P = 0.02; Supplementary potential targets of this amplifi ed region. Fig. S2), more than what we observed when we suppressed CDH1 (E-cadherin). These observations identify SKIL as a Expression of TLOC1 and SKIL Correlates with gene resident in the 3q26 amplicon involved in increased cell 3q26 Copy Number Gain invasion. When we assessed TLOC1 protein expression in the These observations implicated TLOC1 and SKIL as genes HMLE–MEKDD cells by immunoblotting, we found that that contribute to distinct cancer phenotypes. To determine TLOC1 migrated faster than the predicted size of 46 kD whether overexpression of TLOC1 or SKIL suffi ced to confer ( Fig. 2G ). We sequenced the construct used in the screen tumorigenic growth to immortalized cells, we overexpressed and found that it encoded a novel 220-amino acid version of each of these genes in murine embryonic fi broblasts and TLOC1, which represents a splice variant lacking amino acids assessed tumorigenicity in a subcutaneous tumor assay. Simi- 182 to 360 compared with the reported TLOC1 isoform in lar to what we observed when we expressed the breast cancer Ensembl (ENSP00000337688). This splice variant excludes oncogene IKBKE in these cells (12 tumors per 12 sites; ref. exons 6 and 7 and includes a fusion of exon 5 into the middle 9 ), we found that TLOC1 (six tumors per 12 sites) and SKIL of the last exon 8 by intraexonal splicing (GenBank acces- (six tumors per 12 sites) induced tumor formation ( Fig. 2F ). sion number KC005990). To determine whether this smaller These observations show that TLOC1 and SKIL contribute mRNA isoform is expressed in cancer cell lines, we cloned

1048 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

E G H J K IB:V5 IP:Flag IP:Flag 250 4 P = 0.02 150 3 SKIL Control TLOC1 Control IB: SKIL IB: TLOC1 100 2 75 HMEC TLOC1 220 HMEC MDA-MB-231 HCC364 H28 MCF7 T47D HCC1937 HCC95 H1819 DLD1 SW480 COLO320 TE6 FUOV1 100 399 1

50 +/– +/– 3q26 status 100 75 –––– + ++++–– ++ TLOC1 399/actin SKIL 0 75 37 220 50 25 50 TLOC1 399 50 hcRED Short exposure 37 TLOC1 20 37 300 P = 0.04 25 200 25 50 TLOC1 399 100 hcRED 37 TLOC1 220 Long exposure 0 TLOC1 399 TLOC1 220

Actin TLOC1 220/actin F I 25 P = 0.41 8 20

) 15 3 2,500 6 10

2,000 HMEC SKIL HMEC MDA-MB-231 HCC364 H28 MCF7 T47D HCC1937 HCC95 H1819 DLD1 SW480 COLO320 TE6 FUOV1 4 SKIL/actin 5 +/– 1,500 –––– +++++ ––+/– ++ 0

2 P < 0.01 Normal amplified 3q26 1,000 100

Fold colony number 75 500 0 SKIL

Tumor volume (mm 0 50 Actin hcRED SKIL IKBKE TLOC1 Control TLOC1 220 TLOC1 399

Figure 2. (Continued) E, immunoblots for Flag-epitope–tagged SKIL immune complexes (IP) from cells expressing SKIL used in D. F, tumor formation of NIH3T3 cells stably expressing control vector (hcRED), TLOC1, SKIL, or IKBKE as assessed at 8 weeks. G, immunoblots for Flag-epitope–tagged TLOC1 in Flag-bead immune complexes (IP) from TLOC1-overexpressing cells used in D. Arrows indicate specifi c bands. H, V5-immunoblots for V5-tagged 399– or 220–amino acids splice variants of TLOC1 and hcRED in HMLE–MEKDD cells. Arrows indicate specifi c bands. I, the 220- and 399-amino acids splice variants of TLOC1 signifi cantly induced anchorage-independent growth in HMLE–MEKDD as compared with hcRED control (P < 0.01, Student t test). Bars indicate average fold colony number. J, expression levels of endogenous 399- and 220-amino acid forms of TLOC1 (top) and SKIL (bottom) in a panel of cell lines. Cell lines with 3q26 amplifi cation are indicated with red text and a + sign and cells with normal 3q26 status with black text and a − sign. Cell lines that were determined by qRT-PCR on genomic DNA to harbor moderately increased copy number of 3q26 are indicated with a +/− sign. HMEC cells expressing the short form of TLOC1 were loaded in parallel as immunoblot controls. K, comparison in expression levels of the 399- or 220-amino acid forms of TLOC1 or SKIL in cell lines with normal or increased 3q26 copy number (P = 0.02, 0.04, and 0.41, Student t test). Plots showing relative expression levels between cell lines with normal or increased 3q26 copy number. The longer parallel bar represents the mean expression and the whiskers SD. Quantifi cation of TLOC1 and SKIL expression levels relative to actin are from the previous panel. The values are standardized to expression levels in the HMEC cells. n.s., not signifi cant. Error bars, 1 SD. hcRED, Heteractis crispa Red; IP, immunoprecipitation; IB, immunoblot.

TLOC1 from a T47D cell cDNA library and found both the (CCLE; ref. 10 ), which consisted of 967 cancer cell lines. 3q26

short (TLOC1 220) and long (TLOC1 399) splice variants of amplifi cation was defi ned as having a log2 copy number ratio TLOC1 composed of 220 and 399 amino acids, respectively. We of more than 0.3 for SKIL . From this cell line classifi cation, we also showed that expression of either form in HMLE–MEKDD then conducted a comparative marker selection test, based on cells ( Fig. 2H ) had a similar capacity for inducing signifi cant a t test metric, on the gene expression data available for these anchorage-independent growth (P < 0.01; Fig. 2I ). These obser- samples to identify genes that signifi cantly differed between vations indicate that the amino acids encoded by exons 6 and these two classes. We found that TLOC1 and SKIL transcript 7 are not required for TLOC1-induced transformation. levels, out of a total of 20,500 transcripts, were the third and To investigate whether gene expression or protein levels cor- 64th highest differentially expressed between cell lines with related with amplifi cation status, we conducted two analyses. increased or normal 3q26 copy number, respectively (Supple- To assess whether cell lines that harbor 3q26 amplifi cations mentary Table S2). These observations confi rmed that TLOC1 exhibit increased gene expression levels of TLOC1 or SKIL, we and SKIL expression levels correlated with amplifi cation of conducted comparative marker selection using a dataset con- 3q26 in approximately 900 cancer cell lines. sisting of cancer cell lines with matched gene expression data To investigate whether 3q26 amplifi cation also correlated and amplifi cation data from the Cancer Cell Line Encyclopedia with TLOC1 or SKIL protein expression in the cell lines used for

SEPTEMBER 2013CANCER DISCOVERY | 1049

RESEARCH ARTICLE Hagerstrand et al. our screen, we investigated the TLOC1 and SKIL protein expres- ( 13 ). To investigate how TLOC1 induced cell transforma- sion in 13 cell lines. We found signifi cantly higher TLOC1 pro- tion, we isolated TLOC1-interacting proteins by expressing a tein expression levels (both TLOC1 399 amino acids and TLOC1 V5-epitope–tagged version of full-length TLOC1 in HMLE– 220 amino acids) in amplifi ed cell lines as compared with cell MEKDD cells, isolated V5-immune complexes (Fig. 3A ), and lines that harbor normal copy number (P = 0.02 and 0.04; Fig. 2J identifi ed coimmunoprecipitated proteins by mass spec- and K ). We did not fi nd a strict correlation between SKIL protein trometry. We found 127 proteins that uniquely precipitated levels and SKIL copy number (P = 0.41; Fig. 2J and K ). Specifi - with TLOC1 compared with the unrelated protein luciferase cally, we found that two cell lines (H28 and COLO320) that har- (Supplementary Table S3). Eighteen of these proteins were bor moderately increased copy number at 3q26 by quantitative identifi ed by at least four unique peptides (Fig. 3B and Sup- real-time PCR (qRT-PCR) on genomic DNA did not exhibit plementary Table S3). The peptide coverage of these 18 pro- higher expression of SKIL protein (Supplementary Fig. S3). teins ranged from 3% to 33% ( Fig. 3B ). Of these 18 proteins, Together, these observations confi rmed that the transcript levels four have previously been found as interacting proteins with of TLOC1 and SKIL correlated with 3q26 amplifi cation. the Drosophila homolog of TLOC1, Trp1: rb (AP3B1), Trp1 (TLOC1), msk (IPO7), and CGI0077 (DDX5; Supplementary Identifi cation of TLOC1-Interacting Proteins That Table S4). We also identifi ed HSPA5 (Bip, Kar2p), which has Are Essential for Transformation previously been described to associate with TLOC1 ( 14 ). TLOC1 is part of the translocon complex ( 11 ), a channel When we examined the top 18 identifi ed proteins, we found complex in the endoplasmic reticulum (ER) through which enrichment in proteins implicated in translation ( Fig. 3C and newly synthesized proteins are transported into the ER lumen. Supplementary Table S3). Specifi cally, four of 18 proteins have This complex includes SEC61, TLOC1, and SEC63. TLOC1 been reported to be involved in translation: HNRNPM, DDX3X, and SEC63 have been suggested to recruit newly synthesized PABPC1, and EIF3CL ( 15–18 ). To identify proteins essential for polypeptides emerging from the ribosome ( 12 ). These obser- TLOC1-induced cell transformation, we suppressed each of the vations implicate TLOC1 in cotranslational translocation associated proteins with two independent shRNAs and tested for

A V5 IP D 120 100

M.W. V5-LUC V5-TLOC1 80 V5 IP 60 40 Colony number

SYPRO stain V5-LUC V5-TLOC1 20 100 75 0 75 LUC 50 50 TLOC1 399 shIPO7_1 shIPO9_1 37 shIPO9_2 shIPO7_2 shGFP684 shDDX5_1 shDDX5_2 shPKM2_1 shPKM2_2 shLMNA_1 shLMNA_2 shEIF3C_1 37 shEIF3C_2 shTGFBI_1 shTGFBI_2 shHSPA5_2 shHSPA5_1 shDDX3X_2 shDDX3X_1 shCAND1_2 shCAND1_1 shSRRM2_1 shTMOD3_2 shTMOD3_1 shSRRM2_2 shYWHAE_1 shYWHAE_2 shPABPC1_1 shPABPC1_2 shHNRNPM_1 shHNRNPM_2 shATP6V1A_1 shATP6V1A_2 25 hcRED/shGFP684 TLOC1 Silver stain V5-blot IP:V5 C27 N149 N81 Δ Δ B C E F Δ G Unique Total peptides peptides Protein Coverage HNRNPM IP:V5 16 DDX3X 16 27 ATP6V1A 33% PABPC1 14 16 27 HNRNPM 24% EIF3CL 12 n.s. TLOC1 220- LUC TLOC1 399 TLOC1 220- TLOC1 220-

13 23 IPO7 18% LUC TLOC1 10 13 18 LMNA 22% IB:V5 IB:DDX3X 8 P = 0.006 12 15 CAND1 14% 8 15 SEC62 15% 6 7 9 DDX3X 14% 75 4 Colony number 7 9 HSPA5 14% LUC 2 6 7 TNPO1 7% DDX3X 50 TLOC1 399 6 6 SRRM2 3% 0 6 7 PABPC1 12% Translation

Nuclear protein import Δ YFP Δ N81 6 6 PKM2 15% TLOC1 220- C27 Δ C27 hcRed 5 5 EIF3CL 7% Structural 25 TLOC1 220-ΔN81 Δ N149 4 5 IPO9 6% Metabolism 20 Δ 4 4 TGFBI 9% Protein transport TLOC1 220- N149 TLOC1 399 4 7 DDX5 6% Protein stability IB:TLOC1 4 4 TMOD3 19% Splicing TLOC1 75 DDX3X TLOC1 220- TLOC1 220-

4 8 YWHAE 14% Other TLOC1 220-

Figure 3. DDX3X association with the TLOC1 is essential for TLOC1-induced transformation. A, analysis of coimmunoprecipitated TLOC1-associated proteins visualized by silver or SYPRO staining. The V5-immunoblot shows immunoprecipitated V5-tagged TLOC1 from 2% of the lysate. B, 18 TLOC1- associated proteins were identifi ed by at least four unique peptides. C, category distribution of the proteins identifi ed in the mass spectrometry analysis. D, the effect of suppressing TLOC1-associated proteins on TLOC1-induced anchorage-independent growth in HMLE–MEKDD cells. Black bars indicate number of colonies. The top and bottom dashed horizontal lines indicate mean ± 1 SD, respectively. E, DDX3X and TLOC1 immunoblots for V5 and DDX3X immune complexes from V5-TLOC1- or V5-luciferase–overexpressing cells. Rabbit immunoglobulin G was used as a negative control for the DDX3X immunoprecipitation and V5-tagged luciferase for the V5 experiments. F, coimmunoprecipitation of DDX3X with different V5-tagged TLOC1 truncation mutants. TLOC1 mutants and DDX3X were detected by V5 and DDX3X immunoblotting, respectively. G, removal of the 81 fi rst N-terminal amino acids of TLOC1 signifi cantly reduced its ability to induce anchorage-independent growth in HMLE–MEKDD cells (P = 0.006, Student t test). The bars indicate average colony number. Yellow fl uorescent protein (YFP) and Heteractis crispa Red (hcRED) were used as negative controls. For all panels, arrows indicate specifi c bands for the immunoblots. n.s., not signifi cant. Error bars, 1 SD. IB, immunoblot.

1050 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

their effect on anchorage-independent growth in TLOC1-overex- TLOC1 Affects Cap-Dependent versus pressing HMLE–MEKDD cells. Of these candidates, DDX3X and IRES-Dependent Translation SRRM2 signifi cantly reduced anchorage-independent growth To assess whether TLOC1 and DDX3X affected protein ( Fig. 3D ). Because DDX3X previously has been reported to be translation, we manipulated TLOC1 and DDX3X expression involved in translation ( 17 ) and gave the strongest phenotype, we and measured the effects on translation using a bicistronic focused on DDX3X for subsequent experiments. translation reporter system. In this system, cap-dependent To confi rm that DDX3X interacted with TLOC1, we reiso- translation controls Renilla luciferase expression and an inter- lated V5-tagged immune complexes and detected endog- nal ribosome entry site (IRES) regulates Firefl y luciferase enous DDX3X by immunoblotting (Fig. 3E ). To identify expression (ref. 19 ; Fig. 4A ). We found that TLOC1 overex- domains of TLOC1 that were responsible for the associa- pression signifi cantly reduced IRES-dependent translation tion with DDX3X, we generated a set of TLOC1 truncation ( P = 0.002), which resulted in a signifi cant net increase of mutants. These truncation mutants were generated from the the ratio of cap-dependent to IRES-dependent translation 220-amino acid–long splice variant of TLOC1 ( Fig. 3F ). We ( P = 0.01; Fig. 4B ). Furthermore, we tested this reporter sys- found that the 81 N-terminal amino acids were necessary for tem with the TLOC1 truncation mutants and found that the the interaction of TLOC1 with DDX3X. In parallel, we tested full-length form and the shorter C-terminal truncation ver- the transforming ability of each of these truncation mutants sion signifi cantly inhibited the IRES-dependent translation and found that the 81 N-terminal amino acids also were nec- reporter, whereas the 149-amino acid N-terminal truncation essary for TLOC1 to induce anchorage-independent growth mutant, which does not interact with DDX3X, failed to (Fig. 3G ). These observations showed that the 81 N-terminal affect this reporter (Fig. 4C ). We also tested whether sup- amino acids are required for the association of TLOC1 with pression of DDX3X affected translation in cells expressing DDX3X and for TLOC1-induced cell transformation. TLOC1. We found that the increased cap/IRES–dependent

A IRES C 50 70 1.25

60 cap Renilla (RL) Firefly (FL) AAAA 40 1 P = 0.04 50 P = 0.03 B 30 0.75 1.5 1.5 40

12 P = 0.04 P = 0.02

20 30 1 1 8 0.5 20 0.5 10 0.25 0.5 4 10 IRES-dependent translation Cap-dependent translation P = 0.002 (FLU) (RLU) P = 0.01 Cap/IRES-dependent translation

(RLU/FLU) 0 0 0 0 0 0 BFP BFP BFP Cap-dependent translation hcRED hcRED hcRED IRES-dependent translation Cap/IRES-dependent translation TLOC1 399 TLOC1 399 TLOC1 399 TLOC1 399 TLOC1 399 TLOC1 399 TLOC1 220 Δ C27 TLOC1 220 Δ C27 TLOC1 220 Δ C27 TLOC1 220 Δ N149 TLOC1 220 Δ N149 TLOC1 220 Δ N149

D 2 EF7mG- 7mG- G +GFs –GFs Input beads Input beads Input 7mG-beads 1.8 1.6

1.4 BFP TLOC1 BFP TLOC1 BFP TLOC1 BFP TLOC1 BFP TLOC1 BFP TLOC1 BFP TLOC1 BFP TLOC1 1.2 P = 0.03 p-EIF4EBP1

P = 0.006 V5-TLOC1 1 (Thr37/45) 0.8 EIF4E 0.6 V5-BFP EIF4EBP1 0.4 IB:V5 IB:EIF4E 0.2 EIF4G Cap/IRES-dependet translation 0 +GFs –GFs Actin

IB:TLOC1 IB:DDX3X BFP TLOC1 BFP TLOC1 long exposure p-EIF4EBP1 (Ser65) hcRED/shGFP hcRED/shGFP TLOC1/shGFP TLOC1/shGFP V5-TLOC1 TLOC1

TLOC1/shDDX3X_1 EIF4EBP1

IB:TLOC1 IB:actin short exposure

Figure 4. TLOC1 increased the cap- versus IRES-dependent translation ratio by decreasing IRES-dependent translation. A, illustration of the bicistronic translation reporter used to measure translation levels. The Renilla luciferase reporter is driven by cap-dependent translation and the fi refl y luciferase by IRES-dependent translation. B, TLOC1 overexpression signifi cantly increased the ratio of cap- versus IRES-dependent translation by inhibition of IRES- dependent translation (P = 0.01, Student t test). The ratio of cap/IRES–dependent translation was calculated and is illustrated in the right. C, overexpression of the transforming TLOC1 truncation mutant ΔC27 signifi cantly (P = 0.03, Student t test) inhibited IRES-dependent translation as compared with the non- transforming variant ΔN149. D, suppression of DDX3X in TLOC1 overexpressing cells signifi cantly reversed the TLOC1-induced change in translation ratio (P = 0.03, Student t test). E, TLOC1 and DDX3X binds to 7-methylated GTP beads, and TLOC1. F, TLOC1 overexpression increased EIF4G protein expression. G, TLOC1 overexpression in HMLE–MEKDD cells decreased EIF4EBP1-phosphorylation on Threonines 37 and 45 and Serine 65. Lysates were prepared from TLOC1- or BFP-overexpressing HMLE–MEKDD cells, which had been grown with (+GFs) or without (−GFs) growth factors for 24 hours. Error bars, 1 SD. BFP, blue fl uorescent protein; RLU, relative light unit; FLU, fl uorescent light unit; IB, immunoblot.

SEPTEMBER 2013CANCER DISCOVERY | 1051

RESEARCH ARTICLE Hagerstrand et al. translation induced by TLOC1 expression was signifi cantly expression. By comparative marker selection analysis ( 23 ), reversed upon DDX3X suppression ( Fig. 4D ). Together, we identifi ed the top 50 genes that were correlated with these observations show that TLOC1 levels modulate cap/ either high SKIL expression or low SMAD4. Among these IRES–dependent translation, which is reversed by suppres- genes, 13 were present in both lists, which is signifi cantly sion of DDX3X. higher than what is expected by chance (P < 0.001, bino- Eukaryotic initiation factor (EIF) 4G binds to the mRNA mial distribution test; Fig. 5A ; Supplementary Table S5). At cap structure to initiate translation. EIF4G is a scaffold pro- least six of the intersecting genes have been reported to be tein that interacts with EIF4E and promotes the assembly of involved in regulation of invasion, including TCF8 (ZEB1 ; EIF4E and EIF4A (DDX2A) to create the initiation complex ref. 24 ), EHF (25 ), FOXQ1 (26 ), MARVELD3 (27 ), ST14 (28 ), EIF4F. EIF4A, like DDX3X, is also an RNA helicase. Several and TWIST1 (29 ). These observations suggest that high SKIL other proteins, for example, EIF4EBP1, bind to or modify or low SMAD4 gene expression affected signaling pathways this complex to alter its activity and specifi city. Hypophos- implicated in invasion. phorylated EIF4EBP1 is known to inhibit translation by Because we found an overlap between the expression of sequestering EIF4E. Phosphorylation of EIF4EBP1 releases genes related to invasion and epithelial-to-mesenchymal EIF4E and allows EIF4E to assemble the EIF4F complex and transition (EMT) in cells having high SKIL or low SMAD4 induce translation (20 ). EIF4EBP1 phosphorylation levels expression, we tested the role of EMT master regulators in are regulated by the mTOR pathway. invasion. TWIST1, SNAIL ( SNAI1), and SLUG ( SNAI2) are To determine whether TLOC1 and DDX3X bind translation- well-established transcriptional regulators of invasion and related complexes, we isolated 7-methylated GTP-binding EMT ( 30 ) and have been reported to be components of the complexes and assessed the associated levels of EIF4E, EIF4G, TGF-β signaling pathway ( 31 ). To test whether invasion TLOC1, and DDX3X in HMLE–MEKDD cells expressing induced by expression of SKIL was dependent on the expres- TLOC1. We found that all of these proteins bound 7-meth- sion of any of these genes, we suppressed the expression ylated GTP but not actin or blue fl uorescent protein (BFP), of these genes using specifi c shRNAs in SKIL-expressing which served as controls for nonspecifi c binding (Fig. 4E ). HMLE–MEKDD cells. We found that suppression of SLUG We also found that TLOC1 overexpression increased EIF4G signifi cantly inhibited invasion ( P = 0.03; Fig. 5B ) but did levels as well as the amount of EIF4G1 associated with the not affect cell proliferation (Fig. 5C ). We also found that 7-methylated GTP beads (Fig. 4F ). These observations show SLUG protein levels were upregulated in SKIL-expressing that TLOC1 and DDX3X interact with a translation protein cells (Fig. 5D ). To investigate whether SLUG had any effect complex through 7-methylated GTP and that TLOC1 overex- on SKIL-induced gene expression, we investigated how pression increases EIF4G protein levels as well as the levels of SLUG suppression affected a set of TGF-β and invasion- EIF4G that associate with 7-methylated GTP beads. related genes. We found that several genes were upregulated To investigate whether TLOC1 expression had any effects by SKIL and reversed upon SLUG suppression, including on known regulators of translation, we investigated the phos- SNAIL, TWIST1 , PLAUR , and VIM ( Fig. 5E ). Furthermore, phorylation status of EIF4EBP1. We found that TLOC1 upon overexpression of SLUG, the same subset of genes was overexpression led to decreased EIF4EBP1 phosphorylation upregulated ( Fig. 5F ). at threonines 37 and 45 and at serine 65 ( Fig. 4G ). In aggre- Because SKIL has been suggested to inhibit SMAD4 function, gate, these observations provide evidence that overexpression we tested the effect of suppression of SMAD4 in our system. of TLOC1 affects the ratio of cap-dependent translation We found that suppression of SMAD4 induced invasion and through interactions with proteins involved in regulating that suppression of SLUG signifi cantly reversed this phenotype protein translation, including DDX3X. (P = 0.02 and 0.05; Fig. 5G ). These observations implicate SKIL as a regulator of SMAD4-mediated invasion, which required SLUG. SLUG Is Required for SKIL-Induced Cell Invasion SKIL has been reported to inhibit the TGF-β signaling axis TLOC1 and SKIL Cooperate to Induce by binding to SMAD4 and SMAD2, recruiting the transcrip- Transformation tional corepressor N-CoR and suppressing SMAD-induced Because TLOC1 and SKIL frequently are coamplifi ed, we transcription (21 ). SMAD4 is a tumor suppressor gene fre- investigated whether manipulating the expression of these quently inactivated in pancreatic cancer, and has been corre- genes induced cell transformation. We found that overex- lated with an invasive phenotype (22 ). To determine whether pression of TLOC1 and SKIL together induced a signifi cant overexpression of SKIL or suppression of SMAD4 induced increase in anchorage-independent growth (P = 0.003; Fig. 6A ), similar cell states, we analyzed existing gene expression data which failed to be explained by a general increase in cell prolif- from the CCLE ( 10 ). After downloading the data from the eration (Fig. 6B ). CCLE, we calculated the average expression and SD for SKIL Having investigated cooperative effects of SKIL and and SMAD4 on a set of 634 samples (all CCLE cell lines TLOC1, we tested the effects of coexpressing TLOC1 and excluding samples derived from hematologic malignancies). SKIL with PIK3CA or SOX2 from neighboring amplifi cation High SKIL expression was defi ned as one SD above aver- peaks. We found that both PIK3CA and SOX2 overexpres- age SKIL expression and low SMAD4 expression one SD sion induced anchorage-independent growth in HMLE– below average SMAD4 expression. Ninety-three cell lines MEKDD cells but exhibited no cooperative effect with were identifi ed to have high SKIL expression, and 59 cell TLOC1 or SKIL ( Fig. 6C ). These observations suggest that lines exhibited low SMAD4 levels. Of these cell lines, 15 cell TLOC1 and SKIL induce transformation independently of lines possessed both high SKIL expression and low SMAD4 PIK3CA or SOX2.

1052 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

ACLow SMAD4 High SKIL B D expression expression 100 70 profile profile 80 60 hcRED SKIL 50 60 40 SLUG 37 13 37 40 30 P = 0.03 20 GFP Invaded cells sh 20 Cell doublings 10 shSLUG 0 0 TCF8 FOXQ1 SERPINB5 TMEM30B 0 10 Actin EHF EVA1 MARVELD3 RAB25 Days GRHL2 TACSTD2 ST14 SCNN1A sh SLUG TWIST1 sh SNAIL sh TWIST sh GFP 684 EFG 100 SKIL/shGFP 10 25 SKIL/shSLUG 20 10 15 1 10 = 0.02 P P = 0.05 1 5 Invaded cells Gene expression Gene expression 0 0.1 0.1 VIM VIM SLUG SLUG E-Cad E-Cad MMP2 N-Cad SNAIL SNAIL MMP2 N-Cad TWIST TWIST SERPINE1 SERPINE1 sh GFP /sh SLUG- SLUG- SLUG- SLUG- sh SMAD4 /sh GFP

dependent independent induced independent sh SMAD4 /sh SLUG

Figure 5. SKIL induces SLUG, which is required for SKIL-induced invasion. A, a Venn diagram showing the top 50 genes correlating with either high SKIL expression or low SKIL expression and the signifi cant intersect between these two lists (P < 0.001, binomial distribution test). The overlapping genes are listed below the diagram. B, SLUG is required for SKIL-induced Matrigel invasion. Suppression of SLUG signifi cantly reduced SKIL-induced invasion (P = 0.03, Student t test). Bars indicate average number of invaded cells. C, suppression of SLUG in SKIL-overexpressing cells had no effect on proliferation. Graph shows cell doublings over indicated amount of time. An shRNA-targeting GFP was used as negative control. D, SKIL overexpression induced SLUG expression. E, SKIL overexpression increased gene expression of invasion- and EMT-related genes of which a subset could be reversed by SLUG suppression, marked as SLUG-dependent. mRNA levels were determined by qRT-PCR. F, SLUG overexpression increased gene expression of the same set of genes, which were determined to be SLUG-dependent in E. G, suppression of SMAD4 signifi cantly induced Matrigel invasion (P = 0.02, Stu- dent t test). Suppression of SLUG reversed the invasion induced by suppression of SMAD4 (P = 0.05, Student t test). The bars represent average number of invaded cells. The experiments in this fi gure were carried out in HMLE–MEKDD cells except in A. Error bars, 1 SD.

DISCUSSION for its interaction with ribosomes ( 13 ). DDX3X has been reported to be part of a complex including PABPC1 (PABP), 3q26 is amplifi ed in several types of cancer, including lung, which we also identifi ed as a TLOC1-binding partner. This ovarian, and breast, and is correlated with poor prognosis and complex was shown to promote translation of a selected set of an invasive phenotype ( 32 ). Here, we have systematically inter- mRNAs ( 34 ). Taken together, these observations suggest that rogated the function of the genes harbored in the minimal TLOC1 and DDX3X are components of a macromolecular common amplifi ed region by suppressing or overexpressing complex involved in translational regulation. each of these in human cancer cell lines. We found that TLOC1 Several lines of evidence now implicate translation as a key expression induced anchorage-independent growth, whereas process perturbed in cancer cells ( 35 ). For example , MYC- SKIL induced invasion, and both of these genes induced tumor induced transformation has been shown to drive the shift from formation. Coexpression of TLOC1 and SKIL induced cooper- cap- to IRES-dependent translation during the cell cycle and ative cell transformation. Together, these observations identify lead to genomic instability (36 ). The fi nding that TLOC1 and TLOC1 and SKIL as transforming genes resident in 3q26 and DDX3X induced transformation through inhibition of IRES- provide additional evidence that recurrently amplifi ed regions dependent translation is consistent with this model. We note may harbor more than one gene involved in malignant trans- that mutations involving DDX3X have recently been described in formation ( 33 ). Although we have identifi ed TLOC1 and SKIL several cancer types, including head and neck cancer ( 37 ). as two genes resident in the minimal common amplifi ed region It remains unclear whether perturbation of translation con- at 3q26 that contribute to cell transformation, other genes in tributes to transformation through specifi c transcripts or global this region may contribute to other cancer-related phenotypes. dysregulation of translation. Several recent reports suggest that Prior work has implicated TLOC1 to be a component of the subsets of mRNAs are specifi cally regulated. For example, onco- translocon complex, which is an important component in pro- genic mTOR signaling has been shown to control the translation tein translation. Here, we found that TLOC1 interacts with a of a set of proinvasive transcripts in prostate cancer (38 ), and has number of proteins involved in translation, including DDX3X, also been reported to control translation of terminal oligopyri- whose expression was required for TLOC1-induced transfor- midine (TOP)-motif–containing mRNAs ( 39 ). These observa- mation. The interaction of TLOC1 and DDX3X requires the tions suggest that perturbation of translation control likely amino terminal end of TLOC1, a region previously implicated affects specifi c transcripts that contribute to transformation.

SEPTEMBER 2013CANCER DISCOVERY | 1053

RESEARCH ARTICLE Hagerstrand et al.

AB C 700 700 2.5 P < 0.01 BFP 2 600 600 SKIL )

6 1.5 500 500 (10 1 400

400 Cell number P 0.5 = 0.003 300 TLOC1 300 0 Colony number

Colony number 200 200 2.5 BFP 100 2 TLOC1 100 0 ) 1.5 6 0 (10 BFP BFP BFP

1 SKIL – – – SOX2 SOX2 SOX2 – mAKT mAKT mAKT – – – – – – Cell number PIK3CA PIK3CA PIK3CA – – 0.5 – SKIL SKIL

SKIL SKIL hcRED TLOC1 hcRED SKIL–BFP

0 SKIL hcRED TLOC1 hcRED TLOC1 hcRED–BFP TLOC1–BFP hcRED

0 10 20 TLOC1 hcRED–SKIL TLOC1–SKIL SKIL–TLOC1 hcRED–mAKT Days

Figure 6. TLOC1 and SKIL cooperate to induce anchorage-independent growth. A, HMLE–MEKDD expressing different combinations of genes were tested for anchorage-independent growth. The bars illustrate average number of colonies. A signiﬁ cant difference was detected between TLOC1 and control as illustrated in the graph (P = 0.003, Student t test), and was potentiated in combination with SKIL. B, growth rate of cells used in A showed no signiﬁ cant dif- ferences in growth rate. Cells were cultured, counted, and reseeded at indicated time points, and the cumulative increase in cell number is shown. C, TLOC1 or SKIL displayed no combinatorial effect in anchorage-independent growth with PIK3CA and SOX2. Bars display average colony number. Error bars, 1 SD.

We found that TLOC1 overexpression decreased EIF4EBP1 with PIK3CA or SOX2 , we expressed combinations of genes phosphorylation, which is predicted to induce decreased but did not observe any cooperation between PIK3CA or SOX2 translation. Overexpression of EIF4G also induces tumori- and TLOC1 and SKIL . These observations suggest that PIK3CA genic growth of NIH3T3 cells ( 40 ), and an inhibitor target- or SOX2 act independently of TLOC1 and SKIL ; however, we ing EIF4G, 4EGI-1, was shown to inhibit growth of human note that our experiments do not eliminate the possibility breast and melanoma cancer xenografts ( 41 ). The fi nding that these genes cooperate in other assays in vivo . Moreover, we that TLOC1 is not only amplifi ed in a signifi cant fraction of confi rmed that PIK3CA mutations occur at equal frequency in cancers but also contributes directly to cell transformation cells that harbor or lack increased copy number at 3q26. suggests that dysregulation of protein translation by pertur- We found that TLOC1 and SKIL cooperated to induce anchor- bation of TLOC1 or DDX3X contributes to tumorigenicity. age-independent growth. Cooperating cancer drivers in amplifi - We identifi ed a shorter splice variant of TLOC1 that also had cation peaks have previously been described. YAP1 and BIRC2 transforming capacity, an inhibitory effect on translation, and (cIAP1), which reside in the same amplicon in liver cancer, have an association with DDX3X. Because this form lacks the two been shown to have a cooperative effect on tumorigenesis ( 43 ). In transmembrane domains of TLOC1, we predicted that this addition, EGFR is frequently coamplifi ed with the neighboring form does not bind to the ER. However, despite these differ- gene SEC61G in glioma, and both have been shown to be required ences, we found that both isoforms induced cell transforma- for tumor cell survival ( 44 ). Of note, SEC61G and TLOC1 inter- tion. Two mRNA forms of the TLOC1 ortholog in Drosophila act in the translocation channel. In a similar manner, 8p22 has melanogaster, Trp1 , have been shown to be selectively expressed: a been reported to include a cluster of cooperating tumor suppres- 1.6-kb form, which is expressed in the male reproductive system, sors ( 33 ). Although we showed that overexpression of TLOC1 and a 2.2-kb form confi ned to other tissues ( 42 ). The difference affects protein translation and anchorage-independent growth, between the 2.2 and 1.6 kb transcripts is close to the difference and SKIL affects cell invasion, it remains unclear whether these of the 537 bases of the two TLOC1 forms. These observations functions cooperate or act in parallel in vivo. suggest that these two isoforms may have tissue-specifi c effects Moreover, further studies are necessary to determine whether but that both are transforming when inappropriately expressed. targeting these pathways will lead to clinical responses. Indeed, We also identifi ed SKIL in 3q26 as an inducer of inva- 4EGI-1 has been shown to inhibit growth of human breast and sion. SKIL has been reported to inhibit SMAD4 function melanoma cancer xenografts ( 41 ). In addition, inhibitors of by repressing its transcriptional activity by recruitment of DDX3X, which has also been reported to be important for HIV the transcriptional corepressor N-CoR ( 21 ). Although SKIL propagation, have been developed ( 45 ). Further work will be protein levels did not strictly correlate with copy number, we necessary to determine whether 3q26 amplifi cation predicts the showed that both overexpression of SKIL and suppression of response of cancer cells to these inhibitors. SMAD4 induced an invasive phenotype through upregula- In aggregate, these studies extend work suggesting that tion of SLUG, thus illustrating two common genomic events regions of recurrent SCNAs may harbor more than one driver in cancer that lead to inactivation of the TGF-β pathway and gene that participate in different aspects of tumor initiation subsequently induction of an invasive phenotype. or progression. Future efforts to interrogate such regions will PIK3CA and SOX2 are other oncogenes that also reside on not only require systematic interrogation of resident genes but the 3q arm and are often coamplifi ed with 3q26. To investi- also the use of multiple assays to assess potentially comple- gate whether TLOC1 or SKIL acted in a cooperative manner mentary phenotypes.

1054 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

METHODS with a normal amount of growth factors was added to the lower chamber. An shRNA-targeting CDH1 (E-cadherin; ref. 29 ) or overex- Copy Number Analysis pression of YFP served as positive or negative control, respectively. GISTIC profi les were created from Tumorscape with 3,131 tumor Invasion was determined after 24 to 48 hours by staining with Giemsa samples as previously described (1 , 6 ). To create amplifi cation dia- or Hoechst and counting the number of invaded cells with ImageJ grams of tumor samples and the screened cell lines, copy number (W.S. Rasband, U.S. NIH, Bethesda, MD; http://imagej.nih.gov/ij/ ). illustrations were made with the Integrative Genomics Viewer (46 ). The TCGA was also queried for amplifi cation of genes in the minimal Quantitative Real-Time PCR amplifi ed region of 3q26 (2 , 7 ). RNA was harvested using RNeasy (Qiagen). Complementary DNA was prepared using Advantage RT-for-PCR (Clontech). Quantitative Cell Culture, Vectors, and Lentiviruses PCR was conducted using SYBR (Applied Biosystems). Primers are listed HMLE cells expressing hTERT and SV40 Early Region have been in Supplementary Table S8. For the gene suppression experiments (Sup- described previously (47 ). To express or suppress genes, retroviral and plementary Fig. S1), MCF7 cells propagated in 24-well tissue culture lentiviral vectors were used (pBabe, pWzl, pLex, pLKO) as described plates were infected with individual shRNAs at a multiplicity of infec- < previously ( 48, 49 ). pBabe-Puro-Flag-DEST, pLEX-Puro-V5-DEST, and tion (MOI) 1. After infection, the cells were selected with puromycin pWzl-Neo-Flag-DEST constructs were generated by Gateway clon- and passaged to eliminate dead cells. RNA was purifi ed with RNeasy ing from pDONR223 constructs acquired from either the Human (Qiagen) and converted into cDNA by reverse transcription using ORFeome ( 50 ) or from BP cloning reactions of BP-tagged PCR prod- Superscript II (Invitrogen, Life Technologies). Gene expression levels ucts. Truncation mutants of TLOC1 were created by PCR amplifi ca- for each targeted gene and glyceraldehyde-3-phosphate dehydrogenase tion with BP-tagged primers (Supplementary Table S6) and BP ligation (GAPDH ) were determined for each shRNA-infected sample by RT-PCR (Invitrogen) into pDONR223 and subsequent LR ligation (Invitrogen) in a 7300 Real-Time PCR System (Applied Biosystems, Life Tech- into a viral destination vector. The identity of the cell lines used in this nologies). The relative expression level per gene was normalized against study were verifi ed by the CCLE. GAPDH levels and compared with a reference sample where GFP had been targeted as an unrelated shRNA control. Expression was deter- RNAi Experiments mined by single measurements from three different biologic replicates. For genomic qRT-PCR analysis, DNA was purifi ed from cells with shRNAs were obtained from the RNAi Consortium. The corre- QIAamp (Qiagen). Genomic levels were determined by RT-PCR in a sponding shRNAs are listed in Supplementary Table S7. Ten differ- 7300 Real-Time PCR System (Applied Biosystems, Life Technologies) ent GFP-specifi c shRNAs were used as unrelated negative controls. with primers specifi c for 3q26 (Supplementary Table S8). Primers Lentiviral infections were conducted as described previously ( 49 ). The specifi c for genomic LINE-1 were used to normalize for genomic suppression of each shRNA was determined in a parallel experiment input (Supplementary Table S8). where MCF7 cells were infected, selected, and cultured in a similar fashion as in the screen. The shRNA suppression effi ciency was deter- Immunoblotting mined by qRT-PCR. Primers are listed in Supplementary Table S8. The For immunoblot analyses, samples were harvested in 1% NP-40 effects of suppressing genes resident in 3q26 were calculated by RIGER or radioimmunoprecipitation assay buffer supplemented with pro- ( 8 ). RIGER is a method similar to Gene Set Enrichment Analysis ( 51 ) tease and phosphatase inhibitors. Samples were separated by SDS- to summarize the effects of multiple shRNAs into a single per-gene PAGE and transferred to nitrocellulose fi lters with the iBlot system score. First, we weighted each shRNA by its target suppression effi - (Invitrogen, Life Technologies). Antibodies used were: anti-Flag M2 ciency and computed a differential proliferation score (blue lines in (Sigma), anti-V5 (Invitrogen, Life Technologies), TLOC1 (SEC62; Fig. 2A ) for each shRNA according to the difference of mean prolif- #HPA014059), DDX3X (#HPA001648; Prestige Antibodies; Sigma), eration between the cells harboring 3q26 and controls. These scores SNON (SKIL; #4973), SLUG (#9589), EIF4G (#2469), EIF4E (#9742), were sorted from high to low and each gene was assigned an “enrich- EIF4EBP1 (#9644), pEIF4EBP1(Thr37/45; #9459), pEIF4EBP1(Ser65; ment” score (red lines in Fig. 2A ) according to how overrepresented #9456), (Cell Signaling Technology), and SNON (SKIL; H317; sc-9141; their shRNAs were in the sorted list using a Kolmogorov–Smirnov Santa Cruz Biotechnology). weighted statistic. The positive (negative) enrichment scores are normalized using the absolute value of the mean of the positive (negative) Proliferation Experiments values in a permutation-based null distribution. This null distribution HMLE–MEKDD cells expressing vector control or gene of interest was also used to generate nominal P values and FDR (see Fig. 2A ). The or cancer cell lines expressing shRNA against control or gene of inter- FDR is the expected proportion of false-positives among all queries. est were plated in 6-well plates. Cells were counted after indicated time and replated at the same amounts, and this cycle was repeated Anchorage-Independent Colony Growth as indicated. Increase in cell number was calculated and plotted as HMLE cells expressing the different genes were assayed for their accumulated increase in cell number. colony formation capacity as described previously ( 9 ). In addition to the 20 genes, we included myristoylated AKT1, which has been shown to TLOC1 Association Experiments collaborate with MEKDD and transform immortalized breast epithelial One hundred–milligram protein lysates were generated from cells, as a positive control. As negative controls, we used corresponding HMLE–MEKDD cells expressing V5-tagged LUC or TLOC1 by lysis in vectors expressing YFP (yellow fl uorescent protein), hcRED (Heteractis 1% NP-40 supplemented with protease and phosphatase inhibitors. crispa Red), LacZ (β-galactosidase), or LUC (luciferase). To test the func- Immune complexes were precipitated overnight with anti-V5-beads tion of TLOC1-associated proteins, each candidate was targeted with (Invitrogen). A fraction of the precipitated product was analyzed two shRNAs in TLOC1-overexpressing HMLE–MEKDD cells. on PAGE for silver staining and immunoblotting to confi rm that V5-tagged TLOC1 was isolated. The rest of the protein was separated Invasion Experiments on gels and stained by SYPRO Ruby protein gel stain, and lanes were Invasion capacity of cells was determined by plating 50,000 cells isolated. The immunoglobulin G bands were removed from the lanes in the upper chamber of Matrigel (BD Biosciences)-coated Transwell and the rest was submitted for mass spectrometry analysis. invasion chambers according to the manufacturer’s instructions. The Excised gel bands were subjected to in-gel trypsin digestion. Gel cells were seeded in media containing no growth factors, and media pieces were washed and dehydrated with acetonitrile and rehydrated

SEPTEMBER 2013CANCER DISCOVERY | 1055

RESEARCH ARTICLE Hagerstrand et al. in 50 mmol/L ammonium bicarbonate containing 12.5 ng/μL modi- Development of methodology: D. Hagerstrand, S.E. Schumacher, fi ed sequencing-grade trypsin (Promega) at 4°C. After 45 minutes, N. Ilic, Y. Shrestha, N.A. Spardy, D.A. Barbie, C.H. Mermel, J.P. Mesirov, trypsin solution was removed and replaced with 50 mmol/L ammo- W.C. Hahn nium bicarbonate solution and incubated at 37°C for 16 hours. Acquisition of data (provided animals, acquired and managed Peptide-containing ammonium bicarbonate extract was removed, patients, provided facilities, etc.): A. Tong, S.Y. Kim, C.H. Mermel, and remaining peptides were eluted with 50% acetonitrile containing L.A. Garraway 1% formic acid and dried in a speed-vac. Analysis and interpretation of data (e.g., statistical analysis, Samples were reconstituted in 2.5% acetonitrile containing 0.1% biostatistics, computational analysis): D. Hagerstrand, A. Tong, formic acid and separated over a nanoscale reverse-phase high- R.R. Shen, H.W. Cheung, A.O. Giacomelli, J. Rosenbluh, C.H. Mermel, performance liquid chromatography (HPLC) capillary packed with B.A. Weir, P. Tamayo, J.P. Mesirov, W.C. Hahn C18 spherical silica beads ( 52 ). Sample was loaded in an equilibrated Writing, review, and/or revision of the manuscript: D. Hagerstrand, Famos auto sampler (LC Packings). A gradient was formed and pep- A. Tong, F. Vazquez, A.O. Giacomelli, A.C. Schinzel, N.A. Spardy, tides were eluted with increasing concentrations of solvent (97.5% B.A. Weir, J.P. Mesirov, R. Beroukhim, W.C. Hahn acetonitrile containing 0.1% formic acid). Administrative, technical, or material support (i.e., reporting or Eluted peptides were subjected to electrospray ionization and organizing data, constructing databases): W.C. Hahn entered into an LTQ Velos ion-trap mass spectrometer (Thermo Study supervision: W.C. Hahn Fisher Scientifi c). Peptides were detected, isolated, and fragmented to produce a tandem mass spectrum of specifi c fragment ions for each Acknowledgments peptide. Peptide sequences were determined by matching protein The authors thank members of the Hahn and Cichowski labora- databases with the acquired fragmentation pattern by SEQUEST tories for helpful discussions and P. Hammerman, M. Pop, and H. (Thermo Fisher Scientifi c; ref. 53 ). Widlund for useful input. The authors also thank R. Tomaino at the Taplin Mass Spectrometry facility for carrying out the mass spec- Translation Measurements trometry experiments. Cap- and IRES-dependent translation were determined by transient transfection of the bicistronic translation reporter into HMLE–MEKDD Grant Support cells stably expressing BFP as unrelated control or TLOC1. Cells were This work was supported in part by NIH/National Cancer Institute plated in 96-well cell assay plates at 100,000 cells per well, transfected with (NCI) grants R01 CA130988 (to W.C. Hahn), RC2 CA148268 (to W.C. pDL/N, and incubated for 48 hours. Renilla and Firefl y luciferase activity Hahn), and U54 CA112962 (to W.C. Hahn); the Sweden–America Founda- were determined with Dual-Glo Luciferase Assay system (Promega). tion (to D. Hagerstrand), and the Ernst O. Ek Fund (to D. Hagerstrand).

Cap-Binding Assay Received January 14, 2013; revised June 5, 2013; accepted June 6, Cap-associated proteins were precipitated with 7-methylated GTP 2013; published OnlineFirst June 13, 2013. beads from cell lysates derived from TLOC1- or BFP-overexpressing cells. 7-Methylated GTP sepharose beads (GE Healthcare Life Sci- REFERENCES ence) were used according to the manufacturer’s instructions to precipitate cap-complexes. 1. Beroukhim R , Mermel CH , Porter D , Wei G , Raychaudhuri S , Dono - van J , et al. The landscape of somatic copy-number alteration across Gene Expression Analysis human cancers. Nature 2010 ; 463 : 899 – 905 . 2. TCGA . Comprehensive genomic characterization defi nes human For the gene expression analysis between cell lines with or without glioblastoma genes and core pathways. Nature 2008 ; 455 : 1061 – 8 . 3q26 amplifi cation, gene expression and copy number data were 3. Woenckhaus J , Steger K , Werner E , Fenic I , Gamerdinger U , Dreye r T , downloaded from CCLE for 967 cancer cell lines (54 ). Cell lines with et al. Genomic gain of PIK3CA and increased expression of p110alpha a log2 amplifi cation value higher than 0.3 at SKIL were defi ned as are associated with progression of dysplasia into invasive squamous 3q26 amplifi ed and the rest as normal 3q. A comparative marker cell carcinoma. J Pathol 2002 ; 198 : 335 – 42 . selection based on a two-sided t test was conducted between the two 4. Bass AJ , Watanabe H , Mermel CH , Yu S , Perner S , Verhaak RG , et al. classes based on their gene expression in GenePattern ( 55 ). SOX2 is an amplifi ed lineage-survival oncogene in lung and esophageal squamous cell carcinomas. Nat Genet 2009 ; 41 : 1238 – 42 . Tumor Formation Assays 5. Massion PP , Tafl an PM , Jamshedur Rahman SM , Yildiz P , Shyr Y , Edg- erton ME , et al. Signifi cance of p63 amplifi cation and overexpression in Tumor xenograft experiments were carried out as described pre- lung cancer development and prognosis . Cancer Res 2003 ; 63 : 7113 – 21 . viously ( 9 ). Tumor formation was assessed 8 weeks after injection. 6. Broad Institute. Tumorscape: copy number alterations across mul- 3 Tumors were scored when they reached 5 mm . tiple cancer types. [cited 2012 Dec 22] . Available from : http://www. broadinstitute.org/tumorscape. Disclosure of Potential Confl icts of Interest 7. Memorial Sloan-Kettering Cancer Center. cBioPortal for cancer L.A. Garraway has received a commercial research grant from genomics . [cited 2012 March 22] . Available from : http://www.cbioportal Novartis, has ownership interest (including patents) in Foundation .org/public-portal/. Medicine, and is a consultant/advisory board member of Founda- 8. Luo B , Cheung HW , Subramanian A , Sharifnia T , Okamoto M , Yang tion Medicine, Novartis, Millennium/Takeda, and Boehringer Ingel- X , et al. Highly parallel identifi cation of essential genes in cancer cells . heim. R. Beroukhim has received a commercial research grant from Proc Natl Acad Sci U S A 2008 ; 105 : 20380 – 5 . Novartis, ownership interest (including patents) in AstraZeneca, and 9. Boehm JS , Zhao JJ , Yao J , Kim SY , Firestein R , Dunn IF , et al. Integra- tive genomic approaches identify IKBKE as a breast cancer oncogene . is a consultant/advisory board member of Novartis. W.C. Hahn is a Cell 2007 ; 129 : 1065 – 79 . consultant/advisory board member of Novartis. No potential con- 10. Barretina J , Caponigro G , Stransky N , Venkatesan K , Margolin AA , fl icts of interest were disclosed by the other authors. Kim S , et al. The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 2012 ; 483 : 603 – 7 . Authors’ Contributions 11. Rothblatt JA , Deshaies RJ , Sanders SL , Daum G , Schekman R . Conception and design: D. Hagerstrand, R. Beroukhim, W.C. Hahn Multiple genes are required for proper insertion of secretory

1056 | CANCER DISCOVERYSEPTEMBER 2013 www.aacrjournals.org

TLOC1 and SKIL as Cancer Drivers in 3q26 RESEARCH ARTICLE

proteins into the endoplasmic reticulum in yeast. J Cell Biol 1989 ; patients with esophageal squamous cell carcinoma. Ann Surg Oncol 109 : 2641 – 52 . 2008 ; 15 : 2965 – 75 . 12. Wilkinson BM , Regnacq M , Stirling CJ . Protein translocation 33. Xue W , Kitzing T , Roessler S , Zuber J , Krasnitz A , Schultz N , et al. A across the membrane of the endoplasmic reticulum . J Membr Biol cluster of cooperating tumor-suppressor gene candidates in chromo- 1997 ; 155 : 189 – 97 . somal deletions. Proc Natl Acad Sci U S A 2012 ; 109 : 8212 – 7 . 13. Muller L , de Escauriaza MD , Lajoie P , Theis M , Jung M , Muller A , 34. Soto-Rifo R , Rubilar PS , Limousin T , de Breyne S , Decimo D , O hl- et al. Evolutionary gain of function for the ER membrane protein mann T . DEAD-box protein DDX3 associates with eIF4F to promote Sec62 from yeast to humans. Mol Biol Cell 2010 ; 21 : 691 – 703 . translation of selected mRNAs. EMBO J 2012 ; 31 : 3745 – 56 . 14. Panzner S , Dreier L , Hartmann E , Kostka S , Rapoport TA . Post- 35. Silvera D , Formenti SC , Schneider RJ . Translational control in cancer. translational protein transport in yeast reconstituted with a purifi ed Nat Rev Cancer 2010 ; 10 : 254 – 66 . complex of Sec proteins and Kar2p. Cell 1995 ; 81 : 561 – 70 . 36. Barna M , Pusic A , Zollo O , Costa M , Kondrashov N , Rego E , et al. 15. Yao L , Yan X , Dong H , Nelson DR , Liu C , Li X . Express ion of an IRF-3 Suppression of Myc oncogenic activity by ribosomal protein haploin- fusion protein and mouse estrogen receptor, inhibits hepatitis C viral suffi ciency. Nature 2008 ; 456 : 971 – 5 . replication in RIG-I-defi cient Huh 7.5 cells. Virol J 2011 ; 8 : 445 . 37. Stransky N , Egloff AM , Tward AD , Kostic AD , Cibulskis K , Siva chenko 16. Asano K , Kinzy TG , Merrick WC , Hershey JW . Conservation and A , et al. The mutational landscape of head and neck squamous cell diversity of eukaryotic translation initiation factor eIF3 . J Biol Chem carcinoma. Science 2011 ; 333 : 1157 – 60 . 1997 ; 272 : 1101 – 9 . 38. Hsieh AC , Liu Y , Edlind MP , Ingolia NT , Janes MR , Sher A , et al. The 17. Shih JW , Tsai TY , Chao CH , Wu Lee YH . Candidate tumor suppressor translational landscape of mTOR signalling steers cancer initiation DDX3 RNA helicase specifi cally represses cap-dependent translation and metastasis. Nature 2012 ; 485 : 55 – 61 . by acting as an eIF4E inhibitory protein. Oncogene 2008 ; 27 : 700 – 14 . 39. Thoreen CC , Chantranupong L , Keys HR , Wang T , Gray NS , Sabatini 18. Hosoda N , Lejeune F , Maquat LE . Evidence that poly(A) binding DM . A unifying model for mTORC1-mediated regulation of mRNA protein C1 binds nuclear pre-mRNA poly(A) tails . Mol Cell Biol translation. Nature 2012 ; 485 : 109 – 13 . 2006 ; 26 : 3085 – 97 . 40. Fukuchi-Shimogori T , Ishii I , Kashiwagi K , Mashiba H , Ekimoto H , 19. Venkatesan A , Sharma R , Dasgupta A . Cell cycle regulation of hepa- Igarashi K . Malignant transformation by overproduction of transla- titis C and encephalomyocarditis virus internal ribosome entry site- tion initiation factor eIF4G . Cancer Res 1997 ; 57 : 5041 – 4 . mediated translation in human embryonic kidney 293 cells . Virus Res 41. Chen L , Aktas BH , Wang Y , He X , Sahoo R , Zhang N , et al. Tumor 2003 ; 94 : 85 – 95 . suppression by small molecule inhibitors of translation initiation. 20. Gingras AC , Gygi SP , Raught B , Polakiewicz RD , Abraham RT , Oncotarget 2012 ; 3 : 869 – 81 . Hoekstra MF , et al. Regulation of 4E-BP1 phosphorylation: a novel 42. Noel P , Cartwright IL . A Sec62p-related component of the secretory two-step mechanism. Genes Dev 1999 ; 13 : 1422 – 37 . protein translocon from Drosophila displays developmentally com- 21. Stroschein SL , Wang W , Zhou S , Zhou Q , Luo K . Negative feedback plex behavior. EMBO J 1994 ; 13 : 5253 – 61 . regulation of TGF-beta signaling by the SnoN oncoprotein. Science 43. Zender L , Spector MS , Xue W , Flemming P , Cordon-Cardo C , Silke J , 1999 ; 286 : 771 – 4 . et al. Identifi cation and validation of oncogenes in liver cancer using 22. Iacobuzio-Donahue CA , Wilentz RE , Argani P , Yeo CJ , Cameron JL , an integrative oncogenomic approach. Cell 2006 ; 125 : 1253 – 67 . Kern SE , et al. Dpc4 protein in mucinous cystic neoplasms of the 44. Lu Z , Zhou L , Killela P , Rasheed AB , Di C , Poe WE , et al. Glioblastoma pancreas: frequent loss of expression in invasive carcinomas suggests proto-oncogene SEC61gamma is required for tumor cell survival and a role in genetic progression. Am J Surg Pathol 2000 ; 24 : 1544 – 8 . response to endoplasmic reticulum stress. Cancer Res 2009 ; 69 : 9105 – 11 . 23. Gould J , Getz G , Monti S , Reich M , Mesirov JP . Comparative gene 45. Radi M , Falchi F , Garbelli A , Samuele A , Bernardo V , Paolucci S , marker selection suite. Bioinformatics 2006 ; 22 : 1924 – 5 . et al. Discovery of the fi rst small molecule inhibitor of human DDX3 24. Peinado H , Olmeda D , Cano A . Snail, Zeb and bHLH factors in specifi cally designed to target the RNA binding site: towards the next tumour progression: an alliance against the epithelial phenotype? generation HIV-1 inhibitors. Bioorg Med Chem Lett 2012 ; 22 : 2094 – 8 . Nat Rev Cancer 2007 ; 7 : 415 – 28 . 46. Robinson JT , Thorvaldsdottir H , Winckler W , Guttman M , Lander ES , 25. Albino D , Longoni N , Curti L , Mello-Grand M , Pinton S , Civenni G , Getz G , et al. Integrative genomics viewer. Nat Biotechnol 2011 ; 29 : 24 – 6 . et al. ESE3/EHF controls epithelial cell differentiation and its loss 47. Elenbaas B , Spirio L , Koerner F , Fleming MD , Zimonjic DB , Donaher leads to prostate tumors with mesenchymal and stem-like features. JL , et al. Human breast cancer cells generated by oncogenic transforma- Cancer Res 2012 ; 72 : 2889 – 900 . tion of primary mammary epithelial cells . Genes Dev 2001 ; 15 : 50 – 65 . 26. Zhang H , Meng F , Liu G , Zhang B , Zhu J , Wu F , et al. Forkhead tran- 48. Morgenstern JP , Land H . Advanced mammalian gene transfer: high scription factor foxq1 promotes epithelial-mesenchymal transition titre retroviral vectors with multiple drug selection markers and a and breast cancer metastasis. Cancer Res 2011 ; 71 : 1292 – 301 . complementary helper-free packaging cell line . Nucleic Acids Res 27. Kojima T , Takasawa A , Kyuno D , Ito T , Yamaguchi H , Hirata K, et al. 1990 ; 18 : 3587 – 96 . Downregulation of tight junction-associated MARVEL protein mar- 49. Moffat J , Grueneberg DA , Yang X , Kim SY , Kloepfer AM , Hinkle G , velD3 during epithelial-mesenchymal transition in human pancreatic et al. A lentiviral RNAi library for human and mouse genes applied to cancer cells. Exp Cell Res 2011 ; 317 : 2288 – 98 . an arrayed viral high-content screen. Cell 2006 ; 124 : 1283 – 98 . 28. Cheng H , Fukushima T , Takahashi N , Tanaka H , Kataoka H . Hepa- 50. CCSB . The CCSB human ORFeome collection . [cited 2012 Dec 22] . tocyte growth factor activator inhibitor type 1 regulates epithelial to Available from: http://horfdb.dfci.harvard.edu . mesenchymal transition through membrane-bound serine protein- 51. Subramanian A , Tamayo P , Mootha VK , Mukherjee S , Ebert BL , ases. Cancer Res 2009 ; 69 : 1828 – 35 . Gillette MA , et al. Gene set enrichment analysis: a knowledge-based 29. Yang J , Mani SA , Donaher JL , Ramaswamy S , Itzykson RA , Come C , approach for interpreting genome-wide expression profi les. Proc Natl et al. Twist, a master regulator of morphogenesis, plays an essential Acad Sci U S A 2005 ; 102 : 15545 – 50 . role in tumor metastasis. Cell 2004 ; 117 : 927 – 39 . 52. Peng J , Gygi SP . Proteomics: the move to mixtures. J Mass Spectrom 30. Taube JH , Herschkowitz JI , Komurov K , Zhou AY , Gupta S , Yang J , 2001 ; 36 : 1083 – 91 . et al. Core epithelial-to-mesenchymal transition interactome gene- 53. Yates JR , Eng JK , McCormack AL , Schieltz D . Method to correlate expression signature is associated with claudin-low and metaplastic tandem mass spectra of modifi ed peptides to amino acid sequences breast cancer subtypes. Proc Natl Acad Sci U S A 2010 ; 107 : 15449 – 54 . in the protein database. Anal Chem 1995 ; 67 : 1426 – 36 . 31. Nieto MA . The snail superfamily of zinc-fi nger transcription factors . 54. Broad–Novartis Cancer Cell Line Encyclopedia (CCLE) . [cited 2012 Nat Rev Mol Cell Biol 2002 ; 3 : 155 – 66 . Dec 22]. Available from: www.broadinstitute.org/ccle/home . 32. Akagi I , Miyashita M , Makino H , Nomura T , Hagiwara N , Taka- 55. GenePattern . [cited 2012 Dec 22] . Available from : http://genepattern. hashi K , et al. SnoN overexpression is predictive of poor survival in broadinstitute.org.

SEPTEMBER 2013CANCER DISCOVERY | 1057

Systematic Interrogation of 3q26 Identifies TLOC1 and SKIL as Cancer Drivers

Daniel Hagerstrand, Alexander Tong, Steven E. Schumacher, et al.

Cancer Discovery 2013;3:1044-1057. Published OnlineFirst June 13, 2013.

Updated version Access the most recent version of this article at: doi:10.1158/2159-8290.CD-12-0592

Supplementary Access the most recent supplemental material at: Material http://cancerdiscovery.aacrjournals.org/content/suppl/2013/06/13/2159-8290.CD-12-0592.DC1 http://cancerdiscovery.aacrjournals.org/content/suppl/2021/03/01/2159-8290.CD-12-0592.DC2

Cited articles This article cites 50 articles, 18 of which you can access for free at: http://cancerdiscovery.aacrjournals.org/content/3/9/1044.full#ref-list-1

Citing articles This article has been cited by 11 HighWire-hosted articles. Access the articles at: http://cancerdiscovery.aacrjournals.org/content/3/9/1044.full#related-urls

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 Department at Subscriptions [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cancerdiscovery.aacrjournals.org/content/3/9/1044. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.