Novel Transcriptional Targets of the SRY-HMG Box Transcription Factor SOX4 Link Its Expression to the Development of Small Cell Lung Cancer

Published OnlineFirst November 14, 2011; DOI: 10.1158/0008-5472.CAN-11-3506

Cancer Molecular and Cellular Pathobiology Research

Sandra D. Castillo1, Ander Matheu2, Niccolo Mariani1, Julian Carretero3, Fernando Lopez-Rios4, Robin Lovell-Badge2, and Montse Sanchez-Cespedes1

Abstract The HMG box transcription factor SOX4 involved in neuronal development is amplified and overexpressed in a subset of lung cancers, suggesting that it may be a driver oncogene. In this study, we sought to develop this hypothesis including by defining targets of SOX4 that may mediate its involvement in lung cancer. Ablating SOX4 expression in SOX4-amplified lung cancer cells revealed a gene expression signature that included genes involved in neuronal development such as PCDHB, MYB, RBP1, and TEAD2. Direct recruitment of SOX4 to gene promoters was associated with their upregulation upon ectopic overexpression of SOX4. We confirmed upregulation of the SOX4 expression signature in a panel of primary lung tumors, validating their specific response by a comparison using embryonic fibroblasts from Sox4-deficient mice. Interestingly, we found that small cell lung cancer (SCLC), a subtype of lung cancer with neuroendocrine characteristics, was generally characterized by high levels of SOX2, SOX4, and SOX11 along with the SOX4-specific gene expression signature identified. Taken together, our findings identify a functional role for SOX genes in SCLC, particularly for SOX4 and several novel targets defined in this study. Cancer Res; 72(1); 1–11. 2011 AACR.

Introduction on chromosome 6p22, containing the SOX4 gene, which was expressed at very high levels and was the best candidate for an As with other types of cancer, lung cancer is undergoing a oncogene in the amplicon (2). therapeutic revolution characterized by the identification of The SOX4 gene belongs to the SOX family, which is divided novel driver oncogenes and the generation of drugs that inhibit into 8 groups, A to H, according to protein identity (3). Sox4, their activity in a very specific manner. The success of erloti- Sox11, and Sox12 form the Sox-C group, sharing a high degree of nib/gefitinib in epidermal growth factor receptor (EGFR)- identity in the high-mobility group domain, and in a group- mutant tumors and crizotinib in tumors carrying a transloca- specific transactivation domain (4). Sox4 is predominantly tion of the ALK oncogene are some of the current paradigms expressed during embryonic development in the heart, central (1). Because few tumors carry alterations of these genes, effort nervous system, lung and thymus (5–7). SOX4 protein levels are is required to identify additional targetable oncogenes. We increased in several types of carcinoma (8–11), and knocking previously carried out a wide-ranging DNA copy number down SOX4 induces apoptosis and growth suppression in analysis of lung cancer cell lines and identified an amplicon cancer cells (12–14). Providing further evidence of the oncogenic potential of SOX4, we reported that its overexpression in NIH3T3 cells increases the number of foci induced by the 1 Authors' Affiliations: Genes and Cancer Group, Cancer Epigenetics and mutant RHOA-Q63L (2). In addition, several independent Biology Program- (PEBC), Bellvitge Biomedical Research Institute-IDI- BELL, Barcelona, Spain; 2Division of Stem Cell Biology and Developmental studies have shown that Sox4 (also known as ecotropic viral Genetics, MRC National Institute for Medical Research, London, United integration site 16, Evi16) is a frequent target of retroviral Kingdom; 3Department of Physiology, Faculty of Medicine and Odontol- ogy, University of Valencia; and 4Hospital Universitario Madrid-Sanchi- insertional mutagenesis, leading to neoplastic transformation narro, Laboratorio Dianas Terapeuticas, Madrid, Spain in murine hematopoietic cells (15–18). In the particular case of Note: Supplementary data for this article are available at Cancer Research lung cancer, we previously reported the presence of a high level Online (http://cancerres.aacrjournals.org/). of SOX4 amplification in a subset of lung primary tumors and cancer cell lines (2). The relevance of SOX4 to lung cancer Array data deposited in the Gene Expression Omnibus (GEO) under accession number GSE31612. development has also been observed by others. A meta-analysis examining the transcriptional profiles of human tumors found Corresponding Author: Montse Sanchez-Cespedes, Cancer Epigenetics and Biology Program (PEBC), Bellvitge Biomedical Research Institute- SOX4 to be one of 64 genes uniquely upregulated in cancer IDIBELL 08908, Hospitalet de Llobregat, Barcelona 08907, Spain. Phone: thereby making it part of a general gene expression signature of 34-932-607132; Fax: 34-932-607219; E-mail: [email protected] cancer (19). Lung cancer was among the tumors with the doi: 10.1158/0008-5472.CAN-11-3506 greatest levels of SOX4 expression. In addition, Sox4 was among 2011 American Association for Cancer Research. the set of genes overexpressed in the lungs of c-myc transgenic

www.aacrjournals.org OF1

Castillo et al.

mice, and c-myc was also found to be one of the transcription Gene expression microarrays and real-time quantitative factors with overrepresented Sox4-binding sites among the set PCRs of overexpressed genes (20). The mRNA was extracted using conventional methods, and 1 SOX4 is involved in neural development and the main- mg of it was amplified from each sample and used for gene tenance of some stem cell types (21, 22). Recent studies expression microarray analysis. Universal Human Reference have reported SOX4 to be a direct TGF-beta target that RNA (P/N 740000; Stratagene), was used as reference for activates SOX2 transcription while retaining the stem cell hybridization and analysis. For labeling we used the commer- properties of glioma-initiating cells (21). In addition, in cial Two-Color Microarray-Based Gene Expression Analysis Kit normal hair follicles, Sox4 is expressed in the developing (version 5.5). MMLV-RT retrotranscription of sample from a T7 hair germ (23). promoter primer was followed by a T7 RNA polymerase- In spite of this, and although SOX4 was one of the first catalyzed in vitro transcription reaction in the presence of members of the SOX family to be isolated and characterized, either Cy3-CTP or Cy5-CTP fluorophores. Cy3 labeling was including the demonstration that it has separable DNA-bind- used for the reference sample. Hybridization was carried out on ing and transactivation domains, our present knowledge of the the Whole Human Genome Microarray (4 44 K), scanned genes controlled by SOX4 activity is scarce. This paper reports with a G2505B DNA microarray scanner and quantified using on the role of SOX4 in human lung carcinogenesis, focusing Agilent Feature Extraction Software (version 9.5; Agilent). especially on identifying novel targets of SOX4 transcriptional Fluorescence intensity from each array element was subtracted activity. from the local background and data normalized as previously described (24). We further selected transcripts that fulfilled the Materials and Methods following criteria: (i) repressed at least 2 times in the H522Tr- shSOX4-1 at 48 and 96 hours after induction of shSOX4 relative Cancer cell lines and primary tumors to the level at 0 hour and (ii) no changes in gene expression Cancer cell lines were obtained from the American Type between the parental H522 and the H522Tr-shSOX4-1. The Culture Collection and grown under recommended condi- mRNA and genomic DNA were measured by real-time quan- tions.DNAandRNAfromadditionallungcancercelllines titative PCR. DNase-treated RNA was reverse-transcribed. The used for real-time quantitative reverse transcriptase (RT) cDNA and genomic DNA were amplified using an ABI Prism PCR were kindly provided by Luis M. Montuenga and Ruben 7900 Sequence detector (Applied Biosystems), and levels of Pio of the Centro de Investigacion Medica Aplicada (CIMA), genes were measured by SYBR green real-time PCR. Reactions University of Navarra, Spain, and Jun Yokota, National were carried out in triplicate. As controls we used the human Cancer Center Research Institute, Tokyo, Japan. Fresh frozen GAPDH, B-ACTIN, and TATA box–binding protein (TBP)to lung primary tumors were provided by the CNIO Tumour correct for inter-individual/tumor variations. The primer Bank Network, CNIO, Spain, and were selected as previously sequences used are included in Supplementary Table S2. described (24). Antibodies, Western blot analysis, and immunostaining Expression plasmids and reporters For Western blot analysis, cells were scraped from the dishes We stably transfected the H522 cell line with a Tet repressor into the lysis buffer. A total of 25 mg of total protein was (TetR) expression construct, pCMB1b-TrS, kindly provided by separated by SDS-PAGE and blotted with rabbit anti-SOX4 M.V. de Wetering (Hubrecht Institute, Utrecht, The Nether- (A574) 1:5,000 (CS-129-100, Diagenode), mouse anti-SOX2 lands) using hygromycin selection. We examined TetR activity 1:2,500 (R&D Systems); rabbit anti-NSE 1:1,000 (Abcam), or by transfecting the cells with the pcDNA4/TO/Luc construct mouse anti-GAPDH. The secondary goat anti-mouse-IgG:HRP that expressed the firefly luciferase gene under the control of a (horseradish peroxidase) and goat anti-rabbit:HRP antibodies Tet operator and then measured the luciferase activity with the (DAKO) were added at 1:5,000. We carried out immunohisto- Dual Luciferase Assay Kit (Promega). The pRL-Tk plasmid chemical analysis of SOX9 (1:750 dilution; Chemicon) in the (Promega) that constitutively codified for the Renilla luciferase Autostainer Low FLEX (DAKO), using previously described was used to measure the transfection efficiency. Clones that protocols (25). Sections were counterstained with hematoxylin stably expressed the TetR were transfected with pSUPERIOR- and evaluated by 2 independent researchers. SOX4/S1 construct, kindly provided by C.A. Moskaluk (Uni- versity of Virginia, Charlottesville, VA), which expressed a Chromatin immunoprecipitation assays short hairpin against the sequence 50-AAGACGACCTGCTC- Cells were fixed in 1% formaldehyde for 10 minutes at 37C. GACCTGA-30 of the SOX4 coding region under the control of a Cross-linking was quenched by adding 125 mmol/L glycine. Tet operator and selected using geneticin. These oligonucleo- Cells were then washed with cold PBS, harvested and resus- tides specifically inhibited the expression of SOX4 but not of pended in SDS lysis buffer containing a protease inhibitor other SOX family genes (13). Selected clones were analyzed for cocktail. Chromatin was sheared by sonication (average length SOX4 expression by quantitative RT-PCR and Western blot 0.25–1 Kb) and incubated with 60 ml protein A/G agarose/ analysis before and after doxycycline induction. For transient salmon sperm DNA (50% slurry; Millipore) with gentle agita- transfection, full-length human wt SOX4 and the mutant tion for 30 minutes. The supernatant was then immunopreci- SOX4 S395X carrying an HA tag were cloned as previously pitated with anti-SOX4 antibody 1:500 or its matched nonim- described (2). mune crude serum 1:500 (IgG; Diagenode) at 4C overnight.

OF2 Cancer Res; 72(1) January 1, 2012 Cancer Research

Novel SOX4 Targets and Small Cell Lung Cancer Development

Protein A/G agarose (60 mL of a 50% slurry) was then added and expression between lung tumors with and without high levels incubated for 1 hour. Pellets were washed and protein-DNA of SOX4 were determined by the Mann–Whitney U test. cross-links were reversed by overnight incubation at 65C with Correlations between SOX genes were estimated by Spearman proteinase K. DNA was purified following a conventional rank correlation. Values of P < 0.05 were considered statisti- phenol–chloroform protocol and eluted in 50 mL water. At cally significant. For the Gene set enrichment analysis (GSEA), least 2 independent Chromatin immunoprecipitation (ChIP) raw data from the Gene Expression Omnibus (GEO) database experiments were carried out. Real-time quantitative PCR was repository (27) were normalized using the Robust Multiarray carried out using SYBR Green Master Mix on an ABI Prism 7900 Average (RMA) algorithm available in Bioconductor's Affy (Applied Biosystems). The primer sequences used are included package. We used the LIMMA package to obtain LIMMA- in Supplementary Table S2. moderated t statistics to build a ranked list of expression. GSEA was applied to this ranked list. After testing for normality Mouse embryonic fibroblast processing and RNA of the data with the Kolmogorov–Smirnoff test, our gene sets extraction from mouse tissues were considered significantly enriched between classes under Mice were housed in the pathogen-free barrier area of the comparison for values of FDR less than 0.25, a well-established National Institute for Medical Research, London, UK. Tissues cutoff for the identification of biologically relevant gene sets from young (1–2 months) and old (1.5–2 years) C57BL6 mice (28). were homogenized at high speed and total RNA extracted with TRIzol. Mouse embryo fibroblasts (MEF) were isolated and Results þ þ þ cultured as previously described (26) from Sox4 / , Sox4 / , and Sox4 / E13.5 mouse embryos. Mice were genotyped with Differential gene expression upon SOX4 depletion in primers included in Supplementary Table S2. lung cancer cells The NCI-H522 lung cancer cell line (henceforth referred to as Statistical and bioinformatic analysis H522) is known to feature SOX4 gene amplification (>20 Statistical analyses were carried out with SPSS software. copies/nuclei) and SOX4 overexpression (2). To identify tran- Differences between groups were analyzed by the unpaired scriptional targets regulated by SOX4 we generated H522- t test or Fishers exact test, as appropriate. Differences in gene derived cells that downregulate SOX4 in an inducible manner

Figure 1. Changes in gene expression after depletion of SOX4 in the H522Tr-shSOX4-1 cells. A, Western blot analysis of SOX4 of the H522 parental, H522Tr- shSOX4-1–derived cells with (þ) or without () doxycycline (dox; 2.5 ng/mL) after 48 hours. B, representative examples of real-time quantitative RT-PCR of the indicated transcripts relative to GAPDH. The values were also relative to the uninduced control for the H522Tr-shSOX4-1 (doxþ) and to the parental H1299 (pCEFL-mock) for the H1299 pCEFL-SOX4-wt and H1299 pCEFL-SOX4-S395X. Bars, mean SD from triplicates, from at least 2 independent experiments. C, Western blot analysis of SOX4 and SOX2 of the H1299, transfected with pCEFL-mock, pCEFL-SOX4, and pCEFL-SOX4-S395X. GAPDH is also shown for protein loading comparison. D, Western blot analysis of SOX4 and SOX2 of the H522Tr-shSOX4-1. GAPDH is also shown for protein loading comparison. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

www.aacrjournals.org Cancer Res; 72(1) January 1, 2012 OF3

Castillo et al.

with a shSOX4 that was reported to specifically deplete SOX4 (Genomatix; ref. 34) and designed primers flanking the regions expression (13). Several clones were found to downregulate enriched in these sequences (Supplementary Fig. S1). To SOX4 in an inducible-dependent manner. A stable clone, determine the specificity of SOX4 occupancy in the selected H522Tr-shSOX4-1, which downregulates SOX4 expression by regions, we also screened distant regions (>5,000 bp from 90%, 48 hours after the addition of doxycycline, was chosen for the SOX4-binding sequences) that were considered negative further analysis (Fig. 1A). controls. We immunoprecipitated the H522Tr-shSOX4-1 chro- To determine the gene expression profile characteristic of matin with the rabbit polyclonal anti-SOX4 antibody and its SOX4-depleted expression we compared the global gene matched crude serum as IgG control (Fig. 2A) followed by expression of the parental H522 cells with that of H522Tr- quantitative PCR. We confirmed the recruitment of SOX4 to shSOX4-1 cells 0, 48, and 96 hours after inducing shSOX4 with the known targets TEAD2 and TUBB3 (Fig. 2B; refs. 29, 32). We doxycycline. We found 123 genes with a more than 2-fold also determined that the MYB and VASH2 promoters were change of expression (85 downregulated and 38 upregulated) significantly enriched (>4-fold) in the anti-SOX4 immunopre- in the H522Tr-shSOX4-1 cells after shSOX4-inducible expres- cipitated chromatin from the cells prior to doxycycline induc- sion relative to the H522-parental and to the H522Tr-shSOX4-1 tion, compared with cells after depletion of SOX4 expression cells before adding doxycycline (at 0 hour; Supplementary and IgG immunoprecipitates (Fig. 2C). In all cases tested, the Table S1). We then measured the expression of these tran- enrichment was specific to the indicated SOX4-binding scripts by quantitative RT-PCR in the H522Tr-shSOX4-1 cells to regions, and was negative in the control regions. This indicates validate the microarray observations. Half of the 50 down- that SOX4 binds to the MYB and VASH2 promoters. The regulated genes chosen for validation were confirmed and protocadherin genes, PCDHA, PCDHB and PCDHG, are tan- selected for further analysis. The TEAD2 and Tubb3 genes demly arranged in clusters on the same chromosome. To were previously identified as Sox4 transcriptional targets in determine the possible direct involvement of SOX4 in the neural progenitors (29–32). All the transcripts, including the transactivation of the different PCDHB genes we first screened TEAD2 and TUBB3 controls, showed an approximately 50% for binding of SOX4 in regions upstream of individual genes but drop in expression after interfering SOX4 expression (Fig. 1B), found no SOX4 recruitment. While conducting the screening, validating the robustness of our microarray assay. It is of note the PCDHB cluster was found to be transcriptionally regulated that many of the transcripts found to be downregulated after by a control region downstream of the PCDHG cluster (35). inducing shSOX4 (e.g., EGR3, MLLT11, NBEA, PCDHBs, RBP1, Therefore, we searched and found SOX4-binding sequences TMEFF1, TUBB3, and VASH2) are highly specific to neural- within this control region. Next, we tested and confirmed SOX4 derived tissues, including brain, spinal cord, and retina recruitment to this regulatory region (Fig. 2C), although it (GSE7905 from GEO database). remains to be understood how the binding of SOX4 to this Next, we aimed to determine whether these observations region affects the transcriptional activation of the PCDHB could be extended to other lung cancer cell lines. To this end, cluster. Thus, we found evidence that the PCDHB clusters, as we transfected the H1299 cells, which have low levels of well as the MYB and VASH2 genes, are direct SOX4 transcrip- endogenous SOX4, with a construct carrying wild-type SOX4 tional targets. Other genes that are strongly upregulated by or a mutant form (S395X). The latter is devoid of transactiva- SOX4 were not found to recruit SOX4 to their promoters (EGR3, tion activity because of a mutation that eliminates the serine- GPC2, and RBP1), at least in the regions that we selected for the rich C-terminal domain of SOX4 (Fig. 1C; ref. 2). As shown analysis. in Fig. 1B, ectopic expression of the wild type, but not the mutant SOX4 form, leads to a significant, more than 2-fold Expression of SOX4 targets is increased in lung tumors upregulation of most targets. In particular, EGR3 and RBP-1 with SOX4 gene overexpression and is decreased in were strongly upregulated (>16 times) relative to the control Sox4 / MEF and to the SOX4-mutant. We extended our analysis to lung primary tumors selected It has previously been described that SOX4 regulates the on the basis of high (SOX4-H) and low (SOX4-L) levels of SOX4 expression of SOX2 in glioma-initiating cells (21). Although gene expression. We compared the differences in the expres- SOX2 was not downregulated in our microarray analysis after sion of the transcripts between the 2 groups. As depicted in Fig. depletion of SOX4, we specifically tested for SOX2 levels in a 3A, most of the transcripts, especially CCNG2, DLG3, VASH2, Western blot analysis. No change in protein expression levels of PCDHB9, and -11, were expressed at significantly higher levels SOX2 was observed after ectopic expression of SOX4 in the in the SOX4-H group than in the SOX4-L group. We also used H1299 cells or after abrogation of SOX4 in the H522Tr-shSOX4- GSEA to compare our SOX4-specific gene expression signature, 1 cells (Fig. 1C and D). composed of the genes most significantly downregulated in H522Tr-shSOX4-1 with the data set containing the gene Analysis of SOX4 recruitment to the gene promoters expression profile of the SOX4-H and SOX4-L lung primary We developed a quantitative ChIP assay to investigate which tumors (GEO Series accession number GSE8569; ref. 36). The of the genes downregulated after SOX4 depletion were direct GSEA showed that the set of downregulated genes after transcriptional targets of SOX4. The SOX4 HMG-box binds inducing the expression of the shSOX4 were upregulated in preferentially to the 4-bp DNA-motif 50-ACAA-30 sequence (33) the SOX4-H tumors (Fig. 3B). Collectively, these observations and we searched for putative SOX4-binding sequences in the are consistent with these being transcriptional targets of SOX4 promoters of these genes. We used the MatInspector program and having a role in lung carcinogenesis.

OF4 Cancer Res; 72(1) January 1, 2012 Cancer Research

Novel SOX4 Targets and Small Cell Lung Cancer Development

Figure 2. ChIP identiﬁes novel SOX4 transcriptional targets. A, left, Western blot analysis of SOX4, depicting its proper depletion after shSOX4 induction by doxycycline treatment and GAPDH as loading control. Right, Western blot analysis of precipitated SOX4, showing the presence and absence of SOX4 in the precipitates from the H522Tr-shSOX4-1 before and after induction of the shSOX4. B, qPCR to determine DNA enrichment of the indicated regions of known SOX4 targets relative to the input. C, qPCR to determine DNA enrichment of the indicated regions of the new candidate targets relative to the input. The locations of the SOX4-consensus sequences (black boxes) and that of the control regions (CNTL) relative to the transcription start site (TSS) are included below each graph. Error bars, SD from triplicates. GAPDH, glyceraldehyde-3-phosphate dehydrogenase; IP, immunoprecipitation.

þ Sox4 is key to embryogenesis and, while Sox4 / embryos in the lung tissue vary as mice age. We used real-time quan- are viable, Sox4 / embryos die by day 14 of embryonic titative RT-PCR to compare the gene expression levels of Sox4 development (37). Taking advantage of the availability of MEFs in different tissues (lung, spleen, liver, kidney, and heart) of þ þ from Sox4 / and Sox4 / embryos, we characterized the young (1–2 month) and old (1.5–2 years) C57BL6 mice, from 4 expression levels of the newly identified SOX4 targets in this and 5 different individuals, respectively. As a positive control material (Fig. 3C). Compared with mouse lung tissues, some of we included p16/INK4a. We found that Sox4 levels were þ þ the targets were expressed at very low levels in the Sox4 / significantly reduced (>70%) in the lungs of old mice compared MEFs (i.e., Dlg3, Dock4, Egr3, Mta2, Myb, Rbp1, and most Pcdhb- with young individuals (Fig. 4A). We also tested the gene transcripts) and were then excluded from the analysis. Our expression levels in many of the newly identified targets of results showed that the Gpc2, Rnf122, Tead2, Tubb3, and Vash2 Sox4. Again we discarded some of the transcripts because they transcripts exhibited significant downregulation in the Sox4 / were expressed at very low levels in normal lung (e.g., Pcdhb2, MEFs, relative to their wild-type counterparts (Fig. 3D). Pcdhb5, Pcdhb7, Tubb3, and Vash2). Among those depicting measurable levels of gene expression, 10 (Ccng2, Dlg3, Dock4, Ageing is associated with decreased expression of Sox4 Gcp2, Mta2, Nbea, Rbp1, Rnf122, Tead2, and Tmeff1) exhibited and its targets in lung tissue significantly higher levels in the lungs of young mice than in Ageing tissues experience changes in the expression levels of their older counterparts (Fig. 4B). some tumor suppressors and oncogenes that have been associated with a diminished potential for self-renewal of multi- Analysis and characterization of gene expression levels potent progenitors. In particular, an increase in the levels of among the SOX family of genes in human lung cancer p16/INK4a has been reported in several tissues including the We previously reported that SOX4 expression levels were neural system and in beta cells from the pancreas of ageing significantly higher in small cell lung cancer (SCLC) than in mice (38, 39). This background, coupled with the reported non–small cell lung cancer (NSCLC) histopathology, indicating connection of Sox4 with the maintenance of specific stem cell differences in their underlying biology (2). A functional, but properties (23), prompted us to test whether the levels of Sox4 nonoverlapping relationship has been reported between SOX4

www.aacrjournals.org Cancer Res; 72(1) January 1, 2012 OF5

Castillo et al.

A C TEAD2 ** VASH2 *** DLG3 *** 400 10 40

30 200 5 20 MEFs genotype

over GAPDH over +/+ +/+ +/+ +/– –/– +/– +/– 5 1 10 Sox4 wt- Fold change of expression 0 0 0 SOX4-H SOX4-L SOX4-H SOX4-L SOX4-H SOX4-L Knockout-

CCNG2 PCDHB9 PCDHB11 30 *** 15 *** 30 *** Sox4-

10 Gapdh- 15 15 over GAPDH over 5 Fold change of expression

0 0 0 SOX4-H SOX4-L SOX4-H SOX4-L SOX4-H SOX4-L B D

H-SOX4 L-SOX4

14,07 1 8 8 1 0,817 25 1 1,19965 1 1,21 667 7,56 025 3 25,15973 1,2 3418 3 2,308557 0,761113 2,05 2 9 1 9 1 ,173 534 3,04 8 4 9 Lung primary tumors (GSE8569). *** SOX4 *** CCNG2 15484 , 71 15 ,51768 22,4935 9,99 0531 5,250876 10,359 05 1,850017 9,595 17 2,123 52 2764 , 328 4 ,171 412 3,1 5646 Samples stratified according to the SOX4 status * DOCK4 13557 , 13 20,43 74 2 5,81 865 14,18 263 12 ,09 5 7 41 ,71205 5,40924 14 ,345 46 1,03 8 1 2 5 5 ,948 659 5,61 4005 2 ,327 749 (6 samples of each SOX4-H and SOX4-L) +/+ NS MLLT11 0,368527 1,270872 6,451 505 0,925 812 0,25 749 4 0,706519 0,3254 7 0,401385 0,439577 0,132 495 0,384 179 0,706377 NS MYB 0,008 9 3 5 0,016388 0,216 433 0,03 3 8 8 6 0,05 4 5 8 1 0,0193234 0 , 0150,154775 0,04 2 7 3 3 0,163 364 0,184 633 0,381 815 100% 0.05 6,289684 39564 , 33 20,46 482 16 ,7308 3,29 415 3 9,97 288 4 3,7 7198 3 13 ,06 3 7 6 1,465178 6,01 0 2 6 1 1 1 ,07 9 4 6 26,25 839 NS RBP1 Sox4 0.00 * TMEFF1 0,655981 0,870974 0,98379 0,948 351 0,21 689 7 1,005 4 5 9 0,139751 0,352248 0,118776 0,202678 0,217734 0,48 0758 NS EGR3 49648 , 37 92,65 018 6,806284 0,915 498 1,17 425 2 0,55 875 7 0,807287 4,381683 1,704992 1 ,236 204 0,68 688 9,600179 -0.05 ** ** * **** NS GPC2 0,145976 0,049875 0,147 218 0,05 8 7 4 2 0,12 921 8 0,809331 0,288478 0,01 9 2 2 9 0,01 9 8 3 1 0,274 512 0,05 2 9 2 4 0,12 562 -0.10 * MTA2 5,02 6 4 1 2 3885554 , 6,583 564 3,07 6 1 8 1 5,518109 3,63 359 7 2,06 07 3 2 4,486868 1,652 044 2 ,646 903 2 ,833 323 3,416 395 -0.15 ** TEAD2 241,9087 221,9032 7,129 407 106,7631 2,49 872 5 339 ,1119 1,3 7567 6 1840011 , 1,578488 1,835 544 1,308271 3,506243 -0.20 NS TUBB3 1,358391 0,3839 77 0,819 836 0,378 885 0,05 2 5 2 2 0,49471 0,503868 0,453604 0,533 34 1,547 409 0,04 8 3 5 3 0,485 527 -0.25 FDR = 0.006

VASH2 2,09 5 3 1 5 202 , 1 01 9 0,854 039 7,171 455 0,32 632 9 9,66 532 2 0,270005 0,533462 0,169 048 0,202015 0,32 143 0,36 773 -0.30 P < 0.0001

*** Expression over

NS RNF122 6,400061 13 4,009 12,35 993 3 ,147 886 2,0 1 006 1 2 4 ,91041 2,5 5565 3 8,610644 2,804519 2628 , 586 1 ,887 937 4,309189 Enrichment score (ES) -0.35 0% * PCDHB2 8,195939 2,03 6 7 8 3,449 587 3,498074 6,39 647 1 11,39368 0,283211 0,781308 0,146795 0,208 6 5 48 0 ,19436,73 0635 NS PCDHB5 6,705931 5 ,501594 3,050091 2,59 8101 18,89403 13,01 6 5 6 0,443713 1,841254 0,323 023 36,50425 0,162 162 2 ,731 784

** PCDHB7 11511 , 57 1,58 315 8,280389 2,38 0235 0,57 789 4 7,250648 0,452231 1,218245 0,494284 0,592 024 0,03 7 7 08 0,15 9051

*** PCDHB11 9,626 901 23,04 5 3 7 8,34 699 5,66 4048 6,420349 19,42848 0,912599 1,638353 0,390562 1,007 6 01 0,06 7 3 8 6 5 ,44093

*** PCDHB14 12541 , 36 3,733544 4,02 4 3 4 8 4 ,2 00075 3,63 451 3 10,44958 0,416588 1,534702 0,244 042 1 ,106936 0,246891 1,495 889

*** PCDHB9 9,554123 4,937582 4,102666 8,849041 1,938039 6,17 458 9 0,342288 3,225439 0,403878 1,59 6005 0,08 3 9 1 2 1 ,8 7168

Figure 3. Relative levels of gene expression of SOX4 targets in lung cancer and in Sox4 / MEFs. A, representative examples of real-time quantitative RT-PCRs to determine mRNA levels of the indicated genes relative to GAPDH in lung tumors with high (H-SOX4; n ¼ 6) or low (L-SOX4; n ¼ 6) levels of SOX4. The median level of gene expression for each group is also indicated. Bars, SD from triplicates. B, left, heat map depicting the expression levels of all the transcripts in the H-SOX4 and L-SOX4 tumors from data set GSE8569. , P < 0.05; , P < 0.01; , P < 0.005; NS, not signiﬁcant. Mann–Whitney U test. Right, graphical representation of the ranked gene lists derived from the comparison (using GSEA) of data set GSE8569 and the genes downregulated in the H522Tr-shSOX4-1 clone upon induction of the shSOX4. The corresponding P and false discovery rate (FDR) values are indicated. C, top, genotype of Sox4 in the MEFs from the Sox4þ/þ, Sox4þ/, and Sox4/ mice. Bottom, Western blot analysis of MEFs from the Sox4þ/þ, Sox4þ/, and Sox4/ mice. D, real-time quantitative RT-PCR of the indicated genes in the MEFs from Sox4/, relative to Gapdh. The values depicted are relative to the control MEFs (from Sox4þ/þ mice). Bars, SD from triplicates. , P < 0.01; , P < 0.05. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

and other members of the SOX family of transcription factors adenocarcinomas (AC) and SCCs. To overcome this hurdle (6, 29, 31, 32). Thus, we searched for a possible preferential we also tested in primary lung tumors: 40 SCCs and 30 ACs. We expression of different SOX-family members (i.e., SOX2, SOX4, found that levels of SOX2 and SOX9 were significantly higher in SOX9, SOX11, and SOX12) in the distinct and most common SCCs, whereas high levels of SOX4 predominated in the AC type lung cancer histopathologies. First, we tested a panel of lung (Supplementary Fig. S3A). It is important to note that SOX2 is cancer cell lines, using quantitative real-time RT-PCR, and located on chromosome 3q, a region that is frequently ampli- found that the expression of SOX2, SOX4, and SOX11 tran- fied in SCC and also includes the PIK3CA oncogene (24, 40). In scripts was significantly higher in SCLC than in NSCLC our samples, overexpression of SOX2 was concordant with the (Fig. 5A–B and Supplementary Fig. S2). This occurred in the presence of gene amplification, as we previously observed (24). absence of gene amplification (data not shown). The differ- In the case of SOX9, taking advantage of available antibodies, ences were particularly striking in the case of SOX11, in which 9 we carried out immunostaining and confirmed the high pro- of 18 SCLCs had very high levels of SOX11 (P < 0.005; Fig. 5A). tein levels of SOX9 and its association with the SCC histopa- For unknown reasons, commercially available lung cancer cell thology (Fig. 5C). Although more commonly overexpressed in lines are under-represented in the squamous cell carcinoma SCC, strong SOX9 immunostaining was also observed in some (SCC) type, which affects the comparative assessment of ACs. No differences in the levels of SOX12 were observed differences between the 2 main NSCLC histopathologies: among the different types of lung cancer.

OF6 Cancer Res; 72(1) January 1, 2012 Cancer Research

Novel SOX4 Targets and Small Cell Lung Cancer Development

Figure 4. Variation of Sox4 gene expression in aging lung from mice. A, Sox4 and p16/INK4a mRNA levels assessed by real-time quantitative RT-PCR normalized with the internal control Gapdh in lung tissues from old (n ¼ 5) and young (n ¼ 4) C57BL6 mice. The values were relative to the mean values of young mice. B, mRNA levels assessed by real-time quantitative RT-PCR relative to the internal control Gapdh and to the mean values of young mice, of the indicated genes in lungs from young and old C57BL6 mice. Bars, mean SD from the different groups of individuals. , P < 0.05; , P < 0.01; , P < 0.005. GAPDH, glyceraldehyde-3-phosphate dehydrogenase.

Finally, we determined the correlations between the levels of exclusively found in SCLCs. This is puzzling as it suggests the different SOX family genes. In lung cancer cell lines, there that this cell line has a neuroendocrine origin. were significant direct correlations between all pairs of SOX Next, we used the same data set to obtain the gene expres- transcripts, except SOX9 (Supplementary Fig. S3B). The most sion profile of the SCLC cell lines (Fig. 6B) and to conduct significant direct correlation involved SOX4 and SOX2 expres- GSEA. Our results show significant similarities between the sion levels, which occurred in both NSCLC and SCLC cell lines genes downregulated after inducing the expression of shSOX4 and suggests a functional correlation. The exception was in the in the H522Tr-shSOX4-1 and the SCLC gene expression signa- lung primaries of the SCC type, which probably reflects the fact ture (Fig. 6B). Taken together these observations suggest that that SOX2 overexpression in this type of lung cancer is mainly SOX4 is a key element required for the development of SCLC. because of gene amplification. Discussion The gene expression signature of SOX4 relates to that of SCLC We report the changes in global gene expression after Taking into account the following observations: (i) the high depletion of SOX4 in a lung cancer cell line that has SOX4 level of expression of several SOX-family genes in the SCLC overexpression because of gene amplification. Taking this type; (ii) the neural-derived origin of SCLC, and (iii) the neural- approach, we identified novel and high confidence direct and tissue specificity of many of the SOX4 targets (Supplementary indirect targets of SOX4 activity. We validated our findings in Table S1), we decided to examine whether there was a various models, attesting to the robustness of our approach. relationship between the gene expression signature of SOX4 ChIP of SOX4 indicated that some of these are direct targets, and that of SCLC. First, we compared the expression levels of activated by the position of SOX4 in their promoters, whereas the various SOX4 targets identifiedherewithadataset others are indirectly activated by SOX4. Apart from TEAD2 and containing the gene expression profile of a panel of 68 cells TUBB3, previously described as SOX4 direct targets in neural lines (48 and 20 of the NSCLC and SCLC types, respectively) progenitors (29–32), we identified novel genes that recruit extracted from the GEO database (GSE4824). As depicted SOX4 to their promoters, including the MYB and VASH2 genes. in Fig. 6A, most of the SOX4 targets exhibited higher Some human protocadherin (PCDH) genes are clustered expression levels in the SCLC than in the NSCLC type. Two together on chromosome 5 and classified into 3 subfamilies, of the lung AC cell lines, H2009 and H2887, had a profile for PCDHA, PCDHB, and PCDHG. Of special interest are the PCDHB the SOX4 targets with similarities to that of SCLC. One of genes, many of which are downregulated after SOX4 depletion. these cell lines, H2009, had simultaneous mutations at KRAS The PCDH proteins are located, at least in part, at synapses and and RB (http://www.sanger.ac.uk/). While the former muta- several mouse models lacking Pcdhs have a wide variety of tion is characteristic of lung ACs, the latter is almost neural malformations, suggesting that the Pcdhs regulate

www.aacrjournals.org Cancer Res; 72(1) January 1, 2012 OF7

Castillo et al.

A C

SOX11 P < 0.005; SCLC vs. NSCLC 150

100 GAPDH 50

Relative expression over expression Relative 0

SOX2 P < 0.05; SCLC vs. NSCLC

20 GAPDH Relative expression over expression Relative 0 SCLC SCC LCC AC AL

B NSCLC SCLC 50 P = 0.01; AC vs. SCC A427 A549 H1299 H157 H1650 H23 H522 DMS 273 H1672 H1963 H2107 HCC33 N417 25 (–) SOX4 (+)

NSE tumors % no. 0 SOX2 AC (n = 33) SCC (n = 51)

GAPDH

Figure 5. Characterization of gene expression in SOX family members among the different lung cancer histopathologies. A, mRNA levels of SOX11 and SOX2 genes assessed by real-time quantitative RT-PCR, relative to the internal control GAPDH, in lung cancer cell lines of different histopathologies and in normal lung. LCC, large cell carcinoma; NL, normal lung. Bars, means SD from triplicates. B, Western blot analysis of SOX4, SOX2, and from the neural marker NSE (neuron-speciﬁc enolase) of the indicated lung cancer cell lines. GAPDH is also shown for protein loading comparison. The histopathologic subtypes are also indicated. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. C, representative examples of SOX9-positive immunostaining (original magniﬁcation, 200), in 2 different lung primary tumors. Cells from normal tissue that do not express detectable SOX9 protein are included in the images and serve as negative controls. (P ¼ 0.01; Fisher exact test).

neural formation. Each clustered PCDH isoform has its own abnormalities (37). This indicates an involvement of Sox4 in promoter but it has recently been reported that genes of the embryonic development, which is consistent with the observed Pcdh-b family are controlled by a region located downstream of fluctuations of Sox4 levels during embryogenesis; these are the Pcdh-g cluster. This region activates the expression of the significantly reduced in later stages (42). In addition to embryo- Pcdh-b gene cluster in cis, and its deletion dramatically genesis, we have shown that levels of Sox4 and of most Sox4 decreases their expression levels (35). We found that this targets are reduced in the lungs of older mice compared with control region, containing many SOX4-binding sites, strongly those of younger ones, thereby implying a role for Sox4 in lung recruits SOX4, implying that it is an important regulator of the ageing. In parallel, the levels of the p16Ink4a tumor suppressor PCDHB gene family. Although the exact mechanisms that were significantly increased in aged lungs. This agrees with the underlie this regulation remain to be elucidated, the formation previously reported increase in the expression of p16Ink4a in of a looped architecture is a possible explanation that has been the neural system and in beta cells from the pancreas of ageing reported for other transcription factors. Apart from the mice (38, 39) and suggests that genes involved in cancer PCDHBs, other genes upregulated by SOX4, including direct development also have a role during ageing, possibly related and indirect targets (e.g., EGR3, DLG3, DOCK4, MLLT11, MYB to the diminished self-renewal potential of multipotent pro- and TMEFF1), are also predominantly expressed in neural- genitors. In this regard, a crucial role for Sox4 in cell fate derived tissues (i.e., brain, spinal cord, and retina). These decisions and maintenance of stem cell properties has been observations are in agreement with the reported connection described (21, 23). between SOX4 and neural development and developmental A variety of Sox family genes are expressed in the devel- potential (22, 23, 29). Others have searched for SOX4 targets in oping lung, including Sox2, Sox4, Sox9,andSox11,and human hepatocellular and prostate cancer cells (33, 41). The become repressed as embryonic development progresses overall overlap of targets is low, possibly reflecting that SOX4 (32, 42, 43), indicating that this family of genes actively controls specific subset of genes in different cellular contexts. control lung embryogenesis. On the other hand, lung cancer Although it was reported that SOX2 constitutes a direct target comprises several histopathologic lineages that arise from of SOX4 in glioma cells (21), we could not confirm this in lung various genetically distinct cell types (43). In this study, we cancer-derived cells. found strong differences in the pattern of expression of SOX- þ Sox4 / mice are viable but Sox4 homozygous null mutants related genes (SOX2, SOX4, SOX9,andSOX11)amongthe are embryonic lethal because of cardiac defects and other lung cancer types. In NSCLC, increased levels of SOX4

OF8 Cancer Res; 72(1) January 1, 2012 Cancer Research

Novel SOX4 Targets and Small Cell Lung Cancer Development

A H2009 H2887 SCC AC SCLC

Gene

SOX4 142,2 1,48 172 2, 1316,0 1832,4 13 32 5, 2 ,720 6 ,8 2 31 ,37 1 1 59 ,1 37 ,6 6 4,911 1 34 5 ,8 1 ,23 8 ,7591 2 ,281 3 2 76 ,2 1 40 ,9 6 12 76 ,1 786 1, 849,6 1 01 2 ,6 613,3 23 ,3 6 2 7 7 ,5 9 241 ,6 5 61 4,67 11 ,08 5 1 26 ,8 9 501 8,3 2 50 6 ,2 ,952 4 6 ,60 0 2 609 ,7 64 1,2 1 ,095 8 583 3, ,574 2 1361 8, 261 8 ,7 9 ,81 6 22 ,40 5 4,812 63 3 ,1 77 ,6 0 142 8, 1 92 ,7 1666,7 1290,0 748 1, 2813,7 1 5 1 9 ,8 1 46 ,7 6 1 22 ,6 1 5941 8, 4101 ,6 4246,1 10449,7 7403,2 4584,7 4153 ,4 7810,8 7 75 3 0, 3260,2 1 610 ,48152 ,8 3695,6 9952,2 1459,3 10794,5

CCNG2 12,1 2 42 63, ,34 74,7 108,7 65 7, 7 1,0 9,412 ,17 2 ,311 6 ,45 5 1 86 ,7 842 ,9 1 97 ,1 ,64 5 ,15 9 72 ,9 2 460,5 101,9 453,1 162,5 17 ,49,601 44 4 ,1 34,1 420,5 95,0 99 ,7 ,47 1 1 ,49 65,382 539,8 53,9 44 ,4 611,5 18,30,071 21 4 ,8 ,352 8 115,841 80 , 142,9 493,5 7,3 89 8, 41 ,0 6 52,93 ,832 272,8 722 ,7 220,68944 , 183,2 51 ,1 9 ,743 4 513 1, 901 ,3 ,012 3 432,7 142,1913 , 408,3 793,9 76,4 282,95142 , 405,9 247,8 2 8 6 34, ,463

DLG3 24,9 23 ,3 113,4232 , 133,0 ,911 3 9 ,9 9 ,8 2 ,501 4 5,65 1,132 8,02 ,28 1 ,32 1 7 ,6 2 ,63 3 91 ,3 3 ,1 4 6,03 1 ,85 7 1 ,62 51,501 285,0 13 2 ,191 0 6 , 279,4 188,4 22 ,9 4 121,0 1 3 9 ,1 ,371 4 ,04 4521 8 , 366,0 10,8 02 ,2 5 ,73 421 0 ,4 160,1 71 3 ,9 2 8,92 ,121 03,722 23,1 621 ,3 031 ,9 ,37 2 9,551 ,891 25,88 192,99,612 12,8 275,6 435,4 164,1 228,1 338,2 505,7 248,061 65 , 320,0 249,7 251,7 312 98, 91 4 , 280,5 486,8 281,32,822

DOCK4 92,8 7,93 3 6,8 2 9,57 641 4, 3 ,6 4 ,002 9 7 ,2 5 3,96 1 3 ,6 2 3,281 ,751 7 15 ,3 1 54 ,5 ,85 4 15 ,8 ,99 7 6 ,8 9 1 4,50 4 9,0 3 ,6 6 1 24 ,9 7 ,5 9 41 ,7 278,8 132,3723 , 686,4 33,9 4 ,1 4 1 7 0,3 66 9, 2 8 ,3 4 169,7 60,3 47 ,2 72 ,3 2 1 ,73 3 7 7,1 4 ,5 3 ,24 9 56 ,89,01 602,8 309,3 33,3 2 ,9 2 3 ,0 1 8 8 2, 314,5 64,0 ,58 041 04 , 72,8 86,8 291 ,9 309,1 79,9 9 ,2 0 299,8 692,6 6 5 3,8 246,4 88,0 1 ,80 3 41 8 ,9 ,28 9 ,411 8 311,6 80,9

MLLT11 2796,7 1 4,65 42 9,3 2241,5 284,2 026 ,5 41 ,21 8 ,802 1 11 6 8,3 3 ,22 7 181 ,3 1 2 ,123 2 511 ,5 6 3 6,2 22 9 0, 2 33 ,9 3 32 ,2 3 ,00 3 4 ,40 8 39 ,7 901 ,9 8 1 2 ,8 0 35 ,9 8 ,16 5 21 2 ,6 ,602 6 8 ,8 2 33 ,07 4 6 ,8 7 11 ,0 6 ,2732 2 ,852 8 5 ,7 9 8 5 5, 31 9,4 32 0 ,2 ,086 5 ,252 2 01 40 8, 11 2,4 7,041 ,816 8 9 ,93 6 5881 ,7 75 0 6, ,991 8 ,909 5 70,4 237,9 8 5 2 5, 674,2 313 ,2 8 7785,3 4378,6 4437,0 5651,1 2326,4 7569,07258,7 1185,2 5944,8 36 29 ,4 12945,5 3506,2 7824,7 2371,8 4 ,830 4 3 ,495 8 4099,7 2960,7

MYB 12,9 4 2,9 ,501 4 9 ,44 3 7,68 2 6,44 ,55 1 ,42 5 ,57 5 ,39 1 8,53 4 9 9, 2,58 1 ,64 78,34 26 2 ,1 135,1 98 ,6 36 ,5 ,021 2 9 6,3 7 7,7 46 1, 1 2 ,1 1 77 ,4 ,051 9 3 8 ,5 ,29 8 8 5,9 8,83 5,27 4 8 3, 201 2, 4 5 9, 85 6, 4,901 98 ,2 6,75 58 2,94 1 0 9 9,08 15,47 1 ,86 325,5 17,5 72,9 64,6446 , 230,772 ,31 180,2 296,7 373,9 182,5 396,7 53 5 ,6 252,5 470,4 127,4 442,5 204,8 300,7 341 219,5 2 6 1 ,6 355,1 33 5 ,2 005

NBEA 20,5 23 ,9 3,43 ,532 5 3,141 ,531 1 ,161 5 ,15 2 9,76 9,711 ,67 2 ,53 9 ,36 7 ,53 8 ,321 8 ,271 2 ,55 5 ,521 0 ,61 7 7,49 ,73 8 ,501 5 ,521 0 ,92 6 ,481 3 ,08 9 6,86 ,561 7 ,74 1 ,77 2 ,79 9 ,24 2 ,711 5 ,231 7 6,54 ,15 1 9,07 2,311 ,47 5 ,33 99,13 388,6 83,3 203,1 49 0,8 108,8 811 ,4 5,26 ,421 9 ,17 1 ,68 4 ,421 87,421 98,7 325,5 193,4 582,9 415,93324 , 198,9 163 ,07,703 162,4 720,6 194,3 21 4 ,0 408,6 694 86, ,125 90,5

PCDHB6 9,4 2 2 ,9 ,41 7 59,36,39 17,8 73,1 6,35 91,6 97 ,96,79 35,1 7 5 88 6 ,0 2 ,07 5 ,55 8252 , 123,2 71,3 ,74 2 14,4 1 ,4 3 65 6 7 ,9 ,27 95,01 43,1 75 ,6 4 ,1 3 ,02 1 ,12 9 3 ,5 6 ,63 1 99,7 6,9 550 4 1 , 91,7 12,3 03 4, 22,52 ,82 90,3 116,7 57,3 ,84 8 94,4 153 45 78,1 ,03 3 53,6 33,7 117,9 31,1 61,2 19,7 79 ,7 106,9 72,2 117,8 90,5 60,3 301 5, 4 4 ,64 8 168,5 77,3 116,3

PCDHB8 51 ,62 45,1 35,1 32,2 5 3 , ,32 6 ,9 3 4,34 7,54 ,5 5 ,81 9 ,72 8 ,12 9 ,41 2 ,0 8 ,5 38,31 60,3821 , 14,5 3 ,55,6 0,8 40,8 23 ,5 ,6 3 ,3 7 ,31 8 ,91 8 ,9 6 ,3 7 8,7 7,3 ,1 1 9,43 5,3 ,3 4 9,2 ,2 8 ,35 7 ,5 5 ,8 9 23 ,4 1 6,2 26,61 ,16 46,2 101,8 9,7862 , 8,6421 , 22,2 166,7 18,4 60,6785 , 38,1 24 82 22,9 10,689 , 23 ,1 71,5 16,7 45,9 5,2

PCDHB11 5,9 6 ,7 5 ,4 2 ,8 2 3,3 7 8, 8,9 9,62 2,7 ,4 7 ,7 9 ,41 7 ,71 4 5 ,4 ,6 5 6 ,7 ,6 8 15,875 , 7,7 3 ,3 ,4 9 ,2 9 ,4 2 ,22 2 ,8 2 7 4 ,6 2 7 ,7 9 3 7, 9,3 1 9,7 7 ,5 7 11 ,8 ,6 4 3,1 8,2 ,5 1 ,21 2 3,7 2 2,4 2,4 7 2,21 117,3 38,6 23 ,1 ,11 4 ,13 8 ,8 2 ,72 8 2 6,0 7 41,6 ,02 55,2 19,2 63,1 19,9 6,955 , 15 ,8 32 ,5 ,32 6 52,5 9,8

PCDHB12 11,6 86,1 17,4 21 6 , 5,12 ,51 3 2,12 ,53 2 4 ,6 9 3,43 2 ,2 94,42 81,2 ,17 1 64 6, ,6 5 ,42 9 3,21 89,47 ,32 53,6 51 ,5 ,41 4 ,3 6 ,52 1 91 ,2 ,23 7 2,8 ,24 4 ,82 1 ,12 8 22 ,61 4 ,41 9 74 ,2 8,81 ,81 7 53 ,11,11 8,7 11 43, ,81 17,4 25 3, 5 3 8, ,61 3 73 11, ,42 111,3 34,1 63 1, 7,21 93,12 ,14 74 ,97 58,95 8,2 33 73,2 18,6 157,9 122,8 8,578 , 18 ,3883 , 53,7 114,8 55,2

PCDHB13 3,5 ,5 0 ,2 8 5,6 4,4 2,3 7 8, ,4 6 5,75 ,8 0 ,3 2 ,7 5 ,5 4 ,81 1 8 ,9 ,3 7 ,6 1 ,4 5 ,5 3 5 2, 4,9 83,2 ,3 2,7 4,3 ,21 1 ,2 9 4 64, ,2 34,5 5,6 3 ,7 2 ,4 8,6 ,01 0 5 ,2 ,6 0 9 ,8 ,7 1 7,2 5 ,9 ,61 0 ,7 9 6,21 ,6 0 ,31 8 3,8 58,4 79,4 83,7 61,9 5,9 22,7 6,7 12,5 02 58, ,22 4,772 5 , 70,5 12,6 128,1 21,3 4,3422 , 12 ,7 2 ,0 6 5,7 80,0 4,8

PCDHB17 0,9 6,6 4 ,1 ,4 7 ,1 6 5 ,1 8 1 7,8 4,4 6 9 , 5 1,2 2 4 ,1 7 ,2 3 ,4 9 ,1 5 8,1 4,3 ,1 36,4 12,2 4,9 14,4 4,6 0 9, ,2 5 ,6 7 1,4 ,3 1 0 8, ,1 6 61,5 ,2 5,4 16,3 1,881 , 8,8 1,4 3 1, ,2 2 ,2 1 2 6, 5,1 2 2,1 26,4 7,85,3 4,7 26,7 3,4 4,9 2 ,1 3,8 11 ,8 1,3 5,4 22,5 5,4 1 ,4 6 ,6 2 7 ,4 142 ,3 3,9

PLEKHB1 5,6 ,45 7 1 ,0 9 ,42 45,76 334,1 78,6 ,62 5 ,93 9 1 02 0, ,23 7 ,74 9 ,58 0 11 ,0 ,05 4 ,09 1 ,52 4 ,66 9 7 4,3 ,83 4 591 ,0 ,86 9 02 ,8 1 00 3, ,96 7 11 4 ,7 011 ,7 5 ,2 ,902 0 101 ,5 32 9 ,5 159,1 91 ,6 ,84 4 3 ,0 2 ,101 9 84 ,5 5 ,5 1 6 ,0 1 1 9 ,9 0 ,28 8 1 29 5, ,601 2 6 ,7 1 ,88 6 ,13 1 9 8,0 73 4, 1 5 ,3 6 ,711 5 ,202 91,95 90,0 20,6 304,7 187,3 651,1 237,1 573 ,5642 6 , 84,8 437,3 52,6 368,8 119,3 ,252 042 33 , 502,8 331,8 29,4

RBP1 209,92,532 436,0 ,63 2 15 ,8 99,2 ,713 0 ,35 3 ,11 2 5 ,8 6 ,31 9 ,98 6 ,541 0 8 0,4 19 0, 7 ,3 1 51 ,8 5,821 3 8,2 ,33 9 47,6 73 ,3 37 ,9 85,4 8,6 1 47 ,9 ,52 4 ,63 9 ,77 4 9 ,5 3 9,9 65 ,1 74 ,3 05 1 ,4 3,912 ,32 2 9 ,1 9 5,742 3 ,9 4 3 2,2 80,901 ,48 381,9 761 ,0 08 8, 1,03 01 8,7 5 ,4 81,96 3535,6 314,5 3365,6 154,7 2834,8 258,5 7 42 ,6894 6 , 1651,2 621,4 688,2 942 ,1 2558,0 1138,1 16,5 1784,4 619,3 494 ,55,177 1961,6 2332,9

RNF122 62,2 320,5 195,6 02 5 ,7 3,721 12 4 7, 1 2,77 21 2 ,83,661 350,6 175,2 379,7 245,5 041 ,1 65,871 1 ,31 330,73022 , 89 ,77 5 761 ,131 8 ,051 9 ,541 19,521 637,4 135,8 ,211 7 ,07 13,801 306,6 224,8 201 ,8 ,071 9 ,461 751 ,10 238,5 57 0,5 168,2 031 ,2 315,8 502 ,2 ,052 3 164,1 224,7 1 00,5 99 9, ,491 96,461 714 374 ,7 31 3 ,9 7,893 2 8 ,7 7 ,642 94,782 100,7 224,8 436,9 125,96231 , 131,9 324,1 124,8 308,14832 , 480,7 333,7 138,8 197,3

SOX2 18,3 56 8, 9,721 04,76 ,02 45,4 14 ,7 ,9 3 ,41 3 9,12 ,32 3 ,56 9 ,714 5 ,67 2 ,24 1 ,74 4 ,57 4 ,271 1 ,76 8 ,75 1 ,066 4 ,25 17,17 29,8 561 ,9 ,833 82,761 31,8 31 ,7 6,35 5,912 ,6 3 ,04 7 ,431 9 9,25 1,19 2,51 5,61 0,87 2,04 ,522 4 ,58 83,66 22,6 681 5, 5 4 8,0 63,732 ,92 447,2 387,3 1688,9 8,4 33,1 4,2 1142,6 92,8843 , 489,0 44 ,5 516,2 274,1 576,3 74,0 10,2751 , 621,2 95,0 380,9 1226,8 395,7

SOX11 14,9 ,4 9 ,81 0 2,433 , 3,5 ,03 0 ,4 9 ,7 5 3,41 ,41 5 1,4 1,81 ,6 6 4 ,8 ,42 6 ,14 9 03 1, ,3 7 2 ,8 2 ,3 4 3,41 3 5, ,31 6 54 ,8 2 ,511 5 4,53 4 ,9 ,61 7 ,5 5 6,5 ,9 1 41 ,7 ,21 2 1 9,8 7 ,7 7,72 1 ,1 1 2,5 ,8 6 1 3,4 ,56 9 ,75 7 ,7 9 2,1201 3 ,3 8,9 0,8 ,7 06,92 7,1 1556,6 139,8 1830,831551 , 484,4 352 ,2 1736,4 3156,90,7 1671,3 1149,7 1885,0 16,5 987 ,67,31 1845,0 930,9 402,94,771

STMN1 519,3 443 97, ,763 132,96523 , 331,1 83 2,2 9,044 ,062 5 ,622 1 ,49 5 ,786 5 ,591 2 2 ,55 0 ,432 6 ,614 2 158 ,0 723 ,3 2 ,26 4 3,022 ,702 0 422 ,6 ,603 4 ,431 6 03 ,4 3 ,035 1 ,634 2 ,101 3 ,273 1 42 ,1 8 ,712 2 ,043 6 341 ,2 3,233 ,265 6 ,572 5 ,361 0 6,995 ,972 5 1 ,16 3 2 8,38 5,972 ,015 7 ,252 3 5 2 8,8 ,293 5 7,376 9,083 ,092 76,293 521,6 1456,7 509,3 980,6 713,5 542,3 324 6, 3022,9 1790,30,499 2095,417812 , 1348,9 799,9 1772,6 908,9 305 ,9 ,5421 158 ,30 1007,3

TMEFF1 108,7 9 5,2 8,91 100,93,122 134,0 32 ,2 0,87 1 ,98 7 ,81 9 ,25 7 2 ,1 3 ,92 9 1,14 94 0, ,86 5 2 5 ,9 ,93 3 ,701 3 23 ,4 ,21 1 62 ,1 63 ,7 ,33 9 ,73 0 ,68 2 ,32 2 341 1, ,15 9 ,42 6 ,1 7 ,72 2 21 ,9 0 ,93 1 0,07 3 ,0 0 1,04 1 ,1 5 ,3 0 2 ,3 7 3 0 ,0 9,51 9,16 ,011 5 2 ,6 7 2 ,8 4 ,513 6 ,351 7 ,05 37,9 80,78,741 128,6 81,4 747,7 309,7 719,3 413,4 903,8 301,0 774,4 1889,0 689,2 805,1 301,8378 , 397,6 926 ,359 9 ,3 216,2

TUBB3 1970,1 1 66 7 8, 1,5041 911 3 ,242 ,764 1364,8 03 88 ,6 ,9071 5 31 2 ,6 6 7,998 84 ,3 9 ,4632 0 01 95 ,8 ,3831 7 7671 ,7 1 6 ,39 1 ,035 749 ,56 2704,3 7151 ,3 1 ,969 1 7 2 ,4 001 ,315 2392,,2 1 821 3 3651 ,3 1 46 4,4 ,0301 3 2 14 ,0 4 2 ,373 3 1 3 30 ,4 ,029 1 ,949 6 71 ,94 8 9361 ,9 1 57 ,5 8 4,816 13 1 ,1 5 1 ,126 9 3601 ,4 1 1,254 3672 ,6 ,2713 2 ,749 6 2 932 2, ,4012 2 971 3 5, 3,5252 831 2,2 4 ,963 9 7122 4, 5034,1 2369,1 6478,9 2823,2 5142 4, 2 ,947 3 7937,6 6230,5 2135,0 8784,6 4755,5 7683,7 2557,1 4604,4 4 457 ,37,2754 6906,8 3386,9 53 2 2 ,2

VASH2 11,8 32 0, 12,4 13,25,02 20,0 45 ,2 ,21 8 ,62 9 ,04 7 ,31 0 ,42 3 ,44 5 ,2 3 2 ,0 88,1 26 8 ,5 27,5 24 ,7 2 8 ,8 7,61 5,61 ,9 42,83 20,0 ,06 77,1 8,8 4 ,1 1 ,6 3 7,16 ,41 6 ,3 4 5,01 2 ,1 3 23,92 ,72 73,7 61 2, ,02 0 ,9 1 ,23 29,93 6,4 51 ,3 ,8 0 ,22 3 94,12 ,71 31,6 85,80801 , 75,0 172,1 224,7542 6 , 66,45131 , 152,7 47,442 7 , 64,9 211,0 41,9 191,6 27 ,7 153,73221 , 37,7 93,0

B SCLC NSCLC Lung cancer cell lines (dataset GSE4824) Samples stratified according to histopathology: NSCLC vs. SCLC

0.10 0.05 0.00 -0.05 -0.10 -0.15 -0.20 -0.25 FDR = 0.06 -0.30

Enrichment score (ES) P < 0.05 -0.35

-3 -0.62 1.49

Figure 6. Similarities between the SOX4 expression signature and the expression profile of SCLC. A, heat map depicting the comparison of gene expression levels of the indicated transcripts in NSCLC and SCLC. The arrows indicate 2 AC cell lines with a similar profile to those of the SCLC type. B, left, 2D hierarchical clustering of the genes with greatest differential expression between NSCLC and SCLC cell lines. Right, graphical representation of the ranked gene lists derived from the comparison (using GSEA) of the indicated data set and the genes downregulated in the H522Tr-shSOX4-1 clone upon induction of the shSOX4. The P and false discovery rate (FDR) values are indicated. preferentially occurred in ACs whereas SOX2 and SOX9 were family in this lung cancer type. This occurs in the absence of more commonly upregulated in the SCC type. It is important gene amplification and indicates that there are alternative to note that SOX2 is located on chromosome 3q26 and is mechanisms for SOX-gene upregulation in this lung tumor amplified and overexpressed in lung SCCs (24, 40). Moreover, type. Sox4 and Sox2 are known to be TGFb target genes (21) shRNA targeting SOX2 decreased colony formation in SOX2- anditcannotberuledoutthatthehighlevelsofSOX4and amplified cells lines (40), supporting the causative effect of SOX2 in SCLC in the absence of an associated gene alteration SOX2 amplification in lung cancer development. However, may be because of an active TGFb pathway. While the high because the 3q26 amplicon includes another bona fide expression levels of SOX4 in SCLC had been described by us oncogene, PIK3CA (24), it is not possible to determine which and others (2, 44), the strong and unique association is the target for gene amplification in each case. Interest- between high levels of SOX11 and the SCLC type is a new ingly, SOX2, SOX4,andSOX11 exhibited very strong expres- finding. Here, we also report that the SOX4-associated sion levels in most SCLCs, implying specificrolesfortheSOX expression profile has significant similarities with the SCLC

www.aacrjournals.org Cancer Res; 72(1) January 1, 2012 OF9

Castillo et al.

gene expression signature. Collectively, these observations tumors with neuroendocrine characteristics, especially of strongly implicate SOX4 and other SOX family genes in the the SCLC type. development of the SCLC type. The SOX4 and SOX11 proteins are members of the Sox-C group, are molecularly very Disclosure of Potential Conflict of Interest similar (4) and both are critical to the induction of the expression of neuronal traits and the maintenance of neural No potential conflicts of interest were disclosed. stem cells (21, 29). Taken into account this background and the fact that many of the SOX4 targets are related to neural Authors' Contributions tissues, the high levels of SOX4 and SOX11 in SCLC may also indicate specific cells of origin. In fact, it has been suspected S.D. Castillo, A. Matheu, R. Lovell-Badge, and M. Sanchez-Cespedes conceived and designed the experiments. S.D. Castillo, A. Matheu, and N. Mariani carried for a long time that the SCLC type arises from neuroendo- out the experiments. F. Lopez-Rios contributed to the histopathologic and crine cells, commonly found in clusters known as neuroen- immunohistochemical analysis and evaluation of the data. J. Carretero did the bioinformatic analysis. S.D. Castillo and M. Sanchez-Cespedes analyzed the data docrine bodies. Recent observations, using mouse models and wrote the manuscript. for targeted Trp53 and Rb1 inactivation in distinct cell types of the adult lung, support the neuroendocrine origin of SCLC Acknowledgments (45). It is likely that along with their role in the stemness of the neuroendocrine cells of the lung, SOX4 and possibly The authors acknowledge the technical assistance of Albert Coll (PEBC) and SOX11, are required for the development of SCLC. of Orlando Dominguez at the Genomics Unit (CNIO). In conclusion, through the abrogation of SOX4 expression in lung cancer cells carrying SOX4 amplification and over- Grant Support expression, we have identified novel transcriptional targets S.D. Castillo was supported for fellowships from the CAM, AGAUR, and EMBO for SOX4. We show that these are upregulated in lung short-term fellowship-ASTF-125-2009 and A. Matheu by a postdoctoral fellow- tumors with SOX4 overexpression, indicating involvement ship from the HFSP. Work of R. Lovell-Badge is funded by UK-MRC (U117512772) and of M. Sanchez-Cespedes by MICINN (SAF2008-02698), RTICC (RD06/0020/ in lung cancer development. We have also found that SOX4 0062), and EU-Framework Programme (HEALTH-F2-2010-258677). and other SOX family genes are strongly upregulated in The costs of publication of this article were defrayed in part by the payment of the SCLC type. Given the neural-related nature of many page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. of the SOX4 targets identified here and the similarity between the gene expression signatures of SOX4 and SCLC, Received October 25, 2011; revised October 31, 2011; accepted November 2, we propose that SOX4 is involved in the development of lung 2011; published OnlineFirst November 14, 2011.

References 1. Pao W, Girard N. New driver mutations in non–small cell lung cancer. 12. Liu P, Ramachandran S, Ali Seyed M, Scharer CD, Laycock N, Dalton Lancet Oncol 2011;12:175–80. WB, et al. Sex-determining region Y box 4 is a transforming oncogene 2. Medina PP, Castillo SD, Blanco S, Sanz-Garcia M, Largo C, Alvarez S, in human prostate cancer cells. Cancer Res 2006;66:4011–9. et al. The SRY-HMG box gene, SOX4, is a target of gene amplification 13. Pramoonjago P, Baras AS, Moskaluk CA. Knockdown of Sox4 expres- at chromosome 6p in lung cancer. Hum Mol Genet 2009;18:1343–52. sion by RNAi induces apoptosis in ACC3 cells. Oncogene 2006;25: 3. Prior HM, Walter MA. SOX genes: architects of development. Mol Med 5626–39. 1996;2:405–12. 14. Tavazoie SF, Alarcon C, Oskarsson T, Padua D, Wang Q, Bos PD, et al. 4. Penzo-Mendez AI. Critical roles for SoxC transcription factors in Endogenous human microRNAs that suppress breast cancer metas- development and cancer. Int J Biochem Cell Biol 2010;42:425–8. tasis. Nature 2008;451:147–52. 5. Farr CJ, Easty DJ, Ragoussis J, Collignon J, Lovell-Badge R, Good- 15. Du Y, Spence SE, Jenkins NA, Copeland NG. Cooperating cancer- fellow PN. Characterization and mapping of the human SOX4 gene. gene identification through oncogenic-retrovirus-induced insertional Mamm Genome 1993;4:577–84. mutagenesis. Blood 2005;106:2498–505. 6. van de Wetering M, Oosterwegel M, van Norren K, Clevers H. Sox-4, an 16. Lund AH, Turner G, Trubetskoy A, Verhoeven E, Wientjens E, Hulsman Sry-like HMG box protein, is a transcriptional activator in lymphocytes. D, et al. Genome-wide retroviral insertional tagging of genes involved in EMBO J 1993;12:3847–54. cancer in Cdkn2a-deficient mice. Nat Genet 2002;32:160–5. 7. Wegner M. From head to toes: the multiple facets of Sox proteins. 17. Shin MS, Fredrickson TN, Hartley JW, Suzuki T, Akagi K, Morse HC III. Nucleic Acids Res 1999;27:1409–20. High-throughput retroviral tagging for identification of genes involved 8. Aaboe M, Birkenkamp-Demtroder K, Wiuf C, Sorensen FB, Thykjaer in initiation and progression of mouse splenic marginal zone lympho- T, Sauter G, et al. SOX4 expression in bladder carcinoma: clinical mas. Cancer Res 2004;64:4419–27. aspects and in vitro functional characterization. Cancer Res 18. Suzuki T, Shen H, Akagi K, Morse HC, Malley JD, Naiman DQ, et al. 2006;66:3434–42. New genes involved in cancer identified by retroviral tagging. Nat 9. Beer DG, Kardia SL, Huang CC, Giordano TJ, Levin AM, Misek DE, Genet 2002;32:166–74. et al. Gene-expression profiles predict survival of patients with lung 19. Rhodes DR, Yu J, Shanker K, Deshpande N, Varambally R, Ghosh D, adenocarcinoma. Nat Med 2002;8:816–24. et al. Large-scale meta-analysis of cancer microarray data identifies 10. Evans AJ, Gallie BL, Jewett MA, Pond GR, Vandezande K, Underwood common transcriptional profiles of neoplastic transformation and J, et al. Defining a 0.5-mb region of genomic gain on chromosome progression. Proc Natl Acad Sci U S A 2004;101:9309–14. 6p22 in bladder cancer by quantitative-multiplex polymerase chain 20. Reymann S, Borlak J. Transcription profiling of lung adenocarcinomas reaction. Am J Pathol 2004;164:285–93. of c-myc-transgenic mice: identification of the c-myc regulatory gene 11. Garber ME, Troyanskaya OG, Schluens K, Petersen S, Thaesler Z, network. BMC Syst Biol 2008;2:46. Pacyna-Gengelbach M, et al. Diversity of gene expression in adeno- 21. Ikushima H, Todo T, Ino Y, Takahashi M, Miyazawa K, Miyazono K. carcinoma of the lung. Proc Natl Acad Sci U S A 2001;98:13784–9. Autocrine TGF-beta signaling maintains tumorigenicity of glioma-

OF10 Cancer Res; 72(1) January 1, 2012 Cancer Research

Novel SOX4 Targets and Small Cell Lung Cancer Development

initiating cells through Sry-related HMG-box factors. Cell Stem Cell scriptional network in prostate cancer cells. Cancer Res 2009;69: 2009;5:504–14. 709–17. 22. Thein DC, Thalhammer JM, Hartwig AC, Crenshaw EB III, Lefebvre V, 34. Cartharius K, Frech K, Grote K, Klocke B, Haltmeier M, Klingenhoff A, Wegner M, et al. The closely related transcription factors Sox4 and et al. MatInspector and beyond: promoter analysis based on tran- Sox11 function as survival factors during spinal cord development. scription factor binding sites. Bioinformatics 2005;21:2933–42. J Neurochem 2010;115:131–41. 35. Yokota S, Hirayama T, Hirano K, Kaneko R, Toyoda S, Kawamura Y, 23. Lowry WE, Blanpain C, Nowak JA, Guasch G, Lewis L, Fuchs E. et al. Identifi cation of the cluster control region for the protocadherin- Defining the impact of beta-catenin/Tcf transactivation on epithelial {beta} genes located beyond the protocadherin-{gamma} cluster. stem cells. Genes Dev 2005;19:1596–611. J Biol Chem 2011;286:31885–95. 24. Angulo B, Suarez-Gauthier A, Lopez-Rios F, Medina PP, Conde E, 36. Reppe S, Rian E, Jemtland R, Olstad OK, Gautvik VT, Gautvik KM. Sox- Tang M, et al. Expression signatures in lung cancer reveal a profile for 4 messenger RNA is expressed in the embryonic growth plate and EGFR-mutant tumours and identify selective PIK3CA overexpression regulated via the parathyroid hormone/parathyroid hormone-related by gene amplification. J Pathol 2008;214:347–56. protein receptor in osteoblast-like cells. J Bone Miner Res 2000;15: 25. Conde E, Angulo B, Tang M, Morente M, Torres-Lanzas J, Lopez- 2402–12. Encuentra A, et al. Molecular context of the EGFR mutations: evidence 37. Schilham MW, Oosterwegel MA, Moerer P, Ya J, de Boer PA, van de for the activation of mTOR/S6K signaling. Clin Cancer Res 2006;12: Wetering M, et al. Defects in cardiac outflow tract formation and pro-B- 710–7. lymphocyte expansion in mice lacking Sox-4. Nature 1996;380:711–4. 26. Pantoja C, Serrano M. Murine fibroblasts lacking p21 undergo senes- 38. Krishnamurthy J, Ramsey MR, Ligon KL, Torrice C, Koh A, Bonner- cence and are resistant to transformation by oncogenic Ras. Onco- Weir S, et al. p16INK4a induces an age-dependent decline in islet gene 1999;18:4974–82. regenerative potential. Nature 2006;443:453–7. 27. Edgar R, Domrachev M, Lash AE. Gene expression omnibus: NCBI 39. Molofsky AV, Slutsky SG, Joseph NM, He S, Pardal R, Krishnamurthy gene expression and hybridization array data repository. Nucleic Acids J, et al. Increasing p16INK4a expression decreases forebrain progeni- Res 2002;30:207–10. tors and neurogenesis during ageing. Nature 2006;443:448–52. 28. Subramanian A, Tamayo P, Mootha VK, Mukherjee S, Ebert BL, Gillette 40. Bass AJ, Watanabe H, Mermel CH, Yu S, Perner S, Verhaak RG, et al. MA, et al. Gene set enrichment analysis: a knowledge-based approach SOX2 is an amplified lineage-survival oncogene in lung and esoph- for interpreting genome-wide expression profiles. Proc Natl Acad Sci ageal squamous cell carcinomas. Nat Genet 2009;41:1238–42. U S A 2005;102:15545–50. 41. Liao YL, Sun YM, Chau GY, Chau YP, Lai TC, Wang JL, et al. 29. Bergsland M, Werme M, Malewicz M, Perlmann T, Muhr J. The Identification of SOX4 target genes using phylogenetic footprinting- establishment of neuronal properties is controlled by Sox4 and Sox11. based prediction from expression microarrays suggests that over- Genes Dev 2006;20:3475–86. expression of SOX4 potentiates metastasis in hepatocellular carcino- 30. Bhattaram P, Penzo-Mendez A, Sock E, Colmenares C, Kaneko KJ, ma. Oncogene 2008;27:5578–89. Vassilev A, et al. Organogenesis relies on SoxC transcription factors for 42. Mariani TJ, Reed JJ, Shapiro SD. Expression profiling of the develop- the survival of neural and mesenchymal progenitors. Nat Commun ing mouse lung: insights into the establishment of the extracellular 2010;1:9. matrix. Am J Respir Cell Mol Biol 2002;26:541–8. 31. Dy P, Penzo-Mendez A, Wang H, Pedraza CE, Macklin WB, Lefebvre V. 43. Maeda Y, Dave V, Whitsett JA. Transcriptional control of lung mor- The three SoxC proteins–Sox4, Sox11 and Sox12–exhibit overlapping phogenesis. Physiol Rev 2007;87:219–44. expression patterns and molecular properties. Nucleic Acids Res 44. Bangur CS, Switzer A, Fan L, Marton MJ, Meyer MR, Wang T. 2008;36:3101–17. Identification of genes over-expressed in small cell lung carcinoma 32. Hoser M, Potzner MR, Koch JM, Bosl MR, Wegner M, Sock E. Sox12 using suppression subtractive hybridization and cDNA microarray deletion in the mouse reveals nonreciprocal redundancy with the expression analysis. Oncogene 2002;21:3814–25. related Sox4 and Sox11 transcription factors. Mol Cell Biol 45. Sutherland KD, Proost N, Brouns I, Adriaensen D, Song JY, Berns A. 2008;28:4675–87. Cell of origin of small cell lung cancer: inactivation of Trp53 and Rb1 33. Scharer CD, McCabe CD, Ali-Seyed M, Berger MF, Bulyk ML, in distinct cell types of adult mouse lung. Cancer Cell 2011;19: Moreno CS. Genome-wide promoter analysis of the SOX4 tran- 754–64.

www.aacrjournals.org Cancer Res; 72(1) January 1, 2012 OF11

Novel Transcriptional Targets of the SRY-HMG Box Transcription Factor SOX4 Link Its Expression to the Development of Small Cell Lung Cancer

Sandra D. Castillo, Ander Matheu, Niccolo Mariani, et al.

Cancer Res Published OnlineFirst November 14, 2011.

Updated version Access the most recent version of this article at: doi:10.1158/0008-5472.CAN-11-3506

Supplementary Access the most recent supplemental material at: Material http://cancerres.aacrjournals.org/content/suppl/2011/11/14/0008-5472.CAN-11-3506.DC1

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://cancerres.aacrjournals.org/content/early/2011/12/20/0008-5472.CAN-11-3506. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.