Oncogene (2011) 30, 398–409 & 2011 Macmillan Publishers Limited All rights reserved 0950-9232/11 www.nature.com/onc ORIGINAL ARTICLE eIF3m expression influences the regulation of tumorigenesis-related in human colon cancer

S-H Goh1, S-H Hong1, S-H Hong1, B-C Lee1, M-H Ju2, J-S Jeong2, Y-R Cho1, I-H Kim1 and Y-S Lee1

1Division of Convergence Technology, Research Institute, National Cancer Center, Goyang, Republic of Korea and 2Department of Pathology, Dong-A University College of Medicine, Busan, Republic of Korea

Abnormal regulation of expression is essential for mRNA, but the flow of information from gene to tumorigenesis. Recent studies indicate that regulation of in eukaryotes is not immediate enough to oncogene expression and neoplastic transformation are comply with rapid changes in the environment. Eukar- controlled by subunits of eukaryotic initiation yotes solve this problem by maintaining a pool of factors (eIFs). Here we focused on eIF3 performing a mRNAs that are not utilized currently (De Benedetti pivotal role in protein synthesis and the differential ex- and Graff, 2004). In metazoans, translational efficiency pression of its subunits in cancer. The most uncharac- varies over a hundred fold (Koch et al., 1980) and their terized non-core subunit eIF3m was confirmed to be highly translation depends on the prevailing growth conditions. expressed in human cancer cell lines and colon cancer patient According to the filter hypothesis, the trans- tissues. By expression silencing with eIF3m-specific small lation initiation factors recruit certain sets of mRNA interfering RNA (siRNA), we confirmed that eIF3m species to the ribosome and fulfill the requirements for influences cell proliferation, cell cycle progression and cell cell proliferation without causing a drastic shift in the death in human colon cancer cell line HCT-116. Using a existing mRNA pool (Mauro and Edelman, 2002). ribonomics approach, we identified a subset of elF3m- 3 (eIF3) is the influenced genes and showed that the expression of two largest (B800 kDa) and the most complex mammalian highly represented tumorigenesis-related genes, MIF and initiation factor. elF3 has a pivotal role in protein MT2,wereaffectedbyeIF3matthemRNAlevel.Wealso synthesis that bridges the 43S pre-initiation complex confirmed eIF3m-dependent regulation of MT2A down- and eIF4F-bound mRNA (Silvera et al., 2010). It is stream molecule CDC25A, which is necessary for cell cycle composed of 13 non-identical polypeptides designated progression in HCT-116 cells. These results suggest that as eIF3a–m (Zhou et al., 2005). Five of the subunits eIF3m mediates regulation of tumorigenesis-related genes in (eIF3 a, b, c, g and i) are conserved in all eukaryotes human colon cancer. Further investigations on tumorigen- (hence called conserved ‘core’) and the remaining eight esis-related genes and their regulation by eIFs will provide are considered ‘non-core’ (LeFebvre et al., 2006). The clues for designing targeted therapy for cancer. functional roles of individual subunits are not yet Oncogene (2011) 30, 398–409; doi:10.1038/onc.2010.422; known, but aberrant expression of several eIF3 subunits published online 13 September 2010 were detected in various human cancers and reviewed (Dong and Zhang, 2006). For example, eIF3a is Keywords: eukaryotic translation initiation factor; overexpressed in human breast, esophagus, cervix, lung, eIF3m; cell proliferation; cell death; ribonomics stomach and colon cancers; eIF3b in human breast cancers; eIF3c in human testicular cancer; eIF3h in breast, liver, and prostate cancers; and eIF3i in breast cancer. Elevated expression of eIF3i in breast cancer Introduction was suggested to facilitate mTOR-dependent growth transformation (Ahlemann et al., 2006). eIF3e, also Development of neoplasm is characterized by increased called INT6 and originally known as a tumor suppressor cell proliferation, which requires a general increase in (Green et al., 2007) and shown to be expressed in a protein synthesis and a specific increase in the synthesis truncated form in tumors, was intensively investigated of replication-promoting (Rosenwald, 2004). (Mack et al., 2007) and has been implicated in tumor Expression of each gene depends on the abundance of its progression (Morris et al., 2007). eIF3f was reported to be lost in melanoma and pancreatic cancer (Doldan et al., 2008). Recently, it was reported that the Correspondence: Dr Y-S Lee, Functional Branch, Research overexpression of eIF3 subunits leads to malignant Institute, National Cancer Center, 111 Jungbalsan-ro, Ilsandong-gu, transformation of immortal fibroblast cells (Zhang 809 Madu-dong, Goyang, Gyeonggi-do 410-769, Republic of Korea. E-mail: [email protected] et al., 2007). Thus far, there have been no reports of Received 8 June 2009; revised 11 July 2010; accepted 10 August 2010; altered expression of any of the other eIF3 subunits in published online 13 September 2010 cancer and whether these eIF3 subunits are involved in eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 399

Figure 1 Elevated expression of eukaryotic translation initiation factor subunit, elF3m, in cancer cell lines and in human colon tumor tissues. RT–qPCR was done by absolute quantification and the copy number of eIF3m transcript per nanogram of total RNA was plotted with standard deviation from replication (n ¼ 3). (a) Absolute copy numbers of eIF3m transcript in colon cancer cell lines (HCT-116, SNU-C4, SNU-C5 and SNU-81), lung cancer cell line (A549) and breast cancer cell lines (JIMT-1 and SK-BR-3) were much higher than in normal human dermal fibroblast cell line (HDF). (b) Most of the human colon tissues showed higher expression of eIF3m in tumor than in normal counterparts. Western blotting also confirmed higher eIF3m expression in cancer cell line (c) and human colon tissues (d). tumorigenesis is not known. Especially, as the last tumor. For this, we performed reverse defined eIF3 subunit, the function of eIF3m in humans real-time quantitative PCR (RT–qPCR) and meas- is not yet characterized. ured the transcripts of 13 eIF3 subunits in pairs of We hypothesized that the non-core subunit eIF3m normal colon and tumor (adenocarcinoma) tissues may be involved in the process of cell proliferation from 19 patients by relative quantification normalized and tumor progression by affecting the mRNAs of a by GAPDH (Supplementary Figure S1). eIF3b was the specific subset of genes. If any differences are found in only eIF3 core subunit that showed high expression in expression of elF3m between normal and tumor tissues, tumors of most of the patients (18/19). Among the this would indicate the functional importance of elF3m eight non-core subunits, eIF3d, e, h, k and m subunits in tumor progression. showed elevated expression in tumors of more than 50% In this study we confirmed the expression of the of patients. We chose the most uncharacterized eIF3m uncharacterized eIF3m subunit in cancer cell lines as for further study and confirmed the elevated mRNA well as in tumor regions of human colon tissue. We also levels of elF3m in tumors of all patients as well as in demonstrated a change in cell proliferation, cell death seven kinds of cultured cancer cell lines by absolute and cell cycle progression when eIF3m expression was quantification deduced from a standard curve (r240.95) silenced by specific siRNA in human colon cancer cell (Figures 1a and b). In all cancer cell lines including four line HCT-116. Furthermore, we identified target genes colon cancer cell lines (HCT-116, SNU-C4, SNU-C5 regulated by eIF3m expression by employing a ribo- and SNU-81), one lung cancer cell line (A549) and two nomics approach, which was developed for the identi- breast cancer cell lines (Herceptin-resistant JIMT-1 fication of mRNAs in mRNP complexes (Tenenbaum and Herceptin-sensitive SK-BR-3), it showed more than et al., 2002), and confirmed that their expression is 10-fold elevated expression level compared with the regulated in an eIF3m-dependent manner. All together, normal cell line HDF (Figure 1a). Interestingly, JIMT-1 we suggest that eIF3m is involved in the regulation of showed more than double the expression in SK-BR-3 tumorigenesis-related genes in human colon cancer. (Figure 1a). While the relative quantification data showed higher expression in tumors of 12 out of 19 patients, the absolute quantification showed markedly Results elevated expression of eIF3m in tumors in all patients (paired t-test P ¼ 0.00013) (Figure 1b). In accordance Elevated expression of eIF3m subunits in cancer with this, cancer cell lines showed higher protein levels We first examined whether upregulation of eIF3 of eIF3m than normal cell line HDF (Figure 1c). The subunits is related to the progression of human colon difference in eIF3m protein expression between normal

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 400 Tissue specificity of eIF3m mRNA expression We examined the tissue specificity of eIF3m expression at transcription level by Northern hybridization. Among the normal tissues, heart, skeletal muscle, kidney, liver, placenta and peripheral blood leukocytes showed elevated eIF3m expression, but in brain, colon, thymus, small intestine and lung there was no detectable expression of eIF3m (Figure 2a). Among several human cancer cell lines, eIF3m expression in several leukemia cell lines, a lymphoma, a colorectal adenocarcinoma SW480 and in a lung carcinoma cell line A549 was high but was relatively low in melanoma cell line G-361 (Figure 2b). In this hybridization no splice variant of eIF3m transcript was found.

Confirmation of the preferential expression of eIF3m in human colon carcinoma tissues To determine the region(s) of eIF3m expression on human colon tissue, we performed immunohisto- chemistry. We detected markedly high expression of eIF3m in colon carcinoma cells (Figures 3c and f) and in metastatic carcinoma cells in the regional lymph node (Figure 3a). In contrast, we detected much lower signals in the adjacent non-neoplastic epithelial cells (Figure 3b and e). In terms of subcellular localization, eIF3m expression was not detected in the nucleus but was confined to the cytosol in both carcinoma and non- neoplastic epithelial cells. The metastatic colon carcino- ma in liver also showed strong expression of eIF3m (Figures 3g–i). These results suggest that eIF3m is highly expressed in the colon carcinoma cells where cell proliferation is active at a higher level than in the adjacent non-neoplastic epithelial cells.

Silencing eIF3m expression reduces proliferation of human colon cancer cells The low expression of eIF3m in the peritumoral regions confirmed by immunohistochemistry suggests that Figure 2 eIF3m mRNA expression in normal human tissues and human cancer cell lines by northern blot. (a) Normal tissues eIF3m is also involved in reactive proliferation as well showed very weak or no eIF3m signal except in heart compared as in tumor progression. To determine the effect of with b-actin. (b) In cancer cell lines, other than melanoma G-361 eIF3m expression on cell proliferation, we studied the cell line, eIF3m showed very strong signals. No mRNA splice effect of silencing eIF3m mRNA in HCT-116 colon variant was found either in normal tissues or in cancer cell line. cancer cells using siRNA. Silencing of eIF3m expression was confirmed by western blotting (Figure 4a and Supplementary Figure S3a). eIF3m expression was reduced from 24 h after siRNA transfection until 96 h. and tumorous colon tissue was also similar, although In contrast, buffer only (BF) or negative control (NC) did not as dramatic as the mRNA levels (Figure 1d). To not change the level of eIF3m protein. The confluency make sure of its expression difference, we checked after BF and NC rapidly increased and reached a plateau its mRNA and protein levels in 20 additional paired by 72 h (Figure 4b). In contrast, eIF3m siRNA-1 slowed patient tissues (Supplementary Figures S2a and b). Then down the proliferation from 24 h and kept lower we evaluated protein expression difference between confluency until 96 h (Figure 4b). MTT assays confirmed normal and tumor tissues by densitometry (Supple- this by showing not more than threefold increase with mentary Figure S2c). The average density of eIF3m eIF3m siRNA-1 at 72 h compared with sixfold with NC itself in tumor tissues was 1.3-fold higher than in its and fivefold with BF (Figure 4c). This result was normal counterpart (paired t-test P ¼ 0.00098). The fold confirmed again by using another siRNA, siRNA-3 change of normalized density by b-actin was also 1.3 (Supplementary Figure S3b). The silencing efficacies of (P ¼ 0.00070). These analyses suggest that elevated siRNA-1 and siRNA-3 compared with NC at 96 h expression of eIF3m is associated with tumor progression were 46.4% (Student’s t-test P ¼ 1.3 Â 10À5) and 65.3% in cell lines as well as in colon tissues. (Student’s t-test P ¼ 2.8 Â 10À9), respectively. These results

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 401

Figure 3 Immunohistochemical study of eIF3m in human colon carcinoma tissues. Brown spots on whole tissue images represent eIF3m expression in colon carcinoma tissues (a, d) and liver (g) as well as in regional lymph node (a). (b, e) Adjacent non-neoplastic crypts. The epithelial cells lining the crypt adjacent to colon adenocarcinoma display also elevated expression of eIF3m. This region is the place where reactive proliferations occur. (c, f) Higher magnification of colon carcinoma. The individual carcinoma cell shows strong expression of eIF3m in the cytoplasm. (g) Whole tissue of metastatic carcinoma in the liver. Brown spots represent eIF3m expression in metastatic carcinomas. (h) Higher magnification of adjacent liver. Hepatocytes show little or no positive signal. (i) Higher magnification of metastatic colon carcinoma. Metastatic carcinoma cells show strong eIF3m signal as primary colon carcinoma. suggest that proliferation of cancer cells is retarded by BF (31.16%) or NC (29.22%) by 96 h. Taken together, silencing eIF3m expression. these results suggest that silencing of eIF3m expression increased subG0/G1 proportion and led to cell death. eIF3m silencing leads to death of human colon cancer cells In agreement with this, we could confirm the effect of eIF3m silencing on cell death using several markers In the context of the cell proliferation retardation by for apoptosis. First, it showed a steady increase of eIF3m silencing, we analyzed cell cycle progression annexin V-positive cells by eIF3m siRNA-1 until 96 h following eIF3m knockdown in HCT-116 cell lines (Figure 6a). Second, the frequency of abnormal chromo- by flow cytometry. The HCT-116 cells were treated with eIF3m siRNA-1, siRNA-3, BF, or NC controls. At somes with condensed and/or fragmented appearance was significantly higher by eIF3m siRNA-1 (Figure 6b). 24 h of siRNA transfection, there was no significant Third, PARP1, which is known for caspase substrate, difference in the proportion of each mitosis stage almost completely turned into its activated form after treatment with NC, BF or eIF3m siRNA-1 or -3 (89 kDa) from its inactive form (116 kDa) by eIF3m (Figure 5 and Supplementary Figure S4). As the time silencing compared with the slight change by NC point moved from 24 to 96 h, the portion of sub-G0/G1 (Figure 6c). Thus, eIF3m expression seems to be phase in eIF3m siRNA-1-treated cells increased from required for the cells to continue cell-cycle and 1.65 to 12.36%. On the other hand, the proportion of S eventually for cell proliferation. phase of eIF3m siRNA-1 decreased from 26.40 to 19.49%, while it increased in both BF (from 24.67 to 32.22%) and NC (from 26.00 to 30.56%). Similarly, G2/ Identification of EIF3m-associated transcripts M phase also decreased from 31.14 to 21.81%. eIF3m by ribonomics siRNA-3 also showed increased sub-G0/G1 (11.26%) To identify a subset of genes of which expression portion compared with BF (2.52%) or NC (3.31%) and is regulated by eIF3m, we employed a ribonomics decreased G2/M-phase (19.75%) portion compared with strategy. The EIF3m open cloned into

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 402

Figure 4 Silencing of eIF3m expression by siRNA results in reduction of cell proliferation rate. The eIF3m siRNA-1-transfected HCT-116 (wt) cells were incubated for 96 h. Negative controls included buffer alone, no siRNA (BF) and non-human negative control siRNA (NC). (a) Inhibition of eIF3m expression by eIF3m siRNA-1 treatment was confirmed by western blotting. eIF3m expression at translation level dramatically decreased in HCT-116 cells treated with eIF3m siRNA-1 from 24 h. (b) Proliferation patterns of HCT- 116 cells at 24-h intervals. Proliferation of eIF3m siRNA-1-transfected cells was retarded unlike in BF and NC controls. (c) MTT assay of siRNA-treated cells showing that eIF3m siRNA-1 slowed down the proliferation of HCT-116 cells; this pattern is similar to that of cells shown in panel b. MTT assay was performed in replicate with n ¼ 3 and plotted with s.d. The difference between eIF3m siRNA-1 and NC was significant (Student’s t-test P ¼ 1.3 Â 10À5).

pFLAG-CMV2 was expressed in HCT-116 cell lines. control did not (Supplementary Figure S5a). When we After 48 h of transfection, the tagged eIF3m was purified total RNA from immunoprecipitated gel pellet, immunoprecipitated with a FLAG-M2 affinity gel. The the eIF3m expression clone generated the detect- EIF3m-inserted plasmid generated the designated size of able amount of RNA but the blank vector did not eIF3m band on western blot, whereas blank vector (Supplementary Figure S5b). This RNA was used to

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 403

Figure 5 Sub-G0/G1 population of HCT-116 colon cancer cell lines increased when eIF3m expression was silenced. The nuclear contents of siRNA-transfected cells were measured every 24 h after treatment by flow cytometry to observe the cell cycle progression. BF and NC controls did not show any significant cell cycle differences among mitotic cell cycle stages. The eIF3m siRNA-1 treatment increased sub-G0/G1 stage at 24 h and remained elevated until 96 h. make a cDNA library. Among the 344 randomly picked level from 72 h (Figure 7a). The UC1MT antibody clones, 181 successful reads represented 81 kinds of detects both MT1 and MT2 isoforms, but as MT2A was eIF3m-associated genes (Table 1). These sequences the major form in ribonomics, we applied the antibody include 75 genes (41.4%), eIF3m to detect MT2A protein. The protein level of MT2A itself (25.4%), 54 single represented genes (29.8%) and 8 also decreased until 48 h, but it did not return to the unclassified entries (4.4%). Classification of these genes normal level (Figure 7b) unlike MIF. mRNA level by for molecular function by DAVID estimated by RT–qPCR detected mRNA level changes Bioinformatics Resource and SOURCE search revealed (Figure 7c) in a similar pattern of protein level (Figures that most of the eIF3m-associated transcripts encode 7a and b). This suggests that eIF3m influences the genes for protein translation (protein binding, structural expression of a designated subset of genes for cell constituent of the ribosome, translation initiation factor proliferation and tumor progression by affecting the activity and ribonucleoprotein binding) and RNA mRNA level. In addition, as Lim et al. (2009) reported binding (nucleic acid binding, RNA binding and rRNA that MT2A silencing resulted in cell division cycle 25 binding) (Table 1 and Supplementary Figure S6). homolog A (CDC25A) degradation and G1-arrest in Another class of genes noted was related to metal ion- breast cancer cells, we found that eIF3m silencing binding activity (cadmium ion and copper ion binding). immediately decreased CDC25A from 24 h until 72 h For further investigation on these transcripts we (Figure 8). These results further support that eIF3m chose two highly represented genes, macrophage migra- expression influences the cell cycle regulation. tion inhibitory factor (MIF) and metallothionein 2A (MT2A). They have recently been associated with tumorigenesis: MIF gene with tumor growth (Bifulco Discussion et al., 2008; Bach et al., 2009), and MT2A gene with breast cancer (Jin et al., 2004) and chemoresistance in Abnormal cell proliferation and cell cycle progres- ovarian cancers (L’Espe´rance et al., 2006). We examined sion are hallmarks of aggressive malignant neoplasm, whether the expression of MIF and MT2A is influenced and upregulated protein synthesis is important to the by eIF3m in the HCT-116 cells. When eIF3m siRNA-1- etiology of cancer (Rosenwald, 2004). Recent reports treated HCT-116 cell lysate was examined by western confirmed the importance of eIF3s in regulation of blot, the protein level of MIF decreased until 48 h, as translation initiation rate and in the oncogenic roles of was eIF3m expression, but it was revived to the normal several of its subunits (Dong and Zhang, 2006; Zhang

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 404 that elevated expression of eIF3m is an attribute of tumor tissues and/or proliferating cancer cell lines. It was reported that eIF3e in its truncated form leads to persistent hyperplasia and tumorigenesis in mammary alveolar epithelium (Mack et al, 2007) and subunits of eIF3 also showed oncogencity when modified by truncation or phosphorylation. However, in this study we could not detect any variant of eIF3m mRNA that could be considered as truncated. Also, no alternatively spliced forms of transcript was found in either normal tissues or in cancer cell lines, though its expression levels, observed in Northern blot, were different among normal tissues. The high expression of eIF3m was not restricted to the colon cancer cell line but also occurred in several leukemias and carcinomas. This suggests potential role of eIF3m in tumorigenesis in general. This study also demonstrated a low level of eIF3m expression in the normal cell line, HDF, and in the normal counterparts of tumor samples. This raises the following questions: Is eIF3m preferentially highly expressed only in tumor cells? If it is also expressed in normal cells, does it depend on the physiological state of the cell such as cell proliferation? To address these questions, we performed immunohistochemistry on several human colon tumor tissues. The tumor region showed vivid eIF3m expression compared with the peritumoral region. We also found recognizable expression of eIF3m in some other regions, for example, in the lining of colon lumen. These regions are not of the tumor itself, but are places where reactive proliferation occurs and mamma- lian intestinal mucosa undergoes a process of continual Figure 6 Analyses of apoptosis induced by silencing of eIF3m cell turnover essential for maintenance of normal expression in HCT-116 cells. (a) Increase of annexin V-positive function by epithelial restitution with migration, pro- cells by eIF3m silencing. Apoptotic cells were detected by FITC annexin V staining in conjunction with propidium iodide and liferation and differentiation of epithelial cells to form average frequencies (% gated) are plotted with standard error. new crypts (Ramachandran et al., 2000; Humphries and Annexin V-positive cells are rapidly increased until 96 h in eIF3m Wright, 2008). These results therefore indicate that siRNA-1-treated cells, but it was almost similar throughout the eIF3m expression is required more for reactive prolif- time course in NC cells. Representative scatter plot of NC and eIF3m siRNA-1 at 96 h are on the right side. (b) Hoechst 33342 eration of normal cells in response to inflammation and staining of and nuclei. Condensation of nucleus and/ for proliferation of tumors, than for non-proliferating or fragmented in apoptotic cells (indicated by arrow normal cells. heads) was counted in eIF3m-silenced and NC cells after 72 h of eIF3m’s involvement in cell proliferation was con- transfection. Frequency of aberrant chromosome in NC remained firmed by silencing its expression with siRNA in human around 11% but it was over 41% in eIF3m siRNA-treated cells. The significance of difference was assessed by Student’s t-test and colon cancer cell line HCT-116. This retarded the P-value was below 0.001. (c) PARP1 was detected by western progression into G0/G1 phase and the fraction of cells blotting. PARP1 is a main cleavage target of caspase-3 and it gives in subG0/G1 phase increased as the time of siRNA out 89 kDa cleaved form from its 116 kDa full-length form. treatment increased. In accordance with the reduction of Compared with the slight increase of PARP1 in NC, almost complete activation of PARP1 was observed from 72 h in eIF3m cell proliferation, it confirmed the increase of apoptotic siRNA-1 condition. cells by increase in the number of annexin-V-positive cells, fragmented/condensed chromosomes and acti- vated PARP1 protein when eIF3m expression was silenced. This implies that lack of eIF3m leads to the et al., 2007, 2008). As the last defined eIF3 subunit, the apoptosis of proliferating cells. function of eIF3m in humans is not yet characterized. If As the overexpression of eIF4E (Gingras et al., 1999; any differences are found in expression of elF3m Raught and Gingras, 1999) and eIF4G (Fukuchi- between normal and tumor tissues, this would indicate Shimogori et al., 1997; Pyronnet et al., 1999; LeFebvre the functional importance of elF3m in tumor progres- et al., 2006) was reported to be responsible for their sion. Our assessment of mRNA levels of elF3 subunits oncogencity, it was reported that five subunits of elF3 by RT–qPCR showed elevated eIF3m expression in lead to the oncogenic properties of NIH-3T3 cells cancer cell lines and in colon adenocarcinoma tissues. (Zhang et al., 2007). Thus, eIF3m may also influence a The expression difference at protein level was also subset of oncogenic genes by mediating interaction with detected by western blotting. These correlations suggest ribosomal machinery or by regulating mRNA level. We

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 405 Table 1 Ribonomic identification of genes influenced by eIF3m Gene description Gene symbol Possible molecular RefSeq ID functiona

ADAPTOR-RELATED PROTEIN COMPLEX 4, AP4S1 Protein transporter activity NM_007077 SIGMA 1 SUBUNIT ATX1 ANTIOXIDANT PROTEIN 1 HOMOLOG (YEAST) ATOX1 Copper-chaperone activity NM_004045 ATP SYNTHASE, H þ TRANSPORTING, MITOCHON- ATP5G1 Transporter activity NM_005175 DRIAL F0 COMPLEX, SUBUNIT C1 (SUBUNIT 9) ATP SYNTHASE, H þ TRANSPORTING, MITOCHON- ATP5L ATPase activity NM_006476 DRIAL F0 COMPLEX, SUBUNIT G BETA-2-MICROGLOBULIN B2M Structural molecular activity NM_004048 SIMILAR TO RPE-SPONDIN C20ORF199 NAb XM_945311 CHROMOSOME 21 OPEN READING FRAME 45 C21ORF45 NA NM_018944 CHROMOSOME 7 OPEN READING FRAME 30 C7ORF30 NA NM_138446 HYPOTHETICAL PROTEIN FLJ10803 C7ORF44 NA NM_018224 COILED-COIL DOMAIN CONTAINING 72 CCDC72 NA NM_015933 EUKARYOTIC TRANSLATION INITIATION FACTOR 1 EIF1 Translation initiation factor activity NM_005801 EUKARYOTIC TRANSLATION INITIATION FACTOR 2, EIF2S2 Translation initiation factor activity NM_003908 SUBUNIT 2 BETA, 38KDA DENDRITIC CELL PROTEIN EIF3M Translation initiation factor activity NM_006360 ENHANCER OF YELLOW 2 HOMOLOG (DROSOPHILA) ENY2 Nucleic acid-binding activity NM_020189 FAMILY WITH SEQUENCE SIMILARITY 49, MEMBER B FAM49B NA NM_016623 FXYD DOMAIN CONTAINING ION TRANSPORT FXYD5 Cadherin binding NM_014164 REGULATOR 5 CHORIONIC SOMATOMAMMOTROPIN HORMONE 1 GH1 Growth hormone receptor binding NM_022644 (PLACENTAL LACTOGEN) HIGH-MOBILITY GROUP NUCLEOSOME BINDING DO- HMGN1 DNA binding NM_004965 MAIN 1 HEMATOLOGICAL AND NEUROLOGICAL EXPRESSED 1 HN1 NA NM_016185 MACROPHAGE MIGRATION INHIBITORY FACTOR MIF Cytokine, phenylpyruvate tautomerase NM_002415 (GLYCOSYLATION-INHIBITING FACTOR) activity METALLOTHIONEIN 1E (FUNCTIONAL) MT1E Metal ion binding NM_175617 METALLOTHIONEIN 1X MT1X Metal ion binding NM_005952 METALLOTHIONEIN 2A MT2A Metal ion binding NM_005953 NADH DEHYDROGENASE (UBIQUINONE) 1 ALPHA NDUFA4 NADH dehydrogenase activity NM_002489 SUBCOMPLEX, 4, 9KDA PEROXIREDOXIN 1 PRDX1 Peroxiredoxin activity NM_002574 PITUITARY TUMOR-TRANSFORMING 1 PTTG1 Transcription factor activity NM_004219 RIBOSOMAL PROTEIN L10 RPL10 Structural constituent of ribosome NM_006013 RIBOSOMAL PROTEIN L13 RPL13 Structural constituent of ribosome NM_033251 RIBOSOMAL PROTEIN L17 RPL17 Structural constituent of ribosome NM_001035005 RIBOSOMAL PROTEIN L21 RPL21 Structural constituent of ribosome NM_000982 RIBOSOMAL PROTEIN L23 RPL23 Structural constituent of ribosome NM_000978 RIBOSOMAL PROTEIN L23A RPL23A Structural constituent of ribosome NM_000984 RIBOSOMAL PROTEIN L27 RPL27 Structural constituent of ribosome NM_000988 RIBOSOMAL PROTEIN L27A RPL27A Structural constituent of ribosome NM_000990 RIBOSOMAL PROTEIN L28 RPL28 Structural constituent of ribosome NM_000991 RIBOSOMAL PROTEIN L30 RPL30 Structural constituent of ribosome NM_000989 RIBOSOMAL PROTEIN L31 RPL31 Structural constituent of ribosome NM_000993 RIBOSOMAL PROTEIN L32 RPL32 Structural constituent of ribosome NM_000994 RIBOSOMAL PROTEIN L35A RPL35A Structural constituent of ribosome NM_000996 RIBOSOMAL PROTEIN L36 RPL36 Structural constituent of ribosome NM_033643 RIBOSOMAL PROTEIN L37 RPL37 Structural constituent of ribosome NM_000997 RIBOSOMAL PROTEIN L37A RPL37A Structural constituent of ribosome NM_000998 RIBOSOMAL PROTEIN L39 RPL39 Structural constituent of ribosome NM_001000 RIBOSOMAL PROTEIN L41 RPL41 Structural constituent of ribosome NM_021104 RIBOSOMAL PROTEIN L5 RPL5 Structural constituent of ribosome NM_000969 RIBOSOMAL PROTEIN L6 RPL6 Structural constituent of ribosome NM_001024662 RIBOSOMAL PROTEIN L7A RPL7A Structural constituent of ribosome NM_000972 RIBOSOMAL PROTEIN S11 RPS11 Structural constituent of ribosome NM_001015 RIBOSOMAL PROTEIN S13 RPS13 Structural constituent of ribosome NM_001017 RIBOSOMAL PROTEIN S14 RPS14 Structural constituent of ribosome NM_001025071 RIBOSOMAL PROTEIN S15 RPS15 Structural constituent of ribosome NM_001018 RIBOSOMAL PROTEIN S15A RPS15A Structural constituent of ribosome NM_001019 RIBOSOMAL PROTEIN S17 RPS17 Structural constituent of ribosome NM_001021 RIBOSOMAL PROTEIN S21 RPS21 Structural constituent of ribosome NM_001024 RIBOSOMAL PROTEIN S23 RPS23 Structural constituent of ribosome NM_001025 RIBOSOMAL PROTEIN S25 RPS25 Structural constituent of ribosome NM_001028 RIBOSOMAL PROTEIN S27 (METALLOPANSTIMULIN 1) RPS27 Structural constituent of ribosome NM_001030 RIBOSOMAL PROTEIN S28 RPS28 Structural constituent of ribosome NM_001031 RIBOSOMAL PROTEIN S29 RPS29 Structural constituent of ribosome NM_001030001 RIBOSOMAL PROTEIN S4, X-LINKED RPS4X Structural constituent of ribosome NM_001007

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 406 Table 1 Continued Gene description Gene symbol Possible molecular RefSeq ID functiona

RIBOSOMAL PROTEIN S8 RPS8 Structural constituent of ribosome NM_001012 S100 CALCIUM BINDING PROTEIN A6 (CALCYCLIN) S100A6 Growth factor activity NM_014624 SEC61 BETA SUBUNIT SEC61B P–P-bond-hydrolysis-driven protein NM_006808 transmembrane transporter activity SMALL EDRK-RICH FACTOR 2 SERF2 NA NM_016400 SERPIN PEPTIDASE INHIBITOR, CLADE B (OVALBUMIN), SERPINB6 Serine-type endopeptidase inhibitor NM_004568 MEMBER 6 activity SMALL NUCLEAR RIBONUCLEOPROTEIN D2 SNRPD2 Protein binding NM_004597 POLYPEPTIDE 16.5KDA SMALL NUCLEAR RIBONUCLEOPROTEIN SNRPG Protein binding NM_003096 POLYPEPTIDE G TRANSLOCASE OF INNER MITOCHONDRIAL TIMM50 Protein tyrosine phosphatase activity; NM_001001563 MEMBRANE 50 HOMOLOG (YEAST) ribonucleoprotein binding; protein serine/threonine phosphatase activity TRANSMEMBRANE PROTEIN 14C TMEM14C NA NM_016462 TUMOR PROTEIN, TRANSLATIONALLY-CONTROLLED 1 TPT1 Calcium ion binding NM_003295 TRIO AND F-ACTIN BINDING PROTEIN TRIOBP Actin binding; GTP-Rho binding NM_138632 UBIQUITIN A-52 RESIDUE RIBOSOMAL PROTEIN UBA52 Structural constituent of ribosome NM_001033930 FUSION PRODUCT 1 UBIQUITIN-CONJUGATING ENZYME E2L 3 UBE2L3 Ubiquitin-protein ligase activity NM_198157 VACUOLAR PROTEIN SORTING 35 (YEAST) VPS35 Ubiquitin binding; transcription NM_018206 corepressor activity ZONA PELLUCIDA GLYCOPROTEIN 2 ZP2 Coreceptor activity NM_003460 (SPERM RECEPTOR)

aAnnotations on possible molecular functions were retrieved from SOURCE search of each gene (http://source.stanford.edu). bNA denotes annotation is not available.

applied ribonomics strategy to address this question. suggests that eIF3m expression has an effect in Most of the genes identified from this approach were for regulating mRNA levels of a specific subset of genes protein translation itself (Table 1). However, there were associated with cell proliferation and tumor progression. also many genes involved in tumor progression, such as Further support for this hypothesis is provided by the FXYD5 (FXYD domain containing ion transport reduced expression of MT2A resulting from ubiquitin- regulator 5) (Ino et al., 2002; Nam et al., 2007), PTTG1 dependent degradation of CDC25A, which is necessary (pituitary tumor transforming gene 1) (Dominguez for the progression from G1 to S phase of the cell cycle et al., 1998) and S100A6 (S100 calcium binding protein via ATM/Chk2 in a breast cancer model (Lim et al., A6) (Komatsu et al., 2002). Thus, the eIF3m-associated 2009). Although HCT-116 colon cancer cells showed mRNA subset influences the expression of genes subG0/G1 arrest rather than G1 arrest and S phase required for accelerated cell proliferation, transforma- reduction with CDC25A loss as it was reported (Tomko tion of cells and for their journey from their anchored et al., 2009), we supposed that this difference came from position into other places. This concept, obviously, the different cell property of HCT-116 from breast should be confirmed by further studies. cancer cells, MCF-7 or MCF12. Based on our findings, We also evaluated the eIF3m expression-dependent eIF3m expression may affect cell cycle progression regulation of MT2A and MIF, which were also identi- through the regulation of MT2A level (Figure 7b) and fied from ribonomics by silencing eIF3m. MIF is a its downstream pathway molecule, CDC25A (Figure 8). highly conserved multipotential protein that acts var- In summary, we confirmed that eIF3m expression is iously as a pro-inflammatory cytokine, a pituitary elevated in proliferating cells and human colon tumor hormone, and as a cell proliferation and migration tissues. By silencing of eIF3m expression, we showed that factor (Bifulco et al., 2008). It is a unique cytokine eIF3m affects cell proliferation, cell cycle progression and affecting multiple processes fundamental to tumorigen- cell death in colon cancer cells. To identify the eIF3m- esis (Bach et al., 2009). In addition, MIF has prognostic associated gene subset, we applied ribonomics approach for value in human colorectal cancer because its expression the first time to the study of mammalian eukaryotic correlates with aggressiveness of the cancer (Legendre translation initiation factor and found that many of them et al. 2003). MT2A gene belongs to the metallothionein are involved in cell proliferation and transformation. We families, which are important for metal ion homeostasis. confirmed that two highly represented tumorigenesis-related It was reported to be associated with breast cancer (Jin genes from this subset, MIF and MT2, are regulated by et al., 2002) and tumor grade (Jin et al., 2004) and is eIF3m at the mRNA level. It was shown that eIF3m even- related to the chemoresistance of ovarian cancer tually influences MT2A downstream molecule CDC25A, (L’Espe´rance et al., 2006). The protein level of these which is necessary for cell cycle progression. genes affected by silencing eIF3m and their mRNA level The importance of eukaryotic translation initiation also changed in a similar pattern (Figure 7). This factors in cancer development is increasing more and

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 407

Figure 8 The regulated expression of cell division cycle 25 homolog A (CDC25A) protein in HCT-116 cells at 0, 24, 48, 72 and 96 h after eIF3m siRNA-1 transfection. Human b-actin was also detected for the loading control.

Reverse transcription real-time quantitative PCR RT–qPCR analyses were carried out using 10 ng of cDNA transcribed by Superscript RT III (Invitrogen), derived from 2 mg of total RNA, 5 pmol of forward and reverse primers (see Supplementary Table S1) and QuantiFast SYBR Green PCR master mix (Qiagen, Hilden, Germany) with the following thermal profile: 50 1C at 2 min hold, 95 1C at 10 min hold, 95 1C for 30 s, 58 1C for 30 s, 72 1C for30 s repeated for 40 cycles in Light Cycler 480 (Roche Applied Science, Rotkreuz, Switzer- land). Data were analyzed by either relative quantification normalized with GAPDH or ACTB or the absolute quantifi- cation method.

Northern blotting The full-length open reading frame of EIF3m was labeled with [a-32P] dCTP (Perkin-Elmer NEN, Boston, MA, USA) by RediPrime kit (GE Healthcare, Piscataway, NJ, USA) and Figure 7 The protein expression of macrophage migration filtered through G-50 Microcolumn (GE Healthcare, USA). inhibitory factor (MIF)(a) and metallothionein 2A (MT2A)(b) MTN blot (Clontech, Mountain View, CA, USA) was in HCT-116 cells at 0, 24, 48, 72 and 96 h after eIF3m siRNA-1 hybridized in ULTRAhyb solution (Ambion, Austin, TX, transfection detected by western blot. Human b-actin was also USA) at 42 1C overnight and detected according to standard detected for the loading control. (c) mRNA level of MIF and protocol on BioMax MS film (Kodak, Rochester, NY, USA) MT2A detected by RT–qPCR in HCT-116 cells at 0, 24, 48, 72 and with an intensifying screen at À80 1C. 96 h after eIF3m siRNA-1 transfection. mRNA level was analyzed by relative quantification normalized with b-actin mRNA expres- sion. In BF and NC, which allow cells to proliferate, the mRNA Western blotting level was increased. In contrast, eIF3m silencing kept the mRNA Cell and tissue lysates were prepared with protease inhibitor- level constant throughout the time course in MIF and decreased supplemented mammalian protein extraction reagent or tissue the mRNA level in MT2A. The specificity of RT–qPCR reaction was confirmed by sequencing of PCR product. protein extraction reagent lysis buffer (Thermo Scientific, Rockford, IL, USA). The lysates resolved on 4–12% NuPAGE gel (Invitrogen) were transferred onto Immobilon- more. Further investigations on the effect of regulation P membrane (Millipore, Billerica, MA, USA). They were of tumorigenesis-related genes by eukaryotic translation detected with the following antibodies for each condition: anti- initiation factor subunits, similar to eIF3m, on human eIF3m (1:2000, #11423-1-AP, Proteintech Group, Chicago, IL, cancers will provide clues for identifying potential USA); anti-MIF (2a10-4d3) (1:500, #H00004282-M01, Abno- therapeutic targets in human cancer. va, Taiwan); anti-metallothionein monoclonal antibody (UC1MT), which detects all metallothionein isoforms (1:500, #ab12228, Abcam, Cambridge, UK); anti-CDC25A (1:1,000, Materials and methods #3652, Cell Signaling, Danvers, MA, USA); and finally 1:10 000 diluted HRP-conjugated secondary antibody (Sig- Human tissues ma-Aldrich, St Louis, MO, USA). The signal density of the Human colonic tissues were obtained by surgical resection region of interest was measured with background subtraction from consenting patients in accordance with the protocols and using Multi Gauge V3.0 (Fujifilm, Tokyo, Japan). guidelines approved by the Institutional Review Board and the principles of Helsinki. Immunohistochemistry Human tumor tissues were fixed in 10% neutral buffered Cell culture formalin solution and subjected to the standard procedure to All cell lines were obtained from American Type Culture make paraffin blocks. Slide-mounted 4-mm-thick tissues were Collection (ATCC, Manassas, VA, USA) or the Korean Cell pre-treated with proteinase and then stained by a BechMark Line Bank (Seoul, Korea). Each cell line was maintained in XT automated system (Ventana Medical System, Tucson, AZ, designated media (Mediatech Inc., Manassas, VA, USA) USA). Anti-eIF3m antibody (#11423-1-AP, Proteintech Group, supplemented with 10% FBS (Invitrogen, Carlsbad, CA, USA). Chicago, IL, USA) was applied at 1:100 dilutions and the signal

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 408 was detected by biotinylated secondary antibodies, followed by ATCAGAAAGACTCAAAAGGCTG-30) primers and cloned the binding of streptavidin–HRP conjugate. into pFLAG-CMV2 (Sigma-Aldrich). Two million HCT-116 cells transfected with 4 mg of pFLAG-CMV2-eIF3m by Silencing of eIF3m expression Lipofectamine 2000 (Invitrogen) were cultured for 48 h at The effects of eIF3m on cell proliferation and cell cycle 37 1C. The cells were homogenized in Symplekin immuno- were assayed after treatment with eIF3m-specific siRNA precipitation buffer (Kim and Richter, 2006) and cleared lysate (#1: S10043164 50-ATGGATAAGAATACTCCTGTA-30;#3: was precipitated with 40 ml of FLAG-M2 affinity gel (Sigma- S10043178 50-AAGCAAGAAGCTTTGATTGAA-30, Qiagen). Aldrich). Total RNA was purified from immunoprecipitated Two million HCT-116 cells transfected with eIF3m siRNA, or gel pellet by Trizol reagent. In all, 1.575 mg of total RNA was negative control siRNA NC or BF, were cultured for 96 h in a used for making cDNA library by GeneRacer kit (Invitrogen) six-well plate and checked every 24 h. and amplification by 25 cycles of PCR with adapter primers. The PCR product was cloned into pBlueScriptII-KS( þ ) vector (Stratagene, La Jolla, CA, USA) and transformed Cell proliferation and cell cycle analyses DH10B competent cells (Invitrogen). Nucleotide sequences Proliferation was assessed by MTT assay. Cell cycle progres- from randomly chosen colonies were searched against NCBI sion was analyzed in the same manner with approximately 4 GenBank, DAVID Bioinformatics Resource (Huang et al., 5 Â 10 HCT-116 cells stained in propidium iodide staining 2009) and SOURCE search (http://source.stanford.edu). buffer (10 mg/ml DNase-free RNase A, 50 mg/ml propidium iodide in phosphate buffered saline) by BD Cellquest software on FACSCalibur Flow Cytometer (BD Biosciences, San Jose, Assessment of statistical significance CA, USA). The statistical significance was analyzed by Student’s two- tailed t-test for comparison between unpaired groups and Analyses of apoptosis Student’s paired two-tailed t-test for comparison between The reaction to detect apoptotic cells based on plasma normal/tumor tissue pairs. For displaying the error range, we membrane degradation was performed using FITC Annexin adopt standard deviation or standard error mean (s.e.m.) for V apoptosis detection kit (BD Pharmingen, San Diego, CA, each group. USA; #556 547) according to the manufacturer’s protocol and then analyzed on FACSCalibur Flow Cytometer (BD Bioscience). For the observation of chromosome condensation and fragmentation, cells were fixed with 4% paraformaldehyde Conflict of interest in 1 Â phosphate buffered saline for 20 min at room tempera- ture and chromosomes stained with Hoechst 33342 were The authors declare no conflict of interest. observed under a microscope with UV laser (detection at 480 nm). The activated form of poly (ADP-ribose) polymerase (PARP1) by caspase cleavage was detected using anti-PARP1 (1:1,000, #9542, Cell Signaling) antibody. Acknowledgements

Ribonomics We thank Dr Sri Ram for editorial assistance of manuscript. The eIF3m open reading frame was amplified with forward This study was funded by Intramural Research Grants of the (50-CACCATGAGGGTCCCGGC-30) and reverse (50-GGT National Cancer Center (NCC 0710660, NCC 0810160).

References

Ahlemann M, Zeidler R, Lang S, Mack B, Mu¨nz M, Gires O. (2006). Fukuchi-Shimogori T, Ishii I, Kashiwagi K, Mashiba H, Ekimoto H, Carcinoma-associated eIF3i overexpression facilitates mTOR- Igarashi K. (1997). Malignant transformation by overproduction dependent growth transformation. Mol Carcinog 45: 957–967. of translation initiation factor eIF4G. Cancer Res 57: Bach JP, Deuster O, Balzer-Geldsetzer M, Meyer B, Dodel R, 5041–5044. Bacher M. (2009). The role of macrophage inhibitory factor in Gingras AC, Gygi SP, Raught B, Polakiewicz RD, Abraham RT, tumorigenesis and central nervous system tumors. Cancer 115: Hoekstra MF et al. (1999). Regulation of 4E-BP1 phosphorylation: 2031–2040. a novel two-step mechanism. Genes Dev 13: 1422–1437. Bifulco C, McDaniel K, Leng L, Bucala R. (2008). Tumor growth- Green EM, Barrett CF, Bultynck G, Shamah SM, Dolmetsch RE. (2007). promoting properties of macrophage migration inhibitory factor. The tumor suppressor eIF3e mediates calcium-dependent internaliza- Curr Pharm Des 14: 3790–3801. tion of the L-type calcium channel CaV1.2. Neuron 55: 615–632. De Benedetti A, Graff JR. (2004). eIF-4E expression and its role in Huang DW, Sherman BT, Lempicki RA. (2009). Systematic and malignancies and metastases. Oncogene 23: 3189–3199. integrative analysis of large gene lists using DAVID Bioinformatics Doldan A, Chandramouli A, Shanas R, Bhattacharyya A, Cunning- Resources. Nature Protoc 4: 44–57. ham JT, Nelson MA et al. (2008). Loss of the eukaryotic initiation Humphries A, Wright NA. (2008). Colonic crypt organization and factor 3f in pancreatic cancer. Mol Carcinog 47: 235–244. tumorigenesis. Nat Rev Cancer 8: 415–424. Dominguez A, Ramos-Morales F, Romero F, Rios RM, Dreyfus F, Ino Y, Gotoh M, Sakamoto M, Tsukagoshi K, Hirohashi S. (2002). Tortolero M et al.(1998).hPTTG,ahumanhomologueofratpttg,is Dysadherin, a cancer-associated cell membrane glycoprotein, down- overexpressed in hematopoietic neoplasms. Evidence for a transcrip- regulates E-cadherin and promotes metastasis. Proc Natl Acad Sci tional activation function of hPTTG. Oncogene 17: 2187–2193. USA 99: 365–370. Dong Z, Zhang J-T. (2006). Initiation factor eIF3 and regulation of Jin R, Chow VTK, Tan PH, Dheen ST, Duan W, Bay BH. (2002). mRNA translation, cell growth, and cancer. Crit Rev Oncol Metallothionein 2A expression is associated with cell proliferation Hematol 59: 169–180. in breast cancer. Carcinogenesis 23: 81–86.

Oncogene eIF3m influences tumorigenesis-related gene regulation S-H Goh et al 409 Jin R, Huang J, Tan PH, Bay BH. (2004). Clinicopathological Morris C, Wittmann J, Jack HM, Jalinot P. (2007). Human INT6/ significance of metallothioneins in breast cancer. Pathol Oncol Res eIF3e is required for nonsense-mediated mRNA decay. EMBO Rep 10: 74–79. 8: 596–602. Kim JH, Richter JD. (2006). Opposing polymerase-deadenylase activities Nam JS, Hirohashi S, Wakefield LM. (2007). Dysadherin: a new regulate cytoplasmic polyadenylation. Mol Cell 24: 173–183. player in cancer progression. Cancer Lett 255: 161–169. Koch G, Bilello JA, Kruppa J, Koch F, Oppermann H. (1980). Pyronnet S, Imataka H, Gingras AC, Fukunaga R, Hunter T, Amplification of translational control by membrane-mediated Sonnenberg N. (1999). Human eukaryotic translation initiation events: a pleiotropic effect on cellular and viral . factor 4G (eIF4G) recruits mnk1 to phosphorylate eIF4E. EMBO J Ann N Y Acad Sci. 339: 280–306. 18: 270–279. Komatsu K, Andoh A, Ishiguro S, Suzuki N, Hunai H, Ramachandran A, Madesh M, Balasubramanian KA. (2000). Apop- Kobune-Fujiwara Y et al. (2002). Increased expression of S100A6 tosis in the intestinal epithelium: its relevance in normal and (Calcyclin), a calcium-binding protein of the S100 family, in human pathophysiological conditions. J Gastroenterol Hepatol 15: 109–120. colorectal adenocarcinomas. Clin Cancer Res 6: 172–177. Raught B, Gingras AC. (1999). eIF4E activity is regulated at multiple L0Espe´rance S, Popa I, Bachvarova M, Plante M, Patten N, Wu L levels. Int J Biochem Cell Biol 31: 43–57. et al. (2006). Gene expression profiling of paired ovarian tumors Rosenwald IB. (2004). The role of translation in neoplastic transfor- obtained prior to and following adjuvant chemotherapy: molecular mation from a pathologist’s point of view. Oncogene 23: 3230–3247. signatures of chemoresistant tumors. Int J Oncol 29: 5–24. Silvera D, Formenti SC, Schneider RJ. (2010). Translational control in LeFebvre AK, Korneeva NL, Trutschl M, Cvek U, Duzan RD, cancer. Nat Rev Cancer 10: 254–266. Bradley CA et al. (2006). Translation initiation factor eIF4G-1 Tenenbaum SA, lager PJ, Carson CC, Keene JD. (2002). Ribonomics: binds to eIF3 through the eIF3e subunit. J Biol Chem 281: Identifying mRNA subsets in mRNP complexes using antibodies to 22917–22932. RNA-binding proteins and genomic arrays. Methods 26: 191–198. Legendre H, Decaestecker C, Nagy Y, Hendlisz A, Schu¨ring M-P, Tomko RJ, Azang-Njaah NN, Lazo JS. (2009). Nitrosative stress Salmon I et al. (2003). Prognostic values of galectin-3 and the suppresses checkpoint activation after DNA synthesis inhibition. macrophage migration inhibitory factor (MIF) in human colorectal Cell Cycle 8: 299–305. cancers. Mod Pathol 16: 491–504. Zhang L, Pan X, Hershey JW. (2007). Individual overexpression of five Lim D, Jocelyn KMX, Yip GWC, Bay BH. (2009). Silencing the of human translation initiation factor eIF3 promotes malig- metallothionein-2A gene inhibits cell cycle progression from G1- to nant transformation of immortal fibroblast cells. J Biol Chem 282: S-phase involving ATM and cdc25A signaling in breast cancer cells. 5790–5800. Cancer Lett 276: 109–117. Zhang L, Smit-McBride Z, Pan X, Rheinhardt J, Hershey JW. (2008). Mack DL, Boulanger CA, Callahan R, Smith GH. (2007). Expression An oncogenic role for the phosphorylated h-subunit of human of truncated Int6/eIF3e in mammary alveolar epithelium leads to translation initiation factor eIF3. J Biol Chem 283: 24047–24060. persistent hyperplasia and tumorigenesis. Breast Cancer Res 9: R42. Zhou C, Arslan F, Wee S, Krishnan S, Ivanov AR, Oliva A et al. Mauro VP, Edelman GM. (2002). The ribosome filter hypothesis. (2005). PCI proteins eIF3e and eIF3m define distinct translation Proc Natl Acad Sci USA 99: 12031–12036. initiation factor 3 complexes. BMC Biol 3: 14.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene