RNA-binding motif protein 35A is a novel tumor suppressor for colorectal cancer

Abstract

The frequent occurrence of inactivating gene mutations in tumors suggests a tumor suppressor function of the mutated gene. The RNA binding motif protein 35A (RBM35A) is mutated in ~50% of analyzed primary colon tumors with microsatellite instability. The Tet-off regulated ectopic expression of RBM35A gene in RBM35A-null LS180 colon carcinoma cells inhibited anchorage-independent growth in vitro, suppressed tumorigenic potential in vivo and enhanced adhesiveness of these cancer cells. Using microarray hybridization we found that in response to RBM35A expression a small fraction of genes showed a decrease in polysome-associated mRNA. Experiments using cell-free in vitro translation system demonstrated that RBM35A differentially affects translation of luciferase reporter mediated by various 5′untranslated regions (UTR). We found that Gibbs energy value (ΔG) of secondary structure formed by 5′UTRs of mRNAs can account for differential effect of RBM35A on reporter translation efficiency. Targeted mutation in the FOS 5′UTR sequence, which increased the ΔG value of hairpin stem formation, resulted in a stronger inhibitory effect of RBM35A on reporter translation efficiency mediated by this UTR. Immunoblotting revealed that ectopic expression of RBM35A in LS180 cells caused alterations in protein levels for several cancer related genes. Our results demonstrate for the first time that RBM35A functions as a tumor suppressor in colon cancer cells. We propose that RBM35A is involved in posttranscriptional regulation of a number of genes by exerting a differential effect on protein translation via 5′UTRs of mRNAs.

Introduction

RNA binding proteins (RBPs) involved in RNA metabolism directly interact with RNA molecules via RNA binding domains (RBD). The most common class of RBDs is the RNA recognition motif (RRM), found in >50% of the RBPs.^{1, 2} Numerous and diverse RBPs are found in various ribonucleoprotein complexes, which are implicated in pre-mRNA processing, RNA transport, cellular localization and degradation.¹ Genetic and proteomic data provide evidence that aberrations in function of RBPs lead to defects in RNA metabolism and thereby may underlie a broad spectrum of complex human diseases including cancer. In line with known capabilities of RBPs to modulate cell growth and proliferation,² both proto-oncogene and tumor suppressor activities have been reported for various RBPs.^{3, 4}

We have previously found that the LS180 colon carcinoma cell line carries a homozygous frameshift mutation in the (A)₈ coding mononucleotide repeat of the RBM35A gene, which results in degradation of mutant mRNA through nonsense mediated decay pathway.⁵ The RBM35A gene, which is predicted to contain three putative RNA recognition motifs (RRM_1 superfamily) in its polypeptide sequence, has been also found to be mutated in ~50% (11 out 23) of primary colon tumors with microsatellite instability (MSI).⁵ Frameshift mutations in the coding microsatellites are the major mechanism of inactivation of tumor suppressor genes in cancers with MSI.⁶ However, because of the inherent instability of the microsatellite DNA and high frequency of mutations caused by inactivation of the mismatch repair system in cancer cells with MSI, bi-allelic inactivating gene mutations alone do not necessarily indicate a tumor suppressor function of a mutant gene.^{7, 8}

In this study, in order to demonstrate the tumor suppressor capability of the RBM35A, we re-introduced RBM35A into LS180 cells, null for its expression, using a Tet-off tetracycline repressible expression system. Ectopic expression of RBM35A protein resulted in suppression of tumorigenic potential of LS180 colon cancer cells. Microarray analysis showed that ectopic expression of RBM35A resulted in moderate changes in polysomal loading of a number of genes that could indicate involvement of RBM35A in posttranscriptional regulation of gene expression. We further demonstrated that RBM35A exerts a differential effect on reporter RNA translation mediated by various 5′UTRs and that the degree of 5′UTR-mediated translational inhibition by RBM35A was strongly dependent on the complexity of the secondary structure of the 5′UTRs. We propose that RBM35A protein may suppress malignant potential of colorectal cancer cells by exerting a tuning effect on translation of a number of RNA transcripts and that could result in alterations in protein levels of several cancer related genes. To our knowledge this is the first demonstration that RBM35A functions as a tumor suppressor in colon cancer cells.

Results

Ectopically expressed RBM35A protein inhibits tumorigenic potential of LS180 colon cancer cells

To demonstrate the relevance of the RBM35A gene inactivation to colon carcinogenesis, we investigated the effect of ectopic expression of the RBM35A on a malignant phenotype of the RBM35A-null LS180 colon carcinoma cells. To this end, the Tet-off LS180-RBM35A cell line, expressing the flag-tagged wild type RBM35A protein under control of a tetracycline repressible promoter, was established. Expression of RBM35A protein was induced by removal of doxycycline (Dox) from the growth medium and verified by immunoblotting with anti-flag antibody (Suppl. Fig. 1A). RT-PCR analysis showed that the amount of mRNA for the ectopically expressed RBM35A gene in LS180 cells does not exceed the level of endogenous RBM35A mRNA in the cells with a wild type of this gene (Suppl. Fig. 1B), thus indicating physiological level of the RBM35A expression. Hereafter, for simplicity, the LS180 cells expressing RBM35A (maintained in the Dox-free medium) and the LS180 cells lacking this protein (the same cells grown in Dox supplemented medium) will be referred to as RBM35+ and RBM35− cells, respectively.

The restoration of RBM35A expression did not affect the proliferative potential of the LS180 cells under tissue culture conditions, as was determined by colorimetric MTS assay (data not shown). Anchorage-independent colony formation, a hallmark in vitro characteristic of transformation, which closely correlates with the tumorigenicity of epithelial cells, was significantly reduced upon restoration of RBM35A expression. We found that RBM35+ cells formed 46% fewer large colonies in soft agar than RBM35− cells (Fig. 1A). Furthermore, the restoration of RBM35A function significantly reduced the in vivo tumorigenic potential of LS180 cells in the athymic nu/nu mice assay. Three weeks after subcutaneous inoculation of these cells, xenograft tumors formed by RBM35+ cells in mice on Dox-free diet were on average ~4 times smaller than tumors established from control RBM35− cells in mice administered Dox in their drinking water (Fig. 1B). Thus, ectopic re-introduction of RBM35A inhibited the tumorigenic potential of the LS180 cells in two assays for malignant phenotype, strongly indicating a tumor suppressive capability of the RBM35A gene.

Figure 1.

Effects of ectopic re-introduction of RBM35A on malignant phenotype of LS180 colon cancer cells. (A) Anchorage-independent growth of RBM35+ and RBM35− cells as measured by colony growth in soft agar. Columns, mean from one of three experiments performed in triplicate; bars, SD. (B) RBM35+ and RBM35− cells were grown as tumor xenografts in nude mice. Columns, mean from five mice; bars SD. (C) RBM35⁺ cells display a greater adhesiveness to the surface of culture vessels not suitable for growth of cell monolayers. RBM35⁻ and RBM35+ cells (10⁶/well) were plated in non-tissue culture treated dishes. 24–48 h later cells were gently washed with PBS, fixed and stained with Diff-Quick fixative and Stain Set (DADE Behring, Newark, DE). Photographs (left) show well defined monolayer formed by RBM35+ cells in contrast to rounded and merely adhered RBM35− cells, which are easily detached. (D) Immunoblot of whole cell lysates with anti-E-cadherin antibody reveals increased level of E-cadherin in RBM35+ cells. Results in (B–D) are representative of two independent experiments. Statistical analysis was performed using Student’s t-test.

RBM35A-expressing LS180 colon cancer cells display enhanced adhesiveness in vitro

Pathways related to cell-matrix and cell-cell adhesion are known to be altered in colorectal cancer.⁹ Therefore, it is of interest that we observed a strikingly greater ability of RBM35A-expressing cells to attach to the surface of culture vessels, especially to the more hydrophobic non-tissue culture treated dishes not suitable for growth of cell monolayers. Two days after being plated in such dishes, more than 90% of RBM35− cells were floating as clumps and were easily removed during washes and the fixation/staining procedure, whereas most of RBM35+ cells were attached and well spread on the bottom of the wells as seen in Figure 1C. RBM35+ cells also showed an increase in the cellular level of tumor suppressor protein E-cadherin (Fig. 1D), a cell membrane adhesion molecule, which is involved in a maintenance of normal tissue structure.¹⁰ These findings implicate RBM35A in regulation of the cell adhesion pathway in colon cancer cells, further supporting a tumor suppressor activity of this protein.

RBM35A protein expression alters level of polysomal loading for some RNA transcripts

To gain an insight into the molecular mechanisms underlying the tumor suppressive effects of RBM35A gene, we first investigated the intracellular localization of RBM35A protein. Immunofluorescence analysis using anti-flag antibody showed that, in transiently transfected Hela cells, RBM35A is primarily localized in the cytoplasm (Fig. 2). This finding, together with the fact that the RBM35A protein sequence contains putative RNA-binding motifs, prompted the hypothesis that RBM35A might be involved in the regulation of mRNA turnover and/or in translational regulation of cancer related genes. To explore these possibilities, we investigated whether induction of RBM35A expression affects (1) total RNA profile or (2) polysome-associated RNA profile in Tet-off LS180-RBM35A cells using hybridization to Affymetrix U95 Genechip array. The microarray analysis showed that induced expression of RBM35A gene did not significantly alter total RNA profile in the LS180 cells indicating that stabilization or degradation of RNA transcripts may not be affected by the RBM35A protein. However, 55 hybridization probes (corresponding to 48 genes) out of 11,943 probes corresponding to approximately 6,000 genes represented on the Affymetrix U95 Genechip showed a more than two-fold change in polysome-associated RNA between the RBM35+ and RBM35− samples (8 and 47 probes up and down, respectively (Suppl. Table 1)), thus suggesting that RBM35A may be involved in translational regulation of a subset of RNA transcripts. Interestingly, among genes that showed RBM35A induced change in polysomal RNA were many genes implicated in carcinogenesis, including proto-oncogenes MYC and FOS (Suppl. Table 1). The values observed in polysomal RNA were very moderate (slightly exceeding two-fold change threshold level for the majority of the detected transcripts). The form of distribution curve of the fold-change alterations (Suppl. Fig. 2) suggested that none of the detected genes could be assigned as a specific target for the RBM35A protein. Therefore, we next proposed that RBM35A protein could have a general effect on translation of multiple mRNAs. Taking into consideration that ribosomes are recruited and engaged in translation at 5′ untranslated regions of mRNAs, we further hypothesized that RBM35A might exert differential effect on 5′UTR mediated translation initiation of a number of RNA transcripts. To test this hypothesis we next analyzed the effect of RBM35A protein on translational efficiency of luciferase reporter RNA controlled by various 5′UTRs using an in vitro translation system.

Figure 2.

Ectopically expressed flag-tagged RBM35A protein is predominantly localized in cytoplasm of transiently transfected Hela cells. The cells were subjected to indirect immunofluorescence staining with anti-flag antibody.

RBM35A exerts a differential effect on in vitro translation of luciferase reporter RNA controlled by various 5′UTRs

5′UTRs of MYC, FOS, C19orf6, PPP5C, PKM2 and GJB5 mRNAs were inserted in Luciferase SP6 control plasmid (Fig. 3A). Cloned 5′UTRs differed by lengths, complexity of predicted secondary structure and the presence of upstream AUGs. Flag-tagged RBM35A protein was synthesized in vitro as described in the legend for Supplementary Figure 3 and production of the full length protein was confirmed by immunoblotting with anti-flag antibody (Suppl. Fig. 3). The effect of RBM35A protein on in vitro translation of luciferase reporter RNA mediated by various 5′UTRs was analyzed using a Coupled Transcription/Translation Rabbit Reticulocyte Lysate System. In vitro translation reactions were performed in the presence or absence of RBM35A protein followed by measurement of produced luciferase activity. Control reaction analyzing the effect of RBM35A on translation efficiency of the control reporter RNA coded by Luciferase SP6 plasmid without any 5′UTR upstream of luciferase cDNA was performed for each 5′UTR analyzed. This control was aimed to exclude any effects of RBM35A not related to a 5′UTR as well as the effect of variability in translational efficiency between different preparations of Rabbit Reticulocyte lysate. Ratios of luciferase activity produced in the presence of RBM35A to that produced in the absence of the protein were calculated for control (no UTR) plasmid and for experimental constructs containing 5′UTRs. The ratio for control plasmid was set to 100% and ratios for experimental 5′UTR constructs were expressed as % of control.

Figure 3.

(A) (Top) The structures of the Luciferase SP6 constructs containing various 5′UTRs. (Bottom) RBM35A protein differentially affects in vitro translation of luciferase reporter mediated by various 5′UTRs. Luciferase was produced from luciferase SP6 constructs in the presence or absence of RBM35A protein using a coupled transcription/translation system. Data are presented as described in Results and represent mean of two to six independent experiments (n = 6 for MYC; n = 3 for PKM2; n = 3 for PPP5C; n = 3 for C19orf6; n = 2 for GJB5; n = 6 for wt FOS; n = 4 for mut FOS) ±SD. ΔG, Gibbs energy values (kcal/mole). (B) The levels of polysome-bound c-myc and c-fos mRNAs (normalized to GAPDH level) are reduced in RBM35+ cells relative to RBM35− cells as shown by real-time RT-PCR. Results are mean of three for c-myc and of two for c-fos independent experiments ±SD. (C) Luciferase cDNA was first transcribed using an in vitro transcription system and then the effect of RBM35A protein on translation of control (no UTR) and MYC 5′UTR-containing RNA transcripts was assayed using Rabbit Reticulocyte Lysate translation mix. Values are mean of three independent experiments ±SD. p values were calculated using paired one tailed t-test.

Results of in vitro translation studies (Fig. 3A) support our hypothesis that RBM35A differentially affects a reporter translation mediated by distinct 5′UTRs, although a little correlation was observed between the translational effect of RBM35A and the microarray data on RNA polysomal loading (Suppl. Table 1). As seen in Figure 3A, RBM35A dramatically inhibited luciferase activity produced from reporter constructs containing MYC and PKM2 5′UTRs, although MYC and PKM2 mRNAs showed a decrease and an increase in polysomal loading, respectively. RBM35A had no effect on luciferase expression mediated by 5′UTRs of GJB5 and FOS mRNAs, although there was a slight increase and a decrease in polysomal loading of GJB5 and FOS mRNAs, respectively. Only a mild inhibition of luciferase activity by RBM35A was detected for the plasmids containing PPP5C and C19orf6 5′UTRs. These data indicate that a 5′UTR mediated translational effect of RBM35A does not directly correlate with polysomal loading of the respective mRNA. Quantitative real-time RT-PCR analysis for the MYC and FOS mRNAs confirmed a decrease in polysomal loading of their mRNAs (Fig. 3B) and together with the results of in vitro translation studies with MYC and FOS 5′UTRs suggest that RBM35A probably is not involved in association of an mRNA with polysomes. Observed changes in RNA polysomal loading may be a secondary consequence of RBM35A function.

Since in vitro coupled transcription/translation system was used to analyze the effect of RBM35A on reporter translation efficiency we next ruled out the possibility that the observed inhibition of luciferase expression in vitro by RBM35A could be explained by its effect on transcription rather than on translation. To this end, we first synthesized the luciferase RNA transcripts using an in vitro transcription system primed with either control (no UTR) or MYC 5′UTR-containing luciferase SP6 constructs. Consequently, the effect of RBM35A on translation of these RNA transcripts was assayed using Rabbit Reticulocyte Lysate translation mix primed with equal amounts of RNA. As shown in Figure 3C, RBM35A efficiently inhibits translation of luciferase RNA transcript containing MYC 5′UTR in cell-free translation mix, thus confirming that the observed in vitro inhibition of luciferase expression by RBM35A occurs at the level of translation. Together these findings indicate that a 5′untranslated region is a critical determinant in the effect of RBM35A on reporter RNA translation and that the outcome of the RBM35A activity is apparently dependent on the sequence of a particular 5′UTR.

To find an explanation for the observed differential effect of RBM35A on translation of the reporter RNA controlled by various 5′UTRs, we considered length, presence of AUGs in the UTR sequences and the Gibbs energy values (ΔG) (Fig. 3A) of the secondary structure formation calculated using Mfold program.¹¹ The presence of five upstream AUGs in the sequence of PPP5C 5′UTR, on one hand, and only moderate inhibition of reporter translation mediated by this 5′UTR in the presence of RBM35A, on the other hand, ruled out the critical role of upstream AUGs in RBM35A mediated inhibition of translation. Interestingly, we found that the degree of inhibitory effect of RBM35A on in vitro translation correlated with the lengths of 5′UTRs and with the values of ΔG of secondary structure formed by 5′UTRs. The effect of RBM35A on in vitro translation correlated more strongly with the values of ΔG (Pearson coefficient = 0.88) than with the lengths of 5′UTRs (Pearson coefficient = 0.81) suggesting that the complexity rather than the length of a 5′UTR determines the RBM35A mediated translational inhibition.

Theoretical secondary structures of 5′UTRs for analyzed genes can be found in Supplementary Figure 4. Based on the analysis of putative secondary structure of the 5′UTRs, we suggested that the lengths of double stranded GC enriched stems of hairpins in 5′UTRs could serve as determinants in RBM35A inhibitory activity on in vitro translation of luciferase reporter mediated by the 5′UTRs.

To test this hypothesis, we modified the sequence of FOS 5′UTR by substituting CAU trinucleotide for GGG at the location indicated by arrows in Figure 4, using Stratagene mutagenesis kit. This modification changed the ΔG of predicted secondary structure from −47 to −62 kcal/mole, thus resulting in a stronger hairpin stem in the mutated FOS 5′UTR, which contains a stretch of eight GC pairs as compared to the stretch of four GC pairs in the corresponding location of wild type Fos 5′UTR (compare inserts in Fig. 4). As shown in Figure 3A, RBM35A significantly inhibited luciferase production mediated by mutant 5′UTR, while having no effect on reporter translation controlled by the wild type FOS 5′UTR, thereby supporting the hypothesis that the strength of GC enriched hairpin stems in the structure of a 5′UTR determines the degree of inhibition exerted by RBM35A on translation mediated by the 5′UTR.

Figure 4.

Putative secondary structures of wild type (wt) and mutated (mut) FOS 5′UTR, created using Mfold program. Arrows indicate site of mutation. Inserts show enlarged area of mutation localization.

RBM35A exerts 5′UTR-mediated inhibitory effect on the translation of firefly luciferase reporter in vivo

To test whether inhibitory translational effect of RBM35A also takes place in vivo, we chose MYC 5′UTR to analyze the effect of RBM35A on 5′UTR mediated translation of luciferase reporter transfected into cells. MYC 5′UTR was inserted directly upstream of the firefly luciferase coding sequence in pGL3 control vector, creating an experimental construct pGLmyc-UTR (Fig. 5). RBM35+ and RBM35− LS180 cells were transiently transfected with pGL3-C (no UTR) or pGLmycUTR constructs together with Renilla luciferase plasmid as a transfection control and the relative luciferase activities produced from pGL3-C and pGLmyc-UTR constructs were scored in each cell line. The effect of RBM35A on reporter translation was assessed as a ratio of the relative luciferase activity produced in RBM35+ cells to that produced in RBM35− cells calculated for each construct. The ratio for control (no UTR) plasmid was set to 100% and ratio for pGLmycUTR was expressed as % of control. Consistent with our in vitro translation studies (Fig. 3A), relative luciferase activity produced from pGLmycUTR was significantly inhibited in RBM35+ cells relative to that in RBM35−cells (Fig. 5). In addition, RBM35A also showed a similar inhibitory effect on MYC 5′UTR mediated reporter translation in HEK293 cells (Fig. 5), when the cells were cotransfected with either RBM35A−flag expression construct or empty vector together with firefly and Renilla luciferase plasmids. Together these results confirm that a 5′UTR-mediated inhibitory translational effect of RBM35A observed in a cell-free system also takes place in a cellular environment.

Figure 5.

RBM35A protein inhibits translation of luciferase reporter mediated by MYC 5′UTR in Tet-off LS180-RBM35A and HEK293 cells. The expression cassette of the pGL3-C firefly luciferase reporter vector (Promega) was used to measure translation of a reporter RNA in vivo. (Top) Structure of the pGLmycUTR construct. (Middle), Firefly luciferase activity produced from control (no UTR) pGL3-C or pGLmycUTR constructs and Renilla luciferase activity were measured in transiently transfected RBM35− and RBM35+ LS180 cells and in HEK293 cells cotransfected with either RBM35A-flag expression construct or empty vector. Data are presented as described in Results and represent mean of four independent experiments ±SD. (Bottom) Immunoblot of whole cell lysates confirms pattern of RBM35A protein expression in the cells used for luciferase assays. p values were calculated using paired one tailed t-test.

Induction of RBM35A protein expression in the LS180 cells results in alterations in the cell proteome

The final outcome of gene expression regulation, which determines the tumorigenic potential of a cell, is the level of proteins for cancer related genes. We used immunoblotting with available antibodies to analyze the effect of ectopic expression of RBM35A on the protein level of several genes involved in cell cycle progression, signal transduction and adhesion. Data on Figures 1D and 6 show a decrease in the steady-state levels of protein products for cyclin D1, MYC and integrin β1 and an increase in E-cadherin and p21^cip, cyclin-dependent kinase inhibitor 1A, in response to ectopic RBM35A expression in LS180 cells. RBM35+ cells also showed a decrease in protein kinase Cα, although this signal transducer has been shown to play a tumor suppressive role in colon tissue.^12,13 No changes were detected for the retinoblastoma protein, focal adhesion kinase, eukaryotic translation initiation factor 2A, eukaryotic translation initiation factor 4E and 4EBP1, a binding partner of cap-binding protein eIF4E (not shown).

Figure 6.

The restoration of RBM35A expression in LS180 cells results in moderate changes in the cell proteome. Immunoblots of whole cell lysates using indicated antibodies show moderate changes in the steady-state levels of some proteins, involved in cell proliferation, cell cycle progression, cell adhesion and signal transduction. Actin or non-specific bands serve as loading control.

Of note, a decrease in c-myc and cyclin D1 proteins was detected only in subconfluent cell cultures, while there was no significant difference in c-myc and cyclin D1 levels between confluent RBM35+ and RBM35− cells (not shown). Also, c-fos protein was not detected in the cells under the cell culture conditions used. Taken together, our findings indicate that ectopic expression of RBM35A in LS180 cells resulted in differential change in the level of expression of multiple proteins and that the resultant cumulative effect of these changes may affect tumorigenicity of the cells.

Discussion

Perturbations in translational regulation have been implicated in carcinogenesis. Regulation of translation in eukaryotes is still poorly understood.¹⁴ The widely accepted ribosome scanning model suggests that proteins which can bind near the 5′ end of the mRNA and thus can block ribosome entry^{15, 16} as well as cis-acting elements in the 5′UTR’s which produce base-paired structures near the 5′end of GC rich mRNAs may be involved in translational regulation. The majority of RNA transcripts of constitutively expressed genes contain short and unstructured 5′UTRs, which allow efficient ribosome scanning to reach the translation initiation codon. A number of cellular mRNAs have longer 5′UTR sequences with high GC content, which can form complex hairpin structures resulting in impeded ribosome scanning due to a higher energy required to unwind secondary structures. These structured “weak” mRNAs, such as those coding for growth regulatory, survival or angiogenesis factors,¹⁷ are translated inefficiently in normal conditions but their translation can be disproportionally increased in cancer.^{17, 18} For example, overexpression of translation initiation factor eIF4E is known to cause transformation by disproportional increase in translation efficiency of “weak” mRNAs, such as MYC and other cancer related genes.^19–21

In this report, we demonstrate that RBM35A protein suppresses malignant potential of LS180 colon cancer cells (Fig. 1). Results of our in vitro and in vivo reporter translation studies implicate RBM35A in translational regulation of gene expression through a 5′UTR of mRNAs. We found that the extent of translational inhibition by RBM35A depends on complexity and strength of secondary structure formed by an mRNA 5′UTR (Fig. 3A). We showed that of all the 5′UTRs analyzed here, RBM35A inhibited the most in vitro translation mediated by 5′UTRs with high value of ΔG, such as c-MYC (Fig. 3A). These findings implicate RBM35A in negative regulation of translation of mRNAs with highly structured GC rich 5′UTRs, among which are known to be many cancer related genes. In conclusion, we suggest that a moderate decrease in the efficiency of translation of many “weak”, malignancy related mRNAs may result in biologically relevant cumulative tumor suppressor effect. This assumption is supported by the observed changes in the cell proteome of the LS180 cells expressing RBM35A (Figs. 1D and 6). We found a decrease in a few oncogenic proteins (c-myc, cyclin D1, integrin βI) and an increase in two tumor suppressors (p21^cip1, E-cadherin) (Figs. 1D and 6). However, we also observed a decrease in the steady-state level of tumor suppressive PKCα (Fig. 6). Even though these findings might seem to be controversial, they do not contradict the idea that it is a resultant cumulative effect of changes in expression of many proteins induced by RBM35A directly or indirectly that results in reduced tumorigenicity of colon cancer cells.

Taken together, our data introduce a novel tumor suppressor protein RBM35A as a negative regulator of translation, whose effect is determined by the structure of an mRNA 5′UTR. The genomic landscapes of breast and colorectal cancers have been recently proposed to be comprised of a handful of commonly mutated gene “mountains” and a much larger number of gene “hills” that are mutated at lower frequency.⁹ Consistent with this view of cancer, is the idea that a large number of mutations, each associated with a small fitness advantage, drive tumor progression. Based on the evidence presented herein, the loss-of-function mutation in RBM35A gene can be added to the list of “hills” in colorectal cancer.

Materials and Methods

Reagents and antibodies

Doxycycline (Dox) was from Sigma; G418 and hygromycin B were from CellGro (Kansas City, MO). Antibodies used in this study were: flag (M2, F1804), actin (A2066) from Sigma, c-Myc (N262), PKCα (H-7, sc-8393) from Santa Cruz Biotechnology (Santa Cruz, CA), cyclin D1 (rm-9104) from LabVision Corp. (Fremont, CA), p21^cip1 (13436E) from BD Pharmingen, integrin βI from Chemicon International (Billerica, MA), E-cadherin (4065) from Cell Signaling Technology (Beverly, MA); horseradish peroxidase-conjugated secondary antibodies goat anti-rabbit and sheep anti-mouse from Chemicon International and Amersham (Piscataway, NJ), respectively.

Cells

LS180 human colon cancer, HEK293 human kidney embryonic and Hela human cervical carcinoma cell lines were purchased from American Type Culture Collection. Cell lines were cultured in DMEM supplemented with 10% FBS and antibiotics at 37°C, 5% CO₂. All cell culture reagents were purchased from CellGro.

Plasmids

Vectors pCMV-Tag4B and pTRE2hyg were obtained from Stratagene (La Jolla, CA) and BD Biosciences Clontech (Palo Alto, CA), respectively. Detailed protocol for cloning of pCMV-RBM35A-flag and pTRE2hyg-RBM35A-flag expression constructs, Luciferase SP6 constructs containing various 5′UTRS and pGLmyc-UTR luciferase reporter plasmid can be obtained from authors upon request. ptTA-IRES-Neo plasmid (tTA, tet activator, IRES, internal ribosome entry site, Neo, G418 resistance gene)²² was a generous gift of Drs. B. Vogelstein and K.W. Kinzler.

Generation of Tet-off LS180-RBM35A cells

LS180 cells were transfected with ptTA-IRES-Neo plasmid. G418-resistant clones were selected in the presence of 1 mg/ml of G-418. LS180-tet-off clones with a tight regulation of tet-responsive pTRE-lux reporter (BD Biosciences) were chosen to generate inducible cell lines. The LS180-tet-off clone 42 was transfected with a pTRE2hyg-RBM35A-flag and cells were selected in the presence of 150 μg/ml hygromycin B and 2 μg/ml Dox.

Colony formation in soft agar assay

To assess the anchorage-independent growth of cells in soft agar, Tet-off LS180-RBM35A cells were cultured in a two-layer agar system. 7.5 × 10³ cells were suspended in 0.3% agarose (DNA grade, Invitrogen, Carlsbad, CA) (top agar) in growth medium and added to a preset 0.5% bottom agar layer in six-well plates. The top agar cell layers were covered with growth medium, either containing Dox (1 μg/ml) or without it. The top medium was changed every four days. Four to five weeks after starting the culture, colonies were stained with 0.005% crystal violet in 25% methanol and colonies >50 μm in diameter were counted microscopically.

Athymic nu/nu mouse assay

All animal studies were in compliance with experimental animal guidelines in the protocol approved by the Institutional Animal Care and Use Committee at Roswell Park Cancer Institute. Exponentially growing Tet-off LS180-RBM35A cells expressing RBM35A (maintained in Dox-free medium for at least four days) and their parental control cells (the same cells, grown in medium with Dox) were harvested and resuspended in DMEM/matrigel (BD Biosciences). Six-week-old female athymic nu/nu nude mice were given s.c. injections of 1 × 10⁶ cells in the right flank. Animals in a control group received Dox (2 mg/ml) in their drinking water starting three days prior to inoculation and throughout the experiment. Three weeks after inoculation the length and width of developed tumors were measured with a caliper, and the mice were sacrificed. Tumor volumes were calculated using the formula for solitary ellipsoid V = (length × width²)/2.

Protein immunoblots

Cells were lysed and analyzed by Western blotting as described previously.²³ Briefly, cells were harvested in boiling SDS lysis buffer (10 mM Tris-HCl, pH 7.4, 1% SDS), boiled for 5 min, and sheared by 5–10 passes through a 27 g needle. Extracts were cleared by centrifugation (10 min, 10,000 g, RT). Protein concentrations were determined by Bicinchoninic acid (BCA) protein assay reagent (Pierce; Rockford, IL), and lysates were boiled in Laemmli sample buffer for 5 min before being subjected to SDS-PAGE. Equal amounts of proteins were separated using 8% or 10% SDS-polyacrylamide minigels. Proteins were transferred to the nitrocellulose membrane. Blots were routinely stained with 0.1% Fast Green (Sigma) to confirm equal loading and even transfer. Membranes were incubated overnight with primary antibodies followed by incubation with respective horseradish peroxidase -conjugated secondary antibodies. Immune complexes were visualized using the Super Signal chemiluminescence system (Pierce) and CL-XPosure film (Pierce).

Immunofluorescence analysis

Hela cells, transiently transfected with pCMV-RBM35A-flag plasmid, were grown on coverslips, fixed in methanol/acetone mixture and following washes in PBS, were air dried. After being blocked in 5% donkey serum/PBS the cells were incubated in rabbit polyclonal anti-flag antibody (Sigma) followed by incubation in Rhodamine-conjugated donkey anti-rabbit secondary antibody (Jackson ImmunoResearch Labs). Negative controls (no primary antibody) were included in all experiments. Coverslips were viewed with a Zeiss Axioskop epifluorescence microscope, using a 63× Apochromat (1.4 NA) objective lens. The staining patterns observed with polyclonal anti-flag antibody were confirmed by direct fluorescence analysis using mouse monoclonal anti-flag M2-Cy3-conjugated antibody (Sigma). Images were obtained with a Hamamatsu C7780 digital camera and processed for presentation using Adobe Photoshop CS.

Isolation of polysomal RNA

Tet-off LS180-RBM35A cells with restored expression of RBM35A protein (incubated without Dox for at least six days) and their parental cells maintained in the presence of Dox were grown to no more than 80% confluency to ensure good loading of polysomes. Cytosolic lysates were prepared and subjected to centrifugation in 10–40% continuous sucrose gradient. An absorbance profile at 254 nm was used to identify polysome-associated fractions. RNA was isolated from pooled polysome-associated fractions. Detailed protocol for entire procedure can be obtained from authors upon request.

Affymetrix genechip hybridization

Microarray analysis was performed in accordance with MIAMI guidance at the Gene Expression facility of Roswell Park Cancer Institute. Data were submitted to GEO database, accession # GSE13171.

In vitro translation and transcription of luciferase reporter constructs

Luciferase reporter RNA was translated in vitro from various luciferase SP6 constructs using TnT SP6 Coupled Reticulocyte Lysate System (Promega, Madison, WI) according to manufacturer’s instructions. To assess the effect of RBM35A on translation, 2 μl of either in vitro translated RBM35A or negative control reaction were added to the 20 μl Luciferase translation reactions. Luciferase activity was measured using Steady-Glo Luciferase Assay System (Promega). In vitro transcription was performed using RiboMax Large Scale RNA Prodution System-SP6 (Promega) according to manufacturer’s instructions. In vitro transcribed transcripts were then used for translation in Rabbit Reticulocyte Lysate translation mix.

Real-time RT-PCR analysis

Quantitative RT-PCR analysis of polysome-associated c-myc and c-fos mRNA levels was performed using High Capacity cDNA Reverse Transcription kit, TaqMan Gene Expression Master Mix and TaqMan Gene expression Assays for c-myc (Hs00153408), c-fos (Hs00170630) and GAPDH (402869) (Applied Biosystems, Foster City, CA). Reactions were carried out on the ABI 7900 HT Taqman Sequence Detection System. Each sample was run in quadruplicate. C_T values for the target amplicon and endogenous control (GAPDH) were determined for each sample. Quantification was performed using the comparative C_T method (ΔΔC_T).

Transient transfection and luciferase assays

Cells were transfected with luciferase reporter plasmids using Effectene reagent (Qiagen, Chatsworth, CA). Firefly and Renilla luciferase activities were measured using a Dual-Luciferase reporter assay system (Promega) according to manufacturer’s guidelines. Values for firefly luciferase activity were normalized to values for Renilla luciferase.

Abbreviations

DMEM

Dulbecco’s modified eagle’s essential minimal medium

Dox

doxycycline

FBS

fetal bovine serum

MSI

microsatellite instability

MTS

3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium

mut

mutant

PBS

phosphate-buffered saline

RBD

RNA-binding domain

RBM35A

RNA-binding motif protein 35A

RBP

RNA-binding protein

RRM

RNA recognition motif

UTR

untranslated region

wt

wild type