RIN1 Is a Breast Tumor Suppressor Gene

Breast cancer progression is driven by altered gene expression. We show that the RIN1 gene, which encodes a RAS effector regulating epithelial cell properties, is silenced in breast tumor cell lines compared with cultured human mammary epithelial cells. We also report that RIN1 is often reduced in human breast tumor cells compared with morphologically normal breast glandular cells. At least two silencing mechanisms seem to be involved. Overexpression of the transcription repressor SNAI1 (Snail) was observed in ZR75-1 cells, and SNAI1 knockdown restored RIN1 expression. In addition, DNA methylation within the RIN1 promoter and the first exon in KPL-1 cells suggested that epigenetic modifications may contribute to silencing, and demethylation was shown to restore RIN1 expression. Reexpression of RIN1 was shown to inhibit anchorage-independent growth in soft agar. In addition, RIN1 expression inhibited both the initiation and progression of tumorigenesis for two breast tumor cell lines in a mouse model, consistent with a tumor suppressor function. We also show that RIN1 acts as a negative regulator of tumor cell invasive growth and that this requires the ABL kinase–signaling function of RIN1, suggesting a mechanism through which RIN1 silencing may contribute to breast cancer progression. [Cancer Res 2007;67(24):11510–6]

Among the most common loss-of-function events reported for breast cancers are mutations in TP53 and PTEN (1), with less frequent mutations reported in several other tumor suppressor genes (reviewed in 2). In addition to these inactivation mutations, there seems to be a common role for transcriptional silencing of tumor suppressor genes during breast tumor progression and metastasis. Several mechanisms of gene silencing during tumorigenesis have been described. Covalent modification of DNA (e.g., CpG methylation) and/or histones (e.g., methylation and deacetylation) can greatly reduce the rate of transcription initiation for some target genes. Target genes may also be silenced by a reduction in transcription activator proteins or by overexpression of transcription repressor proteins. CDH1 (E-cadherin), which encodes a component of adherens junctions and promotes epithelial over mesenchymal functions during normal development and transformation (reviewed in 3), is subject to mutations as well as silencing by epigenetic mechanisms during tumor progression (4). In addition, transcriptional repressors such as SNAI1 (Snail), which silence CDH1 and other proepithelial genes, are induced during developmental epithelial to mesenchymal transition and overexpressed in some metastatic tumor cells (5 and reviewed in 6). These observations suggest that the regulated expression of proepithelial cell genes, which evolved to orchestrate developmental transitions, can be exploited by tumor cells during expansion and metastasis.

Among the most common events in human tumors are activating mutations in RAS genes (HRAS, KRAS, and NRAS). And although only a small fraction of breast tumors have such mutations, these tumors often have elevated RAS signaling due to overexpression of upstream receptor tyrosine kinases such as ERBB2, epidermal growth factor receptor, MET, and RON (7–10). Signaling pathways downstream of RAS are mediated by at least a half dozen direct RAS effectors (reviewed in 11). Among these, BRAF and PIK3CA (PI3K), which stimulate mitosis and block apoptosis, respectively, are mutationally activated in a variety of tumors and tumor cells including some breast cancer cell lines (reviewed in 12; Cancer Genome Project). Given the diversity of signals from RAS, and the established role of RAS in development, the contribution of tumor suppressor pathways downstream of RAS also needs to be considered.

The RAS effector RIN1 activates ABL tyrosine kinases and RAB5 GTPases to regulate cytoskeletal remodeling and endocytic pathways that promote normal epithelial functions (13). Silencing of RIN1 leads to increased motility of epithelial cells.⁶

Here we report that RIN1 expression is silenced in a large proportion of breast tumor cell lines as well as in tumor tissue samples. Further, we identify multiple mechanisms for RIN1 silencing. We directly confirm the breast tumor suppressor function of RIN1 and provide evidence supporting a role for RIN1 in restraining invasive growth of epithelial cells.

Cell lines and culture conditions. The breast cancer cell lines MDA-MB-231, BT-549, KPL-1, ZR75-1, T47D, Hs578t, BT20, and BT549 were grown in DMEM with 10% fetal bovine serum. MCF10A was cultured in DMEM/F12 with hEGF (20 ng/mL), hydrocortisone (500 ng/mL), insulin (10 μg/mL), cholera toxin (100 ng/mL), and 5% equine serum. Normal human mammary epithelial cells (HMEC) were obtained from Clonetics and were cultured in mammary epithelial growth medium (Clonetics). Lentivirus stock production and viral transduction of cells were performed as previously described (13). Blasticidin (Invitrogen) was used at 20 μg/mL for selection of transduced cells. Zebularine (obtained from the National Cancer Institute) was used at 150 ng/mL.

For soft agar growth assays, 1 × 10⁴ breast tumor cells were seeded into a 10-cm culture dish containing 0.35% low-melting agarose over a 0.7% agarose layer, both in culture medium, and incubated for 3 to 5 weeks at 37°C. Colonies were then stained with 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (250 μg/mL) and counted.

Invasive growth assays were performed using modified Boyden chambers coated with Matrigel (BD Biosciences). The upper chamber was seeded with 2 × 10⁵ cells and these were allowed to migrate at 37°C for 18 h toward a lower reservoir of DMEM with hepatocyte growth factor (10 ng/mL) before fixing (2% paraformaldehyde for 20 min), staining (crystal violet for 1 h), and counting cells that had passed through the membrane separating the chambers.

RNA preparations and real-time PCR. Total RNA was extracted from breast tumor and normal breast cell lines using TRIzol reagent (Invitrogen Life Technologies) following the manufacturer's protocol. Isolated RNA was then used to synthesize cDNA using an iScript cDNA synthesis kit (Bio-Rad). Real-time PCR was performed using the iCycler PCR platform (Bio-Rad). Thermal cycling conditions were as follows: an initial incubation at 95°C for 10 min followed by 40 cycles of 95°C for 30 s, 55°C for 1 min, and 72° C for 30 s. Followed by a final cycle of 95°C for 1 min, 55°C for 30 s, and 95°C for 30 s. IQ SYBR Green Supermix (Bio-Rad) was used in accordance with the manufacturer's instructions. The primers used were as follows: RIN1 5′-GGCAGCAGAGGAGTAGCTTGA and 5′-GCTTGCTGGCGCTAAAAGG, SNAI1 5′-GAAAGGCCTTCAACTGCAAA and 5′-TGACATCTGAGTGGGTCTGG, SNAI2 5′-ATGAGGAATCTGGCTGCTGT-3′ and 5′-CAGGAGAAAATGCCTTTGGA-3′, and β-actin (ACTB) 5′-CATTGCCGACAGGATGCA and 5′-CGCTCAGGAGGAGCAATGAT.

For tissue-based expression analysis, primary tumor epithelial cells and matching normal epithelial cells were isolated by laser capture microdissection (LCM) from sectioned breast tissue obtained from the University of California at Los Angeles Tissue Procurement Core Laboratory. Mammary epithelial cells were isolated onto LCM transfer film (Arcturus), as described (14) using an LCM microscope (Pixcell IIe; Arcturus). RNA was then purified using the PicoPure RNA isolation kit (Arcturus).

Immunoprecipitations, immunoblotting, and immunohistochemistry. For semiquantitative analysis, RIN1 was first immunoprecipitated from breast tumor cell line lysates using polyclonal anti-RIN1 (BD Biosciences) and immunoblotted using monoclonal anti-RIN1 (BD Biosciences). Immunohistochemistry was performed using established protocols (15) with monoclonal anti-RIN1 (1:250 dilution) or polyclonal rabbit anti-human RIN1 (1:400 dilution). Monoclonal anti-CDH1 (E-cadherin; BD Biosciences) antibodies were used at 1:400 dilution. Nonimmune antibody (mouse IgG1) was used as a negative control. Biotinylated Universal Antibody (Vectastain) was used as the secondary stain. Tissue samples were counterstained with H&E.

Viral constructs, gene silencing, and methylation analysis. The self-inactivating lentivirus vector M4 has been previously described (13). M4-blast is a modified version of this vector that includes an SV40 promoter–driven Blasticidin resistance cassette derived from pcDNA6 (Invitrogen Life Technologies). The RIN1-directed shRNA lentivirus construct M4-shRIN1-blast was created by first converting the M4 unique BamHI site to a BstBI site and then removing the cytomegalovirus promoter between the flanking BstBI sites. Oligonucleotide primers, one encoding a RIN1 shRNA, were used to PCR amplify U6 promoter sequence, and the resulting BamHI-EcoRI fragment was cloned into pKS (Stratagene) then moved as an XbaI-EcoRI fragment to M4 cut with NheI and EcoRI. A Blasticidin resistance gene cassette from pcDNA6 (Invitrogen) was then inserted into the EcoRI site of the modified M4 vector. The upstream U6 promoter primer (5′–3′) was GCTGGGATCCCAAGGTCGGGCAGGAAGAGGG-3′ (BamHI). The RIN1-directed shRNA primers (5′–3′) were #753 [GCTTCGGCAGCACATATACTACTGGAGGTCTCTGTGGACACGCGCACTTTTTGGAAGGTGTGCGTGTCCACAGAGATCTCCAGTTTTTGAATTCTG (EcoRI)] and #2264 [GCTTCGGCAGCACATATACTATTCAGCAGTTGTCTCAGACTGTTCCTTGGGGAATAGTCTGAGACAGCTGCTGAATTTTTGAATTCTG (EcoRI)].

SNAI1-directed siRNA (cat#AM16708; Ambion) and scrambled sequence control siRNA (cat#4615; Ambion) were transfected into ZR75-1 cells using siPORT (Ambion).

Methylation sites were identified by sequence analysis of bisulfite-treated genomic DNA and was carried out by Seqwright using the primers: 5′-TGTAAAACGACGGCCAGTTGTGTGGAGGTGGTATTTTTTA-3′ and 5′-CAGGAAACAGCTATGACCCATCCCCAATAAATAACACTTC-3′.

In vivo tumor formation. In vivo tumor assays were carried out using an established protocol (16) with the following modifications. Tumor cells were washed twice with PBS and then incubated in serum-free DMEM for 2 h at 37°C. Cells were subsequently incubated with versene: trypsin (4:1 mixture) for 5 min. Cells were then collected using serum-free DMEM and soybean trypsin inhibitor (Sigma-Aldrich).

Tumor cells (2 × 10⁶ cells transduced with M4-blast or M4-RIN1-blast) were injected into the mammary fat pads of 4- to 6-week-old female nude mice (Charles River Laboratories). Tumor growth was monitored at 3- to 4-day intervals. Upon completion of the assay, tumors were removed and sectioned. Sectioned tumor tissue was immunostained for RIN1 protein using polyclonal anti-RIN1 as described above.

RIN1 is silenced in breast tumor cells. Based on its expression in mammary epithelial cells (13) and its function as a RAS effector, we examined whether RIN1 expression is altered in breast tumors. We initially examined five breast tumor cell lines, all derived from invasive ductal carcinomas, and observed a significant reduction in RIN1 protein levels compared with normal HMECs (Fig. 1A).

In normal mammary tissue, RIN1 protein is enriched in ductal epithelial cells with localization to the cytoplasm and plasma membrane (Fig. 1B). Consistent with the tumor cell line analysis, we observed reduced levels of RIN1 in primary breast tumor cells, although the degree and extent of silencing was not uniform (Fig. 1B). The same technique applied to cell lines confirmed staining specificity (Supplementary Fig. S1A). These findings indicate that reduced RIN1 protein levels occur in breast tumor tissue as well as cell lines derived from such tumors. In addition, lymph node metastatic tumors had RIN1 levels that were low and generally weaker than primary tumors from the same patients (Fig. 1C).

We next examined whether lower RIN1 protein levels in tumor cells reflected a decrease in RIN1 mRNA. RIN1 transcript levels from nine breast tumor cell lines, and normal HMECs were quantified by real-time PCR. The normalized results indicated consistent silencing of RIN1 message in these established breast tumor cell lines (Fig. 2A).

We also tested whether RIN1 message levels were reduced in primary human breast tumor samples compared with glandular epithelial cells from normal human breast tissue. Mammary epithelial cells collected using LCM from tumor and surrounding normal tissue were analyzed by real-time PCR. All six of the patient samples that were tested had reduced RIN1 transcript levels in tumor compared with surrounding normal mammary epithelial tissue (Fig. 2B), indicating that RIN1 silencing occurs at a high frequency in human breast malignancies. Immunohistochemical analysis of the same patient samples showed RIN1 protein levels commensurate with the observed reduction in transcripts (data not shown).

RIN1 silencing mechanisms in breast tumor cells. The human RIN1 promoter includes 20 binding sites for SNAI1, a transcriptional repressor of epithelial genes such as E-cadherin (17), and three of these SNAI1 sites are well conserved in mammals (Supplementary Fig. S3). In addition, overexpression of SNAI1 is seen in multiple types of epithelial-derived tumors (5) and correlates with breast tumor recurrence (18). We therefore examined breast tumor cell lines with low RIN1 levels and found that ZR75-1 cells have nearly 4-fold higher levels of SNAI1 than the normal mammary epithelial cell line MCF10A (Fig. 3A). The closely related transcription repressor SNAI2 (Slug) showed unchanged or reduced expression in the same set of tumor lines (Fig. 3B). To determine whether elevated levels of SNAI1 contribute to the silencing of RIN1 in this cell line, we used a SNAI1-targeted siRNA to reduce levels of SNAI1 in ZR75-1 cells. This treatment reduced SNAI1 mRNA levels and at the same time increased the mRNA levels of RIN1 (Fig. 3C), consistent with SNAI1 acting as a repressor of RIN1 expression.

We next considered whether RIN1 silencing might sometimes occur through DNA methylation, a common gene-silencing mechanism in tumor cells. DNA methylation typically occurs at CpG dinucleotides within the promoter and first exon sequence of a gene, and subsequently leads to the recruitment of chromatin remodeling complexes (19). We used methylation-specific sequence analysis to check for cytosine methylation in the RIN1 promoter region. Several methylated cytosines were found in the promoter and first exon sequences of the RIN1 gene in KPL-1 cells (Fig. 3D). However, no methylation was detected in the same region of the mammary epithelial cell line MCF10A or the breast tumor lines MDA-MB-231, ZR75-1, and BT549. To determine if RIN1 promoter methylation correlated with reduced expression, we next tested for reactivation of RIN1 expression after treatment with the nonnucleoside demethylating agent Zebularine (20). Treatment over the course of three cell doublings increased RIN1 expression in KPL-1 cells (Fig. 3E), suggesting that DNA methylation was involved in silencing of RIN1 in these cells. In contrast, Zebularine treatment of MCF10A cells resulted in a slight reduction of RIN1 expression [MCF10A cells express RIN1 at levels comparable with immortalized HMECs (Supplementary Fig. S1B)]. These results indicate that the recovery of RIN1 expression in KPL-1 cells was not a generalized nonspecific activating effect of Zebularine. The region of the RIN1 promoter methylated in KPL-1 cells does not conform to the standard criteria for a “CpG island”, the name given to sequences with a high density of CpG dinucleotides that are frequently targeted for DNA methylation.⁷

This raises the possibility that targeting factors may be involved in directing methylation to the RIN1 promoter.

Restoration of RIN1 expression inhibits tumor phenotypes. To directly evaluate the tumor suppressor properties of RIN1, we engineered ectopic expression through stable transduction of the breast tumor lines KPL-1 and MDA-MB-231 with a RIN1 expression vector, and expression levels were assessed by immunoblot (Supplementary Fig. S3A). Expression of RIN1 suppressed the growth of KPL-1 cells in soft agar (Fig. 4A), a measure of anchorage independence and a common phenotype of tumor cells. In the well-studied and aggressively metastatic tumor cell line MDA-MB-231, restored RIN1 expression did not noticeably affect the number or size of colonies formed (data not shown). One possible explanation for this is that additional alterations in this cell line, including the KRAS^G13D allele, have rendered it less sensitive to the suppressing effects of RIN1 in this assay. In addition, levels of ectopic RIN1 expression were not as high in MDA-MB-231 cells as they were in KPL-1 cells.

Restoration of RIN1 expression also inhibited the ability of KPL-1 cells to form tumors after injection into the mammary fat pads of nude mice. Compared with cells transduced with empty vector, KPL-1 cells transduced with RIN1 showed approximately a 10-day delayed onset of detectable tumors (Fig. 4B). In addition, cells transduced with RIN1 showed reduced kinetics of tumor growth (15 ± 3 mm³/d) compared with vector-transduced cells (42 ± 5 mm³/d). To eliminate the possibility that the tumors that eventually arose from RIN1-transduced cells might have resulted from silencing of ectopic RIN1 expression, we stained tissue from tumors collected at the termination of the assay. We found that elevated RIN1 expression was retained in the tumors that arose from RIN1-transduced breast tumor cells (Supplementary Fig. S3B), consistent with a role for RIN1 in suppressing, but not completely blocking, tumor formation.

We next examined whether ectopic RIN1 levels would reduce tumor formation by MDA-MB-231 cells. These cells retain more endogenous RIN1 than KPL-1 cells but less than normal mammary epithelial cells (Fig. 1A and Supplementary Fig. S4A). RIN1 expression caused a substantial delay in the formation of palpable tumors from these cells as well, compared with vector-transduced cells (Fig. 4C). Again, a marked reduction in tumor growth kinetics (23 ± 4 mm³/d), compared with control cells (30 ± 2 mm³/d), was observed. These results from breast tumor cell lines that differ in their degree of silencing suggest that RIN1 suppresses an early step in tumor formation as well as a function required for tumor progression.

RIN1 is a negative regulator of tumor cell invasive growth. Previous analysis has shown that RIN1 is an inhibitor of mammary epithelial cell migration (13),⁶ raising the possibility that silencing of RIN1 might promote the invasive growth associated with tumor spread. Indeed, MDA-MB-231 cells stably transduced with RIN1 had a reduced capacity for invasive growth through a Matrigel substrate, compared with vector control MDA-MB-231 cells (Fig. 5A).

Because these cells retain a significant level of RIN1 expression (Figs. 1A and 2A), we asked whether a further reduction of RIN1 might lead to an additional increase in cell invasion potential. Using shRNA to stably silence RIN1 expression in MDA-MB-231 cells, we observed enhanced invasive growth (Fig. 5B) compared with control cells. This is consistent with a role for RIN1 as an invasive growth inhibitor in mammary epithelial cells.

RIN1 is a binding partner and activator of ABL tyrosine kinases (13), which in turn regulate actin remodeling and cell motility. We tested the contribution of this signaling pathway to invasive growth blockade using a RIN1 mutant. RIN1^QM carries tyrosine to phenylalanine substitutions at positions 36, 121, 148, and 295. This mutant shows reduced levels of tyrosine phosphorylation and ABL binding,⁶ severely compromising its ABL activation function (Supplementary Fig. S4C). RIN1^QM was unable to block cell invasion (Fig. 5C), strongly implicating ABL stimulation as a required pathway for RIN1-mediated tumor suppression.

We have shown that the RAS effector RIN1 has the properties of a breast tumor suppressor. As with a growing number of tumor suppressor genes, RIN1 seems to be transcriptionally silenced, rather than deleted, during tumor progression. We identify two potential mechanisms causing reduced expression. These include DNA methylation of the RIN1 promoter and transcriptional repression through elevated levels of SNAI1, an established repressor of genes defining epithelial cell properties (21). Consistent with this observation, RIN1 expression was significantly reduced in mammary epithelial cells after treatment with transforming growth factor β (TGFβ),⁶ a factor that confers mesenchymal phenotypes and promotes tumor metastasis in part through SNAI1.

In addition to transcription repressor–based silencing, DNA methylation of the RIN1 promoter was observed in a breast tumor cell line, suggesting that alterations in chromatin structure may contribute to silencing. Other breast tumor cell lines with low RIN1 expression showed no indication of either mechanism, implying that additional means, such as miRNAs, might be used to silence this locus in a reversible manner. The involvement of alternate transcription-silencing mechanisms, including SNAI1 overexpression and promoter methylation, has also been reported for the breast tumor suppressor CDH1 (5, 22). These observations highlight the limitation of therapeutic approaches that target only specific silencing mechanisms.

The most intensely studied RAS effectors (RAF and PI3K) enhance transformation and are themselves mutationally activated in many tumors (reviewed in 12). There is, however, precedence for a RAS effector with tumor suppressor properties. RASSF1 is frequently silenced in tumors and shows the hallmarks of a tumor suppressor (23). Other members of the RASSF family, which share a RAS association domain, are potential tumor suppressors as well (24–26). RASSF proteins seem to enhance apoptosis (27), and silencing of RASSF genes likely promotes tumor cell survival.

RIN1 functions through two downstream pathways involved in maintenance of epithelial properties. By activating ABL tyrosine kinases, RIN1 blocks the cytoskeletal rearrangements associated with cell dissociation and migration (13). We present evidence that signaling through ABL tyrosine kinases is required for RIN1-mediated blockade of invasive growth. This is consistent with a proposed inhibitory role for ABL in breast cancer cell tumorigenicity (28), although other studies (29, 30) suggest a more complex role for ABL signaling in breast cancer. RIN1 also signals downstream through RAB5 proteins to promote endocytosis (31, 32). This down-regulates growth factor receptors required for directed migration and enhances TGFβ signaling.⁶ RAB5 signaling might also contribute to the initiation and/or progression of primary tumors, but there is as yet no direct evidence for this.

E-cadherin is another protein that contributes to epithelial character in normal cells, although its loss facilitates cell motility and is characteristic of tumor cells (3). Like RIN1, CDH1 is silenced by promoter hypermethylation as well as SNAI1-mediated repression in breast tumors (33). Also like RIN1, loss of CDH1 and other epithelial markers does not strictly correlate with tumor grade or metastatic potential in vivo (34). These observations lend support to a model of tumor progression in which gene expression is dynamically regulated. In early stages, matrix invasion and intravasation of primary tumor cells through endothelial cell barriers should select for the silencing of genes that inhibit mesenchymal phenotypes or are associated with tumor metastasis. Such genes might include CDH1 (35), BRMS1 (36), NDRG1 (Drg-1 and CAP43; ref. 37), NME1 (nm23-H1; ref. 38), CD82 (KAI1; ref. 39), SERPINB5 (maspin; ref. 40), MKK4 (41), RRM1 (42), PEBP1 (RKIP; ref. 43), and RIN1 (this work). Resistance to silencing may partly explain why a relatively moderate increase in RIN1 expression from a viral promoter had such a strong tumor suppressor effect in MDA-MB-231 cells. Subsequently, establishment of secondary tumors may remove, or even reverse, this selective pressure for silencing and lead to restored expression of invasion suppressor genes. Finally, it should be considered that culture conditions may exert their own selective pressure for tumor cell lines to display mesenchymal properties, making them more representative of particular tumor stages.

The loss of epithelial cell characteristics together with a gain of mesenchymal phenotypes, as seen in many cancer cells, has prompted a comparison between tumor progression and the cell transitions that occur during development. Indeed, during organ formation, epithelial cells seem to have a natural plasticity that allows them to adopt mesenchymal characteristics as needed. A striking example is the multiple waves of SNAI1 gene expression that occur during vertebrate morphogenesis (44). And, just as epithelial cells adopt mesenchymal phenotypes only temporarily during development, the metastatic spread of tumor cells may also require forward and reverse transitions between epithelial and mesenchymal characteristics.

It is worth noting that the RIN1 gene is located <1 kb downstream of, and in the same orientation as, the breast cancer metastasis suppressor gene BRMS1 (chr 11: 65,856,118-65,869,158). The relative position of these genes is highly conserved in mammals. A reduction in BRMS1 expression is associated with the metastatic potential of breast tumor cells, and restored expression blocks metastasis (36). The BRMS1 gene product functions as a transcriptional corepressor (45, 46) and normally enhances apoptosis of nonadherent cells. This raises the intriguing possibility that the RIN1 and BRMS1 gene products work in concert to promote stable mammary epithelial cell structures. Combined repression of both the RIN1 and BRMS1 genes may collaboratively facilitate cell migration during development, as well as the spread of tumor cells during metastasis. We are currently exploring this hypothesis.

Grant support: NIH grant NS046787, Department of Defense grant W81XWH-04-1-0443, and California Breast Cancer Research Project fellowship 11GB-0038 (M. Milstein); and Early Detection Research Network National Cancer Institute grant CA-86366 (L. Goodglick).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank Apoor Patel, Delia Adefuin, Christopher Creencia, Lily Zhang, Minh Thai, and Samantha Olson for technical assistance, and Chad Barber and Bob Schultz for technical advice and reagents.