Cooperative Interaction between the MUC1-C Oncoprotein and the Rab31 GTPase in Estrogen Receptor-Positive Breast Cancer Cells

Rab31 is a member of the Ras superfamily of small GTPases that has been linked to poor outcomes in patients with breast cancer. The MUC1-C oncoprotein is aberrantly overexpressed in most human breast cancers and also confers a poor prognosis. The present results demonstrate that MUC1-C induces Rab31 expression in estrogen receptor positive (ER+) breast cancer cells. We show that MUC1-C forms a complex with estrogen receptor α (ERα) on the Rab31 promoter and activates Rab31 gene transcription in an estrogen-dependent manner. In turn, Rab31 contributes to the upregulation of MUC1-C abundance in breast cancer cells by attenuating degradation of MUC1-C in lysosomes. Expression of an inactive Rab31(S20N) mutant in nonmalignant breast epithelial cells confirmed that Rab31 regulates MUC1-C expression. The functional significance of the MUC1-C/Rab31 interaction is supported by the demonstration that Rab31 confers the formation of mammospheres by a MUC1-C-dependent mechanism. Analysis of microarray databases further showed that (i) Rab31 is expressed at higher levels in breast cancers as compared to that in normal breast tissues, (ii) MUC1+ and ER+ breast cancers have increased levels of Rab31 expression, and (iii) patients with Rab31-positive breast tumors have a significantly decreased ten-year overall survival as compared to those with Rab31-negative tumors. These findings indicate that MUC1-C and Rab31 function in an autoinductive loop that contributes to overexpression of MUC1-C in breast cancer cells.


Introduction
Rab proteins are members of the Ras superfamily of small GTPases [1]. The Rab GTPases function in receptor internalization, recycling and signaling [2]. Rab31, a Rab5 subfamily GTPase, is involved in the trafficking of early endosomes [3]. Rab31 blocks insulin-stimulated translocation of the Glut4 glucose transporter from endosomes to the cell membrane [4]. In addition, Rab31 is required for transport of mannose 6-phosphate receptors from the trans-Golgi network to endosomes [5]. Interestingly, high levels of Rab31 expression in tumors from 280 node-negative breast cancer patients were shown to be significantly associated with decreased metastasis-free and overall survival [6]. However, no insights are available regarding a potential functional role for Rab31 in breast cancer.
Mucin 1 (MUC1) is a heterodimeric transmembrane glycoprotein that is overexpressed in about 90% of human breast cancers [7,8,9]. MUC1 consists of two subunits that form a complex at the cell membrane after translation and autocleavage of a single polypeptide [7,9]. The MUC1 N-terminal subunit (MUC1-N) contains glycosylated tandem repeats that are a characteristic of mucin family members. The MUC1 C-terminal subunit (MUC1-C) spans the cell membrane and contains a cytoplasmic domain that interacts with diverse effectors, such as the epidermal growth factor receptor (EGFR) and ErbB2, that have been linked to transformation [7,9]. Overexpression of MUC1-C, as found in human breast cancers, is sufficient to induce anchorage-independent growth and tumorigenicity [10,11]. The overexpression of MUC1 in transgenic mouse models is also associated with the induction of breast tumors [12,13]. MUC1-C is internalized from the cell membrane by clathrin-mediated endocytosis [14] and is then recycled from endosomes back to the cell membrane [15]. The overexpression of MUC1 in breast cancer cells is also associated with accumulation of MUC1-C in the cytoplasm by a mechanism that likely involves retrograde trafficking during endocytosis and movement to the endoplasmic reticulum [7,9]. Cytosolic MUC1-C interacts with importin-b and is transported to the nucleus [16], where it interacts with estrogen receptor a (ERa) and promotes ERa-mediated gene transcription [17].
The present studies demonstrate that MUC1-C induces Rab31 expression in breast cancer cells. MUC1-C forms a complex with ERa on the Rab31 promoter and activates Rab31 gene transcription. In turn, Rab31 contributes to the upregulation of MUC1-C levels in an autoinductive loop. These findings are further supported by the demonstration that Rab31 expression is increased in primary breast cancers that are positive for both ERa and MUC1.

MUC1 Upregulates Rab31 Expression
Gene microarray analysis of MCF-7 breast cancer cells demonstrated that silencing of MUC1 is associated with decreases in Rab31 expression. To confirm this observation, MCF-7 cells stably expressing an empty vector (MCF7/vector) or MUC1-siRNA (MCF-7/MUC1siRNA) were analyzed for Rab31 mRNA levels by RT-PCR. Downregulation of MUC1-C mRNA levels was associated with a decrease in Rab31 transcripts (Fig. 1A, left). Quantitative RT-PCR confirmed that Rab31 transcripts are significantly lower in MCF-7/MUC1siRNA cells as compared to that in MCF-7/vector cells (Fig. 1A, right). Similar results were obtained when mRNA from ZR-75-1 breast cancer cells was analyzed by RT-PCR (Fig. 1B, left) and qRT-PCR (Fig. 1B, right). In concert with these results, MUC1-C silencing was associated with decreases in the abundance of Rab31 protein in MCF-7 ( Fig. 1C) and ZR-75-1 (Fig. 1D) cells. These findings indicate that MUC1-C functions in upregulating Rab31 expression.

Rab31 Promoter is Activated by a MUC1-dependent Mechanism
To assess the effects of MUC1 on Rab31 gene transcription, we cloned a 1874 bp region upstream to the Rab31 transcription start site and inserted those sequences in a luciferase expression vector (pRab31-Luc). Analysis of the Rab31 promoter region incorporated into pRab31-Luc identified putative consensus binding sites for diverse transcription factors, including multiple potential ERa responsive elements (EREs) ( Fig. 2A). Transfection of MCF-7/ vector and MCF-7/MUC1siRNA cells with pRab31-Luc demonstrated that silencing MUC1 results in repression of the Rab31 promoter (Fig. 2B). Similar results were obtained in studies of ZR-75-1 cells, indicating that MUC1-C confers activation of Rab31 transcription (Fig. 2C). Previous work showed that MUC1-C associates with ERa and promotes activation of ERa target genes [17]. To assess dependence on ERa, we transfected MCF-7 cells with control or ERa siRNA pools to transiently decrease ERa levels ( Fig. 2D, left). Silencing of ERa was associated with a decrease in pRab31-Luc activation, consistent with ERa-mediated induction of the Rab31 promoter (Fig. 2D, right). To provide further support for Rab31 gene activation by estrogen signaling, ZR-75-1 cells were transfected with pRab31-Luc, cultured in estrogen-depleted medium, and then stimulated with estradiol (E2). Compared to unstimulated cells, E2 treatment was associated with a 3-fold increase in pRab31-Luc activation (Fig. 2E). These findings indicate that Rab31 promoter is activated by a mechanism dependent on ERa and MUC1-C.

E2 Induces Rab31 Expression by a MUC1-C-dependent Mechanism
In further support for involvement of ERa in Rab31 expression, E2 treatment of MCF-7 cells was associated with a .3-fold increase in Rab31 mRNA levels at 6 h and then a modest decline by 24 h (Fig. 3A, left). Stimulation of ZR-75-1 cells with E2 was also associated with a transient increase in Rab31 mRNA levels at 6 h (Fig. 3A, right). To assess dependence on MUC1-C, E2 treatment studies were performed on MCF-7/vector and MCF-7/ MUC1siRNA cells. Stimulation of MCF-7/vector cells with E2 was associated with increases in Rab31 mRNA levels (Fig. 3B, left).
By contrast, E2 stimulation had no apparent effect on Rab31 expression in MCF-7/MUC1siRNA cells (Fig. 3B, right). Studies performed with ZR-75-1/vector and ZR-75-1/MUC1siRNA cells further demonstrated that MUC1-C is necessary for E2-induced increases in Rab31 transcripts (Fig. 3C, left and right). Consistent with these results, E2 stimulation of MCF-7 and ZR-75-1 cells was associated with increases in Rab31 abundance (Fig. 3D, left and right). Moreover, this response to E2 was attenuated by silencing MUC1-C (Fig. 3E). These findings indicate that E2 stimulates Rab31 expression at the mRNA and protein levels by a MUC1-Cdependent mechanism.

MUC1-C Associates with ERa on the Rab31 Gene Promoter
Coimmunoprecipitation studies performed on lysates from MCF-7 cells demonstrated that MUC1-C associates with ERa and that this interaction is increased in the response to E2 stimulation (Fig. 4A). Analysis of control and E2-stimulated ZR-75-1 cells demonstrated similar results (Fig. 4B). Chromatin immunoprecipitation (ChIP) studies were then performed on control and E2-stimulated MCF-7 cells to assess the presence of ERa and MUC1-C on the Rab31 promoter (Fig. 4C). Stimulation with E2 was associated with a ,5-fold increase in ERa occupancy compared to that obtained with a control IgG (Fig. 4D, left). Re-ChIP studies further demonstrated that MUC1-C is detectable with ERa on the Rab31 promoter (Fig. 4D, right). Analysis of ZR-75-1 cells similarly showed that ERa occupancy of the Rab31 promoter is increased by E2 stimulation (Fig. 4E, left) and that this response is associated with formation of ERa/MUC1-C complexes (Fig. 4E, right). These findings indicate that MUC1-C coactivates Rab31 gene transcription by interacting with ERa on the Rab31 promoter.

Rab31 Upregulates MUC1-C Expression
Rab31 functions in the trafficking of proteins, such as MUC1-C, that are processed in early endosomes. To determine whether Rab31 contributes to the regulation of MUC1-C, we silenced Rab31 in MCF-7 cells. Notably, downregulation of Rab31 was associated with a decrease in MUC1-C abundance (Fig. 5A, left). Silencing of Rab31 in ZR-75-1 cells also resulted in a decline in MUC1-C levels ( Fig. 5A, right). Analysis of MUC1 transcripts by qRT-PCR demonstrated that silencing Rab31 has no apparent effect on MUC1-C expression at the mRNA level (data not shown). Consequently, studies were performed to determine whether Rab31 regulates processing of MUC1-C from the endosome to the lysosome. Indeed, treatment of MCF-7/ Rab31siRNA cells with the lysosome inhibitor chloroquine (CQ) was associated with an increase in MUC1-C abundance (Fig. 5B, left). Similar results were obtained in CQ-treated ZR-75-1/ Rab31siRNA cells (Fig. 5B, right). To extend this analysis, we examined Rab31 expression in non-malignant MCF-10A breast epithelial cells. Compared to that in MCF-7 and ZR-75-1 cells, Rab31 transcripts were substantially lower in MCF-10A cells (Fig. 5C). Moreover, Rab31 and MUC1-C protein levels were in lower abundance in MCF-10A cells relative to that in the breast cancer cells (Fig. 5D). We therefore transfected MCF-10A cells to stably express Rab31 or an inactive Rab31(S20N) mutant (Fig. 5E). Expression of Rab31 in MCF-10A cells was associated with a modest increase in MUC1-C levels (Fig. 5E). In contrast, the Rab31(S20N) mutant completely suppressed MUC1-C expression (Fig. 5E). As a control, levels of EGFR protein were unaffected by expression of Rab31 or Rab31(S20N) (Fig. 5E), supporting the selective effects of Rab31 on MUC1-C expression.

Rab31 Supports Growth of MCF-10A Cells as Mammospheres by a MUC1-C-dependent Mechanism
To assess affects of Rab31 on growth, the MCF-10A cells were first seeded in soft agar. MCF-10A/vector cells failed to form colonies (Fig. 6A, left). In contrast, MCF-10A/Rab31 cells formed colonies and this response was substantially suppressed with the Rab31(S20N) mutant (Fig. 6A, left and right). Mammosphere structures are formed by culturing tumor-initiating cells under nonadherent and nondifferentiating conditions [18,19]. In concert with previous observations with parental MCF-10A cells [20], MCF-10A/vector cells failed to form mammospheres (Fig. 6B, left). However, mammospheres were formed with MCF-10A/ Rab31 cells (Fig. 6B, left). Moreover, for MCF-10A/Rab31(S20N) cells, small clusters were detectable that did not develop into mammosphere structures (Fig. 6B, left). Quantitation of mammo-sphere number confirmed dependence of MCF-10A cells on Rab31 for development of these structures (Fig. 6B, right). Immunoblot of MCF-10A/Rab31 cells growing as a monolayer and as mammospheres demonstrated the marked upregulation of MUC1-C levels under the mammosphere culture conditions (Fig. 6C, left). To assess involvement of MUC1-C in mammosphere formation, MUC1-C was silenced in the MCF-10A/Rab31 cells (Fig. 6C, right). Analysis of the MCF-10A/Rab31 cells expressing a control or MUC1 siRNA demonstrated that silencing MUC1-C is associated with a significant decrease in the formation of mammospheres (Fig. 6E, left and right). Using another approach to assess involvement of MUC1-C, we treated the MCF-10A/Rab31 cells with the cell-penetrating peptide GO-203 that blocks MUC1-C function [21,22]. As a control, cells were also treated with an inactive cell-penetrating peptide CP-2. Treatment

MUC1-C Upregulates Rab31 Expression
The overexpression of MUC1 in human breast cancers is associated with targeting of the MUC1-C subunit to the nucleus where it interacts with ERa [7,9,17]. The present studies demonstrate that silencing of MUC1-C in ER+ breast cancer cells is associated with downregulation of transcripts encoding the Rab31 GTPase. Little is known about the function of Rab31, other than it is involved in the trafficking of proteins in endosomes [3,4,5]. In addition, high levels of Rab31 in tumors from patients with node-negative breast cancers have been associated with decreased survival [6]. However, to our knowledge, there has been no known association between Rab31 and effectors linked to breast cancer. The present findings support a model in which the MUC1-C oncoprotein activates Rab31 expression by an ERamediated mechanism. MUC1-C stabilizes ERa by blocking its ubiquitination and degradation [17]. MUC1-C also associates with ERa on estrogen-responsive promoters, enhances ERa occupancy, increases recruitment of p160 coactivators and antagonizes the inhibitory effects of tamoxifen [17]. In the present studies, analysis of the Rab31 promoter identified potential ERa binding sites and, in ChIP assays, occupancy by ERa. Silencing of ERa suppressed activation of a Rab31 promoter-reporter and decreased Rab31 mRNA levels, indicating that Rab31 gene transcription is driven by ERa signaling. To our knowledge, Rab31 expression had not been previously linked to ERamediated transactivation and this finding provided the basis for assessing the potential involvement of MUC1-C. Indeed, in support of the regulation of Rab31 by MUC1-C, complexes of ERa and MUC1-C were detectable on the Rab31 promoter. Moreover, silencing MUC1-C suppressed (i) Rab31 promoter activation, (ii) Rab31 mRNA levels, and (iii) E2-induced Rab31 expression. There had previously been no available information on how the Rab31 gene is regulated in breast cancer cells, particularly those that express high Rab31 levels. Our results support a model in which the Rab31 gene is activated by both MUC1-C and ERa in breast cancer cells. These findings, however, do not exclude the possibility that MUC1-C could also contribute to Rab31 expression by ERa-independent mechanisms.

Rab31 Cooperates with MUC1-C in a Potential Autoinductive Loop
Expression of the MUC1-C subunit at the cell membrane is regulated by clathrin-mediated endocytosis [14,15]. The role of Rab31 in the trafficking of certain proteins in endosomes invoked the possibility that MUC1-C-mediated induction of Rab31 expression could, in turn, affect MUC1-C levels. Indeed, silencing of Rab31 in breast cancer cells was associated with a decrease in MUC1-C protein, but not transcripts, indicating that Rab31 might prevent processing of endosomal MUC1-C for degradation in lysosomes. In concert with such a model, chloroquine treatment of breast cancer cells silenced for Rab31 was associated with increases in MUC1-C abundance, indicating that Rab31 promotes recycling of MUC1-C, as opposed to its lysosomal degradation. The interaction between Rab31 and MUC1-C was further supported by studies in non-malignant MCF-10A breast epithelial cells, which express low levels of Rab31. Introduction of Rab31 in MCF-10A cells was associated with a modest increase in MUC1-C levels. Strikingly, expression of an inactive Rab31(S20N) mutant resulted in complete suppression of MUC1-C abundance, a response that was selective for MUC1-C in that Rab31(S20N) had no effect on EGFR levels. Rab31, but not Rab31(S20N), also conferred the ability of MCF-10A cells to form mammospheres. Under these nonadherent and nondifferentiating conditions of culturing the MCF-10A/Rab31 cells, the formation of mammospheres was decreased by silencing MUC1-C expression. In addition, treatment of the MCF-10A/Rab31 cells with GO-203, an inhibitor of MUC1-C function, blocked mammosphere formation. These results thus provide support for a cooperative interaction between MUC1-C and Rab31. MUC1-C activates Rab31 expression and, in turn, Rab31 increases MUC1-C levels in a positive feedback autoinductive loop. In this capacity, the functional significance of such a loop is that Rab31 alters the growth characteristics of breast epithelial cells, at least in part, by a MUC1-C-dependent mechanism.

Association of MUC1 and Rab31 Expression in ER+ Breast Cancers
An experimental model of MUC1-C-induced transformation identified the activation of gene families involved in oncogenesis [26]. Application of MUC1-C-induced genes associated with tumorigenesis to breast cancer databases predicted significant decreases in disease-free and overall survival [26]. Rab31 expression in breast cancers is also significantly associated with decreases in overall survival [6]. The present results demonstrate that Rab31 expression is significantly higher in breast tumors as compared to normal breast tissue. In addition and in concert with our finding that ERa activates Rab31 gene transcription, Rab31 expression was also significantly higher in ER+ as compared to ER-breast cancers. As noted above, Rab31 expression is not restricted to ER+ breast cancer cells, indicating that ERa-independent mechanisms can also contribute to Rab31 gene transcription. Our results further demonstrate that Rab31 and MUC1 are significantly co-expressed in ER+ breast cancers. These findings are thus consistent with a model in which MUC1-C coactivates ERa-mediated Rab31 transcription in breast cancer cells growing in vitro and as primary tumors. Other studies have demonstrated that MUC1-C induces genes involved in cholesterol and fatty acid metabolism [27]. The MUC1-C-induced gene set associated with lipid metabolism was applied to two independent databases from patients with ER+ breast tumors who were treated with tamoxifen. The results showed that patients with tumors expressing MUC1 and the lipid metabolic pathways are at higher risk for recurrence and death [27]. A positive correlation was also found between the MUC1-C-induced gene set and the ER signaling pathway [27]. These findings indicate that the autoinductive loop between MUC1-C and Rab31 found in ER+ breast cancer cells in the present studies could contribute, at least in part, to ER+ breast tumors that fail to respond to tamoxifen treatment. A direct inhibitor of MUC1-C is presently under study in a Phase I trial for patients with refractory solid tumors. Based on the present findings, patients with breast cancer that co-express MUC1-C and Rab31, and perhaps are resistant to tamoxifen, could be candidates for the targeting of MUC1-C with this agent.

Cell Culture
Human MCF-7 breast cancer cells were cultured in Dulbecco's modified Eagle's medium with 10% heat-inactivated fetal bovine serum (FBS), 100 U/ml penicillin, 100 mg/ml streptomycin and 2 mM L-glutamine. Human ZR-75-1 cells were grown in RPMI-1640 medium (ATCC) with 10% FBS, antibiotics and Lglutamine. Human MCF-10A mammary epithelial cells were grown in mammary epithelial growth medium (MEGM, Lonza). In certain experiments, cells were cultured in phenol red-free MEM medium containing 2% charcoal dextran-treated calf serum and then treated with 10 nM estradiol (E2; Sigma). Cells were also treated with chloroquine (CQ; Sigma). Stable silencing of MUC1 or Rab31 was performed by transduction of cells with a lentivirus expressing Mission shRNAs (Sigma) and selection in puromycin. For ERa silencing, cells were transfected with smart pool ERa siRNAs (Dharmacon).

RT-PCR and qRT-PCR
Total RNA was isolated from cells using an RNeasy Mini kit (Qiagen). cDNAs were synthesized with 0.3-1 mg RNA using the first-strand cDNA synthesis kit (Invitrogen). Expression of Rab31, MUC1 and b-actin was analyzed with 1 ml of cDNA using Taq DNA polymerase (Promega). For quantitative PCR, the SYBR green qPCR assay kit (Applied Biosystems) was used with 5 ml of 20-fold diluted cDNA from each sample, and the samples were amplified with the ABI Prism 7300 machine (Applied Biosystems). Primers used for RT-PCR and qRT-PCR are listed in Tables S1 and S2, respectively.

Immunoprecipitation and Immunoblot Analysis
Cell lysates were incubated with control IgG or anti-ERa (Santa Cruz Biotechnology) overnight at 4uC. Protein G-Sepharose beads (GE Health Care Life Sciences) were added for another 2 h. The immunoprecipitates and lysates not subjected to precipitation were immunoblotted with anti-Rab31 (C-15; Santa Cruz Biotechnolo-

Rab31 Promoter-reporter Assays
The Rab31 promoter region from 21 to 21874 was cloned into the pGL3 luciferase vector (Promega). Control pGL3 or pRab31-Luc constructs were transfected with the Renilla plasmid into cells in the presence of Lipofectamine. At 48 h after transfection, cells were lysed, and luciferase reporter activity was measured using the Promega Dual Glo kit.

Chromatin Immunoprecipitation (ChIP) Assays
Soluble chromatin was prepared from 2-3610 6 cells as described [28] and precipitated with anti-ERa or a control nonimmune IgG. For re-ChIP assays, complexes from the initial ChIP were eluted and reimmunoprecipitated with anti-MUC1-C as described [28]. For PCR, 2 ml from a 50 ml DNA sample was used with the indicated primers (Table S3) and 25-35 cycles of amplification. Fold enrichment was calculated as described [28].

Colony Formation in Soft Agar
Cells (3610 4 ) were suspended in 2 ml of 0.35% (wt/vol) agar containing DMEM/10% fetal bovine serum and overlaid onto a 0.75% (wt/vol) agar solution in 6-well plates. One ml of fresh medium was added once a week. Colonies .20 cells were counted and imaged after 3 weeks of incubation.

Analysis of Breast Cancer Datasets
Five publicly available datasets were analyzed that contain normal or breast tumor expressional data from 30 [29], 129 [30], 307 [23], 295 [24], and 147 [25] patients. All statistical analyses were performed using JMP 9.0 (SAS Institute Inc., Cary, NC, USA). The raw signal intensity for each probe set ID of interest for each patient was normalized to the average value of the probe set ID across all patients. Multiple probe set IDs for a given gene were averaged for each patient sample to obtain a representative expression value for each gene. MUC1 and Rab31 expression were defined as having a normalized expression value greater than one. Fisher's exact test was used to determine differences in association between two groups. Survival analysis was performed using Kaplan-Meier statistics with log-rank tests to test the null hypothesis of no difference in survival functions between patient groups.