|Home | About | Journals | Submit | Contact Us | Français|
Breast cancer metastasis suppressor 1 (BRMS1) suppresses metastasis of multiple tumor types without blocking tumorigenesis. BRMS1 forms complexes with SIN3, histone deacetylases and selected transcription factors that modify metastasis-associated gene expression (e.g., EGFR, OPN, PI4P5K1A, PLAU). microRNA (miRNA) are a recently discovered class of regulatory, noncoding RNA, some of which are involved in neoplastic progression. Based on these data, we hypothesized that BRMS1 may also exert some of its antimetastatic effects by regulating miRNA expression. Micro-RNA arrays were done comparing small RNAs that were purified from metastatic MDA-MB-231 and MDA-MB-435 and their non-metastatic BRMS1-transfected counterparts. miRNA expression changed by BRMS1 were validated using SYBR Green RT-PCR. BRMS1 decreased metastasis-promoting (miR-10b, -373 and -520c) miRNA, with corresponding reduction of their downstream targets (e.g., RhoC which is downstream of miR-10b). Concurrently, BRMS1 increased expression of metastasis suppressing miRNA (miR-146a, -146b and -335). Collectively, these data show that BRMS1 coordinately regulates expression of multiple metastasis-associated miRNA and suggests that recruitment of BRMS1-containing SIN3:HDAC complexes to, as yet undefined, miRNA promoters might be involved in the regulation of cancer metastasis.
Gene regulation by microRNA (miRNA) is a conserved mechanism in animals and plants.1 Endogenous, miRNA range from 15 to 28 nucleotides (nt) in Homo sapiens and, most commonly, negatively regulate gene expression, although some gene expression is positively regulated by miRNA. miRNA acts as templates for RNA-induced silencing complexes (RISC) to target mRNA. Animal miRNA differs functionally from plant miRNA in that they have imperfect base pairing with their target mRNA and more commonly inhibit protein translation than degrade mRNA.2 Imperfect or promiscuous base pairing allows animal miRNA to target multiple mRNA or even entire cellular pathways.3,4 To date, miRNA has been shown to regulate multiple cellular processes or pathways critical for neoplastic transformation and progression,5 including apoptosis,6,7 cell cycle regulation,8 differentiation,4,9,10 immune function11 and metabolism.12,13 Up- and down-regulation of miRNA expression are correlated with development of multiple cancers,5,14 including breast carcinoma,15–17 and a growing number of miRNA also contribute to promotion and suppression of cancer invasion and metastasis.16,18–27
Metastasis suppressors, defined by their ability to suppress metastasis without blocking orthotopic tumor growth, are an expanding family of >25 molecules.28,29 Breast cancer metastasis suppressor 1 (BRMS1) inhibits breast, melanoma, nonsmall cell lung and ovarian cancer metastasis in xenograft and syngeneic models.30–36 Expression of BRMS1 protein,30,37 but not necessarily mRNA,38,39 expression generally correlates inversely with survival and development of metastasis. BRMS1 associates with SIN3:histone deacetylase complexes36,40,41 which are involved in chromatin structure and selective regulation of gene expression. Collectively, these factors lead to the hypothesis that BRMS1 suppresses metastasis by altering expression of metastasis-associated genes. Previous studies have identified selective regulation of the epidermal growth factor receptor,42 osteopontin,35,36,43,44 connexins 45 and urokinase plasminogen activator.46 We hypothesized that BRMS1 might also regulate recently discovered, metastasis-associated miRNA.
To determine whether BRMS1 regulates miRNA expression, we compared miRNA expression patterns in nonexpressing and BRMS1 re-expressing breast carcinoma cells. miRNA expression was compared using microarrays imprinted with 328 known human miR probes and a selected common subset was further validated using quantitative real-time PCR (RTQ). In this article, we report that BRMS1 alters miRNA expression in metastatic breast carcinoma cells and notably down-regulated 3 of the 4 published metastasis-promoting miRNA22,23,47 and up-regulated all 3 of the known metastasis-suppressing miRNA.16,24
MDA-MB-231 (231) and MDA-MB-435 (435) are human estrogen receptor- and progesterone receptor-negative cell lines derived from pleural effusions of metastatic infiltrating ductal breast carcinomas.48,49 Both cell lines form progressively growing tumors when injected into the mammary fat pads of immunocompromised mice. MDA-MB-435 cells develop macroscopic metastases in lungs and regional lymph nodes by 10–12 weeks postinoculation, but infrequently metastasize after direct injection into the lateral tail vein or following subcutaneous injection. In contrast, MDA-MB-231 cells form macroscopic metastases when injected intravenously, but less commonly following injection into an orthotopic site. Both lines form osteolytic metastases following injection into the left ventricle of the heart.31,50–52 The origin of 435 has been questioned,53 however, this does not affect the interpretation of the results since BRMS1 suppresses metastasis of tumor cell lines from multiple tissue origins.33,34
Parental cell lines were cultured in a mixture (1:1 v:v) of Dulbecco’s modified Eagle’s medium (DMEM) and Ham’s-F12 medium (DMEM/F12; Invitrogen, Carlsbad, CA) supplemented with 2 mM L-glutamine, 0.02 mM nonessential amino acids and 5% fetal bovine serum (Invitrogen). All cultures were maintained without antibiotics or antimycotics on 100-mm tissue culture dishes (Corning, Corning, NY) at 37°C with 5% CO2 in a humidified atmosphere. When cultures reached 80–90% confluence they were passaged using a solution of 2 mM EDTA in Ca2+/Mg2+-free Dulbecco’s phosphate buffered saline (CMF-DPBS; Invitrogen). All cultures were regularly tested and confirmed negative for Mycoplasma spp. infection using a PCR-based test (TaKaRa, Shiga, Japan).
Cells (231 and 435) were transduced with GFP to facilitate tracking of cells in vivo (231GFP/435GFP) as previously described.50 Full-length BRMS1 cDNA was cloned into lentiviral constructs and transduced into 231GFP and 435GFP cells. Single cell clones (231BRMS1 and 435BRMS1) were isolated and BRMS1 mRNA and protein expression were verified. Transduced cells were initially selected with puromycin (500 μg/ml) and maintained in puromycin (100 μg/ml) to ensure stable transduction. For routine culture, no antibiotic selection was used and expression has been verified to be stable for more than 2 years.
Cells were grown to 90% confluence, media aspirated, washed in ice-cold PBS and lysed in acid phenol:chloroform (Ambion, Austin, TX). Small RNA species (≤200 nt) were isolated using mirVana PARIS kit (Ambion) according to manufacturer’s instructions and immediately stored at −80°C. RNA quantification was performed with use of a DU 800 spectrophotometer (Beckman Coulter, Fullerton, CA). Three independently isolated samples were collected for every cell line.
miRNA array profiling was performed at the Vanderbilt Microarray Shared Resource according to the core’s standard operating procedures (http://array.mc.vanderbilt.edu/microarray/expr/protocol.vmsr) as summarized later. RNA species (10–40 nt) were enriched using the flashPAGE fractionater (Ambion). The 3′ ends were polyadenylated with modified amines; RNA from 231GFP and 435GFP were labeled with Cy5 whereas RNA from 231BRMS1 and 435BRMS1 cells were labeled with Cy3. The yields on the coupling reactions were typical of miRNA labeling reactions with no detectable CyDye, but a large increase in measurable RNA indicated that poly A polymerase was effective in adding the modified dNTPs to the miRNA (data not shown). RNA samples were suspended in 3X Hyb™ buffer (Ambion), heated for 95°C for 2 min, allowed to cool at room temperature for 1 min and then loaded onto mirVana miRNA bioarrays v2 (Ambion) with Maui DC cover slips (BioMicro Systems, Salt Lake City, UT). Hybridizations were performed using the Maui hybridization station at 42°C for 16 hr with Mix D settings in Maui A1-A3. Arrays were washed once in Ambion SlideHyb™ low-stringency buffer for 30 sec with mixing, followed by 2 Ambion SlideHyb™ high-stringency buffer washes for 30 sec with mixing. Arrays were spun dry and scanned on the AXON 4000B scanner (Molecular Devices, Sunnyvale, CA).
Raw GeneChip files from GeneChip Operating Software (GCOS, Affymetrix, CA) were uploaded and background was subtracted. Expression changes were normalized to the respective control cells (231GFP or 435GFP) to calculate the intensity ratio/fold changes of the BRMS1-expressing counterparts (Cy5/Cy3). Data were sorted from greatest to least intensity. miR spots with foreground intensities less than 150 were disregarded as signals were too near background median. miR with values of −50 and −100 were also excluded as spot irregularity-affected fluorescence.
Raw data from Axon were background-subtracted, normalized to the respective control cells (231GFP or 435GFP) to calculate the intensity ratio/fold changes of the BRMS1-expressing counterparts (Cy5/Cy3). miR spots with foreground intensities less than 150 were disregarded as signals were too near background median. miR with values of −50 and −100 were also excluded as spot irregularity-affected fluorescence.
BRMS1 monoclonal antibody clone 1a5.7 was used at 1:2,500 and was previously described.36,54 Other antibodies used in this study were purchased and used at the titer indicated: anti-Twist1 (1:1,000), anti-α tubulin (1:1,000) and anti-EGFR (1:1,000; Cell Signaling Technology, Danvers, MA), anti-HoxD10 E20 (1:200; Santa Cruz Biotechnology, Santa Cruz, CA), anti-RhoC (1:1,000; Abcam, Cambridge, MA). Western blotting was performed as previously described.36,54
miRNA expression was determined by collecting total RNA from 70 to 90% confluent cell cultures using Qiazol (Qiagen Inc, Valencia, CA). RNA was purified using miRNeasy (Qiagen Inc). MiScript primers (Qiagen Inc), designed to amplify specifically mature miRNA (i.e., not pre-miR or pri-miR), were as follows: hsa-miR-10b_2, miR-96_1, hsa-miR-30e3p_1, hsa-miR-30e5p, hsa-miR-151, hsa-miR-320, hsa-miR-335_1, hsa-miR-373_1 and hsa-miR-520c. All samples were normalized to small nuclear RNA U6 and fold changes were calculated as previously described.55
Small RNAs (≤40 nt) were enriched from metastatic 231GFP, 435GFP and metastasis-suppressed BRMS1-transduced isogenic counterparts. BRMS1-associated mature miRNA expression changes were determined using miR-microarrays in 3 independent experiments. The top 25 consistently increased and decreased mature miRNA as a result of BRMS1 re-expression from both cell line comparisons are listed in Table I. The arrays revealed changes in mature miRNA expression that could contribute to BRMS1 metastasis suppression. For example, the oncogenic miR-155 decreased in 435 cells and tumor suppressing let-7a increased in 231 cells; however, as the arrays were used as a screen for miRNA changes observed in both cell lines, many miRNA were excluded from further study in this report. In general, the direction of miRNA expression change was consistent during the validations, but the magnitude of the changes was often under-appreciated on the arrays.
The criteria for prioritizing candidates were as follows: (i) technical replicates on the arrays were consistent; (ii) the direction of miRNA expression change was consistent in both 231/231BRMS1 and 435/435BRMS1 cell pairs; (iii) miRNA that target metastasis-associated mRNA (or proteins); and (iv) miRNA previously demonstrated to alter phenotypes associated with invasion and/or metastasis. Based on these criteria 6 miRNA were initially selected for further follow-up: miR-10b, -30e-3p, -30e-5p, -96, -151, -339 (Fig. 1).
Our original goals were to determine whether BRMS1 regulates miRNA and, if so, whether BRMS1-regulated miRNAs are downstream mediators of BRMS1 metastasis suppression. However, as these results were being collected, other laboratories reported on miRNA regulation of invasion and/or metastasis,16,22,23 compelling more detailed analysis of our data with regard to BRMS1 regulation of those miRNA (Table II).
The first miRNA validated to promote metastasis was miR-10b.23 Knockdown of miR-10b in 231 cells decreased in vitro migration and invasion. Additionally, ectopic expression of miR-10b in HMEC and SUM149 cells promoted in vitro migration and invasion. In addition, there was increased dissemination to the lung and formation of macroscopic foci in the peritoneum by SUM159 cells.23 Correspondingly, miR-10b expression was suppressed in 231BRMS1 and 435BRMS1 cells by greater than 50%, which is consistent with prior data showing that BRMS1 suppresses invasion and metastasis. Furthermore, when miR-10b expression was examined in a panel of cells denoting breast cancer progression, BRMS1 reduced miR-10b levels toward those found in nonmetastatic cells (Fig. 2).
Ma et al.23 further showed that miR-10b down-regulated HoxD10 protein which led to an increased expression of RhoC, which is a positive regulator of metastasis.56–58 Although RhoC mRNA expression decreased in both 231BRMS1 and 435BRMS1 cells (Fig. 3), HoxD10 mRNA increased in 435BRMS1 cells but decreased in 231BRMS1 cells. The differences observed when comparing cell lines suggest a nonlinear pathway and/or other (i.e., non-HoxD10) mediators of miR-10b. Because we were unable to detect HoxD10 with the commercially available antibodies, we measured mRNA. The inconsistency of mRNA levels between cell lines is in agreement with the findings reported by Ma et al.23
Relatedly, Huang et al.22 utilized a library of miR-transduced into human breast cancer cells to identify which miRNA promoted invasion in vitro. miR-373 and miR-520c stimulated migration and invasion in vitro and metastasis in vivo. Expression of miR-373 and -520c was decreased in 231BRMS1 and 435BRMS1 cells (Fig. 4 and Table II).
Tavazoie et al.16 compared miRNA expression patterns in paired sets of 231 variants which showed preferential organ-selective metastasis patterns. miR-335, when over-expressed by 250-fold, significantly suppressed lung and bone metastasis. Re-expression of BRMS1 suppressed metastasis to lungs and bone31,34,36 and increased expression of the metastasis-suppressing miR-335 by 6-fold (Fig. 4 and Table II). In the same report, miR-126 was described as metastasis suppressing16; however, it also suppressed orthotopic tumor growth, thereby excluding it as a metastasis suppressor.
Classically metastasis suppressor genes have encoded proteins; however, the demonstration that miR-335 and miR-146a/b suppress metastasis without blocking primary tumor growth shows that the types of molecules considered to be regulating metastasis needs to be more inclusive. Thus far, few pathways have been linked between known metastasis suppressors. In this article, we show for the first time that BRMS1 coordinately regulates expression of multiple miRNA and their corresponding downstream targets [e.g., RhoC (this report) and epidermal growth factor receptor42]. The findings are consistent with the emerging awareness that metastasis involves simultaneous control of multiple genes. Our observations beg the question as to whether other metastasis suppressors also mediate some or all of their affects through miRNA regulation.
Prior studies relating miRNA and metastasis have focused attention primarily toward identifying miRNA targets. Comparatively little is known about miRNA regulation.59,60 Ma et al.23 identified TWIST1, a major regulator of epithelial-mesenchymal transition,61 as a promoter of miR-10b expression.23 We discovered that BRMS1 decreases TWIST expression in whole cell lysates as well as nuclear fractions (Fig. 3). Moreover, BRMS1 mutants that do not suppress metastasis fail to down-regulate TWIST1 in 231 cells. Whether BRMS1 directly binds to the miR-10b promoter is still not known. But it is conceivable that BRMS1 might be part of a corepressor complex(es) recruited to miR-10b because BRMS1 is a component of several SIN3:histone deacetylase complexes.36,41 Similarly, the mechanism(s) by which regulation of other miRNA occurs will be the object of intense future investigation. Given the decrease in such a prominent epithelial-to-mesenchymal transition (EMT) regulator (Twist1), one could speculate that BRMS1 suppresses metastasis by invoking the reverse process (mesenchymal-to-epithelial transition, MET) in metastatic cells; however, some published and unpublished data argue against this possibility. Several recent publications indicate that miR-200 family are associated with increased expression of E-cadherin,20,21,26,62,63 which is typically lost during EMT.64 Data presented in Table I show decreases in miR-200 family members when BRMS1 is expressed. This is the opposite of what would be predicted. Moreover, changes in miR-200c are not consistent between 231 and 435 cells when BRMS1 is re-expressed. Also, we have previously reported that BRMS1 re-expression has no effect on E-cadherin mRNA or protein expression.65 Finally, BRMS1 consistently increases the vimentin expression, which is opposite of what one would expect if BRMS1 were regulating metastasis solely by regulating EMT.
Some miRNA expression changes observed in BRMS1-expressing 231 and 435 cells are large (miR-10b; 0.35/0.36, respectively), whereas others are more subtle (miR-373; 0.69/0.53 and miR-520c; 0.65/0.59, respectively). Nonetheless, it is interesting that nearly all (6 of 7; 86%) of the known metastasis-associated miRNA are regulated in the direction predicted for a metastasis suppressor. The exception was miR-21 which promoted metastasis in breast47 and colorectal cancer66 cells. BRMS1 increased miR-21 expression in 231 by ~ 2-fold, but did not change in 435. It is quite possible that in 231 cells, the combined effects changing expression of the other metastasis-associated miRNA may overwhelm the miR-21 up-regulation or the miR-21 expression change may be an attempt by cells to compensate for other changes.
Among the many miRNA changes observed in BRMS1-expressing cells, miR-146a and -146b expression were increased. Coupled with our prior reports that BRMS1 inhibits NFκB activity46 and EGFR expression,42 NFκB regulation by miR-146a/b25,67 and demonstration that EGFR is a target sequence for miR-146a/b, we tested the hypothesis that miR-146 was a downstream mediator of BRMS1 metastasis suppression. Transfection of miR-146a or miR-146b into 231 cells resulted in a significant suppression of lung metastases in experimental xenograft models.24 We emphasize that, unlike prior studies testing the role of miRNA in the processes of invasion and metastasis, the data reported here are based upon near physiologic expression levels of BRMS1 and the corresponding downstream miRNA. This point is key because off-target effects are more likely if experiments involve gross over-expression.
In summary, we demonstrate that BRMS1 regulates virtually all of the known miRNA regulating invasion and/or metastasis. Given that a single metastasis suppressor regulates several key metastasis-associated miRNA, future studies should focus on which step(s) of the metastatic cascade are dysregulated in human disease and the specific miRNA expression regulatory elements that are regulated by BRMS1 directly.
This work was supported by U.S. Public Health Service Grants CA87728 (D.R.W.), CA13148 (D.C.), UL1RR025777 (D.C.) and F32CA113037 (D.R.H.); a predoctoral fellowship from the U.S. Army Medical Research and Materiel Command W81-XWH-08-1-0786 (M.D.E.); a postdoctoral fellowship from Susan G. Komen for the Cure PDF1122006 (K.S.V.) and a grant from the National Foundation for Cancer Research—Center for Metastasis Research (D.R.W.). This work is submitted in partial fulfilment of the requirements of the Molecular and Cellular Pathology Graduate Program (M.D.E.). The authors thank Dr. Janet Price (University of Texas M.D. Anderson Cancer Center) for providing the MDA-MB-231 and -435 cell lines. They also thank the Vanderbilt Microarray Core Facility for superb technical assistance with the microRNA arrays. Author contributions: M.D.E. and D.R.W. designed research; M.D.E., D.R.H., K.S.V. and L.J.S. performed research; M.D.E., D.R.H., K.S.V., D.C. and D.R.W. analyzed data; D.C. contributed new analytical tools and M.D.E., D.R.H., K.S.V. and D.R.W. wrote the article.
Grant sponsor: U.S. Public Health Service Grants; Grant numbers: CA87728, CA13148, UL1RR025777, F32CA113037; Grant sponsor: U.S. Army Medical Research and Materiel Command; Grant number: W81-XWH-08-1-0786; Grant sponsor: Susan G. Komen for the Cure; Grant number: PDF1122006; Grant sponsor: National Foundation for Cancer Research—Center for Metastasis Research
Conflicts of interest: The authors declare no conflicts of interest.