The MicroRNA-23b/27b/24 Cluster Promotes Breast Cancer Lung Metastasis by Targeting Metastasis-suppressive Gene Prosaposin

Background: The role of microRNA (miRNA)-23b/27b/24 in metastasis is controversial. Results: Ectopic expression of the miRNA-23b/27b/24 cluster enhances metastasis in vivo by direct inhibition of prosaposin (PSAP). Conclusion: PSAP is a metastasis suppressor that is targeted by miRNA-23b/27b/24 in breast cancer. Significance: This finding represents a novel miRNA-target interaction for therapeutic intervention of breast cancer metastasis. MicroRNAs (miRNAs) have been shown to function as key regulators of tumor progression and metastasis. Recent studies have indicated that the miRNAs comprising the miR-23b/27b/24 cluster might influence tumor metastasis, although the precise nature of this regulation remains unclear. Here, expression of the miR-23b/27b/24 cluster is found to correlate with metastatic potential in mouse and human breast cancer cell lines and is elevated in metastatic lung lesions in human breast cancer patients. Ectopic expression of the miRNAs in the weakly metastatic mouse 4TO7 mammary tumor cell line had no effect on proliferation or morphology of tumor cells in vitro but was found to increase lung metastasis in a mouse model of breast cancer metastasis. Furthermore, gene expression profiling analysis of miRNA overexpressing 4TO7 cells revealed the direct targeting of prosaposin (PSAP), which encodes a secreted protein found to be inversely correlated with metastatic progression in human breast cancer patients. Importantly, ectopic expression of PSAP was able to suppress the metastatic phenotype in highly metastatic 4T1 and MDA-MB-231 SCP28 cells, as well as in cells ectopically expressing miR-23b/27b/24. These findings support a metastasis-promoting function of the miR-23b/27b/24 cluster of miRNAs, which functions in part through the direct inhibition of PSAP.


MicroRNAs (miRNAs) have been shown to function as key regulators of tumor progression and metastasis. Recent studies
have indicated that the miRNAs comprising the miR-23b/ 27b/24 cluster might influence tumor metastasis, although the precise nature of this regulation remains unclear. Here, expression of the miR-23b/27b/24 cluster is found to correlate with metastatic potential in mouse and human breast cancer cell lines and is elevated in metastatic lung lesions in human breast cancer patients. Ectopic expression of the miRNAs in the weakly metastatic mouse 4TO7 mammary tumor cell line had no effect on proliferation or morphology of tumor cells in vitro but was found to increase lung metastasis in a mouse model of breast cancer metastasis. Furthermore, gene expression profiling analysis of miRNA overexpressing 4TO7 cells revealed the direct targeting of prosaposin (PSAP), which encodes a secreted protein found to be inversely correlated with metastatic progression in human breast cancer patients. Importantly, ectopic expression of PSAP was able to suppress the metastatic phenotype in highly metastatic 4T1 and MDA-MB-231 SCP28 cells, as well as in cells ectopically expressing miR-23b/27b/24. These findings support a metastasis-promoting function of the miR-23b/27b/24 cluster of miRNAs, which functions in part through the direct inhibition of PSAP.
MicroRNAs (miRNAs) 2 have been recognized as regulatory RNAs with significant roles in multiple pathological processes. Individual miRNAs have been shown to regulate hundreds of downstream genes, and studies have implicated improper miRNA expression as a causal factor in tumorigenesis and metastasis (1)(2)(3). Alterations in miRNA levels are a frequent occurrence in cancer, and miRNA expression patterns can be utilized to stratify cancer subtypes, even in poorly differentiated tumors of ambiguous origins (4,5). In addition, miRNAs have been identified as significant prognostic and predictive markers during tumor progression. For example, miR-10b (6) and the miR-200 family of miRNAs (7) are correlated with breast cancer metastasis, whereas miR-15b is associated with recurrence in melanoma (8).
Previous studies have implicated the miR-23/27/24 miRNAs in the regulation of tumor progression. These miRNAs are expressed in two separately regulated clusters, comprising 23a/ 27a/24 -2 and miR-23b/27b/24 (15). Expression of the miR-23/ 27/24 clusters has been linked to bone morphogenetic protein and TGF␤ signaling, although the exact regulatory mechanism of each cluster remains unclear (16 -18). Sun et al. (16) found that the miRNAs within the miR-23b/27b/24 cluster were independently regulated upon treatment with bone morphogenetic protein-2, with miR-23b decreasing in expression, miR-27b showing no significant changes, and miR-24 levels increasing. In addition, there are conflicting reports regarding the functional role of the miR-23/27/24 clusters during tumor development and metastatic progression. miR-23b has been shown to decrease migration and invasion in vitro (19) and to decrease colon cancer lung metastasis (20) and breast cancer lymph node metastasis in vivo (21). Conversely, ectopic expression of the entire miR-23a/27a/24 cluster increased migration and invasion in breast cancer cells and promoted hepatic metastasis (22). Similarly, miR-24 has been shown to inhibit apoptosis and promote the formation of micrometastases within the lungs of subcutaneously injected mice (23). Finally, although miR-27b has been shown to decrease primary colorectal cancer tumor growth and inhibit angiogenesis (24), miR-27a has been shown to positively correlate with breast cancer progression (25). Thus, the miR-23ab/27ab/24 miRNAs have been shown to possess dichotomous roles during tumor progression. Although it is possible that these contradictory findings are a result of the different model systems employed in the previous reports, further research is required to identify the precise role of the miR-NAs during breast cancer invasion and tumor progression.
In this study, we examined the expression of the miR-23/ 27/24 clusters across multiple isogenic cell line series containing sublines with diverse metastatic propensities. Expression of miR-23b/27b/24 was found to be elevated in the metastatic variants within these progression series, as well as within lung metastasis samples relative to matched primary human breast cancer tissues. Furthermore, ectopic expression of all three miRNAs within the lowly metastatic 4TO7 mouse breast cancer cell line enhanced lung metastasis. Microarray expression analysis for targets of the miRNAs further uncovered prosaposin (PSAP) as a direct target of miR-24 and miR-27b.
Analysis of Clinical Samples and KM-plotter Data Set-Patient samples from the IRCCS Institute of Romagna for the Study and Treatment of Cancer (IRST) were collected and purified as described previously (7). Briefly, women with resected breast cancer were selected from patients followed from 1995 to 2010 in IRST (Meldola, Italy). Tumor specimens were deidentified and were considered exempt samples in accordance with the institutional review board of the Local Ethic Committee (Forlì, Italy). Surgical tumor specimens were fixed in formalin and embedded in paraffin. Ten primary breast tumors and 10 lung metastatic tissues derived from primary breast lesions were collected. For six patients, matched primary-metastatic tissues were available. Total RNA was collected from 20-mm-thick sections from formalin-fixed paraffin-embedded tissue blocks using the formalin-fixed paraffin-embedded RNA/DNA purification kit (Norgen) according to the manufacturer's instructions and subjected to RT-PCR analysis. For KM-Plotter investigation, distant metastasis-free survival analysis was performed utilizing the latest available breast cancer data set, comprising 1610 patient samples and the 200871_s_at PSAP probe (26).
RNA Isolation and Microarray-RNA for qRT-PCR and microarray was isolated from cultured cells using a miRVana total RNA isolation kit (Ambion) according to the manufacturer's instructions. For RNA isolation from tumor lesions, the lesions were macroscopically dissected and pulverized using a liquid nitrogen cooled mortar and pestle, followed by RNA isolation with the miRVana kit. For microarrays, samples were analyzed with the Agilent Whole Mouse Genome 4 ϫ 44k arrays. RNA samples were labeled with Cy5 using the Agilent low RNA input linear amplification kit and were hybridized with the Cy3-labeled human reference RNA (Stratagene). Duplicate arrays were performed for each sample and analyzed with an Agilent G2565BA scanner and Agilent Feature Extraction software (version 9.5). The Cy5/Cy3 ratios were calculated by median signal and normalized by the array median. Probes with Ͼ2.5-fold changes were identified as potential direct targets of miR-23/27/24.
Transwell Migration Assay-10 5 4TO7 control or miRNAoverexpressing cells were seeded into the upper chamber of an 8-m pore transwell migration insert (BD Biosciences) containing serum-free media. Serum-containing DMEM was used as an attractant in the bottom chamber, and cells were allowed to migrate for 12 h. At the conclusion of the experiment, the inserts were removed, and trypisized cells were counted from the bottom chamber, providing a quantitative value for migrated cells across the membrane. For invasion assays, 1 mg/ml Matrigel (BD Biosciences) was polymerized in the top chamber prior to the addition of cells, which were allowed to invade for 20 h before quantification.
Tumor Xenografts and Bioluminescence Analysis-All procedures involving mice and experimental protocols were approved by Institutional Animal Care and Use Committee of Princeton University. For experimental lung metastasis studies, 10 5 tumor cells were injected into the lateral tail vein of female BALB/c mice. For orthotopic injections, 10 6 cells were injected into the mammary gland as described previously (27). Development of primary tumors was analyzed every 4 days, with final measurements performed at end point. Lungs collected from tail vein-injected mice were washed briefly in PBS, followed by overnight fixation in Bouin's fixative (Sigma) and were stored in 70% ethanol. Lungs from orthotopic injections were minced and digested with 300 units/ml type 1A collagenase (Sigma, C2674) and 100 units/ml hyaluronase (Sigma, H3506) for 60 min at 37°C. Dissociated cells were passed through a 70-m strainer and plated into 15-cm plates. 60 M 6-thioguanine was used to select for 4TO7 tumor cells, which were grown for 2 weeks, followed by staining with crystal violet for 30 min for quantification.

miR-23/27/24 Expression Correlates with Breast Cancer
Lung Metastasis-Previous reports have implicated the miR-23ab/27ab/24 miRNAs as key mediators of metastasis, although there exists uncertainty as to the precise role of the miRNAs during breast cancer metastasis. These clusters are thought to originate from a highly conserved duplication event; miR-23a and -23b and miR-27a and -27b contain minute differences within the mature miRNA sequences, although the seed sequences of the related miRNAs are conserved across clusters, resulting in two highly conserved clusters (Fig. 1A). To examine expression of the miRNAs within metastasis, we utilized matched primary tumor and lung metastasis samples isolated from microdissected tumor samples from breast cancer patients. qRT-PCR analysis of these samples revealed a consistent up-regulation of all miRNAs in the metastatic tissues (Fig. 1B).
To better examine the function these miRNAs during tumor progression, three breast cancer isogenic cell line series were selected that model the progression of breast cancer with increasing metastatic ability: the mouse 4T1 cell line series and the human MCF10A series and MDA-MB-231 series. The 4T1 series, consisting of 67NR, 168FARN, 4TO7, 66cl4, and 4T1 cells, are a syngenic model derived from a spontaneously arising BALB/cfC 3 H tumor (28). This series displays varied metastatic potential upon orthotopic injection into the mammary gland of mice, with 4T1 and 66cl4 cells producing spontaneous lung metastases, 4TO7 forming micrometastases in the lungs, and 67NR and 168FARN forming primary tumors but lacking metastases (28 -30). The MCF10A series includes the immortalized, but otherwise normal and non-tumorigenic MCF10A mammary epithelial cell line, as well as in vivo selected derivatives of H-RAS-transformed parental cells with varying metastatic potentials (31). These include the relatively benign MCF10AT cells, which do not typically progress into carcinomas, as well as the low-grade, poorly metastatic carcinoma line MCF10CA1h and the metastatic MCF10CA1a cells. The MDA-MB-231 cell line consists of a weakly metastatic parental line, as well as clonally selected, single cell-derived progenies (SCPs) that feature varying metastatic propensity and organotropism (32,33). Specifically, although SCP4 and SCP6 cells are essentially non-metastatic, the SCP28, 4173, and 4175 lines are capable of forming overly lung metastases (33). miR-23a/27a/24 and miR-23b/27b/24 expression levels were examined in these cells by qRT-PCR. Expression of miR-23a, miR-27a, and miR-24 showed a correlation with metastatic potential within the 4T1 (Fig. 1C), MCF10A (Fig. 1D), and MDA-MB-231 progression series (Fig. 1E). miRNA expression of all three members of the miR-23b/27b/24 cluster showed an overall similar but slightly stronger association with metastatic potential (Fig. 1, C-E). Given the similarity of these two clusters, we chose to focus on functional analysis of the role of the miR-23b/27b/24 cluster.
The mouse miRNAs were stably overexpressed in the weakly metastatic 4TO7 cells using the pMSCV retroviral vector, as verified by qPCR ( Fig. 2A). Expression of all three miRNAs in the miR-23b/27b/24 cluster were elevated by ϳ6 -8 fold relative to vector transduced cells, at a level similar to the endogenous miRNA expression found in 4T1 cells (Fig. 1C). Importantly, the overexpression was specific, with no significant changes observed in miR-23a and miR-27a within these cells. Overexpression of miR-23b/27b/24 had no effect on morphology or proliferation within the 4TO7 cells (Fig. 2, B and C). Interestingly, miRNA overexpressing cells reduced transwell migration and decreased Matrigel invasion (Fig. 2D), consistent with previous reports of decreased migration upon ectopic expression of miR-23b in prostate cancer cell lines (19).
The miR-23b/27b/24 Cluster Promotes Lung Metastasis-To investigate the influence of the miR-23b/27b/24 cluster miRNAs on metastatic colonization in the lungs, mice were inoculated with vector control or miRNA-overexpressing 4TO7 cells via the lateral tail vein and subsequently monitored for the incidence of lung metastasis. Lungs were collected 2 weeks post-injection, and overt lesions were counted. This quantification revealed that miR-23b/27b/24 expression significantly increased the number of lung lesions relative to control (p ϭ 0.0289, Fig. 2, E and F). Interestingly, qRT-PCR analysis revealed that miR-23b/27b/24 levels were modestly elevated in isolated lung lesions beyond that of the in vitro population prior to injection, suggesting that the metastasis process may have selected for clones with elevated expression (Fig. 2G).
To examine the influence of miR-23b/27b/24 on the early steps of metastatic progression, 4TO7 cells were inoculated orthotopically into the mammary fat pad of BALB/c mice. Primary tumor growth was comparable between both parental and miR-23b/27b/24-overexpressing 4TO7 cells (Fig. 2, H and I), both of which spontaneously disseminated to the lungs but failed to develop into overt lung metastases. Instead, the micrometastatic burden was estimated by culturing tumor colonies from dissociated lungs; quantification of crystal violet-stained colonies revealed a trend toward increased metastasis from mice injected with the miR-23b/27b/24-overexpressing cells, although this did not reach significance (Fig. 2, J and K). Together, these in vivo results indicate that the miR-23b/ 27b/24 cluster promotes lung metastasis with most of the influence exerted on the step of colonization.

Genomic Identification of miR-23b/27b/24 Target Genes-
To elucidate the molecular mechanism of miR-23b/27b/24 in enhancing lung metastasis, we employed a combined approach, utilizing microarray analysis and the TargetScan computational target prediction algorithm (34). The microarray analysis detected 24,705 unique genes, revealing 402 genes with Ͼ2.5fold decreased expression in the miRNA-overexpressing cells. To identify potential direct targets of miR-23b/27b/24, the subset of down-regulated genes was analyzed for putative miRNAbinding sites. Candidate target genes were then examined for predicted miRNA binding sites using TargetScan, and eight predicted targets were chosen for further analysis (Fig. 3A). qRT-PCR results validated repression of all eight genes after ectopic miRNA expression (Fig. 3B).
To confirm direct targeting by the miRNAs, the 3Ј-UTRs from the putative target genes were cloned into the pMIR-REPORT luciferase plasmid. Reporter plasmids were co-transfected into HeLa cells along with the miRNAs of interest and a Renilla luciferace plasmid for normalization. Analysis of luciferase expression revealed that only the prosaposin (Psap) reporter expression was suppressed after ectopic miRNA expression (Fig. 3C). PSAP is a secreted protein that has been described previously as both a metastasis promoter and suppressor (35)(36)(37). PSAP mRNA contained three predicted bind-ing sites for the miR-23/27/24 clusters within the 3Ј-UTR, including two for miR-24 and one for miR-27 (Fig. 3D). We confirmed direct targeting through mutagenesis of the predicted miRNA binding sites: mutation of the miR-27 binding site and one of the two miR-24 partially rescued luciferase signal, whereas combined mutagenesis of both miR-24 and miR-27 sites fully rescued luciferase signal (Fig. 3E).
PSAP Expression Is Correlated with Lung Metastasis-Examination of PSAP expression in our cell line models revealed that the mRNA and protein expression level of PSAP decreased in the metastatic variants of the 4T1 and MCF10A isogenic progression series (Fig. 4, A-D). In particular, a significant reduction of PSAP protein and mRNA expression was found in the 4T1 and MCFCA1a cell lines, which represent the most metastatic ones and the only ones capable of forming macrometastases in each series. These findings implicate PSAP as a putative metastasis suppressor, we next sought to examine the effect of PSAP overexpression on metastatic progression. To this end, mouse and human PSAP was ectopically expressed in 4T1 and MDA-MB-231-SCP28 cells, respectively. Ectopic expression was confirmed by qRT-PCR and Western blot analyses (Fig.  5A), and cells were injected via the lateral tail-vein into mice. Analysis of lung lesions revealed that PSAP suppressed metastasis in both 4T1 and SCP28 cells (Fig. 5B). Importantly, when

The miR-23b/27b/24 Cluster Promotes Metastasis
Psap was ectopically expressed in 4TO7-miR-23b/27b/24 cells, there was a significant decrease in lung lesions, indicating that re-expression of Psap was capable of partially neutralizing the metastasis-promoting effects of the phenotype resulted from the overexpression of the miR-23b/27b/24 cluster (Fig. 5C).
To further examine the clinical significance of PSAP as a metastasis factor, we analyzed the expression levels of PSAP in publically available clinical datasets, as well as the microdissected primary breast cancer or lung metastasis patient samples from before. Analysis of breast cancer patients using the KMplotter data set (26) revealed a significant correlation between PSAP expression and increased distant metastasis-free survival, particularly in estrogen receptor-negative breast cancer patients (Fig. 5D). In addition, analysis of patient samples revealed that PSAP expression decreased in lung metastasis samples compared with the primary mammary tumors (Fig.  5E). Together, these results indicate that PSAP is a direct target of the miR-23/27/24 clusters and is reduced during metastasis, both in cell lines and clinical samples.

DISCUSSION
miRNA regulation has been implicated in various critical aspects of tumor progression, both as clinical biomarkers and functional regulators of metastasis. Previous in vitro and in vivo studies have implicated the miR-23/27/24 family of miRNAs as both positive and negative regulators of the metastatic cascade. For example, expression of miR-23b and miR-27b has been shown to inhibit migration and anchorage-independent growth in prostate cancer cells (19), whereas miR-23a/27a/24 increases migration and metastasis in breast cancer lines (22). In this study, we show that the expression of the miR-23b/27b/24 cluster is elevated in lung metastasis in human breast cancer and that ectopic expression of the miRNAs promotes lung metastasis in the 4TO7 mouse mammary cell line. Therefore, the miR-23b/27b/24 is likely to have a metastasis-promoting role in lung metastasis of breast cancer.
It is curious that ectopic expression of the miRNAs was capable of increasing the metastatic potential upon intravenous inoculation while possessing much weaker effects on spontaneous metastasis. It is possible that this effect can be attributed in part to limitations within the assay; both 4TO7 vector and miRNA overexpression cells develop visible lesions upon lateral tail vein injection but were unable to develop into overt metastasis in orthotopic experiments. In addition, it is possible that FIGURE 3. Identification of putative miR-23b/27b/24 cluster targets. A, average gene expression (expressed as fold change) for candidate target genes from microarray analysis of 4TO7-miR-23b/27b/24 cells relative to vector control. B, qRT-PCR mRNA expression levels for putative target genes in 4TO7 vector cells or miR-23b/27b/24 overexpressing cells. **, p Ͻ 0.05; **, p Ͻ 0.01 by Student's t test. Data represent average Ϯ S.D. C, normalized firefly luciferase expression from reporter plasmids containing indicated 3Ј-UTR 24 h after co-transfection with control or miR-23b/27b/24 miRNA oligonucleotides. **, p Ͻ 0.01. Data represent average Ϯ S.D. D, schematic representing the predicted miR-27ab and miR-24 binding sites within the human PSAP 3Ј-UTR. MiR-24 contains two predicted sites, referred to here as 1 and 2, whereas miR-27ab contains one predicted binding site. E, normalized firefly luciferase (Luc) expression from PSAP reporter constructs similar to following site-directed mutagenesis of predicted miRNA binding sites and co-transfection with miR-27b or miR-24 as indicated. *, p Ͻ 0.05; **, p Ͻ 0.01 by Student's t test. Data represent average Ϯ S.D. the observed defect in migration and invasion could offset any increase of metastatic colonization abilities conferred by the miRNAs. Future studies will be necessary to further elucidate the influence of the miRNAs on the stepwise progression of breast cancer lung metastasis.
Multiple targets have been identified for members of the miR-23b/27b/24 cluster, including SPRY, ANXA2, ARHGEF6, CFL2, LIMK2, PIK3R3, PLAU, VEGFC, PTPN9, and PTPRF (21)(22)(23)(24). We have additionally uncovered the direct targeting of PSAP by miR-24 and miR-27b. PSAP, a highly conserved precursor for saposins A, B, C, and D, has been identified previously as a functional regulator of metastatic progression (36,37). Our results show that PSAP mRNA and protein expression is correlated with reduced metastatic potential and that ectopic PSAP expression is capable of inhibiting metastasis. Previous studies found that PSAP was down-regulated in metastatic variants of the PC3 prostate cancer and MDA-MB-231 breast cancer cells (36). Finally, PSAP has been shown to increase thrombospondin 1 expression in PC3 cells, and PSAP or a short PSAP-mimetic peptide (DWLPK) inhibited prostate cancer lung metastasis (38). Future studies will be required to further analyze the functional mechanism of PSAP in suppressing metastatic colonization in the lungs.
Enhanced migration and invasion were canonically believed to increase the metastatic potential of tumor cells, although recent reports have identified highly metastatic cells with decreased migratory capacity (39). Indeed, this is observed within the 4T1 series, in which the most highly metastatic 4T1 cells are actually less migratory (7,29). Thus, it is not possible to predict metastatic potential through migratory capacity alone. In our current study, overexpression of miR-23b/27b/24 inhibits invasion and migration, while dramatically increasing metastatic colonization in the lungs. This observation is similar to what we previously reported about the miR-200 family miRNAs (7). miRNAs have been shown to influence the expression of hun-dreds of downstream genes, so it is perhaps not surprising that ectopic expression of three miRNAs might simultaneously influencing separate aspects of the metastatic cascade. The dichotomous functions of these miRNAs in early and late steps of the metastatic cascade may complicate their utilization as therapeutic targets. One possible approach to sidestep this complication is to use the direct target of miR-23b/27b/24, PSAP, or its peptide mimetics, as therapeutic agents against metastasis. Future studies should assess whether PSAP or its functional peptide mimetics may have a therapeutic influence on metastatic progression. The combination of clinical correlation and direct functional inhibition in preclinical mouse models strongly suggests that PSAP is functioning as a metastasis suppressor. Future studies will be required to asses the utility of PSAP as a therapeutic strategy, both as a diagnostic marker for disease progression and as a direct pharmaceutical target.