Context-dependent bidirectional regulation of the mutS homolog 2 by transforming growth factor β contributes to chemoresistance in breast cancer cells

Cell lines, plasmids and viruses

Human BC cell lines MDA-MB-231 (MDA231), MCF7 and HCC1954, MCF10A mammary epithelial cells and 293 embryonic kidney cells were obtained from American Type Culture Collection (ATCC; Manassas, VA) and cultured in the recommended media in a humidified 5% CO₂ incubator at 37 °C. MCF10A cells stably overexpressing HER2 (MCF10A/HER2) or vector (MCF10A/vec) and primary human mammary epithelial cells (HMEC) were generated and cultured as previously described (20-21). Recombinant human TGF-β1 was purchased from R&D Systems (Minneapolis, MN). The type I/II TGF-β receptor inhibitor LY2109761 was provided by Eli Lilly and Company (Indianapolis, IN). The HER2/EGFR small molecule inhibitor lapatinib ditosylate was purchased from LC Laboratories (Woburn, MA). The MDM2 antagonist nutlin-3 was purchased from Sigma-Aldrich (St. Louis, MO). The expression construct of MSH2 was kindly provided by Dr. Jeremy Stark (City of Hope).

To construct the 3’UTR reporter plasmids, annealed oligonucleotides 5’-tcgataatggaatgaaggtaatattgataagctattgtctg-3’ and 5’-ggcccagacaatagcttatcaatattaccttcattccatta-3’ containing nt 11~47 of MSH2 3’UTR sequence (encompassing the putative miR-21 binding site), 5’-tcgataatggaatgaaggtaatattgatccgctattgtctg-3’ and 5’-ggcccagacaatagcggatcaatattaccttcattccatta-3’ containing the mutated MSH2 3’UTR sequence (MSH2 3’UTR^mut; mutated nucleotides in the miR-21 binding site underlined), or 5’-tcgaagttatttgctatatattattttcaacatgagttttt-3’ and 5’-ggccaaaaactcatgttgaaaataatatatagcaaataact-3’ containing a scrambled sequence, were inserted into the XhoI/NotI sites of psiCHECK^TM-2 vector (Promega, Madison, WI), downstream of the luciferase encoding region. A 224-bp promoter region (-201~+24) of human MSH2 gene (hMSH2) was cloned by PCR using the upstream primer 5’-atatatgctagcaggatgcgcgtctgcgggtttcc-3’ (for SBE wt) or 5’-atatatgctagcaggatgcgcgatcgcgggtttcc-3’ (for SBE mt) and the downstream primer 5’-gcgcgcaagcttacacccactaagctgtttcc-3’, and inserted into the NheI/HindIII sites of pGL3-Basic vector (Promega), upstream of the luciferase reporter gene, to generate the hMSH2 promoter reporters. The small hairpin (sh) RNA targeting the MSH2 mRNA sequence 5’-gggctggtgacagtcaattga-3’ was first constructed by inserting annealed oligonucleotides 5’-gatcgggctggtgacagtcaattgatttgtgtagtcaattgactgtcaccagcccttttttgcatg-3’ and 5’-cccgaccactgtcagttaactaaacacatcagttaactgacagtggtcgggaaaaaac-3’ into the BglII/SphI sites of pU6tetO vector (22) (kindly provided by Dr. John J. Rossi, City of Hope), downstream of a doxycycline (Dox) inducible U6 promoter. The efficiency of gene knockdown by this construct as well as other two constructs targeting different MSH2 sequences was examined in transiently transfected MDA231 cells, where this construct most efficiently reduced MSH2 expression as detected by Western blot (data not shown). The entire U6tetO-MSH2 shRNA region was then subcloned into the NotI/SphI sites of lentiviral vector pTIG (pHIV7-TetR-IRES-GFP) (22) (kindly provided by Dr. Rossi), to generate pTIG-MSH2 shRNA. Viruses expressing MSH2 shRNA or empty vector were produced by co-transfecting 293 cells with the construct and packaging plasmids pCHGP-2, pCMV-Rev and pCMV-G, as previously described (23), and used to infect MDA231 cells, followed by GFP selection.

RNA extraction, reverse transcription (RT) and real time quantitative PCR (qPCR)

RNA was isolated using the TRIzol reagent (Invitrogen, Carlsbad, CA) and subjected to RT using the miScript Reverse Transcription Kit (Qiagen; Valencia, CA), following the manufacturers’ protocols. Gene expression was quantified by qPCR using the miScript SYBR Green PCR Kit (Qiagen) and 25 ng of cDNA per reaction. For miR-21 detection, qPCR was carried out using the miScript Primer Assay for miR-21 or the U6 non-coding small nuclear RNA (snRNA) (as a control) and the miScript Universal Primer (Qiagen). Primers used for detection of MSH2 mRNA and in ChIP assay (for an hMSH2 coding region) are 5’-tcatggctgaaatgttggaa-3’ and 5’-ttggccaaggcagtaagttc-3’, and for GAPDH mRNA 5’-accacagtccatgccatcac-3’ and 5’-tccaccaccctgttgctgta-3’. Primers used for detection of the SBE-p53 region in hMSH2 promoter in the ChIP assay are 5’-atatatgctagcaggatgcgcgtctgcgggtttcc-3’ and 5’-gcgcgcaagcttacacccactaagctgtttcc-3’. An annealing temperature of 55 °C was used for all the primers. PCR reactions were performed in a standard 96-well plate format with the iQ^TM5 multicolor real time PCR detection system (BioRad; Hercules, CA). For the data analysis of qRT-PCR, raw Ct was first normalized to U6 (for miRNA) or GAPDH (for mRNA) for each sample to obtain dCt. The normalized dCt was then calibrated to control cell samples to obtain ddCt. For the data analysis of qChIP-PCR, raw data was calibrated to the corresponding input DNA sample, which was purified from one-twentieth of the cell lysate used in ChIP but without the IP procedure.

Chromatin immunoprecipitation (ChIP) assay and Western blot analysis

Preparation of cell lysates, ChIP and Western blot were carried out as described previously (21, 24). Smad2/4, p53, MSH2, and GAPDH antibodies were purchased from Cell Signaling (Danvers, MA).

Cell transfection and RNA interference (RNAi) studies

DNA transfection was performed using Lipofectamine 2000 (Invitrogen) following the manufacturer's protocol, as described previously (20). The miRIDIAN miRNA hairpin inhibitor and mimic of miR-21 and the negative controls were purchased from Dharmacon (Lafayette, CO). Silencer siRNA against p53 target sequence 5’-aaggaaatttgcgtgtggagt-3’ and the AllStars negative control siRNA were purchased from Qiagen. MiRNA inhibitors/mimics and siRNAs were transfected into cell lines using DharmaFECT Duo Transfection Reagent (Dharmacon) according to the manufacturer's procedures. In 6-well plate format, a final concentration of 25 nM miRNA inhibitors/mimics or 100 nM siRNAs and 6 μL of DharmaFECT Duo Transfection Reagent mixed in 2 mL of serum-free medium were used for each transfection. In the co-transfection with psiCHECK reporter plasmids, 25 nM miRNA mimics/inhibitors and 500 ng reporter plasmid DNA were added.

Luciferase reporter assay

Firefly and Renilla reniformis (as internal controls) luciferase activities were measured at 48 h after cell transfection using the dual luciferase assay system (Promega) as reported previously (21).

MTT (thiazolyl blue tetrazolium bromide) cell viability assay

Cells were seeded in quadruplicate at 5,000 cells per well in 96-well plates, before drugs were added at the indicated concentrations after 24 h. At 72 h after drug treatment, MTT/PBS solution was added to the medium to a final concentration of 0.5 mg/ml. After 4 h incubation at 37 °C, MTT-containing medium was removed and 150 μL of 0.04 N HCl/isopropanol was added to each well. The absorbance at 570 nm (test) and 630 nm (reference) was measured on an ELISA plate reader to obtain sample signal (OD₅₇₀-OD₆₃₀), which was then compared to the signal of untreated control wells.

Statistical correlation analysis

The linear dependence between microarray-determined expression levels of TGFB1 and MSH2 was evaluated by the Pearson correlation coefficient. The gene expression data and associated clinical data were extracted from a published 295 BC dataset (25-26).

TGF-β downregulated MSH2 in transformed cells and negatively correlated with MSH2 in primary breast tumors

We identified MSH2 as a downstream target of TGF-β in a previous study using 2D DIGE to quantitatively measure the proteomic changes in TGF-β-treated MCF10A/HER2 cells (18). The MSH2 protein is reduced by >60% at 24 h and 40 h after TGF-β treatment (Fig. 1A). A significant downregulation of MSH2 at both mRNA and protein levels was also observed in BC cell lines MDA231, MCF7 and HCC1954 (Fig. 1B&1C). Consistently, inhibition of TGF-β signaling with LY2109761, an inhibitor of the TGF-β receptors, elevated MSH2 protein level in MDA231 cells (Fig. 1C). To explore if there is a correlation between TGF-β1 and MSH2 in primary cancer tissue, we analyzed the expression levels of the two genes in a previously reported 295 BC dataset (25-26). A strong negative correlation (p<0.001) was observed between the microarray-determined TGF-β1 and MSH2 expression levels in the whole dataset. When stratifying patients by grade, the linear association is significant among patients with grade 1 (p=0.001) and grade 2 (p=0.002) tumors. When stratifying patients by stage, the correlation is only significant among patients with stage 1 (p=0.001) and stage 2a (p=0.001) tumors (Table 1), suggesting the downregulation of MSH2 by TGF-β may occur during early cancer development.

TGF-β downregulated MSH2 in transformed cells through inducing miR-21

One mechanism to downregulate gene expression at post-transcriptional level is through miRNAs, which are frequently dysregulated in human cancers (27-28). These naturally-occurring non-coding small RNA molecules play crucial functions in cells by base pairing to the 3’UTR of target mRNAs, resulting in mRNA degradation or translation inhibition. To determine if the downregulation of MSH2 involved miRNA-mediated post-transcriptional control, we interrogated the 3’UTR sequence of human MSH2 (hMSH2) gene transcript using TargetScanHuman 5.1 (www.targetscan.org) and miRDB (www.mirdb.org). Both programs predicted a potential miR-21 targeting site at position 33~39, which was highly conserved among various species (Fig. 2A, top panel). We then cloned this putative miR-21 binding region as well as a mutated version downstream to a luciferase reporter gene in psiCHECK vector, and transfected 293 cells with these constructs or a control vector containing a scrambled sequence. Co-transfection with a miR-21 mimic, but not a control miRNA, efficiently suppressed expression of the luciferase reporter followed by the wild-type MSH2 3’UTR sequence, but not the mutated or scrambled sequences (Fig. 2A, middle panel). Transfection of a miR-21 hairpin inhibitor increased MSH2 protein levels in various cell lines (Fig. 2C), consistent with the suppressive function of miR-21 on MSH2 expression. TGF-β has been reported to induce the maturation of miR-21 through Smad2/3, which bind to the primary transcript of miR-21 by interacting with the DROSHA miRNA processing complex and facilitate its function (3). We thereby speculated that TGF-β downregulated MSH2 by inducing miR-21-mediated post-transcriptional repression. Quantitative RT-PCR indicated that the level of miR-21 was indeed increased by TGF-β in various cell lines tested (Fig. 2B). In BC cells, treatment with TGF-β suppressed both MSH2 expression (Fig. 2C) and expression of luciferase reporter containing the miR-21 binding site in MSH2 3’UTR (Fig. 2A, bottom panel). These effects were impaired by transfection of the miR-21 inhibitor (Fig. 2C&2A). However, in non-transformed MCF10A cells, TGF-β increased MSH2 protein level, although it was able to induce miR-21 in these cells (Fig. 2B&2C). Compared to MCF10A/HER2 and BC cells, MCF10A cells expressed a lower level of miR-21 and a relatively high level of MSH2 (Fig. 2D).

TGF-β upregulated MSH2 in normal cells through Smad-mediated, p53-dependent transcriptional activation

It was intriguing that in sharp contrast to the downregulation of MSH2 in BC cells, TGF-β upregulated both MSH2 mRNA and protein levels in non-transformed MCF10A cells and primary HMEC (Fig. 3A). This effect was absent in MCF10A/HER2 cells but could be partially restored by HER2 inhibition using lapatinib, a small molecule inhibitor of HER2 kinase (Fig. 3A). Searching of the upstream promoter sequence of hMSH2 indicated a potential Smad binding element (SBE) located 14 bps upstream of a reported p53 binding site at -154~-174 bps (Fig. 3B, top panel). We previously reported that TGF-β/Smad and p53 synergize to activate the expression of tumor suppressor gene maspin, which requires both the p53 sites and adjacent SBEs in maspin promoter (21). P53 can interact with Smad2 (29), providing physical evidence for a functional synergy between Smads and p53. We thereby cloned the hMSH2 promoter region containing both SBE and p53 site to construct an hMSH2-luciferase reporter. Transfection assays in MCF10A cells indicated that TGF-β, or expression of Smad2/4 or p53, activated wild-type (w.t.) hMSH2 promoter activity, whereas the promoter reporter containing mutated SBE exhibited an impaired response to TGF-β and Smad2/4, but not to p53 (Fig. 3B).

We also observed that the basal activity of transfected w.t. hMSH2 promoter was significantly lower in MCF10A/HER2 cells than that in MCF10A cells (Fig. 3C, left panel). This suggested a lower activity of endogenous hMSH2 promoter in HER2-transformed cells, and was consistent with the lower MSH2 mRNA and protein levels observed in these cells (Fig. 3A). Because HER2 overexpression was reported to induce downregulation of p53 in mammary epithelial cells (30), we examined if modulation of p53 expression/activity in transfected cells influenced MSH2 expression. Knockdown of p53 in MCF10A cells using small interfering (si)RNA markedly decreased basal hMSH2 promoter activity and abolished the induction by TGF-β. Consistently, restoration of p53 in MCF10A/HER2 cells using nutlin-3, a small molecule that induces p53 stabilization by inhibiting MDM2-dependent p53 degradation (31), elevated hMSH2 promoter activity and restored TGF-β's effect (Fig. 3C, left panel). MDA231 cells harbor a point mutation in codon 280 (G/A) of the p53 gene. TGF-β failed to induce hMSH2 promoter activity in transfected MDA231 cells, but this functional deficiency was rescued by cotransfection with a w.t. p53 plasmid (Fig. 3C, right panel). ChIP assay further confirmed the binding of Smads and p53 to the promoter region, but not a coding region of MSH2 in MCF10A cells. This binding was significantly induced by TGF-β (Fig. 3D). Thus, the Smad-mediated MSH2 promoter activation by TGF-β is dependent on an intact p53 function, and is paralyzed in some cancer cells due to p53 inactivation.

The TGF-β/miR-21/MSH2 axis regulated cancer cell response to DNA-damaging agents

MSH2 deficiency has been implicated in the tolerance to DNA damages and chemoresistance. To determine if TGF-β contributes to resistance to DNA-damaging drugs by suppressing MSH2 in cancer cells, we tested the response of MDA231 to cisplatin, methyl methanesulfonate (MMS) and doxorubicin in the presence and absence of TGF-β. Exposure to TGF-β for 24 h increased cell viability upon treatment with these DNA-damaging agents (Fig. 4A). In contrast, transfection of the miR-21 hairpin inhibitor further enhanced the inhibitory effect of cisplatin in MDA231 cells (Fig. 4B). Using a MDA231-derived cell line stably expressing shRNA against MSH2 under the control of a doxycycline-regulated promoter, knock down of MSH2 induced resistance to both cisplatin and doxorubicin, but not docetaxel, a chemotherapy drug targeting the microtubule (Fig. 4C).