Long noncoding RNA PVT1 promotes breast cancer proliferation and metastasis by binding miR-128-3p and UPF1

In this study, blood samples were collected from female patients at the Central Hospital of Wuhan, including 80 BC patients, 70 healthy volunteers, and 70 fibroma patients; for 35 BC patients, paired blood samples (preoperative and postoperative blood in one month after surgery) were collected. None of the patients had received any clinical treatment before sample collection and corresponding tumor specimens were confirmed by histopathological examination. No diseases or injuries were found in the healthy controls. Peripheral blood samples were collected in EDTA-containing tubes and were processed within 4 h by centrifugation at 1000 g for 15 min at 4 °C. The supernatant was then transferred into RNAse-free tubes and stored at − 80 °C. This study was approved by the Medical Ethics Committee of The Central Hospital of Wuhan and all human subject research was performed in accordance with institutional, national, and Declaration of Helsinki requirements.

The human BC cell lines (Hs578t, MCF-7, MDA-MB-231, T47D, Bt549, HCC1806), the noncancerous breast epithelial HBL-100 cells, and the human embryonic kidney (HEK) 293 T cells were purchased from the American Type Culture Collection (ATCC). Cell culture was performed following the recommendations of ATCC. We confirm the authentication of all cell lines used—the full policy and requirements are available in the instructions to authors.

The pSIF–GFP–miR-128-3p precursor plasmid and the precursor control were gifts from Dr. Yong Li in Baylor College of Medicine. miR-128-3p inhibitors and miRNA controls were obtained from GenePharma Technology (Shanghai, China). Small interfering RNAs (siRNAs) targeting PVT1 and a negative control were also obtained from GenePharma Technology (Shanghai, China). Three shRNAs targeting human UPF1 and the shRNA negative control sequence were as follows: shUPF1#1, sense: 5’-GATCCGAGCCACATTGTAAATCATTTCAAGAGAATGATTTA CAATGTGGCTTTTTTTG-3, antisense: 5’-AATTCAAAAAAAGCCACATTGTAAATC ATTCTCTTGAAATGATTTACAATGTGGCTCG-3’; shUPF1#2, sense: 5’-GATCCG GCGAGAAGGACTTCATCATTCAAGAGATGATGAAGTCCTTCTCGCTTTTTTG-3’, antisense: 5’-AATTCAAAAAAGCGAGAAGGACTTCATCATCTCTTGAATGATGAA GTCCTTCTCGCCG-3’; shUPF1#3, sense: 5’-GATCCGGCAGCCACATTGTAAAT.

CATCAAGAGTGATTTACAATGTGGCTGCTTTTTTG-3’, antisense: 5’-AATTCA AAAAAGCAGCCACATTGTAAATCACTCTTGATGATTTACAATGTGGCTGCCG-3’. The shRNA constructs were cloned into Bam HI and Eco RI sites in an RNAi-ready pSIREN-RetroQ vector (Clontech, USA). All DNA constructs were confirmed by Sanger sequencing. Transfections were performed using Lipofectamine® LTX and Plus reagent (Invitrogen, USA) according to the manufacturer’s instructions.

After transfection, Hs578t or MCF-7 cells were seeded into 12-well plates at 500 cells/well and cultured in DMEM with 10% FBS at 37 °C for 2 weeks. Plates were washed twice with PBS, colonies were then fixed with 4% paraformaldehyde for 20 min and stained with 0.1% crystal violet for 30 min. More than 50 cells were counted as a colony.

RIP was performed using a EZ-Magna RIP kit (Millipore, USA) according to the manufacturer’s instructions. In brief, Hs578t or MCF-7 cells were lysed in the complete RIP lysis buffer and then the lysates were incubated with beads-antibody in RIP buffer. Rabbit anti-UPF1 antibody and rabbit IgG control (Millipore, USA) were used. Finally, the co-precipitated RNAs were extracted and detected by qRT-PCR.

Hs578t or MCF-7 cells were seeded in 6-well plates. A wound was made by a sterile 200 μl pipette tip 24 h post transfection. The images (0 h) were taken using a microscope. After incubation at 37 ℃ for 24 h, cells were washed twice with PBS and the wound healing images (24 h) were photographed again.

To investigate the capacities of cell invasion, transwell chamber (Corning, USA) was coated with Matrigel (BD Biosciences, USA). Hs578t or MCF-7 cells (5 × 10⁴) were starved in 200 μL serum-free medium 24 h post transfection and then seeded into the top chamber of each insert (8 μm, Corning, USA). The lower chambers were filled with 600 μL of complete medium. After cultured at 37 ℃ for 24 h, cells migrated to the lower chambers were fixed, stained, and counted.

RNA was extracted using TRIzol reagent (Invitrogen, USA). cDNA was quantified by SYBR green (Applied BioSystems, USA) on the StepOne System (Bio-Rad, USA). U6 was used as the endogenous control for miR-128-3p and GAPDH for PVT1. Each reaction was performed in triplicates and relative expression of target genes was calculated by the 2^–△△Ct method for gene expression in BC cells and the 2^–△Ct method for that in BC plasma. The following primers were used: PVT1 sense (5’-TGAGAACTGTCCTTA CGTGACC-3’) and antisense (5’-AGAGCACCAAGACTGGCTCT-3’); GAPDH sense (5’-GGGAGCCAAAAGGGTCAT-3’) and antisense (5’-GAGTCCTTCC ACGATACCAA-3’).

Western blot analyses were conducted according to the method described previously [12]. The primary antibodies were: rabbit monoclonal anti-E-cadherin (#3195, 1:1000, Cell Signaling Technology), rabbit monoclonal anti-Vimentin (#5741, 1:1000, Cell Signaling Technology), rabbit polyclonal anti-UPF1 (23379-1-AP, 1:1000, ProteinTech), rabbit polyclonal anti-FOXQ1 (23718-1-AP, 1:1000, ProteinTech), mouse monoclonal anti-β-actin (A5316, 1:5000, Sigma-Aldrich), and mouse monoclonal anti-PCNA (sc-25280, 1:1000, Santa Cruz). The blot signal was detected using ECL (GE Healthcare, UK).

For the PVT1 luciferase reporter assay, pRL-TK-PVT1 or pRL-TK-PVT1 mutant vectors and pGL3 control vector (Promega) were co-transfected into HEK293T along with pSIF-GFP-miR-128-3p precursor plasmid using Lipofectamine® LTX and the Plus reagent (Invitrogen).

For miR-128-3p target gene FOXQ1 luciferase reporter assay, pRL-TK-FOXQ1-3΄UTR or pRL-TK-FOXQ1-3΄UTR mutant vectors and pGL3 control vector (Promega) were co-transfected into cells along with pSIF-GFP-miR-128-3p precursor plasmid using Lipofectamine® LTX and the Plus reagent (Invitrogen). Luciferase activity was detected 48 h after transfection using the Dual-Glo luciferase reporter assay system (Promega) according to the manufacturer’s instructions. Experiments were repeated at least three times.

Female athymic 4-week-old BALB/c nude mice were purchased from HFK Bio-Technology. Co., Ltd (Beijing, China). Hs578t (5 × 10⁶ cells) were injected into the mammary fat pad. Tumor volumes were analyzed using the formula V = length × width²/2. When the tumor volumes reached 60 mm³, the mice were randomly divided into a control group (n = 6) and a si-PVT1 group (n = 6). si-PVT1 or the siRNA control (20 μl, 5 nM)) was directly injected into the implanted tumor twice per week. After 3 weeks, all the nude mice were anesthetized by 1% pentobarbital sodium (50 mg/kg) and euthanized through the cervical vertebra acetabular method. The tumors and major organs were fixed in 10%-buffered formalin and embedded in paraffin for immunohistochemistry or hematoxylin & eosin (H&E) staining.

All statistical analyses were carried out using Graphpad Prism 5.0 software. Student’s t-test or one-way analysis of variance (ANOVA) were performed for comparisons between groups. P value < 0.05 indicates statistical significance. *P < 0.05, **P < 0.01, ***P < 0.001.

To explore the expression profile of PVT1 in BC, we first detected PVT1 expression levels in BC cell lines. PVT1 expressions were remarkably higher in BC cells than that in the immortalized mammary epithelial cell line HBL-100 cells (Fig. 1A). Moreover, we tested PVT1 expression levels in plasma samples from 80 BC patients, 70 fibroma patients, and 70 healthy controls. PVT1 levels were significantly increased in BC samples compared with either healthy controls or breast fibroma patients (Fig. 1B). Breast fibromatosis is the most common histologically benign lesion and rarely undergo malignant transformation. The clinical characteristics of the BC patients are listed in Table 1. To further explore the significance of PVT1 expression in BC, we analyzed PVT1 expression in relationship to tumor stages and found that the plasma levels of PVT1 were significantly higher in patients with stage III and IV than those with stage I and II (Fig. 1C). And we also find the expression of PVT1 has no difference among the subtypes (see Additional file 1: Table). In addition, we analyzed 35 paired preoperative and postoperative plasma samples from BC patients, PVT1 expressions were remarkably reduced in the postoperative samples (Fig. 1D). These results suggested that PVT1 is a candidate predictor for BC diagnosis and staging assessment. Because the PVT1 expression was the highest in cell line Hs578t and MCF-7, we chose these two lines for further study.

To investigate the function consequences of PVT1 overexpression on BC, we introduced three specific si-PVT1s into BC cells. si-PVT1 #3 was the most efficient in knocking down PVT1 expression (Fig. 2A, see Additional file 1: Figure A). The colony formation assay showed that knockdown of PVT1 significantly inhibited Hs578t and MCF-7 cell proliferation (Fig. 2B, C). Additionally, the wound healing assay showed that cells transfected with si-PVT1 underwent a slower closing of scratch wound than the control (Fig. 2D, E). Meanwhile, the transwell assay showed that knockdown of PVT1 dramatically decreased Hs578t and MCF-7 cell invasion (Fig. 2F, G). These findings indicate that PVT1 promotes BC cell proliferation, migration and invasion in vitro. We examined protein markers for cancer proliferation, migration and invasion in Hs578t and MCF-7 cells. As shown in Fig. 2H, following the treatment of the si-PVT1, the level of the epithelial marker E-cadherin was upregulated, while that of the mesenchymal marker Vimentin and proliferation-related protein PCNA was downregulated in Hs578t and MCF-7 cells. These results indicate that inhibition of PVT1 suppresses cell proliferation, migration, invasion and EMT in BC.

To further evaluate the effects of PVT1 in vivo, we injected the si-PVT1 or a negative control intratumorally, when the tumors xenografted with Hs578t cells reached an average of 60 mm³ in immunodeficient mice. We found that the volumes of tumors in mice treated with the si-PVT1 were markedly smaller than those in mice injected with the control (Fig. 3A, B). Lung and liver samples were obtained to evaluate tumor metastasis, and the number of metastatic cells in liver and lung were significantly reduced by si-PVT1 (Fig. 3C, D). In addition, immunohistochemistry results (Fig. 3E) showed that the expression levels of Ki-67 and vimentin were much lower and that of E-cadherin was higher in tumors treated with si-PVT1 than that with the control. These results implicate that downregulation of PVT1 significantly reduces tumor growth and metastasis in vivo.

It is known that lncRNAs often function as competing endogenous RNAs (ceRNAs) or molecular sponges to recruit microRNAs (miRNAs), regulating their biological functions [13, 14]. To further explore the mechanism of PVT1 in BC tumorigenesis, we investigated whether miRNAs are involved. By using the online bioinformatics software -starBase (http://​starbase.​sysu.​edu.​cn), we observed that PVT1 contains a potential binding site for miR-128-3p (Fig. 4A). As shown in Fig. 4B, miR-128-3p expression in BC cell lines was lower than that in the immortalized mammary epithelial cell line HBL-100. The plasma level of miR-128-3p was much lower in BC patients than that in fibroma patients and healthy controls (Fig. 4C). We then performed a luciferase reporter assay, in which the Renilla luciferase (Rluc) was upstream of the PVT1 gene in one plasmid and the firefly luciferase as a control in another plasmid; the miR-128-3p binding site in Mut-PVT1 was disrupted. Both plasmids and a miR-128-3p expression plasmid were transfected into 293 T cells. As shown in Fig. 4D, the Rluc activity was significantly reduced with wild-type PVT1 and miR-128-3p. However, the Rluc activity with Mut-PVT1 and miR-128-3p showed no statistically significant changes. We next down-regulated PVT1 expression in Hs578t and MCF-7 cells using siRNAs and found PVT1 down-regulation markedly increased miR-128-3p levels (Fig. 4E). Similar results were also observed in xenografted tumors (Fig. 4F).

To identify target genes of miR-128-3p, we used starBase and TargetScan and found a potential miR-128-3p site in the FOXQ1 3΄UTR (Fig. 5A). FOXQ1 is a transcription factor involved in the EMT, which is associated with tumor growth and metastasis. A luciferase assay showed that miR-128-3p significantly inhibited reporter activity with the FOXQ1 3΄UTR; however, mutation in the putative targeting site in the 3΄UTR resulted in compete abrogation of the repressive effect (Fig. 5B). To examine whether PVT1 regulates FOXQ1 expression, BC cell proliferation and metastasis by directly targeting miR-128-3p, we transfected Hs578t cells with si-PVT1, miR-128-3p inhibitor, or both and evaluated the malignant phenotypes. As shown in Fig. 5C, the colony numbers of Hs578t cells were reduced with si-PVT1 transfected, but were increased with miR-128-3p knockdown. Yet with both miR-128-3p and PVT1 knockdown, the colony numbers were lower than that with miR-128-3p knockdown only. Moreover, knockdown of PVT1 markedly inhibited the wound closure rate and reduced cell invasion; inhibition of miR-128-3p exhibited the opposite effects, yet co-transfection with both inhibitors reversed increased migration and invasion of Hs578t cells mediated by miR-128-3p knockdown (Fig. 5D, E). When PVT1 was knocked down, the protein level of E-cadherin was elevated, whereas PCNA, Vimentin, and FOXQ1 were downregulated. Inhibition of miR-128-3p had opposite effects, and the impact of PVT1 knockdown was partially reversed by co-transfection with both inhibitors (Fig. 5F). These results suggest that PVT1 promotes BC cell proliferation, migration, and invasion, at least partially, by competitively binding to miR-128-3p.

To further study the molecular mechanism underlying PVT1’s role in BC pathogenesis, we identified proteins potentially associated with PVT1 using starBase. Up-frameshift protein 1 (UPF1), a key factor in nonsense-mediated mRNA decay (NMD), is such a protein. UPF1 was upregulated when PVT1 was knocked down in Hs578t and MCF-7 cells (Fig. 6A). UPF1 antibody pulled down PVT1 in a RIP assay (Fig. 6B). To examine whether PVT1 regulates BC cell proliferation and metastasis by binding UPF1, we transfected Hs578t cells with si-PVT1#3, sh-UPF1#2 (sh-UPF1 #2 was the most efficient in knocking down UPF1 expression in Hs578t and MCF-7 cells, see Additional file 1: Figure B), or both, and evaluated the malignant phenotypes. As shown in Fig. 6C, PVT1 knockdown reduced colony formation of Hs578t cells, while UPF1 knockdown increased it. Yet inhibition of both UPF1 and PVT1 reversed the increase in colony formation mediated by UPF1 knockdown. Similar results were observed for migration and invasion (Fig. 6D, E). PVT1 knockdown upregulated E-cadherin, but down-regulated PCNA and vimentin. UPF1 knockdown had opposite effects, and the impact of PVT1 knockdown was partially reversed by concurrent UPF1 knockdown (Fig. 6F). These results suggest that PVT1 promotes BC cell proliferation, migration, and invasion by binding UPF1.