PTBP3-Mediated Regulation of ZEB1 mRNA Stability Promotes Epithelial–Mesenchymal Transition in Breast Cancer

Breast cancer is the most common cancer among women worldwide. Distant metastases are the cause of about 90% of deaths in breast cancer (1). The epithelial-to-mesenchymal transition (EMT) process, first described in embryogenesis, is characterized by changes in cell morphology, behavior, and plasticity (2, 3). The activation of EMT program depends on a diverse array of proteins, including transcription factors, cell signaling regulators, and secretory factors, as well as regulation by long noncoding RNAs (lncRNA). EMT is widely accepted as an essential step for fueling tumor initiation and metastatic spread (4).

EMT is associated with decreased expression of epithelial markers, such as E-cadherin, and increased expression of mesenchymal markers (5). E-cadherin is repressed at the transcription level by multiple EMT inducers, including Snail, Slug, Twist, ZEB1, and ZEB2 (6). ZEB1 can suppress E-cadherin expression by binding to the E-box region of CDH1 gene promoter (7, 8), thereby acting as a master inhibitor of EMT (9). Recent studies have shown that Slug can directly activate ZEB1 at the transcriptional level (10). However, ZEB1 expression is posttranscriptionally repressed by several miRNAs, including mir-200 family, mir-205 (11), and mir-216a (12). It is therefore generally agreed that ZEB1 is an important activator of EMT and may serve as a moderator of breast cancer stem cells (BCSC; ref. 13).

Recently, increasing evidence suggests that diverse RNA-binding proteins (RBP) are involved in cancer process. PTBP1 is a member of RBPs that interacts with regulatory RNAs to regulate mRNA splicing (14), stability (15), localization (16), and translation (17). PTBP1 has two paralogs, PTBP2 (nPTB; neural PTB) and PTBP3 (ROD1; regulator of differentiation 1), with a similar protein architecture and containing four RNA recognition motifs (18). The functions of PTBP1 and PTBP2 have been well studied, whereas the role of PTBP3 has been relatively neglected in a long time. Recent studies have shown that PTBP3 plays a role in nonsense-mediated mRNA decay and functions as a splicing repressor (19). Considering the multiple functions of PTBP1 and PTBP2 in various biological processes and the similar structure of PTBP3 with PTBP1/2, we expected that PTBP3 would have an important nonredundant role in cellular processes.

In this study, we have extensively analyzed the function of PTBP3 using cell-based approaches, mouse xenograft models, and IHC-based human breast cancer correlation studies. The results have uncovered previously unrecognized roles for PTBP3 in EMT and breast cancer progression. We demonstrated that high expression of PTBP3 in cancers was confirmed to be a predictor of poorer survival of breast cancer patients. PTBP3 acted as an inducer of EMT program and promoted cell migration, invasion, proliferation, and breast cancer stemness acquisition. In addition, knockdown of PTBP3 was testified to repress tumor growth and metastasis in vivo. Finally, the important role of ZEB1 for PTBP3-induced EMT was revealed. In mechanism, PTBP3 regulated ZEB1 expression by preventing ZEB1 mRNA degradation through binding to the 3′UTR region of ZEB1 mRNA.

MCF10A, MCF7, MDA-MB-231, BT474, and HEK293T cell lines were obtained from ATCC and immediately expanded and frozen down as master stocks. A vial of working stock was always used to start a new experiment, and the cells were replenished with another working stock after being cultured for no more than 6 weeks. In addition, these cell lines were authenticated by short tandem repeat profiling by ATCC or Genewiz, CN. MCF10A cells were cultured as described previously (20). MCF7 and BT474 cells were cultured in RPMI1640 medium supplemented with 10% FBS. MDA-MB-231 cells were cultured in L15 medium with 10% FBS at 37°C without CO₂. HEK293T cells were cultured in DMEM medium containing 10% FBS.

For siRNA transfections, cells were transfected with 20 nmol/L siRNA (Shanghai GenePharma) by siLentFect (Bio-Rad Laboratories, Inc.). For plasmid transfections, cells were transfected using Lipofectamine 2000 (Life Technologies), according to the manufacturer's instructions. The siRNAs and shRNA sequences are described as follows: siPTBP3#1: CCCAGUAAAUGCACAUUAUTT, siPTBP3#2: GCCCUGUGCUUCGAAUAAUTT, siZEB1#1:GGAUCAGAAUGACUCUGAUTT, siZEB1#2: GGUCUUAUUCUCAACACAUTT, siPTBP1: GCCUCAACGUCAAGUACAATT, shCtrl-For: CCGGCCTAAGGTTAAGTCGCCCTCTCGAGAGCGAGGGCGACTTAACCTTAGGTTTTTG, shPTBP3#1-for: CCGGACCAGGAAATTCTGTTCTACTCTCGAGAGTAGAACAGAATTTCCTGGTTTTTTG, shPTBP3#2-for: CCGGCAGAGACTTCACTCGCTTAGACTCGAGTCTAAGCGAGTGAAGTCTCTGTTTTTG.

The vector pCMV6-XL5-PTBP3 was a kind gift from Dr. Luis F. Congote, McGill University Health Centre (Montreal, Canada; ref. 21). PTBP3 cDNA was cloned by PCR from pCMV6-XL5-PTBP3 using TransStart FastPfu Fly DNA Polymerase Kit (TransGen Biotech) and inserted into pCDH1-CMV-MSC-EF1-GFP-Puro lentivirus vector (System Biosciences) at ECOR1/BamH1 sites. PTBP3 shRNA sequences were cloned to the vector pLko.1 at Age1/ECOR1 sites. PTBP3 overexpression and knockdown lentiviruses were generated by cotransfecting 293T cells with the other two packing vectors pMD2G and psPAX and concentrated as described previously (20). Stable cell lines overexpressing or lacking PTBP3 were generated by being infected with lentivirus and selected with 2 μg/mL puromycin for about 2 weeks. The vectors pCDNA3-Flag-MS2bp and pCDNA3-12 × MS2bs were provided by Dr. Xiaofei Zheng (Academy of Military Medical Sciences, Beijing, China). ZEB1 3′UTR was cloned by PCR and inserted into pCDNA3-12 × MS2bs between ECOR1 and Not1 sites. Primer sequences for vector cloning are available in Supplementary Experimental Procedures.

Immunofluorescence and Western blotting were done as described previously (20). Primary antibodies against the following proteins were used: E-cadherin (BD Biosciences, 610181), N-cadherin (BD Biosciences, 610920), vimentin (BD Biosciences, 550513), fibronectin (BD Biosciences, 610077), ZEB1 (Santa Cruz Biotechnology, sc-25388), ZEB2 (Santa Cruz Biotechnology, sc-271984), AGO2 (Santa Cruz Biotechnology, sc-53521), PTBP3 (Santa Cruz Biotechnology, sc-100845), PTBP1 (Santa Cruz Biotechnology, sc-16547), GAPDH (Santa Cruz Biotechnology, sc-271984), PKM1 (SAB, #21577), PKM2 (SAB, #21578), C-myc (Genescript, A00704), and Flag (Sigma, F1804). Secondary antibodies included horseradish peroxidase (HRP)-goat anti-mouse, HRP-goat anti-rabbit, and HRP-rabbit anti-goat (ZSGB-BIO). Protein bands were detected by Tanon 5200 automatic chemiluminescence imaging analysis system using ECL reagent (Tanon).

CCK-8 assay was applied to measure the cell proliferation according to the Cell Counting Kit-8 manufacturer's protocol (Dojindo). Colony formation assay was performed in 60-mm dishes in which 1,000 cells were seeded and cultured for 2 weeks. Colonies were counted manually after staining with 0.1% crystal violet. Migration assays and Matrigel invasion assays in breast cancer cell lines were conducted as described previously (20), with minor modification. Cells were treated with serum starvation overnight, and then migration and invasion assays were performed as described previously.

TRIzol reagent was used to extract RNA from cells and tissues, and the cDNA was generated with random primers (Promega) using the Reverse Transcription System (Promega). Real-time PCR was carried out on ABI-7500 using SYBR Green Real-Time PCR Master Mix (Vazyme Biotech). GAPDH was used for normalization of qRT-PCR data. Primer sequences are available in Supplementary Experimental Procedures.

Immunoprecipitation was carried out as described previously (22). Briefly, MCF7 lysates were prepared in immunoprecipitation lysis buffer (20 mmol/L Tris-Cl, pH 8.0, 10 mmol/L NaCl, 1 mmol/L EDTA, 0.5% NP-40) containing a protease inhibitor cocktail (Sigma). Two micrograms of cell extracts was precleared with 50 μL protein A/G-agarose (Santa Cruz Biotechnology) at 4°C for 2 hours, and the supernatant was incubated with corresponding antibodies with gently shacking at 4°C overnight, followed by the addition of 50 μL of protein A/G-agarose for another 2 hours. The beads were washed and then resuspended in 60 μL of 1× loading buffer and boiled for 5 minutes, and followed with Western blot detection. RNA pull-down assay was carried out using Flag-MS2bp-MS2bs–based pull-down assay as described previously (20), with minor modification. Specifically, pcDNA3-FlagMS2bp and pcDNA3-ZEB1-3′UTR-MS2bs (or mock vector) were cotransfected with pCMV6-XL5-PTBP3 to 293T cells, and the cells were harvested after 48 hours. About 2 × 10⁷ cells were lysed in lysis buffer (20 mmol/L Tris-Cl, pH 8.0, 10 mmol/L NaCl, 1 mmol/L EDTA, 0.5% NP-40) supplemented with RNasin (80 U/mL, NEB) and a protease inhibitor cocktail (Sigma). A total of 2.5 μg monoclonal anti-Flag(R) M2 antibody (sigma) was added to each binding reaction tube and incubated at 4°C for 4 hours, followed by addition of 50 μL of protein A/G-agarose for another 2 hours. Beads were washed five times with the lysis buffer and boiled in 1× loading buffer for 5 minutes. The retrieved proteins were subjected to SDS-PAGE and detected by Western blot analysis.

A series of 418 human breast tumor specimens were obtained from the First Affiliated Hospital of Nanjing Medical University (Nanjing, China). The clinicopathologic information of patients was obtained from the archive of the pathology department and confirmed by the medical record of the hospital, and informed consent was obtained from all patients. The patient studies were conducted in accordance with Declaration of Helsinki. Five-year clinical follow-up results were available for 123 patients. The use of these specimens and data for research purposes were approved by the Ethics Committee of the Hospital. For IHC, as described previously (23), tissues were stained overnight at 4°C with a 1:100 dilution of PTBP3 antibody (Santa Cruz Biotechnology, sc-100845) or with a 1:200 dilution of ZEB1 antibody (Santa Cruz Biotechnology, sc-25388).

The ZEB1-3′UTR fragment was cloned by PCR and inserted into the luciferase reporter vector psiCHECK2 (Promega) between Xho1 and Not1 sites. Experiments were performed as described previously (20). Primer sequences for vector cloning are available in Supplementary Experimental Procedures.

Experiments were performed as described previously (20).

BALB/c nude mice (4–5 weeks old) were purchased from Beijing HFK Bio-technology. All animal experiments were approved by the Animal Care Committee of the Xuzhou Medical University, Xuzhou, China. Animal experiments were performed as described previously (20).

Statistical analyses were carried out with SPSS 20.0 software. The association between PTBP3 staining and the clinicopathologic parameters of the breast cancer patients were evaluated by a χ² test. The Kaplan–Meier method and log-rank test were used to evaluate the correlation between PTBP3 expression and patient survival. Pearson and Spearman correlation coefficients were used to calculate the correlation between PTBP3 and ZEB1 in tissue microarray (TMA). The unpaired t test was used to determine the statistical significance of differences between groups. Data were presented as mean ± SD. P < 0.05 was considered statistically significant.

To examine the expression of PTBP3 in breast cancer tissue samples, IHC was carried out in TMA slides (Fig. 1A). The correlation between PTBP3 expression and clinicopathologic characteristics in 418 breast cancer samples was analyzed (Supplementary Table S1). Compared with histology grade 1 and 2, the PTBP3 signal was dramatically increased in histology grade 3 (P = 0.004, χ² test, Fig. 1B). In addition, high PTBP3 expression is positively correlated with ER negative (P = 0.002, χ² test, Fig. 1C) and PR negative (P = 0.040, χ² test, Fig. 1D). Interestingly, high PTBP3 expression was significantly correlated with lymph node metastasis (P = 0.008, χ² test, Fig. 1E) and lymph node metastasis numbers (P = 0.004, χ² test, Fig. 1F). The results of Kaplan–Meier survival analysis revealed that high PTBP3 levels are correlated with poor 5-year overall survival (P < 0.001, χ² test, Fig. 1G) and disease-specific patient survival (P < 0.001, χ² test, Fig. 1H).

Next, we addressed whether PTBP3 expression was an independent prognostic factor for breast cancer. The univariate Cox regression analyses revealed that PTBP3 expression was an independent prognostic marker for breast cancer patients overall survival (HR, 4.226; 95% confidence interval (CI), 2.406–7.422; P < 0.001) and disease-specific survival (HR, 4.989; 95% CI, 2.269–10.970; P < 0.001; Supplementary Table S2). In multivariate Cox regression analysis (Supplementary Table S3), we found that PTBP3 expression was also an independent prognostic marker for both 5-year overall survival (HR, 3.768; 95% CI, 2.014–7.051; P < 0.001) and disease-specific survival (HR, 3.700; 95% CI, 1.495–9.158; P = 0.005).

To further evaluate the clinical link between PTBP3 and breast carcinomas, PTBP3 mRNA expression level was analyzed with the Oncomine database (24). Data from 29 (total 29) breast cancer cohorts showed higher mRNA level of PTBP3 in mammary carcinomas than in normal breast tissues (Supplementary Table S4), strongly suggesting that PTBP3 expression was positively associated with breast cancer. Further analysis indicated that the PTBP3 expression was closely correlated with a very poor clinical outcome (Supplementary Table S5). The analysis of PTBP3 mRNA expression levels in a TCGA dataset containing 817 breast tumors (25) indicated that high PTBP3 expression was positively correlated with a poor overall survival and disease-free survival (Supplementary Fig. S1A and S1B). In summary, results from both database and our clinical specimens have suggested PTBP3 expression as a predictor of the poor clinical outcome in breast cancer.

To assess the role of PTBP3 in EMT, a MCF10A-based cell line was created to stably express PTBP3 at levels comparable with those observed in breast cancer cell lines and breast tumor tissues (Fig. 2A). Forced expression of PTBP3 caused MCF10A to undergo EMT, as revealed by the loss of cell–cell adhesion and apical–basal polarity, as well as by the gain of mesenchymal cell phenotype (Fig. 2B). Consistent with these findings, ectopic overexpression of PTBP3 induced the downregulation of epithelial markers (E-cadherin and β-catenin) and the upregulation of mesenchymal markers (vimentin, fibronectin, and N-cadherin) at the protein levels (Fig. 2C and D). Real-time PCR experiments revealed higher levels of mRNA expression, suggesting a predominantly transcription effect (Fig. 2E). Similar to MCF10A cells, overexpressed PTBP3 in MCF7 cells also promotes EMT program; in contrast, decreased expression of PTBP3 in MCF7, MDA-MB-231, and BT474 cells by siRNA or shRNA-mediated knockdown was found to elevate the levels of the epithelial markers and to diminish the mesenchymal markers (Fig. 2F; Supplementary Fig. S2A–S2E). Taken together, these results suggested a role for PTBP3 as a driver of EMT in breast epithelial cells.

The BCSC hypothesis suggests that breast cancer is derived from a single tumor-initiating cell with stem cell–like properties (26). BCSCs are defined as a subpopulation of cells with a CD44^high/CD24^low phenotype. Previous evidence has pointed out that cancer cells undergoing EMT frequently exhibited stem cell–like characteristics, including CD44^high/CD24^low phenotype and ability to form mammosphere. We tested whether PTBP3 is associated with BCSCs. By suspension culture, MCF7 and MDA-MB-231 cell lines were employed to acquire tumorspheres that might enrich BCSCs. We observed that PTBP3 protein level was dramatically increased in tumorspheres in comparison with that observed in adherent breast cancer cells (Fig. 3A). In addition, FACS analysis demonstrated that MCF10A and MCF7 cells overexpressing PTBP3 favored generation of CD44^high/CD24^low stem cell–like subpopulation (Fig. 3B; Supplementary Fig. S3A and S3B). Conversely, silencing PTBP3 in MDA-MB-231 cells partly decreased CD44^high/CD24^low subpopulation (Supplementary Fig. S3C and S3D). Furthermore, overexpression of PTBP3 increased formation of mammospheres by both size and number (Fig. 3C and D). These findings are consistent with a hypothesis suggesting a role for PTBP3 in supporting BCSCs.

PTBP3 has been implicated in promoting cell proliferation in gastric cancer cells (27), and inhibition of PTBP3 induced apoptosis and cell arrest in gastric cancer cells (28). Recent data also showed PTBP3 induced migration (29). We have confirmed and extended these results. Compared with controls, MCF10A and MCF7 cells overexpressing PTBP3 exhibited significantly higher cell proliferation rates (Fig. 4A and B). Conversely, MCF7 and MDA-MB-231 cells depleted of PTBP3 displayed reduced rates of cell growth when compared with controls (Fig. 4B; Supplementary Fig. S3E). In addition, colony formation in MCF7 and MDA-MB-231 cells is positively correlated with the expression levels of PTBP3 (Fig. 4C and D; Supplementary Fig. S3F and S3G). Similar overexpression and silencing experiments revealed a positive role for PTBP3 in promoting cell migration and invasion in MCF10A and MCF7 cells (Fig. 4E and F; Supplementary Fig S3H and S3I). These results demonstrate requirements for PTBP3 in supporting breast cancer cell proliferation, migration, and invasion.

Previous studies showed that RBPs (e.g., PTBP1 and IMP2) were involved in miRNA-mediated gene decay (30–32). Recently, PTBP3 functions in mRNA decay and RNA splicing (19) were also reported. They might work in a similar way as PTBP1. We found PTBP3 was mainly located in nucleus, but also expressed in cytoplasm (Supplementary Fig. S4), implying PTBP3 could function both in nucleus and cytoplasm.

To investigate whether PTBP3 impacts the expression of EMT inducers, we performed real-time PCR analysis in MCF10A and MCF7 cells. Of the five well-characterized EMT inducers, the mRNA levels of ZEB1 and ZEB2 were increased or decreased, corresponding to PTBP3 overexpression or depletion, respectively (Fig. 5A and B). Subsequent immunoblot experiments confirmed this correlation on the protein levels (Fig. 5C).

We next determined whether PTBP3 is associated with ZEB1-3′UTR. For this purpose, ZEB1-3′UTR was cloned into a plasmid containing a 12*MS2bs sequence that binds to MS2bp. HEK293T cells were cotransfected with the first plasmid expressing Flag-MS2bp and the second plasmid that expresses an RNA species containing 12*MS2bs or 12*MS2bs-ZEB1-3′UTR. Anti-Flag immunoprecipitation then followed to isolate Flag-MS2bp/RNA complexes. Immunoblot experiments revealed the presence of PTBP3 and AGO2 (a core component of RISC complex) at levels more abundantly in cells expressing 12*MS2bs-ZEB1-3′UTR than in cells expressing 12*MS2bs only (Fig. 5D and E). Further immunoprecipitation experiments revealed the interactions between PTBP3 and AGO2 (Fig. 5F and G). These findings are consistent with a hypothesis suggesting direct association of PTBP3 with ZEB1-3′UTR. PTBP3 may bind to the AGO2-containing RISC complex to prevent RISC-mediated mRNA degradation.

To provide more direct evidence for PTBP3/ZEB1-3′UTR–mediated transcriptional activity, we developed a psiCheck-2 reporter plasmid containing the ZEB1-3′UTR sequence cloned downstream of Renilla LUC and firefly LUC driven by HSV-TK promoter. We observed Rluc activity in cells overexpressing PTBP at levels significantly higher than those seen in the control cells (Fig. 5H). This result suggests an ability of PTBP3 to activate the expression of ZEB1 by binding to, and enhancing the stability of ZEB1 3′UTR.

In addition, in 259 breast carcinoma samples, the relative protein expression levels of PTBP3 were concordant to ZEB1 (Fig. 5I). Similar observations were made in different tumor subtypes (Supplementary Fig. S5A–S5F). These data suggest PTBP3-mediated activation of ZEB1 in a variety of tumors.

To assess the requirement for ZEB1 in EMT that is initiated by PTBP3, we silenced ZEB1 by siRNA in MCF10A cells overexpressing PTBP3. ZEB1 silencing markedly reduced the levels of fibronectin and partially restored expression of E-cadherin and β-catenin (Fig. 6A). ZEB1 knockdown suppressed the effects of PTBP3 overexpression on cell migration and invasion (Fig. 6B and C). Furthermore, ZEB1 depletion reduced formation of BCSC-like subpopulation cells with CD44^high/CD24^low phenotype (Fig. 6D). These results demonstrated that PTBP3 promoted EMT in breast cancer via a mechanism by activating ZEB1 expression.

To evaluate a role for PTBP3 in tumor growth in vivo, we employed xenograft model using nude mice subcutaneously transplanted with MDA-MB-231 cells that were mock treated or depleted of PTBP3. As shown in Fig. 7A and B, PTBP3 knockdown cells formed smaller tumors than the control. Also, the expression levels of PTBP3 and ZEB1 were low in the PTBP3 knockdown subcutaneous tumors (Fig. 7C). Next, we investigated the effects of PTBP3 in breast cancer cell lung metastasis by the tail vein injection model. Briefly, MDA-MB-231-shPTBP3/shCtrl cells were injected via lateral tail vein to 6-week-old nude mice, and the mice were executed to analyze the lung metastasis status 9 weeks later. As shown in Fig. 7D, although multiple large metastatic foci in lung were found in the control group, fewer were observed in the samples containing MDA-MB-231-shPTBP3 cells (Fig. 7E). These findings provided compelling evidence for the ability of PTBP3 to confer tumorigenesis and metastasis in vivo.

In this work, we provide a comprehensive set of data suggesting critical roles for PTBP3 in EMT and in breast cancer progression. First, PTBP3 was shown to support the proper function of the EMT transcription program, acting to downregulate epithelial markers (such as E-cadherin and β-catenin), but upregulating mesenchymal markers (such as vimentin, fibronectin, and N-cadherin; Fig. 2). Second, it was revealed that PTBP3 mediates the formation of a subpopulation of breast cancer cells named mammospheres that display a CD44^high/CD24^low phenotype (Fig. 3), which has been defined as a stem cell–like property. Third, evidence was presented, demonstrating a supporting role for PTBP3 in breast cancer cell proliferation, migration, and invasion (Fig. 4). Fourth, the results of mouse xenograft experiments revealed a role for PTBP3 in promoting breast cancer tumorigenesis and metastasis in vivo (Fig. 7). Finally, it was found that in human breast cancer patients, high PTBP3 expression correlates with clinicopathologic parameters and poor 5-year overall survival (Fig. 1).

At mechanistic level, PTBP3 was shown to bind to ZEB1 3′UTR (Fig. 5A–E) presumably via its intrinsic RNA-binding activity. The results of reporter assay suggest that this binding activity enhances gene expression (Fig. 5H). These findings suggest that the binding of PTBP3 to ZEB1 3′UTR may have a protective role, resulting in stabilization of ZEB1, a critical EMT inducer. Consistent with these results, ZEB1 was found to be critical for EMT that was driven by PTBP3 overexpression (Fig. 6).

PTBP1 is one of the most investigated RNA-binding protein invertebrates. It plays a comprehensive role in cell differentiation (33), apoptosis (34), cancer initiation, and progression (35, 36). Previous study showed PTPB3 protein level was increased in lung squamous cell carcinomas (37). In this study, we found a significant positive correlation between increased PTBP3 and poor 5-year survival, lymph node metastasis in breast cancer patients (Fig. 1; Supplementary Table S1). PTBP3 was reported to play a role in the regulation of cell proliferation (21), differentiation (38), and migration (29). In our work, we first demonstrated that PTBP3 was required for cell proliferation, migration, and invasion in breast cancer cells (Fig. 4). We also noticed PTBP3 has little effects on the expression of C-myc, PKM1, and PKM2 (Supplementary Fig. S6A), which are known targets of PTBP1 and involved in cancer metastasis (17, 39, 40). In addition, a previous study indicated that only very few PTBP1 targets responded to PTBP3 knockdown in K562 cells (41). Thus, despite sequence conservation (70% identity of amino acid sequence), PTBP1 and PTBP3 may play nonredundant roles with distinct sets of targets. Interestingly, we noticed that PTBP1 was decreased in PTBP3-overexpressed MCF10A cells, and although PTBP1 was repressed by PTBP3 overexpression, PTBP1 silencing could partly repress PTBP3-induced EMT in MCF10A (Supplementary Fig. S6B), but PTBP1 has little effects on the mRNA expressions of ZEB1 and ZEB2 (Supplementary Fig. S6C). These results are in agreement with the previous reports for the role of PTBP1 in EMT (20), and it suggested that PTBP1 may play a role in EMT transition in a different way. Herein, we observed that PTBP1 was downregulated in PTBP3-induced EMT; however, PTBP3 did not affect the expression of PTBP1 targets such as C-myc. It seems a discrepancy emerged in our data. Although C-myc could be regulated by various factors such as Notch1, Stat3, and HIF2α (42–44), it suggested that PTBP1 may not play a critical role in regulating the expressions of C-myc and other target genes in PTBP3-induced EMT in breast cancer. Future investigations will focus on the balance of PTBP3 versus PTBP1 in physiologic conditions. We anticipate that a change in the balance of PTBP3 versus PTBP1 activity will play an important role in biological process.

EMT is accompanied by massive changes in cell behavior, such as cell proliferation, cell differentiation, cell migration, and cell adhesion (45). In this study, we showed ectopic overexpression of PTBP3-induced mammary EMT accompanied with the acquisition of enhanced migration and invasion ability (Fig. 4). In addition, PTBP3 protein level was elevated in suspension-cultivated breast cancer cells, and the PTBP3 overexpression promoted the generation of CD44^high/CD24^low subpopulation cells (Fig. 3). Thus, PTBP3 might be essential for nonstem cells to acquire tumor-initiating capacity and function as a mediator of breast cancer cell stemness. Our conclusion is consistent with the idea that the EMT generates cells with CSC-like activity (46, 47).

ZEB1 is a well-characterized EMT inducer and functions as a key mediator in maintaining the stem cell properties. ZEB1 was proven to be essential for PTBP3-induced EMT in breast cancer here. ZEB1 was upregulated upon PTBP3 overexpression, and ZEB1 silencing partly reversed the PTBP3-induced EMT characters. ZEB1 and ZEB2 were increased at both mRNA and protein levels upon PTBP3 overexpression, while PTBP3 had very little influence on the expression of other EMT-TFs, such as Snail, Slug, and Twist. ZEB1 and ZEB2 were regulated by mir-200 family and formed a ZEB/miR-200 feedback loop (48, 49). PTBP3 might indirectly influence ZEB/mir-200 feedback loop through regulating ZEB1. Whether PTBP3 directly regulates ZEB2 expression requires further investigation.

In conclusion, we have established an important functional role for PTBP3 in breast cancer and have also demonstrated that PTBP3 expression is positively correlated with poor 5-year survival and lymph node metastasis in breast cancer patients. Our clinical analyses point the prospect of PTBP3 as a novel biomarker for prognosis and a potential molecular therapeutic target for highly aggressive breast cancer. We also reveal that PTBP3 regulates an important cell-fate determination event, namely, EMT. In mechanism, PTBP3 regulates ZEB1 by preventing ZEB1 3′UTR degradation. Although more details still need to explore the functions of PTBP3 in a physiologic condition, our data uncover the importance of PTBP3 in breast cancer progression.

No potential conflicts of interest were disclosed.

Conception and design: P. Hou, J. Bai, J. Zheng

Development of methodology: P. Hou, L. Li

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): P. Hou, L. Li, F. Chen, H. Liu

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): P. Hou, L. Li, F. Chen, Y. Chen

Writing, review, and/or revision of the manuscript: P. Hou, J. Li, J. Zheng

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): P. Hou, L. Li, F. Chen

Study supervision: P. Hou, J. Bai, J. Zheng

This work was supported by funding from the National Natural Science Foundation of China (81502280) to Pingfu Hou, 81672845 and 81472663 to Jin Bai; The Research Foundation of Xuzhou Medical University (D2015016) to Pingfu Hou; Education Department of Jiangsu Province (no. 15KJA320006) to Jin Bai. This work also was supported by the Project of Invigorating Health Care through Science, Technology and Education; Jiangsu Provincial Key Medical Discipline (Laboratory); and Jiangsu Provincial Medical Youth Talent Project (QNRC2016776) to Jingjing Li. The authors thank Dr. Luis F. Congote and Xiaofei Zheng for providing the plasmids mentioned in Materials and Methods.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.