Nonphosphorylatable PEA15 mutant inhibits epithelial-mesenchymal transition in triple-negative breast cancer partly through the regulation of IL-8 expression

The human breast adenocarcinoma cell lines MDA-MB-231 and MDA-MB-468 were obtained from the American Type Culture Collection. MDA-MB-231 and MDA-MB-468 cells were grown in DMEM/F12 (Life Technologies) supplemented with 10% FBS and penicillin/streptomycin and maintained in a humidified incubator at 37 °C containing 5% CO₂. CRISPR-edited PEA-15-KO clones in MDA-MB-231 were established.

Cells were washed with PBS (pH 7.4) and then lysed in lysis buffer [20 mmol/L Na₂PO₄ (pH 7.4), 150 mmol/L NaCl, 1% Triton X-100, 1% aprotinin, 1 mmol/L phenylmethylsulfonyl fluoride, 100 mmol/L NaF, and 2 mmol/L Na₃VO₄] as previously described [5]. Further, western blot analysis was performed as described in Supplementary methods.

As described previously [11], MDA-MB-468 and MDA-MB-231cells were transfected with expression plasmids containing empty vector (pcDNA3-HA), PEA15-AA (pcDNA3-HA-PEA15-AA), or PEA15-DD (pcDNA3-HA-PEA15-DD) using FuGENE transfection reagent according to the manufacturer’s instructions. These constructs were kindly provided by Dr. Mark H. Ginsberg (University of California San Diego, La Jolla, CA).

Colony formation and Mammosphere formation assays were performed as described in Supplementary methods.

As described previously [15], migration assay were performed in triplicate using a 24-well micro-chemotaxis chamber. The assays were further conducted as described in Supplementary methods.

Four-to six-week-old female NOD/SCID mice were used to establish MDA-MB-468 breast cancer xenografts stably expressing empty vector or the PEA15 mutants. Cells were prepared in a 1:1 mixture of PBS and growth factor-reduced Matrigel (BD Biosciences) at 4 × 10⁶ cells in 100 µL, and this cell suspension was injected into the mammary fat pads of mice.

MDA-MB-468 cells (5 × 10⁶ cells in 100 μL Matrigel) were inoculated into the mammary fat pads of NOD/SCID mice. When tumor size reached approximately 200 mm³, either Rb-PEA15-AA or Rb-control was intravenously injected into the mice twice a week for 6 weeks. The mice were treated with equimolar amounts of each protein (27 μM/dose).

Tumor volume (mm³) was calculated, and changes in tumor volumes were tested for statistical significance with the Mann–Whitney test or Student's two-tailed t test.

MDA-MB-231 PEA15-KO cells were used to establish stable cell lines overexpressing empty vector, PEA15-WT, PEA15-AA and PEA15-DD. Female athymic nu/nu mice, age 6 weeks, were purchased (Envigo). MDA-MB-231 PEA15-KO cells stably expressing PEA15 constructs (2 × 10⁶ cells/100 µL PBS) were injected into the tail veins of the mice. At 8 weeks after inoculation, animals were euthanized, and the lung tissues were analyzed for metastasis. The experimental metastasis model will allow us to determine the two late events of the metastatic process, extravasation and organ colonization.

As described previously [15], immunohistochemistry was performed as further described in Supplementary methods.

RT² Profiler Human Cancer Inflammation and Immunity Crosstalk PCR Array (Qiagen) was performed using the Bio-Rad CFX96 cycler [16]. The expression levels were quantified relative to the values obtained for housekeeping genes (ACTB, B2M, GAPDH, HPRT1, and RPLP0). Data were analyzed using Qiagen web-based software (https://​www.​qiagen.​com/​us/​shop/​genes-and-pathways/​data-analysis-center-overview-page/​).

Data are given as mean ± SD. Student’s t test or ANOVA was performed to compare the differences between two or more than two groups. P < 0.05 was considered a statistically significant value.

MDA-MB-468 cells were selected because these cells are known to be tumorigenic in a xenograft model [5] and they demonstrated low endogenous expression of PEA15. To further study the effect of the phosphorylation status of PEA15 on metastasis in vivo, we selected a highly metastatic TNBC cell line, MDA-MB-231, even though it has high expression of PEA15 (Figure S1). To circumvent the effect of endogenous PEA15, the endogenous PEA15 was knocked out in the MDA-MB-231 cells (MDA-MB-231 PEA15-KO) using CRISPR/Cas9. We either transiently transfected or established stable cell lines overexpressing empty vector (PEA15-V), wild-type PEA15 (PEA15-WT), nonphosphorylatable mutant (PEA15-AA), and phosphomimetic mutant (PEA15-DD).

We tested the effects of PEA15 phosphorylation status on cell proliferation and observed no significant difference between cells expressing PEA15-AA and PEA15-DD compared to the vector control cells (Figure S2). Next, we assessed the effect of PEA15 phosphorylation on colony formation ability in TNBC cells. PEA15-AA-expressing cells formed fewer colonies than did the PEA15-WT-expressing or PEA15-DD-expressing MDA-MB-468 and MDA-MB-231-PEA15-KO cells (Fig. 1A). The EMT phenotype is strongly correlated with the ability of cells to form mammospheres, which are an indicator of self-renewal and cancer stem cell (CSC) properties [17]. To determine the role of PEA15 phosphorylation on the CSC-like phenotype, we performed a mammosphere formation assay. As shown in Fig. 1B, PEA15-AA significantly decreased the formation of mammospheres compared to PEA15-WT-expressing or PEA15-DD-expressing cells. Taken together, these results strongly suggest that the unphosphorylated PEA15-AA contributed to the suppression of colony formation and stemness phenotype in TNBC cells.

To study the role of phosphorylation of PEA15 in the TNBC cell migratory phenotype, we performed an in vitro migration assay. Compared with control PEA15-WT and PEA15-DD cells, PEA15-AA cells had effective suppression of migration (Fig. 2A). To assess the effect of phosphorylation status of PEA15 on EMT, the protein levels of EMT markers were analyzed. Overexpression of PEA15-AA in MDA-MB-468 cells decreased the expression of fibronectin and vimentin (a mesenchymal markers) (Fig. 2B). Furthermore, the mRNA levels of EMT-inducing transcription factors, Snail and Slug were significantly decreased in PEA15-AA-expressing cells (Fig. 2C). In contrast, we observed that phosphomimetic PEA15-DD-expressing cells had increased expression of fibronectin and vimentin (Fig. 2B) and Snail and Slug (Fig. 2C). Together, these results indicate that unphosphorylated PEA15 suppresses cell motility and EMT of TNBC cells in vitro. These results suggest that one potential pathway by which PEA15 phosphorylation status regulates breast cancer cell migration is through upregulating or downregulating EMT-related molecules.

We first assessed the anchorage-independent growth of the MDA-MB-468 stable overexpressing clones in soft agar as an indirect test of their tumorigenicity. We observed that the PEA15-AA stable transfectants formed fewer colonies (i.e., were less tumorigenic) than did the PEA15-WT- or PEA15-DD-transfected cells (Fig. 3A, left panel).

To identify the most tumor-suppressive form of PEA15, we examined the effects of PEA15-AA- and PEA15-DD-overexpressing stable MDA-MB-468 cells in vivo by injecting them into the mammary fat pads of NOD/SCID mice. At 8 weeks, the tumors in the mice injected with PEA15-AA were significantly smaller when compared with mice that were injected with PEA15-WT and PEA15-DD. Animals were euthanized at 8 weeks, and the expression levels of EMT markers in the tumors were analyzed with immunohistochemical staining. As expected, expression of mesenchymal marker vimentin was inhibited in PEA15-AA-expressing tumors compared with PEA15-WT or PEA15-DD-expressing tumors (Fig. 3B). The level of epithelial marker E-cadherin was decreased in PEA15-DD-expressing tumors. However, PEA15 phosphorylation status did not affect the expression of proliferation marker Ki-67. Next, we assessed the impact of PEA15 phosphorylation on lung metastasis. Vector, PEA-15-WT-, PEA15-AA-, and PEA15-DD-overexpressing stable MDA-MB-231 PEA15-KO cells were injected into the tail veins of nude mice. At 8 weeks after inoculation, mice were euthanized and analyzed for lung metastasis burden. Figure 3C shows representative lung images that display the metastasis burden of animals injected with vector control, PEA15-WT-expressing, PEA15-AA-expressing, and PEA15-DD-expressing cells. Mice injected with PEA15-AA cells showed much less macroscopic metastasis/colonization in the lungs compared with the vector control and PEA15-DD-expressing groups; presence of tumor was confirmed histologically in group DD after H&E staining of formalin-fixed paraffin lung tissue sections. In addition, as shown in Fig. 3D, the weight of lungs from mice injected with PEA15-AA-expressing cells was lower than those from mice injected with PEA15-WT- or PEA15-DD-expressing cells. These results suggest that phosphorylation of PEA15 may stimulate tumor growth by modulating EMT-related molecules and further enhance metastasis of cancer cells. Additionally, we tested the antitumor efficacy of purified PEA15-AA protein combined with an EGFR-specific repebody (rEgH9) and a translocation domain (TDP) in vivo. The delivery platform is composed of a targeting moiety, translocation domain, and PEA15 derivative. The successfully developed rEgH9 and TDP were utilized by Ryou et al. [18] as an effective tool for cytosolic delivery of cargo proteins in an EGFR-specific manner, implying broad uses for cancer therapy. As seen in Fig. 3E, mice implanted with high-EGFR-expressing MDA-MB-468 cells showed significant inhibition of tumor growth when treated with Rb-PEA15-AA protein compared with Rb-control. Taken together, our data demonstrate that unphosphorylated PEA15-AA possesses an antitumor effect in vivo and thus offers a potential therapeutic agent for the treatment of breast cancer.

We next studied the molecular mechanism by which nonphosphorylatable PEA15 inhibits cell migration and EMT. Our group [11] observed a strong antitumorigenic effect of unphosphorylated PEA15-AA in an ovarian cancer xenograft model even though both PEA15-AA and PEA15-DD exhibited a tumor-suppressive phenotype in vitro. Based on that study, we speculated that the tumor microenvironment is a critical factor in regulating tumor growth and progression. Thus, we performed the RT² Profiler Human Cancer Inflammation and Immunity Crosstalk PCR Array to analyze the gene expression changes that occur as a result of phosphorylation status of PEA15. Interestingly, we found that IL-8 was one of the most downregulated genes in PEA15-AA-expressing cells and one of the most upregulated in PEA15-DD-expressing cells (Fig. 4A). As shown in Fig. 4B, we confirmed decreased expression of IL-8 mRNA in PEA15-AA-expressing cells. Further, IL-8 protein expression in these cells were consistently low at different timepoints (Fig. 4C). In addition, we observed an increased expression of IL-8 in PEA15-DD-expressing cells. IL-8 levels increased in PEA15-WT-expressing cells as well, however, this increase did not occur at the same rate as seen in PEA15-DD-expressing cells (Fig. 4C).

IL-8 is also expressed by cancer cells undergoing EMT and shown to promote metastasis [17, 19‐22]. The important role of IL-8 in promoting cell migration, and its reduced expression by PEA15-AA-expressing cells, led us to test whether decreased IL-8 expression in PEA15-AA cells was responsible for reduced migration. We tested whether recombinant IL-8 (rIL-8) could rescue the decreased migration of PEA15-AA-expressing MDA-MB-468 cells. We observed that recombinant IL-8 treatment rescued PEA15-AA-induced inhibition of migration (Fig. 5A). We observed that the PEA15-AA-expressing cells showed a more significant increase in migration (by 9%, P < 0.0001) compared to Vector and PEA-15-WT-expressing cells (Fig. 5A). No increase in migration was observed in PEA15-DD -expressing cells even in the presence of rIL8. Moreover, IL-8 treatment-induced expression of mesenchymal markers fibronectin, Zeb1, vimentin and snail in PEA15-AA-expressing cells (Fig. 5B,C). These results imply that IL-8 treatment is able to partly rescue the reduced migration caused by PEA15-AA.

Next, treating the cells with a neutralizing antibody against IL-8 (anti-IL-8) did not decrease migration in the empty Vector, PEA-15-WT or PEA-15-AA-expressing cells, but did decrease migration by 20% in PEA-15-DD-expressing cells compared to IgG isotype control. This confirmed that the high expression of IL-8 in the PEA-15-DD-expressing cells was partly responsible for its increased migratory capabilities.

To further confirm IL-8 as an important regulator in TNBC migration and EMT, we used IL-8 siRNA to knock down IL-8 expression in MDA-MB-468 cells stably expressing PEA15-DD. RT-qPCR analyses revealed that IL-8 siRNA successfully reduced the expression of IL-8 about 80% compared with control scrambled siRNA (Fig. 5D, left). We found that the migratory ability was partly decreased in PEA15-DD cells upon IL-8 siRNA treatment compared with control treatment (Fig. 5D, right). As expected, knockdown of IL-8 decreased the mesenchymal markers (n-cadherin, Twist, and Zeb1), as shown in Fig. 5E. These data demonstrate that IL-8 downregulation partially abolishes the increased cell migration and reverses the mesenchymal phenotype seen in the PEA15-DD-expressing cells and the effects of IL-8 expression could be altered by the phosphorylation status of PEA15.

It has been reported that ERK downstream molecules cause tumor development and induce IL-8 expression in many cancer types. Furthermore, in neuroblastoma, IL-8 expression was upregulated in an Ets1-dependent manner, suggesting a critical link between Ets-1 and IL-8 in the angiogenesis and metastasis process [23]. Moreover, the IL-8 promoter contains binding sites for Ets-family transcription factors in hematopoietic cells [24]. These findings indicate that interruption of IL-8 signaling may be a potential targeted treatment in metastasis. To determine whether phosphorylation of PEA15 regulates IL-8 expression through ERK downstream signaling, we measured the mRNA levels of the ERK-responsive transcription factor Ets-1. We observed that the mRNA expression level of Ets-1 was significantly decreased in PEA15-AA-expressing cells compared with PEA15-WT- and PEA15-DD-expressing cells. On the other hand, Ets-1 expression was greatly increased in PEA15-DD-expressing cells (Fig. 6A, left).

Considering the possible functional relevance of Ets-1 cellular distribution, we further analyzed the levels of total Ets-1 and phosphorylated Ets-1 in fractionated cell lysates from MDA-MB-468 cells. Compared with PEA15-WT- and PEA15-DD-expressing cells, PEA15-AA-expressing cells showed reduction of phosphorylated Ets-1 in the nuclear fraction. In addition, we observed that the p-ERK level in the cytoplasm was higher in the PEA15-AA-expressing cells than in the PEA15-DD-expressing cells (Fig. 6A, right). To test the effect of Ets-1 on IL-8 expression, we used Ets-1 siRNA to knockdown Ets-1 expression in MDA-MB-468 cells stably expressing PEA15-DD (Fig. 6B, left). Having verified Ets-1 knockdown, we examined whether Ets-1 silencing in PEA15-DD-expressing cells reduces the upregulated IL-8 levels previously observed. IL-8 mRNA expression was significantly decreased by approximately 85% upon Ets-1 knockdown (Fig. 6B, right). Furthermore, silencing of Ets-1 partly inhibited migration of cells stably overexpressing PEA15-DD (Fig. 6C). Taken together, these results suggest that PEA15 phosphorylation may regulate activity of Ets-1, leading to induction of one of its downstream targets, IL-8, thereby contributing to EMT and TNBC migration.