Drugs

Palbociclib (cat# CS-0019), abemaciclib (cat# CS-1230) and ribociclib (cat# CS-1750) were purchased from ChemScene (Monmouth Junction, NJ, USA). Palbociclib was dissolved in water while all other drugs were reconstituted in dimethyl sulfoxide (DMSO) (Sigma-Aldrich, USA).

Cell culture

HR+ breast cancer cell lines MCF7, T47D, and endogenously PEG10-expressing prostate cancer cell line PC3 were purchased from the American Type Culture Collection (Manassas, VA, USA) and maintained in a humidified atmosphere with 5% CO2 at 37 °C. Parental (MCF7 and T47D) and palbociclib-resistant cells (MCF7-PR and T47D-PR) were cultured in RPMI 1640 medium (cat# LM011-01, Welgene Inc., Daegu, Korea) supplemented with 10% heat-inactivated FBS (cat# S 001–01; Welgene Inc., Daegu, Korea) and 1% 100 × P/S solution (cat# LS202-02, Welgene Inc. Daegu, Korea). All the palbociclib-resistant cells were maintained with 5 µM palbociclib, and fresh medium was changed before 24–48 h of experiments.

Generation of resistant cells

Palbociclib-resistant cells (MCF7-PR and T47D-PR) were established after 7 − 9 months by a gradual increase in the palbociclib concentration starting from the IC50, which was 750 and 250 nM for MCF7 and T47D cells, respectively. Detailed methods for generation of palbociclib-resistant cells are explained in our previous study [6].

Gene expression microarray analysis

Gene expression microarray analysis was performed as described previously [6]. In brief, to discover the differentially expressed genes (DEGs), we performed microarray of palbociclib-resistant (MCF7-PR and T47D-PR) versus their corresponding parental (MCF7 and T47D) cells. First, RNA samples were extracted. Then Affymetrix Whole Transcript Expression array process was performed according to the manufacturer’s protocol (GeneChip Whole Transcript PLUS reagent Kit, ThermoFisher Scientific, USA). Signal values were computed using the Affymetrix® GeneChip™ Command Console software. Genes exhibiting at least a two-fold upregulation or downregulation were considered meaningful.

Quantitative real-time polymerase chain reaction (qRT-PCR)

qRT-PCR was performed on total RNA isolated from palbociclib-resistant (MCF7-PR and T47D-PR) and their corresponding parental (MCF7 and T47D) cells to validate the mRNA expression level. cDNA was synthesized using the Takara prime script 1st strand cDNA synthesis kit (cat# 6110A, Takara Bio Inc., Japan) according to the manufacturer’s instructions. qRT-PCR was performed using a ViiA7 Real-time PCR system (Applied Biosystems, Warrington, UK) and a Power-up SYBR Green Master Mix (cat# A25741, ThermoFisher Scientific, USA). The target gene expression was normalized to the β-actin. A list of PCR primers is listed in Supplementary Table S1.

Western blots

Cell protein was extracted with RIPA Lysis Buffer (cat# 89901, ThermoFisher Scientific, USA) combined with a protease inhibitor cocktail (cat# 11873580001, Roche, Germany) and phosphatase inhibitors (cat# 1862495, ThermoFisher Scientific, USA), and the concentration was detected by Pierce™ BCA protein assay (cat# 23228 and cat# 1859078, ThermoFisher Scientific, USA) Kit. Western blotting was performed as described previously [6]. The antibodies used for western blotting are listed in Supplementary Table S2.

Genomics of Drug Sensitivity in Cancer (GDSC) database analysis

Gene expression and drug screening data for the cell lines were downloaded from the GDSC website [25]. A total of 43 breast cancer cell lines were used to analyze the association between PEG10 expression and palbociclib sensitivity. Furthermore, among the 43 breast cancer cell lines, a subset comprising 11 HR+ breast cancer cell lines were also used to analyze the association between PEG10 expression and palbociclib sensitivity. Palbociclib sensitivity was defined as IC50 of < 3.5 µM. The list of 43 breast cancer cell lines is shown in the Supplementary Table S3.

Plasmids and ectopic overexpression studies

PEG10 has two overlapping open reading frames (ORFs), where the translation of ORF1 produces the PEG10 (RF1) gag-like protein that is associated with invasion and drug resistance, whereas the pol-like PEG10 (RF2) protein is synthesized by a programmed –1 frameshift translation [21, 26].

PEG10 plasmid, (PEG10-RF1 and PEG10-fsRF1/RF2), was generously provided by Prof. Dr. Jozsef Tozser, University of Debrecen, Hungary. Flag p21 WT plasmid (Plasmid # 16240) [27] was purchased from Addgene, and pCMV-SPORT6/SIAH1 plasmid (clone number hMU004814) was purchased from Korea Human Gene Bank (Medical Genomics Research Center, KRIBB, Korea), and respective empty vectors were used for MOCK controls. 5 × 105 cells were seeded, and the next day plasmid was transfected using the CalFectin ™ Mammalian DNA Transfection Reagent (cat# SL1000478, SignaGen Laboratories, USA) following the manufacturer’s instructions. After 48 h, cells were collected, and overexpression was confirmed using western blotting.

Migration and invasion assay

Migration and invasion assay was performed as described previously [28]. In brief, for migration and invasion assay, 5 × 105 cells were seeded in insert well with serum-free medium and incubated for 24–48 h at 37 °C. For the invasion assay, 50–100 µL of Matrigel matrix (cat# 254234, Corning, NY, USA) was coated in the insert well before cell seeding.

After incubation for the indicated time, the media was aspirated and unmigrated and non-invaded cells were carefully removed using cotton from the insert well. Furthermore, migrated, and invaded cells were fixed with 2% PFA, and crystal violet staining was performed for 5–10 min. Washed with PBS and water and then dried for 3–4 h before being observed under a light microscope. ImageJ software version 1.53t was used to quantify the area of migratory and invasive cells in five random nonoverlapping microscopic fields of each insert well at 100 × magnification.

Cell viability assay

Cell viability was measured using the thiazolyl blue tetrazolium bromide (MTT; Sigma, St. Louis, MO, USA) assay as previously described [6]. A total of 1,000–1,500 cells per well were seeded in 96-well plates and incubated overnight. Cells were then treated with various concentrations of palbociclib and fixed concentration of PEG10 siRNA (cat# 4392421, Assay ID: s23005, ThermoFisher Scientific, USA) with Lipofectamine RNAiMAX (cat# 13778100, ThermoFisher Scientific, USA) or PEG10-antisense oligonucleotide (PEG10-ASO) with Lipofectamine 2000 (cat# 11668019, ThermoFisher Scientific, USA) in OptiMEM® I reduced-serum medium (cat# 31985070, ThermoFisher Scientific, USA) and incubated for 72 h at 37 °C. Subsequent procedures were performed as described in our previous study [6]. The IC50 and combination index (CI) values were analyzed using CompuSyn software (ComboSyn, NJ, USA).

Double thymidine block cell cycle and fluorescence-activated cell sorting analysis (FACS)

For the double thymidine block experiment, cells were seeded at 5 × 105 in a 60-mm cell culture dish. After overnight incubation, 2 mM thymidine (cat# sc-296542, Santa Cruz Biotechnology, USA) was added for 18 h. Media was replaced with fresh complete media for 9 h; a second thymidine block was then performed for 18 h. Finally, cells were treated with 25 nmol PEG10 siRNA in a time-dependent manner, and cell cycle analysis was performed as described previously [6]. In brief, harvested cells were fixed with 70% ethanol, resuspended in RNAse A (100 μg/ml), and stained with propidium iodide (50 μg/ml) in PBS. Stained cells were assessed using a CytoFLEX flow cytometer (Beckman Coulter, USA), data were analyzed using FlowJo v10 (FlowJo LLC, Ashland, OR 97520, USA), and the percentage of cells in each phase of the cell cycle was determined.

Gene knockdown experiments

To perform siRNA/shRNA knockdown experiments, cells were seeded at 5 × 105 in a 60-mm cell culture dish and allowed to attach overnight. After overnight incubation, cells were transfected with scramble siRNA (cat# Sc-37007, Santa Cruz Biotechnology, USA) or PEG10 siRNA using Lipofectamine RNAiMAX reagent and ZEB1 shRNA (TRC Number: TRCN0000369267, Sigma-Adrich INC) by using Lipofectamine 2000 reagent following the manufacturer’s instructions. After 24–48 h of incubation, cells were isolated and analyzed for further experiments.

To perform ASO knockdown experiments, 5 different PEG10-ASOs and control-ASO were designed and synthesized by Creative Biolabs (USA). PEG10-ASOs were designed with a central gap segment comprising ten 2ʹ-deoxynucleosides and a central gap that was flanked on the 5ʹ and 3ʹ wings by five 2ʹ-MOE modified nucleosides; all internucleoside linkages were phosphorothioate linkages. For in vitro screening of PEG10-ASOs to select the most effective ASO, 5 × 105 cells were seeded in a 60-mm cell culture dish and allowed to attach overnight. Various doses of each ASO (100–500 nmol) were then diluted in 200 µL of OptiMEM® I reduced-serum medium and Lipofectamine 2000 reagent was diluted separately in 200 µL of OptiMEM® I reduced-serum medium, according to the manufacturer’s instructions. Diluted ASOs and Lipofectamine were mixed and incubated for 15 min at room temperature and added to the cells dropwise and then incubated for various times from 24–72 h. Cells were harvested for western blotting and qRT-PCR analysis. The best candidate ASO among the 5 different PEG10-ASOs was selected for further experiments such as in vivo efficacy assay according to the PEG10 knockdown efficacy. The sequence of control-ASO used was (5ʹ CCUUC CCTGAAGGTT CCUCC 3ʹ), which was adopted from Worley et al. [29]. The sequence of the best candidate PEG10-ASO was (5ʹ UUUGG TGCTTTAGGA UGUGU 3ʹ).

Animal studies

For tumor xenograft experiments, 5-week-old female BALB/c-nude mice were purchased from Gem Biosciences (Cheongju, Korea) and kept in a pathogen-free environment. All animal procedures were performed following the IACUC-approved protocol of CHA University. A total of 1 × 107 MCF7-PR cells in 0.1 mL of PBS containing a 50% (v/v) Matrigel solution (cat# 354248, Corning, NY, USA) were subcutaneously injected into the mammary fat pad of the mouse. For the estrogen supplement, estrogen valerate (3 μg/mouse) was subcutaneously injected once a week. When the tumor volume reached 60–80 mm3, the animals were randomized into four groups, each containing five mice. After grouping, mice were treated with control-ASO 15 mg/kg (group 1), palbociclib 100 mg/kg (group 2), PEG10-ASO 15 mg/kg (group 3), and PEG10-ASO with palbociclib (group 4). ASO was dissolved in PBS and injected systematically via intraperitoneal injection once per day for 5 days and then three times per week there after for a total of 18 days. Palbociclib was administered by oral gavage once daily for 22 days. Tumor size was measured three times a week using a vernier caliper, and the volume was calculated by using the formula (tumor length × tumor width2 × 0.5). Additionally, body weight was regularly measured three times a week. At 22 days, all mice were euthanized and the xenografted tumors were excised and preserved for further analysis.

Immunohistochemistry (IHC) staining analysis

We performed the IHC using the xenograft tissues to assess the PEG10 and Ki67 expression, and also using the patients’ tumor samples to assess PEG10 expression. Firstly, to perform the IHC using the xenograft tissues, samples were deparaffinized and antigen retrieval was performed using heat-mediated methods with an IHC antigen retrieval solution (cat# 00–4955-58, Invitrogen, ThermoFisher Scientific, USA) for 25 min. The section was stained with PEG10 primary antibody (cat# 14412–1-AP, Proteintech, USA) or Ki67 primary antibody (cat# ab92742, abcam, USA) at 1:200 dilution for 12 h at 4 °C. Next day, the section was washed, and HRP-conjugated secondary antibody was added for 1 h at room temperature. Finally, slides were counterstained, and mounting was performed, and then examined under a light microscope. To quantify IHC staining, the number of positively stained cells were counted using ImageJ software version 1.53t in five random nonoverlapping 400 × microscopic fields for each tissue section.

Secondly, to evaluate if the PEG10 protein expression level is associated with efficacy of CDK4/6 inhibitor, we performed the IHC using the patients’ tumor samples. Patients’ formalin-fixed, paraffin-embedded tumor samples were prepared from 15 patients with advanced breast cancer. Clinical characteristics of these patients were listed in the Supplementary Table S4. Fourteen patients were HR+/HER2– and one HR+/HER2+. Regarding metastatic sites in the beginning of CDK4/6 inhibitor treatment, 10 patients (66.7%) had visceral metastases. All these patients were treated with combined palbociclib and letrozole or anastrozole as a first-line therapy. During median follow-up of 10 months (1.0–72.5), 11 patients stopped CDK4/6 inhibitor treatment due to disease progression (n = 10) and adverse events (n = 1). All tumor specimens were obtained before CDK4/6 inhibitor treatment at CHA Bundang Medical Center (Seongnam, South Korea). For IHC of patients’ tumor samples, the automatic device Ventana Discovery ULTRA Stainer (Roche, Germany) was used, and the procedure was performed according to the manufacturer’s instructions. In brief, human tissue sections were stained with PEG10 primary antibody at 1:200 dilution (cat# 14412–1-AP, Proteintech, USA) for 60 min. Following incubation, one drop OmniMap anti-Rb HRP (cat# 760–4310, Roche, USA) was added and incubated for 16 min. Finally, after counterstain mounting was performed. Positivity for PEG10 was assessed from 0 to 100% of stained cells by cytoplasmic staining with any intensity. The cut-off of 1% was used for PEG10 positivity as in a previous report [30]. CDK4/6 inhibitor progression-free survival (PFS) was analyzed according to PEG10 positivity.

Terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) assay

TUNEL assay was conducted to evaluate apoptosis in xenograft tissues according to the manufacturer’s instructions (cat# S7101, Merck, MA, USA). Briefly, after dewaxing and hydration, slides were treated with TUNEL and hematoxylin solution for indicated times. To quantify TUNEL assay results, the number of positively stained cells were counted using ImageJ software version 1.53t in five random nonoverlapping 400 × microscopic fields for each tissue section.

Public gene expression profiling data sets in patients with breast cancer

To evaluate if the PEG10 mRNA expression level is associated with prognosis in patients with HR+ breast cancer, we used public mRNA expression data sets of curatively resected HR+ breast cancer. Firstly, of downloadable public data sets, we searched for data sets of minimum 200 estrogen receptor–positive (ER+) patients containing essential clinical factors such as pathologic stages and recurrence to conduct multivariate analyses. We chose GSE 6532 and series matrix files of the data set, which were already normalized by original authors [31], were downloaded for our analyses. Secondly, other data sets were analyzed using online platform Kaplan–Meier plotter [32] (https://kmplot.com/analysis/) without downloading processed raw data. Of the total 24 data sets analyzed, we selected 4 data sets (GSE 2034, E-MTAB-365, GSE 3494, and GSE 25066) with the criteria of ER+ status by IHC and minimum 200 analyzed patients. Affymetrix GeneChip Human Genome U133 Arrays were used for all these 5 GSE data sets. Affymetrix ID 212092 was selected for PEG10. PEG10 mRNA expression level was divided into two groups, “PEG10-High” and “PEG10-Low” with cut-off of upper quartile (GSE 6532) or auto-select best cut-off (GSE 2034, E-MTAB-365, GSE 3494, and GSE 25066). An optimal cut-off point for normalized intensity of PEG10 mRNA was determined using minimum p-value approach in predicting recurrence-free survival (RFS) [33].

Statistical analysis

All statistical analysis were performed using SPSS version 19.0 (IBM SPSS statistics 19.0, NY, USA). Student’s t-test was used to analyze independent two groups comprising continuous variables: the qRT-PCR of mRNA expression of EMT genes, gene expression according to palbociclib sensitivity using GDSC database, cell viability assay, cell cycle analysis, and tumor growth inhibition. IHC, TUNEL assay, invasion, and migration assay data were first quantified using ImageJ software version 1.53t, and p-values were analyzed by Student’s t-test using quantified data. Values are represented as mean ± SD unless otherwise indicated. RFS was defined as the time from curative surgery to breast cancer recurrence or the last date when the patient was known to be free of recurrence (censoring time). PFS was defined as the time from CDK4/6 inhibitor treatment to breast cancer progression or the last date when the patient was known to be free of progression (censoring time). RFS and PFS were calculated using the Kaplan–Meier method. A log-rank test was used to compare RFS and PFS between groups. Significant prognostic variables by univariate analysis, defined as those with a p-value < 0.05 were submitted to multivariate analysis using the Cox proportional hazard regression model. All p-values were two-tailed, and p-values < 0.05 were considered as significant.