FZD5 contributes to TNBC proliferation, DNA damage repair and stemness

Chemotherapy currently remains the standard treatment for triple-negative breast cancer (TNBC). However, TNBC frequently develop chemoresistance, which is responsible for cancer recurrence and distal metastasis. Both DNA damage repair and stemness are related to chemoresistance. FZD5, a member in Frizzled family, was identified to be preferentially expressed in TNBC, and associated with unfavorable prognosis. Loss and gain of function studies revealed that FZD5 contributed to TNBC cell G1/S transition, DNA replication, DNA damage repair, survival, and stemness. Mechanistically, transcription factor FOXM1, which promoted BRCA1 and BIRC5 transcription, acted as a downstream effecter of FZD5 signaling. FOXM1 overexpression in FZD5-deficient/low TNBC cells induced FZD5-associated phenotype. Finally, Wnt7B, a specific ligand for FZD5, was shown to be involved in cell proliferation, DNA damage repair, and stemness. Taken together, FZD5 is a novel target for the development of therapeutic strategies to overcome chemoresistance and prevent recurrence in TNBC.


Introduction
Breast cancer is the most common malignant tumor in women. Early detection and systemic therapies have decreased the mortality in North America and the European Union 1 . Triple-negative breast cancer (TNBC) currently lack established molecular targets, and TNBC patients usually have an unfavorable prognosis. Cytotoxic chemotherapy remains the standard treatment for TNBC 2 . Although TNBC generally have a high response rate to chemotherapy, they frequently develop chemoresistance 3 . Therefore, it is crucial to identify new molecular targets and develop novel strategies by elucidating the mechanisms of chemoresistance.
In response to endogenous and exogenous insults, normal cells utilize a variety of DNA damage sensing and repair mechanisms, also called DNA damage response (DDR), to maintain genomic integrity. Compared to normal cells, malignant cells have a higher DNA damage burden because of oncogene-induced replication stress 4 . Accordingly, malignant cells usually have enhanced DNA repair capacity to cope with DNA damage and to proliferate and survival. The increase in DNA repair capacity weakens the therapeutic effect of radiotherapy and chemotherapy, which induce cell death through DNA damage, thereby causing radioresistance and chemoresistance, and cancer recurrence. In TNBC, several factors such as tripartite motif containing 37 (TRIM37), neuropilin 2 (NRP2), and dynamin 2 (DNM2) have recently been shown to be associated with DNA damage repair, contributing to chemoresistance [5][6][7] .
Here Frizzled 5 (FZD5), a member in FZD family, was identified as a factor promoting DNA damage repair, stemness, and chemoresistance in TNBC cells. Moreover, transcription factor Forkhead Box M1 (FOXM1), which modulated BRCA1 and BIRC5 (Survivin) transcription, was demonstrated to be a downstream effector of FZD5 signaling.

Results
FZD5 is principally expressed in TNBC, and associated with unfavorable prognosis The Cancer Genome Atlas (TCGA) database was interrogated for FZD5 mRNA expression in breast cancer tissues. TNBC expressed a higher level of FZD5 mRNA compared to non-TNBC (Fig. 1A). This finding was supported by the interrogation of GSE2603 database (Fig. 1A). The preferential expression of FZD5 in TNBC was further confirmed by Immunohistochemistry staining (Fig. 1B). 15 of 18 TNBC presented positive staining, while only 4 of 24 non-TNBC appeared FZD5-positive (χ 2 = 18.45, P < 0.0001). The association of FZD5 with survival was analyzed in Kaplan-Meier plotter website. Higher FZD5 expression was correlated with shorter overall survival (OS), recurrence-free survival (RFS), distal metastasis-free survival (DMFS), and post-progression survival (PPS) (Fig. 1C). Gene set enrichment analysis (GSEA) of TCGA database suggested that FZD5 was related to cell cycle transition, DNA replication, and DNA damage repair (Fig. 1D, E). GSEA further indicated that FZD5 was implicated in a series of signaling pathways related to chemo-resistance or radio-resistance (Supplementary Table 1, Supplementary Fig. 1).

FZD5 induces cell growth in vitro and in vivo
Three TNBC cell lines, MDA-MB-231, MDA-MB-468, and Hs-578t with differential FZD5 expression, were selected for the subsequent study. FZD5 expression in MDA-MB-231 cells was stably knockdowned ( Fig. 2A). Cell growth in vitro was assessed by CCK8 and colony formation tests. FZD5 knockdown significantly repressed MDA-MB-231 cell growth (Fig. 2B, C). FZD5 knockdown similarly inhibited MDA-MB-468 cell growth (Supplementary Fig. 2A, B). To the contrary, stable FZD5 overexpression remarkably promoted Hs-578t cell growth ( Fig. 2D-F). MDA-MB-231 cells with or without FZD5 knockdown were inoculated into nude mice to evaluate the effect of FZD5 on cell growth in vivo. Consistent with the finding in vitro, tumors with FZD5 knockdown grew more slowly compared to tumors without FZD5 knockdown (Fig. 2G). Moreover, tumors with FZD5 knockdown exhibited a lower fraction of phosphorylated Histone 3 (p-H3)-positive and Ki67-positive cells, indicating that FZD5 knockdown blocked cell proliferation in vivo (Fig. 2H, Supplementary Fig. 2C).

FZD5 promotes G1/S transition and DNA replication
Flow cytometry was used to investigate the effect of FZD5 on cell cycle. FZD5 knockdown in MDA-MB-231 and MDA-MB-468 cells increased G1 fraction but lowered S fraction, indicating that FZD5 knockdown arrested G1/S transition (Fig. 3A, Supplementary Fig. 3A). Consistent with this finding, FZD5 knockdown reduced the expression of factors related to G1/S transition such as CDK2, Cyclin E2, and Cyclin A2 (Fig. 3B, Supplementary  Fig. 3B). FZD5 knockdown decreased PCNA expression, indicating that DNA replication was suppressed, which was further confirmed by decreased fraction of EDUpositive cells (Fig. 3B, C, Supplementary Fig. 3B, C). Just as expected, FZD5 overexpression in Hs-578t cells diminished G1 fraction but elevated S fraction (Fig. 3D). FZD5 overexpression increased the expression of CDK2, Cyclin E2, Cyclin A2, and PCNA, and the fraction of EDU-positive cells (Fig. 3E, F).

FZD5 enhances DNA damage repair and chemoresistance
As GSEA suggested that FZD5 was related to DNA damage and chemoresistance (Fig. 1D Supplementary Fig. 1), the role of FZD5 in DNA damage repair and chemoresistance was investigated. Adriamycin (ADR) was used to induce DNA damage, which is characterized by γ-H2AX recruitment.    Table 2). These correlations also existed in 28 TNBC cell lines (Supplementary Table 3, Supplementary Fig. 8). Loss and gain of function studies confirmed that FZD5 modulated these factors in breast cancer cells (Fig. 6B, C). Whether FOXM1, a transcription factor, promoted BRCA1 and BIRC5 gene transcription was subsequently determined. The binding sites for FOXM1 were identified in BRCA1 and BIRC5 promoters (Fig. 6D). ChIP in combination with real-time PCR detection verified that FOXM1 could bind to these two gene promoters (Fig. 6E). Furthermore, FOXM1 overexpression in Hs-578t cells induced BRCA1 and BIRC5 expressions (Fig. 6F). FZD5 knockdown reduced active β-catenin level, and treatment with β-catenin inhibitor XAV939 suppressed FOXM1 expression, indicating that FZD5 modulated FOXM1 in a Wnt/β-catenin-dependent manner (Fig. 6G, H).
Discussion FZD belong to the superfamily of G protein-coupled receptor (GPCR). The N-terminus contains a cysteinerich domain (CRD), via which FZD can bind Wnt ligands. The C-terminus binds Dishevelled (Dvl) and interacts with G proteins. 10 FZD have thus far been identified in human. FZD5 induces proliferation of cancer cells by activating β-catenin pathway 25 . FZD5 also mediates β-catenin-independent pathways to increase tumor cell motility 27 . Our study revealed a novel role of FZD5 in TNBC. FZD5 initiates a signaling to enhance DNA damage repair and induce chemoresistance. FZD5 promotes Cyclin A, Cyclin E, CDK2 and PCNA expression, G1/S transition and DNA replication. Furthermore, FZD5 is associated with epithelial-like stemness characterized by hyper-proliferation and higher expression of ALDH1, EPCAM, and CD133 13,14 . TNBC is characterized by aberrant activation of both canonical and non-canonical Wnt pathways, which induces TNBC stemness and metastasis 28,29 . Several other FZDs such as FZD6 and FZD7 have been shown implicated in TNBC 30,31 . Our study identified FOXM1 as a downstream effecter of FZD5 signaling. FOXM1 is one of the members of the Forkhead family proteins. In normal cells, FOXM1 regulates cell cycle transition. Overexpression of FOXM1 has been observed in a variety of cancer types, indicating that FOXM1 plays essential roles in carcinogenesis 32,33 . In breast cancer, RNA-Seq analysis has revealed that FOXM1 is associated with not only proliferation, but also cell cycle transition, apoptosis, regulation of transcription, DNA replication, and DNA damage repair 34 . FOXM1 contributes to chemoresistance in breast cancer by enhancing DNA damage repair [35][36][37][38] . Moreover, suppression of FOXM1 was shown to reduce breast cancer growth in vitro and in vivo 34,39 .
BRCA1 plays a crucial role in the maintenance of DNA stability 40 . Therefore, BRCA1 mutation results in increased genomic instability and risk of developing breast and ovarian cancers due to deficient DNA repair 41 . About 20% TNBC harbor BRCA1/2 mutation, and patients with these tumors have a significantly lower risk of recurrence 42 . The better prognosis may largely be related to the higher sensitivity to anticancer drugs due to deficient DNA damage repair. To the contrary, increased DNA repair capacity contributes to chemoresistance. High BRCA1 expression is correlated with chemoresistance in hepatocellular carcinoma 43 . Tamoxifen-resistant breast cancer cells are resistant to cisplatin and ADR because of high BRCA1 expression 44 . Our study has identified FZD5-FOXM1 signaling as an upstream modulator of BRCA1, which mediates FZD5-induced chemoresistance in TNBC cells without BRCA1 mutation.
BIRC5/Survivin is one of the members of the inhibitor of apoptosis protein family (IAPs). High BIRC5 expression in tumor cells correlates with cell division, apoptosis inhibition, chemoresistance, and stemness 45 . BIRC5 downregulation inhibits cell growth but promotes apoptosis in breast cancer cells 46 . Moreover, BIRC5 silencing induces DNA double-strand breaks and impairs DNA repair capacity 47 . BIRC5 is transcriptionally modulated by FOXM1, and overexpression of FOXM1 and BIRC5 is related to drug resistance and unfavorable prognosis in breast cancer 38 . BIRC5 silencing also inhibits self-renewal and chemoresistance of EPCAM-positive breast CSC 48 .
In summary, our study has demonstrated a novel function of FZD5 in breast cancer, especially in TNBC. FZD5 contributes to G1/S transition, DNA replication, DNA damage repair, chemoresistance, and stemness. As a downstream effector of FZD5 signaling, FOXM1 transcriptionally upregulates BRCA1 and BIRC5, both of which play crucial roles in these processes. Therefore, FZD5-FOXM1 may be a potential target for the development of strategies to increase chemosensitivity and prevent recurrence in TNBC.

Materials and methods
In silico analysis TCGA (TNBC: 115; non-TNBC: 858) and GSE2603 (TNBC: 25; non-TNBC: 71) databases were interrogated for FZD5 mRNA expression in breast cancer samples. The association of FZD5 with survival was analyzed in Kaplan-Meier plotter website (http://kmplot.com/ analysis/) 49 . GSEA was performed using TCGA database. According to the median value of FZD5 mRNA level, the samples were divided into FZD5-high and FZD5-low groups. The gene pathways differentially expressed between FZD5-high and FZD5-low breast cancer samples were analyzed. Cancer Cell Line Encyclopedia (CCLE) database was interrogated for FZD5, FOXM1, BRCA1, and BIRC5 mRNA expression in a series of human breast cancer cell lines (Supplementary Tables 2  and 3). Correlation between two genes was analyzed by Pearson statistics.

Immunohistochemistry
The tissues were fixed in 4% paraformaldehyde for 72 h, embedded in paraffin and then sliced into 4 μm sections. Xylene and gradient alcohol were used to deparaffinize and hydrate, respectively. 3% H 2 O 2 was used to eliminate endogenous peroxidase activity. Citrate buffer was used to repair antigen, and then the sections were blocked by BSA. The sections were incubated with primary antibodies (anti-FZD5, Abcam, ab75234, UK, 1:200; antiphosphorylated Histone 3, ThermoFisher, PA5-17869, USA, 1:200; anti-Ki67, Invitrogen, 14-5699-95, USA, 1:500) overnight at 4°C. Subsequently, the sections were incubated with goat anti-rabbit IgG and streptavidin peroxidase (SP) complex at 37°C for 30 min, and stained with DAB reagent. Finally, the sections were re-stained with hematoxylin, dehydrated with gradient alcohol and xylene, and observed under a microscope (LEICA DM2500 LED). Phosphorylated Histone 3-positive or Ki67-positive cells in five randomly selected fields were counted.

Western blot
Cells were lysed using RIPA lysis buffer with 1% PMSF on ice for 1 h. Protein fragments were centrifuged with 12,000 × g at 4°C for 40 min. A BCA protein assay kit was used to determine the protein concentration. 30 μg protein was separated on 10% SDS-PAGE gel and transferred into a polyvinylidene difluoride (PVDF) membrane in a wet electron transfer device. The membrane was blocked in 5% skimmed milk in Tris-buffered saline (TBS) containing 0.05% Tween 20 for 2 h at room temperature. Subsequently, the membrane was incubated with various primary antibodies at 4°C overnight. After being washed in TBST three times, the membrane was incubated in horseradish peroxidase (HRP)-conjugated goat anti-rabbit for 1.

Quantitative real-time PCR
Total RNA was extracted from cells with TRIZOL Reagent (Takara, 9108/9109, China) according to the standard instructions. Reverse transcription was conducted with the cDNA synthesis Kit (Takara, RR047A, China) with 1 μg RNA. The target cDNA was amplified by TB Green™ Premix Ex Taq II and an ABI PRISM 7300 Sequence Detection system (Applied Biosystems, USA). The relative gene expression was analyzed using the 2 −ΔΔCt method. GAPDH was used as control. The primers used are listed in Table 1.

In vivo animal study
Female BALB/c nude mice (5-6 weeks of age, 18-20 g) were purchased from Weitong Lihua (Beijing, China), and all animals were dealt with according to the Animal Ethics Committee of China Medical University. Before tumor cell inoculation mice were randomized into different groups (five in each group). 5 × 10 6 MDA-MB-231 cells were resuspended in 100 μL PBS with 50% Matrigel, and injected into the mammary fat pad of the mice. Tumor length and width were measured with a vernier caliper every 3 days. Tumor volume was calculated by the formula: V = 1/2 × length × width 2 . The investigator was blinded to the group allocation of the animals during the experiment.
Cell cycle assay 1 × 10 6 cells were harvested and washed with PBS, then fixed with 70% ethanol at 4°C overnight. All samples were treated with 5 μL RNase A and 450 μL propidium iodide (PI) (KeyGEN BioTECH) for 1 h at room temperature in the dark and analyzed by a FACS calibur flow cytometer (BD).

EDU staining
Cells were seeded in 24-well plates (MDA-MB-231: 5 × 10 5 /well; MDA-MB-468: 5 × 10 5 /well; Hs-578t: 4 × 10 5 / well) for 24 h, and then incubated with 50 μM 5-Ethynyl-20-deoxyuridine (EDU, Ribobio) for 2 h according to the manufacturer's instruction. Subsequently, cells were washed with PBS, fixed with 4% paraformaldehyde for 30 min and permeabilized with 0.5% Triton X-100 for 10 min. Cells were then treated with 200 μL 1 × Apollo reaction cocktail for 30 min at room temperature in the dark. At last, the DNA contents were stained by 200 μL 1 × Hoechst 33342 for 30 min at room temperature in the dark. A Laser scanning confocal focus microscope was used to visualize the staining. Positive cells in random five fields were counted.

Immunofluorescence
Cells were seeded in 24-well plates (MDA-MB-231: 5 × 10 5 /well; MDA-MB-468: 5 × 10 5 /well; Hs-578t: 4 × 10 5 / well). After incubation for 24 h, ADR (Solarbio, China) was added into cell cultures at a final concentration of 300 nM for 48 h. The cells were washed with PBS, fixed with 4% paraformaldehyde at room temperature for 30 min, permeabilized with 0.5% Triton X-100 at room temperature for 10 min, and blocked with 5% donkey serum and 0.3% Triton X-100 in PBS at room temperature for 1 h. The cells were then incubated with antibody for γ-H 2 AX (Cell Signaling Technology, #9718, USA, 1:400) at 4°C overnight. Subsequently, the cells were incubated by Alexa Fluor 488-conjugated anti-mouse IgG for 2 h at room temperature in the dark. The nuclei were stained by DAPI at room temperature for 5 min. Immunofluorescence staining was observed under a laser scanning confocal focus microscope.
Apoptosis assay 5×10 5 cells were harvested and washed with PBS. According to the standard protocol, an Annexin V-PE/7-AAD Apoptosis Kit (KeyGEN BioTECH, China) was used to analyze cell apoptosis. Cells were incubated in 500 μL binding buffer with 5 μL 7-AAD at room temperature in the dark for 15 min. The number of apoptotic cells was measured by a FACS Calibur Flow Cytometer (BD).
Tumorisphere formation 1 × 10 4 cells were seeded in a six-well attachment surface polystyrene culture plate (Corning Costar, USA). Cells were cultured in complete MammoCult™ Human Medium (STEMCELL Technologies, USA) at 37°C and 5% CO 2 for 12 days. Spheroids in five randomly selected fields were counted.

Chromatin immunoprecipitation (ChIP)
ChIP was analyzed using an assay kit (Beyotime, China) according to the manufacturer's protocols. Briefly, Cells in 10 cm plates were fixed with 1% formaldehyde for 10 min at 37°C. To shear the chromatin, cells were treated with 1 mM PMSF in SDS lysis buffer for 10 min at 4°C, followed by cell sonication for 15 min at 4°C. After a portion of the cross-linked chromatin was removed as input for the subsequent test, the remaining cell lysis was incubated with 1 μg anti-FOXM1 antibody at 4°C overnight. Then protein A + G agarose was added to precipitate the target protein recognized by anti-FOXM1 antibody for 1 h at 4°C. Anti-IgG antibody was used as a negative control (NC). The beads were then washed off and DNA was collected for subsequent real-time PCR. The enrichment was indicated as % of input. The NC sequence for BRCA1 is AGACAGTAACT, and for BIRC5 is GAGAAGTGAG. The primers for BRCA1 NC are CTAACATGGCGGAC AAAGACA (forward) and GAGGGACAAGTGGTAAG AGCC (reverse), for BIRC5 NC are GGGGCTGGAGG GCTAATA (forward) and TGCTTTGGAACAGGGTG T (reverse).

Statistical analysis
All cell experiments were performed in triplicate. The data are expressed as mean ± SD. GraphPad prism 5 was used to analyze the data. Differences were analyzed by two-sided Student's t test or one-way ANOVA when the variance is similar between the groups. P value < 0.05 was considered statistically significant. No statistical method was used to predetermine the sample size for xenograft mice experiment, which was based on previous experimental observations. The sample size of each experiment is shown in the legend. No data was excluded from the analysis.