Abstract

Purpose: How neddylation pathway functions in breast tumor and regulation of estrogen receptor (ER) expression is rarely reported. The purpose of this study was to identify the role of neddylation in breast cancer and ER expression, and further explore the underlying mechanisms.

Experimental Design: The expression patterns of nedd8-activating enzyme (NAE) and nedd8, two key proteins in the neddylation pathway, were examined in human breast specimens. ER-“ expression was investigated using animal 18F-FES-PET/CT and immunoblotting upon NAE inhibitor MLN4924 treatment. Chromatin immunoprecipitation assay (ChIP), Luciferase reporter promoter assay and the CRISPR-Cas9 system were used to elucidate the mechanism of ER-“ regulation by MLN4924. The ER-positive mouse model was used to determine the synergetic effect of MLN4924 and fulvestrant on tumor growth. All statistical tests were two-sided.

Results: Both NAE1 and nedd8 expressions were higher in ER-positive subgroup. Higher expressions of NAE1 and nedd8 indicated poorer prognosis. Importantly, ER-“ expression was significantly downregulated upon MLN4924 treatment in vitro and in vivo. Mechanistically, MLN4924 treatment delayed serum and glucocorticoid-induced protein kinase (SGK) degradation and induced Forkhead box O3a (FOXO3a) nuclear export as well as decreased binding to the ESR1 promoter. Importantly, MLN4924 single or synergized with fulvestrant significantly suppressed the growth of ER-positive breast cancer in vitro and in vivo.

Conclusions: Our proof of principle study determines the activation of neddylation in breast tumor CyBio automatic dispenser tissues for the first time and reveals a new ER-“ regulatory mechanism as well as further explored an effective approach to improve fulvestrant sensitivity through neddylation inactivation combination.

Keywords: estrogen receptor (ER),neddylation/MLN4924, fulvestrant, breast cancer.

Translational relevance

Neddylation pathway activation status and the role it plays in breast tumor are still unknown. In this study, we found that neddylation was overactivated in breast cancer and predicted poor prognosis. Neddylation was activated more significantly in ER-positive breast cancer and targeting neddylation transcriptionally inhibited ER-“ expression via SGK1-dependent nuclear export of FOXO3a. Inactivation neddylation with MLN4924, a specific NAE inhibitor, significantly increased fulvestrant sensitivity in ER-positive breast cancer.

Introduction

Breast cancer has been ranked as the most common cancer diagnosed among women worldwide (1). Two thirds of breast cancer patients show positive estrogen receptor (ER) expression which is an important biomarker for molecular classification of breast cancer (2). Moreover, ER determines the sensitivity and effectiveness of endocrine therapy and is undoubtedly one of the most successful molecular targets in the history of breast cancer drug discovery (3, 4). The long-term goal in this area is to explore the regulatory mechanisms of ER-“ expression and new therapeutics based on this in breast cancer. Several mechanisms have been reported. At the transcriptional level, transcription factors, co-activators and co-repressors regulate ESR1 transcription by direct binding to the ESR1 promoter (5, 6). In addition to transcriptional regulation, posttranslational modifications, including phosphorylation, ubiquitination, acetylation and sumoylation, influence the stability and transcriptional activity of ER (7-9). Although vast efforts have been made, the molecular mechanisms regulating ER expression are not yet fully understood.Fulvestrant, a pure estrogen receptor antagonist and selective estrogen receptor degrader (SERD), has been approved for the treatment of ER-positive, metastatic breast cancer in postmenopausal women who failed to respond to initial endocrine therapy (i.e., tamoxifen or an aromatase inhibitor)(10, 11). A phase III clinical trial has proven that fulvestrant 500 mg was correlated with extended progression free survival (PFS) without increasing toxicity (12). Furthermore, as the first-line treatment of metastatic breast cancer, fulvestrant 500 mg was superior to anastrozole with a median PFS of 23 versus 13 months (13). However, the effectiveness of fulvestrant remains limited.

Neddylation is a type of posttranslational protein modification, by which the ubiquitin-like molecule nedd8 (neural precursor cell expressed, developmentally downregulated 8) is conjugated to substrates through a process of three-step enzymatic cascade (14). The above process involves nedd8-actvating enzyme (NAE, a heterodimer comprising subunits NAE1 and UBA3, ubiquitin-activating enzyme 3) and nedd8-conjugating enzyme E2 (UBC12) and substrate-specific E3s (15, 16). Cullin family members, the essential subunits of multiunit Cullin RING E3 ligases (CRLs), are the best understood substrates of neddylation (17). CRLs are in charge of the degradation for approximately 20% of ubiquitinated cellular proteins to regulate diverse biological processes, whereas their dysfunctions lead to carcinogenesis (18, 19).

Recently, theneddylation pathway has been reported overactivated in many tumors, including lung cancer (20), liver cancer (21) and glioma (22), and has served as an attractive anticancer target. Inactivation of neddylation pathway with MLN4924, a specific inhibitor of NAE (23), causes accumulation of a mass of CRL E3 substrates, which triggers DNA damage and induces abnormal cell-cycle progression, apoptosis, and/or senescence to suppress the growth of cancer cells in vitro and in vivo (24, 25). MLN4924 shows potent antitumor activity in preclinical models and has been proven to be well tolerated in phase I/II clinical trials for several solid tumors and haematologic malignancies (26). However, limited studies have focused on the function of neddylation in breast cancer. Importantly ，1) the neddylation activation status in breast cancer tissues is still unknown; 2) Medicopsis romeroi whether the activation of neddylation is distinct among different molecular subtypes; 3) the importance of neddylation in regulation of ER expression; 4) the role of neddylation inactivation with MLN4924 in fulvestrant sensitivity in ER-positive breast cancer are largely unknown.In this study, we first studied the expression of neddylation in breast cancer tissues depending on molecular subtypes and revealed a novel regulatory mechanism of ER-“ by neddyaltion and further explored an effective approach to improve fulvestrant sensitivity through neddylation inactivation in ER-positive breast cancer.

Materials and Methods

Cell culture, Reagents and Antibodies

All cell lines used in this study were obtained from the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China) and were authenticated by short tandem repeat profiling and monitoring mycoplasma contamination. Cells were expanded and frozen immediately into numerous aliquots after arrival in 2014. Cells were cultured in RPMI 1640 supplemented with 10% fetal bovine serum plus 5% penicillin/streptomycin at 37°C under 5% CO2. MLN4924 was synthesized and used for in vitro and in vivo studies as previously described (27). Fulvestrant was kindly provided by AstraZeneca (London, UK). Cycloheximide (CHX) and MG132 was purchased from Sigma-Aldrich (Sigma-Aldrich, St. Louis, MO). Leptomycin B (LMB) was purchased from Beyotime (Beyotime, Jiangsu, China). Antibodies for Immunohistochemistry, Immunoblotting and ChIP assays were listed in Supplemental Methods (available online).

Patients’ selection

Human breast cancer tissue arrays were from Shanghai Biochip (Biochip, Shanghai, China). The tissue arrays were from 113 patients with histologically confirmed primary breast cancer who underwent mastectomy in the Shanghai Changhai Hospital between January 2000 and December 2003. Their clinicopathological characteristics and follow-up information were available. This study was approved by the Research Ethics Committee of Shanghai Changhai Hospital. All patients provided written informed consent before their inclusion in this study.

Immunohistochemical Staining

Human breast cancer tissue arrays were stained by immunohistochemistry (IHC) with NAE1, nedd8, ER and PR antibodies. A positive control tissue with invasive breast cancer known to express high levels of each marker was established. For negative control primary antibody was replaced by corresponding IgG. The mouse tumor tissue sections were stained with ER, PR and ki67 antibodies. The tissue array sections were dehydrated and subjected to peroxidase blocking. Primary antibodies were added and incubated at room temperature for 30 minutes on the DAKO AutoStainer using the DakoCytomation EnVision+ System-HRP detection kit (Dakocytomation, Carpinteria, CA). The slides were counterstained with hematoxylin.The stained slides were examined by light microscopy by two independent observers. For the evaluation of immunohistochemical staining, please see theSupplemental Methods(available online).Written informed consent regarding tissue and data use for scientific purpose was obtained from all participants.

Cell Proliferation Assay

Cells were seeded in 96-well plates with 3000 cells per well, in triplicate, and cultured overnight. Cells were treated with the indicated drugs for 72 hours, followed by Cell Counting Kit-8 assay (Dojindo, Kumamoto, Japan), according to the manufacturer’s instructions.The combination therapy assay was performed as previously reported (28). The proliferation was assessed using Cell Counting Kit-8 assay. Growth inhibition values were normalized to the measured value of non-treated group. The combination index (CI) was calculated using CompuSyn.

Cell ClonogenicAssay

Cells were seeded in six-well plates (300 cells per well) in triplicate, and cultured overnight. Cells were treated with the indicated drugs for 14 days. Representative results of three independent experiments with similar trends arepresented.

Isolation of Nuclear and Cytoplasmic Extract

Cells were harvested at 80% confluence through trypsination. Isolation of nuclear and cytoplasmic extract was made using an NE-PER Nuclear Cytoplasmic Extraction Reagent kit (Pierce, Rockford, IL) according to the manufacturer’s instructions.

Immunoblotting

Cell lysates (30μg) were resolved by SDS-PAGE, transferred to PVDF membranes (Millipore, Billerica, USA), and incubated with the indicated primary antibodies. Corresponding protein-antibody complexes were detected using enhanced chemiluminescence system (BIO-RAD ，Molecular Imager® ChemiDOC™ XRS+).The experiments were repeated at least three times.

Real-Time Polymerase Chain Reaction Analyses

The real-time polymerase chain reaction (PCR) was performed using SYBR® Premix Ex Taq™ (TaKaRa, Shiga, Japan) on the ABI 7900 thermocycler (Thermo Fisher Scientific, Rockford, IL) following the manufacturer’s protocol and instrument manual. Sequence of the primers used can be found in the Supplemental Methods (available online).

CRISPR Cas9 assay

CRISPR Cas9 assay was used to generate UBC12, nedd8 and SGK1 knock-out cells. Briefly, single-guide (sg) RNAs targeting the corresponding proteins were designed (http://www.genome-engineering.org/crispr/) and cloned into the lentiGuide-puro vectors (Addgene plasmid no. 52963). HEK293T cells were co-transfected with lentiGuide-puro/sgRNA vector to generate sgRNA virus. After harvest, the viruses were introduced into MCF-7 cells stably expressing Cas9-GFP-luciferase. Forty-eight hours later, the infected cells were subjected to selection with 2 μg/ml puromycin for one week. The KO cells were validated by immunoblotting with corresponding primary antibodies.

ER-Positive Xenograft in Mice

Female BALB/c athymic nude mice (aged 6–8 weeks) were maintained and treated in accordance with established guidelines, and the protocol was approved by an internal animal protocol review committee. All the mice were handled using aseptic procedures and allowed to acclimatize to local conditions for one week before the experimental manipulations. A 0.72-mg-90-day–release- 17β-estradiol pellet (Innovative Research, USA) was implanted subcutaneously into each mouse one week before 1 × 107 ZR-75- 1 or MCF-7 cells with Matrigel (1:1) were injected into the right flank of each mouse. Mice with tumor size ~100 mm3 (i.e., in 2–4 weeks) were randomly into 4 groups and treated with 10% 2-hydroxypropyl-β-cyclodextrin (HPBCD), fulvestrant, MLN4924 or MLN4924 plus fulvestrant with the indicated doses. Tumor xenografts were measured with a caliper twice a week, and tumor volume was determined using the formula: [(Length× Width2)/2].

Animal 18F-FES Static PET/CT Scan

Female BALB/c athymic nude mice (aged 6–8 weeks) were maintained and treated in accordance with established guidelines and the protocol was approved by an internal animal protocol review committee. 18F-FES PET/CT scans were acquired before the mice were euthanasia. All mice were injected intravenously with 7.4 MBq of 18F-FES. PET/CT imaging was conducted 1 hour after the mice were isoflurane anesthesia. 3D static emission scans were obtained at the 20min time point. The OSEM3D/MAP assay was used to reconstruct the attenuation corrected PET/CT fusion image and the region of interest in the tumor and contralateral muscle were outlined for subsequent analysis. 18F-FES uptake at 20min in the tumor and contralateral muscle were noted as SUV hT and SUV hB, respectively. The relative 18F-FES uptake was noted as TBR (TBR is the tumor-to-background ratio, SUV hT/SUV hB).

Chromatin Immunoprecipitation (ChIP) Assay

ChIP assays were performed according to the manufacturer’s instructions (Active Motif, ChIP-IT® Express Chromatin Immunoprecipitation Kits, Carlsbad, CA) using MCF-7 cells grown to 80% confluence treated with MLN4924 or not. The binding of FOXO3a to ESR1 promoter was detected using specific FOXO3a binding site (S4) to ESR1 promoter (Forward: CCACTGGGAAATGAGAGACCTCGT; Reverse: GTGGATCAAATGCCTTACTGGCCCPCR). The enrichment of the ChIP signal was validated via quantitative real-time PCR (signal/input ratio). Reactions were done in duplicates.

Transfection and Luciferase Reporter Promoter Assay

MCF-7 cells were cultured in 96-well plates at a density of 2 × 104 cells per well. The cells were transfected with 800ng of either wild-type or mutant FBE-driven constructs of the promoterpGL3-basic with 4 ng/well of the control Renilla luciferase reporter construct pRL-TK (Promega Biotech, Co, Ltd) using FuGENE HD reagent (Promega Biotech, Co, Ltd) according to the manufacturer’s instructions for 24 hours, MLN4924 was added for another 24 hours. Luciferase assays were performed for firefly/Renilla luciferase assays as recommended by the Dual-Glo Luciferase Assay System (Promega, Southampton, UK).

Statistical analysis

Student’s t-tests were used to determine the statistical significance between two groups. The differences among multiple groups were analyzed by ANOVA. Pearson χ2 test was performed to compare categorical variables. Survival curves were analyzed using the Kaplan–Meier method, and the univariate survival difference was determined using the log-rank test. All the analyses were performed with SPSS 20.0.P values of less than 0.05 were considered statistically significant.For detailed information on all other experimental methods, please see the Supplemental Methods(available online).

Results

Higher NAE1 and nedd8 expressions are positively correlated with ER-positive breast cancer and predict poor prognosis.

To identify the role of neddylation activation in breast cancer, we first determined the expression patterns of NAE1 and nedd8, two key proteins inneddylation pathway, using IHC staining of 28 pairs of primary breast cancer tissues and normal tissues. According to the staining intensity, the samples were divided into five groups (±to +++). The expression intensity including strong (+++), medium (++) and weak (+) expression was shown in Fig. 1A. We found that the expressions of these two proteins were higher in breast tumor tissues than that of in adjacent tissues (Fig. 1B-C), indicating that neddylation pathway was activated in breast cancer.Next, we investigated whether neddylation activation was different depending on molecular subtypes. A total of 113 patients were recruited in our study. The patient and disease characteristics were summarized in Supplemental table1.The expressions of NAE1 and nedd8 were determined by IHC staining of tissue arrays, which were analyzed depending on ER status. Interestingly, both NAE1 and nedd8 expression was higher among ER-positive tumors (p<0.0001 and p<0.0001, respectively) (Fig. 1D-E). Moreover, Kaplan – Meier analysis showed poorer disease free survival (DFS) rate inpatients with elevated NAE1 and nedd8 expression (DFS; P= 0.039 and P=0.029, respectively; log-rank test) (Fig. 1F-G). In addition, nedd8 overexpression was an independent prognostic factor of poor DFS in multivariable survival analysis (HR = 2.462, 95% CI = 1.120 to 5.409, P =0.025) (Supplemental Table2). Furthermore, the Kaplan –Meier analysis stratified by ER status showed that the prognosis of ER-positive/neddylation-high activation (ER+/neddylaiton high) breast cancer was better than that of ER-/neddylation-low breast cancer in the first 80 months, while after 80 months, the prognosis of ER-/neddylation-low breast cancer became better than that of ER+/neddylation-high breast cancer (Fig. 1H-I). Taken together, these data suggest that NAE1 and nedd8 played central role in ER-positive breast cancer and the neddylation pathway may regulate ER-“ expression.

Neddylation pathway inactivation regulates ER-“ expression

To verify whether neddylation pathway regulates ER-“ expression, three ER-positive breast cancer cell lines, MCF-7, T47D and ZR-75- 1, were treated with MLN4924. The expression of ER-“ and its downstream target genes, PR and cathepsin D, were evaluated after treatment. P21(30) and P27(31), two classical substrates of Cullin RING E3 ligases that accumulate upon neddylation-CRLs inactivation, were determined as controls. Interestingly, ER-“ and its downstream target genes were decreased following MLN4924 treatment in time- and dose-dependent manners (Fig. 2A-B). In contrast, MLN4924 treatment did not impact the expression of HER2 and ER- β in the above three ER-positive breast cancer cells as well as triple positive cell BT-474 (Supplemental Fig.1A).

It is reported that ER is transcriptionally activated by phosphorylation which contribute to regulation of multiple biological processes including hormone sensitivity, nuclear localization, DNA binding, protein/chromatin interactions, protein stability and gene transcription (32, 33). The most common phosphorylation sites reported by numerous laboratories are serine 118 (Ser118) and serine 167 (Ser167) (34). We tested the expressions of pER (Ser118) and pER (Ser167) upon MLN4924 treatment.Strikingly, the expressions of pER (Ser118) and pER (Ser167) were significantly downregulated with MLN4924 treatment (Fig. 2A-B). The specificity of the effect of neddylation pathway inactivation with MLN4924 on the expression of pER(Ser118) and pER (Ser167), ER-α and its downstream target genes, PR and cathepsin D, was further validated by knocking down UBC12 and nedd8 expression levels using the CRISPR Cas9 system (Supplemental Fig. 1B-C). Moreover, we silenced UBA3 expression by siRNA and consistently, the expression levels of ER-α and its downstream target genes, PR and cathepsin D, were significantly decreased (Supplemental Fig. 1D).Next, we investigated the efficacy of MLN4924 on ER-α expression in vivo by 18F-FES-PET/CT, an approach frequently used for serial whole-body evaluation of ER expression (35, 36). As shown in Fig. 2C, FES uptake was significantly decreased after MLN4924 treatment. The relative FES uptake (tumour-to-background ratio, TBR) was also significantly decreased after MLN4924 treatment (P=0.026, paired t test) (Fig. 2D). The downregulation of ER-α expression (Fig. 2E) and PR expression (Supplemental Fig. 1E) in xenograft tissues upon MLN4924 treatment was further confirmed by IHC staining. Taken together, these data indicate that the neddylation pathway inactivation inhibits the ER-α expression in ER-positive breast cancer.

Inactivation of neddylation pathway represses the transactivation of ER-“To determine the underlying mechanism of ER-α inhibition by neddylation inactivation, we first applied cycloheximide (CHX) to block protein translation and determined ER-α turnover rate upon MLN4924 treatment. We found that neddylation inactivation by MLN4924 had no effect on ER-α turnover in both MCF-7 (Fig. 3A) and T47D cells (Fig. 3B) when it delayed the degradation of P21 and P27 due to CRLs inhibition. Similarly, treatment of cells with MG132, a classical proteasome inhibitor, also did not change the inhibitory effect of MLN4924 on ER-α expression (Fig. 3C). We next determined whether MLN4924 regulates ER-α expression at the transcriptional level. Interestingly, the transcription of ER-α was significantly inhibited by MLN4924 in MCF-7, T47D and ZR-75- 1 cells in a dose-dependent manner (Fig. 3D). Moreover, the transcription of ER-α target genes, PR (PgR) and cathepsin D (CTSD), were also significantly inhibited after MLN4924 treatment in MCF-7, T47D and ZR-75- 1 cells (Supplemental Fig. 2A-C). Similarly, genetic inactivation of the neddylation pathway by UBA3 knockdown substantially inhibited the transcription of ER-α (Fig. 3E) and its target genes, PgR and CTSD (Supplemental Fig. 2D-F) in MCF-7, T47D and ZR-75- 1 cells. These findings demonstrate that MLN4924 inhibits ER-α transcription.

MLN4924 inhibits ER-“ transcription via serum and glucocorticoid-inducible kinase (SGK) induced FOXO3a nuclear export in MCF-7 and ZR-75-1 cells

It is reported that FOXO3a and FOXM1 transcriptionally regulated ESR1 activity by directly binding to the ESR1 promoter (37, 38) and our study confirmed the positive regulation of FOXO3a and FOXM1 on ER-α (Supplemental Fig. 3A-B). Interestingly, we found that the expression of FOXO3a in the nucleus was significantly decreased upon neddylation inhibition ，while the expression of FOXM1 in the nucleus showed no change upon neddylation inhibition in MCF-7 (Fig. 2′-3′-cyclic GMP-AMP Sodium 4A) and ZR-75- 1 cells (Supplemental Fig. 3C). We therefore hypothesized that MLN4924 might inhibit ER-α expression through regulating FOXO3a nuclear localization. To further test this hypothesis, cells were treated with Leptomycin B (LMB), a specific inhibitor of the nuclear transport receptor, CRM1/XPO1, that blocks the nuclear export of protein. As hypothesized, in the presence of LMB, cells treated with MLN4924 showed an obvious total and nuclear ER-α restore in MCF-7 (Fig. 4B-C) and ZR-75- 1 cells (Supplemental Fig. 3D-E).Next, we explored whether MLN4924 decreased the binding of FOXO3a to the ESR1 promoter to inhibit ESR1 transcription. It is reported that there was a consensus ‘‘TGTTTAC’’ sequence of the FOXO binding element (FBE) in the ESR1 promoter named S4 (39) (Fig. 4D, upper). Chromatin immunoprecipitation (ChIP) assay was used to determine the binding of FOXO3a to the ESR1 promoter after MLN4924 treatment. We found that the binding of FOXO3a to the ESR1 promoter was significantly decreased upon MLN4924 treatment in MCF-7 (Fig. 4D, lower) and ZR-75-1 cells (Fig. 4E), while the binding of FOXM1 to the ESR1 promoter showed no change in the above cells (Supplemental Fig. 3F-G). This finding was also confirmed by steptavidin-agarose pull down assay that the FOXO3a pulled down by ESR1 S4 probe was significantly decreased upon MLN4924 treatment (Fig. 4F),while the FOXM1 pulled down by ESR1 S4 probe was unchanged upon MLN4924 treatment (Supplemental Fig. 3H).

To further determine whether FOXO3a is crucial forESR1 inhibition by MLN4924, we constructed a luciferase reporter driven by either the wild-type FBE from the ESR1 promoter or its mutant form (Fig. 4G, upper), and transfected them into MCF-7 cells, followed by MLN4924 treatment. The result suggested that MLN4924 decreased expression of the reporter driven by the wild-type FBE, but not by the mutant FBE (Fig. 4G, lower), indicating that FOXO3a is required for the inhibition of ESR1 by MLN4924. These findings suggest that MLN4924 promotes FOXO3a nuclear export and further decreases its binding to the ESR1 promoter and finally inhibits ER-“ expression.

Subsequently, we explored how MLN4924 promoted FOXO3a nuclear export. It is reported that FOXO3a phosphorylation is necessary for its nuclear localization (40). In the absence of stimulus, when AKT and serum and glucocorticoid-inducible kinase (SGK1) are inactive, FOXO3a is localized in the nucleus (40). When cells are exposed to stimulus, the AKT or SGK1 pathway is activated and triggered the phosphorylating of FOXO3a at Ser253 and Ser318/321, respectively. Phosphorylated FOXO3a transformed its conformation and revealed its nuclear export signal (NES) domain and was further exported by CRM1/XPO1 from the nucleus (40). Given that SGK1 has been reported as the substrate of CRL E3 ligase (41), we then hypothesized that MLN4924 promoted FOXO3a phosphorylation via blocking SGK1 degradation. Indeed, we found that MLN4924 significantly delayed the degradation of SGK1 (P=0.0214, paired t test) (Fig. 4H). Moreover, MLN4924 markedly increased the expression of SGK1 and phosphorylated FOXO3a at Ser318/321 (Fig. 4I), while the expression of phosphorylated FOXO3a at Ser253 induced by AKT showed no change (Fig. 4I).

To further identify whether MLN4924 promoted FOXO3a nuclear export and inhibited ER-“ expression via SGK1 accumulation, we knocked out SGK1 using six guide RNAs (gRNAs) targeting SGK1 by the CRISPR/Cas9 system. The result showed that the 5th and 6th gRNAs worked more effectively (Supplemental Fig. 3I). As we hypothesized, SGK1 knockdown partially rescued ER-“ inhibition by MLN4924 (Fig. 4J). In addition, we found that FOXO3a phosphorylation at Ser318/321 was increased after MLN4924 treatment, while SGK1 knockout rescued the increase of FOXO3a phosphorylation by MLN4924 (Fig. 4J). To further investigate the necessity of SGK-induced FOXO3a phosphorylation and nuclear export in ER-“ inhibition with MLN4924 treatment, we generated mutants of serine 318/321 site and the nuclear export signal (NES) domain of FOXO3a in which the serine 318/321 residues and the critical leucine or methionine in NES were replaced by alanines (Fig. 4K, upper). Interestingly, the total (Fig. 4K, lower) and nuclear ER-“ expression (Fig. 4L) were partially restored in both S318/321A mutant and NES mutant of FOXO3a-dependent manners. Both mutants of FOXO3a remain localized in the nucleus in the presence of MLN4924 (Fig. 4L). Moreover, the growth inhibition by MLN4924 was partially restored in both S318/321A mutant and NES mutant of FOXO3a-dependent manners (Supplemental Fig. 3J). Altogether, these results indicate that MLN4924 promotes SGK1-induced FOXO3a phosphorylation at Ser318/321 followed by increased FOXO3a nuclear export. As a result, the binding of FOXO3a to the ESR1 promoter is decreased and finally contributes to ER-“inhibition.

Downregulation of ER-α expression is required for MLN4924-induced cell growth suppression

Targeting ER-“ is an effective way to treat ER-positive breast cancer (3, 4). Due to the inhibition of ER-“ expression by MLN4924, we next evaluated the effect of MLN4924 on the growth of ER-positive breast cancer cells (MCF-7, and ZR-75-1). MLN4924 significantly inhibited cell proliferation and clonogenic survival in dose-dependent manners (Fig. 5A-B). To determine whether MLN4924-induced ER-“ inhibition is crucial for inhibiting the cellular malignant phenotype, we transduced full-length cDNA of human ER-“ into MCF-7 and ZR-75- 1 cells (Supplemental Fig. 4A) and evaluated the effect of ER-“ overexpression on MLN4924-induced cell growth inhibition. As shown in Supplemental Fig. 4B, ER-“ overexpression potently blocked the MLN4924-induced reduction of ER-“ expression. Notably, overexpression of ER-“ significantly precluded the inhibition of cell growth and clonogenic survival by MLN4924 in both MCF-7 and ZR-75- 1 cells (Fig. 5C-D). Taken together, these results indicate that MLN4924 suppresses the proliferation of ER-positive breast cancer cells in part via ER-“ downregulation.

MLN4924 synergizeswith fulvestrant to inhibit ER-“ expression and cell growth

Considering the inhibition of ER-α expression by MLN4924, we hypothesized that MLN4924 might increase the sensitivity of breast cancer cells to fulvestrant, a selective ER degrader (SERD), by downregulating ER-α expression synergistically. As shown in Fig. 6A, MLN4924 synergized with fulvestrant to inhibit ER-α expression in MCF-7 and ZR-751 cells. To confirm whether MLN4924 synergized with fulvestrant to suppress cell growth, we treated MCF-7 and ZR-75- 1 cells with increasing concentrations of MLN4924 and fulvestrant alone and in combination (28). Interestingly, we observed a synergistic interaction between MLN4924 and fulvestrant in suppressing cell proliferation, with optimal combination index (CI) of 0.04 and 0.045, respectively (Fig. 6B). Moreover, MLN4924 synergized with fulvestrant to suppress the clonogenic survival of MCF-7and ZR-75- 1 cells (Fig. 6C-D). In contrast, we found that MLN4924 did not synergize with tamoxifen (Supplemental Fig. 5A-B), another ER-α antagonist and the selective ER modulator (SERM), which directly inhibits the ligand E2 and ER-α interactions and is the most commonly used ER antagonist, particularly in first-line therapy (42). Furthermore, we found that tamoxifen had no effect on ER-α expression (Supplemental Fig. 5C). Collectively, these results show that neddylation pathway inactivation with MLN4924 increases sensitivity of the ER-positive breast cancer cells to fulvestrant.

MLN4924 and fulvestrant synergistically suppress ER-positive breast cancer using ZR-75-1 and MCF-7 xenografts

Next, we determined whether MLN4924 and fulvestrant synergized to suppress ER positive breast cancer in vivo. We found that, compared to MLN4924 or fulvestrant treatment alone, the combination of these two agents almost completely inhibited tumor growth derived from ZR-75- 1 cells (Fig. 7A), significantly reduced tumor volumes (Fig. 7B) and tumor weight (Fig. 7C). During the whole experiment, the body weight of the mice in the four groups was not substantially changed (Fig. 7D). In addition, combination of MLN4924 and fulvestrant significantly inhibited tumor growth derived from MCF-7 cells (Supplemental Fig. 6A). The body weight of the mice in the four groups showed no significant decrease after treatment (Supplemental Fig. 6B). Mechanistic studies revealed that the combination of MLN4924 and fulvestrant significantly enhanced inhibition of ER-“ and PR expression in the collected tumor samples, as demonstrated by IHC (Fig. 7E). Consistently, the proliferation index, ki67, was dramatically decreased upon fulvestrant-MLN4924 combined treatment (Fig. 7E). Together, these findings demonstrate that targeting the neddylation pathway represents a novel strategy to sensitize ER-positive breast cancer to fulvestrant.

Discussion

ER, as one of the most successful molecular targets in the history of anti-breast cancer drug discovery, determines the sensitivity and effectiveness of endocrine therapy of breast cancer. There are currently limited studies investigating the mechanism for ER regulation and by which searching effective targets for endocrine-resistant tumor. In this study, we found the neddylation pathway was a novel regulator of ER-“ expression in breast cancer. a) neddylation pathway is overactivated in breast cancer tissues and its activation status is positively correlated with ER-positive breast cancer progression; b) pharmaceutical and genetic inactivation of neddylation pathway significantly inhibits the transactivation of ER-“ and its target genes; c) mechanistically, the SGK1/FOXO3a axis is involved in neddylation pathway inhibition-induced ER-“ downregulation; d) targeting neddyaltion with MLN4924 significantly improves the sensitivity to fulvestrant. Thus, our findings define the neddylation pathway as an intrinsic regulatory signaling of ER-“ and as a sensitizer for fulvestrant treatment in ER-positive breast cancer.In this study, we firstly found that neddylation was activated more significantly in ER-positive disease. It seems counter-intuitive that neddylation overactivation which is associated to poorer prognosis, is more activated in ER-positive disease. Typically, ER-positive disease shows slower growth rate, clear differentiation and longer DFS (43). Patients with ER-positive tumors are at continued risk of relapse for many years after their initial breast cancer diagnosis (44). In a meta-analysis of 10,000 patients, ER-negative tumors were found to have a poorer prognosis in the first few years after diagnosis, but after 5 to 10 years, ER-positive tumors were associated with relatively poorer outcomes (45). In ER-positive subgroup, NAE1 and nedd8 showed no prognostic value in the first 80 months, while after 80 months, NAE1 and nedd8 expressions indicated poor prognosis. The explanation for this result may be that the good prognostic value of ER is superior to the poor prognostic value of neddylation in ER-positive breast cancer for the first 80 months. This is accordance with the overview data from the Early Breast Cancer Trialists Collaborative Group (EBCTCG) that ER-positive disease has better prognosis in the first few years (46). After 80 months, the poor prognostic value of neddylation is superior to the good prognostic value of ER, which contributes to the recurrence and resistance of ER-positive breast cancer.Secondly, we explored a novel mechanism regulating ER-α expression. Neddylation inactivation transcriptionally inhibits ER-α expression, which is contradict to the conclusion of the study by Fan et al in 2003 which suggests that impaired neddylation via direct inhibition of nedd8 leads to stabilization of ER-α protein (47). However, in Fan’s study, they investigated the effect of neddylation inactivation on ER-α degradation in the presence of ligand estradiol in Hela cells, but not in ER-positive breast cancer cells. Although previous studies have reported that ER-α can be degraded via ubiquitin-proteasome system (48), we found that neddylation inactivation by MLN4924 exerted no effect on the degradation of ER-α, in stand, it inhibited ER-α transcription.

Mechanistically, we found that neddylation inactivation regulates the transcription of ER-α partially via SGK1-denpendent cytoplasmic localization of FOXO3a (Fig. 7F). To our knowledge, this is the first study reported that the regulation of ER-α through SGK1. Importantly, this is also the first time we found that neddylation transcriptionally regulates ER-α expression and play essential role in the response of ER-positive breast cancer to fulvestrant treatment.Fulvestrant is an important drug to treat ER-positive breast cancer by promoting the proteasomal degradation of the ER-“ (10). However, the efficacy of fulvestrant was still limited (17- 18). In this study, we found that MLN4924 sensitizes ER-positive breast cancer cells to fulvestrant, but not tamoxifen to suppress ER-positive breast cancer growth. This phenomenon could be interpreted as that fulvestrant targets ER for degradation (10), while tamoxifen competitively binds ER to block the E2-ER signaling pathway (49).In conclusion, our study reveals that neddylation is overactivated in ER-positive breast cancer. Inactivation of neddylation with MLN4924 transcriptionally inhibits ER-“ partially via SGK1-denpendent nuclear export of FOXO3a and improves the sensitivity of ER-positive breast cancer to fulvestrant treatment. Furthermore, it is conceivable that the essential components of the neddylation pathway may also be targeted in combination with endocrine therapy to induce synergistic effects in ER-positive breast cancer.