Introduction
Breast cancer is the most common cancer and the leading cause of cancer mortality among women [
1]. With improved screening, more and more breast cancer patients are being detected early. Although there are many strategies for early breast cancer, recurrence and metastasis are still the primary causes of breast cancer treatment failure. Metastasis is no longer believed to involve a linear cascade of events, but rather multiple signaling pathways, such as those related to the cell cycle [
2] and tumor microenvironment [
3]. According to clinical guidelines, ionizing radiation and chemotherapies are the major choices for postoperative adjuvant therapies in early breast cancer. Their main mechanism involves DNA damage [
4]. However, partial breast cancer therapy may result in endogenous factors antagonizing of the DNA damage caused by ionizing radiation or chemotherapy due to an upregulation of the DNA damage response, which comprises cell cycle checkpoints and DNA repair [
5]. Cell cycle checkpoint kinase inhibitors can sensitize breast cancer cells to radiotherapy or chemotherapy. Unfortunately, it seems that these therapies cannot effectively kill breast cancer cells by themselves, and there is an absence of appropriate biomarkers [
6,
7].
Ring finger protein 126 (RNF126) is an E3 ubiquitin ligase that is involved in diverse biological processes, including cell proliferation [
8], DNA repair [
9], and the cell cycle [
10]. Overexpression of RNF126 can promote proliferation in the tongue [
11] and gastric cancer [
8] cells. In addition, RNF126 facilitates DNA double-stranded break repair by promoting homologous recombination [
12] and nonhomologous end joining [
13]. However, the proliferation regulatory ability of RNF126 in breast cancer metastasis has not yet been evaluated.
Ataxia telangiectasia mutated and Rad3-related kinase (ATR), a cell cycle checkpoint kinase, can be activated in response to replication protein A-coated single-stranded DNA caused by endogenous or exogenous factors [
14]. ATR-phosphorylated CHEK1 results in cell cycle arrest at the G2/M checkpoint, allowing for DNA repair. ATR plays a major role in preventing cells with incomplete DNA replication from undergoing mitosis after exposure to DNA-damaging agents, such as chemotherapeutic drugs or ionizing radiation [
15]. Suppression of ATR signaling increases firing of dormant origins, leading to massive fork collapse and DNA breakage [
16]. Accordingly, ATR inhibitors have been developed and are currently being used in preclinical and clinical studies. In addition, ATR inhibitors paired with genotoxic chemotherapeutics or radiotherapy have synergistic activity in numerous cancer cells [
17‐
19]. However, ATR inhibitor monotherapy in cancer cells has rarely been reported. Moreover, ATR activity itself may not be sufficient as an effective target for ATR inhibitors [
20,
21].
Here, we identify a specific metastasis-related gene expression signature from a microarray dataset comprising early breast cancer patients without lymphatic metastasis (GSE11121) in the Gene Expression Omnibus (GEO) [
22]. The dataset showed that RNF126 is associated with breast cancer metastasis. We applied the signature to The Cancer Genome Atlas (TCGA) to determine RNF126 was highly expressed in the tumor tissues and constructed RNF126-related network modules through weighted gene coexpression network analysis (WGCNA). Gene Set Enrichment Analysis (GSEA) also implied that the cell cycle pathway is enriched in patients overexpressing RNF126. Additionally, we characterized the molecular function of RNF126, examining its proliferation ability in both in vitro and in vivo experiments. Treatment with ATR inhibitors increased cell death by triggering abnormal origin firing in breast cancer cells with higher RNF126 expression. CDK2 may mediate the killing effect of ATR inhibitors on RNF126 high-expression breast cancer cells. In brief, our results suggest that higher expression of RNF126 may accelerate breast cancer metastasis and that RNF126 may be an effective biological target for ATR inhibitors.
Materials and methods
The GSE11121 samples were processed using the R package WGCNA for removing outliers. After clustering, 199 breast cancer samples were included, GSM282518 was excluded. A univariable Cox proportional hazards regression model was used to select metastasis-related genes with |coefficient|> 0.50, P value < 0.05. Metastasis-related differentially expressed genes were analyzed by using Gene Ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses, and the GSEA with the clusterProfiler package of R. The KEGG pathways and GO terms regarding cellular component, molecular function, and biological process with P values and false discovery rates less than 0.05 were considered statistically significant.
Then, WGCNA was performed to identify RNF126-related signal pathways. The similarity matrix was transformed into an adjacency matrix with a network type of signed and a soft threshold of β = 5 and then transformed into a topological matrix with the topological overlap measure (TOM) describing the degree of association between genes. 1-TOM was used as the distance to cluster the genes, and then the dynamic pruning tree was built to identify the modules. We identified 5 modules by setting the merging threshold function at 0.20. The enrichment P values for the GSEA were based on 1000 permutations and adjusted by calculating the false discovery rates. The GSEA results were visualized using the R package enrichplot.
Cell culture and transfection
Human breast cancer cell lines (MCF7 and MDA-MB-231) were cultured in DMEM medium supplemented with 10% fetal bovine serum (FBS), and maintained at 37 °C in a humidified incubator with 5% CO2. All DNA-plasmid transfections were performed using Lipofectamine 2000 according to the manufacturer's recommendations (Invitrogen). The sequences of shRNA used were as follows: shRNF126#1 5′-TGCATGGTTTGTGGCGGAAGA-3′; shRNF126#2 5′-CAACGAGAACGCCACATGGTC-3′.
Quantitative real-time polymerase chain reaction (qRT-PCR)
The total RNA was isolated with the TRIzol reagent (Invitrogen). Novogene (Beijing, China) completed the cDNA library construction. Experiments were carried out in triplicate for each data point. Reactions were performed using SYBR Green mix (Roche) and a MyiQ real-time PCR detection system (Bio-Rad). Relative mRNA levels were calculated using the comparative Ct method (ΔCt).
GAPDH forward primers: 5′-CTCTGCTCCTCCTGTTCGAC-3′; reverse primers: 5′-TTAAAAGCAGCCCTGGTGAC-3′.
RNF126 forward primers: 5′-TATCGAGGAGCTTCCGGAAGAGA-3′; reverse primers: 5′-AAAGCAAACTGTCCGTAGCCCT-3′.
CDK1 forward primers:5′-GGATGTGCTTATGCAGGATTCC-3′; reverse primers: 5′-CATGTACTGACCAGGAGGGATAG-3′.
CDK2 forward primers: 5′-TATTAACACAGAGGGGGCCA -3′; reverse primers: 5′-AAAGATCCGGAAGAGCTGGT-3′.
CDK5 forward primers:5′-GGAAGGCACCTACGGAACTG-3′; reverse primers: 5′-GGCACACCCTCATCATCGT-3′.
Scratch wound assay
The cells were inoculated in a 6-well plate, scraped through each hole with the tip of a sterile 10 μL pipette and washed with phosphate-buffered saline to remove any debris. After 24 h, the cells migrated to the empty space.
Cell invasion and migration assays
Approximately 2 × 104 cells in 300 μL DMEM medium without FBS were seeded into upper transwell chamber (8 μm pore size) to evaluate cell migration. The lower chamber was filled with 800 μL DMEM medium supplemented with 10% FBS. After 24 h, the cells attached to the lower surface of the membrane were fixed with 4% formaldehyde, stained with 0.5% crystal violet, and then counted under a microscope in five random fields. Invasion assays were performed under the same conditions as the migration assays, but in matrigel-coated transwell (Corning, NY, USA) inserts.
Cell viability and calculation of half-maximal inhibitory concentration (IC50)
The cells were seeded in 96-well plates in 100 μL DMEM medium containing 10% FBS, at a density of 2 × 103 cells per well. The cells were exposed to various doses of inhibitors and assayed for viability at indicated times, using the MTT according to the manufacturer's instructions. In brief, MTT (20 mL of 5 mg/mL) was added to each well and cells were incubated for a further 3.5 h in an incubator. MTT solvent was added after removing the medium and the cells in plates were agitated on an orbital shaker for 15 min. The absorbance was read at 590 nm with a reference filter of 620 nm. The absorbance values were normalized with respect to those of untreated control cells. The IC50 was calculated using nonlinear regression analysis in GraphPad Prism 6.0.
In vivo studies
MCF7 and MDA-MB-231 cells (1 × 107) in 150 μL PBS were subcutaneously injected into the right flank of female nude mice. Tumor volume was measured by caliper and calculated as length × width2/2. When tumor volume grew up to 50–100 mm3, the mice were randomly divided into two groups (five mice per group), and then treated with PBS daily, AZD6738 (50 mg/kg, oral, daily). Tumor volume was measured every 3 days. In assays to measure formation of metastases, 107 breast cancer cells were injected into tail veins of mice. The number of metastases was assessed in 3 or 6 weeks, respectively. All the animal experiments were carried out with the approval of the guidelines of Guangxi Medical University Cancer Hospital.
Immunohistochemistry
The formalin-fixed mouse tumor, liver, or lung tissues were embedded with paraffin. The treated tissues were sectioned (3 μm) and stained with the hematoxylin and eosin (H&E). H&E-stained liver or lung sections were imaged using a microscope (Olympus). For immunostaining, slides were heated to 60 °C and then deparaffinized in xylene. The slides were rehydrated in descending alcohol concentrations. Antigen retrieval was performed by incubating slides in a retrieval solution of citrate buffer. Hydrogen peroxide was added to block endogenous peroxidase activity to decrease unwanted background staining. Primary antibody (RNF126, ab234812, Abcam) was added at 1:100 dilution. The substitution of primary antibody performed negative controls with phosphate-buffered saline (PBS). To guarantee consistent IHC evaluation, slides from a reference tumor previously determined as positive were included in each staining procedure. Evaluations of staining reactions were performed in accordance with the immunoreactive score (IRS) proposed by Remmele and Stegner: IRS = staining intensity (SI) X percentage of positive cells (PP). Staining intensity was marked as nongranulated (0); low grade (light yellow; 1); moderate (brownish yellow; 2); or strong (reddish brown; 3). The PP was scored as negative (< 5%; 0); weak (5–10%; 1); moderate (11–50%; 2); strong (51–80%; 3); or very strong (> 81%; 4). Specimens scoring beyond 3 were considered positive overexpression. Slides were studied with the microscopic (Olympus).
Immunoblotting
Cellular extracts were prepared by resuspending cells in radio immunoprecipitation assay (RIPA) buffer. After protein samples were separated by 5%, 12%, or 15% SDS-PAGE, they were transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Burlington, MA). Around 5% BSA was used to incubate the PVDF membrane for 1 h at room temperature, and then at 4 °C overnight with antibodies specific to glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (AP0066, 1:10,000; Bioworld), RNF126 (C-1 SC-376005, 1:100; Santa Cruz Biotechnology); ATR (C-1 SC-515173, 1:100; Santa Cruz Biotechnology); p-ATR (S428 AB178407, 1:100; Abcam); γ-H2AX (Ser139 SC-517348, 1:100; Santa Cruz Biotechnology); p-RPA2 (S4/S8; rabbit polyclonal, BL647, 1: 1000; Bethyl Laboratories); CHEK1 (AB32531, 1:100; Abcam); CDK1 (AB265590, 1:500; Abcam); CDK2 (610,146, 1:200; BD Biosciences); CDK5 (AB40773, 1:500; Abcam); Cleaved PARP (AB4830, 1:200; Abcam). After washing in Tris-buffered saline with 0.1% Tween 20 (TBST) for three times (10 min each time), the PVDF membranes were incubated with secondary antibodies (Goat anti-mouse IgG-horseradish peroxidase (HRP)–conjugated (#7076S, 1:1,000; Cell Signaling Technology), goat anti-rabbit IgG-HRP–conjugated (#7074S, 1:1000; Cell Signaling Technology), and donkey anti-goat IgG-HRP–conjugated (A2216, 1:1,000; Santa Cruz Biotechnology)) for 1 h at room temperature. By using enhanced chemiluminescence blotting reagents, proteins were detected after three TBST washes (FUDE Biological, Hangzhou, China). Signal intensity was assessed by using a Tanon-5500 chemiluminescence detection system (Tanon Science & Technology Ltd, Shanghai, China).
Immunofluorescence analysis
Cells growing on slides were fixed directly in 3–4% paraformaldehyde. Cells were extracted for 5 min on ice with 0.5% Triton X-100 in cytoskeletal (CSK) buffer (10 mmol/L PIPES, 300 mmol/L sucrose, 100 mmol/L NaCl, 3 mmol/L MgCl2; pH = 6.8) supplemented with 1 mmol/L phenylmethylsulfonyl fluoride, 0.5 mmol/L sodium vanadate, and proteasome inhibitor for 10 min at 4 °C. Then, extracted cells were fixed with 3–4% paraformaldehyde. The cells were permeabilized for 10 min with PBS containing 0.5% Triton X-100 for 15 min at room temperature, followed by blocking with 1% BSA, and then incubated with primary antibodies specific to CDC45 (H-300 clone, SC20685, 1:50; Santa Cruz Biotechnology). The bound antibodies were revealed with chicken anti-rabbit IgG Alexa Fluor 488. Slides were viewed at 1,000 magnifications with a NIKON 90i fluorescence microscope (photometric cooled mono CCD camera, NIKON, Tokyo, Japan).
DNA fiber assays
DNA fiber assays were performed as published with some modifications [
23]. Cells were pulse-labeled with 50 mmol/L IdU (Sigma-Aldrich #I7125) for 40 min and then pulse-labeled with 200 mmol/L CldU (Sigma–Aldrich #C6891) for 40 min in the presence or absence of ATR inhibitor. At the end of the CldU pulse, cell suspensions (2.5 mL) were mixed with 7.5 mL of lysis buffer (0.5% SDS, 200 mmol/L Tris–HCl (pH 7.4), 50 mmol/L EDTA). Each mixture was dropped on the top of an uncoated regular glass slide. Slides were inclined at 25° to spread the suspension on the glass. Once dried, DNA spreads were fixed by incubation for 10 min in a 3:1 solution of methanol-acetic acid. The slides were dried and placed in precooled 70% ethanol at 4 °C for at least 1 h. DNA was denatured with 2.5 mol/L HCl for 30 min at 37 °C. The slides were blocked in 1% BSA in PBS for 30 min at room temperature and then incubated with mouse anti-BrdU antibody (BD Biosciences #347,580) at a 1:200 dilution and rat anti-CldU antibody (Abcam #ab6326) at a 1:400 dilution. The slides were incubated with secondary fluorescent antibodies [goat anti-mouse IgG (H + L) Alexa Fluor 594 secondary antibody (A-11032, 1:400; Thermo Fisher Scientific); or chicken anti-rabbit IgG (H + L) Alexa Fluor 488 secondary antibody (A-21441, 1:400); Thermo Fisher Scientific]. Replication fibers were viewed at 1000 magnifications on a NIKON 90i fluorescence microscope (photometric cooled mono CCD camera, NIKON, Tokyo, Japan). Signals were measured using ImageJ software (NCI/NIH), with some modifications made specifically to measure DNA fibers.
Statistical analysis
Statistical analyses were undertaken using the statistical software package, R version 4.0.4. Univariate Cox regression analysis was performed to select genes with P < 0.05 for metastasis. Univariate survival analysis was performed by Kaplan–Meier survival analysis with the log-rank test. Paired t-test, one-way or two-way ANOVA analysis of variance were used to determine the significance of differences between groups. Continuous variables were expressed as mean ± SD. In all the statistical analyses, P < 0.05 was considered as statistically significant.
Discussion
Recurrence and metastasis are the main reasons for treatment failure in early breast cancer, however, the mechanism is still unclear. RNF126 promotes cell proliferation in a variety of cancer cells [
8,
11], but the biological function of RNF126 in promoting breast cancer metastasis in early breast cancer is lacking. A reanalysis of the GSE11121 dataset of early breast cancer patients without lymph node metastasis found that RNF126 could promote breast cancer metastasis (Fig.
1A), and we verified this result in vitro and in vivo experiments (Fig.
2). Furthermore, analysis of 44 cases of early breast cancer patients without lymph node metastasis in TCGA found that RNF126 was highly expressed in tumor samples compared with normal breast tissues (Fig.
1B). These results are close to our previously reported. We have reported that RNF126 may be highly expressed in invasive breast cancer and that its expression is associated with a poor prognosis [
24]. However, both datasets contained only a few samples. Thus, further study with a larger number of early breast cancer cases without lymph node metastasis is required.
Regardless, these results still suggested that RNF126 might be a potential therapeutic biomarker for breast cancer, even though the relationship between RNF126 protein expression and metastasis was still unclear. Using WGCNA, we found some signaling pathways related to the high expression of RNF126, such as cell cycle, proteasome, spliceosome pathway (Fig.
1C and D). It is well established that cell cycle machinery is controlled by such mechanisms, including feedback loops of genes and protein products that display periodic activation and repression, which are associated with cell proliferation and apoptosis [
33]. We postulated that RNF126 may activate the cell cycle pathway and thereby promoted thereby promote breast cancer's malignant evolution. ATR is an important protein in the cell cycle signaling pathway. Targeting of the ATR is a promising strategy for the treatment of cancer. However, there is currently a lack of effective correlates with guiding the use of ATR inhibitors, and the expression level of ATR cannot effectively predict the effectiveness of ATR inhibitors. Moreover, we also found that ATR's expression level cannot predict breast cancer cells' metastasis (Additional file
1: Fig. S13). RNF126 can regulate the expression of CHEK1 protein and can be used as a predictive biomarker of CHEK1 inhibitors. ATR inhibitors are reported to be similar to CHEK1 inhibitors and we speculated that RNF126 might be a determinant of ATR inhibitors. In our model, we determined that breast cancer cells with a higher level of RNF126 are more sensitive to ATR inhibitors than breast cancer cells with a lower level of RNF126.
However, there is considerable crosstalk in the DNA damage response network, and CHEK1 is also activated by claspin after replication stress, independently of ATR [
34]. There is also an ATR-dependent, CHEK1-independent, intra-S phase checkpoint that suppresses origin firing [
35]. Moreover, ATR and CHEK1 have distinct functions and may not act linearly in the kinase cascade [
36]. In addition, the combination of ATR and CHEK1 inhibitors has a synergistic lethal effect, and our results also supported this conclusion (Additional file
1: Figs. S8A, S9A). Interestingly, although we have previously reported that RNF126 knockdown reduces CHEK1 expression (
24), we did not observe a synergistic lethal effect of ATR inhibitors on breast cancer cells with knockdown of RNF126 (Additional file
1: Figs. S8B and S9B). p-ATR can represent the functional status of ATR protein [
25], that was examined p-ATR expression by western blot. The results showed that RNF126 knockdown increased p-ATR expression, which meant that the function of ATR protein had been activated, but ATR inhibitors could not effectively kill breast cancer cells with RNF126 knockdown (Fig.
3). These results implied that RNF126 may be a biomarker of ATR inhibitors.
It has been reported that ATR suppresses oncogene-induced replication stress and that ATR inhibitor monotherapy can effectively damage cells with high replication stress [
37]. We hypothesized that the lower efficiency of ATR inhibitors in breast cancer cells with RNF126 knockdown might be explained by the relative reduction in replication stress [
38,
39]. Activation of oncogenes leads to replication stress, as an excess of ongoing replication forks will consume the limited dNTP pool and causes fork stalling. We demonstrated that breast cancer cells with higher RNF126 expression could stall the replication fork and trigger abnormal replication initiation after ATR inhibitor application. However, the replication process of breast cancer cells with RNF126 knockdown was only mildly affected, with only mild changes in the replication elongation rate and new replication initiation rate (Fig.
4). These results confirmed our hypothesis that their own replication stress causes the ATR inhibitor sensitivity of breast cancer cells with higher RNF126 expression.
Wee1 deactivation leads to increased dNTP demand and replication stress through CDKs-induced firing of dormant replication origins [
40]. Oncogene-deregulated CDKs activity is required to manifest the synthetic lethality of ATR and CHEK1 inhibitors [
26]. Thus, we presumed that CDKs induce endogenous replication stress in breast cancer cells with RNF126. After using CDKs inhibitors to pretreat breast cancer cells with a higher level of RNF126, we observed that the cell-killing effect of ATR inhibitors could be counteracted by the inhibitors, in conjunction with an alleviates replication stress. Given that both NU2058 and Roscovitine have some interaction with CDK1/CDK2/CDK5, we detected the protein and mRNA levels of CDK1/CDK2/CDK5 by knockdown or overexpression of RNF126, we found that the expression level of RNF126 could affect the expression of CDK2, but not CDK1 or CDK5. In addition, overexpression of RNF126 C229A/C232A (the ubiquitination function of RNF126 was inactivated) [
41] could not affect the expression of CDK2. These results showed that RNF126 might regulate the expression of CDK2 through transcription rather than ubiquitination (Fig.
5). E2F1 regulated expression of CDK2 by binding the promoter region of CDK2 [
42]. RNF126 also binds to E2F1 [
12], and the ability of RNF126 to regulate CDK2 might be mediated by E2F1. This hypothesis needs to be further verified. Moreover, overexpression of CDK2 in RNF126 knockdown cells accelerated the cell-killing effect. Thus, CDK2-mediated replication stress in breast cancer with a high level of RNF126 expression might be one of the reasons for the sensitivity of these breast cancer cells to ATR inhibitors.
Notably, despite the use of ATR inhibition as a monotherapy strategy to target chronic lymphocytic leukemia cells with TP53 defects [
43], our results showed that breast cancer cells with high expression of RNF126 had enough endogenous replication stress via CDK2 to mediate and affect the sensitivity to ATR inhibitors in breast cancer cells with or without wild-type TP53 (Fig.
5P). This was consistent with published data reporting that ATR inhibition can target cancer cells as single agents irrespective of TP53 status [
44]. Thus, acting as single agents, the mechanisms by which ATR inhibitors lead to cell death might be distinct. Overall, RNF126 has a greater advantage than ATR, or p-ATR expression alone, as a biomarker of ATR inhibitors.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.