Influence of cyclin D1 splicing variants expression on breast cancer chemoresistance via CDK4/CyclinD1‐pRB‐E2F1 pathway

Abstract Cyclin D1 (CCND1), a mediator of cell cycle control, has a G870A polymorphism which results in the formation of two splicing variants: full‐length CCND1 (CCND1a) and C‐terminally truncated CCND1 species (CCND1b). However, the role of CCND1a and CCND1b variants in cancer chemoresistance remains unknown. Therefore, this study aimed to explore the molecular mechanism of alternative splicing of CCND1 in breast cancer (BC) chemoresistance. To address the contribution of G870A polymorphism to the production of CCND1 variants in BC chemoresistance, we sequenced the G870A polymorphism and analysed the expressions of CCND1a and CCND1b in MCF‐7 and MCF‐7/ADM cells. In comparison with MCF‐7 cells, MCF‐7/ADM cells with the A allele could enhance alternative splicing with the increase of SC‐35, upregulate the ratio of CCND1b/a at both mRNA and protein levels, and activate the CDK4/CyclinD1‐pRB‐E2F1 pathway. Furthermore, CCND1b expression and the downstream signalling pathway were analysed through Western blotting and cell cycle in MCF‐7/ADM cells with knockdown of CCND1b. Knockdown of CCND1b downregulated the ratio of CCND1b/a, demoted cell proliferation, decelerated cell cycle progression, inhibited the CDK4/CyclinD1‐pRB‐E2F1 pathway and thereby decreased the chemoresistance of MCF‐7/ADM cells. Finally, CCND1 G870A polymorphism, the alternative splicing of CCDN1 was detected through Sequenom Mass ARRAY platform, Sanger sequencing, semi‐quantitative RT‐PCR, Western blotting and immunohistochemistry in clinical BC specimens. The increase of the ratio of CCND1b/a caused by G870A polymorphism was involved in BC chemoresistance. Thus, these findings revealed that CCND1b/a ratio caused by the polymorphism is involved in BC chemoresistance via CDK4/CyclinD1‐pRB‐E2F1 pathway.


| INTRODUC TI ON
Breast cancer (BC) is currently the most common malignant tumour worldwide. 1 Up to 10% of BC patients have inherited (germline) DNA mutations, and ~90% of BC cases are caused by acquired (somatic) genetic and epigenetic alterations. 2 Immunosuppressive therapy and androgen steroids can promote the development of BC. 3 Human epidermal growth factor receptor 2 (HER2) is an established molecular prognostic marker, and it is often used to predict the response to endocrine therapy or targeted therapy in BC. 4 Pathological and imaging examinations are helpful to diagnose patients with HER2positive BC. 5 Studies have shown that BC can be subclassified based on genetic defects that reflect prognostic and predictive information to ensure that patients receive personalized treatment and improve treatment efficacy. 6,7 Adriamycin (ADM) and other anthracycline antibiotics were generally classified as first-line drugs for BC chemotherapy. 8,9 However, chemoresistance to drugs has significantly limited its effectiveness. The underlying molecular mechanism is complex and related to multiple processes; the aberrant regulation of cell cycle is one of the factors. 10,11 Currently, the mechanism by which abnormal cell cycle regulation leads to chemoresistance remains elusive. Therefore, studying the molecular mechanism underlying the relationship between the cell cycle and drug resistance is essential to overcoming drug resistance.
Cyclin D1 (CCND1) is a mediator of cell cycle control that regulates the transition from G1 to S phase and contributes to cell cycle progression. 12 The binding of CCND1 to cyclin-dependent kinase 4 (CDK4) results in the phosphorylation of the retinoblastoma protein (pRB). 13 This causes the release of E2F1 transcription factors, allowing the transcription of genes required for cell cycle G1 to S phase progression, enhancing the chemoresistance. [14][15][16] The CCND1 gene undergoes alternative splicing leading to the formation of two splicing variants: full-length CCND1 (CCND1a) and C-terminally truncated CCND1 species (referred to as CCND1b). 17 CCND1a is the common variant containing five exons, while CCND1b derives from the retention of intron 4 and contains a premature termination. 18 CCND1b is lacking the Thr-286 phosphorylation site necessary for nuclear export. This structural difference of CCND1b renders it to localize in the nucleus through the cell cycle, which may increase its oncogenic potency. 19 The G870A polymorphism at the splice donor site of the exon 4/intron 4 boundaries, which is thought to affect the production of CCND1a and CCND1b, was identified as a predictor for increased cancer risk. 18 The CCND1 G870A was reported to modulate CCND1b expression, which is also associated with poor clinical prognosis. 20,21 Furthermore, upregulation of CCND1b has been observed in several cancers including BC, thyroid cancer, glioma cancer, prostate cancer and non-small cell lung cancer. [21][22][23][24][25] To date, there is little evidence regarding the role of alternative splicing of CCND1 in BC chemoresistance. In addition, the influence of G870A polymorphism on the production of CCND1 variants in BC chemoresistance is not fully understood. In this study, we sequenced the G870A polymorphism and analysed the expressions of CCND1a and CCND1b in the ADM-resistant MCF-7 BC cells (MCF-7/ADM) and MCF-7 cells. Furthermore, CCND1b expression and the downstream signalling pathway were analysed through Western blotting and cell cycle in MCF-7/ADM cells with knockdown of CCND1b.
Finally, CCND1 G870A polymorphism, the alternative splicing of CCDN1 was detected through Sequenom Mass ARRAY platform, Sanger sequencing, semi-quantitative RT-PCR, Western blotting and immunohistochemistry in BC clinical specimens. Here, our study provided the first evidence that CCND1b/a ratio caused by polymorphism is involved in BC chemoresistance via the CDK4/CyclinD1-pRB-E2F1 pathway. CCND1b/a ratio could be used as a predictive biomarker and potential target for the therapy of BC patients.

| Study population
A total of 234 peripheral blood specimens were collected from patients who were admitted to the Lanzhou General Hospital of the Lanzhou Military Region between September 2013 and April 2020.
In addition, we randomly obtained 18 snap-frozen tissue specimens that were genotyped, including nine chemosensitive specimens with the GG genotype and nine chemoresistant specimens with the AA genotype. The association of two variants (CCND1a and CCND1b) with chemoresistance was analysed at the mRNA and protein levels by RNA and protein extraction from these 18 tissue specimens. We also collected 24 paraffin-embedded BC tissue specimens, including 12 chemosensitive specimens with the GG genotype and 12 chemoresistant specimens with the AA genotype.
All patients were confirmed by pathological examination and received chemotherapy for at least two courses. All chemotherapy regimens were based on anthracyclines. All specimens that had not received anthracycline treatment or received less than two courses of treatment were excluded. The effectiveness of chemotherapy can be evaluated clinically, such as tumour shrinking, which is an indicator of a good response. 27,28 The chemotherapy efficacy after anthracycline treatment was evaluated and scored according to the RECIST criteria. [29][30][31][32] According to RECIST, the therapy's response is classified as a complete response, partial response, stable disease or progressive disease, then classified into chemoresistant or chemosensitive groups. [33][34][35] The patients were chemosensitivity if there is the loss of all tumour masses or pathological lymph nodes (complete response), and if there is a tumour that becomes at least 30% smaller than the longest diameter of the tumour (partial response).
Patients were chemoresistance if there were tumour increases by at least 20% of the longest diameter (progressive disease) and where the tumour size reduction is not sufficient to generate a partial response, but tumour size does not increase and becomes a progressive disease (stable disease). 27,29,30,36 We classified the

| DNA extraction and genotyping
Genomic DNA was extracted from the cells and 2 mL of ethyl-  Table S1.

| RNA isolation and semi-quantitative RT-PCR and quantitative RT-PCR
Total RNA was extracted from BC tissues (N = 18) and cell lines using TRIzol reagent (Invitrogen), according to the manufacturer's instructions. Then, total RNA was reverse-transcribed using RT MasterMix (Cwbiotech). Gene-specific primers for CCND1a and CCND1b were also designed (Table S1). We used cDNA synthesized by reverse transcription as the template for PCR amplification. The products of PCR were visualized on a 1% agarose gel and analysed by ImageJ software (National Institutes of Health). The PCR-amplified products were verified via sequencing. Quantitative RT-PCR was accomplished using an UltraSYBR Mixture (Cwbiotech). The target gene expression was quantified using the 2 −ΔΔCT method. Each sample was run in triplicate.

| Western blotting and protein-protein docking
Proteins from snap-frozen BC tissues (N = 18) and cell lines were

| Statistical analysis
Each assay was performed and calculated in triplicate (N = 3).
Hardy-Weinberg equilibrium was evaluated in two groups of subjects to detect the genotype distribution of CCND1 G870A polymorphism using online software (http://ihg.gsf.de/cgi-bin/hw/hwa2.pl).
The clinical data and genotyping results were analysed using SPSS 22.0 software (IBM Corp.), independent specimen t-test, chi-square test and Fisher's exact test (specimen size <5) were used. p < 0.05 was considered statistically significant. Data were presented as the Mean ± SD and analysed with GraphPad Prism 5.0. The differences between the two groups were analysed with Student's t-test.

| Resistance of MCF-7/ADM cells to ADM
To study the mechanism of BC chemoresistance, the ADM-resistant BC cell line was generated by processing the parental MCF-7 cells with consecutive rounds of ADM treatment ( Figure 1A). MCF-7 and MCF-7/ADM cells were treated with ADM at different concentrations for 48 h and cell viability was assessed using the Cell Counting Kit-8 (CCK-8) assay. After cells had been treated with ADM, the growth of cells slowed down in a concentration-dependent manner with an IC 50 ( Figure 1B and Table S2). Compared with MCF-7 cells, MCF-7/ADM cells were accompanied by higher IC 50 values ( Figure 1C). The above results indicated that MCF-7/ADM cells were resistant to ADM. MCF-7/ADM cells can provide a more ideal cell model for subsequent study of chemoresistance mechanisms in BC.

| CCND1 G870A polymorphism in MCF-7/ADM cells
To investigate whether chemoresistance could cause alteration in the CCND1 G870A polymorphism, we extracted DNA from MCF-7 and MCF-7/ADM cells. Then, the extracted DNA was PCR amplified with specific primers of CCND1 G870A polymorphism. The PCR amplification products of MCF-7 and MCF-7/ADM cells were shown in Figure 1D. Subsequently, the PCR products were subjected to Sanger sequencing. Sequence alignment analysis showed that MCF-7 cells carry the wild G allele, while MCF-7/ADM cells carry the mutant A allele.

| CCND1 G870A polymorphism results in the formation of CCND1a and CCND1b
CCND1a is preferentially produced when the G allele creates an optimal splice donor site. In contrast, the A allele is expected to hinder the splicing event, allowing intron 4 retention, and production of a Cterminally truncated product of CCND1b. Compared with CCND1a, CCND1b is encoded by the first four exons, lacks the complete exon 5 sequence and retains part of the intron 4 sequence (VSEGD VPG SLA GAY RGR HLV PRK CRG WCQ GPQ G). Using Swiss-Model software, we predicted the protein structure of CCND1b. It is apparent from the protein model that CCND1b has an anomaly of the a-helix at the C-terminus ( Figure 1E). The CCND1b protein has a completely divergent C-terminal domain, lacking the PEST motif and residues (Thr-286) that control nuclear export and protein stability. Therefore, CCND1b is considered to be a more stable constitutive nuclear protein with enhanced capability to regulate CDK activity and cell cycle control.

| Upregulation of CCND1b/a ratio is associated with drug resistance in BC cells
We examined the subcellular localization of nuclear speckle marker SC-35 protein by fluorescent immunohistochemistry to observe whether abnormal alternative splicing was present in MCF-7 and MCF-7/ADM cells. We found that the nuclear speckles were more rounded and more numerous in MCF-7/ADM cells ( Figure 1F). This in-  (Figure 2A,B). Interestingly, the ratio of CCND1b/a was significantly upregulated in MCF-7/ADM cells at mRNA and protein levels. Figure 2C,D were the gray values of mRNA and protein.

| CDK4/CyclinD1-pRB-E2F1 signalling pathway activated in MCF-7/ADM cells
We analysed the correlation between CCND1 and some BC-related drug resistance genes by Assistant for Clinical Information online platform. We focused on the association of some key genes in the CDK4/CyclinD1-pRB-E2F1 signalling pathway with CCND1. The RB1, CDK4 and E2F1 genes were relatively strongly correlated with CCND1 in BC ( Figure 2E). Therefore, our subsequent experiments mainly explored the relationship between the CDK4/CyclinD1-pRB-E2F1 signalling pathway and chemoresistance in BC. The formation of the CCND1/CDK4 complex is a key point in the CDK4/CyclinD1-pRB-E2F1 signalling pathway. We used ZDOCK software to dock CCND1a and CCND1b with CDK4 and performed interaction mode analysis with Pymol2.3.0. It was shown that both CCND1a and CCND1b can interact with CDK4 through different binding sites ( Figure 2F). Furthermore, we analysed the protein expression of p-RB, CDK4 and E2F1 in cells. E2F1 and p-RB were highly expressed in MCF-7/ADM ( Figure 2G,H). However, the expression of CDK4 protein was not significantly different between the two groups. Combined with the experimental results, we speculated that the CCND1 G870A polymorphism affects the ratio of CCND1a and CCND1b. Abnormal degradation mechanism of CCND1b because CCND1b lacks the C-terminal PEST domain and Thr-286. However, CCND1a can be degraded normally. This ultimately leads to the upregulation of CCND1b/a ratio, activation of the CDK4/CyclinD1-pRB-E2F1 signalling pathway and resulting in chemoresistance ( Figure 2I).

| Downregulation of CCND1b could demote cell proliferation and decelerate the cell cycle progression in BC cells
To further confirm the role of CCND1b in chemoresistance, we

| Downregulation of CCND1b could inhibit cellcycle CDK4/CyclinD1-pRB-E2F1 pathway in BC cells
To demonstrate that CCND1b is essential for the activation of the CDK4/CyclinD1-pRB-E2F1 signalling pathway, we knocked down CCND1b in MCF-7/ADM cells to observe the protein change of cellcycle CDK4/CyclinD1-pRB-E2F1signaling pathway. Knockdown of CCND1b resulted in decreased expression of E2F1 and pRB proteins, but no change in CDK4 protein ( Figure 3F). Therefore, we assumed that the knockdown of CCND1b downregulated the ratio of CCND1b/a, inhibited the CDK4/CyclinD1-pRB-E2F1 signalling pathway, prevented tumour cells from passing through the G1-S transition, resulting in cycle arrest, thereby decreasing the chemoresistance of MCF-7/ADM cells.

| CCND1 G870A polymorphism is associated with chemoresistance in clinical BC specimens
The CCND1 G870A polymorphism in all subjects was detected by the Sequenom Mass ARRAY platform, and the association be-  (Table S3). Next, we analysed the correlation between the G870A polymorphism and BC chemoresistance in clinical BC specimens. The frequency distribution of different genotypes of CCND1 G870A polymorphism is shown in Table 2.
In the homozygote model, the AA genotype was significantly different from the GG genotype (OR = 2.647, 95% CI = 1.142-6.135, p = 0.021). The difference was statistically significant in the reces-  Figure 4C. The wild homozygous sequence, heterozygous sequence and mutant homozygous sequence of the CCND1 G870A polymorphism are listed in Figure 4D.
Subsequently, we examined the association between the G870A polymorphism and BC chemoresistance risk in the subgroups of the participants based on the relevant factors of age, menopausal status and tumour size. The stratification analysis by age, menopausal status and tumour size revealed that the distribution of the G870A genotype and allele between chemosensitive and chemoresistant specimens was slightly different ( Figure 4E and Table S4).

| Upregulation of CCND1b/a ratio affected by G870A polymorphism is associated with drug resistance in clinical BC specimens
We randomly obtained 18 snap-frozen tissue specimens that were genotyped to determine the expression of two variants (CCND1a and CCND1b), including nine chemosensitive specimens with the GG genotype and nine chemoresistant specimens with the AA genotype. The PCR products (CCND1a: 259 bp, CCND1b: 224 bp) were verified by sequencing. There was no alteration in the transcription level of CCND1a in the chemosensitive and chemoresistant specimens; however, there was a significant difference in CCND1b expression. The CCND1b/a ratio was higher in chemoresistant BC tissue specimens with the AA genotype ( Figure 5A). Then, we detected the protein expression of CCND1a and CCND1b in 18 snap-frozen tissue specimens using Western blotting. The result of proteins was consistent with the transcription level. CCND1b and the ratio of CCND1b/a were significantly increased in the chemoresistant specimens with the AA genotype ( Figure 5B). Thus, we considered that the upregulation of CCND1b/a ratio affected by G870A polymorphism was associated with chemoresistance in BC. In the  (N = 3). The data are expressed as the mean ± SD. ***p < 0.001; **p < 0.01; *p < 0.05.
present study, we focused on the ratio of CCND1b/a, which is more likely to reflect the role of chemoresistance than CCND1b. To further explore the correlation of CCND1a and CCND1b with chemoresistance in BC patients, we stained chemosensitive (GG genotype) and chemoresistant (AA genotype) tissue specimens for CCND1a and CCND1b by immunohistochemistry. We found a high rate of positivity for CCND1b in chemoresistance specimens. In addition, CCND1a was mainly distributed in the nucleus, whereas CCND1b was distributed in the cytoplasm, cell membrane and nucleus. Figure 5C shows representative immunohistochemical staining for Abbreviations: C-erbB-2, human epidermal growth factor receptor 2; ER, oestrogen receptor; PR, progesterone receptor. . The data are expressed as the mean ± SD. ***p < 0.001; *p < 0.05.

TA B L E 2
The association between CCND1 gene G870A polymorphism and BC chemoresistance. CCND1a and CCND1b. Due to our limited specimen size, no correlation between CCND1a and CCND1b expression and genotype was tentatively found.

| DISCUSS ION
CCND1, a nuclear protein that regulates cell cycle and controls the G1-S transition, is closely related to the occurrence and development of various cancers. [40][41][42] Up to 50% of BC is overexpressed with CCND1, and amplification of the CCND1 gene is associated with prognosis. [42][43][44] Studies have shown that the CCND1 expression was related to poor overall survival and disease-free survival in BC. 45,46 In addition, the overexpression of CCND1 was associated with the response of chemotherapeutic agents such as adriamycin, 5-fluorouracil and cisplatin. 47 In recent years, an increasing number of studies have found that CCND1 gene polymorphism was also associated with the occurrence and development of cancers, especially to chemotherapy response of cancers. [48][49][50] However, the in-depth mechanism between CCND1 gene polymorphism and chemotherapy response remains to be elucidated. CCND1 G870A polymorphism results in the formation of CCND1a and CCND1b, 51 but the role of CCND1a and CCND1b variants in cancer chemoresistance remains unknown. We surmised that the CCND1 variants generated by the G870A polymorphism might be related to drug resistance. To date, there is little evidence regarding the role of CCND1a and CCND1b variants in BC chemoresistance. Therefore, this study aimed to explore the molecular mechanism of CCND1a and CCND1b variants in BC chemoresistance.
The wild G allele was carried in MCF-7 cells, whereas the mutant A allele was carried in MCF-7/ADM cells ( Figure 1D). Then, we collected peripheral blood specimens to explore the relationship be- ( Figure 1E). 17 Comstock et al. 22 cloned the intron 4 sequence containing the G or A allele and inserted it into the full length by minigene, and ultimately proved that the A allele preferentially produces CCND1b. In our study, we found that CCND1b expression can be increased by the A allele. However, the result is inconsistent and Howe et al. 51 found that A allele produced more CCND1a while G allele had more CCND1b in malignant lymphocytes. This inconsistency may be caused by sample size and disease types. Therefore, it is necessary to analyse in a larger cohort of patients and diseases to eliminate the inconsistency of published literature.
CCND1a and CCND1b play an important role in cancer. CCND1a was discovered in 1991 and has high expression in various cancers. 55 Since CCNDb was discovered in 1995, it has been widely studied in cancers. 17 CCND1b is constitutively in the nucleus. 56 26 Moreover, abnormally high expression of CCND1b was particularly associated with poor outcomes in BC. 20,21 Studies showed that CCND1b antagonizes the action of CCND1a and inhibits the growth of breast cancer cells. 61 However, the potential molecular mechanism of CCND1b and CCND1a in BC chemoresistance remains unclear. In our study, reduced mRNA and protein levels of CCND1a were observed in MCF-7/ADM, but the mRNA level of CCND1b was increased in MCF-7/ADM. However, no difference was observed in CCND1b protein expression between the two groups ( Figure 2A,B). Interestingly, the ratio of CCND1b/a was significantly upregulated in MCF-7/ADM cells at mRNA and protein levels. Our research indicated that there seems to be an equal amount of stable CCND1b in MCF-7 and MCF-7/ADM cells, which is due to the abnormal degradation caused by the deletion of PEST domain. 62 The difference of CCND1a level may be related to the difference of G1 and G2 population size between MCF-7 and MCF-7A/DM cells.
As we all know, CCND1 is a mediator of cell cycle control that regulates the transition from G1 to S phase and contributes to cell cycle progression. 12 Studies had reported that the level of CCND1a was low in the S phase and high in G1 and G2 phases. 63,64 Therefore, the difference of G1 and G2 population size between MCF-7 and MCF-7/ ADM may be one of the reasons for the difference of CCND1a level, which may have some impact on our conclusion. In this study, we only observed the difference of CCND1a levels through simple experiments. In the future, we will conduct in-depth molecular mechanism research to determine the specific reasons for this difference.
In cell research, we found that the ratio of CCND1b/a at the mRNA and protein levels is upregulated in MCF-7/ADM cells (Figure 2A-D).
Furthermore, we further performed validation in clinical specimens ( Figure 5A,B). In clinical specimens, this result was consistent with the cell research results. Thus, this indicates that the upregulation expression of the ratio of CCND1b/a caused by G870A polymorphism was involved in BC chemoresistance.
The dysregulated cell cycle progression is associated with chemoresistance. 11 The CDK4/CyclinD1-pRB-E2F1 signalling pathway axis is the key transcriptional mechanism that drives the process of cell cycle. 65 RB by binding to CDK4, triggering the E2F1 required to enter the S phase to promote the cell cycle process, 37 ultimately leading to drug resistance ( Figure 2I). In addition, to confirm that CCND1b is essential for the activation of the CDK4/CyclinD1-pRB-E2F1 signalling pathway, we knocked down CCND1b in MCF-7/ADM cells to observe the protein change of cell-cycle CDK4/CyclinD1-pRB-E2F1signalling pathway. Our results showed that the knockdown of CCND1b downregulated the ratio of CCND1b/a, inhibited the CDK4/CyclinD1-pRB-E2F1 signalling pathway and prevented tumour cells from passing through the G1-S transition resulted in cycle arrest, thereby decreasing the chemoresistance ( Figure 3C-F).

| CON CLUS ION
In conclusion, our study highlights CCND1b/a ratio was caused by G870A polymorphism involved in BC chemoresistance through cell research and clinical specimens' validation ( Figure 6). Our data established CCND1b/a ratio as a potential predictive biomarker for chemoresistance and demonstrated that the CCND1b/a ratio may F I G U R E 6 Schematic representation of the influence of CCND1 splicing variants expression on BC chemoresistance. Upregulation of CCND1b/a ratio affected by G870A polymorphism is associated with chemoresistance.

ACK N O WLE D G E M ENTS
None.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors confirm that there are no conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.