- Research article
- Open access
- Published:
SNRPD1 conveys prognostic value on breast cancer survival and is required for anthracycline sensitivity
BMC Cancer volume 23, Article number: 376 (2023)
Abstract
Background
Cancers harboring spliceosome mutations are highly sensitive to additional perturbations on the spliceosome that leads to the development of onco-therapeutics targeting the spliceosome and opens novel opportunities for managing aggressive tumors lacking effective treatment options such as triple negative breast cancers. Being the core spliceosome associated proteins, SNRPD1 and SNRPE have been both proposed as therapeutic targets for breast cancer management. Yet, their differences regarding their prognostic and therapeutic use as well as roles during carcinogenesis are largely unreported.
Methods
We conducted in silico analysis at gene expression and genetic levels to differentiate the clinical relevance of SNRPD1 and SNRPE, and explored their differential functionalities and molecular mechanistic associations with cancer in vitro.
Results
We showed that high SNRPD1 gene expression was prognostic of poor breast cancer survival whereas SNRPE was not. The SNRPD1 expression quantitative trait loci, rs6733100, was found independently prognostic of breast cancer survival using TCGA data. Silencing either SNRPD1 or SNRPE independently suppressed the growth of breast cancer cells, but decreased migration was only observed in SNRPD1-silenced cells. Knocking down SNRPD1 but not SNRPE triggers doxorubicin resistance in triple negative breast cancer cells. Gene enrichment and network analyses revealed the dynamic regulatory role of SNRPD1 on cell cycle and genome stability, and the preventive role of SNRPE against cancer stemness that may neutralize its promotive role on cancer cell proliferation.
Conclusion
Our results differentiated the functionalities of SNRPD1 and SNRPE at both prognostic and therapeutic levels, and preliminarily explained the driving mechanism that requires additional explorations and validations.
Background
Breast cancers are highly heterogeneous, with the triple negative breast cancer (TNBC) subtype being the most difficult to treat. This particular cohort of malignant cells have higher invasiveness, recurrence rate and poorer patient survival as compared with the other subtypes, and are featured by low expression of many surface markers that render them insensitive to existing hormonal or targeted therapies such as Tamoxifen and Herceptin [1, 2]. This leaves chemotherapies the first-line approach against TNBCs that are known with strong adverse effects as a result of lack of tumor specificity. Intensive effort has been devoted to establish targeted therapies against TNBCs. Dominant strategies include tumor angiogenesis inhibitors such as Ruxolitinib [3] and poly ADP-ribose polymerase (PARP) inhibitors such as Olaparib [4]. Tumor angiogenesis inhibitors are typically accompanied with multifaceted toxicities such as liver damage, anaemia, neutropenia, and thrombocytopenia [5,6,7]. The efficacy of PAPR inhibitors is limited to TNBCs harboring BRCA1 mutations [8] and is adversely affected by many factors such as interactions between PARP1 and Fra-1 (an AP-1 member over-expressed in TNBCs) [4]. Thus, identifying novel targets towards improved TNBC management represents a global concern for researchers working on breast cancers.
Accurate splicing is essential to ensure normal cell functionality, alteration of which affects many cellular processes and is associated with a plethora of diseases including human cancers [9, 10]. Cancer transcriptomes are featured by abnormal RNA splicing as a result of, e.g., recurrent mutations to RNA splicing factors. Thus, cancers harboring spliceosome mutations are highly sensitive to additional perturbations on the spliceosome due to, at least partially, elevated pre-mRNA synthesis and consequent spliceosome burden [11]. This has led to the development of onco-therapeutics targeting the spliceosome [12,13,14] that opens novel opportunities for managing aggressive tumors lacking effective treatment options such as TNBCs [11], and a clinical trial on small-molecule modulators of the spliceosome (NCT02841540).
The spliceosome is a large dynamic macromolecular ribonucleoprotein (RNP) complex catalyzing the splicing of precursor mRNA into mRNA in eukaryotic cells, deregulation of which is associated with the generation of novel mRNA isoforms and protein variants. There are two types of spliceosomes in eukaryotic cells, i.e., U2-dependent (dominant) and U12-dependent spliceosomes. The U2 spliceosome processes approximately 95.5% of all U2-type introns, and the U12 spliceosome functions in the splicing of rare U12-type introns that occur at a frequency of around 0.35% in all human introns [11]. Both types of spliceosomes contain 5 uridine-rich (U-rich) small nuclear RNP (snRNP) units that differs in the types of snRNPs included [15]. While both spliceosomes contain the U5 snRNP, the U2-dependent spliceosome contains U1, U2, U4, U6 snRNPs and the U12-dependent spliceosome contains U11, U12, U4atac and U6atac snRNPs. Each snRNP is comprised of an RNA component, Sm proteins (i.e., SNRPB/B’, SNRPD1, SNRPD2, SNRPD3, SNRPE, SNRPF, SNRPG) that form a 7-member ring core structure encompassing RNA, and varieties of other associated proteins [11].
SNRPD1 and SNRPE, among other Sm proteins, were shown overexpressed in a subset of highly aggressive breast cancers, and were proposed as novel onco-therapeutic targets since the depletion of either one halted breast cancer growth without affecting the quasi-normal MCF10A cells [16]. However, we observed different effects of SNRPD1 and SNRPE on breast cancer cells that motivated us to investigate their differential roles and mechanisms associated with carcinogenesis.
The prognostic value of SNRPD1 on breast cancer outcome was identified with statistical significance using multiple public data sets which, however, was not found for SNRPE. The rare allele of one quantitative trait loci (eQTL) of SNRPD1, rs6733100, was revealed to confer a protective effect on patient outcome using data retrieved from The Cancer Genome Atlas (TCGA). Though suppressed expression of either SNRPD1 or SNRPE led to reduced tumor cell proliferation, SNRPD1 promoted tumor cell migration and sensitized tumor cells to chemotherapy, SNRPE suppressed cancer stemness, and both proteins differ in their dynamic regulations on cell cycle progression. Our results suggest the prognostic value of SNRPD1 but not SNRPE on breast cancer survival, and warrant the use of SNPRD1 in the onco-therapeutic design given its unveiled roles in sensitizing TNBC cells to anthracycline-type of chemotherapies.
Methods
Computational analysis
The following datasets containing information on SNRPD1 and SNRPE and being available at the time when this study was conducted were used in this study.
GSE24450 gene expression data set
The GSE24450 data set, retrieved from Gene Expression Omnibus (GEO) [17], is comprised of 183 primary breast tumor samples containing 39 breast cancer death or distant metastasis events (supplementary Table 1). The maximum follow-up time on disease specific death or distant metastasis was 60 months.
Gene expression profiling was conducted at SCIBLU Genomics Centre, Lund University, Sweden using Illumina HumanHT-12_V3 Expression Bead Chips that contain 24660 Entrez Gene entities following the manufacturer recommendations (http://www.illumina.com).
After quality control and quantile normalization of the microarray raw data, the gene expression matrix was obtained by averaging the probes mapped to the same Entrez Gene IDs. All analyses were processed using ‘BioConductor’ in R (https://www.r-project.org).
GSE1456 gene expression data set
The GSE1456 (GPL96) data set was retrieved from GEO, which is comprised of 159 samples including 40 breast cancer death or relapsed events [18] (supplementary Table 1). The maximum follow-up time on disease free survival was 102 months.
RNA was extracted following the RNeasy mini protocol (Qiagen, Hilden, Germany) [18]. All tumor specimens were assessed using the Affymetrix Human Genome U133A array at Bristol-Myers Squibb (Princeton, New Jersey) [18].
Data was normalized using the global mean method, natural-log-transformed and scaled by adjusting the average intensity of the signal to a target value of log 500 [18].
GSE4922 gene expression data set
The GSE4922 (GPL96) data set was retrieved from GEO, and comprised of 249 samples including 89 breast cancer specific death or relapsed events [19] (supplementary Table 1). The maximum follow-up time on disease free survival was 153 months and truncated at 10 years in the analysis.
RNA was extracted using the RNeasy mini protocol (Qiagen, Hilden, Germany). Tumor samples were profiled using the Affymetrix Human Genome U133A chips at the Genome Institute of Singapore [19].
Data was normalized using the global mean method, natural-log-transformed and scaled by adjusting the mean signal to a target value of log 500 [19].
TCGA gene expression data set
The primary solid breast tumor mRNA expression data of level 3 was retrieved from TCGA (http://cancergenome.nih.gov) on 21st November 2015 (supplementary Table 1). The data included 514 samples, among which 512 had recorded information on overall survival including 53 death events. The mRNA data was lowess-normalized, and the ratio between the two channels was log2-transformed. The maximum follow-up time on overall survival was 226.5 months, which was truncated at 10 years in the analysis.
Kaplan Meier plotter data set
The prognostic values of the evaluated genes were confirmed using the online tool, Kaplan Meier Plotter [20], which included 4142 breast cancer samples. Gene expression data and clinical information were downloaded from GEO, EGA and TCGA, and integrated simultaneously. Only Affymetrix HG-U133A, HG-U133 Plus 2.0 and HG-U133A 2.0 were included.
TCGA genotype data set
The primary solid tumor genotyping data was retrieved from TCGA at http://tcga.cancer.gov/dataportal on 15th January, 2015 (supplementary Table 1). The data set consisted of 906600 SNPs genotyped in 504 samples with an average age at diagnosis of 59.6 years. The follow up data for these cases included 53 overall death events. The TCGA genotype data on 906600 SNPs was produced using the Affymetrix Genome Wide Human SNP array 6.0. The genotypes and confidence scores for each sample and each SNP were generated from the raw data using the birdseed algorithm by fitting two-dimensional Gaussians to the SNP data and coding genotypes with confidence score > 0.1 as the missing data.
TCGA copy number variation data set
Breast cancer CNV data, which was comprised of putative copy-number calls determined using GISTIC 2.0 (-2, -1, 0, 1, 2 represent homozygous deletion, hemizygous deletion, neutral, gain and over-amplification, respectively), was downloaded from TCGA via cBio [21] on 15th January, 2016 (supplementary Table 1). The data set includes 889 samples from the TCGA provisional study, among which 502 overlap with the genotype and gene expression data.
Mass spectrometry proteomic data set
Mass spectrometry data was retrieved from the supplementary file of Johansson H.J. et al. [22] that includes 45 samples and 9995 proteins. The sample cohort covers 5 subtypes, i.e., luminal A, luminal B, HER2 positive, triple negative/basal-like, and normal-like, each containing 9 samples. This proteomic data was produced using HiRIEF nanoLC-MS/MS.
Gene expression association analysis
Cox regression analysis was conducted on patient survival for disease specific death and distant metastasis or relapse at the mRNA level for SNRPD1 and SNRPE using GSE24450, GSE1456 and GSE4922 data sets. The data was binarized by the median into high and low expression. Meta-analysis was conducted using the fixed-effects model for homogenous data (SNRPD1) and the random-effects model whenever heterogeneity exists (SNRPE) by the R package ‘meta’.
Kaplan Meier Plotter (which includes information from TCGA) was employed to check such associations using the log rank test, where the 5-year and 10-year relapse free survival were conducted with median being used as the sample splitting point. The prognostic values of SNRPD1 expression on the clinical outcome among breast cancer patients with and without chemotherapy treatment were examined.
Histopathological association analysis
Histopathological association of SNRPD1 gene expression was analysed using GSE24450 data. Samples were binarized into high and low expression by the median, and the associations with histopathological markers including ER, PR, TP53, HER2, tumor (T), node (N), Ki67 expression, grade, and different subtype classifications, were analysed separately by Chi-square test in R. The kruskal test was conducted to examine potential differences between the expression of genes in tumors categorized by different histopathological markers.
eQTL analysis
The primary solid tumor genotype and level 3 TCGA gene expression data were used for eQTL analysis. TCGA CNV data was retrieved from cBio cancer genomics portal (http://www.cbioportal.org/public-portal/) [21] and was used as the covariate. In total, 502 samples that have available information on genotyping, gene expression and CNV were used in the analysis. The gene expression and genotype data were fitted into a linear model in the eQTL analysis, with and without CNV being adjusted as the covariate. SNPs with p < 0.005 were selected for further analysis on its independent association with patient clinical outcome.
Tagging SNPs for these eQTL loci were retrieved using SNAP (Proxy Search) [23], where Caucasion samples (CEU) from 1000 Genomes Pilot 1 were used, the distance limit was set to 500 and \({\mathrm{r}}^{2}\) was set above 0.8. These SNPs together with their tagging SNPs were mapped to the TCGA genotype data for the survival analysis.
Genetic association analysis
Genetic association study with 10-year overall death as the endpoint was conducted on the eQTL SNPs using TCGA data. For each SNP, the data was fit into an additive, dominant and recessive model using a Cox regression model using the ‘R’ software. SNPs showing statistically significant associations with patient survival and having consistent clinical effects with the expression of corresponding genes were reported.
The gene encompassing or closest to the eQTL SNP (within 500 kb) was retrieved using GRAIL [24], where CEU samples (HapMap release 21) or Human Genome Assembly 17 were used with the ‘Functional Data source’ being set to PubMed April 2011.
Correlation analysis
Genes and proteins highly associated with SNRPD1 or SNRPE regarding expression were analysed using the gene expression data retrieved from TCGA and the mass spectrometry data retrieved from [22]. Genes having the Pearson correlation scores above 0.5 or below -0.5 and p<0.001 were considered significantly correlated with SNRPD1 or SNRPE. A gene or protein was considered significantly associated with SNRPD1 or SNRPE if the absolute value of the correlation score was > 0.5 and the p \(\le\) 0.001. A gene or protein was considered differentially correlated with SNRPD1 or SNRPE if the absolute value of the difference between its correlation scores with SNRPD1 or SNRPE was \(\ge\) 0.3.
Experimental studies
Cell culture
The luminal cell line MCF7 and the TNBC cell lines MDAMB231, HCC1937 were included in this study. All cells were purchased from ATCC, mycoplasma tested. MCF7 and MDAMB231 were cultivated using DMEM containing 10% fetal bovine serum (FBS) (Gibco), and HCC1937 was cultured using RPMI-1640 supplemented with 10% FBS.
Cells were stored in liquid nitrogen in frozen solution that contains 90% FBS and 10% DMSO. Immediately prior to transfection, cells were thawed and washed using the corresponding culture medium and the cell number was counted using a hemocytometer (Thermo).
Gene silencing
Two siRNAs (GenePharma) were designed for SNRPD1 (s13229 and s13230) and SNRPE (s13237 and s13239), respectively, and pooled together before usage. GenePharma Silencer Select Negative Control was used as the negative control. The siRNA products were ordered from Sigma-Aldrich (supplementary Table 2).
CRISPR sgRNA construction
Three enhancer sgRNAs (supplementary Table 3) were designed for SNRPD1 or SNRPE using http://crispr.mit.edu/, and synthesized from GENEWIZ. The sgRNAs were constructed into the sgRNA CRISPR lentivector (#CS30453, ABcam), separately, using BbsI (NEB). One dCas9 synergistic activation mediator lentivector (K015) was purchased from ABcam.
Quantitative real-time PCR
After siRNA transfection, cells were collected and total RNA was extracted using TRIzol reagent (TianGen) 2 days post-transfection. The cDNA was synthesized using PrimeScriptRT reverse transcriptase (Takara). The quantitative real-time PCR (q-PCR) system contains 5μL 2×SYBR Premix Ex Taq, 0.4μL, 10μM forward and reverse primers (supplementary Table 4), 0.2μL ROX Reference Dye, 2μL cDNA and 2μL ddH2O. The q-PCR experiments were conducted using ABI Stepone plus Real-Time PCR System (ABI). Detailed procedure for Q-PCR includes 5min initial denaturation at 95ºC, 5s denaturation at 95ºC for 45 cycles, 30s annealing at 57ºC, and 15s extension at 72ºC. The absorbance values were recorded at the extension stage, and the relative expression levels were computed using the 2-△△Ct approach.
Student’s t-test was computed using R, where the p values were assessed as the two-tailed probability at 95% confidence from the standard Gaussian distribution.
Western blot
Cultured cells were washed twice using prechilled PBS and lysed in RIPA lysis buffer supplemented with protease inhibitors for 5min, centrifuged at 12,000 g for 10min followed by supernatant collection. The protein concentration was examined using the BCA Protein Assay Kit (Tiangen). Proteins (50μg) per lane were resolved by SDS-PAGE and transferred to PVDF membrane. After blocking with 5% non-fat dried milk powder in TBS plus Tween-20 buffer, the membrane was incubated using the appropriate primary antibodies at 4ºC overnight followed by secondary antibodies for 2h at the room temperature. Antibody binding was visualized by developing the blot using enhanced chemiluminescence reagent. The bands were visualized using OmegaLumG (UVP) followed by analysis using the Image J software. The primary and secondary antibodies used were listed in supplementary Table 5.
Doxorubicin resistance assay
Doxorubicin (#D1515, Sigma-Aldrich) at 8 concentrations (i.e., 0.83nM, 10nM, 25nM, 100nM, 250nM, 1000nM, 2500nM, 10000nM) were used in this study. ‘SiRNA+Doxorubicin’ and ‘negative control+Doxorubicin’ were designed for each gene in each 96-well plate, where ‘Doxorubicin’ has 8 concentrations and each sample has 6 replicates. Two siRNAs were designed for each gene and pooled together before usage.
Doxorubicin was added to cells 24h after siRNA transfection, and 10 μL/well CKK-8 was added to cells 96h after siRNA transfection. Luminescence was detected using EZ Read 800 microplate Reader after cells were incubated at 37ºC for 2h.
The dose curves in response to doxorubicin treatment were drawn for each cell line and half-maximum inhibitory concentration (IC50) values were obtained by fitting data to a 4-parameter log-logistic model (LL.4) using the ‘drc’ package in R. Statistical significance on IC50 shift was assessed by student’s t-test using R.
Proliferation assay
8000 cells/well were added to 100μL culture medium and seeded in a 96-well plate (Nalgene # 167008). Cells were incubated overnight until they achieved 30%-50% confluence prior to transfection. 50μL Optimem medium (Gibco) containing SNRPD1, SNRPE or control siRNAs was added to 750nL siRNA-mate (GenePharma) per well and mixed for 15-20min prior to transfection. The mixed content was transferred to 96-well plate, with the final siRNA concentration being 20nM. After transfection, cells were incubated at 37ºC in the presence of 5%CO2 (HERA Cell 150i, Thermo Scientific). For cell proliferation measurement, 10μL per well of CKK-8 (Dojindo) was added 48h post-transfection, and luminescence was detected using EZ Read 800 microplate Reader (Biochrom) after cell incubation at 37ºC for 2h.
Student’s t-test was conducted using R to assess the statistical significance, and the p values were computed as the two-tailed probability at 95% confidence from a standard normal distribution.
Wound healing assay
After siRNA (knock down) or sgRNA (enhance) transfection following the manufacture’s protocol, cells were incubated for 48h until the form of confluent monolayers. Medium was refreshed by serum-free medium before wounding. Wounds were made using a pipette tip, and photographs were taken immediately (time 0) and 12h, 24h and 36h after the wounds were made. Altered distance as measured between the two edges of wounded area were computed at each time point. Results were presented as the migration area.
The student’s t-test was used to assess the statistical significance in R, where the p values were computed as the two-tailed probability at 95% confidence from the standard Gaussian distribution.
Flow cytometry
Cell flow cytometry was performed 48h after siRNA transfection. Cells were collected using EDTA-free trypsin, washed twice using 0.5 ml PBS, suspended in 70% pre-cold ethanol, and stored in 4ºC overnight. Ethanol was removed and cells were re-suspended in PBS the next day. Cells were supplemented with 0.05 mg/ml Propidium Iodide (PI) and kept in darkness on ice for 30 min prior to cell cycle detection using BD Accuri™ C6 flow cytometer. The analysis was performed using flowJo v10 (Tree Star, USA).
In the ALDH analysis, cells were incubated with the antibodies in recommended concentrations at 37 °C for 0.5 h, followed by PBS washing for two times. Cells were sorted and analyzed by BD Accuri™ C6 flow cytometer. The ALDEFLUOR Kit (StemCell Technologies) was used to test ALDH activity following the manufacturer’s protocol. The diethylaminobenzaldehyde (DEAB), a specific ALDH inhibitor, was used as negative control. Flow cytometry data processing were performed using flowJo v10 (Tree Star, USA).
Results
To investigate the differential roles of SNRPD1 and SNRPE on breast cancer prognosis and therapeutics as well as the potential driving mechanisms, we conducted a series of in silico analyses and in vitro experiments that are summarized in supplementary Figure 1.
SNRPD1 over-expression is prognostic of poor breast cancer survival
High SNRPD1 gene expression was significantly associated with increased hazard on breast cancer specific and distant metastasis (GSE24450) or relapse free survival (GSE1456, GSE4922) in three independent patient series (p<0.0001, HR=1.89 from meta-analysis, Fig. 1). However, no significant association was found between SNRPE expression and patient survival (p>0.05, HR=0.93, Table 1).
Analyses on 10-year breast cancer relapse free survival (RFS) using median for data binarization on 3951 breast tumors from Kaplan-Meier plotter (encompassing data from GEO, EGA and TCGA) [20] showed significant association between SNRPD1 over-expression and increased hazard (probe 202690_s_at); whereas no consistent prognostic value was observed for SNRPE (probe 231112_at). Importantly, while SNRPD1 over-expression is prognostic of poor RFS for ER-HER2- (corresponding to TNBC) and ER+HER2- (mostly referring to luminal A) tumors as well as suggests favorable RFS outcome for ER-HER2+ tumors (i.e., HER2+ breast cancers), high level of SNRPE conveys no prognostic value for none of these breast cancer subtypes (Fig. 2).
No significant association was found between SNRPD1 expression and the histopathological parameters using GSE24450 data (supplementary Table 6).
Expression quantitative trait loci of SNRPD1
Given the positive association between SNRPD1 over-expression and poor breast cancer survival, we investigated the expression quantitative trait loci (eQTL) of SNRPD1 using data retrieved from TCGA. 19638 SNPs (2.16% of the total SNPs tested) were found significantly associated with SNRPD1 expression (p ≤ 0.005). These include 5394 and 5185 SNPs from the analysis without and with copy number variation (CNV) adjustment, respectively, and 9059 SNPs significant from both (supplementary Table 7).
The association of these eQTLs with patient clinical outcome was examined using TCGA data, with one SNP (rs6733100) found having an independent prognostic value on patient 10-year overall survival. This SNP was significantly associated with SNRPD1 expression with (p<0.01) or without (p<0.01) CNV adjustment. The minor allele of rs6733100, A, with a frequency of 0.1753, showed significant protective effect on patient 10-year overall survival in the dominant model (Fig. 1F, p <0.05, HR = 0.56). It is a trans-eQTL (genomic location is 2:36293207) and about 62 kb upstream of the closest transcriptional start site of CRIM1.
SNRPD1 and SNRPE promote the proliferation of breast cancer cells
Given the differential associations of SNRPD1 and SNRPE on breast cancer survival, we explored their functionalities on cancer cell growth using in vitro assays. Both SNRPD1 and SNRPE have been effectively knocked down as shown at the mRNA level (reduced to 1/4 of the control in both MDAMB231 and MCF7 cells for both genes, Fig 3A and B) and at the protein level (reduced to 68% in MDAMB231 and to 65% in MCF7 for SNRPD1; to 30% in MDAMB231 and to 47% in MCF7 for SNRPE, Fig. 3C and D). Significantly decreased cell viability was observed for tumor cells when either SNRPD1 (reduced approximately 20% for MDAMB231 and 40% for MCF7) or SNRPE (reduced around 30% for MDAMB231and MCF7) was knocked down regardless of the tumor subtype (Fig. 4A and B).
SNRPD1 promotes the migrative ability of breast cancer cells
We next examined the effects of SNRPD1 and SNRPE on the migrative abilities of breast cancer cells. Significant recession or enhancement on cell migration was observed at 12 h (after scratches were made) in tumor cells with low or high SNRPD1 expression (reduced 40% in MCF7 and almost 30% in MDAMB231 for siRNA-mediated SNRPD1 silencing, enhanced 30% in MCF7 and 20% in MDAMB231 for sgRNA-mediated SNRPD1 enhancing, Fig. 4C and E, supplementary Figure 2) but not in cells with altered SNRPE expression (Fig. 4D and F). SNRPD1 and SNRPE were both effectively over-expressed in MDAMB231 and MCF7 cells as examined at both gene and protein expression levels. That is, p<0.0001 (MDAMB231) and p<0.005 (MCF7) for SNRPD1, and p<0.001 (MDAMB231) and p<0.0001 (MCF7) for SNRPE at the gene expression level (supplementary Figures 3A, 3B); and those at the protein expression level were increased to 3.5 folds (MDAMB231) and 2.2 folds (MCF7) for SNRPD1, and 3.8 folds (MDAMB231) and 2.9 folds (MCF7) for SNRPE (supplementary Figures 3C, 3D).
SNRPD1 sensitizes breast cancer cells to doxorubicin
Enlightened by the oncogenic roles unveiled for SNRPD1 and SNRPE, we investigated their impacts on the efficacy of doxorubicin, an anthracycline type of chemotherapy canonically used for TNBC treatment. Patients with high SNRPD1 expression have poorer RFS as compared with those harboring low SNRPD1 expression (1010 patients, HR = 1.56, p<0.0001 for 10 years), and such a survival disadvantage vanished after receiving chemotherapies (602 patients, HR=1.21, p>0.05 for 10 years, Fig. 5A), suggestive of the role of SNRPD1 in sensitizing breast cancer cells to chemotherapies.
We next examined the sensitivity of TNBCs in response to doxorubicin, a common form of adjuvant chemotherapy for breast cancer treatment especially among TNBCs, after silencing SNRPD1 or SNRPE using MDAMB231 and HCC1937 as the cell models. The drug response curves were significantly right-ward shifted when SNRPD1 was silenced (p<0.00001 in MDAMB231, p<E-9 in HCC1937) that was not observed when SNRPE was knocked down (Table 2, Fig. 5B), implicative of TNBC resistance developed in response to doxorubicin after silencing SNRPD1.
SNRPD1 and SNRPE differ in cell cycle genes they regulate
Provided with the different roles of SNRPD1 and SNRPE in cell migrative abilities and drug sensitivities that may drive their differential prognostic values on breast cancer survival, we examined their correlated genes to investigate differences on their molecular mechanisms.
Genes and proteins highly associated with SNRPD1 or SNRPE expression were analysed using TCGA gene expression data (supplementary Table 8) and mass spectrometry (MS) data retrieved from Johansson H.J. et al. [22] (supplementary Table 9). Correlations of these genes or proteins follow the Gaussian distribution, where the kurtosis and skewness for SNRPD1 were 0.22 and 0.49, and for SNRPE were 0.4 and 0.36, respectively, using the TCGA gene expression data (Fig. 6A); and those for SNRPD1 were 1.04 and 0.71, and for SNRPE were 1.16 and 0.93, respectively, using the MS protein data (Fig. 6B).
Genes highly correlated with SNRPD1 had, in general, higher absolute correlation scores than those with SNRPE (Fig. 6C and D). There were 29 genes and 1 protein differentially associated with SNRPD1 and SNRPE (supplementary Table 10), which were subjected to Gene Ontology (GO) and KEGG enrichment analyses. The results showed that these genes/proteins were significantly enriched in ‘chromatin segregation’, ‘microtubule cytoskeleton organization’ and ‘cell cycle’ (Fig. 6E and F ).
Among these 30 nodes, 63 pairs of protein-protein interactions (PPIs) were constructed using STRING (functional protein association networks) [25], 23 pairs out of which had the combined scores above 0.95 (Fig. 6G, supplementary Table 11). Among these 23 PPIs with high confidence, 10 pairs were experimentally validated (Table 3), with interactions between CENPN and CENPA being assigned with the highest experimental score (0.837), text mining score (0.851) and combined score (0.999). Interestingly, CENPN and CENPA were not highly correlated regarding their expression profiles (co-expression score being 0.68) and shared no homology, suggesting the involvement of additional players in their interactions. Thus, we selected for the follow-up q-PCR experimental validations.
We next explored whether SNRPD1 or SNRPE affected CENPN or CENPA. The q-PCR results showed that CENPA and CENPN were differentially modulated on SNRPD1 and SNRPE silencing (Fig. 7A and B). While knocking down SNRPD1 significantly enhanced CENPN expression (p<0.05), silencing SNRPE increased CENPA expression with statistical significance (p<0.05). Silencing SNRPD1 altered the cell cycle profile by reducing the S phase from 23.1% to 15.3% (Fig. 7B), and knocking down SNRPE considerably enhanced ALDH1 percentage from 4.39% to 15.1%, where MDAMB231 was used as the modelling cell line (Fig. 7C).
Discussion
We differentiate SNRPD1 from SNRPE regarding its clinical implications and the potential underlying mechanism in this study, whereas both genes were proposed as therapeutic targets sharing similar carcinogenetic roles [16]. Specifically, SNRPD1 over-expression is prognostic of poor breast cancer survival whereas SNRPE is not. Interestingly, high level of SNRPD1 conveys poor prognostic value on ER-HER2- (mostly TNBC) and ER+HER2- (corresponding to luminal A) tumors, but implicates favorable clinical outcome on ER-HER2+ (HER2+ tumors) breast cancers. Though we could not exclude the possibility of having an inaccurate observation due to the small sample cohort (i.e., 219 ER-HER2+ cases in Fig. 2), the higher RFS, in general, for HER2+ patients (both ER-HER2+, ER+HER2+) together with the significantly poor RFS for HER2- patients harboring SNRPD1 over-expression suggested the prominent role of HER2 in stratifying breast cancers towards the sensitivity of tumors to SNRPD1 status.
We, in addition, identified one trans-eQTL (rs6733100) of SNRPD1 independently prognostic of breast cancer survival, with the rare allele conferring a protective effect. Given that genes can be regulated in trans by elements on other chromosomes [26], it is possible that somatic non-coding changes of distal regulatory elements can affect breast cancer progression by altering the levels and functions of SNRPD1. Thus, CRIM1, a gene located adjacent to the eQTL variant of SNRPD1 and known with a regulatory role on epithelial-mesenchymal transition (EMT) and lung cancer metastasis [27], may be responsible for the observed role of SNRPD1 on breast cancer cell migration.
Among the spliceosome-related proteins, silencing SNRPD1 or SNRPE remarkably reduced the viabilities of breast, lung and melanoma cancer cells as a result of autophagy [16]. Consistent with these results, we did not observe any difference regarding their roles in cancer cell proliferation but identified a unique role of SNRPD1 in promoting tumor cell migration whereas SNRPE could not. One might argue that the observed influence of SNRPD1 on cell migration might be inflated by its role on cell proliferation. Yet, the data was collected at 12h that was within the time duration cells take to complete one life cycle (24h) and cells were cultured in serum-free medium to remove the influence of cell proliferation on migration. In addition, no significant correlation was found between SNRPD1 expression and KI67 status (a proliferation marker) using GSE24450 (supplementary Table 6). These results collectively suggested that the prognostic value observed for SNRPD1 was driven by its promotive roles on cancer cell migration instead of proliferation.
CENPA is a genomic marker for centromere activity that contributes to the assembly of active kinetochores, overexpression of which promotes genome instability [28]. Specific recognition of centromere-specific histone variant CENPA-containing chromatin by CENPN, a protein dynamically bound to kinetochores during S and G2 and absent from kinetochores during mitosis and G1 [29], is essential in the assembly of the kinetochore complex at centromeres prior to cell division [30]. The fact that silencing SNRPD1 or SNRPE led to specific increase of CENPN and CENPA, respectively, implicated the driving functionality of SNRPD1 on cell cycle that adversely afflicted cells’ genome stability. Flow cytometry assays further consolidated this hypothesis as the S phase of cells was substantially reduced on silencing SNRPD1 (Fig. 7B). Interestingly, increased percentage of ALDH1, a cancer stem cell marker, was observed in cells deficient of SNRPE expression (Fig. 7C), suggestive of the suppressive role of SNRPE on cancer stemness that neutralizes its promotive role on cancer cell proliferation.
Doxorubicin resistance in response to SNRPD1 silencing was observed in TNBC lines (Fig. 5). It is possible that the observed reduction on cells’ sensitivity to chemotherapies was inflated by cells’ slowed growth. However, as SNRPE that reduced cells’ viabilities in a comparable level to SNRPD1 was used as the control gene in this study and silencing SNRPE did not cause curve shift, the observed altered dose response curve after silencing SNRPD1 should reflect its true impact on cells’ drug response. Thus, SNRPD1 over-expression may be a favourable sign on the treatment response of TNBC patients receiving doxorubicin-based chemotherapies. However, this does not imply that patients with high level of SNRPD1 have, in general, better clinical outcome than those with low SNRPD1 expression given the promotive role of SNRPD1 on cancer cell proliferation and migration. As TNBCs are featured by fast growth and thus having been conventionally treated using chemotherapies including doxorubicin, the resistance gained for such tumors on silencing SNRPD1 implicated a reduced chemo-sensitivity of tumors as a result of halted cell cycle progression. Previous reports have suggested therapies targeting spliceosome core machinery such as SNRPD1 and SNRPE16. However, drugs targeting SNRPD1 may not create desirable therapeutic outcome if coupled with anthracycline-like chemotherapies in cancer treatment.
This study is limited by the number of cell lines used for in vitro investigations and lack of in vivo validations. We used MDAMB231 and HCC1937 as the TNBC lines and MCF7 as the luminal breast cancer cells in investigating the effects of SNRPD1 and SNRPE on cancer cell proliferation, apoptosis, migration and drug sensitivity; however, cells representative of other breast cancer subtypes such as HER2-positive tumors were not included. In addition, the key message that SNRPD1 enhances TNBC sensitivity to doxorubicin should be validated in vivo by, e.g., treating SNRPD1-deficient TNBC-carrying mice with doxorubicin and comparing the growth of these tumors with those harboring the wildtype SNRPD1. However, these were not conducted due to the limited number of cell lines available and lack of the animal resource at the time of investigation that awaits to be performed towards generalized conclusions. Additionally, whether the sensitizing role identified for SNRPD1 on TNBC cells in response to doxorubicin could be extended to other types of anthracycline-like chemotherapies needs to be investigated using other therapeutics.
Conclusions
We showed, in this study, that SNRPD1 over-expression was prognostic of poor breast cancer survival, promoted cancer cell proliferation and migration due to, possibly, reduced cell cycle S phase and enhanced genome stability. We also differentiated SNRPD1 and SNRPE regarding their roles on breast cancer progression and onco-therapeutic implications. One eQTL of SNRPD1, rs6733100, was found independently prognostic of breast cancer patient survival. We warranted the combined use of drugs targeting SNRPD1 and anthracycline type of chemotherapies in breast cancer management that may generate undesirable therapeutic outcome as targeting SNRPD1 triggered the resistance of TNBC cells to doxorubicin. Our results differentiated SNRPD1 from SNRPE at both prognostic and therapeutic levels and preliminarily explained the potential driving mechanisms, with the aim of advancing our therapeutic control of breast cancers via targeting these two genes.
Availability of data and materials
The datasets generated during and/or analysed during the current study are available from public repository, all detailed information is recorded in Material and methods.
Abbreviations
- TNBC:
-
Triple negative breast cancer
- PARP:
-
Poly ADP-ribose polymerase
- RNP:
-
Ribonucleoprotein
- snRNP:
-
small nuclear RNP
- eQTL:
-
expression Quantitative Trait Loci
- TCGA:
-
The Cancer Genome Atlas
- GEO:
-
Gene Expression Omnibus
- T:
-
Tumor
- N:
-
Node
- CEU:
-
Caucasion samples
- q-PCR:
-
quantitative real-time PCR
- PI:
-
Propidium Iodide
- DEAB:
-
Diethylaminobenzaldehyde
- RFS:
-
Relapse free survival
- MS:
-
Mass spectrometry
- GO:
-
Gene Ontology
- PPIs:
-
Protein-protein interactions
References
Dai X, Xiang L, Li T, Bai Z. Cancer Hallmarks, Biomarkers and Breast Cancer Molecular Subtypes. J Cancer. 2016;7(10):1281–94.
Dai X, Li T, Bai Z, et al. Breast cancer intrinsic subtype classification, clinical use and future trends. Am J Cancer Res. 2015;5(10):2929–43.
Kim JW, Gautam J, Kim JE, Kim JA, Kang KW. Inhibition of tumor growth and angiogenesis of tamoxifen-resistant breast cancer cells by ruxolitinib, a selective JAK2 inhibitor. Oncol Lett. 2019;17(4):3981–9.
Song D, He H, Sinha I, et al. Blocking Fra-1 sensitizes triple-negative breast cancer to PARP inhibitor. Cancer Lett. 2021;506:23–34.
Velev M, Puszkiel A, Blanchet B, et al. Association between Olaparib Exposure and Early Toxicity in BRCA-Mutated Ovarian Cancer Patients: Results from a Retrospective Multicenter Study. Pharmaceuticals (Basel) 2021; 14(8).
Loap P, Loirat D, Berger F, et al. Combination of Olaparib with radiotherapy for triple-negative breast cancers: One-year toxicity report of the RADIOPARP Phase I trial. Int J Cancer. 2021;149(10):1828–32.
Ricci AD, Rizzo A, Novelli M, et al. Specific Toxicity of Maintenance Olaparib Versus Placebo in Advanced Malignancies: A Systematic Review and Meta-analysis. Anticancer Res. 2020;40(2):597–608.
Mehta AK, Cheney EM, Hartl CA, et al. Targeting immunosuppressive macrophages overcomes PARP inhibitor resistance in BRCA1-associated triple-negative breast cancer. Nat Cancer. 2021;2(1):66–82.
Rahman MA, Krainer AR, Abdel-Wahab O. SnapShot: Splicing Alterations in Cancer. Cell. 2020;180(1):208-e1.
Taylor J, Lee SC. Mutations in spliceosome genes and therapeutic opportunities in myeloid malignancies. Genes, Chromosomes Cancer. 2019;58(12):889–902.
Bowling EA, Wang JH, Gong F, et al. Spliceosome-targeted therapies trigger an antiviral immune response in triple-negative breast cancer. Cell. 2021;184(2):384-403 e21.
Han T, Goralski M, Gaskill N, et al. Anticancer sulfonamides target splicing by inducing RBM39 degradation via recruitment to DCAF15. Science. 2017;356(6336):eaal3755.
Uehara T, Minoshima Y, Sagane K, et al. Selective degradation of splicing factor CAPERalpha by anticancer sulfonamides. Nat Chem Biol. 2017;13(6):675–80.
Sidarovich A, Will CL, Anokhina MM, et al. Identification of a small molecule inhibitor that stalls splicing at an early step of spliceosome activation. Elife. 2017;6:e23533.
Yang H, Beutler B, Zhang D. Emerging roles of spliceosome in cancer and immunity Protein Cell. 2021.
Quidville V, Alsafadi S, Goubar A, et al. Targeting the deregulated spliceosome core machinery in cancer cells triggers mTOR blockade and autophagy. Cancer Res. 2013;73(7):2247–58.
Barrett T, Wilhite SE, Ledoux P, et al. NCBI GEO: archive for functional genomics data sets–update. Nucleic Acids Res. 2013;41(Database issue):D991-5.
Pawitan Y, Bjohle J, Amler L, et al. Gene expression profiling spares early breast cancer patients from adjuvant therapy: derived and validated in two population-based cohorts. Breast Cancer Res. 2005;7(6):R953-64.
Ivshina AV, George J, Senko O, et al. Genetic reclassification of histologic grade delineates new clinical subtypes of breast cancer. Cancer research. 2006;66(21):10292–301.
Lanczky A, Gyorffy B. Web-Based Survival Analysis Tool Tailored for Medical Research (KMplot): Development and Implementation. J Med Internet Res. 2021;23(7):e27633.
Cerami E, Gao J, Dogrusoz U, et al. The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer discovery. 2012;2(5):401–4.
Johansson HJ, Socciarelli F, Vacanti NM, et al. Breast cancer quantitative proteome and proteogenomic landscape. Nat Commun. 2019;10(1):1600.
Monacell JT, Carbone I. Mobyle SNAP Workbench: a web-based analysis portal for population genetics and evolutionary genomics. Bioinformatics. 2014;30(10):1488–90.
Raychaudhuri S, Plenge RM, Rossin EJ, et al. Identifying relationships among genomic disease regions: predicting genes at pathogenic SNP associations and rare deletions. PLoS Genetics. 2009;5(6):e1000534.
Szklarczyk D, Gable AL, Nastou KC, et al. The STRING database in 2021: customizable protein-protein networks, and functional characterization of user-uploaded gene/measurement sets. Nucleic Acids Res. 2021;49(D1):D605–12.
Liu X, Mefford JA, Dahl A, et al. GBAT: a gene-based association test for robust detection of trans-gene regulation. Genome Biol. 2020;21(1):211.
Zeng H, Zhang Y, Yi Q, Wu Y, Wan R, Tang L. CRIM1, a newfound cancer-related player, regulates the adhesion and migration of lung cancer cells. Growth factors. 2015;33(5–6):384–92.
Mahlke MA, Nechemia-Arbely Y. Guarding the Genome: CENP-A-Chromatin in Health and Cancer. Genes (Basel). 2020;11(7):810.
Hellwig D, Emmerth S, Ulbricht T, et al. Dynamics of CENP-N kinetochore binding during the cell cycle. J Cell Sci. 2011;124(Pt 22):3871–83.
Fang J, Liu Y, Wei Y, et al. Structural transitions of centromeric chromatin regulate the cell cycle-dependent recruitment of CENP-N. Genes Dev. 2015;29(10):1058–73.
Acknowledgements
Not applicable.
Funding
This study was funded by the National Natural Science Foundation of China (Grant No. 81972789), the National Science and Technology Major Project of China (Grant No. 2018ZX10302205-004-002), Fundamental Research Funds for the Central Universities (Grant No. JUSRP22011), Technology Development Funding of Wuxi (Grant No. WX18IVJN017). These funding sources have no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Author information
Authors and Affiliations
Contributions
X. D conducted the bioinformatics analysis and drafted the paper. Z. Z and L. C conducted the experiments. X. D, Z. Z, L. C prepared the figures and tables. X. D and J. L provided the financial support. All authors have read and approved the final draft.
Authors’ information
X. Dai, Dean of Plasma Research Laboratory in Shaanxi Provincial Center for Regenerative Medicine and Surgical Engineering, obtained two M.Sc. and two Ph.D. degrees on diverse disciplines in Finland. In the past few years, she has published 93 first-author and/or corresponding-author high-quality SCI papers (accumulated IF ~ 610) with top journals such as Progress in Polymer Science, Advanced Science, Biotechnology Advances, Trends in Biotechnology, Trends in Genetics, Theranostics on cancer biomarker discovery and novel therapeutic design, both genetically and epigenetically, been granted with 2 US patents of invention and 4 Chinese patents of invention as the first inventor. Her current effort and interest focus on combining the multi-effect cold atmospheric plasma with targeted genetic and epigenetic approaches on malignant cancer state control. L. Cai, and Z. Zhang are MSc students of X. Dai. J. Li, MD, MPH, PhD, Vice President of Henan Luoyang Orthopedic Hospital, won the Henan Provincial Youth Science and Technology Award, Henan Provincial Science and Technology Progress Award, etc. Till now he has published 70 papers in top academic internal and international journals, including 48 SCI papers with cumulative impact factor>250, owned 22 domestic and foreign scientific research projects, 21 patents, and 6 scientific and technological achievements awards. His research interest lays on novel approaches for cancer treatment.
Corresponding authors
Ethics declarations
Ethics approval and consent to participate
Not applicable.
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Additional file 1:
Supplementary Table 1. Data sets description for computational prediction at the mRNA and genetic levels. In eQTL analysis, the number of samples overlapping among different types of data and used in the analysis is shown in 'Overlap'. In GEX survival (gene expression association analysis) and SNP survival (association analysis at the genetic level), the sample sizes are shown with the number of events listed in the brackets.
Additional file 2: Supplementary Table 2.
Information on siRNAs purchased for knocking down SNRPD1 and SNRPE and main reagents used in the study.
Additional file 3:
Supplementary Table 3. Information on the sgRNA primers used in the experiments.
Additional file 4:
Supplementary Table 4. Information on the qPCR primers used in the experiments.
Additional file 5:
Supplementary Table 5. Information on all antibodies used in the experiments.
Additional file 6:
Supplementary Table 6. Histopathological association analysis on SNRPD1 gene expression using GSE24450 data.
Additional file 7: Supplementary Table 7.
The eQTL SNPs of SNRPD1 using TCGA data. 'P_TCGA' and 'P_TCGA_coCNV' each shows the p value obtained using TCGA data without and with copy number variation being adjusted.
Additional file 8:
Supplementary Table 8. Genes highly correlated with SNRPD1 or SNRPE.
Additional file 9:
Supplementary Table 9. Genes differentially correlated with SNRPD1 and SNRPE.
Additional file 10:
Supplementary Table 10. Genes differentially correlated with SNRPD1 and SNRPE using both TCGA gene expression and MS protein expression data.
Additional file 11:
Supplementary Table 11. Protein-protein interactions among genes differentially correlated with SNRPD1 and SNRPE. Protein-protein interactions were constructed using STRING.
Additional file 12:
Supplementary Figure 1. Work flow and logic of this study. This study is comprised of investigations on the ‘phenomenon’ of SNRPD1 and SNRPE relevant to breast cancer prognosis and therapeutics, and ‘mechanism’ capable of explaining the observed phenomenon. In each set of investigations, both in silico dry lab analysis and in vitro wet lab experiments were conducted. ‘Green’ and ‘purple’ each represents dry lab analysis and wet lab assays conducted for SNRPD1 and SNRPE, respectively. ‘Black’ represents the analysis or assays. Statements in the brackets are conclusions drawn on SNRPD1 (‘green’) or SNRPE (‘purple’) from the corresponding series of analysis or experiments.
Additional file 13:
Supplementary Figure 2. Cell migration after modulating SNRPD1 at different time points in MDAMB231 and MCF7 cells. (A) Knocking down SNRPD1, (B) Over-expressing SNRPD1.
Additional file 14:
Supplementary Figure 3. The efficiency of over-expressing SNRPD1 or SNRPE in MDAMB231 and MCF7 cells. (A) Over-expressing SNRPD1 or (B) SNRPE as tested at the mRNA level. (C) Over-expressing SNRPD1 or (D) SNRPE as tested at the protein level.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Dai, X., Cai, L., Zhang, Z. et al. SNRPD1 conveys prognostic value on breast cancer survival and is required for anthracycline sensitivity. BMC Cancer 23, 376 (2023). https://doi.org/10.1186/s12885-023-10860-z
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s12885-023-10860-z