Background
Breast carcinoma is a huge socio-economic and clinical challenge, being the second most diagnosed cancer for both sexes combined, the most prevalent female malignancy and the leading cause of female mortality worldwide [
1]. Recent data approximates the global cancer mortality at 7.6 million, and breast cancer with an annual incidence of over one million newly diagnosed cases, accounts for over 6 % of this mortality [
2].
Breast cancer, like many polyetiologic human pathologies, is a product of cumulative genetic, epigenetic, somatic, and endocrine aberrations. The polyaetiologism and constitutive complexity of breast cancer presents a challenge for prevention and treatment of breast malignancies. There are five subtypes of breast cancer [
3], and of these, the hormone receptor-negative basal epithelial, commonly referred to as ‘triple negative breast cancer’ (TNBC), is the most aggressive, most common among younger women of African-American and Latina ancestry and has the worst clinical prognosis [
4]. TNBCs are very invasive breast carcinomas lacking estrogen (ER), progesterone (PR), and human epidermal growth factor receptors (HER2) [
5] and are associated with enhanced cellular proliferation, early disease recurrence, and poor overall survival [
6]. However, despite increased knowledge of the aetiology and mechanism of this breast cancer type, developing an effective anti-TNBC therapeutic strategy is still a clinical challenge.
Long non-coding RNAs (lncRNAs) are transcribed RNA molecules longer than 200 nucleotides and endogenously expressed in mammalian cells. Accumulating evidence indicate that lncRNAs, once considered to be genomic anomalies and functionless, do play significant roles in both physiologic and pathologic human conditions through their regulation of defined target mRNA expression, and their post-transcriptional epigenetic modulation [
7]. Following the work of Okazaki et al., which demonstrated that many mammalian transcriptome are non-protein coding and defined lncRNAs as a significant class of these transcripts [
8], it is estimated that about 11 % of the approximately 180,000 large mouse transcriptome are probably protein-coding [
9], and that the number of lncRNAs far exceeds protein-coding mRNAs in the mammalian transcriptome [
10]. Despite the heightened interest in lncRNA biology, their precise function in cellular processes and promoter regulation remains widely undetermined, thus, our study of the functional significance of lncRNAs, their genomic role and their epigenetic regulation.
We hypothesized that the 8.6 kb lncRNA MALAT1 is epigenetically-regulated, and that this is associated with the modulation of various oncogene expression and activities including cell proliferation, migration, invasion, and metastasis. This is consistent with documented evidence that transient MALAT1 overexpression enhanced tumor proliferation in cell lines and xenograft tumor formation in nude mice, while its attenuation resulted in reduced tumorigenicity [
11,
12].
KDM5B belongs to the histone lysine demethylase family, with the ability to cause transcriptional silencing by specifically demethylating di- and tri- methylated lysine 4 of histone 3 of their target genes, and is overexpressed in several carcinomas, including gastric cancer, glioma and breast cancer [
13,
14]. Observing a concomitant increase in MALAT 1 and KDM5B expression as breast cancer progresses, we investigated and validated the hypotheses that MALAT 1 interacts with KDM5B, and that the MALAT1 expression is positively regulated by that of KDM5B in the highly malignant and clinically challenging TNBC. In addition, since highlighting the problem without proffering a solution was not the intent of our work, we systematically screened for an effective therapeutic approach that not only targets KDM5B or MALAT1 expression and/or activities, but also improve clinical outcome, using a combination of small molecule inhibitors, genetic ablation or sncRNA.
Methods
Tissue samples
Twenty human breast tumor tissue samples collected during reduction mammoplasty and classified into various histological subtypes were obtained from archived samples of the Tissue Bank at Taipei Medical University- Shuang Ho Hospital (TMU-SHH). All of the patients gave signed, informed consent for their tissues to be used for scientific research. Recommendations of the Declaration of Helsinki for biomedical research involving human subjects were also followed. Ethical approval for the study was obtained from Joint Institutional Review Board of the Taipei Medical University (approval number: 201202007/ B201112003).
Cell lines and culture
The panel of selected cell lines used in this study consisted of non-tumorigenic MCF-10A and non-metastatic MCF-7 breast myoepithelial cell lines, as well as six breast carcinoma cell lines, MDA-MB-231, MDA-MB-453, HS578T, T47D, AU565 and SKBr3. All cell lines were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA) and cultured according to established standard conditions using RPMI1640 supplemented by 5 μg/ml insulin (Invitrogen, Thermo Fisher Scientific Inc., Grand Island, NY, USA), 10 ng/ml EGF (Sigma), 10 % FBS (Sigma), Penicillin/Streptomycin (Sigma) in a humidified 5 % CO2 incubator. Cells were passaged at 90 % confluence and the medium changed every 48–72 h.
KDM5B knockdown and overexpression
For KDM5B knockdown, MDA-MB-231 cells were infected with KDM5B short hairpin RNA (shRNA, clone ID -TRCN0000329952) targeting the sequence ATCGCTTGCTTCATCGATATT, GTGCCTGTTTACCGAACTAAT, or GCACCAAATTAGAGAGTCT, for clones I, II or III respectively, or vector (pLKO_TRC005), from National RNAi Core Facility, Academia Sinica, Taiwan, then shRNA expressing cells were selected with 1ug/ml puromycin. KDM5B overexpression in MCF10A cells was via transfection of the human KDM5B (NM_006618.3) cDNA sequence cloned into pCMV6-Entry vector (pCMV-KDM5B; E2384, GeneCopoeia, Inc. Rockville, MD, USA) using LipofectAMINE PLUS reagent (Life Technologies, Thermo Fisher Scientific Inc., NY, USA). MCF10A cells were seeded and cultured in 35 mm diameter dishes until 80 % confluence. On the day of transfection, 1 mg of DNA diluted in 100 μl of serum-free medium, and 6 μl of LipofectAMINE PLUS regents were then added. The DNA-PLUS mix was incubated for 20 min at room temperature, and then 4 μl of LipofectAMINE reagent was added and incubated for an additional 20 min. After incubation, the cells were washed with serum-free medium twice and 800 μl of serum-free transfection medium. The DNA-PLUS–LipofectAMINE reagent mix was then added to the cells and incubated at 37 °C in 5 % humidified CO2 incubator for 3 h. After 3 h, recovery medium with 10 % FBS was added till final volume of 2 ml and incubated. After incubating overnight, the recovery medium was suctioned and fresh DMEM medium containing serum and antibiotics added.
RNA extraction, RT-PCR and real time PCR
Total RNA was isolated using TRI Reagent (Sigma) according to manufacturer’s protocol. RNeasy Mini Kit was used for RNA purity optimization. Total RNA concentration was determined using NanoDrop ND1000 spectrophotometer (Nyxor Biotech, Paris). 1 μg of total RNA was transcribed reversely with 2 μg of random hexamers (Amersham, Taipei, Taiwan) and Superscript III reverse transcriptase (Invitrogen, Thermo Fisher Scientific Inc., Grand Island, NY, USA) according to manufacturer’s instructions. DEPC-treated water was used to dilute cDNA 100 folds and stored at −20 °C. Real-time PCR was done using SYBR Green PCR Master Mix using inbuilt System Software (Applied Biosystems, Life Technologies, Grand Island, NY, USA), 200nM forward and reverse primers, and cDNA equivalent of 0.5ug RNA. The triplicate PCR reaction conditions were as follows: 25 °C–5 min, 42 °C–60 min, 70 °C–5 min; total 45 cycles of 70 °C–10 min. 20ul PCR product was loaded to 1.5 % SYBR Green agarose gel for electrophoresis and checked under UV light. Gene expression was normalised to GAPDH and altered expression measured relative to the control (MDA-MB-231 Vector; shKDM5B-vector infected MDA-MB-231 cells).
Western blot analysis
Total cell lysates were prepared and analyzed by western blot assay. Primary antibodies used included KDM5B polyclonal antibody (H00010765-A01; Abnova, Neihu District, Taipei City, Taiwan), Oct 4 Rabbit mAb (C30A3; Cell Signaling Technology, Inc., Beverly, MA, USA), Survivin (FL-142: sc-10811; Santa Cruz Biotechnology, Inc., Dallas, Texas, USA) and GAPDH. Secondary antibodies were Alexa Fluor 680-conjugated affinity-purified anti-mouse or anti-rabbit IgG (Invitrogen, Thermo Fisher Scientific Inc., Grand Island, NY, USA) detected using the UVP Imaging.
2 × 104 cells were seeded into a 6-well cell culture plate and incubated for 2 weeks at 37 °C after treatment. Then, cells were washed twice with PBS, fixed with cold methanol, stained with 0.005 % crystal violet, washed and air dried. Colonies were then counted. In each well, the total colonies with a diameter ≥ 100 μm were counted over 5 randomly selected fields in triplicate assays.
Matrigel invasion assay
Using the 24-well plate Transwell system, 3 × 104 cells were seeded into the upper chamber of the insert (BD Bioscience, 8 μm pore size) containing medium without serum, and medium containing 10 % FBS in the lower chamber served as chemoattractant. After 24 h of incubation, medium was discarded, cells on filter membrane were fixed with 3.7 % formaldehyde for 1 h and stained with crystal violet staining solution, and cells on the upper side of the insert were removed with a cotton swab. The migrated cells were visualized and migratory capacity was evaluated as the total number of cells on the lower surface of the membrane, as determined by microscopy.
MiRNA profiling and secondary structure prediction
We used TargetScan [
15], PicTar [
16] and miRANDA [
17] for miR profiling and target sorting. The M-FOLD program v 2.3, [
18] was employed to predict the secondary structure of hsa-miR-448. The prediction was done as earlier described by Bellucci M and colleagues [
19].
Access and probe of online cancer data set
Publicly available and freely accessible online cancer data repositories used in this study include TCGA, Oncomine, GEO and CCLE. The Cancer Genome Atlas (TCGA) dataset used was the breast invasive carcinoma (BRCA)-IlluminaHiSeq RNAseq,
N = 1182 [
20]. We downloaded and analyzed the TGCA dataset using the UCSC Cancer Browser [
21] and via the Oncomine interface [
22]. We also used dataset GSE3494, platforms GPL 97 from the Gene Expression Omnibus (GEO) [
23] consisting of freshly frozen breast tumors from a cohort of 315 women which represents 65 % of all breast cancers resected in Uppsala County, Sweden, from 1/1/1987 to 31/12/1989,with their estrogen receptor (ER) status determined using biochemical assay.
Immunohistochemical staining and statistical analyses
A total of 270 patients diagnosed with breast carcinoma between January 1, 2005 and December 31, 2010 in Mackay Memorial Hospital (Taipei City, Taiwan) were enrolled for the study. All of the patients gave signed, informed consent for their tissues to be used for scientific research. Recommendations of the Declaration of Helsinki for biomedical research involving human subjects were also followed. Ethical approval for the study was obtained from Joint Institutional Review Board of the Mackay Memorial Hospital (approval number: 11MMHIS154). Patients’ clinical records were reviewed to determine tumor stage at the time of diagnosis and outcome. H&E–stained sections of the mammoplasty specimens were reviewed to select representative areas of the tumor to carry out KDM5B immunohistochemical detection. The working dilution was 1:200. KDM5B immunohistochemistry was carried out using tissue microarray (TMA) on an automated system for immunostaining (Dako Autostainer), with antigen retrieval at high pH. We graded the intensity of the membrane and cytoplasmic staining as absent, weak, moderate, or intense, after stained sections were counterstained with hematoxylin. However, for subsequent statistical analysis we reclassified the cases as high (moderate or intense) or low (null or weak staining similar to control areas of normal breast tissue). In all cases, sections from normal breast tissue bordering the tumor site were used as negative controls. We carried out survival analysis using the Cox univariate and multivariate analyses of proportional hazards model for KDM5B status and selected clinicopathological predictors of outcome. The multivariate model was produced by assessing KDM5B status with other baseline covariates of clinical relevance, such as tumor size, body weight, lymph node metastasis, and hormone receptor status. Log-rank test was used to evaluate significant survival probability differences, while 95 % confidence interval (CI) and hazard ratio (HR) were derived from the regression coefficients. Data were expressed as mean ± standard error of mean, and compared using one way ANOVA and Student’s t-test. p < 0.05 was considered statistically significant.
Discussion
Triple negative breast cancer (TNBC) defines a group of hormone-deficient breast carcinoma associated with very aggressive tumor biology, dearth of documented targeted therapy and almost invariably, unfavourable clinical outcome [
29]. Recently, the identification, validation and integration of novel biomarkers with current diagnostic or prognostic practices in TNBC clinics have been the subject of several studies, with the aim of proffering better patient stratification and more effective therapeutic strategy. In this study, we demonstrated the critical roles of KDM5B and its downstream target MALAT1 in triple negative breast cancer metastatic phenotype and clinical prognosis, based on the following observations: (i.) KDM5B was more abundant in the proteome of the highly metastatic TNBC cells, (ii.) KDM5B was over-expressed in highly invasive TNBC MDA-MB-231 and other metastatic cell lines, than in non-invasive MCF-7 and non-tumorigenic MCF-10A cells, (iii.) KDM5B ablation attenuated tumor cell migration, invasion, clonogenicity, and mammosphere formation ability of breast cancer cells, (iv.) KDM5B-silencing suppressed MALAT1 expression and inhibited cell proliferation, and (v.) human breast carcinoma samples exhibited the highest expression of KDM5B and MALAT1 in advanced stage (T3 and T4) cancer with associated poorer overall survival. These findings were concordant with our hypotheses that KDM5B through the modulation of MALAT1 expression is associated with breast tumorigenesis, progression and poor prognosis.
Recently, the role of epigenetic regulation in breast cancer biology, especially that of the histone lysine demethylases (KDMs), has been the subject of several studies [
30]. This group of chromatin structure modifiers are increasingly shown to facilitate several steps of cancer progression. Several KDMs, including KDM5B have been implicated in tumor growth, angiogenesis, invasion, metastasis, and more recently, in tumor-related chemoresistance [
31,
32]. KDM5B, a member of the JmjC/ARID domain-containing protein family, with restricted tissue distribution, specifically removes methyl residues from methylated lysine 4 of histone 3 (H3K4), consequently repressing gene transcription. This repression is through the discriminate binding of the AT-rich interactive domain (ARID) of KDM5B to CG-rich DNA sites, or its interaction with proteins at the DNA binding domain [
33,
34]. In contrast to its physiologically low expression in normal adult tissue, except in the testes where it is highly expressed, aberrant expression of KDM5B has been demonstrated in skin, lung, prostate, bladder and recently in breast cancer tissues and cell lines, while KDM5B gene silencing, similar to knockdown of KDM3A, was shown to cause a significant G
1/S transitional lag in MCF-7 breast tumor cells, suggesting its proliferative and tumorigenic activity [
35‐
39]. Similar to KDM1A, KDM5B is involved in the silencing of breast tumor suppressors, including BRCA 1 [
40], BRCA 2 [
41], pRB [
42], CAV1 [
43], HOXA5 [
44] and SFN [
45], in addition to its active role in the signal transduction of hormone-regulated organs such as the ovaries, testes and breasts of gravid females, as well as, in transcriptional activation of androgen receptor (AR) [
31,
33].
Recent studies have shown that the dissemination of malignant cells, as well as disease recurrence, is linked to the activity of a small subpopulation of cancer cells believed to possess tumor-growth initiating abilities [
46,
47]. Based on the role of KDM5B in cancer stem cell-like events such as cell fate determination, self-renewal and enhanced cell motility [
48,
49], we hypothesized that KDM5B is actively involved in the highly metastatic phenotype of TNBC cells.
Our immunohistochemistry analysis of metastatic human breast cancer samples indicated a positive correlation between KDM5B expression, advanced tumor stage and poor clinical outcome of patients (Figs.
1 and
2). This is suggestive of KDM5B’s ability to enhance tumor cell invasion, facilitate their homing and increase the clonogenicity of KDM5B-expressing cells. It is thus not unlikely that secondary site colonization by highly metastatic TNBC cells involves the active participation of KDM5B and its functional substrate, MALAT1. Our data revealed a correlative association between the expression and/or activity of KDM5B, MALAT1 and that of MALAT1-regulated effector genes, vimentin and snail (Figs.
3 and
5). In parallel experiments, KDM5B ablation downregulated MALAT1 expression and also blocked the MALAT1-induced expression of snail and vimentin (Fig.
6). Taken together, these data show that KDM5B via modulation of MALAT1 activity plays a critical role in the maintenance of TNBC invasive phenotype, and that the KDM5B–MALAT1 signalling axis regulates the activity of the metastatic EMT factors, such as snail and vimentin. These findings are consistent with previous findings in which MALAT1 was shown to positively modulated tumor cell motility by concomitantly regulating motility factors, including snail and vimentin [
50,
51].
Furthermore, to validate the structural feasibility of KDM5B-MALAT1 complex formation, we used the GUI PyMOL software to demonstrate ligand docking coupled with limited binding site analysis (Fig.
7). In addition, we utilized the catRAPID graphic and strength RNA-protein interaction validation tool for confirmation of KDM5B-MALAT1 complex formation with KDM5B-MALAT1 interaction strength of 93 % (Fig.
7) based on interaction propensity rank of binding regions in the positive set (10201 interacting pairs). We propound that targeting these factors of breast tumor aggression, or their regulatory genes, might be a more effective therapeutic approach for combating metastasis and disease recurrence in breast cancer patients.
MicroRNAs (miRs) play very critical roles in the regulation of many eukaryotic genes and their associated bioprocesses, however, epigenetic dysregulation of these miR activity or expression has been implicated in several human carcinomas including breast carcinoma [
52]. Differential expression of miRs in tumor versus non-tumor tissues, tumor tissue groups with varying degrees of invasiveness, or amongst tumor samples with favourable compared to poor clinical outcome, currently serve as vital template for the generation of miR signatures with potential diagnostic and/or prognostic value. Nevertheless, the biofunctional relevance of aberrant miR expression in triple negative breast carcinoma remained largely underexplored. Thus, we evaluated the interaction between KDM5B and the miR, hsa-miR-448.
Our results demonstrate that the expression of hsa-miR-448 is inversely correlated with that of KDM5B (Fig.
8). Mechanistically, miRs suppress the expression of their target genes by effecting proximity and facilitating interaction between RNAi-induced silencing complex (RISC) effector proteins and complementary sequences of the target mRNA [
53]. In compliance with the consensual miRNA - mRNA rule of functional binding, nucleotide 2 to 7 of the 5′ region of hsa-miR-448 served as the interaction ‘seed’ region. The seed match of this evolutionarily conserved region of hsa-miR-448 bonding in a complementary manner with nucleotides in the 3′ untranslated region (3′UTR) of KDM5B was sufficient for KDM5B mRNA recognition and eventual degradation (Fig.
8). On the molecular level, we demonstrated that this process combines inhibition of translation, and mRNA degradation. Our data show that hsa-miR-448, with as little as 6 base pair (bp) match, significantly suppressed KDM5B expression (Fig.
8). Consistent with emerging model of miR-mediated gene silencing [
54,
55], it is probable that this hsa-miR-448-induced suppression of KDM5B expression is sequel to the deadenylation and exonucleolytic cleavage of KDM5B mRNA.
We are not unaware of the limitations associated with the use of differential gene expression in the identification of miR targets, such as their link with several indirect alterations in transcript abundance, however, this pattern of gene expression utility in the identification of miR targets is in compliance with several seminal reports, including that which indicates that over 84 % of miR-mediated gene suppression was due to reduced mRNA levels [
56‐
58].
Conclusion
Unravelling the molecular mechanism of KDM5B expression in TNBC cells remains a work in progress, however, taken together, our findings indicate that KDM5B is an epigenetic modulator of MALAT1 activity and its downstream effector genes, SNAI and vimentin, as well as plays a critical role in tumor invasion, survival, and niche colonization by the highly metastatic TNBC cells. We demonstrated that KDM5B is a surrogate prognostic biomarker of breast cancer progression and represents a therapeutic target in metastatic breast cancer. Furthermore, we showed that the microRNA, hsa-miR-448 significantly inhibits MALAT-mediated oncogenic and metastatic potential by directly suppressing the expression of KDM5B in triple negative breast cancer, thus, alluding to the clinicopathophysiologic relevance of these findings. The dearth of an effective TNBC anticancer therapy necessitates the development of new therapeutic strategies such as miR replacement therapy. We posit that systemic hsa-miR-448 inoculation, targeting the expression and/or activities of KDM5B may be a stride in the right direction in combating the menace of triple negative breast cancer.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
OAB: Study conception and design, Cell-based and molecular assays. Collection and assembly of data, Data analysis and interpretation, Bioinformatics and Computational biology, Manuscript writing. WCH: Collection and/or assembly of data, Data analysis and interpretation. WHL: Data analysis and interpretation. AW, LSW & MH: Collection and/or assembly of data. CTY & TYC: Study design, Data analysis and interpretation, Final manuscript review. All authors read and approved the final manuscript.