A novel long non-coding RNA regulates the integrin, ITGA2 in breast cancer

ITGA2 encodes the integrin, α2 which mediates metastatic progression, and is a predictor of poor prognosis and chemoresistance in breast cancer. Decreased ITGA2 promoter methylation is implicated as a driver of increased gene expression in aggressive prostate and pancreatic tumours, however the contribution of altered methylation to ITGA2 expression changes in breast tumours has not been examined. ITGA2 gene methylation and gene expression was examined in publicly available breast cancer datasets, and ITGA2 promoter methylation was mapped by targeted bisulphite sequencing analysis in breast tumour cell lines. The expression of a putative regulatory long noncoding RNA (lncRNA) was examined by qPCR and its’ functionality was investigated using gene knockdown (antisense oligonucleotides) and over expression in breast cancer cell lines. In breast tumours and breast cancer cell lines the ITGA2 promoter is largely unmethylated, with gene expression variable in tumour subtypes, irrespective of promoter methylation. A novel lncRNA (AC025180.1;ENSG00000249899), named herein I2ALR, was identified at the ITGA2 gene locus, and was variably expressed in breast tumours and breast cancer cell subtypes. I2LAR knockdown resulted in upregulation of ITGA2 gene expression, whilst over-expression of I2ALR resulted in downregulation of ITGA2 mRNA. Further, examination of two downstream targets of ITGA2 associated with breast tumor stemness and metastasis (CCND1 and ACLY), revealed concomitant gene expression changes in response to I2ALR modulation. I2ALR represents a novel regulatory molecule targeting ITGA2 expression in breast tumours; a finding of significant and topical interest to the development of therapeutics targeting this integrin.


Introduction
Cell surface molecules, including integrins, are critical for metastasis and invasion. Integrins are adhesion receptors with extracellular and cytoplasmic signalling domains. The integrin ITGA2, or α2 (CD49b), forms the α2β1 collagen type I (col-I) and laminin receptor [1,2], a key mediator of metastasis and marker of poor prognosis in solid tumors including breast, ovarian and pancreatic cancers [3][4][5][6]. In normal breast tissue, α2β1 is expressed in the mammary gland at cell-cell/basement membrane interface [7], and is regulated by oestrogen [8]. Primary breast cancer (BrCa) tumors retain the elevated α2β1 expression and the estrogen receptor positive (ER + ) status of normal breast tissue, with expression of α2β1 and ER often lost in soft tissue metastases [4]. Luminal/ER + BrCa cell lines (MCF-7, T-47D or MCF-10A) have been reported to express elevated ITGA2, relative to basal-metastatic/ER − lines (MDA-MB-231 and MDA-MB-436) [9][10][11]. However, MDA-MB-231 cells cultured on bone matrix highly express α2 relative to MCF-7 and T-47D cells [12], and preferentially metastasize to bone, where they express elevated a2 [13]. In addition, α2β1 Adele F. Holloway and Joanne L. Dickinson have contributed equally to this work. expression promotes stem-cell like behaviour in triple negative BrCa [14].
Epigenetic modifications facilitate the nuanced control of gene expression and play an important role in the regulation of integrins. DNA-methylation, histone modifications, micro RNAs and long non-coding RNAs (LncRNAs) co-ordinate to deliver precise expression of genes, and present exciting chemotherapeutic options. MiRNAs including miR-373-3p and miR-206 are reported to post-transcriptionally regulate ITGA2 protein levels in BrCa [11,14]. The ITGA2 promoter includes a CpG island (CpGI) which in normal prostate tissues and pancreatic cancer is hypomethylated and associated with upregulated ITGA2 expression. Methylation at the promoter is increased in the locally invasive LNCaP PrCa cell line which expresses low levels of ITGA2. Whilst in bone metastatic PC3 cells, decreased promoter methylation is associated with high α2β1 expression [3,15], which is also reflected in PrCa bone metastases [16]. Histone modifications also regulate ITGA2 [17], and it is hypothesised that a combination of these epigenetic mechanisms contribute to control of expression of this gene in different tumor tissue contexts.
In the present study, we demonstrate that methylation at the ITGA2 promoter CpGI in BrCa cells was not associated with changes in ITGA2 expression, in contrast to previous observations in PrCa, suggesting alternative regulatory mechanisms are at play. Examination of the ITGA2 promoter flanking regions revealed a previously uncharacterized lncRNA, herein named I2ALR (AC025180.1, ENSG00000249899, also referred to as lnc-MOCS2-1). This study provides insight into the tissue-specific differential epigenetic mechanisms regulating ITGA2 and provides important context to current efforts to develop ITGA2-targeted therapeutics.

Quantitation of gene expression
Cell pellets were obtained and RNA was extracted using the RNeasy ® Mini Kit (QIAGEN) and converted to cDNA using the iScript™ Reverse Transcription Supermix kit (Bio-Rad Laboratories Inc.). Alternatively, SuperScript ® III First-Strand Synthesis Kit (Invitrogen, Thermo Fisher Scientific) was used to specifically detect poly-A transcripts, using an oligo(dT) 20 primer, and RNase H. Expression levels of genes of interest were quantitated using quantitative PCR and levels are expressed relative to the housekeeping gene GAPDH. Primer sequences are presented in Supplementary Methods Table 2. Analyses were performed in triplicate using Sen-siFAST™ SYBR ® No-Rox (Bioline). Thermal cycling conditions are presented in Supplementary Methods Table 2. Standard curves were generated to determine absolute gene expression. Gel electrophoresed PCR products were extracted with QIAquick ® Gel Extraction Kit (QIAGEN ® ).

Methylation analysis of ITGA2 CpGI
To map CpGI methylation at the ITGA2 promoter, gDNA was extracted from cell pellets using DNeasy Blood & Tissue kit (QIAGEN), and bisulphite converted with the EZ DNA Methylation-Gold™ Kit (ZYMO Research). Amplified regions of the bisulphite converted promoter (see Supplementary Methods Table 3 for primer sequences) were generated using MyTaq™ HS Mix (Bioline). Fragments were ligated with pGEM ® -T Easy Vector System I (Promega) and transformed used SoloPack ® Gold Competent Cells (Agilent Technologies). Ten bacterial colonies were selected per amplified region, cultured and DNA extracted with the QIAprep ® Spin Miniprep Kit (QIAGEN).

Knockdown of lncRNA I2ALR
Knockdown of lncRNA I2ALR in MDA-MB-453 cells was achieved using antisense oligonucleotide (ASO) gapmers. Two targeting ASOs and a scrambled control (ASO-Scr) were designed and synthesised by Integrated DNA Technologies (Supplementary Methods Table 4). ASOs were transfected into cell lines using the INTERFERin ® in vitro siRNA/miRNA transfection reagent protocol (Polyplus-Transfection), at a final optimized ASO concentration of 10 nM (in 10.5 mL of media), with cells at ~ 50% confluency, and cultured for 24 and 48 h before harvesting.

Overexpression of lncRNA I2ALR
To over-express I2ALR, a construct, named I2ALR construct , was synthesised and cloned into pUC57-Kan by GENEWIZ Inc. (USA). The I2ALR construct contained the combined exons for ENST00000505701.5 and ENST00000503559.1, including an uncharacterized 3′ region (U3R) plus the poly-A signal, and 100 bps of polyadenylation following the poly-A signal (I2ALR construct ; Supplementary Methods Table 5). Using NheI-HF and XhoI restriction enzyme sites incorporated at 5′ and 3′ flanks respectively, the I2ALR construct fragment was ligated into pcDNA TM 3.1 (+) mammalian expression vector (Invitrogen) and transformed into SoloPack Gold Competent E. coli cells. PCR screening using 'I2ALR set 5′ primers (Supplementary Methods Table 2) identified positive colonies which were cultured, and plasmid isolated for subsequent transfection. MCF-7 cells were selected for over-expression experiments as they had moderate ITGA2 and I2ALR expression. Cell lines were transfected by electroporation (5.0 × 10 6 cells per cuvette + 10 µg of plasmid, voltage 230 V, 950 µF, ∞ Ω, distance 4.0 mm). Cells were harvested after 48 h.

Statistical analysis
Comparisons of the differences was conducted with Student's t-tests or one-way ANOVA. Student's t-tests were conducted for comparison of two groups or for when equal variance (homoscedasticity) was appropriate. One-way ANOVA analysis was conducted in Rstudio, with Tukey's post-hoc testing on the ANOVA models. All reported p-values are two tailed. Analyses of publicly available datasets revealed that reduced ITGA2 expression was a significant predictor of distant metastases (Student's t-test, P < 0.05, Fig. 1B) in BrCa. Expression of ITGA2 in BrCa and PrCa was elevated in primary tumors and bone metastases and reduced in soft tissue metastases. Statistical significance was evident between the primary tumors and soft tissue metastases in BrCa datasets (Student's t-test: P < 0.001, Fig. 1C; P < 0.01, Fig. 1D), although statistical comparisons were impacted for the BrCa bone metastasis by the small sample size available. In the PrCa dataset, ITGA2 expression differences between bone metastases compared to soft tissue metastases were significant (Student's t-test, P < 0.01, Fig. 1E).

DNA methylation of the ITGA2 promoter CpGI
Methylation mapping of the ITGA2 promoter in BrCa cell lines by bisulphite sequencing revealed hypomethylation irrespective of the level of ITGA2 expression ( Fig. 2A). In the BrCa dataset (TCGA Cell 2015), ITGA2 promoter methylation was elevated in BrCa soft tissue metastases and lower in both bone metastases and primary tumors (Fig. 2B), although not statistically significant. ITGA2 promoter methylation was only weakly inversely associated with expression in the BrCa samples (TCGA Cell 2015) (Fig. 2C). However, a stronger correlation was observed in PrCa samples (TCGA 1 3 Cell 2015) (Fig. 2D). Interestingly, the majority of PrCa samples showed high overall ITGA2 promoter methylation (methylation β-scores > 0.6, and none with scores < 0.4). By comparison, promoter methylation varied across the full spectrum at the ITGA2 promoter in BrCa samples. Statistical analysis revealed these differences between BrCa and PrCa to be significant (Student's t-test, P < 0.001). These data suggest that alternative regulatory mechanisms may be at play.

Expression of lncRNA I2ALR in BrCa and PrCa
In seeking to understand alternative mechanisms employed by BrCa and PrCa cells in the regulation of ITGA2, the gene locus was examined, revealing an uncharacterised lncRNA transcribed in reverse, downstream of the ITGA2 transcription start site. Given that lncRNAs are known to regulate genes in cis, we sought to examine whether this novel lncRNA regulated ITGA2. The uncharacterised lncRNA gene AC025180.1 (ENSG00000249899), was designated ITGA 2 Antisense LncRNA (I2ALR). I2ALR expression assessed by qPCR was relatively low compared to ITGA2 expression, and of the examined BrCa cell lines, I2ALR was predominantly expressed by MDA-MB-453 cells, followed by T-47D and MCF-7, and MDA-MB-231 with the lowest expression. Expression in the PrCa cell lines PC3 and LNCaP is presented for comparison (Fig. 3A). In BrCa cell lines, I2ALR expression levels were weakly inversely correlated with levels of ITGΑ2 mRNA (Fig. 3B). In contrast, and somewhat surprisingly, in PrCa cell lines I2ALR expression was highest in PC3 cells (Fig. 3A) in which ITGA2 is also highly expressed (Fig. 1A).

Knockdown of I2ALR increases ITGA2 expression
To determine whether modulation of I2ALR influenced ITGA2 expression, antisense oligonucleotides (ASOs) were designed to target the lncRNA (Supplementary Results  Fig. 1A). Transient transfection of two ASOs, ASO-160 and ASO-991, targeting regions of I2ALR, into MDA-MB-453 cells at increasing concentrations (10 nM, 25 nM and 50 nM) revealed both ASOs induced I2ALR knockdown relative to the ASO scrambled control (ASO-Scr) after 24-h. I2ALR knockdown was clearly evident at the lowest concentration (10 nM) and this concentration was therefore selected Transfection of MDA-MB453 cells with ASO-160 and ASO-991 at 10 nM revealed effective knockdown of I2ALR ( Fig. 4A; Student's t-tests, P < 0.05; one-way ANOVA, P = 0.055) and an observed increase in ITGA2 mRNA levels ( Fig. 4B; one-way ANOVA, Tukey's post hoc tests, P < 0.05). I2ALR knockdown was also observed at 48-h (data shown in Supplementary Fig. 1D) with concomitant significant upregulation of ITGA2 expression (data shown in Supplementary Fig. 1E).

Overexpression of the putative I2ALR transcript reduces ITGA2 expression
To further confirm an I2ALR regulatory effect on ITGA2, an I2ALR over-expression construct, I2ALR construct was transfected into MCF-7 cells. High expression of I2ALR construct was confirmed in transfected cells compared with the vector control and untreated cells ( Fig. 4C; oneway ANOVA; Tukey's post-hoc tests, P < 0.001). Expression of ITGA2 was reduced by over-expression of I2ALR relative to vector and un-transfected controls ( Fig. 4D; one-way ANOVA, Tukey's post-hoc tests, P < 0.001).

Effect of I2ALR on downstream ITGA2 targets
In order to determine whether I2ALR-mediated knockdown of ITGA2 impacted downstream genes, the expression of two known ITGA2-responsive genes, CCND1 and ACLY was examined. Increased ITGA2 expression upregulates these genes promoting stemness and metastatic progression in BrCa [14]. Both ACLY and CCND1 were downregulated by over-expression of I2ALR in MCF-7 cells. A statistically significant decrease in ACLY expression was observed when compared with controls (P < 0.05, oneway ANOVA Tukey's post-hoc test). A smaller decrease in CCND1 expression also observed (P < 0.05, Student's t-test, although non-significant using one-way ANOVA) ( Fig. 4E and F).
In a second breast cancer cell line, targeted KD of I2ALR by ASO-160 in MDA-MB-453 cells (associated with increased ITGA2 expression) was also observed to result in a concomitant increase in expression of ACLY and CCND1 (P < 0.05; one-way ANOVA, Tukey's post-hoc test) (Fig. 4G and H) at 48 h. These results are consistent with ITGA2dependent regulation of ACLY and CCND1 reported by Adorno-Cruz et al. [14].

Characterisation of I2ALR
A diagrammatic representation of the I2ALR in relation to the ITGA2 promoter is shown in Fig. 5A showing two putative TSSs. The DNA sequence is shown in Fig. 5B, with additional analysis using alternative primer pairs revealing that the classical poly-A signal is the likely preferred motif to the alternative poly-A signal (see Supplementary Results Fig. 2 for additional analyses). Analysis of lncRNA localization using LncLocator predicts subcellular localizations based on lncRNA sequence. Analysis of the combined exons of ENST00000505701.5 and ENST00000503559.1 indicated that I2ALR was predicted to be translocated to the cytoplasm (57%), and inclusion of the uncharacterized 3′-UTR did not substantially alter this prediction (55%). The predicted localization of I2LAR was also consistent with experimental cell line data available in the LncExpDB database for this transcript (HSALNG0041756) [27].
Given that lncRNAs are known to form lncRNA-[RNA binding protein]-mRNA complexes, putative interactions were investigated in silico between the 1112 bp I2ALR construct sequence and ITGA2 mRNA using the IntaRNA 2.0 bioinformatics tool. Interactions with a seed sequence of > 6 ideal base-pairings were considered, yielding five energetically favourable pairings ( Supplementary  Results Fig. 3A). The lncRNA secondary structure was also predicted with RNAfold ( Supplementary Results Fig. 3B), with two hairpin loop domains apparent and a minimum free energy of − 319 kcal/mol.

I2ALR expression and cancer survival
To examine whether I2ALR is associated with clinical outcome, expression of I2ALR was examined in publicly available cancer datasets. Analysis of GEPIA data revealed elevated I2ALR expression (normalized to GAPDH) in BrCa   Fig. 5 Characterization of the I2ALR and correlation with cancer survival. A Diagrammatic representation of the location of I2ALR (5′−3′) in relation to the ITGA2 locus displaying two bioinformatically predicted TSS for I2ALR. B Amplified I2ALR DNA fragment; confirmed DNA sequence is highlighted in grey with the DNA primers in black. Amplification employing a third primer (highlighted dark grey) suggested that the classical poly-A signal (boxed-bolditalicized text) is preferred to the alternate poly-A signal (underlined text) see data shown in Supplementary Results Fig. 2. C-F Overall survival using GEPIA data for I2ALR (ENSG00000249899) expression, normalized to GAPDH for C breast invasive carcinoma (BRCA), D head and neck squamous cell carcinoma (HNSC), E prostate adenocarcinoma (PRAD), and F liver hepatocellular carcinoma (LIHC). High expression cohort = 'circle' and low expression cohort = 'triangle', gene expression is expressed as transcripts per million with dotted lines = 95% CI, P-value (Logrank p) and hazard ratio P values ("p(HR)") displayed, number of samples in each cohort = "n(high)" and "n(low)" ◂ 1 3 (breast invasive carcinoma) was associated with improved overall survival post diagnosis. However, in later stages of disease this effect was no longer evident and was not statistically significant (Fig. 5C). In keeping with previous findings there was little correlation with overall survival in the PRAD (prostate adenocarcinoma) dataset (Fig. 5E). Examination of other tumor types revealed higher I2ALR expression correlated with improved overall survival ( Fig. 5D and F).

Discussion
Several studies have now shown that multiple epigenetic mechanisms regulate ITGA2 expression in solid tumors including altered promoter methylation and post transcriptional regulation by miRNAs [11,28]. In prostate tumors ITGA2 expression is highly correlated with promoter hypermethylation [15], however here it was shown that in breast tumors the promoter is largely hypomethylated, yet ITGA2 expression remains highly variable. Similarly, BrCa cell lines exhibit hypomethylation at the ITGA2 promoter irrespective of expression, consistent with a recent report [29]. Here we describe a novel lncRNA (I2ALR) which is variably expressed in BrCa cell lines and functions in BrCa cells to regulate ITGA2 gene expression. Further we provide evidence that I2ALR activity may be cell-type specific, as its expression did not correlate with ITGA2 expression in prostate tumor cells.
Adorno-Cruz et al. [14] have recently reported ITGA2 expression is variable in breast tumours with high expression associated with reduced survival, and the significance and magnitude of this effect varies between BrCa sub-types. Here we accessed GEPIA data, and consistent with this finding these analyses showed that higher I2ALR expression was associated with improved survival, up to approximately 120 months, over a similar time frame as that reported by Adorno-Cruz et al. [14]. We were not able to examine I2LAR expression in breast tumor sub-types using the GEPIA data, as it did not permit subclassification based on breast tumor sub-type or stage. This may have impacted the ability detect a significant association with survival. More in depth examination of a role for this lncRNA in breast cancer sub-types is warranted. For other cancers, GEPIA data shows that I2ALR expression was significantly associated with improved survival in several other solid tumor types. Targeted approaches conducted using in vitro and in vivo models representing these tumour types may provide new insights into the role of this lncRNA as a putative modulator of tumorigenic behaviour in these cancers.
High levels of ITGA2 expression has been previously associated with stemness and the initiation of metastases in both prostate and breast tumor cells [30]. Recent studies have shown that ITGA2 induced proliferative and metastatic effects are mediated by the metabolic gene ACLY [14]. ACLY is an enzyme involved in fatty acid synthesis, and has an established role in modulation of proliferation, migration and apoptosis in many tumor types including breast cancer [31]. Similarly, overexpression of the CCND1 gene, a key regulator of cell cycle, is frequently observed in breast tumors. I2ALR overexpression and knockdown experiments resulted in reduced, and increased, ITGA2 expression respectively with concomitant expected changes in ACLY and CCND1 gene expression observed in both MCF7 and MDA MB453 cells. These changes were consistent with decreased expression of ACLY and CCND1 in MDA MB231 cells following siRNA knockdown of ITGA2 and provide strong supportive evidence for a role for this lncRNA in ITGA2 regulation in breast cancer cells. Although both knockdown and overexpression of I2ALR resulted in expected gene expression changes in ITGA2, ACLY and CCND1 in two separate breast cell lines, the impact of I2ALR on cell proliferation and cell migration should be directly examined in future studies.
Although lncRNAs are known to guide epigenetic modifications [32,33], our data indicates a direct interaction with ITGA2 mRNA may be occurring given a suppressive effect of I2ALR was observed in a 24 h time frame. The I2ALR TSS is downstream of ITGA2's promoter CpG island, possibly allowing differential regulation of the lncRNA in the presence of the methylated ITGA2 promoter. Deregulation, or sequestering of I2ALR, may permit the increase of ITGA2 expression required for bone metastasis. Like I2ALR, other antisense lncRNAs also regulate expression of their adjacent gene; e.g. HOTTIP repressing HOXΑ13 [34], and BLNCR regulating ITGB1 [35]. I2ALR was found to comprise 5 exons, with a polyadenylated extended terminal exon (U3R). We hypothesise I2ALR forms a lncRNA-[RNA binding protein]-mRNA complex, destabilizing ITGA2 mRNA, in a similar manner to the lncRNAs, RP11 and HOXA11-AS. These lncRNAs bind cytoplasmic hnRN-PA2B1 and STAU1 proteins, and mRNAs FBXO45/SINH1 and KLF2 respectively. The lncRNA-mRNA complementary interactions guide specificity in the case of RP11, to facilitate degradation of their target mRNAs [36,37]. In silico analyses of I2ALR support this proposed mechanism; predicted cytoplasmic localization (as well as experimentally validated expression) was consistent with similar lncRNAs (e.g. LINC01354 [38]) and complementary regions between ITGA2 mRNA and I2ALR were predicted. Interestingly one of these predicted complementary interactions was in the ITGA2 mRNA 5′-UTR while the remaining four were all within the 3′-UTR, regions typically engaged by regulatory RNAs such as miRNAs. The lncRNA RP11 downregulated SIAH1 and FBXO45 mRNAs by complementary interactions within their CDS and 3′UTR regions [36], and ITGB2-AS1 downregulated ITGB2 mRNA via an interaction in the mRNAs 5′UTR [39].
The α2 integrin has been targeted by small molecules and antibodies as potential therapies for cancer and other diseases, some reaching phase II trials [40,41]. There is also interest in lncRNAs as potential novel therapeutics, and the FDA has approved ASO drugs targeting mRNA transcripts to degrade them via RNase H, or as inhibitors of translation [42]. Furthermore, lncRNAs are potential biomarkers of disease, for example prostate cancer antigen 3 in PrCa [43], and HOTAIR as a marker of chemotherapy resistance [44].
To date our limited understanding of mechanisms controlling the complex temporal and context dependent expression of the ITGA2 gene may have hampered our ability to test prospective therapeutics in appropriate contexts. Identification of this lncRNA and its potential role in ITGA2 gene regulation is an important step forward as it offers novel insight into the cell type and context dependent mechanisms controlling α2 integrin expression in tumor development. Whilst the evidence presented suggests that ITGA2 mRNA is the likely direct target of I2ALR, this does not preclude the possibility that other direct targets exist. This could be addressed with experiments examining direct I2ALR interactions. Taken together, these findings shed light on the role of ITGA2 and the nuanced understanding needed for the development and testing of therapeutics targeting this integrin.

Acknowledgements
The authors would like to convey their sincere thanks to participants of the Tasmanian Familial Prostate Cancer study who have generously consented to participate in this study and allow their stored pathology samples to be used for the purposes of this research. The authors would like to acknowledge Dr Kelsie Raspin provided cDNA samples for PrCa cell lines.
Author contributions JLD and AFH conceived and designed the work, TJV performed laboratory experiments, analysed the data. TJV prepared the original draft of the work. TJV, JLD and AFH interpreted the data, and participated in revisions and editing of the manuscript.
Funding This study was supported by the Andree Greenwood Secondary Breast Cancer Fund; the Hobart Police Charity Trust; and the Royal Hobart Hospital Research Foundation. JLD is supported by a Select Foundation Cancer Research Fellowship, and previously supported by an Australian Research Council Future Fellowship.
Data availability All data generated or analysed during this study are included in this published article and its supplementary information files. Publicly available data that support the findings of this study are openly available https:// www. cbiop ortal. org/.