Cis-acting super-enhancer lncRNAs as biomarkers to early-stage breast cancer

Abstract

Background: Increased breast cancer screening over the past four decades has led to a substantial rise in the diagnosis of ductal carcinoma in situ (DCIS). Although DCIS lesions precede invasive ductal carcinoma (IDC), they do not always transform into cancer. The current standard-of-care for DCIS is an aggressive course of therapy to prevent invasive and metastatic disease resulting in over-diagnosis and over-treatment. Thus, there is a critical need to identify functional determinants of progression of DCIS to IDC to allow discrimination between indolent and aggressive disease. Recent studies show that super-enhancers, in addition to promoting other gene transcription, are themselves transcribed producing super-enhancer associated long noncoding RNAs (SE-lncRNAs). These SE-lncRNAs can interact with their associated enhancer regions in cis and influence activities and expression of neighboring genes. Furthermore, they represent a novel, untapped group of therapeutic targets.

Methods: With an integrative analysis of enhancer loci with global expression of SE-lncRNAs in the MCF10A progression series, we have identified differentially expressed SE-lncRNAs which can identify mechanisms for DCIS to IDC progression. Furthermore, cross-referencing these SE-lncRNAs with patient samples in the The Cancer Genome Atlas (TCGA) database, we have unveiled 27 clinically relevant SE-lncRNAs that potentially interact with their enhancer to regulate nearby gene expression. To complement SE-lncRNA expression studies, we conducted an unbiased global analysis of super-enhancers that are acquired or lost in progression.

Results: Here we designate SE-lncRNAs RP11-379F4.4 and RP11-465B22.8 as potential markers of progression of DCIS to IDC through regulation of the expression of their neighboring genes (RARRES1 and miR-200b, respectively). Moreover, we classified 403 super-enhancer regions in MCF10A normal cells, 627 in AT1, 1053 in DCIS, and 320 in CA1 cells. Comparison analysis of acquired/lost super-enhancer regions with super-enhancer regions classified in 47 ER positive patients, 10 triple negative breast cancer (TNBC) patients, and 11 TNBC cell lines reveal critically acquired pathways including STAT signaling and NF-kB signaling. In contrast, protein folding, and local estrogen production are identified as major pathways lost in progression.

Conclusion: Collectively, these analyses identify differentially expressed SE-lncRNAs and acquired/lost super-enhancers in progression of breast cancer important for promoting DCIS lesions to IDC.

Keywords: Breast cancer progression; Ductal carcinoma in situ; Super-enhancer long non-coding RNAs; Super-enhancers.

Fig. 1

Filtering process taken to identify potentially cis-acting SE-lncRNAs from 7753 SE-lncRNAs that may contribute to progression to early-stage breast cancer. a Heatmap of 7753 SE-lncRNAs in MCF10A progression series. Hierarchical Clustering was performed. b Process to identify potentially cis-acting SE-lncRNAs crucial for progression of DCIS to IDC. Red boxes represent SE-lncRNAs that made the “cut.” c Gene Ontology analysis on the list of 138 mRNAs that were neighboring differentially expressed SE-lncRNAs (Fold Change cutoff: 2)

Fig. 2

138 SE-lncRNAs filtered to 27 SE-lncRNAs and their neighboring mRNAs. From 138 SE-lncRNAs, 27 potentially cis-acting SE-lncRNAs and their neighboring mRNA were highlighted

Fig. 3

Expression of 27 potentially cis-acting SE-lncRNAs and their neighboring mRNA and 4 of the highest differentially expressed SE-lncRNAs in the MCF10A progression series. a Expression of 11 SE-lncRNAs and their neighboring mRNAs in progression that correlated in TANRIC. Top are upregulated SE-lncRNAs and their associated mRNA while bottom are down regulated and their associated mRNA. SE-lncRNA-mRNA pairs: FAM83H-AS1 and FAM83H, HOTAIR and HOXC11, CTD-2127H9.1 and OSMR, LINC01057 and SLC44A3, NR2F1-AS1 and NR2F1, HOXA11-AS and HOXA11, RP11-379F4.4 and RARRES1, KB-14601.5 and ZNF706, RP11-403A21.1 and LAMA3, RP3-483K16.4 and ELOVL5, SNHG18 and SEMA5A. b Expression of 16 SE-lncRNAs and their neighboring mRNAs in progression that did not correlate in TANRIC. Top are upregulated SE-lncRNAs and their associated mRNA, while bottom are down regulated and their associated mRNA. SE-lncRNA-mRNA pairs: CTD-2033D15.1 and THBS1, RP11-303E16.3 and CENPN, RP11-346D6.6 and DKK1, RP11-560J1.2 and JARID2, PCAT1 and FAM84B, RP11-61F12.1 and COTL1, RP11-57P1.4 and ADRM1, RP11-326G21.1 and PDE4DIP, RP11-620J15.3 and XRCC6BP1, LINC01125 and ACTR1B, HCG11 and BTN1A1, MIR22HG and WDR81, C4B-AS1 and C4B, CCND2-AS1 and CCND2, FAM13A-AS1 and FAM13A, RP11-137H2.6 and FAM213A. c Expression of the highest differentiated SE-lncRNAs in progression. Top two are upregulated SE-lncRNAs, while bottom two are down regulated. One-way ANOVA with Tukey correction was carried out to evaluate statistical significance of gene expression between cell lines, n = 3, * = P < 0.05, error bars represent standard deviation. Full statistical analysis is presented in Additional file 6: Table S3

Fig. 4

Localization of the potential cis-acting SE-lncRNAs. a Immunoblot of Cell Fractionation of Whole Cell Lysate, Cytoplasmic Fraction, and Nuclear Fraction in MCF10A and CA1 cells. GAPDH was used as control for Cytoplasmic fraction, while Tri-methyl Histone was used as control for Nuclear Fraction. b, c Localization of 14 SE-lncRNAs (11 up-regulated and 3 down-regulated) from our list of 27 potentially cis-acting SE-lncRNAs and 4 highest differentiated that are primarily localized within the nucleus. d Expression level of SE-lncRNA, RP11-379F4.1, and its neighboring mRNA, RARRES1, in MFC10A progression series, n = 3, * = P < 0.05,, one-way ANOVA with Tukey comparison, error bars represent standard deviation. Expression levels of SE-lncRNA RP11-379F4.1 in 24 DCIS and 24 IDC patients (* = P < 0.05), unpaired t test. e Expression level of the highest differentiated SE-lncRNA, RP11-465B22.8, in MCF10A progression series, n = 3, * = P < 0.05, one-way ANOVA with Tukey comparison, error bars represent standard deviation. Expression levels of SE-lncRNA RP11-465B22.8 in 16 DCIS and IDC patients (** = P < 0.005), unpaired t test. f Knockdown of the two target SE-lncRNAs was performed and expression of the SE-lncRNAs and their neighboring mRNAs was determined 48 h post-transfection in DCIS and CA1 cells, n = 3, * = P < 0.05, paired t-test, error bars represent standard deviation

Fig. 5

Super-enhancers Identified in MCF10A progression series. a Super-enhancers quantified in the MCF10A progression series by H3K27ac signal applying the ROSE Algorithm. b Heatmap of super-enhancer regions classified in each cell line in MCF10A progression series and their H3K27ac signal in corresponding cell lines in the series (From Left to Right: MCF10A super-enhancers, AT1 super-enhancers, DCIS.com super-enhancers, and CA1 super-enhancers). Hierarchical Clustering was performed. c H3K27ac signal at AC080013.1 (RP11-379F4.4), the most promising potential cis-acting SE-lncRNA, and AL390719.2 (RP11-465B22.8), the highest up-regulated SE-lncRNA in progression. H3K27ac signal was normalized to mapped reads

Fig. 6

Super-enhancers acquired/lost in progression. Number of newly acquired super-enhancer at each stage in progression in the MCF10A progression series and the number of lost super-enhancer at each stage in progression relative to MCF10A. Super-enhancers that were acquired at a specific stage and were also ranked as super-enhancers in later stages up to CA1 cells were classified as retained. GO analysis of neighboring genes (50 kb up or downstream) for each list of acquired/lost super-enhancer regions classified in the MCF10A progression series. Pathways acquired in progression at each stage are indicated by red arrows. Pathways lost in progression at each stage are indicated by blue arrows

Fig. 7

Comparison of Super-Enhancers Acquired/Lost in Progression with Super-Enhancers Identified in Patient Samples. a Schematic of the comparison between super-enhancers acquired/lost in progression with 47 estrogen receptor positive patients (ER + P), 10 triple-negative breast cancer patients (TNBCP) and 11 triple-negative breast cancer cell lines (TNBCC). Top super-enhancers regions within our comparison are represented by the mRNA nearest the super-enhancer. The stage super-enhancers were acquired/lost at within our progression series, how many patient samples they were present/not present, and the nearest mRNA are represented below the schematic. b 6 of the most interesting super-enhancer regions in our progression series that corresponded with patient samples with their nearest gene highlighted