Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2016 Oct 25.
Published in final edited form as: Cell Rep. 2016 Sep 27;17(1):261–274. doi: 10.1016/j.celrep.2016.08.081

Mammary Tumor Associated RNAs impact tumor cell proliferation, invasion and migration

Sarah D Diermeier 1, Kung-Chi Chang 1,2, Susan M Freier 3, Junyan Song 1,4, Osama El Demerdash 1, Alexander Krasnitz 1, Frank Rigo 3, C Frank Bennett 3, David L Spector 1,2,*
PMCID: PMC5079290  NIHMSID: NIHMS814281  PMID: 27681436

Summary

Long non-coding RNAs (lncRNAs) represent the largest and most diverse class of non-coding RNAs, comprising almost 16,000 currently annotated transcripts in human and 10,000 in mouse. Here, we investigated the role of lncRNAs in mammary tumors by performing RNA-Seq on tumor sections and organoids derived from MMTV-PyMT and MMTV-Neu-NDL mice. We identified several hundred lncRNAs that were over-expressed compared to normal mammary epithelium. Among these potentially oncogenic lncRNAs we prioritized a subset as Mammary Tumor Associated RNAs (MaTARs) and determined their human counterparts, hMaTARs. To functionally validate the role of MaTARs, we performed antisense knockdown and observed reduced cell proliferation, invasion and/or organoid branching in a cancer-specific context. Assessing the expression of hMaTARs in human breast tumors revealed that 19 hMaTARs are significantly up-regulated and many of these correlate with breast cancer subtype and/or hormone receptor status, indicating potential clinical relevance.

Keywords: long non-coding RNAs, transcriptome, breast cancer, mammary tumor, 3D organoid culture, antisense oligonucleotides

Graphical abstract

graphic file with name nihms814281u1.jpg

Introduction

A number of studies over the past decade revealed that only 2% of the human genome encodes for proteins, while as much as 80% can be transcribed in a cell type specific manner (Bertone et al., 2004; Carninci et al., 2005; Djebali et al., 2012; ENCODE Project Consortium, 2012; Kapranov et al., 2007; Katayama et al., 2005; Okazaki et al., 2002). Non-coding transcripts are subdivided into several classes, with long non-coding RNAs (lncRNAs) representing the largest and most diverse class. LncRNAs are defined by length, ranging from 200 nucleotides to 138 kilobases. Like mRNAs, they are transcribed by RNA polymerase II and can be capped, spliced and poly-adenylated. However, lncRNAs lack a significant open reading frame, can be transcribed from the sense or antisense orientation and commonly originate from introns or intergenic regions (Derrien et al., 2012; Iyer et al., 2015; St Laurent et al., 2012). Members of this class of non-coding transcripts have been implicated as regulatory molecules in a variety of cellular functions (for review, see Bergmann and Spector, 2014; Kornienko et al., 2013; Rinn and Chang, 2012).

In recent years, lncRNAs emerged as a novel class of regulatory molecules in cancer (for review, see Cheetham et al., 2013; Prensner and Chinnaiyan, 2011; Shore et al., 2012; Wahlestedt, 2013; Wapinski and Chang, 2011). One of the first examples of a lncRNA that has been studied in the context of breast cancer is the HOX antisense intergenic RNA (HOTAIR) (Rinn et al., 2007). It promotes mammary tumor invasion as well as metastasis and acts as an independent predictor of patient survival rates (Gupta et al., 2010; Rinn et al., 2007). Elevated expression levels of another lncRNA BCAR4 was described to correlate with a higher rate of metastasis and shorter patient survival (Xing et al., 2014). Recently, lncRNA152 and lncRNA67 were shown to play a role in the growth of breast cancer cell lines (Sun et al., 2015). Eleanors, a new class of lncRNAs, were found to be associated with breast cancer resistance and adaptation (Tomita et al., 2015). In addition, genetic knockout or ASO-mediated knockdown of Malat1 results in the differentiation of mammary tumors and significant reduction of metastasis in vivo (Arun et al., 2016). Differential expression of a few additional lncRNAs has been studied in the context of breast cancer, but as for most lncRNAs the functional mechanisms remain elusive (for review, see Hansji et al., 2014; Vikram et al., 2014). As only a handful of the almost 16,000 annotated lncRNAs (GENCODE v24) have been functionally characterized and studied in the context of breast cancer, it is essential to perform an unbiased RNA-Seq screen to identify the complement of lncRNAs that exhibit altered expression in mammary tumors.

Here, we generated a comprehensive compendium of lncRNAs that are differentially expressed in primary mammary tumors compared to the normal mammary gland epithelium. We performed RNA-Seq analysis on three physiologically relevant transgenic mouse models representing the luminal B (MMTV-PyMT) and the HER2/neu-amplified (MMTV-Neu-NDL and MMTV-Cre;Flox-Neo-Neu-NT) subtypes of breast cancer (Andrechek et al., 2000; Guy et al., 1992; Siegel et al., 1999). Among the 290 lncRNAs that are upregulated at least 2-fold in the tumors compared to normal mammary epithelial cells we prioritized 30 previously uncharacterized lncRNAs as Mammary Tumor Associated RNAs (MaTARs). In order to functionally validate the MaTARs we performed antisense knockdown studies in primary cancer cells and 3D ex vivo mammary tumor organoids. Our results indicate that independent knockdown of 20 of the 30 investigated MaTARs significantly reduce mammary tumor cell proliferation, invasion and/or collective cell migration. Notably, we observed phenotypic changes upon MaTAR knockdown in organoids derived from tumors but not in organoids derived from normal mammary epithelial cells, indicating the cancer-specific function of MaTARs. Furthermore, we confirmed the over-expression of 19 human MaTARS (hMaTARs) comparing breast tumors to matched normal controls. Of those, 11 are expressed in a subtype-specific manner. We propose that these previously uncharacterized lncRNAs might act as prognostic indicators and/or drivers of breast tumor progression.

Results

The transcriptome of luminal B and HER2-amplified mammary tumors

Recently, the culturing of mammary organoids in 3D artificial extracellular matrix (ECM) hydrogels has gained popularity over 2D cell culture approaches, especially for studying mammalian development and disease (for review, see Clevers, 2016; Shamir and Ewald, 2014). Previous studies demonstrated that mammary branching morphogenesis can be recapitulated in an organoid system by retaining its epithelial spatial organization (Barcellos-Hoff et al., 1989; Ewald et al., 2008; Fata et al., 2007; Nguyen-Ngoc et al., 2012). Furthermore, organoids developed from tumors have the potential to serve as a rapid in vitro screening system in drug development and “personalized medicine”.

We set out to validate that organoids indeed closely resemble the gene expression signature of the tumors they were derived from. Therefore, we removed mammary tumors from three MMTV-PyMT mice and used one half of each tumor for direct RNA isolation from the tissue sections. The other half was used to generate organoids and RNA was extracted either immediately following organoid preparation (day 0) or after 6 days of culturing in 3D matrigel domes in the presence of FGF2 (day 6). Comparative RNA-Seq analyses revealed that the global gene expression signature of mammary tumor organoids is comparable to the tumor transcriptome (Fig. S1A–B and S2A). Importantly, only 1% of the transcriptome (142 out of the 13,854 expressed genes) showed statistically significant alterations (Fig. 1A–B, Fig. S1C) between the tumor and the day 6 organoids. The affected genes were almost exclusively (85%) protein coding genes (Fig. 1C, Fig. S1D). Several pathways involved in signaling and metabolism appeared altered when taking into account genome-wide fluctuations (Fig. S2B).

Figure 1. The transcriptome of MMTV-PyMT and MMTV-Neu-NDL tumors.

Figure 1

Open in a new tab

A. Venn diagram indicating the overlap of differentially expressed genes across all datasets.

B. Illustration of the number of genes significantly altered comparing tumor, day 0 and day 6 organoid datasets. Percentage of the transcriptome is denoted in brackets, i.e. 142 genes represent 1.0% of the transcriptome in day 6 organoids compared to tumor sections.

C. Transcript biotypes of the 142 differentially expressed genes comparing day 6 organoids to tumors. The majority (85%) of altered transcripts are protein-coding. TEC = To be Experimentally Confirmed (GENCODE classification).

D. Organoids were prepared from MMTV-PyMT, MMTV-Neu-NDL and WT normal mammary epithelia and cultured for 6 days in matrigel (left). UCSC genome browser image (right) displays RNA-Seq reads exemplarily at the Erbb2/neu locus. Vertical viewing range: 0–50.

E. Principal component analysis comparing the transcriptome of organoids derived from WT (purple), MMTV-PyMT tumors (orange) and MMTV-Neu-NDL tumors (green). Three clusters were detected according to the three different mouse models.

F. Examples of GSEA datasets found up-regulated in MMTV-Neu-NDL tumors (upper panel) and MMTV-PyMT tumors (lower panel). FDR < 0.01.

G. Examples of GSEA datasets found down-regulated in MMTV-Neu-NDL tumors (upper panel) and both MMTV-Neu-NDL and MMTV-PyMT tumors (lower panel). FDR < 0.01.

See also Figures S1–S2 and Tables S1–S2.

As our initial comparison demonstrated that organoids are a valid model system to study lncRNA expression, we generated organoids from MMTV-PyMT and MMTV-Neu-NDL mice. After 6 days in matrigel, the transcriptome of the organoids was analyzed using RNA-Seq. To classify cancer-specific changes in the gene expression signature in comparison to normal tissue, we also generated and sequenced organoids from nulliparous mammary glands of three wild type FVB females (Fig. 1D). The RNA-Seq screen revealed significant changes in both the coding and the long non-coding transcriptome, with a total of 4,633 genes differentially expressed in MMTV-PyMT- and 4,322 in MMTV-Neu-NDL-derived organoids. As an internal control, we monitored the expression of Erbb2/neu, which was overexpressed only in the MMTV-Neu-NDL organoids but not in the PyMT or wild type (WT) organoids (Fig. 1D, Fig. S1F). Principal component analysis and Euclidean distance plots comparing WT mammary glands to the luminal B and HER2/neu-dependent mammary tumors revealed distinct clustering according to the tumor-driving transgene (Fig. 1E, Fig. S1E). Furthermore, we observed general and subtype-specific expression changes in cellular pathways (Fig. S2C). Importantly, gene-set enrichment analyses (GSEA) revealed that our mouse mammary tumor organoid data resembled the transcriptome of human breast cancer. For instance, the down-regulated genes of both models correlated well with expression signatures of invasive breast cancer and metastatic tumors. A similar correlation was found for up-regulated genes in both datasets (Fig. 1F–G).

Taken together, these data demonstrate that mouse mammary carcinoma-derived organoids recapitulate the transcriptome of luminal B and HER2/neu-amplified breast cancer and are a reliable model system to study the expression of non-coding RNAs.

Identification of up-regulated lncRNAs in mammary organoids

We set out to generate a compendium of all currently annotated lncRNAs that exhibit altered expression levels in mammary tumors to provide a useful database for future research. We identified 484 lncRNAs in MMTV-PyMT and 402 lncRNAs in MMTV-Neu-NDL with an overlap of 122 between both models that are differentially expressed in organoids derived from mammary tumors compared to WT mammary epithelial cells (Fig. 2A). These genes represent a diverse group in terms of their biotypes and genomic location (Fig. S3A–B). Biotypes include “lincRNA” (long intergenic non-coding RNAs), “processed pseudogenes”, “antisense” transcripts and “processed transcripts.” (Fig. S3A).

Figure 2. Identification and characterization of MaTARs.

Figure 2

Open in a new tab

A. Computational analysis pipeline to identify potentially oncogenic lncRNAs. Numbers indicate genes significantly differentially expressed (FDR adj. p-value < 0.1). Ensembl IDs and log2(fold-change) of all up-regulated (> 2-fold) genes is provided in Tables S1 and S2.

B. Coding probability of all 30 MaTARs according to CPAT. Coding probability is plotted on a scale from 0 to 1, with 0 indicating “no coding potential” and “1” representing “high coding potential”. Insert: 63% of all MaTARs are non-coding (no) whereas 37% are potentially protein-coding (yes).

C. Correlation of the CPAT coding label and lncRNA biotypes. The 37% of transcripts with protein-coding potential in B) are attributed to the pseudogene fraction of MaTARs.

D. Venn diagram illustrating overexpression of the 30 MaTARs in four tumor models. PyMT tumor = tumor sections from MMTV-PyMT sections; Neu tumor = tumor sections from MMTV-Cre;Flox-Neo-Neu-NT. Most MaTARs are up-regulated in several datasets.

E. Length distribution of all 30 MaTARs. The majority (20 MaTARs) is <1500 nt long.

F. Number of exons per MaTAR. More than 50% of MaTARs consist of one or two exons, only three transcripts comprise more than four exons.

G. MaTAR expression (vst counts) across ENCODE tissues, wild type (WT) organoids and tumor organoids and sections. The highest expression level for 18 MaTARs was detected in tumors. All MaTARs are expressed in a tissue-specific manner. PyMT.tumor = tumor sections from MMTV-PyMT sections; Neu.tumor = tumor sections from MMTV-Cre;Flox-Neo-Neu-NT. Counts are scaled per row.

H. Connectivity between genes in the “mammary epithelium” module is plotted against the module’s eigengene correlation. Red dotted lines correspond to first and third quartiles. Six MaTARs are in the fourth quartile, denoted by high eigengene correlation and intra-module connectivity.

See also Figures S1, S3–S4 and Tables S3–S4.

Further, we aimed to focus on potentially oncogenic lncRNAs that might act as drivers of tumor progression and filtered our dataset for lncRNAs that are up-regulated at least 2-fold. We identified 109 up-regulated lncRNAs in MMTV-PyMT (Table S1) and 207 in MMTV-Neu-NDL organoids (Table S2), with an overlap of 26 lncRNAs (Fig. 2A, highlighted in grey in Tables S1 and S2). Interestingly, very few of the identified transcripts have been studied previously. One of these lncRNAs is Trp53cor1 (also known as lincRNA-p21), which is up-regulated at least 4-fold in both mammary tumor models (Tables S1 and S2) and has been implicated in p53/p21-dependent gene regulation (Dimitrova et al., 2014; Huarte et al., 2010) as well as cancer (Chou et al., 2015; Yang et al., 2014). Furthermore, the two Pluripotency-associated transcripts Platr4 and Platr7 were up-regulated more than 4-fold in MMTV-Neu-NDL organoids (Table S2). Notably, Platr lncRNAs were identified in an RNA-seq screen of embryonic stem cells (ESCs) and shown to be correlated to the maintenance of the ESC expression profile (Bergmann et al., 2015). We performed coding potential analysis on all 290 up-regulated lncRNA genes (443 associated transcripts in total) using the Coding-Potential Assessment Tool (CPAT) (Wang et al., 2013). Our results revealed that 75% of the 443 identified transcripts do not exhibit coding potential (Fig. S3C–F) despite including pseudogenes in our analysis. As expected, we observe a good correlation between coding probability and ORF length but not RNA length (Fig. S3E–F).

We further characterized the identified differentially expressed lncRNA genes regarding shared transcription factor binding sites at their promoter regions. To do so, we performed motif analyses in the regions of −900 bp to +100 bp around the TSS of all differentially regulated lncRNA genes, identified in both PyMT and Her2/neu organoids. For down-regulated promoters, we observe the enrichment of a rather rare motif matching Zbtb33/Kaiso as well as motifs for NF-κB and CEBP (Fig. S4A). In contrast, up-regulated lncRNA genes seem to be primarily regulated by androgen receptor (AR) and p63 (Fig. S4B).

To narrow down the long list of candidates to the ones with highest potential to impact mammary tumors, we prioritized the transcripts further based on several stringent criteria: i) statistical significance (FDR adjusted p-value < 0.1), ii) at least 2-fold up-regulation in tumor organoids and/or tumor sections compared to normal mammary organoids, iii) sufficient read coverage per transcript to eliminate very lowly abundant RNAs (Fig. S1G), iv) human conservation based on sequence and/or synteny, v) location in an intergenic genomic region and vi) lack of highly repetitive elements. The last two points ensure knockdown specificity, i.e. that antisense oligonucleotides (ASOs) can be designed for the lncRNA of interest while avoiding potential off-target effects. Out of 290 up-regulated lncRNAs, 30 transcripts fulfilled the above-mentioned criteria and were selected as an initial subset for further characterization. All 30 are overexpressed in at least 4 out of the 5 sequenced PyMT organoid datasets as well as significantly up-regulated in PyMT tumor sections (Fig. 2D). We also included RNA-Seq data from MMTV-Cre;Flox-Neo-Neu-NT tumors, a second model for the HER2/neu-amplified subtype of human breast cancer, to further refine our candidate lncRNAs. We termed the prioritized transcripts Mammary Tumor Associated RNA 1-30 (MaTAR1-30). A complete list of their official gene IDs and expression levels in all evaluated mammary tumor models is presented in Table S3. Of the 30 MaTARs, only MaTAR4 and MaTAR8 have been characterized in the past. MaTAR4 (Hoxa transcript antisense RNA, myeloid-specific 1; Hotairm1) has been implicated in myelopoiesis and myeloid maturation (Zhang et al., 2009; 2014). MaTAR8 (lncRNA-Smad7) has been described as a TGF-β – regulated antisense transcript of Smad7 inhibiting apoptosis in mouse breast cancer cells (Arase et al., 2014). We tested the coding potential of all MaTARs using CPAT (Wang et al., 2013). While 60% of the MaTARs were classified as non-coding, 40% were predicted to be coding, attributed to the pseudogene fraction (Fig. 2B–C). We further characterized the MaTARs according to their total length and number of exons (Fig. 2E–F). MaTARs range in size from 291 – 4867 nt with the majority of MaTARs being shorter than 1500 nt. More than half of all MaTARs consist of one or two exons, only 10% show complex structures with more than 4 exons.

In order to compare the expression levels of MaTARs in mammary tumors to other tissue types, we analyzed the ENCODE expression datasets of adult mouse tissue as well as embryonic stem cells (Bergmann et al., 2015) (Fig. 2G). Our results confirm the general notion of tissue-specific expression of lncRNAs (Cabili et al., 2011; Derrien et al., 2012), with the majority of the MaTARs (18 out of 30) being expressed at very low levels in tissues other than mammary tumors. The remaining 12 MaTARs (MaTAR1, 3, 7, 9, 10, 11, 12, 18, 19, 23, 26 and 27) are also overexpressed in mammary tumors compared to normal mammary glands, but in addition these MaTARs are present in comparable levels in at least one other tissue type.

To further characterize MaTARs in the context of global gene expression and to assess their potential role in mammary tumor progression, we performed weighted gene correlation network analysis (WGCNA) (Langfelder and Horvath, 2008). We identified 39 modules with a median gene number of 274. Remarkably, 17 of the 30 MaTARs are residing within the same module. This particular module (“blue”, Table S4) is enriched for genes that are highly expressed in our RNA-Seq datasets, both in the normal mammary gland organoids as well as mammary tumors and hence was termed the “mammary epithelium” module. It is rather large, comprising 5,744 genes in total. Interestingly, only 2,588 of the module genes are classified as “protein coding”, emphasizing the tissue-specificity of non-coding RNA species as well as their potential functional role.

In order to identify key drivers of the module, we focused on potential “hub genes” with high intra-module connectivity (> 1550) and very good correlation with the module eigengene (≥ 0.9), marked in red in Fig. 2H. At least 6 MaTARs (MaTAR5, 6, 9, 15, 16 and 30) fall into this region, indicating their importance within the module. Notably, 12 of the 17 MaTARs in the “mammary epithelium” module were also expressed higher in tumor tissue compared to normal organs (Fig. 2G). This indicates that MaTARs are excellent marker genes and sufficient to stratify mammary epithelia from other tissue types. GO analysis of the module revealed association with several interesting biological processes that are in agreement with the top “hub genes”, including 33 ribosomal and snoRNA genes, several components of the spliceosome, cyclooxygenase 2 and 3 but also typical marker genes for mammary epithelia like caseins and lactalbumin.

In summary, our findings demonstrate that we identified a module related to mammary epithelial development and/or mammary carcinogenesis, and suggest that MaTARs are likely to play key roles in these functional processes.

Knockdown of MaTARs decreases cell viability and invasion

The results of our computational approach suggested that multiple MaTARs may contribute to mammary cancer progression. We tested this possibility by performing knockdown experiments of all 30 MaTARs using antisense oligonucleotides (ASOs) initially in primary MMTV-PyMT tumor cells. These oligonucleotides are short (20-mers), single stranded DNA molecules containing phosphorothioate-modified nucleotides as well as modifications of the 2′-ribose (5-10-5 2′-MOE gapmer) (for review, see Geary et al., 2015). Importantly, we found that the uptake of ASOs in primary mammary tumor cells is efficient without the use of transfection agents, a mechanism that has been studied in detail in hepatocytes (Koller et al., 2011). One of the candidates, MaTAR2, was not expressed at levels consistent enough for reliable quantification using qRT-PCR and hence was excluded from knockdown experiments. For each of the remaining 29 MaTARs, we tested 10 different specific ASOs and achieved knockdown efficiencies ranging from 38% – 89% after 24 h using 5 μM of the most potent ASO (Fig. S5A).

To initially investigate the functional impact of MaTAR down-regulation on tumor cells, we combined the ASO treatment with cell viability or invasion assays (Fig. 3A–B). Significant changes were defined as a consistent reduction ≥25% of viability or invasion compared to untreated control cells. Interestingly, MTT assays revealed that the knockdown of MaTAR12 leads to a 50% decrease in cell viability (Fig. 3A), an effect comparable to ASO-mediated knockdown of Eg5 that served as a positive control (Koller et al., 2006). Knockdown of three additional candidates (MaTAR11, 18 and 20) resulted in ≥25% decrease of cell viability. We did not observe an effect on cell viability with a scrambled ASO control (scASO) (Fig. 3A). To confirm that our results obtained with the MTT assays are in fact detecting differences in cell viability, we performed assays measuring protease activity within living cells on a subset of the MaTARs and obtained similar results (Fig. S5B).

Figure 3. ASO-mediated knockdown of MaTARs in primary tumor cells.

Figure 3

Open in a new tab

Primary mammary tumor cells were treated with 5 μM of the most potent ASO for 24 h.

A. MTT assay after ASO-mediated knockdown of MaTARs. Cell viability is normalized to untreated cells (“Control”). A scrambled ASO (“scASO”) was used as a negative control; ASO knockdown of Eg5 was used as a positive control. Bars denote the mean of two replicates +/− std. err.

B. Invasion assay after ASO-mediated knockdown of MaTARs. Like A; serum-free medium was used as positive control.

C. Summary of observed effects on cell viability, cell invasion and organoid branching upon ASO-mediated knockdown of MaTARs. Data is normalized to untreated cells (= 100%). Reduction of viability, invasive potential or branching of ≥25% compared to untreated cells was defined as a significant difference. The color legend is indicated.

See also Figure S5.

In addition, we performed invasion assays of tumor cells in 96-well Boyden chamber plates upon ASO-mediated knockdown. We used serum-free medium as a positive control and detected a 25–45% reduction of tumor cell invasion upon independent knockdown of 15 candidates (MaTAR8, 9, 13, 14, 15, 16, 17, 19, 20, 21, 23, 24, 25, 28 and 29; Fig. 3B). While knockdown of several MaTARs impacts either the viability or the invasive potential of mammary tumor cells, indicating specific roles of the different non-coding transcripts in cellular processes, only MaTAR20 showed effects in both assays (Fig. 3C). Of the 29 tested MaTARs, 11 lncRNAs (MaTAR1, 3, 4, 5, 6, 7, 10, 22, 26, 27 and 30) did not exhibit significant effects on cell growth or invasion upon knockdown.

As an example validation of the antisense knockdown experiments, we used the CRISPR/Cas9 system (Ran et al., 2013) to generate a knockout clone of MaTAR25, the lncRNA that exhibited the most dramatic impact on cell invasion (−45%, Fig. 3B). Indeed, the genetic loss of MaTAR25 significantly reduced tumor cell invasion as well (−55%, Fig. S5C–E), recapitulating the ASO-mediated knockdown. Our results indicate that our initial computational selection of up-regulated lncRNAs identified a number of transcripts that have the potential to impact tumor cell growth or invasion.

Knockdown of MaTARs inhibits collective cell migration in organoids

To further elucidate the functional role of MaTARs in a more physiological context, we performed ASO-mediated knockdown experiments in mammary tumor organoids (Barcellos-Hoff et al., 1989; Ewald et al., 2008; Fata et al., 2007; Nguyen-Ngoc et al., 2012). As for primary tumor cells, we found that mammary organoids perform transfection-free uptake of ASOs.

For the 29 tested MaTARs, we observed knockdown efficiencies in organoids ranging from 30 – 68% after 6 days of treatment using 4 μM of the most potent ASO (Fig. S6A). Untreated as well as scASO treated MMTV-PyMT organoids generally exhibit branching in about 70–75% of all organoids. Upon ASO-mediated knockdown, we detected a significant decrease (25–50% less compared to untreated organoids) in branching morphogenesis for 11 candidates (MaTAR6, 11, 12, 16, 17, 18, 19, 20, 25, 26 and 28; Fig. S6B). Interestingly, many of the MaTARs whose knockdown interfered with cell viability and/or invasion in 2D assays also exhibited a branching defect in organoids (Fig. 3C). However, this correlation is not true for all MaTARs, indicating that 2D culture models different aspects of tumor cell growth than organoids.

As expected, we do not see phenotypic changes upon ASO-mediated down-regulation for all MaTARs. While we obtained a knockdown efficiency of 55% for MaTAR8 (ENSMUSG00000092569, Gm20544, lncRNA-Smad7), we did not observe a loss of branching or any other phenotypic change in organoids (Fig. 4D, additional images in Fig. S6D). This is in agreement with previous findings describing that MaTAR8 is not involved in epithelial mesenchymal transition (EMT) (Arase et al., 2014). Our results indicate that different MaTARs might impact different aspects of tumor biology.

Figure 4. ASO-mediated knockdown of MaTARs in mammary organoids.

Figure 4

Open in a new tab

Organoids were treated with 4 μM of specific ASOs for 6 days. Statistical significance was determined using a two-tailed, paired student’s t-test; * p < 0.05, ** p < 0.01, *** p < 0.01. Scale bar = 100 μm

A. DIC images of tumor organoids upon knockdown with the two most potent ASOs targeting MaTAR16, MaTAR18 or MaTAR26 are shown. Knockdown of MaTAR16 and MaTAR26 leads to a loss of branching while knockdown of MaTAR18 results in organoids with tiny protrusions.

B. qRT-PCR to determine the knockdown efficiency of two specific ASOs per MaTAR, normalized to untreated cells (“Control”). Bars denote the mean of three replicates +/− std. dev.

C. Quantification of tumor organoid branching upon knockdown with two specific ASOs per MaTAR. Reduction of organoid branching correlates directly with ASO-mediated knockdown efficiency. Black and grey bars indicate the percentage of branched organoids +/− std. dev.; white bars indicate the percentage of unbranched organoids. The mean of three biological replicates is shown; total number of assayed organoids per treatment = 300.

D. Knockdown of MaTAR8 using two specific ASOs in tumor organoids. Upper panel: qRT-PCR to determine the knockdown efficiency. Bars denote the mean of three replicates +/− std. dev. Middle panel: Quantification of organoid branching. The mean of three biological replicates is shown; total number of assayed organoids per treatment = 300. Lower panel: DIC images of organoids upon knockdown.

E. DIC images of WT normal mammary epithelia organoids upon knockdown with the two most potent ASOs targeting MaTAR16, MaTAR18 or MaTAR26 are shown.

F. DIC images of C57Bl/6 MMTV-PyMT organoids upon knockdown with the two most potent ASOs targeting MaTAR16, MaTAR18 or MaTAR26 are shown.

See also Figures S6–S7.

We further focused on MaTAR16 (ENSMUSG00000086249, Gm12724), MaTAR18 (ENSMUSG00000085873, Ttc39aos1) and MaTAR26 (ENSMUSG00000097378, B230208H11Rik), as these three transcripts showed very distinct organoid phenotypes upon ASO-mediated knockdown (Fig. 4A). The three RNAs represent different transcript biotypes with MaTAR16 being classified as “processed transcript”, MaTAR18 as “antisense” and MaTAR26 as “lincRNA”. Both MaTAR16 and MaTAR26 are part of the “mammary epithelium” module, while MaTAR18 resides in another highly interesting module that is comprised of the gene-expression signature for embryonic stem cells and embryonic liver. This module is associated with GO terms related to replication and the cell cycle.

Knockdown experiments were performed with two different ASOs for each transcript to ensure that the observed phenotypes are not caused by off-target effects. Phenotypes for both ASOs are shown in comparison to organoids that were either untreated or treated with scASO (Fig. 4A with additional images in Fig. S6C). While knockdown of MaTAR16 and MaTAR26 completely abolished organoid branching, down-regulation of MaTAR18 resulted in organoids without defined branches but with a “rough” surface. In addition, we observed that organoids treated with ASOs targeting MaTAR16 tend to be smaller compared to untreated or scASO treated organoids (Fig. 4A). Individual knockdown efficiencies are displayed in Fig. 4B and correlate well with the loss of branching (Fig. 4C). Compared to untreated controls, we detected a reduction in branching of 38% for MaTAR16, 40% for MaTAR18 and 35% for MaTAR26 with the best ASO, respectively.

To further elucidate the role of MaTAR16, MaTAR18 and MaTAR26 we performed ASO-mediated knockdown in control organoids derived from WT nulliparous mammary glands. Knockdown efficiencies of the ASOs are comparable to the experiments in tumor organoids (Fig. S7A). Importantly, we do not see a pronounced loss of branching in the normal mammary gland organoids, strongly indicating the cancer-specific role of these three lncRNAs (Fig. 4E). In addition, this experiment serves as another control, confirming that the observed reduction in branching morphogenesis of tumor organoids is not due to ASO induced toxicity or off-target effects.

To exclude the possibility that our results are dependent on the genomic background of the mice, we confirmed the findings for MaTAR16, MaTAR18 and MaTAR26 in organoids derived from C57Bl/6 MMTV-PyMT tumors (Fig. 4F). Knockdown efficiencies and branching quantifications are shown in Fig. S7B.

Human MaTARs are overexpressed in breast cancer

To confirm the relevance of MaTARs in breast cancer we set out to identify human orthologs of all MaTAR genes. We compared mouse and human transcripts on the level of both sequence conservation and genomic location. If the mouse MaTAR sequence matched an annotated human gene using the BLAT alignment tool, the respective transcript was defined as the human counterpart. For instance, BLAT results for MaTAR1 indicate sequence identity in the human genome as well as a similar genomic context. It matches the last exon of a longer lncRNA (LINC00461, ENSG00000245526) in close spatial proximity of a miRNA gene (Fig. 5A). The sequence of MaTAR1 spans 2419 bp in the human genome with a sequence identity of 92.6% and a BLAT score of 1438. This genomic region is highly conserved in vertebrates as indicated by the PhyloP conservation track (Fig. 5A). The sequence-based approach is not informative for pseudogenes, as the parental protein-coding genes are the best sequence matches, e.g. FAM96B for a Fam96b pseudogene such as hMaTAR2 (Table S5). As many non-coding RNAs are conserved between different species on the level of genomic location rather than based on sequence, possibly implying functional conservation (Diederichs, 2014; Ulitsky et al., 2011), we extended our analysis to synteny by analyzing the nearest neighboring genes of each hMaTAR gene. The complete list of the genomic locations of potential hMaTARs is provided in Table S5. Interestingly, many hMaTARs are located in clusters of two or more non-coding genes in the genome, resulting in a total of 41 identified potential hMaTARs. Experimental validation will be necessary to identify the correct human counterpart unambiguously.

Figure 5. hMaTARs are overexpressed in breast cancer.

Figure 5

Open in a new tab

A. Identification of a human MaTAR gene based on sequence identity. The upper panel represents the mouse genome (mm10), the lower panel the human genome (hg38). MaTAR1 is highlighted (ENSMUST00000052354), ENSMUST00000083490 = miR-9-2.

B. Expression fold-change of hMaTAR1 published TCGA RNA-Seq datasets compared to the respective normal tissue. FDR < 0.00001. KIRC = kidney renal clear cell carcinoma, BRCA = breast invasive carcinoma, LUSC = lung squamous cell carcinoma.

C. Differential gene expression of hMaTAR1 in subtypes of human breast cancer. Her2 = HER2-amplifed, LumA = luminal A, LumB = luminal B.

D. Correlation of hMaTAR1 expression to estrogen receptor (ER, left) and progesterone receptor (PR, right) status.

E. Survival curve for the luminal B subtype, comparing patients in the upper quartile of hMaTAR1 expression to patients in the lower quartile of hMaTAR1 expression.

F. Reactome pathways positively or negatively correlated to hMaTAR1 expression.

See also Figures S8–S9 and Tables S5–S7.

To evaluate the expression status of the potential hMaTARs, we analyzed RNA-seq data of The Cancer Genome Atlas (TCGA) by comparing breast tumors to matched normal tissues of 103 patients. Out of the 41 hMaTARs, we were able to obtain differential gene expression for 40 (Table S6). hMaTAR3 dropped out due to insufficient read coverage. Notably, 28 hMaTARs were found to be up-regulated in breast tumors, 19 of those with high statistical significance (adj.p < 0.05), confirming that our initial RNA-seq screen in mouse models identified multiple lncRNAs that are potentially clinically relevant. We further investigated whether hMaTARs are expressed in a subtype-specific manner by analyzing all TCGA breast tumor samples with subtype information (463 datasets). Of the 19 up-regulated hMaTARs, 11 were expressed significantly different (q < 5.00E-11, same range as PAM50 genes (Parker et al., 2009)) among the five TCGA subtypes “basal”, “Her2-amplified”, “luminal A”, “luminal B” and “normal-like” (Table S7, Fig. S8). In addition, hMaTAR4 and hMaTAR23 showed strong subtype-specificity, indicating that these lncRNAs may be clinically relevant in certain subtypes despite not being significantly up-regulated across the board (Table S6).

Here, we focus on hMaTAR1, as this lncRNA was up-regulated the most in our original screen (80- and >100-fold in Her2-amplified and PyMT tumors, respectively). We found hMaTAR1 to be significantly over-expressed in breast tumors compared to normal breast tissue (Table S6). We confirmed our analysis using previously published TCGA datasets (Yan et al., 2015) and detected a slight, but significant, up-regulation in kidney carcinoma as well as substantial over-expression in lung squamous cell carcinoma (Fig. 5B). Moreover, we observed statistically significant differences in the expression levels of hMaTAR1 comparing different subtypes of breast cancer, with lowest levels in the luminal A and highest levels in the basal subtype (Fig. 5C, Table S7). Notably, we also detected a strong correlation of high hMaTAR1 levels and negative ER/PR status (Fig 5D, Table S7), matching the subtype data. Similar observations were made for 7 additional hMaTARs (Table S7, Fig. S9). Interestingly, the high expression levels in ER−/PR− tumors did not correlate with a genomic amplification of the hMaTAR1 locus (data not shown), indicating that up-regulation of this lncRNA may be caused by changes in gene expression and/or epigenetic factors. Survival analysis revealed that patients with low levels of hMaTAR1 survive on average ~1500 days (approx. 4 years) longer than patients with high levels of this lncRNA (Fig 5E). In addition, we performed pathway analysis of TCGA data and observed a significant positive correlation of pathways such as “cell cycle” and “DNA replication”, hinting to a high proliferative rate of tumors with elevated hMaTAR1 levels.

In summary, we performed an RNA-Seq screen to identify lncRNAs that are up-regulated in mammary tumors and represent potentially oncogenic transcripts. We identified 109 lncRNAs that are up-regulated at least 2-fold in MMTV-PyMT and 207 that are up-regulated at least 2-fold in MMTV-Neu-NDL tumors. Among these transcripts we prioritized 30 as MaTARs and showed that ASO-mediated knockdown of 20 MaTARs significantly reduces cell proliferation and/or invasion in 2D cell culture as well as branching in organoids. Phenotypic changes were only observed in organoids derived from mammary tumors but not normal mammary glands, indicating a specific role of MaTARs in cancer. A significant number of hMaTARs are overexpressed in human breast tumors and their expression signature correlates with subtype and hormone receptor status.

Discussion

In recent years, high-throughput sequencing studies have begun to shed light on the emerging role of non-coding transcripts in cancer. Several studies suggest lncRNAs as novel prognostic biomarkers, tumor suppressors and oncogenes (for review, see Cheetham et al., 2013; Huarte, 2015; Prensner and Chinnaiyan, 2011; Shore et al., 2012; Wahlestedt, 2013; Wapinski and Chang, 2011). However, many global gene expression studies were performed in human cancer cell lines. There are several potential caveats to using cell lines to study tumor biology, such as adaption to growth in vitro on plastic dishes, lack of corresponding normal tissue controls, cell immortalization, lack of a physiological ECM and accumulation of chromosomal aberrations due to continued passaging. Importantly, transcriptome analyses revealed that commonly used cancer cell lines often do not represent tumors well enough and that tumors cluster closer to their corresponding normal tissues than to cell lines (Domcke et al., 2013; van Staveren et al., 2009). Primary 3D ex vivo organoids resemble the organization and physiology of native epithelia much better than cancer cell lines grown in 2D, and additionally model interactions with the ECM (Boj et al., 2015) (for review, see Clevers, 2016; Sachs and Clevers, 2014; Shamir and Ewald, 2014). To investigate whether organoids mirror the tumor on the level of global gene expression as well, we performed the first RNA-Seq screen comparing the transcriptome of organoids to the mammary tumors from which they were derived. Importantly, our data demonstrate that non-coding RNA levels seem to remain virtually unchanged. Our findings support the potential use of patient-derived organoids as a model system to study tumor biology and to model drug responses (“personalized medicine” (Boj et al., 2015); for review, see Lancaster and Knoblich, 2014; Shamir and Ewald, 2014).

We performed an unbiased RNA-Seq screen to categorize the compendium of differentially regulated lncRNAs in mammary tumors compared to normal mammary epithelial cells. While we confirmed several known cancer-associated lncRNAs such as lincRNA-p21 in our screen, the majority of the 764 differentially expressed candidates have not been described previously in the context of cancer. Our study emphasizes the importance of unbiased screening approaches and represents a valuable resource of lncRNAs to pursue in regard to their involvement in mammary carcinogenesis and as prognostic and/or therapeutic targets.

Genome-wide gene expression studies often lack systematic functional validation. Here, we characterized the identified mammary tumor associated lncRNAs using both computational approaches and molecular assays. Motif analysis of all differentially regulated lncRNAs in our study revealed specific regulators for up- and down-regulated non-coding RNA genes. Repressed lncRNAs show an enrichment in their promoter regions for the transcription factor Kaiso, which has been described to act as an oncogene by DNA methylation-dependent silencing of tumor suppressor genes in colon cancer (Lopes et al., 2008). Furthermore, we detect an enrichment of motifs for NF-κB and CEBP, two transcription factors that have been implicated in aggressive breast cancer and EMT (Bundy and Sealy, 2003; Huber et al., 2004). Up-regulated lncRNA promoters are enriched for AR and p63 binding motifs. AR was described to stimulate breast tumor progression but its effect depends on the ER status of the patient (for review, see Chang et al., 2013). The transcription factor p63 is required to initiate collective cell migration and invasion in tumor organoids and in vivo (Cheung et al., 2013). Interestingly, our motif analysis revealed that the same regulators that control known tumor suppressor genes might down-regulate the identified lncRNAs, while up-regulated lncRNAs might be activated by factors that also control known oncogenes.

In addition to computational predictions, we performed ASO-mediated knockdown assays. Of the 290 up-regulated candidates, we prioritized 30 lncRNAs, termed MaTARs, and observed that independent down-regulation of 20 MaTARs leads to a significant reduction of cell proliferation, invasion and/or collective cell migration. Given that the results of individual knockdown experiments were characteristic for the respective non-coding RNA, we propose that each MaTAR might impact different aspects of tumor cell growth and/or metastasis. Interestingly, knockdown of MaTAR16, MaTAR18 and MaTAR26 resulted in reduced branching of tumor organoids, but not normal mammary organoids. Our findings suggest that MaTARs might be dispensable for normal mammary epithelial growth but seem to be essential in the context of cancer.

Based on our extensive analysis in mouse, we propose that the human orthologs of MaTARs are likely impacting human breast cancer progression. In this regard, concerns are often raised about the lack of conservation between mouse and human lncRNAs (Derrien et al., 2012). Unlike protein-coding genes, the functions of lncRNAs are not necessarily directly linked to their nucleotide sequence but are often dependent on RNA secondary structure folding or syntenic transcription (Diederichs, 2014). Synteny between non-coding and protein-coding genes is frequently conserved across species (Ulitsky et al., 2011). Importantly, we identified potential human counterparts for all MaTAR genes and demonstrated that at least 50% of them are over-expressed in breast tumors compared to matched normal controls. The TCGA data analysis confirms that our RNA-seq screen in mouse models identified lncRNAs that act as important regulators of mammary tumor progression and whose functions may be conserved in the human system. Similar to the mouse studies, we revealed that hMaTARs can be expressed in a breast cancer subtype-specific manner or generally up-regulated across all subtypes.

Of the 20 MaTARs that impact cell proliferation, invasion and/or organoid branching, 11 have human orthologs that are up-regulated in breast cancer. We suggest that these lncRNAs are likely drivers of tumor progression and could potentially be of clinical relevance. Future studies will reveal the detailed molecular mechanism by which hMaTARs act as well as their prognostic and therapeutic potential.

Experimental Procedures

Organoid culture

Organoids from wild type nulliparous mammary glands, MMTV-PyMT tumors and MMTV-Neu-NDL tumors were prepared and cultured as described previously (Ewald, 2013). For MMTV-PyMT mice, individual tumors were isolated and organoids were generated in a tumor-specific manner. Mammary epithelial fragments (organoids) were mixed with growth factor-reduced matrigel at a concentration of 5 organoids/μl and plated as 80 μl domes in 24-well dishes. Organoids were grown in DMEM/F12 medium supplemented with 1x ITS (insulin, transferrin, and sodium selenite) media supplement, 1% penicillin/streptomycin and murine FGF2 (2.5 nM).

RNA-Sequencing

Total RNA was isolated either directly from cryosections of the tumor tissue or from organotypic epithelial cultures using TRIzol according to the manufacturer’s instructions. For tissue sections, the tumors were embedded in OCT and cryosectioned. Sections from the middle of the tumor were stained using toluidine blue and assayed regarding the homogeneity of the section. Homogenous 30 μm sections comprising >90% malignant cells were immediately dispersed and homogenized in TRIzol. RNA quality was assayed by running an RNA 6000 Nano chip on a 2100 Bioanalyzer. For high-throughput sequencing, RNA samples were required to have an RNA integrity number (RIN) ≥ 9. TruSeq (Illumina) libraries for polyA+ RNA-Seq were prepared from 0.5 – 1 μg RNA per sample. To ensure efficient cluster generation, an additional gel purification step of the libraries was applied. The libraries were multiplexed (4–6 libraries per lane) and sequenced paired-end 101 bp on the HiSeq2000 platform (Illumina), resulting in an average 40 Mio reads per library.

ASO-mediated knockdown in primary cells

MMTV-PyMT primary cells were seeded at a density of 20,000 cells/well (100 μl per well) into 96-well plates. Transfection-free uptake of ASOs was accomplished by adding 5 μM of either a MaTAR-specific ASO or scrambled ASO (scASO) to the primary cell culture medium immediately after seeding the cells. ASO sequences are provided in Table S8. Cells were incubated for 24 h at 37°C and RNA was isolated using the RNeasy 96 kit ( Qiagen) according to the manufacturer’s instructions. RNA samples were used directly in a one-step 384-well qRT-PCR (QuantiTect SYBR Green RT-PCR Kit, Qiagen) on an Applied Biosystems 7900HT Fast Real-Time PCR System. qRT-PCR conditions were as follows: 30 min at 50°C for reverse transcription, 15 min at 95°C for the initial activation step followed by 40 cycles of 15 sec at 94°C, 30 sec at 60°C. Peptidylprolyl isomerase B (cyclophilin B) was used as an endogenous control to normalize each sample, and relative expression results were calculated using the 2−ΔΔCt method. A list of primers used is provided in Table S9.

ASO-mediated knockdown in organoids

Organoids were seeded at a density of 5 organoids/μl and plated as 80 μl domes in 24-well dishes. Transfection-free uptake of ASOs was accomplished by adding 4 μM of either a MaTAR-specific ASO or scASO to the organoid culture medium 15–20 min after the organoids were plated in matrigel domes. Organoids were incubated for 6 days at 37°C and both medium a nd ASOs were replenished at day 3. ASO sequences are provided in Table S8. Total RNA was isolated using TRIzol. DNase I treatment was performed for 15 min at RT to remove contaminating DNA. cDNA synthesis was carried out using TaqMan Reverse Transcription reagents (Life Technologies) and random hexamers according to the manufacturer’s instructions. Real time quantitative PCR was performed using the Power SYBR Green Master Mix (Life Technologies) in 384-well plates using the Applied Biosystems 7900HT Fast Real-Time PCR System (Thermo Fisher Scientific). Cycling conditions were as follows: 10 min at 95°C followed by 40 cycles of 15 sec at 95°C, 1 min at 60°C. Primer spe cificity was initially tested by agarose gel electrophoresis and subsequently monitored by melting curve analysis. Peptidylprolyl isomerase B (cyclophilin B) was used as an endogenous control to normalize each sample and relative expression results were calculated using the 2−ΔΔCt method. A list of primers used is provided in Table S9. For visualization purposes and quantification of organoid branching, images were acquired using an Axio-Observer Live Cell inverted microscope (Zeiss) at 10x magnification.

TCGA data analysis

Analysis of TCGA data (Cancer Genome Atlas Network, 2012) was performed on the Cancer Genomics Cloud hosted by Seven Bridges Genomics (https://www.sbgenomics.com). For differential gene expression comparing 103 breast tumor samples to matched normal controls, raw RNA-seq reads were mapped to hg38 and counted as described above. The GENCODE v24 GTF was used as a reference. Fold changes calculated for individual patients were averaged and statistical significance was calculated using Wilcoxon rank tests; p-values were adjusted using the Benjamini & Hochberg method. Subtype specific analysis, survival and pathway analysis are described in detail in the Supplemental Experimental Procedures.

Supplementary Material

supplement

Acknowledgments

We would like to thank all members of the Spector Lab for helpful discussions and suggestions throughout the course of this study; especially Allen Yu for generously sharing his pSpCas9(BB)-2A-mCherry vector and Gayatri Arun for critically reviewing the manuscript. We acknowledge Mikala Egeblad, Senthil Muthuswamy (University of Toronto) and William J. Muller (McGill University) for providing mice and/or tumors. We acknowledge the CSHL Animal Shared Resource, Animal and Tissue Imaging Shared Resource, DNA Sequencing Shared Resource, and Microscopy Shared Resource (NCI 2P3OCA45508). This project was funded by NCI 5P01CA013106-Project 3 (D.L.S.), the Manhasset Women’s Coalition Against Breast Cancer (S.D.D.), the Simons Foundation (J.S and A.K.), NCI 5U01CA168409-03 (J.S and A.K.) and NCI 1U01CA188590-01A1 (A.K.).

Footnotes

Accession Numbers

The RNA-Seq data presented in this paper are publicly available at the NCBI Gene Expression Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo) under accession number GSE72823.

Author Contributions

S.D.D. and D.L.S. conceived the study, designed the experiments and wrote the manuscript. S.D.D. performed the computational analyses of all mouse data. S.D.D. and K-C.C. performed experiments and interpreted the data. S.D.D., J.S., O.E. and A.K. analyzed TCGA data. S.M.F., F.R. and C.F.B. designed andprovided ASOs.

Competing Financial Interests

D.L.S. receives research support from, and is a consultant to, Ionis Pharmaceuticals.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Andrechek ER, Hardy WR, Siegel PM, Rudnicki MA, Cardiff RD, Muller WJ. Amplification of the neu/erbB-2 oncogene in a mouse model of mammary tumorigenesis. Proc Natl Acad Sci USa. 2000;97:3444–3449. doi: 10.1073/pnas.050408497. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Arase M, Horiguchi K, Ehata S, Morikawa M, Tsutsumi S, Aburatani H, Miyazono K, Koinuma D. Transforming growth factor-β-induced lncRNA-Smad7 inhibits apoptosis of mouse breast cancer JygMC(A) cells. Cancer Sci. 2014;105:974–982. doi: 10.1111/cas.12454. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Arun G, Diermeier S, Akerman M, Chang KC, Wilkinson JE, Hearn S, Kim Y, MacLeod AR, Krainer AR, Norton L, et al. Differentiation of mammary tumors and reduction in metastasis upon Malat1lncRNA loss. Genes & Development. 2016;30:34–51. doi: 10.1101/gad.270959.115. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Barcellos-Hoff MH, Aggeler J, Ram TG, Bissell MJ. Functional differentiation and alveolar morphogenesis of primary mammary cultures on reconstituted basement membrane. Development. 1989;105:223–235. doi: 10.1242/dev.105.2.223. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Bergmann JH, Spector DL. Long non-coding RNAs: modulators of nuclear structure and function. Curr Opin Cell Biol. 2014;26:10–18. doi: 10.1016/j.ceb.2013.08.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bergmann JH, Li J, Eckersley-Maslin MA, Rigo F, Freier SM, Spector DL. Regulation of the ESC transcriptome by nuclear long noncoding RNAs. Genome Res. 2015;25:1336–1346. doi: 10.1101/gr.189027.114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bertone P, Stolc V, Royce TE, Rozowsky JS, Urban AE, Zhu X, Rinn JL, Tongprasit W, Samanta M, Weissman S, et al. Global identification of human transcribed sequences with genome tiling arrays. Science. 2004;306:2242–2246. doi: 10.1126/science.1103388. [DOI] [PubMed] [Google Scholar]
  8. Boj SF, Hwang CI, Baker LA, Chio IIC, Engle DD, Corbo V, Jager M, Ponz-Sarvise M, Tiriac H, Spector MS, et al. Organoid Models of Human and Mouse Ductal Pancreatic Cancer. Cell. 2015;160:324–338. doi: 10.1016/j.cell.2014.12.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Bundy LM, Sealy L. CCAAT/enhancer binding protein beta (C/EBPβ)-2 transforms normal mammary epithelial cells and induces epithelial to mesenchymal transition in culture. Oncogene. 2003;22:869–883. doi: 10.1038/sj.onc.1206216. [DOI] [PubMed] [Google Scholar]
  10. Cabili MN, Trapnell C, Goff L, Koziol M, Tazon-Vega B, Regev A, Rinn JL. Integrative annotation of human large intergenic noncoding RNAs reveals global properties and specific subclasses. Genes & Development. 2011;25:1915–1927. doi: 10.1101/gad.17446611. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Cancer Genome Atlas Network. Comprehensive molecular portraits of human breast tumours. Nature. 2012;490:61–70. doi: 10.1038/nature11412. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Carninci P, Kasukawa T, Katayama S, Gough J. The transcriptional landscape of the mammalian genome. Science. 2005;309:1559–1563. doi: 10.1126/science.1112014. [DOI] [PubMed] [Google Scholar]
  13. Chang C, Lee SO, Yeh S, Chang TM. Androgen receptor (AR) differential roles in hormone-related tumors including prostate, bladder, kidney, lung, breast and liver. 2013;33:3225–3234. doi: 10.1038/onc.2013.274. [DOI] [PubMed] [Google Scholar]
  14. Cheetham SW, Gruhl F, Mattick JS, Dinger ME. Long noncoding RNAs and the genetics of cancer. British Journal of Cancer. 2013;108:2419–2425. doi: 10.1038/bjc.2013.233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Cheung KJ, Gabrielson E, Werb Z, Ewald AJ. Collective Invasion in Breast Cancer Requires a ConservedBasal Epithelial Program. Cell. 2013;155:1639–1651. doi: 10.1016/j.cell.2013.11.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Chou SD, Murshid A, Eguchi T, Gong J, Calderwood SK. HSF1 regulation of β-catenin in mammary cancer cells through control of HuR/elavL1 expression. Oncogene. 2015;34:2178–2188. doi: 10.1038/onc.2014.177. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Clevers H. Modeling Development and Disease with Organoids. Cell. 2016;165:1586–1597. doi: 10.1016/j.cell.2016.05.082. [DOI] [PubMed] [Google Scholar]
  18. Derrien T, Johnson R, Bussotti G, Tanzer A, Djebali S, Tilgner H, Guernec G, Martin D, Merkel A, Knowles DG, et al. The GENCODE v7 catalog of human long noncoding RNAs: Analysis of their gene structure, evolution, and expression. Genome Res. 2012;22:1775–1789. doi: 10.1101/gr.132159.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Diederichs S. The four dimensions of noncoding RNA conservation. Trends in Genetics. 2014;30:121–123. doi: 10.1016/j.tig.2014.01.004. [DOI] [PubMed] [Google Scholar]
  20. Dimitrova N, Zamudio JR, Jong RM, Soukup D, Resnick R, Sarma K, Ward AJ, Raj A, Lee JT, Sharp PA, et al. LincRNA-p21 activates p21 in cis to promote Polycomb target gene expression and to enforce the G1/S checkpoint. Molecular Cell. 2014;54:777–790. doi: 10.1016/j.molcel.2014.04.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Djebali S, Davis CA, Merkel A, Dobin A, Lassmann T, Mortazavi A, Tanzer A, Lagarde J, Lin W, Schlesinger F, et al. Landscape of transcription in human cells. Nature. 2012;488:101–108. doi: 10.1038/nature11233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Domcke S, Sinha R, Levine DA, Sander C, Schultz N. Evaluating cell lines as tumour models by comparison of genomic profiles. Nature Communications. 2013;4:2126. doi: 10.1038/ncomms3126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489:57–74. doi: 10.1038/nature11247. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Ewald AJ. Isolation of mouse mammary organoids for long-term time-lapse imaging. Cold Spring Harb Protoc. 2013;2013:130–133. doi: 10.1101/pdb.prot072892. [DOI] [PubMed] [Google Scholar]
  25. Ewald AJ, Brenot A, Duong M, Chan BS, Werb Z. Collective Epithelial Migration and Cell Rearrangements Drive Mammary Branching Morphogenesis. Developmental Cell. 2008;14:570–581. doi: 10.1016/j.devcel.2008.03.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Fata JE, Mori H, Ewald AJ, Zhang H, Yao E, Werb Z, Bissell MJ. The MAPKERK-1,2 pathway integrates distinct and antagonistic signals from TGFα and FGF7 in morphogenesis of mouse mammary epithelium. Developmental Biology. 2007;306:193–207. doi: 10.1016/j.ydbio.2007.03.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Geary RS, Norris D, Yu R, Bennett CF. Pharmacokinetics, biodistribution and cell uptake of antisense oligonucleotides. Adv Drug Deliv Rev. 2015;87:46–51. doi: 10.1016/j.addr.2015.01.008. [DOI] [PubMed] [Google Scholar]
  28. Gupta RA, Shah N, Wang KC, Kim J, Horlings HM, Wong DJ, Tsai MC, Hung T, Argani P, Rinn JL, et al. Long non-coding RNA HOTAIR reprograms chromatin state to promote cancer metastasis. Nature. 2010;464:1071–1076. doi: 10.1038/nature08975. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Guy CT, Cardiff RD, Muller WJ. Induction of mammary tumors by expression of polyomavirus middle T oncogene: a transgenic mouse model for metastatic disease. Molecular and Cellular Biology. 1992;12:954–961. doi: 10.1128/mcb.12.3.954. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. Hansji H, Leung EY, Baguley BC, Finlay GJ, Askarian-Amiri ME. Keeping abreast with long non-coding RNAs in mammary gland development and breast cancer. Front Genet. 2014;5:379. doi: 10.3389/fgene.2014.00379. [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Huarte M. The emerging role of lncRNAs in cancer. Nature Medicine. 2015;21:1253–1261. doi: 10.1038/nm.3981. [DOI] [PubMed] [Google Scholar]
  32. Huarte M, Guttman M, Feldser D, Garber M, Koziol MJ, Kenzelmann-Broz D, Khalil AM, Zuk O, Amit I, Rabani M, et al. A large intergenic noncoding RNA induced by p53 mediates global gene repression in the p53 response. Cell. 2010;142:409–419. doi: 10.1016/j.cell.2010.06.040. [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Huber MA, Azoitei N, Baumann B, Grünert S, Sommer A, Pehamberger H, Kraut N, Beug H, Wirth T. NF-κB is essential for epithelial-mesenchymal transition and metastasis in a model of breast cancer progression. J Clin Invest. 2004;114:569–581. doi: 10.1172/JCI21358. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Iyer MK, Niknafs YS, Malik R, Singhal U, Sahu A, Hosono Y, Barrette TR, Prensner JR, Evans JR, Zhao S, et al. The landscape of long noncoding RNAs in the human transcriptome. Nat Genet. 2015;47:199–208. doi: 10.1038/ng.3192. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Kapranov P, Cheng J, Dike S, Nix DA, Duttagupta R, Willingham AT, Stadler PF, Hertel J, Hackermüller J, Hofacker IL, et al. RNA maps reveal new RNA classes and a possible function for pervasive transcription. Science. 2007;316:1484–1488. doi: 10.1126/science.1138341. [DOI] [PubMed] [Google Scholar]
  36. Katayama S, Tomaru Y, Kasukawa T, Waki K, Nakanishi M, Nakamura M, Nishida H, Yap CC, Suzuki M, Kawai J, et al. Antisense transcription in the mammalian transcriptome. Science. 2005;309:1564–1566. doi: 10.1126/science.1112009. [DOI] [PubMed] [Google Scholar]
  37. Koller E, Vincent TM, Chappell A, De S, Manoharan M, Bennett CF. Mechanisms of single-stranded phosphorothioate modified antisense oligonucleotide accumulation in hepatocytes. Nucleic Acids Res. 2011;39:4795–4807. doi: 10.1093/nar/gkr089. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Koller E, Propp S, Zhang H, Zhao C, Xiao X, Chang M, Hirsch SA, Shepard PJ, Koo S, Murphy C, et al. Use of a chemically modified antisense oligonucleotide library to identify and validate Eg5 (kinesin-like 1) as a target for antineoplastic drug development. Cancer Research. 2006;66:2059–2066. doi: 10.1158/0008-5472.CAN-05-1531. [DOI] [PubMed] [Google Scholar]
  39. Kornienko AE, Guenzl PM, Barlow DP, Pauler FM. Gene regulation by the act of long non-coding RNA transcription. BMC Biol. 2013;11:59. doi: 10.1186/1741-7007-11-59. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Lancaster MA, Knoblich JA. Organogenesis in a dish: Modeling development and disease using organoid technologies. Science. 2014;345:1247125–1247125. doi: 10.1126/science.1247125. [DOI] [PubMed] [Google Scholar]
  41. Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network analysis. BMC Bioinformatics. 2008;9:559. doi: 10.1186/1471-2105-9-559. [DOI] [PMC free article] [PubMed] [Google Scholar]
  42. Lopes EC, Valls E, Figueroa ME, Mazur A, Meng FG, Chiosis G, Laird PW, Schreiber-Agus N, Greally JM, Prokhortchouk E, et al. Kaiso Contributes to DNA Methylation-Dependent Silencing of Tumor Suppressor Genes in Colon Cancer Cell Lines. Cancer Research. 2008;68:7258–7263. doi: 10.1158/0008-5472.CAN-08-0344. [DOI] [PubMed] [Google Scholar]
  43. Nguyen-Ngoc KV, Cheung KJ, Brenot A, Shamir ER, Gray RS, Hines WC, Yaswen P, Werb Z, Ewald AJ. ECM microenvironment regulates collective migration and local dissemination in normal and malignant mammary epithelium. Proc Natl Acad Sci USA. 2012;109:E2595–E2604. doi: 10.1073/pnas.1212834109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Okazaki Y, Furuno M, Kasukawa T, Adachi J, Bono H. Analysis of the mouse transcriptome based on functional annotation of 60,770 full-length cDNAs. Nature. 2002;420:563–573. doi: 10.1038/nature01266. [DOI] [PubMed] [Google Scholar]
  45. Parker JS, Mullins M, Cheang MCU, Leung S, Voduc D, Vickery T, Davies S, Fauron C, He X, Hu Z, et al. Supervised Risk Predictor of Breast Cancer Based on Intrinsic Subtypes. Journal of Clinical Oncology. 2009;27:1160–1167. doi: 10.1200/JCO.2008.18.1370. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Prensner JR, Chinnaiyan AM. The Emergence of lncRNAs in Cancer Biology. Cancer Discovery. 2011;1:391–407. doi: 10.1158/2159-8290.CD-11-0209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  47. Ran FA, Hsu PD, Wright J, Agarwala V, Scott DA, Zhang F. Genome engineering using the CRISPR-Cas9 system. Nat Protoc. 2013;8:2281–2308. doi: 10.1038/nprot.2013.143. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Rinn JL, Chang HY. Genome Regulation by Long Noncoding RNAs. Annu Rev Biochem. 2012;81:145–166. doi: 10.1146/annurev-biochem-051410-092902. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Rinn JL, Kertesz M, Wang JK, Squazzo SL, Xu X, Brugmann SA, Goodnough LH, Helms JA, Farnham PJ, Segal E, et al. Functional demarcation of active and silent chromatin domains in human HOX loci by noncoding RNAs. Cell. 2007;129:1311–1323. doi: 10.1016/j.cell.2007.05.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Sachs N, Clevers H. Organoid cultures for the analysis of cancer phenotypes. Curr Opin Genet Dev. 2014;24:68–73. doi: 10.1016/j.gde.2013.11.012. [DOI] [PubMed] [Google Scholar]
  51. Shamir ER, Ewald AJ. Three-dimensional organotypic culture: experimental models of mammalian biology and disease. Nature Reviews Molecular Cell Biology. 2014;15:647–664. doi: 10.1038/nrm3873. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Shore AN, Herschkowitz JI, Rosen JM. Noncoding RNAs involved in mammary gland development and tumorigenesis: there’s a long way to go. J Mammary Gland Biol Neoplasia. 2012;17:43–58. doi: 10.1007/s10911-012-9247-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Siegel PM, Ryan ED, Cardiff RD. Elevated expression of activated forms of Neu/ErbB-2 and ErbB-3 are involved in the induction of mammary tumors in transgenic mice: implications for human breast cancer. The EMBO Journal. 1999;18:2149–2164. doi: 10.1093/emboj/18.8.2149. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. St Laurent G, Shtokalo D, Tackett MR, Yang Z, Eremina T, Wahlestedt C, Urcuqui-Inchima S, Seilheimer B, McCaffrey TA, Kapranov P. Intronic RNAs constitute the major fraction of the non-coding RNA in mammalian cells. BMC Genomics. 2012;13:504. doi: 10.1186/1471-2164-13-504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  55. Sun M, Gadad SS, Kim DS, Kraus WL. Discovery, Annotation, and Functional Analysis of Long Noncoding RNAs Controlling Cell-Cycle Gene Expression and Proliferation in Breast Cancer Cells. Molecular Cell. 2015;59:698–711. doi: 10.1016/j.molcel.2015.06.023. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Tomita S, Abdalla MOA, Fujiwara S, Matsumori H, Maehara K, Ohkawa Y, Iwase H, Saitoh N, Nakao M. A cluster of noncoding RNAs activates the ESR1 locus during breast cancer adaptation. Nature Communications. 2015;6:6966. doi: 10.1038/ncomms7966. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Ulitsky I, Shkumatava A, Jan CH, Sive H, Bartel DP. Conserved Function of lincRNAs in Vertebrate Embryonic Development despite Rapid Sequence Evolution. Cell. 2011;147:1537–1550. doi: 10.1016/j.cell.2011.11.055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. van Staveren WCG, Solís DYW, Hébrant A, Detours V, Dumont JE, Maenhaut C. Human cancer cell lines: Experimental models for cancer cells in situ? For cancer stem cells? Biochim Biophys Acta. 2009;1795:92–103. doi: 10.1016/j.bbcan.2008.12.004. [DOI] [PubMed] [Google Scholar]
  59. Vikram R, Ramachandran R, Abdul KSM. Functional significance of long non-coding RNAs in breast cancer. Breast Cancer. 2014;21:515–521. doi: 10.1007/s12282-014-0554-y. [DOI] [PubMed] [Google Scholar]
  60. Wahlestedt C. Targeting long non-coding RNA to therapeutically upregulate gene expression. Nat Rev Drug Discov. 2013;12:433–446. doi: 10.1038/nrd4018. [DOI] [PubMed] [Google Scholar]
  61. Wang L, Park HJ, Dasari S, Wang S, Kocher JP, Li W. CPAT: Coding-Potential Assessment Tool using an alignment-free logistic regression model. Nucleic Acids Res. 2013;41:e74–e74. doi: 10.1093/nar/gkt006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  62. Wapinski O, Chang HY. Long noncoding RNAs and human disease. Trends in Cell Biology. 2011;21:354–361. doi: 10.1016/j.tcb.2011.04.001. [DOI] [PubMed] [Google Scholar]
  63. Xing Z, Lin A, Li C, Liang K, Wang S, Liu Y, Park PK, Qin L, Wei Y, Hawke DH, et al. lncRNA directs cooperative epigenetic regulation downstream of chemokine signals. Cell. 2014;159:1110–1125. doi: 10.1016/j.cell.2014.10.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Yan X, Hu Z, Feng Y, Hu X, Yuan J, Zhao SD, Zhang Y, Yang L, Shan W, He Q, et al. Comprehensive Genomic Characterization of Long Non-coding RNAs across Human Cancers. Cancer Cell. 2015;28:529–540. doi: 10.1016/j.ccell.2015.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Yang F, Zhang H, Mei Y, Wu M. Reciprocal regulation of HIF-1α and lincRNA-p21 modulates the Warburg effect. Molecular Cell. 2014;53:88–100. doi: 10.1016/j.molcel.2013.11.004. [DOI] [PubMed] [Google Scholar]
  66. Zhang X, Lian Z, Padden C, Gerstein MB, Rozowsky J, Snyder M, Gingeras TR, Kapranov P, Weissman SM, Newburger PE. A myelopoiesis-associated regulatory intergenic noncoding RNA transcript within the human HOXA cluster. Blood. 2009;113:2526–2534. doi: 10.1182/blood-2008-06-162164. [DOI] [PMC free article] [PubMed] [Google Scholar]
  67. Zhang X, Weissman SM, Newburger PE. Long intergenic non-coding RNA HOTAIRM1 regulates cell cycle progression during myeloid maturation in NB4 human promyelocytic leukemia cells. RNA Biology. 2014;11:777–787. doi: 10.4161/rna.28828. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

supplement
NIHMS814281-supplement.pdf (6.8MB, pdf)

RESOURCES