HIF2-Induced Long Noncoding RNA RAB11B-AS1 Promotes Hypoxia-Mediated Angiogenesis and Breast Cancer Metastasis

Metastasis is responsible for about 90% of deaths due to breast cancer and targeting breast cancer metastasis remains a challenge. Long noncoding RNAs (lncRNA) are a group of RNA transcripts longer than 200 nucleotides and have gained increasing attention because of their emerging role in gene regulation, which contributes to a wide range of cellular functions, including development, differentiation, cell fate, and disease pathogenesis (1–4). About 15,000 lncRNA transcripts have been identified in human genome by the GENCODE consortium in the ENCODE project (5). Several lncRNAs have been shown to play a critical role in breast cancer metastasis. For example, HOTAIR is highly expressed in metastatic breast cancer and promotes breast cancer metastasis by regulating polycomb repressive complex 2-dependent gene expression (6). TreRNA promotes breast cancer metastasis by suppressing the translation of E-Cadherin mRNA (7). LincRNA-ROR antagonizes miR-145 to activate the small GTPase ADP-ribosylation factor 6, which downregulates E-cadherin and disrupts cell–cell junctions, leading to invasion of triple-negative breast cancer cells (8). BCAR4 activates the chemokine CCL21-induced Hedgehog pathway and breast cancer metastasis by regulating the SNIP1-PNUTS-p300-RNA polymerase (Pol) II axis (9). Given a critical role of lncRNAs in metastatic breast cancer, it is of great importance to discover new lncRNAs and their role in the pathologic processes of breast cancer metastasis.

Hypoxia is a hallmark of the microenvironment in breast tumors and promotes tumor progression and metastasis (10). The transcription factor hypoxia-inducible factor (HIF) is a key regulator of response to hypoxic stress in metazoans (11). HIF is a heterodimer of an oxygen-regulated α subunit and a stably expressed β subunit (12). In mammals, the α subunit is encoded by three distinct genes: HIF1A, EPAS1, and HIF3A (12–14), whereas the β subunit is encoded by ARNT and ARNT2 genes (12, 15). Under normoxia, HIFα protein is hydroxylated on the proline residues by prolyl hydroxylase 2 and then rapidly degraded at the 26S proteasome by the ubiquitin E3 ligase von Hippel-Lindau/Cullin-2/Elongin-B/C (16). Upon hypoxia, HIFα protein is stabilized and translocated into the nucleus, where it dimerizes with HIFβ, allowing the heterodimer to bind the hypoxia response element (HRE) and then to activate the transcription of the downstream target genes (17). Extensive studies have discovered hundreds of HIF-induced protein-coding genes in cancer cells, whose protein products regulate the important cancer biology, including cancer cell stemness, autophagy, angiogenesis, metabolism, immune evasion, and metastasis (11). Although a recent genome-wide transcriptome study showed that hypoxia induces a vast array of lncRNAs in breast cancer MCF-7 cells (18), little is known about the function of hypoxia-induced lncRNAs in breast cancer.

In this study, we reported that the lncRNA RAB11B-AS1 is transcriptionally induced by hypoxia in breast cancer cells in a HIF2-dependent manner. RAB11B-AS1 enhances hypoxia-induced VEGFA and ANGPTL4 expression in breast cancer cells through RNA Pol II and promotes angiogenesis and distant breast cancer metastasis in mice. These findings uncover a novel mechanism of hypoxia-induced angiogenesis and breast cancer metastasis.

The culture conditions of MDA-MB-231 (from Rolf Brekken at UT Southwestern in 2015), MDA-MB-468 (from ATCC in 2017), SUM159, MCF-7, T47D, BT474, ZR-75-1, HCC1954, HeLa, HEK293FT, and HEK293T (from Gregg L. Semenza at Johns Hopkins in 2014) cells were described previously (19). Cells within six passages were used for experiments. Human umbilical vein endothelial cells (HUVEC) were cultured in M200PRF medium with low serum growth supplement (Thermo Fisher Scientific) at 37°C in a 5% CO₂/95% air incubator. HUVECs within the first three passages were used for experiments. Hypoxic cells were placed in a modular incubator chamber (Billups-Rothenberg) and flushed with a gas mixture of 1% O₂, 5% CO₂, and balanced N₂. Cells were transfected using PolyJet (SignaGen). Lentiviruses encoding scrambled control (SC) or RAB11B-AS1 short hairpin RNAs (shRNA) were generated as described previously (19). All cell lines except for HUVECs were annually tested to be Mycoplasma-free and have been authenticated by STR DNA profiling analysis in 2016-2017.

Total RNA was isolated from the tissues or cultured cells using TRizol (Thermo Fisher Scientific), treated with DNase I (Thermo Fisher Scientific) and then subject to cDNA synthesis with the iScript cDNA Synthesis Kit (Bio-Rad). qPCR was performed with iTaq universal SYBR green supermix (Bio-Rad) and normalized to the internal control 18S RNA or RPL13A as described previously (19). The primers used for qPCR are listed in Supplementary Table S1. RNA-sequencing (RNA-seq) was performed as described previously (19). The deidentified human breast tumors and benign breast tissues were used and their pathology was verified by a breast cancer pathologist (Y.V. Fang). The study was approved by the Institutional Review Board at UT Southwestern Medical Center with informed written consent.

5′ and 3′ rapid amplification of cDNA end (5′- and 3′-RACE) analyses were performed using the SMARTer RACE 5′/3′ Kit (Takara Bio). Total RNA was extracted using the RNeasy Micro Kit (Qiagen) and treated with DNase I. RNA integrity was assessed with the Agilent 4200 TapeStation system. After reverse transcription of total RNA to cDNA, 5′- and 3′-RACE PCRs were performed with the universal and gene-specific primers listed in Supplementary Table S2 and the resulting PCR products were cloned into linearized pRACE vector and sequenced. The nucleotide sequence of RAB11B-AS1 was deposited in the GenBank (MK855053).

Full-length human RAB11B-AS1 cDNA was cloned into pcDNA3.1 or pBluescript II KS (+). RAB11B-AS1 shRNA nucleotides (Supplementary Table S2) were cloned into pLKO.1 vector. HIF1β single gRNA (sgRNA) nucleotides (Supplementary Table S2) were cloned into lentiCRISPRv2 vector. Wild-type (WT) or mutant RAB11B-AS1 HRE nucleotides (Supplementary Table S2) were cloned into MluI/BglII-linearized pGL2p vector (Promega). All plasmids were verified by DNA sanger sequencing. Other plasmid constructs have been described previously (19).

A total of 5 × 10⁴ of HEK293T or HeLa cells were seeded onto 48-well plates precoated with poly-l-lysine a day prior to transfection. Cells were cotransfected with pGL2p-WT RAB11B-AS1 HRE, pGL2p-mutant RAB11B-AS1 HRE, or pGL2p-VEGFA HRE; pSV40-Renilla; and p3XFLAG-HIF2α, pcDNA3.1-RAB11B-AS1, or empty vector (EV) and exposed to 20% or 1% O₂ for 24 hours. Firefly and Renilla luciferase activities were measured using the Dual Luciferase Reporter Assay System (Promega).

Cells were lysed with NETN lysis buffer (150 mmol/L NaCl, 1 mmol/L EDTA, 50 mmol/L Tris-HCl, pH8.0, 0.5% NP-40, and fresh protease inhibitor) and protein samples were separated by SDS-PAGE, transferred to nitrocellulose membrane, and visualized by chemiluminescence with ECL prime (GE Healthcare) as described previously (19). The following antibodies were used: anti-HIF1α (BD Bioscience, catalog no. 610959), anti-HIF2α (Bethyl Laboratories, catalog no. A700-003), anti-HIF1β (Novus Biologicals, catalog no. 100-110), RNA polymerase II (Abcam, catalog no. ab817), and anti-β-actin (Proteintech, catalog no. 66009-1-lg).

Cells were cross-linked with 1% formaldehyde for 10 minutes, quenched in 0.125 mol/L glycine, and lysed with lysis buffer, followed by sonication. Then the lysates were incubated overnight at 4°C with anti-RNA Pol II antibody (Abcam, catalog no. ab817), anti-HIF2α antibody (homemade), anti-HIF1β (Novus Biologicals, catalog no. 100-110), or IgG (Cell Signaling Technology, catalog no. 2729S). After decrosslinking of the protein–DNA complex and DNA purification, precipitated DNA was subject to library preparation for sequencing or qPCR with primers listed in Supplementary Table S1. Bioinformatics analysis of chromatin-immunoprecipitation (ChIP)-sequencing (ChIP-seq) data was performed as described previously (19). The RNA-seq and ChIP-seq data were deposited at the GEO database (GSE139861).

The biotin-labeled sense and antisense of RAB11B-AS1 RNA were synthesized in vitro by T7 RNA polymerase (Thermo Fisher Scientific) using pBluescript-RAB11B-AS1 sense and antisense as a respective template and the biotin RNA labeling mix (Sigma). The RNAs were denatured, refolded in RNA structure buffer (10 mmol/L Tris-HCl, pH 7.0, 0.1 mol/L KCl, and 10 mmol/L MgCl₂), and incubated for 2 hours at 4°C in the presence of 0.1 μg/μL tRNA, 0.025% BSA, and the precleared lysates, which were extracted from MDA-MB-231 cells exposed to 20% or 1% O₂ for 24 hours in RNA immunoprecipitation buffer (150 mmol/L KCl, 25 mmol/L Tris-HCl, pH 7.4, 5 mmol/L EDTA, 0.5% NP-40, 0.5 mmol/L DTT, fresh protease inhibitor, and RNase inhibitor). 0.25% BSA-prebound streptavidin magnetic beads (GE Healthcare) were added for additional 1-hour incubation at 4°C. The proteins bound to biotin-labeled RAB11B-AS1 RNA were separated by SDS-PAGE and analyzed by immunoblot assay.

Cell migration and invasion assays were performed as described previously (19). A total of 2 × 10⁴ of cells were suspended in serum-free media and seeded onto the noncoated (migration) or Matrigel-coated (invasion) upper chamber. 10% FBS in the bottom chamber was used as a chemoattractant. Cells were exposed to 20% or 1% O₂ for 16 hours (migration) or 24 hours (invasion), fixed with methanol, and stained with 0.1% crystal violet. Colonies were imaged using microscope and counted in five randomly selected fields at ×20 magnification.

A total of 4 × 10⁴ of HUVECs were starved in M200PRF medium without low serum growth supplement for 2 hours, resuspended in fresh conditional media with 1% FBS, and seeded onto a 96-well plate coated with growth factor–reduced Matrigel (BD Biosciences). Conditional media were collected by incubating MDA-MB-231-EV or -RAB11B-AS1 cells with serum-free DMEM for 24 hours. HUVECs were imaged using the IncuCyte S3 live-cell analysis system (Sartorius) for 5 hours and the total tube length from five randomly selected fields at ×10 magnification was measured by the Wimasis image analysis.

Six- to eight-week-old female NOD/SCID IL2Rγnull (NSG) mice were used for animal studies, which has been approved and conducted under the oversight of the UT Southwestern Institutional Animal Care and Use Committee. For the orthotopic xenograft model, 2 × 10⁶ of breast cancer cells suspended in PBS/Matrigel (1:1, Corning) were implanted into the second left mammary fat pad of female NSG mice. Tumor growth was monitored with a caliper every 3 days starting 14 days after implantation and calculated with the formula: volume = 0.52 × length × height × width. Mice were euthanized 32 to 40 days after cell implantation and primary tumors were harvested. After perfusion with PBS, the whole lungs were inflated with low-melt agarose and analyzed by hematoxylin and eosin (H&E) staining. The whole liver was also collected for H&E staining. Genomic DNA was extracted from the whole lungs and liver and analyzed by qPCR assay with primers for human HK2 gene and mouse and human 18S rRNA. For the tail vein injection model, 1 × 10⁶ of breast cancer cells suspended in 100 μL PBS were injected into the tail vein of female NSG mice. Three weeks later, the lungs were harvested for H&E staining and qPCR analysis with primers for human HK2 gene and mouse and human 18S rRNA. The area of metastatic foci was quantified by ImageJ and normalized to the total area of lung or liver tissues.

IHC assay was performed by the Dako Autostainer Link 48 system (Dako) as described previously (19). The tissue slides underwent deparaffinication, rehydration, and epitope retrieval and were then incubated with the primary antibody: Endomucin (Santa Cruz Biotechnology, 1:50), cleaved caspase-3 (CC3, Cell Signaling Technology, 1:1,500), and Ki-67 (Proteintech, 1:1,000). Staining was visualized using the EnVision FLEX visualization system (Dako). The endomucin-, Ki-67-, and CC3-positive areas were quantified by ImageJ and normalized to their respective total area.

Statistical analysis was performed by Student t test with or without Welch's correction between two groups, and one-way or two-way ANOVA with multiple testing correction within multiple groups. RNA-seq and ChIP-seq were performed in duplicate and other experiments were repeated at least three times. Data were expressed as mean ± SEM. P < 0.05 is considered significant.

Our RNA-seq studies (GSE108833) showed that RAB11B-AS1 was a hypoxia-inducible lncRNA in human breast cancer MDA-MB-231 cells. To validate, we performed RT-qPCR in a panel of human breast cancer cell lines, including MDA-MB-231, MDA-MB-468, SUM159, MCF-7, T47D, BT474, ZR-75-1, and HCC1954, exposed to 20% or 1% O₂ for 24 hours. RAB11B-AS1 RNA levels were significantly upregulated by hypoxia in all breast cancer cell lines we tested (Fig. 1A). RAB11B-AS1 is an antisense RNA of the RAB11B gene. To annotate the natural transcript of RAB11B-AS1 in breast cancer cells, we performed 5′- and 3′-RACE and detected a specific 1037-nucleotide RAB11B-AS1 transcript in MDA-MB-231 cells (Fig. 1B; Supplementary Fig. S1). The RAB11B-AS1 transcript had 10 more A nucleotides in the poly(A) tail at its 3′ end but missed 7 nucleotides at the 5′ end as compared with the reference transcript deposited in the GenBank (NR_038237.1; Supplementary Fig. S1). In addition, there were two nucleotide mutations in the exons 2&3 (Supplementary Fig. S1). Together, these data indicate that RAB11B-AS1 is a natural lncRNA and induced by hypoxia in breast cancer cells.

To determine whether HIF is required for hypoxia-induced RAB11B-AS1 in breast cancer cells, we performed RT-qPCR assay in parental, HIF1α knockout (KO), HIF2α KO, HIF1α, and HIF2α double knockout (DKO) MDA-MB-231 cells exposed to 20% or 1% O₂ for 24 hours. HIF2α KO, but not HIF1α KO, eliminated hypoxia-induced RAB11B-AS1 expression in MDA-MB-231 cells (Fig. 1C). Similar results were also found in HIF1α and HIF2α DKO cells under hypoxia (Fig. 1C). KO of HIF1α or HIF2α in MDA-MB-231 cells were confirmed by determining the expression of HIF1α and HIF2α (Fig. 1D). Likewise, KO of HIF1β abolished hypoxia-induced RAB11B-AS1 expression in MDA-MB-231 cells (Fig. 1E and F). These data indicate that HIF2 is responsible for hypoxia-induced RAB11B-AS1 expression in breast cancer cells.

To unbiasedly identify the HRE on the RAB11B-AS1 gene, we performed ChIP-seq assay in hypoxic MDA-MB-231 cells and found a robust HIF2α peak annotated with the HRE sequence 5′-ACGTG-3′ at the promoter of the RAB11B-AS1 gene (Fig. 2A). To validate, we performed ChIP-qPCR in MDA-MB-231 cells exposed to 20% or 1% O₂ for 24 hours. HIF2α was enriched at this HRE, but not on the nontarget gene RPL13A (Fig. 2B and C). Hypoxia significantly increased HIF2α occupancy at the RAB11B-AS1 HRE (Fig. 2B). Similarly, HIF1β occupancy was detected at the RAB11B-AS1 HRE in hypoxic MDA-MB-231 cells (Fig. 2D). To examine whether this HRE is functional, a 60-base pair DNA fragment containing the HRE was cloned into the upstream of the SV40 promoter and firefly luciferase gene in a pGL2p reporter vector. HEK293T cells were transfected with pGL2p-EV or pGL2p-RAB11B-AS1 HRE, and the control plasmid pSV40-Renilla, and exposed to 20% or 1% O₂ for 24 hours. The RAB11B-AS1 HRE significantly enhanced firefly luciferase activity in HEK293T cells under hypoxia as compared with the EV, but failed to do so in nonhypoxic cells (Fig. 2E), where no HIF2α protein is present due to its proteasomal degradation, indicating the specificity of the RAB11B-AS1 HRE in response to hypoxia. Mutation of 5′-CGT-3′ into 5′-AAA-3′ within the RAB11B-AS1 HRE completely abolished hypoxia-induced firefly luciferase activity in HEK293T cells (Fig. 2E). Moreover, ectopic expression of HIF2α significantly increased RAB11B-AS1 HRE-driven firefly luciferase activity in nonhypoxic and hypoxic HEK293T cells (Fig. 2F; Supplementary Fig. S2A). In contrast, HIF2α KO significantly inhibited hypoxia-induced RAB11B-AS1 HRE activity in HeLa cells (Fig. 2G; Supplementary Fig. S2B). Collectively, these data indicate that RAB11B-AS1 is a direct HIF2 target gene in human breast cancer cells.

To determine the oncogenic role of RAB11B-AS1 in breast cancer cells, we next generated RAB11B-AS1-overexpressed MDA-MB-231 cells. RT-qPCR assay showed about 14-fold increase in RAB11B-AS1 RNA levels in these cells (Fig. 3A). The Boyden transwell assay showed that ectopic expression of RAB11B-AS1 significantly increased migration and invasion of MDA-MB-231 cells under 20% and 1% O₂ as compared with the EV (Fig. 3B–E). Similar results were also observed in SUM159 cells (Supplementary Fig. S3A–S3E). These data indicate that RAB11B-AS1 overexpression promotes breast cancer cell migration/invasion in vitro.

To complement gain-of-function studies, we next performed loss-of-function studies by generating two independent RAB11B-AS1 knockdown (KD) MDA-MB-231 cell lines via lentiviral infection. About 80% to 90% of RAB11B-AS1 KD efficiency was achieved in either of cell lines under hypoxia (Fig. 3F). Hypoxia increased migration of SC MDA-MB-231 cells by more than 1.5-fold, whereas RAB11B-AS1 KD significantly inhibited cell migration under 20% and 1% O₂ (Fig. 3G and H). Similarly, the number of cells invaded through Matrigel was also significantly decreased by RAB11B-AS1 KD under 20% and 1% O₂ (Fig. 3I and J). Likewise, RAB11B-AS1 KD also significantly reduced migration and invasion of SUM159 or MDA-MB-468 cells under 20% and 1% O₂ (Supplementary Figs. S3F–S3O). Together, gain- and loss-of-function studies indicate that RAB11B-AS1 promotes breast cancer cell motility in vitro.

To investigate the effect of RAB11B-AS1 on breast cancer cell motility in vivo, we implanted EV or RAB11B-AS1-overexpressed MDA-MB-231 cells orthotopically into the mammary fat pad of female NSG mice. Ectopic expression of RAB11B-AS1 in MDA-MB-231 cells did not affect primary tumor growth in mice (Fig. 4A–C). RT-qPCR assay confirmed >20-fold increase in RAB11B-AS1 RNA levels in RAB11B-AS1 tumors (Fig. 4D). The blood vessel marker endomucin-positive area was significantly increased in RAB11B-AS1 tumors as compared with EV tumors (Fig. 4E and F), whereas the area percentage of staining of Ki-67, a marker for cell proliferation, was comparable in these tumors (Fig. 4E and G). RAB11B-AS1 did not alter the percentage of the cell death marker CC3-positive area in tumors (Supplementary Figs. S4A and S4B), suggesting that RAB11B-AS1 has no effect on caspase-3-dependent cell death in breast tumors. Notably, H&E analysis in the lungs revealed more spontaneous metastases in mice bearing RAB11B-AS1 tumors than EV tumors (Fig. 4H and I), which was confirmed by analysis of the presence of human genomic DNA in mouse lungs with qPCR (Fig. 4J). Similarly, ectopic expression of RAB11B-AS1 significantly increased MDA-MB-231 cell metastasis to the liver in mice (Fig. 4K–M).

Next, we performed loss-of-function studies in the orthotopic breast cancer mouse model. In line with gain-of-function studies above (Fig. 4A–C), RAB11B-AS1 KD in MDA-MB-231 cells did not affect primary tumor growth in NSG mice (Fig. 5A–C). RT-qPCR assay confirmed robust RAB11B-AS1 KD in tumors (Fig. 5D). IHC analysis showed the density of endomucin-positive blood vessels was significantly decreased in RAB11B-AS1 KD1 or KD2 tumors as compared with SC tumors (Fig. 5E and F). In contrast, Ki-67 staining was comparable between SC and RAB11B-AS1 KD tumors (Fig. 5E and G), in line with gain-of-function studies (Fig. 4E and G). Both H&E and qPCR assays showed decreased spontaneous breast cancer metastasis to the lungs and liver in mice bearing RAB11B-AS1 KD tumors as compared with those with SC tumors (Fig. 5H–M).

To investigate whether RAB11B-AS1 regulates breast cancer cell colonization to increase distant metastasis, we injected SC or RAB11B-AS1 KD1 or KD2 MDA-MB-231 cells into the tail vein of female NSG mice. The lungs were harvested 3 weeks later. H&E staining detected extensive tumor foci in the lungs from SC tumor bearing mice (Fig. 5N and O). In contrast, fewer breast cancer cells were present in the lungs from mice bearing RAB11B-AS1 KD1 or KD2 tumors (Fig. 5N and O). qPCR assay further confirmed that RAB11B-AS1 KD significantly reduced breast cancer cell colonization in the lungs (Fig. 5P). Collectively, these findings indicate that RAB11B-AS1 increases tumor angiogenesis and promotes breast cancer metastasis to distant organs in mice without affecting primary tumor growth.

To dissect the molecular mechanism of RAB11B-AS1–induced tumor angiogenesis, we investigated whether RAB11B-AS1 regulates the expression of angiogenic factors in breast cancer cells. RNA-seq assay identified 1,526 upregulated genes and 1,677 downregulated genes in RAB11B-AS1–overexpressed MDA-MB-231 cells under hypoxia (FDR >0.05, LogCPM >0, fold change >1.25; Fig. 6A). Gene ontology analysis of upregulated genes revealed that angiogenesis and response to hypoxia were two of top biological processes (Supplementary Table S3). To validate, we performed RT-qPCR assay in EV and RAB11B-AS1–overexpressed MDA-MB-231 cells exposed to 20% or 1% O₂ for 24 hours. Consistently, ectopic expression of RAB11B-AS1 significantly increased the VEGFA mRNA expression in nonhypoxic and hypoxic MDA-MB-231 cells (Fig. 6B). Interestingly, ANGPTL4 mRNA was also enhanced by RAB11B-AS1 overexpression in nonhypoxic and hypoxic MDA-MB-231 cells (Fig. 6B), despite not being selected as the RAB11B-AS1 target gene according to our selection criteria in RNA-seq (Fig. 6A). Similar results were also observed in SUM159 cells (Supplementary Fig. S5A). Complementary to gain-of-function studies, RAB11B-AS1 KD by either of two shRNAs significantly decreased hypoxia-induced VEGFA and ANGPTL4 mRNAs in MDA-MB-231 and MDA-MB-468 cells (Fig. 6C; Supplementary Fig. S5B). Reduction of VEGFA and ANGPTL4 mRNAs was very modest or little in RAB11B-AS1 KD MDA-MB-231 and MDA-MB-468 cells exposed to 20% O₂ (Fig. 6C; Supplementary Fig. S5B). Other angiogenic factors including ANGPT2, CXCR4, FN1, MMP9, and bFGF were not significantly affected by RAB11B-AS1 KD1 or KD2 (Supplementary Fig. S5C). Ectopic expression or KD of RAB11B-AS1 failed to affect the mRNA levels of RAB11B and RPL13A in MDA-MB-231, MDA-MB-468, or SUM159 cells (Fig. 6B and C; Supplementary Fig. S5). These data indicate that RAB11B-AS1 increases hypoxia-induced angiogenic factors including VEGFA and ANGPTL4 in breast cancer cells.

Next, we studied whether angiogenic factors induced by RAB11B-AS1 in breast cancer cells regulates angiogenesis in vitro. Conditional culture media from EV or RAB11B-AS1–overexpressed MDA-MB-231 cells were collected and incubated with HUVECs. After 5-hour incubation, the length of tubes was significantly increased in HUVECs cultured with RAB11B-AS1 conditional media as compared with EV conditional media (Fig. 6D and E). These data indicate that RAB11B-AS1 increases the expression of angiogenic factors in breast cancer cells to promote angiogenesis in vitro, which supports our in vivo findings (Fig. 4E and F).

RNA Pol II binding to the gene promoter is an essential step in transcriptional regulation. Gene ontology of RAB11B-AS1–upregulated genes identified RNA Pol II-dependent gene regulation as the top biological function, implicating that RNA Pol II may be involved in RAB11B-AS1–mediated gene regulation (Supplementary Table S3). To determine the role of RNA Pol II in RAB11B-AS1-mediated VEGFA and ANGPTL4 expression, we studied whether RAB11B-AS1 interacts with RNA Pol II by RNA pull-down assay. In vitro transcribed biotin-labeled RAB11B-AS1 RNA was incubated in the presence of streptavidin magnetic beads with lysates of MDA-MB-231 cells exposed to 20% or 1% O₂ for 24 hours. Endogenous RNA Pol II was strongly pulled down by RAB11B-AS1 RNA, but barely bound to beads only (Fig. 6F). We further studied whether RAB11B-AS1 regulates RNA Pol II binding to VEGFA and ANGPTL4 genes by RNA Pol II ChIP-qPCR assay. RNA Pol II was enriched at the promoter of VEGFA and ANGPTL4 genes and its enrichment was significantly increased by hypoxia in MDA-MB-231 cells (Fig. 6G and H). RAB11B-AS1 KD significantly reduced RNA Pol II occupancy at the promoter of VEGFA and ANGPTL4 genes in hypoxic MDA-MB-231 cells (Fig. 6G and H), which was consistent with their reduced gene expression in RAB11B-AS KD cells under hypoxia (Fig. 6C; Supplementary Fig. S5). RAB11B-AS1 KD had no effect on RNA Pol II binding to the RPL13A gene in MDA-MB-231 cells (Fig. 6I). These findings indicate that RAB11B-AS1 promotes RNA Pol II binding to VEGFA and ANGPTL4 genes under hypoxia to increase their expression in breast cancer cells.

It has been known that HIF induces VEGFA and ANGPTL4 in cancer cells to promote tumor angiogenesis and metastasis (19). We studied the role of HIF in RAB11B-AS1–mediated VEGFA and ANGPTL4 expression. RNA pull-down assay showed that HIF1α and HIF2α both failed to bind to sense or antisense of RAB11B-AS1 RNA (Supplementary Fig. S6A). The stability of HIF1α and HIF2α proteins is critical for HIF transcriptional activity (16). Ectopic expression or KD of RAB11B-AS1 did not affect HIF1α and HIF2α protein levels in MDA-MB-231 cells (Supplementary Fig. S6B and S6C). We further performed HIF luciferase reporter assay to determine whether RAB11B-AS1 regulates HIF transcriptional activity. HEK293T cells were cotransfected with a HIF reporter plasmid, which contains 2x VEGFA HRE upstream of the SV40 promoter and firefly luciferase coding gene, and pSV40-Renilla, and RAB11B-AS1 expression vector or EV, and exposed to 20% or 1% O₂ for 24 hours. Ectopic expression of RAB11B-AS1 had no effect on HIF transcriptional activity in nonhypoxic and hypoxic HEK293T cells (Supplementary Fig. S6D). Together, these data indicate that HIF is not directly involved in RAB11B-AS1–mediated VEGFA and ANGPTL4 expression in hypoxic breast cancer cells.

To study clinical relevance of RAB11B-AS1, we analyzed its expression in human breast tumors and control benign breast tissues. RAB11B-AS1 RNA levels were significantly higher in human breast tumors as compared with control breast tissues (Fig. 7A). Likewise, VEGFA and ANGPTL4 mRNAs were also upregulated in human breast tumors (Fig. 7B and C). The Pearson correlation analysis revealed that RAB11B-AS1 RNA expression was significantly correlated with VEGFA or ANGPTL4 mRNA expression in human breast tumors (Fig. 7D and E). These data indicate that RAB11B-AS1 RNA is upregulated and correlated with VEGFA and ANGPTL4 mRNAs in human breast tumors.

Because thousands of lncRNAs have been identified in human genome by genome-wide gene analysis, it becomes clear that the expression of lncRNAs is differentially regulated by various stimuli in the tissue- and cell type-specific manners. In this study, we for the first time annotated the nucleotide sequence of human RAB11B-AS1 in MDA-MB-231 cells by 5′- and 3′-RACE assays and showed that RAB11B-AS1 is a hypoxia-inducible lncRNA in breast cancer cells. Hypoxia induction of RAB11B-AS1 occurs in all subtypes of human breast cancer cell lines, indicating that hypoxia is a general mechanism of regulation of RAB11B-AS1 expression in breast cancer. Our ChIP, luciferase reporter, and RT-qPCR data revealed that HIF2, but not HIF1, controls RAB11B-AS1 transcription in hypoxic breast cancer cells. Previous studies showed that hypoxia upregulates a subset of lncRNAs in MCF-7 cells (18), and several lncRNAs, including NEAT1, MALAT1, and UCA1, have been characterized to be regulated by HIF1 and/or HIF2 in breast cancer cells (20–22). As lncRNAs display tissue-specific expression patterns (5), further studies are needed to determine whether RAB11B-AS1 is also induced by HIF2 in normal and malignant cells from other tissues besides breast.

Accumulating evidence reveals that lncRNA plays an important role in gene regulation and several mechanistic models have been proposed (23). Our gain- and loss-of-function studies showed that RAB11B-AS1 enhances the expression of angiogenic factors including VEGFA and ANGPTL4 in breast cancer cells exposed to hypoxia. The expression of the RAB11B gene is not regulated by RAB11B-AS1 in breast cancer cells, although both genes are partially overlapped as shown by our RACE data. ANGPTL4 also locates next to the RAB11B-AS1 gene in the genome, but VEGFA locates on the different chromosome 6, suggesting that a trans-regulatory mechanism may contribute to RAB11B-AS1-mediated VEGFA and ANGPTL4 expression. Indeed, RAB11B-AS1 interacts with RNA Pol II and controls RNA Pol II binding to the promoter of VEGFA and ANGPTL4 genes to enhance their expression in hypoxic MDA-MB-231 cells. Although VEGFA and ANGPTL4 are both known HIF target genes (19), RAB11B-AS1 fails to increase HIF transcriptional activity, and thus it is unlikely that HIF is directly involved in RAB11B-AS1–mediated VEGFA and ANGPTL4 expression under hypoxia. Collectively, our findings reveal that the RAB11B-AS1-RNA Pol II axis represents a previously unrecognized mechanism of hypoxia-induced expression of angiogenic factors in breast cancer (Fig. 7F).

Similar to protein-coding genes, lncRNAs participate in multiple cellular signaling pathways, including oncogenic signaling (24). Our present studies showed that RAB11B-AS1 increases the expression of a subset of angiogenic factors including VEGFA and ANGPTL4 in hypoxic breast cancer cells. In line with regulation of these angiogenic factors, RAB11B-AS1 promotes angiogenesis in vitro and in xenograft tumors. Angiogenesis promotes breast cancer metastasis to distant organs (25), suggesting that RAB11B-AS1–mediated angiogenesis may contribute to breast cancer metastasis at least in the xenograft mouse model that mimics spontaneous metastasis. Further studies are needed to investigate its role in the clinical model of metastasis and if additional mechanisms are also involved in RAB11B-AS1–mediated breast cancer metastasis as RAB11B-AS1 controls dysregulation of many genes in breast cancer cells under hypoxia. Notably, primary breast tumor growth is not affected by RAB11B-AS1 overexpression or KD. In accordance with this, RAB11B-AS1 fails to regulate the expression of the cell proliferation marker Ki-67 in xenograft tumors. These findings reveal that RAB11B-AS1 is a metastasis-associated lncRNA in breast cancer. A previous study reported that RAB11B-AS1 abrogates osteosarcoma cell proliferation, migration, and invasion in vitro and suppresses osteosarcoma growth in vivo (26). Thus, RAB11B-AS1 may function as both oncogene and tumor suppressor in a tissue-specific manner. The tissue-specific functions have been similarly observed for other lncRNAs, such as H19 (27, 28).

In summary, we identified a novel HIF2 target lncRNA RAB11B-AS1 in breast cancer cells. RAB11B-AS1 is induced by hypoxia and in turn increases the expression of angiogenic factors including VEGFA and ANGPTL4 in hypoxic breast cancer cells through the recruitment of RNA Pol II, leading to tumor angiogenesis and metastasis (Fig. 7F). These findings provide novel mechanistic insights into breast cancer progression and uncover RAB11B-AS1 as a possible target for the treatment of metastatic breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: Y. Niu, Y. Wang, W. Luo

Development of methodology: Y. Niu, Y.V. Fang, C. Xing, Y. Wang, W. Luo

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): Y. Niu, L. Bao, Y. Chen, C. Wang, M. Luo, B. Zhang, M. Zhou, J.E. Wang, Y.V. Fang, Y. Wang

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Niu, Y.V. Fang, A. Kumar, C. Xing, Y. Wang, W. Luo

Writing, review, and/or revision of the manuscript: Y. Niu, L. Bao, Y. Chen, C. Wang, M. Luo, B. Zhang, M. Zhou, J.E. Wang, Y.V. Fang, C. Xing, Y. Wang, W. Luo

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): Y. Niu, Y.V. Fang, C. Xing, Y. Wang, W. Luo

Study supervision: C. Xing, Y. Wang, W. Luo

Y. Niu (CPRIT); L. Bao (NIH, Welch); Y. Chen (Susan G Komen); C. Wang (CPRIT); M. Luo (Susan G Komen); B. Zhang (CPRIT); M. Zhou (CPRIT); J.E. Wang (NIH); Y.V. Fang (UTSW); A. Kumar (UTSW); C. Xing (UTSW); Y. Wang (NIH, CPRIT, Welch, UTSW); W. Luo (NIH, CPRIT, Welch, Susan G Komen, UTSW). We thank Ondine Cleaver (UT Southwestern) for HUVECs and the UTSW Cancer Center Tissue Resource for assistance in IHC. This work was supported by grants from the Susan G. Komen (CCR16376227), the NIH (R01CA222393, R00CA168746), the CPRIT (RR140036, RP190358), and the Welch Foundation (I-1903) to W. Luo, and the NIH (R00NS078049, R35GM124693, and R01AG066166), the CPRIT (RP170671), and the Welch Foundation (I-1939) to Y. Wang. W. Luo is a CPRIT Scholar in Cancer Research.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.