The lncRNA MIR2052HG regulates ERα levels and aromatase inhibitor resistance through LMTK3 by recruiting EGR1

Human ER-positive breast cancer CAMA-1, HER-positive Au565, triple negative breast cancer (TNBC) MDA-MB-231, and human embryonic kidney cell line 293 T cell lines were obtained from American Type Culture Collection (ATCC). The identities of all cell lines were confirmed by the medical genome facility at Mayo Clinic Center (Rochester MN) using short tandem repeat profiling upon receipt. The breast cancer MCF7/AC1 cell line (stably transfected CYP19A1 gene) was provided by Dr. Angela Brodie (University of Maryland, Baltimore, MD). The cells were authenticated in 2015 by Genetica DNA Laboratories (Cincinnati, OH) using a StemElite ID system that uses short tandem repeat genotyping. CAMA-1cells were cultured in EMEM media (Eagle’s minimum essential medium) (ATCC) with 10% fetal bovine serum (FBS) (Corning) and MCF7/AC1 cells were cultured in IMEM (Improved MEM, no phenol red) (Life Technologies) with 10% heated inactive FBS in the incubator with 5% CO₂ at 37°C. 293T cells were cultured in Dulbecco’s modified Eagle’s medium (Life Technologies) with 10% FBS, and Au565 cells were cultured in RPMI-1640 (Life Technologies) with 10% FBS in the incubator with 5% CO₂ at 37 °C. MDA-MB-231 cells were cultured in L-15 medium with 10% FBS in the incubator without CO₂. Five lymphoblastoid cell lines (LCLs) with MIR2052HG wild-type (WT) SNPs and five LCLs with MIR2025HG variant SNPs were selected. Before androstenedione treatment, ~ 2 × 10⁷ cells from each LCL were cultured in RPMI-1640 media containing 5% charcoal stripped FBS (Invitrogen) for 24 h, followed with RPMI-1640 medium without FBS for additional 24 h. Each LCL was plated into 12-well plates with RPMI-1640 medium containing 0.01, 0.1, 1, and 10 nM androstenedione. After 24 h treatment, increasing concentrations of anastrozole or exemestane were added. The anastrozole and exemestane concentrations ranged from 0.1, 1, 10, to 100 nM. After an additional 24 h, all LCLs were collected for further RNA extraction and qRT-PCR.

Anastrozole and exemestane (Selleckchem) were dissolved in DMSO (Sigma-Aldrich) as 100 mM stock. 4-Androstene-3, 17-dione (Steraloids Inc.) was dissolved in 100% ethanol. FLAG-ERα plasmid was provided by Thomas Spelsberg, Ph.D. (Mayo Clinic). FLAG-ERα S167A mutant was generated from the FLAG-ERα plasmid using QuickChange Site Directed Mutagenesis kit (Agilent). HA-Ub plasmid was provided by Zhenkun Lou, Ph.D. (Mayo Clinic). LMTK3 plasmid was purchased from Origene. Protein kinase C (PKC) inhibitor, Go 6983, was purchased from Sigma. Antibodies against GAPDH, AKT, p-AKT (Ser473), FOXO3, EGR1, EZH2, RSK1, pRSK1, ERK1/2, p-ERK1/2, MEK1/2, p-MEK1/2, and Phospho-(Ser) PKC Substrate Antibody were purchased from Cell Signaling. ERα and ERα-S167 antibodies were obtained from Abcam. LMTK3, BHLHE40, CTCF, EP300, HDAC6, POLR2A, REST, CREBBP, YY1, STAT1, and SNRNP70 antibodies were purchased from Santa Cruz. CHD2 antibody was from Thermo Fisher Scientific. Secondary HRP (horseradish peroxidase)-conjugated anti-rabbit IgG and anti-mouse IgG antibodies were from Cell Signaling.

Antisense oligonucleotides (ASOs) were used to knockdown and study the functions of MIR2052HG. Pools of two ASOs for MIR2052HG produced with locked nucleic acids modification of 5′ and 3′ ends (Exiqon) were validated previously [28]. Negative control ASO was obtained from Exiqon. Lipofectamine RNAi Max (Invitrogen) and OPTI-MEM (Life Technologies) were used for ASOs transfection. Knockdown efficiency was measured using TaqMan qRT-PCR. The sequences of ASOs were as follows: ASO1: 5′-GTTGATTAGATTTGG-3′; ASO2: 5′-ACAGTCCCGATCTTC-3′; negative control: 5′-AACACGTCTATACGC-3′. LMTK3 plasmids (Origene) were transfected into cells using Lipofectamine 2000 reagent (Thermo Fisher Scientific). Total RNA was extracted 48 h after transfection for RNA quantification. Whole cell lysates were collected 48 h after transfection for western blots.

QRT-PCR assays were performed for measuring gene expression using the TaqMan RNA-to-Ct 1-Step Kit (Life Technologies). RNA was extracted using the miRNeasy mini Kit (Qiagen). RNA was measured by NanoDrops300. The TaqMan primers for MIR2052HG, ESR1, and GAPDH were purchased from Life Technologies. Primers for EGR1 targeted genes and ESR1 targeted genes were purchased from IDT. QRT-PCR reactions were prepared following the manufacturer’s protocol. Samples were run using the StepOnePlus real-time PCR system (ABI).

Cells were washed with cold PBS and were lysed in cold NETN buffer (100 mM NaCl, 20 mM Tris·HCl pH = 0.8, 0.5 mM EDTA, NP-40) with proteasome cocktail inhibitor and phosphatase inhibitor PhosSTOP EASY. Cell lysates were diluted with SDS loading buffer (SDS, glycerol, bromic acid, 1 M Tris·HCl) and boiled, centrifuged at 10,000 rpm, and stored at − 20 °C. Equal amounts of protein were subjected to electrophoresis in TGX SDS gels and were transferred to PVDF membranes. Membranes were blocked in TBS with 5% BSA and 0.1% Tween-20 and then incubated overnight at 4 °C with the indicated primary antibodies. Membranes were washed with TBS-T (TBS with 0.1% Tween-20) and then incubated with anti-mouse IgG or anti-rabbit IgG for 1 h at room temperature. All blots were visualized with the Supersignal WestPico or Supersignal WestDura chemiluminescent ECL kits (Thermo Fisher) and blue X-ray films or Gel Doc XR+ Gel documentation system (Biorad).

RNA was prepared from MCF7/AC1 cells using the TRIzol extraction kit (Life Technologies). Genomic DNA was removed using the Ambion DNA-free kit. NuGEN Encore reagents were used for library preparation of total RNA samples. One microgram of total RNA input was used for each sample. The libraries were sequenced on an Illumina HiSeq 2000 sequencing system using 100-bp single-ended reads. After removing the poor-quality bases from FASTQ files for the whole transcriptome sequencing, paired-end reads were aligned by reads that were aligned to the human reference genome UCSC hg19 with Tophat 2.0.14 and the bowtie 2.2.6 aligner option. Transcript abundance was estimated using a count-based method with htseq-count.

Cells were seeded (2000 cells/100 μL/well) in a 96-well plate. The CyQUANT Direct Cell Proliferation Assay kit (Invitrogen) was used to determine the cell viability in six replicates. CyQUANT assays were performed to determine the cell viability every 2 days. Measurements were made using a microplate reader with excitation at 485 nm and emission detection at 530 nm. Each absorbance was normalized to the media control without any cells.

Cell survival assays were carried out in 96-well plates. Cells were seeded (5000 cells/100 μL/well) in a 96-well plate and treated with AIs for 72 h. CyQUANT assay was used to determine the cell viability in six replicates. Each absorbance was normalized to the media control without any cells.

Cells transfected with MIR2052HG ASOs or LMTK3 plasmids were plated (800 ~ 1000 cells/well) in 6-well culture clusters in triplicates. Subsequently, the cells were cultured for up to 14 days at 37 °C, 5% CO₂ to allow colony formation. Colonies were washed with cold PBS, fixed with 4% paraformaldehyde and stained by 0.05% crystal violet. Colonies (> 50 cells) were accounted with the ImageJ software (version 1.42q).

Approximately1.5 μg of HA-ubiquintin plasmid and 1.5 μg of pcDNA 4.1-ERα plasmid with FLAG tag were co-transfected in HEK 293 T cells using lipofectamine 2000. Twenty-four hours later, cells were reversely transfected with the MIR2052HG ASO or negative control. Approximately 2 × 10⁵ ASOs transfected and control cells were subsequently seeded into each 60-mm dishes. After 64 h, MG132 was added at a final concentration of 10 μM for an additional 8 h. Cells were then collected for the ubiquitination assay. Specifically, these cells were washed in cold PBS with NEM (1:100) and lysed in 2% SDS lysis buffer [62.5 mM Tris·HCl pH = 6.8, 10% glycerol (v/v), SDS 2% (g/v)]. Immunoprecipitation assays were performed with the anti-FLAG antibody-conjugated gels (Sigma). After washing, the FLAG gels were dissolved in 2xSDS loading buffer and boiled. These samples were then subjected to western blotting using the anti-ubiquitin antibody and anti-FLAG antibody.

PepTag assay for nonradioactive detection of PKC Activity: MCF7/AC1 and CAMA-1 cells were transfected with indicated ASO or plasmids. Forty-eight hours later, cells were lysed and incubated with PKC reaction mixture (25 μl) according to the manufacturer’s (Promega) protocol at 30 °C for 30 min. The reactions were stopped by heating at 95 °C for 10 min. After adding 80% glycerol (1 μl), phosphorylated and nonphosphorylated PepTag peptides were separated by electrophoresis with 0.8% agarose gel. The negatively charged phosphorylated bands were excised using a razor blade, placed in a graduated microcentrifuge tube, and heated at 95 °C until the gel slice melted. The volume of the solution was adjusted to 250 μl with water. The hot agarose solution (125 μl) was added to a separate tube containing 75 μl of gel solubilization solution and 50 μl of glacial acetic acid. The absorbency was read at 570 nm. Using the absorbance, we calculated the number of units of kinase activity in each slice of agarose as per the manufacturer’s instructions (Promega). Assays were performed in triplicates.

Western Blot analysis of PKC activity was performed using lysates from MCF7/AC1 and CAMA-1 cells transfected with indicated ASO or plasmids. Phosphoserine PKC substrate proteins were detected by incubation overnight at 4 °C with anti-phosphoserine PKC substrate antibody.

ChIP assays were performed using EpiTect ChIP OneDay kit (Qiagen). MCF7/AC1 and CAMA-1 cells were transfected with MIR2052HG ASO for 24 h. Cells were then subjected to ChIP assay as described by the manufacturer. LCLs were cultured in 5% charcoal stripped FBS for 24 h, followed with serum-free medium for additional 24 h. LCLs were then treated with 1 nM androstenedione, 1 nM androstenedione plus 100 nM anastrozole, and 1 nM androstenedione plus 100 nM exemestane for additional 24 h. Approximately 2 × 10⁷ LCLs per every sample (different SNP genotypes with androstenedione or androstenedione plus anastrozole or exemestane treatment groups) were collected for the ChIP assay. Equal amount of chromatin from each sample (~ 2 million cells each IP) and 1 μg control IgG or antibody against EGR1 were used. Q-PCR was carried out, and the result was normalized to input. All primers are listed in Additional file 1: Table S1.

The sequential protein staining and RNA detection were performed as previously described [34, 35]. Briefly, the cells were grown in chamber slides. LMTK3 staining was performed as usual until secondary antibody is labeled in the presence of RNase inhibitor. Slides were then dehydrated by serial treatment of ethanol with different concentrations. The Alexa Fluor 488-labeled RNA probe was obtained using the FISH Tag RNA Kit (Invitrogen). In the first step, in vitro transcription is used to enzymatically incorporate an amine-modified nucleotide into the probe template. The modified nucleotide is UTP having an NH2 group attached through a linker to the C5 position of the base. In the second step, dye labeling of the purified amine-modified RNA is achieved by incubation with amine-reactive dyes. These active ester compounds react with the primary amines incorporated into the probe template, covalently conjugating the dye to the modified nucleotide base. The purified probe is then ready for hybridization to the specimen slides at 37 °C overnight. Signal was then amplified using Tyramide Signal Amplification (TSA) kit (Life Technologies). LMTK3 DNA probe was produced using the FISH Tag DNA Kit (Invitrogen). In the first step, nick translation is used to enzymatically incorporate an amine-modified nucleotide into the probe template. The modified nucleotide is dUTP having an NH2 group attached through a linker to the C5 position of the base. In the second step, dye labeling of the purified amine-modified DNA is achieved by incubation with amine-reactive dyes. These active ester compounds react with the primary amines incorporated into the probe template, covalently conjugating the dye to the modified nucleotide base. The purified probe is then ready for hybridization to the specimen. For dual RNA-DNA-FISH, we used the protocol as previously described [35]. In brief, RNA-FISH was performed by using Nick-translated Alexa Fluor 488-labeled probe and followed by tyramide signal amplification kit as above. After RNA-FISH, the cells were treated by RNase A and denatured. Nick-translated BAC containing LMTK3 was labeled with Alexa Fluor 594 and used as probe. Images were obtained with the LSM 780 inverted confocal microscope runs on Zeiss’s Zen software package.

An RNA-binding protein immunoprecipitation (RIP) assay was performed using the Magna RIP kit (Millipore) according to the manufacturer’s instruction. Cell lysates from 50 million cells and 2–5 μg of control IgG or antibody against BHLHE40, CHD2, CTCF, EGR1, EP300, EZH2, HDAC6, POLR2A, REST, CREBBP, YY1, and STAT1 were used. We validated the RIP assay using the SNRNP70 antibody, which can bind to U1 snRNA. Specifically, cells were washed on the plates twice with 10 mL of PBS, scraped off from plate, and centrifuged at 1500 rpm for 5 min at 4 °C and discard the supernatant. Cell pellet was re-suspended in an equal pellet volume of complete RIP lysis buffer and then incubated on ice for 5 min. Dispense ~ 200 μL each of the lysate into nuclease-free microcentrifuge tubes and store at − 80 °C. Immunoprecipitations were performed using antibodies of interest and IgG control. Anti-SNRNP70 served as controls. Fifty micro liters of magnetic beads was washed and re-suspended in 100 μL of the RIP wash buffer. Two to approximately micrograms of the antibody of interest was added to each reaction and incubated with rotation for 30 min. The beads were then washed three times with RIP wash buffer. Nine hundred microliters of RIP immunoprecipitation buffer was then added to each tube. The RIP lysate were thawed and centrifuged at 14,000 rpm for 10 min at 4 °C, and 100 μL of the supernatant was added to each beads-antibody complex in RIP immunoprecipitation buffer. Ten microliters of the supernatant of RIP lysate was removed as “10% input” and stored at − 80 °C until starting RNA purification. The immunoprecipitations were incubated with rotating overnight at 4 °C, followed by six washes with 500 μL of cold RIP wash buffer. RNA purification was then performed. Each immunoprecipitate was re-suspended in 150 μL of proteinase K buffer. The input samples were thawed and 107 μL of RIP wash buffer, and 15 μL of 10% SDS and 18 μL of proteinase K were added to the tubes. All tubes were incubated at 55 °C for 30 min with shaking to digest the protein and then centrifuged briefly before being placed on the magnetic separator. The supernatant was then transferred into a new tube, together with 250 μL of RIP wash buffer. Four hundred microliters of phenol to chloroform to isoamyl alcohol was then added to each tube, followed by vortex for 15 s and centrifugation at 14000 rpm for 10 min to separate the phases. Three hundred microliters of the aqueous phase was carefully removed and placed in a new tube, together with 400 μL of chloroform. After vortex for 15 s and centrifugation at 14000 rpm for 10 min, the phases were separated. Three hundred microliters of the aqueous phase was carefully removed, and place it in a new tube. 50 μL of salt solution I, 15 μL of salt solution II, 5 μL of precipitate enhancer, and 850 μL of ethanol were added to each tube and kept at − 80 °C overnight to precipitate the RNA. The samples were then centrifuged at 14,000 rpm for 30 min at 4 °C. The pellets were washed once with 80% ethanol, air dried, and re-suspended in 10 to 20 μL of RNase-free water. The RNAs were then analyzed by qRT-PCR.

Transcription activity of EGR1 was measured using the dual luciferase assay with the Cignal EGR1 Reporter Assay Kit (Qiagen). MIR2052HG knocked-down MCF7/AC1 and CAMA-1 cells were transfected with either EGR1 reporter (EGR1-responsive GFP reporter), negative control (GFP reporter construct with GFP expression controlled by a minimal promoter), or positive control (constitutively expressing GFP construct) constructs using the Lipofectamine 2000 transfection reagent. After 24 h of transfection, luciferase assay was performed using the Dual-Glo Luciferase Reporter Assay System (Promega) following the manufacturer’s protocol.

For cell survival, cell proliferation, kinase activity, gene expression, and quantifications, data are represented as the mean ± SEM of three independent experiments. Statistical analyses were performed with Student’s t test. Statistical significance is represented in figures by *p < 0.05 and **p < 0.01.

We previously reported that MIR2052HG sustained ERα levels by promoting AKT/FOXO3-mediated upregulation of ESR1 transcript and by limiting proteasome-dependent degradation of ERα protein [28]. However, the mechanism involved in the regulation of MIR2052HG-mediated AKT activation and ERα ubiquitination remains unknown. Kinome screening previously identified LMTK3 as a potent ERα regulator, acting by decreasing the activity of protein kinase C (PKC) and the phosphorylation of AKT (Ser473), resulting in increased binding of FOXO3 to the ESR1 promoter [24]. LMTK3 also protected ERα from proteasome-mediated degradation [24]. Given that the effects of LMTK3 on ERα were similar to our observations with MIR2052HG [28], we hypothesized that MIR2052HG might regulate LMTK3 to mediate ERα levels and, in turn, response to AIs.

Previous studies have demonstrated that lncRNAs can function in trans to regulate expression of protein-coding genes; therefore, we examined the possibility that MIR2052HG may facilitate AI resistance by regulating LMTK3 expression. Consistent with this hypothesis, we found that knockdown of MIR2052HG using a pooled ASO in human ER-positive CAMA-1 breast cancer cells resulted in a dramatic decrease of LMTK3 expression (Fig. 1a). A similar effect was also observed in an aromatase overexpressing cell line, MCF7/AC1 [36] (Fig. 1b). We also observed that the changes in mRNA levels were confirmed at the protein level by the western blot analysis (Fig. 1c), supporting the notion that MIR2052HG regulates LMTK3 expression. To determine whether LMTK3 is a major downstream target of MIR2052HG in regulating AI response, we first determined the transcriptome changes in MIR2052HG-knockdown MCF7/AC1 cells and collected published RNA-seq data after LMTK3-knockdown [26]. Analysis of the RNA-seq data indicated that the changes induced by MIR2052HG knockdown and LMTK3 knockdown showed a large number genes overlapped, especially almost 1/3 genes regulated by LMTK3 were also regulated by MIR2052HG (Fig. 1d, Additional file 2: Table S2). The common dysregulated genes in both knockdowns included cell cycle genes, oocyte maturation, and oocyte meiosis genes (Additional file 2: Table S2).

To further define the relationship between MIR2052HG and LMTK3, we transfected LMTK3 overexpressing constructs into ERα-positive breast cancer cells with MIR2052HG-knockdown, followed by cell growth and colony forming assays. The cell proliferation and colony formation analysis demonstrated that the cell growth defect caused by downregulation of MIR2052HG could be successfully rescued by LMTK3 overexpression (Fig. 2a–d), indicating that LMTK3 is a major target that mediates the MIR2052HG regulation on cell growth in ER-positive breast cancer.

Previous studies demonstrated that MIR2052HG regulates ERα expression through transcription regulation of ESR1 and ER protein degradation [24, 28]. However, the direct target of MIR2052HG in ERα regulation has not been fully elucidated. Therefore, we examined the role of LMTK3 in ERα regulation. Our previous report indicated that the effect of MIR2052HG on ESR1 transcript is through AKT/FOXO3 [24]. In MCF7/AC1 and CAMA-1 cells, downregulation of MIR2052HG reduced ESR1 mRNA levels by promoting AKT-mediated downregulation of FOXO3 protein level, a transcription factor known to be involved in ESR1 transcription (Fig. 3a, b). LMTK3 overexpression rescued the downregulation of ERα mRNA induced by MIR2052HG silencing (Fig. 3a, b). LMTK3 overexpression resulted in a decrease in phosphorylated AKT (at Ser473) and an increase in FOXO3 protein level but not mRNA level (Fig. 3a, b, Additional file 3: Figure S1a). At the protein level, ERα protein was reduced by MIR2052HG knockdown, whereas LMTK3 overexpression stabilized ERα (Fig. 3a, b, right panel). Our previous results showed that MIR2052HG also regulated ERα stability by regulating its proteasome-dependent degradation process. Here, LMTK3 overexpression could increase protein level by decreasing ERα ubiquitination (Fig. 3c). Together, these data indicate that LMTK3, downstream of MIR2025HG, mediated MIR2052HG effect on the regulation of ESR1 transcription and ERα protein stability.

Next, we investigated the mechanisms involved in MIR2052HG and LMTK3 regulation of ERα protein degradation. ERα phosphorylation, especially increased phosphorylation of ERα at Ser167, has been implicated in ERα proteasome-mediated degradation [24]. To explore the mechanism, we first determined the level of ERα pSer167 in MIR2052HG knockdown ERα-positive breast cancer cells. ERα pSer167 levels increased with MIR2052HG knockdown, despite decreased total ERα amounts (Fig. 4a). Furthermore, we observed that knocking down MIR2052HG increased wild-type (WT) ERα ubiquitination, but not the mutant ERα with serine 167 to alanine (S167A) (Fig. 4b), confirming the involvement of ERα Ser167 in ubiquitin-dependent and proteasome-mediated degradation. The phosphorylation of ERα at Ser167 is regulated by pp90 (RSK1) [22] which is activated by MAPK [37]. We thus hypothesized that MEK/ERK/p90RSK1 might be the signaling pathway that mediates ERα phosphorylation at Ser167 upon MIR2052HG knockdown. We first tested the effect of knockdown of MIR2052HG on MEK/ERK/p90RSK1 activity. As shown in Fig. 4a, in MCF7/AC1 and CAMA-1 cells transfected with MIR2052HG ASO, coinciding with increased ERα pSer167, pMEK, pERK, and pRSK1 levels were also increased, indicating an increased MEK/ERK/p90RSK1 activity. We then examined the role of LMTK3 in MIR2052HG-mediated regulation of MEK/ERK/p90RSK1 activity and found that LMTK3 overexpression abolished increased pMEK/pERK/p90RSK1 levels caused by MIR2052HG silencing, resulting in increased ERα protein (Fig. 4c, d) and its transcriptional activity (Additional file 3: Figure S1b). These data further imply that MIR2052HG regulates LMTK3 expression, which then influences MEK/ERK/p90RSK1 activity, regulating ERα protein levels.

As protein kinase C (PKC) has been implicated to play a role in ERα protein degradation [38] and AKT–FOXO3 regulation [39], and LMTK3 inhibits PKC activity [24], we examined the effects of MIR2052HG and LMTK3 on PKC. Downregulation of MIR2052HG increased PKC activity (Fig. 4e, f, and Additional file 3: Figure S1c), whereas overexpression of LMTK3 decreased PKC activity (Fig. 4e, f, and Additional file 3: Figure S1c). In addition, LMTK3 overexpression dramatically reduced PKC activity that was induced by MIR2052HG silencing (Fig. 4e, f, and Additional file 3: Figure S1c). Inhibition of PKC with the Go 6983 inhibitor reduced MEK/ERK/p90RSK1 activity and ERα pSer167, which in turn, partially rescued ERα levels (Fig. 4g), suggesting that MIR2052HG regulated ERα protein level through the axis of LMTK3/PKC/MEK/ERK/RSK1. Our findings also confirmed that MIR2052HG effects on AKT/FOXO3 activation and downstream ESR1 mRNA level were through the regulation of LMTK3/PKC pathway (Figs. 3 and 4).

Next, we wanted to address how MIR2052HG regulates LMTK3 transcription. First, we examined the localization of the MIR2052HG RNA transcript. RNA-FISH demonstrated that MIR2052HG localized to a limited number of nuclear foci (one to two spots in most cases), suggesting that MIR2052HG had limited targets. We also checked the genomic location of the MIR2052HG transcript by RNA-DNA dual FISH, and the results showed that the MIR2052HG transcript was located at the LMTK3 gene locus (Fig. 5a, Additional file 4: Figure S2). Taken together, these data suggest that MIR2052HG is likely involved in LMTK3 transcription.

LMTK3 expression could be activated by several transcription factors based on the ENCODE database, including BHLHE40, CHD2, CTCF, EGR1, EP300, EZH2, HDAC6, POLR2A, REST, CREBBP, YY1, and STAT1. Therefore, we asked whether any of these transcription factors, together with MIR2052HG might be involved in the regulation of MIR2052HG expression. Immunoprecipitation followed by qRT-PCR analysis demonstrated that MIR2052HG was significantly enriched in the EGR1 immunoprecipitates (Fig. 5b, c). The enrichment of MIR2052HG by the EGR1 antibody was specific, as the antibody did not pull down another lncRNA, LOC102724785 (Fig. 5b, c). Knockdown of MIR2052HG did not change LMTK3 expression in a HER2-positive Au565 and a TNBC MDA-MB-231 cell lines, and no significant enrichment of MIR2052HG by the EGR1 antibody was observed in AU565 and MDA-MB-231 cells (Additional file 5: Figure S3). These data suggest that MIR2052HG regulation of LMTK3 transcription involves EGR1 in ER-positive breast cancer.

EGR1 was highly expressed in The Cancer Genome Atlas (TCGA) [40] ER-positive breast cancer patients (Additional file 6: Figure S4). We then confirmed EGR1 regulation of LMTK3 gene expression in the MCF7/AC1 and CAMA-1 cells. Knockdown of EGR1 reduced LMTK3 mRNA level (Fig. 6a, b). To examine whether binding of EGR1 to the LMTK3 promoter requires MIR2052HG, we first mapped the binding locations of EGR1 on the LMTK3 gene locus (Fig. 6c, chr19:48994366-48994811, chr19:48996320-48996559, chr19:49015095-49015334). ChIP assays demonstrated that EGR1 bound to all three binding sites (Fig. 6d, e). Importantly, knocking down MIR2052HG reduced the EGR1 binding to the LMTK3 gene locus (Fig. 6d, e) without significant effect on the binding of EGR1 to other EGR1 targets (Additional file 7: Figure S5 and Additional file 8: Figure S6). Furthermore, MIR2052HG failed to locate in the LMTK3 gene locus in EGR1 knockdown cells (Fig. 6f). Although EGR1 remained as a transcription factor for LMTK3 in HER2-positive and TNBC cells and knockdown LMTK3 inhibited cell growth (Additional file 9: Figure S7), knocking down MIR2052HG did not change the EGR1 binding to LMTK3 gene locus (Additional file 9: Figure S7 c, d).

Our previous GWAS showed that MIR2025HG SNPs regulate its own gene expression as well as ERα expression in an estrogen or AI-dependent fashion [28]. Based on our findings showing MIR2025HG regulation of LMTK3, we then determined whether the expression of LMTK3 might be also SNP- and AI-dependent using the human LCLs system. This cell line model system, consisting of 300 individual LCLs for which we have extensive genomic and transcriptomic data, has shown repetitively to make it possible for us to study the relationship between common genetic variant and cellular phenotypes [28, 41, 42]. In the presence of androstenedione, LCLs with variant genotypes for both of the MIR2052HG SNPs, rs4476990 and rs3802201, showed dose-dependent increases in LMTK3 expression (Fig. 7a, b). However, addition of AI, either anastrozole (Fig. 7a) or exemestane (Fig. 7b) caused a “reversal” of the expression pattern with increased LMTK3 expression in LCLs with homozygous WT, but a marked decrease in LCLs homozygous for the variant genotypes. Of particular interest was the observation of a direct correlation between this pattern of expression for MIR2052HG and ERα [28] and that of LMTK3 (Fig. 7a, b). Since MIR2052HG regulated LMTK3 expression in a SNP- and AI-dependent fashion (Fig. 7a, b), we determined if EGR1 binding to the LMTK3 promoter region was also SNP- and AI-dependent. In the presence of androstenedione, cells homozygous for the variant SNP genotypes showed increased binding of EGR1 to the LMTK3 promoter (Fig. 7c, d) relative to WT in ChIP assays using the EGR1 antibody. Anastrozole and exemestane could reverse this effect (Fig. 7c, d).

We next sought to determine the functional consequences of the MIR2025HG SNP on response to AIs. LCLs homozygous for the variant SNP, which have low LMTK3 expression (Fig. 7a, b), were more sensitive to anastrozole and exemestane than homozygous WT LCLs (Fig. 7e, f). To assess the role of MIR2052HG-EGR1-LMTK3 axis in AI response, we next used anastrozole-sensitive MCF7/AC1 (because of its high expression of the AI target, CYP19A1), and anastrozole-resistant MCF7/AnaR [43] cell lines to determine the role of MIR2052HG-EGR1-LMTK3 axis in these two settings. In both lines, knockdown of MIR2052HG or EGR1 significantly increased anastrozole sensitivity compared to negative control, whereas overexpression of LMTK3 in MIR2052HG-knockdown or EGR1-knockdown cells resulted in decreased AI sensitivity (Fig. 7g). These results suggest that MIR2052HG facilitates EGR1 recruitment to the LMTK3 promoter region in a SNP-dependent fashion to activate LMTK3 transcription, resulting in AI resistance.