lncRNA-WAL Promotes Triple-Negative Breast Cancer Aggression by Inducing β-Catenin Nuclear Translocation

In all types of breast cancer (BC), triple-negative breast cancer (TNBC) is a refractory molecular subtype (1, 2). TNBC is insensitive to current adjuvant therapy, and survival rates are considerably hindered for this disease (3). In the last few decades, a large number of studies on TNBC have contributed to the development of new therapy strategies, including immune checkpoint inhibitors, antibody‒drug conjugates, PARP inhibitors, and other therapy combinations (4). However, the prognosis of TNBC is the poorest among BC subtypes (5). An effective therapeutic strategy is still lacking for TNBC; hence, screening for effective therapeutic targets for TNBC is urgently needed (6, 7).

Research has revealed that the Wnt/β-catenin pathway is activated in TNBC (8, 9). In our study, we also found that the quantity of β-catenin in the nucleus was increased to 67% in patients with TNBC. It is well known about the activation of the Wnt/β-catenin pathway that β-catenin is not phosphorylated by destruction complexes and, subsequently, β-catenin is translocated to the nucleus and target genes are transcribed (10–14). In many cancers, including colorectal cancer, activation of the Wnt/β-catenin pathway is mainly caused by inactivating mutations of adenomatous polyposis coli (APC) and β-catenin (15, 16). However, somatic mutations (mutations of APC and β-catenin) that activate the wnt/β-catenin pathway are rare in TNBC; hence, there must be other involved mechanisms that are similar to the activation of the Wnt/β-catenin pathway in TNBC (17, 18).

Long noncoding RNAs (lncRNA) are an endogenous class of lncRNAs longer than 200 nucleotides that are not translated into proteins and are involved in diverse pathological processes, including cancer, inflammation, and other human diseases (19–22). In previous studies, lncRNAs could modulate various biological and pathological processes by targeting the wnt/β-catenin signaling pathway (23–25). lnc-LALR1 inhibited AXIN1 expression by recruiting CTCF to the AXIN1 promoter region and activated the wnt/β-catenin signaling pathway (26). lnc-00673-v4 activated Wnt/β-catenin signaling by augmenting the interaction between DDX3 and CK1 and consequently increased the aggressiveness of lung adenocarcinoma (23). In our previous study, lnc-NKILA could directly interact with functional domains of the signaling protein NF-κB/IκB and prevent the activation of NF-κB to suppress BC metastasis (27).

Herein, we report an lnc-WAL that was aberrantly expressed and pivotal for the function of the Wnt/β-catenin pathway in TNBC. Our study showed that lnc-WAL specifically bound β-catenin and AXIN, which resulted in dephosphorylation of β-catenin and nuclear translocation, sequentially enhancing Wnt/β-catenin pathway activation. In turn, the activation of wnt/β-catenin signaling promoted the production of lnc-WAL, demonstrating positive feedback between lnc-WAL and the Wnt/β-catenin pathway. Therefore, lnc-WAL promoted the aggressiveness of TNBC by activating the Wnt/β-catenin signaling pathway.

In situ hybridization (ISH) staining for lnc-WAL and immunohistochemistry (IHC) staining for β-catenin were performed on the tissue samples of primary TNBC (189 cases), luminal (120 cases), and HER2+ (120 cases) from Sun Yat-sen Memorial Hospital (SYSMH), Sun Yat-sen University (Guangzhou, China), and the First Affiliated Hospital of Chongqing Medical University (Chongqing, China) from May 2012 to March 2017. RNA levels of lnc-WAL were detected in the fresh frozen tissues of triple-negative (15 cases Wnt/β-catenin signaling activation and 15 cases Wnt/β-catenin signaling inactivation), luminal (30 cases), and HER2+ (30 cases) samples from SYSMH, Sun Yat-sen University (Guangzhou, China). This study was approved by the Institutional Ethics Committee of the SYSMH, Sun Yat-sen University, and First Affiliated Hospital of Chongqing Medical University.

A panel of breast cancer cell lines including TNBC cell lines (MDA-MB-231, MDA-MB-468, BT-549, Hs 578T, and HCC1937) and non-TNBC cell lines (MCF-7, T47D, BT-474, and SK-BR-3) was used. Human normal breast epithelial cell line MCF-10A was used for sphere formation experiment. Above breast cancer cell lines were cultured at 37 °C in DMEM with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA). MCF10A cells were cultured at 37 °C in DMEM-F12 × 1:1 (Gibco) with 10% FBS, including 4 mg/mL insulin (Sigma), 0.4% bovine serum albumin (Sigma), 20 ng/mL EGF (PeproTech), and B-27 (1:50, Gibco).

lncRNA-WAL siRNA kits were purchased from Guangzhou Gene Denovo Biotechnology Corporation. Transfections were performed using Lipofectamine 2000 (Invitrogen, 11668027) according to the manufacture’s instructions with a concentration of 10 nM siRNA. The cells were collected for subsequent molecular biology experiments.

A microarray assay was conducted to examine the lncRNA changes in different TNBC tissue RNA samples, and cyanine-3 (cy3)-labeled cRNA was prepared from 0.5 μg RNA using the One-Color Low RNA Input Linear Amplification PLUS kit (Agilent) according to the manufacturer’s instructions, followed by RNA column purification (QIAGEN, Valencia, CA). Dye incorporation and cRNA yield were checked with the NanoDrop ND-1000 Spectrophotometer. We ran a microarray significance analysis to generate a list of lncRNAs, and differentially expressed lncRNAs were analyzed by random variance model. The raw sequencing data from this microarray have been deposited in the Gene Expression Omnibus under the accession number GSE270306.

The expression of lnc-WAL was detected in paraffin-embedded tissue sections and cell lines by using ISH and fluorescence ISH (FISH) arrays. ISH was performed using a digoxigenin-labeled oligonucleotide lnc-WAL probe (TCA​GCA​CTG​TCA​TCA​TTA​CAT​T; Takara, Japan) according to previous literature (27). For FISH, we used a FISH kit (Ribo, Guangzhou, China) according to the manufacturer’s instructions. In the kit, the 18S probe was used as a cytoplasmic control, and U6 was used as a nuclear control.

IHC was conducted as per the standard protocol according to the literature (28). The protein quantification was evaluated by two independent pathologists.

The ISH and IHC results were evaluated by two blinded individuals using a quick scoring system from 0 to 12 by combining the staining intensity and the positive percentage, in which the staining intensity was recorded on a scale of 0 (no staining), 1 (light staining), 2 (moderate staining), and 3 (strong staining). The positive percentage was graded as follows: 0 (0%), 1 (<25%), 2 (25%–50%), 3 (50%–70%), and 4 (>75%). The results were calculated as follows: staining intensity × positive percentage. This scoring system was based on previous literature (29).

To explore those lncRNAs that bound to β-catenin, we established MDA-MB-231 cells overexpressing β-catenin by transiently transfecting β-catenin/Flag plasmid. Approximately 2 × 10⁷ cells were lysed by using immunoprecipitation (IP) lysis buffer (Invitrogen, 87787), and the obtained lysates were incubated with Flag antibody and Dynabeads Protein G (Invitrogen, 10004D) overnight at 4°C with 360° rotation. RNA was extracted using phenol-alcohol. After the coprecipitated RNAs were extracted, ribosomal RNAs (rRNA) were removed. Then the remaining RNAs were detected by library construction and sequencing at Guangzhou Gene Denovo Biotechnology Corporation. The RNA immunoprecipitation sequencing (RIP-seq) was performed using Illumina Hiseq6000 platform. The raw sequencing data from this RIP-seq have been deposited in the Genome Sequence Archive in BIG Data Center (https://ngdc.cncb.ac.cn/gsa-human/s/A1gz6ISM), National Genomics Data Center (NGDC), under the accession number HRA007309.

To explore the effect of lnc-WAL on the wnt/β-catenin pathway, we performed RNA-sequencing upon lnc-WAL knockdown cells. We knocked down lnc-WAL in MDA-MB-231 cells by transiently transfecting siRNA. The protocol for RNA extraction and detection are the same as detailed for RIP-sequencing. Raw RNA-sequencing data together with RIP-sequencing data have been deposited in the Genome Sequence Archive in BIG Data Center (https://ngdc.cncb.ac.cn/gsa-human/s/A1gz6ISM), National Genomics Data Center (NGDC), under the accession number: HRA007309.

We used the NE-PER Nuclear and Cytoplasmic Extraction Kit (Thermo, 78833) to detect the ratio of lnc-WAL in the nucleus and cytoplasm. The expression of lnc-WAL in the nucleus and cytoplasm was measured by qRT–PCR. MALAT1 was used as a nuclear positive control, and β-actin was used as a cytoplasmic positive control. The associated gene primer sequences are listed in Supplementary Table S1.

Rapid amplification of cDNA ends was conducted using a SMARTer Kit (Takara, 634858) according to the manufacturer’s instructions, and the detailed procedure was referenced in the literature (23). The associated 5′/3′ primer sequences are shown in Supplementary Table S2.

The detailed process of the sphere formation experiment was performed as described in our previous instructions (30). MCF10A cells (800 cells/mL) were cultured in ultralow adhesion plates (Corning) in serum-free Dulbecco's Modified Eagle Medium/Nutrient Mixture F-12 (DMEM-F12) = 1:1 (Gibco), including 4 mg/mL insulin (Sigma), 0.4% bovine serum albumin (Sigma), 20 ng/mL EGF (PeproTech), and B-27 (1:50, Gibco). After culturing for 7 to 10 days according to the design of the experiment, mammospheres with diameters up to 75 μm were harvested, and three independent experiments were conducted in each analysis.

To explore the motor and metastasis abilities of TNBC cells, we used Transwell chambers to imitate the cell barrier. For the invasion assay, a layer of Matrigel (BD Bioscience) was placed above the chamber. The indicated cells were pretreated with silnc-WAL (RiboBio, Guangzhou) for 48 hours and subsequently trypsinized and washed in PBS. Two thousand cells were seeded into the upper chambers in DMEM with 5% FBS, and DMEM with 10% FBS was added to the lower chambers. After approximately 8 hours, the chambers were collected and quantified by photographing three random fields.

A dual-luciferase reporter assay was used to detect the effect of lnc-WAL on β-catenin activity. β-catenin-Luc plus 5 ng pRL-TK-Renilla was transfected into the cells with Lipofectamine 3000 (Invitrogen, L3000-015) in 96-well plates. After 8 hours, siRNAs were transfected into the cells with Lipofectamine 2000 (Invitrogen, 11668027). After another 8 hours, a dual-luciferase reporter assay was performed via the Dual-Luciferase Reporter Assay System (Promega, E1910) as previously described. Luciferase activity was normalized to Renilla luciferase activity for each sample.

A co-IP assay was used to detect the interaction among different proteins. The indicated cells were washed in cold PBS and lysed with IP Lysis Buffer (Thermo, 87787) with protease inhibitor, where the detailed processes were previously described (28). The lysates were immunoprecipitated with primary antibodies for incubation overnight at 4°C, and Dynabeads Protein G (Invitrogen, 10004D) was added to the lysates for another 2 hours at room temperature with 360°C rotation. After washing three times with lysis buffer, the beads were adsorbed by a magnet and resuspended in protein loading buffer. Finally, the complex was analyzed by Western blot (WB).

The animal experiments were approved by the Animal Care and Use Committee of Sun Yat-sen University. We purchased 4-week-old female BALB/c nude mice from GemPharmatech Corporation (Nanjing, China). We established stable lnc-WAL knockout in MDA-MB-231 cells by transducing lentivirus, and vector was used as a negative control. A total of 5 × 10⁶ cells in 100 μL of PBS were injected into the third right mammary fat pads (six nude mice per group). Once tumors reached 5 mm in diameter according to the standard modified formula, volume (mm³) = (length × height²)/2, IWR1 (0.8 µmol/L, Wnt/β-catenin pathway inhibitor) was used to deliver therapy to the tumors once every 3 days in vivo. After 32 days of treatment, the tumors were excised, and tumor tissues were individually preserved for IHC, ISH, RNA, and protein extraction. We also used antisense oligonucleotides (ASO) to target lnc-WAL to explore the functional role of lnc-WAL in the malignant progression of TNBC. The detailed procedure was referenced in the literature (31).

For the metastasis model, we transduced luciferase lentivirus into MDA-MB-231 cells, and 1 × 10⁶ cells were mixed in 200 μL of PBS and intravenously injected into the tail vein (three nude mice per group). The lung, liver, and bone were excised to examine the metastases using the IVIS Lumina Imaging System.

To explore the effect of lnc-WAL on overall survival in TNBC in vivo, we performed KM survival experiment with 10 nude mice per group in three groups (shvector, shlnc-WAL^#1, and shlnc-WAL^#2), and 1 × 10⁶ cells were mixed in 200 μL of PBS and intravenously injected into the tail vein. We observed the status of these mice every 3 days until 2 months and drew the KM overall survival curve by using GraphPad 5.0.

All data are expressed as the mean ± SD values, and P values were calculated by the two-tailed Student t test or one-way ANOVA followed by GraphPad Prism 5. Significant differences (*, P ≤ 0.05; **, P ≤ 0.01; ***, P ≤ 0.001) are indicated. Each independent in vitro experiment included three independent replicates. Error bars reflect independent experiments.

The Wnt/β-catenin pathway was abnormally activated in TNBC, and the reason remains unclear (32). We detected the location and expression of β-catenin by IHC in our cohort of BC tissue samples. Nuclear and cytoplasmic β-catenin was most prominent in TNBC, and membrane β-catenin was observed predominantly in luminal A, luminal B, and HER2+ tumors (Figs. 1A and B), consistent with original reports. To detect the clinical significance of its activation in TNBC, the location of β-catenin was detected by IHC in 189 samples of patients with TNBC. The location of β-catenin was used to divide these patients into activation (β-catenin^{Cytoplasm+Nuclear}) and inactivation (β-catenin^Membrane) subgroups (Supplementary Fig. S1A and S1B). Survival analysis revealed that patients in the activation subgroup had poorer overall survival (OS) and disease-free survival (DFS) than those in the inactivation subgroup (Supplementary Fig. S1C and S1D). These results indicated that the wnt/β-catenin pathway may be a key factor in TNBC progression.

In TNBC, to further identify Wnt/β-catenin pathway activation-associated lncRNAs, we carried out one lncRNA tissue microarray to compare the differences between the activation group (n = 5) and inactivation group (n = 5) of TNBC tissues. We identified 56 lncRNAs that were upregulated and 28 that were downregulated by more than fourfold (Supplementary Table S3). Among them, lnc-WAL, an lncRNA encoded by a gene at chromosome 19q13, was consistently found to be the first upregulated lncRNA in the wnt/β-catenin pathway activation group by microarray profiling (Fig. 1C). To screen the lncRNAs that bound to β-catenin, we also detected the β-catenin location in cell lines by IF. The localization of β-catenin was prominent in the nucleus in TNBC cells compared with other subtypes (Fig. 1D; Supplementary Fig. S1E). Next, we transiently transfected the Flag/β-catenin plasmid into MDA-MB-231 cells to overexpress β-catenin. The working efficiency of the Flag/β-catenin plasmid was tested by detecting active β-catenin and total β-catenin expression (Fig. 1E; Supplementary Fig. S1F). We screened the lncRNAs that bound to β-catenin between Flag/β-catenin and IgG in MDA-MB-231 cells by RIP-seq, and we also discovered that lnc-WAL was the strongest combination of β-catenin compared with IgG (Fig. 1F). We further verified RIP-seq by RIP-qRT–PCR in MDA-MB-231 cells (Supplementary Fig. S1G). In clinical samples of patients with BC, we examined the expression and location of β-catenin by IHC and the expression of lnc-WAL by ISH (Fig. 1G). We also performed statistical analysis of the correlation between β-catenin location and lnc-WAL expression in 189 clinical TNBC samples and found that the expression of lnc-WAL was higher in the activation samples than in the inactivation samples (Supplementary Fig. S1H), and there was a positive relationship between the expression of lnc-WAL and β-catenin (Fig. 1H). In MCF7 and T47D (Wnt/β-catenin pathway inactivation cell lines), qRT–PCR and NB arrays confirmed that lnc-WAL was upregulated when the cells were treated with LICL (Wnt/β-catenin pathway agonist; Figs. 1I and J), and the expression of lnc-WAL in response to LICL was time and dose dependent (Supplementary Fig. S1I and S1J). We tested the work efficiency of Wnt/β-catenin pathway agonists, including LICL and AZD2858, by detecting active β-catenin expression (Supplementary Fig. S1K). Meanwhile, we detected lnc-WAL expression when β-catenin was knocked down and found that lnc-WAL expression was inhibited upon β-catenin knockdown (Supplementary Fig. S1L and S1M). All these data indicated that lnc-WAL was a Wnt/β-catenin pathway activation-associated lncRNA, and the functional characteristics of lnc-WAL have not been reported in previous research.

Given that lnc-WAL was associated with Wnt/β-catenin signaling pathway activation in TNBC, we inspected the expression of lnc-WAL in different subtypes of BC, in which the level of lnc-WAL in TNBC was significantly elevated compared with luminal and HER2+ patients with BC by qRT–PCR (Fig. 2A). Moreover, high levels of lnc-WAL were associated with Wnt/β-catenin signaling activation by qRT‒PCR (Fig. 2B). Meanwhile, lnc-WAL expression in TNBC cell lines was significantly elevated compared with that in the other subtypes of BC cell lines by qRT‒PCR and NB (Figs. 2C and D).

In this study, after the nuclear and cytoplasmic fractionations of cells, qRT‒PCR and RT‒PCR showed that lnc-WAL mainly localized in the cytoplasm (Figs. 2E and F), which we also verified by RNA FISH with 18S and U6 as cytoplasmic and nuclear controls (Fig. 2G). Lnc-WAL was mainly localized in the nucleus, and its expression was significantly increased upon the addition of Wnt/β-catenin pathway agonists, including Lithium Chloride (LICL), IC261, SKL2001, and AR2000, and was inhibited upon the addition of a Wnt/β-catenin pathway inhibitor, including IWR1 (Fig. 2H). We amplified lnc-WAL in the UCSC Genome Browser, where we performed 5′ and 3′ rapid amplification of cDNA ends assays and found that the full length of lnc-WAL was 4,300 nt (Supplementary Fig. S2A). These results suggest that the expression of lnc-WAL is higher in TNBC than in other subtypes, especially in TNBC with Wnt/β-catenin signaling activation.

We detected lnc-WAL using ISH in 189 TNBC, 120 luminal, and 120 HER2+ tissue samples, and these patients were divided into lnc-WAL^High and lnc-WAL^Low subgroups by the staining intensity and positive percentage of lnc-WAL. We analyzed the relationship between lnc-WAL and the clinicopathological features of 189 TNBC tissue samples and found that higher lnc-WAL expression was significantly related to tumor size, metastasis and local recurrence, Ki67 level, and Wnt/β-catenin pathway activation status but was not associated with age, menopausal status, or histological grade (Supplementary Table S4). Kaplan‒Meier analysis indicated that patients with TNBC with high lnc-WAL had poorer OS and DFS than those with low lnc-WAL expression (P < 0.001; Figs. 2I and J). Univariate and multivariate Cox proportional hazard analyses showed that lnc-WAL (P < 0.001) was a poor independent prognostic factor for OS and DFS (Supplementary Table S5 and S6). In luminal and HER2+ subtypes of patients with BC, we did not find the same prognostic value of lnc-WAL (Supplementary Fig. S2B and S2C). These results suggested that lnc-WAL levels are closely related to the malignant development of TNBC and can be used to predict the outcomes of OS and DFS in TNBC.

Next, multiple datasets were retrieved. In the Gene Expression Profiling Interactive Analysis (GEPIA) datasets, higher lnc-WAL levels were associated with poorer OS (P = 0.03) and DFS (P = 0.0067) in BRCA (Supplementary Fig. S2D and S2E). In the KM plot TNBC dataset, higher lnc-WAL levels were associated with poorer DFS (P = 0.034; Supplementary Fig. S2F). Strikingly, in multiple GEPIA cancer datasets, high lnc-WAL levels were significantly associated with poor survival in liver, stomach, and lung cancers (Supplementary Fig. S2G–S2J). Additionally, in the stomach cancer KM plot, higher lnc-WAL levels were associated with poorer OS (P = 7.6e−09; HR = 1.91; 95% CI, 1.53–2.38; Supplementary Fig. S2K). The above publicly available datasets were consistent with our survival data of BRCA samples.

This result shows that lnc-WAL plays a key role in maintaining the TNBC cell phenotype. The special characteristics of TNBC are epithelial–mesenchymal transition (EMT) and the relatively high percentage of the CD44⁺/CD24⁻ cell population. To directly assess the role of lnc-WAL in EMT, siRNAs and overexpression plasmids against lnc-WAL were transiently transfected into cells, in which we found that E-cad (an epithelial cell marker) expression was increased or inhibited upon loss or gain of function of lnc-WAL by IF and WB and vice versa for vimentin (a mesenchymal cell marker; Figs. 3A and B; Supplementary Fig. S3A and S3B).

To directly evaluate the role of lnc-WAL in BC stem cell (BCSC), we used immortalized human mammary epithelial MCF-10A cells to detect the influence of lnc-WAL on the proportion of BCSCs (33). We found that lnc-WAL-depleted cells generated smaller and fewer mammospheres than siNC-treated cells, whereas lnc-WAL promoted the mammosphere-forming capacity of MCF-10A cells (Fig. 3C; Supplementary Fig. S3C). As expected, lnc-WAL further increased the BCSC-enriched CD44^high/CD24^low cell population, as shown by flow cytometry (Fig. 3D), and lnc-WAL increased CD44 and reduced CD24 expression, as shown by Western blotting (Fig. 3E; Supplementary Fig. S3D). Next, we isolated a CD44⁺/CD24⁻ cell population and a non-CD44⁺/CD24⁻ cell population from MCF-10A cells and found that the expression of lnc-WAL was higher in the CD44⁺/CD24⁻ cell population than in the non-CD44⁺/CD24⁻ cell population (Supplementary Fig. S3E and S3F). To further assess whether the gain of BCSC properties of Wnt/β-catenin-associated lncRNA (WAL)-overexpressing MCF-10A cells could strengthen tumorigenicity, we carried out an extreme-limiting dilution assay in nude mice. WAL-overexpressing and vector MCF-10A cells were injected in limiting dilutions subcutaneously into nude mice. A total of 1 × 10⁴ WAL-overexpressing cells formed tumors in four of six mice, but no tumors formed in the 1 × 10⁴ WAL-vector cell group (Figs. 3F and G). Thus, lnc-WAL potentially drives the expansion and tumorigenicity of BCSCs in vitro and in vivo.

Here, we also found that lnc-WAL promoted cell proliferation, colony formation, migration, and invasion. The abovementioned functional experiments were carried out by using loss or gain of function of lnc-WAL. The proliferation indices (optical density values) of cells were significantly decreased in both silnc-WAL^#1 and silnc-WAL^#2 cells compared with siNC cells and increased in WAL-overexpressing cells compared with WAL-vector cells (Fig. 3H; Supplementary Fig. S3G). lnc-WAL knockdown also inhibited the colony formation capacity of cells on day 14 after siRNA transfection, and WAL overexpression increased the colony formation capacity of MCF-7 cells (Fig. 3I; Supplementary Fig. S3H and S3I). Both cell migration and invasion were significantly reduced in both silnc-WAL^#1- and silnc-WAL^#2-treated cells compared with siNC-treated cells by migration and invasion assays, and WAL overexpression increased the migration and invasion capacities of MCF-7 cells (Fig. 3J; Supplementary Fig. S3J and S3K). The knockdown and overexpression efficiencies of lnc-WAL were determined by qRT‒PCR (Supplementary Fig. S3L). These results indicate that lnc-WAL can promote the malignant progression of TNBC.

Lnc-WAL was screened from those lncRNAs that bound to β-catenin by RIP-seq. To verify whether lnc-WAL combined with β-catenin, we performed RIP–PCR in endogenous and exogenous systems. We carried out an incubation assay between purified β-catenin and lnc-WAL in vitro to verify the direct interaction of β-catenin with WAL. In addition, we also performed an RNA pull-down assay to prove the interaction. Our results indicated that lnc-WAL specifically interacted with β-catenin (Figs. 4A–D). Meanwhile, FISH and IF staining of lnc-WAL and β-catenin demonstrated that there was prominent colocalization between lnc-WAL and β-catenin in the different cell lines (Fig. 4E). RNA pull-down assays indicated that lnc-WAL mutants containing nucleotides 1,500 to 3,000 retained the capability to bind to β-catenin as efficiently as full-length lnc-WAL (Fig. 4F).

Translocation of β-catenin into the nucleus is mediated by β-catenin phosphorylation via a destruction complex that consists of GSK3, AXIN, APC, and CK1α (34, 35). To explore the role of lnc-WAL in wnt/β-catenin signaling activation, after knocking down lnc-WAL, we found that β-catenin phosphorylation was increased and active and that total β-catenin was decreased (Fig. 4G; Supplementary Fig. S4A). We further verified the colocalization between lnc-WAL and β-catenin protein in the cytoplasm, with subsequent knockdown of lnc-WAL, which decreased β-catenin nuclear translocation, by using FISH and IF (Fig. 4H; Supplementary Fig. S4B). We detected the knockdown efficiency of siRNA by qRT‒PCR (Supplementary Fig. S4C). To further explore the effects of lnc-WAL on β-catenin, we detected the effects of lnc-WAL on the protein stability of β-catenin by a half-life assay with or without CHX (0.1 μg/mL), in which the half-life of β-catenin protein was calculated with and without silnc-WAL. It was found that active, total-β-catenin protein levels decreased by ∼50% within 7 and 7.5 hours in siNC-treated cells. However, active, total-β-catenin protein levels decreased by ∼50% within 3 and 3.5 hours in the silnc-WAL-treated cells. The half-life of active, total-β-catenin protein was significantly shortened to 3 and 3.5 hours (Fig. 4I; Supplementary Fig. S4D and S4E). These data indicated that lnc-WAL increased β-catenin protein stability and prolonged its half-life. To evaluate the effect of lnc-WAL on the Wnt/β-catenin pathway, we performed RNA sequencing upon lnc-WAL knockdown. Gene Set Enrichment Analysis signaling pathway analysis indicated that depletion of lnc-WAL could impair the Wnt/β-catenin pathway (Fig. 4J). Meanwhile, we detected lnc-WAL expression upon β-catenin overexpression, and we found that the expression of lnc-WAL was increased along with β-catenin overexpression by Northern blot, FISH, and qRT–PCR (Figs. 4K and L; Supplementary Fig. S4F). The above results indicate that lnc-WAL activates the Wnt/β-catenin pathway by specifically combining and inhibiting the phosphorylation and subsequent nuclear translocation of β-catenin.

It is well known that β-catenin protein is phosphorylated by kinase destruction complexes, including GSK3, AXIN, APC, and CK1α (36). To identify potential kinases involved in lnc-WAL-induced inhibition of β-catenin phosphorylation, we performed rescue assays in vitro using different kinase inhibitors. We found that SKL2001 (AXIN inhibitor) could rescue the inhibition of β-catenin induced by silnc-WAL. However, the CK1α inhibitor (IC261) and GSK3 inhibitor (LICL and AR2000) could not rescue the inhibition of β-catenin (Fig. 5A; Supplementary Fig. S5A and S5B). To further confirm that lnc-WAL inhibited β-catenin phosphorylation by the kinase AXIN, we silenced AXIN by siRNA after silencing WAL. Subsequently, lnc-WAL-induced inhibition of β-catenin phosphorylation was abrogated when AXIN was silenced by two siRNAs (Fig. 5B; Supplementary Fig. S5C and S5D). On the other hand, due to the phosphorylation site of β-catenin at serine 45 (S45) by AXIN (36), point mutations at the AXIN binding site (S45A) of β-catenin rescued silnc-WAL-induced phosphorylation of β-catenin (Fig. 5C; Supplementary Fig. S5E and S5F). The siRNA sequences in vitro are shown in Supplementary Table S7.

To further understand whether lnc-WAL disturbs the binding of β-catenin and its kinase AXIN, we performed co-IP and found that knockdown of lnc-WAL enhanced the interaction between β-catenin and AXIN (Fig. 5D). As it has been verified that binding between β-catenin and AXIN induces the phosphorylation of β-catenin and reduces the nuclear accumulation of β-catenin (37), we further explored whether the activation of the Wnt/β-catenin pathway occurs through lnc-WAL to block the binding between β-catenin and AXIN. To this end, we knocked down lnc-WAL and found that it could activate the Wnt/β-catenin pathway by a luciferase reporter assay. LICL-treated cells were utilized as a positive control (Fig. 5E). We examined whether lnc-WAL affects downstream genes of the Wnt/β-catenin pathway, where the expression levels of c-myc, VEGF, cyclinD1, MMP7, MMP9, β-catenin, and CD44 were decreased in cells where lnc-WAL was knocked down and overexpressed by WB (Fig. 5F; Supplementary Fig. S5G) and qRT‒PCR (Fig. 5G; Supplementary Fig. S5H). We performed rescue experiments to further verify whether lnc-WAL promotes malignant progression by activating the wnt/β-catenin pathway. SKL2001 (a Wnt/β-catenin pathway agonist) rescued the inhibitory effects upon WAL knockdown, including migration and Mesenchymal to Epithelial Transition (MET), and IWR1 (a Wnt/β-catenin pathway inhibitor) inhibited the promoting effects upon WAL overexpression, including migration and EMT (Fig. 5H; Supplementary Fig. S5I). Finally, we used many Wnt/β-catenin pathway agonists, including LICL, IC261, SKL2001, and AR2000. We detected the expression of total β-catenin and active β-catenin by WB and qRT‒PCR, and the results showed that Wnt/β-catenin pathway agonists could activate the Wnt/β-catenin pathway (Supplementary Fig. S5J). Therefore, the results provide further evidence that lnc-WAL inhibits the phosphorylation of β-catenin by disturbing the binding of β-catenin and its kinase AXIN, which then induces active β-catenin nuclear translocation, and lnc-WAL plays a role as an oncogenic gene by activating the Wnt/β-catenin pathway.

To evaluate the therapeutic potential of targeting lnc-WAL in vivo, the indicated MDA-MB-231 cells (shvector, shlnc-WAL^#1, and shlnc-WAL^#2) were subcutaneously injected into 4-week-old female nude mice, IWR1, which works by stabilizing AXIN, was intraperitoneally injected (Fig. 6A), and the efficiency of shlnc-WAL^#1 or shlnc-WAL^#2 was up to 80% by using qRT–PCR (Supplementary Fig. S6A). After 4 weeks of injection, we found that both the volume and weight of tumors derived from the lnc-WAL-depleted group were clearly reduced compared with those of tumors derived from the vector group. More importantly, the volume and weight of tumors derived from lnc-WAL depleted together with IWR1 injection groups were further reduced compared with tumors derived from other groups (Figs. 6B–D). Moreover, H&E staining, ISH of lnc-WAL and IHC of β-catenin, Ki67, and immunofluorescence staining of terminal deoxynucleotidyl transferase–mediated dUTP nick end labeling assay showed that lnc-WAL knockdown in vivo reduced cell proliferation and β-catenin activation and increased apoptosis of tumor cells, whereas the effects were more significant in the groups (lnc-WAL depleted together with IWR1; Fig. 6E; Supplementary Fig. S6B). We also examined the expression of lnc-WAL and target genes and found that the levels of target genes were decreased in tumor samples derived from the therapeutic groups compared with the untreated group (Fig. 6F; Supplementary Fig. S6C and S6D). To evaluate the clinical translational significance of targeting lnc-WAL and the Wnt/β-catenin pathway, we established another xenograft nude mouse model by using ASOs. We subcutaneously injected MDA-MB-231 cells into nude mice. When the tumor diameter reached 5 mm (n = 6/group), we established a therapeutic model by intravenous injection of WAL-ASOs with or without IWR1 (a Wnt/β-catenin pathway inhibitor). Tumor growth was significantly inhibited upon combination of WAL-ASOs with IWR1, whereas the mice treated with control WAL-ASOs continued to grow into large tumors (Supplementary Fig. S6E).

The metastasis rate of TNBC is up to 20% (2). Therefore, we established a mouse metastasis model using luciferase (Luc)-expressing MDA-MB-231 cells to further validate the metastasis-promoting ability of lnc-WAL (Fig. 6G). We established stable knockout lnc-WAL by using two shRNAs in Luc-expressing MDA-MB-231 cells, and 10⁶ cells described above per mouse (shvector, shlnc-WAL^#1, and shlnc-WAL^#2) were intravenously injected into 4-week-old female nude mice (n = 3). After 2 months, all three mice developed obvious lung metastasis, one of which also had obvious bone metastasis in the knee joint. In contrast, the mice injected with shlnc-WAL^#1 and shlnc-WAL^#2 cells did not develop any metastasis (Fig. 6H). Immunohistochemistry (IHC) staining of lung and knee joint serial sections also indicated that lnc-WAL promoted tumor metastasis, as demonstrated by high lnc-WAL expression, more active β-catenin expression, more proliferation (Ki67), and less apoptosis (caspase 2; Fig. 6I; Supplementary Fig. S6F). The expression of lnc-WAL was detected in the lung and knee joint tissues by using qPCR, and the results showed that lnc-WAL was highly expressed in the tumor metastasis tissues of control group (Supplementary Fig. S6G). Moreover, silencing lnc-WAL in MDA-MB-231 xenografts extended mouse survival (Supplementary Fig. S6H). All antibodies and reagents are shown in Supplementary Table S8.

These results suggest that lnc-WAL is a promising therapeutic candidate that controls tumor growth and distant metastasis by regulating the Wnt/β-catenin signaling pathway in vivo.

Our study revealed the functional role of lnc-WAL in the malignant progression of TNBC by activating Wnt/β-catenin signaling, and lnc-WAL was positively related to the progression of TNBC. Regarding the molecular mechanism, we found that lnc-WAL specifically perturbed the combination of AXIN with β-catenin at serine 45 to translocate the signaling protein β-catenin into the nucleus. Under LICL stimulation, lnc-WAL expression was upregulated in MCF-7 and T47D cells. Therefore, under the condition of Wnt/β-catenin signaling hyperactivation, such as TNBC, lnc-WAL may function as a positive regulator of Wnt/β-catenin signaling. A positive feedback loop was formed between Wnt/β-catenin signaling and lnc-WAL.

The Wnt/β-catenin signaling pathway is a vital regulator in the development of all kinds of species. Whether in growth-related disease or different types of cancer, Wnt/β-catenin pathway is crucial, which suggests that it is a noticeable target for disease therapy (34, 38–41). However, there are many side effects if directly targeting the Wnt/β-catenin pathway is used as a cancer therapeutic strategy. The dephosphorylation of β-catenin by destruction complexes (AXIN, GSK3, and CK1) and accumulation in the nucleus have been well recognized as a crucial part of Wnt/β-catenin signaling activation (42, 43). To explore the molecular mechanism of Wnt/β-catenin signaling activation in TNBC, we performed a novel screening of the differentially expressed lncRNAs between activation and inactivation. Meanwhile, we searched for lncRNAs that bind to β-catenin by RIP-seq and identified lnc-WAL as the most prominent lncRNA using the two methods mentioned above.

It has been reported that many lncRNAs regulate the Wnt/β-catenin signaling pathway. For example, lnc-CRNDE promoted colon cancer cell malignant evolution by inhibiting miR-181a-5p expression, which indirectly influenced Wnt/β-catenin signaling (44). By targeting miR-125b and miR-100, MIR100HG promotes cetuximab resistance by inhibiting negative regulators of the Wnt/β-catenin pathway (25). In hepatobiliary carcinogenesis, activation of the Wnt/β-catenin pathway promoted the expression of lncRNA uc.158, and it may act as an endogenous competing lncRNA for the pro-apoptotic miR-193b, resulting in cancer cell survival (45). However, these articles only indicated a positive or negative relationship between lncRNAs and the wnt/β-catenin pathway. Interestingly, lnc-00673-v4 acted as a scaffold to directly combine DDX3 with CK1 and then promoted the aggressiveness of lung adenocarcinoma (23); lnc-CCAT1 could bind miR-204/211/148a/152 and inhibit their expression, leading to the upregulation of TCF4 and DNMT1, subsequently initiating Wnt/β-catenin signaling in BC (46). Nevertheless, it is unclear whether there is a functional relationship among lncRNAs, the signal protein β-catenin, and the kinase destruction complex (AXIN, GSK3, and CK1), especially in TNBC.

Unlike previously reported lncRNAs, lnc-WAL could directly bind signal protein β-catenin in the cytoplasm by RNA pull-down, RIP, and RNA FISH experiments, in which it inhibited β-catenin phosphorylation by AXIN kinase at serine 45 by utilizing kinase inhibition and point mutation assays. Co-IP assays demonstrated that lnc-WAL knockdown significantly enhanced β-catenin and AXIN interactions. Finally, lnc-WAL activated the Wnt/β-catenin pathway and promoted targeted gene production, as shown by luciferase reporter assays. Of note, we think it is not impossible that lnc-WAL activates the Wnt/β-catenin signaling pathway through more than one mechanism because β-catenin is phosphorylated by different kinases at different phosphorylation sites, in addition to the activation of β-catenin being a dynamic process from the cytoplasm to nucleus. Hence, it is beyond question that there will be different lncRNAs involved in different steps. Thus, whether there are any relationships among these lncRNAs that regulate the Wnt/β-catenin pathway through the above putative mechanism also requires additional studies.

Increasing evidence has demonstrated that differential expression levels of lncRNAs could be correlated with clinical prognosis in various cancers, which supports the possibility of lncRNAs as a targeted therapy in disease. An increasing number of studies have indicated that lncRNAs could be valuable biomarkers and promising drug therapeutic targets for cancer (31). Here, we found that lnc-WAL was significantly upregulated in TNBC tissue samples, especially in Wnt/β-catenin signaling activation samples. The ectopic expression of lnc-WAL in TNBC tissue samples was closely associated with metastasis and relapse in vitro and in vivo. Our results indicated that lnc-WAL might be a promising therapeutic target for cancer. Recently, siRNAs and ASOs that target lncRNAs have been under intensive investigation and create promising treatment opportunities in vivo (47, 48). In our present study, we used siRNAs and ASOs to effectively silence lnc-WAL in vitro and in vivo. Together, our current study provides a clue that targeting the Wnt/β-catenin-associated lncRNA-WAL could possibly be achieved in anticancer therapy.

In summary, lnc-WAL is a newly reported lncRNA that regulates β-catenin phosphorylation by the kinase AXIN (Supplementary Fig. S6I). This finding not only amplifies the regulatory mechanisms for Wnt/β-catenin signaling but also draws attention to many lncRNAs that exert their function by influencing the posttranslational modification of signaling proteins. We conclude that lnc-WAL is indispensable in the Wnt/β-catenin signaling activation of TNBC and that the regulatory role in this pathway is crucial for the aggressive features of TNBC. lnc-WAL could become a new therapeutic target for patients with TNBC, and there will be fewer side effects compared with targeting Wnt/β-catenin signaling. We will also design more experiments and translational research to confirm the advantage of lnc-WAL in the treatment of TNBC.

No disclosures were reported.

H. Huang: Conceptualization, resources, data curation, funding acquisition, methodology, writing–original draft. H. Jin: Conceptualization, data curation, methodology. R. Lei: Conceptualization, resources. Z. He: Conceptualization, resources. S. He: Conceptualization, resources. J. Chen: Conceptualization, resources. P.E. Saw: Writing–review and editing. Z. Qiu: Resources, data curation. G. Ren: Project administration, writing–review and editing. Y. Nie: Investigation, project administration, writing–review and editing.

We would like to thank Genedenovo Biotechnology Co., Ltd. (Guangzhou, China), for the RIP-seq services and bioinformatics analysis. We are particularly grateful to Professor Erwei Song and Professor Hai Hu from SYSMH for supervision of the study design. This study was supported by the Natural Science Foundation of China (No. 82002782, 82173054, 81872158), Guangdong Science and Technology Department (2022B1515020048), and Guangzhou Science Technology and Innovation Commission (202102010148).