Patient tissue specimens were collected in The First Affiliated Hospital of USTC. Before collecting the tissues, written informed consent was acquired from volunteering patients. Additionally, ethical approval was granted from the Institutional Research Ethics Committee. All operations followed the guidelines of the ethics committee of The First Affiliated Hospital of USTC, as well as the Declaration of Helsinki (2008) of the World Medical Association and its subsequent amendments or equivalent ethical standards. The information of clinicopathological characteristics of ER⁺ breast cancer specimens are shown in Supplementary Table S1 .

The human breast cancer cell lines MCF-7 and T47D were obtained from FuHeng Cell Center (Shanghai, China), and DMEM medium (Gibco, California, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, California, USA) and 1% penicillin/streptomycin antibiotic (Gibco, California, USA) were employed to culture the cells. All cell cultures were maintained at 37°C in a constant-humidified incubator containing 5% CO₂.

The LINC01133-overexpressing plasmid (LINC01133) and LINC01133-knockdown lentiviral plasmid (sh#1 and sh#2) were prepared by Umine Biotechnology Co., LTD (Guangzhou, China). Subsequently, we accomplished transfection of the plasmids using the Lipofectamine 3000 reagent (Invitrogen, Carlsbad, USA) following the instructions. The lentiviral plasmids were used to package the viral particles. Finally, we infected the cells with the virus and screened the stable cell lines using 1 μg/ml of puromycin (Gibco, California, USA) for 10 days.

Total RNA extraction from tissues or cells was performed using TRIzol reagent (Invitrogen, Carlsbad, USA) following the instructions. PrimeScript™ RT reagent Kit (Takara, Kyoto, Japan) was applied to synthesize the cDNA. qRT‐PCR was completed using the SYBR Premix Ex Taq kit (Takara, Kyoto, Japan). We used GAPDH as the internal control to calculate mRNA levels. The RNA primers are as follows: LINC01133, forward, 5′-CCTAATCTCACCACAGCCTGG-3′, reverse, 5′-TCAGAGGCACTGATGTTGGG-3′; GAPDH, forward, 5′-GTCTCCTCTGACTTCAACAGCG-3′, reverse, 5′-ACCACCCTGTTGCTGTAGCCAA-3′.

ISH was performed in paraffin-embedded ER⁺ breast cancer samples to estimate LINC01133 expression according to the method previously described (21). The sequence of biotin-labeled LINC01133 probe is as follows: 5′-GGAGGTAAAGAGTAGAAGACAGTATCAAGAATCCAGAG-3′.

First, cells were cultivated using six-well plates for 24 h. Subsequently, 5 μg/ml of actinomycin D (Cat: HY-17559; Med Chem Express, Monmouth Junction, NJ, USA) was used to treat cells to inhibit gene transcription. RNA extraction was completed, and examined by RT-qPCR assay. Finally, RNA levels of different groups were measured at different time points. In addition, the levels of GAPDH mRNA also served as controls.

All animal experiments were priorly approved by the institutional animal use and care committee of The First Affiliated Hospital of USTC and conducted in line with the protocol established by this committee. Female 4-week-old BALB/c mice were obtained from Zhuhai BesTest Bio-Tech Co., Ltd. (Zhuhai, China), fed in a pathogen-free room at 22°C with 60% relative humidity, and subjected to a 12-h light/dark cycle.

For tumor growth analysis, 5 × 10⁶ cells were subcutaneously injected into the flank of each mouse. For metastatic potential analysis, every mouse tail vein was injected with 5 × 10⁶ cells. Tumor volumes were evaluated every 3 days following the equation (L × W²)/2, where L represents the longest diameter, and W indicates the shortest diameter perpendicular to L.

RIP, RNA pull-down, and mass spectrometry were accomplished following the methods reported previously (8).

RIP assays were performed using the EZ-Magna RIP kit (Merck, Darmstadt, Germany) following the manufacturer’s protocol. Briefly, 2 × 10⁷ prostate cells were lysed in RIPA buffer (Beyotime, China) supplemented with protease and RNase inhibitors. Ten percent of the lysate was reserved as pre-immunoprecipitation input. Beads were washed three times with RIP buffer, resuspended in 500 μl of RIP buffer, and incubated with 5 μg of anti-GTF2F2 antibody (#MA5-44917, Thermo Fisher Scientific, Waltham, MA, USA) at 4°C for 8 h. After centrifugation at 3,000 rpm for 2 min to remove supernatants, 300 μl of cell lysate was added to the antibody-bound beads and incubated overnight at 4°C. IgG served as the negative control. Immunoprecipitated RNAs were subsequently eluted and quantified by qRT-PCR.

RNA pull‐down assay was performed following the protocol. Briefly, 1 × 10⁷ cells were lysed on ice for 10 min in 500 μl of polysome extraction buffer [100 mM KCl, 5 mM MgCl₂, 5% NP-40, and 20 mM Tris-HCl (pH 7.4)]. The supernatant was collected and incubated with 1–2 μg of biotin-labeled probes at room temperature for 30 min to allow RNA–protein complex formation. Pre-treated streptavidin magnetic beads (BioLabs, Massachusetts, USA) were then added to the reaction mixture and incubated at room temperature for an additional 30 min. Following six washes with RNA washing buffer, the precipitated RNA–protein complexes were analyzed by Western blotting. Bound proteins were subsequently eluted and analyzed using mass spectrometry (LTQ Orbitrap Velos Pro, Thermo Scientific, Waltham, MA, USA).

Total protein extraction from cells or tissues was accomplished using RIPA (Sigma-Aldrich, St. Louis, MO, USA). We used the Pierce™ BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA) to evaluate the concentration of protein. To separate the protein, 10% SDS-PAGE gel (BeyoGel™ Plus PAGE, Beyotime, Shanghai, China) was employed. The process of Western blotting assay was executed following an established protocol (22). In this research, the primary antibodies employed were anti-IGF2BP2 (Cat: MA5-44917; Thermo Fisher Scientific, Waltham, MA, USA) and anti-GAPDH (Cat: 437000, Invitrogen, Carlsbad, USA). GAPDH was employed as an internal loading control, which can normalize and verify equal protein loading among all experimental samples.

To research the functions of LINC01133 on breast cancer in vitro, we performed some cellular functional assays including colony formation, CCK-8, EdU, and Transwell following previously established protocols (23).

All experiments were repeated more than three times. We accomplished the statistical analysis via SPSS 23.0 software (IBM, USA). The mean ± standard deviation was employed to display the statistical analysis results. Differences between paired tissues were analyzed using Student’s paired t-test, whereas comparisons among more than two groups were examined via variance (ANOVA), along with Dunnett’s test. A value of p below 0.05 was considered statistically significant.

To assess LINC01133 levels in breast cancer, we initially analyzed the data derived from The Cancer Genome Atlas (TCGA) database. As exhibited in Figures 1a–c , LINC01133 was significantly downregulated in tumor tissues compared to their matched adjacent normal breast tissue (ANT; Figure 1a ). In addition, LINC01133 levels in breast cancer tissues are much lower than those in normal breast tissues ( Figure 1b ). Later, we compared LINC01133 expression in ER⁻ and ER⁺ breast cancers. The results revealed a progressive decrease in LINC01133 levels from normal breast tissues, ER⁻ breast cancer tissues, to ER⁺ breast cancer tissues ( Figure 1c ). Given the notable reduction in ER⁺ breast cancer tissues, we then investigated its expression and role in this subtype. Likewise, real-time PCR assay uncovered that LINC01133 was robustly decreased in ER⁺ breast cancer (BCa, n = 12) relative to corresponding adjacent normal breast tissues (ANT, n = 12) ( Figure 1d ). To test whether LINC01133 is linked to metastasis of ER⁺ breast cancer, its expression was examined in 14 cases of ER⁺ breast cancer with metastasis (BCa/M, n = 14) and 12 cases of ER⁺ breast cancer without metastasis (BCa/nM, n = 12). The results revealed that LINC01133 was significantly abnormally downregulated in BCa/M compared with BCa/nM ( Figure 1e ). Consistently, ISH assay illustrated a significant decrease in LINC01133 in ER⁺ breast cancer specimens compared to ANT, and in BCa/M compared to BCa/nM, respectively ( Figures 1f, g ). Additionally, low LINC01133 predicted shorter overall survival and metastasis-free survival compared to those with high LINC01133 in ER⁺ breast cancer patients ( Figure 1h ).

Collectively, these findings demonstrated that a notable reduction in LINC01133 might play an indispensable function in the breast cancer progression, especially in ER⁺ breast cancer.

To elucidate the roles of LINC01133 in ER⁺ breast cancer, researchers performed a gene set enrichment analysis (GSEA) on ER⁺ breast cancer samples sourced from TCGA. GSEA analysis revealed a notable trend of proliferation traits focused on the subset of ER⁺ breast cancer exhibiting low levels of LINC01133 ( Figure 2a ) suggesting that LINC01133 might mediate the proliferation of ER⁺ breast cancer. Subsequently, a stable cell line upregulating LINC01133 was constructed using MCF-7 cells ( Figure 2b ). Conversely, a stable cell line exhibiting low levels of LINC01133 was established using T47D cells ( Figure 2b ). Functional experiments of cell proliferation, including CCK8 ( Figure 2c ), colony formation ( Figure 2d ), and EdU ( Figure 2e ), demonstrated that inhibition of LINC01133 notably promoted, while overexpression of LINC01133 suppressed, the proliferation of ER⁺ breast cancer cells. Additionally, we subcutaneously injected luciferase-labeled MCF-7 cells with or without LINC01133 overexpression into BALB/c nude mice. All mice were euthanized at 28 days. Downregulation of LINC01133 significantly inhibited tumor growth, as indicated by the decreased volume and weight ( Figure 2f ).

The GSEA assay indicated that LINC01133 may also be involved in tumor metastasis ( Figure 2a ), which was initially examined in vitro. Transwell assay demonstrated that the downregulation of LINC01133 importantly strengthened the migration and invasion capabilities of ER⁺ breast cancer cells, whereas LINC01133 upregulation substantially reduced these abilities ( Figures 2g, h ). Subsequently, the function of LINC01133 in cancer metastasis was assessed in vivo. Luciferase-labeled MCF-7 cells were introduced to the tail vein of the mice, with or without LINC01133 overexpression. Notably, bioluminescence imaging and H&E staining demonstrated that the overexpression of LINC01133 effectively hindered lung metastasis ( Figures 2i–k ).

Collectively, the abovementioned findings uncovered that LINC01133 regulates the aggressiveness of ER⁺ breast cancer.

Since the localization of lncRNA is critical to their underlying mechanisms (24), we detected the localization of LINC01133 applying subcellular fractionation and RNA-FISH assays. These assays showed that LINC0113 is located in both the nucleus and cytoplasm ( Figures 3a, b ). LncRNAs fulfill multiple functions by directly interacting with their targeting proteins (25). To identify the interaction proteins of LINC01133, the possible interaction proteins were pulled using biotinylated probes of LINC01133, and the different band in silver staining was identified by mass spectrometry. Mass spectrometry analysis demonstrated a potential direct interaction between IGF2BP2 and LINC01133 ( Figure 3c ). This interaction was subsequently validated through independent experiments, including an RNA pull-down assay ( Figure 3d ) and RIP experiments ( Figure 3e ). To further clarify the specific fragments of LINC01133 interacting with IGF2BP2, we predicted the secondary structure of LINC01133 using an online tool RNAfold ( Figure 3f ). Based on the secondary structure analysis of LINC01133, we designed and constructed three deletion mutants while preserving their essential RNA hairpin structures. RNA pull-down experiments demonstrated that the F5 fragment (spanning nucleotides 811–1,154) is crucial for mediating the association between LINC01133 and IGF2BP2 ( Figure 3g ).

To investigate whether the interaction of LINC01133 with IGF2BP2 influences the protein level of IGF2BP2, several experiments were performed. The results clarified that overexpression of LINC01133 inhibited, while LINC01133 knockdown increased, the protein level of IGF2BP2 ( Figure 4a ). However, either increasing or decreasing LINC01133 hardly influences the RNA levels of IGF2BP2 ( Figure 4b ) inferring that LINC01133 might participate in the protein, rather than RNA, mediation process of IGF2BP2. Thus, we proposed a hypothesis that LINC01133 might regulate the protein stability of IGF2BP2. Then, we tested our hypothesis. We treated the cells using the proteasome inhibitor MG132 (10 µM), and detected the ubiquitination level of IGF2BP2, which reflects its degradation level by the 26S proteasome. As illustrated in Figures 4c, d , overexpression of LINC01133 sharply promoted the ubiquitination and degradation levels of IGF2BP2, while downregulation of LINC01133 had the opposite effect. These findings demonstrated that LINC01133 modulates the IGF2BP2 stability through a ubiquitination-dependent mechanism. However, knockdown of IGF2BP2 does not affect LINC01133 expression levels ( Supplementary Figure S1a ), RNA stability ( Supplementary Figures S1b, c ), or subcellular localization ( Supplementary Figure S1d ).

To further evaluate the influence of IGF2BP2 on LINC01133-induced proliferation and metastasis of ER⁺ breast cancer, IGF2BP2 was transiently transfected into LINC01133-upregulated cells, and siRNA targeting IGF2BP2 was transiently transfected into LINC01133-downregulated cells. CCK8 ( Figure 5a ), colony formation ( Figure 5b ), EdU ( Figure 5c ), migration ( Figure 5d ), and invasion assays ( Figure 5e ) showed that IGF2BP2 could partially rescue the suppressive effects induced by LINC01133 upregulation. Conversely, IGF2BP2 silencing could partially attenuate the promoting effects resulting from LINC01133 downregulation on proliferation, migration, and invasion of ER⁺ breast cancer cells, respectively.

We subsequently researched the underlying mechanisms that lead to the decline of LINC01133 in ER⁺ breast cancer. Since m⁶A is the most widely reported RNA modification (26–29), we predicted whether the m⁶A modification site is located on the LINC01133 transcript using the online tool SRAMP. The prediction results showed that three sites on the LINC01133 transcript might be modified by m⁶A ( Figure 6a ). We further explored which writer is involved in the m⁶A modification of LINC01133 using the online tool RM2Target. The results showed that METTL3 might modulate the m⁶A modification of LINC01133 ( Figure 6b ). Analysis of data from TCGA showed that the LINC01133 level was negatively correlated with the METTL3 level ( Figure 6c ) inferring that METTL3-mediated m⁶A modification of LINC01133 might decrease its level. We then tested our hypothesis and found that downregulation of METTL3 could significantly inhibit the m⁶A level of LINC01133 ( Figure 6d ) and promote the LINC01133 expression ( Figure 6e ). Moreover, we examined the influence of METTL3 downregulation on LINC01133 stability, and a significant increase was observed in the RNA half-life of LINC01133 ( Figure 6f ) inferring that METTL3 regulated the stability of LINC01133 via m⁶A modification. Finally, the online tool SRAMP predicted that m⁶A modification might be located at three sites, and we mutated the three sites, respectively, to investigate which site is responsible for the m⁶A-induced RNA stability ( Figure 6g ). Luciferase reporter assay showed that site 3 was essential for METTL3-mediated m⁶A modification of LINC01133 ( Figure 6h ).

Since m⁶A modification requires a reader to recognize the m⁶A base, we next examined which reader protein is responsible for this function. Through the online tool RM2Target, we found that YTHDF2, an m⁶A reader, might interact with LINC01133 ( Figure 6i ). Moreover, data from TCGA also showed that YTHDF2 was significantly increased in ER⁺ breast cancer tissues relative to ANT ( Figure 6j ). Further assays indicated that downregulation of YTHDF2 hardly influences the m⁶A modification level of LINC01133 ( Figure 6k ), but significantly increased the RNA level of LINC01133 ( Figure 6l ). In addition, RIP assay showed that YTHDF2 was co-precipitated with LINC01133 ( Figure 6m ). These results suggested that YTHDF2 might regulate LINC01133 expression through m⁶A modification. To verify this hypothesis, we investigated the concrete molecular mechanisms of YTHDF2 on the regulation of LINC01133. Our assay showed that YTHDF2 knockdown significantly increased the stability of LINC01133 ( Figure 6n ). Further assays indicated that YTHDF2 regulated the LINC01133 level through recognizing m⁶A modification at site 3 ( Figure 6o ). Especially, as shown in Supplementary Figure S1e , we found that YTHDF2 is the only m⁶A reader binding to LINC01133 in ER⁺ breast cancer cells. Additionally, as shown in Supplementary Figure S1f , Western blotting analysis revealed a negative correlation between the protein levels of METTL3/YTHDF2 and LINC01133 expression using ER⁺ breast cancer tissues consistent with our mRNA data. These results further strengthen the biological relevance of the interaction between the m⁶A machinery and LINC01133 in breast cancer.

Collectively, our findings showed that METTL3 and YTHDF2 regulates the m⁶A level of LINC01133 and that m⁶A modification plays a critical function in modulating LINC01133 stability.