CRISPR activation screens identify oncogenic lncRNAs that are susceptible to CDK4/6 inhibitor treatment

Yifei Wang; Yueshan Zhao; Jiaxin Hu; Zehua Wang; Dhamotharan Pattarayan; Sihan Li; Yu Zhang; Xiaofei Wang; Yue Wang; Wen Xie; Min Zhang; Da Yang

doi:10.1038/s41467-026-70816-2

. 2026 Apr 9;17:5512. doi: 10.1038/s41467-026-70816-2

CRISPR activation screens identify oncogenic lncRNAs that are susceptible to CDK4/6 inhibitor treatment

Yifei Wang ^1,^#, Yueshan Zhao ^1,^#, Jiaxin Hu ^1,^#, Zehua Wang ¹, Dhamotharan Pattarayan ¹, Sihan Li ¹, Yu Zhang ¹, Xiaofei Wang ¹, Yue Wang ¹, Wen Xie ¹, Min Zhang ^2,^✉, Da Yang ^1,^3,^4,^✉

PMCID: PMC13287569 PMID: 41957359

Abstract

The roles of long non-coding RNAs (lncRNAs) in tumorigenesis and therapeutic response remain largely unknown. Here we perform genome-wide and focused CRISPR activation screens to identify lncRNAs regulating palbociclib response in breast cancer cells. A synchronized two-stage proliferation screen not only characterizes tumor growth-regulating lncRNAs, but also reveals a strong negative correlation between lncRNA-mediated regulation of tumor proliferation and CDK4/6 inhibitor sensitivity. By integrating CRISPRa screen results with drug response data from 815 cancer cell lines, we identify and functionally validate that TENM3-AS1, LINC01117, and ENSG00000226706 can increase breast cancer sensitivity to CDK4/6i while promoting tumor proliferation. In breast cancer patients, all three lncRNA signatures are associated with CDK4/6 inhibitor response. Mechanistically, we have shown that lncRNA TENM3-AS1 is a potential ERα-interacting lncRNA, and its regulation of CDK4/6 inhibitor sensitivity is dependent on ERα expression. Our integrated strategy characterizes oncogenic lncRNAs as potential therapeutic biomarkers for CDK4/6 inhibitor treatment in cancer.

Subject terms: Cancer genetics, Targeted therapies

Long non-coding RNAs (lncRNAs) can have cancer-promoting roles. Here, the authors perform genome-wide and focused CRISPR activation screens in breast cancer cells and identify three lncRNAs that promote tumour cell proliferation and sensitise cancer cells to CDK4/6 inhibitor treatment.

Introduction

Tremendous efforts have been made to better characterize novel cancer genes that can serve as targets and biomarkers of cancer treatment. Those large-scale, high-throughput genomics efforts, mainly focusing on protein-coding components of the genome, have led to many insightful discoveries^1,2, but also new questions: few novel cancer genes have been identified to fully explain the molecular and clinical heterogeneity of tumor initiation, progression, and therapeutic response³. Recent large-scale studies, such as the Encyclopedia of DNA Elements (ENCODE), have identified numerous long non-coding RNAs (lncRNAs)^4,5. Further bioinformatics analyses of cancer transcriptome and epigenome have revealed that lncRNAs are among the most prevalent transcriptional changes in cancer^6–9. Functional characterizations of the lncRNAs have suggested that some of them play important roles in tumorigenesis^5,10–14, metastasis¹⁵, and cancer treatment response^9,16–20. For example, lncRNA HOTAIR has been found to augment ER signaling, leading to tamoxifen resistance in breast cancer patients²¹. Our previous research found that the oncogenic lncRNA EPIC1 contributes to Bromodomain and Extra-Terminal inhibitors (iBET) resistance by enhancing the transcriptional activity of the MYC protein^9,22. Moreover, lncRNA EGFR-AS1 and MIR205HG are shown to regulate erlotinib sensitivity through in vitro studies and lung cancer patient cohorts¹⁶. Previous studies have also shown that lncRNA LEIGC inhibits the epithelial-to-mesenchymal transition and enhances the sensitivity to 5-fluorouracil in gastric cancer²³.

Despite these advancements, the function of the vast majority of lncRNAs in cancer initiation, progression, and treatment response remains elusive. Most of the discoveries on novel lncRNAs start from computational analysis, which primarily detects associations rather than functional causality. High-throughput functional genomic studies, such as CRISPR screens, provide a robust method to comprehensively characterize novel genes’ roles in specific phenotypes. Numerous loss-of-function high-throughput screening methods, either based on shRNA or CRISPR technology, have been demonstrated to be an effective effort to systematically explore functionally uncharacterized protein-coding genes. Compared with protein-coding genes, lncRNAs are cell line, species, and disease-specific^24–26. They are also expressed at lower levels with highly diverse splicing isoforms. In this regard, loss-of-function CRISPR screens on lncRNAs suffer from low efficacy and power²⁷. In addition, loss-of-function screens effectively identify negative regulator genes of the phenotype, such as essential genes or tumor suppressor genes, but are less efficient in oncogene discovery. Compared with loss-of-function assays, the gain-of-function CRISPR-SAM activation (CRISPRa) system increases the sensitivity and specificity of screens, especially considering lncRNA’s relatively low expression level and diverse splicing isoforms¹⁹.

To date, three different CDK4/6 inhibitors (Cyclin-dependent kinase 4/6 inhibitors) (i.e., Abemaciclib, Palbociclib, and Ribociclib) have been approved by the U.S. FDA for breast cancer treatment. CDK4/6 inhibitors are targeted therapies used to treat estrogen receptor-positive (ER+), human epidermal growth factor receptor 2-negative (HER2−) breast cancer. Principal mechanism of CDK4/6 inhibitors is suppressing breast cancer cells’ proliferation by blocking cell cycle progression mediated by CDK4 and CDK6 proteins. CDK4/6 inhibitors were first designed for hormone therapy-resistant patients who continue to maintain high expression of the estrogen receptor. Clinically, CDK4/6 inhibitors are used as a first-line treatment method when combined with hormone therapy^28,29. However, lncRNAs’ functions in patients responding to CDK4/6 inhibitors are not well documented. More importantly, there is a lack of treatment biomarkers to better identify patients who may respond to the regimen.

In this work, we conduct two-stage activation screens using the CRISPRa system to functionally characterize lncRNAs involved in CDK4/6i response. Subsequent analyses identify potential lncRNAs that modulate tumor sensitivity to FDA-approved CDK4/6 inhibitors. In a parallel effort to screen for oncogenic lncRNAs in the same system, we observe a strong negative correlation between lncRNA-mediated regulation of tumor proliferation and CDK4/6i response. Integrating screening results with cancer cell line pharmacogenomic and patient data identifies three oncogenic lncRNAs, including TENM3-AS1, LINC01117, and ENSG00000226706, which promote tumor progression by accelerating cell-cycle transitions. Mechanistically, these lncRNAs enhance CDK4/6 inhibitor sensitivity in part through activation of estrogen receptor alpha signaling.

Results

Characterization of CDK4/6i response-regulating lncRNAs by a two-stage CRISPRa screen

To systematically explore the functional roles of lncRNAs in CDK4/6i response, we conducted a genome-wide lncRNA activation screen. Briefly, we established the CRISPR-SAM activation system (Fig. 1A and Supplementary Fig. 1a) in MCF-7 breast cancer cells, as we previously described¹⁹. We then transduced the cells with a 96,458 sgRNA lentiviral library³⁰ targeting 9744 transcriptional start sites of lncRNAs (Supplementary Fig. 1b). The screen included a library control (P0), a non-treatment control sample (P14), and three palbociclib-treated biological replicates (T14a/b/c). These sample groups were well separated in the principal component analysis (PCA) plot, and biological replicates demonstrated good reproducibility (Fig. 1B, left). By comparing the palbociclib-treated samples with the non-treatment control sample (P14), we identified lncRNA genes that may regulate cellular response to palbociclib when activated (Fig. 1C and Supplementary dataset 1).

Fig. 1 — A The overall workflow of CRISPR activation screen using both genome-wide and focused sgRNA libraries to identify lncRNAs regulating palbociclib response. The figure is created with BioRender.com. B PCA plots show the consistency in replicates and difference among sgRNA library control (P0), non-treatment control samples (P14a, b), and palbociclib-treated samples (T14a, b, c). C Positive and negative lncRNA selections identified by MAGeCK in the genome-wide palbociclib screen. D Integration of the two-stage screens leads to consistent positive (red) and negative (blue) lncRNA selections. The dot size represents the average MAGeCK beta score for the two screens. E Volcano plot for significant positive gene selection (red) and negative gene selection (blue) in focused palbociclib screen. Cell-cycle-associated protein-coding genes (PCGs) *CCND1* and *CCND2* are marked in yellow. F MTT assay of MCF-7 cells in two non-targeting (NT) control groups and *CCND1/CCND2* activated groups treated with palbociclib for 72 h at 9 different dosages (0, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, and 8 μM) (n = 4 technical replicates, mean ± SD). G Representative colony formation image of palbociclib-treated two non-targeting controls (sgNT1 and sgNT2) and two PCG-activated groups (sgCCND1 and sgCCND2) at four concentrations (0, 0.125, 0.5, and 2 μM) on Day 7 (left). Quantification of the relative growth area of each group (right) (n = 3 technical replicates, mean ± SD).

In addition to the genome-wide CRISPRa screen on the MCF-7 cell line, we performed a correlation analysis between gene expression and palbociclib response across 41 breast cancer cell lines, using data from the Cancer Cell Line Encyclopedia (CCLE)³¹ and Genomics of Drug Sensitivity in Cancer (GDSC)³² databases (see “Methods”). This analysis included both protein-coding genes and non-coding genes, and successfully recapitulated known CDK4/6i response-modulating genes, such as PIK3CA³³, PTEN³⁴, and MUC16³⁵, as well as genes involved in cell-cycle regulation pathways (Supplementary Fig. 1c). In addition, the analysis also revealed lncRNAs (Supplementary dataset 1) whose expression levels were associated with palbociclib response (P < 0.05).

To validate and prioritize top hits, we further developed a focused sgRNA library specifically activating 127 lncRNAs and 33 PCGs, curated from the genome-wide screen, gene expression-drug response associations, and known cell cycle regulators such as CDK2^36,37, CDK4³⁸, CCNB1³⁹, CCND2⁴⁰, and TP53⁴¹ (see “Methods”, Fig. 1A and Supplementary Fig. 1d). In the focused screen, we designed sgRNAs independently from the genome-wide sgRNA library (Supplementary Fig. 1d) to mitigate potential false positives arising from sgRNA off-target effects and activation efficiency variations. Samples from the focused screens were well grouped and showed good reproducibility between biological replicates (Fig. 1B, right). We observed 58 lncRNAs that exhibited consistent regulation of palbociclib response, showing the same direction of selection in both genome-wide and focused screens (see “Methods” and Fig. 1D, Supplementary dataset 2).

Among our discoveries of CDK4/6i regulators in the focused CRISPR screen are CCND1 (FDR = 0.07, Fig. 1E) and CCND2 (FDR = 0.03, Fig. 1E). Both genes encode cyclin D, which interacts with CDK4/6 proteins directly. Single-gene activation of CCND1 (Supplementary Fig. 1e, f) and CCND2 (Supplementary Fig. 1g, h) demonstrated that the CRISPRa-sgRNA system can lead to at least threefold overexpression for each gene. Further drug-treated MTT and colony formation assays indicated that upregulation of cyclin D genes enhanced the response to palbociclib in MCF-7 cells (Fig. 1F, G and Supplementary Fig. 1i).

LncRNAs that confer susceptibility to palbociclib treatment tend to promote cell proliferation

CDK4/6 inhibitors can block the cell cycle progression from the G1 to the S phase and have been used to target rapidly proliferating cancer cells⁴². After identifying lncRNAs that can modulate palbociclib response, we seek to investigate their roles in cancer cell proliferation. By comparing the sgRNA distribution of cells after 14 days of culture (P14) with cells right after infection (P0), we identified lncRNA genes (Supplementary dataset 1) that may regulate cell proliferation when activated in the genome-wide screen. Further focused screen identified TP53 as a proliferation suppressor (FDR = 0.03), while CCNB1 (FDR = 0.11) and CCND2 (FDR = 0.11) were identified as proliferation promoters (Fig. 2A and Supplementary Fig. 2b). Indeed, the single-gene CRISPRa activations of CCNB1 and CCND2 (Supplementary Figs. 2a and 1g) increased cell growth (Fig. 2B) and proliferation (Fig. 2C). Additionally, the overexpression of CCNB1 and CCND2 also resulted in decreased G0/G1 cell-cycle phase arrest but promoted S, G2, and M phases progression (Fig. 2D).

Fig. 2 — A Volcano plot for significant positive gene selection (red) and negative gene selection (blue) in focused proliferation screen. Cell-cycle-associated PCGs *TP53, CCNB1*, and *CCND2* are marked in yellow. **B–D** Representative figures of cell proliferation assay (B), colony formation assay and quantification (C), and cell-cycle analysis (D) of two non-targeting controls (sgNT1 and sgNT2) and two PCG-activated (sgCCNB1 and sgCCND2) MCF-7 cells. (n = 6/3/3 technical replicates in (B–D), mean ± SD). Experiments were conducted independently twice for cell proliferation assay and cell-cycle analysis, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. E The common significant genes identified from both drug and proliferation screens using the genome-wide library (left) and focused library (right). Most identified genes conferring drug resistance (red) repress cell proliferation, while most of the identified drug sensitivity genes (blue) promote cell proliferation. The width of each stream represents the number of genes indicated on the y-axis. “ns” denotes genes without a significant effect on drug response or cell proliferation. F, G Opposite log2-fold change of the sgRNAs in proliferation (red dot) and palbociclib (blue dot) focused screens for cell-cycle associated PCGs (F) and three lncRNAs (G). MAGeCK Wald FDR for each gene is shown in focused proliferation (red) and drug (blue) screens. H A scatterplot demonstrates the association between the rank of lncRNA expression and the rank of drug response (AUC) in CCLE breast cancer cell lines measured by two-sided Spearman’s correlation. The negative correlation indicates that higher lncRNA expression is associated with higher drug sensitivity (lower AUC). (*TENM3-AS1*, *Rho* = −0.37, P = 0.016; *LINC01117*, *Rho* = −0.51, P = 0.0008; *ENSG00000226706*, *Rho* = −0.3, P = 0.04).

Integrating our findings on proliferation-regulating and CDK4/6i response-modulating lncRNAs, we observed a strong negative correlation between lncRNAs’ regulation of cell proliferation and CDK4/6i response (Fig. 2E). Specifically, 72.7% of drug-resistant hits were found to suppress cell proliferation, whereas 61.5% of drug-sensitive hits enhanced cell proliferation (Fig. 2E and Supplementary dataset 1). This observation highlights a strong link between cell proliferation regulation and the cellular response to palbociclib. Similar trends were observed in the focused screen (Fig. 2E and Supplementary dataset 2). Additionally, this strong negative correlation between cell proliferation and CDK4/6i sensitivity was also observed for protein-coding genes. For example, cell cycle-related genes such as CCND1 and CCND2 were shown to induce sensitivity to palbociclib treatment while promoting cell proliferation (Fig. 2F).

Given the observed inverse relationship between tumor proliferation and CDK4/6i sensitivity in breast cancer, we hypothesized that certain lncRNAs function as oncogenes by increasing tumor growth and could serve as potential therapeutic markers for CDK4/6i treatment. Based on their expression in breast cancer patients and cell lines, along with their association with patient survival, three lncRNAs (TENM3-AS1, LINC01117, and ENSG00000226706) were selected for functional validation (see “Methods” and Supplementary dataset 2 and Supplementary Fig. 2c). All three lncRNAs were identified to sensitize cells to palbociclib treatment, while promoting cell proliferation in both genome-wide (Supplementary Fig. 2d) and focused screens (Fig. 2G). In CCLE and GDSC databases, the selected lncRNAs’ endogenous expression was significantly correlated with CDK4/6i sensitivity across 41 breast cancer cell lines (P < 0.05, Fig. 2H). Furthermore, these lncRNAs’ expressions were also correlated with CDK4/6i sensitivity in other cancer types, such as bladder cancer, myeloma, and neuroblastoma (Supplementary Fig. 2e). These observations suggested that the tumors overexpressing the selected lncRNAs might be susceptible to CDK4/6i therapy.

TENM3-AS1, LINC01117, and ENSG00000226706 are oncogenic lncRNAs that promote cell proliferation, tumorigenesis, while serving as therapeutic markers for CDK4/6 inhibitors

To functionally validate the selected lncRNAs’ impact on palbociclib’s response, we specifically activated each of the three candidate lncRNAs (TENM3-AS1, LINC01117, and ENSG00000226706) in breast cancer cell lines MCF-7 and T-47D. The CRISPRa system was able to activate corresponding lncRNAs by at least 2.5-fold (Supplementary Fig. 3a). We observed that the activation of the lncRNAs augment palbociclib sensitivity after 48 h (Supplementary Fig. 3b, c) or 72 h (Fig. 3A, B) of treatment in both cell lines. Meanwhile, a reduction in colony formation was observed after being treated with palbociclib at different dosages in each lncRNA-activated cancer cell lines (Fig. 3C, D). In addition to CRISPR activation, siRNA-mediated knockdown of TENM3-AS1, LINC01117, or ENSG00000226706 led to resistance to palbociclib in both 48 h and 72 h treatments (Fig. 3E).

Fig. 3 — A, B Representative figures of MCF-7 (A) and T-47D (B) cells MTT assay of non-targeting control groups (sgNT) and three lncRNA-activated groups (sgTAS, sg1117, and sg706) treated with 9 different dosages of palbociclib (0, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, and 8 μM) for 72 h (n = 4 technical replicates, mean ± SD). Experiments were conducted independently five times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. C, D Representative colony formation image (left) and quantification of the relative growth area of each group (right) of control group (sgNT) and three lncRNA-activated groups on Day 7 treated with palbociclib in MCF-7 (C, concentration: 0, 0.125, 0.5, and 2 μM) and T-47D cells (D, concentration: 0, 0.5, 2, 8 μM). (n = 3 technical replicates, mean ± SD). Experiments were conducted independently at least three times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. E Representative figures of MCF-7 cell MTT assay of the siRNA control group (siCon) and lncRNA knockdown groups (siTAS, si1117, and si706) treated with 9 different dosages of palbociclib (0, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, and 8 μM) for 48 h and 72 h (n = 4 technical replicates, mean ± SD). Experiments were conducted independently four times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file.

After validating all three lncRNAs’ susceptibilities to palbociclib treatment, we move on to investigate their regulation of cell proliferation and tumorigenesis. Cell growth assays (Fig. 4A) and the colony formation assay (Fig. 4B) demonstrated that the activation of candidate lncRNAs can increase ER+ breast cancer cells’ proliferation in both MCF-7 and T-47D cells. Further cell cycle assay showed that the activation of three lncRNAs led to an increase in the G2/M phase but a decrease in the G0/G1 resting phase (Fig. 4C), which suggested an accelerated cell cycle progression.

Fig. 4 — A Representative figures of cell proliferation assay of non-targeting control (sgNT) and three lncRNA-activated (sgTAS, sg1117, and sg706) MCF-7 and T-47D cells (n = 6 technical replicates, mean ± SD). Experiments were conducted independently three times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. B Representative colony formation image (upper panel) and quantification (lower panel) of non-targeting control (sgNT) and three lncRNA-activated (sgTAS, sg1117, and sg706) MCF-7 and T-47D cells. (n = 3 technical replicates, mean ± SD). Experiments were conducted independently twice in T-47D cells, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. C Cell-cycle analysis of non-targeting control (sgNT) and three lncRNA-activated (sgTAS, sg1117, and sg706) MCF-7 cells (n = 3 biological replicates, mean ± SD, P value was obtained from one-way ANOVA followed by Tukey multiple comparisons test. D, E Representative figures of cell proliferation assay of the control (siCon), *TENM3-AS1* knockdown (siTAS) (D), and *LINC01117* knockdown (si1117) (E) groups in MCF-7, HS578T, and U2OS cells, respectively (n = 6 technical replicates, mean ± SD). F Representative colony formation image (upper panel) and quantification (lower panel) of the control (siCon), *TENM3-AS1* knockdown (siTAS), *LINC01117* knockdown (si1117), and *ENSG00000226706* knockdown (si706) MCF-7 cells (n = 3 technical replicates, mean ± SD). For (D–F), all MCF-7 experiments were independently performed four times with consistent results. All results from independent replicates are included in the corresponding Source Data file.

To determine if knockdown of the identified lncRNAs will suppress tumor proliferation, we first assessed the expression of lncRNAs in nine cell lines and identified four cell lines that have high endogenous levels of the identified lncRNAs. Specifically, HS578T, SKOV3, U2OS, and MCF-7 showed higher expression levels of TENM3-AS1 and LINC01117, while RKO and MCF-7 have increased endogenous expression of ENSG00000226706 (Supplementary Fig. 4a). We designed siRNAs specifically for each lncRNA and knocked down the endogenous lncRNA expression in up to three cell lines with high expression of corresponding lncRNAs (Supplementary Fig. 4b–d). Consistent with our observation in lncRNA activation, the knockdown of lncRNAs TENM3-AS1 (Fig. 4D), LINC01117 (Fig. 4E), and ENSG00000226706 (Supplementary Fig. 4e) led to decreased cell proliferation across multiple cancer models. Further colony formation assays confirmed that suppressing each of the three lncRNAs led to a reduction of colonies (Fig. 4F).

To determine if these lncRNAs can promote breast tumor progression in vivo, we inoculated the lncRNA-activated and control MCF-7 breast cancer cells into the nude mice. Compared with the non-targeting control sgRNA group (NT), activation of each of the three lncRNAs led to increased tumor growth, as indicated by tumor volume (Fig. 5A and Supplementary Fig. 5a) and tumor weight (Fig. 5B). In addition, immunofluorescence analysis of these tumor tissues showed that the activation of the three lncRNAs upregulated Ki-67 expression, a known proliferation marker, in breast cancer tumors (Fig. 5C).

Fig. 5 — A The tumor growth rate of the control sgNT1 (n = 7) and sgNT2 (n = 8), and three lncRNA-activated sgTAS (n = 10), sg1117 (n = 10), and sg706 (n = 10) MCF-7 cells in the xenograft model (left) and the image of representative xenograft tumors from each group (right). The relative tumor growth ratio at each time point is normalized to Day 0 in each group (mean ± SEM, P value was obtained from two-way ANOVA followed by Tukey multiple comparisons test). B Tumor weight of MCF-7 xenograft mouse model in different groups of (A), sgNT1 (n = 7), sgNT2 (n = 8), sgTAS (n = 10), sg1117 (n = 10), and sg706 (n = 10) (mean ± SD, P value was obtained from one-way ANOVA followed by Tukey multiple comparisons test). C Representative tissue immunofluorescence image of Ki-67 expression of MCF-7 xenograft model (n = 3 biological replicates, scale bar = 40 µm). D Representative figures of cell proliferation assay demonstrate the effect of the control (siCon) and *TENM3* knockdown (siTENM3) on cell growth of non-targeting, and *TENM3-AS1* activated MCF-7 cells (n = 6 technical replicates, mean ± SD). E Representative PCR result for cDNA overexpression efficiency for *TENM3-AS1* in T-47D cells compared with the control group (Vector). F Representative cell proliferation assay of the control (Vector) and *TENM3-AS1* overexpression (TAS-O/E) T-47D cells (n = 6 technical replicates, mean ± SD). Experiments were conducted independently four times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file. G Representative colony formation image and quantification of the control (Vector) and *TENM3-AS1* overexpression (TAS-O/E) groups in T-47D cells (n = 3 technical replicates, mean ± SD). Experiments were conducted independently four times, consistently producing similar results. All results from independent replicates are included in the corresponding Source Data file.

Among the three validated lncRNAs, TENM3-AS1 is classified as an antisense lncRNA. The sgRNAs activating TENM3-AS1 may lead to off-target activation of its neighboring protein-coding gene TENM3. To test this possibility, we designed siRNA to knock down TENM3 mRNA in the sgTENM3-AS1 cells. This analysis revealed that knockdown of the TENM3 (Supplementary Fig. 5b) does not affect breast cancer cells’ proliferation (Supplementary Fig. 5c), nor abolish TENM3-AS1’s regulation of cell growth (Fig. 5D). In addition to TENM3, TENM3-AS1’s annotated location suggested it may share the promoter with another gene, ENSG00000248266, which is not affected by TENM3-AS1’s sgRNA activation (Supplementary Fig. 5d). These observations indicate that lncRNA TENM3-AS1’s regulation of cell proliferation is not mediated by collateral activation of TENM3 or ENSG00000248266. To further corroborate this observation, we overexpressed TENM3-AS1 cDNA (805 bp, see “Methods”) in breast cancer T-47D cells (Fig. 5E). Similar to the sgTENM3-AS1 activation, overexpression of TENM3-AS1 cDNA promoted cell proliferation, as indicated by augmented cell growth (Fig. 5F) and colony formation abilities (Fig. 5G). Altogether, our results suggest that tumors overexpressing oncogenic lncRNAs, despite showing a higher proliferation rate, may be vulnerable to the FDA-approved palbociclib treatment.

RNA-seq analysis revealed that lncRNA TENM3-AS1 activates the ER pathway

To determine the potential mechanism by which these oncogenic lncRNAs reshape tumor proliferation and palbociclib sensitivity, we performed RNA-seq analysis after the activation of oncogenic lncRNAs. This analysis characterized 466, 521, and 522 differentially expressed genes in response to the activation of TENM3-AS1, LINC01117, and ENSG00000226706, respectively (Supplementary dataset 3). Pathway analysis highlighted the upregulation of “Estrogen Response Early” (TENM3-AS1 FDR = 0.015, LINC01117 FDR = 0.04, ENSG00000226706 FDR = 0.034, Fig. 6A and Supplementary Fig. 6a, b) and “MYC Targets V1” pathways (TENM3-AS1 FDR = 0.011, LINC01117 FDR = 0.02, ENSG00000226706 FDR = 0.2, Supplementary dataset 4, Fig. 6A, and Supplementary Fig. 6a), suggesting that these lncRNAs may mediate pro-proliferative and oncogenic effects. Meanwhile, knockdown of the oncogenic lncRNAs led to the reversion of 48.9%, 48.7%, and 47.3% of gene expression changes induced by TENM3-AS1, LINC01117, and ENSG00000226706, respectively (Supplementary dataset 4). Notably, both “Estrogen Response Early” and “MYC Targets V1” were downregulated after lncRNAs knockdown (FDR < 0.05, Fig. 6A).

Specifically, CRISPRa activation of TENM3-AS1 in MCF-7 (Fig. 6B) and T-47D cells (Supplementary Fig. 6c) upregulates ERα and its targets MYC, GREB1, and TFF1. This activation of the ERα targets was recapitulated in the TENM3-AS1 cDNA overexpressed T-47D cells (Fig. 6C). Moreover, the ERα and MYC expression were also increased in the TENM3-AS1-activated MCF-7 mouse model (Fig. 6D). Consistently, knockdown of TENM3-AS1 downregulated the expression of these genes (Fig. 6E). To further investigate if ERα mediates TENM3-AS1’s oncogenic role, we used siRNA to knockdown ERα in the control cells and TENM3-AS1-activated MCF-7 cells. Suppression of ERα abolished TENM3-AS1’s regulation of its downstream targets (Fig. 7A). In addition to the target expression, inhibition of ERα offset the function of TENM3-AS1 in cell proliferation (Fig. 7B) as indicated by the MTT assay. Besides TENM3-AS1, we observed that LINC01117 and ENSG00000226706 activation didn’t change ERα or TFF1’s expression, but led to upregulated expression of MYC and GREB1 (Supplementary Fig. 6b and d). These results suggest that three oncogenic lncRNAs, especially TENM3-AS1, promote cell growth and tumorigenesis through activating the ERα pathway.

Fig. 7 — A Representative mRNA expression (left) of *ESR1* and its downstream targets (*MYC*, *TFF1*, and *GREB1*) (n = 3 technical replicates, mean ± SD), and protein expression (right) level (ERα, MYC, and GREB1) in Control (siCon) and ERα siRNA-treated (siERα) MCF-7 control (sgNTs) and *TENM3-AS1*-activated cells. B Representative cell proliferation assay of the control (siCon) and ERα knockdown (siERα) in MCF-7 sgNTs and *TENM3-AS1* sgRNA-activated cells (n = 4 technical replicates, mean ± SD). C Expression of lncRNA-regulated genes in CDK4/6i-sensitive and -resistant organoids (left; n = 4 vs 13) and xenografts (right; n = 13 vs 24). Box plots indicate median, interquartile range, and minimum-maximum values. Statistical significance was assessed by two-sided Welch’s t-test. D Association between lncRNA activation signatures and pathway scores in single-cell data. Median lncRNA signature scores are shown across bins of G2/M (top) and Estrogen Response (bottom) scores; error bars indicate ± SEM. E Changes in lncRNA signature scores pre- and post-treatment for letrozole-only (responders n = 5, non-responders n = 5) and combination therapy (responders n = 2, non-responders n = 5). Solid and dashed lines denote responders and non-responders; error bars show ± SEM. F Representative *TENM3-AS1*, U6, and *GAPDH* expression (n = 3 technical replicates, mean ± SD) and western blot of fractionation assay in MCF-7 cells. U1 (SNRNP70), GAPDH, and Total H3 served as nuclear (Nuc), cytoplasmic (Cyto), chromatin (Chrom) markers in whole-cell lysates (WCL). G Representative western blot of ERα IP efficacy (bottom). RT-PCR result (top) of RIP assay of *TENM3-AS1* enriched by ERα protein in MCF-7 *TENM3-AS1*-activated cells. GAPDH and U6 served as negative controls (n = 3 technical replicates, mean ± SD). H Representative MTT assay of siRNA control (siCon) and ERα knockdown (siERα) MCF-7 cells with or without *TENM3-AS1* activation treated with different dosages of palbociclib (0, 0.0625, 0.125, 0.25, 0.5, 1, 2, 4, and 8 μM) for 48 and 72 h (n = 4 technical replicates, mean ± SD). For (A, B, F, and G), experiments were performed independently three times, and for (H), four times, with consistent results across replicates. All results from independent replicates are included in the corresponding Source Data file.

LncRNA TENM3-AS1 activates the ERα signaling to induce cancer cell proliferation and regulate CDK4/6i response

The estrogen receptor alpha (ERα) is the major driver for ER+ breast cancer^43,44. Previous clinical and pre-clinical studies have indicated the importance of ERα in maintaining patient response to CDK4/6 inhibitors^45–47. With the observations that three oncogenic lncRNAs activate the ERα signaling, we speculated that lncRNAs’ regulation of CDK4/6i sensitivity may also be partially mediated by the ERα-MYC axis.

To test this hypothesis, we constructed lncRNA activation scores based on genes that are significantly changed upon each oncogenic lncRNA activation (see “Methods”). Using gene expression data from patient-derived organoids⁴⁸, we showed that the palbociclib-sensitive samples exhibited higher activation scores of oncogenic lncRNAs than resistant samples (Fig. 7C). A similar pattern was also observed in patient-derived xenografts, where 13 palbociclib-sensitive samples showed elevated lncRNA activation scores compared with 24 resistant samples⁴⁸ (Fig. 7C). To corroborate this observation, we further applied the lncRNA activation signatures to single-cell transcriptomes generated from serial tumor biopsies of 24 patients with early-stage breast cancer enrolled in the FELINE clinical trial⁴⁷. In this study, patients were treated with either endocrine therapy (i.e., letrozole) alone or in combination with the CDK4/6 inhibitor ribociclib. Consistent with our previous observations, tumor cells with higher lncRNA activation scores exhibited elevated ER pathway activity (Fig. 7D) and higher G2/M transition (Fig. 7D). These observations supported the potential role of the oncogenic lncRNAs in promoting cell cycle progression through activating the ER pathway. Notably, patients who responded to the letrozole + ribociclib combination therapy consistently showed higher lncRNA activation scores (Fig. 7E), aligning with our screening results. These results suggest that the oncogenic lncRNAs may be exploited as therapeutic biomarkers for the FDA-approved combination regimen.

Since the TENM3-AS1’s regulation of cell growth and CDK4/6 inhibitors exhibited a relatively high association with ERα, we next sought to investigate how lncRNA TENM3-AS1 activates ERα. We first performed the fractionation assay, which suggested that lncRNA TENM3-AS1 is mainly localized in the chromatin section (Fig. 7F). Further RNA immunoprecipitation (RIP) assay demonstrated that RNAs co-immunoprecipitated with ERα protein are enriched with TENM3-AS1 (Fig. 7G). In contrast, TENM3-AS1 RNA is not enriched in the MYC RIP sample (Supplementary Fig. 7a), suggesting TENM3-AS1 may specifically interact with the ERα protein. To further determine if TENM3-AS1’s regulation of CDK4/6 inhibitors’ sensitivity is dependent on ERα expression, we knocked down ERα in TENM3-AS1-activated cells and then treated them with various doses of palbociclib. At both 48 h and 72 h (Fig. 7H) after palbociclib treatment, we observed that knockdown of ERα abolished TENM3-AS1 and the other two lncRNAs, LINC01117 and ENSG00000226706-mediated (Supplementary Fig. 7b) sensitivity to palbociclib. Overall, our results suggest that TENM3-AS1 is an oncogenic lncRNA that promotes breast cancer proliferation by activating the ERα targets.

Discussion

In this study, we performed both genome-wide and focused CRISPRa screens to identify lncRNAs that regulate palbociclib response in ER+/HER2− breast cancer cells. In parallel, a two-stage CRISPRa proliferation screen characterized lncRNAs involved in tumor growth. Notably, we observed a negative correlation between lncRNA-mediated regulation of tumor proliferation and CDK4/6i sensitivity. Further integrating the screening results with endogenous lncRNA expression in breast cancer cell lines and patient samples identified three oncogenic lncRNAs: TENM3-AS1, LINC01117, and ENSG00000226706. Loss-of-function and gain-of-function studies demonstrated that three lncRNAs can promote breast cancer proliferation by accelerating G1 to G2/M cell cycle transition in multiple cancer models. We have further validated that these oncogenic lncRNAs drive tumors toward a fast-proliferating but CDK4/6i-sensitive phenotype. We evaluated the clinical relevance of the oncogenic lncRNAs by constructing activation scores based on genes significantly altered upon their activation. Using patient-derived organoid models and single-cell transcriptomes from serial biopsies in the FELINE clinical trial, we observed that tumor cells with higher lncRNA activation scores demonstrated accelerated G2/M transition. Importantly, patients who responded to the combination of letrozole and ribociclib consistently exhibited higher lncRNA activation scores, in agreement with our screening results. Together, these findings support the role of oncogenic lncRNAs in promoting cell cycle progression and highlight their potential as predictive biomarkers for response to FDA-approved CDK4/6 inhibitor-based therapies.

CDK4/6 inhibitors have shown promising therapeutic effects in multiple cancer types^49–51. However, CDK4/6 inhibitors are only approved for postmenopausal women with HR+/HER2− advanced or metastatic breast cancer patients^52,53. Numerous preclinical and clinical studies are exploring the potential of CDK4/6i across more cancer types, either as monotherapy or in combination. These studies revealed that patient outcomes to CDK4/6i treatment are highly heterogeneous. Although intrinsic RB1 loss⁵⁴ or CDK4 amplification⁵⁵ could predict CDK4/6 inhibitors’ resistance in patients, clinically actionable biomarkers are urgently needed for designing and expanding the patient population that can benefit from CDK4/6i treatment. In this study, we have shown that lncRNA TENM3-AS1, LINC01117, and ENSG00000226706 are overexpressed in multiple cancer types, including lung, lymphoma, and gastrointestinal cancers. Some of them are under consideration or in clinical trials for CDK4/6i treatment^49,51,56,57. Our results suggest that these lncRNAs’ regulation of CDK4/6i sensitivity may be applied to other cancer types. Future work is warranted to determine if those lncRNAs can help screen patients for CDK4/6i therapy.

Although CRISPRa enables systematic interrogation of lncRNA function, several caveats should be considered when analyzing these findings. Because many lncRNAs act through highly context-dependent mechanisms, such as chromatin remodeling, or isoform-specific situations, artificial activation by the CRISPR activation system may not fully reproduce their endogenous regulatory complexity³⁰. Moreover, CRISPRa can induce off-target transcriptional effects or alter neighboring gene expression, particularly for cis-regulation or transcribed from bidirectional promoters⁵⁸. We also observed modest phenotypic differences between two non-targeting sgRNAs (sgNT1 and sgNT2), which may reflect sequence-dependent chromatin accessibility, sudden promoter activation, or random variation in viral integration. These observations indicate the importance of including multiple non-targeting controls and multiple validation strategies when applying CRISPRa to study lncRNA biology.

Mechanistically, our findings demonstrated that three oncogenic lncRNAs enhance ERα signaling and drive MYC overexpression. ERα regulates key components of tumor progression in multiple cancer types^59–64 and has been implicated in shaping responses to CDK4/6 inhibitors^65,66. Using activation signatures derived from these lncRNAs, we showed that tumors or cells with elevated lncRNA activation signatures exhibited enhanced ER pathway activity and accelerated G2/M progression across patient-derived organoids, xenografts, and single-cell data from the FELINE clinical trial. Our study also identified TENM3-AS1 as a chromatin-associated lncRNA that is enriched in the RNAs co-immunoprecipitated with ERα protein. Knockdown of ERα diminished the ability of TENM3-AS1 to regulate cell growth and palbociclib sensitivity, indicating that its effects are at least partially ERα-dependent.

Recent studies have revealed that many canonical transcription factors (TFs) directly bind RNA, and that these RNA-TF interactions can influence chromatin occupancy, complex assembly, and protein function^67,68. For instance, our recent work demonstrated that MYC directly binds a broad set of RNAs with sequence preferences, and that RNA depletion reduces MYC chromatin occupancy⁶⁹. Similarly, we have shown that nuclear receptor TF, PXR, can bind cytoplasmic RNAs to enhance mRNA stability⁷⁰. Consistent with these findings, ERα has also been reported to exhibit RNA-binding activity. Ruggero and colleagues demonstrated that cytoplasmic ERα can bind mRNAs to promote their translation⁷¹, while Notani and colleagues further showed that nuclear RNAs interact with ERα to increase its chromatin occupancy and transcriptional activity⁷². In line with these observations, our study identifies TENM3-AS1 as an ERα-associated lncRNA that increases the expression of ERα targets. In addition, we have shown that MYC expression is consistently regulated by all three lncRNAs in both gain- and loss-of-function assays. Future studies are required to determine the structural and molecular basis of how these oncogenic lncRNAs activate ERα signaling and MYC expression.

Methods

Institutional approval

This study complies with all relevant ethical regulations, and all protocols were approved by the Institutional Biosafety Committee of the University of Pittsburgh. Experimental protocols involving mice were approved by the Institutional Animal Care and Use Committee (IACUC) of the University of Pittsburgh.

Cell lines and reagents

Human cancer cell lines MCF-7, HS578T, HEK293T, T-47D, RKO, and U2OS were purchased from the American Type Culture Collection (ATCC). HEK293FT cell was purchased from Invitrogen (R70007). HEK293FT cell was cultured in DMEM medium supplemented with GlutaMax (Gibco, 10569010), 10% FBS, and 1% penicillin-streptomycin. MCF-7, 293T cells were cultured in DMEM (Hyclone, SH30243.01) supplemented with 10% FBS (Gibco, 10438-026) and 1% penicillin-streptomycin (Corning, 30-002-CI). T-47D was cultured in the RPMI-1640 medium (Hyclone, SH30027.01) with 10% FBS, 1% penicillin-streptomycin, and 10 μg/mL insulin (Sigma, I1882). RKO was cultured in the MEM medium (Corning, 10-009-CV) supplemented with 10% FBS and 1% penicillin-streptomycin. HS578T cells were cultured in DMEM medium supplemented with 10% FBS, 1% penicillin-streptomycin, and a final concentration of 0.01 mg/mL insulin. U2OS cells were cultured in McCOY’ 5A medium (Hyclone, SH30200.01) supplemented with 10% FBS and 1% penicillin-streptomycin.

Focused screen library construction and amplification

The preparation of the focused screen library followed the previously published protocol⁷³. In brief, after we have selected the targets according to Supplementary Fig. 1c. Then, we designed our sgRNAs utilizing CRISPick, a genetic perturbation platform invented by Broad Institute, which ranks and picks candidate CRISPRko/a/i sgRNA sequences to maximize on-target activity for the target(s) provided⁷³. For each target gene, we picked the top six ranked sgRNAs, and together with 40 non-target (NT) sgRNAs (from the global screen library), we had a library containing 1000 sgRNAs targeting 160 genes. Since we used the Lenti-sgRNA (MS2)-Puro plasmid (Addgene, 73795) as the backbone and inserted sgRNA at the BsmBI site, we designed the oligo with 5′ flank sequence partially overlapped with the U6 promoter and a 3′ flank sequence overlapped with the MS2 sequence from the backbone. Then, these oligos were synthesized by Twist Bioscience (Supplementary dataset 5). The pooled oligo library was amplified by PCR following the protocol. Forward primer: 5′-TAACTTGAAAGTATTTCGATTTCTTGGCTTTATATATCTTGTGGAAAGGACGAAACACCG-3′; Reverse primer: 5′-TGTTGGCCTAGCTCTAAAAC-3′. The PCR product is purified by QIAquick PCR purification kit (Qiagen, 28106) according to the manufacturer’s directions and quantified the product with a NanoDrop UV spectrophotometer. The purified oligo products were loaded into a 2% low-melting agarose gel (Bio-rad, 161-3102), and the oligos were extracted using QIAquick Gel Extraction Kit (Qiagen, 28706) according to the manufacturer’s directions and quantified by NanoDrop. The extracted pooled oligos were further purified and concentrated by DNA Clean & Concentrator ^TM-5 kit (Zymo Research, D4014). The Lenti-sgRNA (MS2)-Puro plasmid was digested with Esp3I (BsmBI) enzyme (Thermo Fisher, FD0454) and treated with FastAP Thermosensitive Alkaline phosphatase (Thermo Fisher, EF0651) according to the protocol at 37 °C for 1 h. The digest reactions were loaded into 2% low-melting agarose gel. The backbone was extracted from the gel using the QIAquiick Gel Extraction Kit according to the manufacturer’s protocol and quantified by NanoDrop. The digested library plasmid backbone and pooled oligo were assembled through Gibson assembly using Gibson Assembly Master Mix, 2× (NEB, M5510A) according to the protocol and incubated at 50 °C for 2 h. The Gibson Assembly reaction mix was purified and concentrated by Isopropanol according to the protocol. The precipitated plasmid DNA was washed with cold 80% ethanol twice and dissolved in 6 mL of TE buffer at 55 °C for 10 min and quantified by NanoDrop.

Then, 2 mL of 50 ng/mL oligo library plasmid was used for one electroporate into Endura ElectroCompetent cells (LGC Biosearch Technologies, 60242-2) according to the manufacturer’s direction. Each electroporation (2 mL in total) was plated on the large LB agar plates (245-mm square bioassay dish with ampicillin). At the same time, prepare 1:100, 1:1000, and 1:10,000 dilutions for calculating the transformation efficiency for both sgRNA library and control groups. After incubating all LB agar plates overnight at 37 °C for 12–14 h, then calculated electroporation efficiency by counting the number of colonies on the 1:10,000 dilution plate. Proceed to the next step only if there are more than 500 colonies per sgRNA in the library for the customized library (in our case, > 500,000 colonies) and there are at least 20 times more colonies per electroporation in the library compared with the control Gibson assembly reaction. The colonies were harvested from the LB agar plates, and the OD₆₀₀ value was measured to determine the number of the Maxipreps as follows: $\frac{OD value Total volume of colonies harvested}{1, 200}$ and performed the Maxipreps using Macherey-Nagel NucleoBond Xtra Maxi EF kit (Macherey-Nagel, 740424.10) according to the manufacturer’s directions. The pooled library plasmid was quantified by NanoDrop.

sgRNA library lentivirus production and transfection

Global screen

The preparation of the library and cells followed the previously published protocol and previous publication⁷³. Human CRISPR lncRNA Activation Pooled Library (SAM − 3 plasmid system) was purchased from Addgene (1000000106). Based on ENCODE V19 and reannotation, the library targets 9744 transcription start sites of lncRNAs, with 96,458 sgRNAs in total. The library also contained 500 control sgRNAs (NT sgRNAs). The purchased sgRNA library was first amplified by electroporation to the Endura ElctroCompetent cells according to the manufacturer’s directions, which were shown in the front. The colony number is greater than 100 colonies per sgRNA in the library according to the protocol. The amplified library plasmids were extracted using Maherey-Nagel NucleoBond Xtra Maxi EF Kit according to the manufacturer’s directions. The extracted plasmids were quantified by NanoDrop and ready to use. The antibiotic killing curve was performed in MCF-7 cells to determine the concentration of three antibiotics (zeocin, blasticidin, and hygromycin), which matched with three plasmids (sgRNA library, dCas9-VP64-Blast, and MS2-P65-HSF1), respectively. The concentration for selection was indicated below: 200 μg/mL zeocin (Invivogen, ant-zn-05), 4 μg/mL blasticidin (Sigma, 15205), and 300 μg/mL hygromycin (RPI, 31282-04-9). Before transducing the lncRNA library, we generated the dCas9-VP64 and MS2-p65-HSF1-Hygro stable MCF-7 cells with the multiplicity of infection (MOI) of lentivirus 0.5 and selected with the proper concentration of blasticidin and hygromycin which indicated above, respectively. HEK293FT cells were cultured in T225 flasks and maintained at 70%–80% confluency. The transfection HEK293FT cells were seeded the previous day and reached 80–90% confluency on the next day for transfection. Transfection reagents, including the Lipofectamine 2000 (Invitrogen, 11668-019) and the PLUS reagent (Invitrogen, 11514-015), were used according to the protocol⁷³ referred to the amount per T225 flask. For each T225 flask, (15.3 μg pMD2.G (Addgene, 12259) and 23.4 μg psPAX2 (Addgene, 12260) as helper plasmids), and 30.6 μg sgRNA library plasmids were mixed with 2250 μL of Opti-MEM (Gibco, 11058-021). The medium was replaced with a prewarmed DMEM medium after eight hours. After 48 h and 72 h, the lentivirus was harvested and filtered with 0.45 μm filters. The MOI value of the lncRNA library lentivirus was determined and calculated⁷³ by infecting the MCF-7 cells with the titer of the library lentivirus, including 0, 25, 50, 100, 200, and 400 mL. (One group was treated with zeocin based on the killing curve indicated in the front and another group was not treated with zeocin). Finally, MCF-7 cells transduced with the SAM CRISPR system were infected with lncRNA sgRNA library lentivirus at an MOI around 0.3, maintaining coverage of > 500 cells expressing each sgRNA. Cells were selected with zeocin for 10 days and then expanded for the next step, CRISPR screening.

Focused screen

The focused screen process was very similar to the global screen, which also followed the protocol⁷³. The minor differences from the global screen are shown below. The customized library contained 1000 sgRNAs targeting 160 genes along with 40 non-targeting control sgRNAs. The constructed sgRNA library was described in the previous section. The colony number is greater than 500 colonies per sgRNA in the library for customized library according to the protocol. The customized library used Lenti-sgRNA (MS2)-Puro plasmid as the backbone for library construction. The antibiotic killing curve was performed in MCF-7 cells to determine the concentration of three antibiotics (puromycin, blasticidin, and hygromycin), which matched with three plasmids (sgRNA library, dCas9-VP64, and MS2-P65-HSF1), respectively. The concentration for selection was indicated below: 0.5 μg/mL puromycin (Gibco, A11139-03), 4 μg/mL blasticidin (Sigma, 15205), and 300 μg/mL hygromycin (RPI, 31282-04-9) for MCF-7 cells. Finally, MCF-7 cells transduced with the SAM CRISPR system were infected with lncRNA sgRNA library lentivirus at an MOI of 0.11, maintaining coverage of > 500 cells expressing each sgRNA. Cells were selected with puromycin for 5 days and then expanded for the next step CRISPR screening.

Global and focused CRISPRa screens for cell proliferation and palbociclib response

For global screen, the proliferation screen contains two samples. One sample is a zero-passage library; another was cultured for 14 days. For palbociclib screen, the palbociclib (Sigma, PZ0383) concentration for MCF-7 cells was determined by MTT and colony formation assay. The final concentration for screening was 0.826 μM. There are a total of five samples, which includes one zero-passage library sample (P0); one sample that cultured for 14 days but was not treated with palbociclib (P14); three samples that cultured for 14 days but treated with palbociclib (T14a, T14b, T14c). Each sample was seeded in three 15-cm dishes with 5 × 10⁷ cells/dish. P0 was directly collected from the dish and extracted genomic DNA (5 × 10⁷ cells). Other samples were harvested after 14 days with at least 5 × 10⁷ cells per sample for further genomic DNA extraction and sgRNA amplification.

Regarding the focused screen, the proliferation screen contains three samples. One sample is a zero-passage library; another two samples were cultured for 14 days. The palbociclib concentration for MCF-7 cells followed the global screen concentration (0.826 μM). There are a total of six samples for each cell line, which includes one zero-passage library sample (P0); two samples when cultured for 14 days but were not treated with palbociclib (P14a and P14b); three samples that were cultured for 14 days but treated with palbociclib (T14a, T14b, and T14c). Each sample was seeded in a 15-cm dish with 1 × 10⁷ cells. P0 was directly collected from the dish and extracted genomic DNA (1 × 10⁷ cells). Other samples were harvested after 14 days with at least 1 × 10⁷ cells per sample for further genomic DNA extraction and sgRNA amplification.

Genomic DNA extraction and sample preparation for NGS

The amplification of sgRNA, along with the primer sequence and preparation of samples, followed the previously published protocol⁷³. Cell numbers for global and focused screen were described above. Briefly, genomic DNA was extracted from the cells after screening with the maintained coverage of > 500 cells expressing per sgRNA using the Zymo Research Quick-DNA Midiprep plus kit (Zymo Research, FD4075) according to the manufacturer’s instruction. Then, sgRNAs were amplified using PCR with the NEBNext High Fidelity 2× PCR Master Mix (NEB, M0541S). The primers were listed in the published protocol⁷³. Each sample will be amplified using ten forward primers mix, but only one reverse primer that contains a unique barcode. For global screen, five samples of MCF-7 cells matched with the reverse primers as listed here: P0-R1, P14-R2, T1-R3, T2-R4, and T3-R5. For focused screen, MCF-7 group focused library plasmid-R1, P0-R2, P14A-R3, P14B-R4, T1-R5, T2-R6, T3-R7. Both the global screen library plasmid and customized screen library plasmid used 20 ng genomic DNA as templates for sgRNA amplification; other samples used 1 μg genomic DNA as templates. PCR reactions were performed according to the protocol⁷³. PCR products were pooled for each sample and then purified by a QIAquick PCR purification kit (Qiagen, 28106). Quantification was determined by Nanodrop. 2 μg of purified PCR products were loaded into 2% agarose gel. The final product was around 280 bp after running the gel. Gel extraction was performed using the QIAquick Gel Extraction kit (Qiagen, 28706) according to the manufacturer’s directions and quantified by Nanodrop and Qubit (Invitrogen, 033226). Then, these samples were ready for NGS sequencing.

CRISPRa screen data analysis

We first reannotated the Human CRISPR lncRNA Activation Pooled Library as we previously reported¹⁹ using GENCODE v19 annotation⁷⁴. MAGeCK (v0.5.9.4) was used for sgRNA quantification, and the MAGeCK-mle algorithm⁷⁵ was then used for the candidate gene selection with 500 non-targeting control sgRNAs for normalization and generation of the null distribution. PCA was first performed based on sgRNA read counts in each sample to ensure the quality of screening. Comparison between sgRNA-transduced control samples with different culture periods was considered as proliferation screen, and three biological replicates that were treated with palbociclib for 14 days were compared with samples without drug treatment as drug treatment screen. Parameters were set as defaulted.

After the genome-wide screen analysis, we ranked the genes based on their FDR values for both positive and negative selections. The selection criteria for genes in the genome-wide drug treatment screen were set to FDR (BH-adjusted Wald p-values) less than 20%. To develop a focused sgRNA library, we integrated the results from genome-wide CRISPRa screen with cell line-based pharmacogenomics analyses (see below). This library included 25 lncRNAs that overlapped significant hits between the CRISPRa screening (FDR < 0.2), pharmacogenomics analyses (P < 0.05), and the top 50 lncRNAs uniquely identified in the CRISPRa drug screen. In addition, we included 33 PCGs known to regulate cell-cycle progression or palbociclib response, along with two additional lncRNAs previously reported to modulate palbociclib response.

After the two-stage CRISPRa screen, we have integrated the selected lncRNA’s expression in both cell lines (i.e., CCLE) and patient tumors (i.e., TCGA) to help remove false positives (Supplementary dataset 2). Specifically, the criteria of including lncRNAs for next step validation are 1. lncRNAs are significantly selected in both genome-wide and focused drug screens; 2. lncRNAs with detectable expression in the MCF-7 cell line; 3. lncRNAs with detectable expression in BRCA patient tumor samples; 4. The lncRNAs are significantly overexpressed/underexpressed in cancer cell lines and tumor samples (P < 0.05); 5. The lncRNAs’ expression are significantly associated with patient survival (P < 0.05). In total, we selected top three lncRNAs from positive/negative selection for further validation.

LncRNA expression, drug response, and patient survival association analyses across patient tumors and human cancer cell lines

Gene expression across cancer cell lines from CCLE was downloaded from DepMap. For drug response analysis, the expression of genes across 41 breast cancer cell lines from CCLE with matched palbociclib sensitivity data from GDSC³² was used. Spearman correlation coefficients between palbociclib area under the dose-response curve (AUC) values and gene expression levels were calculated, and p values were obtained. Candidate genes with P < 0.05 in both CRISPR screen and drug response analysis were selected as hits in our genome-wide screen.

For patient tumor analysis, TCGA⁷⁶ RNA-seq and clinical data were obtained from the Genomic Data Commons portal. The annotation for the transcriptome data was performed on the human reference genome GRCh38, and the gene expression level was then quantified by fragments per kilobase of transcript per million mapped reads upper quartile. LncRNA expression was compared between TCGA tumor samples and normal samples from the Genotype-Tissue Expression database⁷⁷ using a two-tailed unpaired Student’s t-test.

Exploratory RNA-seq analysis

Total RNAs were extracted from cells as described in the “RNA extraction” section. Then, RNAs were purified with the RNA Clean & Concentrator^TM-5 kit (Zymo Research, R1016). Then, at least 3 mg purified RNA was sent to Genewiz (lncRNA activation batch) or Novogene (lncRNA knockdown batch) for quality control, library construction, and sequencing. Reads were mapped to the human reference genome GRCh38 using spliced transcript alignment to a reference⁷⁸. Descriptive gene expression quantification and differential expression analysis were performed using RNA-seq by expectation maximization⁷⁹ and Cuffdiff⁸⁰ on the activation samples against their controls. Gene set enrichment analysis was then performed on the preranked gene lists by signed log-transformed Cuffdiff P values using the cancer hallmark gene set from the Molecular Signatures Database (MSigDB).

LncRNA activation signature analysis

Differentially expressed genes between lncRNA-activated samples in comparison with NT control sample were collected. A lncRNA activation signature was generated using upregulated genes and downregulated genes with Cuffdiff P < 0.05. We then calculated the weight $k_{i}$ for each selected signature gene using $k_{i} = \frac{w_{i}}{\max (∣w∣)}$ , where $w_{i}$ is the Cuffdiff test statistic for the ith selected gene and $k_{i}$ ∈ [−1,1]. With a batch of RNA-seq profiles in cell lines, we first obtained expression z scores for all of the genes across samples. Then, a lncRNA activation signature score S was calculated as the weighted sum expression of the signature genes using the equation S = ∑(k_i × z_i), where z_i denotes the expression level of the ith gene.

Single-nuclei RNA-seq and palbociclib treatment sample analyses

Expression profiles for PDX tumors and organoids derived from endocrine therapy-resistant and CDK4/6i-sensitive and palbociclib-resistant patients were obtained from GSE229235 and GSE229146, respectively. LncRNA activation scores were compared between palbociclib-resistant and sensitive samples using a two-tailed unpaired Student’s t-test.

To complement the bulk RNA-seq analysis, we analyzed publicly available single-nuclei RNA-seq (snRNA-seq) data from postmenopausal women with ER+ and/or PR+, HER2-negative breast tumors. Patients were randomized to receive either endocrine therapy alone (letrozole plus placebo) or continuous low-dose combination therapy (letrozole plus ribociclib, 400 mg/day) and tumor biopsies were collected at baseline (day 0) and post-treatment (day 14 or day 180). Processed data were obtained from GSE158724.

Only tumor cells were included for downstream analysis. Cells were filtered based on standard quality control thresholds: nFeature_RNA between 200 and 5000, and percent.mt below 5%. Samples with fewer than 200 high-quality cells were excluded. To integrate samples across patients, we applied Seurat’s reciprocal PCA method, as originally described in the dataset’s processing pipeline. After integration, we computed gene module scores for each cell using the AddModuleScore function in Seurat. Specifically, we evaluated: (1) an ER signaling signature based on the Hallmark Estrogen Response Early gene set from MSigDB, (2) custom-defined lncRNA activation signatures, and (3) cell cycle phase scores, including S phase and G2/M phase scores using Seurat’s default CellCycleScoring function. To examine the relationship between lncRNA activation and cell cycle regulation, we stratified tumor cells into ten bins based on their S score and G2/M score distributions and calculated the average lncRNA signature score within each bin. For treatment response analysis, we grouped patients by response status (responder vs. non-responder) within each treatment arm and compared the average lncRNA signature scores before and after treatment.

Single-sgRNA cloning and stable cell line generation

As for single sgRNA activation stable cell line, sgRNAs of selected lncRNAs and control NT sgRNAs were cloned into Lenti-sgRNA (MS2)-Puro plasmid (Addgene, 73795) at BsmBI sites. The sgRNA sequence for activating target lncRNAs and control sgRNAs is listed in Supplementary dataset 6. The top six ranked sgRNAs from both global and focused libraries were picked for each selected gene according to the analysis (see “Methods”). HEK293T cells were used to produce sgRNA lentivirus. HEK293T cells were seeded one day before the transfection and reached 70%–80% confluency on the next day. The transfection was performed using the method from our previously published paper¹⁹. Lipofectamine 2000 transfection reagent (Invitrogen, 11668-019) was used according to the manufacturer’s instructions. The plasmid mix contained 4.5 μg of psPAX2 (Addgene, 12260), 1.5 μg of pMD2.G (Addgene, 12259), and 6 μg of sgRNA plasmids (each gene used six sgRNA plasmids with 1:1 ratio mix). After 48 h of transfection, the lentivirus was harvested and filtered with 0.45 μm filters. Then, MCF-7-VP64-MPH and T-47D-VP64-MPH cells were infected with lentivirus corresponding to different genes’ activation for 24 h, then selected with puromycin until no negative-control cells (no virus-infected cells) survived. The stable cells were ready to validate and use.

RNA interference

siControl and siRNA targeted TENM3-AS1, LINC01117, and ENSG00000226706 were transfected to the MCF-7, RKO, U2OS, and HS578T cells using Lipofectamine RNAiMAX (Invitrogen, 13778-150) according to the manufacturer’s instructions. In summary, 60 pmol of each siRNA for the targeted gene or control was given to each well of the 6-well plate. Cells were seeded one day before the transfection and reached 40% confluency on the next day. The cell culture medium was replaced before the transfection. After 48 h, RNA and protein were separately extracted from each group of cells. The siRNA sequences were listed as follows: siControl: Santa Cruz (sc-37007); siTENM3 (three siRNAs): MCE (HY-RS14364); siTENM3-AS1-A: 5′-GCCGGCUUUCGGCUCUGAU-3′; siTENM3-AS1-B: 5′-GACGGAUUCCGAGGCAGAA-3′; siLINC01117-A: 5′-GGCAGGAGAGAGAGCACAC-3′; siLINC01117-B: 5′-GGCGAUACGUGAUCAUUUA-3′; si226706-A: 5′-CAGCAAGCGCCCACUCGUA-3′; si226706-C: 5′- GAGCAAACGCACAUGGCAA-3′; siESR1: 5′-CAGGCACAUGAGUAACAAA-3′.

Generation of the stable cell line

The TENM3-AS1 cDNA was synthesized by Twist Bioscience. First, synthesized cDNA was amplified by PCR with primers designed with XbaI and XhoI restriction sites included. Then, amplified fragments were assembled into the pCDH-CMV-MCS-EF1-Puro backbone (System Biosciences, CD510B-1) after treated with XbaI and XhoI restriction enzymes using Gibson Assembly® Master Mix (NEB, E2611). Recombinant plasmids were transformed into DH5α competent cells, and the sequence from positive clones were verified by Sanger sequencing. Lentiviral particles were produced in HEK293T cells. Cells were seeded one day before transfection and reached 70 ~ 80% confluency at the time of transfection in a 6 cm dish. A plasmid mix containing 4.5 μg psPAX2 (Addgene, 12260), 1.5 μg pMD2.G (Addgene, 12259), and 6 μg TENM3-AS1/ZsGreen (control) cDNA plasmid was transfected using Lipofectamine 2000 (Invitrogen, 11668-019) according to the manufacturer’s instructions. Viral supernatants were harvested 48 h post-transfection and filtered with a 0.45-μm strainer. T-47D cells were infected in the presence of 8 μg/mL polybrene and selected with puromycin. Stable TENM3-AS1 expression was confirmed by PCR. PCR was performed using DreamTaq™ PCR Master Mix (2X) (Thermo Scientific, K1061) according to the manufacturer’s instructions to confirm the presence of TENM3-AS1 transcripts.

RNA extraction and quantitative real-time PCR (qRT-PCR)

Total RNA was extracted with TRIzol. After centrifuging at 12,000 × g for 15 min at 4 °C, the supernatant was acquired and treated with isopropanol to precipitate the RNA. RNA pellets were washed twice with 75% ethanol. After air drying of the RNA pellets, nuclease-free water was used to dissolve RNA, and then quantified by Nanodrop. As for cDNA synthesis, 1 μg of RNA was used from above, and then a prepared mix using Reverse Transcription Kit (Applied Biosystems, 4368813) according to the manufacturer’s directions, and then performed reverse transcription in a PCR machine. cDNA was diluted in a 1:5 ratio, and qPCR was performed using SYBR Green PCR 2′ Master Mix (Applied Biosystems, 4367659) on a 384-well plate. Then. The plate was inserted in the QuantStudioTM 6 Pro Real-Time PCR equipment (Applied Biosystems). The relative RNA expression and fold change were determined by normalizing to GAPDH. The primer sequences are listed below: Human GAPDH forward, 5′-GGTGAAGGTCGGAGTCAACG-3′; reverse, 5′-TGGGTGGAATCATATTGGAACA-3′. Human TENM3-AS1 forward, 5′-ACACGCACCAACAAAATGGG-3′; reverse, 5′-GAATCCGTCTTCCCGGCAG-3′. Human LINC01117 forward, 5′-CCCGATCCTCGGCATCCTT-3′; reverse, 5′-GGCCTTCTTTTGCAGCAAGA-3′. Human ENSG00000226706 forward, 5′-CATCAGCGCAAGAGCAAACG-3′; reverse, 5′-TGCAGACATCCACCGTTCCA-3′. Human CCND1 forward, 5′-TCTACACCGACAACTCCATCCG-3′; reverse, 5′-TCTGGCATTTTGGAGAGGAAGTG-3′. Human CCND2 forward, 5′- TGATGTTGGGCAACTCTGCG-3′; reverse, 5′- GTGCAACCCGTCTCGTCT-3′. Human CCNB1 forward, 5′-GACCTGTGTCAGGCTTTCTCTG-3′; reverse, 5′-GGTATTTTGGTCTGACTGCTTGC-3′. Human ESR1 forward, 5′-GCTTACTGACCAACCTGGCAGA-3′; reverse, 5′-GGATCTCTAGCCAGGCACATTC′. Human MYC forward, 5′-AATGAAAAGGCCCCCAAGGTAG-3′; reverse, 5′-GTCGTTTCCGCAACAAGTCCT-3′. Human TFF1 forward, 5′-CATCGACGTCCCTCCAGAAGAG-3′; reverse, 5′-CTCTGGGACTAATCACCGTGCTG-3′. Human GREB1 forward, 5′-ATCAGCTGCTCGGACTTGCTG-3′; reverse, 5′-TGAGCTCCGGTCCTGACAGAT-3′. Human TENM3 forward, 5′- GGACAAGGCTATCACAGTGGAC-3′; reverse, 5′- TTCTGAGGGAGCCGTCATAACC-3′.

Western blot

Whole cell lysate was extracted from cells using RIPA lysis buffer supplemented with freshly added protease inhibitor (MCE, HY-K0010). After centrifuging tubes at 16,000 × g, 15 min at 4 °C, cell lysate was acquired, and the concentration was determined by the BCA protein assay kit (ThermoFisher, 23225) and incubated at 37 °C for 30 min. Loading samples were prepared according to the concentration and supplemented with the 5′ SDS loading buffer. Samples were then denatured at 98 °C for 10 min, loaded to the SDS-PAGE gel, and further transferred to the PVDF membrane (Bio-Rad, 162-0177). Membranes were incubated in the block solution, 5% non-fat milk (LabScientific, M0841) in phosphate-buffered saline containing 0.2% Triton X-100 (PBST) solution for 1 h at room temperature on a shaker. Then, membranes were incubated in the primary antibody overnight at 4 °C (see “Method” antibodies section). On the next day, membranes were washed with 1 × PBST three times and then incubated in HRP-conjugated secondary antibodies for 1 h at room temperature on a shaker. After another three times washing with PBST, membranes were treated with chemiluminescent substrate (ECL) (ThermoFisher, F32106) and then developed the band signal on the X-ray films using an AX 700LE film processor (ALPHATEk).

RNA Immunoprecipitation (RIP)

RNA immunoprecipitation (RIP) was performed to detect RNAs associated with ERα and MYC, with normal mouse IgG and rabbit IgG as controls based on our previous publication¹⁹. Approximately 2 × 10⁷ cells were harvested, washed once with PBS, and crosslinked with 0.3% formaldehyde (diluted with 16% formaldehyde, ThermoFisher, 28906) for 10 min at room temperature with rotation. Crosslinking was quenched with 1.25 M glycine (Bio-Rad, 1610718) for 5 min. Pellets were collected by centrifugation at 3200 × g for 3 min at 4 °C. Cells were lysed in the lysis buffer (50 mM Tris-HCl, pH 7.4), 100 mM NaCl, 1% NP-40, 0.1% SDS, and 0.5% sodium deoxycholate) supplemented with protease (MCE, HY-K0010) and RNase inhibitor (NEB, M0314L). Then, cells were treated with deoxyribonuclease (2 μL/mL; LifeTech, AM2239) and incubated for 20 min at room temperature. After centrifuging, the supernatant was saved, and the concentration was determined using the BCA Protein Assay Kit (Thermo Fisher Scientific, 23225). Magnetic beads were pre-incubated with 5 µg antibody (ERα, MYC, or IgG) overnight at 4 °C. Sheep anti-mouse IgG beads (Invitrogen, 11201) were used for ERα, and Protein G (Invitrogen, 11201) beads for MYC. Antibody-bead complexes were then incubated with 2 mg cell lysate overnight at 4 °C with rotation. Beads were collected using a magnetic stand and washed six times with the lysis buffer. Approximately one-tenth of the beads was mixed 1:1 with 2 × SDS sample buffer (Bio-Rad, 1610737) and analyzed by Western blotting, as described above. The remaining beads were incubated in RIPA buffer at 70 °C for 1 h, followed by RNA extraction and analysis by qRT-PCR for RNA enrichment, as described above.

Antibodies

Antibodies listed below were used for Western blot flow cytometry and RNA immunoprecipitation. For immunoblotting, rabbit anti-CCND1 (CST, 555067), rabbit anti-CCND2 (CST, 3741), rabbit anti-CCNB1 (CST, 4138), rabbit anti-MYC (CST, 13987), rabbit anti-GREB1 (CST, 65171), mouse anti-ERα (Santa Cruz, sc-8002), mouse anti-β-actin (Sigma, A5441), Goat anti-Mouse IgG (H + L) Secondary Antibody, HRP (Invitrogen, 31430), Goat anti-Rabbit IgG (H + L) Secondary Antibody, HRP (Invitrogen, 31460). For RNA immunoprecipitation, rabbit anti-MYC (CST, 13987), mouse anti-ERα (Santa Cruz, sc-8002), normal rabbit IgG (CST, 2729), and normal mouse IgG (Santa Cruz, sc-2025).

MTT assay

With palbociclib

A total of 3000 cells per well were seeded into the 96-well plates with at least three biological replicates for each drug concentration for each targeted gene-activated cell line for MCF-7 cells (5000 cells per well for siRNA-treated knockdown experiments). For T-47D cells, 7000 cells were seeded into the 96-well plates. On the second day, cells were treated with different concentrations of palbociclib. Two siRNAs shown in Supplementary Fig. 3c were used in a 1:1 mix ratio for the siRNA-treated experiments. After 48 h and 72 h, MTT reagent (Biosynth, T-3450) was added to each well with the final concentration of 1×. Cells were incubated at 37 °C and 5% CO₂ for 4 h. Then MTT solubilization solution (40% dimethylformamide (DMF), 2% acetic acid, 16% SDS) was treated to each well. After overnight incubation, the absorbance of the plate was measured by a microtiter plate reader at 570 nm and 630 nm wavelengths. Then, the data was analyzed by normalizing with the zero palbociclib concentration group for each cell line.

Cell proliferation

For targeted gene activation, 1000 MCF-7 cells and 2000 T-47D cells were seeded into the 96-well plates. For siRNA-treated knockdown experiments, 3000 MCF-7 cells, 7000 HS578T cells, 5000 U2OS cells, and 6500 RKO cells were seeded into 96-well plates. A total of six plates were used for each cell line and harvested every day continuously (including day 0). The rest of the process was the same as described above. Then, the data was analyzed by normalizing with the day one readout and compared with the siControl group.

Colony formation assay

With palbociclib

A total of 10,000 MCF-7 cells stably activated targeted genes or controls were seeded into 12-well plates and allowed to attach overnight. On the second day, cells were treated with different concentrations of palbociclib (0, 0.125, 0.5, and 2 μM). For T-47D cells, 25,000 cells were seeded into the 12-well plates, and cells were treated with different concentrations of palbociclib (0, 0.5, 2, and 8 μM). Medium and drug will be replaced every 2 days. All the cells will be cultured for 7 days. Then, cells were washed with PBS once and fixed with 4% formaldehyde for 10–15 min at room temperature. Then, cells will be washed with PBS once and stained with 0.01% crystal violet-25% methanol solution for at least 20 min. Plates with crystal violet staining will be scanned and analyzed. The quantification of the colony was performed through ImageJ software, and each stable cell line will be normalized with the group that was not treated with palbociclib.

Cell proliferation

The same cell number of MCF-7 and T-47D cells described above was seeded into 12-well plates as described above, but each cell line was seeded into two plates. One plate will be collected on the second day, as day 0, for further analysis. The other plate will be cultured for 7 days and then collected as described above. The medium will be changed every two days. Each stable cell line will be normalized with the data from day 0 of its own.

Cell cycle assay

1 × 10⁶ MCF-7 cells with stably activated targeted genes were collected and washed with PPBS once. Tubes were centrifuged at 200 × g for 10 min at 4 °C and aspirated the supernatant after centrifuging. Cells were fixed with 75% ice-cold ethanol by adding drop by drop to the slowly, continuously vortex tubes. Then, incubated cells for at least 2 h or overnight. Cells were washed with PBS once and then washed with the FBS Stain buffer (BD Biosciences, 554656) once. Cells were centrifuged at 200 × g for 10 min at 4 °C. After aspirating the supernatant, cell pellets were resuspended in 0.5 mL of PI/RNase staining buffer (BD Biosciences, 550825) and incubated for 15 min at room temperature. Flow cytometry was then performed, and data was analyzed with the FlowJo software.

In vivo xenograft model

All animal studies were performed following the institutional guidelines. All related experiments were done following the animal protocols approved by the IACUC of the University of Pittsburgh, protocol #22030771 (subsequently renewed as protocol #25036427). All mice were housed in specific pathogen-free facilities under controlled conditions (22–24 °C, 45% humidity, 12/12 light/dark cycle), with food and water available ad libitum. Around 6-week-old female athymic nude mice were purchased from Charles River lab (Wilmington, MA). Early passage of MCF-7 sgNTs and lncRNA-activated cells were cultured. 100 µL of five million MCF-7 cells were mixed with Matrigel (Corning, 354234) with a 1:1 ratio. Then, 200 µL in total were inoculated through fat pad injection. The tumor size was measured using a caliper every three days and tumor volume was calculated using the formula: length × width² × 0.5. Mice were sacrificed, and tumors were harvested from mice reaching predefined endpoints. Tumor weight was also determined after scarification. The ethical endpoint was defined as tumors reaching 20 mm or more in any dimension or a tumor volume exceeding 1000 mm³. At no point did tumor size or burden exceed the limits approved by the IACUC. Only female mice were used in this study, as MCF-7 xenograft tumor establishment and growth are estrogen-dependent, and breast cancer occurs predominantly in biological females ( > 99% according to WHO estimates).

Tissue immunofluorescence

After tumors were harvested from mice, the partial tumor will be placed in the tissue cassette (Fisherbrand, 15182701k) and perform tissue processing. Then, the tissue will be embedded into the disposable base molds (Fisherbrand, 22363553) with paraffin. Then, paraffin blocks will be sliced (4–5 nm) and placed on the coverglass (Leica, 3800145ACS). Then, the paraffin will be melted and wiped out by triple incubation in the xylene (Fisher Chemical, X3P-1GAL). Then, tissues will be rehydrated with the following steps: two times 100% Ethanol, two times 95% Ethanol, one time 70% Ethanol, 50% Ethanol, 30% Ethanol, and pure water. Tissues will then be boiled in Sodium citrate buffer (Sigma, W302600) and washed three times with PBS. Then, tissues will be blocked with 5% goat serum for 1 h. Ki-67 samples will be permeabilized with 0.5% Triton X-100 simultaneously. Tissues will be washed with PBS another three times and then incubated with primary antibodies in 2% BSA at 4 °C overnight. Primary antibodies information: Ki-67(CST, 96344), MYC (CST, 13987), ERα (Santa Cruz, sc-8002) with 1:100 dilution. Secondary antibodies information: Goat Anti-Rabbit IgG H&L (Cy3®) preadsorbed (Abcam, ab97075), Goat Anti-Mouse IgG H&L - Alexa Fluor® 488 (Abcam, ab150117). The next day, tissues were washed with PBS three times and mounted with media containing DAPI and fluorescence antifade agent (Invitrogen, P36935). Tissues will be sealed with cover glasses and detected the signal using a fluorescence microscope.

Statistical analysis

The association between gene expression and overall survival rates was assessed by Cox regression analyses⁸¹ in TCGA cancer patient samples. Analysis was performed using Python 3.8.0 and GraphPad Prism 10.6.1. The statistical test used is indicated in each figure legend. The significance threshold was defined as P < 0.05. The Benjamini–Hochberg adjustment was used to calculate the false discovery rates.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Supplementary information

Supplementary information^{(14.8MB, pdf)}

41467_2026_70816_MOESM2_ESM.pdf^{(195.2KB, pdf)}

Description of Additional Supplementary Files

Supplementary Dataset1^{(1.6MB, xlsx)}

Supplementary Dataset2^{(55.7KB, xlsx)}

Supplementary Dataset3^{(3.5MB, xlsx)}

Supplementary Dataset4^{(26.2KB, xlsx)}

Supplementary Dataset5^{(67.8KB, xlsx)}

Supplementary Dataset6^{(11.5KB, xlsx)}

Reporting Summary^{(3.7MB, pdf)}

Transparent Peer Review file^{(8.8MB, pdf)}

Acknowledgements

This study was supported by the Shear Family Foundation (to D.Y.) and National Cancer Institute (R01CA255196, R01CA272866, and R01CA282704 to D.Y.). This research was partly supported by the University of Pittsburgh Center for Research Computing, RRID: SCR_022735, through the resources provided. Specifically, this work used the HTC cluster, supported by NIH award number S10OD028483.

Author contributions

Conceptualization: D.Y., Yifei W., Yueshan Z. Methodology: Yifei W., Yueshan Z., Z.W., S.L., M.Z., and D.Y. Validation: Yifei W., J.H., Z.W., D.P., Yu Z., and X.W. Formal analysis: Yueshan Z. and M.Z. Resources: S.L., Yue W., and W.X. Visualization: Yifei W., Yueshan Z., M.Z., and D.Y. Supervision: D.Y. and M.Z. Writing (original draft): Yifei W., Yueshan Z., D.Y., and M.Z. Writing (review and editing): Yifei W., Yueshan Z., J.H., D.Y., and M.Z.

Peer review

Peer review information

Nature Communications thanks Juliet French, Yin Shen, and the other anonymous reviewers for their contribution to the peer review of this work. [A peer review file is available].

Data availability

The sequencing data generated in this study have been deposited in the Gene Expression Omnibus (GEO) under accession numbers GSE253290, GSE254283, and GSE279981. Source data are provided with this paper.

Code availability

The code used for CRISPRa screen hit analysis, RNA-seq pathway analysis, and snRNA-seq integration in this study is available from GitHub (https://github.com/dyanglab/CRISPRa_CDKi)⁸².

Competing interests

The authors declare no competing interests.

Footnotes

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

These authors contributed equally: Yifei Wang, Yueshan Zhao, Jiaxin Hu.

Contributor Information

Min Zhang, Email: miz45@pitt.edu.

Da Yang, Email: dyang@pitt.edu.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-70816-2.

References

1.Perou, C. M. et al. Molecular portraits of human breast tumours. Nature406, 747–752 (2000). [DOI] [PubMed] [Google Scholar]
2.Cancer Genome Atlas, N. Comprehensive molecular portraits of human breast tumours. Nature490, 61–70 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
3.Cancer Genome Atlas Research, N. Integrated genomic characterization of papillary thyroid carcinoma. Cell159, 676–690 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
4.Djebali, S. et al. Landscape of transcription in human cells. Nature489, 101–108 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
5.Schmitt, A. M. & Chang, H. Y. Long noncoding RNAs in cancer pathways. Cancer Cell29, 452–463 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Iyer, M. K. et al. The landscape of long noncoding RNAs in the human transcriptome. Nat. Genet.47, 199–208 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
7.Reon, B. J. et al. Expression of lncRNAs in low-grade gliomas and glioblastoma multiforme: an in silico analysis. PLoS Med.13, e1002192 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
8.Verma, A. et al. Transcriptome sequencing reveals thousands of novel long non-coding RNAs in B cell lymphoma. Genome Med.7, 110 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
9.Wang, Z. et al. lncRNA epigenetic landscape analysis identifies EPIC1 as an oncogenic lncRNA that interacts with MYC and promotes cell-cycle progression in cancer. Cancer Cell33, 706–720.e709 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Wang, D. et al. LncRNA MALAT1 enhances oncogenic activities of EZH2 in castration-resistant prostate cancer. Oncotarget6, 41045–41055 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
11.Marques Howarth, M. et al. Long noncoding RNA EWSAT1-mediated gene repression facilitates Ewing sarcoma oncogenesis. J. Clin. Invest.124, 5275–5290 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Agrelo, R. et al. SATB1 defines the developmental context for gene silencing by Xist in lymphoma and embryonic cells. Dev. Cell16, 507–516 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Lee, S. et al. Noncoding RNA NORAD regulates genomic stability by sequestering PUMILIO Proteins. Cell164, 69–80 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
14.Ling, H. et al. CCAT2, a novel noncoding RNA mapping to 8q24, underlies metastatic progression and chromosomal instability in colon cancer. Genome Res.23, 1446–1461 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Tsung, K. et al. CRISPRi screen of long non-coding RNAs identifies LINC03045 regulating glioblastoma invasion. PLoS Genet.20, e1011314 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
16.Nath, A. et al. Discovering long noncoding RNA predictors of anticancer drug sensitivity beyond protein-coding genes. Proc. Natl. Acad. Sci. USA116, 22020–22029 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
17.Deng, X. et al. Long noncoding RNA PiHL regulates p53 protein stability through GRWD1/RPL11/MDM2 axis in colorectal cancer. Theranostics10, 265–280 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Fang, Z. et al. ERINA is an estrogen-responsive LncRNA that drives breast cancer through the E2F1/RB1 Pathway. Cancer Res.80, 4399–4413 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Wang, Y. et al. Genome-wide gain-of-function screening characterized lncRNA regulators for tumor immune response. Sci. Adv.8, eadd0005 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Guo, W. et al. LincRNA-immunity landscape analysis identifies EPIC1 as a regulator of tumor immune evasion and immunotherapy resistance. Sci. Adv.7, eabb3555 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
21.Xue, X. et al. LncRNA HOTAIR enhances ER signaling and confers tamoxifen resistance in breast cancer. Oncogene35, 2746–2755 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
22.Wang, Y. et al. Systematic identification of non-coding pharmacogenomic landscape in cancer. Nat. Commun.9, 3192 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Han, Y. et al. LEIGC long non-coding RNA acts as a tumor suppressor in gastric carcinoma by inhibiting the epithelial-to-mesenchymal transition. BMC Cancer14, 932 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]
24.Replogle, J. M. et al. Mapping information-rich genotype-phenotype landscapes with genome-scale perturb-seq. Cell185, 2559–2575.e2528 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Seifuddin, F. et al. lncRNAKB, a knowledgebase of tissue-specific functional annotation and trait association of long noncoding RNA. Sci. Data7, 326 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Johnsson, P. et al. Transcriptional kinetics and molecular functions of long noncoding RNAs. Nat. Genet.54, 306–317 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Doench, J. G. Am I ready for CRISPR? A user’s guide to genetic screens. Nat. Rev. Genet.19, 67–80 (2018). [DOI] [PubMed] [Google Scholar]
28.Huang, J., Zheng, L., Sun, Z. & Li, J. CDK4/6 inhibitor resistance mechanisms and treatment strategies (Review). Int. J. Mol. Med.50, 128 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Wang, X. et al. Recent progress of CDK4/6 inhibitors’ current practice in breast cancer. Cancer Gene Ther.31, 1283–1291 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
30.Joung, J. et al. Genome-scale activation screen identifies a lncRNA locus regulating a gene neighbourhood. Nature548, 343–346 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
31.Barretina, J. et al. The cancer cell line encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature483, 603–607 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Yang, W. et al. Genomics of drug sensitivity in cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells. Nucleic Acids Res.41, D955–D961 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Park, Y. H. et al. Longitudinal multi-omics study of palbociclib resistance in HR-positive/HER2-negative metastatic breast cancer. Genome Med.15, 55 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Costa, C. et al. PTEN loss mediates clinical cross-resistance to CDK4/6 and PI3Kα inhibitors in breast cancer. Cancer Discov.10, 72–85 (2020). [DOI] [PubMed] [Google Scholar]
35.Kietzman, W. B. et al. Short- and long-term Effects of CDK4/6 inhibition on early-stage breast cancer. Mol. Cancer Ther.18, 2220–2232 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]
36.Spencer, S. abrinaL. et al. The proliferation-quiescence decision is controlled by a bifurcation in CDK2 activity at mitotic exit. Cell155, 369–383 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
37.Malumbres, M. et al. Mammalian cells cycle without the D-type cyclin-dependent kinases Cdk4 and Cdk6. Cell118, 493–504 (2004). [DOI] [PubMed] [Google Scholar]
38.Baker, S. J. & Reddy, E. P. CDK4: a key player in the cell cycle, development, and cancer. Genes Cancer3, 658–669 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Zhang, H. et al. Effect of CCNB1 silencing on cell cycle, senescence, and apoptosis through the p53 signaling pathway in pancreatic cancer. J. Cell. Physiol.234, 619–631 (2019). [DOI] [PubMed] [Google Scholar]
40.Büschges, R. et al. Amplification and expression of cyclin D genes (CCND1 CCND2 and CCND3) in human malignant gliomas. Brain Pathol.9, 435–442 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Bénard, J., Douc-Rasy, S. & Ahomadegbe, J.-C. TP53 family members and human cancers. Hum. Mutat.21, 182–191 (2003). [DOI] [PubMed] [Google Scholar]
42.O’Leary, B., Finn, R. S. & Turner, N. C. Treating cancer with selective CDK4/6 inhibitors. Nat. Rev. Clin. Oncol.13, 417–430 (2016). [DOI] [PubMed] [Google Scholar]
43.Cao, J. et al. Targeting estrogen-regulated system xc− promotes ferroptosis and endocrine sensitivity of ER+ breast cancer. Cell Death Dis.16, 30 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Nagandla, H. & Thomas, C. A.-O. Estrogen signals through ERβ in breast cancer; what we have learned since the discovery of the receptor. Receptors (Basel)3, 182–200 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Murphy, C. G. & Dickler, M. N. The role of CDK4/6 inhibition in breast cancer. Oncologist20, 483–490 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Álvarez-Fernández, M. & Malumbres, M. Mechanisms of sensitivity and resistance to CDK4/6 inhibition. Cancer Cell37, 514–529 (2020). [DOI] [PubMed] [Google Scholar]
47.Griffiths, J. I. et al. Serial single-cell genomics reveals convergent subclonal evolution of resistance as patients with early-stage breast cancer progress on endocrine plus CDK4/6 therapy. Nat. Cancer2, 658–671 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
48.Navarro-Yepes, J. et al. Abemaciclib is effective in palbociclib-resistant hormone receptor-positive metastatic breast cancers. Cancer Res.83, 3264–3283 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Goldman, J. W. et al. Treatment rationale and study design for the JUNIPER study: a randomized phase III study of abemaciclib with best supportive care versus erlotinib with best supportive care in patients with stage IV non-small-cell lung cancer with a detectable KRAS mutation whose disease has progressed after platinum-based chemotherapy. Clin. Lung Cancer17, 80–84 (2015). [DOI] [PubMed] [Google Scholar]
50.Geoerger, B. et al. A phase I study of the CDK4/6 inhibitor ribociclib (LEE011) in pediatric patients with malignant rhabdoid tumors, neuroblastoma, and other solid tumors. Clin. Cancer Res.23, 2433–2441 (2017). [DOI] [PubMed] [Google Scholar]
51.Patnaik, A. et al. Efficacy and safety of abemaciclib, an inhibitor of CDK4 and CDK6, for patients with breast cancer, non-small cell lung cancer, and other solid tumors. Cancer Discov.6, 740–753 (2016). [DOI] [PubMed] [Google Scholar]
52.Finn Richard, S. et al. Palbociclib and letrozole in advanced breast cancer. N. Engl. J. Med.375, 1925–1936 (2016). [DOI] [PubMed] [Google Scholar]
53.Turner, N. C. et al. Palbociclib in hormone-receptor-positive advanced breast cancer. N. Engl. J. Med.373, 209–219 (2015). [DOI] [PubMed] [Google Scholar]
54.Finn, R. S. et al. PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human breast cancer cell lines in vitro. Breast Cancer Res.11, R77 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]
55.Olanich, M. E. et al. CDK4 amplification reduces sensitivity to CDK4/6 inhibition in fusion-positive rhabdomyosarcoma. Clin. Cancer Res.21, 4947–4959 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
56.Toulmonde, M. et al. Activity and safety of palbociclib in patients with advanced gastrointestinal stromal tumors refractory to imatinib and sunitinib: a biomarker-driven phase II study. Clin. Cancer Res.25, 4611–4615 (2019). [DOI] [PubMed] [Google Scholar]
57.Leonard, J. P. et al. Selective CDK4/6 inhibition with tumor responses by PD0332991 in patients with mantle cell lymphoma. Blood119, 4597–4607 (2012). [DOI] [PubMed] [Google Scholar]
58.Sanson, K. R. et al. Optimized libraries for CRISPR-Cas9 genetic screens with multiple modalities. Nat. Commun.9, 5416 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]
59.Andersen, C. L. et al. Active estrogen receptor-alpha signaling in ovarian cancer models and clinical specimens. Clin. Cancer Res.23, 3802–3812 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
60.Chen, Z. et al. Estrogen-ERα signaling and DNA hypomethylation co-regulate expression of stem cell protein PIWIL1 in ERα-positive endometrial cancer cells. Cell Commun. Signal.18, 84 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
61.Lafront, C. et al. The estrogen signaling pathway reprograms prostate cancer cell metabolism and supports proliferation and disease progression. J. Clin. Invest.10.1172/JCI170809 (2024). [DOI] [PMC free article] [PubMed]
62.He, M. et al. Estrogen receptor α promotes lung cancer cell invasion via increase of and cross-talk with infiltrated macrophages through the CCL2/CCR2/MMP9 and CXCL12/CXCR4 signaling pathways. Mol. Oncol.14, 1779–1799 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]
63.You, H., Meng, K. & Wang, Z.-Y. The ER-α36/EGFR signaling loop promotes growth of hepatocellular carcinoma cells. Steroids134, 78–87 (2018). [DOI] [PubMed] [Google Scholar]
64.Topi, G. et al. High Oestrogen receptor alpha expression correlates with adverse prognosis and promotes metastasis in colorectal cancer. Cell Commun. Signal.22, 198 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
65.JavanMoghadam, S., Weihua, Z., Hunt, K. K. & Keyomarsi, K. Estrogen receptor alpha is cell cycle-regulated and regulates the cell cycle in a ligand-dependent fashion. Cell Cycle15, 1579–1590 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]
66.Heldring, M. M. et al. Interdependency of estradiol-mediated ERα activation and subsequent PR and GREB1 induction to control cell cycle progression. Heliyon10, e38406 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
67.Henninger, J. E. & Young, R. A. An RNA-centric view of transcription and genome organization. Mol. Cell84, 3627–3643 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
68.Oksuz, O. et al. Transcription factors interact with RNA to regulate genes. Mol. Cell83, 2449–2463.e2413 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]
69.Li, S. et al. Integrative characterization of MYC RNA-binding function. Cell Genomics10.1016/j.xgen.2025.100878 (2025). [DOI] [PMC free article] [PubMed]
70.Wang, X. et al. Cytoplasmic PXR regulates glucose metabolism by binding mRNAs and modulating their stability. Nat. Struct. Mol. Biol.10.1038/s41594-025-01614-5 (2025). [DOI] [PMC free article] [PubMed]
71.Xu, Y. et al. ERa is an RNA-binding protein sustaining tumor cell survival drug resistance. Cell184, 5215–5229.e5217 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]
72.Soota, D., Saravanan, B., Mann, R., Kharbanda, T. & Notani, D. RNA fine-tunes estrogen receptor-alpha binding on low-affinity DNA motifs for transcriptional regulation. EMBO J.43, 5186–5210-5210 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]
73.Joung, J. et al. Genome-scale CRISPR-Cas9 knockout and transcriptional activation screening. Nat. Protoc.12, 828–863 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]
74.Fujita, P. A. et al. The UCSC Genome Browser database: update 2011. Nucleic Acids Res.39, D876–D882 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
75.Li, W. et al. Quality control, modeling, and visualization of CRISPR screens with MAGeCK-VISPR. Genome Biol.16, 281 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]
76.Chang, K. et al. The Cancer Genome Atlas Pan-cancer analysis project. Nat. Genet.45, 1113–1120 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
77.Lonsdale, J. et al. The genotype-tissue expression (GTEx) project. Nat. Genet.45, 580–585 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
78.Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics29, 15–21 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]
79.Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinforma.12, 323 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]
80.Anders, S. & Huber, W. Differential expression analysis for sequence count data. Genome Biol.11, R106 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]
81.Cox, D. R. Regression models and life-tables. J. R. Stat. Soc.: Ser. B (Methodol.)34, 187–202 (1972). [Google Scholar]
82.Zhao, Y. Code Repository for CRISPRa_cdkscreen. GitHub10.5281/zenodo.18118851 (2026).

Associated Data

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

Supplementary Materials

Supplementary information^{(14.8MB, pdf)}

41467_2026_70816_MOESM2_ESM.pdf^{(195.2KB, pdf)}

Description of Additional Supplementary Files

Supplementary Dataset1^{(1.6MB, xlsx)}

Supplementary Dataset2^{(55.7KB, xlsx)}

Supplementary Dataset3^{(3.5MB, xlsx)}

Supplementary Dataset4^{(26.2KB, xlsx)}

Supplementary Dataset5^{(67.8KB, xlsx)}

Supplementary Dataset6^{(11.5KB, xlsx)}

Reporting Summary^{(3.7MB, pdf)}

Transparent Peer Review file^{(8.8MB, pdf)}

Data Availability Statement

The code used for CRISPRa screen hit analysis, RNA-seq pathway analysis, and snRNA-seq integration in this study is available from GitHub (https://github.com/dyanglab/CRISPRa_CDKi)⁸².

[CR1] 1.Perou, C. M. et al. Molecular portraits of human breast tumours. Nature406, 747–752 (2000). [DOI] [PubMed] [Google Scholar]

[CR2] 2.Cancer Genome Atlas, N. Comprehensive molecular portraits of human breast tumours. Nature490, 61–70 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR3] 3.Cancer Genome Atlas Research, N. Integrated genomic characterization of papillary thyroid carcinoma. Cell159, 676–690 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR4] 4.Djebali, S. et al. Landscape of transcription in human cells. Nature489, 101–108 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR5] 5.Schmitt, A. M. & Chang, H. Y. Long noncoding RNAs in cancer pathways. Cancer Cell29, 452–463 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR6] 6.Iyer, M. K. et al. The landscape of long noncoding RNAs in the human transcriptome. Nat. Genet.47, 199–208 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR7] 7.Reon, B. J. et al. Expression of lncRNAs in low-grade gliomas and glioblastoma multiforme: an in silico analysis. PLoS Med.13, e1002192 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR8] 8.Verma, A. et al. Transcriptome sequencing reveals thousands of novel long non-coding RNAs in B cell lymphoma. Genome Med.7, 110 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR9] 9.Wang, Z. et al. lncRNA epigenetic landscape analysis identifies EPIC1 as an oncogenic lncRNA that interacts with MYC and promotes cell-cycle progression in cancer. Cancer Cell33, 706–720.e709 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR10] 10.Wang, D. et al. LncRNA MALAT1 enhances oncogenic activities of EZH2 in castration-resistant prostate cancer. Oncotarget6, 41045–41055 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR11] 11.Marques Howarth, M. et al. Long noncoding RNA EWSAT1-mediated gene repression facilitates Ewing sarcoma oncogenesis. J. Clin. Invest.124, 5275–5290 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR12] 12.Agrelo, R. et al. SATB1 defines the developmental context for gene silencing by Xist in lymphoma and embryonic cells. Dev. Cell16, 507–516 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR13] 13.Lee, S. et al. Noncoding RNA NORAD regulates genomic stability by sequestering PUMILIO Proteins. Cell164, 69–80 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR14] 14.Ling, H. et al. CCAT2, a novel noncoding RNA mapping to 8q24, underlies metastatic progression and chromosomal instability in colon cancer. Genome Res.23, 1446–1461 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR15] 15.Tsung, K. et al. CRISPRi screen of long non-coding RNAs identifies LINC03045 regulating glioblastoma invasion. PLoS Genet.20, e1011314 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR16] 16.Nath, A. et al. Discovering long noncoding RNA predictors of anticancer drug sensitivity beyond protein-coding genes. Proc. Natl. Acad. Sci. USA116, 22020–22029 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR17] 17.Deng, X. et al. Long noncoding RNA PiHL regulates p53 protein stability through GRWD1/RPL11/MDM2 axis in colorectal cancer. Theranostics10, 265–280 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR18] 18.Fang, Z. et al. ERINA is an estrogen-responsive LncRNA that drives breast cancer through the E2F1/RB1 Pathway. Cancer Res.80, 4399–4413 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR19] 19.Wang, Y. et al. Genome-wide gain-of-function screening characterized lncRNA regulators for tumor immune response. Sci. Adv.8, eadd0005 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR20] 20.Guo, W. et al. LincRNA-immunity landscape analysis identifies EPIC1 as a regulator of tumor immune evasion and immunotherapy resistance. Sci. Adv.7, eabb3555 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR21] 21.Xue, X. et al. LncRNA HOTAIR enhances ER signaling and confers tamoxifen resistance in breast cancer. Oncogene35, 2746–2755 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR22] 22.Wang, Y. et al. Systematic identification of non-coding pharmacogenomic landscape in cancer. Nat. Commun.9, 3192 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR23] 23.Han, Y. et al. LEIGC long non-coding RNA acts as a tumor suppressor in gastric carcinoma by inhibiting the epithelial-to-mesenchymal transition. BMC Cancer14, 932 (2014). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR24] 24.Replogle, J. M. et al. Mapping information-rich genotype-phenotype landscapes with genome-scale perturb-seq. Cell185, 2559–2575.e2528 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR25] 25.Seifuddin, F. et al. lncRNAKB, a knowledgebase of tissue-specific functional annotation and trait association of long noncoding RNA. Sci. Data7, 326 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR26] 26.Johnsson, P. et al. Transcriptional kinetics and molecular functions of long noncoding RNAs. Nat. Genet.54, 306–317 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR27] 27.Doench, J. G. Am I ready for CRISPR? A user’s guide to genetic screens. Nat. Rev. Genet.19, 67–80 (2018). [DOI] [PubMed] [Google Scholar]

[CR28] 28.Huang, J., Zheng, L., Sun, Z. & Li, J. CDK4/6 inhibitor resistance mechanisms and treatment strategies (Review). Int. J. Mol. Med.50, 128 (2022). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR29] 29.Wang, X. et al. Recent progress of CDK4/6 inhibitors’ current practice in breast cancer. Cancer Gene Ther.31, 1283–1291 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR30] 30.Joung, J. et al. Genome-scale activation screen identifies a lncRNA locus regulating a gene neighbourhood. Nature548, 343–346 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR31] 31.Barretina, J. et al. The cancer cell line encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature483, 603–607 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR32] 32.Yang, W. et al. Genomics of drug sensitivity in cancer (GDSC): a resource for therapeutic biomarker discovery in cancer cells. Nucleic Acids Res.41, D955–D961 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR33] 33.Park, Y. H. et al. Longitudinal multi-omics study of palbociclib resistance in HR-positive/HER2-negative metastatic breast cancer. Genome Med.15, 55 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR34] 34.Costa, C. et al. PTEN loss mediates clinical cross-resistance to CDK4/6 and PI3Kα inhibitors in breast cancer. Cancer Discov.10, 72–85 (2020). [DOI] [PubMed] [Google Scholar]

[CR35] 35.Kietzman, W. B. et al. Short- and long-term Effects of CDK4/6 inhibition on early-stage breast cancer. Mol. Cancer Ther.18, 2220–2232 (2019). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR36] 36.Spencer, S. abrinaL. et al. The proliferation-quiescence decision is controlled by a bifurcation in CDK2 activity at mitotic exit. Cell155, 369–383 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR37] 37.Malumbres, M. et al. Mammalian cells cycle without the D-type cyclin-dependent kinases Cdk4 and Cdk6. Cell118, 493–504 (2004). [DOI] [PubMed] [Google Scholar]

[CR38] 38.Baker, S. J. & Reddy, E. P. CDK4: a key player in the cell cycle, development, and cancer. Genes Cancer3, 658–669 (2012). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR39] 39.Zhang, H. et al. Effect of CCNB1 silencing on cell cycle, senescence, and apoptosis through the p53 signaling pathway in pancreatic cancer. J. Cell. Physiol.234, 619–631 (2019). [DOI] [PubMed] [Google Scholar]

[CR40] 40.Büschges, R. et al. Amplification and expression of cyclin D genes (CCND1 CCND2 and CCND3) in human malignant gliomas. Brain Pathol.9, 435–442 (1999). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR41] 41.Bénard, J., Douc-Rasy, S. & Ahomadegbe, J.-C. TP53 family members and human cancers. Hum. Mutat.21, 182–191 (2003). [DOI] [PubMed] [Google Scholar]

[CR42] 42.O’Leary, B., Finn, R. S. & Turner, N. C. Treating cancer with selective CDK4/6 inhibitors. Nat. Rev. Clin. Oncol.13, 417–430 (2016). [DOI] [PubMed] [Google Scholar]

[CR43] 43.Cao, J. et al. Targeting estrogen-regulated system xc− promotes ferroptosis and endocrine sensitivity of ER+ breast cancer. Cell Death Dis.16, 30 (2025). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR44] 44.Nagandla, H. & Thomas, C. A.-O. Estrogen signals through ERβ in breast cancer; what we have learned since the discovery of the receptor. Receptors (Basel)3, 182–200 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR45] 45.Murphy, C. G. & Dickler, M. N. The role of CDK4/6 inhibition in breast cancer. Oncologist20, 483–490 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR46] 46.Álvarez-Fernández, M. & Malumbres, M. Mechanisms of sensitivity and resistance to CDK4/6 inhibition. Cancer Cell37, 514–529 (2020). [DOI] [PubMed] [Google Scholar]

[CR47] 47.Griffiths, J. I. et al. Serial single-cell genomics reveals convergent subclonal evolution of resistance as patients with early-stage breast cancer progress on endocrine plus CDK4/6 therapy. Nat. Cancer2, 658–671 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR48] 48.Navarro-Yepes, J. et al. Abemaciclib is effective in palbociclib-resistant hormone receptor-positive metastatic breast cancers. Cancer Res.83, 3264–3283 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR49] 49.Goldman, J. W. et al. Treatment rationale and study design for the JUNIPER study: a randomized phase III study of abemaciclib with best supportive care versus erlotinib with best supportive care in patients with stage IV non-small-cell lung cancer with a detectable KRAS mutation whose disease has progressed after platinum-based chemotherapy. Clin. Lung Cancer17, 80–84 (2015). [DOI] [PubMed] [Google Scholar]

[CR50] 50.Geoerger, B. et al. A phase I study of the CDK4/6 inhibitor ribociclib (LEE011) in pediatric patients with malignant rhabdoid tumors, neuroblastoma, and other solid tumors. Clin. Cancer Res.23, 2433–2441 (2017). [DOI] [PubMed] [Google Scholar]

[CR51] 51.Patnaik, A. et al. Efficacy and safety of abemaciclib, an inhibitor of CDK4 and CDK6, for patients with breast cancer, non-small cell lung cancer, and other solid tumors. Cancer Discov.6, 740–753 (2016). [DOI] [PubMed] [Google Scholar]

[CR52] 52.Finn Richard, S. et al. Palbociclib and letrozole in advanced breast cancer. N. Engl. J. Med.375, 1925–1936 (2016). [DOI] [PubMed] [Google Scholar]

[CR53] 53.Turner, N. C. et al. Palbociclib in hormone-receptor-positive advanced breast cancer. N. Engl. J. Med.373, 209–219 (2015). [DOI] [PubMed] [Google Scholar]

[CR54] 54.Finn, R. S. et al. PD 0332991, a selective cyclin D kinase 4/6 inhibitor, preferentially inhibits proliferation of luminal estrogen receptor-positive human breast cancer cell lines in vitro. Breast Cancer Res.11, R77 (2009). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR55] 55.Olanich, M. E. et al. CDK4 amplification reduces sensitivity to CDK4/6 inhibition in fusion-positive rhabdomyosarcoma. Clin. Cancer Res.21, 4947–4959 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR56] 56.Toulmonde, M. et al. Activity and safety of palbociclib in patients with advanced gastrointestinal stromal tumors refractory to imatinib and sunitinib: a biomarker-driven phase II study. Clin. Cancer Res.25, 4611–4615 (2019). [DOI] [PubMed] [Google Scholar]

[CR57] 57.Leonard, J. P. et al. Selective CDK4/6 inhibition with tumor responses by PD0332991 in patients with mantle cell lymphoma. Blood119, 4597–4607 (2012). [DOI] [PubMed] [Google Scholar]

[CR58] 58.Sanson, K. R. et al. Optimized libraries for CRISPR-Cas9 genetic screens with multiple modalities. Nat. Commun.9, 5416 (2018). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR59] 59.Andersen, C. L. et al. Active estrogen receptor-alpha signaling in ovarian cancer models and clinical specimens. Clin. Cancer Res.23, 3802–3812 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR60] 60.Chen, Z. et al. Estrogen-ERα signaling and DNA hypomethylation co-regulate expression of stem cell protein PIWIL1 in ERα-positive endometrial cancer cells. Cell Commun. Signal.18, 84 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR61] 61.Lafront, C. et al. The estrogen signaling pathway reprograms prostate cancer cell metabolism and supports proliferation and disease progression. J. Clin. Invest.10.1172/JCI170809 (2024). [DOI] [PMC free article] [PubMed]

[CR62] 62.He, M. et al. Estrogen receptor α promotes lung cancer cell invasion via increase of and cross-talk with infiltrated macrophages through the CCL2/CCR2/MMP9 and CXCL12/CXCR4 signaling pathways. Mol. Oncol.14, 1779–1799 (2020). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR63] 63.You, H., Meng, K. & Wang, Z.-Y. The ER-α36/EGFR signaling loop promotes growth of hepatocellular carcinoma cells. Steroids134, 78–87 (2018). [DOI] [PubMed] [Google Scholar]

[CR64] 64.Topi, G. et al. High Oestrogen receptor alpha expression correlates with adverse prognosis and promotes metastasis in colorectal cancer. Cell Commun. Signal.22, 198 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR65] 65.JavanMoghadam, S., Weihua, Z., Hunt, K. K. & Keyomarsi, K. Estrogen receptor alpha is cell cycle-regulated and regulates the cell cycle in a ligand-dependent fashion. Cell Cycle15, 1579–1590 (2016). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR66] 66.Heldring, M. M. et al. Interdependency of estradiol-mediated ERα activation and subsequent PR and GREB1 induction to control cell cycle progression. Heliyon10, e38406 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR67] 67.Henninger, J. E. & Young, R. A. An RNA-centric view of transcription and genome organization. Mol. Cell84, 3627–3643 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR68] 68.Oksuz, O. et al. Transcription factors interact with RNA to regulate genes. Mol. Cell83, 2449–2463.e2413 (2023). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR69] 69.Li, S. et al. Integrative characterization of MYC RNA-binding function. Cell Genomics10.1016/j.xgen.2025.100878 (2025). [DOI] [PMC free article] [PubMed]

[CR70] 70.Wang, X. et al. Cytoplasmic PXR regulates glucose metabolism by binding mRNAs and modulating their stability. Nat. Struct. Mol. Biol.10.1038/s41594-025-01614-5 (2025). [DOI] [PMC free article] [PubMed]

[CR71] 71.Xu, Y. et al. ERa is an RNA-binding protein sustaining tumor cell survival drug resistance. Cell184, 5215–5229.e5217 (2021). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR72] 72.Soota, D., Saravanan, B., Mann, R., Kharbanda, T. & Notani, D. RNA fine-tunes estrogen receptor-alpha binding on low-affinity DNA motifs for transcriptional regulation. EMBO J.43, 5186–5210-5210 (2024). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR73] 73.Joung, J. et al. Genome-scale CRISPR-Cas9 knockout and transcriptional activation screening. Nat. Protoc.12, 828–863 (2017). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR74] 74.Fujita, P. A. et al. The UCSC Genome Browser database: update 2011. Nucleic Acids Res.39, D876–D882 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR75] 75.Li, W. et al. Quality control, modeling, and visualization of CRISPR screens with MAGeCK-VISPR. Genome Biol.16, 281 (2015). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR76] 76.Chang, K. et al. The Cancer Genome Atlas Pan-cancer analysis project. Nat. Genet.45, 1113–1120 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR77] 77.Lonsdale, J. et al. The genotype-tissue expression (GTEx) project. Nat. Genet.45, 580–585 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR78] 78.Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics29, 15–21 (2013). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR79] 79.Li, B. & Dewey, C. N. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinforma.12, 323 (2011). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR80] 80.Anders, S. & Huber, W. Differential expression analysis for sequence count data. Genome Biol.11, R106 (2010). [DOI] [PMC free article] [PubMed] [Google Scholar]

[CR81] 81.Cox, D. R. Regression models and life-tables. J. R. Stat. Soc.: Ser. B (Methodol.)34, 187–202 (1972). [Google Scholar]

[CR82] 82.Zhao, Y. Code Repository for CRISPRa_cdkscreen. GitHub10.5281/zenodo.18118851 (2026).

PERMALINK

CRISPR activation screens identify oncogenic lncRNAs that are susceptible to CDK4/6 inhibitor treatment

Yifei Wang

Yueshan Zhao

Jiaxin Hu

Zehua Wang

Dhamotharan Pattarayan

Sihan Li

Yu Zhang

Xiaofei Wang

Yue Wang

Wen Xie

Min Zhang

Da Yang

Abstract

Introduction

Results

Characterization of CDK4/6i response-regulating lncRNAs by a two-stage CRISPRa screen

Fig. 1. Genome-wide and focused CRISPR activation screens identified lncRNAs regulating CDK4/6i response in breast cancer.

LncRNAs that confer susceptibility to palbociclib treatment tend to promote cell proliferation

Fig. 2. LncRNAs that confer susceptibility to palbociclib treatment tend to promote cell proliferation.

TENM3-AS1, LINC01117, and ENSG00000226706 are oncogenic lncRNAs that promote cell proliferation, tumorigenesis, while serving as therapeutic markers for CDK4/6 inhibitors

Fig. 3. Activation of lncRNAs TENM3-AS1, LINC01117, and ENSG00000226706 leads to palbociclib sensitivity.

Fig. 4. TENM3-AS1, LINC01117, and ENSG00000226706 are oncogenic lncRNAs that promote cell proliferation and cell-cycle progression.

Fig. 5. TENM3-AS1, LINC01117, and ENSG00000226706 are oncogenic lncRNAs that promote cell proliferation and tumorigenesis.

RNA-seq analysis revealed that lncRNA TENM3-AS1 activates the ER pathway

Fig. 6. RNA-seq analysis revealed that lncRNA TENM3-AS1, LINC011117, and ENSG00000226706 activate the ERα pathway.

Fig. 7. LncRNA TENM3-AS1 activates the ERα-MYC axis to induce cancer cell proliferation and regulate CDK4/6i response.

LncRNA TENM3-AS1 activates the ERα signaling to induce cancer cell proliferation and regulate CDK4/6i response

Discussion

Methods

Institutional approval

Cell lines and reagents

Focused screen library construction and amplification

sgRNA library lentivirus production and transfection

Global screen

Focused screen

Global and focused CRISPRa screens for cell proliferation and palbociclib response

Genomic DNA extraction and sample preparation for NGS

CRISPRa screen data analysis

LncRNA expression, drug response, and patient survival association analyses across patient tumors and human cancer cell lines

Exploratory RNA-seq analysis

LncRNA activation signature analysis

Single-nuclei RNA-seq and palbociclib treatment sample analyses

Single-sgRNA cloning and stable cell line generation

RNA interference

Generation of the stable cell line

RNA extraction and quantitative real-time PCR (qRT-PCR)

Western blot

RNA Immunoprecipitation (RIP)

Antibodies

MTT assay

With palbociclib

Cell proliferation

Colony formation assay

With palbociclib

Cell proliferation

Cell cycle assay

In vivo xenograft model

Tissue immunofluorescence

Statistical analysis

Reporting summary

Supplementary information

Acknowledgements

Author contributions

Peer review

Peer review information

Data availability

Code availability

Competing interests

Footnotes

Contributor Information

Supplementary information

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES