Enzymatic activity-independent NANS stabilizes LATS2 to drive growth and therapeutic resistance in HR+/HER2- breast cancer

Abstract

Hormone receptor-positive, human epidermal growth factor receptor 2-negative (HR+ /HER2-) breast cancer is the most common subtype in females, frequently challenged by resistance to endocrine therapy and CDK4/6 inhibitors. Here we show that N-acetylneuraminate synthase (NANS) is significantly upregulated in female HR+ /HER2- breast cancer patients, inversely correlating with patient prognosis. Functionally, NANS promotes tumor growth and confers resistance to tamoxifen and CDK4/6 inhibitors independently of its canonical enzymatic activity. Mechanistically, NANS recruits ubiquitin-specific protease 7 (USP7) to deubiquitinate and stabilize large tumor suppressor kinase 2 (LATS2), leading to canonical Hippo pathway activation and upregulation of estrogen receptor α and CDK4. Reintroducing LATS2 into NANS-depleted cells partially rescues the impaired growth-promoting and drug-resistant phenotypes in vitro and in vivo. Furthermore, the clinical relevance of the NANS-USP7/LATS2 axis is confirmed in patient-derived female HR+ /HER2- breast cancer samples. Together, our findings unveil an enzymatic activity-independent role for NANS in driving therapy resistance, nominating it as a promising therapeutic target.

LATS2 is known to regulate treatment response in breast cancer, but the mechanisms controlling its stability remain mostly elusive. Here, the authors discover that NANS-mediated recruitment of USP7 stabilizes LATS2, driving tumor growth and resistance to endocrine therapy and CDK4/6 inhibitors in breast cancer.

Introduction

Breast cancer (BC) is a highly heterogeneous disease and includes distinct molecular subtypes that are intimately linked to different therapeutic strategies and clinical outcomes¹. BC is generally classified into three primary molecular subtypes on the basis of the expression status of estrogen and progesterone receptors (collectively referred to as hormone receptors, HR) and human epidermal growth factor receptor 2 (HER2). These subtypes include HR-positive and HER2-negative (HR+ /HER2-), HER2-positive (HER2+ ), and triple-negative breast cancer (TNBC)^2,3. Among them, HR+ /HER2- is the most common subtype, accounting for approximately two-thirds of all primary BC cases⁴.

Clinically, endocrine therapy (ET), such as the selective estrogen receptor (ER) modulator tamoxifen, the ER degrader fulvestrant, and aromatase inhibitors that block estrogen production, is the mainstay of treatment for patients with HR+ /HER2- BC⁵. However, ~ 20% of patients with early-stage disease and nearly all patients with advanced/metastatic disease eventually develop endocrine resistance⁵. Accumulating evidence has shown that abnormal activation of cyclin-dependent kinases 4 and 6 (CDK4/6) and their binding partner cyclin D is among the major mechanisms of endocrine resistance^6–8. Consequently, the combination of ET with a selective CDK4/6 inhibitor (abemaciclib, palbociclib, or ribociclib) has become the standard first-line treatment for advanced/metastatic HR+ /HER2- BC^4,9,10. Although CDK4/6 inhibitors are effective in these clinical settings, nearly all patients eventually develop resistance and succumb to the disease⁹. However, the molecular mechanisms driving resistance to ET and CDK4/6 inhibitors are still not fully understood and warrant further investigation.

N-acetylneuraminate synthase (NANS) is a key enzyme in the biosynthesis of sialic acids and plays critical roles in brain and skeletal development^11,12. Accordingly, NANS mutations disrupt endogenous sialic acid biosynthesis and impair the polysialylation of proteins, thus leading to widespread dysregulation of neuronal proliferation, migration, differentiation, and synaptogenesis¹². In addition, mutations in NANS interfere with ribosomal function, compromising protein translation¹². Interestingly, the latest emerging evidence indicates that NANS is a potential prognostic marker for patients with prostate¹³, endometrial¹⁴ and colon¹⁵ cancers. By analyzing our recently published large-scale multiomics dataset of HR+ /HER2- BC samples¹⁶, we found that NANS is highly expressed in HR+ /HER2- BC and that its high expression correlates negatively with patient outcomes. However, the potential biological functions and related mechanisms of action of NANS in cancer progression and therapeutic response remain unexplored to date.

The Hippo pathway is among the most commonly dysregulated signaling pathways in human cancer cells and has been implicated in the initiation and progression of various tumors^17–19. As a core component of this pathway in mammals, large tumor suppressor kinase 2 (LATS2) acts as a negative regulator of its downstream transcriptional effectors, Yes-associated protein (YAP) and transcriptional coactivator with PDZ-binding motif (TAZ), to direct transcriptional programs during numerous biological processes^17,19,20. Although LATS2 was initially thought to be a tumor suppressor gene^21–24, accumulating evidence has shown that LATS2 exerts tumor-promoting effects in the context of breast cancer^25–29. In particular, LATS1/2 promotes ER-positive (ER+ ) breast cancer growth by maintaining ER expression via YAP/TAZ inhibition^26,28, as YAP suppresses ER expression and ER+ breast cancer growth^27,28. Consistently, pharmacological inhibition of LATS1/2 blocks ER+ breast cancer growth and enhances the sensitivity of ER+ breast cancer cells to the CDK4/6 inhibitor palbociclib²⁸. In addition, LATS2 has been reported to promote the epithelial-to-mesenchymal transition and invasive capacity of TNBC cells independent of its effects on YAP or TAZ²⁹, and knockout of LATS1/2 in the mouse TNBC cell line 4T1 inhibits tumor growth in mice²⁵. Given the functional importance of LATS2 in breast cancer, an in-depth understanding of its regulatory mechanisms may lead to therapeutic interventions for this disease. Although recent work has shown that LATS2 undergoes both proteasomal^30–32 and autophagic³³ degradation, its upstream regulatory mechanisms in HR+ /HER2- BC remain largely unknown.

In this study, we report that NANS acts as an oncogene in HR+ /HER2- BC to promote tumor growth and confer resistance to tamoxifen and CDK4/6 inhibitors and that these functions exerted by NANS are independent of its enzymatic activity. Mechanistic investigations revealed that NANS stabilizes LATS2 by enhancing its interaction with ubiquitin-specific protease 7 (USP7), thus activating the Hippo pathway to induce ERα and CDK4 expression. Importantly, the NANS-USP7/LATS2 axis in HR+ /HER2- BC was validated by functional rescue assays, pharmacological inhibition of USP7, and histological analysis of clinical samples. Collectively, these findings suggest that NANS is a potential therapeutic target and a predictive biomarker of resistance to ET and CDK4/6 inhibitors in HR+ /HER2- BC patients.

Results

NANS is highly expressed in HR+ /HER2- BC and is associated with poor patient prognosis

To identify molecular aberrations driving the pathogenesis of HR+ /HER2- BC, we conducted integrative analyses of our recently published large-scale multiomics HR+ /HER2- BC dataset¹⁶ and an RNA-seq dataset of HR+ /HER2- BC from TCGA (Fig. 1A and Supplementary Fig. S1A). First, we identified differentially expressed genes (DEGs) between tumor samples and adjacent normal tissues by analyzing RNA-seq and quantitative proteomics datasets from our center (FUSCC)¹⁶ and an RNA-seq dataset from the TCGA. In accordance with the stringent screening criteria (log₂FC > 0.7 and adjusted p < 0.05), we found that 85 genes were significantly upregulated in HR+ /HER2- BC tissues (Supplementary Fig. S1A). Second, we analyzed a single–cell RNA sequencing (scRNA-seq) dataset from FUSCC¹⁶ with the screening criteria of log₂FC > 0.3 and adjusted p < 0.05 to identify DEGs that were specifically expressed in breast cancer cells but not in stromal cells, and found that 38 out of the 85 DEGs identified above were highly expressed in HR+ /HER2- BC cells (Supplementary Fig. S1A). Third, we performed weighted gene co-expression network analysis (WGCNA) to further prioritize functionally relevant genes using an RNA-seq dataset of HR+ /HER2- BC samples from FUSCC¹⁶, and found that 6 out of the 38 candidate genes were most strongly associated with HR+ /HER2- BC (Supplementary Fig. S1A). Fourth, we performed Kaplan‒Meier survival analysis using the HR+ /HER2- BC dataset from FUSCC¹⁶ and the TCGA dataset and found that high expression of 4 genes, namely, NANS, CERS6 (ceramide synthase 6), MORF4L2 (mortality factor 4 like 2), and HMGB3 (high mobility group box 3), was associated with poor overall survival (OS) in HR+ /HER2- BC patients (Supplementary Fig. S1A). Finally, we selected NANS for further investigations because its functional and mechanistic roles in HR + /HER2- BC pathogenesis have not yet been explored.

Fig. 1. NANS is highly expressed in HR+ /HER2- breast cancer (BC) and is negatively associated with patient prognosis.

A Flowchart by which NANS was identified as a candidate oncogene in HR+ /HER2- BC. Certain icons were created in BioRender. Cai, J. (2026) https://BioRender.com/r5i9oxj. B, C Protein expression levels of NANS in normal tissue and three molecular subtypes of BC in proteomic datasets from Chinese Breast Cancer Genome Atlas (CBCGA) (B) and CPTAC (C). For the CPTAC dataset, expression levels are presented as Z-values representing standard deviations from the median. For all box plots, the center line indicates the median, the bounds of the box represent the 25th and 75th percentiles, and the whiskers extend to 1.5 × the interquartile range (IQR) from the box boundaries. Minima and maxima are represented by the individual data points (jitter). The exact sample sizes are as follows: for the CBCGA dataset (B), n = 8 (Normal), n = 150 (HR+ /HER2-), n = 64 (HER2+ ), and n = 64 (TNBC) biologically independent samples; for the CPTAC dataset (C), n = 18 (Normal), n = 64 (Luminal), n = 10 (HER2 Positive), and n = 16 (TNBC) biologically independent samples. D Immunoblotting analysis of the protein expression levels of NANS in normal breast epithelial cell lines and breast cancer cell lines with different molecular subtypes. E–G Association of NANS mRNA expression levels with overall survival (OS) (E), relapse-free survival (RFS) (F), and distant metastasis-free survival (DMFS) (G) in patients in the HR+ /HER2- BC cohort from FUSCC (n = 569 biologically independent patients). H Quantitative PCR analysis of NANS mRNA levels in 12 pairs of HR+ /HER2- BC samples and adjacent normal tissues. (n = 12 biologically independent patient pairs). I, J Immunoblotting analysis of NANS protein levels in 19 pairs of HR+ /HER2- BC samples and adjacent normal tissues (I). The relative NANS expression levels were quantified using ImageJ software (J). Data are presented as mean ± SD. n = 19 biologically independent patient pairs. For the representative immunoblot in (D), experiments were repeated independently three times, yielding similar results. Statistical significance was determined using a two-sided Wilcoxon rank-sum test (B), a two-sided unpaired Student’s t test (C), and a two-sided paired Student’s t test (H, J). For survival analyses (E–G), exact p-values were determined using a two-sided log-rank test, and hazard ratios (HR) were calculated using univariate Cox proportional hazards regression. For all statistical comparisons, exact p-values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

Next, we conducted pan-cancer analyses of NANS expression status in normal and tumor tissues using the TIMER2.0 (http://timer.cistrome.org/) database³⁴ and found that NANS expression was significantly upregulated in tumor samples compared with normal tissues in most cancer types, including breast cancer (Supplementary Fig. S1B). We then analyzed the expression patterns of NANS in normal breast tissues and tissues from the three primary molecular subtypes (HR+ /HER2-, HER2+ , and TNBC) of breast cancer^2,3 using proteomic datasets from the CBCGA (Chinese Breast Cancer Genome Atlas)³⁵, alongside CPTAC (Clinical Proteomic Tumor Analysis Consortium) data³⁶ accessed via the UALCAN web portal (https://ualcan.path.uab.edu/analysis-prot.html). The results revealed that the expression levels of NANS were highest in the HR+ /HER2- subtype in the CBCGA dataset (Fig. 1B) and in the luminal subtype in the CPTAC dataset (Fig. 1C). Similarly, NANS mRNA expression in the HR+/HER2- subtype was significantly higher than that in TNBC across both the TCGA and METABRIC (Molecular Taxonomy of Breast Cancer International Consortium) datasets, and also significantly higher than that in normal breast tissues within the TCGA cohort (Supplementary Fig. S1C, D). It is worth mentioning that the METABRIC dataset lacks corresponding normal sample data. Consistent with these findings, immunoblotting assays revealed that NANS expression was significantly upregulated in HR+ /HER2- BC cell lines (MCF7 and T47D) compared with that in normal mammary epithelial cell lines and HER2+ and TNBC cell lines (Fig. 1D). Multiplex immunofluorescence staining of HR+ /HER2- breast cancer tissues indicated that NANS primarily co-localized with the tumor epithelial marker pancytokeratin (panCK), rather than with fibroblast activation protein (FAP), a marker for stromal cancer-associated fibroblasts (Supplementary Fig. S2A). These results suggest that NANS is localized mainly in tumor cells rather than in stromal components.

To further analyze the relationship between the expression levels of NANS and patient prognosis, we conducted a survival analysis using a large-scale HR+ /HER2- BC cohort from FUSCC¹⁶. High NANS expression was associated with poor overall survival (OS), relapse-free survival (RFS), and distant metastasis-free survival (DMFS) in HR+ /HER2- BC patients (Fig. 1E–G, respectively). Similarly, elevated expression of NANS was correlated with poor prognosis in HR+ /HER2- BC patients in the TCGA and METABRIC datasets, as well as in analyses performed via the Kaplan‒Meier plotter online web portal (Supplementary Fig. S2B–E, respectively).

To validate the above results, we performed RT‒qPCR and immunoblotting assays using tumor samples and paired adjacent normal tissues from HR+ /HER2- BC patients. The results revealed that the expression levels of NANS were significantly greater in tumor samples than in adjacent normal tissues at both the mRNA (Fig. 1H) and protein (Fig. 1I, J) levels. Among the 19 protein samples analyzed, 12 were obtained from the same patients used for the RT-qPCR analysis.

Taken together, these results suggest that NANS is highly expressed in HR+ /HER2- BC tissues and is associated with poor patient prognosis.

NANS promotes the growth of HR+ /HER2- BC cells both in vitro and in vivo

To investigate the biological functions of NANS in HR+ /HER2- BC cells, we stably depleted endogenous NANS or overexpressed Flag-NANS in T47D and MCF7 cells and validated the expression status of NANS in these established cell lines by immunoblotting (Fig. 2A, B). CCK8 and colony formation assays demonstrated that depletion of NANS using two independent sgRNAs targeting NANS (sgNANS #2 and #3) significantly reduced the viability and colony formation ability of MCF7 and T47D cells (Fig. 2C–E). Conversely, ectopic expression of NANS increased the viability and colony formation ability of both MCF7 and T47D cells (Fig. 2F–H). Moreover, reintroduction of NANS into NANS-depleted MCF7 and T47D cells rescued the impaired viability and colony-forming capacity caused by NANS depletion (Fig. 2I, J and Supplementary Fig. S3A, B). Flow cytometry and Transwell assays further demonstrated that NANS knockdown significantly promoted apoptosis and markedly reduced the migratory and invasive abilities of these cells (Supplementary Fig. S3C–J).

Fig. 2. NANS promotes the growth of HR+ /HER2- breast cancer cells both in vitro and in vivo.

A Immunoblotting analysis of NANS expression levels in MCF7 and T47D cells stably expressing sgNC and sgNANS (#2 and #3). B Immunoblotting analysis of the expression status of NANS in MCF7 and T47D cells stably expressing pLVX or Flag-NANS. The samples derive from the same experiment, but different gels for Flag, and another for NANS and β-tubulin were processed in parallel. C Assessment of the viability of MCF7 and T47D cells stably expressing sgNC or sgNANS (#2 and #3) determined by CCK-8 assays. Data are presented as mean ± SD of n = 3 biologically independent experiments. D, E Colony formation assays were conducted to evaluate the clonogenic capacity of MCF7 and T47D cells stably expressing sgNC or sgNANS (#2 and #3). Representative images of surviving colonies (D) and the corresponding quantitative results (E) are shown. Data are presented as mean ± SD of n = 3 biologically independent experiments. F Assessment of the viability of MCF7 and T47D cells stably expressing pLVX or Flag-NANS determined by CCK-8 assays. Data are presented as mean ± SD of n = 3 biologically independent experiments. G, H Colony formation assays were conducted to evaluate the clonogenic capacity of MCF7 and T47D cells stably expressing pLVX or Flag-NANS. Representative images of surviving colonies (G) and the corresponding quantitative results (H) are shown. Data are presented as mean ± SD of n = 3 biologically independent experiments. I MCF7 and T47D cells stably expressing sgNC or sgNANS #2 alone or in combination with Flag-NANS were subjected to immunoblotting. The samples derive from the same experiment, but different gels for Flag, and another for NANS and β-tubulin were processed in parallel. J MCF7 and T47D cells stably expressing sgNC or sgNANS #2 alone or in combination with Flag-NANS were subjected to a CCK-8 assay. Data are presented as mean ± SD of n = 4 biologically independent experiments. K–M MCF7 cells stably expressing sgNC or sgNANS (#2 and #3) were orthotopically injected into the mammary fat pads of female BALB/c nude mice (n = 10 biologically independent animals per group). Images of harvested xenograft tumors (K), tumor weights (L), and tumor volume (M) are shown. N, O Immunohistochemical (IHC) staining was performed to assess Ki-67 expression levels in mouse xenograft tumors. Representative IHC staining images (N) and the corresponding quantitative results (O) are presented. Scale bar, 50 μm. Data are presented as mean ± SD of n = 3 biologically independent animals. For representative images and blots in (A, B, D, G, I), experiments were repeated independently three times, yielding similar results. Statistical significance for cell viability at the final time point(C and J) and clonogenic capacity/tumor weight/tumor volume /IHC (E, L, M, O) was determined using a one-way ANOVA. For (F) (at the final time point) and (H), the p-value was calculated using a two-sided unpaired Student’s t-test. For all statistical comparisons, exact p- values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

Next, we established xenograft tumor models in nude mice to assess the impact of NANS knockdown on the growth of HR+ /HER2- BC cells in vivo. As shown in Fig. 2K–M, knockdown of NANS expression significantly reduced xenograft tumor weight and volume in nude mice. Consistent with these findings, immunohistochemical (IHC) staining revealed that knockdown of NANS decreased the expression levels of the putative cell proliferation marker Ki-67³⁷ in xenograft tumors (Fig. 2N, O). Collectively, these results suggest that NANS accelerates the growth of HR+ /HER2- BC cells both in vitro and in vivo.

NANS confers resistance to tamoxifen and CDK4/6 inhibitors in HR+ /HER2- BC

ET alone or combined with a CDK4/6 inhibitor is the current standard-of-care first-line therapy for patients with HR+ /HER2- BC^4,38–40. By analyzing HR+ /HER2- BC datasets from both the FUSCC cohort¹⁶ and Kaplan‒Meier plotter web portal⁴¹, we found that high NANS expression was associated with poor survival (OS) in patients who received ET (Supplementary Fig. S4A, B). These results indicate that NANS may contribute to endocrine resistance in HR+ /HER2- BC.

On the basis of these findings, we next examined whether NANS affects the sensitivity of HR+ /HER2- BC cells to the most commonly used ET drug, tamoxifen (Tam), and the clinically approved CDK4/6 inhibitors abemaciclib (Abema) and palbociclib (Palbo)^9,10. CCK-8 cell viability assays revealed that knockdown of NANS expression significantly increased the sensitivity of MCF7 and T47D cells to Tam (Fig. 3A), Abema (Fig. 3B), and Palbo (Fig. 3C), as indicated by the marked decreases in the half inhibitory concentration (IC50) values. Consistently, colony formation assays revealed that depletion of NANS led to more pronounced inhibitory effects on colony formation after treatment with Tam (Fig. 3D, E), Abema (Fig. 3F, G), and Palbo (Fig. 3H, I).

Fig. 3. NANS confers resistance to tamoxifen and CDK4/6 inhibitors in HR+ /HER2- BC.

A–C MCF7 and T47D cells stably expressing sgNC or sgNANS (#2 and #3) were treated with or without increasing concentrations of tamoxifen (Tam) (A), abemaciclib (Abema) (B), or palbociclib (Palbo) (C) for 72 h. The IC50 values were determined by CCK-8 assays. Data are presented as mean ± SD of n = 4 biologically independent experiments for T47D cells in (A), and n = 3 biologically independent experiments for all other conditions. D–I Colony formation assays were conducted using MCF7 and T47D cells stably expressing sgNC or sgNANS (#2 and #3) in the presence or absence of increasing concentrations of Tam (D, E), Abema (F, G), and Palbo (H, I) for 7–9 days. Representative images (D, F, H) and statistical analysis of relative surviving colonies (E, G, I) are shown. Data in (E, G, I) are presented as mean ± SD of n = 3 biologically independent experiments. J The expression levels of NANS in two HR+ /HER2- BC patient-derived organoids (PDOs) were evaluated by immunohistochemical (IHC) staining of postoperative tumor tissue sections from the same patients. The quantitative results are shown. Data are presented as mean ± SD of n = 3 independent tissue sections per patient sample. Representative IHC staining images are presented in Supplementary Fig. S6G. K Two HR+ /HER2-BC patient-derived organoids (PDOs) were treated with vehicle, Tam (2 μM), Abema (0.5 μM), or Palbo (1 μM). CellTiter-Glo 3D assays were subsequently used to assess PDO viability. The quantitative results are shown. Data are presented as mean ± SD of n = 4 biologically independent experiments. Representative images of PDOs are presented in Supplementary Fig. S6H. For representative images in (D, F, H), experiments were repeated independently three times, yielding similar results. Statistical significance for IC50 values based on nonlinear regression curve fitting (A–C) was determined using the extra sum-of-squares F test. Statistical significance for surviving colonies at the highest concentration (E, G, I) was determined using a one-way ANOVA. For (J, K), the p-values were calculated using a two-sided unpaired Student’s t test. For all statistical comparisons, exact p-values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

Moreover, reintroduction of NANS into NANS‑depleted MCF7 and T47D cells reversed the enhanced sensitivity to tamoxifen (Tam) and abemaciclib (Abema) induced by NANS depletion (Supplementary Fig. S4C, D). Similarly, NANS knockdown significantly sensitized tamoxifen‑resistant (TamR) and palbociclib‑resistant (PalboR) MCF7 cells to their respective drugs (Supplementary Fig. S4E–G).

To further mimic clinical therapeutic settings, we established models with dual-resistance to tamoxifen and palbociclib (TPR) in both T47D and MCF7 cell lines, with their resistance profiles validated via drug sensitivity assays (Supplementary Fig. S4H, I). Notably, NANS depletion markedly sensitized both T47D-TPR and MCF7-TPR cells to combined tamoxifen and palbociclib treatment (Supplementary Fig. S4J, K). Furthermore, NANS deficiency significantly promoted apoptosis and attenuated the migratory and invasive capacities of these dual-resistant cells (Supplementary Fig. S5A–H).

We next generated aromatase‑overexpressing cell lines (MCF7‑aro and T47D‑aro) by stable transfection of aromatase-encoding gene CYP19A1 to assess aromatase inhibitor sensitivity (Supplementary Fig. S5I). In these cells, NANS knockdown markedly increased sensitivity to the aromatase inhibitors letrozole and exemestane (Supplementary Fig. S5J, K). Furthermore, NANS depletion increased cellular sensitivity to multiple combination regimens, including tamoxifen plus abemaciclib (Supplementary Fig. S6A) or palbociclib (Supplementary Fig. S6B), letrozole plus abemaciclib (Supplementary Fig. S6C) or palbociclib (Supplementary Fig. S6D), and exemestane plus abemaciclib (Supplementary Fig. S6E) or palbociclib (Supplementary Fig. S6F).

To validate the above findings, we conducted drug sensitivity experiments on patient-derived organoids (PDOs) from HR+ /HER2- BC patients. The expression levels of NANS in these PDOs were assessed by IHC staining of postoperative pathological tissue sections derived from the same patients to establish PDOs (Fig. 3J and Supplementary Fig. S6G). The results revealed that compared with PDO #221 with low NANS expression, PDO #333 with high NANS expression displayed decreased sensitivity to Tam, Abema, and Palbo (Fig. 3K and Supplementary Fig. S6H). These results collectively suggest that NANS contributes to resistance to Tam and CDK4/6 inhibitors in HR+ /HER2- BC.

NANS exerts its functions in HR+ /HER2- BC independently of its enzymatic activity

In mammals, NANS catalyzes the condensation of phosphoenolpyruvate (PEP) with N-acetylmannosamine 6-phosphate (ManNAc-6-P) to produce N-acetylneuraminic acid (NeuAc) and, subsequently, cytidine monophosphate (CMP)-NeuAc (the active form of NeuAc)^42,43. Given that previous studies have shown that the biological functions of NANS are largely dependent on its enzymatic activity^11,12, we next investigated whether the noted functions of NANS in HR+ /HER2- BC depend on its enzymatic activity.

To this aim, we first analyzed the metabolomic dataset from 384 patients with HR+ /HER2- BC in the FUSCC cohort¹⁶ to compare the differences in the levels of the catalytic products of NANS (NeuAc and CMP-NeuAc) between patients with high and low expression of NANS. Unexpectedly, we found no significant differences in the levels of NeuAc or CMP-NeuAc between patients with high and low expression of NANS at either the mRNA or protein level (Supplementary Fig. S7A, B). We subsequently conducted gene set variation analysis (GSVA), specifically focusing on cell cycle-related pathways to compare pathway enrichment between patients with high and low NANS expression. The results revealed that cell cycle-related pathways were significantly enriched in the high NANS expression group (Supplementary Fig. S7C). Furthermore, comparisons of responders and nonresponders to endocrine therapy in the HR+ /HER2− breast cancer cohort from the FUSCC¹⁶ revealed no differential enrichment of pathways related to sialic acid metabolism between the two groups (Supplementary Fig. S7D).

Previous studies have shown that mutation of any of the four sites responsible for enzymatic activity (K131Q, G133V, Y188H, and P189L) in NANS can impair its catalytic activity¹¹. We therefore constructed an NANS expression vector in which all four enzymatic activity sites were mutated according to the above literature (named NANS-Mut) and stably overexpressed wild-type (WT) and mutant NANS in MCF7 and T47D cells to perform functional assays (Supplementary Fig. S7E). The catalytic activity of NANS was assessed by measuring its direct product, Neu5Ac-9-P, using an ELISA kit. Overexpression of wild-type NANS led to a significant increase in intracellular Neu5Ac-9-P levels, whereas expression of the catalytic mutant had no appreciable effects (Supplementary Fig. S7F). These results indicate that these mutations abolish NANS enzymatic activity.

CCK-8 and colony formation assays revealed that mutation of the sites responsible for enzymatic activity in NANS did not significantly affect its ability to promote the viability (Supplementary Fig. S7G) or colony formation ability (Supplementary Fig. S7H, I) of MCF7 and T47D cells. Moreover, ectopic expression of both NANS-WT and NANS-Mut reduced the sensitivity of MCF7 and T47D cells to Tam and the CDK4/6 inhibitors Abema and Palbo (Supplementary Fig. S7J–L). Together, these results suggest that the functions of NANS in HR+ /HER2- BC are not dependent on its enzymatic activity.

NANS blocks the ubiquitin-dependent proteasomal degradation of LATS2

To further elucidate the underlying mechanisms through which NANS promotes the growth and therapeutic resistance of HR+ /HER2- BC cells, we conducted quantitative proteomics analyses with data-independent acquisition-based mass spectrometry (DIA-MS)⁴⁴ using cellular lysates from MCF7 cells stably expressing sgNC and sgNANS #2. We found that 177 proteins were upregulated and 122 proteins were downregulated following knockdown of NANS on the basis of the screening criteria of a fold change threshold of 1.2 and p < 0.05 (Fig. 4A). KEGG (Kyoto Encyclopedia of Genes and Genomes) pathway enrichment analysis revealed that the downregulated 122 proteins following knockdown of NANS were significantly enriched in the Hippo signaling pathway (Fig. 4B). Gene Set Enrichment Analysis (GSEA) also revealed that the Hippo signaling pathway activity was lower in sgNANS #2-expressing cells than in cells expressing sgNC (Supplementary Fig. S8A).

Fig. 4. NANS suppresses the ubiquitin-dependent proteasomal degradation of LATS2.

A Lysates from MCF7 cells stably expressing sgNC or sgNANS #2 were subjected to quantitative proteomics analyses via data-independent acquisition mass spectrometry (DIA‒MS) (n = 3 biologically independent samples per group). The differentially expressed proteins between cells expressing sgNC and sgNANS #2 are presented in a volcano plot. B Downregulated differentially expressed proteins were subjected to KEGG enrichment analysis. C The protein expression levels of LATS2 and PAK1 between cells expressing sgNC and sgNANS #2, based on DIA quantitative proteomics analysis, are presented as box plots. For the box plots, the center line indicates the median, the bounds of the box represent the 25th and 75th percentiles, and the whiskers extend to 1.5 × the interquartile range (IQR) from the box boundaries. Minima and maxima are represented by individual data points. n = 3 biologically independent samples per group. D Immunoblotting analysis of the protein expression levels of LATS2, LATS1, and PAK1 in MCF7 and T47D cells stably expressing sgNC and sgNANS (#2 and #3). The samples derive from the same experiment, but different gels for LATS1 and PAK1, and another for LATS2, NANS, and β-tubulin were processed in parallel. E Immunoblotting analysis of the protein expression levels of LATS2 and LATS1 in MCF7 and T47D cells stably expressing pLVX or Flag-NANS. The samples derive from the same experiment, but different gels for LATS1 and another for LATS2, NANS, and β-tubulin were processed in parallel. F, G MCF7 and T47D cells stably expressing sgNC or sgNANS #2 were treated with or without 100 μg/mL CHX for the indicated durations and then subjected to immunoblotting with the indicated antibodies (F). The relative protein levels of LATS2 (LATS2/β-tubulin) are shown in (G). H, I MCF7 and T47D cells were treated with or without 10 μM MG132 (H) or 200 nM Baf-A1 (I) for the indicated durations and then subjected to immunoblotting. For panel I, the samples derive from the same experiment, but different gels for p62 and another for LATS2, and β-tubulin were processed in parallel. J, K MCF7 and T47D cells stably expressing sgNC or sgNANS #2 were treated with or without 10 μM MG132 (J) or 200 nM Baf-A1 (K) for 12 h and then subjected to immunoblotting. The numbers below the corresponding immunoblots indicate the relative protein expression levels normalized to β-tubulin. For panel K, the samples derive from the same experiment, but different gels for p62, and another for LATS2, NANS, and β-tubulin were processed in parallel. L, M HEK293T cells were transfected with the indicated expression vectors for 48 h, followed by sequential IP and immunoblotting with the indicated antibodies, with input and IP samples processed in parallel on separate gels. For the IP samples in panels L and M, the samples derive from the same experiment, but different gels for V5-Ub and another for Flag were processed in parallel. For the Input samples in panel (L), different gels for V5-Ub, and another for Flag, HA, and β-tubulin were processed in parallel. For the Input samples in panel (M), different gels for V5-Ub, and another for Flag, NANS, and β-tubulin were processed in parallel. For representative immunoblots in (D, E, H–M), experiments were repeated independently three times, yielding similar results. Statistical significance for (C) was determined using a two-sided unpaired Student’s t test. For all statistical comparisons, exact p-values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

With respect to the Hippo pathway, we found that both LATS2 and PAK1 (p21-activated kinase 1) were significantly downregulated in MCF7 cells following knockdown of NANS (Fig. 4C). Immunoblotting assays demonstrated that knockdown of NANS significantly downregulated the expression levels of LATS2, in contrast to PAK1, in MCF7 and T47D cells (Fig. 4D). In addition, overexpression of NANS led to a significant upregulation of LATS2 in MCF7 and T47D cells (Fig. 4E). Notably, neither knockdown nor overexpression of NANS affected LATS1 expression (Fig. 4D, E). These results suggest that NANS positively regulates LATS2 expression in HR+ /HER2- BC cells.

To further investigate how NANS regulates LATS2, we performed RT‒qPCR assays and found that knockdown of NANS did not significantly affect the mRNA levels of LATS2 (Supplementary Fig. S8B). Cycloheximide (CHX) chase assays revealed that knockdown of NANS shortened the half-life of the LATS2 protein (Fig. 4F, G). These results indicate that NANS could increase the protein stability of LATS2. Notably, previous studies have shown that LATS2 undergoes both proteasomal^30–32 and autophagic³³ degradation in other model systems. However, treatment of T47D and MCF7 cells with the proteasome inhibitor MG132 but not the lysosome inhibitor Baf-A1 resulted in increased LATS2 protein levels (Fig. 4H, I). These results suggest that LATS2 is degraded in HR+ /HER2- BC cells mainly through the ubiquitin–proteasome pathway. Moreover, the reduced levels of LATS2 caused by NANS knockdown were partially restored by treatment with MG132 but not with Baf-A1 (Fig. 4J, K). These results highlight that NANS can block the proteasomal degradation of LATS2. In support of this notion, ubiquitination assays revealed that the ubiquitination levels of LATS2 decreased upon overexpression of NANS (Fig. 4L) but increased following NANS knockdown (Fig. 4M). Collectively, these results suggest that NANS blocks the ubiquitin-dependent proteasomal degradation of LATS2.

NANS recruits the deubiquitinase USP7 to stabilize LATS2

To elucidate the underlying mechanism through which NANS blocks the proteasomal degradation of LATS2, we next performed immunoprecipitation coupled with liquid chromatography‒tandem mass spectrometry (LC‒MS/MS) assays using cellular lysates from HEK293T cells expressing pCDH and Flag-LATS2 to identify the binding partners of LATS2 (Fig. 5A). Using this approach, we identified a total of 166 LATS2-interacting proteins (Supplementary Fig. S9A). Among them, we found that the putative deubiquitinase USP7 (ubiquitin-specific peptidase 7)⁴⁵ was a potential binding partner of LATS2 (Supplementary Fig. S9B). Thus, we first used in vivo ubiquitination assays to determine whether USP7 affects LATS2 ubiquitination. The results revealed that ectopic expression of wild-type USP7, but not its catalytically inactive mutant USP7^C223S46–48, led to a marked decrease in LATS2 ubiquitination (Fig. 5B). Conversely, knockdown of endogenous USP7 increased LATS2 ubiquitination levels (Fig. 5C). These results suggest that USP7 acts as a deubiquitinase of LATS2 to promote LATS2 deubiquitination in a manner that depends on its deubiquitinase activity. Moreover, knockdown of endogenous USP7 downregulated LATS2 expression at the protein level but not at the mRNA level (Fig. 5D and Supplementary Fig. S9C).

Fig. 5. NANS recruits the deubiquitinase USP7 to stabilize LATS2.

A HEK293T cells expressing pCDH and Flag-LATS2 were subjected to IP assays using anti-Flag magnetic beads, followed by SDS‒PAGE. The resulting gels were stained with Coomassie Brilliant Blue solution. B, C HEK293T cells were transfected with the indicated expression vectors for 48 h, followed by treatment with 10 μM MG132 for another 6 h. The cell lysates were subsequently subjected to sequential IP and immunoblotting with the indicated antibodies. For the IP samples in panels B and C, the samples derive from the same experiment, but different gels for V5-Ub and another for Flag were processed in parallel. For the Input samples in panel B, different gels for V5-Ub, another for Flag, and another for HA and β-tubulin were processed in parallel. For the Input samples in panel C, different gels for V5-Ub, another for Flag, and another for USP7 and β-tubulin were processed in parallel. D Immunoblotting analysis of LATS2 expression levels in MCF7 and T47D cells expressing shNC or shUSP7 (#1 and #2). The samples derive from the same experiment, but different gels for LATS2, and another for USP7 and β-tubulin were processed in parallel. E Immunoblotting analysis of LATS2 expression levels in cells expressing Flag-NANS alone or in combination with either shNC or shUSP7 (#1 and #2). The samples derive from the same experiment, but different gels for NANS, LATS2, and β-tubulin, and another for USP7 and Flag were processed in parallel. F HEK293T cells were transfected with the indicated expression vectors for 48 h, followed by treatment with 10 μM MG132 for another 6 h. The cell lysates were subsequently subjected to sequential IP and immunoblotting with the indicated antibodies. For the IP samples in panel F, the samples derive from the same experiment, but different gels for V5-Ub and another for Flag were processed in parallel. For the Input samples, different gels for V5-Ub, another for Flag, and another for USP7, HA, and β-tubulin were processed in parallel. G Ubiquitinated Flag-LATS2 was incubated with His-USP7 alone or in combination with His-NANS in deubiquitination buffer, and the reaction products were analyzed by sequential immunoblotting using the indicated antibodies. For the IP samples in panel G, the samples derive from the same experiment, but different gels for V5-Ub and another for Flag were processed in parallel. For the Input samples, different gels for Flag, and another for His-USP7 and His-NANS were processed in parallel. H HEK293T cells were transfected with the indicated plasmids for 48 h and subjected to sequential IP and immunoblotting with the indicated antibodies. For the IP samples in panel H, the samples derive from the same experiment, but different gels for Flag, and another for USP7 and NANS were processed in parallel. For the Input samples, different gels for Flag and another for USP7, NANS, and β-tubulin were processed in parallel. I, J HEK293T cells were transfected with the indicated expression vectors for 48 h, followed by sequential IP and immunoblotting with the indicated antibodies. The input and IP samples were processed in parallel on separate gels. The numbers below the corresponding immunoblots indicate the relative protein expression levels normalized to Flag. For the IP samples in panels I and J, the samples derive from the same experiment, but different gels for Flag and another for USP7 were processed in parallel. For the Input samples in panel I, different gels for Flag, and another for USP7, HA, and β-tubulin were processed in parallel. For the Input samples in panel (J), different gels for Flag and another for USP7, NANS, and β-tubulin were processed in parallel. For representative immunoblots (B–J), experiments were repeated independently three times, yielding similar results. Source data are provided as a Source Data file.

Next, we investigated whether the NANS-mediated stabilization of LATS2 is dependent on USP7. As expected, immunoblotting and ubiquitination assays revealed that knockdown of USP7 partially attenuated the NANS-induced upregulation of LATS2 (Fig. 5E) and impaired the NANS-mediated deubiquitination of LATS2 (Fig. 5F). Furthermore, in vitro deubiquitination assays demonstrated that USP7 reduced LATS2 ubiquitination and that NANS enhanced the USP7-mediated deubiquitination of LATS2 (Fig. 5G). These results suggest that USP7 contributes to the inhibitory effects of NANS on the ubiquitin-dependent proteasomal degradation of LATS2.

We investigated how NANS regulates LATS2 via USP7. Immunoblotting analyses revealed that neither overexpression nor knockdown of NANS affected the expression level of USP7 (Supplementary Fig. S9D, E). Thus, we hypothesized that NANS stabilizes LATS2 by influencing the interaction between USP7 and LATS2. Reciprocal IP assays demonstrated that ectopically expressed Flag-LATS2 interacted with endogenous USP7 and NANS (Supplementary Fig. S9F), whereas exogenously expressed HA-USP7 bound to endogenous LATS2 and NANS (Supplementary Fig. S9G). Similarly, exogenously expressed Flag-NANS interacted with endogenous LATS2 and USP7 in HEK293T, T47D, and MCF7 cells (Supplementary Fig. S9H). Importantly, endogenous NANS also interacted with endogenous LATS2 and USP7 in T47D and MCF7 cells (Supplementary Fig. S9I).

Next, we generated LATS2 truncation constructs to perform domain mapping experiments. Immunoprecipitation assays revealed that NANS interacts with Flag-LATS2 (301-600 aa), whereas USP7 binds to Flag-LATS2 (601-1088 aa) (Fig. 5H). To validate these direct interactions in vitro, we conducted GST pull-down assays. The results revealed that GST-LATS2 (480-1088 aa) can directly bind full-length His-USP7 and full-length His-NANS and that GST-NANS can interact with full-length His-USP7 (Supplementary Fig. S9J–L). These findings provide strong evidence that NANS, USP7, and LATS2 can directly interact with one another. Moreover, overexpression of NANS enhanced, whereas knockdown of NANS reduced, the interaction between LATS2 and USP7 (Fig. 5I, J). Taken together, these results suggest that NANS forms a ternary complex with USP7 and LATS2 and enhances the interaction between LATS2 and USP7 to stabilize LATS2.

NANS promotes the growth of and therapeutic resistance in HR+ /HER2- BC partially by stabilizing LATS2 both in vitro and in vivo

To investigate whether LATS2 is involved in the noted functions of NANS in HR+ /HER2- BC, we conducted functional rescue experiments both in vitro and in vivo. For this purpose, we reintroduced LATS2 into NANS-depleted cells and verified the expression status of NANS and LATS2 in these established stable cell lines by immunoblotting (Fig. 6A).

Fig. 6. NANS promotes HR+ /HER2- breast cancer growth and resistance to tamoxifen and CDK4/6 inhibitors partially by stabilizing LATS2 both in vitro and in vivo.

A MCF7 and T47D cells stably expressing sgNC or sgNANS #2 alone or in combination with Flag-LATS2 were subjected to immunoblotting. The blot shown is representative of n = 3 biologically independent experiments yielding similar results. The numbers below the corresponding immunoblots indicate the relative protein expression levels normalized to β-tubulin. B, C The cells shown in Panel A were subjected to CCK-8 (B) and colony formation (C) assays. The quantitative results for the surviving colonies are shown in (C), and representative images of the surviving colonies are presented in Supplementary Fig. S10A. Data are presented as mean ± SD of n = 3 biologically independent experiments. D The cells shown in Panel (A) were treated with increasing concentrations of tamoxifen (Tam) for 72 h. CCK-8 assays were performed to determine the IC50 values. Data are mean ± SD of n = 4 (T47D) or n = 3 (MCF7) biologically independent experiments. E The cells shown in Panel (A) were subjected to colony formation assays in the presence or absence of the indicated concentrations of Tam for 7–9 days. The quantitative results of the relative surviving colonies are shown. Representative images of surviving colonies are shown in Supplementary Fig. S10B. Data are presented as mean ± SD of n = 3 biologically independent experiments. F The cells shown in Panel A were treated with increasing concentrations of abemaciclib (Abema) for 72 h. CCK-8 assays were performed to determine the IC50 values. Data are presented as mean ± SD of n = 3 biologically independent experiments. G The cells shown in Panel A were subjected to colony formation assays in the presence or absence of the indicated concentrations of Abema for 7–9 days. The quantitative results of the relative surviving colonies are shown. Representative images of surviving colonies are shown in Supplementary Fig. S10C. Data are presented as mean ± SD of n = 3 biologically independent experiments. H The cells shown in Panel A were treated with increasing concentrations of palbociclib (Palbo) for 72 h. CCK-8 assays were performed to determine the IC50 values. Data are presented as mean ± SD of n = 3 biologically independent experiments. I The cells shown in Panel A were subjected to colony formation assays in the presence or absence of the indicated concentrations of Palbo for 7–9 days. The quantitative results of the relative surviving colonies are presented. Representative images of surviving colonies are shown in Supplementary Fig. S10D. Data are presented as mean ± SD of n = 3 biologically independent experiments. J–L Female BALB/c nude mice (n = 6 per group) were orthotopically injected with MCF7 cells stably expressing sgNC or sgNANS #2 alone or in combination with Flag-LATS2. After 20 days of incubation, the mice were treated with vehicle, Tam (50 mg/kg daily), or Abema (25 mg/kg every other day) for 20 days. Images of xenograft tumors (J), tumor weights (K), and tumor weight inhibition rates (L) are shown. Data in (K, L) are presented as mean ± SD of n = 6 biologically independent animals per group. The unit of study is the individual animal. Statistical significance for IC50 values based on nonlinear regression curve fitting (D, F, H) was determined using the extra sum-of-squares F test. Statistical significance for cell viability at the final time point (B), clonogenic capacity (C), surviving colonies at the highest concentration (E, G, I), tumor weights (K), and inhibitory rates (L) was determined using a one-way ANOVA. For all statistical comparisons, exact p-values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

CCK-8 and colony formation assays demonstrated that re-expression of LATS2 into NANS-depleted cells partially reversed the impaired cell viability (Fig. 6B) and colony-forming ability (Fig. 6C and Supplementary Fig. S10A) of MCF7 and T47D cells caused by NANS knockdown. Furthermore, in vitro drug sensitivity assays revealed that the reintroduction of LATS2 into NANS-depleted MCF7 and T47D cells partially attenuated the increased sensitivity to Tam (Fig. 6D, E and Supplementary Fig. S10B) and the CDK4/6 inhibitors Abema (Fig. 6F, G and Supplementary Fig. S10C) and Palbo (Fig. 6H, I and Supplementary Fig. S10D) caused by depletion of NANS. Moreover, restoring LATS2 expression markedly restored the resistance phenotype in both Tam-resistant and Palbo-resistant NANS-depleted MCF7 cells (Supplementary Fig. S10E–G). Importantly, these rescue effects were fully recapitulated in the more clinically representative dual-resistant (TPR) models (Supplementary Fig. S10H, I). Consistently, in vivo xenograft tumor assays demonstrated that the re-expression of LATS2 in NANS-depleted cells partially reversed the decreased tumor growth and increased sensitivity to Tam and Abema caused by the knockdown of NANS (Fig. 6J–L and Supplementary Fig. S10J, K). Collectively, these results highlight that NANS promotes growth and confers resistance to tamoxifen and CDK4/6 inhibitors in HR + /HER2- BC partially through the regulation of LATS2 both in vitro and in vivo.

NANS regulates ERα and CDK4 expression by activating the LATS2-mediated canonical Hippo pathway

Previous studies have documented that LATS2 is a central component of the Hippo pathway that phosphorylates its transcriptional effectors YAP and TAZ. Phosphorylated YAP and TAZ are sequestered in the cytoplasm or degraded, thus regulating downstream transcriptional programs involved in numerous biological processes^17,19,20. In this context, LATS1/2 are essential for maintaining ERα expression in ER+ breast cancer cells through the negative regulation of YAP/TAZ in a kinase activity-dependent manner^26,28, and decreased YAP1 expression in ER+ breast cancer is associated with more aggressive clinicopathologic phenotypes, increased ER expression, and decreased sensitivity to tamoxifen⁴⁹. In addition, CDK6 is regulated by Hippo signaling in ER+ breast cancer cells⁵⁰. Thus, we next investigated the effects of NANS knockdown on the activation of the Hippo pathway. As shown in Supplementary Fig. S11A, we found that knockdown of NANS expression significantly reduced the expression levels of LATS2 (but not LATS1), phosphorylated YAP (p-YAP), phosphorylated TAZ (p-TAZ), CDK4, CDK6, and ERα (compare Lane 2 and Lane 1). Moreover, the re-expression of LATS2 in NANS-depleted cells partially restored the reduced expression of p-YAP, p-TAZ, CDK4 and ERα but not that of CDK6 (compare Lane 3 and Lane 2). Furthermore, to assess changes in YAP phosphorylation more sensitively, we performed Phos-tag immunoblotting assays. Consistent with the above results, knockdown of NANS markedly decreased p-YAP levels, accompanied by an increase in the levels of nonphosphorylated YAP. Notably, the re-expression of LATS2 in NANS-depleted cells partially reversed the reduction in p-YAP levels (Supplementary Fig. S11B). These results suggest that NANS regulates CDK4 and ERα expression via LATS2. Furthermore, immunofluorescence staining revealed that knockdown of NANS led to a marked increase in the nuclear localization of YAP and TAZ, and these effects were attenuated in NANS-depleted cells with LATS2 re-expression (Supplementary Fig. S11C–F). Notably, knockdown of YAP1 in NANS-depleted cells partially restored the reduced expression levels of CDK4 and ERα caused by NANS knockdown (Supplementary Fig. S11G). Conversely, re-expression of the constitutively active YAP mutant (YAP1−5SA), in which five serine residues (positions 61, 109, 127, 164, and 381) were replaced by alanine⁵¹, in NANS-overexpressing MCF7 and T47D cells significantly reduced the altered CDK4 and ERα expression levels mediated by NANS (Supplementary Fig. S11H). Together, these findings suggest that NANS regulates ERα and CDK4 expression by activating the LATS2-mediated Hippo pathway.

As CDK4 plays a pivotal role in the cell cycle transition from the G1 phase to the S phase⁵², we next examined whether NANS affects cell cycle progression. Flow cytometry analysis revealed that knockdown of NANS led to cell cycle arrest at the G1/S phase (Supplementary Fig. S12A, B). Moreover, immunoblotting revealed the downregulation of G1/S checkpoint proteins, including CDK4 and CDK6, as well as their downstream targets, including phosphorylated retinoblastoma protein (Rb) at serine residues 807 and 811 (phospho-Rb S807/811) and cyclin A2⁹, following knockdown of NANS (Supplementary Fig. S12C). Taken together, these findings suggest that NANS promotes the cell cycle transition from the G1 phase to the S phase.

The NANS-USP7/LATS2 axis is clinically relevant in HR+ /HER2- BC specimens

To evaluate the clinical relevance of the NANS-USP7/LATS2 axis in HR+ /HER2- BC, we first performed polychromatic immunofluorescence staining⁵³ of 45 HR+ /HER2- BC samples (Fig. 7A). The results revealed strong colocalization of NANS, USP7, and LATS2 in these samples, supporting their spatial and functional association in HR+ /HER2- BC tissues (Fig. 7B). Immunohistochemical (IHC) staining of tumor samples from 20 patients with HR+ /HER2- BC (Fig. 7C) revealed a positive correlation between the protein expression levels of NANS and LATS2 (Fig. 7D) as well as between those of USP7 and LATS2 (Fig. 7E). Moreover, CDK4 expression levels were positively correlated with NANS, USP7, and LATS2 expression in the same samples (Fig. 7F–H). Notably, we did not evaluate ERα expression in these samples because the positive relationship between LATS2 and ERα in ER+ breast cancer has been well documented previously^26,28. Survival analyses using an RNA-seq dataset of HR+ /HER2- BC samples from the FUSCC¹⁶ revealed that high USP7 expression was associated with poor OS, RFS, and DMFS in patients who received ET (Supplementary Fig. S13A–C). Similarly, in the external METABRIC dataset with well-annotated endocrine therapy information, high USP7 expression was also associated with poor OS and RFS in patients receiving endocrine therapy (Supplementary Fig. S13D, E). In addition, high expression of NANS was associated with poor OS in patients who received ET (Supplementary Fig. S4A), and the expression levels of CDK4 were negatively associated with RFS in patients who received ET (Supplementary Fig. S13F, G) in the FUSCC and METABRIC datasets.

Fig. 7. The NANS-USP7/LATS2 axis is clinically relevant in HR+ /HER2- breast cancer specimens.

A Polychromatic immunofluorescence staining was conducted using tissue samples from HR+ /HER2- BC patients to detect the colocalization of NANS, USP7, and LATS2. Scale bars: 50 μm. B Pairwise colocalization analysis of NANS, USP7, and LATS2 was performed using the Colocalization Finder plugin in ImageJ. Scatter plots and corresponding Pearson’s Rr values are presented. C Immunohistochemical (IHC) staining was conducted to assess the expression levels of NANS, LATS2, USP7, and CDK4 in 20 HR+ /HER2- BC samples. Scale bars: 50 μm. Representative IHC staining images are shown from n = 20 biologically independent patient samples. D–H Pearson correlation analysis was performed to evaluate the associations between the protein levels of LATS2 and NANS (D), LATS2 and USP7 (E), and CDK4 and NANS (F), LATS2 (G), and USP7 (H) in the samples presented in Panel (C). I–J CellTiter-Glo 3D viability assays were conducted in two HR+ /HER2- BC patient-derived organoids (PDOs) treated with or without P5091, followed by treatment with tamoxifen (Tam) (2 μM), abemaciclib (Abema) (0.5 μM), palbociclib (Palbo) (1 μM), or the combination of Tam and Abema at the concentrations indicated above. Representative images of the posttreated organoids are shown in (I), and the corresponding quantitative results for cell viability are presented in (J). Scale bar: 100 μm. Data in (J) are presented as mean ± SD of n = 4 biologically independent experiments. K–O Female BALB/c nude mice were orthotopically injected with MCF7 cells stably expressing sgNC or sgNANS #2 (n = 6 biologically independent animals per group). After 20 days of incubation, the mice were treated with vehicle, Tam (50 mg/kg daily), or Abema (25 mg/kg every other day), either alone or in combination with P5091 (15 mg/kg daily) for 20 days. Images of xenograft tumors (K), tumor weights (L), tumor weight inhibition rates (M), tumor volumes (N), and tumor volume inhibition rates (O) are shown. Data in (L–O) are presented as mean ± SD. The unit of study is the individual animal. P The proposed working model. NANS is highly expressed in HR+ /HER2- BC cells and increases LATS2 stability through the recruitment of the deubiquitinase USP7. Stabilized LATS2 induces ERα and CDK4 expression via its negative effector YAP, resulting in tumor progression. Representative images in (A) are shown from n = 45 biologically independent patient samples. For representative images in (I), experiments were repeated independently four times, yielding similar results. Statistical significance for cell viability (J), tumor weights (L), inhibitory rates (M, O), and tumor volumes (N) was determined using a one-way ANOVA. For all statistical comparisons, exact p- values are indicated directly within the respective figure panels, and 95% confidence intervals (CIs) are provided in the Source Data. Source data are provided as a Source Data file.

Finally, we constructed a four-gene signature comprising NANS, USP7, LATS2, and CDK4 using multivariate Cox proportional hazards regression analysis among patients who received endocrine therapy (ET) in the FUSCC cohort. On the basis of this signature, patients were stratified into high-risk and low-risk groups, with the high-risk group exhibiting significantly poorer overall survival (OS) and relapse-free survival (RFS) (Supplementary Fig. S13H, I). We then validated the signature in the METABRIC cohort, in which high-risk patients who received ET similarly demonstrated worse prognosis (Supplementary Fig. S13J, K). These findings suggest that the NANS-USP7/LATS2 axis is clinically relevant in HR + /HER2- BC specimens.

It has been well documented that USP7 is overexpressed in various types of human cancer and is an important regulator of tumorigenesis and an emerging drug target for cancer treatment⁵⁴. In accordance with these data, selective USP7 inhibitors, such as P5091, exhibit broad-spectrum antitumor activity against multiple malignancies⁵⁵. On the basis of these results and our findings, we evaluated the effect of the USP7 inhibitor P5091 on drug sensitivity in tamoxifen-resistant and palbociclib-resistant MCF7 cells. P5091 treatment significantly increased the sensitivity of TamR cells to tamoxifen and of PalboR cells to palbociclib, producing effects comparable to those observed with NANS knockdown (Supplementary Fig. S4E–G). Moreover, we examined the effects of the USP7 inhibitor P5091 on drug sensitivity in the PDOs shown in Supplementary Fig. S6G, H. The results demonstrated that the administration of P5091 effectively increased the sensitivity of drug-resistant PDO #333 to Tam and the CDK4/6 inhibitors Abema and Palbo (Fig. 7I, J). These findings were further validated in xenograft models, in which P5091 treatment markedly suppressed tumor growth and enhanced the sensitivity of the tumor to tamoxifen and abemaciclib, with inhibitory effects comparable to NANS inhibition (Fig. 7K–O). Furthermore, to better recapitulate the clinical treatment setting, we employed an exemestane- and abemaciclib-resistant patient-derived xenograft (PDX) model for additional in vivo drug sensitivity studies. The results demonstrated that the addition of P5091 significantly enhanced the sensitivity of the resistant PDX tumors to exemestane and abemaciclib combination therapy (Supplementary Fig. S14A–C). Together, these preclinical data strongly suggest that pharmacological targeting of USP7 can effectively overcome resistance to endocrine and CDK4/6 inhibitor therapies in HR+ /HER2- breast cancer.

Collectively, these results suggest that the NANS-USP7/LATS2 axis is clinically relevant in HR+ /HER2- BC and that targeting USP7 might overcome NANS-mediated resistance to ET and CDK4/6 inhibitors in this disease (Fig. 7P).

Discussion

ET is the cornerstone of systemic treatment for HR+ /HER2- BC patients and substantially reduces both disease recurrence and patient mortality^56,57. However, prolonged ET inevitably induces resistance, thus driving extensive research into alternative targeted therapies such as CDK4/6 inhibitors. Clinical evidence has shown that ET combined with CDK4/6 inhibitors significantly prolongs progression-free survival. However, the development of resistance represents a major clinical challenge^40,58,59. Therefore, deciphering the mechanisms of resistance to these therapies and discovering therapeutic targets or predictive biomarkers of the therapeutic response are crucial priorities in the management of HR+ /HER2- BC patients. In this study, we report several interesting findings concerning previously unrecognized functional and mechanistic roles of NANS in driving tumor growth and conferring therapeutic resistance in HR+ /HER2- BC cells.

First, we found that NANS is highly expressed in HR+ /HER2- BC and promotes tumor growth and resistance to ET and CDK4/6 inhibitors in an enzymatic activity-independent manner. While sialic acid metabolism has been thoroughly characterized in various tumor contexts, these studies focused mainly on elucidating how sialylation processes contribute to tumorigenesis and disease progression^60–62. For instance, cytidine monophosphate N-acetylneuraminic acid synthetase (CMAS), a rate-limiting enzyme in sialic acid metabolism, promotes breast cancer progression through a sialylation-dependent mechanism^63,64. Despite its fundamental role in sialic acid biosynthesis, NANS remains poorly characterized in human cancers, with several limited reports in prostate, endometrial and colorectal cancers^14,15,65. In this study, we demonstrated that NANS exhibits strong tumor-promoting activity in HR+ /HER2- BC, increasing tumor growth and resistance to ET and CDK4/6 inhibitors (Figs. 2, 3 and Supplementary Figs. S3–S6). Interestingly, these functions of NANS in HR+ /HER2- BC cells are independent of its enzymatic activity (Supplementary Fig. S7).

Second, we discovered that NANS performs its oncogenic functions in HR+ /HER2- BC partially by stabilizing LATS2 both in vitro and in vivo. Dysregulation of the Hippo pathway has been extensively reported in various cancers^17–19. LATS2, a core component of this pathway, was initially recognized as a classical tumor suppressor^{17,20,66–68}. However, emerging evidence shows that LATS2 may exhibit oncogenic functions in certain cancer contexts^69–71. Notably, multiple studies have demonstrated that LATS2 functions as an oncogenic factor, while its downstream effector, YAP1, acts as a tumor suppressor in HR+ /HER2- BC^25–28. Consistent with these results, our study demonstrated that NANS promotes tumor growth and confers resistance to tamoxifen and CDK4/6 inhibitors in HR+ /HER2- BC partially by stabilizing LATS2 (Figs. 4–6 and Supplementary Fig. S8–S10). Previous studies have shown that LATS2 can be regulated by the E3 ubiquitin ligases DCAF1⁷², FZR1⁷³, and SIAH2³⁰ as well as by the deubiquitinase USP9X³² in human cancer, thereby affecting its protein stability. In this study, we revealed that USP7 acts as a deubiquitinase of LATS2 that mediates its stability and that NANS enhances the interaction between USP7 and LATS2 to stabilize LATS2 (Fig. 5 and Supplementary Fig. S9). Importantly, the reintroduction of LATS2 into NANS-depleted cells partially rescued the impaired functional phenotypes caused by NANS knockdown (Fig. 6 and Supplementary Fig. S10), and the USP7 inhibitor P5091 effectively reversed NANS-induced resistance to tamoxifen or CDK4/6 inhibitors alone or in combination in the PDO model, cell line-derived xenograft, and patient-derived xenograft (PDX) models (Fig. 7 and Supplementary Fig. S14). Moreover, polychromatic immunofluorescence and immunohistochemical staining revealed that NANS, LATS2, and USP7 were colocalized in clinical HR+ /HER2- BC samples and that there were positive correlations between the protein expression levels of NANS and LATS2 as well as between those of USP7 and LATS2 (Fig. 7). These results highlight that the NANS-USP7/LATS2 axis is clinically relevant in HR+ /HER2- BC.

Third, we demonstrated that NANS regulates ERα and CDK4 expression by activating the LATS2-mediated canonical Hippo pathway. Previous studies have shown that LATS1/2 promotes ER-positive (ER+ ) breast cancer growth by maintaining ER expression via YAP/TAZ inhibition^26,28 and that pharmacological inhibition of LATS1/2 blocks ER+ breast cancer growth and increases the sensitivity of ER+ breast cancer cells to the CDK4/6 inhibitor palbociclib²⁸. In addition, CDK4 overexpression has been detected in approximately 15% of breast cancer cases⁷⁴, and its overexpression contributes to resistance to both ET and CDK4/6 inhibitors^9,75–79. Consistent with previous studies, we demonstrated that NANS regulates ERα expression via the LATS2-YAP pathway (Supplementary Fig. S11). In addition, we report that LATS2 regulates CDK4 expression via the LATS2-mediated canonical Hippo pathway, providing insights into the oncogenic functions of Hippo signaling in breast cancer.

In conclusion, the findings presented here reveal that NANS is highly overexpressed in HR+ /HER2- BC and functions as an oncogene to promote growth and therapeutic resistance by stabilizing LATS2. NANS may serve as a promising therapeutic target and predictive biomarker of the treatment response in HR+ /HER2- BC patients.

Methods

Ethics compliance and consent

This study was conducted in accordance with the Declaration of Helsinki and was approved by the Medical Ethics Committee of Fudan University Shanghai Cancer Center (FUSCC). All experiments involving human patient-derived samples were performed with informed consent obtained from all participants. Animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of FUSCC (Approval numbers: FUSCC-IACUC-2024311, FUSCC-IACUC-2025584, FUSCC-IACUC-2026113, and FUSCC-IACUC-2024548).

Cell culture and chemicals

The human embryonic kidney 293 T (HEK293T) cell line, human mammary epithelial cell lines (HMEC and MCF10A), and human breast cancer cell lines (MCF7, T47D, HCC1954, MDA-MB-231, LM2-4175, and BT549) were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China) and Shanghai Key Laboratory of Breast Cancer. MCF10A and HMEC cells were maintained in DMEM (#L110, BasalMedia) supplemented with 5% donor horse serum (#16050-114, Gibco), 10 μg/mL insulin (#40107ES76, Yeasen Biotech), 20 ng/mL epidermal growth factor (#10605-HNAE, Sino Biol), and 0.5 μg/mL hydrocortisone (#40109ES08, Yeasen Biotech). The other cell lines were cultured in DMEM supplemented with 10% fetal bovine serum (#10270-106; Gibco) and 1% penicillin‒streptomycin (#S110B; BasalMedia).

The tamoxifen-resistant (TamR) MCF7 cell line and its parental counterpart were kindly provided by Prof. Cuixia Yang (Shanghai Jiao Tong University School of Medicine, Shanghai, China)⁸⁰. The palbociclib-resistant (PalboR) MCF7 cell line and its corresponding parental line were purchased from MEISEN CELL (Shanghai, China). The tamoxifen (Tam) and palbociclib (Palbo) dual-resistant (TPR) MCF7 cell line was established by continuous exposure of PalboR MCF7 cells to tamoxifen for 3–4 months. The tamoxifen- and palbociclib-resistant (TPR) T47D cell line was generated by continuous treatment with both tamoxifen and palbociclib for 3–4 months. To maintain the drug-resistant phenotypes of these cells, MCF7-TamR and MCF7-PalboR cells were cultured in DMEM supplemented with 1 µM tamoxifen (Tam) or 1 µM palbociclib (Palbo), respectively. MCF7-TPR and T47D-TPR cells were cultured in DMEM supplemented with 0.5 µM tamoxifen (Tam) and 0.2 µM palbociclib (Palbo). MCF7 and T47D cells were stably transfected with the aromatase gene CYP19A1 or corresponding empty vectors to establish MCF7-aro and T47D-aro cell lines for use in aromatase inhibitor sensitivity assays. Cells were cultured in phenol red-free DMEM (BasalMedia, #L140KJ) supplemented with 10% charcoal-stripped fetal bovine serum (#12676-029, Gibco), 1% penicillin–streptomycin, and androstenedione (10⁻⁷ mol/L; GC19559, GLPBIO). All cell lines were authenticated by short tandem repeat (STR) profiling. All cell lines were routinely tested and confirmed to be negative for mycoplasma contamination. Detailed information regarding all the reagents used in this study is provided in Supplementary Table S1.

Clinical samples and datasets

The multiomics HR+ /HER2- BC datasets from Fudan University Shanghai Cancer Center (FUSCC), including RNA sequencing (RNA-seq), single-cell RNA sequencing (scRNA-seq), metabolome, and quantitative proteomics data, and the clinical information of these patients have been described previously¹⁶. HR+ /HER2- BC tissue samples and adjacent normal breast tissues were collected from female patients who underwent surgery at the Department of Breast Surgery of FUSCC and did not receive any treatment. All participants provided written informed consent. The experimental procedures were approved by the Medical Ethics Committee of FUSCC. An RNA-seq HR+ /HER2- BC dataset from the TCGA (The Cancer Genome Atlas) was downloaded from https://portal.gdc.cancer.gov/.

Construction of expression vectors

The expression vectors encoding Flag-NANS, HA-NANS, Flag-NANS-Mut, HA-USP7, HA-USP7 (C223S), YAP1 (5SA), Flag-LATS2 (amino acids 1-300), Flag-LATS2 (amino acids 301-600), Flag-LATS2 (amino acids 601-1088), and single guide RNAs (sgRNAs) targeting NANS (sgNANS #2 and #3) and YAP1 (sgYAP1) were synthesized by Syn-Biotech (Suzhou, China). Flag-LATS2 was synthesized by MiaoLing Biol (Wuhan, China). The sequences of short hairpin RNAs (shRNAs) targeting USP7 (shUSP7 #1 and #2) were obtained from Sigma‒Aldrich and cloned into the pLKO.1-TRC vector (#10878; Addgene). The sequences of the sgRNAs and shRNAs used in this study are presented in Supplementary Tables S2 and S3, respectively.

Plasmid transfection and lentiviral infection

Plasmid transfection was performed using Neofect DNA transfection reagent (#TF201201; Neofect Biotech) according to the manufacturer’s protocol. For lentiviral infection, the packaging plasmids and target plasmids were transfected into HEK293T cells for 48 h. The supernatants were filtered, collected, and used for cell infection along with polybrene (#H9268, Sigma‒Aldrich). After 48 h of viral infection, the transduced cells were subjected to antibiotic selection using 2 μg/mL puromycin (#S7417, Selleck) or 0.5 mg/mL geneticin (G418) (#A600958-0005, Sangon Biotech).

RNA isolation and real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from cultured cells and tissues using TRIzol reagent (#9109; Takara), followed by reverse transcription using HiScript III RT SuperMix (#R323-01; Vazyme). Quantitative PCR (qPCR) assays were performed using ChamQ Universal SYBR qPCR Master Mix (#Q711-03, Vazyme). Relative gene expression levels were calculated using the 2^−ΔΔCT method. The sequences of the primers used for the qPCR assays are listed in Supplementary Table S4.

Immunoblotting and immunoprecipitation (IP) assays

Cellular proteins were extracted using RIPA lysis buffer supplemented with both protease and phosphatase inhibitor cocktails (#B14002 and #B15003, Bimake). After quantification using a BCA protein quantification kit (#20201ES; Yeasen Biotech), equal amounts of protein were subjected to SDS-PAGE and then transferred to PVDF membranes (#IPVH00010; Millipore). The membranes were blocked with 5% bovine serum albumin and incubated sequentially with primary and HRP-conjugated secondary antibodies. Signal detection was performed using a chemiluminescent substrate kit (#180-5001E, Tanon). Phos-tag reagents were purchased from GLPBIO (#GK10005), and Phos-tag SDS–PAGE gels were prepared according to the manufacturer’s instructions. For exogenous IP assays, cells were lysed using NP-40 lysis buffer and incubated with anti-HA (#GNI4510-HA; Shanghai Genomics Technology) or anti-Flag (#B23102; Bimake) beads for more than 3 h. For endogenous IP assays, cellular lysates were incubated with primary antibodies overnight at 4 °C, followed by incubation with protein A/G magnetic beads (#B23202; Bimake) for 3 h. After being washed three times, the immunocomplexes were subjected to immunoblotting. The information for all antibodies used in this study is provided in Supplementary Data 1.

GST pull-down assays

GST-tagged full-length NANS, His-tagged full-length USP7, and His-tagged full-length NANS were expressed in E. coli BL21 (DE3). After cells were lysed, the proteins were purified via affinity chromatography using glutathione Sepharose for GST-tagged proteins and Ni-NTA resin for His-tagged proteins. GST protein (#11213-HNAE) and GST-tagged C-terminal LATS2 (amino acids 480-1088; #L02-35G) were obtained commercially from Sino Biological. For GST pull-down assays, the proteins were incubated in binding buffer (20 mM Tris pH7.5, 500 mM NaCl, 0.5% NP-40, and 10% glycerol) overnight at 4 °C to allow binding. The beads were then washed three times with wash buffer (10 mM Tris pH7.5, 150 mM NaCl, and 0.1% NP-40), resuspended in 2 × SDS loading buffer, and analyzed by immunoblotting with the indicated antibodies.

Immunofluorescence (IF) staining

Cells were washed with PBS and fixed with 4% paraformaldehyde for 10 min, followed by permeabilization with 0.5% Triton X-100 (#93443; Sigma‒Aldrich). After being blocked with 5% BSA, the cells were incubated with primary antibodies overnight at 4 °C. After being washed with PBST three times, the cells were incubated with Alexa Fluor 488- or Alexa Fluor 555-conjugated secondary antibodies for 1 h. The nuclei were counterstained with DAPI (#ab104139; Abcam). Images were captured using a Leica fluorescence microscope (Leica Microsystems, Buffalo Grove, USA).

Immunohistochemical (IHC) staining

Human breast cancer tissues and mouse xenograft tumor tissues were fixed in formalin, embedded in paraffin, and sectioned. After deparaffinization and rehydration, antigen retrieval was performed using 10 mM sodium citrate buffer (pH 6.0). The sections were incubated with 3% hydrogen peroxide for 10 min to block endogenous peroxidase activity, followed by serum blocking for 1 h. The tissues were then incubated with primary antibodies overnight at 4 °C, followed by incubation with secondary antibodies (1:200) at room temperature for 1 h. Signal detection was performed using a DAB (3,3’-diaminobenzidine) staining kit (#G1212-200T; Servicebio), followed by hematoxylin counterstaining.

Proteomic analysis

For DIA quantitative proteomics, total proteins were extracted from MCF7 cells (sgNC vs. sgNANS #2, n = 3 biological replicates per group). The peptides were desalted and analyzed via LC-MS/MS on an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific) (Shanghai Bioprofile Biotech). To investigate LATS2-interacting proteins, cellular lysates from HEK293T cells (pCDH vs. Flag-LATS2) were immunoprecipitated using anti-Flag magnetic beads (Bimake). The complexes were separated by SDS-PAGE and stained with Coomassie Brilliant Blue. The protein bands were excised and analyzed via LC-MS/MS using an Orbitrap Exploris 480 mass spectrometer (Thermo Fisher Scientific) (Westlake Omics, Hangzhou, China). Protein identification was performed using Fragpipe (v19.1) and the MSFragger search engine against the UniProt Swissprot_Human database (containing 20,456 reviewed proteins). Search parameters included: trypsin digestion with a maximum of two missed cleavages; precursor and fragment mass tolerances were set at 20 ppm; and peptide FDR was set at < 0.01.

In vivo and in vitro ubiquitination assays

For the in vivo ubiquitination assays, HEK293T cells were transiently transfected with the indicated plasmids for 48 h, followed by treatment with the proteasome inhibitor MG132 (10 μM; #S2619, Selleck) for 6 h. Cells were then lysed using denaturing buffer and subjected to IP assays.

For the in vitro deubiquitination assays, HEK293T cells were co-transfected with V5-tagged ubiquitin and Flag-tagged LATS2 and then treated with MG132 for 6 h. Ubiquitinated Flag-LATS2 was immunoprecipitated, and the isolated protein was incubated with His-USP7 alone or together with His-NANS in deubiquitination buffer (60 mM HEPES, 5 mM MgCl₂, and 4% glycerol, pH7.5) at 30 °C for 4 h. Deubiquitinated proteins were then analyzed by immunoblotting.

Cycloheximide (CHX) chase assays

Cells were treated with 100 μg/mL CHX for the indicated durations and then subjected to immunoblotting assays with the indicated antibodies.

Cell Counting Kit-8 (CCK-8) assays

Cells were seeded into 96-well plates, and cell viability was determined by measuring the optical density at 450 nm (OD450) using a CCK-8 kit (#40203ES92; Yeasen Biotech).

Cell cycle analysis

Cells were fixed in 70% ethanol for 2 h and subsequently washed with PBS. Cell staining was performed using a cell cycle analysis kit (#40301ES50; Yeasen Biotech) according to the manufacturer’s instructions. Flow cytometry was then used to analyze the cell cycle distribution.

Flow cytometry analysis of apoptosis

Apoptosis was assessed with an Annexin V–FITC/PI Apoptosis Detection Kit (#40302ES60; Yeasen Biotech) according to the manufacturer’s protocol. Collected cells were washed twice with ice‑cold PBS, resuspended in binding buffer, and stained with Annexin V–FITC and propidium iodide (PI) for 10–15 min at room temperature in the dark. Flow cytometric analysis was performed within 1 h after staining. The sequential gating strategies for flow cytometry are shown in Supplementary Fig. S15.

Cell migration and invasion assays

Migration and invasion assays were performed using Transwell chambers (Matrigel-coated for invasion) placed in 24-well plates. The lower chambers contained DMEM supplemented with 10% fetal bovine serum as a chemoattractant. Cells were resuspended in serum-free DMEM and seeded into the upper chambers. After 24 h of incubation, cells that had migrated or invaded were fixed with methanol and stained with 0.2% crystal violet for quantification.

Drug sensitivity assays

For the cell viability assays, cells were seeded into 96-well plates and treated with or without P5091, letrozole, tamoxifen, abemaciclib, or palbociclib. Cell viability was assessed by measuring the OD450 using a CCK-8 kit. The half maximal inhibitory concentration (IC50) was determined using a three-parameter nonlinear regression model (log[inhibitor] versus response) in GraphPad Prism 9.5.1. For the colony formation assays, the cells were plated in 12-well plates and treated as indicated for 7–9 days. Surviving colonies were then fixed with methanol and stained with 0.2% crystal violet.

Patient-derived organoids (PDOs)

PDOs were established following previously described methods⁸¹. Briefly, fresh tumor tissues were obtained from HR+ /HER2- BC patients at FUSCC, minced, and digested with collagenase (#C9407; Sigma‒Aldrich) and hyaluronidase (#37326-33-3; Sigma‒Aldrich) at 37 °C for 1–3 h. The digested mixture was filtered through a 100 µm cell strainer and centrifuged. Red blood cells were lysed using the lysis buffer. The resulting organoid pellets were resuspended in 1 mL of basement membrane extract type-2 (#3533-010-02, Trevigen) and seeded into 384-well plates (#GN781900, Greiner). After 5-7 days of drug treatment, morphological changes were observed under a microscope, and organoid viability was assessed using CellTiter-Glo 3D assays (#G9683; Promega).

In vivo xenograft tumor models

All animal experimental procedures were approved by the Institutional Animal Care and Use Committee (IACUC) of FUSCC (Approval number: FUSCC-IACUC-2024311, FUSCC-IACUC-2025584, FUSCC-IACUC-2026113 and FUSCC-IACUC-2024548). The maximal tumor size allowed by the IACUC of FUSCC was 2000 mm³ (or 20 mm in diameter). We confirm that the tumor burden in our experiments did not exceed this limit. For cell line–derived xenograft models, a total of 5 × 10⁶ mycoplasma-free MCF7 cells in the logarithmic growth stage were suspended in PBS and mixed with an equal volume of Matrigel (#356334, Corning Falcon). Following subcutaneous implantation of 17β-estradiol pellets (#SE-121, Innovative Research of America), the cell suspension was injected into the mammary fat pads of female BALB/c nude mice. When the tumor volume reached 50–100 mm³, the mice were treated with P5091 (15 mg/kg, dissolved in 6.5% DMSO, 3% Tween-80 and 20% PEG400 in ddH₂O, administered daily via intraperitoneal injection), tamoxifen (50 mg/kg, dissolved in corn oil, administered daily via intraperitoneal injection), or abemaciclib (25 mg/kg, dissolved in corn oil, administered every other day via oral gavage). The tumor diameters were measured regularly, and the tumor volumes were calculated using the formula 0.5 × length × width². The experiments were terminated, and mice were euthanized, after 20 days of treatment or earlier if humane endpoints were reached. Humane endpoints were defined as a maximum tumor burden of 2000 mm³ or 20 mm in diameter, rapid weight loss exceeding 10% of the initial body weight, or severe tumor ulceration.

For patient-derived xenograft (PDX) models, four- to six-week-old female NCG mice were ovariectomized and allowed to recover from surgery for one week prior to tumor implantation. Androstenedione (100 μg, dissolved in corn oil) was administered subcutaneously to each mouse 5 days per week throughout the experiment to provide a substrate for aromatase. The dual-resistant PDX model was derived from an HR + /HER2- BC patient who developed liver and bone metastases within two months of receiving combined exemestane and abemaciclib therapy. As previously described^82,83, the PDX fragments were subcutaneously implanted into the flanks of the NCG mice. Tumor dimensions were monitored twice weekly, and tumor volumes were calculated using the formula: V = 0.5 × length × width². When the average tumor volume reached approximately 50–100 mm³, the mice were assigned into four treatment groups (n = 6 mice/group): (1) Vehicle control; (2) Exemestane (15 mg/kg, dissolved in 6.5% DMSO, 3% Tween-80, and 40% PEG400 in ddH₂O, administered daily via intraperitoneal injection) plus Abemaciclib (25 mg/kg, administered every other day via oral gavage); (3) P5091 (15 mg/kg, administered daily via intraperitoneal injection); and (4) the combination of exemestane, abemaciclib, and P5091. All treatments were administered continuously for 20 days. The experiments were terminated at this point, or earlier if mice met any of the predefined humane endpoint criteria, including a maximum tumor burden of 2000 mm³ or 20 mm in diameter, rapid weight loss exceeding 10% of the initial body weight, or severe tumor ulceration. At the endpoint, the mice were euthanized, and xenograft tumors were excised, photographed, and weighed for subsequent analyses.

Bioinformatic analyses

The differentially expressed genes (DEGs) between two groups were analyzed using the R package limma with a log₂FC (fold change) > 0.7 and an adjusted p-value < 0.05 as the cutoff criteria. Gene clustering was performed on the basis of similar expression patterns in HR+ /HER2- BC samples from the FUSCC cohort using the WGCNA (weighted gene coexpression network analysis) package in R version 4.3.0. A minimum threshold was applied to construct a dendrogram, and genes with similar characteristics were grouped into the same module. Multivariate Cox proportional hazards regression was conducted to assess the association between the expression of multiple genes and patient survival.

Sex and gender considerations

This study focuses on HR+ /HER2- breast cancer, a disease that predominantly affects females. Therefore, only female human participants, female-derived cell lines, and female mice were included in the experimental design. The sex of the human research participants was determined based on self-report in the clinical records. As this research applies to a single sex, sex- and gender-based analyses were not performed.

Statistics & reproducibility

All data are presented as the mean ± standard deviation (SD) from at least three independent experiments or biological replicates. Differences between the two groups were analyzed using an paired or unpaired Student’s t test, whereas one-way ANOVA was used for comparisons among multiple groups. Kaplan‒Meier survival curves (log-rank test) were used to assess differences in survival. Statistical analyses were performed using GraphPad Prism 9.5.1, R (version 4.3.0), or ImageJ software (version 1.54). A p-value less than 0.05 was considered to indicate statistical significance (Exact p-values are provided directly within the figures whenever possible; otherwise, significance is denoted as (*, p < 0.05; **, p < 0.01; ***, p < 0.001). No statistical method was used to predetermine sample size. No data were excluded from the analyses. The investigators were not blinded to allocation during experiments and outcome assessment. All experiments were performed at least three times with consistent results, or representative data are shown. For in vivo studies, mice were assigned to experimental groups to ensure comparable tumor volumes at the start of treatment.

Reporting summary

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

Author contributions

J.Y.C.: data curation, formal analysis, investigation, and writing the manuscript draft. M.Y.H.: data curation, formal analysis, and investigation. S.Y.Y., Y.L.Z., F.L.Z., L.A. and Q.Z.: data analysis and investigation. A.Y.C.: clinical samples and associated research resources. D.Q.L. and Z.M.S.: supervision, funding acquisition, project administration, and manuscript editing.

Peer review

Peer review information

Nature Communications thanks Carla Alves, Nelson Soares, Mark von Itzstein and the other anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by the National Natural Science Foundation of China (Grant Nos. 82372633 and 82573776 to D.Q.L.), the Shanghai Key Laboratory of Breast Cancer (Grant No. 12DZ2260100 to Z.M.S.), and the SHDC Municipal Project for Developing Emerging and Frontier Technology in Shanghai Hospitals (Grant No. SHDC12021103 to Z.M.S.).

Data availability

The RNA sequencing data and metabolome data for the FUSCC HR+ /HER2- breast cancer cohort have been deposited in the Genome Sequence Archive (GSA) database under accession code PRJCA017539. TMT-based mass spectrometry (MS)-quantified protein data for the FUSCC HR+ /HER2- breast cancer cohort have been submitted to iProX under accession code IPX0006535000. The raw data-independent acquisition (DIA) quantitative proteomics data and LC-MS/MS protein identification raw data for our study have been deposited in the PRIDE database under accession codes PXD065351 and PXD065340. The TCGA and METABRIC data were downloaded from the cBioPortal website [https://www.cbioportal.org/]. The CPTAC data were analyzed using the UALCAN portal [https://ualcan.path.uab.edu/analysis-prot.html]. All remaining data can be found in the Article, Supplementary, and Source Data files. Source data are provided in this paper.

Code availability

All code used for bioinformatics and statistical analyses in this study relies on standard, publicly available algorithms and R packages. No mathematical algorithms central to the conclusions were generated.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75386-x.