TM9SF1 drives the lipophagic flux via AMPK-ULK1 signaling to sustain metabolic fitness in HER2-positive breast cancer

Transcriptomic data (RNA-seq) for all available 180 female HER2 + BC patients and 22 normal controls were obtained from The Cancer Genome Atlas-Breast Cancer (TCGA-BRCA) database [28]. Additionally, a cohort of 78 HER2 + BC specimens was consecutively collected at the Fujian Cancer Hospital during our study period, of which 63 were paired with adjacent normal tissues (2 cm periphery of resected tumors). For human specimens, inclusion criteria were female patients with a confirmed diagnosis of HER2+ breast cancer as defined by an immunohistochemistry score of 3+ or a fluorescence in situ hybridization score ≥ 2.0. No pre-established exclusion criteria were applied, and no samples were excluded from the analysis during the study. The study was approved by the Ethics Committee of Fujian Cancer Hospital (K2024-521-01), and informed consent was obtained from all participants.

TCGA gene expression data were analyzed for differential expression using the 'limma' R package. Gene set enrichment analysis (GSEA) was performed using the Molecular Signatures Database to identify enriched pathways [29]. The prognostic significance of TM9SF1 in HER2 + BC was assessed using the Kaplan-Meier plotter (http://kmplot.com/analysis/↗).

MCF10A, MDA-MB-361, SKBR3, and HEK293T cells were sourced from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences, and JIMT1 and AU565 cells were acquired from ProCell (Wuhan, China). The MCF10A cells were cultured in a specific medium (ProCell, CM-0525). SKBR3, HEK293T, JIMT1, and AU565 cells were cultured in Dulbecco's modified Eagle medium (ProCell, PM153312) with 10% fetal bovine serum (ProCell, 164210) at 37 °C with 5% CO₂, and MDA-MB-361 cells were cultured in Leibovitz's L-15 (ProCell, PM151010) medium with 20% fetal bovine serum (ProCell, 164210) at 37 °C with 100% air. All cells were authenticated by STR and tested for mycoplasma. Earle's balanced salt solution (EBSS) was customized by ProCell (PB180337).

For stable cell line generation, lentiviral vectors encoding specific shRNAs (Supplementary Table) or the TM9SF1 coding sequence (GeneChem) were produced in HEK293T cells. Target cells were transduced and selected with puromycin for 7–10 days. Knockdown or overexpression efficiency was validated by real-time quantitative PCR (RT-qPCR) and western blotting (WB). S1

Total RNA was extracted using TRIzol reagent (Invitrogen, 10296-010). First-strand cDNA was synthesized using a reverse transcription kit (ABM, G592). RT-qPCR was performed using the SYBR Green Mix (Vazyme, Q111) on a real-time PCR system (Thermo Fisher Scientific, 4351107). GAPDH served as an endogenous control. Primer sequences are presented in Supplementary Table. Relative gene expression was calculated using the ΔΔCt method. S2

Cells were lysed in RIPA buffer (Beyotime, P0013B) with protease (Beyotime, P1005) and phosphatase inhibitors (Beyotime, P1082). Equal amounts of protein (30 µg) were resolved using sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred onto a polyvinylidene fluoride membrane. Membranes were blocked with 5% non-fat milk and incubated with primary antibodies at 4 °C overnight, followed by incubation with horseradish peroxidase-conjugated-tagged secondary antibodies. ECL (Biosharp, BL520B) was used to detect protein bands. A list of the antibodies used is provided in Supplementary Table. Unedited WB images are shown in Supplementary Information. S3

Cell proliferation was assessed using the cell counting kit 8 (CCK8) assay, colony formation assay, and 5-ethynyl-2'-deoxyuridine (EdU) incorporation assay. For CCK8, cells were seeded at 3000 cells/well in 96-well plates. Cells were assayed using 10 µL of CCK8 (MCE, HY-K0301) for 2 h at 37 °C. The absorbance was measured at 450 nm. For colony formation, 1000 cells/well were seeded in 6-well plates and cultured for 14-20 days and followed by staining with a crystal violet solution (Beyotime, C0121) for 15 min. For EdU assays, cells were incubated with 10 µM EdU (Beyotime, C0078S) for 2 h before detection. EdU incorporation was quantified as a percentage of total Hoechst-stained cells. All assays were performed according to the manufacturer's instructions.

For visualization of lysosomes and LDs, cells were co-stained with LysoTracker Red (Beyotime, C1046) (1:10000) and BODIPY 493/503 (Beyotime, C2053S) (1×) for 30 min at room temperature. For LD quantification, cells were stained with Nile Red (1:1000) (Beyotime, C2051S) for 15 min in the dark or Oil Red O solution (Beyotime, C0158S) for 20 min at room temperature. For autophagosome detection, cells were stained with 0.1 μM Monodansylcadaverine (MDC)- propidium iodide (PI) (Beyotime, C3019S) for 30 min at 37 °C. Images were acquired on a laser scanning confocal or fluorescence microscope and analyzed using ImageJ.

Paraffin-embedded tumor sections were deparaffinized, subjected to heat-induced antigen retrieval in EDTA buffer (MXB, MVS-0099), and blocked with 5% bovine serum albumin for 60 min. Sections were incubated with primary antibody at 4 °C overnight, followed by a horseradish peroxidase-conjugated secondary antibody and visualization with a DAB (MXB, DAB-0031) substrate kit. Nuclei were counterstained with hematoxylin. The expression levels of the proteins of interest were assessed by quantifying the intensity and distribution of staining within tumor sections.

The sample size (n = 5 per group) aligns with widely accepted standards for preliminary xenograft studies, ensuring statistical feasibility while adhering to the 3 R principles (reduction, refinement, and replacement). The allocation of animals to experimental groups was randomized, and the investigators were not blinded to group allocation during the experiment or outcome assessment. Cells (5 × 10⁶) with genetic modification were suspended in 100 µL PBS and randomly injected subcutaneously into the mammary fat pad of five-week-old BALB/c female nude mice (GemPharmatech). Tumor volumes were measured every 3–4 days using the formula: volume = (length × width²)/2. At the end of the study, tumors were excised and weighed. All animal experiments were approved by the Ethics Committee of Fujian Medical University (IACUC FJMU 2024-0145).

Cells were fixed with 2.5% glutaraldehyde at 4 °C for 2 h, followed by 1% osmium tetroxide post-fixation for 1 h. Cells were dehydrated using a graded ethanol series, embedded in epoxy resin, and then cut into 50 nm sections. These sections were stained with uranyl acetate and lead citrate for 15 min. The sections were examined using TEM to identify autophagosomes based on their characteristic double-membrane structures.

Cells were placed in lysis buffer supplemented with protease inhibitors on ice for 30 min. The lysate was centrifuged (14,000×g, 15 min, 4 °C), and then pre-cleared with Protein A/G magnetic beads (MCE, #HY-K0202) for 1 h at 4 °C. Subsequently, 1–5 µg of target-specific antibody or isotype-matched control IgG was added and incubated overnight at 4 °C, followed by incubation with Protein A/G beads for 2 h. The beads were pelleted, washed with buffer, and eluted with 2× sodium dodecyl sulfate-polyacrylamide gel electrophoresis loading buffer at room temperature. Protein complexes were analyzed by WB.

Lipid extracts were analyzed using a UPLC system coupled with tandem mass spectrometry [30]. Chromatographic separation was performed using a Thermo Accucore C30 column with a gradient elution. Lipid quantification was performed using the Multiple Reaction Monitoring mode of a triple quadrupole mass spectrometer.

Cells were cultured at 3 × 10⁴ per well in XF24 plates and then washed with Seahorse XF medium (Agilent, 103672-100) and equilibrated in a CO₂-free incubator for 60 min. The oxygen consumption rate was determined using a Seahorse XFe24 Analyzer, with etomoxir added, to evaluate the impact of FAO on respiration [31].

Cells were lysed in ATP extraction buffer on ice for 10–15 min, followed by centrifugation at 12,000×g for 10 min at 4 °C. ATP levels were quantified using an ATP assay kit (Solarbio, BC0305).

An assay kit (Solarbio, BC0595) was used to extract FFA. Absorbance was measured at 550 nm, and a standard curve was used to determine the FFA content in the samples.

For quantitative proteomics, 100 µg of protein lysate from each group was reduced with 5 mM DTT, alkylated with 11 mM iodoacetamide, and digested overnight with trypsin (Promega). The resulting peptides were desalted using C18 resin (Millipore) and separated on a Vanquish Neo UHPLC system. Data-independent acquisition mass spectrometry was performed on an Orbitrap Astral mass spectrometer (Thermo Fisher Scientific). Raw files were analyzed using DIA-NN v1.8.1 against the UniProt human database (1% false discovery rate). Proteins with a fold change ≥ 1.5 and a p-value < 0.05 were considered differentially expressed and were subjected to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis.

Data are presented as mean ± standard error of the mean (SEM). Unless otherwise specified, all experiments were repeated three times independently. Statistical comparisons between two groups were performed using a two-tailed Student's t-test or Mann–Whitney U-test. Comparisons among multiple groups were performed using one-way ANOVA followed by an appropriate post-hoc test. A p-value < 0.05 was considered statistically significant (^*p < 0.05, ^**p < 0.01, ^***p < 0.001).

Consistent with this hypothesis, RT-qPCR and WB assays demonstrated significantly higher expression of core autophagy markers, such as Beclin1 (BECN1) and microtubule-associated protein 1 A/1B-light chain 3B (LC3B), in HER2 + BC cells compared to the non-tumorigenic mammary epithelial cells MCF10A (Fig. 1D–E and Supplementary Fig. S1A). Furthermore, the mRNA expression of LIPA [32] and other key enzymes in lipid degradation and β-oxidation (CPT1A [33], CPT1B [34], and ACSL3 [35]) was markedly upregulated in HER2 + BC cells (Fig. 1F). Fluorescence co-localization assays confirmed increased interaction between lysosomes and LDs in HER2 + BC cells vs controls, indicative of heightened lipophagic flux (Fig. 1G, H). Consequently, starvation-induced LD clearance was accelerated in HER2 + BC cells, signifying rapid LD catabolism (Fig. 1I, J).

To ascertain the HER2-dependency of this phenotype, we examined the effects of HER2 inhibition with trastuzumab. In trastuzumab-sensitive SKBR3 cells, LC3B-II/I ratios exhibited a biphasic response, while lipid catabolic genes were sustainably upregulated. In contrast, trastuzumab-resistant JIMT1 cells showed a progressive increase in LC3B-II/I ratios but only a transient induction of lipid catabolic genes (Supplementary Fig.). These cell-type-specific patterns suggest that HER2 signaling dynamically coordinates lipophagic responses to metabolic stress. S2A–B

To validate these findings clinically, immunohistochemistry analysis of our patient cohort confirmed that TM9SF1 expression in HER2 + BC tissues was significantly higher than that in adjacent normal tissues (tumor = 78 vs normal = 63) (Fig. 2E–G). Receiver operating characteristic curve analysis demonstrated that TM9SF1 is a strong diagnostic biomarker for HER2 + BC, with an area under the curve of 0.881 (Fig. 2H). High TM9SF1 expression was also associated with high-risk clinical features, including younger age and high Ki67 indices (Fig. 2I). To explore the prognostic relevance of TM9SF1, we analyzed its association with overall survival in two independent publicly available datasets. In both the GSE1456 (n = 23) and GSE45255 (n = 31) cohorts, Kaplan-Meier analysis revealed a consistent trend where high TM9SF1 expression was associated with poorer patient outcomes (Fig. 2J). Consistent with its role as a stress-responsive mediator, TM9SF1 expression showed biphasic dynamics following HER2 inhibition (Supplementary Fig. S3B). Taken together, these data identify TM9SF1 as a clinically relevant, lipophagy-associated prognostic biomarker in HER2 + BC.

Concurrently, TM9SF1 knockdown suppressed the expression of key lipolytic genes, including LIPA, CPT1A, CPT1B, and ACSL3 (Fig. 4F). This transcriptional suppression manifested phenotypically as a significant accumulation of LDs, as observed by Oil Red O (Fig. 4G) and Nile Red staining (Fig. 4H). Conversely, TM9SF1 overexpression promoted LD clearance. Importantly, the anti-proliferative effect of TM9SF1 knockdown was rescued by the autophagy inducer rapamycin, while the pro-proliferative effect of TM9SF1 overexpression was abrogated by BafA1 (Fig. 4I–J). Collectively, these results demonstrate that TM9SF1 enhances autophagic flux and promotes the lysosomal degradation of LDs, while also facilitating HER2 + BC cell proliferation through enhanced autophagy.

To elucidate the underlying mechanism by which TM9SF1 regulates lipophagy, we performed quantitative proteomics on SKBR3-shTM9SF1 cells. Functional enrichment analysis identified a significant enrichment of differential expression proteins in the AMPK signaling pathway (Fig. 5F). The AMPK/ULK1 signaling pathway plays a critical role in autophagy as a key upstream regulator. Based on these findings, we hypothesized that TM9SF1 may influence lipophagy levels through the AMPK/ULK1 pathway. Immunohistochemical analysis of clinical samples revealed significantly higher levels of phosphorylated AMPK (p-AMPK) and ULK1 (p-ULK1) in HER2 + BC tissues compared to normal tissues (n = 20) (Fig. 5G). TCGA-BRCA transcriptomic data further showed a positive correlation between TM9SF1 and the expression of AMPK subunits (PRKAA1, PRKAA2) and ULK1 expression (Fig. 5H). Crucially, WB analysis confirmed this regulatory link: knockdown of TM9SF1 attenuated the phosphorylation of AMPK at Thr172 and consequently suppressed the phosphorylation of its downstream target ULK1 at Ser555 (Fig. 5I and Supplementary Fig. S1E). This regulatory cascade was recapitulated in tumor xenografts (Fig. 5J). Treatment with the AMPK activator AICAR [40] rescued the lipophagic defect and reversed LD accumulation in shTM9SF1 cells (Fig. 5K, L). Although co-immunoprecipitation assays did not detect a direct interaction between TM9SF1 and AMPK (Supplementary Fig. S4A), these findings strongly suggest that TM9SF1 sustains lipophagy by maintaining the activity of the AMPK-ULK1 signaling axis.

As lipophagy is critical for maintaining cellular energy balance, we next assessed its bioenergetic impact. A Seahorse metabolic flux analysis revealed that TM9SF1 promotes fatty acid β-oxidation to generate energy flux, a process that was inhibited by BafA1 (Fig. 6D–F). Consistent with this, ATP assays demonstrated that TM9SF1 knockdown led to a significant decrease in ATP production in both cell types (Fig. 6G) and in vivo mouse tumor tissues (Fig. 6H), confirming that TM9SF1-driven lipophagy is essential for energy generation.