Identification of Bruceine A as a novel HSP90AB1 inhibitor for suppressing hepatocellular carcinoma growth.

Introduction
Hepatocellular carcinoma (HCC) represents one of the most common forms of liver cancer, accounting for 80–90 % of primary liver cancer cases worldwide and resulting in approximately 1 million deaths annually [1]. Despite progress in localized treatments such as surgical resection and liver transplantation, these options remain viable only for a small proportion of patients diagnosed at early stages [2]. Locoregional therapies, including transarterial approaches, are vital for managing the majority of HCC cases but are often limited by the degree of liver dysfunction [3]. Most patients present with advanced disease, where systemic therapies, despite improvements, often fail due to late-stage diagnosis and tumor resistance [4]. Consequently, the prognosis for HCC remains poor, with a five-year survival rate of just 18 % [5], emphasizing the urgent need for more effective and targeted treatments.
Natural products (NPs) are a key source of innovative drug development, particularly in the fields of oncology and antimicrobial therapy [6], due to their complex structures and broad biological activities [7]. Currently, over 50 % of FDA-approved small-molecule drugs are derived from NPs [8]. Recent clinical trials for HCC treatment have incorporated several NPs, including polyphenols such as icaritin, resveratrol, and silybin, saponins like ginsenoside Rg3, alkaloids such as irinotecan, and inorganic compounds like arsenic trioxide [9]. Bruceine A (BRA), a natural compound derived from the seeds of Brucea javanica, has garnered attention for its promising medicinal properties [10]. Extensive research demonstrates BRA's efficacy against various cancers, including pancreatic cancer, triple-negative breast cancer, and colon cancer [[11], [12], [13], [14]]. However, studies on BRA in HCC remain scarce, and the precise mechanisms of its anticancer effects in this context are not well understood, highlighting the need for further investigation.
Identifying the molecular targets of bioactive NPs is crucial for elucidating their mechanisms of action, optimizing current therapies, and facilitating the development of novel therapeutic agents [15]. The absence of identified targets for BRA presents a significant challenge in advancing both research and clinical applications. Therefore, uncovering BRA's molecular targets is essential for understanding its therapeutic potential and advancing drug development efforts.
This study demonstrates that BRA exhibits potent anti-HCC activity both in vitro and in vivo, utilizing preclinical patient-derived organoid (PDO) and xenograft (PDX) models. It was further shown that BRA exerts substantial antitumor effects by directly targeting HSP90AB1, a chaperone protein crucial for HCC cell survival. These findings highlight the therapeutic promise of BRA as a novel HSP90AB1 inhibitor and suggest potential strategies for targeting HSP90AB1 in HCC treatment.

Materials and methods
Detailed Materials and Methods were presented in the Supporting Information.

Chemicals and reagents
Bruceine A (BRA, purity ≥ 98 %) was obtained from Shanghai Yuanye Bio-Technology Co., Ltd. Fetal bovine serum (FBS) and Dulbecco's Minimal Essential Medium (DMEM) were sourced from Gibco (NY, USA). Phosphate-buffered saline (PBS), RIPA lysis buffer, penicillin–streptomycin, Super ECL Detection Reagent, the Annexin V-FITC Apoptosis Detection Kit, and the Coomassie Blue Fast Staining Kit were acquired from Beyotime Biotechnology (Jiangsu, China). Primary antibodies against EGFR, PIK3CG, HSP90AB1, PCNA, p21, c-Myc, Bcl-2, cleaved PARP1, cleaved Caspase-3, and Ki-67 were also sourced from Beyotime Biotechnology. The Cell Counting Kit-8 (CCK-8) was procured from TargetMol (Boston, USA), while multicolor protein markers were obtained from Invitrogen (26617, CA, USA) or Epizyme (WJ102, Shanghai, China). Polyvinylidene fluoride (PVDF) membranes were acquired from Merck Millipore (IPVH00010, Darmstadt, Germany). Recombinant human Hsp90 beta protein (Active) was obtained from Abcam (ab80033, MA, USA).

Ethics statement
All animal experiments were conducted in accordance with the Guide for the Care and Use of Laboratory Animals and were approved by the Animal Ethical and Welfare Committee of The Second Xiangya Hospital, Central South University (Approval no. 20240600). The use of clinical HCC specimens adhered to the principles set forth in the Declaration of Helsinki and received approval from the Ethics Committee of The Second Xiangya Hospital, Central South University (Approval no. LYF20240243). Detailed clinical information for the patients is provided in Table S1 (Supporting Information). Tumor specimens for PDX and PDO models were obtained from six patients with HCC during surgical resections at Xiangya Hospital, with approval from the Ethics Committee of Xiangya Hospital, Central South University (Approval no. 202103124). Informed consent or its equivalent was obtained from all patients for inclusion in the study.

Target protein identification for BRA
Target protein identification for BRA was carried out using modified techniques from previous studies [16,17]. HCC-LM3 cells were harvested and lysed through brief sonication in lysis buffer consisting of 50 mM Tris⋅HCl (pH 7.5), 100 mM NaCl, 2 mM MgCl2, 2 mM CaCl2, 0.2 % Nonidet P-40, 0.8 % Triton X-100, 1 mM β-mercaptoethanol, and a 1 × protease inhibitor cocktail. Following centrifugation at 18,000g at 4 °C for 10 min, the supernatant was collected and rotated with 20 μL of prewashed streptavidin agarose (#S1638; Sigma) at 4 °C for 2 h. Protein concentrations were adjusted to 5  mg/mL using the BCA assay. Biotin (2 μM), biotin-BRA (2 μM), or biotin-BRA (2 μM) plus BRA (10 μM) were added, and each tube was incubated at 4 °C for 12 h, with 10 μL of supernatant reserved as the input control. Following incubation, 40 μL of prewashed streptavidin agarose beads was added and rotated for 4 h at 4 °C. The beads were washed six times with 1 mL of PBS containing 0.05 % Tween 20, and bound proteins were eluted by boiling in the SDS sample buffer. For target identification, proteins were separated by SDS-PAGE (4–10 % gel) and stained with Coomassie Brilliant Blue. Gel lanes were excised and analyzed via protein mass spectrometry, as previously described [18].

Cellular thermal shift assay (CETSA)
CETSA was performed as outlined in prior studies [19]. HCC-LM3 cells were incubated with either solvent control (DMSO) or BRA (50 nM) for 2 h, washed with PBS, and subsequently lysed. Proteins in PBS containing protease inhibitors were divided into seven aliquots, heated at 40, 43, 46, 49, 52, 55, and 58 °C for 30 s, followed by freezing in liquid nitrogen for 3 min. Each sample underwent three freeze–thaw cycles before being centrifuged at 12,000 rpm for 15 min. The supernatants were combined with a loading buffer, and HSP90AB1 protein levels were assessed via Western blot analysis.

Surface plasmon resonance (SPR) analysis
SPR experiments were conducted using a Biacore T200 instrument (GE Healthcare) with CM5 sensor chips at 37 °C [20]. Purified recombinant human HSP90AB1 protein (50 μg·mL−1) was injected into the Sensor chip for immobilization. Different concentrations of BRA in a running buffer (1 × PBS, 5 % DMSO) were then passed over the chip to generate response signals. The association and dissociation rate constants were calculated using the GE Biacore T2000 Evaluation software.

Microscale thermophoresis (MST) assay
The MST assay was performed using a Monolith NT.115 (NanoTemper), and the equilibrium dissociation constant (Kd) values for recombinant human HSP90AB1 with BRA were calculated using NanoTemper analysis software, as previously described [21].

Malachite green assay
The ATPase enzymatic activity of HSP90AB1 was measured using a malachite green-phosphate assay, following established methods [22]. Briefly, 10 μL of recombinant human HSP90AB1 protein stock (0.1 mg/mL) was incubated with varying concentrations of BRA (100 nM to 0.5 nM) and 2.5 mM ATP in 25 μL of assay buffer for 3 h at 37 °C. Following incubation, 80 μL of malachite green reagent (10009325, Cayman) and 10 μL of 34 % sodium citrate were added to each well. The plate was shaken and allowed to stand at room temperature for approximately 15 min. Absorbance at 620 nm was measured using a SpectraMax iD5 microplate reader (Molecular Devices, USA).

TMT-based quantitative proteomics
TMT-based quantitative proteomics was used to investigate the differential protein expression profiles of SK-HEP-1 cells with or without HSP90AB1 knockdown, as previously outlined [23]. Briefly, cells were harvested, total proteins were extracted via filter-aided sample preparation, followed by tagging the peptides with TMT reagent (Thermo Fisher Scientific, Waltham, MA, USA). The TMT-labeled peptides were subsequently fractionated using reversed-phase chromatography with the Agilent 1260 Infinity II HPLC and analyzed through liquid chromatography-mass spectrometry (LC-MS/MS). All LC-MS studies were conducted using a Q Exactive Plus mass spectrometer (Thermo Fisher Scientific, USA), and the original data were processed using MaxQuant software.

PDO tumor models
Human HCC tumor tissues were collected for the establishment of human organoids, as described in earlier work [24,25]. Briefly, tumor tissues were sectioned, washed with PBS, and then subjected to digestion using an Accuroid Tissue Dissociation Solution (Accurate International Biotechnology, Guangzhou, China). Subsequent to a filtration process through a 100 μm filter, the mixture underwent centrifugation and erythrocyte lysis. The resultant cells were then resuspended in Advanced DMEM/F12 medium (Gibco) and mixed with an equal volume of Matrigel. Subsequently, Accuroid Organoid Culture Medium-Liver Cancer (Accurate International Biotechnology) was applied to overlay the Matrigel-cell suspension, and the cells were incubated in a CO2 incubator maintained at 37 °C. The culture medium was refreshed twice daily, with the organoids being maintained for a duration of 7 to 14 days before being subcultured for further experimental procedures.

Tumor xenograft experiments
Five-week-old male BALB/c nude mice (Beijing IDMO Co., Ltd., Beijing, China), weighing 18–22 g, were acclimatized for 7 days in an SPF-grade animal facility before experimentation. A 0.1 mL cell suspension containing 2 × 106 HCC-LM3 cells was subcutaneously injected into the disinfected right axillary region. When the tumor volume reached approximately 50 mm3, the mice were randomly divided into two groups (n = 6 per group): control group (vehicle) and BRA 50 mg/kg group, see supporting materials for dispensing methods. The mice were administered the corresponding treatments once daily via oral gavage using an 8-gauge gavage needle attached to a 1 mL syringe for two weeks. Body weight and tumor growth were closely monitored during treatment. To determine the tumor volume using an external caliper, the greatest longitudinal diameter (a) and the greatest transverse diameter (b) of the tumor were measured. The tumor volume calculated from these caliper measurements was derived using the formula: volume (mm3) = a × b2/2. At the end of the treatment period, tumor tissues were harvested for subsequent hematoxylin-eosin (H&E) staining and immunohistochemical (IHC) analysis, as detailed in prior research [26].

PDX tumor models
Male BALB/c nude mice (Beijing IDMO Co., Ltd., Beijing, China) were employed for PDX models, following previously established protocols [27]. Fresh liver tumor tissues were sectioned into appropriately sized fragments (∼3 mm) and transplanted into the right subcutaneous area of the nude mice. Upon reaching a tumor volume of ∼ 50 mm3, mice were subjected to daily oral gavage of control or BRA at a dose of 50 mg/kg. Mice were euthanized when tumor volume reached 1500 mm3. Xenograft tissue was preserved by either snap freezing at −80 °C or formalin fixation.

Histological analysis
Tumor specimens and PDOs underwent hematoxylin-eosin (H&E) staining, and immunohistochemical (IHC) staining was performed as previously described [28]. For HE staining, the paraffin sections were stained with hematoxylin for 3 min, rinsed, stained with eosin for 5 min, rinsed again, cleared, and mounted for microscopy. For IHC staining, the paraffin sections underwent deparaffinization, rehydration, blocking, and overnight immunostaining at 4 °C with primary antibodies against HSP90AB1, Ki-67, cleaved caspase-3, PIK3CG, EGFR, and KDM5C. Slides were subsequently incubated with a secondary antibody for 60 min and visualized using light microscopy (BX43, Olympus, Japan). Histological images of tissue sections were captured using an Olympus BX43 light microscope (Japan).

Western blot analysis
Western blot analysis was conducted as previously outlined [29]. Briefly, cellular samples were subjected to homogenization in ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer, which was enriched with protease and phosphatase inhibitors. The resulting lysates underwent separation via 8–12 % SDS-PAGE and were subsequently transferred to polyvinylidene fluoride (PVDF) membranes (IPVH00010; Millipore, USA). These membranes were then blocked using a 5 % nonfat milk solution and incubated with primary antibodies overnight at 4 °C. Following this, secondary antibodies were introduced and allowed to incubate for one hour at ambient temperature. Finally, the membranes were processed with an enhanced chemiluminescence (ECL) kit (Beyotime, China) and visualized using a Tanon5200 imaging system (Shanghai, China).

Co-immunoprecipitation (Co-IP)
Co-IP of endogenous HSP90AB1 from HCC-LM3 and SK-HEP-1 cell lysates was performed according to established protocols [30]. Cellular HSP90AB1 was immunoprecipitated using a mouse anti-HSP90AB1 monoclonal antibody, with mouse IgG serving as a negative control. Immune complexes were captured using Protein G Dynabeads (Beyotime Biotechnology), and immunoprecipitates were analyzed by SDS-PAGE followed by immunoblotting with rabbit antibodies targeting HSP90AB1, EGFR, PIK3CG, and KDM5C. Immunoreactive bands were visualized using enhanced chemiluminescence.

Statistical analysis
Data from cell and animal experiments were analyzed using two-tailed Student's t-test or one-way ANOVA with GraphPad Prism 10.0. Results are presented as means ± SD from a minimum of three independent experiments. Statistical significance was indicated as *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001; n.s., not significant.

Results

BRA remarkably inhibited the proliferation and growth of HCC cells both in vitro and in vivo
To investigate the anti-proliferative effects of BRA on HCC in vitro, cell viability, colony formation, and EdU incorporation assays were conducted. In the cell viability assay, seven human HCC cell lines (HepG2, JHH-7, Huh-7, SK-HEP-1, Li-7, HCC-LM3, and MHCC-97H) were treated with specified concentrations of BRA for 72 h. The results demonstrated that BRA significantly inhibited the viability of HCC cell lines in a dose-dependent manner (Fig. 1A), with half-maximum inhibitory concentration (IC50) values ranging from 16.57 to 86.75 nmol/L (Supporting InformationTable S1). In contrast, normal human hepatocytes (THLE-2) exhibited significantly higher IC50 values compared to the HCC cell lines. Based on these findings, SK-HEP-1, Li-7, and HCC-LM3 cells, which showed a strong inhibitory response to BRA, were selected for further analysis. As expected, BRA inhibited cell viability in SK-HEP-1, Li-7, and HCC-LM3 cells in a time-dependent manner (Fig. 1B).
In the colony formation assay, BRA treatment significantly reduced both the number and size of colonies formed by SK-HEP-1, Li-7, and HCC-LM3 cells, compared to the control group, indicating concentration-dependent inhibition of clonogenic potential (Fig. 1C, D). The EdU incorporation assay further assessed the proliferation of these cells post-BRA treatment. The proportion of EdU-positive cells in the BRA treatment group decreased significantly in a dose-dependent manner, compared to the control group (Fig. 1E, F).
Apoptosis levels in BRA-treated SK-HEP-1, Li-7, and HCC-LM3 cells were subsequently evaluated using flow cytometry, revealing a significant increase in the proportion of apoptotic cells (Figs. 1G, H). Flow cytometric analysis of the cell cycle indicated that BRA increased the proportion of cells in the S phase (Supporting Information
Fig. S1A). To corroborate these findings, Western blot analysis for proliferation, cell cycle, and apoptosis-related proteins was performed. BRA treatment increased the expression of p21, cleaved PARP, and cleaved CASP3 while decreasing the levels of PCNA, c-MYC, Cyclin A, and Bcl-2 in HCC cells (Fig. 1I).
We conducted a preliminary evaluation on the acute toxicity of BRA in BALB/c mice. After the intervention, histopathological examination of liver, spleen, and kidney tissues, combined with serum biochemical analysis, revealed that within the 50 mg/kg dose range, there were no significant changes in body weight, tissue morphology, or hepatic and renal function (Supporting Information
Fig. S2). To evaluate the in vivo antitumor activity of BRA, a xenograft mouse model was established by subcutaneously injecting HCC-LM3 cells into nude mice (Fig. 1J). Significant inhibition of HCC-LM3 xenograft tumor growth was observed with BRA treatment, evidenced by reduced tumor volumes and weights following oral gavage of BRA at doses of 50 mg/kg daily for 14 days (Fig. 1K–M). Throughout the experiment, body weights were monitored, showing no significant differences between the control and 50 mg/kg groups (Fig. 1N).
Further pathological examinations, including H&E staining and Ki-67 and cleaved CASP3 IHC staining, demonstrated that BRA treatment significantly decreased the number of proliferating cells and induced apoptosis in tumor tissues of HCC-LM3 xenograft models (Fig. 1O and Supporting Information
Fig. S3).
Collectively, these findings indicate that BRA markedly inhibits HCC progression in vivo.

BRA strongly suppressed HCC tumor growth in PDOs and PDX models
To evaluate the anti-HCC efficacy of BRA, PDO and PDX models were employed, serving as optimal preclinical platforms for assessing treatment responses in cancer drug discovery (Fig. 2A). Tumor tissues from six patients with HCC were initially used to generate PDO models, which were then utilized to assess the anti-proliferative effects of BRA. Results demonstrated that BRA significantly decreased the viability of HCC PDOs, with IC50 values ranging from 47.18 to 145.8 nmol/L (Fig. 2B and Supporting Information
Table S2, 3). Notably, BRA exerted the most dose-dependent inhibition of viability in PDO#HCC1016.
H&E staining further revealed significant tumor cell death in HCC1016 following BRA treatment, in a dose-dependent manner. IHC analysis confirmed that BRA reduced Ki-67 expression while upregulating cleaved CASP3, indicative of substantial antitumor activity in the PDO models (Fig. 2C).
The therapeutic potential of BRA against HCC was also assessed in vivo using a clinically relevant PDX model. Mice bearing PDX tumors were treated with either vehicle control or BRA (50 mg/kg/day), and tumor growth was monitored over 8 days. BRA administration resulted in marked tumor growth inhibition, accompanied by significant reductions in both tumor volumes and weights (Fig. 2D–F). Histopathological examination confirmed extensive tumor cell death following BRA treatment. IHC analysis revealed a significant reduction in Ki-67 expression and a notable increase in Cleaved CASP3 levels (Fig. 2G and Supporting Information
Fig. S4).
In summary, these findings demonstrate that BRA exerts potent pharmacological inhibition of HCC in both PDO and PDX preclinical models.

HSP90AB1 is the direct target of BRA in HCC
The identification of molecular targets is a pivotal step in the development of NP-based therapeutics [31]. This study aimed to identify the molecular targets of BRA responsible for its growth-inhibitory effects, as illustrated in Fig. 3A.
A biotinylated derivative of BRA (BRA-Biotin, Fig. 3B) was synthesized by labeling BRA with biotin, and its biological activity was assessed. Viability assays showed that BRA-Biotin maintained its biological activity, albeit to a lesser extent (Supporting Information
Fig. S5A). Subsequently, HCC-LM3 cell lysates were incubated with BRA-Biotin to capture potential target proteins, which were separated by SDS-PAGE and visualized by Coomassie brilliant blue staining. A prominent protein band near 100 kDa was significantly more intense in the BRA-Biotin group compared to the control biotin group (Fig. 3C). This band exhibited diminished intensity when excess BRA was used to compete for binding (Fig. 3C). The gel band was excised and identified via liquid chromatography-mass spectrometry/mass spectrometry (LC-MS/MS), identifying HSP90AB1 as a potential BRA target (Supporting Information
Fig. S5B). Western blot analysis further confirmed the presence of this protein using an anti-HSP90AB1 antibody (Fig. 3C). BRA treatment had only minor effects on the protein levels of HSP90AB1 in HCC cells (Fig. 1I). We further examined the mRNA expression of HSP90AB1 in Li-7 and HCC-LM3 cells after BRA treatment. Evidently, BRA treatment exerted no significant impact on the mRNA levels of HSP90AB1 (Supporting Information
Fig. S1B). In vitro enzymatic assays showed that BRA treatment significantly inhibited the ATPase activity of recombinant human HSP90AB1 in a dose-dependent manner, suggesting that HSP90AB1 is a direct target of BRA (Fig. 3D). Notably, this inhibitory effect was observed without altering the protein levels of recombinant human HSP90AB1 (Supporting Information
Fig. S6). Similarly, Biotin-BRA also inhibited HSP90 ATPase activity in a dose-dependent manner (Supporting Information
Fig. S7).
To confirm HSP90AB1 as a direct BRA target, MST, SPR, and CETSA assays were performed to evaluate BRA's binding affinity to HSP90AB1 in vitro. The MST experiment demonstrated direct binding of BRA to HSP90AB1 with a Kd of 88.9 nM (Fig. 3E). The SPR analysis showed positive affinity interactions between BRA (Kd = 66.3 nM) and recombinant human HSP90AB1 (Fig. 3F). Additionally, the results of CETSA proved that the thermal stability of HSP90AB1 was enhanced after BRA treatment, which was confirmed by the shift of the melting curve and the decrease of protein degradation at elevated temperature. (Fig. 3G).
To assess whether BRA binding to HSP90AB1 induces anticancer effects, lentiviral short hairpin RNA (lenti-shRNA) vectors were used to knock down HSP90AB1 in SK-HEP-1, Li-7, and HCC-LM3 cells. Western blot analysis confirmed knockdown efficiency (Supporting Information
Fig. S9). Cell viability assays revealed that HSP90AB1 knockdown significantly attenuated BRA's inhibitory effect on cell viability compared to the control scrambled group (Fig. 3H).
In summary, these findings suggest that HSP90AB1 is the primary cellular target of BRA.

Key residues in the interaction of BRA and HSP90AB1
Our findings confirm that HSP90AB1 is the direct target of BRA in the treatment of HCC. Consequently, further investigation into the binding modes and interaction sites between BRA and HSP90AB1 was undertaken to elucidate the regulatory mechanisms involved.
Molecular docking simulations using AutoDock Vina were first performed to assess the BRA-HSP90AB1 interaction based on the binding energy. The affinity of BRA and HSP90AB1 was −7.978 kcal/mol, indicating a solid binding interaction between BRA and HSP90AB1. Then, the key amino acids involved in the BRA-HSP90AB1 interaction were predicted. The analysis revealed that five amino acid residues—THR-110, SER-108, ASN-46, PHE-133, and GLY-132—on the B chain of the HSP90AB1 protein (ATPase domain) form conventional hydrogen bonds with BRA. Additionally, MET-93, TYR-134, and PHE-129 likely contribute to hydrophobic interactions (π-alkyl interactions) with BRA (Fig. 4A, B).
Molecular dynamics (MD) simulations were performed to further evaluate the stability of the docked BRA-HSP90AB1 complex and its interactions. Analysis of the 100 ns simulation trajectory, key parameters including Rg, RMSD, H-bonds, and RMSF confirmed the binding stability between BRA and HSP90AB1 (Supporting Information
Fig.S8A-D). Furthermore, we performed principal component analysis (PCA) to analyze Gibbs free energy landscapes (FELs) of the BRA-HSP90AB1 complex to comprehensively characterize its conformational dynamics. As shown in Supporting Information
Fig.S8E and F, the Gibbs free energy landscape exhibited a single, well-defined energy cluster with minimal energy regions, indicating a stable and energetically favorable conformation for the complex.
To validate the importance of these residues in BRA binding, mutations were introduced at these sites (Fig. 4C). Both wild-type and mutant HSP90AB1 constructs were overexpressed in HCC-LM3 cells. As shown in Fig. 4D, the expression levels of both wild-type and mutant HSP90AB1 were comparable and higher than endogenous levels. A BRA-biotin pull-down assay was conducted to assess the impact of these mutations on BRA binding. Notably, differential binding affinities were observed for the HSP90AB1 mutants. The SER-108 mutant (S108A) exhibited the weakest binding to BRA compared to the wild-type and other mutants (Fig. 4D), highlighting SER-108 as a critical residue for BRA interaction.
To further investigate the role of SER-108 in mediating BRA’s anti-HCC effects, HSP90AB1-depleted HCC cells were transfected with shRNA-resistant pCMV3 plasmids expressing either wild-type (WT) HSP90AB1 or the S108A mutant. Western blot analysis confirmed successful transfection (Supporting Information
Fig. S10). CETSA analysis revealed that the S108A mutation significantly weakened the binding affinity of HSP90AB1 for BRA, resulting in minimal thermal shift changes (Fig. 4E). Additionally, a CCK-8 assay demonstrated a significantly higher IC50 for the HSP90AB1 S108A mutant compared to the wild-type, confirming that the SER-108 residue is a key binding site for BRA on HSP90AB1 (Fig. 4F).
Therefore, these findings suggest that SER-108 is the key amino acid in the interaction of BRA and HSP90AB1.

HSP90AB1 is upregulated in liver cancer tumors and predicts poor prognosis
To investigate the clinical relevance of HSP90AB1 in HCC, HSP90AB1 mRNA expression was analyzed using data from 369 HCC tissues, 160 normal tissues, and 50 matched pairs of HCC and adjacent normal tissues from the TCGA databases. The results revealed that HSP90AB1 mRNA expression was significantly higher in HCC samples compared to normal tissues (Fig. 5A). Additionally, RNA-seq data from the matched cohort indicated notable overexpression of HSP90AB1 in the majority of HCC tissues (Fig. 5B).
Kaplan-Meier survival analysis of TCGA-LIHC data demonstrated a significant correlation between high HSP90AB1 expression and poorer overall survival compared to low expression levels (P = 0.0035; Fig. 5C). This finding was further supported by spatial transcriptomic data from HCCDB v2.0, which showed H&E staining of HCC-4 and the spatial distribution of tissue clusters. Four tissue spot types were identified: tumor (red), normal (green), stromal (yellow), and immune (blue) (Fig.5D-F). Elevated HSP90AB1 mRNA expression was observed in HCC tumor regions relative to adjacent normal tissues (Fig. 5G, H).
IHC analysis of 20 pairs of HCC tumors and matched non-tumor liver tissues revealed marked overexpression of HSP90AB1 protein in tumor tissues compared to adjacent non-tumor tissues (Fig. 5I, J). Moreover, the CPTAC database confirmed higher HSP90AB1 protein levels in tumor tissues (Fig. 5K).
To explore the cellular expression profile of HSP90AB1 in HCC, single-cell sequencing data (GSE149614) from 21 samples of 10 patients were analyzed. Since HCC arises from hepatic epithelial cells, single cells were classified into distinct clusters, including epithelial cells, based on their genomic features (Fig. 5L). As expected, HSP90AB1 was highly expressed in both epithelial and T cells (Fig. 5M), with significant upregulation in tumors compared to normal tissues, predominantly in epithelial cells (Fig. 5N, O).
Collectively, these findings indicate that HSP90AB1 is overexpressed in liver cancer and is associated with poor prognosis.

HSP90AB1 is required for HCC cell survival both in vitro and in vivo
To evaluate the biological effects of HSP90AB1 upregulation on HCC progression, the impact of HSP90AB1 knockdown on cell proliferation was assessed using various in vitro assays. CCK-8 assays demonstrated that HSP90AB1 knockdown effectively inhibited HCC cell proliferation (Fig. 6A–C). Likewise, colony formation assays showed a significant reduction in colony formation by HSP90AB1-depleted HCC cells compared to control cells (Fig. 6D, E). EdU incorporation assays further confirmed that downregulation of HSP90AB1 resulted in a substantial decrease in the percentage of EdU-positive cells (Fig. 6F, G). Apoptosis assays revealed an increase in apoptosis in HCC cells following HSP90AB1 knockdown (Fig. 6H, I).
Western blot analysis indicated that HSP90AB1 knockdown significantly reduced the protein levels of PCNA, c-MYC, and Cyclin A while elevating levels of p21, Cleaved PARP, and Cleaved CASP3, and reducing Bcl-2 expression (Fig. 6J).
To further assess the impact of HSP90AB1 on HCC tumor growth, both parental HCC-LM3 cells and HSP90AB1-knockdown HCC-LM3 cells were implanted into nude mice for in vivo evaluation (Fig. 6K). After 31 days, analysis of tumor growth curves, sizes, and weights showed that HSP90AB1 knockdown effectively inhibited subcutaneous tumor growth (Fig. 6L–N). IHC staining revealed a significant reduction in cancer cell proliferation (Ki-67) and an increase in apoptosis (Cleaved CASP3) in HSP90AB1-knockdown tumors, consistent with the in vitro findings (Fig. 6O and Supporting Information
Fig. S11).
These results demonstrate that HSP90AB1 is crucial for the survival of HCC cells both in vitro and in vivo.

TMT-based proteomic analysis reveals candidate downstream targets of HSP90AB1
To identify the downstream proteins regulated by HSP90AB1, TMT-based quantitative proteomic analysis was performed to assess protein expression levels in HCC-LM3 cells transfected with either a negative control lentivirus or an HSP90AB1-targeting lentivirus (shHSP90AB1), as depicted in Fig. 7A. Statistically significant results, defined by an adjusted P-value ≤ 0.05 and a fold-change cutoff greater than 1.5 or less than 1/1.5, revealed 71 differentially expressed proteins (DEPs). Among these, 35 proteins were significantly upregulated, while 36 proteins were significantly downregulated following HSP90AB1 knockdown (Supporting Information
Table S4). A heatmap further emphasized the changes in the downregulated DEPs (Fig. 7B).
Gene Ontology (GO) analysis categorized the downregulated DEPs into three functional groups: cell component (CC), molecular function (MF), and biological process (BP). These proteins were involved in regulating several biological processes, including “autophagy,” “regulation of cell death,” “regulation of apoptotic processes,” and “regulation of programmed cell death” (Fig. 7C). KEGG pathway analysis of the downregulated DEPs identified their involvement in pathways related to “Inositol phosphate metabolism,” “Autophagy,” “Apoptosis,” and “PI3K-AKT signaling” (Fig. 7D).
Protein-protein interaction (PPI) network analysis of the downregulated DEPs was performed using the STRING database (version 11.0) (Fig. 7E). Among the identified proteins, PIK3CG, EGFR, and KDM5C were recognized as known HSP90 chaperone proteins, selected for further validation [[32], [33], [34]]. CO-IP assays confirmed direct interactions between HSP90AB1 and PIK3CG, EGFR, and KDM5C in both HCC-LM3 and SK-HEP-1 cells, supporting the TMT findings (Supporting Information
Fig. S12).
To assess whether the oncogenic functions of HSP90AB1 involve PIK3CG, EGFR, and KDM5C, protein expression levels were evaluated in HSP90AB1 knockdown SK-HEP-1, Li-7, and HCC-LM3 cells. Consistent with the TMT data, HSP90AB1 knockdown led to decreased expression of PIK3CG, EGFR, and KDM5C (Fig. 7F). Furthermore, BRA treatment also reduced the protein levels of these three proteins in a dose-dependent manner (Fig. 7G). Supporting this finding, BRA treatment promoted the degradation of PIK3CG, EGFR, and KDM5C in a cycloheximide (CHX) chase assay (Supporting Information
Fig. S13). IHC analysis further demonstrated that the expression of PIK3CG, EGFR, and KDM5C proteins was inhibited in vivo following either BRA treatment or HSP90AB1 knockdown (Supporting Information
Fig. S14). These results indicate that BRA may promote the protein degradation of PIK3CG, EGFR, and KDM5C by inhibiting HSP90AB1 activity in HCC.

The knockdown of PIK3CG, EGFR, and KDM5C inhibits HCC cell proliferation and promotes apoptosis in vitro
To further investigate the roles of PIK3CG, EGFR, and KDM5C in HCC cell proliferation, proliferation assays (CCK-8, colony formation, and EdU) were performed on SK-HEP-1 and HCC-LM3 cells following validation of knockdown efficiency by Western blot (Supporting Information
Fig. S15).
CCK-8 assay results demonstrated that PIK3CG knockdown significantly decreased HCC cell viability, with similar effects observed for EGFR and KDM5C knockdown (Fig. 8A, B). Colony formation assays revealed a substantial reduction in colony formation capacity after silencing PIK3CG, EGFR, and KDM5C (Fig. 8C). Additionally, the EdU proliferation assay indicated a marked reduction in HCC cell proliferation following knockdown (Fig. 8D). Flow cytometry analysis showed an increased proportion of Annexin V-FITC positive cells (indicative of apoptosis) in scrambled siRNA-treated SK-HEP-1 and HCC-LM3 cells compared to control cells (Fig. 8E). Furthermore, protein levels of PCNA, c-MYC, and Cyclin A were significantly reduced following knockdown of PIK3CG, EGFR, and KDM5C. The knockdown also resulted in lower levels of Bcl-2 and elevated levels of Cleaved PARP and Cleaved CASP3, suggesting enhanced apoptosis in HCC cells (Fig. 8F, G).
Collectively, these findings confirm that silencing PIK3CG, EGFR, and KDM5C effectively inhibits HCC cell proliferation and promotes apoptosis in vitro.

Discussion
Current therapeutic strategies for HCC are often limited by the advanced stage at which the disease is diagnosed, underscoring the need for continued research to identify novel therapeutic targets and improve patient outcomes [35]. This study provides the first evidence that BRA significantly inhibits HCC growth both in vitro and in vivo. In addition to conventional cell line models, PDO and PDX models were utilized to further validate BRA’s anticancer effects in inhibiting HCC progression. These results offer critical insights for future developments and clinical applications, as PDO and PDX models maintain the histopathological features of the original tumors, making them essential translational systems for predicting anticancer drug responses [36]. Thus, BRA may represent a promising therapeutic agent for human HCC, warranting further investigation.
Identifying protein targets (“target deconvolution”) and determining the binding affinity of drug targets (“target engagement”) are pivotal steps in the discovery and development of novel NP-based therapies [37,38]. Using chemical proteomics approaches and experimental assays, HSP90AB1 was identified as the direct target of BRA in the treatment of HCC, a novel finding. HSP90AB1, a member of the HSP90 molecular chaperone family, plays a critical role in facilitating oncogene activation and promoting cancer cell survival. The chaperone activity of HSP90AB1 in cancer cells prevents the misfolding and degradation of overexpressed and mutated oncogenic proteins [39]. Abnormal and constitutive activation of HSP90AB1 is observed in various cancers, including liver, gastric, and lung cancers, contributing to poor patient prognosis [[40], [41], [42]]. Multi-omics analysis of HCC tissues, including TCGA data, single-cell RNA sequencing, and tumor samples, revealed that HSP90AB1 is overexpressed in patients with HCC and correlates with poor survival outcomes. Notably, the binding affinity between BRA and HSP90AB1 was confirmed through direct binding assays, including CETSA, SPR, and MST. Furthermore, it was demonstrated that BRA directly binds to HSP90AB1, with SER-108 playing a key role in this interaction. Moreover, knockdown of HSP90AB1 significantly inhibited HCC cell proliferation both in vitro and in vivo, suggesting that HSP90AB1 could serve as a valuable biomarker and potential therapeutic target in HCC.
Mechanistically, HSP90AB1 interacts with low-density lipoprotein receptor-related protein 5 (LRP5), preventing its ubiquitin-mediated degradation and activating the AKT and Wnt/β-catenin signaling pathways, thereby promoting gastric cancer progression [42]. Additionally, HSP90AB1 has been implicated in enhancing endothelial cell-dependent tumor angiogenesis by activating VEGFR transcription in HCC [40]. However, the precise mechanisms through which HSP90AB1 influences HCC progression remain poorly understood. Using TMT-based quantitative proteomics, this study revealed that HSP90AB1 knockdown led to a reduction in the expression levels of PIK3CG, EGFR, and KDM5C, a novel finding. Moreover, both HSP90AB1 knockdown and BRA treatment in vitro significantly reduced the expression of these proteins, suggesting that PIK3CG, EGFR, and KDM5C are downstream targets of HSP90AB1 and are inhibited by BRA. Increasing evidence supports that these proteins are upregulated and associated with poor overall survival in patients with HCC, highlighting their critical role in the disease [[43], [44], [45]]. Notably, silencing these proteins has been shown to inhibit HCC cell proliferation. However, the mechanisms by which HSP90AB1 regulates these proteins remain unclear and warrant further investigation.
This study has several limitations. Emerging evidence indicates that HSP90AB1 modulates the tumor microenvironment by enhancing immune evasion, promoting tumor cell–cancer-associated fibroblast (CAF) cross-talk, and remodeling stromal cells [46,47]. Consequently, further exploration of BRA's impact on the tumor microenvironment is needed. While molecular docking and mutagenesis studies have demonstrated the binding of HSP90AB1 to BRA, the precise chemical structure of this interaction remains undefined. Future studies will utilize cryo-electron microscopy (cryo-EM) to resolve the structure of human HSP90AB1 bound to BRA, providing insights into the interactions at the ligand-binding pocket and the conformational changes induced by the agonist, as described previously [48]. Modifying and optimizing the chemical structure of BRA in the future may help improve the curative effect while reducing the toxicity and side effects on normal tissues [49]. Additionally, as NPs often exert pharmacological effects through interactions with multiple targets or pathways, developing novel strategies for NP target identification is crucial [50].

Conclusion
This research highlights the significant inhibitory effect of BRA on HCC cell proliferation, positioning HSP90AB1 as a key molecular target. These findings not only enhance the understanding of BRA’s anti-tumor mechanisms but also suggest a promising pathway for targeted therapies in HCC. The demonstrated relationship between BRA and HSP90AB1 underscores the potential for innovative therapeutic strategies aimed at improving patient outcomes. Future clinical validation of these findings, along with exploration of BRA and its derivatives in treating other malignancies, will contribute to the evolving landscape of targeted cancer therapies.

Consent for publication
All co-authors have reviewed and approved the final version of the manuscript and its submission to this journal.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Compliance with ethics requirements
The present study was approved by the Ethics Committee of Xiangya Hospital and The Second Xiangya Hospital, Central South University, and was performed according to the Declaration of Helsinki.

CRediT authorship contribution statement
Weijun Peng: Conceptualization, Funding acquisition, Resources, Project administration, Writing – original draft, Writing – review & editing, Supervision. Dazun Shi: Funding acquisition, Resources, Project administration, Data curation, Conceptualization. Die Xu: Methodology, Validation, Visualization, Writing – original draft. Xiaowei Wang: Methodology, Resources, Validation, Visualization. Yisi Cai: Methodology, Validation, Visualization, Writing – original draft. Yejun Tan: Data curation, Formal analysis, Visualization. Yuqing Liu: Data curation, Formal analysis, Visualization. Yajuan Cui: Funding acquisition, Validation, Methodology. Lemei Zhu: Resources, Validation. Ke Ye: Resources, Validation. Kuan Hu: Resources, Validation. Jun Fu: Conceptualization, Investigation, Validation, Methodology, Data curation, Visualization, Writing – review & editing, Supervision.

Ethics approval and consent to participate
This study was approved by the Ethics Committee of Xiangya Hospital and the Second Xiangya Hospital, Central South University, and conducted in accordance with the Declaration of Helsinki.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.