Abstract

   Osteosarcoma (OS) presents significant treatment challenges due to
   chemoresistance. This study explores the molecular mechanisms
   underlying chemoresistance in OS, focusing on the novel fusion gene
   RPS27-RPS24. Using single-cell multi-omics techniques, we identified a
   significant upregulation of RPS27-RPS24 in chemoresistant OS cells. Our
   analyses revealed that RPS27-RPS24 enhances glutaminase (GLS)-mediated
   glutamine metabolism and inhibits copper-induced cell death, thereby
   promoting chemoresistance. In vitro experiments with
   adriamycin-resistant (ADMR) OS cells confirmed that overexpression of
   RPS27-RPS24 leads to increased cell viability and proliferation under
   chemotherapy. In vivo studies further validated these findings,
   demonstrating that targeting glutamine metabolism can reverse
   chemoresistance. Our results suggest that the RPS27-RPS24 fusion gene
   plays a critical role in OS chemoresistance through metabolic
   reprogramming, providing a potential therapeutic target for improving
   OS treatment outcomes.

   graphic file with name 41420_2025_2487_Figa_HTML.jpg
   [34]Open in a new tab

   The application of multiple analytical techniques in this study (as
   shown in the upper image) and the hypothesized mechanism (as shown in
   the lower image).

   Subject terms: Cancer, Molecular biology

Introduction

   Osteosarcoma (OS) is a highly lethal malignant tumor with
   chemoresistance, presenting significant challenges in its treatment
   [[35]1, [36]2]. The development of chemoresistance in the treatment of
   OS is increasingly evident, yet its mechanisms remain unclear (CITE).
   Previous studies have indicated the crucial role of fusion genes in
   tumor biology [[37]3–[38]5]. Consequently, investigating the function
   and mechanisms of fusion genes in OS is of great significance in
   improving its therapeutic efficacy [[39]6–[40]8].

   Fusion genes are formed by fusion events of two distinct genes and have
   diverse origins and important functions [[41]9]. In tumorigenesis and
   progression, fusion genes play a vital role in cell cycle regulation,
   transcription regulation, and epigenetic regulation [[42]10, [43]11].
   In OS, several fusion genes associated with occurrence and development
   have been identified, such as BCR-ABL, EWS-FLI1, etc. [[44]3, [45]12,
   [46]13]. Although fusion genes have been extensively studied, their
   specific role in the mechanism of chemoresistance in OS remains unclear
   [[47]3, [48]12, [49]13].

   Single-cell multi-omics techniques, such as single-cell RNA sequencing
   (scRNA-seq) and proteomics and metabolomics analysis, can reveal the
   molecular mechanisms of complex diseases, including the formation of
   chemoresistance [[50]14, [51]15]. The application of these techniques
   can overcome the limitations of traditional research and provide new
   perspectives for a deeper understanding of the mechanisms underlying
   chemoresistance in OS [[52]16, [53]17]. By utilizing these techniques,
   we can obtain information on gene expression, protein, and metabolite
   levels in individual cells, thus elucidating the role of fusion genes
   in chemoresistance in OS.

   This study employed scRNA-seq technology, along with data analysis and
   cell annotation methods, to screen the malignant cell markers in tumor
   tissues of OS patients. Key fusion genes associated with
   chemoresistance were identified through gene co-expression network
   analysis and prediction of fusion genes using online databases.
   Additionally, functional validation was performed using in vitro cell
   experiments and CRISPR/Cas9 editing technology to investigate the
   impact of key genes, glutamine metabolism, and copper death on
   chemoresistance in OS cells. To validate the experimental results, a
   subcutaneous xenograft mouse model of OS was constructed for in vivo
   experiments.

   The aim of this study was to provide a comprehensive understanding of
   the molecular mechanisms underlying chemoresistance in OS, particularly
   those related to the fusion gene RPS27-RPS24. Uncovering these
   mechanisms is crucial for understanding chemoresistance in OS,
   developing new therapeutic strategies, and improving patient prognosis.
   By delving into the functions and regulatory mechanisms of the fusion
   gene RPS27-RPS24, we hope to identify new targets and approaches that
   can offer more effective strategies and tools for personalized
   treatment and precision medicine. Despite some limitations, such as
   limited sample size and the absence of an animal model, we believe
   these findings will provide new directions and insights for future OS
   research and translational studies in other tumors.

Results

Characterization of cellular heterogeneity and malignancy markers in OS

   OS is a highly aggressive and metastatic malignant tumor originating
   from bone or soft tissues, posing a longstanding challenge in cancer
   treatment. This disease has a higher incidence rate in children and
   adolescents, causing significant physiological and psychological stress
   to patients and their families. Therefore, understanding the mechanisms
   underlying the development of OS is of critical importance for saving
   patients’ lives [[54]18, [55]19]. Cellular changes in the tumor
   microenvironment play a vital role in the malignant progression of OS
   [[56]20, [57]21].

   To gain a deeper understanding of the development process of OS and
   changes in the tumor microenvironment, we obtained OS-related scRNA-seq
   data from the GEO database, specifically the [58]GSE162454 dataset.
   This dataset consisted of tumor tissue samples from six OS patients,
   and the data analysis workflow is summarized in Fig. [59]1A. Firstly,
   we used the Seurat package to integrate the sequencing data and examine
   the gene counts (nFeature_RNA), the number of mRNA molecules
   (nCount_RNA), and the percentage of mitochondrial genes (percent.mt) in
   all cells of the scRNA-seq data (Fig. S[60]1A). A quality control step
   was performed by filtering the data based on the criteria of
   200 < nFeature_RNA < 5000, 200 < nCount_RNA < 20,000, and
   percent.mt < 25, resulting in a total of 37,432 cells. The calculation
   of sequencing depth correlation revealed a correlation coefficient of
   r = −0.1 between nCount_RNA and percent.mt, and a correlation
   coefficient of r = 0.88 between nCount_RNA and nFeature_RNA (Fig.
   S[61]1B), indicating good quality of the filtered cell data. Next, we
   identified highly variable genes in the filtered cells from OS,
   including a total of 25,841 genes, and performed gene expression
   variance analysis to identify the top 2,000 variable genes for
   downstream analysis (Fig. S[62]1C).

Fig. 1. UMAP clustering, cell annotation, and CNV analysis of scRNA-seq data.

   [63]Fig. 1
   [64]Open in a new tab

   A Workflow for scRNA-seq data analysis. B UMAP clustering analysis
   classified cells into 17 cell clusters (shown by sample origin). C 17
   cell clusters annotated as 9 cell types, with suspected OS cells
   highlighted in the red box. D Bubble plot of marker genes for the 17
   cell clusters, with suspected OS cell marker genes highlighted in the
   red box. E Violin plot showing expression of suspected OS cell marker
   genes across the 9 cell types. F CNV analysis of scRNA-seq data, with
   blue indicating DNA copy loss, red indicating DNA copy gain, and normal
   cell samples highlighted in the red box. G Comparison of CNV scores
   after K-means clustering.

   The CellCycleScoring function was used to calculate the cell cycle of
   the samples (Fig. S[65]1D), and the data was normalized. The RunPCA
   function was employed for PCA on the top 2000 variable genes to reduce
   the dimensionality of the data. The expression plot of the main
   correlated genes of the first two principal components is shown
   selectively (Fig. S[66]1E). An ElbowPlot was generated to rank the
   principal components by standard deviation, where the “important”
   components had higher standard deviations (Fig. S[67]1F). Additionally,
   the JackStrawPlot function was used to visualize the P distribution of
   the top 50 principal components compared to the mean distribution (Fig.
   S[68]1G). The “important” PCs typically had small P (represented by
   solid lines above the dashed line) and reflected the information
   contained in the highly variable genes selected earlier. Based on these
   results, the first 30 principal components adequately represented the
   information in the selected highly variable genes and were chosen for
   subsequent UMAP analysis.

   Using UMAP clustering analysis, we identified 17 cell clusters, where
   the cellular composition did not differ significantly between the
   sample sources but varied in proportions within the cell clusters (Fig.
   [69]1B). By searching relevant literature and consulting the CellMarker
   online database, we annotated the 17 cell clusters into nine cell types
   (Fig. [70]1C). These included five immune cell types: B cells (marked
   by MS4A1 and CD79A), monocytes (marked by LYZ, CD14, and FCGR3A), mast
   cells (marked by TPSB2 and CPA3), NK/T cells (marked by CD3D, CD3E,
   CD3G, GNLY, NKG7, KLRD1, and KLRB1), and plasmocytes (marked by IGHG1
   and MZB1) (Fig. [71]1D). Additionally, four non-immune cell types were
   identified: endothelial cells (marked by PECAM1 and VWF), osteoclasts
   (marked by ACP5 and CTSK), BMSCs/osteoblastic OS cells (marked by
   RUNX2, THY1, IBSP, and ALPL), and chondroblastic OS cells (marked by
   ACAN and SOX9) (Fig. [72]1D, E).

   To accurately identify malignant cells from BMSCs/osteoblastic OS cells
   and chondroblastic OS cells, we employed the “inferCNV” package to
   detect widespread CNV, using NK/T cells as a control. The results
   demonstrated that a majority of BMSCs/osteoblastic OS cells and
   chondroblastic OS cells exhibited CNV (Fig. [73]1F). Subsequently, the
   cells in “Observation” were clustered into six classes, revealing that
   the fourth class included all normal cells with the lowest CNV score
   (Fig. [74]1G). After excluding the fourth class, a total of 6443
   malignant OS cells were obtained. Subsequently, we applied a threshold
   of avg_log2FC > 1 and P < 0.05 to select 3805 genes as markers for
   further analysis.

Identification of key genes and fusion genes in chemoresistance in OS

   To identify key genes closely associated with chemoresistance in OS, we
   obtained RNA-seq data from the [75]GSE154540 dataset in the GEO
   database and constructed a scale-free co-expression network using the R
   package WGCNA. Initially, we performed hierarchical clustering on tumor
   tissue samples from 50 OS patients and removed outliers (Fig. S[76]2A).
   A soft threshold of β = 11 was chosen to establish the scale-free
   network (Fig. S[77]2B). The parameters minModule size and
   mergeCutHeight were set to 300 and 0.25, respectively, resulting in the
   division of the network into 18 modules (Fig. S[78]2C). Furthermore,
   module-trait correlation analysis revealed that the blue module was
   significantly associated with chemoresistance in OS (r = 0.41,
   P = 0.004) (Figs. [79]2A and S[80]2D).

Fig. 2. Identification of key genes and fusion genes in chemoresistance in OS
using single-cell integrated transcriptomic data.

   [81]Fig. 2
   [82]Open in a new tab

   A Correlation heatmap between module feature genes and clinical
   features, each cell contains the corresponding correlation and p-value.
   B Venn diagram showing the intersection between blue module genes and
   malignant OS cell marker genes. C Forest plot showing the prediction of
   overall survival in OS patients using 5 high-risk genes (N = 96). D
   Survival curve of patients with high expression of RPS27 (N = 48) and
   low expression of RPS27 (N = 48). E ROC curve showing the prediction of
   mortality rate in OS patients using 5 high-risk genes (N = 37). F ROC
   curve showing the prediction of tumor recurrence in OS patients using 5
   high-risk genes (N = 37). G Correlation analysis of expression levels
   between RPS27 and RPS24 (N = 96). H Survival curve of patients with
   high expression of RPS24 (N = 48) and low expression of RPS24 (N = 48).
   I ROC curve showing the prediction of overall survival at 1, 3, and 5
   years using RPS24 expression (N = 96). J Interaction network of the
   fusion between RPS27 and RPS24 genes.

   Next, we extracted 1230 genes from the blue module and intersected them
   with the previously identified 3805 malignant OS cell marker genes,
   resulting in 76 overlapping genes (Fig. [83]2B). To investigate the
   association between these intersection genes and the prognosis of OS
   patients, we downloaded RNA-seq data (TARGET-OS) and corresponding
   survival information from 96 OS tumor tissue samples in the TARGET
   database. Univariate survival analysis of the 76 intersection genes
   revealed that 5 high-risk genes (IGF1R, S100A13, FDPS, RPS27, FABP3)
   were significantly correlated with overall survival in OS patients
   (Fig. [84]2C, D). In addition, using RNA-seq data from 37 OS patients
   in the GEO dataset [85]GSE39055, along with clinical information on
   necrosis and tumor recurrence, we performed ROC analysis on the 5
   high-risk genes and found that only RPS27 showed good predictive value
   for necrosis and tumor recurrence in OS patients (Fig. [86]2E, F).
   These findings highlight RPS27 as a key gene in chemoresistance in OS
   identified through single-cell transcriptomic analysis.

   RPS27 (Ribosomal Protein S27), also known as Metallopanstimulin-1 or
   MPS-1, is a ribosomal gene that encodes a component of the 40S subunit.
   Previous studies have demonstrated the upregulation of RPS27 expression
   in various human malignancies, such as melanoma [[87]22] and gastric
   cancer [[88]23]. However, the role and mechanism of RPS27 in OS or
   chemoresistance have yet to be elucidated, making it a suitable target
   for our study. Gene fusion events are commonly observed in tumors and
   have important implications for diagnosis, prognosis, and treatment
   [[89]24]. Additionally, several studies have reported a close
   association between fusion genes and tumor chemoresistance
   [[90]25–[91]28].

   To investigate whether there are RPS27-related gene fusions in OS, we
   first used the FusionGDB2, ChimerDB 3.0, and ChiTaRS 5.0 databases to
   predict fusion gene pairs involving RPS27. These databases predicted 6,
   1, and 3 fusion gene pairs, respectively, and we obtained the
   intersection of these predictions, resulting in the identification of
   RPS27-RPS24 as a fusion gene pair. We performed correlation analysis on
   the expression levels of RPS27 and RPS24 in the TARGET-OS data, which
   showed a significant positive correlation between them (Fig. [92]2G).
   Furthermore, OS patients with high expression of RPS24 exhibited poorer
   prognosis (Fig. [93]2H), and RPS24 demonstrated good predictive value
   for 3-year and 5-year overall survival in OS patients (Fig. [94]2I).
   The prediction results from ChimerDB 3.0 also revealed the gene fusion
   interaction process between RPS27 and RPS24 (Fig. [95]2J). Therefore,
   we hypothesize that the RPS27-RPS24 fusion gene may be involved in
   chemoresistance in OS.

Overexpression of RPS27-RPS24 confers chemoresistance to OS cells

   We selected the most commonly used chemotherapy drug in OS treatment,
   ADM [[96]29, [97]30]. U2OS and MG63 cells were cultured in a medium
   supplemented with ADM at a concentration of 1 µg/ml for 6 months,
   resulting in the establishment of U2OS/ADMR and MG63/ADMR cells.
   Subsequently, we assessed the response of these OS cells to ADM and
   explored the potential correlation between the expression of the fusion
   gene RPS27-RPS24 and ADM resistance.

   The cells were exposed to different concentrations of ADM (0, 0.5, 1,
   and 2 µg/ml) for 48 h, and the expression of RPS27 was then detected
   through Western blot analysis. The results showed a significant
   increase in the levels of RPS27 protein in U2OS/ADMR and MG63/ADMR
   cells after exposure to different concentrations (1 and 2 µg/ml) of ADM
   for 48 h, compared to U2OS and MG63 cells treated with different
   concentrations of ADM. It is worth noting that during the detection of
   RPS27 protein expression, we observed a distinct protein band above,
   which exhibited changes parallel to the expression trend of RPS27. This
   upper band could possibly correspond to RPS27-RPS24 (Figs. [98]3A and
   S[99]3A). Next, we constructed an overexpression plasmid for
   RPS27-RPS24 and transfected it into U2OS/ADMR and MG63/ADMR cells. We
   observed the detection of a smaller protein band, consistent in size
   with the protein detected by the anti-Flag antibody and the upper band
   mentioned earlier (Figs. [100]3A and S[101]3A). Furthermore, the
   expression of this protein was significantly increased in the cells
   overexpressing RPS27-RPS24 (Figs. [102]3B and S[103]3B). Additionally,
   IP experiments demonstrated that in cells modified with the fusion gene
   RPS27-RPS24, a larger protein band was detected by the RPS24 antibody,
   with a size of ~15–20 kDa larger than the natural RPS27 protein,
   confirming the expression of RPS27-RPS24 protein (Figs. [104]3C and
   S[105]3C). Overall, these results indicate an upregulation of the
   RPS27-RPS24 fusion gene expression in OS ADM-resistant cells.

Fig. 3. Impact of RPS27-RPS24 on chemoresistance in U2OS cells.

   [106]Fig. 3
   [107]Open in a new tab

   A Western blot analysis of RPS27 protein expression in U2OS and
   U2OS/ADMR cells in different groups. B Western blot analysis of RPS27
   protein expression in U2OS/ADMR cells transfected with either Vector or
   RPS27-RPS24 overexpression plasmids. C IP experiment detecting the
   expression of RPS27-RPS24 protein in U2OS/ADMR cells transfected with
   either Vector or RPS27-RPS24 overexpression plasmids. D Clonogenic
   assay measuring cell proliferation of U2OS and U2OS/ADMR cells after
   treatment with 5 µg/ml ADM, along with the corresponding statistical
   graph. E Sensitivity of U2OS and U2OS/ADMR cells to ADM. F MTT assay
   evaluating cell viability of U2OS/ADMR cells treated with different
   concentrations of ADM for 48 h. G Clonogenic assay measuring cell
   proliferation of U2OS/ADMR cells treated with 5 µg/ml ADM after
   transfection with either Vector or RPS27-RPS24 overexpression plasmids,
   along with the corresponding statistical graph. H EdU assay assessing
   cell proliferation of U2OS/ADMR cells treated with 5 µg/ml ADM after
   transfection with either Vector or RPS27-RPS24 overexpression plasmids
   (scale bar: 50 μm), along with the corresponding statistical graph. I
   Treatment of U2OS/ADMR cells with 1 µM Thapsigargin for 24 h, followed
   by MTT assays to assess cell proliferation after transfection with
   Vector or RPS27-RPS24 overexpression plasmids. An independent sample
   t-test was used for comparing two groups, while two-way ANOVA was used
   for comparisons across different time points. * Indicates P < 0.05
   compared to the U2OS group or Vector group. Cell experiments were
   performed in triplicates.

   Furthermore, as expected, colony formation experiments showed that
   compared to U2OS and MG63 cells, treatment with ADM (5 µg/ml) did not
   lead to a decrease in the colony formation ability or a reduction in
   the size of U2OS/ADMR and MG63/ADMR colonies (Figs. [108]3D and
   S[109]3D). Moreover, cell viability assays confirmed that U2OS/ADMR and
   MG63/ADMR cells were able to tolerate higher concentrations of ADM
   compared to parental U2OS and MG63 cells. The IC50 of ADM in U2OS and
   MG63 cells was significantly lower than that in MG63/ADMR and U2OS/ADMR
   cells, suggesting the potential role of increased RPS27-RPS24
   expression in both intrinsic and acquired ADM resistance in OS cells
   (Figs. [110]3E and S[111]3E). Thus, it was evident that ADM induced
   upregulation of the fusion gene RPS27-RPS24 and that increased
   expression of RPS27-RPS24 may partially contribute to the intrinsic and
   acquired resistance of OS cells to ADM.

   Subsequently, we examined the effect of overexpressing RPS27-RPS24 on
   the chemoresistance of OS cells using MTT, colony formation, and EdU
   assays. The results revealed a significant attenuation of the response
   of U2OS/ADMR and MG63/ADMR cells to ADM after overexpression of
   RPS27-RPS24, characterized by a weakened inhibitory effect of the drug
   on cell viability and proliferation (Figs. [112]3F–H and S[113]3F-H).
   These findings suggested that the upregulation of RPS27-RPS24 during
   ADM treatment in OS cells impedes the cytotoxic effect of the drug. We
   performed MTT assays to assess the ability of RPS27-RPS24
   overexpression to resist endoplasmic reticulum (ER) stress in OS cells.
   The results demonstrated that overexpression of RPS27-RPS24
   significantly enhanced the ability of U2OS/ADMR and MG63/ADMR cells to
   resist the ER stress inducer thapsigargin (Figs. [114]3I and S[115]3I).

RPS27-RPS24 regulates a network of DEPs in OS cells

   We analyzed U2OS/ADMR cell samples with stable transfection of Vector
   or RPS27-RPS24 using the TMT quantitative proteomics technique. The
   analysis workflow is shown in Fig. [116]4A. Through this study, we have
   revealed that the impact of RPS27-RPS24 on chemoresistance in OS may be
   related to differential protein expression.

Fig. 4. TMT-labeled quantitative proteomic analysis under RPS27-RPS24
intervention.

   [117]Fig. 4
   [118]Open in a new tab

   A Schematic diagram of the proteomics workflow in this study. B Volcano
   plot analysis of the differences between the Vector group (N = 3) and
   RPS27-RPS24 group (N = 3) U2OS/ADMR cell samples, with red representing
   upregulated proteins, blue representing downregulated proteins, and
   gray indicating proteins with insignificant differential expression. C
   Expression heatmap of 23 DEPs in the Vector group (N = 3) and
   RPS27-RPS24 group (N = 3) U2OS/ADMR cell samples. D Bubble plot of the
   WikiPathways enrichment analysis results for DEPs. E Bar plot of the
   KEGG enrichment analysis results for DEPs. F Dendrogram of the GO
   enrichment analysis results for DEPs. G Protein–protein interaction
   network of DEPs. H Western blot analysis of GLS protein expression in
   cells from each group. I Western blot analysis of FDX1, LIAS, and
   Lipoy-DLAT protein expression in cells from each group, with the cell
   experiment repeated three times.

   Initially, we identified 7141 protein molecules, applying a threshold
   of |log2FC| > 1 and P < 0.05. We filtered out 23 DEPs, including 19
   upregulated proteins (GLS, RPL19, RPS27A, NR3C1, HSCB, RPL38, PIPOX,
   RCAN3, CDKN2A, MRM1, COX4I1, GABBR1, TRMT61B, ITGA7, LYRM4, THNSL2,
   ENO2, FDX1, and SIRT5) and 4 downregulated proteins (CHAMP1, PHF6,
   ATP7B, and SLC31A1) (Fig. [119]4B, C). Functional enrichment analysis
   of these 23 DEPs, with a threshold of P < 0.05, revealed that they were
   primarily involved in pathways such as chemoresistance, ribosome, and
   copper metabolism (Fig. [120]4D, E). Furthermore, they were enriched in
   BP, such as copper ion transport, copper ion homeostasis, and
   alpha-amino acid catabolic processes, CC such as cytoplasmic ribosome,
   polysome, ribosomal subunit, and cytoplasmic large ribosomal subunit,
   as well as MF such as rRNA methyltransferase activity, catalytic
   activity acting on rRNA, and transition metal ion transmembrane
   transporter activity (Fig. [121]4F). These findings indicate a close
   association between RPS27-RPS24 and OS cell copper death.

   Subsequently, we constructed a protein–protein interaction (PPI)
   network for the DEPs and identified the core proteins through
   topological analysis, with GLS being located at the central position
   (Fig. [122]4G). Western blot analysis confirmed a significant
   upregulation in the expression of GLS in RPS27-RPS24 group OS cells
   compared to the Vector group (Fig. [123]4H), further validating the
   accuracy of the proteomic analysis.

   Copper death is characterized by the loss of Fe–S cluster proteins FDX1
   and LIAS, as well as reduced protein lipoylation of DLAT. Therefore, we
   further examined the protein expression of FDX1, LIAS, and Lipoy-DLAT.
   The results showed a significant upregulation in the expression of
   FDX1, LIAS, and Lipoy-DLAT in RPS27-RPS24 group OS cells compared to
   the Vector group (Fig. [124]4I). Recent evidence suggests that GLS, as
   a glutaminase, participates in tumor progression by influencing the
   copper death process [[125]31–[126]33]. This suggests that RPS27-RPS24
   may affect OS-resistant cells’ copper death by promoting GLS
   expression.

   In conclusion, we have uncovered the role of a series of DEPs upon
   overexpression of RPS27-RPS24, with particular emphasis on the central
   role of GLS and the association of RPS27-RPS24 with copper death. This
   discovery enhances our understanding of the molecular mechanisms by
   which RPS27-RPS24 affects chemoresistance in OS, and further
   investigations will be conducted in follow-up experiments.

RPS27-RPS24 promotes glutamine metabolism and chemoresistance

   To investigate the potential mechanism by which RPS27-RPS24 functions
   further, we collected culture supernatant samples from U2OS/ADMR cells
   transfected with a stable vector or RPS27-RPS24 for untargeted
   metabolomics analysis. The analysis process is outlined in Fig.
   S[127]4A. Firstly, we conducted PCA using EZinfo software to analyze
   the acquired mass spectrometry data, as shown in Fig. S[128]4B, C.
   Differential metabolites (highlighted in red boxes) were selected based
   on a VIP > 1 cutoff (Fig. S[129]4D). Subsequently, data were further
   corrected using the MetaboAnalyst website (Fig. S[130]4E), fold change
   and P were calculated, and differentially expressed metabolites were
   identified using a threshold of |log2FC| > 2 and P < 0.05, resulting in
   32 differential metabolites (Fig. S[131]4F, G). The VIP values of each
   metabolite were calculated using the OPLSDA algorithm, and the top 15
   metabolites with the highest VIP values were presented (Fig. [132]5A,
   B). Pathway and functional enrichment analysis revealed that the
   glutamine metabolism pathway was prominently enriched (Fig. [133]5C,
   D). These results suggest that RPS27-RPS24 may be involved in
   OSchemoresistance by regulating glutamine metabolism.

Fig. 5. Metabolomic analysis under RPS27-RPS24 intervention.

   [134]Fig. 5
   [135]Open in a new tab

   A OPLSDA-PCA-2D plot, with a T score on the x-axis and an Orthogonal T
   score on the y-axis. B Bubble plot of the top 15 metabolites based on
   VIP value. C Pathway enrichment analysis results of the differential
   metabolites in the MetaboAnalyst database. D Functional enrichment
   analysis results of the differential metabolites in the MetaboAnalyst
   database. E Glutamine uptake content in cells from each group. F–H
   Metabolic levels of glutamate (F), α-ketoglutarate (G), and ammonia (H)
   in cells after 8 h of glutamine supplementation. The comparison of two
   sets of data was conducted using an independent samples t-test.
   *Indicates P < 0.05 compared to the Vector group, with the cell
   experiment repeated three times.

   Glutamine is a non-essential amino acid that plays a crucial role in
   cellular metabolism. The metabolic products of glutamine include: (1)
   glutamate, which is converted from glutamine through the action of GLS;
   (2) α-ketoglutarate, a key intermediate of the tricarboxylic acid (TCA)
   cycle that is generated from glutamine by further deamination; (3)
   ammonia, which is released during glutamine conversion to
   α-ketoglutarate [[136]34]. To assess the level of glutamine intake in
   the culture medium, we used a YSI 7100 multi-channel biochemical
   analyzer. Additionally, the concentrations of metabolites (glutamate,
   α-ketoglutarate, and ammonia) were measured using high-performance
   liquid chromatography (HPLC), and the changes in metabolite
   concentrations relative to fresh growth medium were normalized to
   protein content per well. We found that compared to the Vector Group,
   the RPS27-RPS24 group had higher levels of glutamine intake and
   significantly increased concentrations of glutamate, α-ketoglutarate,
   and ammonia (Fig. [137]5E–H), consistent with the results of the
   metabolomics analysis. These findings suggest that overexpression of
   RPS27-RPS24 promotes glutamine metabolism.

   In line with previous studies linking GLS-driven glutamine catabolism
   to the occurrence and progression of prostate cancer and pancreatic
   cancer [[138]35, [139]36], as well as enhanced glutamine hydrolysis
   driving hypoxia-induced chemoresistance in pancreatic cancer [[140]37],
   we hypothesize that RPS27-RPS24 may promote glutamine metabolism by
   upregulating GLS expression, which may be closely associated with
   OSchemoresistance and copper death.

Reversal of chemoresistance by GLS knockout and inhibition of copper death in
OS cells

   Cell grouping was performed as shown in Fig. [141]6A (using U2OS/ADMR
   as an example). Firstly, U2OS/ADMR and MG63/ADMR cell lines with GLS
   knockout were constructed using CRISPR/Cas9 gene editing technology
   (Fig. S[142]5A). The efficiency of gene knockout was validated through
   RT-qPCR and Western blot analysis, which revealed a significant
   reduction in GLS expression levels in GLS-KO U2OS/ADMR and MG63/ADMR
   cells compared to GLS-WT cells (Fig. S[143]5B, C). This demonstrates
   the successful knockout of GLS. The impact of GLS knockout on
   chemoresistance in OS cells was assessed through MTT, colony formation,
   and EdU assays. The results revealed a significant enhancement in the
   response of GLS-KO cells to ADM compared to GLS-WT cells, evidenced by
   the increased inhibitory effect of the drug on cell viability and
   proliferation (Fig. S[144]6A–C). This indicates that the knockout of
   GLS considerably weakens chemoresistance in OS cells.

Fig. 6. The impact of RPS27-RPS24 on GLS-mediated glutamine metabolism on
U2OS/ADMR cells in copper death and chemoresistance.

   [145]Fig. 6
   [146]Open in a new tab

   A Schematic representation of the cell grouping process, using
   U2OS/ADMR as an example. B–D Represents the metabolic levels of
   glutamate (B), α-ketoglutarate (C), and ammonia (D) in the cells after
   8 h of culture in the presence of glutamine medium. E Western blot
   analysis of FDX1, LIAS, and Lipoy-DLAT protein expression in the cells
   of each group. F Copper ion content in the cells of each group. G Cell
   viability was determined by CCK-8 assay after 72 h of treatment with
   different concentrations of elesclomol–Cu (ratio = 1:1). H Cell
   migration capability assessed by scratch experiment (upper image)
   (scale bar: 200 μm) and cell invasion ability assessed by Transwell
   assay (lower image) (scale bar: 100 μm) after 72 h of treatment with
   50 mM elesclomol–Cu (ratio = 1:1). I Cell viability of GLS-WT and
   GLS-KO U2OS/ADMR cells determined by MTT assay. J Cell proliferation of
   GLS-WT and GLS-KO groups assessed by clone formation experiment and
   statistical analysis after treatment with 5 µg/ml ADM (scale bar:
   50 μm). K Cell proliferation of GLS-WT and GLS-KO groups assessed by
   EdU assay and statistical analysis after treatment with 5 µg/ml ADM
   (scale bar: 50 μm). L Cell viability of GLS-KO + RPS27-RPS24 + PBS and
   GLS-KO + RPS27-RPS24 + TTM groups determined by MTT assay. M Cell
   proliferation of GLS-KO + RPS27-RPS24 + PBS and
   GLS-KO + RPS27-RPS24 + TTM groups assessed by clone formation
   experiment and statistical analysis after treatment with 5 µg/ml ADM. N
   Cell proliferation of GLS-KO + RPS27-RPS24 + PBS and
   GLS-KO + RPS27-RPS24 + TTM groups assessed by EdU assay and statistical
   analysis after treatment with 5 µg/ml ADM. For the comparison of data
   across different time periods or factors, a two-way ANOVA was
   conducted. *Indicates P < 0.05 compared to the Vector group or the
   GLS-KO + RPS27-RPS24 + PBS group, and all cell experiments were
   repeated three times.

   We also measured the concentrations of glutamine metabolism products
   (glutamate, α-ketoglutarate, and ammonia) in the different groups of
   cells. The results showed a significant increase in the concentrations
   of these metabolites in the RPS27-RPS24 group of GLS-WT cells compared
   to the Vector group. However, there were no significant differences in
   the concentrations of these metabolites between the Vector group and
   the RPS27-RPS24 group in GLS-KO cells (Figs. [147]6B–D and S[148]7A–C).
   These findings demonstrate that the knockout of GLS effectively
   reverses the promoting effect of RPS27-RPS24 overexpression on
   glutamine metabolism.

   Next, we evaluated the protein expression levels of copper
   death-related factors and the copper ion content in the different
   groups of cells. The results showed a significant upregulation of FDX1,
   LIAS, and Lipoy-DLAT protein expression and a significant reduction in
   copper ion levels in the RPS27-RPS24 group of GLS-WT cells compared to
   the Vector group. However, no significant differences were observed in
   protein expression levels between the Vector group and the RPS27-RPS24
   group in GLS-KO cells (Figs. [149]6E, F and S[150]7D, E). Additionally,
   we treated the cells with different concentrations of elesclomol–Cu
   (ratio = 1:1) for 72 h and assessed cell viability through CCK-8
   assays, cell migration through scratch assays, and cell invasion using
   Transwell assays. The results showed that the RPS27-RPS24 group of
   GLS-WT cells exhibited higher cell viability, enhanced migration, and
   invasion capabilities compared to the Vector group. However, there were
   no significant differences in cell viability, migration, and invasion
   capabilities between the Vector group and the RPS27-RPS24 group in
   GLS-KO cells (Figs. [151]6G, H and S[152]7F, G). This indicates that
   the knockout of GLS significantly reverses the inhibitory effect of
   RPS27-RPS24 overexpression on copper death in drug-resistant OS cells.

   To further confirm whether RPS27-RPS24 affects copper death through
   GLS-mediated glutamine metabolism and ultimately promotes
   chemoresistance, we treated cells overexpressing RPS27-RPS24 with the
   glutamine metabolism inhibitor CB-839 [[153]38]. The cells were divided
   into the RPS27-RPS24 + PBS group and the RPS27-RPS24 + CB-839 group,
   and changes in copper death-related indicators were assessed. It was
   found that compared to the RPS27-RPS24 + PBS group, the
   RPS27-RPS24 + CB-839 group exhibited significant downregulation of
   FDX1, LIAS, and Lipoy-DLAT protein expression, a significant increase
   in copper ion levels, lower cell viability after elesclomol-Cu
   treatment, and reduced migration and invasion capabilities (Fig.
   S[154]8A–D). This provides further evidence that RPS27-RPS24 promotes
   the inhibition of copper death through GLS-mediated glutamine
   metabolism.

   The results of chemoresistance-related assays showed that the
   RPS27-RPS24 group of GLS-WT cells exhibited significantly reduced
   response to ADM compared to the Vector group, resulting in a weaker
   inhibitory effect of the drug on cell viability and proliferation.
   However, there were no significant differences in response to ADM
   between the Vector group and the RPS27-RPS24 group in GLS-KO cells
   (Figs. [155]6I–K, and S[156]7H–J). These results indicate that the
   knockout of GLS significantly reverses the promoting effect of
   RPS27-RPS24 overexpression on chemoresistance in OS cells.

   Finally, to clarify the impact of copper death on chemoresistance and
   whether RPS27-RPS24 promotes chemoresistance through regulating GLS
   expression to inhibit copper death, GLS-KO cells transfected with
   RPS27-RPS24 plasmids were treated with PBS or the copper death
   inhibitor TTM. The results showed a significantly reduced response to
   ADM in the GLS-KO + RPS27-RPS24 + TTM group compared to the
   GLS-KO + RPS27-RPS24 + PBS group (Figs. [157]6L–N and S[158]7K–M). This
   indicates that the inhibition of copper death promotes chemoresistance
   in OS cells, and the specific mechanism involves RPS27-RPS24 promoting
   GLS-mediated glutamine metabolism to inhibit copper death and
   ultimately promote chemoresistance.

RPS27-RPS24 promotes in vivo tumor growth and chemoresistance via enhanced
glutamine metabolism

   The above paragraph describes the results of an in vitro cell
   experiment investigating the role of RPS27-RPS24 in promoting
   GLS-mediated glutamine metabolism inhibition and chemoresistance in OS
   cells. To validate whether this mechanism affects the tumorigenic
   capability of OS cells in vivo, we injected OS cells overexpressing
   RPS27-RPS24 into the axilla of mice to establish subcutaneous xenograft
   tumor models. The mice were treated with either the glutamine
   metabolism inhibitor CB-839 or the copper death inhibitor TTM, starting
   from day 7, via intraperitoneal injection of a PBS solution or 5 mg/kg
   ADM treatment (Fig. [159]7A).

Fig. 7. Impact of RPS27-RPS24 on tumorigenesis and chemoresistance in U2OS
cells through regulation of glutamine metabolism and copper death.

   [160]Fig. 7
   [161]Open in a new tab

   A Schematic representation of the animal grouping process. B Western
   blot analysis of RPS27-RPS24, GLS, FDX1, LIAS, and Lipoy-DLAT protein
   expression levels in the tumor tissues of each group of mice. C–E
   Metabolic levels of glutamate (C), α-ketoglutarate (D), and ammonia (E)
   in the tumor tissues of each group of mice. F Measurement of tumor
   growth at different time points using bioluminescence intensity. G
   Morphology of tumor tissues from each group, with three representative
   examples shown for each group. H Immunohistochemical staining to
   determine protein expression levels of Ki67 in the tumor tissues of
   each group (scale bar = 50 μm). I Tumor tissue weights of each group. J
   Statistical analysis of Ki67-positive expression. For comparing data
   between different groups, one-way ANOVA was utilized. *Indicates
   P < 0.05 compared to the Vector + PBS + PBS group or the
   RPS27-RPS24 + PBS + PBS group or the RPS27-RPS24 + CB-839 + PBS group,
   with six mice per group.

   Firstly, Western blot analysis revealed that compared to the
   Vector+PBS + PBS group, the RPS27-RPS24 + PBS + PBS group of mice
   showed a significant increase in the expression levels of RPS27-RPS24,
   GLS, FDX1, LIAS, and Lipoy-DLAT proteins in tumor tissues. In
   comparison to the RPS27-RPS24 + PBS + PBS group, the
   RPS27-RPS24 + CB-839 + PBS group exhibited no significant change in
   RPS27-RPS24 expression and a significant decrease in the expression of
   GLS, FDX1, LIAS, and Lipoy-DLAT proteins. Additionally, compared to the
   RPS27-RPS24 + CB-839 + PBS group, the RPS27-RPS24 + CB-839 + TTM group
   showed no significant changes in RPS27-RPS24 and GLS expression but had
   a significant increase in the expression of FDX1, LIAS, and Lipoy-DLAT
   proteins (Figs. [162]7B and S[163]9A). These results suggest that
   RPS27-RPS24 overexpression can promote GLS expression and inhibit
   copper death, while the glutamine metabolism inhibitor can reverse the
   inhibitory effect of RPS27-RPS24 overexpression on copper death.
   Furthermore, the addition of the copper death inhibitor can further
   reverse the effect of the glutamine metabolism inhibitor.

   We also measured the concentrations of glutamine metabolism products in
   the tumor tissues of each group of mice. The results showed that
   compared to the Vector+PBS + PBS group, the RPS27-RPS24 + PBS + PBS
   group of mice had significantly increased levels of glutamate,
   α-ketoglutarate, and ammonia in tumor tissues. In comparison to the
   RPS27-RPS24 + PBS + PBS group, the RPS27-RPS24 + CB-839 + PBS group
   exhibited significant decreases in the concentrations of all
   metabolites. There were no significant changes in the concentrations of
   these metabolites between the RPS27-RPS24 + CB-839 + PBS group and the
   RPS27-RPS24 + CB-839 + TTM group (Figs. [164]7C–E and S[165]9B–D).
   These findings indicate that RPS27-RPS24 overexpression can promote
   glutamine metabolism and increase the production of glutamate,
   α-ketoglutarate, and ammonia. The glutamine metabolism inhibitor can
   reverse the effect of RPS27-RPS24 overexpression, while the addition of
   the copper death inhibitor has no effect on glutamine metabolism. This
   suggests that RPS27-RPS24 inhibits copper death by promoting glutamine
   metabolism.

   After one and two weeks, tumor growth in mice was assessed using
   bioluminescent imaging. At week 4, tumor tissue was excised to observe
   tumor morphology, measure tumor weight, and perform Ki67
   immunohistochemical staining. In the PBS treatment group, compared to
   the Vector+PBS + PBS group, the RPS27-RPS24 + PBS + PBS group of mice
   exhibited faster tumor growth, a significant increase in tumor weight,
   and a significant increase in Ki67 protein positivity. In comparison to
   the RPS27-RPS24 + PBS + PBS group, the tumor growth rate, tumor weight,
   and Ki67 protein positivity were significantly reduced in the
   RPS27-RPS24 + CB-839 + PBS group. Moreover, compared to the
   RPS27-RPS24 + CB-839 + PBS group, the RPS27-RPS24 + CB-839 + TTM group
   showed faster tumor growth, increased tumor weight, and a significant
   increase in Ki67 protein positivity (Figs. [166]7F, J and S[167]9E–I).
   When compared to the PBS treatment group, all the groups treated with
   ADM showed slower tumor growth, lighter tumor weight, and lower Ki67
   protein positivity, with similar changes as observed in the PBS group
   (Figs. [168]7F, J and S[169]9E–I). These results demonstrate that
   RPS27-RPS24 overexpression can promote tumor growth and chemoresistance
   in OS cells. The glutamine metabolism inhibitor can reverse the
   promoting effect of RPS27-RPS24 overexpression on tumor growth and
   chemoresistance, while the addition of the copper death inhibitor can
   further reverse the effect of the glutamine metabolism inhibitor.

   In summary, RPS27-RPS24 overexpression can promote glutamine
   metabolism, inhibit copper death, and thereby promote tumor growth and
   chemoresistance in OS cells.

Discussion

   One of the significant challenges in OS is its resistance to
   chemotherapy, which has been widely documented [[170]39, [171]40]. The
   development of chemoresistance is a major cause of treatment failure
   and recurrence in tumors [[172]41, [173]42]. Therefore, investigating
   the molecular mechanisms underlying chemoresistance is of great
   significance in enhancing OS treatment efficacy [[174]43, [175]44]. The
   aim of this study is to explore the molecular mechanisms of a novel
   fusion gene, RPS27-RPS24, in chemoresistance, providing new insights
   and approaches for the prevention and treatment of chemoresistance in
   OS.

   To date, several key genes related to chemoresistance have been
   identified in previous studies [[176]45–[177]47]. However, in
   comparison to previous research, our study discovered a new fusion
   gene, RPS27-RPS24. Comparative analysis revealed differences in the
   mechanism of action of RPS27-RPS24 in chemoresistance in OS when
   compared to previously discovered key genes. Furthermore, we have
   utilized data from previous studies to support and compare our study
   findings.

   In previous research, various metabolic regulation mechanisms and
   signaling pathways related to chemoresistance in OS have been described
   [[178]48, [179]49]. However, in this study, we unveil new findings
   regarding the role of RPS27-RPS24 in GLS-mediated glutamine metabolism
   regulation and copper death. These novel discoveries provide important
   clues and new research directions for further understanding the
   mechanisms of chemoresistance in OS.

   There are differences in experimental methods and techniques between
   previous studies and this study. Compared to previous research, this
   study utilizes more advanced and precise experimental methods to
   comprehensively and accurately analyze the regulatory role of
   RPS27-RPS24 in chemoresistance in OS. When analyzing the consistency
   and differences in research results, we found that the novelty of this
   study lies in uncovering a novel regulatory mechanism of RPS27-RPS24 in
   chemoresistance.

   In this study, fusion genes related to OS prognosis and chemoresistance
   were screened using single-cell transcriptome data. Downstream key
   proteins and metabolic pathways were identified through proteomics and
   metabolomics techniques. By integrating in vivo and in vitro
   experimental results, we have come to the following conclusions: the
   novel fusion gene RPS27-RPS24 promotes GLS-mediated glutamine
   metabolism, inhibits copper death, and ultimately enhances OS cell
   chemoresistance, thereby promoting the progression of OS. Comparing and
   evaluating the results with previous studies, we found some
   similarities with previous research and also revealed some
   inconsistencies. These findings are important not only for the
   understanding of OS development and treatment but also for providing
   new ideas and directions for the development of new therapeutic
   strategies and drug targets.

   Our study demonstrates that RPS27-RPS24 influences copper-induced cell
   death mechanisms by promoting GLS-mediated glutamine metabolism. Copper
   concentration within the cell is closely associated with mitochondrial
   function. Elevated copper ion levels may alter mitochondrial membrane
   potential, which is often linked to cell death processes [[180]50].
   Specifically, copper-induced cell death may lead to oxidative stress
   and mitochondrial damage, causing a loss of mitochondrial membrane
   potential [[181]51]. Upregulation of GLS could further affect
   mitochondrial health by modulating energy metabolism, influencing cell
   survival and death pathways. Based on our findings, RPS27-RPS24 may
   play a crucial role in copper-induced cell death through glutamine
   metabolism regulation, linking this process to the maintenance of
   mitochondrial membrane potential. This discovery suggests that
   exploring the relationship between copper ions and cell metabolism
   could offer new approaches to overcoming OS chemoresistance.

   In this study, our focus was on the role of RPS27 in cell metabolism,
   not directly on its relationship with autophagy or mitophagy. However,
   existing research highlights that autophagy is vital for cellular
   homeostasis and metabolic regulation, with RPS27 potentially playing a
   significant role in this process [[182]52]. For instance, RPS27
   upregulation may influence autophagy through the mTOR pathway,
   promoting cell survival under stress [[183]53]. Mitophagy, a specific
   form of autophagy, is essential for removing damaged mitochondria to
   maintain cellular energy balance and function. Studies suggest that
   RPS27 might indirectly affect mitophagy by regulating the cellular
   response to oxidative stress [[184]54]. Future research should explore
   the connections between RPS27, autophagy, and mitophagy to
   comprehensively understand its biological functions.

   We also investigated the combined effects of ADM and other treatments
   on OS cell lines U2OS and MG-63, focusing on copper-induced cell death
   and ER stress. Our experiments reveal that the RPS27-RPS24 fusion gene
   plays a critical role in regulating glutamine metabolism and
   copper-induced cell death. Specifically, ADM treatment followed by GLS
   knockdown and RPS27-RPS24 overexpression experiments demonstrated that
   the copper death inhibitor TTM significantly weakened the cells’
   response to ADM. This indicates that inhibiting copper-induced cell
   death may be a key mechanism of chemoresistance in OS cells. The
   RPS27-RPS24 fusion gene enhances resistance to copper-induced cell
   death through GLS-mediated glutamine metabolism, affecting the efficacy
   of chemotherapy. These findings align with current literature showing
   that glutamine metabolism is crucial for tumor cell survival and
   resistance [[185]55]. Additionally, our results suggest that the
   RPS27-RPS24 fusion gene enhances resistance to ER stress, indicating
   that it may regulate ER stress responses and influence chemotherapy
   sensitivity. ER stress has been shown to play a vital role in tumor
   cell survival and resistance [[186]56]. Thus, the role of RPS27-RPS24
   in OS chemoresistance involves complex interactions between
   copper-induced cell death and ER stress regulation. We plan to further
   investigate these mechanisms in future studies to reveal potential
   therapeutic applications for RPS27-RPS24 in OS treatment.

   Although we have not specifically identified the post-translational
   modifications (PTMs) of the RPS27-RPS24 fusion protein in this study,
   the literature indicates that PTMs are crucial in regulating protein
   function, stability, and interactions [[187]57]. Known PTMs include
   phosphorylation, ubiquitination, acetylation, and methylation, which
   may influence the biological function of RPS27-RPS24 [[188]58].
   Phosphorylation, for example, plays a key role in cell signaling and
   protein activity regulation [[189]59], while ubiquitination is closely
   related to protein degradation and cell cycle control [[190]60].
   Understanding how these PTMs affect the fusion protein’s regulation of
   GLS-mediated glutamine metabolism and copper-induced cell death will be
   essential for elucidating its role in chemoresistance. Future studies
   should focus on identifying these PTMs and investigating their effects
   on the fusion protein’s functions and its role in cancer cell survival
   and stress responses. Moreover, the RPS27-RPS24 fusion protein may
   participate in forming multiprotein complexes, interacting with other
   ribosomal and signaling proteins to regulate metabolic pathways and
   chemoresistance. Exploring these interactions will further our
   understanding of RPS27-RPS24’s role in tumor biology.

   While important discoveries have been made in this study, there are
   limitations in the methods and data. Due to the limited sample size,
   our research results may have certain biases. Additionally, we face
   technical and time limitations. To further validate and strengthen the
   results of this study, we will focus on expanding the sample size,
   adopting more advanced experimental techniques, and improving the
   accuracy of data analysis.

   In summary, this study has discovered the regulatory role of the novel
   fusion gene RPS27-RPS24 in chemoresistance in OS and revealed its
   specific molecular mechanism. This discovery deepens our understanding
   of the mechanisms underlying chemoresistance in OS and has significant
   scientific value. These results provide new targets and strategies for
   OS treatment and offer new directions and potential for precision
   medicine development.

   The results of this study have important clinical applications,
   especially in chemoresistance reversal and precision treatment. Based
   on a deep understanding of RPS27-RPS24 regulatory mechanisms, we can
   develop new treatment strategies and drug designs targeting this fusion
   gene. However, further research and challenges are still needed in the
   future to overcome the limitations of this study and to enhance
   improvements and developments in OS treatment. We will commit to
   conducting more clinical trials and in-depth research to fully
   understand the clinical potential of RPS27-RPS24 in OS and to explore
   more targeted treatment methods and strategies to provide patients with
   more effective and personalized treatment options.

Materials and methods

Ethical statement

   All experiments involving mice were approved by Jiangxi Cancer Hospital
   (2024ky058).

Public data download

   OS-related scRNA-seq data [191]GSE162454 and transcriptome RNA
   sequencing (RNA-seq) data [192]GSE154540, [193]GSE39055 were obtained
   from the Gene Expression Omnibus (GEO,
   [194]https://www.ncbi.nlm.nih.gov/geo/). The [195]GSE162454 dataset
   includes tumor tissue samples from 6 OS patients. The [196]GSE154540
   dataset contains tumor tissue samples from 29 chemotherapy-sensitive OS
   patients and 21 chemoresistance OS patients. The [197]GSE39055 dataset
   consists of tumor tissue samples from 37 OS patients. In addition,
   RNA-seq data (TARGET-OS) of tumor tissue from 96 OS patients, along
   with clinical information, were downloaded from the TARGET database
   ([198]https://ocg.cancer.gov/programs/target) (Table S[199]1)
   [[200]61]. Since these data are sourced from public databases, ethical
   committee approval is not required.

scRNA-seq data analysis

   The scRNA-seq data ([201]GSE162454) was analyzed using the R software
   package Seurat [[202]62]. First, the data was subjected to quality
   control based on the following criteria: 200 < nFeature_RNA < 5000,
   200 < nCount_RNA < 20,000, and percent.mt < 25. Next, batch effects
   were eliminated using the CCA method. The data, which was normalized
   using the LogNormalize function, underwent principal component analysis
   (PCA) using the RunPCA function. The most significant principal
   components were selected for UMAP clustering analysis through the use
   of the JackStrawPlot and ElbowPlot functions. CellMarker database
   ([203]http://bio-bigdata.hrbmu.edu.cn/CellMarker/) was used for cell
   annotation [[204]63] by obtaining marker genes for each cell cluster
   with the FindAllMarkers function. DotPlot and VlnPlot functions were
   employed to visualize the expression of marker genes in different cell
   clusters. Additionally, putative malignant cells were identified using
   the subset function. These cells underwent copy number variation (CNV)
   analysis using the inferCNV package in R. K-means algorithm was
   utilized to exclude non-malignant cells, resulting in the
   identification of marker genes specific to malignant cells.

Bioinformatics analysis

   The gene expression clustering analysis and phenotype association
   analysis of genes in the [205]GSE154540 dataset were performed using
   the R software package WGCNA, known as weighted gene co-expression
   network analysis. This analysis aimed to identify important gene
   modules related to chemoresistance in OS and extract the genes within
   those modules [[206]64].

   The intersection of the module genes and malignant OS cell marker genes
   was obtained through the jvenn website. Subsequently, single-gene
   survival analysis was conducted on the intersection genes based on the
   TARGET-OS data, with overall survival as the predicted parameter. The
   survival package in R was used to perform the proportional hazards
   assumption test and to establish a fitted survival regression model.
   The results were visualized using the survminer and ggplot2 packages in
   R. Log-rank tests were employed to display survival curves and compare
   the overall survival differences between the high-expression group and
   the low-expression group. A significance level of P < 0.05 was
   considered [[207]65]. Additionally, the candidate genes were subjected
   to receiver operating characteristic (ROC) analysis using the R package
   pROC, based on the GEO dataset [208]GSE39055, to evaluate their
   predictive value in OS patient mortality and tumor recurrence.

   FusionGDB2 ([209]https://compbio.uth.edu/FusionGDB2/index.html),
   ChimerDB 3.0 ([210]https://www.kobic.re.kr/chimerdbv3/mindex.cdb), and
   ChiTaRS 5.0 ([211]http://biosrv.org/chmb/search) databases were
   utilized to retrieve fusion gene pairs related to RPS27
   [[212]66–[213]68]. Moreover, the ChimerDB 3.0 database was used to
   obtain the gene fusion interaction network involving RPS27 and RPS24.
   Subsequently, based on the TARGET-OS data, R software package corplot
   was employed to analyze the correlation between the expression levels
   of RPS27 and RPS24.

Cell culture

   The human OS cell lines U2OS (HTB-96) and MG-63 (CRL-1427) were
   obtained from ATCC (USA). U2OS and MG-63 cells were cultured in McCoy’s
   5A medium (16600082, Gibco, USA) supplemented with 10% FBS (10099141,
   Gibco, USA), 10 μg/ml streptomycin, and 100 U/ml penicillin (15140148,
   Gibco, USA).

   The 293 T cell line was purchased from ATCC (CRL-3216) and cultured in
   DMEM medium (11965092, Gibco, USA) supplemented with 10% FBS, 10 μg/ml
   streptomycin, and 100 U/ml penicillin. All cells were cultured in a
   humidified incubator at 37 °C with 5% CO[2] (Heracell™ Vios 160i CR
   CO[2] incubator, 51033770, Thermo Scientific™, Germany). When the cells
   reached 80–90% confluency, subculturing was performed [[214]69].

   To establish U2OS/adriamycin-resistant (ADMR) and MG63/ADMR cells, U2OS
   and MG63 cells were cultured in a medium supplemented with ADM (D5220,
   Sigma-Aldrich, USA) at a concentration of 1 µg/ml for 6 months
   [[215]70].

Construction and transfection of fusion genes

   Specific primers were designed to amplify the corresponding regions of
   RPS27 and RPS24 genes. The PCR products were purified, and the digested
   fragments of RPS27 and RPS24 were obtained using the respective
   restriction enzymes. The digested fragments of RPS27 and RPS24 were
   ligated together to form the fusion gene RPS27-RPS24. The fusion gene
   was then inserted into the mammalian expression vector pCDNA3.1 to
   generate the recombinant expression vector. Finally, the bacterial
   transformation and selection of resistant clones were conducted,
   followed by sequencing validation to confirm the correct expression
   vector of the RPS27-RPS24 fusion gene [[216]13].

   The RPS27-RPS24 coding sequences were cloned into the
   pAd-DsRed-IRES-EGFP adenovirus reporter vector. 293 T cells were
   cultured in a 6-well plate for 2–3 weeks and infected with the
   amplified adenovirus. After the production of adenovirus, the lysates
   were harvested. The constructed RPS27-RPS24 fusion gene adenovirus
   (MOI = 10, virus titer 1 × 10^8 TU/mL) was then transfected into
   appropriate U2OS/ADMR and MG63/ADMR cells. The infected cells were
   screened with puromycin (2 μg/ml, HY-K1057, Med Chem Express, USA) for
   7 days and validated through protein-level experiments [[217]12].

Validation of the Binding of RPS27 and RPS27-RPS24 through
Co-immunoprecipitation (Co-IP)

   U2OS/ADMR or MG-63/ADMR cells were seeded in 6 cm culture dishes and
   treated with 10 μM proteasome inhibitor MG132 (HY-13259, Med Chem
   Express, USA) for 6 h. The cells were then collected and lysed using
   NP-40 lysis buffer (P0013F, Bioteke Corporation, Shanghai, China).
   Forty micrograms of total protein were prepared as the input group, and
   the remaining protein was adjusted to a concentration of 1 mg/mL in
   1 mL and divided into three tubes. Anti-RPS27 antibody (10 μg, 114838,
   Beijing NuoPu Bio-Technology Co., Ltd., China) or anti-Flag antibody
   (ab205606, Abcam, UK) or IgG (ab6715, 1:1000, Abcam, UK) antibody
   (10 μg) was added to the total protein and incubated with slow
   agitation overnight at 4 °C. Protein A/G agarose (sc-2003, Santa Cruz
   Biotechnology, USA) was added and incubated for 4 hours at 4 °C,
   followed by three washes with pre-chilled TBS solution.
   Immunoprecipitated proteins were detected by protein immunoblot
   analysis [[218]71, [219]72].

Protein sample preparation and assessment

   U2OS/ADMR cells overexpressing Stable Trans Vector (N = 3) or
   RPS27-RPS24 (N = 3) were transferred to 5 cm^3 centrifuge tubes.
   Subsequently, ultrasonication under ice bath conditions was performed
   using a SCIENTZ-IID ultrasonic cell disruptor (Scientz, Ningbo, China)
   with a phenol extraction buffer containing 10 mM DTT (R0861, Solarbio,
   Beijing, China), 1% protease inhibitor cocktail (P6731, Solarbio,
   Beijing, China) and 2 mM EDTA (E1170, Solarbio, Beijing, China). This
   step was repeated 8 times. Next, an equal volume of Tris-saturated
   phenol (pH 8.0) (HC1380, Vander Biological, Beijing, China) was added,
   and the mixture was vortexed for 4 min. Subsequently, centrifugation at
   5000×g for 10 minutes at 4 °C was performed to transfer the upper phase
   of phenol to a new centrifuge tube. The phenol solution was
   supplemented with 0.1 M ammonium sulfate (101217, Merck, USA) and
   saturated methanol (106035, Merck, USA) in a 1:5 volume ratio and left
   overnight to precipitate the proteins. Afterward, centrifugation at
   4 °C for 10 min was carried out to remove the supernatant. Finally, the
   remaining precipitate was washed once with cold methanol and three
   times with cold acetone. The washed protein was then dissolved in 8 M
   urea (U8020, Solarbio, Beijing, China), and its concentration was
   determined using the BCA assay kit (P0012, BiyunTian, Shanghai, China)
   according to the manufacturer’s instructions [[220]73].

Protease digestion, peptide labeling, and nano-LC–MS/MS analysis

   For each sample, 50 µg of protein was subjected to enzymatic digestion.
   The protein solution was mixed with DTT to a final concentration of
   5 mM and incubated at 56 °C for 30 min. Then, acetonitrile was added to
   achieve a concentration of 11 mM, and the mixture was incubated at room
   temperature for 15 minutes. Finally, the urea concentration of the
   sample was diluted to less than 2 M, and trypsin (25200056, Thermo
   Fisher Scientific, USA) was added at a 1:50 (w/w) ratio to the protein
   and incubated overnight at 37 °C. Subsequently, trypsin digestion
   continued for an additional 4 h at a 1:100 (trypsin:protein) ratio.

   Following trypsin digestion, peptide desalting was performed using a
   HyperSep™ C18 purification column (60108-302, Thermo Fisher Scientific,
   USA), followed by vacuum drying. The peptides were then reconstituted
   in 0.5 M TEAB (90114, Thermo Fisher Scientific, USA) and processed
   according to the manufacturer’s instructions for the TMT reagent kit
   (90064CH, Thermo Fisher Scientific, USA). In brief, one unit of the TMT
   reagent was thawed and reconstituted in acetonitrile (113212, Merck,
   USA). The peptide mixture was incubated at room temperature for 2 h,
   followed by desalting and drying using a vacuum centrifugal
   concentrator. Equal amounts of each labeled peptide fraction from
   different groups were combined, and the dried peptides were resuspended
   in 0.1% formic acid (159002, Merck, USA).

   For nano-LC separation, 2 µg of peptides from each sample were loaded
   onto an Easy nLC 1200 nano-UPLC system (Thermo Fisher Scientific, USA).
   The sample was first trapped onto a Trap C18 column (100 µm × 20 mm,
   5 µm) and then separated on a C18 analytical column (75 µm × 150 mm,
   3 µm) using a gradient elution at a flow rate of 300 nL/min. The mobile
   phase included 0.1% formic acid in water as solvent A and 0.1% formic
   acid in water-acetonitrile (containing 95% acetonitrile) as solvent B.
   The gradient elution program was as follows: 0 → 2 min, 2% → 8% B;
   2 → 71 min, 8% → 28% B; 71 → 79 min, 28% → 40% B; 79 → 81 min,
   40% → 100% B; 81 → 90 min, 100% B. The separated peptides were then
   subjected to mass spectrometric analysis on a Q-Exactive HFX mass
   spectrometer (Thermo Fisher Scientific, USA). The analysis lasted for
   60 min with an electrospray voltage of 2.1 kV. The detection mode was
   positive ion mode with a survey scan range of 350–1200m/z. The
   resolution of the primary mass spectrometry was set at 60,000 (FWHM at
   m/z 200) with an AGC target of 3e6 and a maximum injection time of
   30 ms. The resolution of the secondary mass spectrometry was set at
   15,000 (FWHM at m/z 200) with an AGC target of 1e6 and a maximum
   injection time of 25 ms. HCD was used as the activation type for MS2,
   with an isolation window of 20 Th and a normalized collision energy of
   32 [[221]73, [222]74].

Proteomic data analysis

   The LC–MS/MS data obtained were processed using MaxQuant software
   (v.1.5.2.8), which included peptide identification and protein
   quantification. For tandem mass spectrometry searching, UniProt 14.1
   (2009)—Gossypium hirsutum database and the reversed decoy database were
   used. Trypsin was selected as the cleavage enzyme, with up to 2 missed
   cleavages allowed. The initial search used a mass tolerance of 20 ppm,
   followed by a main search using 5 ppm and a fragment ion mass tolerance
   of 0.02 Da. Peptide false discovery rate (FDR) ≤ 0.01, protein
   FDR ≤ 0.01, and peptide score distribution were used as filtering
   criteria for database searching [[223]73, [224]75].

   To identify differentially expressed proteins (DEPs) between the Vector
   and RPS27-RPS24 groups, we employed the R package limma. The criteria
   for differential expression were set as |log2FC | > 1 and P < 0.05. The
   identified DEPs were subjected to the STRING database
   ([225]https://string-db.org/), with the species limited to “Homo
   sapiens”. A minimum required interaction score of ≥0.4 was set as the
   threshold to construct a protein-protein interaction network, which was
   visualized using Cytoscape software (v3.8.2) [[226]76].

   To investigate the functions of the DEPs further, we utilized the R
   package clusterProfiler and performed enrichment analysis for gene
   ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG), and
   WikiPathways. The significance threshold was set at P < 0.05. The GO
   terms included biological processes (BP), cellular components (CC), and
   molecular function (MF) [[227]77].

Metabolomic analysis

   Metabolomic analysis was conducted using the LC20
   ultra-high-performance liquid chromatography system (Shimadzu, Japan)
   coupled with a Triple TOF-6600 mass spectrometer (AB Sciex). U2OS/ADMR
   cell samples from the stable Vector group (N = 6) and RPS27-RPS24 group
   (N = 6) were targeted for analysis. Chromatographic separation was
   performed using a Waters ACQUITY UPLC HSS T3 C18 column (100 × 2.1 mm,
   1.8 μm). The column temperature was maintained at 40 °C, and the flow
   rate was set at 0.4 mL/min. The mobile phase consisted of an
   acetonitrile-water solution containing 0.1% formic acid. The gradient
   elution program for mobile phase B was as follows: 5%, 0.0–11.0 min;
   90%, 11.0–12.0 min; 5%, 12.1–14 min. The eluent was directly introduced
   to the mass spectrometer without splitting [[228]78].

   For positive/negative ion mode mass spectrometry, the conditions were
   set as follows: ionization voltage of 5500 V, capillary temperature of
   550 °C, spray gas flow rate of 50 psi, and auxiliary heating gas flow
   rate of 60 psi. To prevent overfitting, the preprocessed data were
   analyzed using orthogonal partial least-squares discriminant analysis
   (OPLS-DA) and permutation tests (100 permutations). Metabolites with
   VIP scores > 1 and P < 0.05 in the OPLS-DA model were identified as
   differential metabolites (DMs). Additionally, differential metabolites
   were selected based on the combination of multivariate analysis with
   fold change ≥ 2 and ≤0.5 and univariate analysis with P < 0.05 in the
   Student’s t-test. Metabolic pathways related to the identified
   metabolites were determined using the MetaboAnalyst software (v5.0)
   [[229]79].

CRISPR/Cas9 editing technology

   The sgRNA used to construct GLS-KO cells through CRISPR/Cas9 technology
   is as follows: GLS-sgRNA: F: 5’-TCCTTCAGCTCCTGCTG-3’ (PAM: AGG); R:
   5’-TGACAATAATAGTTTCAACA-3’ (PAM: TGG). The sgRNA was inserted into the
   Lenti-CRISPR v2 vector (obtained from Hanheng Biotech, Shanghai, China)
   containing the Streptococcus pyogenes Cas9 nuclease gene. Cells were
   transduced with the lentiviral Lenti-CRISPR v2 vector and subsequently
   subjected to a CRISPR/Cas9 editing system to generate GLS-KO cells.
   Transfected cells with sgRNA plasmid and donor sequence were selected
   using 4 μg/mL puromycin. Surviving cells were screened by limiting
   dilution cloning and confirmed as GLS-KO cells through RT-qPCR and
   Western blot analysis [[230]80].

RT-qPCR

   Total RNA was extracted using the Trizol reagent kit (T9424,
   Sigma-Aldrich, USA), and the quality and concentration of the RNA were
   assessed using UV–visible spectrophotometry (ND-1000, Nanodrop, Thermo
   Fisher, USA). Reverse transcription was performed using the
   PrimeScript™ RT-qPCR kit (RR086A, TaKaRa, Mountain View, CA, USA).
   Real-time quantitative reverse transcription polymerase chain reaction
   (RT-qPCR) was conducted using the SYBR Green PCR Master Mix (4364344,
   Applied Biosystems, USA) and ABI PRISM 7500 Sequence Detection System
   (Applied Biosystems, USA). GAPDH was used as an endogenous control for
   mRNA. Primer design was provided by Shanghai General Biotechnology Co.,
   Ltd. (Table S[231]2 for details). The fold change in expression of the
   target gene between the experimental and control groups was determined
   using the formula 2^−ΔΔCt (ΔΔCT = ΔCt experimental group−ΔCt control
   group, where ΔCt = target gene Ct−endogenous control Ct) [[232]81].

Western blot

   First, tissues or cells were collected and lysed using an enhanced RIPA
   lysis buffer containing a protease inhibitor (P0013B, Beyotime,
   Shanghai, China). Protein concentration was determined using the BCA
   protein quantification kit (P0012, Beyotime, Shanghai, China). The
   proteins were separated by 10% SDS-PAGE and transferred onto a PVDF
   membrane. The membrane was then blocked with 5% BSA at room temperature
   for 2 h to prevent nonspecific binding. Diluted primary antibodies
   (Table S[233]3 for details) were added and incubated at room
   temperature for 1 h. All primary antibodies were rabbit anti-human
   antibodies. After washing the membrane, HRP-conjugated secondary
   antibodies, goat anti-rabbit IgG (ab6721, 1:2000, Abcam, UK) or goat
   anti-mouse IgG (ab6785, 1:1000, Abcam, UK), were added and incubated at
   room temperature for 1 h. An equal amount of A and B reagents from
   Pierce™ ECL Western blot substrate (32209, Thermo Scientific™, Germany)
   were mixed and dropped onto the membrane, followed by exposure using a
   gel imaging system [[234]82]. The Western blot images were captured
   using the Bio-Rad imaging system (BIO-RAD, USA), and the band
   intensities were quantified using Image J analysis software, with GAPDH
   as the endogenous control. Each experiment was repeated three times.
   The Full and uncropped western blots can be found at the supplementary
   materials.

Cell grouping

   The U2OS/ADMR and MG63/ADMR cell groups were organized as follows: (1)
   GLS-WT group (control cells), further divided into the Vector group
   (stable Vector transfected control cells), and the RPS27-RPS24 group
   (stable RPS27-RPS24 transfected control cells); (2) GLS-KO group (GLS
   knockout cells), further divided into the Vector group (stable Vector
   transfected GLS knockout cells) and the RPS27-RPS24 group (stable
   RPS27-RPS24 transfected GLS knockout cells); (3)
   GLS-KO + RPS27-RPS24 + PBS group (GLS knockout cells with stable
   RPS27-RPS24 transfection, along with the addition of PBS); (4)
   GLS-KO + RPS27-RPS24 + TTM group (GLS knockout cells with stable
   RPS27-RPS24 transfection, along with the addition of the copper death
   inhibitor tetrathiomolybdate, TTM); (5) RPS27-RPS24 + PBS group (stable
   RPS27-RPS24 transfected cells with the addition of PBS); (6)
   RPS27-RPS24 + CB-839 group (stable RPS27-RPS24 transfected cells with
   the addition of the glutamine metabolism inhibitor CB-839). CB-839
   (catalog number: HY-12248, 1 μM, treated for 72 h) and TTM (catalog
   number: HY-128530, 20 μM, treated for 12 h) were purchased from Med
   Chem Express [[235]83]. In addition, for copper death-related testing,
   we treated the above cell groups with different concentrations of the
   copper ion carrier elesclomol–Cu (ratio = 1:1) for 72 h [[236]84].
   Afterward, we used the CCK-8 assay to measure cell viability, the
   scratch test to assess cell migration, and the Transwell assay to
   evaluate cell invasion. The elesclomol compound (catalog number:
   HY-12040) was purchased from Med Chem Express.

MTT and CCK-8 assays for cell viability

   MTT Assay: The test cells were seeded in a 96-well cell culture plate
   at a density of 3–5 × 10^4 cells/mL and incubated for 48 h. MTT
   solution (CT02, Sigma Aldrich, USA) was added to the cell suspension
   and incubated for 4 h. Then, dimethyl sulfoxide (DMSO) was added, and
   the mixture was shaken for 10 min. The absorbance (OD 490 nm) was
   measured using a spectrophotometer (Laspec, China). The resistance to
   ADM was determined as the half inhibitory concentration (IC50) of ADM
   [[237]85].

   CCK-8 Assay: Each cell group was digested and resuspended, and the cell
   density was adjusted to 1 × 10^5 cells/mL. Next, 100 μL/well of the
   cell suspension was seeded in a 96-well plate for routine culture.
   After the cells adhered to the well surface, drug treatment was
   applied, followed by overnight incubation. On the first and fourth days
   after incubation, the cell viability was assessed using the
   instructions provided with the CCK-8 assay kit (C0041, Beyotime,
   Shanghai, China). In each assessment, 10 μL of CCK-8 detection solution
   was added to each well and incubated for 4 h in a cell culture
   incubator. Subsequently, the absorbance at 450 nm was measured using an
   enzyme-linked immunosorbent assay reader, and the cell survival rate
   was calculated [[238]86].

Clonal formation assay for evaluating the capacity of cell cloning

   Cells in the exponential growth phase were selected for the experiment.
   Conventional digestion and passage were used to prepare cell
   suspension, ensuring a single-cell rate of more than 95%. Subsequently,
   the cells were counted and diluted to an appropriate concentration.
   Each plate was seeded with 5 mL of cell suspension to achieve a uniform
   distribution of cells in a 60 mm culture dish. Gentle shaking in a
   crosswise direction was performed to facilitate dispersion. The culture
   dishes were then incubated at 37 °C with 5% CO[2] for 2–3 weeks. Once
   visible clones appeared in the dishes, the cultivation was terminated.
   After discarding the culture medium, the dishes were carefully washed
   twice with PBS and air-dried. Samples were fixed with methanol for
   15 min, followed by removal of methanol and air-drying. Subsequently,
   the samples were stained with Giemsa solution for 10 minutes, gently
   rinsed with running water, and air-dried again. Finally, the clones
   were visually counted using naked eyes or a low-power microscope,
   ensuring a count of more than 10 clones. The clonal formation rate was
   calculated using the following formula: Clonal formation rate = (number
   of clones/number of seeded cells) × 100% [[239]87].

EdU assay for cell proliferation

   The test cells were seeded in a 24-well plate, and the culture medium
   was supplemented with EdU (C10310-2, Guangzhou Libo Biological
   Technology Co., Ltd., Guangzhou) to achieve a concentration of
   10 µmol/L. After incubating in the CO[2] incubator for 2 h, the culture
   medium was aspirated, and the cells were fixed with a PBS solution
   containing 4% paraformaldehyde at room temperature for 15 minutes.
   Subsequently, the cells were washed twice with PBS containing 3% BSA.
   Then, PBS containing 0.5% Triton-100 was added to each well and
   incubated at room temperature for 20 min, followed by two washes with
   PBS containing 3% BSA. Each well was then treated with 100 µL of
   staining solution and incubated in the dark at room temperature for
   30 minutes. Afterward, DAPI was added for nuclear staining for
   5 minutes. The samples were finally mounted on slides, and 6–10 random
   fields were observed under a fluorescence microscope (BX63, Olympus,
   Japan). The number of positive cells in each field was recorded. The
   EdU labeling rate (%) was calculated as the number of positive cells
   divided by the sum of positive and negative cells multiplied by 100%
   [[240]88]. Each experiment was repeated three times.

Scratch experiment to assess cell migration ability

   Uniform horizontal lines were drawn on the back of a 6six-well plate
   using a marker, with each line spaced 1 cm apart, crossing through the
   wells. Approximately 5 × 10^5 cells were seeded in each well and
   cultured until they reached a confluence of 100%. Subsequently,
   scratches were made perpendicular to the plane of the cells along the
   lines on the cell layer on the back of the plate. After completing the
   scratches, the cells were washed three times with sterile PBS to remove
   non-adherent cells, making the gaps created by the scratches clearly
   visible. Fresh serum-free culture medium was then replaced, and the
   cells were placed in a 37 °C, 5% CO[2] incubator for static culture.
   After 24 h, the width of the scratches was observed and measured under
   a microscope, and the results were recorded by taking photographs
   [[241]89]. Finally, migration rate analysis was performed using Image J
   software.

Transwell assay for cell invasion

   Cell invasion was evaluated using Transwell chambers with 8 µm pore
   size (3391, Corning, USA). For the invasion assay, Matrigel (354277, BD
   Biosciences, USA) was added to each well according to the
   manufacturer’s instructions, and cells from each group (2 × 10^5 cells)
   were added to the serum-free culture medium and placed in the upper
   chamber of the Transwell. After incubating at 37 °C for 24 h, cells
   that had migrated were stained with 0.5% crystal violet at room
   temperature for 20 min. Using a Nikon Eclipse Ci optical microscope
   (Nikon, Tokyo, Japan), five random areas (top, bottom, center, left,
   and right) were selected for observation and photography. The number of
   positively stained cells in the five fields presenting a purple color
   was then counted and subjected to statistical analysis [[242]90].

In vivo animal experiment

   Male NOD/SCID mice aged 6–8 weeks and weighing between 20 and 30 g were
   purchased from Hunan Slacc Jingda Experimental Animals Co., Ltd.,
   China. The mice were housed in specific pathogen-free (SPF) animal
   facilities with a humidity range of 60–65%, temperature maintained at
   22–25 °C, and provided ad libitum access to food and water under a 12-h
   light–dark cycle. After one week of acclimatization, the mice’s health
   status was observed before initiating the experiment. All animal
   procedures followed the guidelines for the care and use of laboratory
   animals [[243]91].

   A total of 5 × 10^6 pretreated, co-cultured U2OS/ADMR or MG-63/ADMR
   cells, stably transfected with Vector or RPS27-RPS24, were injected
   subcutaneously in the left axillary region of the mice. Tumor growth
   was monitored and photographed regularly. Tumor volume was measured
   every other day, and once the subcutaneous tumor reached a diameter of
   approximately 3 mm, the mice received intraperitoneal injections of ADM
   twice a week at a dose of 5 mg/kg. Two weeks after drug injection, the
   fluorescent signals of the implanted pretreated OS cell lines were
   analyzed using the CRi Maestro in vivo imaging system (Cambridge
   Research & Instrumentation, Massachusetts, USA) [[244]92]. On day 28,
   after the drug injection, the mice were euthanized with CO[2]
   inhalation, and the tumors were excised and weighed. The collected
   tumor tissue was used for Western blot and immunohistochemical
   analysis.

   The mice were randomly divided into 16 groups with 6 mice in each
   group: (1) Vector + PBS + PBS group: U2OS/ADMR or MG63/ADMR cells
   transfected with Vector were injected, followed by PBS injections; (2)
   RPS27-RPS24 + PBS + PBS group: U2OS/ADMR or MG63/ADMR cells transfected
   with RPS27-RPS24 were injected, followed by PBS injections; (3)
   RPS27-RPS24 + CB-839 + PBS group: U2OS/ADMR or MG63/ADMR cells
   transfected with RPS27-RPS24 were injected, followed by CB-839 and PBS
   injections of equal volume; (4) RPS27-RPS24 + CB-839 + TTM group:
   U2OS/ADMR or MG63/ADMR cells transfected with RPS27-RPS24 were
   injected, followed by oral administration of CB-839 (200 mg/kg, twice
   daily) or TTM (0.2 mg/day, once daily) [[245]83, [246]93]. In each
   group, mice were treated with PBS or ADM.

Immunohistochemical staining

   Immunohistochemical (IHC) staining is a technique used to detect the
   expression of the Ki67 protein in tumor tissue. The specific steps are
   as follows: Firstly, subcutaneous transplant tumor tissue from mice is
   fixed overnight with 4% paraformaldehyde. It is then dehydrated,
   embedded, and prepared into 4 μm-thick paraffin sections. Next, the
   sections are deparaffinized with xylene and hydrated with a gradient of
   alcohol (including anhydrous ethanol, 95% ethanol, and 75% ethanol,
   each for 3 min). Subsequently, the sections are boiled in 0.01 M
   citrate buffer for 15–20 min for antigen retrieval. After that, they
   are incubated with 3% H[2]O[2] at room temperature for 30 min to
   inactivate endogenous peroxidase. Then, goat serum-blocking solution is
   added and left for 20 min to remove excess fluid. Subsequently, a
   primary antibody against Ki67 (ab16667, 1:200, Abcam, UK) is added and
   incubated at room temperature for 1 h, followed by washing with PBS.
   Next, the sections are incubated with a secondary antibody (ab150077,
   1:1000, Abcam, UK) at 37 °C for 20 min, then washed with PBS and
   treated with streptavidin-peroxidase (SP) at 37 °C for 30 min, followed
   by PBS washing. Then, DAB (P0202, Shanghai, China) from the Bi Yun Tian
   company is dropped on the sections, and after 5–10 min of staining, the
   reaction is stopped by washing with water for 10 min. Subsequently, the
   sections are counterstained with hematoxylin (C0107, Shanghai, China)
   from the Bi Yun Tian company for 2 min, followed by differentiation
   with hydrochloric acid ethanol and a final water rinse for 10 min. The
   sections are then dehydrated with gradient alcohol (using xylene for
   transparency) and finally mounted with 2–3 drops of neutral resin.
   Finally, the results are observed, and the following statistical
   analysis is performed under a light microscope: five high-power fields
   are randomly selected from each section, with 100 cells observed in
   each field, and the ratio of Ki67-positive cells is calculated
   [[247]94].

Statistical software and data analysis methods

   In our study, we utilized R language, version 4.2.1, for data analysis
   and implemented compilation through the RStudio integrated development
   environment, version 2022.12.0-353. For file processing, we employed
   Perl language, version 5.30.0. Additionally, GraphPad Prism software,
   version 8.0, was utilized.

   Quantitative data were presented as mean ± standard deviation. For
   comparing two groups of data, an independent sample t-test was
   employed. One-way analysis of variance (ANOVA) was used to compare data
   among different groups, while two-way ANOVA was utilized to compare
   data among different time points of multiple groups. Bonferroni
   correction was employed for post hoc analysis. The significance level
   was set at P < 0.05 [[248]95].

Supplementary information

   [249]Full and uncropped western blots^ (2.4MB, docx)
   [250]Supplementary Materials^ (2.1MB, docx)

Author contributions

   Zhiwei Tao conceived and designed the study, supervised the
   experiments, and wrote the manuscript. Zou Pingan and Zhengxu Yang
   performed the single-cell multi-omics analyses and contributed to data
   interpretation. Tao Xiong conducted the in vitro experiments and
   analyzed the chemoresistance assays. Zhi Deng contributed to the in
   vivo studies and the metabolic pathway analysis. Qinchan Chen provided
   statistical analysis and helped with manuscript revisions. All authors
   reviewed and approved the final version of the manuscript.

Data availability

   The datasets generated and/or analyzed during the current study are
   available from the corresponding author on reasonable request.

Competing interests

   The author declares no competing interests.

Footnotes

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

   These authors contributed equally: Zhiwei Tao, Pingan Zou.

Supplementary information

   The online version contains supplementary material available at
   10.1038/s41420-025-02487-9.

References