Abstract

   Osteosarcoma, the most prevalent malignant bone tumor among
   adolescents, frequently exhibits limited responsiveness to
   immunotherapy, a challenge attributed to poorly understood underlying
   mechanisms. Here, we identify enhanced polyamine biosynthesis as a key
   driver of osteosarcoma progression and immunotherapy resistance. We
   show that osteosarcoma cell proliferation and tumor growth rely on
   polyamine availability and that disruption of polyamine synthesis
   significantly boosts the cytotoxic efficacy of TCR-engineered T cells
   against osteosarcoma cells. Mechanistically, we reveal that the
   knockdown of antizyme inhibitor 1 (AZIN1) or suppression of polyamine
   production reduces MYC expression, leading to diminished tumor cell
   viability via the downregulation of cell cycle-related genes.
   Furthermore, reduced MYC levels are associated with changes in the
   expression of immunomodulatory cytokines and human leukocyte antigen
   molecules, pointing to a potential link with enhanced T-cell-mediated
   cytotoxicity. Collectively, our findings establish a pivotal role for
   the AZIN1-polyamine axis in osteosarcoma proliferation and immune
   evasion, and support the development of novel immunotherapeutic
   strategies targeting polyamine biosynthesis to combat this aggressive
   cancer.

   Subject terms: Cancer metabolism, Immunosurveillance, Immunotherapy

Introduction

   Osteosarcoma is the most common malignant bone tumor in adolescents.
   While conventional treatments, such as radiotherapy and chemotherapy,
   are effective in approximately 70% of non-metastatic cases, patients
   who present with metastatic disease at diagnosis face considerably
   worse five-year survival rates, underscoring the need for improved
   therapeutic strategies [[46]1, [47]2]. Although immunotherapy has
   emerged as a promising approach for various cancers, its success in
   osteosarcoma has been limited [[48]3, [49]4]. This limitation
   highlights the importance of gaining a deeper understanding of the
   tumor’s immune microenvironment.

   Recent studies have emphasized the profound impact of metabolic
   dysregulation on immunotherapy effectiveness [[50]5]. For instance,
   mutations in isocitrate dehydrogenase (IDH), which lead to elevated
   levels of the oncometabolite 2-hydroxyglutarate (2HG), can markedly
   influence immune responses [[51]6]. However, the full spectrum of
   metabolic alterations in osteosarcoma remains insufficiently
   characterized. Of particular interest, polyamine metabolism
   (encompassing putrescine, spermidine, and spermine) plays an essential
   role in proliferation, differentiation, apoptosis, and immune
   regulation [[52]7, [53]8]. Despite the recognized significance of these
   metabolites, their specific impact on immunotherapeutic outcomes in
   osteosarcoma has yet to be thoroughly investigated.

   A central enzyme in polyamine biosynthesis is ornithine decarboxylase
   (ODC1), which converts ornithine into putrescine and serves as the
   rate-limiting step in this pathway [[54]9]. Notably, ODC1 activity is
   regulated by antizyme inhibitor 1 (AZIN1), which stabilizes ODC1 by
   inhibiting antizyme-mediated degradation [[55]10]. This regulatory axis
   presents a potential target for modulating polyamine levels in tumor
   cells. However, a comprehensive picture of the metabolic profiles and
   regulatory mechanisms controlling polyamine dynamics in osteosarcoma
   remains incomplete.

   In this study, we investigated the metabolic landscape of osteosarcoma
   and uncovered a dysregulated polyamine biosynthesis pathway driving
   both tumor cell proliferation and immune evasion. Mechanistically, we
   identified AZIN1 as a key regulator of this process. Reducing AZIN1
   levels diminished MYC expression and suppressed cell cycle progression,
   ultimately inhibiting tumor growth. Furthermore, reduced MYC levels are
   associated with increased human leukocyte antigen (HLA) expression,
   suggesting a potential link to enhanced T-cell-mediated cytotoxicity.
   Collectively, these findings underscore the pivotal role of the
   AZIN1-polyamine axis in osteosarcoma and support developing
   immunotherapeutic strategies that target polyamine metabolism to
   overcome resistance and enhance treatment efficacy.

Materials and methods

Collection of osteosarcoma specimens and adjacent normal tissue

   Osteosarcoma specimens and their corresponding normal adjacent tissues
   were collected during surgical procedures at the Department of Bone
   Tumor, the First Affiliated Hospital of Sun Yat-sen University,
   following protocols approved by the institutional review board. Written
   informed consent was obtained from each participant after providing a
   detailed explanation of the procedure and its associated risks, in
   compliance with the Declaration of Helsinki. All collected specimens
   were subjected to histopathological examination by a qualified
   pathologist to confirm the tumor type and grade.

Non-targeted metabolomics

   Osteosarcoma adjacent normal tissues were retrieved from liquid
   nitrogen storage and thawed on ice before being weighed. Samples of
   50 mg were transferred into 2 mL centrifuge tubes, to which 800 µL of
   80% methanol was added. The samples were homogenized for 90s, vortexed
   for one minute, and then ultrasonicated for 30 min at 4 ℃. After
   ultrasonication, samples were incubated at -40 ℃ for one hour, vortexed
   again for 30 s, and allowed to settle at 4 ℃ for 30 min without
   disturbance. Following this, the samples were mixed by inversion and
   centrifuged at 12,000×g at 4 ℃ for 15 min. Subsequently, 200 µL of the
   supernatant was transferred to an autosampler vial for LC–MS
   non-targeted metabolomics analysis conducted by Shanghai Sensichip
   Infotech Co., Shanghai, China.

   Raw data were processed using Compound Discoverer 3.1 (CD) software to
   perform multivariate statistical analyses. These analyses included
   qualitative and quantitative assessments, quality control measures,
   orthogonal projections to latent structures-discriminant analysis
   (OPLS-DA), and differential metabolite analysis. Results from the
   differential metabolite analysis were subjected to pathway enrichment
   analysis using MetaboAnalyst 6.0 ([56]http://www.metaboanalyst.ca).
   P-values less than 0.05 were considered statistically significant.
   Details of metabolite abundance are provided in Table [57]S3.

Animals

   Female Balb/c nude mice and NSFG mice (NOD-Prkdc^scid-B6-IL2rg^nullNh),
   weighing 18 ± 2 g and aged 4–6 weeks, were procured from the Model
   Animal Research Center of Nanjing University and RUIYE Laboratory,
   Guangzhou, China, respectively. All experimental procedures involving
   these animals were conducted in strict accordance with the guidelines
   approved by the Ethical Committee of the First Affiliated Hospital of
   Sun Yat-sen University.

Orthotopic xenograft model

   A single-cell suspension of osteosarcoma cells (20 µL) containing
   1–5 × 10^5 cells was orthotopically injected into the right tibial
   medulla of 5-week-old nude mice or NSFG mice, which were anesthetized
   using chloral hydrate. A customized arginine-deprived mouse diet was
   obtained from MolDiets, Beijing, China. 0.5% DFMO drinking water was
   prepared by dissolving 1 g of DFMO (Invivochem) with 200 mL sterile
   water and sterilized using a 0.22 μm PES (polyethersulfone) filter.
   Three to six weeks post-injection, tumor growth was monitored using an
   in vivo imaging system (IVIS, Xenogen) under isoflurane anesthesia.
   After orthotopic tumors became visible, their dimensions were measured
   with calipers to record both length and width. Tumor volumes were
   calculated using the formula: volume (mm^3) = (length × width^2)/2.
   When the tumor volumes reached 20–50 mm^3, mice were randomly allocated
   to either TCR-T cell treatment groups or control groups receiving
   non-transduced T cells, ensuring an equal initial tumor burden. Mice
   with tumors smaller than 20 mm^3 were excluded from the study. The
   maximal tumor (2 cm diameter) size/burden was not exceeded in all
   experiments.

Cell culture methods

   Human osteosarcoma cell lines 143B, U2-OS, SJSA-1, MNNG, and human
   embryonic kidney 293T cells were maintained in DMEM medium (Corning,
   USA) supplemented with 10% fetal bovine serum (FBS) (BI, USA).
   Customized arginine-deprived DMEM medium was obtained from BOSTER
   Biological Technology, Ltd. (Wuhan, China). Cultures were incubated at
   37 ℃ in a humidified atmosphere containing 5% CO[2]. Human peripheral
   blood mononuclear cells (PBMCs) obtained from LDEBIO, Guangzhou, along
   with T cells, were cultured in a T cell medium (TCM). This medium
   consisted of RPMI-1640 (Gibco), enriched with 10% FBS (BI, USA), 25 mM
   HEPES (Corning), 2 mM GlutaMax (Gibco), 50 µM β-mercaptoethanol
   (Sigma), and 50 IU/mL recombinant human Interleukin-2 (rhIL-2)
   (Peprotech).

Cell viability assay

   Cell proliferation was quantified using the CellTiter-Glo® Luminescent
   Cell Viability Assay kit (Promega), following the manufacturer’s
   protocol. Briefly, 2000 cells suspended in 100 µL of complete culture
   medium were seeded into each well of a 96-well culture plate. At 24,
   48, 72, and 96 h post-inoculation, an equal volume of CellTiter-Glo®
   Reagent was added to each well to lyse the cells. The contents were
   then transferred to an opaque 96-well plate, and luminescence was
   measured using a Biotek Synergy 2 plate reader.

Construction of plasmids and viral packaging

   Short hairpin RNAs (shRNAs) targeting specific genes were cloned into
   the pLVX-puro lentiviral vector. The MAGE-A4/HLA-A*02:01 specific TCR
   sequence was obtained from a patent (WO 2017/174824 A1) held by
   Adaptimmune Ltd. For lentiviral particle production, the plasmids
   expressing target sequences were co-transfected with the packaging
   plasmids pVSV-G and psPAX2 into human embryonic kidney 293T cells using
   StarFect II Transfection Reagent (GenStar, China). The medium
   containing lentiviral particles was collected at 36 and 72 h
   post-transfection. The sequences of the shRNA targets are provided in
   Table [58]S1. Additionally, lentiviral particles T cells transduction
   were concentrated 50-fold by centrifugation at 3000g for 1 h at 4 ℃
   with 8.5% PEG6000 and 0.3 M NaCl. Particle pellets were resuspended in
   TCM (without rhIL-2) and stored at −80 ℃ until required.

Western blotting (WB)

   Cells were lysed in RIPA buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl,
   1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with
   protease and phosphatase inhibitors (Roche). The lysates were incubated
   on ice for 30 min and then clarified by centrifugation at 12,000g for
   15 min at 4 ℃. The protein concentration in the supernatant was
   determined using the BCA Protein Assay Kit (Thermo Fisher Scientific).
   Equal amounts of protein (20 μg) were resolved by SDS-PAGE on a 10%
   polyacrylamide gel and transferred to a polyvinylidene difluoride
   (PVDF) membrane (Millipore). Membranes were blocked in 5% non-fat dry
   milk in Tris-buffered saline with 0.1% Tween-20 (TBST) for 1 h at room
   temperature. The membranes were then incubated overnight at 4 ℃ with
   primary antibodies diluted in 5% bovine serum albumin (BSA) in TBST.
   After washing three times with TBST, the membranes were incubated with
   horseradish peroxidase (HRP)-conjugated secondary antibodies diluted in
   5% non-fat dry milk in TBST for 1 h at room temperature. Following
   three additional washes with TBST, the protein bands were visualized
   using an enhanced chemiluminescence (ECL) detection system (Thermo
   Fisher Scientific) and imaged using a ChemiDoc MP Imaging System
   (Bio-Rad). The primary antibodies used in this study included CDK2
   (78B2) Rabbit mAb (#2546, CST), Cyclin D1 (92G2) Rabbit mAb (#2978,
   CST), p21 Waf1/Cip1 (12D1) Rabbit mAb (#2947, CST), CDK4 (D9G3E) Rabbit
   mAb (#12790, CST), p18 INK4C (DCS118) Mouse mAb (#2896, CST),
   Phospho-RB-S807/811 Rabbit pAb (#AP0484, Abclonal), Mouse monoclonal
   anti-β-actin (#60008, Proteintech), and Anti-PD-L1 antibody
   [[59]EPR19759] (#ab213524, abcam).

Quantitative real-time PCR (qPCR)

   Total RNA was extracted from cells using TRIzol reagent according to
   the manufacturer’s instructions. cDNA was generated from mRNA using
   HiScript II Q RT SuperMix (#R223, Vazyme) following the removal of gDNA
   according to the manufacturer’s instructions. Quantitative real-time
   PCR was carried out on the Applied Biosystems QuantStudio 6 Flex
   Real-Time PCR system (Thermo Fisher Scientific) with ChamQ Universal
   SYBR qPCR Master Mix (#Q711, Vazyme). The primers used are shown in
   Table [60]S2.

Human T cell transduction

   Primary human T cells were activated ex vivo by seeding peripheral
   blood mononuclear cells (PBMCs) in an anti-CD3 (OKT3) antibody-coated
   plate along with anti-CD28 antibody (1 µg/mL) in T cell medium (TCM)
   for two days. Following activation, the cells were pelleted and
   resuspended at a concentration of 5 × 10^5 cells/mL in fresh TCM
   supplemented with lentiviral supernatant and 8 µg/mL polybrene.
   Transduction efficiency was assessed three days post-transduction.

Flow cytometry

   For the detection of MAGE-A4/HLA-A*02:01 (A2M4) specific TCR
   transduction efficiency, transduced T cells were washed once with 1%
   BSA-PBS wash buffer and stained with PerCP-Cyanine5.5 anti-human TCR
   Vα24 antibody (Biolegend) at the recommended dilution, following the
   manufacturer’s instructions. Non-transduced, activated T cells stained
   with the same antibody were used as negative controls. For cell cycle
   analysis, 30 μM BrdU was added to the culture medium and incubated at
   37 ℃ for 3 h. After incubation, cells were washed twice with PBS,
   dissociated into single cells, and fixed in 70% ethanol. BrdU staining
   was performed using the FITC-BrdU Cell Proliferation Detection Kit
   (#B502, Solarbio; #KGA9201, KeyGEN BioTECH, China), following the
   manufacturer’s instructions. For HLA-A/B/C and CD3 staining, tumors
   were harvested 7 days after 0.5% DFMO treatment and 3 days after
   A2M4-TCR T cells injection, respectively. Tumors were cut into pieces,
   digested with 1 mg/mL collagenase Type IV (#17104019, Gibco) and
   20 µg/mL DNaseI (#1121MG010, Biofroxx), ground and filtered using 70 µm
   cell strainers (#CSS013070, JETBIOFIL). Single cells were stained with
   Tonbo™ Ghost Dye™ Violet 510 (#13-0870-T100, eBioscience), fixed in 4%
   paraformaldehyde solution, blocked with mouse IgG, and stained with
   antibodies including HLA class I A/B/C antibody (#66013-1-Ig,
   proteintech), Goat anti-mouse IgG Alexa Fluor™ 488 (#A28175,
   Invitrogen) and anti-Human CD3-APC (#F1100303, MULTI SCIENCES). Flow
   cytometry analysis was conducted using a CytoFLEX instrument (Beckman
   Coulter).

In vitro cytotoxicity assay

   U2-OS or 143B cells expressing HLA-A*02:01 and MAGE-A4 were transduced
   with either control shRNA or shRNA targeting AZIN1, or treated with PBS
   or 5 mM DFMO for 48 h prior to co-incubation with TCR-T cells. Both
   tumor cells and TCR-T cells were washed once with PBS and co-cultured
   at specified TCR-T cells-to-tumor cells ratios (E:T ratio) in T cell
   assay medium (TAM). TAM consists of TCM supplemented with 5% FBS but
   lacks IL-2. The co-culture was maintained for approximately 16 h. For
   PD-L1 blocking experiments, tumor cells were pre-treated with 10 µg/mL
   anti-PD-L1 antibody (Atezolizumab, #B2016, BIOINTRON) or an isotype
   control antibody at 37 ℃ for one hour prior to co-culturing with TCR-T
   cells. Specific cytotoxicity was assessed by measuring lactate
   dehydrogenase (LDH) release using the LDH Cytotoxicity Assay Kit
   (Beyotime), following the manufacturer’s instructions.

ELISA

   The levels of arginine, ornithine, putrescine, spermidine, and IFN-γ
   were quantified using ELISA kits (MEIMIAN and Cloud-Clone Corp.,
   China). For metabolite detection, tumor cells were harvested, washed
   three times with PBS, resuspended in PBS, and lysed by sonication.
   Cellular debris was removed by centrifugation, and the supernatant was
   stored at −20 ℃ until further analysis. For IFN-γ detection, the
   co-culture medium was centrifuged at 500×g for 5 min, and the
   supernatant was collected and stored at −20 ℃ until analysis. On the
   day of the experiment, all reagents and samples were equilibrated to
   room temperature. A total of 50 μL of each sample was added to ELISA
   wells in duplicate, and subsequent steps were performed according to
   the manufacturer’s instructions. Absorbance was measured at 450 nm
   using a Synergy2 plate reader (BioTek).

RNA sequencing

   Total RNA was extracted using TRIzol reagent (Life Technologies, USA)
   according to the manufacturer’s instructions and quantified using a
   Nanodrop 2000 spectrophotometer. RNA integrity was assessed using the
   Agilent Bioanalyzer 2100. Paired-end sequencing was performed by
   Novogene (Novogene, China) using the Illumina NovaSeq X Plus platform.
   Raw sequencing reads were cleaned using Trim Galore to remove
   low-quality reads and adapter sequences. Cleaned reads were then
   aligned to the human reference genome (build hg38) using HISAT2
   software. Differential expression analysis was performed using DESeq2,
   with a significance threshold of p < 0.05. Gene set enrichment analysis
   (GSEA) was conducted using the clusterProfiler, ReactomePA, and
   enrichplot packages to identify enriched pathways.

Quantification and statistical analysis

   All in vitro experiments were conducted in at least triplicate.
   Statistical analyses were performed using GraphPad Prism version 8.0
   (GraphPad Software, USA). Data are presented as mean ± SD. For
   comparisons between two groups, unpaired or paired two-tailed t-tests
   were used to assess statistical significance. For comparisons involving
   three or more groups, one-way or two-way ANOVA was applied. P-values
   less than 0.05 were considered statistically significant and denoted as
   ‘*’; p-values less than 0.01 and 0.001 were indicated as ‘**’ and
   ‘***’, respectively. Non-significant results were labeled as “n.s.”

Results

Association of elevated polyamine synthesis with osteosarcoma progression

   To investigate metabolic alterations in osteosarcoma, we performed
   untargeted metabolomics on both tumor and adjacent non-tumor tissues
   from 30 osteosarcoma patients (Fig. [61]1A). Orthogonal projections to
   latent structures-discriminant analysis (OPLS-DA) revealed a clear
   metabolic distinction between osteosarcoma and non-tumor tissues, as
   shown by the OPLS-DA score plot (Fig. [62]1B). Most samples clustered
   within the 95% confidence interval of Hotelling’s T-squared ellipse,
   indicating the robustness of our model.

Fig. 1. Upregulation of arginine metabolism and polyamine biosynthesis in
osteosarcoma.

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

   A Schematic representation of the experimental design for non-targeted
   metabolomics to investigate metabolic alterations in osteosarcoma. B
   Score scatter plot from the orthogonal partial least squares
   discriminant analysis (OPLS-DA) model comparing osteosarcoma tissues
   (OS, n = 30) to normal adjacent tissues (NT, n = 30). C Analysis of
   differential metabolites between osteosarcoma (OS) and normal adjacent
   tissues (NT) derived from non-targeted metabolomics. Statistical
   analyses were conducted using paired t-test. D Pathway enrichment
   analysis based on differential metabolites identified in the
   non-targeted metabolomics data. E Quantitative analysis of metabolites
   involved in the urea cycle, as determined by non-targeted metabolomics.
   Data are expressed as mean ± standard deviation (SD). Statistical
   analyses were conducted using paired t-test. F, G Scatter plots
   illustrate the Pearson correlation between levels of arginine and
   polyamines, highlighting their interdependencies in osteosarcoma
   metabolism. Significance levels are denoted as follows: ***P ≤ 0.001,
   **P ≤ 0.01, *P ≤ 0.05, n.s. not significant.

   Using multivariate statistical analysis, we focused on metabolites with
   high variable importance in the projection (VIP) scores from the
   OPLS-DA model and p-values < 0.05 (Student’s t-tests). We identified 20
   metabolites that were downregulated and 67 that were upregulated in
   osteosarcoma tissues (Fig. [65]1C). Enrichment analysis with
   MetaboAnalyst 6.0 underscored several significantly altered pathways,
   including the urea cycle, aspartate metabolism, methionine metabolism,
   and the biosynthesis of spermidine and spermine, along with arginine
   and proline metabolism (Fig. [66]1D).

   A comprehensive analysis of these pathways revealed an aberrant
   increase in arginine levels, driving metabolic flux preferentially
   toward polyamine synthesis rather than guanidinoacetate production
   (Fig. [67]1E). Correlation analyses further demonstrated strong
   positive associations between arginine, a key substrate for polyamine
   biosynthesis, and polyamines such as putrescine and spermidine (Fig.
   [68]1F, G). Although glutamine is also an important precursor for
   ornithine and polyamine production [[69]11, [70]12], its levels were
   significantly reduced in osteosarcoma relative to non-tumor tissues
   (Fig. [71]S1A, [72]S1B). Collectively, these findings indicate that
   enhanced polyamine synthesis is closely linked to arginine level.

Targeting polyamine synthesis to mitigate osteosarcoma growth

   To assess the role of polyamine in tumor proliferation, we cultured
   osteosarcoma cell lines in an arginine-depleted medium and monitored
   cell growth via viability assays over 24, 48, 72, and 96 h. Compared to
   cells grown in standard DMEM, arginine-deprived cells exhibited
   markedly reduced proliferation in vitro (Fig. [73]2A–D), underscoring
   the essential contribution of polyamine to osteosarcoma cell expansion.

Fig. 2. Inhibition of polyamine synthesis suppresses osteosarcoma growth.

   [74]Fig. 2
   [75]Open in a new tab

   A–D Relative growth curves of osteosarcoma cell lines 143B, MNNG,
   SJSA-1, and U2-OS cultured in normal control medium or under arginine
   deprivation. Data are expressed as mean ± standard deviation (SD), with
   each condition replicated three times (n = 3). E Schematic
   representation of the experimental setup for evaluating the in vivo
   growth of 143B-luc under a normal control diet compared to an
   arginine-deprived diet. F Representative images of orthotopic 143B-luc
   tumors in mice, with four subjects per treatment group (n = 4). G
   Quantification of in vivo luciferase signal intensity from 143B-luc
   tumors, indicating tumor growth dynamics. Data are presented as
   mean ± SD, with four animals per group (n = 4). Statistical analyses
   were conducted using paired t-test. H–J Measurement of arginine,
   ornithine, and putrescine concentrations in orthotopic 143B-luc tumors,
   showcasing metabolic changes associated with diet. Results are
   expressed as mean ± SD for four subjects per group (n = 4). Statistical
   analyses were conducted using paired t-test. Significance levels are
   denoted as follows: ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05, n.s. not
   significant.

   To validate these findings in vivo, we employed an orthotopic 143B
   xenograft model in which mice were fed either a standard or an
   arginine-deficient diet (Fig. [76]2E). Tumor measurements revealed that
   mice on the arginine-deficient regimen developed significantly smaller
   tumors than those receiving the standard diet (Fig. [77]2F, G).
   Biochemical analyses confirmed notable alterations in arginine,
   ornithine, and polyamine levels within the tumor microenvironment (Fig.
   [78]2H–J). These data indicate that restricting polyamine can be a
   powerful strategy to suppress osteosarcoma tumor expansion.

Targeting polyamine synthesis enhances TCR-T cell cytotoxicity against
osteosarcoma

   Previous studies have demonstrated that suppressing polyamine synthesis
   can increase the effectiveness of immune checkpoint therapies in
   certain cancers [[79]13, [80]14]. Building on this insight, we
   hypothesized that targeting polyamine metabolism could similarly
   improve osteosarcoma immunogenicity and the efficacy of T-cell–based
   therapies. To test this, we conducted in vitro cytotoxicity assays with
   HLA-A*02:01 and MAGEA4 (A2M4) TCR-engineered T cells (TCR-T) against
   the U2-OS osteosarcoma cell line, which naturally expresses HLA-A*02:01
   and MAGEA4. Specifically, we employed T cells engineered to recognize
   MAGEA4 presented by HLA-A*02:01. Pretreating U2-OS cells with the ODC1
   inhibitor difluoromethylornithine (DFMO) significantly enhanced A2M4
   TCR-T cell-mediated cytotoxicity, as indicated by LDH release (Figs.
   [81]3A and [82]S2). A similar increase in T-cell-induced killing was
   observed in 143B-A2M4 cells (143B cells engineered to overexpress
   HLA-A*02:01 and MAGEA4) following DFMO pretreatment (Fig. [83]3A).
   Consistent with these results, polyamine measurements in DFMO-treated
   U2-OS and 143B-A2M4 cultures revealed a substantial decrease in
   putrescine and spermidine at two days post-DFMO treatment (Fig.
   [84]3B).

Fig. 3. Inhibition of polyamine synthesis enhances TCR-T cell-mediated
cytotoxicity in osteosarcoma.

   [85]Fig. 3
   [86]Open in a new tab

   A TCR-T cell-mediated cytotoxicity against U2-OS and 143B-A2M4 cells at
   different E:T ratio (effector cells:target cells), with and without
   DFMO pretreatment (5 mM). Data are expressed as mean ± SD, with three
   biological replicates per condition (n = 3). Statistical significance
   was assessed using an unpaired t-test. B. Levels of putrescine and
   spermidine in U2-OS and 143B-A2M4 cells following DFMO treatment. Data
   are presented as mean ± SD, n = 3 per group. Statistical analysis was
   performed using two-way ANOVA. C Schematic depiction of the
   experimental protocol for assessing in vivo TCR-T cell-mediated
   cytotoxicity of 143B-A2M4-luc cells, with or without DFMO pretreatment
   (0.5% in drinking water). D Representative images and quantification of
   in vivo luciferase signal intensity from orthotopic 143B-luc tumors
   treated with transduced (TCR-T) or non-transduced (NT-T) T cells. Data
   are presented as mean ± SD for four experimental subjects per group
   (n = 4). Statistical evaluation utilized one-way ANOVA. E
   Representative images of orthotopic 143B-A2M4-luc tumors and subsequent
   tumor weight analysis post-treatment. Data are shown as mean ± SD, with
   four subjects per treatment group (n = 4). Statistical analysis was
   conducted using one-way ANOVA. F Representative images of CD3 staining
   in 143B-A2M4 orthotopic tumors, gated on live cells, with corresponding
   quantification analysis. Data are presented as mean ± SD (n = 4).
   Statistical significance was determined using an unpaired t-test.
   Significance levels are indicated as follows: ***P ≤ 0.001, **P ≤ 0.01,
   *P ≤ 0.05, n.s. not significant.

   To evaluate the therapeutic benefit of inhibiting polyamine synthesis
   in vivo, we used an orthotopic model in immunodeficient NSFG mice
   implanted intratibially with luciferase-expressing 143B-A2M4 cells
   (Fig. [87]3C). After two weeks of tumor growth, mice received either a
   0.5% DFMO solution or normal water for seven days, followed by a single
   intravenous injection of A2M4 TCR-T cells at the end of the third week.
   Imaging and tumor analyses during the fifth week revealed that while
   TCR-T cell therapy alone modestly attenuated tumor growth, this effect
   was significantly enhanced by short-term DFMO pretreatment (Fig.
   [88]3D, E). Moreover, we observed increased T-cell infiltration in
   DFMO-treated tumors (Fig. [89]3F). These data highlight the potential
   of polyamine synthesis inhibition not only to impede osteosarcoma
   progression but also to bolster the immune microenvironment, thereby
   improving T-cell-mediated cytotoxicity.

Impact of AZIN1-mediated polyamine synthesis on osteosarcoma progression

   To elucidate the mechanisms underlying aberrant polyamine synthesis in
   osteosarcoma, we examined transcriptional profiles of key genes in the
   arginine–polyamine metabolic pathway using publicly available RNA-seq
   data. Although the expression levels of ARG2 and ODC1, an essential
   enzyme for polyamine synthesis [[90]7], were comparable to those in
   normal tissues, we detected a marked upregulation of AZIN1 (Figs.
   [91]4A and [92]S3). Given AZIN1’s known role in regulating ODC1
   stability and activity [[93]15], we hypothesized that AZIN1
   overexpression underlies the dysregulated arginine-polyamine metabolism
   characteristic of osteosarcoma.

Fig. 4. AZIN1 is critical for polyamine production and cell proliferation in
osteosarcoma.

   [94]Fig. 4
   [95]Open in a new tab

   A Schematic representation integrating non-targeted metabolomics and
   RNA-seq data to show the expression of arginine metabolism-related
   genes in osteosarcoma and bone marrow stromal cells (BMSC) using
   publicly available datasets ([96]GSE87624 and E-MTAB-7925). Data are
   presented as mean ± SD, with sample sizes of n = 12 for BMSC and n = 44
   for OS. Statistical analysis was performed using an unpaired t-test. B
   Relative in vitro growth curves of 143B and U2-OS osteosarcoma cell
   lines with or without AZIN1 knockdown. Data are expressed as mean ± SD,
   n = 3. Statistical analysis utilized one-way ANOVA. C Levels of
   putrescine and spermidine in 143B and U2-OS cells following AZIN1
   knockdown. Data are shown as mean ± SD, with three replicates per group
   (n = 3). Two-way ANOVA was used for statistical analysis. D
   Representative images and quantification of in vivo luciferase signal
   intensity in orthotopic 143B-luc tumors with or without AZIN1
   knockdown. Data are presented as mean ± SD, n = 4 per group.
   Statistical evaluation was conducted using one-way ANOVA. E
   Kaplan-Meier survival analysis comparing patients stratified by high
   (n = 18) and low (n = 67) AZIN1 expression levels from the TARGET_OS
   dataset. Statistical evaluation was conducted using the log-rank test.
   Significance levels are indicated as follows: ***P ≤ 0.001, **P ≤ 0.01,
   *P ≤ 0.05, n.s. not significant.

   To test this hypothesis, we generated 143B and U2-OS osteosarcoma cell
   lines with stable AZIN1 knockdown and conducted cell viability assays.
   Both cell lines displayed substantially reduced proliferation following
   AZIN1 knockdown (Figs. [97]4B and [98]S4). Consistent with AZIN1’s
   regulatory role, we observed a significant decrease in intracellular
   polyamine levels in these knockdown cells (Fig. [99]4C). Further
   supporting these findings, an orthotopic tumor model in mice revealed
   that AZIN1 knockdown markedly inhibited tumor growth in vivo (Fig.
   [100]4D). Clinically, Kaplan-Meier analysis of patient data indicated
   that high AZIN1 expression correlates with significantly shorter
   disease-free survival compared to low expression (Fig. [101]4E). Taken
   together, these results highlight the pivotal function of AZIN1 in
   osteosarcoma proliferation and underscore its potential as a
   therapeutic target.

AZIN1 regulation of polyamine synthesis modulates cell cycle dynamics in
osteosarcoma

   To investigate how AZIN1 influences osteosarcoma cell proliferation, we
   performed RNA sequencing on 143B, and U2-OS cells stably knocked down
   for AZIN1 and compared the results to control cells. Gene Set
   Enrichment Analysis (GSEA) revealed marked downregulation of gene sets
   tied to cell proliferation, particularly those associated with the cell
   cycle, nuclear division, and DNA replication (Figs. [102]5A and
   [103]S5A). Correspondingly, crucial cell cycle regulatory genes,
   including CDC6, CDC7, MCM2, and MCM3 [[104]16–[105]18], were
   substantially suppressed following AZIN1 knockdown (Figs. [106]5B and
   [107]S5B). A similar expression pattern emerged in 143B cells cultured
   in an arginine-depleted medium, and flow cytometry also showed that the
   cell cycle was infected (Fig. [108]S5C, [109]D).

Fig. 5. AZIN1 facilitates cell cycle progression in osteosarcoma cells in a
polyamine-dependent manner.

   [110]Fig. 5
   [111]Open in a new tab

   A Representative gene set enrichment analysis (GSEA) results from
   RNA-seq data of 143B cells with or without AZIN1 knockdown,
   highlighting changes in cell cycle-related gene expression. B Gene
   expression heatmap displaying the transcriptional levels of cell
   cycle-regulating genes in 143B cells, comparing conditions with and
   without AZIN1 knockdown. C Representative flow cytometry plots of cell
   cycle analysis in 143B cells following AZIN1 knockdown and
   supplementation with 10 μM putrescine and 5 μM spermidine, assessed by
   BrdU pulse-labeling and PI staining. Quantification of cell cycle phase
   distribution is on the right. Data are mean ± SD (n = 3). Statistical
   significance was determined using one-way ANOVA. D Western blot
   analysis of cell cycle regulators in 143B cells with AZIN1 knockdown,
   supplemented with putrescine and spermidine. E In vitro growth curves
   of 143B cells with AZIN1 knockdown and subsequent supplementation with
   putrescine and spermidine. Data are mean ± SD (n = 3). Statistical
   analysis was performed using one-way ANOVA. F Normalized expression
   levels of MYC from RNA-seq data of 143B and U2-OS cells with and
   without AZIN1 knockdown. Data are mean ± SD (n = 3). Statistical
   significance was assessed using an unpaired t-test. G Representative
   IGV tracks showing MYC ChIP-seq binding on promoters of cell
   cycle-regulating genes in U2-OS cells, derived from the ChIP-Atlas
   database. Significance levels are indicated as follows: ***P ≤ 0.001,
   **P ≤ 0.01, *P ≤ 0.05, n.s. not significant.

   Given the established function of polyamines in cell cycle regulation,
   we hypothesized that AZIN1 promotes the osteosarcoma cell cycle by
   modulating ODC1 activity and, consequently, polyamine synthesis. Flow
   cytometric analysis of AZIN1-knockdown 143B and U2-OS cells revealed an
   increased proportion of cells in G[0]/G[1] and a concomitant decrease
   in S phase, indicative of cell cycle arrest. Supplementing these cells
   with putrescine or spermidine partly mitigated the G[0]/G[1] arrest
   (Figs. [112]5C and [113]S5E). Consistent with this finding, AZIN1
   knockdown led to reduced levels of phosphorylated retinoblastoma
   protein (p-Rb), CDK2, CDK4, and Cyclin D1, alongside elevated levels of
   the cyclin-dependent kinase inhibitors p18 and p21. Notably, these
   molecular changes were partly reversed by the addition of putrescine or
   spermidine (Figs. [114]5D and [115]S5F). Restoring polyamine levels in
   AZIN1-knockdown cells further led to a resurgence in proliferation in
   both 143B and U2-OS cell lines (Figs. [116]5E and [117]S5G).

   Extending these observations, we found that either AZIN1 knockdown or
   inhibition of polyamine synthesis diminished MYC expression (Fig.
   [118]5F). Furthermore, chromatin immunoprecipitation and promoter
   analyses demonstrated that MYC may serve as a regulatory hub for key
   cell cycle genes (Fig. [119]5G). Collectively, these results highlight
   the pivotal role of the AZIN1-polyamine-MYC axis in governing cell
   cycle progression in osteosarcoma.

Impairment of osteosarcoma immunogenicity by AZIN1

   We performed a comprehensive analysis of RNA-seq data from 143B and
   U2-OS osteosarcoma cells with AZIN1 knockdown to investigate how
   polyamine metabolism modulates the osteosarcoma immune
   microenvironment. Gene set enrichment analysis (GSEA) revealed marked
   upregulation of immune-related gene sets, including inflammatory
   responses, leukocyte proliferation, migration, activation, and
   T-cell-mediated immunity, following AZIN1 knockdown (Figs. [120]6A and
   [121]S6A). Further examination showed increased expression of key
   T-cell activation markers HLA-A/B/C, IL1A, and IL1B (Figs. [122]6B and
   [123]S6B). Accordingly, A2M4 TCR-T cells co-cultured with
   AZIN1-knockdown osteosarcoma cells secreted significantly more IFN-γ
   than those co-cultured with shControl cells (Figs. [124]6C and
   [125]S6C). In parallel, flow cytometric analysis of DFMO-treated tumors
   demonstrated elevated MHC-I (HLA-A/B/C) expression (Fig. [126]6D).

Fig. 6. AZIN1-mediated polyamine production alters osteosarcoma
immunogenicity and enhances T-cell-mediated cytotoxicity.

   [127]Fig. 6
   [128]Open in a new tab

   A Gene set enrichment analysis (GSEA) of RNA-seq data from 143B cells,
   showing changes in immune response-related gene expression with and
   without AZIN1 knockdown. B Heatmap of transcriptional levels of genes
   regulating T cell responses in 143B cells, under conditions with and
   without AZIN1 knockdown. C ELISA quantification of IFN-γ levels in T
   cell co-cultures with 143B-A2M4 cells, comparing AZIN1 knockdown with
   control conditions. Data are expressed as mean ± SD (n = 3).
   Statistical evaluation was performed using one-way ANOVA. D
   Representative image of HLA-A/B/C staining in 143B-A2M4 orthotopic
   tumors, gated on live and HLA-A/B/C positive cells, with mean
   fluorescence intensity (MFI) quantification analysis. Data are
   presented as mean ± SD (n = 4). Statistical significance was determined
   using an unpaired t-test. E Quantitative PCR (qPCR) analysis of T cell
   activation-related gene expression in 143B cells with and without AZIN1
   knockdown. Data are expressed as mean ± SD, n = 3. Statistical analysis
   was performed using two-way ANOVA. F qPCR analysis of HLA-A and IL1A
   expression levels in 143B cells post-AZIN1 knockdown and supplemented
   with either putrescine (put) or spermidine (spd). Data are expressed as
   mean ± SD, n = 3. Statistical analysis was conducted using one-way
   ANOVA. G Assessment of TCR-T cell-mediated cytotoxicity against
   143B-A2M4 cells with AZIN1 knockdown and supplementation with 10 μM
   putrescine (put) or 5 μM spermidine (spd). Data are presented as
   mean ± SD, n = 3. Statistical evaluation utilized one-way ANOVA. H
   Normalized expression levels of NFKBIA in RNA-seq data from 143B and
   U2-OS cells under conditions with and without AZIN1 knockdown. Data are
   presented as mean ± SD (n = 3). Statistical significance was assessed
   using an unpaired t-test. Significance levels are denoted as follows:
   ***P ≤ 0.001, **P ≤ 0.01, *P ≤ 0.05, n.s. not significant.

   To validate these observations, we conducted quantitative PCR (qPCR) on
   AZIN1-knockdown U2-OS and 143B cells, confirming the upregulation of
   HLA-A/B/C, IL1A, and IL1B [[129]19] (Figs. [130]6E and [131]S6D).
   ChIP-seq analyses further revealed that MYC may serve as a regulatory
   hub for these immune-related genes (Fig. [132]S6E).

   We next tested whether these effects were dependent on polyamine
   levels. Supplementation of AZIN1-knockdown cells with putrescine or
   spermidine partially reversed the upregulation of immune-related genes
   (Figs. [133]6F and [134]S6F). Concordantly, TCR-T cytotoxicity assays
   showed that exogenous polyamine supplementation mitigated the enhanced
   T-cell-mediated killing of AZIN1-knockdown 143B-A2M4 and U2-OS cells
   (Figs. [135]6G and [136]S6G). Interestingly, AZIN1 knockdown also
   triggered a polyamine-dependent increase in PD-L1 expression (Fig.
   [137]S7A). Nonetheless, the elevated T-cell cytotoxicity observed after
   AZIN1 knockdown was further potentiated by anti-PD-L1 treatment
   relative to an isotype control (Fig. [138]S7B).

   Lastly, to explore the underlying mechanism, we reanalyzed RNA-seq data
   from 143B and U2-OS cells subjected to AZIN1 knockdown. Both cell lines
   showed significantly increased expression of NFKBIA (Fig. [139]6H),
   which encodes IκBα, a negative feedback regulator preventing NF-κB
   overactivation [[140]20]. Taken together, these findings highlight the
   critical role of AZIN1-driven polyamine synthesis in modulating the
   immune microenvironment, which in turn influences the immunogenicity of
   osteosarcoma and its vulnerability to T-cell-mediated cytotoxicity.

Discussion

   Our comprehensive metabolomic profiling provides compelling evidence
   that osteosarcoma exhibits significant activation of the polyamine
   synthesis pathway. This upregulation is primarily driven by the
   overexpression of AZIN1, a critical regulatory protein that governs
   polyamine biosynthesis. Notably, the knockdown of AZIN1 effectively
   curtailed tumor growth and heightened the tumor’s sensitivity to
   immunotherapy (Fig. [141]7). These findings suggest that osteosarcoma’s
   characteristic “cold” immune phenotype [[142]21] may, at least in part,
   stem from dysregulated polyamine metabolism and that targeting this
   pathway could improve immunotherapeutic outcomes in this challenging
   cancer.

Fig. 7. Schematic representation of the proposed hypothesis: elevated
AZIN1-driven polyamine production promotes osteosarcoma proliferation and
impairs immunotherapy efficacy.

   [143]Fig. 7
   [144]Open in a new tab

   In osteosarcoma cells, increased levels of arginine and AZIN1
   upregulate the metabolism of arginine into polyamines through
   AZIN1-mediated activation of ornithine decarboxylase 1 (ODC1). This
   enhancement in polyamine production supports tumor cell proliferation
   and impedes the efficacy of immunotherapy. Conversely, when polyamine
   synthesis is inhibited—either by AZIN1 knockdown or through
   difluoromethylornithine (DFMO)-mediated inhibition of ODC1-polyamine
   levels decrease, leading to cell cycle arrest and the upregulation of
   human leukocyte antigen (HLA) expression and T cell-activating
   cytokines. These changes augment the cytotoxic efficacy of
   TCR-engineered T cells against osteosarcoma cells.

   Our metabolomic analysis offers deeper insights into the metabolic
   reprogramming in osteosarcoma, particularly highlighting a strong
   preference for the arginine-driven pathway in polyamine production. We
   detected markedly elevated levels of arginine, ornithine, and
   putrescine, accompanied by a notable decrease in glutamine. While
   polyamines can be synthesized via both arginine- and glutamine-derived
   pathways [[145]22], our data suggest that osteosarcoma cells
   predominantly rely on the arginine route. Nonetheless, future studies
   employing isotope-labeled metabolite tracing will be crucial to
   validate and further solidify this conclusion.

   Polyamines have long been recognized as critical mediators of
   tumorigenesis and immune modulation, owing to their essential role in
   promoting cell proliferation and survival [[146]7, [147]8]. This aligns
   with our observation of elevated putrescine and spermidine in
   osteosarcoma tissues. Mechanistically, the AZIN1–polyamine axis emerges
   as particularly pivotal: increased polyamine availability has been
   reported to influence chromatin accessibility at the MYC promoter or
   its regulatory elements, potentially enhancing MYC transcription
   [[148]23–[149]25]. In turn, MYC is known to regulate genes involved in
   cell cycle progression [[150]26–[151]28] and may influence
   immunosuppressive molecules and cytokines [[152]29, [153]30], which
   could contribute to tumor growth and immune evasion.

   Our findings further highlight the central role of MYC in orchestrating
   polyamine-mediated proliferation and survival in osteosarcoma.
   Knockdown of AZIN1 diminishes intracellular MYC levels, leading to
   reduced expression of cyclins and cyclin-dependent kinases and,
   consequently, curbing cell cycle progression. In parallel, lower MYC
   activity may attenuate anti-apoptotic gene expression [[154]31],
   thereby restraining tumor growth. Notably, polyamine supplementation
   can restore these tumorigenic features that were lost upon AZIN1
   knockdown, underscoring the tight functional interplay between
   polyamines and MYC in osteosarcoma pathophysiology. These observations
   support a potential role for MYC within the polyamine-mediated network
   of osteosarcoma, and highlight it as a candidate for future therapeutic
   exploration.

   Beyond fueling proliferation, polyamines also contribute to immune
   evasion in osteosarcoma. By shaping immunosuppressive populations, such
   as regulatory T cells, tumor-associated macrophages (TAMs),
   myeloid-derived suppressor cells (MDSCs), and dendritic cells (DCs),
   excess polyamines create a tumor microenvironment that impairs
   effective antitumor immunity [[155]8, [156]32]. Mechanistically, the
   AZIN1-mediated surge in polyamines is associated with altered cytokine
   and HLA expression, possibly involving MYC-dependent pathways,
   ultimately skewing local cytokine profiles and reducing immune cell
   activation. This suite of changes underlies osteosarcoma’s
   characteristically poor immunogenicity, highlighting how metabolic
   reprogramming can intersect with immune dysregulation to promote tumor
   progression.

   In light of these insights, targeting the AZIN1-polyamine axis presents
   a promising strategy for combating osteosarcoma. Given MYC’s downstream
   involvement, interventions that modulate MYC activity may also hold
   therapeutic value, although further studies are needed to validate its
   role in immune modulation. Future research should focus on refining
   pharmacologic inhibitors of polyamine biosynthesis, determining optimal
   dosing regimens, and investigating potential synergies between these
   agents and established or emerging immunotherapies.

Supplementary information

   [157]Uncropped images for WB figures^ (65.2MB, pptx)
   [158]Supplementary Figures and Tables^ (1.6MB, docx)
   [159]Supplementary Table3^ (56.5KB, csv)

Acknowledgements