USP5-mediated CD73 deubiquitination drives osimertinib resistance via PI3K/AKT and glycolysis activation in LUAD

Summary

Ubiquitin-specific protease 5 (USP5) is frequently overexpressed in lung adenocarcinoma (LUAD) and correlates with advanced stage and poor prognosis. This study demonstrates that USP5 binds directly to CD73 and removes K48-linked polyubiquitin chains, thereby blocking its proteasomal degradation and increasing CD73 protein stability. In contrast, the E3 ligase tripartite motif-containing protein 28 (TRIM28) promotes CD73 ubiquitination and turnover. Functionally, USP5 enhances LUAD cell proliferation, migration, invasion, and tumor growth in vivo in a CD73-dependent manner. Metabolomic profiling and Seahorse assays reveal that the USP5/CD73 axis activates PI3K/AKT/mTOR signaling and drives glycolytic reprogramming, augmenting lactate production. Moreover, this axis contributes to acquired resistance to osimertinib, an epidermal growth factor receptor-tyrosine kinase inhibitor (EGFR-TKI); combined inhibition of USP5 and osimertinib synergistically induces apoptosis and suppresses tumor growth in vitro and in vivo. These findings establish USP5-mediated stabilization of CD73 as a central mechanism underlying glycolytic metabolism and osimertinib resistance in LUAD, highlighting the USP5/CD73 pathway as a promising prognostic indicator and therapeutic target for LUAD treatment.

Graphical abstract

Highlights

• •USP5 deubiquitinates and stabilizes CD73 in LUAD
• •The USP5/CD73 axis correlates with a poor prognosis in LUAD
• •USP5/CD73 drives glycolytic reprogramming in LUAD
• •Targeting USP5 overcomes osimertinib resistance in LUAD

Biochemistry; Cancer; Metabolomics

Introduction

Lung cancer remains the leading cause of cancer-related mortality worldwide, with non-small cell lung cancer (NSCLC) and its predominant subtype, lung adenocarcinoma (LUAD), representing a major therapeutic challenge.¹ The advent of tyrosine kinase inhibitors (TKIs) targeting epidermal growth factor receptor (EGFR) mutations, such as osimertinib, has markedly improved survival outcomes; however, the eventual emergence of acquired resistance continues to be a primary obstacle in clinical management.² This resistance is increasingly linked to tumor metabolic reprogramming. Notably, hyperglycolysis (the Warburg effect) not only furnishes energy and biosynthesis precursors for rapid proliferation but also actively drives therapeutic resistance by shaping an immunosuppressive tumor microenvironment (TME) and activating pro-survival pathways.³ In this context, EGFR mutations have been shown to enhance HIF-1α expression, leading to the upregulation of glycolytic enzymes and increased glycolytic flux, thereby contributing to TKI resistance.⁴ Consequently, inhibiting glycolysis has emerged as a promising strategy to augment the efficacy of EGFR-TKIs and overcome resistance, as demonstrated by studies reporting that targeting key glycolytic enzymes can resensitize osimertinib-resistant NSCLC cells.⁵

A pivotal mediator at the intersection of metabolism and tumor progression is ecto-5′-nucleotidase (CD73/NT5E), a rate-limiting enzyme in the generation of extracellular adenosine. By hydrolyzing AMP to adenosine, CD73 fosters tumor progression through a dual mechanism: adenosine signaling via A2A/A2B receptors directly activates downstream oncogenic pathways to promote proliferation,⁶ while simultaneously inducing a potent immunosuppressive microenvironment by suppressing T cell activity and expanding regulatory T cells (Tregs) and myeloid-derived suppressor cells.⁷ CD73 is frequently overexpressed in NSCLC cells and is strongly correlated with a poor prognosis; specifically, it promotes cancer cell proliferation, tumor growth, angiogenesis, and metastasis in NSCLC.⁸ Its role is further underscored by studies showing CD73 acts as an effector of EGF/EGFR-mediated invasion in head and neck cancer⁹ and that the CD73/adenosine pathway represents a potential therapeutic target in the immunologically inert EGFR-mutant NSCLC subtype.¹⁰ Moreover, CD73 upregulation constitutes a common resistance mechanism to diverse therapies, including chemotherapy, radiotherapy, targeted therapy, and immunotherapy.¹¹ Its influence extends to maintaining cancer stem cell traits, as evidenced by its role in reversing lenvatinib resistance in liver cancer via SOX9 stabilization.¹² These findings collectively position CD73 and the adenosine pathway as highly promising antitumor targets.^13,14,15,16 Nonetheless, clinical development of CD73-targeted agents has been hampered by complexities such as the Hook effect and suboptimal trial designs. A fundamental gap—the incomplete understanding of the upstream regulatory mechanisms governing CD73 overexpression—remains a critical barrier to advancing these therapeutic strategies.

Protein abundance is precisely controlled by the ubiquitin-proteasome system, a highly conserved post-translational regulatory mechanism. Ubiquitination, enacted by a cascade of E1, E2, and E3 enzymes, dictates protein stability, localization, and function.¹⁷ Conversely, deubiquitinating enzymes (DUBs), including ubiquitin-specific protease (USP) family members, counteract this process by removing ubiquitin chains, thereby stabilizing oncoproteins and driving malignant phenotypes, including cell cycle progression and metabolic reprogramming.^18,19 Dysregulation of ubiquitination also underpins therapeutic resistance by modulating DNA repair, immune evasion, and the TME.^20,21 The clinical relevance of this system is validated by therapeutics like the proteasome inhibitor bortezomib, highlighting the potential of targeting ubiquitin-system enzymes.²² USP5, a prominent DUB known as the “guardian of the ubiquitin pool,” is implicated in cancer progression. Our preliminary data established a significant correlation between elevated USP5 expression and poor prognosis in LUAD. Recent studies have revealed that USP5 stabilizes YTHDF1 to modulate cancer immune surveillance,²³ promotes ripretinib resistance in gastrointestinal stromal tumors by antagonizing MDH2 degradation,²⁴ and regulates aberrant glucose metabolism in hepatocellular carcinoma and triple-negative breast cancer.^25,26 However, the full scope of USP5’s mechanistic roles and substrate network in LUAD remains largely unexplored.

Interestingly, a previous CRISPR-Cas9 kinase library screening study suggested tripartite motif-containing protein 28 (TRIM28) as a potential regulator of CD73 expression, though the precise mechanism was not elucidated.²⁷ TRIM28. which functions as a transcriptional corepressor and an E3 ubiquitin ligase, is involved in DNA damage repair and cancer stemness maintenance and facilitates the progression of breast, lung, and liver cancers by promoting substrate degradation.^28,29,30,31 Intriguingly, despite its prevalent oncogenic role, TRIM28 can also exert context-dependent tumor-suppressive functions.³²

In this study, we further validated the upregulation of USP5 in LUAD and its association with poor prognosis. We systematically investigated the functional and mechanistic contributions of USP5 to LUAD progression. We demonstrate, for the first time, that USP5 stabilizes CD73 through deubiquitination, thereby promoting LUAD proliferation, migration, invasion, and osimertinib resistance. Furthermore, we identified TRIM28 as the E3 ubiquitin ligase that promotes CD73 degradation via ubiquitination, revealing a novel post-translational regulatory axis for CD73. The ensuing USP5-CD73 signaling cascade activates the PI3K/AKT/mTOR pathway, enhancing the Warburg effect. Our findings, thus, provide novel insights into LUAD progression and unveil a promising combinatorial therapeutic avenue for overcoming targeted therapy resistance.

Results

Elevated USP5 expression correlates with adverse prognosis in LUAD

Comprehensive pan-cancer analysis of curated TCGA datasets revealed significantly elevated USP5 expression in multiple malignancies (Figure 1A). In LUAD cohorts (TCGA-LUAD and GSE31547), USP5 demonstrated marked upregulation in tumor tissues versus normal controls (Figures 1B and 1C). High USP5 expression correlated with adverse clinical outcomes, manifesting as reduced overall survival (OS; Figures 1D and 1F) and disease-specific survival (DSS; Figure 1E). USP5 levels positively associated with advanced clinical staging (Figure 1G). Functional enrichment analysis implicated USP5 in epithelial-mesenchymal transition (EMT) and glycolytic processes in LUAD (Figure S1A). KEGG pathway interrogation further linked USP5 to pentose phosphate pathway activation and glycolytic reprogramming (Figure 1H). Drug sensitivity profiling in the Cancer Therapeutics Response Portal (CTRP) database revealed positive correlations between USP5 expression and the IC₅₀ of EGFR-TKIs—including gefitinib (1^st generation) and afatinib (2^nd generation)—suggesting USP5-mediated resistance to EGFR-targeted therapies (Figure 1I).

Figure 1.

Elevated USP5 expression correlates with poor prognosis and altered therapeutic responses in LUAD

(A) Pan-cancer analysis of USP5 expression from TCGA database, showing its level across multiple cancer types.

(B and C) USP5 mRNA expression is significantly upregulated in LUAD tissues relative to adjacent normal tissues in the TCGA dataset (B) and GSE31547 dataset (C).

(D and E) Kaplan-Meier survival curves demonstrating that high USP5 expression is associated with shortened OS (D) DSS (E) in the TCGA-LUAD cohort.

(F) Validation of the negative correlation between USP5 expression and OS in the GSE72094 cohort.

(G) USP5 expression levels positively correlate with advanced clinical stage in LUAD patients.

(H) KEGG pathway enrichment analysis of genes co-expressed with USP5, highlighting its association with metabolic pathways.

(I) Correlation analysis between USP5 expression and the half-maximal inhibitory concentration (IC₅₀) of various chemotherapeutic drugs in the CTRP database.

∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

OS, overall survival; DSS, disease-specific survival; CTRP, Cancer Therapeutics Response Portal.

USP5 drives LUAD cell proliferation, migration, invasion and tumor growth

To elucidate the oncogenic function of USP5 in LUAD progression, we established USP5-knockdown and -overexpression models across distinct LUAD cell lines. Lentiviral transduction with shUSP5#1/shUSP5#2 significantly reduced USP5 protein levels in H1299 and PC9 cells, while oeUSP5 transduction markedly elevated USP5 expression in H1650 cells (Figure 2A). EdU incorporation assays demonstrated that USP5 knockdown suppressed cellular proliferation, whereas USP5 overexpression enhanced the proliferative capacity in LUAD cells (Figure 2B). Consistent findings were observed in CCK-8 assays (Figure 2C). Colony formation analysis revealed that USP5 depletion profoundly inhibited clonogenicity in H1299 and PC9 cells, while USP5 overexpression amplified colony formation in H1650 cells (Figure 2D). In murine xenograft models, USP5-knockdown tumors exhibited significantly attenuated growth kinetics, with concomitant reductions in the final tumor volume and weight compared with controls. Conversely, USP5-overexpressing tumors displayed accelerated growth and increased mass (Figures 2E–2G). Collectively, these data establish USP5 as a potent driver of LUAD proliferation and tumorigenesis.

Figure 2.

USP5 promotes malignant proliferation and tumor growth in LUAD

(A) Western blot analysis of USP5 protein levels in H1299, PC9, and H1650 cells following lentiviral-mediated knockdown or overexpression.

(B) Representative images and quantification of EdU incorporation assays evaluating proliferative activity upon USP5 modulation.

(C) Cell viability measured by CCK-8 assays in LUAD cells with USP5 knockdown or overexpression over time.

(D) Colony formation capacity of LUAD cells after USP5 modulation (representative images and quantified results).

(E) Representative excised tumors from nude mice at the experimental endpoint (day 24 post-injection).

(F) Tumor volume measured every 4 days from day 7 post-injection.

(G) Final tumor weights and volumes from the xenograft mouse model.

Data are expressed as the mean ± SD; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001; triplicate experiments.

Wound healing assays revealed impaired migratory capacity in USP5-knockdown LUAD cells versus controls, while USP5 overexpression exerted opposing effects in H1650 cells (Figures 3A and S1B). Consistently, transwell assays demonstrated that USP5 silencing significantly attenuated cell migration and invasion, whereas USP5 overexpression enhanced these malignant phenotypes (Figures 3B, S1C, and S1D). Western blot analysis indicated that USP5 depletion reduced vimentin and N-cadherin expression while elevating E-cadherin levels, suggesting USP5-mediated modulation of EMT in LUAD metastasis (Figure 3D).

Figure 3.

USP5 drives aggressive malignant phenotypes in LUAD by enhancing invasion, suppressing apoptosis, and inducing EMT

(A) Wound healing assays demonstrating that USP5 knockdown impairs migration in H1299 cells, while its overexpression enhances migration in H1650 cells. Scale bars: 100 μm.

(B) Transwell invasion assays showing that USP5 knockdown reduces invasion, whereas its overexpression increases invasion in the respective cell lines. Scale bars: 100 μm.

(C) Flow cytometric analysis indicating that USP5 knockdown promotes apoptosis in H1299 and PC9 cells.

(D) Western blot analysis of epithelial-mesenchymal transition and apoptosis-related protein expression following USP5 knockdown in H1299 cells.

Data are presented as the mean ± SD from n = 3 independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus the respective control group.

USP5 binds to and stabilizes CD73

To delineate the molecular mechanism underlying USP5-driven LUAD progression, immunoprecipitation coupled with mass spectrometry (IP-MS) was performed on lysates derived from H1299 cells. This identified high-confidence, USP5-interacting proteins (Figure 4A). Among top candidates, co-immunoprecipitation (coIP) validation excluded the metabolic enzymes TPI1, PKM, and PGK1 as direct binders (Figure 4B), while confirming robust USP5-CD73 complex formation (Figure 4C). Reciprocal coIP assays across multiple LUAD cell lines (Figure 4D) and in H1299/293T cells co-expressing Flag-USP5 and His-CD73 (Figures 4E–4G) consistently confirmed this interaction. Comparative analysis revealed significantly elevated USP5 and CD73 protein levels in LUAD cell lines (H1299, A549, H1650, PC9, HCC827, and H358) versus normal bronchial epithelium (HBE135-E6E7) (Figure 4H). Notably, USP5 knockdown or overexpression did not alter CD73 mRNA levels as measured by RT-qPCR (Figure 4I) but markedly changed CD73 protein expression as observed in western blot analyses (Figures 4J–4M). These results demonstrate that USP5 post-translationally regulates CD73 stability to drive LUAD progression.

Figure 4.

USP5 interacts with and post-transcriptionally stabilizes CD73 in LUAD

(A) Co-immunoprecipitation (CoIP) of USP5 from H1299 lysates followed by mass spectrometry (MS) analysis of Coomassie-stained bands identifies CD73 as a potential interacting partner.

(B) Specificity control: CoIP shows no interaction between USP5 and the glycolytic enzymes TPI1, PKM, or PGK1.

(C and D) Endogenous coIP confirms the USP5-CD73 interaction in H1299 LUAD cells (C) and PC9 LUAD cells (D).

(E–G) Exogenous coIP validates the direct interaction in HEK-293T cells (E and F) and LUAD cells (G) co-expressing Flag-USP5 and His-CD73.

(H) USP5 and CD73 protein levels are coordinately upregulated in LUAD cell lines compared to normal bronchial epithelial cells (HBE135-E6E7).

(I) USP5 knockdown or overexpression does not alter CD73 mRNA levels (RT-qPCR).

(J–M) USP5 positively regulates CD73 protein expression, as shown by its levels upon USP5 knockdown or overexpression. (J and K) CD73 levels decreased upon USP5 knockdown. (L and M) CD73 levels increased upon USP5 overexpression.

Data are presented as the mean ± SD from n = 3 independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

Immunofluorescence assay localized USP5 (green) and CD73 (red) in H1299, PC9, and H1650 cells, with nuclei counterstained using DAPI (blue) (Figure 5A). Analysis of surgically resected LUAD specimens revealed consistent upregulation of USP5 and CD73 in tumor tissues versus paired adjacent non-tumor tissues by both immunohistochemistry (IHC; Figure 5B) and western blotting (Figures 5C–5E). Critically, CD73 expression demonstrated a positive correlation with USP5 levels (Pearson’s r = 0.37, p = 0.04; Figure 5F). Cycloheximide (CHX) chase assays established USP5-dependent regulation of CD73 stability: USP5 knockdown reduced CD73 half-life in H1299 cells, while USP5 overexpression prolonged CD73 half-life in H1650 cells (Figures 5G–H).

Figure 5.

USP5 stabilizes the CD73 protein and their expressions are correlated in LUAD

(A) Immunofluorescence imaging showing subcellular localization of endogenous USP5 (green) and CD73 (red) in LUAD cells. Nuclei are counterstained with DAPI (blue).

(B–E) Validation of USP5 and CD73 co-expression in clinical LUAD specimens by a representative IHC image (B) and western blotting of paired tumor (T) and adjacent normal (N) tissues (C–E).

(F) Scatterplot showing a significant positive correlation between USP5 and CD73 protein levels in LUAD patient samples (Pearson’s r = 0.37, p = 0.04).

(G and H) Cycloheximide chase assays in H1299 and H1650 cells demonstrating that USP5 overexpression slows the degradation of CD73 protein, thereby extending its half-life; data are presented as the mean ± SD from n = 3 independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001.

USP5 deubiquitinates CD73

To determine whether USP5 stabilizes CD73 via its deubiquitinase activity, we assessed CD73 ubiquitination in LUAD cells upon USP5 knockdown or overexpression combined with 20 mM MG132 treatment. As demonstrated in Figures 6A and 6B, USP5 knockdown markedly enhanced CD73 ubiquitination, whereas USP5 overexpression suppressed it. We further profiled USP5-mediated CD73 deubiquitination using antibodies specific for K48- or K63-linked polyubiquitin chains (Figure 6C). The results indicated that USP5 preferentially cleaves K48-linked ubiquitin chains from CD73. Collectively, these data establish that USP5 stabilizes CD73 through a ubiquitination-dependent mechanism. Furthermore, overexpression of the catalytically inactive mutant C335A-USP5 did not alter CD73 ubiquitination, unlike the wild-type protein (Figure S1E).

Figure 6.

USP5-mediated deubiquitination stabilizes CD73

(A and B) Polyubiquitination level of CD73 with USP5 knockdown or overexpression. (A) USP5 knockdown increases the polyubiquitination level of CD73. (B) USP5 overexpression decreases the polyubiquitination level of CD73.

(C) Immunoprecipitation with chain-specific ubiquitin antibodies reveals that USP5 primarily cleaves K48-linked polyubiquitin chains on CD73.

(D) Schematic of the domain architectures of USP5 and CD73 truncation mutants.

(E and F) Co-immunoprecipitation assays map the direct interaction between the catalytic domain of USP5 and the intracellular domain of CD73. Data are representative of at least three independent experiments.

To delineate the USP5-CD73 interaction domains, we designed two USP5 truncation mutants based on UniProt (https://www.uniprot.org/): USP5-F1 (1–283) and USP5-F2 (283–858). Experimental evidence revealed that the CD73-interacting domain resides within the structure of USP5-F2 (Figure 6E). Similarly, we generated two CD73 domain-deletion mutants: CD73-F1 del (29–310) and CD73-F2 del (341–512). This identified the USP5-binding domain within the structure of CD73-F1 (Figure 6F).

USP5 governs glycolytic reprogramming in LUAD

To determine whether USP5 influences LUAD metabolism, we performed metabolomic profiling in H1299 cells with stable USP5 knockdown (shUSP5) and control cells. Principal component analysis (PCA) revealed distinct principal component patterns between the USP5 knockdown and control groups (Figure 7A). Analysis of relative metabolite abundance identified 245 significantly decreased and 214 significantly increased metabolites in USP5 knockdown cells versus controls, visualized by a volcano plot (Figure 7B). A heatmap further illustrated differential metabolite expression between the groups (Figure 7C). Specifically, USP5 knockdown attenuated glycolytic flux, evidenced by reduced intracellular glucose along with diminished glycolytic intermediates and end products (including G6P, F6P, pyruvate, and lactate) compared with controls (Figure 7D).

Figure 7.

USP5 reprograms glycolytic metabolism in LUAD

(A) Principal component analysis of global metabolic profiles shows a clear separation between the control and USP5-knockdown H1299 cells.

(B and C) Unsupervised analysis identifies significantly altered metabolites upon USP5 knockdown, displayed in a volcano plot (B) and heatmap (C).

(D) Targeted metabolomics reveals an impaired glycolytic flux in USP5-knockdown cells, as indicated by the concurrent depletion of key glycolytic intermediates, including glucose-6-phosphate (G6P), fructose-6-phosphate (F6P), pyruvate, and lactate.

(E–I) Functional assessment of glycolysis using a Seahorse XF Analyzer. USP5 knockdown reduces the proton efflux rate (PER), basal glycolysis, and compensatory glycolysis in H1299 cells (E), PC9 cells (F), and HCC827-OR cells (I); USP5 overexpression enhances these parameters in H1650 cells (G) and HCC827 cells (H).

(J) Consistent with the functional data, extracellular lactate secretion decreased by USP5 knockdown and increased by USP5 overexpression.

Data are presented as the mean ± SD from n = 3 independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus the respective control group.

To define USP5’s role in aerobic glycolysis, we quantified the glycolytic flux (proton efflux rate, PER), using Agilent Seahorse XF Glycolysis Rate Assay, and measured lactate production via enzymatic colorimetry. USP5 knockdown significantly suppressed basal glycolysis and compensatory glycolytic rate (Figures 7E, 7F, and 7I), while USP5 overexpression enhanced these parameters (Figures 7G and 7H). Consistent with these findings, USP5 levels substantially altered intracellular lactate accumulation (Figure 7J).

USP5-CD73 axis regulates glycolysis in LUAD cells by targeting HIF-1α and c-Myc to control GLUT1 and LDHA expression

We next assessed USP5-mediated regulation of glycolytic enzymes. USP5 knockdown downregulated HIF-1α (hypoxia-inducible factor 1-alpha), c-Myc, GLUT1 (glucose transporter type 1), and lactate dehydrogenase A (LDHA) protein expression, whereas USP5 overexpression upregulated these factors (Figure 8A). CD73 depletion similarly altered this enzyme network (Figure 8B). Comparative analysis revealed that osimertinib-resistant HCC827 cells (HCC827-RO) exhibited upregulated expressions of USP5, CD73, and key glycolytic effectors—including HIF-1α, c-Myc, GLUT1, LDHA, and MCT4—relative to parental cells (Figure 8C). Collectively, these results establish enhanced glycolysis as a metabolic basis of EGFR-TKI resistance and position the USP5-CD73 axis as a significant regulator of this process in LUAD.

Figure 8.

The USP5-CD73 axis drives glycolytic flux by modulating key glycolytic effectors

(A) USP5 knockdown reduces, while its overexpression increases, the protein levels of CD73, phosphorylated EGFR (p-EGFR), and key glycolytic enzymes.

(B) CD73 knockdown recapitulates the effect of USP5 knockdown on p-EGFR and glycolytic effector expression.

(C) Osimertinib-resistant HCC827-RO cells exhibit elevated levels of USP5, CD73, p-EGFR, and glycolytic proteins compared to their sensitive counterparts.

(D and E) Functional assessment of glycolysis using a Seahorse XF Analyzer. (D) CD73 knockdown reduces the proton efflux rate (PER), basal glycolysis, and compensatory glycolysis in H1299 cells. (E) CD73 knockdown reduces the PER, basal glycolysis, and compensatory glycolysis in PC9 cells.

(F and G) Reconstitution of CD73 expression in USP5-knockdown H1299 cells (F) and USP5-knockdown PC9 cells (G) rescues the impaired glycolytic function.

(H and I) Consistent with the functional assays, CD73 knockdown decreases extracellular lactate secretion (H), while CD73 overexpression restores extracellular lactate secretion (I) in USP5-deficient cells.

Data are presented as the mean ± SD from n = 3 independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus the respective control group.

Furthermore, assessment of glycolytic function via Seahorse extracellular flux analysis (measuring the PER) and enzymatic lactate quantification revealed that CD73 knockdown significantly attenuated both glycolytic flux and lactate accumulation in LUAD cells (Figures 8D, 8E, and 8H). Critically, overexpression of CD73 rescued the impaired glycolysis induced by USP5 knockdown in these cells. These findings indicate that USP5 likely modulates cellular glycolytic flux and lactate production by regulating CD73 (Figures 8F, 8G, and 8I). Collectively, the data demonstrate that the USP5-CD73 axis promotes aerobic glycolysis in LUAD by regulating GLUT1 and LDHA expression through HIF-1α and c-Myc.

USP5 promotes osimertinib resistance in LUAD cells through CD73

To investigate the roles of USP5 and CD73 in osimertinib sensitivity in LUAD, we determined the 72-h IC₅₀ values for osimertinib in multiple LUAD cell lines following USP5 or CD73 knockdown or overexpression. USP5 knockdown significantly sensitized LUAD cells to osimertinib, reducing IC₅₀ values from 11.05 μM to 6.847 μM in H1299, from 18.13 nM to 10.73 nM in PC9, and from 6.038 μM to 3.755 μM in HCC827-RO cells. Conversely, USP5 overexpression markedly increased IC₅₀ values from 4.72 μM to 6.283 μM in H1650 and from 0.037 μM to 0.088 μM in HCC827 cells (Figure 9A). Critically, CD73 overexpression substantially decreased osimertinib sensitivity (increased IC₅₀), thereby rescuing the enhanced therapeutic effect induced by USP5 knockdown (Figure 9A).

Figure 9.

USP5-CD73 axis drives osimertinib resistance in LUAD

(A) Modulation of the USP5-CD73 axis alters osimertinib sensitivity, as shown by the 72-h IC₅₀ values in various LUAD cell lines.

(B) Flow cytometric analysis of apoptosis in H1299 and HCC827-RO cells treated with DMSO, a USP5 inhibitor, osimertinib, or their combination. Co-treatment synergistically induces apoptosis.

(C) In vivo xenograft model demonstrating that inhibition of the USP5-CD73 axis, in combination with osimertinib, significantly suppresses tumor growth and reduces final tumor weight.

(D) Immunohistochemical (IHC) analysis of xenograft tumors shows downregulation of USP5, CD73, and the proliferation marker Ki67 in the combination treatment group. Representative H&E-stained lung sections are also shown.

(E) Western blot analysis confirms that perturbation of the USP5-CD73 axis modulates the expression of key glycolytic enzymes and p-EGFR levels in LUAD cells.

Data are presented as the mean ± SD from n = 3 independent experiments (A, B, and E) or n = 5 mice per group (C). ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 versus the indicated control groups.

Knockdown of USP5 induced pronounced apoptosis in both H1299 and PC9 cell lines (Figure 3C). Furthermore, USP5 knockdown in H1299 cells induced a pro-apoptotic shift, evidenced by an upregulated expression of apoptotic effectors (Bax, cleaved caspase-3, and cleaved caspase-9) and concomitant downregulation of antiapoptotic regulators (Bcl-2, caspase-3, and caspase-9) (Figure 3D). We further assessed apoptosis induction by the USP5 inhibitor USP5-IN-1 and osimertinib, alone or in combination, using flow cytometry in LUAD cells. Treatment with 25 μM USP5-IN-1 or 5 μM osimertinib for 24 h induced apoptosis in H1299 and HCC827-RO cells, with the combination treatment demonstrating significantly enhanced pro-apoptotic effects compared with either of the agents (Figure 9B). Additionally, western blot analysis revealed that USP5 knockdown reduced p-EGFR expression in LUAD cells, an effect rescued by CD73 overexpression (Figure 9E).

To investigate the targeting of USP5 by its inhibitor, USP5-IN-1, we performed western blot analysis. The results showed that treatment with USP5-IN-1 led to a marked decrease in the protein levels of both USP5 and CD73 in LUAD cells (Figure S1F). Furthermore, treatment of LUAD cells (H1299 and PC9) with USP5-IN-1 for 48 h resulted in significant dose-dependent cell death, as detected by flow cytometry (Figures S2A, S2B, and S2E). Concurrently, a notable loss of mitochondrial membrane potential was observed in the treated cells (Figures S2C, S2D, and S2F). At the molecular level, USP5-IN-1 treatment markedly enhanced the cleavage of key apoptosis executioners, including PARP, caspase-3, and caspase-9, accompanied by an upregulation of the pro-apoptotic protein Bax and downregulation of the antiapoptotic protein Bcl-2 (Figure S2G). Based on these findings, we further hypothesized that USP5-IN-1 might directly induce DNA damage. Western blot analysis revealed that the treatment with USP5-IN-1 led to a significant reduction in key DNA repair proteins, such as CHK1, 53BP1, and RAD51, along with a pronounced increase in the DNA damage marker γ-H2AX (Figure S2H), supporting the notion that USP5-IN-1 triggers DNA damage response pathways.

USP5 promotes proliferation, migration, and invasion in LUAD cells by targeting CD73

Given the established roles of USP5 in promoting malignant progression and regulating CD73 expression in LUAD, we hypothesized that USP5 drives LUAD progression partially through CD73. To test this, we induced CD73 overexpression in USP5-knockdown H1299 and PC9 cells. Cell migration and invasion were analyzed by transwell and wound healing assays, while proliferation was quantified via colony formation and CCK-8 assays. Functional analyses demonstrated that CD73 overexpression enhanced LUAD cell proliferation, migration, and invasion. Critically, these pro-tumorigenic effects were substantially reversed by USP5 knockdown (Figures 10A–10D). Collectively, these results indicate that USP5 accelerates LUAD malignancy by targeting CD73.

Figure 10.

CD73 is a critical downstream effector for USP5-mediated oncogenic phenotypes in LUAD

(A and C) USP5 knockdown in H1299 cells (A) and PC9 cells (C) impairs migration, invasion, and colony formation. These phenotypic defects are rescued by the concomitant overexpression of CD73, as assessed by transwell, wound healing, and colony formation assays (scale bars: 100 μm).

(B and D) Cell viability (CCK-8 assays) in H1299 cells (B) and PC9 cells (D) is reduced upon USP5 knockdown but is restored by CD73 overexpression.

Data are presented as the mean ± SD from n = 3 independent experiments; ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001. Statistical comparisons are shown for relevant groups.

To investigate the impact of the USP5-CD73 axis on malignant progression in LUAD, xenograft tumor models were established in nude mice. Stably transfected PC9 cells were subcutaneously injected into the mice. Compared to the control (shCtrl) group, USP5 knockdown (shUSP5) significantly reduced tumor volume and weight. Conversely, CD73 overexpression markedly promoted tumor growth. Critically, USP5 depletion reversed the pro-tumorigenic effect of CD73 overexpression (Figure 9C). Immunohistochemical staining revealed that USP5 suppression led to decreased levels of the proliferation marker Ki67 in subcutaneous tumors, while CD73 overexpression produced the opposite effect (Figure 9D). Furthermore, elevated CD73 expression significantly counteracted the USP5 knockdown-triggered reduction in lung metastasis (Figure 9D). Collectively, these in vivo findings establish that USP5 modulates tumor growth and metastasis in LUAD through CD73.

TRIM28 acts as an E3 ubiquitin ligase to promote CD73 degradation

Knockdown of TRIM28 in LUAD cells increased the expression level of CD73 (Figure 11A), while overexpression of TRIM28 reduced CD73 expression (Figure 11B). CoIP assays further demonstrated a robust protein-protein interaction between TRIM28 and CD73 in LUAD cells (Figure 11C). CHX chase assays indicated that TRIM28 knockdown extended the half-life of CD73 in H1299 cells, whereas TRIM28 overexpression shortened the half-life of CD73 in HCC827 cells (Figures 11D–11E). Moreover, TRIM28 knockdown markedly attenuated the ubiquitination of CD73, and conversely, TRIM28 overexpression enhanced its ubiquitination level (Figures 11F–11G). Building on these findings, western blot analysis revealed that the increase in CD73 expression induced by TRIM28 knockdown could be partially rescued by the concurrent knockdown of USP5 (Figure S1G).

Figure 11.

TRIM28 functions as an E3 ubiquitin ligase for CD73 and promotes its proteasomal degradation

(A and B) CD73 protein expression is increased upon TRIM28 knockdown (A) and decreased upon TRIM28 overexpression (B) in LUAD cells.

(C) Endogenous co-immunoprecipitation confirms a physical interaction between TRIM28 and CD73.

(D and E) Cycloheximide chase assays demonstrate that TRIM28 overexpression shortens the half-life of CD73 (D), whereas TRIM28 knockdown extends it (E).

(F and G) TRIM28 knockdown reduces the polyubiquitination level of CD73 (F), while its overexpression enhances the polyubiquitination level of CD73 (G).

Data are representative of three independent experiments.

USP5 promotes LUAD progression via the CD73-mediated PI3K/AKT/mTOR signaling pathway

Finally, we investigated the mechanism by which USP5 promotes growth and metastasis in LUAD cells. Compared to controls, USP5 knockdown reduced phosphorylation levels of mTOR, PI3K, and AKT, whereas USP5 overexpression enhanced their phosphorylation (Figure 12A). However, total protein levels of mTOR, PI3K, and AKT remained unaltered. Similarly, CD73 knockdown recapitulated these phosphoprotein changes (Figure 12B), and CD73 overexpression rescued the signaling alterations induced by USP5 depletion (Figure 12C). These results demonstrate that the USP5-CD73 axis modulates the PI3K/AKT/mTOR signaling pathway. Notably, compared with the parental HCC827 cells, the osimertinib-resistant HCC827-RO line also exhibited elevated phosphorylation of mTOR, PI3K, and AKT without changes in total protein levels (Figure 12B), suggesting a strong association between osimertinib resistance and activation of this pathway.

Figure 12.

The USP5-CD73 axis activates the PI3K/AKT/mTOR pathway to drive LUAD progression and confers poor clinical outcomes

(A–C) Western blot analysis shows that perturbation of the USP5-CD73 axis modulates the activation of key components within the PI3K/AKT/mTOR signaling pathway.

(D) In vivo xenograft model demonstrates that co-targeting USP5 (with USP5-In-1) and EGFR (with osimertinib) synergistically inhibits tumor growth (tumor volume and weight are presented as the mean ± SD).

(E) Representative immunohistochemistry images of USP5 and CD73 expression in human LUAD tissues.

(F) Quantitative analysis confirms the significant upregulation of both USP5 and CD73 in tumor tissues compared with paired adjacent normal tissues from 58 LUAD patients.

(G) Pearson’s correlation analysis revealing a significant association between USP5 and CD73 expression levels in 58 paired LUAD tissues.

(H) Kaplan-Meier survival curves demonstrating that high expression of both USP5 and CD73 correlates negatively with overall survival in LUAD patients (log-rank test).

(I) Schematic of the proposed mechanism by which the USP5-CD73 axis drives LUAD progression. (p < 0.05, ∗p < 0.01, ∗∗p < 0.001).

Xenograft tumor models were established using PC9 cells in nude mice. The mice received daily intraperitoneal injections of 0.9% NaCl (control), USP5-IN-1 (15 mg/kg), or osimertinib (5 mg/kg) for three weeks. Compared with the control, the treatment with either USP5-IN-1 or osimertinib significantly suppressed tumor growth. Furthermore, the combination of both agents exhibited greater suppression (Figure 12D).

IHC analysis was performed on 58 paired paraffin-embedded specimens of LUAD and matched adjacent normal tissues obtained from patients who underwent surgical resection at the Second Affiliated Hospital of Nanchang University. The staining results showed marked upregulation of both USP5 and CD73 protein expression in LUAD tissues compared with their normal counterparts (Figures 12E and 12F). Furthermore, a statistically significant positive correlation was observed between USP5 and CD73 expression levels in LUAD specimens (r = 0.3145, p = 0.0162). We next evaluated the association of USP5 and CD73 expression with clinicopathological parameters. High USP5 expression was significantly correlated with patients’ age, metastasis, and EGFR expression status, whereas elevated CD73 levels were associated with metastasis and EGFR expression (Table 1). Kaplan-Meier survival analysis revealed that a high expression of either USP5 (p = 0.0058) or CD73 (p = 0.016) predicted reduced overall survival in the LUAD cohort (Figure 12H). Collectively, these clinical findings support the proposed model in which the USP5-CD73 axis contributes to LUAD pathogenesis, as summarized in Figure 12I.

Table 1.

Correlations between the USP5 and CD73 expression levels and clinical characteristics

Characteristics	Total (N = 58)	USP5		p value	CD73		p value
		Low	High		Low	High
Age (year)				0.031			0.746
≤65	43	20	23		18	25
>65	15	2	13		7	8
Gender				0.985			0.562
Male	37	14	23		17	20
Female	21	8	13		8	13
Smoker				0.1115			0.1529
Yes	34	10	24		12	22
No	24	12	12		13	11
Metastasis				0.014			0.004
Positive	33	8	25		8	25
Negative	25	14	11		17	8
Clinical stage				0.905			0.915
I&II	39	15	24		17	22
III	19	7	12		8	11
Tumor size				0.955			0.853
≤3.5 cm	34	13	21		15	19
>3.5 cm	24	9	15		10	14
EGFR				0.021			0.014
+	27	6	21		7	20
-	31	16	15		18	13

Discussion

Mounting evidence underscores the critical role of DUBs in lung cancer progression. This study systematically establishes the significant oncogenic function of USP5 in LUAD. Analysis of public databases and clinical specimens revealed that USP5 is aberrantly overexpressed in LUAD tissues, and its elevated expression correlates significantly with shortened overall patient survival and advanced clinical tumor stage, indicating its crucial role in LUAD progression. Experimentally, USP5 was shown to critically regulate multiple oncogenic processes: USP5 overexpression enhanced cancer cell proliferation and migration, while USP5 depletion markedly suppressed proliferation, migration, and invasion in vitro, as well as inhibited tumor growth and metastasis in vivo. Consequently, USP5 emerges as a promising biomarker and therapeutic target in LUAD. Subsequent proteomic screening identified CD73 as a key downstream substrate of USP5. Sufficient experimental evidence demonstrates that USP5 binds to CD73, effectively stabilizing it by removing K48-linked polyubiquitin chains and protecting it from proteasomal degradation. Furthermore, we confirmed that USP5 promotes LUAD malignancy and confers resistance to EGFR-TKIs by activating the CD73-mediated PI3K/AKT/mTOR signaling pathway. Additionally, the USP5-CD73 axis enhances aerobic glycolysis in LUAD by upregulating HIF-1α and c-Myc, leading to increased expression of the glycolytic enzymes GLUT1 and LDHA. Collectively, these findings demonstrate that the USP5-CD73 axis plays a pivotal role in LUAD malignant progression, highlighting its potential as a therapeutic target and suggest that effective inhibitors targeting this axis may serve as an important strategy for LUAD treatment.

Functionally, USP5—a member of the deubiquitinating enzyme family—primarily exerts its biological effects by stabilizing and deubiquitinating downstream target proteins. Zhang et al. demonstrated that USP5 is overexpressed in bladder cancer and promotes progression by stabilizing c-Jun.³³ Similarly, Meng et al. reported that USP5 stabilizes SLUG to facilitate EMT in hepatocellular carcinoma.³⁴ USP5 also binds and stabilizes EphA2 to confer radioresistance in nasopharyngeal carcinoma.³⁵ In this study, we identified CD73 as a direct substrate of USP5 through coIP and protein stability assays. USP5 specifically interacts with CD73 via its C-terminal domain and utilizes its deubiquitinase activity to remove K48-linked ubiquitin chains from CD73. This process effectively blocks CD73 degradation via the ubiquitin-proteasome pathway, significantly prolonging its half-life and enhancing protein stability. These findings provide a novel mechanistic explanation for CD73 overexpression in LUAD tissues: USP5-mediated deubiquitination critically regulates CD73 homeostasis. Experimental validation confirmed this regulatory mechanism: USP5 knockdown significantly reduced CD73 protein levels while increasing its ubiquitination. Conversely, USP5 overexpression in LUAD cells upregulated CD73 and decreased its ubiquitination. Furthermore, USP5 overexpression substantially extended CD73 half-life upon CHX treatment, underscoring the importance of USP5-mediated deubiquitination for CD73 stability.

Notably, the specific binding domains governing USP5-substrate interactions remain poorly characterized. By cleaving the USP5 sequence into two domain peptides, F1 (1–283) and F2 (283–858), we found that only F2 can bind to CD73. Correspondingly, we synthesized two domain-deficient peptides, F1-del (29–310) and F2-del (341–512), based on the CD73 sequence. Experimental results showed that only F1 could bind to USP5. Functionally, USP5 significantly enhances LUAD cell proliferation, migration, invasion, and xenograft tumor growth in nude mice by stabilizing CD73. Notably, rescue experiments demonstrated that CD73 overexpression in USP5-knockdown LUAD cells reversed the suppressed proliferation and metastasis phenotypes, indicating that USP5 promotes LUAD malignancy predominantly through CD73 stabilization.

This study identifies the USP5-CD73 axis as a critical nexus for metabolic reprogramming in LUAD. CD73 (ecto-5′-nucleotidase), a cell surface-bound nucleotidase, serves as a key metabolic and immune checkpoint. Research indicates that CD73 is widely expressed on immune cells, functioning as a T cell co-stimulatory molecule and an adenosine-producing factor, thereby playing pivotal roles in immune evasion, cell adhesion, and migration. Moreover, CD73 regulates NAD transport and metabolism, modulating tumorigenic processes to promote growth and proliferation.^6,36 Cancer cells frequently exhibit enhanced aerobic glycolysis (the Warburg effect), which fuels tumor invasion, immune evasion, and chemotherapeutic drug efflux by rapidly generating ATP, accumulating macromolecule precursors, and acidifying the TME, ultimately contributing to therapy tolerance and metastasis.^37,38 Our metabolomic analysis revealed that USP5 knockdown significantly reduced glycolytic intermediates and end products in LUAD cells, particularly intracellular glucose content, suggesting impaired glucose uptake. Corroborating this finding, USP5 knockdown attenuated both basal and compensatory glycolysis rates, as well as lactate production. Conversely, USP5 overexpression potentiated the Warburg effect phenotype in LUAD cell lines. Furthermore, we validated the expression changes of nine core glycolytic genes following USP5 or CD73 knockdown or overexpression.^38,39 We found that the USP5-CD73 axis specifically drives the expression of GLUT1 and LDHA. GLUT1, the principal regulator of glucose uptake in tumor cells, accelerates glucose influx to drive the Warburg effect. LDHA, a central executor of the Warburg effect, maintains glycolytic flux by converting pyruvate to lactate, simultaneously driving TME acidification, immunosuppression, and pre-metastatic niche formation, thereby directly coupling metabolic reprogramming with malignant progression.⁴⁰ Intriguingly, the USP5-CD73 axis also promoted the protein expression of HIF-1α and c-Myc, key transcription factors facilitating tumor cell metabolic adaptation by regulating glycolytic gene expression.⁴¹ Collectively, our findings demonstrate that the USP5-CD73 axis promotes the Warburg effect, significantly contributing to LUAD malignant progression.

Interestingly, this study proposes a novel regulatory mechanism involving TRIM28-mediated degradation of CD73. Given the pleiotropic nature of TRIM28 in oncogenic signaling and DNA damage response,³² its regulation of CD73 may further influence tumor cell metabolic adaptability, invasive potential, and responses to chemotherapeutic agents. These findings provide a theoretical foundation for targeting the TRIM28-CD73 axis in lung cancer and other malignancies. Future studies should place greater emphasis on in vivo experiments to further elucidate the mechanisms of this axis within the TME. Such investigations could facilitate the development of novel combination treatment strategies, potentially enhancing immunotherapy efficacy through coordinated disruption of CD73-mediated immunosuppression.

The etiology and pathogenesis of LUAD remain incompletely understood. However, substantial evidence implicates dysregulation of diverse intracellular signaling pathways. Among the most frequently dysregulated pathways observed in cancer patients is the PI3K/AKT/mTOR signaling cascade, which governs cellular responses related to proliferation, differentiation, energy metabolism, and growth.⁴² Under physiological conditions, PI3K signaling is tightly regulated; however, alterations in this balance induced by oncoproteins or tumor suppressors disrupt cellular metabolism and signal regulation, thereby promoting tumor progression.⁴³ Notably, mutations in receptor tyrosine kinases (RTKs) represent one of the most prominent alterations in NSCLC and serve as key initiators of PI3K/AKT/mTOR pathway activation. Studies indicate that 67% of patients harboring EGFR mutations—the primary driver and therapeutic target in LUAD—exhibit concomitant activation of the AKT/mTOR pathway, underscoring its significant role in LUAD progression.⁴⁴ In this study, we found that the knockdown or overexpression of USP5 led to significant decreases or increases, respectively, in the phosphorylation levels of PI3K, AKT, and mTOR proteins, without necessarily altering their total protein expression. Similar observations were obtained following CD73 knockdown in LUAD cells. Furthermore, key proteins within the PI3K/AKT/mTOR pathway exhibited these pronounced changes in HCC827 osimertinib-resistant cells compared with their parental sensitive counterparts. Collectively, our findings demonstrate that USP5 regulates PI3K/AKT/mTOR pathway activation by stabilizing CD73, potentially mediating LUAD cell proliferation, migration, invasion, and osimertinib resistance.

A notable experimental observation in our study was that despite the high resistance index (RI = 163) of HCC827-RO cells, phosphorylated EGFR (p-EGFR) levels were significantly reduced. This phenomenon reveals a core mechanism of osimertinib resistance: diminished EGFR signaling dependence coupled with compensatory activation of bypass or downstream pathways. Two central mechanisms underlie this compensatory bypass kinase signaling: MET overexpression activating the downstream PI3K/AKT pathway, substituting for EGFR function⁴⁵; and HER2/HER3 amplification, where HER2 heterodimers activate RAS/RAF/MEK, circumventing EGFR inhibition.⁴⁶ A previous study indicated that in EGFR-mutant, TKI-resistant lung cancer, MET amplification can upregulate CD73 to suppress tumor cell STING induction and T cell reactivity.⁴⁷ Crucially, in this study, the USP5-CD73 axis was demonstrated to activate the PI3K/AKT signaling pathway.

Furthermore, EMT constitutes another key mechanism of EGFR-TKI resistance. This is characterized by the upregulation of mesenchymal markers (vimentin and N-cadherin) and downregulation of the epithelial marker E-cadherin, frequently featuring reduced p-EGFR alongside enhanced cancer cell invasiveness.⁴⁸ Furthermore, western blot analysis revealed that USP5 or CD73 knockdown significantly suppressed vimentin and N-cadherin expression while enhancing E-cadherin levels, indicating the involvement of USP5-CD73 in the EMT of LUAD cells. Concordantly, osimertinib-resistant HCC827 cells exhibited higher vimentin and N-cadherin expression and lower E-cadherin levels compared with their parental sensitive counterparts (Figure S1H). Moreover, USP5 knockdown markedly enhanced osimertinib sensitivity in LUAD cells, whereas USP5 overexpression conversely reduced sensitivity. Critically, CD73 overexpression rescued the therapeutic effect of osimertinib in USP5-knockdown cells.

Collectively, our findings demonstrate that the USP5/CD73 axis drives osimertinib resistance through four interconnected mechanisms: (1) metabolic reprogramming-induced lactate accumulation that acidifies the TME to promote invasion; (2) compensatory activation of PI3K/AKT/mTOR-HIF-1α signaling crosstalk⁴⁹; (3) adenosine- and lactate-mediated immune editing characterized by suppression of effector T cell function and recruitment of Tregs to establish an immune-privileged niche; and (4) enhanced tumor cell invasiveness via EMT. Collectively, this work provides compelling rationale for dual targeting of the USP5-CD73 axis as a promising therapeutic strategy against EGFR-TKI resistance.

The complex heterogeneity of tumors and the persistent emergence of drug resistance pose significant challenges in clinical cancer therapy. While researchers have continually been developing novel agents targeting resistance mechanisms, overcoming resistance remains a persistent challenge. Within the ubiquitin-proteasome system, proteasome inhibitors (e.g., bortezomib, carfilzomib, oprozomib, and ixazomib) have been successfully developed for clinical use, demonstrating substantial efficacy.⁵⁰ Recent years have also witnessed significant progress in elucidating the roles of USPs in human cancer progression. Multiple USPs contribute to tumorigenesis and progression by modulating key cancer-associated signaling pathways, including p53, NF-κB, and PI3K/AKT.⁵⁰ These insights have spurred the development of potent, small-molecule USP inhibitors with anticancer activity. For instance, PROTAC-based potent USP7 inhibitors selectively target USP7, impacting tumor progression in lung and prostate cancers.⁵¹ The USP28 inhibitor AZ1 induces DNA damage in NSCLC cells and synergizes with cisplatin to enhance therapeutic efficacy.⁵² The USP1 inhibitor pimozide disrupts glioma stem cell maintenance and radioresistance.⁵³ Furthermore, mitoxantrone has been identified as a USP11 inhibitor, effectively suppressing pancreatic cancer cell proliferation.⁵⁴ Currently developed USP5 inhibitors, including WP1130,⁵⁵ EOAI3402143,⁵⁶ and USP5-IN-1,³⁵ have demonstrated antitumor effects in both in vitro and in vivo models. In this study, we identified that the USP5-CD73 axis promotes LUAD resistance to osimertinib. Critically, combining the USP5 inhibitor USP5-IN-1 with osimertinib synergistically enhanced apoptosis induction in vitro and exerted superior tumor growth suppression in vivo compared with either monotherapy. Therefore, targeting USP5, either alone or in combination with CD73, represents a promising therapeutic strategy to overcome EGFR-TKI resistance in LUAD when combined with existing EGFR-TKIs.

This study unveils, for the first time, the pivotal role of USP5 as a deubiquitinating regulator upstream of CD73, elucidating its dual-mechanism orchestration of osimertinib resistance in LUAD. Mechanistically, the USP5-CD73 axis activates the PI3K/AKT/mTOR cascade, suppressing drug-induced apoptosis; concurrently, it enhances glycolytic flux, providing survival-essential energy substrates for tumors and fostering an immunosuppressive TME. Collectively, high co-expression of USP5 and CD73 emerges as a novel biomarker for prognostic stratification and prediction of osimertinib response in LUAD patients. Therapeutically, targeting the USP5-CD73 axis offers a dual approach: (1) developing inhibitors against the USP5-CD73 protein-protein interaction interface, using structure-guided peptidomimetics, and (2) implementing combinatorial strategies such as pairing USP5 inhibitors with osimertinib or concurrently blocking downstream effectors (e.g., the PI3Kβ/δ inhibitor AZD8186 plus LDHA inhibition).

In this study, we demonstrate that the USP5-CD73 axis is upregulated in LUAD and correlates with poor patient prognosis. Mechanistically, USP5-mediated CD73 stabilization concurrently activates PI3K/AKT/mTOR signaling and enhances aerobic glycolysis, ultimately promoting osimertinib resistance and malignant progression. Targeting this USP5-CD73 cascade, thus, represents a promising therapeutic strategy for LUAD.

Limitations of the study

While our study established a novel USP5-CD73 regulatory axis in LUAD progression and osimertinib resistance, it has limitations that warrant discussion. First, the competitive dynamics between USP5 and TRIM28 for CD73 binding remain to be fully elucidated; future studies should employ competitive coIP and dual modulation assays to dissect this ubiquitination-deubiquitination switch. Second, although the dose-dependent effects of USP5-IN-1 support its on-target activity, definitive validation using a drug-resistant USP5 mutant is needed and has been planned in follow-up studies. Furthermore, our mechanistic focus on cell-intrinsic metabolic functions precluded the exploration of CD73-derived adenosine in the TME. Future work employing immunocompetent models will be essential to unravel how this axis modulates immunosuppression and therapeutic outcomes. Notwithstanding these points, our findings offer a strong preclinical rationale for targeting this axis. The differential overexpression of USP5 and CD73 in tumors suggests a potential therapeutic window. We propose that USP5/CD73 expression could inform patient stratification and that a phase 1b trial of a USP5 inhibitor combined with osimertinib represents a logical next step for overcoming resistance in advanced LUAD.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Junping Xie (junpingxie2023@126.com).

Materials availability

The study did not generate new materials.

Data and code availability

• •All data reported in this paper are available from the lead contact upon request.
• •The metabolomic dataset is available online at Figshare dataset: https://doi.org/10.6084/m9.figshare.30929006. If you use the metabolomic dataset, please cite both the dataset and the primary research article: Chen R, Han XH, Zhang Z, Han XJ, Xie JP. (2026). Metabolomic Data on USP5 Knockdown in Lung Adenocarcinoma Cells.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

We acknowledge the use of Figdraw (https://www.Figdraw.com) to create the mechanism diagram. The present study was funded by the National Natural Science Foundation of China (grant nos. 81160294 and 81960425) and the Jiangxi Province Postgraduate Innovation Special Fund Project (grant no. YC2024-B071).

Author contributions

R.C. conceived the study, designed and performed experiments, analyzed data, and drafted the manuscript; X.-H.H. assisted in completing experiments in vitro; X.-J.H. and Z.Z. critically reviewed and edited the manuscript; J.X. supervised the research and substantively revised the manuscript. All authors have approved the final version.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
USP5	Proteintech	Cat# 10473-1-AP; RRID:AB_2272754
CD73	Abcam	Cat# ab317462; RRID:AB_3739483
TRIM28/KAP1	Proteintech	Cat# 66630-1-Ig; RRID:AB_2732886
GAPDH	ABclonal	Cat# A19056; RRID:AB_2862549
β-Actin	Proteintech	Cat# 66009-1-Ig; RRID:AB_2687938
Flag/DYKDDDDK tag	Proteintech	Cat# 20543-1-AP; RRID:AB_11232216
His-Tag	Proteintech	Cat# 66005-1-Ig; RRID:AB_11232599
Ubiquitin	Proteintech	Cat# 10201-2-AP; RRID:AB_671515
Anti-Ubiquitin(linkage-specific K48)	Abcam	Cat# ab140601; RRID:AB_2783797
Anti-Ubiquitin(linkage-specific K63)	Abcam	Cat# ab179434; RRID:AB_2895239
VeriBlot for IP Detection(HRP)	Abcam	Cat# ab131366; RRID:AB_2892718
E-cadherin	Proteintech	Cat# 20874-1-AP; RRID:AB_10697811
N-cadherin	Proteintech	Cat# 22018-1-AP; RRID:AB_2813891
Vimentin	Proteintech	Cat# 10366-1-AP; RRID:AB_2273020
Caspase 3	Abways Technology	CY5051; RRID:AB_3739491
Cleaved-caspase 3	Abcam	Cat# ab32042; RRID:AB_725947
Caspase 9	Abways Technology	CY5782; RRID:AB_3739492
Cleaved-caspase 9	Abways Technology	CY5682; RRID:AB_3739493
HIF-1α	Proteintech	Cat# 20960-1-AP; RRID:AB_10732601
c-Myc	Proteintech	Cat# 10828-1-AP; RRID:AB_2148585
p-EGFR(Y1197)	Abways Technology	CY5614; RRID:AB_3739494
GLUT1	Proteintech	Cat# 21829-1-AP; RRID:AB_10837075
Hexokinase II (HK2)	Zen-Bioscience	Cat# R24552; RRID:AB_3719314
PFKM	Proteintech	Cat# 55028-1-AP; RRID:AB_10858390
PKM2	Proteintech	Cat# 15822-1-AP; RRID:AB_1851537
GPI	Proteintech	Cat# 15171-1-AP; RRID:AB_2263537
ENO1	Proteintech	Cat# 11204-1-AP; RRID:AB_2099064
PGK1	Proteintech	Cat# 17811-1-AP; RRID:AB_2161218
LDHA	Proteintech	Cat# 19987-1-AP; RRID:AB_10646429
MCT4	Proteintech	Cat# 22787-1-AP; RRID:AB_11182479
PKM	Abways Technology	Cat# CY5764; RRID:AB_3675888
TPI1	Proteintech	Cat# 10713-1-AP; RRID:AB_2207716
Bax	Proteintech	Cat# 50599-2-Ig; RRID:AB_2061561
Bcl-2	Proteintech	Cat# 26593-1-AP; RRID:AB_2818996
PARP1	Proteintech	Cat# 13371-1-AP; RRID:AB_2160459
Cleaved-PARP 1 (D214)	Abways Technology	CY5035; RRID:AB_3739495
FANCD2	Proteintech	Cat# 28619-1-AP; RRID:AB_2881182
RAD51	Proteintech	Cat# 14961-1-AP; RRID:AB_2177083
XRCC1	Proteintech	Cat# 21468-1-AP; RRID:AB_2878865
CHK1	Proteintech	Cat# 60277-1-Ig; RRID:AB_2881396
53BP1	Abways Technology	CY6831; RRID:AB_3739497
γ-H2AX	Proteintech	Cat# 68888-1-Ig; RRID:AB_3670446
PCNA	Proteintech	Cat# 10205-2-AP; RRID:AB_2160330
mTOR	Proteintech	Cat# 66888-1-Ig; RRID:AB_2882219
p-mTOR(Ser2448)	Proteintech	Cat# 67778-1-Ig; RRID:AB_2889842
AKT	Affinity Biosciences	Cat# AF6261; RRID:AB_2835121
p-AKT(Ser473)	Affinity Biosciences	Cat# AF0016; RRID:AB_2810275
PI3K	Affinity Biosciences	Cat# AF6241; RRID:AB_2835340
p-PI3K(Tyr607)	Affinity Biosciences	Cat# AF3241; RRID:AB_2834667
Ki-67	Proteintech	Cat# 27309-1-AP; RRID:AB_2756525
Rabbit IgG control Polyclonal antibody	Proteintech	Cat# 30000-0-AP; RRID:AB_2819035
Goat Anti-Rabbit IgG H&L(HRP)	Zen-Bioscience	Cat# 511203; RRID:AB_2927753
Goat Anti-Mouse IgG H&L(HRP)	Zen-Bioscience	Cat# 511103; RRID:AB_2893489
Bacterial and virus strains
pHBLV-CMV-MCS-3FLAG-EF1-ZsGreen-T2A-PURO	Hanbio Tech (Shanghai, China)	N/A
pHBLV-U6-MCS-CMV-ZsGreen-PGK-PURO	Hanbio Tech (Shanghai, China)	N/A
LV17(EF-1a/Luciferase17&Puro)-6∗his	GenePharma (Shanghai, China)	N/A
Biological samples
Surgical specimens from patients with lung adenocarcinoma	The Second Affiliated Hospital of Nanchang University	N/A
Chemicals, peptides, and recombinant proteins
Puromycin Dihydrochloride	Beyotime	Cat#ST551
Protein A/G Magnetic Beads	MedChemExpress	Cat#HY-K0202
Anti-Flag Magnetic Beads	MedChemExpress	Cat#HY-K0207
Anti-His Magnetic Beads	MedChemExpress	Cat#HY-K0209
MG132	MedChemExpress	Cat#HY-13259
Cycloheximide (CHX)	MedChemExpress	Cat#HY-12320
Osimertinib	MedChemExpress	Cat#HY-15772
USP5-IN-1	Targetmol	Cat#T60130
Lipofectamine 3000 Transfection kit	Invitrogen	Cat#L3000-015
Critical commercial assays
L-Lactic Acid Colorimetric Assay Kit	Elabscience	Cat#E-BC-K044-M
Seahorse XF Glycolysis Rate Assay Kit	Agilent Technologies	Cat#103344-100
RNA-Quick Purification Kit	Esun Bio	Cat#RN001-50
YF® 594 Click-iT™ EdU Kit	UElandy	Cat#C6045L
FITC-Annexin V/propidium iodide (PI) Kit	UElandy	Cat#F6012L
Experimental models: Cell lines
HBE135-E6E7 cell line	Senbeijia Biotechnology Co., Ltd (Nanjing, China)	Cat#BC-C-HU-012
H1299 cell line	The National Collection of Authenticated Cell Cultures	Serial:TCHu160 Identifier:CSTR:19375.09.3101HUMTCHu160
H1650 cell line	The National Collection of Authenticated Cell Cultures	Serial:SCSP-592 Identifier:CSTR:19375.09.3101HUMSCSP592
H358 cell line	The National Collection of Authenticated Cell Cultures	Serial:SCSP-583 Identifier:CSTR:19375.09.3101HUMSCSP583
PC9 cell line	ServiceBio Co., Ltd. (Hubei, China)	Cat#STCC10210P
HCC827 cell line	ServiceBio Co., Ltd. (Hubei, China)	Cat#STCC10212P
A549 cell line	ServiceBio Co., Ltd. (Hubei, China)	Cat#STCC10201P
Experimental models: Organisms/strains
Mouse: BALB/cJFoxn1<nu>/J	Sibeifu Biotechnology Co., Ltd. (Beijing, China)	N/A
Oligonucleotides
siRNA targeting sequence: siCD73#1: CGGAUGAAAUGUUCUGGAATT; siCD73#2: GGAUACACUUCCAAAGAAATT	GenePharma (Shanghai, China)	N/A
Primer: USP5 Forward: AGAAGACAGACAAGACGATGACT Reverse: ACCTGGACCACAGAGTTGAG	This paper	N/A
Primer: GAPDH Forward:GGTGTGAACCATGAGAAGTATGA Reverse: GAGTCCTTCCACGATACCAAAG	This paper	N/A
Primer: CD73 Forward: AGTACCAGGGCACTATCTGGT Reverse: TGAGGAGTGGCTCGATCAGT	This paper	N/A
Recombinant DNA
Plasmid: pcDNA3.1-EF1a-mcs-3flag-CMV-EGFP	Hanbio Tech (Shanghai, China)	N/A
pHBLV-h-shRNA-USP5#1: GGAGCTGACGTGTACTCAT; pHBLV-h-shRNA-USP5#2: CTTTGCCTTCATTAGTCACAT	Hanbio Tech (Shanghai, China)	N/A
pHBLV-h-shRNA-TRIM28#1: GCTGAAGAAGCTGGACAAGAA; pHBLV-h-shRNA-TRIM28#2: CCTGAGATTGACAGAAAGTAA	Hanbio Tech (Shanghai, China)	N/A
Software and algorithms
ImageJ	NIH	https://imagej.nih.gov/ij/; RRID: SCR_003070
Prism (Graphpad)	GraphPad software Version 10.1.2	https://www.graphpad.com/
Seahorse Wave(Agilent)	Agilent Seahorse Wave Desktop software (version 2.6.1; Agilent Technologies)	https://www.agilent.com/en/product/cell-analysis/real-time-cell-metabolic-analysis/xf-software/seahorse-wave-desktop-software-740897
R	R (version 4.3.3)	https://cran.r-project.org/
CytExpert (Beckman Coulter)	Beckman Coulter software Version 2.4	https://www.beckman.com/flow-cytometry/software/cytexpert
Adobe Illustrator	Adobe Illustrator (2023)	https://www.adobe.com/creativecloud.html
Deposited data
Metabolomics data available online at Figshare	Figshare	https://doi.org/10.6084/m9.figshare.30929006

Experimental model and study participant details

Human tissue samples

A cohort of 58 patients (37 male and 21 female) diagnosed with LUAD, who underwent surgical resection at our institution between January 2018 and December 2019 and had complete follow-up data, were enrolled in this study. Critically, none of the patients had received preoperative radiotherapy, chemotherapy, targeted therapy, immunotherapy, or any other anti-tumor treatment prior to surgery. Table 1 summarizes the correlation of USP5 and CD73 expression levels with clinicopathological characteristics in LUAD patients. Additionally, a total of 31 paired fresh-frozen LUAD tissues and adjacent non-cancerous tissue specimens were collected for Western blot analysis. Informed consent was obtained from all participants prior to sample and clinical data collection. Human participant involvement, data, and tissue usage were approved by the Medical Research Ethics Committee of The Second Affiliated Hospital of Nanchang University and conducted in accordance with the declaration of Helsinki.

Mice

Male BALB/c nude mice (4 weeks old) were procured from Sibeifu Biotechnology Co., Ltd. (Beijing, China). All nude mice were housed in a specific pathogen-free (SPF) facility under controlled conditions (temperature: 22 ± 2 °C, humidity: 55 ± 10%, 12-h light/dark cycle) with ad libitum access to food and water. All animal procedures were approved by the Institutional Animal Care and Use Committee of Nanchang Royo Biotech Co., Ltd. (Approval No.: RYE2024102802) and conducted in compliance with the AVMA Guidelines for the Euthanasia of Animals (2020).

Cell line and culture

The following human cell lines were used in this study: the NSCLC lines H1299, H1650, and H358 from the National Collection of Authenticated Cell Cultures; the NSCLC lines PC9, HCC827, and A549 from ServiceBio Co., Ltd. (Hubei, China); and the normal bronchial epithelial cell line HBE135-E6E7 from Senbeijia Biotechnology Co., Ltd. (Nanjing, China). A549 and H358 cells were cultured in DMEM medium (Solarbio, China) supplemented with 10% fetal bovine serum (FBS; ExCell Bio, China) and 1% penicillin/streptomycin (P/S). All other NSCLC cell lines were maintained in RPMI 1640 medium (Solarbio, China) supplemented with 10% FBS (ExCell Bio, China) and 1% P/S. HBE135-E6E7 cells were cultured in Keratinocyte Serum-Free Medium (Zhong Qiao Xin Zhou Biotechnology, Shanghai, China), supplemented with 0.005 mg/mL insulin, 500 ng/mL hydrocortisone, and 1% P/S. Furthermore, the osimertinib-resistant HCC827 cell line used in this study was previously established by our group through stepwise dose escalation of osimertinib. Osimertinib (1 μM) was added to the culture medium every three passages to sustain resistance phenotypes. All cell lines were cultured at 37 °C in a humidified incubator with 5% CO₂. Authentication of all cell lines was performed using short tandem repeat (STR) profiling, with mycoplasma testing yielding negative results.

Method details

Immunohistochemical (IHC) staining

Tissue specimens were fixed in 4% paraformaldehyde, paraffin-embedded, and sectioned. After deparaffinization at 70°C, sections were cleared in xylene and rehydrated through graded ethanol. Antigen retrieval was performed in EDTA buffer, followed by quenching of endogenous peroxidase with 3% H₂O₂. Sections were blocked with 5% goat serum and incubated overnight at 4°C with primary antibodies against USP5 (1:400 dilution; Proteintech, China) and CD73 (1:100 dilution; Abcam, UK). Biotin-conjugated secondary antibodies were then applied, and chromogenic detection was performed using 3,3'-diaminobenzidine (DAB). Counterstaining with hematoxylin, dehydration in xylene, and microscopic examination completed the procedure. Immunoreactivity was semi-quantitatively assessed using the H-score system: Staining intensity (0 = negative, 1 = weak, 2 = moderate, 3 = strong) multiplied by percentage of positive cells (0–100%). Final scores (range: 0–300) categorized samples as low expression (≤100) or high expression (>100).

Lentiviral, small interfering RNA, and plasmid transfection

Lentiviral particles for USP5 knockdown (shRNA) and USP5 overexpression (OE) were obtained from Hanbio Tech (Shanghai, China). Lentiviral particles for CD73 overexpression (OE) and TRIM28 knockdown (shRNA) were purchased from GenePharma (Shanghai, China). Stable cell lines with USP5 knockdown, USP5 overexpression, CD73 overexpression, or TRIM28 knockdown, along with their matched controls, were generated by transducing LUAD cell lines with the respective lentiviral particles. For transient transfection assays, small interfering RNA (siRNA) targeting CD73 was synthesized by GenePharma (Shanghai, China). Plasmids encoding USP5 truncated mutants and CD73 domain-deletion mutants were purchased from Hanbio Biotechnology Co., Ltd. (Shanghai, China). Cells were transiently transfected with siRNA or the indicated plasmids using Lipofectamine 3000 (Invitrogen; Thermo Fisher Scientific, Carlsbad, CA, USA), following the manufacturer's protocol. The oligonucleotide sequences for USP5 shRNA, TRIM28 shRNA, and CD73 siRNA are listed in the key resources table.

Quantitative real-time PCR (qRT‒PCR)

LUAD cells were harvested at 70-90% confluency. Total RNA was isolated using the RNA-Quick Purification Kit (Esun Bio, Shanghai, China). Subsequently, the extracted RNA was retrotranscribed into complementary DNA (cDNA) using the PrimeScript™ RT Reagent Kit (TaKaRa Bio, Japan) according to the manufacturer's protocol. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using TB Green® Premix Ex Taq™ II (Tli RNaseH Plus) (TaKaRa Bio, Japan) on a thermal cycler system. Relative gene expression was calculated using the 2ˆ(−ΔΔCT) method, with GAPDH serving as the internal reference control. Primer sequences used for detection are listed in the key resources table.

Western blotting

Total protein was extracted from LUAD tumor tissues or cultured LUAD cells following cell lysis. Protein concentration was subsequently quantified using a bicinchoninic acid (BCA) assay kit (Beyotime Biotechnology, Shanghai, China). Equal amounts of protein were separated by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) and electrophoretically transferred onto polyvinylidene fluoride (PVDF) membranes (Millipore, Sigma, MA, USA). Membranes were blocked for 1 hour at room temperature with 5% bovine serum albumin (BSA) in Tris-buffered saline containing 0.1% Tween-20 (TBST). Following blocking, membranes were incubated overnight at 4°C with primary antibodies and subsequently with appropriate horseradish peroxidase (HRP)-conjugated secondary antibodies for 1.5 hours at room temperature (detailed antibody information provided in the key resources table). Protein bands were visualized by incubating the membranes with an ultrasensitive chemiluminescent substrate (UElandy, Jiangsu, China) and captured using a digital imaging system. Band intensities were quantified with ImageJ software and normalized to the housekeeping proteins GAPDH or β-actin to account for variations in total protein loading. Densitometry data are presented as the mean from n=3 independent biological replicates. Uncropped images of all blots are provided in the Supplementary file (Data S1).

Colony formation assay, cell counting kit-8 (CCK-8) assay, and EdU incorporation assay

For the cell proliferation assay, LUAD cells were plated in 96-well flat-bottom plates at 1 × 10³ cells/well in 100 μL complete medium and cultured under standard conditions (37°C, 5% CO₂, humidified atmosphere). Following cell adhesion, 10 μL of CCK-8 solution (GLPBIO, Montclair, CA, USA) was administered to each well at the indicated time points (0, 24, 48, 72, and 96 hours). After a 2-hour incubation at 37°C, the optical density (OD) at 450 nm was determined using a microplate reader.

For colony formation assays, LUAD cells were plated in 6-well plates at 1,000 cells/well and cultured for 10-14 days under standard conditions (37°C, 5% CO₂) until colonies contained >50 cells. Following fixation with 4% paraformaldehyde (Servicebio, Wuhan, China) for 30 min at room temperature, cells were rinsed twice with PBS and stained with 1% crystal violet (Solarbio, Beijing, China) for 30 min. After thorough washing with distilled water and air-drying, colonies were imaged and quantified.

Cell proliferation was assessed using the YF® 594 Click-iT™ EdU Kit (UElandy, Jiangsu, China) according to the manufacturer's instructions. Briefly, LUAD cells were incubated with the EdU reagent for 2 hours under standard culture conditions. Following incubation, cells were fixed with 4% paraformaldehyde, permeabilized with 0.5% Triton X-100 in PBS for 20 minutes at room temperature, and then incubated with the Click-iT® reaction cocktail for 30 minutes at room temperature, protected from light. Finally, cell nuclei were counterstained with Hoechst 33342 for 10 minutes at room temperature. Fluorescent images were acquired using a suitable microscope system equipped for YF® 594 and Hoechst 33342 detection.

Wound healing assay and transwell assay

For wound healing assays, LUAD cells were seeded uniformly in 6-well plates. At 80% confluency, scratch wounds were created using a sterile 200-μl pipette tip. After removing detached cells by washing with PBS, cells were maintained in medium containing 3% FBS. Wound areas were imaged at 0, 24, and 48 h using phase-contrast microscopy. Migration rates were calculated based on wound closure relative to 0 h. Cell mobility rate = [1 − (current wound size/initial wound size)] × 100%.

For invasion assays, Matrigel® (BD Biosciences) was diluted 1:8 in serum-free medium, and 40 μL dilution coated onto 8-μm pore Transwell inserts. The upper chamber received 200 μL serum-free medium containing 2×10⁴ cells; the lower chamber contained 700 μL basal medium with 20% FBS. After 36 h incubation (37°C, 5% CO₂), invaded cells were fixed with 4% PFA and stained with 1% crystal violet. Migration assays followed identical protocols using uncoated Transwell inserts.

In vivo tumor growth and metastasis assays

Stably transduced exponentially growing LUAD cells or control cells were subcutaneously injected into the right flank of BALB/c nude mice (5×10⁶ cells/mouse; n=5/group). To evaluate the inhibitory effects of Osimertinib (HY-15772; MedChemExpress, USA) or USP5-IN-1³⁵ (T60130; TargetMol, Shanghai, China) on tumor growth in vivo, mice were randomized into four treatment groups. The single-drug treatment group received intraperitoneal injections of Osimertinib (5 mg/kg) or USP5-IN-1 (15 mg/kg), the combination therapy group received both drugs, and the control group received intraperitoneal injections of 0.9% NaCl. All groups were administered the drugs once daily for three consecutive weeks. After tumor formation, mouse weight and tumor size were recorded every 4 days. Tumor volume was calculated using the formula: (length × width²)/2. Approximately 4 weeks after cell injection, mice were euthanized via controlled CO₂ asphyxiation. Specifically, animals were placed in a transparent chamber prefilled with 100% medical-grade CO₂ at a flow rate displacing 20-30% chamber volume/minute. Exposure continued until cessation of breathing and motor activity (≥2 min post-respiratory arrest). Cervical dislocation was immediately performed under confirmed unconsciousness to ensure death. Absence of vital signs (cardiac arrest, apnea, and loss of corneal reflex) was verified prior to tissue collection. This two-step protocol ensures rapid unconsciousness (<30 sec), minimizes distress, and complies with AVMA requirements for rodent euthanasia. Then, xenograft tumors were excised, weighed, photographed, and fixed in 4% paraformaldehyde for immunohistochemical (IHC) analysis.

For in vivo metastasis experiments, 5×10⁵ LUAD cells in 150 μL PBS were intravenously injected into male BALB/c nude mice via tail vein. Lungs were harvested after 4 weeks, fixed in 4% PFA, and H&E-stained to quantify metastatic nodules.

Protein half-life assays

The protein synthesis inhibitor cycloheximide (CHX; HY-12320, MedChemExpress) was employed to assess protein stability of CD73 after USP5 knockdown or overexpression in LUAD cells. Upon reaching 80% confluency, cells were treated with 100 μg/mL CHX in complete medium to inhibit nascent protein synthesis. Whole-cell lysates were harvested at 0, 3, 6, 9, and 12-hour intervals for total protein extraction. USP5 and CD73 protein levels were evaluated by Western blotting.

Mass spectrometry (MS)

H1299 cells were lysed in IP-grade lysis buffer containing protease and phosphatase inhibitors. Antibody-bead conjugates were prepared by incubating Protein A/G Magnetic Beads (HY-K0202, MedChemExpress) with anti-USP5 antibody (10473-1-AP, Proteintech) on a rotator for 4h at 4°C. Following washes, the conjugates were incubated with cell lysate supernatants overnight at 4°C. After gentle washing, beads were resuspended in 50 μL loading buffer and denatured at 95°C for 10 min using a heating block. Immunoprecipitated proteins were resolved by SDS-PAGE and visualized with Coomassie Blue staining. Entire protein lanes were excised and digested with trypsin. Resulting peptides were analyzed by liquid chromatography-tandem mass spectrometry (LC-MS/MS). The identification of peptide mixtures was performed by Genechem (Shanghai, China). Refer to Supplementary Table for further details.

Ubiquitylation assay and Co-immunoprecipitation (CoIP)

For ubiquitination assays, cells were transfected with lentivirus for 48 hours, treated with 20 μM MG-132 for 6-8 hours, and subjected to CoIP. Ubiquitination modifications were detected using anti-ubiquitin (Ub), anti-K48-linked Ub, and anti-K63-linked Ub antibodies (key resources table). For CoIP assays, cells were lysed in NP-40 buffer supplemented with 20 μM MG-132 and protease inhibitor cocktail. Cleared lysates were obtained by centrifugation at 12,000 × g for 15 min at 4°C. Subsequent immunoprecipitation procedures followed the protocol described in Section 3.10. Immunoprecipitated complexes were washed three times with lysis buffer and analyzed by Western blotting.

Untargeted metabolomics,L-Lactic acid colorimetric assay and seahorse XF metabolic flux analysis

USP5-knockdown and control H1299 cells were expanded to 80% confluency, harvested by rapid quenching in ice-cold 80% methanol (-20°C), and flash-frozen in liquid nitrogen. Samples were transported on dry ice to HaploX Genomics Center (Jiangxi, China) for untargeted metabolomic profiling. Metabolite separation and detection were performed using ultra-high-performance liquid chromatography Orbitrap mass spectrometer (Thermo Fisher Scientific, USA). Compound identification was achieved through spectral matching against the Human Metabolome Database (HMDB version 5.0), with quantification based on integrated chromatographic peak areas normalized to total protein content. Six biological replicates per group, each comprising 1 × 10⁷ cells, were processed for comparative statistical analysis.

Lactate production was quantified using the L-Lactic Acid Colorimetric Assay Kit (E-BC-K044-M, Elabscience, China) in accordance with the manufacturer's protocol. Briefly, cellular homogenates were prepared from 5×10⁶ cells in 200 μL ice-cold PBS, followed by centrifugation at 10,000 × g for 10 min at 4°C to obtain clarified supernatants for subsequent analysis. For the assay, 5 μL of standards or test samples were dispensed into 96-well plates, to which 100 μL of enzyme working solution and 20 μL of chromogenic agent were sequentially added. After 10 min of incubation at 37°C under light-protected conditions, 180 μL of stop solution was introduced. Absorbance at 530 nm was measured immediately following 5-sec orbital shaking using a microplate reader. Concurrently, total protein concentration was determined by BCA assay for normalization purposes. All measurements were performed in technical triplicates across three independent biological replicates.

Glycolytic flux was quantified via real-time proton efflux rate (PER) measurements using the Seahorse XF Glycolysis Rate Assay Kit (103344-100, Agilent Technologies, USA) on an XF96 Analyzer. LUAD cells were seeded at 1×10⁴ cells/well in XF96 microplates and allowed to adhere for 24 h. Cells were equilibrated in unbuffered RPMI 1640 (103681-100, Agilent Technologies, USA) to establish baseline PER, followed by sequential injections of mitochondrial inhibitors (0.5 μM antimycin A/rotenone) and 50 mM 2-deoxy-D-glucose (2-DG) to induce compensatory glycolysis. Basal glycolysis was calculated as the last PER measurement prior to rotenone/antimycin A injection. Compensatory capacity = (maximum PER after oligomycin) - (basal PER). All PER values were normalized to total cellular protein quantified by BCA assay.

Immunofluorescence

Cells grown on glass coverslips were fixed with 4% paraformaldehyde (PFA) for 30 min at room temperature. After three washes with phosphate-buffered saline (PBS), cells were permeabilized with 0.2% Triton X-100 in PBS for 10 min at room temperature and subsequently blocked with 5% bovine serum albumin (BSA) in PBS for 1 h at room temperature. Cells were then incubated overnight at 4°C with primary antibodies diluted in blocking buffer (5% BSA in PBS) within a humidified chamber. Following five washes (5 min each) with PBS, cells were incubated with fluorophore-conjugated secondary antibodies diluted in blocking buffer for 2 h at room temperature in the dark. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; 1μg/mL in PBS) for 5 min at room temperature. Finally, observe the fluorescence under a microscope and take photographs.

Flow cytometry

For apoptosis assessment, transfected or drug-treated cells were stained with FITC-Annexin V/propidium iodide (PI) Kit (UElandy, Jiangsu, China) for 15 min at room temperature in the dark. Flow cytometry was performed within 1 h using the Beckman Coulter CytoFLEX S flow cytometer. The relevant data analysis was performed using CytExpert 2.4 software.

Detection of mitochondrial membrane potential

Following a 24-hour treatment with USP5-In-1 or the vehicle control (0.1% DMSO), changes in the mitochondrial membrane potential of H1299 and PC9 cells were evaluated. The JC-1 fluorescent probe (Beyotime, Shanghai, China) was employed to monitor mitochondrial depolarization, and fluorescence was measured according to the supplier's manual.

Bioinformatics analysis

All bioinformatics analyses were performed using R (v4.3.3) and public databases. Gene expression data were obtained from curated TCGA-LUAD datasets and GEO repositories (GSE31547, GSE72094). Kaplan-Meier survival analysis correlating USP5 expression with clinical outcomes was conducted using the “survival” and “survminer” packages, with significance determined by log-rank testing. Drug sensitivity prediction employed the pRRophetic algorithm,^57,58 which calculates area under the dose-response curve (AUC) values derived from the Cancer Therapeutics Response Portal (CTRP) to quantify chemotherapeutic susceptibility. Pathway enrichment utilized KEGG annotations, with gene set activity quantified via GSVA using z-score normalization. Differential expression analysis (limma package; threshold: |log₂FC|>1, FDR<0.05) preceded gene set enrichment analysis (clusterProfiler). All visualizations were generated with “ggplot2”.

Statistical analysis

All statistical analyses were performed using GraphPad Prism software (version 10.1.2). Data Presentation and Measures: Continuous data derived from in vitro experiments (e.g., densitometry, cell viability, migration/invasion counts, metabolite levels, glycolytic parameters) are presented as the mean ± standard deviation (SD) of n independent biological replicates. The “n” represents the number of independently performed experiments, each with its own technical replicates. Data from in vivo xenograft studies (e.g., tumor volume, weight) are presented as the mean ± SD for n mice per group. Statistical Testing: Comparisons between two groups were conducted using an unpaired, two-tailed Student’s t-test. Comparisons among three or more groups were performed using one-way analysis of variance (ANOVA) followed by appropriate post-hoc tests as indicated in the figure legends. Associations between USP5/CD73 expression (categorized as high/low based on median or optimal cut-off) and clinicopathological features in patient samples were evaluated using Pearson’s χ² test or Fisher’s exact test, as appropriate. Patient survival analysis was performed using the Kaplan-Meier method, and differences between survival curves were assessed with the log-rank test. Significance Thresholds: Statistical significance was defined as ∗p < 0.05. The following asterisk notation is used throughout the figures: ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001; “ns” denotes not significant (p ≥ 0.05).

Published: February 5, 2026