PURPOSE:Blocking the CD47-SIRPα checkpoint reactivates macrophage-driven elimination of malignant cells, although CD47 antagonists have progressed substantially in clinical trials, their implementation is hindered by on-target hematologic toxicity, particularly unintended binding to red blood cells (RBCs). The present study was therefore undertaken to develop a humanized anti-CD47 antibody that achieves robust antitumor activity with an enhanced safety margin. METHODS:A humanized anti-CD47 monoclonal antibody was engineered with a functionally silenced IgG4 Fc to reduce Fc effector function. Binding kinetics and species cross-reactivity were quantified by surface plasmon resonance and flow cytometry; functional blockade of CD47-SIRPα interaction was evaluated through hSIRPα competitive blocking and signaling pathway blocking analyses. The antitumor activity was investigated by primary macrophage phagocytosis assays and in vivo efficacy studies utilizing xenograft animal models encompassing hematologic and solid tumor cells. Translational safety was evaluated through ex vivo binding assays on peripheral blood mononuclear cells (PBMCs) and RBCs, hemagglutination assay, and toxicology studies in cynomolgus monkeys, which included maximum tolerated dose (MTD) determination and hematologic monitoring. RESULTS:A novel humanized anti-CD47 antibody LD002 was engineered. It bound human CD47 with a high affinity (KD 1.5 nM), thus it effectively suppressed CD47-SIRPα signaling and boosted macrophage-driven phagocytic activity to target tumor cells in vitro. LD002 elicited significant antitumor efficacy against lymphoma and small cell lung cancer xenograft models, achieving 79% tumor growth inhibition (TGI) and 100% complete clearance, respectively, at 3 mg/kg, with good tolerability. Although binding to PBMCs and RBCs was observed, LD002 did not exhibit any dose-dependent hemagglutination, distinguishing it from the benchmark antibody Hu5F9-G4. Moreover, in non-human primates, the single-dose maximum tolerance dose (MTD) reached 744 mg/kg, and repeat-dosing up to 150 mg/kg once weekly revealed no clinically meaningful adverse findings, underscoring a widened therapeutic window and manageable hematologic profile that supports its further clinical development. CONCLUSIONS:LD002 addresses key translational bottlenecks in anti-CD47 therapy by decoupling target blockade from RBCs hemagglutination, thereby largely mitigating hematotoxicity while preserving robust antitumor efficacy. These preclinical data position LD002 as a differentiated, promising CD47 inhibitor with significant translational potential.

Bioactive natural products (NPs) are valuable sources for biomedical and various applications. Comprehensive information on their species sources and quantitative properties (including compositions, bioactivities, and toxicities) is of great utility to the research community. Here, we introduced NPASS 3.0 with major updates to significantly expand the previous version and add new data derived from emerging technologies. Firstly, we manually curated new records from 1822 publications to provide: (i) an additional 87 507 quantitative composition records for 4873 NPs in 1030 species and (ii) 878 NPs produced by 341 symbiont organisms and 164 elicitor-responsive organisms, together with an additional 109 engineered organisms and 43 co-culture organisms. Secondly, 34 975 new quantitative toxicity records for 3662 NPs and 9713 quantitative ADME records for 744 NPs. Thirdly, this update substantially increased the records of NPASS, with 9.37%–206.30% data expansion over NPASS 2.0. Currently, NPASS includes 204 023 NPs, 48 940 organisms, 8764 targets, 1 048 756 experimental activity records, 34 975 toxicity records, 9713 quantitative ADME records, 208 415 NP composition records, and 1 117 269 organism–NP pairs. Furthermore, bioactivities were grouped into three classes for better serving pharmaceutical research. We also updated the NPASS web interface, which includes an enhanced search option and a submission page for crowdsourcing curation. NPASS 3.0 is accessible at https://bidd.group/NPASS/index.php.

Background: Non-small cell lung cancer (NSCLC) remains a leading cause of cancer-related mortality, and although PD-1/PD-L1 immune checkpoint blockade has improved outcomes in some patients, therapeutic responses remain heterogeneous. Tumor-intrinsic heterogeneity within malignant epithelial populations is increasingly recognized as a critical determinant of disease progression and therapy response. Methods: Here, we constructed a comprehensive single-cell atlas of NSCLC by integrating 650,461 cells from 216 tumor and normal samples. Tumor-derived epithelial cells were reclustered to identify transcriptionally distinct subpopulations. Pseudotime analysis, functional experiments, and in vivo validation using a humanized xenograft model were performed to investigate the role of COL3A1. Results: Reclustering of tumor-derived epithelial cells revealed 25 transcriptionally distinct subpopulations. Among these, a high-risk cluster exhibited coordinated activation of epithelial-mesenchymal transition (EMT) and angiogenesis programs and was associated with poor patient survival. Within this aggressive subpopulation, Collagen type III alpha 1 (COL3A1) emerged as a tumor-intrinsic gene associated with extracellular matrix remodeling and angiogenic signaling. Pseudotime analysis indicated that COL3A1+ cells represent a late-stage, poorly differentiated malignant state. Functional experiments demonstrated that COL3A1 knockdown impaired NSCLC cell proliferation, migration, and invasion. Virtual knockout further suggested that COL3A1 may be associated with transcriptional programs involved in PD-L1 upstream signaling pathways, indicating a potential indirect link between tumor-intrinsic states and immune regulatory networks. Consistently, in vivo silencing of COL3A1 enhanced the antitumor efficacy of PD-L1 blockade. Conclusions: Collectively, our study identifies COL3A1 as a tumor-intrinsic gene enriched in malignant epithelial cells with mesenchymal features and a potential therapeutic target. These findings provide a rationale for exploring combinatorial strategies integrating tumor-intrinsic pathway inhibition with immune checkpoint blockade in NSCLC.

Backgroud:Nephrotic syndrome (NS) is a major cause of end-stage renal disease. Treating NS relies on immunosuppressants, which have numerous side effects. Therefore, there is an urgent need to identify effective and safe alternative treatments for NS. Angiopoietin-like protein 3 (ANGPTL3) exacerbates proteinuria, whereas interleukin (IL)-22 has a reparative effect on renal cells. Methods:In the present study, we developed a bifunctional anti-ANGPTL3/IL-22 fusion protein and validated its efficacy in an adriamycin-induced nephropathy in mice. Results:The fusion protein significantly decreased the urinary albumin-to-creatinine ratio, serum creatinine, blood urea nitrogen, and total cholesterol levels while increasing serum albumin levels. Pathological renal damage was also alleviated. These therapeutic effects were accompanied by the preservation of mitochondrial integrity, reduced apoptosis, and inhibited autophagy. Finally, we humanized the fusion protein to facilitate its potential clinical translation. Conclusions:In conclusion, our results revealed that the anti-ANGPTL3/IL-22 bifunctional fusion protein ameliorates NS by protecting mitochondria, inhibiting apoptosis, and suppressing autophagy, highlighting a novel therapeutic approach for NS.

The clinical translation of ex vivo CAR-T therapy is constrained by its complex manufacturing processes, limited treatment accessibility, and the toxicities associated with lymphodepleting chemotherapy. In vivo CAR-T strategies, which utilize viral or non-viral vectors to reprogram a patient’s T cells directly in situ genetically, offer a promising alternative. This review provides a systematic examination of the critical barriers impeding the advancement of in vivo CAR-T therapy. We first summarize the factors influencing the targeting ability of both the delivery system and the CAR molecule. As for the vectors, we detail how the composition and physicochemical properties of LNPs shift their tropism from the liver to the spleen and provide a critical comparison of active targeting strategies based on different T-cell surface antigens. As for the CAR molecule, target selection is primarily dictated by the intended disease. Furthermore, we discuss the determinants of poor CAR-T cell persistence, including unstable mRNA payloads and exhaustion-associated signaling pathways. Addressing the challenges of solid tumor treatment, we dissect the multifaceted role of the tumor microenvironment. Finally, this review illustrates the mechanisms underlying overactivation and immunogenicity triggered by both viral and non-viral vector systems.

ABSTRACT Current therapies for diabetic kidney disease (DKD) targeting hyperglycemia and hypertension fail to halt progression, urging exploration of additional drivers. Renal lipotoxicity and chronic inflammation form a self‐perpetuating cycle driving DKD, yet multitarget interventions remain underexplored. We hypothesized that simultaneous inhibition of ANGPTL3 and IL‐33, two key regulators of lipid metabolism and the inflammatory‐fibrotic axis, would ameliorate DKD, and that integrating both antagonistic activities into a single bifunctional molecule offers translational advantages. We engineered a bifunctional fusion protein, FD03‐sST2, comprising an anti‐ANGPTL3 nanobody fused to the IL‐33 decoy receptor sST2. In high‐fat diet‐fed db/db mice, FD03‐sST2 significantly improved renal function (ACR, BUN, urine volume), reduced serum/hepatic lipids, and attenuated renal lipid accumulation. Mechanistically, it suppressed renal inflammation via NF‐κB/NLRP3 inhibition and ameliorated fibrosis by suppressing the IL‐33/ST2/ILC2 axis (reducing renal IL‐33/GATA3 signals and inhibiting IL‐33‐induced profibrotic factors from ILC2s). Transcriptomic and metabolomic analyses confirmed attenuation of DKD‐associated dysregulation. This study identifies a DKD cascade wherein lipotoxicity triggers IL‐33 release, amplifying injury through inflammation and fibrosis. By targeting ANGPTL3 and IL‐33 simultaneously, FD03‐sST2 interrupts this vicious cycle at two nodes—improving lipid metabolism while suppressing downstream inflammatory and fibrotic signaling—providing an integrated alternative to separate biologics.

Epidermal Growth Factor Receptor (EGFR) is crucial for cell proliferation and survival, and has emerged as a promising therapeutic target for colorectal cancer (CRC). Anti-EGFR monoclonal antibodies (mAbs) have demonstrated success in various clinical practice in CRC patients. However, they are only effective in patients with RAS wild-type, and more importantly, most patients eventually develop resistance to anti-EGFR therapy due to the mutations of the effector molecules in EGFR downstream signaling pathways or/and the activation of compensatory feedback loop signaling, which limits the long-term efficacy of anti-EGFR mAbs. Advances in antibody engineering and nanotechnology have brought new hope for cancer treatment, particularly through bi-/tri-specific antibodies, antibody-drug conjugates(ADCs), and antibody-functionalized nanoparticles(AFNPs), which can simultaneously inhibit the activation of EGFR and its compensatory pathways—or even directly eradicate cancer cells bypassing EGFR-mediated signaling—these innovations hold promise for addressing drug resistance and leading to better response rates of EGFR-targeted therapy in CRC. In this review, we provide an overview of the design strategy of EGFR-targeted therapies based on antibody engineering and nanotechnology for CRC using a mechanistic framework, focusing on analyzing the optimization strategies for representative drugs or studies and their potential in overcoming heterogeneous drug resistance in CRC, in order to provide useful information for researchers developing EGFR-targeted drugs for CRC.

Concomitant liver and kidney injury is a critical pathological feature of metabolic disorders, but current organ-specific therapies often fail to provide cross-protection. Lipotoxicity is a core mechanism linking damage in both organs. Therefore, this study aimed to investigate whether simultaneously targeting ANGPTL3 and IL-1β could attenuate lipotoxicity and thereby ameliorate concomitant liver and kidney injury. A novel bispecific antibody (BsAb) targeting both ANGPTL3 and IL-1β was generated and characterized by SDS-PAGE, SEC-HPLC, thermal stability analysis, SPR and in vitro bioassay. Then, its protective effects were subsequently studied in the db/db mouse model and the underlying mechanisms were revealed by biochemical examinations, histopathological analysis, immunofluorescence (IF), ELISA, RNA-seq. Administration of the BsAb in db/db mice effectively improved liver and kidney function with alleviated liver steatosis and inflammation, as well as reduced kidney glomerular injury. Furthermore, the treatment attenuated lipotoxicity in both organs and ameliorated glycolipid metabolism disturbance including restored hepatic glycogen reserves and enhanced renal utilization of fatty acids. The results demonstrate that the anti-ANGPTL3/IL-1β BsAb alleviates concomitant liver and kidney injury in db/db mice by attenuating lipotoxicity and regulating glycolipid metabolism, which highlights a promising therapeutic approach for addressing multi-organ damage in metabolic disorders.

Extensive interactions between microbiota and active substances are health- and disease-relevant. Mechanistic understanding from genomic perspective of these interactions and potential impacts is important for biomedical and pharmaceutical research. However, current data repositories often lack systematic integration from a genomic perspective. Here we describe an update of the MASI microbiota-active substance interactions database. This update includes new data of (1) genomic-derived 166766 microbiota-drug interactions and 205505 microbiota-food interactions linked by 415 biosynthetic gene clusters (BGCs), 59 metabolic gene clusters (MGCs), and 7250 genome-scale metabolic network models (GEMs) of ∼1200 microbiota species, and (2) 1848 microbiota-microbiota interaction records mediated by 39 quorum sensing languages, and (3) 46717 microbiota-disease associations between 640 species and 59 diseases. Overall, this update provides 44643 interasctions derived from ∼2000 publications and 380571 genome-derived interactions, covering 1867 microbe species, 1576 therapeutic substances, 357 dietary substances, which is freely accessible at https://www.aiddlab.com/MASI2025/index.html.

The miRNA-based therapeutics held great promise for the treatment of diseases associated with aberrant gene expression. However, the development of miRNA-based drugs still faces many obstacles, including stability, targetability, tissue penetration, and induction of immune responses. To overcome these challenges, researchers developed various miRNA modification methods and miRNA-based delivery systems, which can protect miRNA from degradation and facilitate their transport across biological barriers. Here, we give an overview of the latest advancements in a variety of delivery systems, including virus, lipid, polymer, inorganic, and exosome-based nanosystems, which have been proved as versatile and valuable carriers for miRNA. Moreover, the use of stimuli-responsive materials allows for the controlled release of oligonucleotides in response to specific triggers. Importantly, the current miRNA-based therapeutics in clinical trials are summarized in this review. The combination of these advancements has the potential to promote more effective and safer treatment designs for a wide range of diseases. This review will contribute to developing more precise targeted delivery systems of miRNA, ultimately facilitating its clinical development.

Antibody-drug conjugates (ADCs) have revolutionized cancer treatment by merging the targeting precision of antibodies with the cell-killing power of cytotoxic drugs. Yet traditional IgG-based ADCs suffer from toxicity issues, poor biodistribution, and high manufacturing cost. To overcome these hurdles, we created an innovative Fc-free bispecific nanobody drug conjugate (dtNDC) directed against Nectin4 and Trop2, two antigens highly expressed across many malignant cancers. The dtNDC, equipped with a C18 fatty acid modification, displayed excellent pharmacokinetics with a 20-h plasma half-life and negligible kidney accumulation while achieving rapid tumor uptake (peaking just 2 h post intravenous injection), striking an optimal balance between systemic persistence and tissue penetration. In nonclinical evaluations, Eribulin-loaded dtNDC demonstrated potent antitumor efficacy across three xenograft models. Notably, in MDA-MB-468 cell-based tumor bearing mice, two subcutaneous injection induced tumor free status in all mice; and it could also result in in complete remission following two intravenous administrations in HCC1954 cell xenograft tumors. Histopathological analysis confirmed absence of treatment-related toxicity in vital mouse organs (liver, lung, kidney, heart) at therapeutic doses. These findings position dtNDC as a promising therapeutic combining durable tumor regression with exceptional tolerability. Moreover, its simplified structure and cost-effective manufacturing process make this dtNDC a compelling next-generation therapeutic for cancers.

Immune checkpoint inhibitors (ICIs) have significantly advanced the field of cancer immunotherapy. However, clinical data has shown that many patients have a low response rate or even resistance to immune checkpoint inhibitor alone. The underlying reasons for its poor efficacy include the deficiency of immune infiltration and effective CD28/CD80 costimulatory signal in tumor. Discoidin domain receptor 1 (DDR1) has been reported to be negatively related to immune cell infiltration in tumors. Herein, we constructed a soluble fusion protein using CD80, the natural ligand of CD28, in combination with DDR1 inhibitor. Our results demonstrated that CD80-Fc effectively activated T cells and inhibited tumor growth in vivo, even in tumors with poor efficacy of ICIs. Importantly, CD80-Fc fusion protein had a milder affinity against the targets which suggested a potential higher safety than CD28 agonists. Further, in order to promote tumor immune infiltration, we attempted to combine CD80-Fc fusion protein with DDR1 inhibitor for treatment. Our results indicated that using CD80-Fc fusion protein along with DDR1 inhibitor significantly promoted T cell infiltration in tumor microenvironment and more strongly inhibited tumor growth. Therefore, the combination use of CD80 fusion protein and DDR1 inhibitor could become an effective tumor immunotherapy strategy, potentially benefiting a larger number of patients. • We successfully constructed, expressed, and purified the recombinant CD80-Fc fusion protein • We demonstrated that CD80-Fc fusion protein has good safety and anti-tumor activity • We demonstrated that using CD80-Fc fusion protein along with DDR1 inhibitor can significantly promote immune infiltration of T cells in tumor microenvironment and more strongly inhibit tumor growth

The gut microbiota plays a crucial role in modulating drug metabolism, efficacy, and toxicity. However, experimental strategies heavily suffered from the technic challenges of the isolation and in vitro culture of individual microbiota species. Predicting gut microbiota-drug associations (GutMDA) is therefore essential for advancing microbiome-informed pharmacology. In this study, we proposed a graph convolutional network-based model GutMDA that utilizes chemical structure similarity for drug representation, and integrates gut microbiota and disease information to enable efficient and accurate prediction of drugmicrobiota associations. Benchmarking results on curated datasets show superior predictive performance compared to existing approaches. Additionally, the case studies showed that more than 90 percent of top 20 predicted associations have been validated experimentally in recent publications, which further demonstrates the accuracy of GutMDA. In a word, GutMDA provided an effective and interpretable tool for gut microbiotadrug associations prediction.

Fibroblast growth factor 21 (FGF21) is a crucial regulator of glucose and lipid metabolism, showing significant therapeutic promise for metabolic disorders. However, its clinical application is limited by poor pharmacokinetics. One potential strategy to improve its half-life is to facilitate albumin binding through fatty acid derivation. Despite this promise, achieving site-specific modifications of FGF21 while preserving its biological activity has been challenging. In this study, we applied a rational design approach to create site-specific fatty acid derivatives of FGF21, guided by the structure of the FGF21-receptor complex. This strategy successfully enhances albumin binding without interfering with receptor interactions. The modified FGF21 derivatives exhibited dramatically extended half-lives in mice, increasing from 0.73 h to 11.36 and 13.36 h, respectively. Furthermore, these analogues showed superior biological activity in the presence of albumin, outperforming the C-terminal-derived variant zalfermin. This rational design approach not only improves the pharmacokinetic profile of FGF21 but also provides a framework for enhancing the therapeutic potential of other small proteins.

Abstract Motivation Many associations among gut microbiota, diseases, and therapeutic drugs have been reported in various clinical settings, providing a new layer of clues for understanding the variations of therapeutic features among patients. However, experimental strategies heavily suffered from the technic challenges of the isolation and in vitro culture of individual microbiota species. Results In this study, we developed a multiple graph convolutional networks model (GutMDA) for predicting gut micro-biota-drug associations to complement experimental strategies and existing prediction methods. We constructed gut microbiota species-species based on phylogenetical-based similarity, drug-drug network based on chemical structural similarity, and disease-disease association networks based on MESH-derived semantic similarity. These networks were subsequently integrated into a microbiota-disease-drug triple network and trained on graph convolutional network models to predict microbiota-drug associations. The results showed that GutMDA demonstrates superior performance with the average AUC and AUPR over 0.99 on MASI dataset in five-fold cross validations. The case studies showed that the most of top-20 predicted associations have been validated experimentally in recent publications, which further demonstrated the effectiveness of GutMDA. In summary, GutMDA provided an effective option for predicting gut microbiota-drug associations. Availability and implementation Original data and codes are available at https://github.com/Wang-Shuaiqi/GutMDA.

The complexity of the tumor microenvironment (TME) makes cancer therapy challenging. Multi-targeting strategies often exhibit superior clinical benefits and dominate ongoing cancer clinical trials. However, existing multi-targeting strategies largely rely on empirical approaches. The rational design of target pairs (TPs) for multi-target cancer therapy development is a high priority but remains a significant challenge, which may benefit from single-cell omics technologies to decode the TME. However, there has been no thorough survey of clinically relevant cancer TPs and their characterization at the single-cell level. Here, we established the TargetPair database to address this gap by manually annotating TPs from drug combinations and multi-targeting drugs whose anti-tumor efficacy has been evaluated, and calculating their omics features at the single-cell level. The TargetPair database provides (1) 60 approved, 1,580 clinical-stage, and 7,424 experimental-stage TPs manually annotated from 3,470 clinical trials and 1,396 publications, and (2) the omics features of TPs (co-expression, distribution, expressed cell fractions, pathways, and single-cell gene networks) from manually curated scRNA-seq datasets derived from 55 pieces of literature (including 3,022,556 single cells, 13 cell types, 16 cancer types, and 588 patients). It can be freely accessed at https://www.targetpair.aiddlab.com via several search and browsing modes.

Cytokines are promising in cancer immunotherapy, but their pleiotropic effects limit specificity and clinical utility. Through binding to IL-18Rα and IL-18Rβ, interleukin-18 (IL-18) stimulates innate lymphocytes and effector T cells for antitumor immunity. However, clinical trials of recombinant IL-18 have been hampered by IL-18 binding protein (IL-18BP), a secreted high-affinity decoy receptor. Here, we developed decoy-resistant bispecific nanobodies that maintain IL-18 signaling potential. Based on agonistic nanobodies targeting IL-18Rα and IL-18Rβ, bispecific nanobody A4B2-mdFc effectively enhanced CD8+ T cell responses with distinct transcriptomic profiles. Systemic delivery of A4B2-mdFc boosted CD8+ T cell infiltration and activation, demonstrating dose-dependent antitumor efficacy in cell-line-derived xenograft and patient-derived xenograft models. Interestingly, PDCD1 and CTLA4 expressions were drastically increased on CD8+ T cells when treated with A4B2-mdFc. Injection of A4B2-mdFc significantly improved the antitumor efficacy of immune checkpoint inhibitors (ICIs) targeting both PD-1 and CTLA-4. Our findings demonstrated that nanobody-based bispecific IL-18 mimetics elicited superior antitumor activity via CD8+ T cell activation and IL-18BP resistance, providing the potential application of cytokine-targeting bispecific nanobody monotherapy or in combination with ICIs for cancer immunotherapy.

Acute myeloid leukemia (AML) is a severe blood cancer with an urgent need for novel therapies for refractory or relapsed patients. Leukocyte-associated immunoglobulin-like receptor 1 (LAIR1), an immune suppressive receptor expressed on immune cells and AML blasts but minimally on hematopoietic stem cells (HSCs), represents a potential therapeutic target. But there has been limited research on therapies targeting LAIR1 for AML and no published reports on LAIR1 antibody-drug conjugate (ADC). We developed LA-057-MMAE, a first-in-class ADC, and evaluated its antitumor potential. LA-057-MMAE demonstrated strong binding to human LAIR1 with an affinity of 3.9 nM, efficient internalization of approximately 70 % within 4 h, and remarkable cytotoxicity against AML cells, with IC50 values of 0.22 nM for MV-4-11, 0.02 nM for U937, and 0.09 nM for HL-60 cells, respectively. In vivo, it achieved complete tumor regression in 100 % of MV-4-11 xenograft mice at 6 mg/kg, extending survival beyond 60 days. Our findings suggest that LA-057-MMAE, as a first-in-class treatment distinct from existing LAIR1 monoclonal therapies, could provide a groundbreaking therapeutic strategy for AML.

Uricase is a key enzyme in purine metabolism that catalyzes the oxidation of uric acid to allantoin, widely used in the treatment of hyperuricemia and gout. In this study, error-prone PCR and one high-throughput screening method were employed to generate uricase mutants with enhanced enzymatic activity from Aspergillus flavus and Candida utilis. After several rounds of mutation and selection, an A. flavus uricase mutant, af-UAM, with activity of 46.21 U/mg, and a C. utilis uricase mutant, cu-UAM, with activity of 31.43 U/mg, were obtained—representing the highest uricase activities reported up to date. Site-directed mutagenesis revealed that the Thr231Ala substitution in A. flavus uricase and the Val234Met substitution in C. utilis uricase were key factors driving their enhanced activities. Furthermore, in vivo experiments demonstrated significant clinical potential of these mutants. These findings offer new insights into the structure-function relationship of uricase and present promising candidates for therapeutic applications in hyperuricemia treatment.

Immunotherapy with interleukin-2 (IL-2) in treating cancers is subject to several limitations such as systemic side effects and reduced efficacy against tumors with low immune cell infiltration despite its promise. To address these challenges, IL-2-So-Lipo, a novel liposomal formulation combining IL-2 with sorafenib derivative, was developed as an anti-angiogenic drug that inhibits the growth of new blood vessels which play crucial roles in tumor growth. Sorafenib derivatives could target at melanoma-specific receptors, further enhancing liposomal specificity at the tumor site. Our results demonstrated that the prepared IL-2-So-Lipo significantly enhanced anti-tumor activity compared to IL-2 or sorafenib monotherapies, as well as their combination. In a B16F10 melanoma model, IL-2-So-Lipo was found to significantly inhibit tumor progression (tumor volume of 108.01 ± 62.99 mm3) compared to the control group (tumor volume of 1,397.13 ± 75.55 mm3), improving the therapeutic efficacy. This enhanced efficacy is attributed to the targeted delivery of IL-2 which promoted the infiltration and activation of cytotoxic T lymphocytes. Additionally, liposomal encapsulation of sorafenib derivatives enhanced its delivery efficiency, promoting tumor cell apoptosis and suppressing angiogenesis. Mechanistically, IL-2-So-Lipo could kill tumors by inducing a shift towards an anti-tumor immune response via facilitating the polarization of macrophages towards the M1 phenotype. Furthermore, IL-2-So-Lipo downregulated several key proteins in the MAPK signaling pathway, exerting a significant role in mediating tumor resistance to sorafenib. These findings underscore the potential of IL-2-So-Lipo as a promising strategy to improve the therapeutic efficacy of immunotherapy and targeted therapy in cancers. Moreover, the combination of IL-2 and sorafenib in a liposomal delivery system overcame the limitations of conventional IL-2 therapy, offering a synergistic approach to improve therapeutic outcomes for solid tumors.