Antioxidant malic-acid-derived lipid nanoparticles delivering secretory uricase circRNA for hyperuricemia and complication management in mice Authors: Keqian Du, Yiming Guo, Junjie Zhou, Yaping Liu, Lin Jiang, Dawei Hua, Di Zhang, Yuqi Zhang, Zijian Deng, Xingyu Zheng, Jianping Guo, Jing Ni, Yingjie Li, Shichao Sun, Zhiming Lin, Jianlin Huang, Lin-Ping Wu Published: 2026 ABSTRACT Hyperuricemia and its complications, including gouty arthritis and hyperuricemic nephropathy, are driven by elevated serum urate levels and crystal deposition. While exogenous uricase offers therapeutic potential, its utility is limited by rapid clearance and immunogenicity. Here we show a malic acid-derived lipid nanoparticle platform with antioxidant and low-immunogenicity properties for delivering circular RNA encoding secretory uricase to treat hyperuricemia and its complications. Leveraging the safety and anti-inflammatory and antioxidant properties of malic acid, a combinatorial library of ionizable lipids is developed. After systematic screening and optimization, LMA2-C10 LNPs are selected to deliver circRNA for its high transfection efficiency, low-immunogenicity, and antioxidant activity. In male animal models, the system achieves sustained urate reduction, attenuation of tissue inflammation and fibrosis, alleviation of arthritic pain, and preservation of renal function. This integrated strategy simultaneously addresses hepatic urate catabolism, peripheral crystal clearance, and long-lasting pharmacological activity, presenting an alternative strategy to resolve the core challenges in hyperuricemia management. Introduction Hyperuricemia is a globally prevalent metabolic disorder characterised by persistently elevated levels of uric acid (UA), which is driven by disrupted uric acid homeostasis1. Global epidemiological surveillance indicates a progressive rise in hyperuricemia and gout prevalence, correlating with dietary modifications2. Although it is considered a manageable benign disorder in rheumatology, the therapeutic outcomes are often suboptimal due to insufficient public awareness, leading to poor patient adherence3. Typically, treatment is only sought during acute episodes, and long-term medication compliance is rarely maintained4,5. Current management strategies predominantly rely on enzyme inhibitors and uricosurics, which require frequent dosing and pose safety risks in patients with renal or cardiovascular comorbidities6. Uricase, the key enzyme responsible for uric acid metabolism, catalyzes the conversion of slightly soluble uric acid into soluble allantoin7. However, unlike most mammals, the uricase gene has been evolutionarily silenced in humans and primates8. Exogenous uricase has been investigated as a potential therapeutic intervention for refractory gout, yet its clinical utility remains limited by rapid clearance, antidrug antibodies, and inherent immunogenicity9. Although PEGylation and nanomaterial-encapsulated uricase has been demonstrated to prolong the enzyme’s half-life, limitations in persistent immunogenicity and efficacy against multi-organ urate deposition concerns continue to hinder clinical translation9–11. Messenger RNA (mRNA) therapeutics have been clinically validated as protein replacement strategies for metabolic disorders12, wherein messenger RNA encoding therapeutic enzymes is delivered and translated by endogenous ribosomal machinery, thereby mitigating immunogenicity concerns associated with exogenous enzyme administration13. Successful mRNA therapeutic implementation critically depends on delivery systems. Lipid nanoparticles (LNPs) represent the clinically most advanced platform14, with demonstrated safety and efficacy in mRNA vaccines15. Unlike vaccines, high dose repeated administrations are often required in protein replacement therapy. Consequently, the safety profile and immunogenicity of lipid nanoparticles is subjected to more stringent requirements16. Exacerbation of inflammatory activation by LNPs may occur in inflammatory diseases, potentially aggravating pathological progression16,17. Malic acid, a naturally occurring dicarboxylic acid, serves as an essential intermediate in cellular metabolism, particularly within the tricarboxylic acid cycle18. It is considered to be a safer substance and is more readily degraded within the human body19. Its therapeutic potential has been experimentally established, with documented anti-inflammatory20, antioxidant21, hepatoprotective22, and metabolic regulatory properties23. Moreover, malic acid has been documented to facilitate renal urate elimination24. Until now, malic acid has not been explored as a scaffold core of ionizable lipid for lipid nanoparticle-based delivery application. This system may represent a potential delivery strategy characterised by high safety and antioxidant properties. Chronic hyperuricemia is managed as a long-term condition, necessitating sustained administration of therapeutic agents to maintain serum urate levels within a target range25. Beyond conventional linear mRNA, circular RNA (circRNA) has emerged as a translationally competent platform, with its covalently closed topology conferring exceptional biostability26,27. It has been proposed that circRNA exhibits substantially prolonged half-life and reduced immunogenicity, attributed to its evasion of cytosolic RNA sensors28,29. The progressive pathogenesis of chronic hyperuricemia is characterised by the systemic deposition of mono-sodium urate (MSU) crystals, which in turn predominantly affects articular and renal systems, culminating in gout and nephropathy30,31. This pathogenicity is mediated by the NOD-like receptor pyrin domain containing 3 (NLRP3) inflammasome cascade activated upon crystal deposition, which promotes endothelial dysfunction and fibrotic remodeling via interleukin (IL)-1β-driven inflammation and reactive oxygen species (ROS) overproduction32. This complexity necessitates therapeutic strategies capable of concurrently suppressing hepatic urate biosynthesis and dissolving existing crystalline deposits. Herein, a promising strategy is proposed for the delivery of circRNA encoding secretory uricase (Uox) to hepatocytes using malic acid-based antioxidant lipid nanoparticles. This approach is designed to facilitate uric acid metabolism within hepatocytes while concurrently promoting the clearance of peripheral urate deposits. Briefly, leveraging the intrinsic properties of malic acid, a structurally diversified library of malate-backboned ionizable lipids was rationally designed and synthesized, with particular attention given to three structural parameters: the stereo-chemistry of the malic acid center, the length of the hydrophobic tail, and the parity of the diamine linker. These factors were considered critical for mRNA transfection performance33,34. Through combinatorial screening and design of experiments (DOE)-guided formulation optimization, a delivery system exhibiting high safety, antioxidant capacity, and potent expression in hepatocytes was identified. The efficacy of this malic acid-based lipid platform for delivering secretory uricase was evaluated in mouse models of acute gouty arthritis (GA) and chronic hyperuricemia (UOX-KO) mice. Therapeutic outcomes were assessed for hyperuricemia and its complications, including gout, metabolic syndrome, and hyperuricemia nephropathy. Results Rational design and synthesis of malic acid-derived lipids A series of malic acid-derived lipids were designed and synthesized, featuring a malic acid core composed of either L-malic acid (LMA) or D-malic acid (DMA), conjugated with diamine linkers and hydrophobic tails of saturated or unsaturated alkyl chains, enabling the construction of diverse lipid architectures. Initially, dimethyl malate, used as a protected derivative of malic acid, was reacted with diamine cores in an amide exchange reaction at 60 °C, using a molar ratio of 1:4, affording malic acid- functionalized amine cores (Fig. 1a and Supplementary Fig. 1). The excess diamine was then removed under reduced pressure, enabling the rapid preparation of a range of malic acid-functionalized amine cores. Subsequently, a library of 30 malic acid-derived lipids was synthesized via reductive amination of the malic acid-bearing amine cores with aldehyde-alkyl chains (Fig. 1a and Supplementary Fig. 2). The resulting products were characterised by 1H NMR spectroscopy (Supplementary Figs. 3–32) and ESI-MS analysis.Fig. 1A combinatorial library of malic acid-derived ionizable lipids for mRNA delivery.a The synthetic pathway of malic acid-derived lipids (LMAX-CX or DMAX-CX), and the starting materials for LMAX-CX or DMAX-CX synthesis. In LMAX-CX or DMAX-CX, the first ‘X’ indicates the carbon count of the diamine unit attached to LMA or DMA, and the second that of the terminal alkyl chain; ‘diene’ is appended when the tail contains two double bonds. b Size distribution and PDI of the formulated MA-LNPs measured by dynamic light scattering (n = 3 technical replicates). c EE of MA-LNPs measured by Quant-it RiboGreen RNA assay (n = 3 technical replicates). d Quantification of bioluminescence intensity for each specific organ as a percentage of the overall organ system (n = 2 mice per group). e Quantitative analysis of bioluminescence intensity for liver 6 h after injection (n = 2 mice per group). f Representative bioluminescence images of major organs 6 h after injection of LMA2-C10 LNPs with different RNA-to-lipid ratios (n = 3 mice per group). g Quantitative analysis of bioluminescence intensity in liver shown in (f) (n = 3 mice per group). h Representative bioluminescence images of major organs of mice 6 h after injection of LMA2-C10 LNPs with different helper lipids (n = 3 mice per group). i Quantified hepatic luciferase expression shown in (h) (n = 3 mice per group). All LNP formulations were administered intravenously at an mRNA dose of 0.75 mg/kg. Data are shown as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by one-way ANOVA with Dunnett’s test for multiple comparisons or Student’s t test. Abbreviations: H, heart; Li, liver; S, spleen; Lu, lung; K, kidney; LN, lymph node. Source data are provided as a Source Data file. A combinatorial library of malic acid-derived ionizable lipids for mRNA delivery. a The synthetic pathway of malic acid-derived lipids (LMAX-CX or DMAX-CX), and the starting materials for LMAX-CX or DMAX-CX synthesis. In LMAX-CX or DMAX-CX, the first ‘X’ indicates the carbon count of the diamine unit attached to LMA or DMA, and the second that of the terminal alkyl chain; ‘diene’ is appended when the tail contains two double bonds. b Size distribution and PDI of the formulated MA-LNPs measured by dynamic light scattering (n = 3 technical replicates). c EE of MA-LNPs measured by Quant-it RiboGreen RNA assay (n = 3 technical replicates). d Quantification of bioluminescence intensity for each specific organ as a percentage of the overall organ system (n = 2 mice per group). e Quantitative analysis of bioluminescence intensity for liver 6 h after injection (n = 2 mice per group). f Representative bioluminescence images of major organs 6 h after injection of LMA2-C10 LNPs with different RNA-to-lipid ratios (n = 3 mice per group). g Quantitative analysis of bioluminescence intensity in liver shown in (f) (n = 3 mice per group). h Representative bioluminescence images of major organs of mice 6 h after injection of LMA2-C10 LNPs with different helper lipids (n = 3 mice per group). i Quantified hepatic luciferase expression shown in (h) (n = 3 mice per group). All LNP formulations were administered intravenously at an mRNA dose of 0.75 mg/kg. Data are shown as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by one-way ANOVA with Dunnett’s test for multiple comparisons or Student’s t test. Abbreviations: H, heart; Li, liver; S, spleen; Lu, lung; K, kidney; LN, lymph node. Source data are provided as a Source Data file. Screening of MA LNPs for mRNA delivery in vivo The transfection efficiency of malic acid-derived lipids (MA) for mRNA delivery was evaluated through the formulation of LNPs composed of malic acid-derived lipid, 1,2-Distearoyl-sn-glycero-3-phosphorylcholine (DSPC), cholesterol, and DMG-PEG2000 encapsulating firefly luciferase (Fluc) mRNA. A library comprising 30 distinct MA LNP formulations was constructed. The resulting MA LNPs exhibited hydrodynamic diameters ranging from 79 to 152 nm, demonstrating favourable monodispersity with over 97% of formulations displaying polydispersity index (PDI) values under 0.2 (Fig. 1b). Encapsulation efficiency (EE) exceeding 75% was achieved in 80% of the prepared MA LNPs (Fig. 1c). To evaluate mRNA delivery efficiency and biodistribution, 24 structurally distinct MA LNPs were individually administered to mice via intravenous injection. In vivo bioluminescence confirmed predominant hepatic and splenic localization of firefly luciferase expression (Fig. 1d). A clear structure–activity relationship was identified between the hydrophobic tail length of malic acid-derived lipids and their hepatic accumulation. Specifically, a majority of lipids with shorter hydrophobic tails exhibited enhanced liver-targeting efficiency, indicating that tail length is a key determinant of organ-level biodistribution (Fig. 1d, e). In addition, quantitative analysis revealed that lipids incorporating diamine linkers with an even number of methylene units (e.g., ethylenediamine and butanediamine) mediated superior mRNA transfection compared to those with odd-numbered linkers (e.g., propanediamine and pentanediamine) (Fig. 1e). Furthermore, the stereo-chemistry of the malic acid center was found to critically influence transfection efficiency, as LNPs formulated with DMA lipids exhibited markedly reduced mRNA delivery relative to their L-enantiomer counterparts, thereby establishing molecular chirality as a key determinant of lipid nanoparticle performance (Fig. 1e). Comparative analysis demonstrated superior hepatic transfection efficiency of the top-performing LMA LNPs relative to commercial ionizable lipids including SM102 and DLin-MC3-DMA (Fig. 1e). Specifically, LMA2-C10 LNPs achieved 1.5-fold and 3.3-fold higher transfection efficiency than SM102 LNPs and DLin-MC3-DMA LNPs, respectively (Fig. 1e). Optimization of LMA2-C10 LNPs for liver-specific RNA delivery Substantial modulation of RNA delivery efficiency by lipid nanoparticle formulations has been well established35. To streamline experimental workflows and minimize animal usage in accordance with the 3R principles including replacement, reduction, and refinement, rational optimization was performed on the preferred LMA2-C10 platform through systematic adjustment of helper lipids, RNA-to-lipid ratios, and quaternary compositions. Progressive attenuation of hepatic luciferase expression was observed with increasing RNA-to-lipid mass ratios (Fig. 1f, g and Supplementary Fig. 33). Subsequent comparative analysis of DSPC versus 1,2-Dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) helper lipids demonstrated significantly enhanced hepatic luciferase expression upon DOPE substitution for DSPC (Fig. 1h), yielding a 1.33-fold increase relative to standard DSPC formulations (Fig. 1i). Based on hepatic mRNA delivery efficiency and targeting specificity, the RNA-to-lipid mass ratio of 3.9% and DOPE helper lipid were selected for subsequent formulation optimization. DOE methodology was employed to systematically investigate multiple formulation parameters concurrently. 15 LMA2-C10 LNP formulations with different lipid compositions were designed using DOE software, encompassing combinatorial variations of core components, with the original formulation (#1) and commercial SM102 LNPs serving as comparative controls (Fig. 2a, Table S2). Microfluidic-fabricated LNPs exhibited hydrodynamic diameters of 73–116 nm with PDI consistently under 0.21 (Fig. 2b), while Fluc mRNA encapsulation efficiencies exceeded 80% for all formulations (Fig. 2c). In vivo bioluminescence imaging revealed significant formulation-dependent variations in transfection efficiency (Fig. 2d and Supplementary Fig. 34), with LNP#15 demonstrating 1.33-fold and 2.20-fold enhancement relative to the original formulation (#1) and SM102 LNPs, respectively (Fig. 2e).Fig. 2Optimising the Liver-targeting LNP system for maximal RNA transfection using DOE analysis.a Molar ratios of the LMA2-C10, DOPE, cholesterol and DMG-PEG2000. b Size distribution and PDI of the LNPs formulated according to DOE measured by dynamic light scattering (n = 3 technical replicates). c EE of LNPs formulated according to DOE measured by Quant-it RiboGreen RNA assay (n = 3 technical replicates). d In vivo imaging of bioluminescence in the major organs was conducted by IVIS imaging System after mice were intravenously injected with the different formulations of LNPs shown in (a) after 6 h post-administration (n = 3 mice per group). e Quantified luciferase expression in the liver from the 16 representative LNPs with different DOE formulations. Data are shown as mean ± s.e.m. (n = 3 mice per group). f Response surface plots modeled from the data presented in (e), showing the LNP compositions predicted capacity for liver transfection efficiency. g List of all formulation variations used to validate the optimised LNP formulations. h In vivo imaging of bioluminescence in the major organs was conducted by IVIS imaging System after mice were intravenously injected with different formulations of LNP shown in (g) after 6 h post-administration during the DOE validation phase (n = 3 mice per group). i Quantified luciferase expression in the liver shown in (g). Data are shown as mean ± s.e.m. (n = 3 mice per group). ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by one-way ANOVA with Dunnett’s test for multiple comparisons. Abbreviations: H, heart; Li, liver; S, spleen; Lu, lung; K, kidney; LN, lymph node. Source data are provided as a Source Data file. Optimising the Liver-targeting LNP system for maximal RNA transfection using DOE analysis. a Molar ratios of the LMA2-C10, DOPE, cholesterol and DMG-PEG2000. b Size distribution and PDI of the LNPs formulated according to DOE measured by dynamic light scattering (n = 3 technical replicates). c EE of LNPs formulated according to DOE measured by Quant-it RiboGreen RNA assay (n = 3 technical replicates). d In vivo imaging of bioluminescence in the major organs was conducted by IVIS imaging System after mice were intravenously injected with the different formulations of LNPs shown in (a) after 6 h post-administration (n = 3 mice per group). e Quantified luciferase expression in the liver from the 16 representative LNPs with different DOE formulations. Data are shown as mean ± s.e.m. (n = 3 mice per group). f Response surface plots modeled from the data presented in (e), showing the LNP compositions predicted capacity for liver transfection efficiency. g List of all formulation variations used to validate the optimised LNP formulations. h In vivo imaging of bioluminescence in the major organs was conducted by IVIS imaging System after mice were intravenously injected with different formulations of LNP shown in (g) after 6 h post-administration during the DOE validation phase (n = 3 mice per group). i Quantified luciferase expression in the liver shown in (g). Data are shown as mean ± s.e.m. (n = 3 mice per group). ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by one-way ANOVA with Dunnett’s test for multiple comparisons. Abbreviations: H, heart; Li, liver; S, spleen; Lu, lung; K, kidney; LN, lymph node. Source data are provided as a Source Data file. Key physicochemical parameters, including particle size, PDI, EE, and quantitative in vivo luminescence intensities of abdomen and liver, were incorporated into Design-Expert software for predictive modeling. All generated models achieved statistical significance (p < 0.05). Scatter plots of predicted versus experimental values (Supplementary Fig. 35) yielded R2 values exceeding 0.9, confirming robust model fitting. The response surface methodology (RSM) visualised lipid compositional effects on LNPs characteristics, with blue-to-red colour gradients indicating low-to-high response values for size, PDI, EE, and transfection efficiency (Fig. 2f and Supplementary Fig. 36). Given the favourable particle size and EE profiles across all 15 formulations, primary optimization focus was directed toward in vivo transfection efficiency, particularly hepatic targeting. Five optimal lipid compositions were identified through response surface methodology-derived modeling (Fig. 2g and Supplementary Fig. 37a, b) and subsequently validated in vivo. Bioluminescence imaging revealed superior transfection efficacy for LNP#2-5 and LNP#2-2 (Fig. 2h and Supplementary Fig. 37c), demonstrating 2.21-fold and 1.58-fold enhancement relative to commercial SM102 LNPs, respectively (Fig. 2i). These experimental findings demonstrated close alignment with computational predictions. Consequently, LMA2-C10 #2-5 LNPs were established as the optimal hepatocyte-targeting formulation. In vitro transfection efficiency The transfection efficiency of the optimised LMA2-C10 LNPs was evaluated in vitro. Results demonstrated that the transfection performance of LMA2-C10 LNPs in HepG2 cells was 3.7-fold higher than that of SM102 LNPs (Supplementary Fig. 38a). Cell viability was assessed using the cell counting kit-8 (CCK-8 assay). LMA2-C10 LNPs exhibited improved safety profiles compared to SM102 (Supplementary Fig. 38b). At a concentration of 5 μg/mL, cell viability remained nearly 99% (Supplementary Fig. 38b). To investigate the mechanism underlying the superior transfection efficiency of LMA2-C10 LNPs, lysosomal escape capacity of LNPs was evaluated. Compared with SM102 LNPs, LMA2-C10 LNPs exhibited evident endosomal escape beginning at 4 h post-incubation, with more pronounced escape observed at 6 h (Fig. 3a). This observation was corroborated by a significant decrease in Manders’ colocalization coefficient (Fig. 3b). In contrast, SM102 LNPs showed only a modest reduction in Manders’ coefficient at 4 h and 6 h relative to 2 h, with values remaining substantially higher than those of LMA2-C10 LNPs (Fig. 3a, b). These results suggested that the enhanced transfection efficiency of LMA2-C10 LNPs was attributable, at least in part, to more efficient endosomal escape.Fig. 3LMA2-C10 LNPs with antioxidant and anti-inflammatory capabilities mediates efficient mRNA expression in Urate-Biosynthetic Cells.a Assessment of endosomal escape efficiency of LMA2-C10 and SM102 LNPs. Representative fluorescence imaging of HepG2 cells incubated with Cy5-labeled mRNA encapsulated in either LMA2-C10 or SM102 LNPs. Cells were imaged at 2, 4, and 6 h post-incubation. Green: Lysosomes, Red: Cy5-labeled mRNA, arrows indicate escaped Cy5-mRNA. Scale bar: 10 μm. b Quantitative evaluation of endosomal escape was performed using Manders’ colocalization coefficient (n = 5 technical replicates). c The transfection efficiency of the mRNAEGFP@LMA2-C10 LNPs towards different liver cells in vivo was detected by flow cytometry after 6 h post-administration. d Quantification of EGFP+ cells in different cell population (PBS group n = 4 mice, LMA2-C10 LNPs group n = 6 mice). e Percentage of different cells in EGFP-positive cells. f Representative immunofluorescence images of various liver cells from three mice per group after dosing with EGFP LNPs. Scale bar: 50 μm. Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by two-sided Student’s t test or one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. LMA2-C10 LNPs with antioxidant and anti-inflammatory capabilities mediates efficient mRNA expression in Urate-Biosynthetic Cells. a Assessment of endosomal escape efficiency of LMA2-C10 and SM102 LNPs. Representative fluorescence imaging of HepG2 cells incubated with Cy5-labeled mRNA encapsulated in either LMA2-C10 or SM102 LNPs. Cells were imaged at 2, 4, and 6 h post-incubation. Green: Lysosomes, Red: Cy5-labeled mRNA, arrows indicate escaped Cy5-mRNA. Scale bar: 10 μm. b Quantitative evaluation of endosomal escape was performed using Manders’ colocalization coefficient (n = 5 technical replicates). c The transfection efficiency of the mRNAEGFP@LMA2-C10 LNPs towards different liver cells in vivo was detected by flow cytometry after 6 h post-administration. d Quantification of EGFP+ cells in different cell population (PBS group n = 4 mice, LMA2-C10 LNPs group n = 6 mice). e Percentage of different cells in EGFP-positive cells. f Representative immunofluorescence images of various liver cells from three mice per group after dosing with EGFP LNPs. Scale bar: 50 μm. Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by two-sided Student’s t test or one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. Hepatic tropism of LMA2-C10 LNPs toward urate-biosynthetic cells Transfection efficiency in distinct hepatic cell populations was evaluated by flow cytometry. The analysis demonstrated transfection in 50% of hepatocytes and 88% of endothelial cells, with negligible transfection in immune cells (Fig. 3c, d). Among transfected liver cells, hepatocytes, which are recognized as primary sites for urate biosynthesis36, accounted for 68.03 ± 4.49%, whereas endothelial cells constituted 28.75 ± 4.91% (Fig. 3e). Confocal microscopy independently validated these findings, confirming predominant mRNAEGFP@LMA2-C10 LNPs transfection in hepatocytes and endothelial cells (Fig. 3f). Substantially reduced transfection efficiency was observed in hepatic stellate cells, immune cells, and cholangiocytes (Fig. 3f). Antioxidant and low-immunogenicity properties of LMA2-C10 LNPs To evaluate whether the LMA2‑C10 LNPs retains the inherent antioxidant properties of malic acid, intracellular L‑malate levels were first measured. Following incubation with LMA2‑C10 LNPs, a marked increase in L‑malate was observed, peaking at 24 h at approximately 471 μM and returning to near‑baseline levels by 48 h (Fig.4a). This concentration falls within the 100-1000 μM range reported to exert anti‑inflammatory and antioxidant effects37–39, and is consistent with levels shown to drive nicotinamide adenine dinucleotide phosphate hydrogen (NADPH) regeneration40. To further assess the antioxidant capacity, HepG2 and RAW264.7 cells were pretreated with hydrogen peroxide (H₂O₂), followed by incubation with various LNPs. Intracellular ROS levels were detected using the ROS-sensitive fluorescent probe, 2′,7′-dichlorofluorescin diacetate (DCFH-DA). Elevated ROS levels were observed in cells treated with SM102 and DLin-MC3-DMA LNPs compared to the H₂O₂ group (Fig. 4b). In contrast, LMA2-C10 LNPs treatment significantly reduced intracellular ROS (Fig. 4b). Quantitative analysis using a microplate reader further confirmed the potent antioxidant activity of LMA2-C10 LNPs (Fig. 4c, d). To determine whether the observed ROS scavenging conferred functional cytoprotection, malondialdehyde (MDA), a terminal marker of lipid peroxidation, was measured. Cells treated with LMA2‑C10 LNPs exhibited a pronounced reduction in MDA levels relative to those exposed to H₂O₂ alone or in combination with SM102 or DLin-MC3-DMA LNPs (Fig. 4e). This decrease confirmed that the antioxidant activity of LMA2‑C10 LNPs mitigated substantive oxidative damage. Given that malate serves as a metabolic precursor for cellular NADPH synthesis41,42, changes in the NADPH/nicotinamide adenine dinucleotide phosphate (NADP⁺) and glutathione (GSH)/glutathione disulfide (GSSG) ratios were examined. After LMA2‑C10 LNPs treatment, the NADPH/NADP⁺ and GSH/GSSG ratio was markedly increased (Fig.4f, g). This shift elevated the reduced glutathione pool, thereby enhancing cellular antioxidant capacity.Fig. 4Antioxidant and Anti-inflammatory Effects of LMA2-C10 LNPs in vitro.a Changes in cellular L-malate levels over time following treatment of HepG2 cells with empty LMA2-C10 LNPs (n = 3 biologically independent samples). b Fluorescent imaging of ROS levels in RAW264.7 and HepG2 cells following treatment with H₂O₂ and various LNPs. Intracellular ROS were labeled with the green fluorescent probe DCFH-DA. Scale bar: 50 μm. c-d Quantitative analysis of ROS levels in RAW264.7 (c) and HepG2 (d) cells measured using a microplate reader (n = 3 biologically independent samples), data were normalised to the blank control (set as 100%). e Intracellular MDA levels in HepG2 cells following treatment with H₂O₂ and various LNPs (n = 3 biologically independent samples). f-g The relative NADPH/NADP+ ratio (f) and the cellular GSH/GSSG ratio (g) in HepG2 cells following treatment with H₂O₂ and various LNPs (n = 3 biologically independent samples). h-j Polarisation of macrophage following different LNPs treatment. RAW264.7 cells were treated with PBS, SM102, DLin-MC3-DMA, or LMA2-C10 LNPs for 24 h. CD86⁺ (M1) and CD206⁺ (M2) macrophage phenotypes were subsequently assessed. Quantitative analysis of the percentage of CD86⁺ macrophages (i) and CD206⁺ macrophages (j) after treatment with various LNPs (n = 3 biologically independent samples). k-l Expression levels of the inflammatory cytokines IL-1β (k) and IL-6 (l) in RAW264.7 cells after treatment with lipopolysaccharide (LPS) and different LNPs, as detected by qPCR (n = 3 biologically independent samples). m-n Inflammatory cytokine induction in vivo following LNPs administration. Serum inflammatory cytokine levels of IL-1β (m), IL-6 (n) after intravenously administrated LMA2-C10 LNPs or SM102 LNPs at an RNA dose of 0.75 mg/kg. After 12 h, plasma is collected for inflammatory cytokine ELISA detection (n = 3 mice per group). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by two-sided one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. Antioxidant and Anti-inflammatory Effects of LMA2-C10 LNPs in vitro. a Changes in cellular L-malate levels over time following treatment of HepG2 cells with empty LMA2-C10 LNPs (n = 3 biologically independent samples). b Fluorescent imaging of ROS levels in RAW264.7 and HepG2 cells following treatment with H₂O₂ and various LNPs. Intracellular ROS were labeled with the green fluorescent probe DCFH-DA. Scale bar: 50 μm. c-d Quantitative analysis of ROS levels in RAW264.7 (c) and HepG2 (d) cells measured using a microplate reader (n = 3 biologically independent samples), data were normalised to the blank control (set as 100%). e Intracellular MDA levels in HepG2 cells following treatment with H₂O₂ and various LNPs (n = 3 biologically independent samples). f-g The relative NADPH/NADP+ ratio (f) and the cellular GSH/GSSG ratio (g) in HepG2 cells following treatment with H₂O₂ and various LNPs (n = 3 biologically independent samples). h-j Polarisation of macrophage following different LNPs treatment. RAW264.7 cells were treated with PBS, SM102, DLin-MC3-DMA, or LMA2-C10 LNPs for 24 h. CD86⁺ (M1) and CD206⁺ (M2) macrophage phenotypes were subsequently assessed. Quantitative analysis of the percentage of CD86⁺ macrophages (i) and CD206⁺ macrophages (j) after treatment with various LNPs (n = 3 biologically independent samples). k-l Expression levels of the inflammatory cytokines IL-1β (k) and IL-6 (l) in RAW264.7 cells after treatment with lipopolysaccharide (LPS) and different LNPs, as detected by qPCR (n = 3 biologically independent samples). m-n Inflammatory cytokine induction in vivo following LNPs administration. Serum inflammatory cytokine levels of IL-1β (m), IL-6 (n) after intravenously administrated LMA2-C10 LNPs or SM102 LNPs at an RNA dose of 0.75 mg/kg. After 12 h, plasma is collected for inflammatory cytokine ELISA detection (n = 3 mice per group). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001 versus between groups by two-sided one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. Malic acid has been reported to influence macrophage polarization21. To evaluate this effect for LMA2-C10 LNPs, RAW264.7 cells were treated with different LNPs for 24 h. All LNPs induced different degree of M1 polarization, with SM102 LNPs showing the strongest effect, followed by DLin-MC3-DMA LNPs (Fig.4h, i). LMA2-C10 LNPs also led to a slight increase in M1 markers, but this did not reach statistical significance compared to the PBS group (Fig.4h, i). A modest increase in M2 macrophages was noted, though this change did not reach statistical significance (Fig.4h, j). This shift may be attributed to upregulated phagocytic activity induced by LNPs treatment. The immunostimulatory potential of LMA2-C10 LNPs was evaluated under inflammatory conditions. Enhanced expression of both IL-1β and IL-6 was observed following macrophage exposure to DLin-MC3-DMA LNPs (Fig. 4k, l). A modest elevation in IL-1β expression was also induced by SM102 LNPs. In contrast, LMA2-C10 LNPs not only failed to potentiate inflammatory cytokine production but also demonstrated inhibitory activity against IL-1β, a principal mediator of MSU-induced inflammation (Fig. 4k, l). The inflammatory potential of different LNPs was further examined in vivo. At 12 h post‑administration, serum IL‑1β levels reached 275 pg/mL in the SM102 LNPs group, which was statistically significant compared with other groups (Fig.4m). However, a modest elevation in IL‑1β was observed following LMA2‑C10 LNPs treatment, approximately 50 pg/mL, with no statistically significant difference relative to the PBS control (Fig.4m). In contrast, neither LNP induced a significant change in IL-6 levels relative to the PBS group (Fig.4n). Collectively, these findings support the characterization of LMA2-C10 LNPs as a delivery system with both antioxidant and low immunogenicity activity. Production and evaluation of circRNA and mRNA Template plasmids for circRNA and mRNA were constructed, with circRNA synthesized via permuted intron-exon (PIE) splicing strategy27. Following synthesis, circRNA was purified via an ÄKTA™ Avant 25 instrument. In vivo evaluation of circRNA@LMA2-C10 LNPs versus 1-methylpseudouridine (m1Ψ) -modified mRNA@LMA2-C10 LNPs demonstrated significantly prolonged expression from circRNA without detectable hepatorenal toxicity (Supplementary Fig. 41). At elevated doses, hepatic and renal function parameters remained within physiological ranges (Supplementary Fig. 42b). Splenomegaly was induced in the m1Ψ-mRNA group, whereas no such effect was detected with circRNA administration (Supplementary Fig. 42a). In the liver, elevated interferon‑alpha (IFN‑α) expression was observed following treatment with both circRNA and m1Ψ-mRNA relative to the PBS control group, with circRNA inducing slightly lower levels than m1Ψ-mRNA (Supplementary Fig. 42c). Expression of the interferon-stimulated gene (ISG) pathway markers Isg15 and radical s-adenosyl methionine domain containing 2 (Rsad2) was significantly upregulated in m1Ψ-mRNA-treated mice compared with the PBS group, whereas no significant increase was detected following circRNA treatment (Supplementary Fig. 42d, e). Notably, no activation of the retinoic acid-inducible gene I (RIG‑I) pathway was detected in the liver under any condition (Supplementary Fig. 42f, g). In the spleen, no elevation of IFN‑α or ISGs was observed (Supplementary Fig. 42h-j). Instead, activation of the RIG‑I/melanoma differentiation-associated gene 5 (MDA5) pathway was evident. Within this pathway, upregulation of interferon induced with helicase C domain 1 (Ifih1) was more pronounced following m1Ψ-mRNA administration (Supplementary Fig. 42k, l). CircRNA encoding urate oxidase (circRNAUox) was synthesized utilizing the PIE method and subsequently purified (Supplementary Fig. 43a, b, sequences detailed in Supplementary Data 1). As expected, circRNAUox demonstrated significantly greater resistance to RNase R digestion compared to linear RNAUox (Supplementary Fig. 43c). Fragments spanning the predicted junction sites were amplified and subjected to Sanger sequencing, which confirmed both successful circularization (Supplementary Fig. 43d). Western blot analysis of transfected HepG2 cells revealed marginally higher uricase expression from circRNAUox compared to m1Ψ-mRNAUox at 24 h, significantly exceeding unmodified mRNA and linear precursor controls (Fig. 5a, b). By 48–72 h, circRNAUox maintained stable expression, whereas m1Ψ-mRNA progressively declined (Fig. 5a, b). Enzymatic activity assays verified functional uricase production (Fig. 5c). CircRNAUox exhibited superior enzymatic activity levels and sustained stability over m1Ψ-mRNAUox, consistent with Western blot data (Fig. 5c).Fig. 5Design and functionality of circRNAspUox.a The expression levels of circRNAUox in vitro. Western blot analysis showing the expression level of uricase in HepG2 cells transfected with circRNAUox, linear RNA precursorUox, m1Ψ-modified mRNAUox, mRNAUox. Untreated and circRNAFluc were considered as negative controls. b Quantitative analysis of uricase expression. Relative expression was calculated as the ratio of uricase/α-tubulin, which was presented as fold change versus mRNAUox (n = 3 biologically independent samples). c Enzyme activity assay following in vitro transfection with circRNAUox. The enzyme activity was detected in HepG2 cells at 24 h, 48 h, and 72 h following transfection with circRNAUox, linear RNA precursorUox, m1Ψ-modified mRNAUox, mRNAUox. Untreated and circRNAFluc were considered as negative controls (n = 3 biologically independent samples). d Schematic diagram of PIE(Ana)-mediated RNA circularization. Coding sequences for Uox and signal peptide sequence (sp) were cloned into circRNA plasmids. e Purification of circRNAspUox. AKTA chromatogram of circRNAspUox via an ÄKTA™ Avant 25 instrument. f Agarose-gel electrophoresis result of the collected fractions in (e), where B represents the purified circRNA. g RNase R cleavage assay of the circRNAspUox. Agarose-gel electrophoresis result of linear RNA and circRNA treated with RNase R for 10 min. h Sanger sequencing result of the junction site of the circRNAspUox. i-j In vitro enzymatic activity assays of supernatant (i) and intracellular (j) fractions after transfection with varying ratios of circRNAUox and circRNAspUox. HepG2 cells were transfected with circRNAUox and circRNAspUox at mass ratios of 4:0, 3:1, 2:2, 1:3, and 0:4, followed by measurement of enzymatic activity at 24 h, 48 h, and 72 h post-transfection (n = 3 biologically independent samples). k Ratio of supernatant to intracellular enzymatic activity after transfection with different circRNAUox: circRNAspUox ratios (n = 3 biologically independent samples). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided repeated-measures ANOVA with Geisser–Greenhouse correction and Tukey’s post hoc test for multiple comparisons. Source data are provided as a Source Data file. Design and functionality of circRNAspUox. a The expression levels of circRNAUox in vitro. Western blot analysis showing the expression level of uricase in HepG2 cells transfected with circRNAUox, linear RNA precursorUox, m1Ψ-modified mRNAUox, mRNAUox. Untreated and circRNAFluc were considered as negative controls. b Quantitative analysis of uricase expression. Relative expression was calculated as the ratio of uricase/α-tubulin, which was presented as fold change versus mRNAUox (n = 3 biologically independent samples). c Enzyme activity assay following in vitro transfection with circRNAUox. The enzyme activity was detected in HepG2 cells at 24 h, 48 h, and 72 h following transfection with circRNAUox, linear RNA precursorUox, m1Ψ-modified mRNAUox, mRNAUox. Untreated and circRNAFluc were considered as negative controls (n = 3 biologically independent samples). d Schematic diagram of PIE(Ana)-mediated RNA circularization. Coding sequences for Uox and signal peptide sequence (sp) were cloned into circRNA plasmids. e Purification of circRNAspUox. AKTA chromatogram of circRNAspUox via an ÄKTA™ Avant 25 instrument. f Agarose-gel electrophoresis result of the collected fractions in (e), where B represents the purified circRNA. g RNase R cleavage assay of the circRNAspUox. Agarose-gel electrophoresis result of linear RNA and circRNA treated with RNase R for 10 min. h Sanger sequencing result of the junction site of the circRNAspUox. i-j In vitro enzymatic activity assays of supernatant (i) and intracellular (j) fractions after transfection with varying ratios of circRNAUox and circRNAspUox. HepG2 cells were transfected with circRNAUox and circRNAspUox at mass ratios of 4:0, 3:1, 2:2, 1:3, and 0:4, followed by measurement of enzymatic activity at 24 h, 48 h, and 72 h post-transfection (n = 3 biologically independent samples). k Ratio of supernatant to intracellular enzymatic activity after transfection with different circRNAUox: circRNAspUox ratios (n = 3 biologically independent samples). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided repeated-measures ANOVA with Geisser–Greenhouse correction and Tukey’s post hoc test for multiple comparisons. Source data are provided as a Source Data file. A secreted uricase variant (circRNAspUox) (Fig. 5d and sequence detailed in Supplementary Data 1) was engineered using a luciferase-derived secretory peptide previously demonstrated to facilitate efficient protein secretion43,44. This construct was designed to concurrently reduce local urate production and clear pre-existing peripheral deposits. Following production, circRNAspUox underwent ÄKTA™ purification and validation (Fig. 5e–h). Enzymatic activity assays confirmed efficient extra-cellular secretion of functional spUox (Fig. 5i, j). Gradient co-delivery of circRNAUox and circRNAspUox demonstrated dose-dependent intracellular and extra-cellular enzymatic activity profiles (Fig. 5i–k). Uricase activity was detected in renal tissues in vivo following intravenous administration of circRNAspUox@LMA2-C10 LNPs (Supplementary Fig. 44). circRNAspUox@LMA2-C10 LNPs alleviates joint inflammation and reduces serum urate in acute gouty arthritis mice model The therapeutic efficacy of RNA@LMA2-C10 LNPs was subsequently evaluated in a GA model (Fig. 6a). Following MSU crystal injection, pronounced ankle swelling and erythema were observed (Fig. 6b, d). Histopathological analysis revealed extensive inflammatory cell infiltration, granuloma formation, acute synovial congestion with villous hyperplasia at articular surfaces (Fig. 6e). While allopurinol and uricase treatments marginally reduced serum urate levels (Fig. 6c), limited resolution of peripheral urate deposits was noted, which were attributable to allopurinol’s delayed onset and rapid uricase clearance. Consequently, only partial reduction in ankle swelling was achieved, with persistent histopathological lesions (Fig. 6d, e). Comparative assessment demonstrated inferior hypouricemic efficacy and gout resolution capacity of intracellular Uox (both mRNA and circRNA formats) versus secreted spUox (Fig. 6c–e). Mild joint swelling persisted after intracellular Uox treatment, with histopathology revealing residual inflammation and granulomatous foci despite normal synovial architecture (Fig. 6d, e). CircRNA-based delivery significantly attenuated inflammatory infiltration relative to mRNA counterparts (Fig. 6e). CircRNAspUox monotherapy achieved comparable gout resolution to the circRNAUox+spUox combination regimen, while demonstrating superior hypouricemic efficacy as monotherapy (Fig. 6c).Fig. 6circRNAspUox@LMA2-C10 LNPs mitigates inflammation and joint damage in acute gouty arthritis.a Schematic diagram of GA therapy procedure. The model was established by intra-articular injection of MSU crystals. The allopurinol group received oral administration at -1, 23, and 47 h, while other groups received intravenous injections of PBS, uricase (5 U/kg), or different RNA@LMA2-C10 LNPs (RNA dose = 0.6 mg/kg) at -1 h. b Ankle diameter measurements following intra-articular injection of MSU crystals or PBS across different treatment groups (n = 6 mice per group). c Serum urate concentrations at 48 h after intra-articular injection of MSU crystals in different treatment groups (n = 6 mice per group). d Representative macroscopic images of ankle joints from different therapeutic groups. The grid provides the scale, with each square measuring 1 × 1 cm. e Representative H&E-stained sections of ankle joint regions in MSU-induced acute gouty arthritis. Scale bar: 500 μm. f Gait functional assessment of GA mice using dynamic footprint analysis. Representative curves depict dynamic variations in hindpaw contact area during stride progression. g-i Quantitative gait parameters including stride length CV (g), paw angular displacement (h), and stance/swing phase ratio (i) across therapeutic groups (n = 6 mice per group, with both hind limbs analysed, yielding 12 measurements). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided repeated-measures ANOVA with Geisser–Greenhouse correction and Tukey’s post hoc test or one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. circRNAspUox@LMA2-C10 LNPs mitigates inflammation and joint damage in acute gouty arthritis. a Schematic diagram of GA therapy procedure. The model was established by intra-articular injection of MSU crystals. The allopurinol group received oral administration at -1, 23, and 47 h, while other groups received intravenous injections of PBS, uricase (5 U/kg), or different RNA@LMA2-C10 LNPs (RNA dose = 0.6 mg/kg) at -1 h. b Ankle diameter measurements following intra-articular injection of MSU crystals or PBS across different treatment groups (n = 6 mice per group). c Serum urate concentrations at 48 h after intra-articular injection of MSU crystals in different treatment groups (n = 6 mice per group). d Representative macroscopic images of ankle joints from different therapeutic groups. The grid provides the scale, with each square measuring 1 × 1 cm. e Representative H&E-stained sections of ankle joint regions in MSU-induced acute gouty arthritis. Scale bar: 500 μm. f Gait functional assessment of GA mice using dynamic footprint analysis. Representative curves depict dynamic variations in hindpaw contact area during stride progression. g-i Quantitative gait parameters including stride length CV (g), paw angular displacement (h), and stance/swing phase ratio (i) across therapeutic groups (n = 6 mice per group, with both hind limbs analysed, yielding 12 measurements). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided repeated-measures ANOVA with Geisser–Greenhouse correction and Tukey’s post hoc test or one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. Gait analysis was performed to assess hindlimb pain and locomotor impairment in GA mice. Following MSU crystal injection, model mice exhibited irregular stride patterns due to ankle joint pain (Fig. 6f), as evidenced by increased coefficient of variation (CV) in stride length (Fig. 6g). Concurrently, the hindpaw angle was significantly increased as a consequence of pain avoidance behaviour and joint swelling (Fig. 6h). A corresponding elevation in the stance/swing ratio was consistently observed, indicating pronounced gait compensation (Fig. 6i). Administration of circRNAspUox@LMA2-C10 LNPs restored normal stride regularity and significantly reduced the coefficient of variation in stride length (Fig. 6f, g). The treatment also normalised hindpaw angle and stance/swing ratio to baseline levels (Fig. 6h, i). All RNA@LMA2-C10 LNPs formulations demonstrated substantial therapeutic efficacy in gait recovery. While allopurinol and uricase treatments showed modest improvement in gait parameters, the effects did not reach statistical significance compared to untreated controls (Fig. 6f–i). Given that circRNAFluc@LMA2-C10 LNPs also exhibited mild urate-lowering effects (Fig. 6c), a direct comparison between RNA-loaded and empty LNPs was performed (Supplementary Fig. 45). In the acute gouty arthritis model, treatment with empty LMA2-C10 LNPs conferred therapeutic benefits nearly equivalent to those of circRNAFluc@LMA2-C10 LNPs (Supplementary Fig. 45). These results indicate that the urate-lowering and anti-inflammatory effects observed in the circRNAFluc@LMA2-C10 group were largely attributable to the LMA2-C10 LNPs itself, rather than the encapsulated Fluc RNA. circRNAspUox@LMA2-C10 LNPs ameliorates chronic hyperuricemia and hyperuricemic nephropathy To further evaluate the therapeutic effect of circRNAspUox@LMA2-C10 LNPs on chronic hyperuricemia, associated metabolic dysfunction and its complication hyperuricemic nephropathy (HN), UOX-KO mice were utilized as an experimental model45 (Fig. 7a, Supplementary Fig. 46). UOX-KO mice, generated by deleting the uricase gene, effectively mimic human hyperuricemia, exhibiting persistent hyperuricemia, dyslipidemia, and abnormal glucose metabolism46. In later disease stages, significant renal dysfunction, tubulointerstitial lesions, and obstructive nephropathy develop46.Fig. 7circRNAspUox@LMA2-C10 LNPs alleviates chronic hyperuricemia and hyperuricemic nephropathy.a Schematic diagram of the HN treatment protocol. UOX-KO mice were administered daily oral allopurinol or intravenous injections (every 5 days) of PBS or uricase (5 U/kg) or different RNA@LMA2-C10 LNPs (RNA dose = 0.2 mg/kg). Blood samples were collected periodically for serum UA, BUN, and creatinine measurements. Mice were sacrificed on day 29. C57BL/6 mice served as controls. b Changes in serum uric acid levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). c Changes in serum creatinine levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). d Changes in BUN levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). e Representative images of renal histopathology (H&E, PAS, and Masson’s trichrome staining) in UOX-KO mice. Scale bars: white (250 μm); black (1 cm). Green asterisks, glomerular collapse; hash signs, tubular dilation; blue arrows, sloughed epithelial cells and cellular debris within tubular lumina; black arrows, cast formation; white asterisks, extensive inflammatory cell infiltration; red arrows, collagen deposition and tubulointerstitial fibrosis. f Renal tubular injury scores in hyperuricemic nephropathy mice (n = 5 mice per group). g Quantitative analysis of Masson’s trichrome-positive areas in kidney tissues (n = 5 mice per group). h–k Post-treatment serum lipid profiles in UOX-KO mice, including LDL-C (h), TG (i), TC (j), HDL-C (k) (n = 5 mice per group). l Blood glucose levels in UOX-KO mice (n = 5 mice per group). m–o Serum inflammatory cytokine levels of IL-1β (m), TNF-α (n), IL-6 (o) in treated UOX-KO mice (n = 5 mice per group). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. circRNAspUox@LMA2-C10 LNPs alleviates chronic hyperuricemia and hyperuricemic nephropathy. a Schematic diagram of the HN treatment protocol. UOX-KO mice were administered daily oral allopurinol or intravenous injections (every 5 days) of PBS or uricase (5 U/kg) or different RNA@LMA2-C10 LNPs (RNA dose = 0.2 mg/kg). Blood samples were collected periodically for serum UA, BUN, and creatinine measurements. Mice were sacrificed on day 29. C57BL/6 mice served as controls. b Changes in serum uric acid levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). c Changes in serum creatinine levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). d Changes in BUN levels in UOX-KO mice across different treatment groups over time (n = 5 mice per group). e Representative images of renal histopathology (H&E, PAS, and Masson’s trichrome staining) in UOX-KO mice. Scale bars: white (250 μm); black (1 cm). Green asterisks, glomerular collapse; hash signs, tubular dilation; blue arrows, sloughed epithelial cells and cellular debris within tubular lumina; black arrows, cast formation; white asterisks, extensive inflammatory cell infiltration; red arrows, collagen deposition and tubulointerstitial fibrosis. f Renal tubular injury scores in hyperuricemic nephropathy mice (n = 5 mice per group). g Quantitative analysis of Masson’s trichrome-positive areas in kidney tissues (n = 5 mice per group). h–k Post-treatment serum lipid profiles in UOX-KO mice, including LDL-C (h), TG (i), TC (j), HDL-C (k) (n = 5 mice per group). l Blood glucose levels in UOX-KO mice (n = 5 mice per group). m–o Serum inflammatory cytokine levels of IL-1β (m), TNF-α (n), IL-6 (o) in treated UOX-KO mice (n = 5 mice per group). Data are presented as mean ± s.e.m. ns, p > 0.05; *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 versus between groups by two-sided one-way ANOVA with Dunnett’s test for multiple comparisons. Source data are provided as a Source Data file. To model the clinical challenge of treating advanced refractory hyperuricemia with renal impairment, therapeutic administration was initiated in 12-week-old mice. At this stage, blood samples from UOX-KO mice revealed markedly elevated serum urate levels, approximately five times higher than those in wild-type mice (Fig. 7b). After four weeks of treatment, low-dose (0.2 mg/kg) RNA@LMA2-C10 LNPs administration substantially reduced serum urate levels in UOX-KO mice. Among all treatment groups, circRNAspUox@LMA2-C10 LNPs demonstrated the most potent urate-lowering effect, followed by the circRNAUox+spUox combination group (Fig. 7b). Secreted uricase constructs exhibited superior efficacy in reducing serum urate compared to non-secreted forms, and circRNA-based therapies outperformed mRNA-based treatments. Although daily oral allopurinol administration effectively controlled serum urate levels, its efficacy remained slightly inferior to circRNAspUox@LMA2-C10 LNPs. Exogenous uricase initially reduced serum urate, but its therapeutic effect diminished with repeated dosing (Fig. 7b). Chronic hyperuricemia was found to induce renal urate deposition, resulting in progressive kidney injury47. Renal function assessment revealed significantly elevated serum creatinine (Scr) and blood urea nitrogen (BUN) in PBS-treated UOX-KO mice (Fig. 7c, d). CircRNAspUox@LMA2-C10 LNPs administration produced measurable renal functional improvement within 4 days of treatment (Fig. 7c, d). Secreted mRNA constructs and uricase therapy initiated creatinine reduction after two doses, whereas allopurinol and non-secreted RNA formulations showed minimal short-term improvement in renal function (Fig. 7c, d). These findings demonstrate the superior therapeutic potential of circRNAspUox@LMA2-C10 LNPs in ameliorating hyperuricemia and its associated renal complications. These findings were corroborated by parallel improvements in renal histopathology. Histological examination of kidney sections revealed significant morphological alterations in UOX-KO mice, including both renal atrophy and hypertrophy, with near complete loss of normal renal architecture (Fig. 7e). Obstructive nephropathy was observed in UOX-KO mice, characterised by prominent cavitation and parenchymal compression. Various treatment groups elicited partial restoration of renal pathology post-intervention. Notably, circRNAspUox@LMA2-C10 LNPs treatment demonstrated superior efficacy in renal morphological restoration (Fig. 7e). High-power microscopic analysis identified extensive lymphocytic infiltration, interstitial expansion, and parenchymal collapse in UOX-KO kidneys (Fig. 7e, f). Periodic acid–Schiff (PAS) staining revealed thickening of the renal basement membrane and the presence of protein casts in renal tubules (Fig. 7e). Significant tubulointerstitial fibrosis was further confirmed by Masson’s trichrome staining (Fig. 7e). Notably, a significant reduction in lymphocyte infiltration and marked alleviation of interstitial expansion were observed following administration of circRNA and mRNA constructs encoding secretory uricase, with results superior to those achieved with non-secretory uricase RNA, allopurinol, or uricase treatment groups (Fig. 7e, f). A pronounced decrease in tubular casts was evident across all treatment groups after therapeutic intervention (Fig. 7e). Regarding fibrosis mitigation, all RNA@LMA2-C10 LNPs treatment groups, with the exception of mRNAUox@LMA2-C10 LNPs, exhibited effective reduction in fibrotic areas (Fig. 7e, g). Superior outcomes were achieved with secretory uricase constructs compared to non-secretory variants, and circRNA-based formulations outperformed their mRNA counterparts (Fig. 7e–g). The therapeutic effects were particularly pronounced in the circRNAspUox@LMA2-C10 LNPs treatment group, which showed near-complete resolution of inflammatory infiltrates, significant fibrosis was absent, and substantial preservation of renal micro-structure (Fig. 7e–g). circRNAspUox@LMA2-C10 LNPs attenuates hyperuricemia associated metabolic dysfunction Blood low-density lipoprotein cholesterol (LDL-C) levels revealed significant reductions in all treatment groups except the uricase group, with trends closely paralleling serum urate outcomes (Fig. 7h). Total cholesterol (TC) and triglycerides (TG) measurements demonstrated superior lipid-lowering efficacy of circRNAspUox@LMA2-C10 LNPs, where both secretory uricase and circRNA formulations exhibited marked advantages (Fig. 7i, j). No statistically significant differences were observed in high-density lipoprotein cholesterol (HDL-C) levels across treatment groups (Fig. 7k). Glucose regulation was improved to varying degrees in all therapeutic groups, with circRNAspUox@LMA2-C10 LNPs showing particularly potent hypoglycemic effects that surpassed both allopurinol and uricase positive controls (Fig. 7l). Given the potential inflammatory activation from sustained hyperuricemia, circulating inflammatory markers were quantitatively assessed. UOX-KO mice exhibited significantly elevated serum tumor necrosis factor-alpha (TNF-α), IL-6 and IL-1β compared to wild-type controls (Fig. 7m–o). All treatment regimens reduced these proinflammatory cytokines to comparable levels, with no statistically distinguishable differences observed between therapeutic approaches (Fig. 7m–o). In vivo safety evaluation Acute toxicity assessment of circRNAspUox@LMA2-C10 LNPs in the GA model was performed through comprehensive histopathological and biochemical analyses. Histological examination by hematoxylin and eosin (H&E) staining revealed no pathological abnormalities in major organs across all treatment groups (Supplementary Fig. 47). Assessment of hepatic and renal function markers, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), and Scr revealed no statistically significant differences from normal controls (Supplementary Fig. 48a–d). Although BUN levels exhibited a slight reduction, the change was not considered clinically relevant (Supplementary Fig. 48e). Levels of LDL-C and TC were within the normal range in all groups (Supplementary Fig. 48f, g). Furthermore, a modest yet statistically significant increase in HDL-C, which is considered a beneficial effect, was specifically observed in the mRNAspUox group (Supplementary Fig. 48h). These findings suggested favourable lipid-modulating effects without adverse metabolic consequences (Supplementary Fig. 48f–h). In the chronic hyperuricemia UOX-KO mouse model, an end-of-study safety evaluation was conducted through systematic examination of vital organs and blood parameters following repeated administration. Consistent with acute observations, no treatment-related pathological alterations were detected in any organ systems (Supplementary Fig. 49). Notably, renal function improvement was documented without evidence of hepatotoxicity (Fig. 7e), as evidenced by stable AST, ALT, and ALP levels that remained within normal physiological ranges (Supplementary Fig. 50a–c). No anti-PEG antibodies were detected in any treatment group after six repeated doses, with levels remaining comparable to wild-type controls (Supplementary Fig. 50d). This finding is particularly relevant given that anti-PEG antibody formation is a known limitation of clinically used PEGylated uricase. These collective data demonstrate the excellent safety profile and biocompatibility of circRNAspUox@LMA2-C10 LNP formulation, supporting its potential for therapeutic application in hyperuricemia management. Discussion The clinical management of hyperuricemia remains constrained by frequent dosing requirements and immunogenicity associated with exogenous enzyme therapies, resulting in fluctuating serum urate levels25. Protein replacement strategies utilizing endogenous translational machinery to express uricase represent an emerging therapeutic paradigm, wherein uric acid is catalytically degraded to soluble allantoin. This approach offers a promising solution to current therapeutic limitations. A critical challenge involves achieving sustained, progressive release of therapeutic agents while simultaneously facilitating the clearance of preexisting urate crystals to mitigate crystal-induced inflammation and fibrosis. In this investigation, a promising strategy was developed utilizing low-toxicity, low-immunogenicity and antioxidative malic acid-derived LNPs for the delivery of circRNA encoding a secretory uricase. This system enabled hepatic metabolism of uric acid while simultaneously promoting clearance of peripherally deposited urate crystals, thereby offering a therapeutic intervention for chronic hyperuricemia and its associated complications. A combinatorial library of malic acid-derived ionizable lipids was synthesized, leveraging the inherent low-immunogenicity and antioxidant properties of the malic acid backbone. LNPs were subsequently formulated from these MA lipids. Through systematic screening, LMA2-C10 LNPs was identified as the lead formulation, demonstrating 2.21-fold greater hepatic expression efficiency than the commercial benchmark SM102 following DOE optimization. Delivery of circRNA encoding secretory uricase mediated potent urate-lowering effects in hyperuricemic models. Furthermore, significant amelioration of disease complications was observed, with pronounced suppression of articular inflammation in acute gouty arthritis and significant mitigation of fibrotic pathology in hyperuricemic nephropathy. The widespread application of current LNPs systems in protein replacement therapy remains constrained by their inherent immunogenicity and long-term safety concerns48. Polymalic acid, recognized for its favourable biocompatibility and biodegradability, is widely employed as a drug delivery vehicle49. However, its clinical translation has been limited by challenges such as low production yield, high cost, and difficulties in controlling polymerisation degree19. Notably, malic acid has not yet been explored as a structural backbone for ionizable lipids in nucleic acid delivery. In this study, a class of malic acid-derived lipids were designed by exploiting the unique structural characteristics of malic acid (Fig. 1a). These LNPs based on malic acid-derived lipids exhibited superior in vivo expression performance (Fig. 1d, e). Four lead malic acid-derived lipids demonstrated enhanced expression efficiency relative to both SM102 and DLin-MC3-DMA controls (Fig. 1e). A systematic structure-activity relationship analysis identified key design parameters that influence transfection efficiency and targeting specificity, which has also been documented in previous studies50. Beyond its superior transfection properties, the inherent anti-inflammatory and antioxidant capacity of malic acid was identified as a principal rationale for the design of malic acid-based lipid formulations. Uric acid is acknowledged as an antioxidant, yet it may also act as a pro-oxidant under conditions of excessive concentration51. Accordingly, increased oxidative stress and systemic inflammation are consistently detected in hyperuricemic patients52. Our results demonstrated that, in contrast to the pro-oxidant characteristics associated with SM102 and DLin-MC3-DMA LNPs, the malic acid-derived LMA2-C10 LNPs retained the inherent antioxidant and low-immunogenicity activities of malic acid (Fig. 4). This antioxidant capacity was primarily attributed to its malate-based backbone, which undergoes intracellular degradation to release malic acid (Fig. 4a). As a metabolic precursor for NADPH41,53, malic acid facilitates the conversion of NADP⁺ to NADPH (Fig. 4f), thereby supplying reducing equivalents for the generation of reduced glutathione (Fig. 4g). This mechanism enhanced the NADPH/GSH pool54 and conferred protection against oxidative stress (Fig. 4b-e). Furthermore, compared with commercially available lipid formulations, LMA2-C10 LNPs induced markedly reduced the degree of M1-type macrophage polarization and attenuated activation of inflammatory cytokines, indicating a lower immunogenic profile (Fig. 4h–o). Similarly, lipoic acid-derived lipid nanoparticles retain their inherent anti-inflammatory and antioxidant capacity, as demonstrated in psoriasis treatment models55. Previous research has demonstrated that the pro-inflammatory properties of SM102 and ALC-0315 can be attenuated through substitution of PEG with Poly(carboxybetaine) or by incorporation of dexamethasone56,57. Lipids containing phenolic hydroxyl groups are known to confer antioxidant activity, yielding anti-inflammatory and antioxidant effects that are significantly lower than those of DLin-MC3-DMA LNPs but comparable to the LMA2-C10 LNPs58. Furthermore, in models of acute gouty arthritis and chronic hyperuricemia, administration of LMA2-C10 LNPs led to a moderate decrease in uric acid levels, along with a reduction in inflammatory cytokines (Fig. 5 and 6), an effect that may be attributed to its malic acid-based backbone. In this investigation, hepatocytes were prioritized as the primary site for uricase expression due to their central role in systemic urate production. Direct modulation at this biosynthetic origin was strategically pursued. Flow cytometric analysis revealed that LMA2-C10 LNPs-mediated transfection predominantly occurred in hepatocytes (68% of EGFP-positive cells), with endothelial cells representing the secondary target population (Fig. 3c–f). That targeting profile is distinctly different from other hepatotropic lipid nanoparticles developed for hyperuricemia treatment, which are predominantly accumulated in Kupffer cells59. Therefore, the LMA2-C10 LNPs platform has been demonstrated to represent a viable delivery system for hyperuricemia therapeutics. Uricase has been recognized as an ancestral enzyme capable of degrading uric acid. However, the clinical application of unmodified uricase is significantly limited due to its inherent immunogenicity as a foreign protein and rapid clearance from circulation, which represent common challenges for protein-based therapeutics60. To address these limitations, modification strategies involving PEGylation or poly(oligo(ethylene glycol) methyl ether methacrylate) (POEGMA) conjugates have been explored61. However, clinical studies have demonstrated that such approaches frequently induce significant anti-drug antibody responses, posing risks for repeated administration11,61. Minimizing dosing frequency and achieving effective clearance of peripheral urate deposits to prevent complications represent two principal clinical challenges in hyperuricemia management. Alternative approaches utilizing nanomaterial-encapsulated uricase have been developed for hyperuricemia treatment62. While nanoparticle encapsulation has been shown to extend the circulating half-life of uricase, the therapeutic effect on urate reduction is typically maintained for only 24–72 h9,62–64. The reported mRNA-based uricase therapy achieved significant serum urate reduction for 72–96 h at an RNA dose of 0.5 mg/kg59. By comparison, in this study circRNAspUox@LMA2‑C10 LNPs, administered at an RNA dose of 0.2 mg/kg every five days, conferred comparable and more durable urate control (Fig. 7a-b). Zhang et al. did not evaluate repeated dosing regimens or the long-term maintenance of uric acid homeostasis59. Additionally, their investigation provided only limited evidence regarding therapeutic efficacy against hyperuricemic nephropathy and refractory gout induced by chronic hyperuricemia59. In this study, these two challenges were addressed by employing a rational RNA design that combines the prolonged expression duration with the biological activity of secretory uricase and the antioxidant and low-immunogenicity activities of LNPs. To achieve prolonged urate-lowering effects with gradual release kinetics, circRNA had been employed as a stable alternative to conventional mRNA in this study. Experimental results have confirmed that circRNA-mediated expression exhibits significantly extended duration compared to modified mRNA, with both sustained expression profiles and increased total protein production (Fig. 5a–c). Incorporation of the secretion peptide allowed uricase to be secreted into the extra-cellular space (Fig. 5i–k) and to penetrate into tissues (Supplementary Fig. 44). This provides an effective means for the clearance of peripheral crystals. Following intravenous administration, serum urate levels have been maintained within the therapeutic range for up to 5 days (Fig. 7b). A dosing interval of 5 days has been demonstrated to sustain stable urate control. More potent urate-lowering effects were consistently observed with circRNA-based interventions in both acute gouty arthritis and chronic hyperuricemia models (Figs. 6c and 7b). As anticipated, circRNA constructs incorporating secretory peptide sequences demonstrated superior therapeutic efficacy in managing disease complications compared to both intracellularly retained uricase and co-expressed variants (Figs. 6 and 7). In hyperuricemic nephropathy models, circRNAspUox@LMA2-C10 LNPs administration produced measurable renal functional recovery within 4 days, while allopurinol or uricase showed minimal short-term improvement (Fig. 7c, d). Renal histopathology demonstrated significant mitigation of lymphocytic infiltration, interstitial expansion, and fibrotic progression in treated groups (Fig. 7e–g). Consistently, analogous therapeutic outcomes were demonstrated in the arthritis model, where joint pathology was ameliorated (Fig. 6e). The observed amelioration of hyperuricemia-associated complications by circRNAFluc@LMA2-C10 LNPs and by empty LMA2-C10 LNPs suggests that these effects were partially mediated by the LNP carrier itself, independent of uricase-dependent mechanisms (Supplementary Fig. 45). Although membrane-coating strategies improve enzyme stability, the dependence on localised injections or daily intravenous infusions nevertheless severely limits their clinical potential9,10,63. This enhanced performance was attributed, on the one hand, to the systemic biodistribution of catalytically active enzyme, which facilitated crystal dissolution at distal pathological sites. Allopurinol maintained moderate serum urate reduction with daily administration but showed limited capacity for peripheral crystal dissolution46. Uricase demonstrated better crystal clearance despite inferior systemic urate-lowering efficacy11. The protein replacement approach, particularly utilizing circRNA platforms incorporating secretory peptide sequences, combined the advantages of both small molecules and biologics while overcoming their respective limitations through sustained enzyme expression. Higher cumulative enzyme production and improved tissue penetration contribute to crystal clearance. On the other hand, it was attributed to the antioxidant and low-immunogenicity properties of the LNPs, which allow for repeated administration without loss of efficacy. This multifaceted advantage addresses the clinical need for therapies that simultaneously maintain stable serum urate levels, resolve existing crystal deposits, and prevent end-organ damage. Although this study definitively demonstrates the enhanced therapeutic efficacy of circRNAspUox@LMA2-C10 LNPs against chronic hyperuricemia and its complications, several limitations ought to be considered. Given that uricase-based therapies in clinical practice are primarily limited by the development of anti-drug antibodies and anti-PEG antibodies, which may pose a challenge in future translational development. No anti-PEG antibodies were detected throughout the dosing regimen. However, due to technical limitations of the assay, anti-uricase antibodies were not directly assessed. The therapeutic efficacy and safety profile need to be systematically evaluated in large animal models. Additionally, comprehensive optimization of the uricase coding sequence has not been thoroughly explored, which could potentially enhance both expression levels and duration of enzymatic activity. The uricase sequence itself contains a canonical C-terminal peroxisomal targeting signal (PTS1), which may compete with the secretion peptide for intracellular trafficking. Removal of the C-terminal three amino acids could potentially eliminate this competition and enhance uricase secretion. Methods Materials Cholesterol and Collagen IV was purchased from Sigma (USA), 1,2-distearoyl-sn-glycero-3-phosphocholine (DSPC), 1,2-dimyristoyl-rac-glycero-3-methoxypolyethylene glycol-2000 (DMG-PEG 2000), were obtained from Avanti (USA). 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE) and Heptadecan-9-yl 8- ((2-hydroxyethyl) (6-oxo-6- (undecyloxy) hexyl) amino) octanoate (SM-102) were purchased from XIAMEN Sinopeg Biotech Co., Ltd (China). DLin-MC3-DMA (MC3), Pronase E and allopurinol were purchased from MedChemExpress (USA). Quant-it RiboGreen RNA assay was purchased from Thermo Fisher. FastPure EndoFree Plasmid Mini Plus Kit, ChamQ SYBR aPCR Master mix, fetal bovine serum (FBS) (F102) and D-luciferin were purchased from Vazyme Biotech Co.,Ltd (China). Zombie Aqua™ Fixable Viability Kit (423102), Brilliant Violet 421™ anti-mouse CD45 (103134), PE anti-mouse CD31 (102407), PE anti-mouse F4/80 (123109), PE anti-mouse CD86 (105007), APC anti-mouse CD206 (141707), LEGEND MAX™ Mouse TNF-α ELISA Kit, and LEGEND MAX™ Mouse IL-6 ELISA Kit were purchased from BioLegend (USA). Anti-alpha smooth muscle Actin antibody (ab124964), anti-Albumin antibody (ab207327) were purchased from abcam (UK). Anti-Cytokeratin 19 antibody (14965-1-AP) and anti-CD45 (60287-1) antibody were purchased from proteintech. Anti-Mouse CD16/CD32 was purchased from BD Biosciences. Anti-CD31 antibody (GB113151) was purchased from Wuhan Servicebio Technology Co., Ltd. (China). Uricase antibody (sc-166214, mouse monoclonal antibody) for Western blotting was purchased from Santa Cruz Biotechnology (USA). DNase Ⅰ, and Rnase R were purchased from Kactus Biosystems (China). Mouse IL-1β ELISA Kit (CSB-E08054m) was purchased from CUSABIO (China), https://www.cusabio.com/. Mouse anti-PEG antibody ELISA Kit was purchased from Jiangsu Meimian Industrial Co., Ltd. Uricase activity test kit (BC4435), HE Staining Kit and HBSS were purchased from Beijing Solarbio Science & Technology Co., Ltd. (China). Uric acid test kit was purchased from Elabscience Biotechnology Co., Ltd. (China). EZ Cap™ Cy5 Firefly Luciferase mRNA (5-moUTP) was purchased from APExBIO (Houston, USA). Confocal Dish (1051000) was purchased from Saining Biotechnology Co., Ltd. (China). L-Malic Acid Assay Kit, MDA Assay Kit, NADP⁺/NADPH Assay Kit, GSH/GSSG Assay Kit and Amplex Red Uric Acid/Uricase Assay Kit were all purchased from Beyotime Biotechnology (China). Chemical synthesis of malic acid-derived lipids The chemical synthesis of all malic acid-derived lipids (LMAX-CX and DMAX-CX) is described in the Supporting Information (Supplementary Fig. 1–32). Briefly, malic acid-bearing amine cores (LMAX and DMAX) were prepared via amine–ester exchange of dimethyl L- or D-malate with the corresponding diamines. These intermediates were subsequently coupled with aldehyde-alkyl chains via reductive amination to afford the target lipids. Lipid nanoparticle formulation and characterization In the initial phase, the LMAX-CX lipid or DMAX-CX lipid, DSPC or DOPE, cholesterol, and DMG-PEG-2000 were mixed at a molar mass ratio of 48.5:10:40:1.5 in ethanol. mRNA or circRNA was dissolved in 0.05 M sodium acetate solution at pH = 5.0. The lipid mixture and RNA were mixed through the microfluidic mixing device (I Nano L, Micro&Nano Biologics Co., Ltd. China) at a ratio of 1:3 and a total flow rate of 16 mL/min. The mass ratio of lipids to RNA was 100:3.9. The LNP mixture was dialysed to remove ethanol using ultra-filtration tubes (10000Da MWCO, Merck Millipore, Germany) against phosphate-buffered saline (PBS, pH 7.4). In the DOE experimental phase, LMA2-C10 lipid, DOPE, cholesterol, and DMG-PEG-2000 were mixed at a molar mass ratio shown in Supplementary Table 2 and Fig. 2g. Other conditions are as mentioned above. Particle size and PDI of LNPs were measured with Dynamic light scattering (DLS) analyzer (Zetasizer Nano ZS, Malvern Panalytical, USA). The amount of encapsulated mRNA and encapsulation efficiency of LNPs were determined using Quant-it™ RiboGreen RNA Assay Kit. Synthesis and purification of mRNA The coding sequences for murine urate oxidase (Uox) (NCBI gene ID: 22262) were chemically synthesized and cloned into the transcription vector pcDNA3.1 by Guangzhou Ige Biotechnology Co., Ltd. The pcDNA3.1 vector contains the T7 promoter, 5′ and 3′ untranslated regions (UTRs), and a poly(A) tail (120 nucleotides). mRNA was produced using a T7 polymerase-mediated in vitro transcription (IVT) system from linearized pcDNA3.1 plasmid vectors. During the transcription reaction, caps and nucleoside triphosphate (NTP) were added. The m1Ψ-modified mRNAs were synthesized with uridine triphosphate (UTP) entirely replaced by 1-methylpseudouridine UTP. Following IVT, the DNA template was degraded by treating the reaction mixture with DNase I. The final transcribed mRNA was purified with a LiCl solution, and the final concentration of the purified mRNA was subsequently quantified by Nanodrop. Purified mRNA was analysed by agarose gel electrophoresis and stored at −80 °C until further use. The sequence of mRNAUox and mRNAspUox were provided in Supplementary Data 1. Synthesis and purification of circRNA The intron-exon alignment (PIE) system for manufacturing circRNA was employed in this experiment65. The pcDNA3.1 plasmids were constructed by Guangzhou Ige Biotechnology Co., Ltd. The synthesis of circRNA was initiated using a linearized circRNA plasmid as a template, which was transcribed in vitro via a T7 enzyme-mediated transcription system. Following IVT, the RNA product was incubated with DNase Ⅰ for 30 min to digest the DNA template. After purification with LiCl, the linear precursors of circRNA were isolated. Next, guanosine triphosphate (GTP) was added to the reaction mixture to a final concentration of 2 mM and incubated at 55 °C for 45 min to facilitate the cyclization of the RNA. The resulting circRNA was then purified using ÄKTA™ Avant purification system. The sequences of the circRNAs are provided in Supplementary Data 1. RNase R cleavage assay The purified circRNA and the mRNA were heated at 65 °C for 5 min and then cooled on ice. The reaction mixture was subsequently incubated with the RNase R at 37 °C for 10 min. The loading dye was added to stop reaction and the RNA was then resolved using agarose gel electrophoresis. Cell line HepG2 (cat# SCSP-510) were purchased it from the Cell Bank of the Chinese Academy of Sciences, and cultured with MEM medium containing 10% FBS,1% Glutamax, 1% Non-essential Amino Acids, and 1% Sodium Pyruvata Solution. RAW264.7 were purchased from Procell Life Science&Technology Co., Ltd. (Wuhan, China), and cultured with DMEM medium containing 10% FBS. Lysosomal escape assay HepG2 cells were seeded at a density of 2 × 10⁵ cells per confocal dish. After 12 h, lysosomes were labeled using CellLight™ Reagents BacMam 2.0 (lysosomes–RFP). Cells were then incubated with Cy5-labeled mRNA@LNPs (RNA concentration = 500 ng/mL) at 37 °C for 2–6 h. Following incubation, cells were washed three times with PBS, and nuclei were counterstained with Hoechst 33342 (10 μg/mL) for 10 min. Fluorescence images were acquired using a Zeiss Elyra 7 microscope. Manders’ colocalization coefficients were calculated using Image J software to quantify the degree of lysosomal escape. Measurement of ROS, MDA, GSH/GSSG ratio, NADP + /NADPH ratio in vitro HepG2 or RAW264.7 cells were seeded and stimulated with 200 μM H₂O₂, followed by a 6-h incubation with LNPs encapsulating luciferase-encoding mRNA. Cells without H₂O₂ stimulation or LNPs treatment served as the negative control. Cells stimulated with H₂O₂ but without LNP treatment were used as the positive control. Subsequently, the cells were gently washed with serum-free medium and incubated with 10 μM DCFH-DA in the dark at 37 °C for 30 min. After incubation, the cells were washed with PBS to remove excess probe. Fluorescent images were acquired using an inverted fluorescence microscope (DMi8, Leica, Germany). Fluorescence intensity was quantified with a microplate reader (Varioskan LUX, Thermo Scientific, USA) at excitation/emission wavelengths of 488/525 nm. HepG2 cells were seeded and cultured, followed by stimulation with 200 μM hydrogen peroxide. Cells were then incubated with empty LNPs for 6 h. Intracellular levels of MDA, NADPH/NADP⁺ ratio, and GSH/GSSG ratio were determined using commercial assay kits, according to the manufacturers’ instructions. Macrophage polarization RAW264.7 cells were seeded and cultured. After incubation with empty LNPs for 24 h, cells were collected and stained. Briefly, viability staining was performed using the Zombie Aqua™ Fixable Viability Kit for 30 min, followed by Fc receptor blockade with anti-CD16/32 antibody for 10 min. Surface marker staining was subsequently conducted. Samples were acquired on an ABcyte cytometer (ABclonal, China), and data were analysed using FlowJo V10 software. qPCR analysis RAW264.7 cells were seeded and stimulated with 1 μg/mL LPS for 24 h, followed by a 24-h incubation with LNPs encapsulating luciferase-encoding mRNA. Mice were administered a single intravenous injection of circRNA or mRNA LNPs at an RNA concentration of 2.5 mg/kg. After 7 days, the liver and spleen were harvested. Subsequently, total RNA was isolated from the cells using a TRIeasy™ Total RNA Extraction Reagent (19202, Yeasen, Shanghai, China). Complementary DNA (cDNA) was synthesized from equal amounts of total RNA with the iScript cDNA Synthesis Kit (1708891, Bio-Rad, USA) on a Thermal Cycler (Bio-Gener, USA). qPCR was performed using iQ SYBR Green Supermix on a CFX Connect Real-Time System (Bio-Rad, USA). The primer sequences used for qPCR amplification are listed in Table S3. Target mRNA expression levels were normalised to the endogenous reference genes glyceraldehyde-3-phosphate dehydrogenase (Gapdh) for murine cells. In vitro expression of mRNA or circRNA encoding uricase HepG2 cells were seeded at a density of 8 × 105 cells per well in a 6-well plate and incubated overnight for the subsequent transfection experiment. The cells were transfected with circRNAUox, linear precursorUox, mRNAUox, 1mΨ-modified mRNAUox, and circRNAFluc using Lipofectamine MessengerMax (Invitrogen). At 24–72 h post-transfection, the cell lysates were collected for Western blotting and enzyme activity assays. Uricase activity assays were measured by Uricase Activity Test Kit (Solarbio Science & Technology Co., Ltd., China). In vitro expression of circRNA encoding secretory uricase HepG2 cells were seeded at a density of 8 × 105 cells per well in a 6-well plate and incubated overnight for the subsequent transfection experiment. The cells were transfected with circRNAUox, and circRNAspUox at different weight ratio using Lipofectamine MessengerMax (Invitrogen). At 24–72 h post-transfection, the cell lysates were collected for Western blotting and enzyme activity assays. Uricase activity assays were measured by Uricase Activity Test Kit. Western blotting Proteins were extracted from HepG2 cells using a radioimmunoprecipitation assay (RIPA) lysis buffer (Beyotime, China). The proteins were then separated by standard sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis and subsequently transferred onto polyvinylidene fluoride (PVDF) membranes. The membranes were blocked for 1 h with non-fat milk in Tris-buffered saline/Tween-20 (TBST) buffer. After blocking, the membranes were incubated overnight at 4 °C with a uricase antibody at a dilution of 1:1000. The following day, the membrane was washed and then incubated for 1 h at room temperature with a secondary antibody at a dilution of 1:5000. After the secondary antibody was washed off, the proteins were detected using the Tanon™ High-sig ECL Western Blotting Substrate (Tanon Science & Technology Co., Ltd., China) and visualised on a chemiluminescent imaging system. Western blotting results were quantified by densitometry using ImageJ, with the values normalised to the control level. Animals Healthy 8 weeks old Balb/c mice and male 8 weeks old C57BL/6 mice were obtained from GuangDong GemPharmatech Co., Ltd. (Guangdong, China). Male 12 weeks old UOX-KO mice were obtained from Changzhou Cavens model animal Co., Ltd. (Ghangzhou, China). Animals were housed in specific pathogen-free conditions at a standard temperature of 22 ± 2 °C and a relative humidity of 55% (45%–70%) in a 12:12-h light-dark cycle. All mice were fed with a standard rodent diet (1010083, Jiangsu Co-operative Bio Co., Ltd., China). Animal welfare was monitored daily for signs of pain or distress, including reduced mobility, piloerection, hunched posture, decreased appetite, and weight loss >20%. Animals showing severe signs of distress were humanely euthanized and excluded from the study. At the end of each experiment, all animals were euthanized by CO₂ inhalation followed by cervical dislocation under isoflurane anesthesia. All efforts were made to minimize suffering. All animal studies were conducted ethically under protocols approved by the Guangzhou Institute of Biomedicine and Health, Chinese Academy of Sciences Institutional Animal Care and Use Committee (IACUC, No. 2024107) in facilities accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC). This study has been reported in accordance with the ARRIVE guidelines. Biodistribution imaging of LNPs Mice were administered a single intravenous (i.v.) injection of Fluc mRNA LNPs at an RNA concentration of 0.75 mg/kg. After 6 h post-injection, mice were injected with D-luciferin at a dose of 150 mg/kg via intraperitoneal (i.p.) injection. Subsequently, luciferase expression was quantified using the IVIS Spectrum system (PerkinElmer, USA). Following sacrifice, the organs (heart, lung, liver, spleen, kidney, and lymph node) were excised, and bioluminescence images of these organs were captured. DOE design and analyse The experimental design and analysis in this study were conducted using Design-Expert software (Version 13.0.1.0, Stat-Ease Inc., MN). A four-factor mixed D-optimization design was utilized to construct a polynomial model for the optimization of the lipid formulation of LMA2-C10 LNPs. The total molar percentage of the four factors, including ionizable lipid, DOPE, cholesterol, and DMG-PEG 2000, was fixed at 100%, with each factor assigned low and high levels as specified in Table S1. The response variables affected by these factors included the size, PDI, EE, and in vivo transfection efficiency. In addition, 5 lack of fit points were included in the design to estimate error. A cubic model was employed to generate 15 different lipid formulations (Table S2), using the original formulation as a control. Polynomial models for each response were proposed following ANOVA analysis, model p-values and lack-of-fit p-values which has been described in detail in the result. The predictive accuracy of the models was evaluated using scatter plots of predicted versus experimental values and the correlation coefficient R2. Finally, the optimization of the best formulation was validated based on the model’s predictions. Cellular distribution of LMA2-C10 LNPs in the liver Mice were sacrificed 6 h after being injected with EGFP mRNA LNPs to investigate the primary cell population in the liver that expresses the EGFP protein using flow cytometry. Single-cell suspensions of liver cells were obtained as previously described66, filtered through a 70 μm filter, and erythrocytes were lysed using red blood cell (RBC) lysis buffer. The cells were stained with the Zombie Aqua™ Fixable Viability Kit for 30 min and then incubated with antibodies included Brilliant Violet 421TM anti-mouse CD45 (dilution 1:100), PE anti-mouse CD31 (dilution 1:20), PE anti-mouse F4/80 (dilution 1:20) for an additional 30 min. Flow cytometric detection was performed using a BD LSR Fortessa cytometer (USA), and data analysis was conducted with FlowJo V10. The FACS gating strategy for analyzing LMA2-C10 LNPs in liver cells was shown in Supplementary Fig. 40. For immunofluorescence, mice were sacrificed 6 h after being injected with EGFP mRNA LNPs and livers were embedded in Tissue-Tek OCT, sectioned into 10-15 µm cryosections, and blocked with 10% normal goat serum in PBS. The sections were incubated overnight with primary antibodies, including anti-αSMA antibody (ab124964, dilution 1:300), anti-Albumin antibody (ab207327, dilution 1:500), anti-Cytokeratin 19 antibody (14965-1-AP, dilution 1:100), anti-CD45 antibody (60287-1, dilution 1:100), and Anti-CD31 antibody (GB113151, dilution 1:200). Subsequently, the sections were washed with PBST and incubated with the corresponding secondary antibody for 1 h at room temperature. After staining the cell nuclei with 4',6-Diamidino-2-phenylindole (DAPI), fluorescent images were acquired using a confocal microscope (LSM800, Carl Zeiss, USA). Acute gouty arthritis model and efficacy study MSU crystals were prepared by alkaline dissolution followed by pH adjustment to 7.2, cooling at 4 °C for 24 h, centrifugation at 2000 × g for 10 min, washing three times with sterile water, drying at 37 °C for 24 h, and sterilization by autoclaving67. Mouse model of acute gout arthritis induced by ankle injection of MSU crystals at a concentration of 50 mg/mL (50 μL). The acute gouty arthritis mouse model was treated daily with allopurinol via oral gavage or administered a single intravenous injection of either RNA@LMA2-C10 LNPs or uricase. Joint swelling was measured using vernier calipers at 1 h, 24 h, and 48 h after MSU crystal injection. Gait dynamics analysis was performed using the DigiGait imaging system (Mouse Specifics) at the 24-h time point following model induction. Mice were allowed to run on motorized transparent treadmill belts with a speed setting of 20 cm/s. Spatiotemporal gait parameters for each paw movement were measured according to gait phase using the DigiGait analysis software (Mouse Specifics). At the 48-h time point following model induction, blood samples were collected from the mice prior to euthanasia. Joint tissues and major organs were subsequently harvested for pathological examination. Hyperuricemia mouse model and efficacy study The UOX-KO mice were administered daily oral doses of allopurinol or received intravenous injections of RNA@ LMA2-C10 LNPs, uricase, or PBS every five days. C57BL/6 mice served as negative controls. Blood samples were collected at predetermined time points for the measurement of serum UA, Scr, and BUN levels. Four days following the final administration, blood was collected prior to euthanasia. Major organs were subsequently harvested for subsequent pathological examination and protein analysis. Pathological evaluation Hind limbs were collected from euthanized gouty arthritis mice and fixed in 4% paraformaldehyde for 24 h. Tissue decalcification was conducted in 14% EDTA (Servicebio, G1105) for 28 days. Joint pathology was assessed by hematoxylin and eosin (H&E) staining to evaluate inflammatory infiltration, synovial damage, and granuloma formation. Kidneys from UOX-KO mice were harvested post-treatment and fixed in 4% paraformaldehyde for 24 h, followed by dehydration, embedding, and sectioning. Histopathological analysis was performed on kidney sections using H&E, Masson’s trichrome, and PAS staining. Renal tubular injury and fibrosis were assessed blinded to the treatment groups. Five non-overlapping fields were selected randomly per section for renal tubular injury scoring and fibrosis assessment. Tubular injury was graded on a scale of 0 to 4, where 0 indicates no injury; 1 represents less than 25% involvement; 2 corresponds to 26%–50% involvement; 3 indicates 51%–75% involvement; and 4 indicates involvement of 76% or more injured tubules. Quantification of Masson’s trichrome/hematoxylin-positive stained areas was performed using ImageJ. Biocompatibility analysis Mice were administered a single intravenous (i.v.) injection of circRNA-LNPs at an RNA concentration of 2.5 mg/kg and 0.75 mg/kg. After 7 days post-injection, blood samples and major organs were harvested from the mice. The serum was isolated and the levels of ALT, AST, ALP, Scr, and BUN were quantified. The harvested organs were fixed in 4% paraformaldehyde and embedded in paraffin Subsequently, the embedded sections were stained with H&E for pathological evaluation. Statistical analysis and reproducibility Data are presented as mean ± Standard Error of the Mean (s.e.m.). Statistical analysis of the experimental results was performed using SPSS 25.0 software. Unpaired two-tailed Student’s t-tests were used for comparisons between two groups. For multiple comparisons, either one-way analysis of variance (ANOVA) with Dunnett’s test or repeated-measures ANOVA with Geisser-Greenhouse correction and Tukey’s post hoc test as applied. A p-value of <0.05, <0.01, <0.001, or <0.0001 was deemed statistically significant and indicated with *, **, ***, or ****, respectively. For immunofluorescence analysis of LMA2-C10 LNPs targeting in different liver cell types, experiments were performed on three mice per group with consistent results. Representative images from one mouse are shown. AKTA purification and agarose gel electrophoresis of circRNA were independently repeated five times with consistent results; representative data from one experiment are shown. RNase R digestion assays were independently repeated three times with consistent results; representative data from one experiment are shown. For ankle joint photography, H&E staining, and gait analysis in the acute gouty arthritis mouse model, experiments were performed on six mice per group with consistent results. Representative results from one mouse are shown. Reporting summary Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article. Supplementary information Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Transparent Peer Review file Supplementary Information Description of Additional Supplementary Files Supplementary Data 1 Reporting Summary Transparent Peer Review file Source data Source Data Source Data