Targeted inhibition of inflammatory liver macrophages with a multifunctional nanoparticle ameliorates metabolic dysfunction–associated steatohepatitis

Abstract

Despite metabolic dysfunction–associated steatohepatitis (MASH) being a significant global health concern, there are few effective ways to prevent or treat MASH. Evidence from preclinical and clinical studies suggests that inflammatory polarization of Kupffer cells (KCs) and infiltration of monocyte-derived macrophages (MDMs) in liver macrophages play an essential role in the initiation and progression of MASH. Herein, we reported a liposomal nanoparticle (NP-OMg) that targets suppressive oligodeoxynucleotide (sODN) and magnesium ions (Mg²⁺) to KCs and MDMs. sODN acted as a potent inhibitor of multiple pattern recognition receptors (PRRs), while Mg²⁺ was chosen for its anti-oxidative and anti-pyroptotic effects, and its capability of coordinating with sODN in NP-OMg. Intravenously administrated NP-OMg was primarily engulfed by both KCs and MDMs in mouse liver. Therapeutically, NP-OMg mitigated hepatic steatosis, inflammation and fibrosis, and restored liver functions in mouse models of metabolic dysfunction–associated steatotic liver disease (MASLD)/MASH. Mechanistically, RNA-sequencing analysis revealed that NP-OMg largely reversed the aberrant gene expression and signaling pathways involved in the pathogenesis of MASLD/MASH. Through comprehensive targeting of inflammatory pathways in liver macrophages, NP-OMg treatment significantly reduced production of inflammatory cytokines/chemokines and decreased the liver recruitment of Ly6c^high inflammatory MDMs. These findings demonstrate the potential of a new liver macrophage-targeted nanomedicine strategy for the treatment of MASH.

Graphical abstract

Liver macrophages-targeted NP-OMg interferes directly or indirectly with multiple PRRs-mediated inflammatory signaling pathways involving TLR9, cGAS-STING-NF-κB, NLRP3 and GSDMD. Therapeutically, NP-OMg ameliorates MASH and restores the liver functions.

Highlights

• •Inflammatory liver macrophages drive progression of MASLD/MASH.
• •A liposomal nanoparticle (NP-OMg) targets suppressive oligodeoxynucleotide (sODN) and magnesium ions to liver macrophages.
• •sODN inhibits multiple pattern recognition receptors (PRRs), while Mg²⁺ exerts anti-oxidative and anti-pyroptotic effects.
• •NP-OMg mitigates hepatic steatosis, inflammation and fibrosis in mouse MASLD/MASH models.

1. Introduction

Metabolic dysfunction–associated steatotic liver disease (MASLD) is the most common liver disease in western countries, which, without effective interventions, will likely progress to steatohepatitis, cirrhosis and even deadly hepatic cellular carcinoma (HCC). Although the exact mechanisms underlying the pathogenesis of MASH are not fully understood, liver macrophages have long been recognized to play a crucial role in the initiation and progression of MASH [1,2]. Kupffer cells (KCs), the liver resident macrophages, vigilantly sense the danger signals encountered in liver through their pattern recognition receptors (PRRs). After binding to damage-associated molecular patterns (DAMPs) or pathogen-associated molecular patterns (PAMPs), the PRRs activate various signaling pathways to produce inflammatory cytokines and chemokines. Of which, tumor necrosis factor-alpha (TNF-α) and interleukin-1 (IL-1) are considered key drivers of steatohepatitis [2,3], while C-C motif chemokine (CCL2) secreted by KCs recruits Ly6c^high inflammatory MDMs to liver, further promoting the disease progression [4,5]. Inflammatory liver macrophages in steatohepatitis can elicit liver fibrosis by enhancing the survival of activated hepatic stellate cells via mediators such as TNF-α and IL-1 [6].

Multiple pattern recognition receptors (PRRs)-mediated inflammatory responses in liver macrophages have been implicated in the development of MASH [[7], [8], [9]]. Oxidized mitochondrial DNA (mtDNA), released from injured hepatocytes, has been identified as a major source of damage-associated molecular patterns (DAMPs) to elicit the PRRs-mediated signaling in liver macrophages [[7], [8], [9]]. Upon binding to toll like receptor 9 (TLR9), mtDNA activates macrophages to produce various inflammatory cytokines and chemokines, promoting MASH progression [7]. Cytosolic mtDNA or double strand DNA (dsDNA) is also recognized and processed by cyclic GMP-AMP synthase (cGAS) to trigger stimulator of interferon genes (STING) signaling [10,11]. Recent clinical evidence indicates that STING-mediated inflammation in liver macrophages contributes to MASH progression in patients [12,13]. There is another cytosolic DNA-sensing mechanism via AIM2 (absent in melanoma 2), which senses cytosolic dsDNA/mtDNA and then forms a caspase-1-activating inflammasome to mediate the maturation of IL-1β and IL-18 [9,14]. Caspase-1 also acts on gasdermin D (GSDMD) to execute pyroptosis and promote secretion of IL-1β and IL-18 [15,16]. GSDMD-driven pyroptosis has been reported to play a role in steatohepatitis by mediating macrophage infiltration [17]. All of these suggest that multiple inflammatory pathways mediated by liver macrophages are critically involved in the pathogenesis of MASH, and thus comprehensive inhibition of these pathways is necessary when we design new therapeutics against MASH.

A suppressive oligodeoxynucleotide (sODN), comprising of four repeats of TTAGGG motif, has been shown to competitively block the binding of cytosolic dsDNA to TLR9, AIM2 or cGAS [[18], [19], [20]]. We therefore hypothesized that sODN may be developed as a useful therapeutic against MASH owing to its ability to suppress multiple PRRs-mediated inflammatory responses. To improve its in vivo bioavailability and targetability to liver macrophages, we employed a phosphatidylserine (PS)-coated liposomal nanocarrier that exhibited excellent targeting of macrophage [21,22]. We screened a series of metal ions for potential synergy with sODN and found that Mg²⁺ exerted its anti-inflammatory effects by reducing mitochondrial oxidative stress and inhibiting GSDMD-mediated pyroptosis. Magnesium deficiency is often associated with MASH patients and low magnesium promotes inflammation and thus exacerbates the disease [23,24]. We herein co-loaded Mg²⁺ with sODNs into the PS-NP (NP-OMg), in which they form a stable coordination complex while releasing sODNs and Mg²⁺ at a low pH encountered in endosome. Utilizing the diet-induced mouse models of MASLD/MASH, we found that intravenous NP-OMg ameliorated steatohepatitis and restored the liver functions. Comprehensive transcriptome profiling and immunobiological assays provided insights into the mode of action of NP-OMg.

2. Materials and methods

2.1. Preparation and characterization of NP-Omg

The lipids 1,2-Distearoyl-sn-glycero-3-phosphocholine,18:0 PC (DSPC), cholesterol, brain l-α-phosphatidylserine (PS), 1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N-[amino (polyethylene glycol)-2000] (DSPE-PEG₂₀₀₀) and DiR were purchased from Avanti Polar Lipids. sODN (ODN TTAGGG) and fluorophore 6-carboxyfluorescein (FAM) labelled sODN were purchased from Tsingke Biotechnology Co., Ltd.

NP-OMg was prepared in two steps using the water-in-oil reverse microemulsion method [25]. We first prepared two identical 2.5 ml of Oil phase 1 and 2 by mixing cyclohexane (Fisher Chemical, C556-4) and Igepal co-520 (Sigma, 238643) in a ratio of 4:1. A mixture of 25 μL brain l-α-phosphatidylserine (PS), 20 μL 1,2-Distearoyl-sn-glycero-3-phosphocholine,18:0 PC (DSPC) and 5 μL cholesterol was add to Oil phase 1. Subsequently, 37.5 μL Na₂HPO₄ (25 mM in ddH₂O, pH = 9.0) and 15 μL of sODN (1 mg/mL in ddH₂O) was added to Oil phase 1. In parallel, 37.5 μL MgCl₂ (2.5 M in ddH₂O, pH = 7.0) and 15 μL of sODN (1 mg/mL in ddH₂O) were added to Oil phase 2, followed by stirring at 1000 rpm for 5 min. The MgP core with single-layer anionic lipid coating was prepared by mixing Oil phase 1 and Oil phase 2 and keep stirring for 30 min. The mixture was then added with 5 ml ethanol and centrifuged at 14,000 × g for 15 min, followed by washing with ethanol three times. The collected MgP cores were dispersed in 1 ml chloroform and centrifuged at 2000 × g for 5 min to remove MgP precipitates without lipid coatings. The supernatants containing a single layer of lipid-coated MgP were further combined with 35.4 μL of a lipid mixture (20 mM, brain PS:DSPC:cholesterol: DSPE-PEG2000 at a molar ratio of 12.8:16:3.3:3.3 in chloroform). Chloroform was subsequently evaporated to generate a lipid film. The bilayer lipid-coated MgP nanoparticles were then prepared by adding 1 mL of PBS (0.1 M, pH 7.4), followed by rehydration through water bath sonication for 5 min and probe sonication at 20 W for an additional 2 min at 70°C.

2.2. Characterization of NP-OMg

The average size and zeta-potential of nanoparticles were measured by Malvern Zetasizer Nano ZS90. Transmission electron microscopy (TEM) measurements were performed on an FEI Tecnai Bio Twin TEM. The concentration of NP-OMg was determined by Nanoparticle Tracking Analysis (NTA) after 1000 times dilution. To determine the loading efficiency, sODN was released from NP-OMg after incubating with 1 M hydrochloric acid and quantitated by Agilent 1100 High-Performance Liquid Chromatography (HPLC). The mobile phases consisted of methanol and an aqueous solution containing 2% (v/v) 1,1,1,3,3,3-hexafluoro-2-propanol (HFIP) and 0.1% (v/v) triethylamine (TEA). The sODN concentration was determined based on the corresponding HPLC peak area using a standard curve generated with free sODN. The encapsulation efficiency was calculated based on the quantified amount of sODN relative to the initial amount used for nanoparticle preparation.

2.3. Assessing metal ions for anti-inflammatory effects

Powder of different metal salts were purchased from Sigma-Aldrich (ZnCl₂, 208086; MgCl₂ 6H₂O, M0250; CuCl₂, 222011; NiCl₂, 339350, CaCl₂, C5080; MgSO₄ 2H₂O, M7634; MnCl₂, 244589). The iBMDM cells (1 × 10⁴) were planted in 96 well plates and cells were cultured in Ca²⁺/Mg²⁺-free DMEM/high-glucose medium (Cytiva, SH320262.01) to minimize interference from basal Ca²⁺ and Mg²⁺, different concentrations of metal ions were added to the plate and incubated for 24 h (0.05 mM, 0.1 mM, 0.5 mM, 1 mM and 2 mM). Then cell viability was determined with MTT assay (3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide).

To determine if metal ions protect the cells from LPS and dsDNA-induced toxicity, 1 × 10⁴ iBMDM cells were planted in 96 well plates and cultured in Ca²⁺/Mg²⁺-free DMEM/high-glucose medium supplemented with 1 mM Ca²⁺. The indicated metal ions were individually added at a final concentration of 1 mM. 100 ng/mL LPS was added to prime the cells for 2 h, followed by polydA:dT (1 μg/mL)-lipofectamine 2000 (Invitrogen, 11668-027) for 4 h. Cell viability was determined with CytoTox 96 Non-radioactive cytotoxicity assay (Promega, G1780).

2.4. Cell culture

The iBMDM murine macrophage cell line was kindly provided by Dr. Hui-Kuan Lin at Duke University. Mouse vascular endothelial cell bEnd.3, murine hepatocyte cell line AML-12 and human embryonic kidney cell 293 TA were obtained from the American Type Culture Collection (ATCC). Cells were maintained in Dulbecco's modified Eagle medium or RPMI-1640 medium (Gibco, 11875-093) supplemented with 10% fetal bovine serum (HyClone, SH30910.03) and 1% penicillin/streptomycin in a humidified atmosphere containing 5% CO₂ at 37°C. AML-12 was maintained in DMEM/F12 medium (Gibco, 11330-032) supplement with 10 μg/mL insulin, 5.5 μg/mL transferrin, 5 ng/mL selenium (Gibco, 51500-056), 40 ng/mL dexamethasone, 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen). Murine Kupffer cells and hepatocytes were isolated from the liver of C57BL/6 mice. Briefly, C57BL/6 mice were perfused with EDTA (0.5 mM, MP Biomedicals LLC, 195173) and collagenase type IV (0.075%, Gibco, 17104-019), then apply gradient centrifugation with Percoll (20% and 50%, GE Healthcare, 10257134) to have higher yield. Isolated Kupffer cells cultured in Dulbecco's modified Eagle medium supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Invitrogen) for following experiment.

2.5. Mitochondrial ROS assessment

To monitor the mitochondrial ROS level, different concentrations (1 mM, 2 mM) of Ca²⁺ or Mg²⁺ ions were added to the medium of iBMDMs planted in 24 well plate and incubated for 24 h. LPS (100 ng/mL) was applied to prime the cells for 2 h, then activated with polydA:dT (1 μg/mL)-lipofectamine 2000 for 2 h. The medium was replaced with 0.5 μM mitoSOX superoxide indicator (Invitrogen, M36008) in DMEM and incubated for 30 min. After washing with PBS and fixing with 4% paraformaldehyde, the cells were counterstained with DAPI (4′,6-diamidino-2-phenylindole) and observed using a fluorescence microscope. ROS levels were judged by measuring red fluorescence signal intensity.

2.6. Extraction of mitochondrial DNA

AML-12 and HEK-293TA cell lines were used to extract mtDNA with Mitochondrial DNA Isolation Kit (Abcam, ab65321) and the amount of mtDNA was quantitated. To oxidize mtDNA (ox-mtDNA), the extracted mtDNAs were treated under ultraviolet-C (UVC) (250 mJ/cm2) for 30 min. For palmitic acid-treated mtDNA (PA-mtDNA), cells were incubated with 200 μM PA (Sigma-Aldrich, P9767) for 24 h, then mtDNA was extracted.

2.7. Priming and activation of macrophages and in vitro treatment with NP-Omg

Kupffer cells (2 × 10⁴) or iBMDM (1 × 10⁴) cells were planted in 96 well plate, first primed with LPS (100 ng/mL) for 2 h in DMEM without Mg²⁺ (Cytiva, SH320262.0 supplemented with 1 mM Ca²⁺), then incubated with MgP, sODN or NP-OMg for 1 h (0.5 mM MgP and 0.33 μM sODN). Subsequently, cells were transfected with ox-mtDNA or PA-mtDNA (2 μg/mL)-lipofectamine 2000 for 6 h. The cells and supernatants were collected for the following biological and biochemical analyses.

2.8. Real-time PCR

The total RNAs were isolated from iBMDM or KCs after treatment using acid guanidinium thiocyanate-phenol-chloroform. For tissue samples, tissues were homogenized in TRIzol reagent (Thermo Fisher) and total RNAs were extracted. One microgram of RNA was transcribed into complementary DNA (cDNA) using high-capacity RNA-to-cDNA kit (Applied biosystems, 4387406), and mRNA expression level was determined using SYBRGreen (Applied biosystems, A25742) with primers listed in Supplementary Table 2.

2.9. Caspase-1 activity assay

The supernatants of iBMDMs or KCs after treatment were analyzed with Caspase-Glo1 inflammasome assay (Promega, G9951) to determine activity of caspase-1 by following the provided instructions.

2.10. ELISA

Cytokine levels in cell culture medium or in blood collected from experimental animals were analyzed with ELISA MaxTM Deluxe Set for mouse cytokines: IL-6 (BioLegend, 431304), TNF-α (BioLegend, 430904) and IL-1β (BioLegend, 432604) following the provider's instructions.

2.11. Western blotting

Cultured KCs and iBMDM cells, and mouse liver tissues were lysed in RIPA buffer containing a protease and phosphatase inhibitor cocktail (Thermo, 1861280). Plasma membrane proteins of cells were extracted using a Minute Plasma Membrane/Protein isolation and cell fractionation kit (Invent Biotechnology Inc. SM-005). To detect proteins in the supernatants of cultured cells, the supernatants were concentrated with methanol and chloroform. Concentrations of collected protein were determined with BCA. Equal amounts of proteins were electrophoresed, and the membranes were visualized by enhanced chemiluminescence (ECL) system (Bio-Rad). A detailed list of the antibodies is provided in Supplementary Table 3.

2.12. In vitro cellular uptake of DiR-NPs

Briefly, 5000 bEnd.3, iBMDM cells, and primary hepatocytes and KCs were seeded in 24-well plates with poly-lysine-coated coverslips for 24 h. Then, the media were replaced with the complete medium containing DiR labelled PS-coated NP-MgP or DOTAP-coated NP-MgP and cultured for another 0.5 h. Cells were then washed twice with cold PBS and fixed with 4% formaldehyde for 15 min. After 1% Triton X-100 treatment for 15 min, the cells were stained with Alexa Fluor 488 phalloidin (Thermo Fisher) for 30 min and counterstained with 1 μg/mL DAPI for 5 min. The cellular uptake of NPs was observed by fluorescence microscope.

2.13. MASLD and MASH mouse models

All animal experiments complied with all relevant ethical regulations for animal testing and research and were performed with approval by the Institutional Animal Care and Use Committee (IACUC) at the Wake Forest University School of Medicine. Male and female C57BL/6 mice were purchased from Charles River Laboratories International, Inc. STING-knockout mice (B6(Cg)-Sting1tm1.2Camb/J) and GSDMD-knockout mice (C57BL/6J-Gsdmdem1Vnce/J) were purchased from the Jackson Laboratory and homozygous breeding in the animal facility. We employed two well-established diet-induced mouse models including:

The MCD model: randomly assigned 6 weeks old female C57/BL6 mice were fed ad libitum a MCD diet (Methionine- and Choline-Deficient L-Amino Acid Diet, Research Diets, A02082002BR) and control mice were fed ad libitum chow diet with access to normal drinking water. The composition of MCD diet is listed in Supplementary Table 1A. For the methionine- and choline-deficient (MCD) model, 6 weeks old female C57/BL6 mice were fed a chow diet (CD) or a MCD diet for 3 weeks and then randomly grouped: (1) chow diet (CD); (2) MCD with PBS; (3) MCD with liposomal MgP (i.v. 6.25 mg/kg Mg²⁺; twice/wk); (4) MCD with free sODN (i.p. 2.0 mg/kg, twice/wk); (5) MCD with NP-OMg (i.v. sODN 2.0 mg/kg sODN and 6.25 mg/kg Mg²⁺; twice/wk). The mice were continuously treated for 4 wks and then euthanized on week 8.

The WD (western-style diet) model: randomly assigned 8 weeks old male C57/BL6 mice were fed ad libitum a high-fat high-fructose diet (Rodent Diet With 40 kcal% Fat (Mostly Palm Oil), 20 kcal% Fructose and 2% Cholesterol, Research Diet, D09100310) with access to drinking water supplemented with a high fructose (23.1 g/L, thermos scientific, 161350010) and glucose (18.9 g/L, Sigma, G8270). Control mice were fed ad libitum chow diet with access to normal drinking water. STING^−/− mice (B6(Cg)-Sting1tm1.2Camb/J) and GSDMD^−/− mice (C57BL/6J-Gsdmdem1Vnce/J) were also fed ad libitum a high-fat high-fructose diet with access to drinking water supplemented with a high fructose (23.1 g/L) and glucose (18.9 g/L). The composition of WD diet is listed in Supplementary Table 1B. For the western-style diet (WD)-induced MASLD model, 8 weeks old male C57/BL6 mice were fed a chow diet or a high-fat high-fructose diet with access to drinking water supplemented with a high fructose and glucose for 7 weeks. The mice were then randomly grouped as follows: (1) CD control; (2) WD with PBS; (3) WD with sODN (i.p. 2.0 mg/kg, twice/wk); (4) WD with NP-OMg (i.v. sODN 2.0 mg/kg sODN and 6.25 mg/kg Mg²⁺; twice/wk). The mice were continuously treated for 7 wks and then euthanized on week 14.

Our choices of gender and study endpoints in these two models were based on previously published studies, in which female C57/BL6 mice fed a MCD diet for 8 wks developed MASH [8], while male C57/BL6 mice developed MASLD after 12 wk feeding of a WD diet [7,12].

2.14. Bulk RNAseq analysis

For bulk RNAseq analysis of WD-induced MASLD model, total RNA was extracted from liver tissue with acid guanidinium thiocyanate-phenol-chloroform. RNA quality was first assessed with NanoDrop for purity and concentration and then by Novogene's standard QC pipeline prior to library construction. All sequencing was conducted by Novogene Corporation Inc. Libraries were prepared using poly(A) enrichment to capture eukaryotic mRNA, followed by fragmentation, cDNA synthesis, adapter ligation, and PCR amplification. The qualified libraries were sequenced on an Illumina NovaSeq X Plus platform using paired-end 150 bp reads (PE150). Clean reads were aligned to the reference genome (Mus musculus, GRCm39/mm39). Gene expression levels were quantified as transcripts per million (TPM) or fragments per kilobase of transcript per million mapped reads (FPKM), and differential expression analysis was performed using NovoMagic cloud platform, with log2FC > 1 and adjusted p-value < 0.05 considered significant. Functional enrichment (GO and KEGG) was carried out using NovoMagic cloud platform.

2.15. Biochemical analysis

The serum samples collected at the end of animal experiments were subjected to biochemical analysis. Briefly, the blood was collected in tubes pre-treated with heparin (Thermo scientific, A16198) and centrifuge at 2000×g for 10 min to separate serum. The activities of aspartate transaminase (AST) and alanine transaminase (ALT) were analyzed with EnzyChrom™ Aspartate Transaminase Assay Kit (BioAssay Systems, EASTER-100) and EnzyChrom™ Alanine Transaminase Assay Kit (BioAssay Systems, EALT-100) following the provided instructions, respectively.

Liver tissues were collected from experimental animals. Approximately 0.2 g liver tissues were homogenized in 1 mL PBS and then centrifuged at 4000×g for 10 min. The supernatants were collected and analyzed with Triglyceride Colorimetric Assay kit (Cayman Chemical, 10010303) and Cholesterol Colorimetric Assay kit (LSBio Lifespan Biosciences, Inc. LS-K304-100) following the provider's instructions. The concentration of cholesterol and triglyceride in collected liver tissues were normalized by liver weights.

2.16. Flow cytometry analysis

Liver tissues were digested with collagenase type IV (0.5 mg/mL) and DNase I (5 U/mL) in 37°C. After further grinding, the cells were first centrifuged in a slow speed centrifuge (50×g) then were further collected with two high speed centrifuges (500×g), collected cells were incubated with RBC lysis buffer and blocked with anti-mouse CD16/CD32 (Biolegend, 101302). Flow cytometry was performed on a BD Canto II flow cytometer, and data were collected by FACSDiva software (6.1.3), and analyzed using the FlowJo software (10.1, BD Biosciences). The live/dead cells were distinguished with Zombie Aqua Dye (Biolegend, 77143). Subsequently, cells were incubated with fluorescently labelled antibodies listed in Supplementary Table 4. Gates and quadrants were set based on isotype control staining, and gating strategies for stratification of cellular compositions in liver tissues are presented in Supplementary Fig. S15-17.

2.17. In vivo bio-distribution of NP-Omg

Healthy mice (fed a chow diet) and MASH mice (fed a MCD diet for 8 weeks) were i.v. administered with DiR labelled NP-MgP (DiR-NP), the bio-distribution of labelled nanoparticle was longitudinally monitored at 2, 24 and 48 h post-injection using the IVIS Lumina system. Randomly selected animals were euthanized at 24 and 48 h (n = 3), and major organs (lymph node, spleen, kidney, intestine, lung, heart and liver) were harvested and imaged ex vivo using IVIS. After imaging at 24 and 48 h, the mice were euthanized, and liver tissues were collected then digested with collagenase type IV (0.5 mg/mL) and DNase I (5 U/mL) in 37°C. After further grinding, the cells were first collected in a slow speed centrifuge (50×g) and considered as “non-CD45 cells”, then cells were further collected with two high speed centrifuges (500×g), the DiR⁺ cells in different cell populations were assessed by flow cytometry.

We also packed FAM-labelled sODN (FAM-sODN) into NP-MgP to investigate the distribution of sODN, MASH mice (fed a MCD diet for 8 weeks) were i.v. administered with NP-FAM-sODN or i.p. administered with free sODN (30 μg/mice). At 24 h after administration, the mice were euthanized, and liver tissues were collected and processed for flow cytometry analysis. The mean fluorescence intensity (ΔMFI) values of FAM⁺ cells in different cell populations were calculated. Antibodies involved were listed in Supplementary Table 4. Flow cytometry was performed on a BD Canto II flow cytometer and analyzed using the FlowJo software (BD Biosciences).

2.18. Histology and immunohistochemistry

H&E staining was performed on paraffin-embedded tissue sections of heart, lung, liver, kidney and intestine from healthy or diet-induced MASLD/MASH mice. Liver tissue sections were also stained with Sirius Red for collagen. Pathological analyses were conducted by Dr. Jingmei Li. Frozen sections of liver tissues obtained from healthy or MASLD/MASH mice in the above in vivo biodistribution studies of DiR-NP were immunostained with anti-mouse F4/80 antibody (1:200, BioLend) followed by goat anti-rat Cy2 (1:500, Jackson Immuno). The sections were then counterstained with DAPI and observed under fluorescence microscope to correlate DiR signals with F4/80⁺ macrophages.

2.19. Bioinformatics clinical data analysis

Transcriptome profile of liver tissues from MASH patients was imported from the Gene Expression Omnibus (GEO) repository dataset (GSE193084) and processed with GEO2R. The data was analyzed with GSDMD (Gene ID: 79792) and STING (Gene ID: 340061). The expression profiles of GSDMD and STING were classified according to the MASH score (from 1 to 8) and the Fibrosis stage (from 0 to 3).

2.20. Statistics and reproducibility

Data are presented as mean ± SD and analyzed using Prism8 (GraphPad Software). P < 0.05 was considered statistically significant. Differences between two groups were analyzed using an unpaired two-tailed t-test or Fisher's exact test. For assessing the differences among multiple groups, a one-way ANOVA (Analysis of Variance) with Tukey's test or Dunnett's multiple-comparison test was employed.

3. Results

3.1. Liver macrophages in MASLD are enriched for pro-inflammatory responses

We utilized a well-established MASLD mouse model that recapitulates human MASLD pathogenesis [26,27]. C57BL/6 mice were fed a western-style diet (WD) containing high-fat and high-cholesterol and supplemented with high fructose in drinking water. After 14 wks, mice developed characteristics of MASLD with hepatic steatosis (Fig. 1A), and elevated levels of hepatic triglyceride (TG) and cholesterol (CHL) and serum aspartate aminotransferase (AST) and alanine aminotransferase (ALT) (Fig. 1B). As compared to chow diet (CD)-fed mice, increased levels of collagen fibers were clearly visualized in liver tissues, albeit less advanced fibrosis developed at this stage of MASLD mice (Fig. 1A). To decipher transcriptome signatures in MASLD, we undertook comprehensive bulk RNA-sequencing on liver tissues. A total of 1510 genes exhibited significant upregulation, while 1322 genes were downregulated in MASLD mice, compared to CD mice (Fig. 1C). Gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of biological pathways identified a significant enrichment of genes associated with lipid metabolism, fibrosis, cell adhesion, inflammatory and immune responses (Fig. 1D). Of note, various inflammatory and innate immune activation pathways including PRR signaling pathway (Cd36, Tlrs, Clec7a), activation of innate immune response (Cd86, Cd40, Adgre1, Slc15a3), macrophage cytokine production and cytokine-mediated signaling pathway (Eda2r, Pparg, Tnf, Tnfsf15, Il1), pyroptosis (Map3k, Gbp2b, Naip6, Nlrp3) and chemotaxis (Cxcl9, Cxcl10, Ccl2, Ccl3, Ccl4, Ccl5) were significantly upregulated. Macrophages are the most abundant innate immune cells in the liver. To corroborate the inflammatory polarization of liver macrophages, we isolated liver macrophages (F4/80⁺) from MASLD mice and conducted real-time PCR, which revealed significantly amplified expression of Il1b, Tnf, Il6, Il12b, Il18b, Ccl2 and Cxcl10 (Fig. 1E). Chemokines such as Ccl2, Ccl5 and Cxcl10 are responsible for recruiting monocyte-derived inflammatory macrophages, which play a critical role in promoting steatohepatitis and liver fibrosis [1,4,5]. Indeed, flow cytometry revealed a marked increase in the number of CD11b⁺ Ly6c^high MDMs while a reduction in the number of KCs (CD11b⁺ F4/80^high), consequently a significant decrease in the ratio of KCs/inflammatory MDMs in the livers of MASLD mice (Fig. 1F), indicating the inflammatory polarization of liver macrophages. Moreover, ELISA assays detected drastically elevated levels of inflammatory cytokines TNF-α, IL-6 and IL-1β in the blood of MASLD mice (Fig. 1G).

Fig. 1.

The activation of proinflammatory signaling and innate immune responses, and inflammatory polarization of liver macrophages are evidence in WD-induced MASLD mice. C57/BL6 mice were fed WD or CD for 14 wks. (A) H&E and Sirius Red staining assessing hepatic steatosis and fibrosis (Scale bar = 50 μm). (B) Measurements of levels of triglyceride and cholesterol in livers and serum levels of AST and ALT in WD- and CD-fed mice. (C) Bulk RNA-seq revealing differential expressions of selected marker genes associated with inflammation, fibrosis and lipid metabolism and Gene Ontology terms and Kyoto Encyclopedia of Genes and Genomes (D) pathways significantly enriched for upregulated genes in livers of WD versus CD mice (n = 3/group). Adjusted p-value < 0.05 is considered significant. (E) Realtime PCR for mRNA expression levels of inflammatory cytokines and chemokines in liver macrophages (F4/80⁺) isolated from WD-fed mice compared with CD-fed mice (n = 3). (F) Flow cytometry analysis of liver macrophage populations of inflammatory monocytes-derived macrophages (MDM, CD45⁺CD11b⁺Ly6c^high) and Kupffer cells (CD45⁺CD11b⁺F4/80^high), showing a significant difference in the ratio of KC to inflammatory MDM between CD and WD mice (n = 3). (G) Serum levels of TNF-α, IL-6 and IL-1β measured by ELISA (n = 3). Data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001.

The aberrant inflammatory gene expression and signaling pathways observed in this study are in good agreement with previous studies in the MASLD mouse model [4,5,8,28]. Importantly, these genomic abnormalities are highly relevant to human MASLD, as reported previously in comparative studies [13,17,29,30]. Together, these data demonstrate that liver macrophages are enriched for proinflammatory and profibrotic responses critically involved in pathogenesis of steatohepatitis through the activation of pro-inflammatory and pro-fibrotic pathways. Therefore, targeted inhibition of inflammatory signaling in liver macrophages may be a rational therapeutic strategy for MASLD treatment.

3.2. Mg²⁺ coordinated to sODN in nanoparticle augments anti-inflammatory effects

Despite sODN being a potent blocker of multiple PRRs [[18], [19], [20]], its ability to inhibit PRR-mediated inflammatory responses in liver macrophages and further prevent the progression of MASLD/MASH remains untested. To enable liver macrophages-targeted delivery of sODN, we employed a phosphatidylserine (PS)-coated liposomal nanocarrier (PS-NP) that has previously been established by us [21,22]. The outer layer-exposed PS is actively recognized by scavenger receptors on macrophages and subsequently internalized into the cells. We have also shown that such divalent metal ions as Ca²⁺ improve the loading efficiency of oligonucleotides by forming the stable coordination complex in the NP, which releases nucleotides in response to an acidic pH in endosome [21,22].

Searching for a metal ion that can coordinate with sODN and potentially enhance its anti-inflammatory effects, we screened a series of divalent metals including Ni²⁺, Mn²⁺, Mg²⁺, Cu²⁺, Zn²⁺ and Ca²⁺. Among them, Ca²⁺ and Mg²⁺ were found to cause a minimum toxicity to mouse immortalized bone marrow-derived macrophages (iBMDMs) at a concentration up to 2 mM (Fig. 2A). To directly compare the effects of individual divalent metal ions without interference from basal Ca²⁺ and Mg²⁺, these experiments were performed in Ca²⁺/Mg²⁺-free DMEM with the indicated ions added back at defined concentrations. Both Ca²⁺ and Mg²⁺ ions form stable complex with sODN, but only Mg²⁺ showed protective effects against lipopolysaccharide (LPS)-primed, dsDNA polydA:dT-induced cell toxicity (Fig. 2B). Mechanistically, MitoSOX staining of mitochondrial reactive oxygen species (ROS) showed a marked decrease in the ROS level of LPS and dsDNA-treated iBMDMs when Mg²⁺ but not Ca²⁺ was added into culture medium (Mg²⁺- and Ca²⁺-free; Fig. 2C). In a parallel study, iBMDMs were treated with LPS followed by oxidized mtDNA (ox-mtDNA) in a medium containing 0 or 2 mM Mg²⁺. Western blotting detected a lower level of membrane GSDMD-N terminal fragments (NT) in the cells cultured with Mg²⁺ than those without Mg²⁺ (Fig. 2D), and a further decrease in GSDMD-NT in the cells co-cultured with magnesium phosphate-loaded PS-NPs (MgP, 2 mM; Fig. 2D). It is known that GSDMD is cleaved by inflammatory caspases to release GSDMD-NT in response to a pyroptotic stimulus such as dsDNA or ox-mtDNA [31]. GSDMD-NT then migrates to plasma membrane and oligomerizes to form pores, through which inflammatory cytokines IL-1β and IL-18 are released. In a recent study, Wang et al. reported that GSDMD-NT assembly in plasma membrane required Ca²⁺ influx, which could be antagonized with increased intracellular [Mg²⁺] [32]. Together, these data indicate that Mg²⁺ exerts its anti-inflammatory effects via inhibiting both mitochondrial ROS production and GSDMD-mediated inflammatory responses.

Fig. 2.

Incorporation of Mg²⁺ into NP enhances anti-inflammatory effects. (A) After incubating mouse iBMDMs cultured in Ca²⁺/Mg²⁺-free DMEM with various divalent metal ions (Ni²⁺, Mn²⁺, Mg²⁺, Cu²⁺, Zn²⁺ and Ca²⁺) at different concentrations for 24 h, heatmap showing cell viability by MTT assay. (B) iBMDM cells in Ca²⁺/Mg²⁺-free DMEM supplemented with 1 mM Ca²⁺ were primed with 100 ng/mL LPS for 2 h, then activated with dsDNA (1 μg/mL in lipofectamine) for 4 h in the presence of different metal ions at 1 mM concentration. Cell viability determined with LDH assay. (C) iBMDMs were treated with LPS plus dsDNA in the presence of different concentrations of Ca²⁺ or Mg²⁺ ions (1 mM or 2 mM). MitoSOX staining (red) monitors the levels of mitochondrial ROS in the cells. Scale bar 50 μm. (D) Plasma membrane proteins were extracted from iBMDMs after treatment with LPS plus ox-mtDNA in the presence of Mg²⁺ (2 mM) or PS-NP-loaded magnesium phosphate (MgP, 2 mM Mg²⁺). Western blotting detects the level of GSDMD-N-Terminal (GSDMD-NT) with transferrin receptor 1 (TFR1) as a housekeeping protein. (E, F) Size distribution of NP-OMg assessed by dynamic light scattering (DLS) and transmission electronic microscope (TEM). Scale bar 100 nm. (G) The release kinetics of sODN from NP-OMg under pH 7.4, 6.5 or 5.0. Data are shown as mean ± SD of n = 3 biologically independent experiments. *p < 0.05, ****p < 0.0001 by One-way ANOVA.

These findings prompted us to develop a novel liposomal NP-OMg that targets sODN and Mg²⁺ to liver macrophages. NP-OMg was prepared using the water-in-oil reverse microemulsion method to load the complex of sODN-magnesium phosphate [25]. NP-OMg had an average diameter of 169.6 ± 2.52 nm (Fig. 2E and F) with concentration of 1.66 × 10¹¹ particles/mL (Supplementary Fig. S1), and a zeta potential of −19.2 ± 1.56 mV. The encapsulation efficiency of sODN was ∼51%, and the concentration of sODN and Mg²⁺ was 18.8 μg/mg lipid and 61.1 μg/mg respectively. NP-OMg was stable over time at pH 7.4 while releasing sODN at an acidic pH (Fig. 2G).

To assess its macrophage-targeting specificity, 1,1′-dioctadecyl-3,3,3′,'3′-tetramethylindotricarbocyanine iodide (DiR)-labelled NP-OMg (DiR-NP) was incubated with various types of cells. After 30 min, robust DiR signals were observed within iBMDM and KCs. In contrast, little was seen in hepatocytes or vascular endothelial cells (Fig. 3A). The macrophage uptake was dampened with pre-incubation of anti-PS antibody for 2 h (Supplementary Fig. S2A). Moreover, we substituted PS with DOTAP (1,2-dioleoyl-3-trimethylammonium-propane), a cationic lipid, on the outer layer of liposome. Clearly, a weaker uptake of DiR-DOTAP-NPs by iBMDMs or KCs was observed (Supplementary Fig. S2B). These data demonstrate that NP-OMg effectively targets both liver resident KCs and bone marrow-derived macrophages.

Fig. 3.

NP-OMg potently suppresses inflammatory responses in mtDNA-stimulated iBMDM and Kupffer cells. (A) After incubating DiR labelled MgP-nanoparticles (DiR-NPs) vascular endothelial cells (ECs), primary hepatocytes, iBMDMs or primary KCs for 30 min, fluorescence microscopy imaging showing DiR-NPs engulfed by iBMDM and KC but not EC or hepatocyte. Top row: merged images of DiR-NP (red), Phalloidin (green), dapi (blue). Bottom row: DiR-NP (red). (B) iBMDMs planted in 96 well plate were primed with LPS (100 ng/mL) for 2 h in Mg²⁺-free DMEM, then incubated with PS-NP-loaded MgP, sODN or NP-OMg (at indicated doses) for 1 h before the cells were transfected with oxidized mtDNA (ox-mtDNA; 2 μg/mL) by lipofectamine. Six hours later, cell viability determined by LDH assay (n = 3). The dose of NP-OMg containing Mg²⁺ 0.5 mM and sODN 0.33 μM was applied in the following experiments. (C) Western blotting revealing the levels of phosphorylated phos-p65 in whole cell lysates and cleaved caspase-1 and mature IL-1β in supernatants of iBMDMs under indicated treatment. β-actin used as a housekeeping protein and plots showing quantitative data of the level of phos-p65 (n = 3, D) and the activity of caspase-1 (n = 5). ELISA assay detecting serum levels of pro-inflammatory cytokines TNF-α, IL-6 and IL-1β (n = 4). (E) The above experiments were repeated in KCs stimulated with LPS plus palmitate (PA)-induced ox-mtDNA. Cell viability by LDH assay, caspase-1 activity, serum levels of TNF α, IL-1β and IL-6. Data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

To study anti-inflammatory effects of NP-OMg, we applied various concentrations of nanoparticle MgP, free sODN or NP-OMg to iBMDMs that were primed with LPS followed by ox-mtDNA activation. As shown in Fig. 3B, MgP at a higher (1 mM) concentration and free sODN at all the tested concentrations were found to have some protective effects against cell death; NP-OMg was clearly superior to sODN and MgP, further improving the survival of cells in a dose-dependent manner. At a concentration containing 0.5 mM Mg²⁺ and 0.33 mM sODN, NP-OMg essentially rescued the cells to the same level of viability as in the PBS control cells (Fig. 3B). We thus used this concentration of NP-OMg in our following studies. Incubation of NP-OMg with iBMDMs treated with LPS and ox-mtDNA resulted in a marked decrease in the phosphorylation state of p65 in whole cell lysates, the levels of mature IL-1β and activated caspase-1 in supernatants (Fig. 3C), and the mRNA and protein levels of proinflammatory cytokines TNF-α, IL-1β and IL-6 (Fig. 3D; Supplementary Fig. S3A–C). The above experiments were repeated with KCs that were isolated from C57BL/6 mice treated with LPS and palmitate-mtDNA (PA-mtDNA). Consistently, compared to free sODN or MgP, NP-OMg significantly improved protection of KCs from LPS and palmitate-mtDNA-induced cytotoxicity, inhibited the activity of caspase-1, and reduced the mRNA and protein levels of inflammatory cytokines of TNF-α, IL-1β and IL-6 (Fig. 3E; Supplementary Fig. S3D–H). The intracellular pro-forms of IL-1β and caspase-1 showed an opposite trend to their extracellular cleaved counterparts (Supplementary Fig. S4). Collectively, these data support our endeavor to develop nano-encapsulated sODN-Mg for targeted suppression of inflammatory liver macrophages.

3.3. Intravenous NP-OMgs are primarily engulfed by liver macrophages

Having demonstrated its excellent macrophage-targeting specificity in vitro, we then studied in vivo pharmacokinetics and biodistribution of NP-OMg. We injected DiR-NPs (100 μl) via a tail vein into healthy mice or MASH mice and conducted fluorescence imaging over time using an in vivo imaging system (IVIS). In both cases, most DiR signals were captured from abdominal space, which sustained for at least 48 h (Fig. 4A and B; Supplementary Fig. S5A and B; Supplementary Fig. S6). Ex vivo IVIS imaging of excised organs and tissues confirmed that the liver was the predominant site of DiR-NP accumulation, accounting for approximately 68% of the total detected fluorescence signal at 24 h. A secondary signal was also detected in the lung, accounting for approximately 17% of the total signal (Fig. 4C and D; Supplementary Fig. S5C and D). Immunohistochemical staining of liver tissue sections showed that DiR-NPs were clearly identified within F4/80⁺ liver macrophages in both healthy and MASH mice (Fig. 4E). Consistently, flow cytometry analysis of cell type-based uptake revealed that DiR-NPs were primarily engulfed by liver macrophages including KCs and MDMs, while little was found in CD45⁻ hepatocytes or other immune cells such as T cells and NK cells in liver tissues (Fig. 4F; Supplementary Fig. S5E). The gating strategy for distribution of NP-OMg among cell populations of mouse liver has been shown in Supplementary Fig. S15 , after CD45 ± cells were defined, CD45⁺ cells including T cells (CD45⁺ CD3⁺), dendritic cells (CD45⁺ CD11c⁺), natural killer cells (CD45⁺ CD49b⁺), inflammatory monocytes-derived macrophages (CD45⁺CD11b^hiLy6c^hi) and Kupffer cells (CD45⁺CD11b⁺F4/80^hi) were defined to determine the distribution of NP-OMg. Intriguingly, DiR-NPs were also found in another type of phagocytes, dendritic cell (DC), despite its small population in liver (Fig. 4F). Furthermore, we compared in vivo biodistribution of fluorophore 6-carboxyfluorescein (6-FAM)-labelled sODN (FAM-sODN) with NP-loaded FAM-sODN in mouse liver. As expected, a significantly higher uptake of NP-FAM-sODN than free FAM-sODN by KCs and MDMs was observed by flow cytometry (Fig. 4G). These results demonstrate that intravenously administered NP-OMg preferentially accumulates in the liver and is predominantly taken up by resident KCs and monocyte-derived macrophages within the hepatic compartment.

Fig. 4.

Intravenously administered NP-OMg targets liver macrophages. DiR labelled MgP-nanoparticles (DiR-NPs) administered i.v into healthy mice and MASH mice. (A) IVIS imaging of in vivo biodistribution of DiR-NP in MASH mice (n = 3) over time, and signal intensity from abdomen at each time point is quantitated (B). (C, D) Ex vivo optical imaging of major organs (lymph node, spleen, kidney, intestine, lung, heart and liver) at 24h and 48h, and corresponding signal intensity from each organ at 24h plotted in the pie chart (n = 3). (E) After imaging, liver tissues (n = 3/group) were sectioned and co-stained with anti-F4/80 antibody, representative fluorescence microscopic images revealing DiR signals (red) merged with F4/80+ macrophages (green) and nuclei (dapi, blue). (F) Flow cytometry analysis of cell type-based uptake of DiR-NP in livers of MASH mice at 24 h and 48 h (n = 3/time). (G) Flow cytometry comparison of cell type-based uptake of fluorophore 6-carboxyfluorescein (FAM)-labelled sODN (FAM-sODN) and FAM-sODN-loaded NPs (NP-FAM-sODN) in livers of MASH mice (n = 3/group). Data are presented as mean ± SD; ***P < 0.001, ****P < 0.0001. n.s., not significant.

3.4. NP-OMg potently suppresses inflammatory signaling and protects against MASLD development and HCC progression

We next investigated if intravenous NP-OMg could inhibit liver macrophages-mediated inflammatory signaling and effectively treat steatohepatitis. The above high fat and high fructose-induced MASLD mouse model was tested here. After feeding for 7 wks to induce MASLD, we started to treat the mice with either free sODN (2.0 mg/kg, i.p. twice a week) or NP-OMg (2.0 mg/kg sODN and 6.25 mg/kg Mg²⁺, i.v. twice a week; Fig. 5A, Supplementary Fig. S7A). At week 14, the mice were sacrificed, and livers were dissected. Compared to the non-treated WD-fed mice, there was a significant reduction in the livers’ weights in NP-OMg-treated mice, while a slight decrease was recorded in the mice treated with free sODN (Fig. 5B and C); consistently, NP-OMg treatment led to a significant decrease in the levels of hepatic triglyceride (TG) and cholesterol (CHL; Fig. 5D). Additionally, NP-OMg markedly reduced hepatic enzymes AST and ALT (Fig. 5D), and serum levels of inflammatory cytokines, TNF-α, IL-6 and IL-1β (Fig. 5E). Furthermore, histological analysis (Dr. Jingmei Lin), by adopting the clinical MASLD activity score system [33], showed that NP-OMg drastically improved the score (1.3 ± 0.5) as compared to the WD group (6.7 ± 0.6; p < 0.001; Fig. 5F). Sirius Red staining also revealed a reduced level of collagen fibers in the NP-OMg-treated livers (Fig. 5G). Of note, the WD mice at wk 7 when the treatment was initiated were histologically confirmed to develop MASLD and had a MASLD activity score of 2.5 (Fig. 5G and F).

Fig. 5.

NP-OMg protects against WD-induced steatohepatitis. (A) Eight-week-old male C57/BL6 mice were fed ad libitum a high-fat high-fructose diet with access to drinking water supplemented with high fructose and glucose (WD). Control mice were fed ad libitum chow diet (CD) with access to normal drinking water. After feeding for 7 weeks, mice were grouped (n = 5/group) and treated as follows: (I) Chow diet; (II) WD diet, (III) WD with sODN (2.0 mg/kg, i.p twice/wk); (IV) WD with NP-OMg (2.0 mg/kg sODN, 6.25 mg/kg Mg²⁺, i.v twice/wk). At week 14, mice were euthanized, and the livers harvested and weighed (B-C). (D) The levels of triglyceride and cholesterol determined in liver tissues. ELISA assay revealing serum levels of liver enzymes AST and ALT, and (E) pro-inflammatory cytokines TNF-α, IL-6 and Il-1β. Histological evaluation of MASLD activity score (F) based on H&E staining (G), and collagen fibers by Sirius Red staining (G) for CD-fed mice, WD-fed mice at wk 7 and WD-fed mice at wk 14, sODN treatment or NP-OMg treatment. Data are presented as mean ± SD; *P < 0.05, **P < 0.01, and ****P < 0.0001.

To gain mechanistic insights into its therapeutic effects, we conducted the bulk RNA-seq on liver tissues to investigate transcriptomic changes elicited by NP-OMg. RNA-seq data analysis revealed significant upregulation in 275 genes and downregulation in 597 genes in the NP-OMg-treated livers vs WD livers. Of which, expression of genes associated with PRRs (Tlrs, Clec7a, Nlrp3 and Sting1), inflammatory cytokines and chemokines (Tnf, Tnfrsf, Il-1, Cd86, Ccl2, Cxcl10, Ccl7, Ccl22, Cxcl2), lipogenesis (Cd36, Pde4d, Lipg, Elovl7, Acot9 and Eda2r) and liver fibrosis (Anxa2, Col1a1, Col3a1, Timp1 and Fgf21) was significantly decreased, whereas genes linked to lipolysis (Tff3, Saa4, Apom4, Hess6 and Irf2bpl) and fibrinolysis (Serpine1, Serpinf2 and Dkk4) were significantly upregulated (Fig. 6A). Notably, GO and KEGG pathway analyses revealed that various pathways related to inflammatory and innate immune activation, lipogenesis and fibrosis, which were found to be activated in WD-induced MASLD (Fig. 1D), were significantly suppressed by NP-OMg treatment (Fig. 6B). Additionally, important inflammatory signaling pathways such as canonical NF-κB signal transduction, TNF superfamily cytokine production, TLR signaling, NOD-like receptor signaling, MAPK signaling and PI3K-Akt signaling were significantly inhibited by NP-OMg (Fig. 6B). Most strikingly, when we plotted transcript changes of individual genes in response to NP-OMg (vs. WD-MASLD) in comparison with the changes of the same genes induced in WD-MASLD (vs. CD control) mice, we identified a strong inverse correlation between them (Fig. 6C; Supplementary Fig. S8), indicating that NP-OMg treatment effectively reverses the aberrant gene expression involved in the pathogenesis of MASLD. Given the emerging role of TREM2⁺ disease-associated macrophages in MASH, we further examined the expression of a panel of TREM2⁺ macrophage-associated genes in the bulk RNA-seq dataset (Supplementary Fig. S9). NP-OMg treatment reduced the expression of multiple genes associated with this macrophage state, including Trem2, Gpnmb, Lgals3, Lpl, Spp1, Plin2, and Tyrobp, suggesting attenuation of the TREM2⁺ macrophage-associated transcriptional program.

Fig. 6.

NP-OMg treatment reverses aberrant gene expression involved in the pathogenesis of MASLD and inhibits hepatic accumulation of inflammatory MDMs. (A) Clustering heatmap generated from RNA-seq displays differential expressions of selected marker genes in livers of WD vs WD with NP-OMg (n = 3/group). (B) Dot plots showing Gene Ontology terms and Kyoto Encyclopedia of Genes and Genomes pathways significantly enriched for downregulated genes in NP-OMg-treated WD vs. WD mice. (C) Scatter plot comparison of individual gene expression changes (log2 FoldChange) between NP-OMg vs. WD and WD vs. CD showing a significant negative correlation (Pearson's R = −0.57, p < 2.2e-16). Points are colored to classify genes with opposite (significant in both groups and with opposite directions; red, n = 496), partial opposite (significant in one condition; orange, n = 2639), or shared (significant in both condition with same directions; blue, n = 23) regulation. Grey dots represent genes not meeting classification thresholds. (D) Flow cytometry quantifies liver inflammatory monocytes-derived macrophages and Kupffer cells in CD45⁺ cells (n = 5). (E) Representative flow plots of inflammatory MDMs (CD45⁺CD11b⁺Ly6c^high) and KCs (CD45⁺CD11b⁺F4/80^high) in livers of mice under indicated treatment. (F) Western blotting of liver tissues expressing phos-p65, p65, phos-TBK1, TBK1, GSDMD, and GSDMD-N terminal, and (G) quantitative data of phos-p65/p65, phos-TBK1/TBK1 (n = 3). Data are presented as mean ± SD; **P < 0.01, ***P < 0.001, and ****P < 0.0001.

Since inflammatory polarization of liver macrophages and infiltration of inflammatory MDMs are highly associated with the progression of MASH [1,4,5], we conducted flow cytometry analysis of liver macrophage populations post NP-OMg treatment. Clearly, NP-OMg but not free sODN treatment significantly decreased the number of Ly6c^high MDMs while largely replenishing the population of KCs (Fig. 6D and E).

Previous studies have indicated the critical roles of activated signaling of STING-NF-κB and GSDMD in MASH patients [13,17], which was also supported by our analysis of a clinical GEO (Gene Expression Omnibus) dataset of MASH patients (Supplementary Fig. S10). Moreover, we observed that Sting^gt or Gsdmd−/− mice developed less severe steatohepatitis after they were fed the WD for 14 wks (Supplementary Figs. S11 and 12). To determine if NP-OMg exerts inhibitory effects on these signaling pathways, we conducted western blotting to detect the expression levels of key proteins involved in the pathways. In line with the RNA-seq data, the levels of phosphorylated NF-κB p65 and tank-binding kinase 1 (p-TBK1) and cleaved GSDMD-NT markedly increased in livers of WD-MASLD mice, which, however, significantly declined after NP-OMg treatment (Fig. 6F and G). It is known that activated TBK1 is indispensable in the activation of both STING and NF-κB signaling [12,13,34], and the interplay between STING, NF-κB and GSDMD has been implicated in the macrophage-mediated inflammatory cascade [35,36]. Together, these results provide in-depth insights into the mode of action of NP-OMg against WD-induced MASLD, which plausibly results from its potent suppression of liver macrophages-mediated multiple inflammatory pathways.

3.5. NP-OMg mitigates MCD diet-induced MASH

To demonstrate the effectiveness and reproducibility of NP-OMg against MASH, we employed another well-known MASH model induced by methionine- and choline-deficient (MCD) diet [27]. Compared to the WD-MASH model, mice fed the MCD diet developed MASH in much shorter time. At week 8, serum levels of TNF-α, IL-6 and IL-1β, and liver enzymes, AST and ALT, as well as liver TG and CHL contents were significantly elevated (Fig. 7A). However, unlike the WD-MASLD model, the MCD-MASH mice decreased in both their body and liver weights over time (Supplementary Fig. S7B; Fig. 7B and C) but exhibited characteristic hepatic steatosis and fibrosis histopathologically. We started to treat the MCD-fed mice on week 3 with free sODN, nanoparticle MgP or NP-OMg twice a week, and continued the treatment for 4 weeks (Fig. 7A). At week 8, NP-OMg treatment significantly lowered serum levels of inflammatory cytokines and liver enzymes (Fig. 7D), and hepatic contents of TG and CHL (Fig. 7E), while modest improvements were observed in the mice treated with MgP or free sODN (Fig. 7D). We then conducted RT-PCR analysis of liver tissues, and our data showed that in addition to suppressing genes (Tnf, Il-6 and Il-1b) encoding inflammatory cytokines, NP-OMg treatment markedly increased expression of lipolytic genes, Ppara and Acaca, while decreasing the level of lipogenic gene, Cd36 (Fig. 7F). Histological examinations confirmed that NP-OMg drastically mitigated hepatic steatosis and fibrosis (Fig. 7G). Furthermore, flow cytometry revealed that NP-OMg repolarized liver macrophages by replenishing the KC population while inhibiting the accumulation of inflammatory Ly6c^high MDMs (Fig. 7H). NP-OMg treatment also decreased the phosphorylation level of NF-κB p65 and the level of GSDMD-N terminal fragments (Supplementary Fig. S13). These data were consistent with our prior findings in the WD-MASLD model, suggesting the utility of liver macrophages-targeting NP-OMg for MASLD/MASH treatment. Intravenous administration of NP-OMg was well tolerated over the course of treatment in WD- or MCD-fed mice. Safety studies showed no abnormality in blood liver enzyme levels or morphological changes in major organs of healthy mice treated with NP-OMg (Supplementary Fig. S14).

Fig. 7.

NP-OMg ameliorates MCD-induced MASH. (A) Six-week-old female C57/BL6 mice were fed ad libitum a MCD diet (Methionine- and Choline-Deficient L-Amino Acid Diet) while control mice were fed ad libitum chow diet (CD). After feeding for 3 weeks, mice were grouped (n = 6/group) and treated as follows: (I) CD diet; (II) MCD diet; (III) MCD with NP-MgP (6.25 mg/kg Mg²⁺, i.v, twice/wk), (IV) MCD with free sODN (2 mg/kg, i.p, twice/wk), (V) MCD with NP-OMg (2 mg/kg sODN, 6.25 mg/kg Mg²⁺, i.v, twice/wk). At week 8, mice were euthanized, and livers harvested (B) and weighed (C). (D) ELISA assay detecting serum levels of pro-inflammatory cytokines TNF-α, IL-6 and IL-1β, and liver enzymes AST and ALT. (E) Bar plots showing levels of triglyceride and cholesterol detected in livers. (F) Real time PCR of mRNA expression in liver tissues of the mice from the above treatment groups (n = 3/group). (G) H&E staining and Sirius Red staining assess hepatic steatosis and fibrosis. Scale bar = 50 μm. (H) Flow cytometry analysis of liver macrophage populations of inflammatory MDMs (CD11b ⁺ Ly6c^high) and KCs (CD11b⁺F4/80^high) in each treatment group. Data are presented as mean ± SD; *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001.

4. Discussion and conclusion

In this study, we reported a novel nanoparticle therapeutic, NP-OMg, that demonstrated excellent protection against diet-induced MASLD/MASH in mice. NP-OMg targets both suppressive oligodeoxynucleotides and magnesium to liver macrophages, where sODNs exert comprehensive inhibition of DNA sensing PRRs primarily involving TLR9, cGAS-STING-NF-κB and AIM2, while Mg²⁺ reduces mitochondrial ROS production and impedes the localization of GSDMD-NT to cell membrane. As a result of the suppression of inflammatory cytokines and chemokines by NP-OMg, liver recruitment of inflammatory MDMs is also significantly inhibited.

It has been reported that nanoparticles larger than 150 nm in diameter and with a negative surface charge are specially favored by KCs, while positively charged nanoparticles tend to be taken up by hepatocytes [37]. Previous studies have also shown that if nanoparticles are functionalized with a macrophage-targeting moiety, their specific targeting of KCs can be significantly enhanced [38,39]. The sugar mannose has been used in these studies to decorate the nanoparticles since KCs express the mannose scavenger receptor CD206. While CD206 is a useful surface marker for targeting macrophages, it expresses in a fraction (20%) of KCs and is a common biomarker of M2, the anti-inflammatory subtype of macrophages. Our in vivo biodistribution studies and cell type-based analyses revealed that PS-coated NPs were taken up predominantly by liver macrophages, in which KCs and Ly6C^high MDMs exhibited a similar uptake ratio (Fig. 4E–G). These data demonstrate the ability of NP-OMg to target both liver resident KCs and inflammatory MDMs. As a result, NP-OMg markedly reduced the hepatic infiltration of Ly6C^high inflammatory MDMs and largely restored the KC population (Fig. 6D and E; Fig. 7H). The switch of inflammatory Ly6C^high MDMs to restorative Ly6C^low macrophages has been shown to be essential for reversal of fibrosis and injury resolution [1,5]. In addition to liver, ex vivo IVIV imaging detected a low level of DiR-NPs (17%) in the lungs (Fig. 4C). This observation concurs with prior literature, in which systemically administered NPs are also trapped in lungs, mainly phagocytosed by macrophages [40]. To study the safety of intravenous NP-OMg, we conducted toxicology studies and found no abnormality in lab results or histological examinations of major organs including the lung after a single dose or 4 doses of NP-OMg (Supplementary Fig. S14).

To investigate the therapeutic potential of macrophage-targeted nano-delivery of anti-inflammatory drugs for MASH treatment, Barbro et al. used the mannose-nanoparticles to target dexamethasone (Dex) to KCs in inflammatory liver disease and showed that Dex-NPs effectively inhibited ROS and TNF-α production in liver tissues. However, Dex-NP was found to accelerate fibrogenesis by increasing collagen 1 and III depositions and tissue inhibitor of metalloproteinase-1 (TIMP-1) mRNA expression [38]. In contrast, NP-OMg significantly suppressed the transcription of fibrosis-related genes, e.g., Col1a, Col3a and Timp1 (Fig. 6A), which may account for its anti-hepatic fibrosis in addition to anti-inflammatory effects observed in the NP-OMg-treated MASH mice (Figs. 5G and 7G).

Recent studies have identified Trem2+ macrophages as lipid-associated macrophages, which play an important role in the clearance of apoptotic and lipid-laden hepatocytes in the MASH liver [41,42]. Loss of TREM2 promotes hepatic inflammation and accelerates the progression of MASH in a high fat diet (HFD) mouse model [42,43]. However, increased expression of TREM2 in liver macrophages and serum level of soluble TREM2 have been associated with MAFLD progression and advanced stages of MASH in clinical patients [44,45]. These seemingly paradoxical results suggest the likelihood of heterogeneous populations of TREM2+ liver macrophages that exert opposite effects in the MASH pathogenesis [46]. Indeed, several studies using single-cell RNA sequencing analysis revealed the population of macrophages overexpressing TREM2 and GPNMB, which were highly associated with the severity of MASLD and liver fibrosis [46,47]. Moreover, Xiong et al. studied dynamic changes of Trem2+ Gpnmb + macrophages and found a drastic reduction in hepatic expression of Trem2 and Gpnmb after MASH was reversed by pharmacological intervention [48]. In good agreement, our data showed that NP-OMg treatment successfully reversed MASLD pathologies, which was accompanied by a significant decrease in hepatic Trem2 and Gpnmd mRNA expression (Supplementary Fig. S9). Together, these findings suggest that TREM2+ macrophages may act as an adaptive and protective mechanism during the development of MASLD/MASH.

Administration of free sODN was found to alleviate steatohepatitis, but to a much lesser extent than NP-OMg (Fig. 5 and Fig. 7). It should be noted, however, that free sODN and NP-OMg were administered through different routes (i.p. and i.v., respectively). While a robust suppression of inflammatory responses by free sODN has been observed in previous in vitro studies by others and confirmed in this study, systemic delivery of sODN for in vivo applications remains challenging. Intraperitoneal administration of free sODN has been adopted exclusively in preclinical studies, though it is clinically impractical. Thus, a major goal of this study is to develop a carrier system enabling intravenous administration of sODN. Moreover, we designed NP-OMg to target sODN to liver macrophages (Fig. 4E). Although intravenously administered FAM-sODNs were also detected in KCs, the uptake was significantly lower than that of NP-OMgs (Fig. 4G). Inclusion of Mg²⁺ salts improves the loading efficiency of oligonucleotides and stabilizes oligonucleotides by forming precipitates in NP (Fig. 2F). Magnesium is an essential element in the human body, and Mg²⁺ deficiency is often observed in MASH patients. While therapeutic effects of supplemental Mg²⁺ in MASH are unclear, low levels of cellular Mg²⁺ have been correlated with MASH progression by promoting inflammation, lipid accumulation and HCC [23,49]. Nevertheless, our data showed that intravenous administration of NP-OMg enabled liver macrophages-targeted delivery of sODN and consequently mitigated inflammatory responses during the development of MASLD/MASH.

Oxidized DNA, released by mitochondria, can also activate NLRP3 inflammasome [50]. Indeed, our RNA-seq analysis revealed that Nlrp3 expression and pyroptosis signaling were significantly upregulated in livers of MASH mice, which, however, were largely suppressed by NP-OMg (Fig. 1, Fig. 6). While previous studies suggested that sODN does not directly interact with NLRP3, NP-OMg likely exerted its inhibitory effects on NLRP3 signaling through the following mechanisms. First, as shown in our genomic and immunological data, NP-OMg significantly inhibited inflammatory signaling of NF-kB and TLRs, reducing the levels of inflammatory cytokines, TNF-α and IL-1β (Fig. 6A–C), which are prominent priming triggers required for transcriptional expression and posttranslational modifications of NLRP3 and pro-IL-1β [51]. Secondly, to activate inflammasome signaling, signal 2, e.g., Ca²⁺ influx or K⁺ efflux is indispensable [52]. Given Mg²⁺ being the main antagonist of Ca²⁺, intracellular delivery of Mg²⁺ by NP-OMg could prohibit Ca²⁺ influx and/or K⁺ efflux, thus attenuating the activation signal. Additionally, concurring with previous studies [24,32], our data showed that Mg²⁺ alleviated the mitochondrial oxidative stress (Fig. 2C), which is another well-recognized factor to elicit NLRP3 inflammasome activation [53]. Collectively, NP-OMg augments anti-inflammatory effects plausibly through its direct and indirect interference with various DNA sensing pathways. Future studies using isolated hepatic cell populations or single-cell approaches will be required to more precisely define the effects of NP-OMg on individual liver cell populations.

The emergence of semaglutide and resmetirom is highly promising for MASH treatment. However, these new drugs reduce fibrosis only in a subset of patients and have not yet shown benefits in advanced fibrosis stage. Combining these therapies with new therapeutic agents that target divergent pathogenic mechanisms and cell types may be necessary for treatment of advanced stages of MASH. NP-OMg, which mainly targets inflammatory signals that are critical to promote HSC-mediated fibrosis, may be complementary to resmetirom treatment. Our future studies will explore the therapeutic effects of NP-OMg combined with resmetirom or semaglutide in preclinical diet-induced MASH models. While the mouse models used in this study displayed marked steatohepatitis, they were still at early stages of MASH without significant fibrosis. Thus, a mouse model at advanced stage of fibrosis will be needed to investigate anti-fibrotic effects of the new drugs [54].

Because PS is the same molecule in all mammalian species, the PS-coated nanoplatform showing a high specificity to mouse macrophages in this study can also be applied for targeting human liver macrophages. Moreover, TTAGGG motifs, a common structure in telomeres of both human and mouse, have been shown to suppress the innate immune response in both human and mouse cells. Thus, it is plausible that NP-OMg may be extrapolated to MASH treatment in humans.

CRediT authorship contribution statement

Kaige Chen: Writing – review & editing, Writing – original draft, Validation, Methodology, Investigation, Formal analysis, Data curation. Lulu Wang: Writing – review & editing, Visualization, Validation, Project administration, Methodology, Investigation, Formal analysis, Data curation. Liang Liu: Writing – review & editing, Formal analysis, Data curation. Chad A. Arledge: Writing – review & editing, Visualization, Data curation. Hui-Kuan Lin: Resources, Methodology. Jingmei Lin: Methodology, Formal analysis. Amyn A. Habib: Resources, Methodology. Dawen Zhao: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Resources, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization.

Ethics approval and consent to participate

All animal experiments complied with all relevant ethical regulations for animal testing and research and were performed with approval by the Institutional Animal Care and Use Committee (IACUC) at the Wake Forest University School of Medicine (approval number A26-073).

Declaration of competing interest

The authors declare that they have no competing interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: