GalNAc-conjugated siRNA targeting C/EBPβ reverses metabolic dysfunction and restores liver homeostasis in a murine MASLD model

TL;DR

A new siRNA therapy targeting C/EBPβ improves liver health and metabolism in mice with a liver disease linked to metabolic dysfunction.

Contribution

GalNAc-conjugated siRNA targeting C/EBPβ is shown to reverse metabolic dysfunction and liver damage in a MASLD model.

Findings

GalNAc-siCEBPβ reduced liver steatosis and improved metabolic parameters in mice on a high-fat diet.

The therapy restored liver function markers without causing hepatotoxicity.

C/EBPβ silencing occurred with high efficiency both in vitro and in vivo.

Abstract

CCAAT/enhancer-binding protein beta (C/EBPβ) is a master regulator of hepatic metabolism, inflammation, and fibrosis, making it an attractive but underexploited target for metabolic dysfunction-associated steatotic liver disease (MASLD). Here, we demonstrate that GalNAc-conjugated small interfering RNA (siRNA) targeting C/EBPβ (GalNAc-siCEBPβ) significantly improves liver function and metabolic parameters in a high fat diet (HFD) murine model. In vitro, GalNAc-siCEBPβ achieved dose-dependent C/EBPβ mRNA silencing (∼80% knockdown at 0.1 μM) in primary mouse hepatocytes. In vivo, subcutaneous administration (10 mg/kg) reduced hepatic C/EBPβ expression by 45% (p < 0.01), concomitant with a marked reduction in liver steatosis and improved metabolic profile (15% less weight gain, 20% lower glucose, 25% reduced triglycerides), and restored liver function (18% higher albumin, 22% lower bilirubin)—all without hepatotoxicity (ALT/AST unchanged). Notably, these effects occurred despite continued HFD feeding, suggesting disease-modifying potential. By leveraging the precision of RNAi and hepatocyte-specific GalNAc delivery, GalNAc-siCEBPβ addresses key limitations of current MASLD therapies by targeting both metabolism and fibrosis. Our findings support clinical translation for MASLD and its complications, including hepatocellular carcinoma.

This preclinical study demonstrates that GalNAc-mediated delivery of C/EBPβ siRNA silences a key transcriptional driver of MASLD, resulting in multi-faceted therapeutic benefits. These include steatosis reduction, systemic metabolic improvements, and a restoration of hepatic function (synthetic and detoxification), offering a promising therapeutic strategy for clinical development in advanced liver stage patients.

Figures

Efficacy of GalNAc-siCEBPβ: from *in vitro* target engagement to *in vivo* steatosis reduction(A) *In vitro* dose-dependent silencing of C/EBPβ mRNA by GalNAc-siCEBPβ in primary mouse hepatocytes. Primary mouse hepatocytes were treated with escalating doses (0.001–5 μM) of GalNAc-siCEBPβ for 48 h. C/EBPβ mRNA levels were quantified by qPCR and normalized to untreated controls. Data show mean ± SD (*n* = 3 biological replicates). Dashed line indicates 50% knockdown achieved at 6 nM by 48 h (B) *In vivo* knockdown of hepatic C/EBPβ mRNA by GalNAc-siCEBPβ in HFD-fed mice. HFD mice received subcut

Metabolic improvements in HFD-fed mice treated with GalNAc-siCEBPβ.Metabolic biometrics measured at 12 weeks in HFD mice after receiving subcutaneous injections on day 1, 3, 5, 8, 12, 15 and 19 of GalNAc-siCEBPβ (10 mg/kg, *n* = 6), GalNAc-siFLUC (non-targeting control, 10 mg/kg, *n* = 8), or PBS (*n* = 8). Encompassing body weight and white adipose tissue (WAT) mass (g), lipid profile (HDL, LDL, cholesterol, and triglycerides) (mg/dL), glucose (mg/dL), HbA1c (%), insulin (μU/mL), and C-peptide (ng/mL). Data are mean ± SD. Significance in respect to the PBS group: ∗*p* < 0.05 (unpaired *t* tes

GalNAc-siCEBPβ restores liver function without toxicity.Liver biochemistry at 12 weeks in HFD-fed mice after receiving subcutaneous injections on day 1, 3, 5, 8, 12, 15, and 19 of GalNAc-siCEBPβ (10 mg/kg, *n* = 6), GalNAc-siFLUC (non-targeting control, 10 mg/kg, *n* = 8), or PBS (*n* = 8). Parameters measured include alanine transaminase (ALT) (IU/L) and aspartate transaminase (AST) (IU/L) to assess for liver injury/hepatotoxicity. Albumin (mg/dL) was used to assess liver synthetic function and bilirubin (mg/dL) to assess liver detoxification capabilities. Data are mean ± SD. Significance in

Full text

Introduction

The CCAAT/enhancer-binding proteins (C/EBPs) are a family of transcription factors that regulate hepatic metabolism, inflammation, fibrosis, and its regeneration, by binding to CCAAT motifs in gene promoters.1 Among these, C/EBPβ is a key regulator of lipogenesis and fibrogenesis, driving expression of PPARγ and SREBP1c to promote lipid accumulation2 and transforming growth factor β (TGF-β) to exacerbate fibrosis.3 Its activity is modulated by isoforms: the full-length LAP variant activates target genes, while the truncated LIP isoform acts as a dominant-negative inhibitor.4 In metabolic liver disease, C/EBPβ is upregulated, skewing the LAP/LIP ratio toward lipogenic and pro-inflammatory signaling,5 a phenomenon recapitulated in our murine high fat diet (HFD) model.

Metabolic-dysfunction-associated steatotic liver disease (MASLD) affects 25% of adults globally, with its progressive form metabolic-dysfunction-associated steatohepatitis (MASH) being the fastest-growing cause of cirrhosis and hepatocellular carcinoma (HCC).6 Current pharmacotherapies—GLP-1 receptor agonists (e.g., semaglutide) and thyroid receptor hormone (THR-β) agonists (e.g., resmetirom)—improve steatosis but show limited efficacy in advanced fibrosis or cirrhosis.7^,^8 Moreover, they do not restore synthetic liver function (e.g., albumin production) or detoxification (e.g., bilirubin clearance), thereby leaving decompensated patients ineligible for HCC treatment.9 Gene-specific therapies like small interfering RNA (siRNA) could bridge this gap by directly targeting pathogenic drivers like C/EBPβ.

Therapeutic RNA interference (RNAi) offers gene targeting precision, but delivery to hepatocytes remains challenging. GalNAc conjugation solves this by exploiting the asialoglycoprotein receptor (ASGPR) for hepatocyte-specific uptake, enabling subcutaneous dosing with month-long silencing effect.10 Clinical success of GalNAc-siRNAs (e.g., inclisiran for hypercholesterolemia) validates this platform, as yet, there are no therapeutic candidates that target C/EBPβ despite its mechanistic centrality in MASLD. Here, we test the hypothesis that GalNAc-siCEBPβ can reverse metabolic dysfunction and restore liver function in HFD-induced steatosis, a precursor to MASH and HCC.

Results

In vitro silencing activity of GalNAc-siCEBPβ in primary mouse hepatocytes

To assess the gene-silencing activity of GalNAc-siCEBPβ, cryopreserved primary mouse hepatocytes (PMHs) were treated with escalating doses (0.001–5 μM) of the siRNA conjugate. Quantitative PCR (qPCR) analysis revealed a dose-dependent reduction in C/EBPβ mRNA levels, with near-maximal silencing (∼80% knockdown relative to untreated controls) achieved at 0.1 μM by 48 h. Notably, the EC50 was ∼6 nM, demonstrating high potency (Figure 1A). No cytotoxicity was observed at any concentration (data not shown), supporting the suitability of this construct for in vivo testing.Figure 1. Efficacy of GalNAc-siCEBPβ: from in vitro target engagement to in vivo steatosis reduction(A) In vitro dose-dependent silencing of C/EBPβ mRNA by GalNAc-siCEBPβ in primary mouse hepatocytes. Primary mouse hepatocytes were treated with escalating doses (0.001–5 μM) of GalNAc-siCEBPβ for 48 h. C/EBPβ mRNA levels were quantified by qPCR and normalized to untreated controls. Data show mean ± SD (n = 3 biological replicates). Dashed line indicates 50% knockdown achieved at 6 nM by 48 h (B) In vivo knockdown of hepatic C/EBPβ mRNA by GalNAc-siCEBPβ in HFD-fed mice. HFD mice received subcutaneous injections on day 1, 3, 5, 8, 12, 15, and 19 of GalNAc-siCEBPβ (10 mg/kg, n = 6), GalNAc-siFLUC (non-targeting control, 10 mg/kg, n = 8), or PBS (n = 8). Liver C/EBPβ mRNA was measured by qPCR at 12 weeks. Data represent mean ± SD, ∗∗p < 0.01 vs. PBS (unpaired t test with Welch’s correction). (C–E) GalNAc-siCEBPβ attenuates hepatic steatosis in HFD mice. (C) PBS- and (D) GalNAc-siFLUC-treated controls show extensive macrovesicular steatosis throughout the liver section, as shown by the accumulation of large fat droplets (red on stain) in hepatocytes that displaces the nucleus, while (E) GalNAc-siCEBPβ-treated livers exhibit minimal lipid accumulation (a decrease in red color intensity) and no inflammatory infiltrates (oil red O stain, 20× magnification).

In vivo knockdown of hepatic C/EBPβ in HFD mice

Subcutaneous administration of GalNAc-siCEBPβ (10 mg/kg) in HFD mice resulted in significant reduction of hepatic C/EBPβ mRNA (∼45%, p < 0.01) compared to PBS and GalNAc-siFLUC (non-targeting control) groups at 12 weeks (Figure 1B). The silencing effect was liver-specific, with no detectable mRNA changes in extrahepatic tissues (e.g., adipose and skeletal muscle; data not shown). This confirms efficient GalNAc-mediated hepatocyte delivery and target engagement, even in steatotic livers with potential ASGPR downregulation.11

GalNAc-siCEBPβ attenuates hepatic steatosis and inflammation

Histological analysis of HFD mouse livers revealed a marked reduction in large fat droplet formation in hepatocytes from GalNAc-siCEBPβ-treated animals compared to GalNAc-siFLUC controls (Figures 1C–1E), with a decrease in hepatic steatosis from severe (>60%) to mild/minimal (<25%). Furthermore, treated livers exhibited minimal inflammatory infiltrates. These findings align with C/EBPβ′s role in promoting lipid metabolism and inflammation.2

Metabolic improvements in HFD mice

GalNAc-siCEBPβ treatment induced a number of systemic metabolic benefits (Figure 2).Figure 2. Metabolic improvements in HFD-fed mice treated with GalNAc-siCEBPβ.Metabolic biometrics measured at 12 weeks in HFD mice after receiving subcutaneous injections on day 1, 3, 5, 8, 12, 15 and 19 of GalNAc-siCEBPβ (10 mg/kg, n = 6), GalNAc-siFLUC (non-targeting control, 10 mg/kg, n = 8), or PBS (n = 8). Encompassing body weight and white adipose tissue (WAT) mass (g), lipid profile (HDL, LDL, cholesterol, and triglycerides) (mg/dL), glucose (mg/dL), HbA1c (%), insulin (μU/mL), and C-peptide (ng/mL). Data are mean ± SD. Significance in respect to the PBS group: ∗p < 0.05 (unpaired t test with Welch’s correction).

First, regarding body weight, treated mice gained 15% less weight than controls (p < 0.05) despite equivalent caloric intake. Of note, there was no significant difference in baseline weights between mice groups. Second, adiposity was less, with white adipose tissue (WAT) mass reduced by ∼30% (p < 0.05). Third, there was an improvement in the beneficial lipid profile, with triglyceride levels dropping by 25%, while HDL/LDL ratio increased by 1.5-fold (p < 0.05). Fourth, glucose homeostasis was improved with fasting glucose, and HbA1c levels decreased by 20% and 12%, respectively (p < 0.05). Notably, insulin and C-peptide levels remained unchanged, suggesting these metabolic improvements were independent of pancreatic β-cell function and likely driven by hepatic C/EBPβ silencing. Additionally, the use of the C57BL/6J mouse substrain, which carries the Nnt mutation and has impaired glucose stimulated insulin secretion, strengthens the observation that GalNAc-siCEBPβ acts primarily via hepatic mechanisms to improve glucose homeostasis.

Restoration of liver function without toxicity

Serum biochemistry analysis demonstrated no hepatotoxicity, as alanine transaminase/aspartate transaminase (ALT/AST) levels were comparable to controls (Figure 3), ruling out any evidence of drug-induced liver injury. There was also improved synthetic function, with albumin levels increasing by 18% (p < 0.05), and enhanced liver detoxification, with bilirubin decreasing by 22% (p < 0.01), both indices reflecting enhanced hepatocyte health. These improvements in the ALBI score components (albumin and bilirubin) suggest potential applicability of GalNAc-siCEBPβ in managing advanced liver disease.12Figure 3. GalNAc-siCEBPβ restores liver function without toxicity.Liver biochemistry at 12 weeks in HFD-fed mice after receiving subcutaneous injections on day 1, 3, 5, 8, 12, 15, and 19 of GalNAc-siCEBPβ (10 mg/kg, n = 6), GalNAc-siFLUC (non-targeting control, 10 mg/kg, n = 8), or PBS (n = 8). Parameters measured include alanine transaminase (ALT) (IU/L) and aspartate transaminase (AST) (IU/L) to assess for liver injury/hepatotoxicity. Albumin (mg/dL) was used to assess liver synthetic function and bilirubin (mg/dL) to assess liver detoxification capabilities. Data are mean ± SD. Significance in respect to the PBS group: ∗p < 0.05, ∗∗p < 0.01; unpaired t test with Welch’s correction. Dotted line represents “normal” to highlight how a change in bilirubin and albumin can impact decision-making in liver cancer management.

Discussion

In this study, we demonstrate that GalNAc-conjugated siCEBPβ effectively improves metabolic dysfunction, reduces hepatic steatosis, and restores liver synthetic/detoxification functions in an HFD murine model. Our findings position GalNAc-siCEBPβ as a promising candidate for clinical translation, particularly for metabolic-dysfunction-associated steatotic liver disease (MASLD) and its advanced stages (MASH), where current therapies face significant limitations.

Unlike peptide and small molecule drugs (e.g., GLP-1 receptor agonists and THR-β agonists) that broadly modulate metabolic pathways, an siRNA offers gene-specific silencing of C/EBPβ, a master regulator of lipogenesis, inflammation, and fibrosis. This precision minimizes off-target effects, a useful advantage in cirrhotic patients with compromised liver function, where systemic toxicity is a major concern.13 Compared to antisense oligonucleotides (ASOs), siRNA’s catalytic mRNA degradation also provides sustained effects at lower doses.14 Our data demonstrate ∼45% C/EBPβ mRNA knockdown persisting for weeks post-dose, aligning with the durability seen in other clinical GalNAc-siRNA platforms in use (e.g., inclisiran for primary hypercholesterolemia and lumasiran for primary hyperoxaluria type 1).

GalNAc conjugation leverages the liver’s asialoglycoprotein receptor (ASGPR) for hepatocyte-specific uptake, avoiding the toxicity and immunogenic risks of viral vectors such as AAVs.10 Notably, our construct retained efficacy despite reduced ASGPR expression in steatotic livers,11 a hurdle for other targeted therapies. In addition, subcutaneous administration and infrequent dosing (e.g., quarterly regimens in humans) contrasts with daily oral drugs (e.g., resmetirom) or weekly GLP-1 injections, potentially improving patient experience and compliance.15

Current therapies (e.g., semaglutide, a GLP-1 agonist) are contraindicated in decompensated cirrhosis (Child-Pugh B/C) due to lack of efficacy in restoring synthetic function.7 In contrast, GalNAc-siCEBPβ normalized both albumin and bilirubin (Figure 3), variables of the ALBI score, a prognostic marker used in decision-making for hepatocellular carcinoma management.12 This suggests a potential to “rescue” liver function in patients otherwise ineligible for cancer treatment. Of note the reduction in steatosis without transaminitis (Figures 1 and 3) also underscores a role in hepatoprotection, in contrast to the paradoxical liver injury reported with some THR-β agonists.8

There are also potentially a number of metabolic benefits beyond just the changes observed with liver steatosis. GalNAc-siCEBPβ improved systemic metabolism (weight, glucose, and lipids; Figure 2), akin to GLP-1 agonists, but with a distinct molecular mechanism: silencing C/EBPβ maybe shifting hepatocyte energetics by downregulating PPARγ-driven lipogenesis while upregulating C/EBPα, a tumor suppressor. This dual effect could potentially benefit both MASLD and HCC prevention.2^,^16

While our HFD model recapitulates early MASLD, the improvements in albumin and bilirubin hint at utility in decompensated cirrhosis: a population with no pharmacologic options beyond liver transplantation. In conclusion, GalNAc-siCEBPβ merges the precision of RNAi with the clinical practicality of GalNAc delivery, addressing unmet needs in MASLD/MASH: reversing steatosis, restoring liver function, and mitigating metabolic dysregulation. Its potential to bridge patients to cancer therapy or transplant warrants accelerated clinical development.

Looking ahead, future studies to build upon this proof-of-concept study will focus on (1) assessing GalNAc-siCEBPβ in advanced fibrosis/cirrhosis models (e.g., CCl_4_ and STAM) to define its disease-modifying potential, (2) comprehensive biodistribution and long-term safety monitoring, given C/EBPβ′s role in regeneration,5 and (3) detailed mechanistic analyses, including effects on C/EBPβ isoforms (LAP/LIP) and key downstream transcriptional targets (e.g., PPARγ, SREBP1c), to further elucidate its mode of action.

Materials and methods

siRNA design and GalNAc conjugation

An siRNA sequence designed to target murine/human C/EBPβ (NCBI Accession: NM_009883.4) using cross-species conserved regions was used, as previously described.17 The 5′ end of siRNA sense strand was conjugated to trivalent N-acetylgalactosamine (GalNAc) ligand via an aminohexyl linker to ensure ASGPR-mediated hepatocyte uptake.18 Conjugation efficiency (>95%) is confirmed by HPLC (data not shown).

In vitro siRNA activity analysis

For dose-response analysis, cryopreserved primary mouse hepatocytes (PMH) (Thermo Fisher Scientific Catalog number MSCP10) were thawed and plated at 25,000 cells/well (96-well plate) in Williams’ E medium +10% FBS as per recommended maintenance protocol. Cells were exposed to GalNAc-siCEBPβ (0.001–1 μM) for 48 h with siRNA treatment performed in quadruplicate for each data point. Target mRNA silencing was assessed using branched DNA assay. IC50 value was calculated using the XLfit tool. For qPCR mRNA knockdown analysis, total RNA was extracted using Qiagen RNAeasy (Cat no./ID. 74104), with C/EBPβmRNA quantified using SYBR Green PCR Master Mix (Applied Biosystems) and target-specific Qiagen QuantiTect Primer Assays (Cat no./ID. 249900) (Mm_Cebpb_1_SG, QT00320313). Gene expression levels were normalized to GAPDH and calculated using the ddCt method.

In vivo HFD model and dosing

Animals used in this experiment were 8-week-old C57BL/6J(B6) male mice fed a 60% high fat diet (HFD, Research Diets D12492) for 9 weeks to induce the HFD phenotype. All experimental procedures were approved by and carried out in compliance with the guidelines set by the Animal Care and Use Committee of the National Taiwan University College of Medicine. Animals were housed in groups of three per cage. Animals were maintained under standard controlled environmental conditions of 22^o^C ± 3^o^C, 50% ± 20% humidity, a 12-h night/dark cycle, and 15–20 fresh air changes per hour. After 9 weeks of an HFD, animals were randomly split into the following three groups: (1) GalNAc-siCEBPβ (experimental siRNA, n = 6), (2) GalNAc-siFLUC (non-targeting control, n = 8), and (3) PBS (vehicle control, n = 8). Dosing of GalNAc-siRNA conjugates was 10 mg/kg given subcutaneously (s/c) on days 1, 3, 5, 8, 12, 15, and 19. At the end of week 12, animals were euthanized and weighed. White adipose tissue (WAT) was removed from the abdominal wall. Tissue and blood samples were collected for later downstream analysis.

Histology

Frozen sections (5 μm) of liver (snap-frozen) were air dried for 30–60 min at room temperature and then fixed in ice-cold 10% formalin. The sections were stained with freshly prepared oil red O (cat. no. 1024190250; Sigma-Aldrich) working solution for 15 min and then rinsed with 60% isopropanol to visualize neutral lipid content. Slides were lightly stained with hematoxylin (cat. no. H3136; Sigma-Aldrich) for 1 min, then mounted with glycerin jelly or another aqueous mounting medium. A semi-quantitative scoring system was employed to estimate the percentage of hepatic parenchyma occupied by macrovesicular lipid droplets: grade 1 (mild, <33%), grade 2 (moderate, 34%–66%), or grade 3 (severe, >66%). This scoring approach is standard for the qualitative assessment of ORO-stained sections in metabolic liver disease research.

Metabolic and liver function assays

Serum samples from all animals was collected after euthanasia and analyzed for cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL), triglycerides (TGs), fasting glucose, hemoglobin A1C (HbA1C), alanine transaminase (IU/L), and aspartate transaminase (IU/L) using a Clinical Chemistry Analyzer (AU5800; Beckman Coulter Inc., Brea, CA). For insulin and C-peptide levels an immunoassay analyzer was used.

Statistical analysis

Data are presented as mean ± SD. Comparisons are performed using an unpaired t test (Welch’s correction) or one-way ANOVA + Dunnett’s test (GraphPad Prism v.9). p < 0.05 = significant. Power analysis indicated that n = 6–8/group ensured 80% power to detect 30% difference (α = 0.05).

Data and code availability

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Acknowledgments

None.

Author contributions

H.N. and R.J. developed the concept. H.K.-W. performed in vivo assays. V.D. performed laboratory work. H.N., R.J., H.K.-W., K.S.E., N.R., R.V., N.J., S.M., D.A., S.P., and V.D. prepared the manuscript.

Declaration of interests

R.J. and H.N. are founders and owners of Apterna Ltd. N.J. and H.K.-W. own shares in Apterna Ltd.

Declaration of generative AI and AI-assisted technologies in the writing process

In the preparation of this work, the author(s) used DeepSeek in editing to improve readability and language. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.