A unimolecular GLP-1 and FGF21 dual agonist for treatment of metabolic dysfunction-associated steatohepatitis

Abstract

Background

Metabolic dysfunction–associated steatohepatitis is a progressive liver disease characterized by inflammation and fibrosis, for which effective pharmacological treatments remain limited. Hormones that regulate metabolism, such as glucagon-like peptide-1 and fibroblast growth factor-21, have shown therapeutic promise through complementary metabolic and hepatoprotective actions. This study aims to develop and evaluate a single molecular therapy that engages both pathways to address multiple drivers of disease progression.

Methods

We designed a unimolecular dual agonist by linking glucagon-like peptide-1 and fibroblast growth factor-21 using a thermally responsive elastin-like polypeptide linker to enable sustained drug exposure after subcutaneous administration. Receptor activity was assessed in cell-based assays. Therapeutic efficacy was evaluated in male C57Bl6/J mice with diet-induced advanced steatohepatitis and fibrosis using metabolic measurements, gene expression analyses, protein quantification, and histological assessment of liver tissue.

Results

Here we show that the dual agonist retains potent activity at both target receptors and forms a reversible subcutaneous depot that prolongs systemic exposure. Treatment improves body weight, liver mass, blood glucose, and cholesterol levels in mice with advanced liver disease. We further observe reduced liver inflammation and fibrosis, accompanied by decreased expression of inflammatory and fibrotic markers, reduced fat accumulation, and increased hepatocyte proliferation.

Conclusions

These findings demonstrate that a single long-acting molecule engaging both metabolic hormone pathways improves metabolic and hepatic features of advanced steatohepatitis in a preclinical model. This work supports continued development of unimolecular multi-pathway therapies as a potential strategy for chronic liver disease.

Plain Language Summary

Fatty liver disease can worsen over time and lead to liver inflammation and scarring, which can seriously affect health. Current treatments are limited, especially for people with advanced disease. This study aimed to develop a new type of medicine that targets more than one disease pathway at the same time. The researchers created a single long-acting drug by combining two natural hormones that help regulate metabolism and protect the liver. This drug was tested in mice with advanced fatty liver disease to see how it affected body weight, blood sugar, cholesterol, and liver health. The treatment improved metabolic health and reduced liver inflammation and scarring. These findings suggest that combining multiple beneficial actions into one medicine could be a promising approach for future treatments of advanced fatty liver disease.

Sirohi et.al. develop a unimolecular dual agonist targeting both GLP-1 and FGF21 for the treatment of advanced metabolic liver disease. The dual agonist improves metabolic parameters, reduces liver inflammation and fibrosis, and promotes liver repair in a preclinical murine model.

Introduction

Metabolic Dysfunction-Associated Steatohepatitis (MASH) has become a major global health challenge due to its increasing prevalence and association with liver-related complications, including cirrhosis and hepatocellular carcinoma¹. Histologically, MASH is characterized by steatosis, hepatocyte ballooning, lobular inflammation, and varying degrees of fibrosis, which is the strongest predictor of long-term outcomes such as liver failure and mortality². Despite the recent approval of Rezdiffra (resmetirom), which marks a significant milestone as the first FDA-approved therapy for non-cirrhotic MASH with moderate to advanced fibrosis, there remains a pressing clinical need for additional therapeutic options. MASH is a multifactorial disease characterized by metabolic dysfunction, inflammation, hepatocellular injury, and progressive fibrosis—components that may not be fully addressed by a single pathway-targeting agent. Rezdiffra primarily acts by reducing hepatic fat via thyroid hormone receptor activation³, but it does not directly target fibrosis reversal or promote liver regeneration⁴. Moreover, patient heterogeneity, varied disease progression rates, and potential comorbidities highlight the need for multi-targeted or combination approaches⁵. Therefore, the development of next-generation therapeutics that can simultaneously modulate multiple pathological processes is urgently needed to more comprehensively treat MASH and prevent progression to cirrhosis or liver failure.

Glucagon-like peptide-1 (GLP-1) receptor agonists, originally developed to treat type 2 diabetes, have shown beneficial effects in MASH patients due to their ability to reduce body weight, improve insulin sensitivity, and decrease hepatic steatosis⁶. These agents exert their metabolic effects primarily through appetite suppression and delayed gastric emptying, resulting in reduced caloric intake and improved glycemic control⁷. Consistent with these benefits, semaglutide (Wegovy) was recently granted accelerated approval by the FDA for the treatment of MASH, based on interim clinical data demonstrating resolution of steatohepatitis and signals of fibrosis improvement pending confirmatory outcomes. However, while clinical trials of GLP-1 analogs like liraglutide and semaglutide have reported histological improvements in steatohepatitis, these effects are thought to arise largely from systemic metabolic improvement and weight loss rather than direct modulation of fibrogenic pathways, and their ability to regress established fibrosis has been modest². This limitation has prompted the search for complementary agents that can target fibrogenesis as well.

Fibroblast growth factor-21 (FGF21) is a hormone-like protein produced primarily by the liver in response to metabolic stress^8,9. It plays a central role in regulating lipid oxidation, ketogenesis, and insulin sensitivity, and has demonstrated therapeutic potential in both preclinical and early clinical studies of MASH^10,11. In addition to its metabolic actions, multiple studies have shown that FGF21 signaling influences pathways relevant to hepatic inflammation and fibrosis, including suppression of pro-fibrotic cytokines, modulation of stellate cell activation, and attenuation of inflammatory signaling^12,13. These effects have been observed in the context of improved hepatic metabolic homeostasis rather than as isolated, metabolism-independent antifibrotic actions. Given the overlapping yet distinct actions of GLP-1 and FGF21, combining them into a single molecule offers a promising strategy for addressing multiple drivers of MASH progression¹⁴. Although the present study does not include single-agent comparator groups, extensive prior literature supports the complementary benefits of GLP-1 and FGF21 pathway activation across metabolic, inflammatory, and hepatic endpoints. We hence hypothesized that a unimolecular GLP1–FGF21 dual agonist may achieve more robust and coordinated control over metabolic, inflammatory, and fibrotic disease drivers while simplifying dosing and improving adherence compared to co-administration of individual agents¹⁵.

To construct the unimolecular GLP1-FGF21 dual agonist, we use an elastin-like polypeptide (ELP)—a repetitive sequence of VPGXG, where X is any amino acid except Proline—as a linker between the GLP-1 and FGF21 domains¹⁶. We chose an ELP to link GLP-1 with FGF21 for several reasons: 1) ELPs are intrinsically disordered polypeptides, which make them flexible linkers that should allow both the GLP-1 and FGF21 to engage with their respective receptors; 2) ELPs exhibit LCST phase behavior, such that they can be designed—at their sequence level—to be soluble in the syringe at room temperature, but undergo thermally driven phase separation into an insoluble coacervate depot upon subcutaneous (s.c.) injection, driven by body heat^17,18; 3) the kinetics of dissolution of the depot can be programmed by control of the hydrophobicity of the ELP sequence and its molecular weight^19,20; and 4) upon dissolution of the depot, the soluble GEF molecules released from the depot will have an extended circulation half-life compared to the individual drugs. These unique features make ELPs attractive for the design of a long-acting protein therapeutics²¹.

Numerous mouse models have been developed to replicate various features of human MASH, including dietary models such as the methionine- and choline-deficient (MCD) diet, the Western diet (high in fat, cholesterol, and sugars), and genetic models like ob/ob and foz/foz mice²¹. However, many of these models either fail to develop significant fibrosis or confound the analysis of fibrosis by introducing obesity and insulin resistance as major variables. In this study, we utilized 24-week-old mice and fed them a choline-deficient, L-amino acid-defined high-fat diet (CDA-HFD) for 16 continuous weeks, which reliably induces advanced hepatic fibrosis while minimizing confounding metabolic phenotypes seen in obesity or diabetes²². We choose this animal model because both the GLP-1R agonists and FGF21 are known to correct systemic metabolic dysfunction but their effect on MASH is unknown. Hence, this approach enabled us to evaluate the impact of dual GLP-1/FGF21 pathway activation on hepatic inflammation and fibrosis in the relative absence of obesity and type 2 diabetes, thereby isolating disease-relevant liver outcomes.

In this study, we design and characterize a unimolecular dual agonist that integrates GLP-1 and FGF21 signaling within a single long-acting molecule. We show that this dual agonist retains functional activity at both the target receptors and exhibits extended pharmacokinetics following subcutaneous administration. In a diet-induced murine model of advanced metabolic dysfunction–associated steatohepatitis with established fibrosis, treatment leads to significant improvements in body weight, liver mass, and circulating metabolic parameters. We further demonstrate reductions in hepatic inflammatory and fibrotic gene expression, accompanied by decreased steatosis and improved liver histology. Collectively, these findings support the feasibility of a single-agent, multi-pathway therapeutic strategy for addressing key metabolic, inflammatory, and fibrotic features of advanced liver disease.

Methods

Expression and purification of GLP1-ELP-FGF21 fusion protein

To produce the final GLP1-ELP-FGF21 construct, the fusion protein MSKGPG-tev-GLP1-ELP-FGF21 was first expressed and purified. The N-terminal MSKGPG leader sequence enhances protein expression yield in E. coli²³, and a TEV protease recognition site (tev = ENLYFQG) enables site-specific and seamless cleavage of the leader sequence leaving the GLP-1 active on the N-terminus of the fusion protein. The TEV protease was also recombinantly produced as a his₆-ELP-TEV fusion where ELP enabled proper folding of the TEV during expression in E. coli, and the histidine (his-6x) tag assisted in purification of the protein from lysate^24–26.

Expression and purification of MSKGPG-tev-GLP1-ELP-FGF21

The plasmid encoding the gene for MSKGPG-tev-GLP1-ELP-FGF21 fusion was transformed into SHuffle T7 Express cells (New England Biolabs), which support enhanced disulfide bond formation in the cytoplasm. The growth media was prepared by supplementing Terrific broth (TB; 55 g/L; VWR) with 0.4% glycerol and 45 µg/mL kanamycin. A 50 mL overnight culture was grown at 37 °C with shaking at 250 rpm overnight. This pre-culture was pelleted by centrifugation at 3365 rcf, resuspended in fresh growth media, and used to inoculate three 1-liter cultures in 6-liter Erlenmeyer flasks. Cultures were grown at 30 °C, with shaking at 200 rpm in an orbital shaker until an optical density (OD₆₀₀) of ~0.5 was reached, at which point protein expression was induced by adding 250 µM isopropyl-β-D-thiogalactopyranoside (IPTG). The temperature was then lowered to 16 °C, and cultures were incubated overnight. Cells were harvested by centrifugation at 4000 rcf for 10 min at 4 °C, and the pellets were resuspended in ice-cold PBS followed by cell lyses by sonication (QSonica sonicator, Newtown, CT) using 10 s on / 40 s off cycle at 75% amplitude for a total on-time of 3 min. To precipitate nucleic acid contaminants in the lysate, 10% polyethyleneimine (PEI) (Millipore Sigma, Burlington, MA) was added and the lysate was clarified by centrifugation at 23,000 rcf, 4 °C for 10 min. The PEI precipitation step was repeated 2–3 additional times until the cold lysate was clear with minimal turbidity. The supernatant was retained for Inverse Transition Cycling (ITC)-based purification, exploiting the phase behavior of ELP^19,27. Purification was initiated by raising the temperature of the clarified lysate to 25 °C and adding 0.2 M ammonium sulfate ((NH₄)₂SO₄; Millipore Sigma) to induce the ELP phase transition. The resulting coacervate suspension was centrifuged at 25 °C for 15 min at 23,000 rcf (“hot spin”), and the pellet containing aggregated ELP fusion protein was collected. This pellet was then resuspended in cold PBS and gently rotated overnight at 4 °C (R4045 RotoBot Programmable Rotator, Benchmark Scientific, Sayreville, NJ). Insoluble debris was removed by centrifugation at 4 °C for 10 min at 23,000 rcf (“cold spin”). This ITC cycle—consisting of salt-induced coacervation, centrifugation at elevated temperature, resolubilization at 4 °C, and clarification—was repeated twice more to ensure high purity of the fusion proteins. The final protein preparation was analyzed by SDS-PAGE stained with Coomassie Blue to confirm expected molecular weight and purity.

Expression and purification of (His)₆-ELP-TEV protease

The (His)₆-ELP-TEV protease was expressed using a similar protocol as GEF, but after harvest, the cells were resuspended in equilibration buffer (EQ = 20 mM Tris-HCl pH 8, 200 mM NaCl, 10 mM imidazole). Lysis was performed by sonication on ice (10 s on/40 s off, 75% amplitude, 3 min total) followed by centrifugation. The clarified lysate was applied to a HisPur Cobalt Resin (Thermofisher) column equilibrated with EQ buffer for purification by immobilized metal affinity chromatography. The resin was washed once with EQ buffer followed by wash buffer (20 mM Tris-HCl pH 8, 200 mM NaCl, 25 mM imidazole) until no further protein was detected by the absorbance at 280 nm (A₂₈₀). Elution was carried out using 200 mM imidazole in the EQ buffer and the purity of elution fractions were analyzed by SDS-PAGE. The pure fractions were combined and dialyzed into 50 mM Tris–HCl (pH 8.0), 200 mM NaCl, 0.5 mM EDTA for storage at −20 °C.

TEV cleavage reaction and purification of GLP1-ELP-FGF21 (GEF)

Cleavage of the fusion protein was carried out by mixing MSKGPG-tev-GLP1-ELP-FGF21 with (His)₆-ELP-TEV protease in 50 mM Tris–HCl (pH 8.0), 200 mM NaCl, 0.5 mM EDTA, 3 mM glutathione, at a protease to substrate (GEF) molar ratio of ~1:12. The reaction was incubated overnight at 4 °C with gentle mixing and completion of the reaction was monitored by SDS-PAGE. Post-cleavage, the TEV reaction buffer was exchanged with 20 mM Tris (pH 8) and the mixture containing cleaved GEF and his-ELP-TEV was passed through cobalt resin to obtain pure GEF in the flow through. Final purification was achieved by anion exchange chromatography where the sample was applied to a HiTrap™ Q HP column (Cytiva) equilibrated with 20 mM Tris-HCl (pH 8.0), and elution was performed using a linear gradient up to 50% of a high-salt buffer (20 mM Tris-HCl, 1 M NaCl, pH 8) on an AKTA protein purification system (Cytiva). Peak fractions containing pure GEF were confirmed by SDS-PAGE and A_260/280 ratio measurements, pooled, concentrated in PBS using ultracentrifugal filtration (30 kDa MWCO Amicon, Millipore Sigma), and stored at –20 °C.

Endotoxin removal and quantification

To minimize potential immune response in downstream applications, the recombinant GEF was subjected to endotoxin clearance using Acrodisc syringe filters (Pall Corporation, Port Washington, NY). The residual endotoxin levels were quantified using the Endosafe Nexgen-PTS system (Charles River Laboratories, Wilmington, MA) and were below the FDA limit of 5 EU per kg mouse body weight in all GEF samples injected into mice.

In vitro characterization of GEF

Thermal phase behavior

The protein was resuspended in 1× PBS at 200 µM concentration at 4 °C, and a serial dilution in PBS was performed resulting in 200, 50, 25, 12.5 µM as the concentrations to be tested. The optical density (OD) of the solution was monitored at 600 nm as a function of temperature on a temperature-controlled UV-vis spectrophotometer (Cary 300 Bio, Varian instruments). Starting at 20 °C, the temperature of the samples in the cuvette was increased at a rate of 0.3 °C/min until ~48 °C and the absorbance data was recorded at each 0.6 °C intervals. A sharp increase in the OD with temperature is indicative of the phase transition and the temperature at the inflection point of the optical density is defined as the T_t. The OD vs temperature was plotted and the temperature at the inflection point that defines the T_t was determined by finding the maximum of the first derivative of the OD versus temperature using GraphPad Prism software.

Formation and reversibility of coacervate droplets by light microscopy

The GEF protein was resuspended in PBS at 200 µM. For imaging, a 3 µL drop of the sample was placed on a glass slide with double-sided sticky tape lining the sample area to allow sufficient space for phase separation to occur between the slide and coverslip. The sample was imaged at 20× magnification on a Zeiss microscope (Axio Imager.D2m) with a custom heating insert, at a temperature above the T_t (35 °C) of the fusion protein and the temperature was then decreased to below T_t (25 °C) to visualize the reversibility of the phase transition.

Quantification of GLP-1R activation by cAMP-responsive luciferase assay

To determine the activity of GLP-1 in the fusion protein, a HEK293 (RRID: CVCL_0045) cell line stably expressing the GLP-1 receptor and a cAMP-inducible luciferase reporter gene was used. The cell line was confirmed to be contamination free by culturing and passaging them at least two times before starting the assay. For the assay, cells were plated at 1 × 10⁵ cells/cm² in 96-well plates and incubated overnight. In parallel, GEF was pretreated with dipeptidyl peptidase-4 (DPP4, ProSpec-Tany) at a 1:500 molar ratio (DPP4:GLP-1) and incubated overnight at 4 °C to cleave the leader peptide and reveal the biologically active N-terminus. The following morning, cell culture media were replaced with an induction buffer containing 129 mM NaCl, 4.8 mM KCl, 1.2 mM MgSO₄, 1.2 mM KH₂PO₄, 2.5 mM CaCl₂, 5 mM NaHCO₃, 10 mM HEPES, 0.5% BSA, and 50 µM IBMX (3-isobutyl-1-methylxanthine). Cells were then treated with serial dilutions of GLP-1R agonist for 5 h. After incubation, media were replaced with Bright-Glo luciferase substrate (Promega, Madison, WI), and the luminescence was measured using a Victor X3 plate reader (PerkinElmer). Signals were normalized against vehicle-treated controls, and dose-response curves were fit using GraphPad Prism 8 to calculate EC₅₀ values via a three-parameter logistic fit.

Assessment of FGF21 bioactivity via ERK1/2 phosphorylation

The in vitro activity of FGF21 in GEF was evaluated by measuring the phosphorylation of extracellular signal–regulated kinases 1/2 (ERK1/2) in a HEK293 (RRID: CVCL_0045) cell line engineered to stably express murine FGFR1 and β-Klotho²⁰. The cell line was confirmed to be contamination free by culturing and passaging them at least two times before starting the assay. For the assay, cells were seeded at a density of 5 × 10⁴ cells/cm² and allowed to adhere overnight. Following a 6-h period of serum deprivation, cells were stimulated with a concentration gradient of either the fusion protein or recombinant mouse FGF21 (ProSpec-Tany, East Brunswick, NJ) for 5 min. Post-treatment, cell lysates were analyzed for levels of phosphorylated and total ERK1/2 using the AlphaLISA SureFire Ultra assay kits (PerkinElmer). Signal quantification was performed with an EnSpire Alpha plate reader (PerkinElmer). Phosphorylated ERK1/2 values were normalized to total ERK1/2, and the resulting dose-response data were analyzed using GraphPad Prism 8 software (GraphPad Software, La Jolla, CA) to derive EC₅₀ values via a three-parameter logistic fit.

Assessment of GEF for hepatocyte proliferation

AML12 (RRID: CVCL_0140) cells—a healthy murine hepatocyte cell line—were purchased from ATCC for the in vitro proliferation assay and cultured as per supplier’s instructions. For the assay, 90 µL of cell suspension, corresponding to 5 × 10³ cells/well, were plated in a 96-well format a day before the assay. Cells were treated with 4 µM of GEF or PBS control in 30 µL media and were incubated for 4, 24, and 48 h. 20 µL of CellTiter 96 AQueous (Promega) reagent was added to each well and incubated for 1 h before measurement of the absorbance at 490 nm. The assay measures the reduction of tetrazolium reagent by metabolically active cells to determine cell viability at each time point after the treatment.

Therapeutic efficacy of the GEF in a model of advanced MASH

Animals

24-week adult male C57Bl6/J were purchased from the Jackson laboratory (#000664, Jackson Laboratory, Bar Harbor, ME) and were maintained in a temperature-controlled, pathogen-free room on 12-h light and dark cycles with ad libitum access to water and diet. All mice (n = 18) were fed a choline-deficient L-amino acid defined high-fat diet (CDA-HFD, A06071302, Research Diet, New Brunswick, NJ) for 16 weeks. The mice received s.c. injection of either vehicle control (PBS, Veh; n = 9) or GEF at a 750 nmol/kg dose (200 µM, GEF; n = 9) once a week for the last 4-weeks. At the start of week 17, mice were euthanized for whole tissue harvest. Slices of liver were formalin-fixed for paraffin embedding, and the remainder were snap frozen in liquid nitrogen for RNA and protein analysis. The blood samples were collected in Microvette 500 (Sarstedt Inc, Newton, NC) following the manufacturer’s instructions—centrifuged at 10,000 rcf for 10 min at room temperature and supernatants were transferred to a fresh tube for analysis. Animal care and diet procedures were conducted in compliance with The Duke Animal Care and Use Program with an approved protocol by The Institutional Animal Care and Use Committee (IACUC – protocol #A200-21-09). All animal experiments were conducted in compliance with ARRIVE guidelines and those set forth in the “Guide for the Care and Use of Laboratory Animals” as published by the National Research Council.

Serum analysis

Measurements in mouse serum for glucose, total cholesterol, HDL-cholesterol, and triglycerides were performed on a Beckman-Coulter DxC 600 clinical analyzer using reagents also from Beckman (Brea, CA). An immunoassay for mouse insulin was run using a kit and QuickPlex imager from Meso Scale Discovery (Rockville, MD).

qRT-PCR

Total RNA was extracted from whole liver chips using TRIzol (LifeTechnologies, Carlsbad, CA) according to the manufacturer’s instructions. First-strand cDNA was prepared by reverse transcription, using 5 µg of DNA-free RNA as template for the High-Capacity cDNA Reverse Transcription kit (Applied Biosystems). For qRT-PCR, 1% of the first-strand reaction was amplified using the StepOne Plus Real-Time PCR Platform (ABI/Life Technologies) and specific intron-spanning oligonucleotide primers for target sequences, as well as the ribosomal S9 housekeeping gene as reference control. qRT-PCR parameters were as follows: denaturing at 95 °C for 3 min, followed by 40 cycles of denaturing at 95 °C for 10 s and annealing/extension at the optimal primer temperature for 60 s. Threshold cycles (Ct) and product melt temperature were automatically calculated by the StepOne Plus Real-Time Detection System. Target gene levels are presented as a ratio to levels detected in the corresponding control tissue or cells according to the ΔΔCt method. Primer sequences are listed in Table 1.

Table 1.

Primer sequences

Gene name	Sequence
TGF-β	5ʹ-TTG CCC TCT ACA ACC AAC ACA A
	5ʹ-GGC TTG CGA CCC ACG TAG TA
TNF-α	5ʹ-CGT CAG CCG ATT TGC TAT CT
	5ʹ-CGG ACT CCG CAA AGT CTA AG
IL-6	5ʹ-GAG CCC ACC AAG AAC GAT AG
	5ʹ-TCC ACG ATT TCC CAG AGA AC
Il-18	5ʹ-CAG GCC TGA CAT CTT CTG CAA
	5ʹ-TCT GAC ATG GCA GCC ATT GT
IL-33	5ʹ-TCC ACG GGA TTC TAG GAA GA
	5ʹ-GAG GCA GGA GAC TGT GTT AAA
HIF-1α	5ʹ-CTC ATC CAA GGA GCC TTA ACC
	5ʹ-TTC GCT TCC TCT GAG CAT TC
HNF4 α	5ʹ-TCA ACG ACC GGC AGT ACG AC
	5ʹ-CTG GCA GAC CCT CCG AGA AG
AFP	5ʹ-CCG AGG AGG AAG TGA ACA AA
	5ʹ-GGC TTT CTA AAC ACC CAT CG
Cyc E	5ʹ-CAG AGC AGC GAG CAG GAG C
	5ʹ-GCA GCT GCT TCC ACA CCA CT
S9	5ʹ-GGG CCT GAA GAT TGA GGA TT
	5ʹ-CGG GCA TGG TGA ATA GAT TT

Protein quantification by ELISA

For the detection of IL-6, the whole liver tissue was lysed by RIPA buffer with proteinase inhibitor (P8340, Millipore Sigma). The 40 μg of lysate protein was detected for IL-6 protein by IL-6 ELISA kit (KMC0061, Life tech., Frederick, MD) following the manufacturer’s instructions.

Immunohistochemistry and quantitation

Liver tissues were fresh frozen or fixed in formalin and embedded in paraffin blocks using standard methods. Lipid accumulation in liver was evaluated by Oil Red O (Cat# O0625-25G, Sigma-Aldrich, St. Louis, MO). Briefly, fresh tissue samples were fixed with 10% formalin, stained with Oil Red O (10 min), and counterstained with hematoxylin (2 min). Results were examined by light microscopy. Immunohistochemistry was used to detect the target proteins using the following antibodies: Rabbit-anti-Ki67 (ab16667, Abcam, Cambridge, MA). Secondary antibodies were HRP-conjugated anti-rabbit (K4003, Agilent, Santa Clara, CA). Blocking and chromogenic detection were performed using the Envision System (Agilent) according to the manufacturer’s protocol, with DAB substrate (K3466, Agilent). Tissue sections were counterstained with Harris Hematoxylin (Leica, Richmond, IL). Negative controls included liver specimens exposed to 1% bovine serum albumin instead of the respective primary antibodies. The numbers of Ki67 immuno-reactive nuclei were quantified by counting 6 randomly chosen, 20x fields per section per mouse.

Statistics

All data is expressed as mean ± SEM. Statistical analysis was performed using Student’s t test as indicated. Differences with p ≤ 0.05 were considered statistically significant.

Results and discussion

Design, production, and in vitro functional validation of a unimolecular GLP-1–ELP–FGF21 dual agonist

To generate a long-acting dual agonist capable of activating both GLP-1 and FGF21 pathways, we designed a single fusion protein incorporating active GLP-1 at the N-terminus, followed by an elastin-like polypeptide (ELP) linker, and FGF21 at the C-terminus—hereafter referred to as GEF (GLP1-ELP-FGF21). We fused the C-terminus of GLP-1 to the ELP and the C-terminus of ELP was fused to FGF21 at its N-terminus terminus because this configuration preserved the receptor-binding capabilities of each hormone, while the linear design facilitated efficient production of the fusion protein in a bacterial expression system¹⁵. To enhance translational efficiency in E. coli, a leader sequence (MSKGPG) was added upstream of GLP-1^23,28. Because GLP-1 requires a free N-terminus for bioactivity, a Tobacco Etch Virus (TEV) protease recognition site was inserted between the leader and the GEF sequence²⁹. Following protein purification, the leader was cleaved by TEV protease, leaving a glycine residue. The remaining Gly-Ala (GA) dipeptide then serves as a substrate for endogenous Dipeptidyl peptidase 4 (DPP4), which cleaves after alanine, yielding a scarless and bioactive N-terminus on GEF³⁰ (Fig. 1A). This approach allowed for high-yield production of the fully functional fusion construct and the purified protein exhibited the expected molecular weight of 72 kDa, as confirmed by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 1B), indicating successful expression and purification of the GEF fusion protein without major degradation products.

Fig. 1. Production and in vitro characterization of the unimolecular GLP1-ELP-FGF21 (GEF) dual agonist.

a Schematics of the fusion protein with a leader sequence at N-terminus for better translation yield, which is then cleaved-off using TEV Protease to produce the final fusion protein composed of active GLP-1 at the N-terminus and FGF21 at the C-terminus. b SDS-PAGE gel showing purified final 72 kDa GEF protein. c Thermal phase behavior of GEF examined by recording absorbance of the protein while heating it at a fixed rate. The sharp increase in absorbance on the heating curve shows transition temperature (T_t) of the fusion protein which is inversely proportional to the protein concentration. d Micrograph of GEF above its T_t showing protein liquid droplets (left) that dissolves on cooling the protein below its T_t (right). This is visual representation for the reversible phase transition behavior of GEF. e, f GEF functions through both the GLP-1R and FGF21R/β-klotho pathways in vitro. e GLP-1 activity of the GEF was quantified by measuring cAMP production in HEK293 cells stably expressing the GLP-1R and the cAMP-inducible luciferase reporter. A potent GLP-1R agonist—Exendin-4—was used as a positive control for this in vitro assay. f FGF21 activity of GEF was quantified by measuring ERK1/2 phosphorylation in HEK293 cells stably expressing FGFR1& β-klotho, and by normalizing phospho-ERK1/2 to total ERK1/2. Commercially available mouse FGF21 was used as the positive control. Data is presented as Mean ± SEM, n = 3.

A critical design element of GEF is the inclusion of the ELP linker, which beyond linking the two proteins, imparts temperature-sensitive phase behavior to the fusion protein. The ELP sequence in the GEF protein was strategically selected for the fusion protein to exhibit a transition temperature (T_t) below body temperature. The thermal phase behavior of GEF was characterized by measuring its absorbance while gradually heating the protein solution. The temperature at which a sharp increase in turbidity was observed is defined as the T_t, and is consistent with the LCST phase behavior of ELPs³¹. We measured the T_t—defined as the temperature at the inflection point of the absorbance versus temperature curves—as a function of GEF concentration, and observed that the T_t increased with decreasing protein concentration, which is consistent with the behavior of ELPs and their fusion proteins in the dilute and semi-dilute regime of their phase diagram^19,20,31 (Fig. 1C). Brightfield microscopy revealed that GEF undergoes phase separation into two immiscible phases—a dense phase of micron-size liquid droplets enriched in GEF and a dilute phase depleted in GEF—above its T_t that readily dissolve into single phase upon cooling the protein below its T_t (Fig. 1D). This phase transition behavior imparted by the ELP linker to GEF creates a depot upon subcutaneous (s.c.) administration but also enhances systemic circulation half-life by reducing renal clearance because of the increased molecular weight provided by the size of the ELP linker—two pharmacokinetic advantages that are critical for therapeutic efficacy of protein-based biologics.

To verify that both arms of the fusion protein retain bioactivity, we carried out pathway-specific in vitro cell-based activity assays. We used a HEK293 cell line that stably expresses GLP-1 receptor (GLP-1R) and a cAMP-responsive luciferase reporter to test the activity of the GLP-1 arm of the dual agonist³². GEF demonstrated robust activation of the GLP-1R, as seen by the dose-dependent increase in luminescence levels corresponding to the intracellular cAMP level, with an EC₅₀ of 2.2 ± 0.3 nM (Fig. 1E). We ran another potent GLP-1R agonist—Exendin-4—in the same assay as a positive control, and as expected, the activity of the GEF fusion protein was lower compared to Exendin-1 due to the steric hinderance posed by the large molecule—an effect we have previously seen with other ELP fusion proteins^18,32. Separately, to test the activity of the FGF21 arm of the dual agonist, we utilized an engineered HEK293 cell line co-expressing FGFR1 and β-Klotho—the canonical receptor complex for FGF21²⁸. This assay system was selected based on extensive prior literature establishing FGFR1/β-klotho as the dominant functional receptor complex mediating FGF21’s metabolic effects in vivo³³. Using this platform, GEF activated the FGF21 signaling axis in a dose-dependent manner²⁰, as shown by phosphorylation of ERK1/2 with an EC₅₀ of 10 ± 4 nM (Fig. 1F). The signaling magnitude of GEF was only 10-fold lower as compared to the receptor activation by recombinant mouse FGF21 consistent with the effect of steric hindrance and confirms that the C-terminal FGF21 domain in GEF is correctly folded and functional. Together, these in vitro findings establish that GEF is a bifunctional fusion protein with bioactivity for both GLP-1 and FGF21 receptors and that it exhibits phase transition behavior typical of ELPs.

GEF treatment reduces liver mass and improves systemic metabolic parameters in a murine model of advanced MASH

To evaluate the therapeutic efficacy of the unimolecular dual agonist in a model of advanced MASH, we employed a chronic dietary induction protocol using 24-week-old C57Bl/6J mice. Animals were placed on a choline-deficient, high-fat diet (CDA-HFD) for 16 weeks, a well-established model that drives steatohepatitis and hepatic fibrosis in the absence of obesity. After 12 weeks of diet-induced liver injury, mice were randomized into two groups and administered either saline or GEF at a dose of 750 nmol/kg once weekly by s.c. injection over the final 4 weeks of the diet period (Fig. 2A). Body weight was recorded daily, and animals were euthanized at week 17 for liver and serum collection. Despite the advanced disease state at the onset of treatment, GEF administration led to a consistent reduction in body weight with each injection over the period of treatment, whereas saline-treated controls maintained their weight (Fig. 2B). This reduction in weight occurred without any observable signs of toxicity or stress and was driven by the central and peripheral metabolic effects of GLP-1R activation, as has been well documented in previous studies¹⁵. The observation that GEF retains the anorectic and weight-lowering activity of GLP-1 in vivo further validates the functional integrity of the GLP-1 moiety in the fusion construct.

Fig. 2. In vivo experimental design and effects of the dual agonist treatment on body weight and liver weight of the mice.

a 24-week-old C57Bl6/J mice were fed CDA-HFD diet for a total 16-weeks and after 12-weeks, mice were injected s.c. with either GEF at 750 nmol/kg dose or saline as a negative control (n = 9), weekly for four weeks. Body weight was measured daily, and mice were euthanized to harvest their liver and serum after 16-weeks (figure created in BioRender). b Percent change in body weight after treatment, which was calculated for each mouse in comparison to its body weight the day before treatment started. Data is presented as mean ± SEM for each treatment group. c Percent liver weight normalized to the body weight of the mouse on the day of euthanization. Each data point represents an individual mouse in each group. Data was analyzed for statistical significance with an unpaired t-test. **P < 0.01.

In addition to reducing body weight, GEF treatment had a marked effect on liver mass. The liver-to-body weight ratio—an indirect indicator of hepatic steatosis and inflammation—was significantly lower in GEF-treated mice compared to controls at the end of the 16-week study (Fig. 2C). The observed reduction in relative liver weight likely reflects the combined effects of FGF21-mediated improvements in lipid metabolism and GLP-1-driven modulation of systemic energy balance³⁴. The observed reductions in body and liver weight after just four weeks of treatment demonstrate that GEF produces sustained metabolic improvements and underscore the potential for this dual agonist to modulate key disease features in a mouse model of diet-induced MASH with advanced liver injury.

To further evaluate the potential of GEF on systemic metabolism, we performed serum biomarker analysis at the end of the 16-week study period by quantifying the circulating levels of total cholesterol, high density lipoprotein (HDL), insulin, and glucose in each animal. Mice treated with GEF exhibited significantly lower levels of total cholesterol compared to the saline controls (Fig. 3A), suggesting improved lipid handling and reduced hepatic lipid accumulation. Interestingly, HDL cholesterol levels were significantly higher in the GEF group, which is often considered cardioprotective³⁵ (Fig. 3B). GEF treatment also produced notable effects on glucose homeostasis, as serum insulin levels were significantly higher, and glucose levels were significantly lower in GEF-treated animals as compared to the controls (Fig. 3C, D). The combined actions of GLP-1 and FGF21 within a single molecular entity appear to act cooperatively in reducing lipid burden, improving glycemic control, and restoring endocrine balance. These metabolic benefits provide a mechanistic basis for the observed reductions in liver pathology and body weight and support the therapeutic promise of GEF for multifactorial diseases like MASH, where multiple physiological pathways must be targeted simultaneously.

Fig. 3. End point serum analysis revealed benefits of the dual agonist in improving metabolic parameters.

a Total cholesterol, b HDL, c insulin, and d glucose levels in serum. Each data point represents an individual mouse in each group. Data was analyzed for statistical significance with an unpaired t test. *P < 0.05, **P < 0.01.

GEF attenuates hepatic inflammatory signaling in advanced MASH

To investigate the impact of GEF treatment on hepatic inflammation and fibrosis—two hallmarks of MASH progression—we analyzed the expression of pro-fibrotic and pro-inflammatory genes in liver tissue using quantitative reverse transcription polymerase chain reaction (qRT-PCR). Gene expression analysis revealed that TGF-β—a central mediator of fibrogenesis and hepatic stellate cell activation—was significantly downregulated in livers of GEF-treated mice compared to saline controls (Fig. 4A). This reduction is consistent with attenuation of fibrogenic signaling pathways and aligns with prior reports demonstrating that FGF21 signaling can influence molecular processes relevant to fibrosis progression, including suppression of pro-fibrotic cytokine expression and modulation of stellate cell activation, particularly within a metabolically improved hepatic environment^36–38.

Fig. 4. Liver analysis by RT-PCR and ELISA revealed benefits of the dual agonist in modestly reducing liver fibrosis and inflammation.

mRNA levels of a TGF-β, b TNFα, c IL-6, d IL-18, e IL-33, and f HIF1α in comparison to the ribosomal S9 housekeeping gene as reference control. g IL-6 protein levels in liver as determined by ELISA. Each data point represents an individual mouse in each group and data was analyzed for statistical significance with an unpaired t test. *P < 0.05, **P < 0.01.

In parallel, transcripts encoding inflammatory cytokines—including TNFα, IL-6, IL-18, and IL-33—were reduced in the GEF-treated group as compared to the saline controls (Fig. 4B–E). These markers are associated with hepatic macrophage activation, innate immune signaling, and inflammatory stress that contribute to liver injury and fibrotic remodeling in MASH³⁹. The coordinated downregulation of multiple inflammatory mediators suggests that GEF broadly dampens hepatic inflammatory signaling, likely reflecting the combined effects of improved systemic metabolic status and local hepatoprotective signaling.

To confirm that transcriptional changes translated to protein-level effects, we measured IL-6 protein concentrations in liver lysates by ELISA. Consistent with the qRT-PCR results, IL-6 protein levels were significantly reduced in GEF-treated animals (Fig. 4G), confirming that the transcriptional changes translate into protein, and reinforcing the anti-inflammatory effect of the treatment. Additionally, expression of HIF1α—a transcription factor upregulated under hypoxic and inflammatory conditions and implicated in fibrogenic progression —was modestly decreased in the livers of GEF-treated animals (Fig. 4F). This downregulation further supports the notion that GEF improves the hepatic microenvironment, potentially reducing oxygen stress and limiting downstream activation of fibrotic pathways.

Beyond mitigating inflammation and fibrosis, a critical goal in the treatment of MASH is to promote hepatic repair and restore normal liver function. To evaluate whether GEF exerts pro-regenerative or hepatoprotective effects, we analyzed the expression of key genes involved in hepatocyte identity, liver regeneration, and cell cycle progression by qRT-PCR.

We first assessed hepatic mRNA levels of HNF4α—a master transcription factor essential for maintaining hepatocyte differentiation and liver-specific gene expression. GEF-treated mice showed a significant increase in HNF4α expression compared to controls (Fig. 5A), suggesting that the dual agonist may help preserve or restore hepatocellular identity in the diseased liver⁴⁰. This is particularly important in the context of MASH, where hepatocyte de-differentiation and loss of HNF4α expression are associated with worsening pathology and impaired liver function⁴¹. Additionally, Alpha-fetoprotein (AFP)—a fetal liver marker that is often re-expressed during hepatic injury and regeneration, was significantly decreased in the GEF group (Fig. 5B). Lower AFP suggests less liver injury as there is less stimulus to regenerate. The reduced AFP is also consistent with higher HNF4α as some of AFP positive cells are thought to arise from more mature HNF4α positive hepatocytes that de-differentiate to re-acquire a progenitor-like state⁴². Interestingly, Cyclin E—a critical regulator of the G1/S transition in the cell cycle—levels were upregulated (Fig. 5C). Elevated Cyclin E transcript levels suggest increased hepatocyte proliferation, potentially contributing to tissue renewal and functional recovery⁴³.

Fig. 5. Benefits of the dual agonist on overall liver health and in vitro healthy hepatocyte proliferation.

mRNA levels of a HNF4α, b AFP, and c Cyclin E in comparison to the ribosomal S9 housekeeping gene as reference control. Each data point represents an individual mouse in each group. d in vitro cell proliferation of AML12 cells over the course of 48 h after incubation with GEF or negative control PBS. Data was analyzed for statistical significance with an unpaired t-test. *P < 0.05, **P < 0.01, ***P < 0.001.

GEF reduces hepatic lipid accumulation and increases hepatocyte proliferation

To directly assess the pro-proliferative effects of GEF on hepatocytes, we performed an in vitro proliferation assay using AML12 cells—a healthy murine hepatocyte cell line. Cells incubated with GEF exhibited significantly greater proliferation over a 48-h period compared to PBS-treated controls (Fig. 5D). This finding corroborates the gene expression data and indicates that GEF can stimulate hepatocyte growth in a non-inflammatory, regenerative context. Taken together, these results point to benefits of GEF treatment in the promotion of liver repair and regeneration. By activating cell cycle genes, and enhancing the expression hepatocyte identity markers, GEF appears to support not only the resolution of injury but also the recovery of functional liver tissue.

To complement our molecular and functional analyses, we performed histological evaluation of liver tissues to directly visualize the effects of GEF treatment on hepatocyte proliferation and lipid accumulation. Ki67 immunohistochemical staining was used to assess hepatocyte proliferation in situ. Representative liver sections from GEF-treated mice (Fig. 6B) showed a markedly increased number of Ki67-positive nuclei compared to those from saline-treated controls (Fig. 6A). We quantified this increase by counting the positive nuclei in 6 randomly chosen, 20× fields per section for each mouse and then plotting the average number of Ki67-positive nuclei in GEF treated group as compared to the control (Fig. 6C). This significantly increased Ki67-positive nuclei evidence supports our earlier findings of elevated Cyclin E expression and enhanced in vitro AML12 proliferation, indicating that GEF promotes active hepatocyte cell cycling and tissue renewal at the cellular level^44,45. Importantly, this increase in proliferation occurred in a background of reduced inflammation and fibrosis, suggesting a favorable regenerative rather than pathological proliferation.

Fig. 6. Liver tissue histology revealed benefits of the dual agonist in improving hepatocyte proliferation and reducing lipid accumulation in the liver.

Ki67 stained micrograph of a control, and b GEF. c Average number of Ki67 positive cells in GEF treated group as compared to the saline control. Oil red stained micrograph of d control, and e GEF. f Averaged percent neutral lipid positive area in GEF treated group as compared to the saline control. Histology results were quantified from 6 randomly chosen, 20× fields per section for each mouse and data is plotted as Mean ± SEM. The scale bar is 100 μm in all the micrographs. Data was analyzed for statistical significance with an unpaired t test. *P < 0.05, ****P < 0.0001.

To assess hepatic lipid burden, we performed Oil Red O staining, which selectively labels neutral lipids within tissue sections. Liver samples from GEF-treated mice (Fig. 6E) exhibited a notable reduction in lipid content, with visibly fewer and smaller lipid droplets as compared to saline-treated animals (Fig. 6D) that displayed dense and widespread lipid droplet accumulation. To quantify this effect, we calculated the percent neutral lipid positive area by measuring the area occupied by red droplets relative to the total area of the micrograph for each mouse. The average percent neutral lipid positive area for the GEF treated mice was significantly lower as compared to the control group (Fig. 6F). This reduction in hepatic steatosis is consistent with the observed decrease in serum cholesterol and glucose and likely reflects enhanced lipid utilization and decreased hepatic lipogenesis mediated by FGF21 activity in the dual agonist. These histological results provide visual confirmation of the biochemical and molecular improvements induced by GEF therapy.

Conclusions

In this study, we present the rational design, characterization, and preclinical evaluation of a unimolecular dual agonist—GLP1-ELP-FGF21 (GEF)—that simultaneously activates GLP-1 and FGF21 signaling pathways. By leveraging an ELP linker, the dual agonist was engineered to possess not only receptor-targeted biological activity but also favorable pharmacokinetic properties through its reversible thermal phase behavior. This platform enables a single molecule to deliver multi-hormonal functionality with enhanced systemic stability. In a diet-induced model of advanced MASH with fibrosis, GEF treatment led to significant improvements in body weight, liver mass, and serum metabolic parameters, including cholesterol, insulin, and glucose. At the molecular level, GEF attenuated hepatic inflammation and fibrosis-associated pathways, as evidenced by reduced expression of pro-inflammatory cytokines and fibrotic markers. By concurrently modulating inflammatory and fibrogenic mediators at both the gene and protein levels, GEF impacts multiple key drivers of disease progression.

Histological and functional analyses further revealed increased hepatocyte proliferation and decreased lipid accumulation in liver tissue, suggesting that GEF not only slows disease progression but also supports hepatic repair. Together with extensive prior evidence demonstrating the complementary metabolic and hepatic effects of GLP-1 and FGF21 signaling, these findings highlight the therapeutic potential of dual GLP-1 and FGF21 activation in reversing liver pathology even at a late stage of disease, and provide a mechanistic rationale for continued development of this unimolecular strategy in chronic liver disease.

Looking forward, further development of GEF will benefit from extended pharmacodynamic studies, dose optimization, and safety profiling in larger animal models. In addition, mechanistic studies in primary human hepatocytes and MASH patient-derived organoids could provide translational insights into receptor dynamics and tissue-specific effects. Ultimately, this unimolecular platform may be adapted to include additional hormone modules or tailored to different disease stages, providing a versatile framework for next-generation therapies targeting multifactorial metabolic disorders.

Author contributions

P.S. and A.C. conceived and designed the research. P.S., S.H.O., C.G.P., J.T., H.C., D.L., and R.K.D. performed the experiments. P.S., S.H.O., C.A.G., A.M.D. contributed to experimental design. P.S., S.H.O., A.M.D. and A.C. interpreted the results and wrote the manuscript. All authors participated in the discussion of the data and commented on the manuscript.

Peer review

Peer review information

Communications Medicine thanks Alessandro Pocai, Gloria Alvarez-Sola and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Data availability

The raw data that support the findings of this study are available from the corresponding author upon reasonable request, and the source data underlying Figs. 1–6 are readily available in the Figshare repository at https://doi.org/10.6084/m9.figshare.30120508. Additionally, the uncropped version of the SDS-PAGE gel from Fig. 1 is provided as the supplementary information.

Competing interests

P.S., S.H.O., A.M.D. and A.C. have submitted invention disclosure for technology titled “A unimolecular GLP1 and FGF21 dual agonist for Non-Alcoholic Steatohepatitis”. All other authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s43856-026-01712-5.