Lung-targeted hepatocyte growth factor mRNA restores alveolar structure in experimental emphysema

Graphical abstract

Overview of the study. HGF: hepatocyte growth factor; c-Met: cellular mesenchymal–epithelial transition factor; AT1/2: alveolar type I/II cells; LNPs: lipid nanoparticles.

Abstract

Background

Emphysema, a major component of COPD characterised by progressive alveolar destruction, lacks effective medical therapies. Hepatocyte growth factor (HGF) possesses potent regenerative functions, but its therapeutic potential remains unrealised due to challenges in achieving targeted delivery and sustained lung expression.

Methods

We first assessed associations between HGF expression and emphysema severity using human datasets, lung tissue, and both elastase-induced and cigarette-smoke-induced murine models. We repurposed a clinical-stage SM102 lipid nanoparticles platform to deliver human HGF mRNA in murine models, evaluating therapeutic efficacy via intratracheal instillation in the elastase model. After optimising nebulisation, we assessed efficacy in the cigarette-smoke model. We investigated underlying mechanisms via single-cell RNA sequencing (scRNA-seq), which we validated in patient-derived lung organoids.

Results

HGF expression displayed a biphasic pattern across the emphysema spectrum, with upregulation in milder disease states and marked reduction in advanced emphysema. Intratracheal delivery of HGF mRNA lipid nanoparticles restored lung function and attenuated alveolar destruction in the elastase model. Nebulised delivery achieved efficient pulmonary distribution and demonstrated comparable therapeutic efficacy in the cigarette-smoke model, including improved lung function, reduced inflammation, and decreased apoptosis. scRNA-seq analysis detected enhanced alveolar type II (AT2) cell proliferation and differentiation in the elastase model and human organoids.

Conclusion

This study provides proof-of-concept evidence for a therapeutic strategy for emphysema. Using a clinical-stage lipid nanoparticle platform, we demonstrate that HGF mRNA therapy is effective via both direct instillation and optimised nebulisation, prompting structural and functional recovery by activating endogenous repair pathways in AT2 cells.

Shareable abstract

HGF mRNA therapy delivered by a clinical-stage LNP platform promotes alveolar repair and improves lung function in experimental emphysema, including via nebulisation, supporting a translational regenerative strategy for COPD https://bit.ly/3OucmTe

Introduction

Emphysema, a major pathological phenotype of COPD, is characterised by irreversible alveolar destruction and loss of lung elastic recoil [1]. Lacking medical therapies capable of restoring structural damage or facilitating alveolar regeneration [2, 3], and with invasive lung volume reduction procedures limited to small patient subsets [4], emphysema is a critical unmet need. Strategies to achieve efficient and sustained delivery of regenerative therapeutics to the lungs are urgently needed.

Emphysema progresses due to imbalance between tissue destruction and adequate repair. In susceptible individuals, inhaled irritants, including cigarette smoke, trigger sustained inflammation and protease–antiprotease imbalance, degrading extracellular matrix, destroying alveolar walls and inducing epithelial apoptosis [5, 6]. However, emphysema also progresses due to compromised repair processes, which become insufficient as disease severity progresses. Alveolar type II (AT2) cells, the principal epithelial progenitor cells, exhibit reduced proliferative capacity and impaired differentiation into alveolar type I (AT1) cells [7]. Supportive niches are also dysfunctional, with downregulation of key WNT/β-catenin regenerative pathways [5, 8]. This failure in re-epithelialisation points to the therapeutic promise of strategies to enhance AT2 cell function and restore epithelial integrity.

Hepatocyte growth factor (HGF) is pivotal for alveolar homeostasis and epithelial repair [9, 10]. Via its receptor, cellular mesenchymal–epithelial transition factor (c-Met), HGF promotes cellular proliferation, migration and survival [11, 12]. Intact HGF signalling is indispensable for alveolar regeneration, as conditional knockout of c-Met in AT2 cells displays impaired epithelial repair and emphysematous pathology [9]. However, despite compelling pre-clinical evidence, translating HGF into a therapy has been hampered by inadequate delivery methods. Current strategies, including cell therapies, viral vectors, and recombinant protein administration, are hindered by invasive delivery requirements, inadequate targeting efficiency, transient expression, and systemic toxicity [9, 13–16]. Developing a safe and efficient platform for sustained, lung-specific HGF expression remains the central obstacle.

Achieving targeted lung delivery requires a safe and precisely controllable platform. mRNA-based therapy fits this need perfectly, allowing tunable protein expression free from the genomic integration risks of DNA-based therapies. However, its therapeutic potential rests on an effective delivery vehicle. Lipid nanoparticles are an attractive vehicle to protect against degradation and facilitate efficient cellular uptake and endosomal escape [17]. The success of mRNA lipid nanoparticles vaccines during the COVID-19 pandemic supports both their safety profile and therapeutic efficacy [18]. Although novel lipid nanoparticles are being explored for improved lung targeting, their clinical translation faces challenges in scalable manufacturing and safety re-evaluation [19, 20]. In contrast, repurposing established lipid nanoparticles platforms offers a compelling strategy to accelerate therapeutic development [21]. The SM102-based lipid nanoparticles platform, with its clinical safety profile and regulatory acceptance in systemic vaccination, provides an ideal starting point for developing a lung-targeted therapeutic [22].

We hypothesised that pulmonary delivery of HGF mRNA via the SM102 lipid nanoparticles platform could overcome delivery barriers and stimulate AT2 cell-mediated repair in emphysema. Here, we evaluate this novel therapeutic strategy in established emphysema models and human lung tissues.

Materials and methods

Detailed methodologies are described in the supplementary methods.

Study design

This study integrated bioinformatic analysis of human datasets, validation using clinical lung tissues, and therapeutic evaluation in two established mouse models of emphysema (elastase-induced and cigarette-smoke-induced). The objective was to assess the reparative effects of pulmonary delivery of HGF mRNA encapsulated in lipid nanoparticles. Long-term safety and repeat-dose tolerability were also assessed.

Ethical compliance

All procedures involving human participants and animals were approved by the institutional review boards and animal ethics committees of Peking University Third Hospital and Hebei Provincial People's Hospital. Written informed consent was obtained from all human participants.

Experimental models and treatment

The association between HGF and emphysema severity was first assessed in public Gene Expression Omnibus datasets and human lung tissues. Therapeutic efficacy was then evaluated in male C57BL/6J mice with experimentally induced emphysema. Mice were randomly allocated to receive intratracheal administration of PBS solution (vehicle control), firefly luciferase (Fluc) mRNA lipid nanoparticles (negative control), HGF mRNA lipid nanoparticles, or recombinant HGF (rhHGF) protein (positive control). The primary lipid nanoparticles platform used was the clinical-stage SM102 formulation.

Key analyses

Outcomes included lung function measurement (FlexiVent system), histological assessment of airspace enlargement (mean linear intercept), single-cell RNA sequencing (scRNA-seq) of human and murine lung cells, and functional validation in patient-derived lung organoids and vascular organoids.

Statistical analyses

Data are presented as mean±sd. Group comparisons were performed using unpaired t-tests, one-way ANOVA, or two-way ANOVA where appropriate, followed by post hoc multiple-comparison testing. A p-value <0.05 was considered statistically significant.

Results

Fibroblast-derived HGF exhibits stage-dependent biphasic regulation in COPD lungs

We first investigated lung expression of HGF and MET in COPD using publicly available human datasets. In the GSE27597 dataset, expression of both HGF and MET was significantly elevated in lung tissues from COPD patients compared to controls (figure 1a). After within-dataset z-score normalisation of GSE37768 and GSE69818, HGF expression increased from never-smokers to smokers and to COPD, but declined progressively with increasing Global Initiative for Chronic Obstructive Lung Disease (GOLD) stage (figure 1b). Congruently, human fibroblast MRC5 cells in the GSE252801 dataset stimulated by cigarette-smoke condensate exhibited markedly reduced HGF expression (supplementary figure S1a). We corroborated these findings in our murine models of emphysema induced by elastase (ELA) and cigarette smoke, in which lung tissue expression of Hgf was similarly reduced (figure 1e, f). However, neither HGF nor MET expression in peripheral blood differed between controls and COPD patients (GSE76705) (supplementary figure S1b).

FIGURE 1.

Stage-dependent regulation of hepatocyte growth factor (HGF) and cellular mesenchymal–epithelial transition factor (c-Met) expression in COPD lungs. a) Normalised mRNA expression of HGF and MET (mesenchymal–epithelial transition factor) in human lung tissues from GSE27597 (control: n=16; COPD: n=48). b) Integrated visualisation of HGF expression across GSE37768 (never-smokers: n=9; smokers: n=11; COPD: n=18) and GSE69818 (Global Initiative for Chronic Obstructive Lung Disease (GOLD) 1: n=11; GOLD 2 and 3: n=50; GOLD 4: n=9) following within-dataset z-score normalisation. c) Representative Western blot of HGF protein expression in lung tissues from our clinical cohort, including never-smokers (n=20), smokers without COPD (n=20), COPD without emphysema (n=10), and COPD with emphysema (n=10). Western blot bands shown in this figure are from three samples per group, and protein expression quantification was based on six samples per group. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) served as loading control. d) Quantification of HGF protein expression shown in c). e) Hgf mRNA levels (log₂ transformed) in lung tissues from mice following elastase (ELA) or PBS instillation (n=6 mice per group). f) Hgf mRNA levels (log₂ transformed) in lung tissues from mice exposed to cigarette smoke (CS) or fresh air (FA) (n=6 mice per group). g, h) Quantification of g) HGF and h) c-Met immunohistochemical staining scores across clinical groups (total n=60). All clinical cohort samples were included in the immunohistochemistry quantification. i) Representative multiplex immunofluorescence staining of HGF (red) and c-Met (red) in human lung tissues, co-stained with markers for mesenchymal (VIMENTIN, green), epithelial (EPCAM, white), endothelial (CD31, cyan), smooth muscle actin (α-SMA, magenta), and myeloid (CD68, yellow) cells. Scale bars=100 μm. For microarray datasets, data are presented as median (interquartile range) and differential gene expression analyses were performed using the Limma package in R. For comparisons between two groups, unpaired two-tailed t-tests were applied. For comparisons between multiple groups, one-way ANOVA followed by appropriate post hoc testing was performed. Data from the clinical cohort and murine models are presented as mean±sd, unless otherwise stated. Each dot represents an individual subject, sample or mouse. ns: not significant; *: p<0.05, **: p<0.01, ***: p<0.001.

For validation, we analysed histologically normal lung parenchyma from 60 patients undergoing clinically indicated pulmonary nodule resections, including (n=20) each never-smokers (all confirmed to be benign), smokers without COPD, and COPD patients, half without emphysema (clinical details are presented in supplementary table S3). Western blot demonstrated progressively increased HGF from never-smokers to smokers without COPD and COPD without emphysema, but a significant reduction in COPD with emphysema (figure 1c, d). Immunohistochemistry (IHC) showed the same pattern for HGF (figure 1g), and alveolar septal staining inversely correlated with emphysema severity as measured by mean linear intercept (MLI) (supplementary figure S1c, d). By region-resolved analysis, HGF displayed similar stage-dependent changes across alveolar, perivascular, and peribronchial stromal compartments (supplementary figure S1g, h). In contrast, c-Met staining increased in COPD without emphysema compared with never-smokers across alveolar, vascular and bronchial regions, but did not significantly decline in COPD with emphysema (figure 1h, supplementary figure S1i). Together, these findings establish biphasic, stage-dependent expression of endogenous HGF during COPD progression.

Multiplex immunofluorescence localised HGF predominantly to VIMENTIN⁺ mesenchymal cells within stromal regions, with minor endothelial or myeloid expression (figure 1i). c-Met was primarily detected in epithelial cells, with lesser endothelial or myeloid expression (figure 1i). Quantitative IHC showed that the overall fraction of HGF⁺ cells declined across disease stages (supplementary figure S1e), whereas intracellular HGF staining intensity within HGF⁺ cells increased in smokers, but decreased in emphysema (supplementary figure S1f).

Single-cell RNA sequencing of an independent human cohort (never-smoker, smoker and COPD with emphysema) confirmed enrichment of HGF transcripts in fibroblasts (supplementary figure S2a, b), whereas MET expression was predominantly observed in epithelial cells (supplementary figure S2c, d). Across groups, fibroblast HGF expression increased in smokers and declined in COPD with emphysema (supplementary figure S2e), while fibroblast proportion progressively decreased from never-smokers to COPD with emphysema (supplementary figure S2f). Gene module analysis revealed parallel stage-dependent alterations in mitogen-activated protein kinase (MAPK) pathway activity within fibroblasts (supplementary figure S2g), consistent with prior evidence of HGF synthesis regulation in lung fibroblasts [23]. Thus, the biphasic bulk pattern of HGF expression reflects stage-dependent changes in both fibroblast abundance and fibroblast HGF expression.

SM102 lipid nanoparticles mediate efficient and lung-specific mRNA delivery

To achieve lung-targeted HGF mRNA delivery, we evaluated lipid nanoparticles formulated with ionisable lipids possessing established clinical safety profiles. We selected SM102, ALC0315 and MC3 lipids to prepare mRNA-loaded lipid nanoparticles via microfluidic mixing (figure 2a). All three lipid nanoparticles formulations encapsulating Fluc mRNA exhibited narrow size distribution (polydispersity index <0.2) (figure 2b), high encapsulation efficiency (>95%) (figure 2c), mildly positive ζ potent­ial (∼12–15 mV) (figure 2d), and mean particle diameters between 64 and 84 nm (figure 2e).

FIGURE 2.

SM102 (1-octylnonyl8-[(2-hydroxyethyl)[6-oxo-6-(undecyloxy)hexyl]amino]-octanoate) lipid nanoparticles (LNPs) enable efficient and lung-targeted mRNA delivery. a) Schematic illustration of LNPs preparation via microfluidic mixing of the ethanol phase (containing ionisable lipid, DSPC (1,2-dioctadecanoyl-sn-glycero-3-phosphocholine), DMG-PEG2000 (1,2-dimyristoyl-rac-glycero-3-methoxypoly­ethylene glycol-2000), cholesterol) and aqueous phase (mRNA in sodium citrate buffer). b–e) Analysis of firefly luciferase (Fluc) mRNA encapsulated in SM102, ALC0315 (((4-hydroxybutyl)azanediyl)bis(hexane-6,1-diyl)bis(2-hexyldecanoate)), and MC3 ((6Z,9Z,28Z,31Z)-heptatriaconta-6,9,28,31-tetraen-19-yl 4-(dimethylamino)butanoate) LNPs (n=3 independent LNP preparations): b) polydispersity index (PDI); c) encapsulation efficiency; d) ζ potential; e) size distribution. f) Schematic of in vivo bioluminescence imaging workflow: intratracheal instillation of Fluc mRNA LNPs in mice, followed by d-luciferin administration and imaging of heart, lungs, liver, kidneys and spleen. g) Representative bioluminescence images of different organs at 24 h post-i.t. instillation of Fluc mRNA encapsulated in SM102, ALC0315, and MC3 LNPs. h) Quantification of total photon flux from lung bioluminescence in mice treated as in g) (n=4 mice per group). i) Transmission electron microscopy images of SM102 LNPs. Top: low magnification, scale bar=100 nm; bottom: high-magnification view of the boxed region, scale bar=50 nm. j–l) In vivo analysis of Cre-mediated recombination in Rosa26-LSL-Fluc-GFP (green fluorescent protein) reporter mice post-i.t. instillation of Cre mRNA SM102 LNPs: j) schematic; k) representative bioluminescence images of different organs; l) flow cytometric quantification of GFP⁺ cells across indicated pulmonary cell populations (n=6 mice per group). In all quantification panels, each dot represents an individual mouse. Data are presented as mean±sd.

Next, we assessed delivery performance following i.t. administration of lipid nanoparticles loaded with 30 µg Fluc mRNA. Bioluminescence imaging at 24 h revealed robust, lung-confined expression, without detectable peripheral organ off-target signal (figure 2f, g). SM102 lipid nanoparticles yielded a three- to 14-fold higher luminescence than ALC0315- or MC3-based formulations (figure 2g, h). Dose-ranging and kinetic analyses confirmed detectable expression at doses as low as 5 µg, with signal evident by 6 h and returning to baseline by 48 h (supplementary figure S3a–e). Transmission electron microscopy revealed spherical morphology with uniform size (figure 2i).

To map cellular tropism, we administered Cre mRNA-loaded SM102 lipid nanoparticles to Rosa26-LSL-Fluc-GFP reporter mice (figure 2j). Recombination, indicated by Fluc/GFP co-expression, was detected in multiple lung cell types (figure 2k, l). Flow cytometry showed that SM102 lipid nanoparticles preferentially transfected alveolar epithelial cells (∼33%), followed by airway epithelial cells (∼21%) and fibroblasts (∼15%) (figure 2l, supplementary figure S3f).

Optimised lung-targeted delivery of HGF mRNA lipid nanoparticles reverses emphysema pathophysiology

Using this SM102 lipid nanoparticles platform, we evaluated a strategy to deliver HGF mRNA to treat established emphysema. We designed an mRNA construct encoding human HGF, which shows conserved coding sequence with mice (supplementary figure S4), with optimised 5′ and 3′ untranslated regions to enhance protein expression. In healthy mice, after a single i.t. instillation of lipid nanoparticles containing 10 µg HGF mRNA, human HGF protein was detectable in lung homogenates at 24 h, declining to baseline by 72 h (supplementary figure S5a).

To determine a therapeutically effective dose, we induced emphysema using a single dose of ELA, with analysis at 4 weeks. During the final week, some groups received two i.t. instillations (72 h apart) of SM102 lipid nanoparticles encapsulating HGF mRNA at 5, 10 or 20 µg doses. Pulmonary function tests showed multiple parameters of markedly impaired lung function in emphysematous mice (ELA, PBS) compared to controls (PBS, PBS); treatment with 20 µg HGF mRNA significantly reversed these functional abnormalities, whereas lower doses did not (supplementary figure S6a–f). Because SM102 lipid nanoparticles containing 20 µg HGF mRNA also attenuated ELA-induced airspace enlargement (supplementary figure S6g), this dose was selected.

We next contrasted the efficacy of i.t.-delivered HGF mRNA lipid nanoparticles versus rhHGF protein in the ELA model (supplementary figure S5b) [9]. To exclude intrinsic nanoparticle effects, we used Fluc mRNA lipid nanoparticles, which produced lung function parameters indistinguishable from the ELA, PBS group (supplementary figures S5c–f and S7a, b). Although rhHGF protein elicited modest functional improvements, HGF mRNA lipid nanoparticles more robustly restored multiple parameters. Pressure–volume loops showed a rightward and upward shift in ELA, PBS group mice, which was corrected partially by rhHGF protein and more substantially by HGF mRNA lipid nanoparticles (supplementary figure S7c). HGF mRNA lipid nanoparticles also more significantly restored alveolar architecture (supplementary figure S5g, h).

To distinguish between cessation of disease progression versus alveolar repair, we included a comparison treatment-initiation group (ELA 3W, PBS). Pulmonary function testing parameters and MLI values did not differ significantly between ELA 3W and ELA 4W groups, indicating that alveolar destruction had largely plateaued by treatment initiation (supplementary figure S8a, b). Importantly, HGF mRNA lipid nanoparticles treatment significantly improved lung function and reduced MLI at week 4 beyond the week-3 baseline values (supplementary figure S8a, b). Thus, HGF mRNA lipid nanoparticles actively promote structural and functional recovery.

HGF mRNA lipid nanoparticles also significantly suppressed pulmonary mRNA and protein expression of interleukin (IL)-6, IL-1β, and tumour necrosis factor (TNF)-α compared to all other groups (supplementary figures S5i–k and S7d–f). Furthermore, HGF mRNA lipid nanoparticles increased parenchyma proportions of proliferating (MKI67⁺) cells and decreased apoptotic (terminal deoxynucleotidyl transferase dUTP nick-end labelling (TUNEL)⁺) cells more effectively than rhHGF protein (supplementary figures S5l, m and S7g), without evidence of systemic toxicity by complete blood counts and comprehensive metabolic panels (supplementary figure S9a–j).

Aerosolised HGF mRNA lipid nanoparticles ameliorate cigarette-smoke-induced emphysema

Next, we evaluated a clinically relevant aerosol delivery method for HGF mRNA lipid nanoparticles to mice. Using a microsprayer, aerosolised Fluc mRNA lipid nanoparticles generated robust, lung-specific luciferase expression, peaking at 24 h and returning to baseline by 72 h (figure 3a). Compared to i.t. instillation, this method enhanced signal intensity across all five pulmonary lobes (figure 3b, c). Pharmacokinetic analysis showed that aerosolised delivery of HGF mRNA lipid nanoparticles enhanced both the peak level and duration of HGF protein expression in murine lungs, compared to i.t. instillation, without HGF protein detection in serum (figure 3d, supplementary figure S10a).

FIGURE 3.

Aerosolised hepatocyte growth factor (HGF) mRNA lipid nanoparticles (LNPs) enhance pulmonary delivery and exert therapeutic effects in cigarette-smoke (CS)-induced emphysema. a–d) Firefly luciferase (Fluc) mRNA LNPs were administered by intratracheal spray to untreated mice, which were assayed by bioluminescence imaging 24, 48, and 72 h later. a) Schematic of workflow, and total bioluminescent flux in lungs (n=3 mice per time point); and representative bioluminescence images of different organs. b) Representative bioluminescence images of lung lobes at 24 h after i.t. instillation or i.t. spray of Fluc mRNA LNPs (n=5 mice per group). c) Quantification of the Fluc signal-positive region percentage and total bioluminescent flux in each lung lobe (cranial, middle, caudal, accessory, left) 24 h post-treatment (n=5 mice per group). d) Kinetic analysis of lung HGF protein concentration after i.t. instillation versus i.t. spray of HGF mRNA LNPs (n=5 mice per time point). e–j) Mice were exposed to CS or fresh air (FA) for 8 months as described in the supplementary methods; during the final week they were treated in five experimental groups: FA, PBS (grey); CS, PBS (blue); CS, Fluc mRNA LNPs (red); CS, HGF mRNA LNPs (yellow); CS, recombinant human (rh)HGF protein (green) (n=6 mice per group). e–h) Pulmonary function parameters in mice, including respiratory system e) elastance, f) compliance, g) quasi-static compliance, h) forced expiratory volume in 0.1 s (FEV_0.1)/forced vital capacity (FVC) ratio. i) Representative haematoxylin and eosin (H&E)-stained lung sections from experimental groups (n=6 mice per group). Scale bars=500 µm, inset=200 µm. j) Quantification of mean linear intercept (MLI) across experimental groups. In all quantification panels, each dot represents an individual mouse. Data are presented as mean±sd. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test. ns: nonsignificant. *: p<0.05, **: p<0.01, ***: p<0.001

Employing this dosing regimen in our established cigarette-smoke-induced COPD model (supplementary figure S10b), aerosolised HGF mRNA lipid nanoparticles significantly reversed pulmonary dysfunction (figure 3e–h, supplementary figure S10c–e), restored lung mechanics by pressure–volume analysis (supplementary figure S10f), and attenuated cigarette-smoke-induced alveolar destruction, septal rupture, and macrophage infiltration more effectively than rhHGF protein (figure 3i, j).

In addition, HGF mRNA lipid nanoparticles decreased total cell and neutrophil counts in bronchoalveolar lavage fluid (BALF) compared to cigarette-smoke, PBS and cigarette-smoke, Fluc mRNA lipid nanoparticle groups (supplementary figure S10g). Furthermore, HGF mRNA lipid nanoparticles significantly suppressed expression of pro-inflammatory cytokine (IL-6, IL-1β and TNF-α) mRNA in lung tissue and protein in BALF (supplementary figure S10h–m). HGF mRNA lipid nanoparticles increased proportions of MKI67⁺ proliferating cells and decreased TUNEL⁺ apoptotic cells (supplementary figure S10n–p).

HGF mRNA lipid nanoparticles drive epithelial remodelling in emphysema

To define the cellular mechanisms underlying HGF mRNA lipid nanoparticle-mediated lung repair, we utilised the ELA model, due to its pronounced and consistent alveolar destruction. scRNA-seq analysis indicated that HGF mRNA lipid nanoparticles significantly increased the proportion of epithelial cells relative to both the PBS, PBS and ELA, PBS groups (figure 4a, b). Ratios of observed-to-expected frequency of specific cell types confirmed that HGF mRNA lipid nanoparticle treatment markedly expanded epithelial populations (figure 4c). By Augur analysis, epithelial cells exhibited the most substantial transcriptional changes across all conditions, including comparisons reflecting both disease progression and therapeutic response (figure 4d–f). Consistent with enhanced HGF signalling, MET receptor expression also significantly increased in epithelial cells after HGF mRNA lipid nanoparticle treatment (supplementary figure S12a).

FIGURE 4.

Single-cell RNA sequencing (scRNA-seq) reveals hepatocyte growth factor (HGF) mRNA lipid nanoparticles (LNPs) mediate lung tissue repair via epithelial remodelling. a) Uniform manifold approximation and projection (UMAP) visualisation of major lung cell lineages from experimental groups, including epithelials, natural killer cells (NKs)/T-cells (TCs), B-cells (BCs), myeloids, endothelials, fibroblasts, smooth muscle cells (SMCs) and neutrophils. b) Proportional distribution of major cell lineages across treatment groups (PBS, PBS; elastase (ELA), PBS; ELA, HGF mRNA LNPs). c) Ratio of observed to expected (R_o/e) cell frequency, quantifying the expansion of epithelial cells following HGF mRNA LNPs treatment. d–f) Augur analysis predicting cell-type-specific transcriptional distinctiveness, presented as area under the curve (AUC) values: d) across all groups, e) between ELA, PBS and PBS, PBS (disease effect), and f) between ELA, HGF mRNA LNPs and ELA, PBS (therapeutic effect). g) UMAP visualisation of epithelial cell subpopulation. h) Cellular composition of the epithelial compartment. Pie chart: contribution of each treatment group to the total epithelial cell population; bar plot: for each epithelial subtype, the percentage contribution of cells from each treatment group to the total of that subtype. i) Proportional distribution of epithelial subpopulations across treatment groups. j) Comparative gene set enrichment analysis (GSEA) in epithelial cells between ELA, PBS and ELA, HGF mRNA LNPs groups, showing HGF-mediated pathway modulation. Colour scale represents log₂ fold change of gene set activity; dot size indicates the range of false discovery rate (FDR). Statistical analysis for Augur and GSEA is detailed in the supplementary methods. PNEC: pulmonary neuroendocrine cell; AT1/2: alveolar type I/II cells; logFC, log₂ fold change.

To dissect epithelial heterogeneity in treatment effects, we classified pulmonary epithelial cells as airway (basal, goblet, club, ciliated and pulmonary neuroendocrine cells) versus alveolar (AT1 and AT2) cells (figure 4g). HGF mRNA lipid nanoparticle-treated mice contributed 57.9% of all epithelial cells, compared to 22.7% in ELA-injured and 19.4% from PBS controls (figure 4h). Analysing relative contributions of each cell type to its total population demonstrated that HGF mRNA lipid nanoparticle treatment was the primary source for all epithelial cells in the post-injury landscape (figure 4h), consistent with increased MKI67⁺ cells following HGF mRNA lipid nanoparticle treatment in both ELA and cigarette-smoke models.

This analysis also showed that ELA injury diminished AT1 and AT2 cell populations, while expanding club cells (figure 4i). HGF mRNA lipid nanoparticle treatment reversed these alterations, increasing AT1 and ciliated cell proportions and reducing club cells. In the context of overall epithelial expansion, treatment resulted decreased relative AT2 cell proportions, but increased their absolute numbers (figure 4i).

Comparative gene set activity analysis of lung epithelial cells between ELA, PBS versus ELA, HGF mRNA lipid nanoparticle-treated mice revealed functional pathway modulation patterns under HGF regulation (figure 4j). Consistent with known HGF properties, HGF mRNA lipid nanoparticles triggered coordinated pathway alterations in epithelial cells: suppression of inflammatory cascades (IL-6–JAK–STAT3 signalling, inflammatory response, TNF-α signalling via NF-κB), inhibition of apoptosis, and activation of proliferative programmes including mitotic spindle assembly.

HGF mRNA lipid nanoparticles reconstitute a functional alveolar niche

Next, we investigated structural reorganisation of the alveolar niche in the ELA model. IHC showed significantly reduced proportions of RAGE⁺ AT1 cells in both ELA, PBS and ELA, Fluc mRNA lipid nanoparticle-treated groups compared to the PBS, PBS controls, which was partially reversed by HGF mRNA lipid nanoparticles, but not by rhHGF protein (figure 5a, b). ELA, PBS and ELA, Fluc mRNA lipid nanoparticle-treated groups also showed reduced surfactant protein C positive AT2 cells, indicating compromised regenerative capacity (figure 5a, c); HGF mRNA lipid nanoparticles increased absolute numbers of AT2 cells with a decrease in their relative proportion, consistent with the AT1 increases observed in both scRNA-seq and immunohistology results (figures 4h, i and 5a, c). Focal clusters of proliferating MKI67⁺ AT2 cells surrounding injured alveoli were evident in the HGF mRNA lipid nanoparticle-treated group, revealing localised repair activity at sites of remodelling (figure 5a, d).

FIGURE 5.

Hepatocyte growth factor (HGF) mRNA lipid nanoparticles promote coordinated reconstruction of the alveolar niche. a) Representative immunofluorescence images of lung sections stained for alveolar type I (AT1) cells (receptor for advanced glycation end products (RAGE), red), AT2 cells (surfactant protein C (SFTPC), green), proliferating cells (MKI67, white) and nuclei (4′,6-diamidino-2-phenylindole (DAPI), blue). Scale bar=20 µm; b–d) Quantification of the percentage of b) RAGE⁺ AT1 cells, c) SFTPC⁺ AT2 cells, and d) MKI67⁺SFTPC⁺ proliferating AT2 cells in lung tissues (n=6 mice per group). e) Representative immunofluorescence images of human lung organoids derived from emphysema patients, cultured with or without recombinant (rh)HGF protein, stained for AT2 cells (SFTPB, red), proliferating cells (MKI67, green), cytoskeleton (phalloidin, white), and nuclei (DAPI, blue). Scale bars=5 µm. f, g) Quantification of the percentage of f) SFTPB⁺ AT2 cells and g) MKI67⁺SFTPB⁺ proliferating AT2 cells within organoids (n=3 patient donors). h) Representative multiplex immunofluorescence images of lung sections stained for endothelial cells (CD31, cyan), smooth muscle cells (α-SMA, magenta), mesenchymal cells (VIMENTIN, yellow), AT1 cells (RAGE, red), AT2 cells (SFTPC, green), proliferating cells (MKI67, white) and nuclei (DAPI, blue). Reconstituted microvessels are indicated by white asterisks. Scale bars=50 µm. i) Quantification of CD31⁺ area fraction in lung tissues (n=6 mice per group). j) Quantification of vascular density in human vascular organoids under control, cigarette smoke extract (CSE), and CSE+rhHGF conditions (n=3 independent organoid experiments). k) Representative immunofluorescence images of vascular organoids showing endothelial cells (CD31, red) and smooth muscle cells (α-SMA, green) under the indicated conditions. Scale bars=50 µm and 100 µm. In all quantification panels, each dot represents one mouse, donor or experimental replicate. Data in all panels are presented as mean±sd. ns: nonsignificant. Statistical significance was determined by b–d, i, j) one-way ANOVA followed by Tukey's post hoc test or f, g) unpaired two-tailed t-test. *: p<0.05, **: p<0.01, ***: p<0.001.

To test whether HGF promotes AT2 proliferation even in emphysema, we utilised an ex vivo organoid model derived from adult stem cells isolated from human peripheral lung tissues (supplementary figure S11a). Immunofluorescence confirmed alveolar cell composition (supplementary figure S11b), and flow cytometry quantified reduced proportions of both AQP5⁺ AT1 and SFTPB⁺ AT2 cells in emphysema-derived organoids compared to those from never-smokers (supplementary figure S11c, d). Treating emphysema-derived organoids with exogenous rhHGF protein improved organoid forming efficiency and spheroid diameter (supplementary figure S11e–g), significantly increasing both total AT2 proportions and the fraction of MKI67⁺ proliferating AT2 cells (figure 5e–g).

Additionally, we assessed whether HGF mRNA lipid nanoparticles reconstituted the alveolar vasculature. In untreated ELA lungs, capillary networks were disrupted and septal architecture was disorganised (figure 5h). HGF mRNA lipid nanoparticles restored organised CD31⁺ capillary structures within septa, with a significant increase in CD31⁺ area fraction (figure 5i). VIMENTIN⁺ mesenchymal cells were appropriately redistributed along reconstructed septa, reconstituting mesenchymal support (figure 5h). To directly test HGF's pro-angiogenic capacity, we employed a human vascular organoid system. rhHGF protein treatment significantly increased CD31⁺ vascular density and enhanced endothelial–smooth muscle organisation within organoids (figure 5j, k), providing mechanistic support for the in vivo vascular findings. Consistent with this pro-angiogenic effect, HGF mRNA lipid nanoparticles increased endothelial proliferation (MKI67⁺CD31⁺) within pulmonary vasculature, whereas proliferation in vascular smooth muscle cells and peribronchial regions remained unchanged (supplementary figure S11h–k).

HGF mRNA lipid nanoparticles activate WNT signalling via the AKT-GSK3β-β-catenin axis

To elucidate how HGF mRNA lipid nanoparticles promote alveolar repair, we analysed transcriptional changes in AT2 cells following treatment. Gene set enrichment analysis (GSEA) confirmed that HGF mRNA lipid nanoparticles upregulated WNT signalling and proliferation pathways (figure 6a), and reversed downregulation of proliferation markers (Cdk1, Cenpa, Mki67) in AT2 cells from emphysematous lungs (supplementary figure S12b and c). Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses of AT2 cells identified functional terms involved in wound healing, tissue regeneration, lung epithelium development, and Wnt signalling pathway, among others (supplementary figure S12d–f).

FIGURE 6.

Hepatocyte growth factor (HGF) engages the AKT–GSK3β–β-catenin axis to activate WNT signalling. a) Gene set enrichment analysis (GSEA) plots showing enrichment of WNT signalling pathway and proliferation pathway in AT2 cells from elastase (ELA), HGF mRNA lipid nanoparticles group versus ELA, PBS group. b) Distribution of HGF-regulated genes identified by bulk RNA sequencing of emphysema patient-derived lung organoids treated with PBS (Ctrl), recombinant (rh)HGF, or rhHGF combined with AKT inhibitor (MK-2206, 5 μM). Among HGF-upregulated genes, 67.6% were reversed by AKT inhibition (reversal rate=0.676). c, d) ssGSEA of c) PI3K_AKT_MTOR signalling and d) HALLMARK_WNT_BETA_CATENIN signalling across Ctrl, rhHGF and rhHGF+MK-2206 groups. e) WNT module score comparison among experimental groups. f) Organoid forming efficiency under indicated conditions. g) Quantification of organoid diameter under indicated conditions. h) Representative bright-field images of emphysema patient-derived lung organoids under Ctrl, rhHGF and rhHGF+MK-2206 conditions. Scale bars=500 µm. i) Western blot analysis of pAKT (Ser473) and total AKT in organoids treated as indicated. j) Western blot analysis of pGSK3β (Ser9), total GSK3β, nuclear non-phosphorylated β-catenin (n-p-β-catenin), and total β-catenin. k) Quantification of pAKT/AKT ratio. l) Quantification of pGSK3β/GSK3β ratio. m) Quantification of n-p-β-catenin/β-catenin ratio. In all quantification panels, each dot represents an independent patient donor. ns: nonsignificant. Data are presented as mean±sd. Statistical significance was determined by one-way ANOVA followed by Tukey's post hoc test. *: p<0.05, **: p<0.01, ***: p<0.001.

Because HGF activates PI3K-AKT signalling, we asked whether AKT mediates HGF-induced WNT activation. We treated emphysema patient-derived lung organoids with PBS (Ctrl), rhHGF, or rhHGF plus an AKT inhibitor (MK-2206). Bulk RNA sequencing showed that AKT inhibition substantially reversed rhHGF-regulated gene expression changes (figure 6b). ssGSEA confirmed activation of PI3K_AKT_MTOR signalling by rhHGF, which was suppressed by MK-2206 (figure 6c). HALLMARK_WNT_BETA_CATENIN signalling and WNT module scores increased with rhHGF and were attenuated by MK-2206 (figure 6d, e). Functionally, rhHGF increased organoid forming efficiency and spheroid diameter, whereas MK-2206 reduced these effects (figure 6f–h).

At the protein level, rhHGF induced phosphorylation of AKT at Ser473, while MK-2206 suppressed pAKT (figure 6i, k). rhHGF increased phosphorylation of GSK3β at Ser9 and elevated nuclear nonphosphorylated β-catenin, effects abrogated by MK-2206 (figure 6j, l, m). WNT pathway activation was also confirmed in the cigarette-smoke model, where HGF mRNA lipid nanoparticles increased nuclear β-catenin⁺ AT2 cells (supplementary figure S12g, h). Collectively, these data support an HGF/c-Met-AKT-GSK3β-β-catenin axis mediating regenerative responses downstream of HGF mRNA lipid nanoparticle treatment.

Durable efficacy and safety profile of HGF mRNA lipid nanoparticles

To assess durability of therapeutic effect, we extended post-treatment observation in the ELA model. Pulmonary function testing parameters and MLI values remained significantly improved, compared to PBS-treated ELA controls, in HGF mRNA lipid nanoparticle-treated mice at both 6 and 8 weeks, indicating persistent benefit (supplementary figure S13a, b). Comprehensive haematological and biochemical analyses at these time points revealed no abnormalities (supplementary figure S14a–j), and airway wall thickness and subepithelial collagen deposition were similarly unaffected (supplementary figure S15a, b).

In a separate 12-month safety study, mice receiving two doses of HGF mRNA lipid nanoparticles exhibited neither bodyweight changes nor histological abnormalities in lung, liver or small intestine (supplementary figure S16a–e). Nor was there evidence of sustained proliferation or epithelial–mesenchymal transition (assessed by MKI67, E-cadherin, and N-cadherin expression) in these organs (supplementary figures S17–S19).

Repeat-dose tolerability was evaluated in mice receiving up to four intratracheal administrations. No cumulative toxicity was observed, with stable BALF markers of epithelial injury and inflammation, absent complement activation, and preserved lung histology (supplementary figure S20a–j). Complete blood counts and serum biochemical parameters remained stable across all dosing groups, without dose-dependent trends (supplementary figure S21a–j).

Given the association of COPD with pulmonary vascular and airway remodelling, we examined these parameters in both emphysema models. In the ELA model, which exhibited minimal vascular remodelling at baseline, HGF mRNA lipid nanoparticles did not induce vascular changes (supplementary figure S22a, c, e, f and i). In the cigarette-smoke model, where vascular remodelling was evident, HGF mRNA lipid nanoparticles treatment did not exacerbate medial wall thickness, collagen deposition, or muscularisation, nor did it significantly reverse established remodelling (supplementary figure S22b, d, g, h and j). Airway wall thickness and subepithelial collagen deposition were similarly unaffected in both models (supplementary figure S23a–f).

We also assessed systemic inflammatory responses. Serum cytokine levels (IL-6, IL-1β, TNF-α) were not elevated following HGF mRNA lipid nanoparticle treatment in either the ELA or cigarette-smoke model; in the cigarette-smoke model, TNF-α was significantly reduced compared to controls (supplementary figure S24a, b). Comprehensive haematological and biochemical analyses in the cigarette-smoke model confirmed no abnormalities (supplementary figure S24c–l). Collectively, with the previous shown data for the ELA model (supplementary figure S9), these results demonstrate that HGF mRNA lipid nanoparticles achieve durable therapeutic effects without evidence of pulmonary vascular remodelling, airway changes, systemic toxicity or safety concerns across acute, repeat-dose and long-term settings.

Discussion

Using publicly available human datasets and clinical lung samples (including organoids), and two murine models, we provide pre-clinical evidence that established emphysema could be treated by restoring HGF/c-Met signalling via mRNA delivery. We demonstrate an initial compensatory upregulation in lung HGF expression induced by smoking and mild COPD, followed by a significant decline in advanced emphysema. To address this late-stage deficiency, we repurposed the clinical-stage SM102 lipid nanoparticles platform to deliver HGF mRNA to the lungs. In ELA- and cigarette-smoke-induced emphysema models by either i.t. instillation or a clinically translatable aerosol route, HGF mRNA lipid nanoparticles restored lung structure and function, enhanced epithelial repair and microvascular regeneration, suppressed inflammation and apoptosis, and promoted AT2 cell proliferation, outperforming conventional recombinant protein therapy. Initial safety testing, including repeat-dose administration and monitoring out to 12 months post-treatment, revealed no cumulative toxicity, though we acknowledge that additional investigation will be needed before human trials. Our findings underscore the translational potential of combining approved delivery systems with mRNA therapeutics, offering a promising approach for patients with advanced emphysema.

While confirming elevated lung HGF expression in smokers and mild-to-moderate COPD [24, 25], consistent with initial upregulated repair mediators [26], a novel aspect of our data is the decline in lung HGF expression with advancing GOLD stage and emphysema severity. This transition suggests exhaustion of repair mechanisms in individuals susceptible to development of emphysema, a specific phenotype seen in only some COPD patients. Mechanistically, this pattern of late HGF deficiency reflected contributions during emphysema development from both reduced fibroblast abundance and stage-dependent regulation of HGF expression within fibroblasts. The parallel changes we identified in lung fibroblast MAPK pathway activity are consistent with the known MAPK-dependence of their HGF synthesis [23]. The strong negative correlation we identified in clinical specimens between alveolar septa HGF protein expression and alveolar damage quantified by MLI, a finding not captured by bulk homogenate measurements [27], illustrates the importance of spatial analysis of lung pathology across the disease spectrum.

HGF is a pleiotropic cytokine that mediates alveolar regeneration via effects on epithelial proliferation, migration, and survival [9, 12]. By scRNA-seq, the predominant response to HGF mRNA lipid nanoparticles by epithelial cells, particularly AT2, aligns with our finding that MET, the HGF receptor, is expressed primarily by this cell type. We showed three mechanisms by which local HGF delivery induced epithelial remodelling: 1) global epithelial expansion; 2) altered lineage proportions; and 3) molecular reprogramming. Increased AT2 cell abundance alongside their decreased relative proportions suggests active proliferation and differentiation, which is supported by MKI67⁺ AT2 cell clusters and elevated AT1 cell proportions near damaged alveoli. Although club cells are established facultative progenitors in some contexts [28, 29], our results suggest their limited role in these models of emphysema repair. Instead, the inverse relationship between club and AT2 cells after injury, and the predominant AT2 response to HGF, imply that alveolar repair primarily occurred through AT2 cell expansion and differentiation. Mechanistically, pathway analysis revealed coordinated modulation of WNT signalling in AT2 cells, a pathway known to regulate proliferation and differentiation [30]. Using human emphysema-derived lung organoids, we validated HGF activation of WNT signalling via an AKT–GSK3β–β-catenin axis, with AKT inhibition abrogating both transcriptional and functional responses to HGF. These findings support a direct mechanistic link between HGF treatment and activation of regenerative programmes in human alveolar epithelium.

Beyond these epithelial effects, HGF also promoted vascular regeneration within the alveolar niche, restoring organised CD31⁺ capillary networks and increasing vascular density [31]. We confirmed pro-angiogenic activity in human vascular organoids, where HGF directly enhanced endothelial network formation and endothelial-smooth muscle organisation [32, 33]. Notably, this regenerative response was compartment-specific, as vascular smooth muscle and peribronchial regions remained unaffected. Together, these coordinated epithelial and vascular responses contributed to reconstitution of a functionally integrated alveolar microenvironment.

Previous attempts to leverage HGF therapeutically have included cell-based approaches (e.g. HGF-secreting adipose tissue-derived stromal cells implanted via scaffolds or administered intravenously) [13, 14], gene therapy [15, 16], and protein administration via osmotic pumps or instillation [9]. Although demonstrating promise, these strategies face considerable limitations, including invasive procedures, poor lung targeting, and the short half-life of recombinant protein [9, 13–16].

To overcome these limitations, we employed mRNA technology, which enables robust in vivo translation of therapeutic proteins without nuclear entry or genomic integration [34]. Compared with previous approaches, mRNA offers simplified production, rapid action, and enhanced safety through transient expression.

A key aspect of our strategy, repurposing the clinical-stage SM102 lipid nanoparticles platform, contrasts with the predominant focus in the field on developing novel ionisable lipids for lung targeting [19, 20, 34, 35]. Although promising, such novel lipids face challenges in scalable manufacturing and require extensive safety re-evaluation. By contrast, SM102 lipid nanoparticles have a safety profile validated by billions of vaccine doses [22, 36], scalable good manufacturing practice production, and a clear regulatory path. Respiratory administration is crucial to translating this approach to an approved therapy, to avoid systemic delivery and hepatic accumulation [37]. Given the transient expression profile of mRNA therapeutics, achieving sustained benefit in a chronic disease such as emphysema may ultimately require repeated administration. In the present study, the repeat-dose regimens in mice were used to assess the feasibility and tolerability of repeated pulmonary delivery, rather than to define a long-term maintenance regimen for clinical use. Such “old platform, new use” to shorten development timelines could be generalised for other respiratory diseases.

Admittedly, lipid nanoparticles possess inherent immunostimulatory properties and may trigger mild innate immune activation depending on formulation and dosing [38]. Indeed, our vehicle control, Fluc mRNA lipid nanoparticles, modestly increased inflammatory parameters. Importantly by contrast, HGF expression markedly suppressed inflammatory markers compared to both disease controls and Fluc mRNA lipid nanoparticles, implying that HGF's anti-inflammatory activity counterbalances the intrinsic immunostimulatory potential of the vehicle. These findings support that the observed therapeutic benefits are attributable to HGF-mediated biological activity rather than to nonspecific effects of the lipid nanoparticle formulation.

Our study has several limitations. First, although we systematically evaluated pulmonary vascular structure, direct haemodynamic measurements such as pulmonary arterial pressure were not performed; thus, we cannot definitively conclude whether HGF mRNA lipid nanoparticles influence pulmonary hypertension. Second, while our 12-month safety study in healthy mice showed no evidence of tumorigenicity, evaluation in tumour-prone or carcinogen-primed models would provide more rigorous assessment of oncogenic risk, particularly given the elevated baseline lung cancer incidence in COPD populations. Third, we included only male human samples and male mice, limiting generalisability in a disease with known sex-dependent differences. Fourth, although SM102 lipid nanoparticles demonstrated efficient lung delivery and tolerability in healthy mice, their performance in chronically diseased, mucus-enriched human airways may differ and requires further investigation. Finally, translation to humans requires validation in large animal models or advanced human ex vivo platforms to account for species-specific differences in lung anatomy and disease progression.

In conclusion, this pre-clinical study presents a novel mRNA-based therapeutic strategy for emphysema leveraging the clinical-stage SM102 lipid nanoparticles platform for pulmonary-targeted delivery of HGF mRNA.

Supplementary material

Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.

Supplementary material

ERJ-02323-2025.Supplement

Data availability

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