Skip to main content
NCBI home page
International Journal of Molecular Sciences logoLink to International Journal of Molecular Sciences
. 2026 Jun 16;27(12):5453. doi: 10.3390/ijms27125453

Integrative Single-Cell Transcriptomic, Mendelian Randomization and In Silico Perturbation Analyses Prioritize MUC20 as a Candidate Gene Associated with Osteoporosis and Metabolic Dysfunction-Associated Steatotic Liver Disease in the Liver–Bone Axis

Hui Jin 1,2, Xiangting Ye 2, Gonghui Jian 1,*,†, Hui Xiong 1,*,†
Editor: Giovanni Malerba
PMCID: PMC13300151  PMID: 42353169

Abstract

Metabolic dysfunction-associated steatotic liver disease (MASLD) and osteoporosis (OP) are epidemiologically linked, but shared cell-type-specific molecular features remain unclear. We integrated public single-cell/single-nucleus transcriptomic datasets for OP (GSE147287) and MASLD (GSE289173) with two-sample Mendelian randomization (MR), colocalization, network annotation, macrophage-focused in silico perturbation, and exploratory serum assessment. After quality control, 13,753 OP cells and 42,438 MASLD cells/nuclei were analyzed. Macrophages were consistently identified in both datasets and showed disease-associated expansion. Directionally concordant macrophage differentially expressed genes yielded 147 shared candidate genes, with enrichment mainly involving lipid/sterol metabolism, extracellular matrix and adhesion processes, immune presentation, lysosomal processing, and phagocytic pathways. MR prioritized MUC20 as the only candidate with concordant odds ratios greater than 1 for both OP (OR = 1.044, 95% CI 1.003–1.086) and MASLD (OR = 1.111, 95% CI 1.038–1.189). Colocalization supported shared genetic signals for MUC20 in OP (PP.H4 = 0.855) and MASLD (PP.H4 = 0.816). In silico perturbation suggested a limited but pathway-enriched predicted transcriptional footprint. Serum MUC20 was higher in patients with OP+MASLD than in healthy controls. This integrative analysis identified shared macrophage-associated transcriptional themes and prioritized MUC20 as a candidate molecule for future liver–bone crosstalk studies.

Keywords: metabolic dysfunction-associated steatotic liver disease, osteoporosis, single-cell transcriptomics, mendelian randomization, in silico perturbation

1. Introduction

Metabolic dysfunction-associated steatotic liver disease (MASLD) and osteoporosis (OP) are common metabolic disorders affecting the liver and skeleton, respectively, and both impose substantial health burdens worldwide. Large systematic reviews estimate that MASLD affects nearly one-third of the global adult population [1,2]. OP affects approximately 200 million people worldwide and contributes to approximately 8.9 million osteoporotic fractures each year [3]. Among adults aged 50 years or older, fracture risk increases markedly with age and remains an important cause of disability and mortality [4]. Observational studies and systematic reviews suggest that MASLD is associated with lower bone mineral density and a higher risk of OP or fractures [5,6]. However, the strength and direction of this association vary across studies, likely because of differences in population characteristics, metabolic comorbidities, and disease definitions [7]. These inconsistencies highlight the need for evidence with higher cellular resolution and complementary genetic support.

Metabolic stress and chronic low-grade inflammation are key features of both MASLD progression and abnormal bone remodeling [8,9]. Myeloid-lineage cells, especially monocytes and macrophages, regulate metabolic inflammation, local tissue remodeling, and immune responses, and may provide a biologically relevant interface between hepatic inflammatory niches and skeletal microenvironmental changes [10,11]. However, much of the current evidence comes from phenotypic associations or bulk transcriptomic studies, which cannot clearly distinguish changes in cell composition from transcriptional changes within specific cell types [12]. Single-cell and single-nucleus transcriptomics can resolve disease-associated molecular changes at the level of individual cell types and cell states, providing a useful approach for identifying cross-disease cellular programs [13]. Recent multi-omics studies in MASLD have also shown coordinated changes in hepatic metabolic, inflammatory, and tissue-remodeling networks, further supporting the value of integrative omics approaches [14].

Observational associations are also limited by confounding and reverse causation, which makes directional interpretation difficult. Mendelian randomization (MR) can provide complementary genetic evidence by using genetic variants as instruments to evaluate exposure–outcome relationships [15]. Although MASLD–OP associations and liver–bone immunometabolic crosstalk have been increasingly discussed, shared myeloid programs and genetically supported candidate genes remain insufficiently defined. Here, we integrated cell-type-resolved single-cell evidence, MR-based genetic prioritization, colocalization analysis, network annotation, computational perturbation, and exploratory serum assessment to identify shared macrophage-related programs and prioritize candidate genes for subsequent validation.

2. Results

2.1. Single-Cell Analysis in Osteoporosis

In the OP dataset (GSE147287), low-quality cells were removed using the following quality control thresholds: nFeature_RNA 200–5000, nCount_RNA 500–86,000, and percent.mt < 15%. After filtering, 13,753 cells were retained, including 6549 control cells and 7204 OP cells. Post-filtering metrics were within the expected ranges: nFeature_RNA 201–4997, nCount_RNA 520–85,588, and percent.mt 0–14.995% (Figure 1A).

Figure 1.

Figure 1

Open in a new tab

Single-cell analysis of osteoporosis (OP). (A) Quality control assessment after filtering. (B) UMAP dimensionality reduction and clustering. (C) Cell-type annotation based on canonical marker genes. (D) Cell-type distribution in control and OP samples. (E) Cell-type proportions in control and OP samples. (F) Dot plot showing representative marker gene expression across annotated cell types. (G) Differentially expressed genes across cell types in OP versus control samples.

Using 2000 highly variable genes, we performed principal component analysis (PCA) and selected 13 principal components for downstream analysis. These components explained 91.23% of the cumulative variance. Unsupervised clustering at a resolution of 0.4 identified 17 clusters, and uniform manifold approximation and projection (UMAP) visualization showed separation of the major clusters (Figure 1B). Based on cluster-specific marker genes, canonical lineage markers, and CellMarker 2.0 reference information [16], we annotated nine major cell types: monocytes, macrophages, dendritic cells, B cells, T cells, plasma cells, erythroid cells, mesenchymal stem/progenitor cells (MSCs), and osteoblast-lineage cells (Figure 1C). Projection of cell-type labels onto the UMAP showed the distribution of these lineages in control and OP samples (Figure 1D). DotPlot analysis further supported the marker-based annotation of the major cell populations (Figure 1F).

Cell composition analysis showed a higher proportion of macrophages in OP than in controls (25.02% vs. 18.17%) and a lower proportion of osteoblast-lineage cells (35.27% vs. 42.52%) (both p < 0.05; Figure 1E). These results suggested disease-related changes in myeloid and bone-forming cell populations. Cell-type-stratified differential expression analysis also showed transcriptional differences across lineages, as summarized by the distribution of upregulated and downregulated genes in each cell type (Figure 1G). These findings provided the basis for subsequent cross-disease comparison.

2.2. Single-Cell Analysis in MASLD

In the MASLD dataset (GSE289173), eight samples were merged and filtered using the following criteria: nFeature_RNA 300–5000, nCount_RNA 499–18,023, and percent.mt < 15%. After filtering, 42,438 cells/nuclei were retained, including 22,805 control cells/nuclei and 19,633 MASLD cells/nuclei. Post-filtering metrics were nFeature_RNA 313–4999, nCount_RNA 502–18,023, and percent.mt 0–14.91% (Figure 2A).

Figure 2.

Figure 2

Open in a new tab

Single-cell analysis of metabolic dysfunction-associated steatotic liver disease. (A) Quality control assessment after filtering. (B) UMAP dimensionality reduction and clustering. (C) Cell-type annotation based on canonical marker genes. (D) Cell-type distribution in control and MASLD samples. (E) Cell-type proportions in control and MASLD samples. (F) Dot plot showing representative marker gene expression across annotated cell types. (G) Differentially expressed genes across cell types in MASLD versus control samples.

PCA was performed using 2000 highly variable genes, and 12 principal components were selected for downstream analysis. These components explained 91.46% of the cumulative variance. Unsupervised clustering identified 18 clusters, and UMAP visualization showed separation of the major clusters (Figure 2B). Using canonical marker genes and CellMarker 2.0 reference information, we annotated seven major cell types: hepatocytes, endothelial cells, mesenchymal cells, monocytes, macrophages, T cells, and B cells (Figure 2C,D). DotPlot visualization showed the expected lineage-specific marker expression patterns and supported the cell-type annotation (Figure 2F).

Cell composition analysis showed a higher macrophage fraction in MASLD than in controls (37.59% vs. 28.98%) and a lower hepatocyte fraction (48.95% vs. 63.57%) (both p < 0.05; Figure 2E). These results suggested disease-related myeloid enrichment and parenchymal cell reduction. Cell-type-stratified differential expression analysis further showed transcriptional differences across immune and parenchymal lineages (Figure 2G).

2.3. Shared Candidate Genes Between Osteoporosis and MASLD

To compare transcriptional features across diseases, we focused on macrophages. This myeloid lineage was consistently annotated in both datasets and showed disease-associated expansion in both OP and MASLD. Because the OP and MASLD datasets came from different tissue contexts, disease-control differential expression analyses were performed separately within each dataset. We then compared directionally concordant macrophage differentially expressed genes (DEGs) at the gene-set level. Within macrophages, DEGs were identified using the same criteria in both datasets: false discovery rate (FDR) < 0.05 and |log2FC| > 0.5.

After stratifying DEGs by direction and intersecting gene sets across diseases, we identified 76 commonly downregulated genes (MASLD: 1662; OP: 864; Figure 3A) and 71 commonly upregulated genes (MASLD: 688; OP: 2457; Figure 3B). These 147 shared candidate genes were used for downstream functional annotation and genetic prioritization. Macrophage subcluster marker expression supporting the macrophage-lineage annotation is shown in Supplementary Figure S1.

Figure 3.

Figure 3

Open in a new tab

Shared macrophage-associated candidate genes and functional enrichment between MASLD and OP. (A) Venn diagram showing the overlap of downregulated macrophage DEGs between MASLD and OP. (B) Venn diagram showing the overlap of upregulated macrophage DEGs between MASLD and OP. (C–E) GO enrichment of the 147 shared candidate genes across biological process (C), cellular component (D), and molecular function (E), shown by GOCircle plots. (F) KEGG pathway enrichment of the shared candidate genes.

Functional enrichment analysis of the 147 shared candidates showed that Gene Ontology (GO) biological process terms were mainly related to cholesterol, sterol, and steroid biosynthesis, responses to oxygen levels or hypoxia, and amoeboidal-type cell migration (Figure 3C). GO cellular component terms included cell–substrate junctions, focal adhesions, collagen-containing extracellular matrix, and vesicle or granule lumen-related structures (Figure 3D). GO molecular function terms included extracellular matrix structural constituents, glycosaminoglycan binding, long-chain fatty acid binding, and redox-related activities (Figure 3E). Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment included steroid biosynthesis, terpenoid backbone biosynthesis, integrin signaling, antigen processing and presentation, intestinal immune network for immunoglobulin A production, and hematopoietic cell lineage (Figure 3F). Overall, these enrichment results suggested overlapping functional annotations related to lipid or sterol metabolism, extracellular matrix and adhesion processes, and immune-presentation pathways.

2.4. Mendelian Randomization and Colocalization Prioritize MUC20 as a Candidate Gene

Using expression quantitative trait locus (eQTL)-based instruments derived from shared macrophage candidate genes, we performed two-sample MR. Inverse-variance weighting (IVW) was used as the primary estimator. The IVW analysis identified 10 genes associated with osteoporosis (OP) and nine genes associated with MASLD at p < 0.05 (Figure 4A,B). The corresponding IVW estimates and robustness diagnostics, including MR-Egger intercept and heterogeneity tests, are summarized in Table 1.

Figure 4.

Figure 4

Open in a new tab

Forest plot of IVW-significant gene-level MR estimates for OP and MASLD. (A) Gene-level MR estimates for OP. (B) Gene-level MR estimates for MASLD. Estimates were obtained using IVW as the primary MR method. Points indicate ORs and horizontal lines indicate 95% CIs.

Table 1.

Summary of gene-level MR estimates and robustness diagnostics for OP and MASLD.

Gene Outcome nSNP IVW OR 95% CI IVW p-Value MR-Egger Intercept MR-Egger Intercept p-Value Heterogeneity p-Value
ACSM3 OP 9 0.763 0.613–0.948 0.0149 −0.0302 0.2676 0.5757
ANK3 OP 54 0.922 0.870–0.977 0.0061 −0.0144 0.1700 0.8992
ARHGEF10 OP 51 1.086 1.029–1.145 0.0026 0.0041 0.6843 0.4687
CD22 OP 5 1.463 1.135–1.884 0.0033 0.0037 0.9017 0.9042
ECHDC3 OP 34 0.929 0.867–0.997 0.0401 −0.0104 0.3907 0.9676
GGH OP 19 0.765 0.659–0.887 0.0004 0.0200 0.4288 0.8960
KCNE1 OP 15 1.220 1.068–1.393 0.0035 0.0063 0.6985 0.6754
LAP3 OP 67 0.955 0.914–0.998 0.0384 0.0056 0.5452 0.6946
MUC20 OP 82 1.044 1.003–1.086 0.0344 0.0056 0.4499 0.1506
TNK2 OP 5 1.529 1.129–2.072 0.0061 0.0147 0.7745 0.9893
ABCC3 MASLD 59 0.909 0.831–0.995 0.0377 −0.0108 0.5566 0.9680
ANK3 MASLD 54 1.134 1.019–1.261 0.0213 −0.0226 0.2422 0.9388
ECHDC3 MASLD 34 1.139 1.001–1.297 0.0483 −0.0111 0.6224 0.4608
ETV7 MASLD 46 1.161 1.041–1.296 0.0075 −0.0118 0.5768 0.4864
IGFBP7 MASLD 27 0.712 0.575–0.881 0.0018 −0.0246 0.4940 0.2447
KYNU MASLD 20 1.296 1.095–1.535 0.0026 0.0420 0.2028 0.7332
MUC20 MASLD 82 1.111 1.038–1.189 0.0024 0.0141 0.2685 0.9742
THBS1 MASLD 38 0.879 0.776–0.995 0.0415 −0.0155 0.4696 0.5772
VSIG10 MASLD 36 0.855 0.747–0.978 0.0225 0.0013 0.9549 0.5811
Open in a new tab

Note: Genes shown in this table reached nominal significance in the IVW analysis (IVW p < 0.05) for OP or MASLD. MR, Mendelian randomization; IVW, inverse variance-weighted; OR, odds ratio; CI, confidence interval; nSNP, number of instrumental variables; OP, osteoporosis; MASLD, metabolic dysfunction-associated steatotic liver disease. The MR-Egger intercept was used to assess directional horizontal pleiotropy. Heterogeneity p-values were calculated using Cochran’s Q test.

Among these genes, MUC20 was the only candidate with concordant odds ratios (ORs) > 1 for both OP and MASLD. The IVW estimate was OR = 1.044 for OP (95% confidence interval [CI] 1.003–1.086; p = 0.034; nSNP = 82) and OR = 1.111 for MASLD (95% CI 1.038–1.189; p = 0.002; nSNP = 82). Sensitivity analyses showed similar effect directions across complementary MR methods. Leave-one-out analyses did not suggest that the estimates were driven by a single variant. Detailed single-nucleotide polymorphism (SNP)-level robustness analyses for MUC20 are provided in Supplementary Figure S2. Retained instrument characteristics are provided in Supplementary Tables S1–S4.

We further performed colocalization analysis to assess whether MUC20 eQTL and disease genome-wide association study (GWAS) signals shared genetic signals. For OP, the posterior probability supporting a shared causal variant was PP.H4 = 0.855, whereas the probability supporting distinct causal variants was PP.H3 = 0.014. For MASLD, PP.H4 was 0.816 and PP.H3 was 0.022 (Figure 5; Table 2). Together, these MR and colocalization results prioritized MUC20 for subsequent functional interpretation in OP–MASLD comorbidity.

Figure 5.

Figure 5

Open in a new tab

Colocalization analysis of MUC20 in OP and MASLD. (A) Scatter plot comparing MUC20 eQTL significance and OP GWAS significance across shared SNPs. (B) Scatter plot comparing MUC20 eQTL significance and MASLD GWAS significance across shared SNPs. The x-axis indicates MUC20 eQTL −log10(p), and the y-axis indicates disease GWAS −log10(p).

Table 2.

Colocalization analysis for MUC20 in OP and MASLD.

Gene Outcome nsnps PP.H0 PP.H1 PP.H2 PP.H3 PP.H4
MUC20 OP 82 0.000 0.061 0.000 0.014 0.855
MUC20 MASLD 82 0.000 0.076 0.000 0.022 0.816
Open in a new tab

Note: PP.H0–PP.H4 represent posterior probabilities for five hypotheses: no association with either trait, association with MUC20 expression only, association with disease only, association with both traits through distinct causal variants, and association with both traits through a shared causal variant, respectively.

2.5. Network-Based Functional Annotation of MUC20

After MR prioritization of MUC20, we performed multi-source network annotation to describe its functional context. GeneMANIA placed MUC20 within a network involving MET, HGF, SOS1, and several glycosylation-related genes. The enriched functions were mainly related to glycosylation and protein O-linked glycosylation (Figure 6A). Comparative Toxicogenomics Database (CTD) annotations linked MUC20 to liver injury, metabolic liver disease, OP, and metabolic phenotypes (Figure 6B). miRDB predictions and CTD chemical association data provided additional database-derived information on possible upstream regulatory and environmental contexts (Figure 6C,D). Overall, these database-derived observations provided functional context for MUC20 and helped define hypotheses for future experimental evaluation.

Figure 6.

Figure 6

Open in a new tab

Network-based and knowledge-based characterization of MUC20. (A) GeneMANIA interaction network of MUC20 and its functional neighborhood, highlighting glycosylation-related modules. (B) Gene–disease associations for MUC20 curated from the CTD. (C) Predicted miRNA regulators of MUC20 from miRDB. (D) Gene–chemical associations for MUC20 from CTD. These analyses provide functional and translational context for subsequent hypothesis testing.

2.6. In Silico Perturbation Suggests Candidate Lysosomal, Efferocytosis-Related, and Osteoclast-Associated Programs

To explore candidate downstream transcriptional programs associated with MUC20, we performed macrophage-focused in silico perturbation of MUC20 in OP macrophages. At the single-cell level, MUC20 expression was higher in disease macrophages than in control macrophages (Figure 7A). After simulated perturbation, a small fraction of genes reached statistical significance, accounting for approximately 2.9% of analyzed genes (Figure 7B). Representative predicted downstream genes included ACAP1, SPIB, JCHAIN, BCL11A, IRF8, LILRA4, PLD4, FAM129C, NEAT1, ISG20, and IL3RA (Figure 7C). High-effect predicted downstream genes were mainly annotated to modules related to the complement/Fc receptor axis, lysosome–phagocytosis, myeloid identity, and osteoclast-associated pathways (Figure 7D).

Figure 7.

Figure 7

Open in a new tab

Macrophage-focused in silico perturbation of MUC20 in osteoporosis and predicted downstream transcriptional programs. (A) MUC20 expression in macrophages from control and OP samples. (B) Proportion of significantly altered genes following simulated MUC20 perturbation. (C) Distribution of predicted downstream transcriptional changes after MUC20 perturbation, with representative genes labeled. (D) Top predicted downstream genes ranked by perturbation effect size and grouped by functional modules. (E) GO enrichment of perturbation-associated genes. (F) KEGG enrichment of perturbation-associated genes. ** p < 0.01, *** p < 0.001, **** p < 0.0001.

Enrichment analysis of perturbation-associated genes showed related functional patterns. Gene Ontology (GO) biological process terms included positive regulation of cytokine production, immune response regulation and activation, and myeloid leukocyte activation. GO cellular component terms included vacuolar lumen, lysosomal lumen, primary lysosome, and azurophil granule. GO molecular function terms included proteoglycan binding, peptidase regulation or binding, and immunoglobulin binding (Figure 7E). KEGG pathways included lysosome, osteoclast differentiation, efferocytosis, complement and coagulation cascades, phagosome, sphingolipid metabolism, cholesterol metabolism, and natural killer cell-mediated cytotoxicity (Figure 7F). Overall, the in silico perturbation analysis indicated a limited but pathway-enriched predicted transcriptional footprint associated with MUC20. These results support further functional assessment of MUC20 in macrophage-related models of OP–MASLD comorbidity.

2.7. Exploratory Serum Assessment of MUC20 in the Clinical Cohort

A total of 52 participants were included in this exploratory clinical cohort, including 20 healthy controls and 32 patients with OP+MASLD. Baseline characteristics are summarized in Table 3. Age and body mass index (BMI) were comparable between the two groups. The OP+MASLD group had a lower lumbar spine T-score and higher levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), triglycerides (TG), and total cholesterol (TC) than healthy controls. These differences were consistent with the skeletal and metabolic liver disease features of the cohort.

Table 3.

Baseline characteristics of participants in the exploratory clinical cohort.

Characteristic Healthy Controls (n = 20) OP+MASLD (n = 32) p-Value
Age, years 53.00 ± 7.21 57.44 ± 12.21 0.106
BMI, kg/m2 23.15 ± 3.45 24.19 ± 3.13 0.282
Lumbar spine T-score 0.03 ± 0.56 −3.19 ± 0.42 <0.001
ALT, U/L 37.43 ± 13.77 56.16 ± 9.36 <0.001
AST, U/L 17.90 (15.75, 20.73) 42.05 (32.27, 49.80) <0.001
TG, mmol/L 1.16 (0.80, 1.44) 3.59 (2.94, 4.13) <0.001
TC, mmol/L 4.12 ± 0.66 6.26 ± 0.46 <0.001
Open in a new tab

Note: Data are presented as mean ± SD or median (IQR), as appropriate. p-values were calculated using the independent-samples t test or Mann–Whitney U test.

Serum MUC20 levels were measured by enzyme-linked immunosorbent assay (ELISA). The mean serum MUC20 level was 1.02 ± 0.90 in healthy controls and 6.37 ± 2.57 in the OP+MASLD group. In this two-group exploratory comparison, serum MUC20 levels were higher in patients with OP+MASLD than in healthy controls (p < 0.001; Figure 8). This finding was consistent with the bioinformatics prioritization of MUC20 at the serum level.

Figure 8.

Figure 8

Open in a new tab

Comparison of serum MUC20 levels between the two groups. **** p < 0.0001.

3. Discussion

The association between MASLD and low bone mass or OP has been reported in epidemiological studies and systematic reviews. However, this relationship is difficult to interpret because obesity, insulin resistance, systemic inflammation, and other metabolic comorbidities are closely linked to both conditions. The direction of this association also remains uncertain [17]. In this study, we examined MASLD and OP at cellular and molecular levels. We focused on macrophages as a shared myeloid context and assessed whether the two diseases showed directionally concordant transcriptional features. We then used genetic and computational analyses to prioritize candidate molecules for further study [6,7].

Previous genetic studies have suggested shared genetic architecture or possible directional relationships between MASLD-related traits and bone outcomes, including bone mineral density (BMD), OP, and fractures. However, their conclusions have varied because of differences in phenotype definitions, instrument selection, outcome datasets, and adjustment strategies [18]. In some analyses, effect estimates were attenuated after adjustment for body mass index or glycemic traits, suggesting that the MASLD–bone relationship may be embedded in a broader metabolic and inflammatory network rather than driven by a single pathway [19]. Therefore, our study aimed to identify shared myeloid transcriptional features at cell-type resolution and to use genetic evidence to prioritize candidate genes for downstream validation [20].

Myeloid cells are important regulators of metabolic stress, inflammation, and tissue remodeling. Single-cell studies of MASLD and metabolic dysfunction-associated steatohepatitis (MASH) have described hepatic macrophage states related to lipid handling, inflammation, repair, and fibrosis [21]. TREM2-associated lipid-associated macrophage programs have also been linked to immune homeostasis under metabolic stress and fibrotic niche remodeling [22,23]. In addition, lysosomal lipid load, efferocytosis, and inflammation-resolution capacity have been proposed as key myeloid functions in steatohepatitis progression and regression [24,25,26]. In bone, myeloid-lineage cells give rise to osteoclasts and also contribute to immune regulation and tissue remodeling [27,28]. Changes in bone-resident macrophages and monocyte-lineage states may affect the balance between bone resorption and formation, especially under inflammatory conditions [29]. These findings support the use of macrophages as a relevant cell type for cross-disease comparison.

Within this matched macrophage lineage, we identified 147 directionally concordant genes between MASLD and OP. Because the datasets differed in tissue source, cell composition, metabolic environment, and matrix structure, this shared gene set should not be interpreted as evidence of identical macrophage states across tissues. Instead, it may reflect overlapping myeloid transcriptional themes under chronic metabolic stress and tissue remodeling [30,31]. Functional annotation of these genes highlighted lipid handling, lysosomal processing, phagocytic clearance, inflammatory regulation, extracellular matrix remodeling, and adhesion-related pathways [28]. These pathways are biologically plausible in both liver injury and bone remodeling contexts, but they should be viewed as functional annotations rather than disease-specific mechanisms.

We next used two-sample MR to add a genetic layer to the transcriptomic findings. Unlike previous MR studies that mainly tested relationships between MASLD phenotypes and bone outcomes [32], our analysis used MR to help prioritize genes emerging from shared macrophage-associated modules [33]. Following Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization (STROBE-MR) recommendations [34], we combined inverse variance-weighted estimates with sensitivity analyses, heterogeneity and pleiotropy tests, and MUC20-focused colocalization. MUC20 was the only gene with concordant odds ratios greater than 1 for both outcomes in the primary analyses. Colocalization further supported shared genetic signals between MUC20 expression and both OP and MASLD. These findings support MUC20 as a candidate gene for further functional evaluation, but they do not establish a large biological effect or a confirmed disease mechanism [35].

We then used macrophage-focused in silico perturbation to explore the functional context of MUC20. This analysis was intended to identify predicted downstream programs, not to validate a mechanism [36]. The perturbation footprint was modest, with approximately 2.9% of genes reaching statistical significance. The predicted changes were mainly annotated to myeloid programs related to phagocytosis, lysosomal processing, inflammatory responses, and osteoclast differentiation. This pattern was consistent with the enrichment results from the shared candidate genes and suggested a possible link between MUC20 and macrophage-related clearance or remodeling programs [26]. However, in silico perturbation depends on model assumptions and data coverage. Future experiments should test MUC20 knockdown or overexpression in macrophages and assess efferocytosis, lysosomal acidification, lipid load, inflammatory mediator production, and osteoclastogenic capacity [37].

MUC20 is a transmembrane mucin family member that has been associated with receptor signaling and stress-related cellular responses. Transmembrane mucins can participate in barrier function, adhesion, and immune-response regulation through their glycosylated extracellular domains and cytoplasmic tails [20]. Previous studies have suggested a possible relationship between MUC20 and hepatocyte growth factor (HGF)/c-MET signaling, a pathway involved in liver injury, repair, inflammation, and chemotaxis [38,39,40]. The cyclic GMP-AMP synthase–stimulator of interferon genes (cGAS–STING) axis may also provide a stress inflammation interface in metabolic disease contexts [41]. Public expression resources suggest that baseline MUC20 expression is generally low in peripheral immune cells and may be tissue- or cell-type-specific [42]. Therefore, the macrophage-associated signal observed here may reflect disease-related expression changes or enrichment of specific macrophage states. Independent replication and tissue-level localization are needed. Overall, these findings suggest that MUC20 may be linked to macrophage-associated signaling and clearance programs in MASLD–OP comorbidity, but this hypothesis requires direct testing in myeloid systems and animal models [43,44,45].

From a cross-organ perspective, MASLD is increasingly viewed as a systemic metabolic condition. Its associations with low bone mass, OP, and fracture risk have been described in reviews and meta-analyses, although effect sizes vary across populations and metabolic subgroups [46]. Current liver–bone axis concepts suggest that hepatokines, inflammatory mediators, and lipid-metabolic stress may affect the balance between bone formation and resorption, while bone-derived signals may also influence hepatic metabolism [47]. In this context, our analysis provides a candidate molecular entry point centered on macrophage-associated programs and MUC20, complementing studies based mainly on phenotypic associations [48]. Outputs from the Comparative Toxicogenomics Database and GeneMANIA should be interpreted as database-derived functional context rather than evidence of clinical efficacy [49]. A practical next step is to test whether MUC20 perturbation affects macrophage immune-metabolic phenotypes, followed by in vivo assessment of liver pathology and bone mass or remodeling endpoints [50,51]. In the exploratory clinical cohort, serum MUC20 levels were higher in patients with OP+MASLD than in healthy controls. This finding was consistent with the bioinformatics prioritization of MUC20, but it should be interpreted as preliminary serum-level evidence.

Several limitations should be considered. First, the public single-cell datasets differed in donor scale, disease definition, tissue source, and sequencing strategy [30]. Cross-tissue comparison cannot fully remove the influence of tissue niche, sample composition, and platform differences [31]. Second, the shared candidate gene set was derived from cell-level differential expression analysis. Because cells from the same donor are not fully independent observations, this approach may introduce pseudoreplication, and some identified genes may reflect donor or sample imbalance rather than genuine disease-associated transcriptional changes [52]. Future studies should validate the shared macrophage signatures using donor-level pseudobulk or mixed-model frameworks in independent cohorts [53]. Third, MR inference remains affected by instrument validity, linkage disequilibrium structure, expression quantitative trait locus tissue context, and assumptions about horizontal pleiotropy. Although we used sensitivity analyses, heterogeneity and pleiotropy tests, and MUC20-focused colocalization, residual bias cannot be excluded. Fourth, in silico perturbation provides computational evidence only and cannot replace experimental knockdown or overexpression. Fifth, the exploratory serum analysis was based on a single-center, two-group cohort that included healthy controls and patients with OP+MASLD. Because OP-only and MASLD-only groups were not included, we could not determine the relative contributions of OP, MASLD, combined disease status, metabolic abnormalities, systemic inflammation, medication exposure, or other clinical covariates. Larger and better-characterized cohorts are needed. Finally, cell–cell communication analysis and spatial localization may help determine whether MUC20-associated myeloid states are enriched in fibrotic liver niches or bone remodeling sites.

In summary, we identified shared macrophage-associated immunometabolic transcriptional themes between MASLD and OP and prioritized MUC20 as a candidate gene using complementary evidence from MR, colocalization, and computational perturbation. These findings provide a basis for future studies of MUC20-related myeloid programs in MASLD–OP comorbidity. Further work should replicate the shared macrophage module in independent and spatially resolved datasets and test the functional relevance of MUC20 in efferocytosis, lysosomal homeostasis, inflammatory output, and bone remodeling-related macrophage functions.

4. Materials and Methods

4.1. Data Sources and Study Design

The overall study design and analytical workflow are summarized in Figure 9. Single-cell transcriptomic data for OP and MASLD were obtained from the Gene Expression Omnibus (GEO; https://www.ncbi.nlm.nih.gov/geo/; accessed on 18 January 2026) under accession numbers GSE147287 and GSE289173, respectively [54,55,56]. We used an integrative framework that combined cell-type-resolved single-cell analysis, two-sample MR, colocalization analysis, network-based functional annotation, and macrophage-focused computational perturbation [57,58].

Figure 9.

Figure 9

Open in a new tab

Study design and analytical workflow.

Each single-cell dataset was processed independently for quality control, cell-type annotation, and cell-type-stratified differential expression analysis [12]. Shared candidate genes were identified within immune lineages present in both diseases. eQTL instruments for candidate genes were used as exposures, and GWAS summary statistics for OP and MASLD were used as outcomes in two-sample MR analyses [59,60,61]. For MR-prioritized genes, we performed network-based annotation and assessed predicted downstream transcriptional effects in macrophages [62]. This study used publicly available de-identified single-cell and summary-level genetic data, together with a single-center exploratory serum cohort. No intervention was performed, and the clinical component was limited to peripheral blood sampling.

4.2. Single-Cell Data Processing

The OP single-cell dataset GSE147287 and the MASLD single-cell or single-nucleus dataset GSE289173 were processed independently in R version 4.2.2 (R Foundation for Statistical Computing, Vienna, Austria) using Seurat version 5.4.0. Seurat objects were generated from raw count matrices. Standard quality control (QC) metrics were calculated, including the number of detected features (nFeature_RNA), total counts (nCount_RNA), and the percentage of mitochondrial transcripts (percent.mt). Dataset-specific QC thresholds were selected based on metric distributions to remove low-quality cells or nuclei and outliers with abnormally high counts.

Data were normalized using LogNormalize with a scale factor of 10,000. Highly variable genes were selected using the variance-stabilizing transformation method, and the top 2000 features were retained. The data were then scaled, and PCA was performed. For multi-sample datasets, batch effects were addressed using Seurat’s anchor-based integration workflow, including FindIntegrationAnchors and IntegrateData. A shared nearest-neighbor (SNN) graph was constructed in the integrated space, and clusters were identified using Louvain community detection. Clusters were visualized using UMAP. Cell types were annotated using canonical marker expression and cluster-specific marker genes.

4.3. Differential Expression and Functional Enrichment Analyses

Within each major cell type or subpopulation, disease-control differential expression analysis was performed using Seurat’s FindMarkers function with the Wilcoxon rank-sum test. Multiple testing was controlled using the Benjamini–Hochberg procedure. DEGs were defined as genes with an FDR-adjusted p-value < 0.05 and |log2 fold change| > 0.5.

To identify cross-disease candidate signatures, we first aligned cell-type labels between datasets. The analysis was then restricted to immune lineages present in both diseases, mainly macrophages, monocytes, T cells, and B cells. For each matched lineage, DEGs were stratified as upregulated or downregulated. Directionally concordant DEGs were intersected across diseases. The union of common upregulated and common downregulated genes was defined as the shared candidate gene set. Functional enrichment analysis was performed using clusterProfiler for GO biological process, cellular component, and molecular function terms, as well as KEGG pathway analysis [63,64,65]. Enrichment significance was defined as FDR-adjusted p < 0.05.

4.4. Two-Sample Mendelian Randomization and Colocalization Analyses

Two-sample MR analyses were performed in R version 4.2.2 using the TwoSampleMR package version 0.6.29. Genetic instruments were derived from candidate gene-related eQTL associations obtained from eQTLGen (https://www.eqtlgen.org/; accessed on 18 January 2026) and OpenGWAS (https://opengwas.io/; accessed on 18 January 2026) [61]. GWAS summary statistics for OP and MASLD were obtained from FinnGen/OpenGWAS resources (FinnGen: https://www.finngen.fi/en; OpenGWAS: https://opengwas.io/; all accessed on 18 January 2026). The OpenGWAS eQTL identifiers ranged from eqtl-a-ENSG00000000003 to eqtl-a-ENSG00000270149.

Single-nucleotide polymorphisms (SNPs) associated with gene expression at genome-wide significance were selected using p < 5 × 10−8. To reduce linkage disequilibrium (LD) among instruments, clumping was performed using an r2 threshold of 0.001 within a 10,000 kb window. Instrument strength was assessed using the F-statistic, and only instruments with F-statistics > 10 were retained. Exposure and outcome datasets were harmonized to ensure consistent effect–allele orientation [59]. Detailed SNP-level information, including exposure and outcome effect estimates, standard errors, effect allele frequencies, allele alignment, palindromic status, and F-statistics, is provided in the Supplementary Tables. The primary MR estimate was obtained using IVW and reported as ORs with 95% CIs. Robustness was evaluated using complementary MR estimators, including MR-Egger, weighted median, simple mode, and weighted mode, with emphasis on consistency in effect direction. Potential bias was assessed using Cochran’s Q test for heterogeneity, the MR-Egger intercept for horizontal pleiotropy, and leave-one-out analysis.

Bayesian colocalization analysis was performed using the coloc package version 5.2.3 in R to assess whether MUC20 expression and disease associations shared genetic signals. Shared SNPs between the MUC20 eQTL dataset and the corresponding OP or MASLD GWAS dataset were aligned and analyzed. Posterior probabilities were calculated for five hypotheses: no association with either trait (H0), association with MUC20 expression only (H1), association with disease only (H2), association with both traits through distinct causal variants (H3), and association with both traits through a shared causal variant (H4). PP.H4 was used to assess support for a shared genetic signal between MUC20 expression and each disease outcome.

4.5. Network-Based Functional Annotation of Prioritized Genes

Network- and knowledge-based annotation of MR-prioritized genes was performed using GeneMANIA (https://genemania.org/; accessed on 18 January 2026), miRDB (https://mirdb.org/; accessed on 18 January 2026), and the Comparative Toxicogenomics Database (CTD; http://ctdbase.org/; accessed on 18 January 2026). Gene–gene interaction networks were constructed using GeneMANIA to describe functional neighborhoods [62]. High-confidence microRNA–messenger RNA regulatory relationships were predicted using miRDB [66]. Gene–disease and gene–chemical associations were extracted from CTD [49]. These analyses were used to provide database-derived functional context for prioritized genes and to support hypothesis generation.

4.6. In Silico Perturbation of MUC20 and Downstream Effect Assessment

Macrophage-focused in silico perturbation of MUC20 was performed using the scTenifoldKnk framework version 1.0.2 [36]. The OP disease-state macrophage subset was used as the perturbation context. Raw count data were extracted from the Seurat object, and highly variable genes were identified using the variance-stabilizing transformation method. The input matrix included the top 20,000 highly variable genes and MUC20. MUC20 was retained during feature filtering to ensure inclusion of the perturbation target. Quality control within the perturbation workflow used a mitochondrial transcript threshold of 0.1 and a minimum library size of 1000. Genes detected in at least 5% of cells were retained when the number of cells exceeded 500. MUC20 was retained regardless of this filtering criterion. Gene regulatory networks were reconstructed using 10 subsampled networks with 500 cells per network, q = 0.9, and three principal components. Tensor decomposition was performed with K = 3, a maximum of 1000 iterations, and a convergence error threshold of 1 × 10−5. MUC20 perturbation was simulated by setting the MUC20 regulatory row to zero in the reconstructed network, followed by manifold alignment between the original and perturbed networks. Differentially regulated genes after MUC20 perturbation were identified from the scTenifoldKnk differential regulation output. Genes with adjusted p < 0.05 were considered perturbation-associated downstream genes. The proportion of significant genes was calculated relative to all genes included in the perturbation output after excluding MUC20 itself. Perturbation-associated genes were then used for GO and KEGG enrichment analyses with Benjamini–Hochberg correction. This analysis was used to prioritize candidate downstream transcriptional programs associated with MUC20 perturbation.

4.7. Exploratory Serum Assessment Using Clinical Samples

To provide exploratory serum-level support for the bioinformatics findings, postmenopausal women consecutively recruited from the Second Affiliated Hospital of Hunan University of Chinese Medicine between December 2025 and March 2026 were enrolled. Participants were assigned to an OP+MASLD group or a healthy control group. OP was defined as a lumbar spine T-score ≤ −2.5 on dual-energy X-ray absorptiometry (DXA). MASLD was diagnosed based on imaging-confirmed hepatic steatosis with concomitant metabolic abnormalities. Individuals with other definite chronic liver diseases, malignancy, acute or chronic infection, severe cardiac or renal dysfunction, or recent use of medications that could substantially affect bone or lipid metabolism were excluded.

Peripheral venous blood was collected in the morning after overnight fasting. Serum was separated by centrifugation and stored at −80 °C until analysis. Serum MUC20 levels were measured using a Human MUC20 QuickTest ELISA Kit (FineTest, Wuhan, China, Cat. No. QT-EH1227) according to the manufacturer’s instructions. All samples were assayed in duplicate. Laboratory personnel were blinded to group assignment, and measurements were performed in the same batch whenever possible. The study was approved by the Ethics Committee of the Second Affiliated Hospital of Hunan University of Chinese Medicine before participant enrollment and sample collection (approval no. 2024-KY03; approval date: 15 March 2024). Written informed consent was obtained from all participants before enrollment.

4.8. Statistical Analysis and Software

Single-cell analyses, enrichment analyses, and MR analyses were performed in R version 4.2.2. Figure assembly was completed using Adobe Illustrator 2023. Clinical data and serum MUC20 levels from the exploratory cohort were also analyzed in R. Continuous variables are presented as mean ± standard deviation (SD) or median (interquartile range [IQR]), as appropriate. Normality was assessed using the Shapiro–Wilk test. Between-group comparisons were performed using Student’s t test or the Mann–Whitney U test, as appropriate. Categorical variables were compared using the chi-square test or Fisher’s exact test, where applicable. Multiple testing was controlled using FDR-adjusted p-values for differential expression and enrichment analyses. Unless otherwise specified, all tests were two-sided, and p < 0.05 was considered statistically significant.

5. Conclusions

Using an integrative framework combining cell-type-resolved transcriptomics, two-sample Mendelian randomization, colocalization analysis, and macrophage-focused computational perturbation, we identified shared macrophage-associated immunometabolic transcriptional themes between metabolic dysfunction-associated steatotic liver disease (MASLD) and osteoporosis (OP). MUC20 was prioritized as a candidate gene associated with both outcomes. Exploratory serum assessment by enzyme-linked immunosorbent assay (ELISA) showed higher circulating MUC20 levels in patients with OP+MASLD than in healthy controls. Overall, these findings support MUC20 as a candidate molecule for future studies of liver–bone crosstalk, with further experimental and clinical validation needed to clarify its functional relevance.

Abbreviations

Abbreviation Full term
MASLD Metabolic dysfunction-associated steatotic liver disease
OP Osteoporosis
GEO Gene Expression Omnibus
GO Gene Ontology
BP Biological process
CC Cellular component
MF Molecular function
KEGG Kyoto Encyclopedia of Genes and Genomes
MR Mendelian randomization
GWAS Genome-wide association study
eQTL Expression quantitative trait locus
LD Linkage disequilibrium
KO Knockout
CTD Comparative Toxicogenomics Database
miRDB microRNA target prediction database
STROBE-MR Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization
MUC20 Mucin 20, cell surface associated
TREM2 Triggering receptor expressed on myeloid cells 2
HGF Hepatocyte growth factor
MET MET proto-oncogene, receptor tyrosine kinase
cGAS Cyclic GMP-AMP synthase
STING Stimulator of interferon genes
Open in a new tab

Supplementary Materials

The following supporting information can be downloaded at: https://www.mdpi.com/article/10.3390/ijms27125453/s1.

ijms-27-05453-s001.zip (1.8MB, zip)

Author Contributions

G.J. and H.X. contributed to the study conception and design, and supervised the entire project. H.J. and X.Y. collected and curated the research data. H.J. performed the statistical analysis. H.J. drafted the initial manuscript. G.J. and H.X. revised the manuscript critically for important intellectual content. All authors have read and agreed to the published version of the manuscript.

Institutional Review Board Statement

The study was conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the Second Affiliated Hospital of Hunan University of Chinese Medicine (protocol code: 2024-KY03; date of approval: 15 March 2024).

Informed Consent Statement

Written informed consent was obtained from all participants prior to enrollment.

Data Availability Statement

The public single-cell transcriptomic datasets analyzed in this study are available from the Gene Expression Omnibus under accession numbers GSE147287 and GSE289173. Genetic association and eQTL data used for Mendelian randomization were obtained from publicly available FinnGen/OpenGWAS and eQTLGen resources.

Conflicts of Interest

The authors declare no conflicts of interest.

Funding Statement

This work was supported by the Hunan Provincial Natural Science Foundation (Youth Project; No. 2024JJ6342) and the Changsha Municipal Natural Science Foundation (No. kq2402177).

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

References

  • 1.Quek J., Chan K.E., Wong Z.Y., Tan C., Tan B., Lim W.H., Tan D.J.H., Tang A.S.P., Tay P., Xiao J., et al. Global prevalence of non-alcoholic fatty liver disease and non-alcoholic steatohepatitis in the overweight and obese population: A systematic review and meta-analysis. Lancet Gastroenterol. Hepatol. 2023;8:20–30. doi: 10.1016/s2468-1253(22)00317-x. [DOI] [PubMed] [Google Scholar]
  • 2.Le M.H., Yeo Y.H., Li X., Li J., Zou B., Wu Y., Ye Q., Huang D.Q., Zhao C., Zhang J., et al. 2019 Global NAFLD Prevalence: A Systematic Review and Meta-analysis. Clin. Gastroenterol. Hepatol. 2022;20:2809–2817.e28. doi: 10.1016/j.cgh.2021.12.002. [DOI] [PubMed] [Google Scholar]
  • 3.Castagna A., Pujia C., Mazza E., Maurotti S., Ferro Y., Rizzo V., Formica M., Conforto R., Mercuri C., Sciacqua A., et al. Underlying Mechanisms of Osteoporosis in the Context of Multimorbidity: Clinical Challenges and Management Strategies. Nutrients. 2026;18:262. doi: 10.3390/nu18020262. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.GBD 2019 Fracture Collaborators Global, regional, and national burden of bone fractures in 204 countries and territories, 1990–2019: A systematic analysis from the Global Burden of Disease Study 2019. Lancet Healthy Longev. 2021;2:e580–e592. doi: 10.1016/s2666-7568(21)00172-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Li X., Luo W., Chen K., Peng Y. Association of metabolic dysfunction-associated steatotic liver disease with bone health in adults: A systematic review and meta-analysis of observational studies. Front. Endocrinol. 2025;16:1717852. doi: 10.3389/fendo.2025.1717852. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Pan B., Cai J., Zhao P., Liu J., Fu S., Jing G., Niu Q., Li Q. Relationship between prevalence and risk of osteoporosis or osteoporotic fracture with non-alcoholic fatty liver disease: A systematic review and meta-analysis. Osteoporos. Int. 2022;33:2275–2286. doi: 10.1007/s00198-022-06459-y. [DOI] [PubMed] [Google Scholar]
  • 7.Zheng M., Xu J., Feng Z. Association between nonalcoholic fatty liver disease and bone mineral density: Mendelian randomization and mediation analysis. Bone Rep. 2024;22:101785. doi: 10.1016/j.bonr.2024.101785. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Alabdulaali B., Al-Rashed F., Al-Onaizi M., Kandari A., Razafiarison J., Tonui D., Williams M.R., Blériot C., Ahmad R., Alzaid F. Macrophages and the development and progression of non-alcoholic fatty liver disease. Front. Immunol. 2023;14:1195699. doi: 10.3389/fimmu.2023.1195699. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Zhang W., Lang R. Macrophage metabolism in nonalcoholic fatty liver disease. Front. Immunol. 2023;14:1257596. doi: 10.3389/fimmu.2023.1257596. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Weivoda M.M., Bradley E.W. Macrophages and Bone Remodeling. J. Bone Miner. Res. 2023;38:359–369. doi: 10.1002/jbmr.4773. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Miron R.J., Bohner M., Zhang Y., Bosshardt D.D. Osteoinduction and osteoimmunology: Emerging concepts. Periodontology. 2000;94:9–26. doi: 10.1111/prd.12519. [DOI] [PubMed] [Google Scholar]
  • 12.Hao Y., Hao S., Andersen-Nissen E., Mauck W.M., 3rd, Zheng S., Butler A., Lee M.J., Wilk A.J., Darby C., Zager M., et al. Integrated analysis of multimodal single-cell data. Cell. 2021;184:3573–3587.e29. doi: 10.1016/j.cell.2021.04.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Giger N.V., Wehrle E. Advances in Spatial Transcriptomics in Bone. Curr. Osteoporos. Rep. 2026;24:3. doi: 10.1007/s11914-025-00949-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Abo Nahas H.H., Al-Dakhil A., Mohamed D.I., Yousef T.A., Almaaty A.H.A., El-Deen I.M., Sembawa H.A.M., Saied E.M. Rebamipide Reprograms Hepatic Networks to Prevent and Reverse Metabolic-Dysfunction-Associated Steatotic Liver Disease: Multi-Omics Insights and Histological Validation. Pharmaceuticals. 2026;19:559. doi: 10.3390/ph19040559. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Skrivankova V.W., Richmond R.C., Woolf B.A.R., Yarmolinsky J., Davies N.M., Swanson S.A., VanderWeele T.J., Higgins J.P.T., Timpson N.J., Dimou N., et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement. JAMA. 2021;326:1614–1621. doi: 10.1001/jama.2021.18236. [DOI] [PubMed] [Google Scholar]
  • 16.Hu C., Li T., Xu Y., Zhang X., Li F., Bai J., Chen J., Jiang W., Yang K., Ou Q., et al. CellMarker 2.0: An updated database of manually curated cell markers in human/mouse and web tools based on scRNA-seq data. Nucleic Acids Res. 2023;51:D870–D876. doi: 10.1093/nar/gkac947. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Upala S., Jaruvongvanich V., Wijarnpreecha K., Sanguankeo A. Nonalcoholic fatty liver disease and osteoporosis: A systematic review and meta-analysis. J. Bone Miner. Metab. 2017;35:685–693. doi: 10.1007/s00774-016-0807-2. [DOI] [PubMed] [Google Scholar]
  • 18.Wei Y., Li J. Mendelian randomization analysis reveals no causal association between non-alcoholic fatty liver disease and hepatocellular carcinoma: Implications for lipid metabolomics and shared pathophysiological mechanisms. Discov. Oncol. 2025;16:1765. doi: 10.1007/s12672-025-03582-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Chen J., Zhou H., Jin H., Liu K. Role of Inflammatory Factors in Mediating the Effect of Lipids on Nonalcoholic Fatty Liver Disease: A Two-Step, Multivariable Mendelian Randomization Study. Nutrients. 2022;14:4434. doi: 10.3390/nu14204434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Montero P., Roger I., Milara J., Cortijo J. Immunomodulatory properties of transmembrane mucins: From chronic diseases to cancer. Physiol. Rev. 2025;105:2233–2304. doi: 10.1152/physrev.00034.2024. [DOI] [PubMed] [Google Scholar]
  • 21.Wang W., Li S., Liu Y., Ding X., Yang Y., Chen S., Cao J., Tacke F., Dong W., Lan T. Macrophage heterogeneity in liver fibrosis. Front. Immunol. 2025;16:1639455. doi: 10.3389/fimmu.2025.1639455. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Ganguly S., Rosenthal S.B., Ishizuka K., Troutman T.D., Rohm T.V., Khader N., Aleman-Muench G., Sano Y., Archilei S., Soroosh P., et al. Lipid-associated macrophages’ promotion of fibrosis resolution during MASH regression requires TREM2. Proc. Natl. Acad. Sci. USA. 2024;121:e2405746121. doi: 10.1073/pnas.2405746121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Hundertmark J., Berger H., Tacke F. Single Cell RNA Sequencing in NASH. Methods Mol. Biol. 2022;2455:181–202. doi: 10.1007/978-1-0716-2128-8_15. [DOI] [PubMed] [Google Scholar]
  • 24.Wang Y., Wang J., Zhang J., Wang Y., Wang Y., Kang H., Zhao W., Bai W., Miao N., Wang J. Stiffness sensing via Piezo1 enhances macrophage efferocytosis and promotes the resolution of liver fibrosis. Sci. Adv. 2024;10:eadj3289. doi: 10.1126/sciadv.adj3289. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 25.Hipolito V.E.B., Ospina-Escobar E., Botelho R.J. Lysosome remodelling and adaptation during phagocyte activation. Cell. Microbiol. 2018;20:e12824. doi: 10.1111/cmi.12824. [DOI] [PubMed] [Google Scholar]
  • 26.Tran S., Baba I., Poupel L., Dussaud S., Moreau M., Gélineau A., Marcelin G., Magréau-Davy E., Ouhachi M., Lesnik P., et al. Impaired Kupffer Cell Self-Renewal Alters the Liver Response to Lipid Overload during Non-alcoholic Steatohepatitis. Immunity. 2020;53:627–640.e5. doi: 10.1016/j.immuni.2020.06.003. [DOI] [PubMed] [Google Scholar]
  • 27.Kannan R., Carruthers N.J., Koh A.J., Kleer G.G., Nakamura K., Parker S.C.J., McCauley L.K., Roca H. Efferocytosis-induced metabolic shift in bone macrophages drives lactate production and modulates inflammation and osteoclastogenesis. Front. Immunol. 2025;16:1650465. doi: 10.3389/fimmu.2025.1650465. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Orvalho J.M., Fernandes J.C.H., Moraes Castilho R., Fernandes G.V.O. The Macrophage’s Role on Bone Remodeling and Osteogenesis: A Systematic Review. Clin. Rev. Bone Miner. Metab. 2023;21:1–13. doi: 10.1007/s12018-023-09286-9. [DOI] [Google Scholar]
  • 29.Batoon L., Millard S.M., Raggatt L.J., Wu A.C., Kaur S., Sun L.W.H., Williams K., Sandrock C., Ng P.Y., Irvine K.M., et al. Osteal macrophages support osteoclast-mediated resorption and contribute to bone pathology in a postmenopausal osteoporosis mouse model. J. Bone Miner. Res. 2021;36:2214–2228. doi: 10.1002/jbmr.4413. [DOI] [PubMed] [Google Scholar]
  • 30.Luecken M.D., Büttner M., Chaichoompu K., Danese A., Interlandi M., Mueller M.F., Strobl D.C., Zappia L., Dugas M., Colomé-Tatché M., et al. Benchmarking atlas-level data integration in single-cell genomics. Nat. Methods. 2022;19:41–50. doi: 10.1038/s41592-021-01336-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Tran H.T.N., Ang K.S., Chevrier M., Zhang X., Lee N.Y.S., Goh M., Chen J. A benchmark of batch-effect correction methods for single-cell RNA sequencing data. Genome Biol. 2020;21:12. doi: 10.1186/s13059-019-1850-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Zhou Y., Ni Y., Wang Z., Prud’homme G.J., Wang Q. Causal effects of non-alcoholic fatty liver disease on osteoporosis: A Mendelian randomization study. Front. Endocrinol. 2023;14:1283739. doi: 10.3389/fendo.2023.1283739. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Bowden J., Davey Smith G., Burgess S. Mendelian randomization with invalid instruments: Effect estimation and bias detection through Egger regression. Int. J. Epidemiol. 2015;44:512–525. doi: 10.1093/ije/dyv080. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Skrivankova V.W., Richmond R.C., Woolf B.A.R., Davies N.M., Swanson S.A., VanderWeele T.J., Timpson N.J., Higgins J.P.T., Dimou N., Langenberg C., et al. Strengthening the reporting of observational studies in epidemiology using mendelian randomisation (STROBE-MR): Explanation and elaboration. BMJ. 2021;375:n2233. doi: 10.1136/bmj.n2233. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Verbanck M., Chen C.Y., Neale B., Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat. Genet. 2018;50:693–698. doi: 10.1038/s41588-018-0099-7. Erratum in Nat. Genet. 2018, 50, 1196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Osorio D., Zhong Y., Li G., Xu Q., Yang Y., Tian Y., Chapkin R.S., Huang J.Z., Cai J.J. scTenifoldKnk: An efficient virtual knockout tool for gene function predictions via single-cell gene regulatory network perturbation. Patterns. 2022;3:100434. doi: 10.1016/j.patter.2022.100434. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Kamimoto K., Stringa B., Hoffmann C.M., Jindal K., Solnica-Krezel L., Morris S.A. Dissecting cell identity via network inference and in silico gene perturbation. Nature. 2023;614:742–751. doi: 10.1038/s41586-022-05688-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Zhao Y., Ye W., Wang Y.D., Chen W.D. HGF/c-Met: A Key Promoter in Liver Regeneration. Front. Pharmacol. 2022;13:808855. doi: 10.3389/fphar.2022.808855. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Wang X., Shi Y., Shi H., Liu X., Liao A., Liu Z., Orlowski R.Z., Zhang R., Wang H. MUC20 regulated by extrachromosomal circular DNA attenuates proteasome inhibitor resistance of multiple myeloma by modulating cuproptosis. J. Exp. Clin. Cancer Res. 2024;43:68. doi: 10.1186/s13046-024-02972-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Xue B., Guo W.M., Jia J.D., Kadeerhan G., Liu H.P., Bai T., Shao Y., Wang D.W. MUC20 as a novel prognostic biomarker in ccRCC correlating with tumor immune microenvironment modulation. Am. J. Cancer Res. 2022;12:695–712. [PMC free article] [PubMed] [Google Scholar]
  • 41.Fu L., Yonemura A., Yasuda-Yoshihara N., Umemoto T., Zhang J., Yasuda T., Uchihara T., Akiyama T., Kitamura F., Yamashita K., et al. Intracellular MUC20 variant 2 maintains mitochondrial calcium homeostasis and enhances drug resistance in gastric cancer. Gastric Cancer. 2022;25:542–557. doi: 10.1007/s10120-022-01283-z. [DOI] [PubMed] [Google Scholar]
  • 42.Woodward A.M., Argüeso P. Expression analysis of the transmembrane mucin MUC20 in human corneal and conjunctival epithelia. Investig. Ophthalmol. Vis. Sci. 2014;55:6132–6138. doi: 10.1167/iovs.14-15269. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Luo X.X., Li S.Z., Wang L., Luo A.L., Qiu H., Yuan X.L. Prognostic role of MUCIN family and its relationship with immune characteristics and tumor biology in diffuse-type gastric cancer. Heliyon. 2024;10:e31403. doi: 10.1016/j.heliyon.2024.e31403. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Su Y.H., Chien K.L., Yang S.H., Chia W.T., Chen J.H., Chen Y.C. Nonalcoholic Fatty Liver Disease Is Associated With Decreased Bone Mineral Density in Adults: A Systematic Review and Meta-Analysis. J. Bone Miner. Res. 2023;38:1092–1103. doi: 10.1002/jbmr.4862. [DOI] [PubMed] [Google Scholar]
  • 45.Chen S.T., Kuo T.C., Liao Y.Y., Lin M.C., Tien Y.W., Huang M.C. Silencing of MUC20 suppresses the malignant character of pancreatic ductal adenocarcinoma cells through inhibition of the HGF/MET pathway. Oncogene. 2018;37:6041–6053. doi: 10.1038/s41388-018-0403-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Khanmohammadi S., Kuchay M.S. Effects of Metabolic Dysfunction-Associated Steatotic Liver Disease on Bone Density and Fragility Fractures: Associations and Mechanisms. J. Obes. Metab. Syndr. 2024;33:108–120. doi: 10.7570/jomes24004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Gao H., Peng X., Li N., Gou L., Xu T., Wang Y., Qin J., Liang H., Ma P., Li S., et al. Emerging role of liver-bone axis in osteoporosis. J. Orthop. Transl. 2024;48:217–231. doi: 10.1016/j.jot.2024.07.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Wang X., Wei W., Krzeszinski J.Y., Wang Y., Wan Y. A Liver-Bone Endocrine Relay by IGFBP1 Promotes Osteoclastogenesis and Mediates FGF21-Induced Bone Resorption. Cell Metab. 2015;22:811–824. doi: 10.1016/j.cmet.2015.09.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Davis A.P., Wiegers T.C., Johnson R.J., Sciaky D., Wiegers J., Mattingly C.J. Comparative Toxicogenomics Database (CTD): Update 2023. Nucleic Acids Res. 2023;51:D1257–D1262. doi: 10.1093/nar/gkac833. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Wang S.S., Wang C.C., Wang C.L., Lin Y.C., Tung C.W. Incorporating Tissue-Specific Gene Expression Data to Improve Chemical-Disease Inference of in Silico Toxicogenomics Methods. J. Xenobiotics. 2024;14:1023–1035. doi: 10.3390/jox14030057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Cheng F., Desai R.J., Handy D.E., Wang R., Schneeweiss S., Barabási A.L., Loscalzo J. Network-based approach to prediction and population-based validation of in silico drug repurposing. Nat. Commun. 2018;9:2691. doi: 10.1038/s41467-018-05116-5. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Iglesias-Velazquez O., Tresguerres F.G.F., Tresguerres I.F., Leco-Berrocal I., Lopez-Pintor R., Baca L., Torres J. OsteoMac: A new player on the bone biology scene. Ann. Anat.-Anat. Anz. 2024;254:152244. doi: 10.1016/j.aanat.2024.152244. [DOI] [PubMed] [Google Scholar]
  • 53.Squair J.W., Gautier M., Kathe C., Anderson M.A., James N.D., Hutson T.H., Hudelle R., Qaiser T., Matson K.J.E., Barraud Q., et al. Confronting false discoveries in single-cell differential expression. Nat. Commun. 2021;12:5692. doi: 10.1038/s41467-021-25960-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Clough E., Barrett T., Wilhite S.E., Ledoux P., Evangelista C., Kim I.F., Tomashevsky M., Marshall K.A., Phillippy K.H., Sherman P.M., et al. NCBI GEO: Archive for gene expression and epigenomics data sets: 23-year update. Nucleic Acids Res. 2024;52:D138–D144. doi: 10.1093/nar/gkad965. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 55.Wang Z., Li X., Yang J., Gong Y., Zhang H., Qiu X., Liu Y., Zhou C., Chen Y., Greenbaum J., et al. Single-cell RNA sequencing deconvolutes the in vivo heterogeneity of human bone marrow-derived mesenchymal stem cells. Int. J. Biol. Sci. 2021;17:4192–4206. doi: 10.7150/ijbs.61950. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Hong S.E., Mun S.J., Lee Y.J., Yoo T., Suh K.S., Kang K.W., Son M.J., Kim W., Choi M. Single-cell eQTL analysis identifies genetic variation underlying metabolic dysfunction-associated steatohepatitis. Nat. Genet. 2025;57:1638–1648. doi: 10.1038/s41588-025-02237-8. [DOI] [PubMed] [Google Scholar]
  • 57.Sanderson E., Glymour M.M., Holmes M.V., Kang H., Morrison J., Munafò M.R., Palmer T., Schooling C.M., Wallace C., Zhao Q., et al. Mendelian randomization. Nat. Rev. Methods Primers. 2022;2:6. doi: 10.1038/s43586-022-00099-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Kurki M.I., Karjalainen J., Palta P., Sipilä T.P., Kristiansson K., Donner K.M., Reeve M.P., Laivuori H., Aavikko M., Kaunisto M.A., et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613:508–518. doi: 10.1038/s41586-022-05473-8. Correction in Nature 2023, 615, E19. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Kerimov N., Hayhurst J.D., Peikova K., Manning J.R., Walter P., Kolberg L., Samoviča M., Sakthivel M.P., Kuzmin I., Trevanion S.J., et al. A compendium of uniformly processed human gene expression and splicing quantitative trait loci. Nat. Genet. 2021;53:1290–1299. doi: 10.1038/s41588-021-00924-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Kerimov N., Tambets R., Hayhurst J.D., Rahu I., Kolberg P., Raudvere U., Kuzmin I., Chowdhary A., Vija A., Teras H.J., et al. eQTL Catalogue 2023: New datasets, X chromosome QTLs, and improved detection and visualisation of transcript-level QTLs. PLoS Genet. 2023;19:e1010932. doi: 10.1371/journal.pgen.1010932. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 61.Haycock P.C., Borges M.C., Burrows K., Lemaitre R.N., Harrison S., Burgess S., Chang X., Westra J., Khankari N.K., Tsilidis K.K., et al. Design and quality control of large-scale two-sample Mendelian randomization studies. Int. J. Epidemiol. 2023;52:1498–1521. doi: 10.1093/ije/dyad018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 62.Szklarczyk D., Kirsch R., Koutrouli M., Nastou K., Mehryary F., Hachilif R., Gable A.L., Fang T., Doncheva N.T., Pyysalo S., et al. The STRING database in 2023: Protein-protein association networks and functional enrichment analyses for any sequenced genome of interest. Nucleic Acids Res. 2023;51:D638–D646. doi: 10.1093/nar/gkac1000. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Wu T., Hu E., Xu S., Chen M., Guo P., Dai Z., Feng T., Zhou L., Tang W., Zhan L., et al. clusterProfiler 4.0: A universal enrichment tool for interpreting omics data. Innovation. 2021;2:100141. doi: 10.1016/j.xinn.2021.100141. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Aleksander S.A., Balhoff J., Carbon S., Cherry J.M., Drabkin H.J., Ebert D., Feuermann M., Gaudet P., Harris N.L., Hill D.P., et al. The Gene Ontology knowledgebase in 2023. Genetics. 2023;224:iyad031. doi: 10.1093/genetics/iyad031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 65.Kanehisa M., Furumichi M., Sato Y., Kawashima M., Ishiguro-Watanabe M. KEGG for taxonomy-based analysis of pathways and genomes. Nucleic Acids Res. 2023;51:D587–D592. doi: 10.1093/nar/gkac963. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Huang H.Y., Lin Y.C., Cui S., Huang Y., Tang Y., Xu J., Bao J., Li Y., Wen J., Zuo H., et al. miRTarBase update 2022: An informative resource for experimentally validated miRNA-target interactions. Nucleic Acids Res. 2022;50:D222–D230. doi: 10.1093/nar/gkab1079. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ijms-27-05453-s001.zip (1.8MB, zip)

Data Availability Statement

The public single-cell transcriptomic datasets analyzed in this study are available from the Gene Expression Omnibus under accession numbers GSE147287 and GSE289173. Genetic association and eQTL data used for Mendelian randomization were obtained from publicly available FinnGen/OpenGWAS and eQTLGen resources.

Articles from International Journal of Molecular Sciences are provided here courtesy of Multidisciplinary Digital Publishing Institute (MDPI)

RESOURCES