IntroductionObesity has emerged as a global crisis, with projections indicating that more than 1 billion individuals will be living with the condition by 2030 (Bathina and Armamento-Villareal, 2023[5]). As a complex metabolic disorder, obesity is intricately associated with a wide range of comorbidities, including type 2 diabetes, metabolic dysfunction-associated fatty liver disease (MAFLD), lower circulating 25‑hydroxyvitamin D (25(OH)D), and disrupted bone homeostasis (Kawai et al., 2021[30]). Paradoxically, despite higher body weight and increased mechanical loading, obese individuals are at greater risk for fractures (Compston et al., 2011[12]; Prieto-Alhambra et al., 2012[42]; Gkastaris et al., 2020[25]), suggesting that factors beyond body weight influence skeletal integrity. Clinical and experimental evidence increasingly points to a complex interplay between MAFLD, obesity, and bone fragility (Zhao et al., 2023[63]; Gao et al., 2024[21]). However, the mechanisms underlying the elevated fracture risk associated with obesity remain unknown.
Vitamin D, a steroid hormone synthesized in the skin via UVB-induced conversion or obtained from dietary sources, plays a crucial role in regulating calcium-phosphate homeostasis and maintaining skeletal integrity (Akter et al., 2022[2]). The liver plays a central role in vitamin D metabolism by catalyzing the first hydroxylation step, converting vitamin D into 25(OH)D (Wintermeyer et al., 2016[59]), the major circulating form and substrate for renal 1α-hydroxylation to generate the active metabolite 1,25-dihydroxyvitamin D (1,25(OH)₂D). This active form regulates calcium-phosphate homeostasis and bone remodeling (Wintermeyer et al., 2016[59]; Ehnert et al., 2019[16]). Clinically, 25(OH)D is the most commonly used biomarker of vitamin D status. However, renal 1α‑hydroxylation is tightly regulated by phosphate, parathyroid hormone (PTH), and fibroblast growth factor‑23 (FGF23), and therefore does not necessarily correlate with 25(OH)D across physiological or disease states (Jacquillet and Unwin, 2019[28]; Latic and Erben, 2021[32]). Therefore, inferences about functional vitamin D sufficiency from 25(OH)D alone must be made with caution (Serdar et al., 2024[48]). Nevertheless, lower 25(OH)D levels are frequently observed in obesity, with a substantial proportion of individuals falling below 30 ng/mL (Via, 2012[57]; Turer et al., 2013[55]). Observational studies suggest that every 10 ng/mL increase in serum 25(OH)D concentration is associated with a 7 % reduction in the risk of any fracture and a 20 % reduction in the risk of hip fracture (Yao et al., 2019[60]). Several mechanisms have been proposed to explain this consistent reduction in circulating 25(OH)D, including the sequestration of vitamin D by excess body fat, reduced sun exposure, chronic inflammation, and gut microbiota dysbiosis (Bennour et al., 2022[6]). However, the evidence remains limited and inconclusive.
Beyond vitamin D metabolism, MAFLD-related hepatic dysfunction may influence bone health through additional mechanisms, including chronic inflammation, altered lipid handling, and impaired bile acid synthesis, which contributes to dietary vitamin D absorption. These processes, combined with systemic metabolic stress, may disrupt bone remodeling and favor resorption. Despite these associations, the mechanistic link between long-term WD-induced hepatic injury and bone remodeling remains poorly understood. Most preclinical studies employ short-term dietary interventions, which fail to capture chronic metabolic adaptations relevant to human disease. To address these gaps, we established a 48-week high-fat, high-carbohydrate WD model that induces obesity and MAFLD-like hepatic injury. The WD model has been comprehensively characterized in a previous study by Ghallab et al. (2021[24]), who identified a sequence of pathological events during MAFLD progression in the same cohort of mice, closely mirroring the progression of human disease. Building on this established dataset not only ensures translational relevance but also adheres to the principles of the 3Rs by reducing the need for additional animal cohorts.
In this study, we investigated the hepatic vitamin D 25-hydroxylases, vitamin D-binding protein (DBP) expression, circulating concentrations of 25(OH)D, skeletal VDR abundance and associated bone alterations. Our objective is to elucidate how hepatic dysfunction contributes to disordered vitamin D metabolism and skeletal fragility in individuals with obesity and MAFLD, thereby highlighting potential therapeutic targets for the clinical management of obesity-related osteoporosis.
Materials and MethodsAnimals24 male, 8 weeks old C57BL/6N mice were obtained from Janvier Labs, Le Genest-Saint-Isle, France. Mice were randomly assigned to either the SD or WD group using a random number generator. 12 were fed a SD (Ssniff R/M-H, 10 mm Standard diet (Ssniff, Soest, Germany)), 12 were fed a high-carbohydrate, high-fat and high-cholesterol WD (Research diets, Inc., #D16022301, New Brunswick, NJ, USA) (Ghallab et al., 2021[24]). Table 1(Tab. 1) summarizes the nutrient composition per 100 g as provided by the manufacturer. The mice were housed individually with free access to food and water, 25 °C temperature, and under 12 h light/dark cycles. They were weighed weekly and sacrificed at week 48. The femora and tibiae were harvested, soft tissue was removed, the bones were weighed and measured, and were individually frozen in Eppendorf tubes at -80 °C. All experiments were approved by the local governmental animal welfare committee (LANUV, North Rhine-Westphalia, Germany; application number: 81-02.04. 2020.A304).
Plasma biochemical and metabolite analysisAlanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) were quantified in freshly collected heparinized cardiac blood using a Piccolo Xpress Chemistry Analyzer (Abaxis/Hitado) together with the Piccolo General Chemistry 13 Panel Kit (Hitado, AB-114-400-0029). Plasma concentration of 25(OH)D was determined using commercial ELISA kit (Immunodiagnostic Systems, AC-57SF1) following the manufacturers' instructions. Absorbance was recorded at 450 nm using an Omega plate reader, and analyte concentrations were calculated based on standard curves generated from known 25(OH)D standards.
µCT analysisThe right femora were stored in Eppendorf tubes and scanned using a Skyscan (1172, Bruker, Billerica, MA USA) at a spatial resolution of 9 µm. A standardized scanning setup was used: tube voltage: 50 kV, exposure time: 1150 ms, current: 500 µA, rotation steps: 0.30 °, filter: 0.5 mm aluminum. Images were reconstructed using nRecon (Bruker, Billerica, MA, US). For analysis, CTAnalyser (Bruker, Billerica, MA, USA) was used. For calibration of gray values phantom rods (calcium hydroxyapatite (CaHA)) with known bone mineral density values (0.250 g and 0.750 g (CaHA/cm3)) were used. The distal femoral growth plate was identified as reference. The region of interest (ROI) for cancellous bone analysis was starting 0.45 mm (50 slices) proximal to the reference and extending 0.54 mm (60 slices). The ROIs were contoured manually and interpolated between each transversal slide. The tissue volume (TV, (µm3)), bone volume (BV, (µm3)), the bone volume fraction of the tissue volume (BV/TV), the trabecular number (Tb. N (1/μm)), the trabecular separation (Tb. Sp (μm)) and the trabecular thickness (Tb. Th (μm)) from trabecular ROIs were determined individually.
Biomechanical analysisThe right tibiae were brought to room temperature. Biomechanical testing was performed by a three-point bending device (Zwick/Roell Z2.5, ZwickRoell GmbH & Co. KG, Ulm, Germany). For standardized measuring conditions, all tibiae were placed on the ventral side downwards with a 10 mm distance between the supports. The measurement was carried out with a constant speed of 0.15 mm/s, orthogonal to the axis of the tibiae. Loading was automatically stopped when the bones got fractured.
HistomorphometryFollowing µCT scanning, the right femora were subjected to histological analysis. The bones were initially fixed in IHC zinc fixative (BD Pharmingen, San Diego, CA) for 24 h. Subsequently, they were decalcified in 13 % EDTA solution for two weeks and then embedded in paraffin. Longitudinal sections, 5 µm thick, were cut from the bones for further staining. For immunohistochemical staining, sections were incubated for 1 h with a monoclonal rabbit anti-mouse antibody against perilipin (1:200; Cell Signaling Technology, Danvers, MA, USA) and a rabbit anti-CD163 (1:300; Abcam, Cambridge, UK) as the primary antibody. After washing, the primary antibody was detected using A goat anti-rabbit peroxidase-labelled secondary antibody (1:100; Jackson ImmunoResearch Laboratories, West Grove, PA, USA).
Images were acquired using an EVOS® Digital Inverted Microscope at 40× magnification. Four to eight fields from the proximal femur were randomly selected and analyzed using QuPath (Version 0.5.1). Background correction was applied using the “estimate stain vectors” function. Positive cells were defined as nuclei surrounded by specific immunoreactive staining and detected using the “Positive Cell Detection” tool. For nuclear size quantification, images were imported into ImageJ (NIH, Bethesda, MD, USA). Eight nuclei per section were randomly selected and manually segmented using the Wand tool. Nuclear cross-sectional area was measured, and mean values were calculated for each group. This approach was chosen because cytoplasmic boundaries are indistinct in bone marrow tissue, making the nuclear area a reliable surrogate for cell size.
RNA isolation and RT-PCR analysisThe left femora and tibiae were frosted with liquid nitrogen in a pre-cooled mortar and crushed into small pieces. The pieces were transferred into a tube and covered with 500 µL of self-made Trifast (0.8 M guanidine thiocyanate, 0.4 M ammonium thiocyanate, 0.1 M sodium acetate solution, 5 % glycerol, 38 % phenol). After overnight freezing at -80°C, 100 µL of chloroform (per 500 µL of self-made Trifast) were added, thoroughly mixed and incubated for 5 min. After centrifugation at 14,000 × g for 10 min, the upper clear phase was mixed with 100 % ethanol to a final ethanol concentration of 40 %. The RNA isolation was continued with the All-In-One DNA/RNA/Protein Miniprep Kit (Bio Basic Inc., Canada) and carried out according to the manufacturer's instructions. During the isolation genomic DNA was digested with DNAse I (EN0521, Thermo Fisher, USA). RNA was quantified with the LVIS Plate on an Omega Plate Reader (BMG Labtech). RNA was transcribed into cDNA with the First Strand cDNA Synthesis Kit (Thermo Fisher) according to the manufacturer's instructions. RT-PCR was performed using 2× Red Taq Master Mix (Biozym, Oldendorf, Germany). Primer pairs were designed from NCBI GenBank sequences using Primer-BLAST, and optimized amplification conditions for each gene target are summarized in Supplementary information (Supplementary Table 1). Each biological sample was analyzed in technical duplicate, and the mean normalized gray value was used for statistical analysis. This semi-quantitative approach allowed for relative comparison of gene expression levels between experimental groups.
Western blotThe right tibiae were crushed into small pieces with liquid nitrogen in a pre-cooled mortar. The pieces were then transferred into microcentrifuge tubes and resuspended in 300 μL of ice-cold RIPA lysis buffer (10 mM TRIS, 100 mM NaCl, 0.5 % Tergitol NP40, 0.5 % deoxycholic acid, 10 mM EDTA, protease and phosphatase inhibitors, pH = 7.6). Samples were incubated on ice for 30 minutes to ensure complete lysis, followed by centrifugation at 13,000 × g for 10 minutes at 4 °C to remove insoluble debris. Protein concentrations in the supernatant were measured using the Lowry assay. Equal amounts of protein (40 μg per lane) were loaded on 10 % SDS-PAGE gels and electro-transferred onto nitrocellulose membranes using a standard wet-transfer system. Membranes were stained with Ponceau S to verify transfer efficiency and subsequently blocked in 5 % bovine serum albumin (BSA) dissolved in TBS-T for 1 hour at room temperature. Primary antibody incubation was performed overnight at 4 °C using antibodies against VDR (1:500, Santa Cruz Biotechnology, #13133) and HPRT (1:500, Santa Cruz Biotechnology, #376938), the latter serving as the loading control. After washing with TBS-T, membranes were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies anti-mouse lgG Kappa BP (1:5000, Santa Cruz Biotechnology, #516102) for 2 hours. Protein bands were visualized using an enhanced chemiluminescence (ECL) detection system and imaged with the INTAS Chemocam (INTAS, Göttingen, Germany).
RNA-Seq analysis and bioinformaticsLiver tissues were collected from SD and WD-fed mice (N = 5 per group). Sample preparation and RNA-Seq analysis were performed as previously described (Ghallab et al., 2021[24]). Differential gene expression analysis was conducted using the DESeq2 package (v1.46.0) in R (v4.4.2). Genes with an absolute log2 fold change (|log2FC|) ≥ 2 and a false discovery rate (FDR)-adjusted p-value ≤ 0.001 were considered DEGs. For functional enrichment analysis, DEGs were analyzed using the GO and KEGG databases through the g:Profiler (Kolberg et al., 2023[31]). Additionally, the Gene Set Enrichment Analysis (GSEA) was performed using the software (v4.4.0), with defined vitamin D metabolism gene set (Cyp2r1, Cyp27a1, Cyp24a1, Gc, Lrp2, Dhcr7, Rxra, Vdr, Cyp27b1) (Subramanian et al., 2005[53]).
StatisticsStatistical analyses were performed using GraphPad Prism (GraphPad Software 9.0.0, La Jolla, CA, USA). Outliers were identified and removed using the ROUT method (Q = 1 %). Data are presented as box-and-whisker plots. Group comparisons were conducted using the Mann-Whitney U test. P<0.05 was considered statistically significant. Post-hoc power analyses were conducted for each experiment, and the exact sample sizes for all analyses are provided in Supplementary information (Supplementary Table 2).
ResultsLong-term WD feeding led to significant weight gain and liver damage compared to SD controls. Bone structure, particularly trabecular microarchitecture, worsened, but three-point bending tests suggested no significant impairment in cortical bone mechanical properties. Molecular studies revealed increased osteoclast activity without notable changes in osteoblast markers. Histological analyses further demonstrated an expansion of bone marrow adipose tissue and an increased abundance of CD163-positive macrophage lineage cells in WD-fed mice. Liver analysis showed downregulation of vitamin D 25-hydroxylases and related pathways, which was accompanied by reduced circulating 25-hydroxyvitamin D levels. Additionally, bone tissue demonstrated reduced VDR protein levels.
Long-term WD feeding induces obesity and hepatic injuryProlonged WD feeding led to a significant increase in body weight starting from week 3 compared to the SD control group (Figure 2A(Fig. 2)). At week 48, WD-fed mice exhibited a significantly elevated liver-to-body weight ratio (Figure 2B(Fig. 2)), accompanied by a trend toward increased random blood glucose levels (Figure 2C(Fig. 2)), suggestive of hepatic dysfunction. Biochemical assays confirmed hepatic injury, as evidenced by markedly increased plasma levels of alanine aminotransferase (ALT), aspartate aminotransferase (AST), and alkaline phosphatase (ALP) in WD-fed mice (Figure 2D-F(Fig. 2)).
WD impairs trabecular microarchitecture and reduces bone mineral densityTo evaluate the impact of obesity and hepatic injury on bone structure, we performed µCT on the distal femur. Representative µCT and 3D reconstructed images (Figure 3A-B(Fig. 3)) showed significantly damaged microarchitecture of the femur. Quantitative analysis revealed significant reductions in bone mineral density (BMD), tissue volume (TV), bone volume (BV), and the BV/TV ratio, along with a lower trabecular number and increased trabecular separation in the WD group (Figure 3C(Fig. 3)).
WD had no impact on bone mechanical propertiesA three-point bending test was performed to further assess alterations in cortical bone mechanical properties (Figure 4(Fig. 4)). A trend towards increased bending stiffness in WD-fed mice was shown, but no statistical significance was reached (Figure 4A(Fig. 4)). Consistently, the yield load-defined as the force required to initiate permanent deformation-was comparable between groups (Figure 4B(Fig. 4)). Likewise, the ultimate load, representing the maximum force sustained before fracture, showed no significant intergroup variation (Figure 4C(Fig. 4)). In addition, the work-to-fracture, an indicator of bone toughness reflecting the total energy absorbed before fracture, also remained unchanged between the two groups (Figure 4D(Fig. 4)).
WD enhances osteoclastogenesis without affecting osteoblast markersTo explore cellular changes associated with bone loss, we analyzed the expression of bone remodeling-related markers in femoral tissue. Consistent with the trabecular bone parameters observed in WD-fed mice, RT-PCR analysis showed that the expression of Csf1 (M-CSF) and Tnfrsf11a (RANK) was significantly increased in the WD group (Figure 5A-B(Fig. 5)). Acp5 (TRAP) expression exhibited a modest increasing trend; however, this difference did not reach statistical significance (Figure 5C(Fig. 5)). In contrast, the expression levels of osteoblast-related genes remained comparable between SD- and WD-fed mice (Figure 5E-F(Fig. 5)).
WD alters the bone marrow microenvironmentHistological analysis revealed that WD feeding markedly altered the bone marrow microenvironment. Perilipin immunostaining demonstrated a clear increase in marrow adipocytes in WD-fed mice compared with SD controls, with adipocytes appearing more abundant and clustered in WD samples (Figure 6A(Fig. 6)). Quantitative analyses confirmed significant elevations in perilipin-positive adipocytes in the WD group (Figure 6B(Fig. 6)). CD163 immunohistochemistry also showed an increase in CD163-positive macrophages in WD-fed mice, indicating enhanced monocyte/macrophage lineage presence within the marrow (Figure 6C(Fig. 6)). Quantitative analyses confirmed significant elevations in CD163-positive macrophages in the WD group (Figure 6D(Fig. 6)).
Liver dysfunction impairs vitamin D biosynthesis, transport and absorptionGiven that obesity alters the complex interplay between the liver and bone, we next investigated whether liver dysfunction in WD-fed mice contributes to the observed skeletal phenotype. RNA sequencing was performed on liver samples from both groups. Differential gene expression analysis was conducted using the DESeq2 package. A total of 864 differentially expressed genes (DEGs) (Supplementary information, Supplementary Table 3) were identified between the WD and SD groups.
To explore the potential biological functions of these DEGs between obese and control mice, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were conducted. In the biological process (BP) category of GO analysis, the downregulated genes in obese mice were significantly enriched in lipid metabolic processes, cholesterol biosynthetic processes, and sterol biosynthetic processes (Figure 7A(Fig. 7)). Consistently, KEGG pathway analysis revealed significant enrichment in cholesterol and steroid hormone biosynthesis pathways (Figure 7B(Fig. 7)), suggesting suppressed hepatic lipid metabolism in obesity.
In contrast, the upregulated genes were primarily enriched in immune-related pathways, including immune system processes, chemokine signaling, and osteoclast differentiation (Supplementary information, Supplementary Figure 1A and 1B).
To test the hypothesis that the observed downregulation of cholesterol and steroid hormone biosynthesis genes may imply compromised hepatic vitamin D metabolism, GSEA with a defined vitamin D metabolism gene set (Cyp2r1, Cyp27a1, Cyp24a1, Gc, Lrp2, Dhcr7, Rxra, Vdr, Cyp27b1) was performed. GSEA results revealed that this gene set was significantly enriched at the top of the ranked gene list (NES = 1.43, FDR < 0.05), suggesting a coordinated downregulation of vitamin D metabolic genes in the WD group (Figure 7C(Fig. 7)).
Further hepatic transcriptome profiling identified significant downregulation of key vitamin D metabolic enzymes, including Cyp2r1, Cyp27a1, and Gc (vitamin D-binding protein, DBP), in the WD group (Figure 7D(Fig. 7)). In addition, expression of Cyp7a1, the rate-limiting enzyme in the bile acid biosynthetic pathway in the liver, was also significantly decreased.
Impaired hepatic vitamin D metabolism disrupts systemic vitamin D availability and skeletal VDR signalingTo validate that hepatic metabolic dysfunction compromises vitamin D biosynthesis and transport, circulating 25(OH)D levels were quantified. As expected, plasma 25(OH)D concentrations were significantly decreased in the WD group compared with the SD controls (Figure 8A(Fig. 8)), consistent with the transcriptional suppression of hepatic vitamin D metabolic enzymes observed in WD-fed mice.
To further determine whether reduced systemic vitamin D availability affects skeletal vitamin D signaling, Vdr gene expression and protein levels were assessed. RT-PCR analysis revealed a significant upregulation of Vdr mRNA expression in the WD group compared to the SD group (Figure 8B(Fig. 8)). However, Western blot analysis demonstrated a pronounced reduction in VDR protein abundance in bone from obese mice (Figure 8C(Fig. 8)).
DiscussionObesity has emerged as a global health crisis and is paradoxically associated with an increased risk of bone fractures. Despite the higher mechanical loading typically expected to promote bone formation, clinical and experimental studies have consistently reported compromised bone quality in obese individuals, particularly in non-weight-bearing bones such as the upper limbs (Prieto-Alhambra et al., 2012[42]; Dimitri, 2019[15]). In this study, we utilized a long-term WD model to investigate the structural, biomechanical, and molecular alterations underlying skeletal fragility in obesity, with particular focus on the liver-vitamin D-bone axis.
The WD employed in this study contained elevated saturated fats (primarily from lard), cholesterol, refined carbohydrates (fructose and sucrose), and a high cholesterol content, alongside a moderately reduced choline level. While choline is essential for phosphatidylcholine synthesis and very-low-density lipoprotein export from hepatocytes (Sherriff et al., 2016[49]), the magnitude of reduction in this model is modest. In contrast, the 2 % cholesterol content represents a substantial dietary intervention, specifically designed to overcome the low intestinal cholesterol absorption in rodents and induce hepatic lipid accumulation (Eng and Estall, 2021[18]). This combination of dietary cholesterol and simple sugars is known to synergistically promote steatohepatitis, systemic inflammation, and metabolic dysfunction (Liang et al., 2018[34]; Stephenson et al., 2018[52]; Parisse et al., 2025[39]).
This distinction between diet composition effects and obesity per se is crucial for interpreting our findings. The bone deterioration observed likely reflects the convergent influence of both dietary components (high fat, cholesterol, refined sugars) and obesity-related factors (adipokines, systemic inflammation, insulin resistance). Recent studies support this multi-factorial model, demonstrating that different high-fat diet compositions can produce distinct metabolic and phenotypic outcomes even when caloric content is controlled (Janoschek et al., 2023[29]; Li et al., 2024[33]).
Long-term WD feeding led to significant deterioration of trabecular bone microarchitecture with reduced trabecular number and volume, increased trabecular separation, and decreased BMD. These structural impairments are hallmark features of osteoporosis (Chen and Kubo, 2014[9]). These µCT measurements reflect changes in the distal femur trabecular compartment, which is highly sensitive to metabolic disturbances. In contrast, three-point bending assessed cortical bone at the tibial diaphysis and showed no significant mechanical impairment. This divergence likely reflects the differing responsiveness of trabecular and cortical bone, as well as the potential protective influence of increased mechanical loading in obesity. Clinically, similar patterns are observed, with obesity often preserving cortical strength at weight-bearing sites despite compromised trabecular quality (Farr and Dimitri, 2017[19]). These findings highlight that trabecular deficits emerge earlier, whereas cortical mechanical changes may be delayed or site dependent, emphasizing the need to evaluate both bone quantity and quality in obesity-related skeletal fragility.
At the cellular level, the increased expression of Csf1 (M-CSF) and Tnfrsf11a (RANK) observed in femoral tissue of WD-fed mice suggests a shift toward a microenvironment that may favor osteoclast differentiation and activity. Although Acp5 (TRAP) expression showed only a non-significant upward trend, the combined changes in osteoclast-associated markers are consistent with the observed reduction in trabecular bone parameters. Notably, the expression of osteoblast-related genes was not altered, indicating that trabecular bone loss under long-term WD feeding may be primarily associated with changes in osteoclast-related pathways rather than impaired osteoblast gene expression. This imbalance was accompanied by significant expansion of bone marrow adipocytes, as evidenced by increased perilipin-positive adipocytes. The expansion of bone marrow adipose tissue (BMAT) represents more than simple space competition with osteoblasts. Bone marrow adipocytes actively suppress osteoblastogenesis through paracrine mechanisms, including NF-κB-mediated inhibition of BMP signaling (Abdallah, 2017[1]). Additionally, BMAT serves as a source of inflammatory mediators that further promote osteoclastic bone resorption (Muruganandan et al., 2020[38]). The elevated CD163 staining observed in WD-fed mice supports the notion of enhanced osteoclastogenic potential, as CD163⁺ cells represent monocyte-macrophage lineage populations that include osteoclast precursors (Gomez-Brouchet et al., 2021[26]; Bai et al., 2024[4]; Qian et al., 2024[43]). An increase in this precursor pool is likely to promote osteoclast formation and accelerate bone resorption in WD mice. This heightened resorptive activity is also consistent with the reduced bone mineral content detected by µCT analysis.
Emerging evidence has highlighted the liver-bone axis in regulating skeletal homeostasis, with reports suggesting that up to 75 % of patients with chronic liver disease exhibit severe osteoporosis (Ehnert et al., 2019[16]). The hepatic transcriptomic analysis indicated impaired hepatic vitamin D metabolic pathways in WD-fed mice. The hepatic suppression of vitamin D metabolism likely reflects the inflammatory and metabolic stress characteristic of MAFLD (Badmus et al., 2022[3], Portincasa et al., 2024[41]). Chronic inflammation is known to impair CYP enzymes activity, thereby disrupting vitamin D biotransformation. The suppression of Cyp2r1 and Cyp27a1 represents a critical bottleneck in vitamin D bioactivation, potentially linking liver metabolic impairment to bone loss (Roizen et al., 2018[44]). Several mechanisms have been proposed to explain the lower vitamin D levels observed in obese individuals, such as sequestration of vitamin D within expanded adipose tissue, reduced cutaneous synthesis due to decreased sunlight exposure, and impaired intestinal absorption (Bennour et al., 2022[6]). However, direct evidence linking hepatic metabolic impairment to vitamin D deficiency has remained limited. Vitamin D 25‐hydroxylation has been considered a constitutive, unregulated step until recently (Roizen et al., 2019[45]; Elkhwanky et al., 2020[17]). Recent human studies following gastric bypass surgery have shown that weight loss increases CYP2R1 expression in subcutaneous adipose tissue, suggesting recovery from obesity-induced suppression (Elkhwanky et al., 2020[17]). Our results align with emerging evidence that hepatic vitamin D hydroxylases are dynamically regulated and may be key drivers of systemic vitamin D deficiency in metabolic disease (Zhu et al., 2021[64]). Additionally, the expression of DBP, the main transporter of 25(OH)D in circulation, was significantly reduced in WD-fed mice. Since DBP is synthesized predominantly in the liver and is essential for vitamin D transport and bioavailability, its downregulation may exacerbate systemic vitamin D insufficiency (Safadi et al., 1999[46]; Bouillon et al., 2019[8]). Clinical studies have reported that low serum DBP levels correlate with reduced BMD (Martínez-Aguilar et al., 2019[36]), further implicating impaired hepatic vitamin D metabolism in skeletal fragility.
Dietary vitamin D (contributing 10-20 % of total in humans and nearly 100 % in UVB-deprived laboratory mice) requires bile acid-mediated micellar solubilization for intestinal absorption (Deepika et al., 2025[13]; Ghallab et al., 2025[23]; Sahithi et al., 2025[47]). Notably, Cyp27a1, is not only an ancillary vitamin D 25-hydroxylase but also a key enzyme initiating the alternative (acidic) pathway of bile acid synthesis (Chiang and Ferrell, 2020[10]). High-fat diets and MAFLD are frequently associated with activation of the intestinal FXR-FGF15/19 axis, which markedly suppresses hepatic Cyp7a1 and can also attenuate Cyp27a1 expression (Wei et al., 2024[58]). These changes disrupt bile acid synthesis and shift the bile acid pool toward a more hydrophobic profile, a compositional pattern known to reduce micellar solubilization capacity and thereby impair the intestinal absorption of vitamin D (Ferslew et al., 2015[20]; Clifford et al., 2021[11]). The marked downregulation of hepatic Cyp27a1 and Cyp7a1 observed in WD-fed mice therefore raises the possibility that impaired bile acid homeostasis may further exacerbate systemic vitamin D insufficiency by reducing the efficiency of dietary vitamin D uptake. Circulating 25(OH)D levels were significantly reduced in WD-fed mice, supporting the presence of vitamin D deficiency. Hepatic vitamin D deficiency lowers circulating 25(OH)D, thereby restricting substrate availability for renal 1α-hydroxylase and ultimately diminishing systemic 1,25(OH)₂D production.
The paradoxical reduction in VDR protein despite elevated Vdr mRNA indicates an important post-transcriptional regulatory mechanism. VDR protein stability is highly dependent on ligand binding, as the receptor undergoes conformational changes upon binding 1,25(OH)₂D, thereby protecting it from proteasomal degradation (Zenata and Vrzal, 2017[62]). This deficiency promotes VDR degradation and may explain the reduced bone VDR protein observed.
VDR serves as the sole nuclear receptor for 1,25 (OH)₂D (Pike et al., 2017[40]). Upon ligand binding, VDR forms a heterodimer with retinoid X receptor alpha and regulates target gene transcription by binding to vitamin D response elements (van Driel and van Leeuwen, 2017[56]). This pathway influences both osteoblast differentiation and osteoclast activity. Reduced VDR levels may therefore disrupt this balance and contribute to increased bone resorption. Increased expression of osteoclastogenic genes in WD mice is consistent with this and supports the finding that reduced VDR causes exacerbated bone loss with enhanced osteoclastic activity (Starczak et al., 2018[51], 2021[50]; Gasperini et al., 2023[22]).
Genetic and clinical evidence further support a causal role for impaired VDR signaling in the skeletal abnormalities observed in our WD-fed mice. Global VDR-knockout mice develop hypocalcemia, secondary hyperparathyroidism, disorganized growth plates, and severe osteomalacia, defects that are largely corrected by calcium/phosphate rescue diets and therefore demonstrate the essential role of VDR signaling in bone mineralization (Yoshizawa et al., 1997[61]; Bikle, 2012[7]). Consistent with these findings, VDR loss-of-function mutations in humans, termed Vitamin D-Dependent Rickets type-II (VDDR-II), present with severe rickets/osteomalacia and elevated 1,25(OH)₂D despite normal ligand supply, underscoring the indispensability of VDR protein for skeletal responsiveness to vitamin D (Malloy et al., 2014[35]). Together, these genetic data support our conclusion that reduced VDR abundance can uncouple bone remodeling and contribute to skeletal fragility. In contrast, clinical trials using vitamin D analogs, such as calcitriol, alfacalcidol, and eldecalcitol, have demonstrated beneficial effects on BMD and fracture prevention (Tilyard et al., 1992[54]; Matsumoto et al., 2011[37]; Harada et al., 2012[27]).
Several limitations should be acknowledged. First, our model does not allow for the separation of obesity and hepatic injury as independent variables. Future studies using targeted models (e.g., diet-induced obesity without liver injury, or liver-specific knockouts) are needed to dissect their individual contributions. Second, our study did not directly measure renal 1α-hydroxylase activity, despite the well-established role of the kidney in converting 25(OH)D to its active form (Deluca, 2014[14]). It is therefore possible that compensatory upregulation of renal 1α-hydroxylation may occur in response to hepatic insufficiency. However, emerging evidence suggests that obesity and MAFLD are frequently associated with elevated PTH levels, which may reflect secondary hyperparathyroidism. Elevated PTH can stimulate renal 1α-hydroxylase, increasing 1,25(OH)2D production, but it can also promote bone resorption and impair cortical bone integrity. Thus, the observed skeletal deterioration may result not only from reduced hepatic 25-hydroxylation but also from compensatory endocrine changes that disrupt bone remodeling.
In summary, our study demonstrates that long-term WD feeding induces obesity and hepatic dysfunction, which together impair vitamin D metabolism and bone homeostasis. The resulting downregulation of hepatic vitamin D activation and bone VDR protein levels promotes osteoclastogenesis and skeletal fragility. These findings highlight the importance of the liver-bone axis and suggest that restoring vitamin D signaling- either through dietary modification or pharmacological intervention -may represent a promising strategy to prevent bone loss in obesity and MAFLD.
DeclarationData availabilityThe data that support the findings of this study are available from the corresponding author upon reasonable request.
AcknowledgmentsWe thank Dr. Caren Linnemann for her essential help in isolating bone RNA and protein. The presented data are in part from Pengcheng Zhou's medical dissertation and Mohammad Majd Hammour's PhD thesis. Pengcheng Zhou is a China Scholarship Council fellow. We acknowledge support from the Open Access Publishing Fund of the University of Tübingen. The work was partially funded by institutional funds of the Siegfried Weller Institute, the DFG (HE2509/24-1; 517010379& 457840828), and the BMBF (031L0314D).
Author contributionPengcheng Zhou (PZ) and Mohammad Majd Hammour (MMH) contributed equally to this work and share first authorship.
Jan G. Hengstler (JGH), Andreas K. Nüssler (AKN), and Tanja C. Maisenbacher (TCM) jointly supervised the study.
Conceptualization and Study Design: JGH, AKN
Experimental Work: PZ, MMH, Romina H. Aspera-Werz (RAW), Sandra Hans (SH)
Data Analysis and Interpretation: PZ, MMH, TCM, Sabrina Ehnert (SE), Matthias W. Laschke (MWL), Ahmed Ghallab (AG)
RNA-Seq and Bioinformatics: Karolina Edlund (KE), MMH
Methodology Development: Maiju Myllys (MM), Zaynab Hobloss (ZH), Reham Hassan (ReH), Daniela González (DG), Rama Hendawi (RaH)
Manuscript Writing - Original Draft: PZ, MMH, AKN, TCM
Manuscript Review and Editing: RAW, SE, MWL, SH, AG, MM, ZH, ReH, DG, RaH, KE, JGH, AKN
Supervision and Project Administration: AG, JGH, AKN
Funding Acquisition: AKN, AG, JGH
All authors contributed to drafting the article or revising it critically for important intellectual content. All authors gave final approval of the version to be submitted.
Conflict of interestThe authors declare no conflict of interest.