Drug Repositioning Based on Single-Cell Transcriptomics Data Identifies Quinidine as a Potential Drug for Treating Osteoporosis

1. Introduction

Osteoporosis (OP) is a systemic skeletal disease characterized by reduced bone mass and deterioration of bone tissue microstructure, leading to decreased bone strength and stability. Its most serious consequence, fracture can result in patient disability or even death [1]. With the acceleration of global population aging, the incidence and prevalence of OP are increasing year by year. It is estimated that by 2050, osteoporotic fractures in China will rise to 5.91 million cases, and the treatment cost will reach $27.48 billion [2], posing a huge economic and medical burden.

The pathogenesis of osteoporosis is complex, primarily involving an imbalance in bone remodeling, changes in endocrine hormones, and dysregulation of the immune microenvironment [3,4,5]. Bone remodeling is a dynamic balance between bone formation and bone resorption. In osteoporosis, a net decrease in bone mass occurs when osteoclast-mediated bone resorption exceeds osteoblast-mediated bone formation [6]. Multiple signaling pathways are involved in the regulation of bone metabolism, including the Wnt/β-catenin signaling pathway, the RANKL/RANK/OPG signaling pathway, and the NF-κB signaling pathway, among others [7,8]. Changes in endocrine hormone levels (such as the sharp decline in estrogen secretion after menopause and the long-term use of glucocorticoids) and disturbances in the immune microenvironment can both lead to osteoporosis by affecting bone remodeling [9,10].

Therapeutic agents of osteoporosis are broadly classified into anti-resorptive and bone-forming drugs. Anti-resorptive drugs mainly incl6de bisphosphonates, estrogens, calcitonin, and RANKL inhibitors. Bone-forming drugs, such as the parathyroid hormone analogue Teriparatide, promote the proliferation, differentiation, and activity of osteoblasts [11]. Although these drugs can, to a certain extent, slow down bone resorption or promote bone formation, long-term use presents certain limitations and side effects. For example, prolonged use of bisphosphonates can lead to osteonecrosis of the jaw [12] and atypical femoral fractures [13], and long-term use of bone-forming drugs may also cause severe adverse events such as nausea, dizziness, and leg cramps [14]. Therefore, the discovery of new drug targets and safe, effective therapeutic strategies is an urgent necessity.

Drug repositioning (or drug repurposing) is the process of redeveloping drugs already approved for treating one disease to treat other diseases. Compared to traditional new drug development, which requires 9–12 years and an approximate cost of $1.24 billion [15,16], drug repositioning can significantly reduce both time and cost. The computational pipeline for drug repositioning based on single-cell data, ASGARD [17], offers a unique advantage. This method considers the heterogeneity of all cell clusters and inputs differentially expressed genes into the L1000 drug response dataset to identify drugs that can significantly reverse the gene expression profile. Compared to methods developed by Alakwaa et al. [18]and Wang Z et al.[19], ASGARD has demonstrated higher AUC values across multiple datasets, indicating superior reliability and robustness.

The zebrafish is an ideal model for osteoporosis research due to its strong developmental capacity, low rearing cost, and short experimental cycle. Zebrafish share approximately 70% gene homology with humans [20], and their skeletal development and molecular regulation mechanisms are similar to those in humans. Dexamethasone is a commonly used drug for inducing an osteoporosis model in zebrafish [21].

This study aims to screen candidate drugs for osteoporosis treatment using a computational pipeline for drug repositioning based on single-cell transcriptomics data. Subsequently, Mendelian Randomization analysis will be used to infer the causal relationship between the expression of drug target gene and osteoporosis. Finally, experimental validation will be performed using a dexamethasone-induced osteoporosis zebrafish model. This comprehensive approach, spanning from computational prediction to in vivo validation, will evaluate the potential of candidate drug in the prevention and treatment of osteoporosis.

2. Materials and Methods

2.1. Acquisition and Analysis of Single-Cell Transcriptomics Data for Osteoporosis

Single-cell transcriptomics dataset for non-union fractures (GSE217792) [21]was retrieved from the Gene Expression Omnibus (GEO) database. Two osteoporotic fracture samples (GSM6726765 and GSM6726769) were selected as the experimental group, and three healthy samples (GSM6726762, GSM6726763, and GSM6726764) were selected as the control group.

The single-cell dataset was subjected to quality control and normalization using the Seurat R package. Cells with fewer than 200 or more than 6000 detected genes, and those with a mitochondrial gene proportion exceeding 10%, were filtered out to remove low-quality cells and technical noise.

The ElbowPlot function was used to observe the variance contribution rate of the principal components. The top 42 principal components were selected for dimensionality reduction. Cell clustering was performed with a resolution of 0.6, and visualization was achieved using UMAP. Differentially Expressed Genes (DEGs) were identified using the FindAllMarkers function (Logfc.threshold=0.25). Cell clusters were annotated by integrating results with existing literature and the CellMarker database(http://bio-bigdata.hrbmu.edu.cn/CellMarker/).

Gene Ontology (GO) functional enrichment analysis was performed on the DEGs of the osteoblast cluster (|LogFC|>1, adjusted P-value < 0.05) to investigate their biological processes, molecular functions.

2.2. Drug Repositioning Analysis Based on Single-Cell Transcriptomics Data

Candidate drugs were screened using the computational pipeline for drug repositioning based on single-cell transcriptomics data (ASGARD) [18]. The disease group and the healthy group were paired for analysis to identify DEGs (P < 0.05, |LogFC| > 1). The set of osteoblast DEGs was then compared against the L1000 drug response dataset to identify drugs capable of significantly reversing the differential gene expression profile. The screened potential candidate drugs were ranked according to a defined drug score formula. The drug score formula is as follows:

$$\mathrm{Drug}\mathrm{score}={\displaystyle \sum _{k=1}^n\left(\frac{\mathrm{Num}{(Cell)}_k}{\mathrm{Num}(Total.Cell)}*(-{\log }_{10}^{FD{R}_k})*\frac{\mathrm{Num}{(ReversedGene)}_k}{\mathrm{Num}{(DiseasedGene)}_k}\right)}$$

where k represents a specific single-cell cluster, n represents all disease-associated single-cell clusters, and the normalized drug score ranges from 0 to 1, with the formula as follows:

$$\mathrm{Standardized}\mathrm{Drug}\mathrm{Score}=1-{\displaystyle \frac{\mathrm{Rank}(Drug)}{\mathrm{Total}\mathrm{Num}(Drug)}}$$

2.3. Mendelian Randomization Analysis

To infer the causal relationship between the expression of candidate drug target gene and osteoporosis, Mendelian Randomization analysis was performed. The target genes of the candidate drug were queried on the DrugBank website (https://go.drugbank.com/). The Ensembl sequence numbers for the target genes were retrieved from the NCBI website(https://www.ncbi.nlm.nih.gov/). Expression Quantitative Trait Loci (eQTLs) for the target genes were then obtained from the IEU OpenGWAS database(https://gwas.mrcieu.ac.uk/).The criteria for selecting instrumental variables were: a P-value threshold of P < 5e-06 for genome-wide significance, an r² threshold of 0.1, and a physical distance threshold greater than 10000kb. This ensured the selection of independent SNPs to serve as instrumental variables. The Genome-Wide Association Study (GWAS) data for osteoporosis was obtained from the IEU OpenGWAS database, comprising 455,386 controls and 7,547 osteoporosis cases.

The analysis was performed using the TwoSampleMR R package [23]. Three primary methods were employed: Inverse Variance Weighted (IVW), MR-Egger, and Weighted Median (WM). The IVW method was used as the main approach to assess the causal relationship between the drug target and osteoporosis, with the other two methods serving as robustness checks.

Cochran’s Q test [24]was used to assess heterogeneity, and the MR-Egger intercept test [25] was employed to evaluate pleiotropy. A P-value greater than 0.05 indicates the absence of significant heterogeneity or pleiotropy, while a P-value less than 0.05 indicates the presence of significant heterogeneity or pleiotropy.

2.4. Zebrafish Animal Model Experimental Validation

To validate the osteogenic effect of candidate drug, we used a dexamethasone-induced osteoporosis zebrafish model. Tu wild-type zebrafish and Tg(Ola.Sp7:nlsGFP) transgenic zebrafish were used. The zebrafish were maintained at 28℃, with the pH kept between 7 and 8, and a light-dark cycle of 14 hours light / 10 hours dark. The zebrafish were fed with brine shrimp three times a day. The transgenic zebrafish Tg(Ola.Sp7:nlsGFP) express green fluorescent protein (nlsGFP) in their osteoblasts, allowing for observation of osteoblast activity and bone mineralization.

Three-day-old wild-type Tu zebrafish and Tg(Ola.Sp7:nlsGFP) transgenic zebrafish were selected for the study. The experimental design included a control group treated with 0.2% DMSO and a model group treated with 10μmol/L dexamethasone. Additionally, three combination treatment groups were established, consisting of 10μmol/L dexamethasone plus varying concentrations of quinidine (100μmol/L, 10μmol/L, and 1μmol/L), as well as three separate quinidine treatment groups at concentrations of 100μmol/L, 10μmol/L, and 1μmol/L. Each group contained 20 larvae, and the treatments were administered continuously for 5 days, with the treatment solutions replaced every 24 hours.

Tu wild-type zebrafish were stained with Alizarin Red. Images were captured using a stereo microscope, and the ventral side of the zebrafish larvae was observed. Bone mineralization area and grayscale values were quantified using Image J (Version 1.52a) software. Tg(Ola.Sp7:nlsGFP) transgenic zebrafish were subjected to fluorescent stereo microscopy imaging. The fluorescent area and fluorescent intensity of the skull were quantitatively analyzed, which represents the degree of bone mineralization.

Cranial bone tissue from each group of zebrafish was collected, and total RNA was extracted using the Trizol method, followed by reverse transcription to synthesize cDNA. Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR) was used to measure the expression levels of osteoblast marker genes (runx2b, col2a1a, and sp7) and osteoclast marker genes (spp1, mmp13, ctsk, and opg). β-actin was used as the internal reference gene. To verify the effect of quinidine on mature bone tissue, an adult zebrafish bone mineral density (BMD) detection experiment was established. Adult zebrafish, aged 3-5 months, were selected. After one week of acclimatization, they were randomly divided into the control group, the dexamethasone group, and the quinidine drug treatment group. An osteoporosis model was established by continuous induction with 10μmol/L dexamethasone for 14 days, with simultaneous administration of the corresponding drug intervention. After the experiment, Micro-CT analysis was performed to evaluate the effect of quinidine on the BMD and structure of the adult zebrafish.

2.5. Statistical Analysis

Statistical analysis was performed using GraphPad Prism (Version 9.0.0) software. The unpaired t-test was used for comparisons between two groups, and one-way analysis of variance (ANOVA) was used for comparisons among multiple groups. All data are presented as mean ± standard deviation (SD). A P-value of < 0.05 was considered to indicate a statistically significant difference. Each treatment group was performed in triplicate, with a sample size of n=15 zebrafish.

3. Result

3.1. Single-Cell Transcriptomic Landscape of Osteoporotic Fractures

For single-cell transcriptomic analysis, three control subjects and two osteoporotic fracture subjects were selected in this study (Figure 1A). After quality control, a total of 29,742 cells were analyzed, comprising 14,410 cells from control group and 15,332 cells from osteoporotic fracture group (Figure 1B). Using the criteria of PC=42 and a resolution of 0.6, cell clustering was performed, resulting in the identification of 31 distinct cell clusters.

Based on the DEGs of each cell cluster, combined with evidence from published literature [22,26] and the CellMarker database, we systematically identified specific marker genes for individual cell clusters (Figure 1C). The 31 cell clusters were annotated into 16 distinct cell types: Osteoblasts, Chondrocytes, T cells, Monocytes, Neutrophils, Macrophages, Smooth muscle cells, B cells, Erythrocytes, Myeloid cells, Endothelial cells, Natural killer cells, pre-B cell CD34+, Plasma cells, Hematopoietic stem cells, and Plasmacytoid dendritic cells (Figure 1D).

A total of 1,647 DEGs were identified between the osteoblast cluster of the osteoporotic fracture group and the control group (|LogFC| > 1 and adjusted P-value < 0.05). Gene Ontology analysis showed the DEGs were primarily enriched in the following biological processes: extracellular matrix organization, chemotaxis, ossification, connective tissue development, myeloid leukocyte migration, and cartilage development(Figure 1E). The enriched molecular functions included collagen metabolic process and collagen fibril organization.We also constructed a gene-pathway interaction network (Figure 1F). This network revealed that upregulated genes (including COL1A1, SPP1, and MMP13)were closely associated with extracellular matrix organization and Wnt signaling, while downregulated genes such as BMP4 and LEF1 were simultaneously enriched in opposing functional cascades. These findings collectively delineate the core molecular dysregulation underlying osteoporosis pathogenesis.

3.2. Identification of Quinidine as a Potential Therapeutic Agent via Drug Repositioning and Mendelian Randomization

Using the ASGARD computational pipeline for drug repositioning on the differentially expressed genes of the osteoblast cluster, four potential candidate drugs for osteoporosis treatment were screened: perphenazine, imatinib, quinidine, and vinblastine (Table 1). Among these, perphenazine, imatinib, and vinblastine have been previously reported in relation to osteoporosis. Quinidine, a sodium channel inhibitor, has been less studied in the field of osteoporosis. Therefore, this study will focus on its in-depth investigation and validation.

Quinidine has seven drug targets: SCN5A, KCNK1, KCNK6, KCNH2, ADRA1A, ADRA1B, and ADRA1D. Among these, SCN5A, KCNK6, and KCNH2 have corresponding eQTLs available in the IEU OpenGWAS database. For the SCN5A drug target, four SNPs (rs6810361, rs61558743, rs11720524 and rs7355975) were selected as instrumental variables.

To further prioritize the candidate drugs, we first constructed a drug-target interaction network (Figure 2A), which identified SCN5A as a primary target of quinidine. MR analysis subsequently revealed a significant causal relationship between the SCN5A drug target with osteoporosis (Figure 2B), while no causal relationship was found between KCNK6/KCNH2 with osteoporosis (Figure 2C). For every one standard deviation increase in SCN5A expression, the risk of developing osteoporosis increases by 0.005 standard deviations. Given that quinidine inhibits the expression of SCN5A, it is thus inferred that quinidine has a potential therapeutic effect on osteoporosis.

3.3. Quinidine Ameliorates Dexamethasone-Induced Osteoporosis

To verify the therapeutic efficacy of quinidine, we treated dexamethasone (Dex)-induced osteoporotic zebrafish with gradient concentrations of quinidine. Alizarin Red staining revealed that quinidine concentrations effectively rescued Dex-impaired cranial bone mineralization. Specifically, compared with the Dex-induced model group, the 10 μmol/L quinidine group achieved the most significant improvements, with a 35.7% increase in cranial mineralization area and a 31.8% increase in intensity (P < 0.05). Consistently, transgenic Tg(Ola.Sp7:nlsGFP) zebrafish exhibited increased osteogenic fluorescent area and intensity after quinidine intervention, with the 10μmol/L quinidine group showing the optimal therapeutic effect. Overall, quinidine markedly ameliorated Dex-induced osteoporosis by restoring bone mineralization in key cranial regions, and exhibited a concentration-dependent therapeutic advantage at the medium concentration (Figure 3 A-N).

qRT-PCR analysis revealed the regulatory effects of quinidine on bone metabolism-related gene expression (Figure 3O). All three quinidine treatment groups increased the expression of osteoblast marker genes (runx2b, col2a1a, and sp7). Notably, the upregulation of osteogenic genes was not statistically significant in the high-concentration quinidine group, whereas low and medium quinidine concentrations significantly elevated col2a1a expression (P < 0.05). For osteoclast markers, spp1 and ctsk were markedly downregulated across all quinidine intervention groups (P < 0.05). These findings indicate that quinidine ameliorates dexamethasone-induced osteoporosis in larval zebrafish by simultaneously promoting osteoblast-related gene expression and suppressing osteoclast-associated gene transcription, thereby restoring bone homeostasis.

Micro-CT of adult zebrafish was further performed to evaluate bone structural changes (Figure 3P-S). Compared with the control group, the dexamethasone-treated group displayed obvious bone loss and structural deterioration in the skull and spine. Quinidine treatment markedly rescued these osteoporotic phenotypes and restored bone architecture. Quantitative analysis verified that quinidine significantly increased bone mineral density (BMD) and bone volume fraction (BV/TV), further confirming its potent osteogenic protective effect in adult zebrafish.

3.4. Quinidine Monotherapy Exerts Osteoprotective Effects

To further investigate the pharmacological activity of quinidine, we treated 3-day-old wild-type Tu and Tg(Ola.Sp7:nlsGFP) transgenic zebrafish with quinidine at gradient concentrations for five consecutive days. Alizarin Red staining and Image J quantitative analysis demonstrated that low and medium concentrations of quinidine exerted obvious osteogenic effects. In wild-type zebrafish, 1μmol/L and 10μmol/L quinidine significantly increased both cranial bone mineralization area and intensity (all P < 0.05), while 100μmol/L quinidine produced no significant changes. Consistent results were observed in transgenic zebrafish: 1μmol/L and 10μmol/L quinidine markedly enhanced cranial osteogenic fluorescent area and intensity (P < 0.05), whereas 100μmol/L quinidine showed no statistical difference (Figure 4A-H，I-L). Collectively, these results confirm that low to medium concentrations of quinidine effectively promote cranial bone mineralization in zebrafish.

qRT-PCR results showed that all three quinidine concentrations slightly increased the expression of osteoblast markers (col2a1a and sp7) without statistical significance (P > 0.05), but significantly downregulated osteoclast-related genes (spp1 and opg) (P < 0.05)(Figure 4M). These data indicate that quinidine monotherapy exerts osteoprotective effects mainly by suppressing osteoclast activity.

To further elucidate the underlying therapeutic mechanism, Molecular docking analysis verified the stable binding affinity of quinidine to the SCN5A protein, with a binding affinity of −6.8 kcal/mol (Figure 4N). By synthesizing the results from computational drug repositioning, MR-based genetic causal inference, and in vivo qRT-PCR validation, an integrative SCN5A-centered regulatory network was constructed (Figure 4O). Network profiling revealed elevated osteogenic genes (runx2b, col2a1a) and decreased bone resorption genes (mmp13, ctsk), which was highly consistent with our qRT-PCR results.

4. Discussion

This study successfully identified and validated quinidine as a potential therapeutic agent for osteoporosis by comprehensively employing computational biology methods, including single-cell transcriptomics analysis, drug repositioning, Mendelian Randomization, and an animal model experiment.

Firstly, through single-cell transcriptomics analysis, 16 cell types were successfully annotated, and 1,647 differentially expressed genes (DEGs) in the osteoblast cluster were screened. GO enrichment analysis revealed that these genes are primarily involved in bone metabolism-related processes, such as extracellular matrix organization, ossification, cartilage development, and collagen metabolic process. Osteoblasts play a critical role in maintaining skeletal homeostasis, and their dysfunction is a central element in the development of osteoporosis [27]. Secondly, by employing the computational pipeline for drug repositioning based on single-cell data (ASGARD) [17], we rapidly identified four potential drugs, including quinidine, from a massive drug library. Among these, perphenazine, imatinib, and vinblastine have been previously reported in relevant studies. Wang M et al.[28] found that the antipsychotic drug perphenazine can lead to reduced bone mineral density and increased incidence of osteoporosis, suggesting a need for caution regarding bone health risks if used clinically. Imatinib, as a tyrosine kinase inhibitor, can decrease osteoclast number and activity, leading to bone metabolism disorders [29]. Ohya K et al. [30] experimentally demonstrated that vinblastine can cause hypocalcemia, possibly related to its interference with the regulatory mechanisms of calcium homeostasis in bone cells and its disruption of microtubules. Ross R et al. [30]also proved through relevant experiments that vinblastine can affect osteoblast function and the synthesis of bone matrix. These cases demonstrate the reliability of the ASGARD method, which significantly shortens the drug development cycle and reduces costs and risks, making it particularly suitable for discovering new therapies for complex diseases. Quinidine, on the other hand, is a Class Ia antiarrhythmic drug primarily used to treat Brugada syndrome and idiopathic ventricular fibrillation, with no current reports linking it to osteoporosis research. This study is the first to associate quinidine with osteoporosis, offering a new direction for drug repurposing.

Furthermore, the Mendelian Randomization (MR) analysis method is an important approach for inferring the causal relationship between an exposure (such as a biomarker, environmental factor, or drug target) and a disease [31]. The results showed a significant positive causal relationship between the quinidine drug target SCN5A and osteoporosis. Since quinidine is an inhibitor of SCN5A, this genetic evidence strongly suggests the potential of quinidine for the clinical treatment of osteoporosis.

From a mechanistic perspective, Quinidine is a Class Ia antiarrhythmic agent that primarily functions by inhibiting the voltage-gated sodium channel NaV1.5, which is encoded by the SCN5A gene. Although traditionally associated with cardiac electrophysiology, recent research has underscored the critical role of bioelectric signaling in skeletal homeostasis. Voltage-gated sodium channels have been found to be functionally expressed in both osteoblasts and osteoclasts, where they regulate membrane potential [32]. They regulate membrane potential, which influences calcium signaling—a critical pathway for osteoblast differentiation and mineralization. By inhibiting sodium influx, quinidine may hyperpolarize the cell membrane or alter depolarization-dependent calcium entry, thereby promoting osteoblast activity (upregulation of runx2b and sp7) while suppressing osteoclastogenesis (downregulation of mmp13 and ctsk). This dual action restores bone remodeling balance [33]. Further mechanistic investigations, including SCN5A knockout models and electrophysiological studies in bone cells, are warranted to fully elucidate these pathways. These findings position quinidine as a multifaceted agent bridging cardiac ion channel pharmacology and skeletal health [34].

Finally, the experimental results from the zebrafish model further confirmed the pharmacological activity of quinidine. The zebrafish is an ideal model for osteoporosis research due to its convenient genetic manipulation, low rearing cost, short experimental cycle, and the many similarities between its skeletal development process and molecular regulation mechanisms and those in humans [35]. In this study, the osteoporosis model was established using 10μmol/L Dex, which induces osteoporosis by inhibiting osteoblast differentiation and promoting osteoclastogenesis. Alizarin Red staining showed that Dex treatment significantly reduced the cranial bone mineralization area and intensity of the zebrafish. qRT-PCR confirmed the downregulation of osteoblast marker genes (runx2b, col2a1a, sp7) and the upregulation of osteoclast marker genes (spp1, ctsk), thus successfully establishing the osteoporosis model. The quinidine treatment results showed that, when combined with Dexamethasone, the 10μmol/L Dex + 10μmol/L Quinidine treatment group demonstrated the most pronounced improvement in cranial bone mineralization area and intensity, effectively alleviating the symptoms of osteoporosis.

The innovation of this study primarily lies in the novel application of single-cell drug repositioning to screen for osteoporosis drugs, integrating Mendelian Randomization analysis for causal verification of the drug target, and finally establishing a closed-loop of computational prediction and experimental validation through dual experiments in both larvae and adult zebrafish. However, this study also has several limitations: First, the single-cell data source and sample size are limited, as the analysis was solely based on the GSE217792 dataset with a small sample size. Future research should integrate more single-cell osteoporosis datasets for meta-analysis to enhance the robustness of the results. Second, the integration of multi-omics data is insufficient; data such as proteomics and metabolomics were not included. Multi-omics integration could more comprehensively elucidate the mechanism of action of quinidine. Finally, quinidine carries a risk of cardiotoxicity. Further studies should validate its osteoprotective effect in animal models, while simultaneously exploring a safe dosage and the underlying mechanistic pathways. In summary, this study systematically revealed the potential role of quinidine in the prevention and treatment of osteoporosis through single-cell drug repositioning, Mendelian Randomization, and zebrafish experimental validation. Quinidine can reverse the pathological transcriptome features of osteoblasts, promote bone mineralization, and increase bone mineral density, providing a new perspective for drug repositioning research in osteoporosis.

5. Conclusions

Based on single-cell transcriptomics analysis, drug repositioning, Mendelian Randomization analysis, and zebrafish experimental validation, this study successfully identified and validated quinidine as a potential therapeutic drug for osteoporosis. This study establishes a robust methodological framework for drug repositioning and provides compelling evidence for quinidine as a novel therapeutic option for osteoporosis, offering new insights for the future development of anti-osteoporosis drugs.