ALKBH5 regulates key lipid metabolites to promote osteoporosis

Abstract

Osteoporosis is a prevalent metabolic bone disease. Research has found a link between N6-methyladenosine (m6A) methylation and bone metabolism. The aim of this study is to investigate the relationship between m6A methylation modification-related genes and osteoporosis, and to explore the role of plasma metabolites in this relationship. Exposure was determined using cis-expression quantitative trait loci from the expression quantitative trait locus gen consortium, and a two-sample Mendelian randomization (MR) approach was employed to analyze the causal relationship between m6A methylation-related genes and osteoporosis. Summary-data-based Mendelian randomization (SMR) analysis was conducted to enhance the reliability of the Mendelian randomization results. Using the single-cell expression quantitative trait loci dataset, we explored the relationship between m6A methylation-related genes and osteoporosis in 14 types of immune cells. Mediation analysis was performed to investigate the effects of plasma metabolites and AlkB Homolog 5 (ALKBH5) on osteoporosis. Our study identified a causal association between the ALKBH5 gene and osteoporosis (odds ratio [OR] = 1.2742, 95% confidence interval: 1.1492–1.4128, P < .05). Moreover, this relationship persists in dendritic cells (OR = 1.5527, 95% confidence interval: 1.1838–2.0365, P < .05). The levels of 1-(1-enyl-palmitoyl)-GPE (P-16:0) and the ratio of Benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] play a mediating role in the relationship between ALKBH5 and osteoporosis, with mediation effects accounting for 15.1% and 8.7% of the total effect, respectively. These results reveal a potential link between m6A methylation-related genes and osteoporosis, providing new evidence for the investigation of pathogenic mechanisms and the exploration of therapeutic targets.

1. Introduction

Osteoporosis is a prevalent systemic metabolic bone disease primarily characterized by a decrease in bone mineral density, which leads to increased skeletal fragility in the elderly and a heightened susceptibility to fractures.^[1] The global prevalence of osteoporosis is estimated at 18.3%.^[2] This condition not only imposes substantial emotional, physical, and economic burdens on patients but also places considerable pressure on healthcare systems, presenting significant challenges to public health.^[3] The primary pathological mechanism underlying osteoporosis is an imbalance in bone metabolism, specifically between bone resorption and formation, resulting in reduced bone mass and degradation of trabecular structure.^[4] Given the complexity of osteoporosis pathogenesis, various clinical therapeutic agents have been developed, including calcitonin, selective estrogen receptor modulators, bisphosphonates, and molecular targeted therapies.^[5] However, many of these medications are associated with severe side effects or are unsuitable for long-term use.^[6] Therefore, it is crucial to explore more effective molecular drug targets for the treatment of osteoporosis.

N6-methyladenosine (m6A) methylation is a widespread RNA modification in eukaryotes that influences various biological functions and the pathogenesis of diseases by regulating processes such as RNA stability, localization, translation efficiency, and degradation.^[7] Research indicates that abnormal alterations in m6A modification can serve as diagnostic and prognostic biomarkers for lung cancer, highlighting its potential utility in clinical settings.^[8] Furthermore, m6A modification significantly impacts several metabolic-related diseases, affecting the onset and progression of type 2 diabetes, obesity, metabolic syndrome, and nonalcoholic fatty liver disease by modulating glucose metabolism and insulin sensitivity.^[9] During the differentiation of bone marrow mesenchymal stem cells into osteoblasts, m6A methylation plays a pivotal role. Specifically, the Methyltransferase-like 14 (METTL14) gene enhances the expression of T cell factor 1 through m6A modification, which subsequently elevates the protein levels of Runt-related transcription factor 2, thereby facilitating osteogenic differentiation and bone formation.^[10] Additionally, m6A methylation is implicated in the metabolic regulation of osteoclasts, where the WTAP gene inhibits osteoclast differentiation through m6A modification, thus preserving the balance of bone metabolism.^[11]

Plasma metabolites are small molecular compounds present in the plasma, primarily comprising amino acids, carbohydrates, nucleotides, and lipids. These metabolites function as products or intermediates of metabolic processes within the organism, reflecting the physiological condition, metabolic level, and disease status of the body.^[12] m6A methylation indirectly regulates the generation and function of metabolites by influencing the stability, translation efficiency, and degradation of RNA. Studies have demonstrated that m6A methylation modulates mitochondrial function by enhancing the translation of nuclear-encoded mitochondrial complex subunit RNAs, thereby affecting the levels of metabolites associated with energy metabolism.^[13] Furthermore, serum levels of 27-hydroxycholesterol and 24S-hydroxycholesterol are significantly elevated in patients with mild cognitive impairment, a condition associated with abnormalities in m6A modification.^[14] Additionally, m6A modification impacts cholesterol metabolism homeostasis by regulating the expression of genes involved in cholesterol synthesis, uptake, and efflux.^[15] Plasma metabolites play a crucial role in the onset, progression, and treatment of osteoporosis. Research indicates that metabolites such as triethanolamine, linoleic acid, and specific phospholipids are significantly reduced in patients with osteoporosis and positively correlate with bone mineral density/T-scores.^[16] Another study reported that levels of 4-methoxy cinnamic acid and 3,4,5-trimethoxy cinnamic acid were downregulated in the plasma of female osteoporosis patients.^[17] However, the role of plasma metabolites in the relationship between m6A methylation modification-related genes and osteoporosis has yet to be thoroughly explored.

Based on cis-expression quantitative trait loci (cis-eQTL) data, we conducted a correlation analysis between several m6A methylation modification-related genes and osteoporosis. Through summary-data-based Mendelian randomization (SMR), we identified m6A methylation-related genes associated with osteoporosis. Utilizing the Single-cell expression quantitative trait loci (sceQTL) dataset, we further explored the relationship between m6A methylation-related genes and osteoporosis across 14 types of immune cells. Additionally, we performed a mediation analysis to examine the potential role of plasma metabolites in the relationship between m6A methylation modification-related genes and osteoporosis. These findings provide a reference for subsequent mechanistic studies of osteoporosis and offer new potential therapeutic targets.

2. Methods

2.1. Study design

Figure 1 illustrates the analytical workflow of this study.

Figure 1.

The workflow of this study. MR = Mendelian randomization, m6A = N6-methyladenosine, SMR = summary-data-based Mendelian randomization, a = the influence of the exposure on the mediator, b = the impact of the mediator on the outcome, c = the total effect, c’ = the direct effect; c’ = c - a*b.

In this research, the m6A methylation modification-related genes, serving as exposure variables, were sourced from the Expression Quantitative Trait Locus Gen (eQTLGen) consortium. All data utilized pertained to cis-eQTL derived from blood samples. Osteoporosis was designated as the outcome variable, with data selected from the FinnGen database. A two-sample Mendelian randomization (MR) analysis was conducted to explore the causal relationship between the expression of m6A methylation modification-related genes and osteoporosis. Additionally, an SMR analysis was performed to further validate the impact of these m6A methylation-related genes on osteoporosis outcomes. Furthermore, we investigated the effects of these genes on osteoporosis across various types of immune cells. Finally, we conducted a mediation analysis to examine whether each plasma metabolite mediated the causal relationship between m6A methylation modification-related genes and osteoporosis.

2.2. Data source

We obtained the blood eQTL dataset through the eQTLGen consortium (https://eqtlgen.org/), which includes cis-eQTLs for all genes derived from blood samples of 31,684 healthy individuals of European ancestry. By reviewing previous studies and literature,^[18–21] we identified 19 genes related to m6A methylation modifications. We examined the eQTLGen consortium database, which comprises 15,695 genes, in search of 19 m6A-related genes. We identified data for only 12 of these genes, while the remaining 7 were not present in the database. (Fig. 2) The genome-wide association study (GWAS) data for plasma metabolites were sourced from the study conducted by Yiheng Chen, which involved a genome-wide association analysis of 1091 blood metabolites and 309 metabolite ratios across 8299 individuals.^[22] The genetic data pertaining to plasma metabolites were selected and retrieved from the GWAS Catalog online (http://ftp.ebi.ac.uk/pub/databases/gwas/summary_statistics/), with data IDs ranging from GCST 90199621 to GCST 90201020. Summary statistics for osteoporosis were obtained from the FinnGen biobank, which consists of 10,461 case subjects and 473,264 controls.^[23] Osteoporosis diagnoses were made according to the International Classification of Diseases criteria, utilizing the 10th revision code (International Classification of Diseases-10). Further details can be found at the following link: https://risteys.finngen.fi/endpoints/M13_OSTEOPOROSIS. All data utilized in this study are publicly accessible, and the GWAS participants were of European descent (Table S1, Supplemental Digital Content, https://links.lww.com/MD/R769).

Figure 2.

Identified 19 genes related to m6A methylation modification and extracted eQTL data for 12 of them. eQTL = expression quantitative trait loci, m6A = N6-methyladenosine.

2.3. Selection of instrumental variables

For the purpose of conducting MR analysis, the selection of instrumental variables (IVs) must adhere to the 3 key assumptions of MR: Assumption states that the single nucleotide polymorphisms (SNPs) used as IVs are strongly associated with the exposure; Assumption indicates that the SNPs are not confounded by factors related to both the exposure and the outcome; and Assumption posits that the SNPs influence the outcome solely through the exposure and have no direct effect on the outcome.^[24]

In this study, the IVs were screened according to specific parameters, applying strict filtering to the eQTL data of each m6A methylation-related gene. Firstly, in the cis-eQTL data, we selected SNPs with P-values below 5.0 × 10⁻8. Subsequently, utilizing data from the European population of the 1000 Genomes Project, we clustered SNPs for each gene, setting the linkage disequilibrium threshold at r² < 0.1 and the clustering window at 10,000 kb to ensure the independence of the IVs. We then harmonized the SNPs to ensure consistency in the effect alleles between the exposure and outcome datasets. Palindromic SNPs were excluded if the allele alignment could not be reliably determined. Based on assumptions, the identified SNPs were submitted to the linkage disequilibrium link database to investigate the associated phenotypes of each genetic variant (https://ldlink.nih.gov/?tab=home). We explored potential risk factors for osteoporosis and selected gender, age, and body mass index as confounding variables. Subsequently, we systematically screened and excluded SNPs linked to gender, age, and body mass index. By filtering out these confounding SNPs, we aimed to minimize the risk of residual confounding in the MR analysis and the mediation analysis. Finally, we calculated the F-statistic to mitigate bias introduced by weak instruments in the study. The F-statistic was computed using the formula F = (beta/se),^[2] and SNPs with an F-statistic <10 were removed.

2.4. Mendelian randomization analysis

This study utilized the two-SampleMR package (R package version 0.6.6) for conducting a two-sample MR analysis. When two or more SNPs were employed as IVs, the inverse-variance weighted (IVW) method was primarily utilized, supplemented by the Mendelian randomization-Egger (MR-Egger), weighted median, simple mode, and weighted mode methods, to ascertain the causal relationship between m6A methylation-related genes and osteoporosis. The IVW method aggregates the ratio estimates of each SNP through meta-analysis under the assumption of valid IVs, serving as the primary approach to determine the effect of exposure on the outcome and to yield reliable results.^[25] In instances where only one SNP was used as the IV in the MR analysis, the Wald ratio method was employed to estimate the causal relationship. To further enhance the reliability of the research findings, the MR-Egger regression test was conducted to assess pleiotropy; a P-value of the MR-Egger regression intercept <.05 indicates the presence of pleiotropy in the study results.^[26] Additionally, the MR pleiotropy residual sum and outlier (MR-PRESSO) method was applied to detect and correct any SNP outliers that may reflect potential pleiotropic bias.^[27] For the global test, the absence of horizontal pleiotropy outliers is indicated by a P-value >0.05. To evaluate the heterogeneity among the causal effects of the IVs, Cochran Q test was utilized, where a P-value <.05 suggests the presence of heterogeneity.^[28] Furthermore, sensitivity analysis was conducted using the leave-one-out method to further assess the robustness of the results.

2.5. Summary-data-based Mendelian randomization

To further validate the causal relationship between m6A methylation-related genes and osteoporosis, we conducted an SMR analysis. We downloaded version 1.3.1 of the SMR software from its official website (https://yanglab.westlake.edu.cn/software/smr/#Overview) and utilized the default parameter settings provided during the analysis.^[29] The auxiliary heterogeneity in dependent instruments (HEIDI) test was employed to confirm that the causal relationship between exposure and outcome was not confounded by linkage disequilibrium.^[29] The SMR and HEIDI methods demonstrate strong capabilities in distinguishing between pleiotropy models and linkage models.^[30] A significance level of P < .05 in the HEIDI test indicates that the observed association is attributable to shared genetic variants.

2.6. The impact of genes on osteoporosis in different types of immune cells

Through the OneK1K cohort (https://onek1k.org/), we acquired and organized a sceQTL dataset encompassing 14 types of immune cells. This dataset includes cis-eQTLs for all genes derived from blood samples of 982 healthy individuals of European ancestry.^[31] We employed a two-sample MR approach to investigate the impact of m6A methylation-related genes on osteoporosis across these 14 immune cell types.

2.7. Mediation analysis

A two-step MR approach was employed to investigate the role of plasma metabolites in mediating the effect of m6A methylation-related genes on osteoporosis. In the first step, we utilized the blood eQTL dataset and applied the IVW method to obtain β values, which were used to assess the association between each m6A methylation-related gene and osteoporosis. These β values represent the total effect (c) of the exposure factor on the outcome variable. In the second step, we again employed the IVW method. First, we obtained the β values between plasma metabolites and osteoporosis to assess the impact of the mediator on the outcome variable (b). Next, we obtained the β values between m6A methylation-related genes and plasma metabolites to evaluate the influence of the exposure factor on the mediator (a). The mediation effect was calculated using the formula a*b, while the direct effect of the exposure factor on the outcome variable was calculated using the formula c’ = c - a*b. The proportion of the mediation effect was calculated using the formula (a * b)/c. In mediation analysis, the proportion mediated quantifies the significance of the indirect pathway. A higher proportion indicates a more substantial mediating effect of the mediator. Although there is no universally accepted threshold, epidemiological and genetic studies suggest that a proportion exceeding 5% is generally regarded as indicative of a non-negligible mediating effect.^[32–34] In this study, if the P-values from the IVW method in the exposure-outcome, mediator-outcome, and exposure-mediator analyses are all <.05, and the calculated proportion of the mediating effect exceeds 5%, then the mediator can be determined to have a significant mediating effect.^[35]

3. Results

3.1. Effects of m6A methylation-related genes on osteoporosis

Using a two-sample MR analysis, we investigated the causal relationship between m6A methylation-related genes and osteoporosis. Among all m6A methylation-related genes, the AlkB Homolog 5 (ALKBH5) and METTL14 genes were significantly associated with osteoporosis in the blood eQTL (P < .05) (Fig. 3). Specifically, the ALKBH5 gene demonstrated a positive causal association with osteoporosis risk (IVW odds ratio [OR] = 1.2742, 95% confidence interval (CI): 1.1492–1.4128), while the METTL14 gene exhibited a negative causal association with osteoporosis risk (IVW OR = 0.8493, 95% CI: 0.7591–0.9502). Additionally, sensitivity analyses were conducted, and the results of Cochran Q test indicated no heterogeneity. The leave-one-out analysis revealed that no single SNP significantly impacted the final results (Figure S1, Supplemental Digital Content, https://links.lww.com/MD/R768). The MR-Egger regression analysis detected no evidence of pleiotropy, and the MR-PRESSO test identified no outliers, further supporting the reliability of our MR analysis results (Table 1). Detailed results of the MR analysis are provided in Tables S2–S3, Supplemental Digital Content, https://links.lww.com/MD/R769.

Figure 3.

Forest plots showed the causal associations between m6A methylation-related genes and osteoporosis. OR = odds ratio, CI = confidence interval, m6A = N6-methyladenosine.

Table 1.

Sensitivity analysis of the associations between identified m6A methylation-related genes on osteoporosis.

Exposure	Outcome	Pleiotropy				Heterogenenity
		MR_PRESSO_Global Test_P-value	MR_egger_intercept	MR_egger_SE	MR_egger_P-value	IVW_Q	IVW_Q_P-value	MR_egger_Q	MR_egger_Q_P-value
ALKBH5	Osteoporosis	.233	0.033	0.019	.138	11.87	.105	7.98	.239
FTO	Osteoporosis	.681	0.002	0.013	.909	8.48	.671	8.46	.583
METTL14	Osteoporosis	.611	0.003	0.013	.842	8.85	.546	8.81	.455
RBM15	Osteoporosis	.436	−0.022	0.032	.527	5.33	.377	4.76	.312
WTAP	Osteoporosis	.745	−0.001	0.008	.937	6.07	.733	6.06	.641
YTHDF3	Osteoporosis	.876	−0.008	0.016	.631	2.91	.821	2.65	.753

IVW = inverse variance weighted, m6A = N6-methyladenosine, MR_PRESSO = Mendelian randomization pleiotropy residual sum and outlier, MR-Egger = the Mendelian randomization-Egger, Q = The Q statistic of Cochran Q test, SE = standard error.

3.2. Summary-data-based Mendelian randomization analysis

SMR analysis and the HEIDI test were conducted on the eQTL of ALKBH5 and METTL14. The results indicated a causal association between the ALKBH5 gene and osteoporosis (SMR P < .05) (Table 2). Consistent with the direction of the aforementioned two-sample MR analysis, the ALKBH5 gene exhibited a positive causal association with the risk of osteoporosis (SMR OR = 1.2519, 95% CI: 1.0913–1.4361). In contrast, no causal association was found between the METTL14 gene and the risk of osteoporosis (SMR P > .05). The results of the HEIDI test suggest that the causal relationship between the ALKBH5 gene and osteoporosis is not influenced by linkage disequilibrium (HEIDI P > .05). Through SMR analysis and the HEIDI test, the impact of pleiotropy and related imbalances was further mitigated, demonstrating a significant causal relationship between the ALKBH5 gene and an increased risk of osteoporosis.

Table 2.

The results of SMR and HEIDI test between m6A methylation-related genes and osteoporosis.

probeID	Gene	topSNP	A1	A2	b_SMR	se_SMR	p_SMR	p_HEIDI	nSNP_HEIDI	OR	or_lci95	or_uci95
ENSG00000091542	ALKBH5	rs2925138	A	G	0.224	0.071	0.001	0.635	20	1.251	1.091	1.436
ENSG00000145388	METTL14	rs6828070	G	A	-0.121	0.104	0.245	0.991	20	0.886	0.722	1.086

A1 = effect allele, A2 = other allele, HEIDI = heterogeneity in dependent instruments, OR = odds ratio, SE = standard error, SMR = summary-data-based Mendelian randomization.

3.3. The impact of m6A methylation-related genes on osteoporosis in immune cells

Using the OneK1K cohort, we obtained sceQTL data for 14 immune cell types and extracted sceQTL for the ALKBH5 and METTL14 genes. Following a two-sample MR analysis, our results revealed a positive causal association between the ALKBH5 gene and the risk of osteoporosis in dendritic cells, with a Wald ratio P-value of <.05 (OR = 1.5527, 95% CI: 1.1838–2.0365) (Fig. 4). The detailed results of the above analysis are presented in Tables S8–S10, Supplemental Digital Content, https://links.lww.com/MD/R769.

Figure 4.

Forest plots showed the causal associations between ALKBH5, METTL14 and osteoporosis in immune cells. OR = odds ratio, CI = confidence interval, bmem = TCL1A- FCER2- B cell, cd4et = CD4 + KLRB1 + T cell, cd8nc = CD8 + LTB + T cell, cd8s100b = CD8 + S100B + T cell, dc = Dendritic cell, bin = TCL1A + FCER2 + B cell, cd4nc = CD4 + KLRB1- T cell, cd4sox4 = CD4 + SOX4 + T cell, monoc = Monocyte CD14 + cell, nk = XCL1- NK cell.

3.4. Mediation analysis

Through the aforementioned two-sample MR analysis steps, we concluded that the ALKBH5 gene exhibits a significant positive causal relationship with osteoporosis. In the IVW method, the β value of 0.2423 represents the total effect of the exposure factor on the outcome variable. Subsequently, we conducted a two-sample MR analysis using the IVW method to investigate the relationship between plasma metabolites and osteoporosis, revealing that 61 plasma metabolites have a causal association with osteoporosis (P < .05). Among these, serine levels (IVW OR = 1.1023, 95% CI: 1.0229–1.1879) and dimethylglycine levels (IVW OR = 1.0624, 95% CI: 1.0048–1.1234), along with 28 other metabolites, showed a positive correlation with the outcome. In contrast, androsterone sulfate levels (IVW OR = 0.9541, 95% CI: 0.9307–0.9782) and the glutamine to asparagine ratio (IVW OR = 0.9374, 95% CI: 0.8968–0.9799), along with 33 other metabolites, exhibited a negative correlation with the outcome (Table S4, Supplemental Digital Content, https://links.lww.com/MD/R769). We conducted a formal multiple testing correction for the metabolome-wide association analysis involving 61 metabolites. After applying the false discovery rate (FDR) method to adjust the IVW P-values, 59 metabolites remained statistically significant. This included the metabolic mediator 1-(1-enyl-palmitoyl)-GPE (P-16:0) levels (P-value = .0393; FDR P-value = .0421) and the ratio of benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] (P-value = .0227; FDR P-value = .0316) (Table S4, Supplemental Digital Content, https://links.lww.com/MD/R769). Following this, we conducted a two-sample MR analysis using the IVW method to examine the causal relationship between the ALKBH5 gene and the aforementioned 61 plasma metabolites. The results indicated a causal association between the ALKBH5 gene and 10 plasma metabolites. The ALKBH5 gene demonstrated positive correlations with 5 metabolites and negative correlations with the remaining 5 metabolites (Table S5, Supplemental Digital Content, https://links.lww.com/MD/R769). In the two-step MR mediation analysis, sensitivity analyses were performed for each step of the analysis. The results of Cochran Q test indicated the absence of heterogeneity. Our MR-Egger intercept test revealed no evidence of directional pleiotropy (P > .05), and the MR-PRESSO global test similarly indicated the absence of horizontal pleiotropy (P > .05). These findings suggest a low likelihood of potential confounding factors affecting the results at each stage of the mediation analysis, thereby further supporting the reliability of the MR analysis results (Table S6–S7, Supplemental Digital Content, https://links.lww.com/MD/R769).

Integrating the results from the aforementioned analyses allows for the calculation of the mediating effect, direct effect, and the proportion of the mediating effect. In this study, the proportion of mediating effects for candidate metabolites ranged from 1% to 10%. A clear distinction was made between influential mediators (proportion of mediating effect >5%) and less influential mediators (proportion of mediating effect <3%) as shown in Table S11, Supplemental Digital Content, https://links.lww.com/MD/R769. Consequently, we selected mediators with a proportion of mediating effect >5% for further interpretation and mechanistic exploration. The findings indicate that 1-(1-enyl-palmitoyl)-GPE (P-16:0) levels and the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio play a significant mediating role in the relationship between the ALKBH5 gene and osteoporosis. Specifically, the direct effect values are 0.2057 and 0.2211, respectively, while the proportions of the mediating effect to the total effect are 15.1% and 8.7% (Table 3).

Table 3.

Mediation analysis results.

Exposure	Mediator	Outcome	Total effect	Indirect effect	Direct effect	Mediation proportion
ALKBH5	1-(1-enyl-palmitoyl)-GPE (P-16:0) levels	Osteoporosis	0.242	0.036	0.205	15.1%
ALKBH5	Benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio	Osteoporosis	0.242	0.021	0.221	8.7%

4. Discussion

In this study, we compiled eQTL data from 12 m6A methylation-related genes and performed a correlation analysis with osteoporosis. We further validated the association of m6A methylation-related genes with osteoporosis through SMR analysis. Additionally, we investigated the relationship between m6A methylation-related genes and osteoporosis in 14 types of immune cells. Furthermore, we conducted a mediation analysis to explore the role of plasma metabolites in the relationship between m6A methylation-related genes and osteoporosis. Our findings indicate a significant association between the ALKBH5 gene and an increased risk of osteoporosis. This research suggests that the ALKBH5 gene may facilitate the mechanisms underlying osteoporosis by modulating the levels of 1-(1-enyl-palmitoyl)-GPE (P-16:0) and the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio.

The ALKBH5 gene encodes an m6A demethylase that catalyzes the demethylation of m6A, thereby promoting mRNA nuclear export and splicing.^[36] ALKBH5 dynamically regulates RNA stability, translation efficiency, and degradation, influencing various biological processes.^[37] Studies have demonstrated that in colorectal cancer, ALKBH5 activates the Wnt/β-catenin signaling pathway by demethylating AXIN2 mRNA, which induces DKK1 expression and subsequently recruits myeloid-derived suppressor cells.^[38] These myeloid-derived suppressor cells inhibit the functions of natural killer cells and cytotoxic T cells, leading to immunosuppression.^[38] In the context of heart failure, the upregulation of ALKBH5 expression may impair cardiomyocyte function by modulating m6A levels.^[39] In psoriasis, ALKBH5 exacerbates skin inflammation by promoting angiogenesis.^[40] In patients with glycogen storage disease type Ib, ALKBH5 influences the inflammatory response by inhibiting inflammasome activation through the regulation of m6A modification of NLRP3 mRNA.^[41] Additionally, ALKBH5 affects the senescence and osteogenic differentiation capacity of bone marrow mesenchymal stem cells by modulating the m6A modification of Voltage-dependent anion-selective channel 3.^[42] In diabetic osteoporosis, ALKBH5 expression is significantly elevated under high glucose conditions, suppressing the stability of fibroblast growth factor 21 mRNA through demethylation, which in turn affects the expression of bone formation-related genes.^[43] Our findings indicate a causal relationship between the ALKBH5 gene and the increased risk of osteoporosis, providing valuable references and a basis for potential therapeutic targets in the treatment of osteoporosis.

Osteoporosis is a systemic bone disease characterized by reduced bone mass and deterioration of bone microarchitecture, which leads to an increased risk of fractures.^[44] Its etiology and pathological mechanisms are complex, involving multiple biological processes and interactions within physiological metabolism.^[44] Serine, a nonessential amino acid, exerts a dual influence on bone metabolism by modulating the balance between bone formation and resorption.^[45] Studies indicate that D-serine reduces bone resorption by inhibiting the expression of osteoclast markers, such as cathepsin K.^[45] As a naturally occurring N-methylated glycine, dimethylglycine plays a significant role in the development and progression of osteoporosis. A large-scale community-based health study found that participants with the lowest plasma dimethylglycine levels had a 68% increased risk of low bone density compared to those with the highest levels.^[46] Our study demonstrates that both serine and dimethylglycine levels have a positive causal relationship with osteoporosis, contributing to an increased risk of developing the condition. Additionally, studies show that androsterone sulfate, an endogenous androgen metabolite, has a causal effect on hip bone density, with an odds ratio of 1.114, indicating that for every unit increase in its level, the probability of increased hip bone density rises by 11.4%.^[47] Glutamine serves as a crucial substrate for cellular energy metabolism and protein synthesis, while asparagine plays a regulatory role in osteoblast differentiation through glutamine-dependent asparagine synthetase.^[48] Our research indicates that androsterone sulfate levels and the glutamine to asparagine ratio can reduce the risk of developing osteoporosis. Further clinical and experimental studies are needed to explore the role of plasma metabolites in the diagnosis and prognosis of osteoporosis.

1-(1-enyl-palmitoyl)-GPE (P-16:0) is a specific type of plasmalogen.^[22] The designation “1-(1-enyl-palmitoyl)” indicates the fatty acid chain at the sn-1 position of the glycerol backbone. Here, “1-enyl” signifies the presence of a vinyl ether linkage (-O-CH=CH-) at the sn-1 position, while “palmitoyl” denotes that the fatty acyl group connected by the ether bond is derived from palmitic acid (16 carbons, 0 double bonds). The abbreviation “P-16:0” represents plasmalogen (p) and palmitic acid (16:0). The term “GPE” describes the head group of the glycerol backbone, where “G” denotes glycerol, “P” stands for phosphate, and “E” indicates ethanolamine, thereby designating GPE as glycerophosphoethanolamine.^[49] The compound 1-(1-enyl-palmitoyl)-GPE (P-16:0) plays a crucial role in the metabolic processes of organisms, particularly in the synthesis and degradation of phospholipids.^[50] Direct research on 1-(1-enyl-palmitoyl)-GPE (P-16:0) in bone metabolism is limited; however, the lipid class to which it belongs, plasmalogens, and its related metabolism are considered critical for skeletal homeostasis. During the differentiation of bone marrow mesenchymal stem cells into osteoblasts, these cells undergo extensive metabolic and membrane structural remodeling.^[51] A systematic investigation into the osteogenic differentiation of human adipose-derived mesenchymal stem cells (hAMSCs) has revealed that plasmalogens are among the lipid molecules significantly altered during this process, indicating their involvement in osteoblast differentiation and function.^[52] The relationship between plasmalogens and bone metabolism has also been substantiated in human genetic disorders. The biosynthesis of plasmalogens primarily occurs within the peroxisomes of cellular organelles. Impairment of peroxisomal function or biosynthesis leads directly to plasmalogen deficiency and triggers a series of severe skeletal pathologies.^[53] As a crucial molecular component of the head group in 1-(1-enyl-palmitoyl)-GPE (P-16:0), glycerophosphoethanolamine also plays a significant role in bone metabolism. Studies have demonstrated that, compared to non-insulin-resistant individuals, the levels of GPE are significantly elevated in patients with insulin resistance and type 2 diabetes, indicating its potential as a biomarker for metabolic disorders.^[54] Abnormal metabolism of glycerophosphoethanolamine may also disrupt the balance of bone metabolism. Research indicates that abnormal bone metabolism in patients with osteoporosis is often accompanied by changes in metabolites, with GPE levels potentially reflecting the state of bone cell function.^[55] Furthermore, certain traditional Chinese medicine preparations have been shown to improve osteoporosis and alleviate associated pain by regulating the glycerophosphoethanolamine metabolic pathway.^[56] However, whether the glycerophosphoethanolamine molecule functions similarly in bone metabolism when serving as a specific phospholipid headgroup remains to be verified through rigorous experimental investigations in the future. Our findings suggest that 1-(1-enyl-palmitoyl)-GPE (P-16:0) levels have a causal relationship with osteoporosis, where an increase in these levels can reduce the risk of developing the condition. Additionally, the levels of 1-(1-enyl-palmitoyl)-GPE (P-16:0) mediate the positive causal relationship between the ALKBH5 gene and osteoporosis. These analytical findings provide a foundation for further research into the pathological mechanisms underlying osteoporosis.

The ratio of benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] represents the ratio of benzoate to a specific diacylglycerol concentration.^[22] “Benzoate” refers to the salt form of benzoic acid, a small-molecule metabolite. “Oleoyl-linoleoyl-glycerol (18:1 to 18:2)” denotes a type of diacylglycerol, where “Oleoyl-” signifies the oleoyl group, a monounsaturated fatty acid characterized by an 18-carbon chain and one double bond (18:1). In contrast, “Linoleoyl-” indicates the linoleoyl group, a polyunsaturated fatty acid with an 18-carbon chain and two double bonds (18:2). The term “Glycerol” refers to the glycerol backbone. The notation “[2]” indicates the specific isomeric position. This ratio serves as a highly specific biochemical marker in lipid metabolism research.^[57] There is insufficient direct evidence for the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio in the context of bone metabolism research. However, both oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2], classified as diacylglycerols, and benzoate play crucial roles in bone metabolism. As an essential messenger molecule within the regulatory network of bone metabolism, diglycerides modulate osteoblast anabolism and osteoclast-mediated bone resorption through the PERK pathway, thus serving as a vital link in the maintenance of bone homeostasis.^[58] Under pathological conditions, imbalances in diglyceride metabolism or abnormalities in associated signaling pathways contribute to inflammatory bone destruction and metabolic bone diseases.^[59] Compared to high-fat diets rich in triglycerides, mice fed a high-fat diet abundant in diglycerides displayed lower body weight, increased bone mineral density, and enhanced bone microstructural parameters.^[60] This mechanism may involve the diglyceride diet improving systemic metabolism by reducing blood glucose, insulin, and cholesterol levels while upregulating the key osteogenic transcription factor Runx2 and downregulating the adipogenic factor PPARγ in bone marrow cells.^[60] Additionally, a certain association exists between benzoic acid and its derivatives (benzoates) and bone metabolism. Studies have indicated that specific benzoic acid derivatives can effectively stimulate osteoblasts to produce bone morphogenetic protein-2 and promote osteoblast differentiation in vitro.^[61] Furthermore, a phenolic extract from the rhizome of a medicinal herb contains benzoic acid compounds, such as 2-amino-3,4-dimethyl-benzoic acid, which exhibit promising anti-osteoporotic properties in vitro.^[62] Our findings reveal a causal relationship between the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio and osteoporosis, suggesting that an increase in this ratio elevates the risk of osteoporosis. Furthermore, the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio mediates the positive causal relationship between the ALKBH5 gene and osteoporosis. Future studies should further validate the mediating role of the benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio in relation to the ALKBH5 gene and osteoporosis.

In recent years, significant progress has been made in the research of molecular targeted drugs for osteoporosis, primarily focusing on key signaling pathways and cellular processes that regulate bone metabolism.^[63] Dendritic cells, which are antigen-presenting cells originating from hematopoietic progenitor cells in the bone marrow, are widely distributed throughout the body.^[64] In the pathological process of inflammatory bone loss, dendritic cells indirectly promote the formation and activity of osteoclasts by secreting pro-inflammatory cytokines, such as IL-17 and TNF-α, and modulating T-cell responses.^[65] Targeting the Toll-like receptor 4 signaling pathway in dendritic cells can reduce bone destruction in inflammatory bone diseases.^[66] Dendritic cell-derived exosomes regulate local immune responses by carrying specific molecules, including TGF-β and IL-10, which inhibit osteoclast activity and reduce bone loss.^[67] Furthermore, studies have demonstrated that inhibiting a specific molecule in dendritic cells, known as dendritic cell-specific transmembrane protein, effectively reduces osteoclast formation and bone resorption activity.^[68] In this study, we found that the expression of the ALKBH5 gene in dendritic cells increases the risk of osteoporosis, suggesting that the regulation of ALKBH5 in dendritic cells may provide a reference direction for the development of targeted drugs for osteoporosis.

Our research employs cis-eQTL data to perform MR analysis, effectively mitigating the influence of confounding factors and reducing the common reverse causation bias prevalent in observational studies. The application of SMR analysis further enhances the reliability of the MR results. Through mediation analysis, we investigated the role of lipid metabolites in the relationship between the ALKBH5 gene and osteoporosis, thereby laying a foundation for further exploration of the pathological mechanisms underlying osteoporosis. Additionally, the incorporation of sceQTL data offers a basis for the development of targeted pharmacological interventions.

This study, however, has certain limitations. First, the genome-wide association study data for the exposure factors, specifically m6A-related genes, mediating variables, including plasma metabolite levels, and outcome measures, namely osteoporosis, were all derived from participants of European ancestry. While utilizing samples with consistent genetic backgrounds effectively mitigates biases caused by population stratification, the applicability of the causal association conclusions obtained in this study to non-European populations, such as those in Latin America or Asia, necessitates rigorous validation.^[69] The IVs for exposure, mediating factors, and outcomes may be influenced by changes in lifestyle habits, gene-environment interactions, and variations in the genetic architecture of diseases.^[70] These factors restrict the generalizability of our findings beyond European populations. Our future research aims to develop cross-ancestry genetic instruments through trans-ancestry GWAS meta-analyses involving broader populations, such as Latin American cohorts and East Asian biobanks, including those from Japan.^[71] This approach is intended to validate and enhance the relevance of our observed results. Second, although our study provides valuable insights for the development of drug targets for osteoporosis treatment, it is imperative to conduct clinical trials in the future to validate these findings. Third, the 5% threshold used to define a meaningful mediation proportion, while based on the observed data distribution and common practice, remains exploratory. Future studies with larger sample sizes may help establish more precise, clinically relevant thresholds for mediation effects in the context of osteoporosis. Finally, further experimental research is essential, particularly studies utilizing cellular and animal models, to elucidate the specific mechanisms by which plasma metabolites mediate the effect of the ALKBH5 gene on osteoporosis.

5. Conclusion

In summary, our research reveals that ALKBH5 may promote osteoporosis by regulating 1-(1-enyl-palmitoyl)-GPE (P-16:0) levels and benzoate to oleoyl-linoleoyl-glycerol (18:1 to 18:2) [2] ratio. This finding provides new insights into the correlation between m6A methylation-related genes and osteoporosis, potentially offering a new foundation for subsequent studies on the mechanisms underlying osteoporosis and the exploration of therapeutic targets.

Acknowledgments

We sincerely thank every author for their hard work.

Author contributions

Conceptualization: Wenliang Wei, Jianzhong Xu.

Data curation: Wenliang Wei.

Formal analysis: Wenliang Wei.

Investigation: Wenliang Wei.

Methodology: Wenliang Wei.

Project administration: Wenliang Wei.

Resources: Wenliang Wei.

Software: Wenliang Wei.

Validation: Wenliang Wei, Jianzhong Xu.

Visualization: Wenliang Wei.

Funding acquisition: Jianzhong Xu.

Supervision: Jianzhong Xu.

Writing – original draft: Wenliang Wei.

Writing – review & editing: Wenliang Wei.

Supplementary Material

medi-105-e48506-s001.xlsx ^{(114KB, xlsx)}

medi-105-e48506-s002.pdf ^{(428.8KB, pdf)}