Skip to main content
Download PDF

Multi-ancestry whole genome sequencing analysis of lean body mass

Genome Biology volume 26, Article number: 106 (2025) Cite this article

Abstract

Background

Lean body mass is a crucial physiological component of body composition. Although lean body mass has a high heritability, studies evaluating the genetic determinants of lean mass (LM) have to date been limited largely to genome-wide association studies (GWAS) and common variants. Using whole genome sequencing (WGS)-based studies, we aimed to discover novel genetic variants associated with LM in population-based cohorts with multiple ancestries.

Results

We describe the largest WGS-based meta-analysis of lean body mass to date, encompassing 10,729 WGS samples from six TOPMed cohorts and the Louisiana Osteoporosis Study (LOS) cohort, measured with dual-energy X-ray absorptiometry. We identify seven genome-wide loci significantly associated with LM not reported by previous GWAS. We partially replicate these associations in UK Biobank samples. In rare variant analysis, we discover one novel protein-coding gene, DMAC1, associated with both whole-body LM and appendicular LM in females, and a long non-coding RNA gene linked to appendicular LM in males. Both genes exhibit notably high expression levels in skeletal muscle tissue. We investigate the functional roles of two novel lean-mass-related genes, EMP2 and SSUH2, in animal models. EMP2 deficiency in Drosophila leads to significantly reduced mobility without altering muscle tissue or body fat morphology, whereas an SSUH2 gene mutation in zebrafish stimulates muscle fiber growth.

Conclusions

Our comprehensive analysis, encompassing a large-scale WGS meta-analysis and functional investigations, reveals novel genomic loci and genes associated with lean mass traits, shedding new insights into pathways influencing muscle metabolism and muscle mass regulation.

Background

Lean body mass (LM), which includes but is not limited to the body’s muscle mass, serves as an important physiological component of body composition. Low lean body mass reflects a lower amount of muscle tissue, which is associated with functional impairment and physical disability and is a major modifiable cause of frailty in the elderly population [1]. Low lean body mass is also associated with higher surgical and post-operative complications, longer length of hospital stay, lower physical function, poorer quality of life [2], malnutrition [3], and mortality [2, 4], which makes it an important measure in clinical practice. Lean mass can be measured by dual-energy X-ray absorptiometry (DXA) or bioelectrical impedance analysis (BIA). Two measurements of LM are usually reported: whole-body lean mass (WB-LM) and appendicular LM (A-LM), the latter being the lean mass in the arms and legs. While A-LM consists largely of skeletal muscle as well as some other connective tissues, WB-LM is determined by skeletal muscle, parenchymatous organs, cardiac muscle, and blood vessels. Studying both WB-LM and A-LM provides a comprehensive understanding of image-based markers of muscle health by capturing systemic and limb-specific changes. Specifically, A-LM may better reflect muscle mass because the limbs have fewer organs included in the lean mass derivation. Additionally, WB-LM provides a more complete picture of overall body composition. This dual approach enhances the assessment of body composition and offers robust insights into sarcopenia and related health outcomes.

Lean body mass has a significant genetic component, as evidenced by a high heritability of 50–80% observed in twins and in families studies [5]. To identify variants associated with this phenotype, we previously performed a large meta-analysis of GWASs that amassed 20 cohorts of European ancestry with a total sample size of > 38,000 for WB-LM and > 28,000 for A-LM [6]. Despite the large sample used, the percentage of phenotypic variance explained by the identified SNPs was very small, suggesting that common variants do not explain most of the heritability of lean body mass. A more recent GWAS of A-LM, conducted in 85,750 middle-aged (aged 38–49 years) individuals from the UK Biobank (UKBB) [7], identified a total of 182 loci, 78% of which were replicated in an independent set of 181,862 older (aged 60–74 years) individuals from the same UKBB cohort [1]. Pei et al. [1] performed a GWAS of A-LM in the full UKBB European-descent cohort and identified > 1000 independent variants meeting genome-wide significance. Although GWAS have identified loci associated with WB-LM and A-LM, they are not designed to identify novel rare variations that may have larger effect sizes than common variants. Also, most GWAS were conducted in populations of European-descent and have limited coverage of variants uncommon in European ancestry populations. Furthermore, sex-specific analyses have not been extensively explored in these studies, potentially overlooking important genetic differences between sexes.

Here we aimed to achieve a better understanding of the genetic etiology of muscle mass by attempting to discover novel genetic variants associated with lean mass. We postulated that deep whole genome sequencing performed in population-based cohorts with multiple ancestries may provide new insights into pathways influencing muscle metabolism and muscle integrity regulation. This knowledge in turn is important to identify druggable targets and/or predict adverse effects of treatments [8].

Results

Single-variant association, conditional, and replication analysis

Meta-analysis was performed for ~ 45,000,000 SNPs in the discovery sample (n = 10,726 for WB-LM and 10,672 for A-LM), as well as for sex-specific subgroups (Table 1). The genomic inflation factors ranged from 0.9953 to 1.0023, indicating sufficient control of population stratification and relatedness (Additional file 2: Fig. S1). In total, we identified seven genomic loci that were genome-wide significantly associated with LM (p < 5 × 10−8), five (ENSG00000233359, LINC01661/PRMT6, EMP2, ZCCHC14-DT, PA2G4P2/LINC01722) for WB-LM and two (SSUH2, RCC2P8/COL25A1) for A-LM (Table 2, Additional file 2: Figs. S2 and S3). Two of these loci (LINC01661/PRMT6 and RCC2P8/COL25A1) were sex-specific signals, LINC01661/PRMT6 was identified from WB-LM female analysis, and RCC2P8/COL25A1 was identified from A-LM male analysis. Among these seven identified loci, there were 12 genome-wide significant variants for WB-LM and 3 genome-wide significant variants for A-LM (Additional file 3: Table S1). All of our identified genome-wide significant loci were conditionally distinct based on our stepwise conditional analysis (Additional file 3: Tables S2–S7). Only one of our identified loci, SSUH2, had a previously reported variant (rs6763944, associated with A-LM in a GWAS study [1]) nearby within a ± 0.5 Mb range. After conditioning on the previously known signal, SSUH2 remained genome-wide significant (Additional file 3: Table S8). Therefore, all of our identified genome-wide significant loci were novel. For the UKBB replication samples, one lead variant (rs140266099) from our identified locus, RCC2P8/COL25A, reached nominal significance (i.e., p < 0.05) using the imputed SNP-Chip data (Table 2). After meta-analysis with UKBB, all variants (rs116652927, rs79764157, rs77796060, rs182466396, and rs140266099) remained genome-wide significant, excluding two variants that were not available in the replication samples (Table 2).

Table 1 Study characteristics
Full size table
Table 2 Single-variant association results for WB-LM and A-LM
Full size table

Rare variants aggregated association and replication analysis

We applied five strategies for grouping QC-passed rare variants with minor allele frequency (MAF) ≤ 1% for both lean mass phenotypes to assess the cumulative effect of rare variants within a specific region. A total of three loci were significantly associated with LM (p < Bonferroni corrected p value) (Table 3). All of them were identified through sex-stratified analysis. One protein coding gene DMAC1 was identified from both WB-LM and A-LM female samples. The other lncRNA gene ENSG00000273183 was selected based on A-LM male samples. The two identified genes were also highly expressed in skeletal muscle tissue based on GTEx portal data [9]. However, despite this observation, they were not replicated in the UKBB samples.

Table 3 Rare variants association results
Full size table

Functional annotation using multi-omics data with bioinformatics approaches

Sixty variants were found with LD greater than 0.7 as outlined in Additional file 3: Table S6. Among them, five variants were located in regions relevant to enhancers specific to hMSC, myotubes, and myoblasts cell types, including rs138235889, rs138353434, rs181902470, rs1169376222, and rs187757389 (Additional file 3: Table S9). Through cell-type specific Hi-C data, enhancer-promoter interactions among these five variants were explored, which led to the identification of five genes likely regulated by the five variants, namely FBXO31, MAP1LC3B, ZCCHC14, ETNPPL, and C16orf95 (Additional file 3: Table S10).

For the Gene Ontology (GO) and pathway enrichment analysis, four genes, namely EMP2, PRMT6, COL25A1, and SSUH2, were mapped to a total of 86 GO terms. Out of these, 29 GO terms and two pathways showed a statistically significant association with three of the four genes, EMP2, PRMT6, and COL25A1, with an adjusted p value ≤ 0.05 (Additional file 2: Fig. S4, Additional file 3: Tables S11 and S12). For the transcription factor binding sites (TFBS) annotation, we have observed that rs183684601, which is in high LD with rs182466396 (annotated as near gene SSUH2), is overlapping with open chromatin histone protein (H3) lysine (K) 27 acetylation (H3K27ac), enhancer chromatin H3K4-monomethylation (H3K4me1), and CCCTC-binding factor (CTCF) binding in muscle-relevant cells [10] among the seven variants related to LM (Additional file 2: Fig. S4, Additional file 3: Tables S11 and S12). Moreover, TRANSFAC identified two predicted allele-specific TFBSs for SNP rs183684601, where Kruppel-like factor 6 (KLF6) is predicted to bind at its reference allele, G, and helicase-like transcription factor (HLTF) is predicted to bind at its alternative allele, A [11]. Also, another SNP rs113293310, which is a proxy for rs182466396, is also annotated with histone modification marks and chromatin states in muscle-relevant cells, but with no predicted TFBS for either allele.

Functional validation through animal models

We prioritized two genes based on the following reasoning: one, EMP2 (modeled in Drosophila), was selected because the sentinel SNP was a missense coding variant; the other, SSUH2 (modeled in zebrafish), was a causal gene for myopathy, distal, Tateyama type, and rippling muscle disease 2. While the conservation score was low (2 on DIOPT v8.0, sequence similarity = 43%), the protein encoded by Drosophila CG4984 is considered to be the major ortholog of human EMP2 by containing the claudin (= PMPP2) domain of EMP2, supported by two protein databases (Panther [12] and PhylomeDB [13]).

Drosophila model

Silencing CG4984, homolog of EMP2, in Drosophila muscle reduced mobility at the larval-pupal-adult stages.

Mobility is an important feature of both larval and adult flies. Typical Drosophila larvae move away from each other, known as horizontal movement. However, this movement was reduced in the CG4984-silenced flies (Fig. 1A). At the end of the 3rd instar, typical larvae will climb up the vial wall and spread out to become pupae. However, the CG4984-silenced flies could not climb to the higher vial wall, thus all the pupae located at the lower region. Pupation height as a reflection of the climbing muscle function; therefore, quantification of the height between each pupa and the food surface level in the vial showed that the average pupation height was significantly reduced (Fig. 1B). In adult flies, flying ability can be tested by a geotaxis assay, which revealed a significant reduction in flight ability in the CG4984-silenced flies compared to control flies (Fig. 1C). This indicated that the flight muscle function was dramatically restricted by silencing CG4984.

Fig. 1
figure 1

Silencing of CG4984 in Drosophila muscles led to functional decline and myofibrillar defect. A–C Mef2 > CG4984-IR (Mef2-Gal4:UAS-CG4984-RNAi) flies compared to control (Mef2-Gal4) flies. Mef2, myocyte enhancer factor 2. A Illustration of the concentric circles on the Petri dish that form the zones to determine locomotion. Graph displays the locomotion data based on the number of larvae that crossed each zone within a 1-min time interval. Each data point represents one larva. For each genotype, data are shown for three experiments of 10 larvae each (n = 30 larvae in total). Error bars correspond to SD. ** denotes significance of p < 0.01. B The vials show the difference in pupation height between the control and CG4984-RNAi flies. Quantitation in the bar graph shows the average pupation height in mm. Error bars correspond to SD. **p < 0.01. C Adult mobility measured by negative geotaxis. Three groups of ten female adults from each genotype were tested. The flies were tapped to the bottom. After 10 s, the number of flies above the 8 cm line was recorded. Averaged data are represented by percentages where the number of flies above the 8 cm mark is divided by the total number of flies tested within each group. Error bars correspond to SD. **p < 0.01

Full size image

Silencing CG4984, homolog of EMP2, in Drosophila did not affect muscle volume or morphology, nor lipid mass

Next, we looked at muscle morphology. Given the association of EMP2 with lean muscle mass, we then used flies with muscle-specific deficiency for CG4984 (Mef2-Gal4:UAS-CG4983-RNAi) to study its effect on muscle morphology. The flies showed no significant morphological defects in muscle of the second segment when comparing with controls (Mef2-Gal4) and CG4984-deficient (Mef2 > CG4984-RNAi) larvae, including (1) the organization of muscle bundles (Additional file 2: Fig. S5A); (2) the sarcomere structure of body wall muscle (Additional file 2: Fig. S5B); and (3) the volume of the muscle bundle (Additional file 2: Fig. S5C). To study a potential effect of CG4984 deficiency on lipid mass, we looked at the fat body, which is the major tissue of lipid and energy storage in Drosophila. At the 3rd instar larval stage, the fat body is a single layer of cells, which facilitates the detection of changes in lipid storage. Using Nile red staining of neutral lipid droplet within the fat body, we found no significant difference in either fat body mass or in lipid droplets relative to the area between control (Lsp2-Gal4/ +) and CG4984-RNAi (Lsp > CG4984-RNAi) flies (Additional file 2: Fig. S5D–F).

These findings indicate no obvious role for EMP2-homolog CG4984 on muscle morphology or fat body in flies. In summary, these findings demonstrate that while muscle volume did not change in the CG4984-RNAi flies, deficiency for EMP2-homolog CG4984 did cause significant defects in muscle function evident in reduced mobility, from larval to adult.

Zebrafish model

Multi-locus targeted SSUH2 gene mutation causes excessive skeletal muscle in F0

We sought to decipher the effect of a multi-locus targeted KO of the SSUH2 gene in skeletal muscle morphology of zebrafish (F0 generation). The CRISPR/Cas9-based multi-locus targeted ssuh2 F0 mutant was established based on a high-efficiency method proposed for large genetic screens in zebrafish [14].

To assess the skeletal muscle morphology, we performed hematoxylin and eosin staining. We discovered that ssuh2 F0 mutant zebrafish demonstrated larger muscle fiber area and perimeter (both p < 0.00001) in contrast to the wild type controls at the age of 2.5 months post fertilization (Fig. 2A–B, Additional file 2: Fig. S6). Our results indicate that multi-locus targeted KO of the SSUH2 gene could activate the growth of muscle fibers. We also tested muscle for excess fat by visualization of neutral lipids using Oil Red O staining (Fig. 3A–B, Additional file 2: Fig. S7A–B). The total amount of neutral lipids was apparently not different at dorsal and trunk muscle regions in ssuh2 F0 mutant in contrast to WT fish.

Fig. 2
figure 2

Multi-locus targeted ssuh2 gene mutation causes excessive skeletal muscle fibers growth in zebrafish. A Histological images of skeletal muscle of adult zebrafish obtained by hematoxylin and eosin staining. Wild type (WT) and ssuh2 F0 KO genotypes (left and right panels), respectively. B Muscle fiber area and perimeter measured in WT and ssuh2 F0 with n = 3 animals per genotype at 2.5 months post fertilization. Each dot on the graph corresponds to the muscle fiber area (491 vs 502 fibers per genotype) and perimeter (493 vs 501 fibers per genotype). Statistical analysis performed using Mann–Whitney test, ****p = 0.0001. Scale bar: 20 µM

Full size image
Fig. 3
figure 3

CRISPR/Cas9-based multi-locus targeted ssuh2 gene in zebrafish revealed no effect in the total amount of neutral lipids in skeletal muscle. A Schematic representation of dorsal and trunk muscle fibers’ source in adult zebrafish for lipid staining. Region of interest of dorsal and trunk muscle fibers was marked as indicated in a color code. Visualization of lipid content in dorsal muscle (B) and trunk muscles (C) of zebrafish by Oil Red O (ORO) staining, wild type (top) and ssuh2 F0 KO (bottom), respectively. The intensity of red color marks the amount of lipids in skeletal muscle. Histological sections of skeletal muscle representative of n = 3 animals per group at 2.5 months post fertilization. Scale bar: 20 µM

Full size image

In regard to the effect of the ssuh2 gene knockout in zebrafish mobility, we investigated swimming activity in the WT and crispants, with and without acid exposure (stress), using the distance moved (cm) as the dependent variable in a cumulative link mixed effects model approach. Acute change in pH (such as acidity) is among other mild environmental stressors applied to make fish rapidly change their swimming pattern by trying to escape the stressor [15]. Observational time was utilized as a time bin in our cumulative link mixed effects model to account for repeated measurements and temporal variations in swimming activity. Three different models were employed to account for various factors and to assess the robustness of our findings (Additional file 2: Fig. S8): The reference model (model 1) was employed to assess the impact of gene knockout on mobility, accounting for acid exposure and observational time. Models 2 and 3, excluding acid stress, were utilized to investigate the effect of gene knockout across different observational periods. The results indicate that the ssuh2 gene knockout did not have a significant impact on mobility across all models (model 1: p = 0.45; model 2: p = 0.26; model 3: p = 0.25). In the reference model (model 1), the effect of acid exposure was found to be significant and negatively associated with mobility (estimate =  − 0.21, p = 0.001), while the observational time also showed a highly significant negative effect (estimate =  − 0.24, p < 2 × 10−16). When excluding the effect of acid and considering selected fish (model 2) or selected time bins (model 3), the ssuh2 gene knockout still had no significant effect on mobility. Overall, our findings suggest that the ssuh2 gene knockout does not have a significant positive impact on zebrafish mobility, while acid exposure and observational time exhibit significant negative associations with effect sizes of − 0.21 and − 0.24, respectively.

Discussion

In this WGS study, we have identified seven distinct genomic loci (ENSG00000233359, LINC01661/PRMT6, EMP2, ZCCHC14-DT, PA2G4P2/LINC01722, SSUH2, and RCC2P8/COL25A1) that were significantly associated with LM. After meta-analysis with the UK Biobank replication samples, all available variants remained genome-wide significant. Although rare variants aggregated analysis is primarily challenged by sample size limitations [16], we were able to discover two genes (DMAC1 and ENSG00000273183) through our grouping strategies. These two identified genes were also highly expressed in the skeletal muscle tissue according to public databases. In addition, the identified genes were further investigated through our functional follow-up using bioinformatic approaches and animal models. Such genetic knowledge is important to identify druggable targets for sarcopenia, since on one hand, targets supported by genetic associations for the drug’s lead indication are 2–3 × more likely to pass through clinical development than the target of a drug without this genetic backing [8, 17]. On the other, early discontinuation of clinical trials is attributed to the absence of genetic/omic evidence for a drug target in question [18].

Functional annotation for novel discoveries suggested potential roles for the newly identified genes. We have observed that KLF6, a transcriptional activator, plays a role in myoblast/muscle function in the TFBS annotation. Studies have shown that KLF6 and MEF2D co-localize in the nuclei of myogenic cells and that the MEF2 cis element is an important component of the KLF6 promoter region. TGFβ has been found to enhance KLF6 protein levels in myoblasts, and inhibition of Smad3 represses this effect. Depletion of KLF6 has been shown to enhance myogenic differentiation and reduce myoblast proliferation in response to TGFβ. The findings have important implications for understanding muscle development and various muscle pathologies [19]. HLTF (helicase-like transcription factor) is a DNA helicase and a member of the SNF2 family of chromatin-remodeling proteins. While there is currently limited research on the specific role of HLTF in muscle, it has been found to play a role in DNA damage response and repair [20], which may be relevant in the context of muscle regeneration and degeneration.

COL25A1 (collagen type XXV alpha 1 chain) is involved in congenital fibrosis of the extraocular muscles and in arthrogryposis multiplex congenita; recently, a single-cell transcriptomic atlas of human skeletal muscle aging found its expression to be associated with myofiber typing [21]. PRMT6 (protein arginine methyltransferase 6) encodes a protein that methylates arginine residues in proteins, resulting in specific epigenetic tags for transcriptional repression. As shown in Additional file 2: Fig. S4, PRMT6 is involved in various methyltransferase activities and has a profound effect on gene regulation through DNA methylation. A recent study has highlighted the importance of PRMT6 in regulating muscle phenotypes in the context of spinobulbar muscular atrophy (SBMA). Specifically, PRMT6 is overexpressed in an androgen-dependent manner in the skeletal muscle of patients and mice with SBMA [22].

EMP2 (epithelial membrane protein 2) is an encoded protein that regulates cell membrane composition and is involved in various functions such as endocytosis, cell signaling, and cell proliferation [23]. As shown in Additional file 2: Fig. S4, EMP2 is associated with 19 biological processes and one pathway. It is also worth noting that both EMP2 and COL25A1 are significantly involved in the GO pathway referred to as “supramolecular fiber organization” (GO: 0097435), resembling a structure of actin filament and myosin [24]. SSUH2 (Ssu-2 homolog (Caenorhabditis elegans)) is a gene associated with distal myopathy, Tateyama type, as well as rippling muscle disease 2, which is a form of limb-girdle muscular dystrophy. It is also involved in the dentin dysplasia type I, thus is suspected to play a role in odontogenesis. The integration of these top genes, known or suspected to have a function in muscle mass or function, reveals that there is currently no direct evidence linking them to muscle mass phenotypes. Particularly, two of these genes, EMP2 and SSUH2, represent novel signals and are associated with the LM in sex-combined samples. Therefore, they are believed to play a role in the general population and have been prioritized for functional follow-up in animal models.

The PMP22_Claudin domain makes up nearly the entire epithelial membrane protein 2, EMP2 (167 amino acids) and is conserved from flies to humans. The domain regulates many processes, including the formation of tight junctions, cell–cell adhesion, and cellular contraction [25, 26]. Notably, the PMP22_Claudin domain only makes up about a third of the fly homolog (CG4984, 447 amino acids), raising the possibility that CG4984 conveys additional functions. This notion is further supported by its homology to two calcium channels in humans—CACNG5 and CAGNG7. Furthermore, our Drosophila assays showed significantly decreased mobility during larval development and in adult flies deficient for CG4984. Since we could not detect morphological differences in either the muscle tissue or the fat body, the mobility defect could be of neuronal origin. Further research into this gene is warranted, to determine any effects of CG4984/EMP2 deficiency on neurons or the neuromuscular junctions.

In parallel to the EMP2 mutagenesis, using crispant (F0 generation mutant) zebrafish, we sought to decipher the effect of targeted SSUH2 gene knock-out on skeletal muscle morphology and function. Thus, we discovered that ssuh2 F0 mutant zebrafish demonstrated larger muscle fiber area and perimeter (both p < 0.00001) in contrast to the wild type controls. Our results thus indicate that SSUH2 gene mutation could activate the growth of muscle fibers. We also tested muscle for excess fat by visualization of neutral lipids in dorsal and trunk muscle regions, but found no difference in this parameter, suggesting the larger fibers are not due to excess fat. To evaluate the functional consequence of gene knockout, we investigated swimming performance in the WT and crispant fish, in normal conditions and under stress (acid exposure). The ssuh2 gene knockout had no significant effect on mobility, either speed or distance moved. Overall, our findings suggest that the ssuh2 gene knockout affects muscle fiber bulk (a phenotype similar to human A-LM the gene was associated with) but does not have an impact on zebrafish mobility. Further research is necessary to elucidate the underlying biological mechanisms by which SSUH2 regulates skeletal muscle morphology, thereby enhancing our understanding of its causal and functional role in muscle development.

Our comparison of results with previous GWAS signals reveals significant challenges in replication. Previous GWAS studies predominantly relied on imputed genotypes and focused only on individuals of European ancestry. This lack of replication may be attributed to variations in study designs (Additional file 3: Table S13) and discrepancies in statistical models and sample sizes. Additionally, many of the variants we identified are relatively rare, and our replication cohort, the UK Biobank, employed WGS data for both WB-LM and A-LM, which may account for the modest sample size and limit our ability to replicate results for gene-centered aggregated rarer variants. While we observed several previously identified variants near our significant loci with p values < 10^ − 6 (Additional file 3: Table S8), only one variant (rs6763944) is in proximity to our genome-wide significant variants (p value < 5 × 10^ − 8) [1]. Moreover, it is worth noting that most GWAS employed BIA to assess lean mass. In contrast, our study utilized DXA machines, which may contribute to the observed differences in results.

Our analyses have identified five lncRNAs associated with lean mass: ENSG00000233359, LINC01661, ZCCHC14-DT, LINC01722, and ENSG00000273183. Although these lncRNAs are relatively unexplored and their functional roles are not well understood, the lack of high genomic conservation in animal models limits our ability to validate these findings. Despite this challenge, the identification of these lncRNAs in our study represents a significant discovery. Future research involving loss-of-function and gain-of-function experiments, along with transcriptomic and proteomic analyses, will be crucial for elucidating their precise biological functions.

This study has multiple strengths. First, in contrast with most GWAS published to date, which were conducted in populations of European-descent and have limited coverage of variants uncommon in European ancestry populations, our WGS samples had greater diversity in sample ethnicity, therefore being generalizable to a diverse population of the USA. Second, through our analysis we were able to identify sex-specific signals associated with LM. The chromatin states are not the same in male and female skeletal muscles, as recent methylome and transcriptome integration analysis [27] revealed that skeletal muscle omics manifest profound sex differences. Thus, sex-specificity in both muscle gross phenotype and muscle physiology is well appreciated.

Several aspects of our study can be improved. First, in this study we only analyzed muscle mass (WB-LM and A-LM). But muscle strength and physical performance are important components for understanding of musculoskeletal diseases, such as sarcopenia. The diagnosis of sarcopenia, defined as age‐related loss of skeletal muscle, is based on an assessment of skeletal muscle mass together with low muscle strength and low physical performance. Further association study on those phenotypes or even bivariate analysis is worthwhile. Second, our replication cohort (UK Biobank) has a relatively modest sample size and is primarily composed of individuals of European ancestry, especially for the WGS data. This may explain why the rare variant results were not replicated. A larger multi-ethnicity cohort for replication could be needed. Third, although a large sample size in our studies helps mitigate the impact of unmeasured confounders by randomly distributing them across genotypes, we acknowledge a limited ability to adjust for some important confounders, such as physical activity, diet, or sex steroid hormonal concentrations, which have a significant influence on lean mass. Within the range of participating studies, some either lack these measurements or use different methods for collecting the data. This need for homogenizing data on these potential covariates requires follow-up studies to arrange such analyses. Last, due to limited resources, we only validate two identified genes through our animal models. More comprehensive validation studies can be done in the future.

Conclusions

Through deep WGS data, bioinformatic tools, and functional follow-up in animal models, this study provides new insights into pathways influencing muscle metabolism and muscle mass regulation and informs future studies dedicated to this important metric of the organism.

Methods

Discovery cohorts: TOPMed and LOS

This study included a total of 10,726 participants with WB-LM (n = 10,726) or A-LM (n = 10,672) measurements who were from the Trans-Omics for Precision Medicine (TOPMed) Consortium and the Louisiana Osteoporosis Study (LOS). The participating TOPMed cohorts included the Genetics of Cardiometabolic Health in the Amish [28] (Amish, n = 487 for WB-LM [478 for A-LM]), Cardiovascular Health Study [29] (CHS, n = 1054 [1054]), Framingham Heart Study [30, 31] (FHS, n = 2863 [2863]), San Antonio Family Osteoporosis Study [32, 33] (SAFOS, n = 409 [361]), and Women’s Health Initiative [34] (WHI, n = 934 [934]) participants. The LOS included 4982 participants, whose ethnic composition is ~ 72% White, ~ 23% Black, and ~ 4% Hispanic/Latino (Additional file 3: Table S14) [35]. Basic characteristics of each study, including sample size, sex composition, and mean (standard deviation, SD), of age, weight, height, total fat, WB-LM, and A-LM, are shown in Table 1. All participants provided informed consent and the appropriate institutional review boards approved all studies (Additional file 2: Table S15). Please refer to Additional file 1: SI 1, SI 2, and SI 3, for more information about each study.

Replication cohort: UK Biobank

In total, 4720 participants (4720 genotyped on the SNP-Chip and an overlapping 1115 with available whole genome sequencing (WGS) data) were included with measurements for both WB-LM and A-LM from the UK Biobank (UKBB) [36]. The ethnic composition of these participants is ~ 97% White, 2.2% Asian, 0.9% Black, and 0.2% Hispanic/Latino. The related basic characteristics of this cohort are also shown in Table 1. UKBB has continuously renewed ethical approval from the North West Multi-center Research Ethics Committee, and all studies were carried out in accordance with the appropriate project’s Material Transfer Agreement. Informed consent was obtained from all participants. The UKBB application license number associated with the data and research in this study is 69,804.

Lean mass phenotype measurements

TOPMed

Lean mass was measured in all TOPMed cohorts using dual-energy X-ray absorptiometry (DXA) (Hologic 4500W, Hologic QDR-2000, Lunar/Prodigy) (Additional file 3: Table S15). DXA can be used to measure body composition, including bone mineral, fat, and fat-free soft tissue. For this study, we used fat-free soft tissue (i.e., lean mass) as our phenotype. We included two types of lean mass: WB-LM and A-LM. The latter includes only lean mass in the arms and legs, which has been demonstrated to be a valid estimate of skeletal muscle mass, especially since the arms and legs do not contain visceral organs [37].

LOS

Both fat mass and lean mass were measured with a DXA machine (Hologic QDR-4500 Discovery DXA scanner, Hologic Inc., Bedford, MA, USA) by trained and certified research staff. The machine was calibrated daily, and software and hardware were kept up-to-date during the data collection process. The two lean mass phenotypes derived using manufacturer’s image analysis protocols, including WB-LM and A-LM, as demonstrated elsewhere [6], were used in the following analyses.

UKBB

Lean mass was measured in UKBB participants using the Lunar-GE iDXA dual-energy X-ray absorptiometry device (GE-Lunar, Madison, WI, USA). Scans were analyzed by a radiographer using the iDXA device at or shortly after acquisition, generating numerical measures of body composition split into fat mass and lean mass (fat-free soft tissue). We again included two lean mass phenotypes, WB-LM (UKBB data field number 23280) and A-LM (sum of UKBB field numbers 23275 and 23,258), defined in the same way as described for the TOPMed studies.

Sequencing data and quality control

TOPMed

Whole genome sequencing (WGS) data for the TOPMed program are acquired by multiple sequencing centers over time [38]. The TOPMed Informatics Research Center (IRC) performs joint variant identification and genotype calling on all available samples periodically and the resulting call set is referred to as a genotype “Freeze.” In this study, we used TOPMed Freeze 8, ~ 30 × WGS data with ≥ 95% of the genome covered to 10 × or greater. The reads were aligned to human genome build GRCh38 using a common pipeline across all centers. For the variant quality control (QC), the inferred pedigree of related and duplicated samples was used to calculate the Mendelian consistency and to train a support vector machine (SVM) classifier. Variants with excess heterozygosity or Mendelian discordance are filtered out. The sample QC included concordance between annotated and genetic sex inferred from the WGS data, concordance between prior SNP array genotypes and WGS-derived genotypes, and comparisons of observed and expected relatedness from pedigrees. Discordant samples were either excluded or were resolved through prior genotyping comparisons and/or pedigree checks. And the estimated sample contamination was below 10% in our analysis. Please refer to https://topmed.nhlbi.nih.gov/topmed-whole-genome-sequencing-methods-freeze-8 and each study’s dbGaP accession [39] for more details about TOPMed WGS data and quality control.

LOS

A total of 5002 samples with genomic DNA were collected and underwent WGS using a BGISEQ-500 sequencer (Beijing Genomics Institute (BGI Group), Shenzhen, China) to generate two sequencing runs of paired-end 350 bp reads with an average sequencing depth of ~ 21 × and 92.29% of the genome covered to at least 10 × coverage. The aligned and cleaned data of each sample were mapped to the human reference genome (GRCh38/hg38) by the use of the Burrows-Wheeler Aligner (BWA) [40] software following the recommended Best Practices for variant analysis with the Genome Analysis Toolkit (GATK version 3.7, https://www.broadinstitute.org/gatk/guide/best-practices) to ensure accurate variant calling [41]. The details of the WGS variant analysis including identification of marker duplicates, recalibration for base quality scores, realignment of indels, and variant calling using variant quality score recalibration are described in Additional file 1: SI 4.

UKBB

For the SNP chip genotyping data, the genotypes of UKBB participants (UKBB field number 22828) were determined for ~ 90% of participants via Affymetrix UK Biobank Axiom Array (Santa Clara, CA, USA) and for the other ~ 10% of participants using the Affymetrix UK BiLEVE Axiom Array [8]. Genotypes were further imputed with the Haplotype Reference Consortium (HRC) panel [42], to ultimately obtain data on ~ 96 million genotypes mapped to GRCh37, with quality control and imputation details as previously described [43]. LiftOver was used to map the genotypes from GRCh37 to GRCh38 coordinates [44]. WGS of 150,119 UKBB participants was performed by two sequencing centers (deCODE Genetics and Wellcome Trust Sanger Institute) on stored blood samples’ DNA, using Illumina NovaSeq machines with an average coverage of 32.5 × , and coverage across all samples ranging from 23.5 × to ~ 50 × [45]. Sequence reads were mapped to GRCh38 and SNPs were jointly called across all individuals in the dataset with GraphTyper [46]. No further QC filters were applied. More details on WGS of UKBB participants, including sequencing center batch effects, sample concordance QC, and GraphTyper parameters (though data in this study come from an earlier WGS release and were not filtered by AAscore), were previously published [45] or described in Additional file 1: SI 5.

Single-variant association analysis

We performed single-variant association tests using linear mixed models on the two lean mass phenotypes, WB-LM and A-LM separately. A two-stage procedure of association test was conducted. Within each study, we first performed linear regression of lean body mass as a function of age (years), age squared, sex, weight (kg), height (cm), total fat (kg), and study specific covariates (e.g., ethnicity) to generate study-specific residuals. Adjusting for height and total fat in our model ensures that the identified SNPs contribute to lean mass independently of their effects on total fat or height [6, 47]. Second, we performed inverse normal transformation on the generated residuals and fit them as a null model with all PCs for TOPMed samples or significant PCs for LOS samples (Additional file 3: Table S16). The output from this second stage was used to perform genome-wide score tests of genetic association for all QC-passed individual variants. We then meta-analyzed association results from TOPMed and LOS through an inverse variance weighted approach and focused on variants with minor allele frequency (MAF) ≥ 0.1% and minor allele count (MAC) > 40. The genome-wide significance threshold \(p<5\times {10}^{-8}\) was used as our significant level. We also conducted a sex-stratified analysis with the same procedure for WB-LM and A-LM after excluding the sex covariate in the regression model at the first stage. The same procedure was applied to the UKBB cohort using WGS data or imputed SNP-Chip data (if WGS data not available) on significant variants as a replication. Finally, association results between discovery cohorts and the replication cohort were meta-analyzed through inverse-variance approach. The related software we used is provided in Additional file 1: SI 6.

Conditional analysis

We performed two sets of conditional analysis to pinpoint a short list of important variants. The first approach was conditional on our own findings to identify a set of distinct signals. A stepwise conditional analysis was performed on our variants with p < 1 × 10−6. Taking WB-LM as an example, for each chromosome, we identified the most significant variant as the “peak variant” and then fit a new model adjusted for both the covariates as well as this peak variant and calculated new p values for the rest of variants on that chromosome. If more than one variant was significant at the 1 × 10−6 level in the new result, we performed a second round of conditional analysis, re-fitting the model, adjusting for the new peak variant along with the first peak variant and the original covariates. We continued this procedure iteratively until no additional variants were significant at a p value threshold of < 1 × 10−6. The second approach was conditional on previously reported signals to determine whether our identified signals were novel. We considered previously known signals within a ± 0.5 Mb range for each of our signals to perform conditional analysis. Previously identified associated variants were downloaded from the GWAS catalog (version: All studies v1.0.2, https://www.ebi.ac.uk/gwas/api/search/downloads/studies_alternative) and were matched to our variants using either rsid or genomic locus and alleles. We included five GWAS studies for WB-LM or A-LM phenotypes (Additional file 3: Table S17) [1, 6, 7, 48, 49]. We also performed these two types of conditional analysis on males and females separately.

Rare variants aggregated association analysis

We applied five strategies for grouping rare variants for both lean mass phenotypes to assess the cumulative effect of rare variants within a specific region. Two of them are based on all genes, one including loss of function and missense variants and the other including coding and non-coding regulatory variants. The other three strategies are specific to muscle-related regions. These include muscle-specific lncRNAs from lncRNAKB (http://psychiatry.som.jhmi.edu/lncrnakb/tissues/index.php?tissue=Muscle), differentially methylated regions (DMRs) in human skeletal muscle and muscle cell related transcription start sites (TSSs). More details about how we defined these muscle-related regions are described in Additional file 1: SI 6. We included variants with MAF ≤ 1% that passed QC to aggregate based on the above strategies. The association model was the same as the single variant analysis except we tested each set of aggregated variants here instead of each variant. We applied the same SKAT test for aggregated variants association analysis to TOPMed and LOS participants. We used the probit method to meta-analyze association p values from TOPMed and LOS studies. We used a Bonferroni correction to determine the significance, adjusting for the number of aggregated regions. The same approach was also performed on males and females separately. For the replication, we performed the same procedure using WGS data of UKBB except that we only focused on identified regions from discovery cohorts and checked if those regions were nominal significant (i.e., \(p<0.05\)).

Functional annotation using multi-omics data with bioinformatics approaches

We utilized a linkage disequilibrium (LD) value of greater than 0.7 with the lead SNPs as the cutoff to identify potential causal variants from our single-variant association results. To ascertain the cell-type specific regulatory function of these potential causal variants, we employed Assay for Transposase-Accessible Chromatin using sequencing (ATAC-seq) on human mesenchymal stem cells (hMSC), myotubes and myoblasts ATAC-seq [50] and DNase-seq from ENCODE [51]. Our aim was to identify the open chromatin regions, transcription factor binding sites, and enhancer regions, with the aid of ChromHMM [52], specific to hMSC, myotubes, and myoblasts cell types. Moreover, cell-type specific Hi-C data were applied to discern enhancer-promoter interactions.

To comprehend the functions of LM-related SNPs, we conducted gene enrichment analysis using Enrichr [53] and multiple pathway databases such as WikiPathways [54], KEGG [55], and Reactome [56], as well as the GO database [57]. The threshold for significance in GO term and pathway analysis is set at an adjusted p value of ≤ 0.05 using the Benjamini-Hochberg (BH) method. Additionally, we employed muscle-relevant functional annotation and regulatory information from HaploReg [58] and TRANSFAC [11] databases. To identify potential causal variants, we performed HaploReg to select variants within an LD block based on the 1000 Genome Project. Subsequently, we focused on candidate SNPs with muscle-related annotations in histone modification and chromatin status [10] and searched for predicted transcription factor binding sites (TFBS) using the TRANSFAC database [11]. More details can be found in Additional file 1: SI 7.

Functional validation through animal models

We considered two types of animal models, including Drosophila and zebrafish, to validate our findings. To consider for functional validation, we prioritized genes in which we identified missense variants and genes more likely to be functionally related to musculoskeletal phenotypes. We focused on non-sex specific genes. Furthermore, we utilized Open Targets Genetics web portal (https://genetics.opentargets.org/) to verify the association of our identified variants with the genes we selected for knocking out. Approximately 61% of disease-causing genes in humans have functional homologs in Drosophila [59], and the fly can be used to study the function of these genes and the consequences of mutations. Zebrafish also have become an attractive animal model for musculoskeletal-specific genetic modeling [60, 61], given both a conserved genetics of the zebrafish as well as their muscle’s similarity to human’s [62]. Effective gene editing with CRISPR/Cas9 system has been a major tool for the functional study of genes in zebrafish, making it more rapid, by enabling phenotypic analysis on the first generation (G0, also known as crispants) [14, 63], saving valuable time and resources. We briefly describe the animal models we used below, and more details about these two animal models are provided in Additional file 1: SI 8 and SI 9.

Drosophila model

The Drosophila model was employed because of its well-established genetic tools and conserved mechanisms of muscle development, providing valuable insights into the regulation of muscle mass that can be extrapolated to human biology [64, 65]. Protein encoded by Drosophila CG4984 is orthologous to the PMPP2_Claudin domain of EMP2. The UAS-CG4984-RNAi stock flies were crossed with the myocyte enhancer factor 2 (Mef2)-Gal4 (muscle-specific) or with the larval serum protein 2 (Lsp2)-Gal4 fly lines for fat body-specific silencing of CG4984. Larvae (3 days old) underwent phalloidin staining; body wall muscles were imaged and the number of myofibers obtained. Neutral lipid quantitation was obtained by Nile red staining; lipid droplet size and number were measured. Pupation height and larval locomotion was recorded, followed by the adult locomotion measured by negative geotaxis, and compared between the genotype groups.

Zebrafish model

We utilized the zebrafish model due to its high genetic homology to humans, with approximately 80% of human disease-related genes conserved, making it an ideal organism for high-throughput genetic knock-out studies to investigate gene functions related to human lean muscle mass [61, 66]. We generated ssuh2 G0 knockout mutants (crispants) by multi-locus targeted CRISPR/Cas9 technology (to assure high efficiency of the mutagenesis). Using ssuh2 crispant zebrafish in comparison to the wild type (WT) controls, we then measured skeletal muscle fiber area and perimeter in young adult (2.5 months old) zebrafish. We also tested muscle for excess fat by visualization of neutral lipids in dorsal and trunk muscle regions. Furthermore, we investigated the swimming distance and swimming behavior, in normal conditions and under stress (acid exposure). Data were tested for normality using the Shapiro–Wilk test (α = 0.05). Normally distributed data were analyzed by Student’s t-test (two genotype groups). Non-normal distributed data were analyzed by a Mann–Whitney test (two groups). Statistical significance was defined as p < 0.05. The cumulative link mixed effect model approach was used to compare the effect of ssuh2 gene knockout on mobility.

Data availability

The datasets analyzed from TOPMed Freeze 8 are available in the dbGaP repository [accession numbers: phs00956.v1.p1 for Amish (28), phs001368.v2.p2 for CHS (29), phs000974.v1.p1 for FHS (30, 31), phs001215.v4.p2 for SAFOS (32, 33), and phs001237.v2.p1 for WHI (34)]. Instructions for accessing TOPMed data can be found at https://www.nhlbiwgs.org/topmed-data-access-scientific-community. The datasets from LOS that support this study’s findings are available from the principal investigator (H.W.D., hdeng2@tulane.edu) upon reasonable request. Access will be granted for academic research purposes, subject to IRB approval and the completion of a data use agreement. Additionally, the LOS WGS data is in the process of being deposited in the AgingResearchBiobank (https://agingresearchbiobank.nia.nih.gov/), where it will be made available to qualified researchers upon application and approval. The UK Biobank data used in this study are available to researchers upon application and approval through the UK Biobank’s access management system (https://www.ukbiobank.ac.uk/enable‐your‐research/apply‐for‐access). The analytical scripts and experimental data can be accessed at the Mendeley Data (67) [https://doiorg.publicaciones.saludcastillayleon.es/10.17632/mk32tnmwwt.1https://doiorg.publicaciones.saludcastillayleon.es/10.17632/mk32tnmwwt.1]. Summary statistics can be found in the Musculoskeletal Knowledge Portal [https://msk.hugeamp.org/].

References

  1. Pei YF, Liu YZ, Yang XL, Zhang H, Feng GJ, Wei XT, et al. The genetic architecture of appendicular lean mass characterized by association analysis in the UK Biobank study. Commun Biol. 2020;3(1):608.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  2. Prado CM, Purcell SA, Alish C, Pereira SL, Deutz NE, Heyland DK, et al. Implications of low muscle mass across the continuum of care: a narrative review. Ann Med. 2018;50(8):675–93.

    Article  PubMed  Google Scholar 

  3. Mareschal J, Achamrah N, Norman K, Genton L. Clinical value of muscle mass assessment in clinical conditions associated with malnutrition. J Clin Med. 2019;8(7):1040.

    Article  PubMed  PubMed Central  Google Scholar 

  4. Lee DH, Keum N, Hu FB, Orav EJ, Rimm EB, Willett WC, et al. Predicted lean body mass, fat mass, and all cause and cause specific mortality in men: prospective US cohort study. BMJ. 2018;362: k2575.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Hsu FC, Lenchik L, Nicklas BJ, Lohman K, Register TC, Mychaleckyj J, et al. Heritability of body composition measured by DXA in the diabetes heart study. Obes Res. 2005;13(2):312–9.

    Article  PubMed  Google Scholar 

  6. Zillikens MC, Demissie S, Hsu YH, Yerges-Armstrong LM, Chou WC, Stolk L, et al. Large meta-analysis of genome-wide association studies identifies five loci for lean body mass. Nat Commun. 2017;8(1):80.

    Article  PubMed  PubMed Central  Google Scholar 

  7. Hernandez Cordero AI, Gonzales NM, Parker CC, Sokoloff G, Vandenbergh DJ, Cheng R, et al. Genome-wide associations reveal human-mouse genetic convergence and modifiers of myogenesis, CPNE1 and STC2. Am J Hum Genet. 2020;106(1):138.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Minikel EV, Painter JL, Dong CC, Nelson MR. Refining the impact of genetic evidence on clinical success. Nature. 2024;629(8012):624–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. GTEx Consortium. The Genotype-Tissue Expression (GTEx) project. Nat Genet. 2013;45(6):580–5.

    Article  Google Scholar 

  10. Roadmap Epigenomics C, Kundaje A, Meuleman W, Ernst J, Bilenky M, Yen A, et al. Integrative analysis of 111 reference human epigenomes. Nature. 2015;518(7539):317–30.

    Article  Google Scholar 

  11. Wingender E, Chen X, Hehl R, Karas H, Liebich I, Matys V, et al. TRANSFAC: an integrated system for gene expression regulation. Nucleic Acids Res. 2000;28(1):316–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Thomas PD, Ebert D, Muruganujan A, Mushayahama T, Albou LP, Mi H. PANTHER: making genome-scale phylogenetics accessible to all. Protein Sci. 2022;31(1):8–22.

    Article  CAS  PubMed  Google Scholar 

  13. Fuentes D, Molina M, Chorostecki U, Capella-Gutierrez S, Marcet-Houben M, Gabaldon T. PhylomeDB V5: an expanding repository for genome-wide catalogues of annotated gene phylogenies. Nucleic Acids Res. 2022;50(D1):D1062–8.

    Article  CAS  PubMed  Google Scholar 

  14. Kroll F, Powell GT, Ghosh M, Gestri G, Antinucci P, Hearn TJ, et al. A simple and effective F0 knockout method for rapid screening of behaviour and other complex phenotypes. Elife. 2021;10:10.

    Article  Google Scholar 

  15. Lee HB, Schwab TL, Sigafoos AN, Gauerke JL, Krug RG 2nd, Serres MR, et al. Novel zebrafish behavioral assay to identify modifiers of the rapid, nongenomic stress response. Genes Brain Behav. 2019;18(2):e12549.

    Article  PubMed  Google Scholar 

  16. Young KL, Fisher V, Deng X, Brody JA, Graff M, Lim E, et al. Whole-exome sequence analysis of anthropometric traits illustrates challenges in identifying effects of rare genetic variants. HGG Adv. 2023;4(1): 100163.

    CAS  PubMed  Google Scholar 

  17. Trajanoska K, Bherer C, Taliun D, Zhou S, Richards JB, Mooser V. From target discovery to clinical drug development with human genetics. Nature. 2023;620(7975):737–45.

    Article  CAS  PubMed  Google Scholar 

  18. Razuvayevskaya O, Lopez I, Dunham I, Ochoa D. Genetic factors associated with reasons for clinical trial stoppage. Nat Genet. 2024;56(9):1862–7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Dionyssiou MG, Salma J, Bevzyuk M, Wales S, Zakharyan L, McDermott JC. Kruppel-like factor 6 (KLF6) promotes cell proliferation in skeletal myoblasts in response to TGFbeta/Smad3 signaling. Skelet Muscle. 2013;3(1):7.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. van Toorn M, Turkyilmaz Y, Han S, Zhou D, Kim HS, Salas-Armenteros I, et al. Active DNA damage eviction by HLTF stimulates nucleotide excision repair. Mol Cell. 2022;82(7):1343-58 e8.

    Article  PubMed  PubMed Central  Google Scholar 

  21. Kedlian VR, Wang Y, Liu T, Chen X, Bolt L, Shen Z, et al. Human skeletal muscle ageing atlas. 2022.

  22. Prakasam R, Bonadiman A, Andreotti R, Zuccaro E, Dalfovo D, Marchioretti C, et al. LSD1/PRMT6-targeting gene therapy to attenuate androgen receptor toxic gain-of-function ameliorates spinobulbar muscular atrophy phenotypes in flies and mice. Nat Commun. 2023;14(1):603.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  23. O’Leary NA, Wright MW, Brister JR, Ciufo S, Haddad D, McVeigh R, et al. Reference sequence (RefSeq) database at NCBI: current status, taxonomic expansion, and functional annotation. Nucleic Acids Res. 2016;44(D1):D733–45.

    Article  CAS  PubMed  Google Scholar 

  24. Zhang Z, Cheng L, Zhao J, Zhang H, Zhao X, Liu Y, et al. Muscle-mimetic synergistic covalent and supramolecular polymers: phototriggered formation leads to mechanical performance boost. J Am Chem Soc. 2021;143(2):902–11.

    Article  CAS  PubMed  Google Scholar 

  25. Dong Y, Simske JS. Vertebrate claudin/PMP22/EMP22/MP20 family protein TMEM47 regulates epithelial cell junction maturation and morphogenesis. Dev Dyn. 2016;245(6):653–66.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Morales SA, Telander DG, Mareninov S, Nagy A, Wadehra M, Braun J, et al. Anti-EMP2 diabody blocks epithelial membrane protein 2 (EMP2) and FAK mediated collagen gel contraction in ARPE-19 cells. Exp Eye Res. 2012;102:10–6.

    Article  CAS  PubMed  Google Scholar 

  27. Landen S, Jacques M, Hiam D, Alvarez-Romero J, Harvey NR, Haupt LM, et al. Skeletal muscle methylome and transcriptome integration reveals profound sex differences related to muscle function and substrate metabolism. Clin Epigenetics. 2021;13(1):202.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  28. Streeten EA, Ryan KA, McBride DJ, Pollin TI, Shuldiner AR, Mitchell BD. The relationship between parity and bone mineral density in women characterized by a homogeneous lifestyle and high parity. J Clin Endocrinol Metab. 2005;90(8):4536–41.

    Article  CAS  PubMed  Google Scholar 

  29. Fried LP, Borhani NO, Enright P, Furberg CD, Gardin JM, Kronmal RA, et al. The cardiovascular health study: design and rationale. Ann Epidemiol. 1991;1(3):263–76.

    Article  CAS  PubMed  Google Scholar 

  30. Dawber TR, Kannel WB, Lyell LP. An approach to longitudinal studies in a community: the Framingham study. Ann N Y Acad Sci. 1963;107:539–56.

    Article  CAS  PubMed  Google Scholar 

  31. Dawber TR, Meadors GF, Moore FE Jr. Epidemiological approaches to heart disease: the Framingham study. Am J Public Health Nations Health. 1951;41(3):279–81.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  32. Mitchell BD, Kammerer CM, Schneider JL, Perez R, Bauer RL. Genetic and environmental determinants of bone mineral density in Mexican Americans: results from the San Antonio Family Osteoporosis Study. Bone. 2003;33(5):839–46.

    Article  PubMed  Google Scholar 

  33. Mitchell BD, Kammerer CM, Blangero J, Mahaney MC, Rainwater DL, Dyke B, et al. Genetic and environmental contributions to cardiovascular risk factors in Mexican Americans. The San Antonio Family Heart Study. Circulation. 1996;94(9):2159–70.

    Article  CAS  PubMed  Google Scholar 

  34. Jackson RD, LaCroix AZ, Cauley JA, McGowan J. The Women’s Health Initiative calcium-vitamin D trial: overview and baseline characteristics of participants. Ann Epidemiol. 2003;13(9 Suppl):S98-106.

    Article  PubMed  Google Scholar 

  35. He H, Liu Y, Tian Q, Papasian CJ, Hu T, Deng HW. Relationship of sarcopenia and body composition with osteoporosis. Osteoporos Int. 2016;27(2):473–82.

    Article  CAS  PubMed  Google Scholar 

  36. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018;562(7726):203–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  37. Chen Z, Wang Z, Lohman T, Heymsfield SB, Outwater E, Nicholas JS, et al. Dual-energy X-ray absorptiometry is a valid tool for assessing skeletal muscle mass in older women. J Nutr. 2007;137(12):2775–80.

    Article  CAS  PubMed  Google Scholar 

  38. Taliun D, Harris DN, Kessler MD, Carlson J, Szpiech ZA, Torres R, et al. Sequencing of 53,831 diverse genomes from the NHLBI TOPMed Program. Nature. 2021;590(7845):290–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Regier AA, Farjoun Y, Larson DE, Krasheninina O, Kang HM, Howrigan DP, et al. Functional equivalence of genome sequencing analysis pipelines enables harmonized variant calling across human genetics projects. Nat Commun. 2018;9(1):4038.

    Article  PubMed  PubMed Central  Google Scholar 

  40. Li H, Durbin R. Fast and accurate short read alignment with Burrows-Wheeler transform. Bioinformatics. 2009;25(14):1754–60.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  41. McKenna A, Hanna M, Banks E, Sivachenko A, Cibulskis K, Kernytsky A, et al. The Genome Analysis Toolkit: a MapReduce framework for analyzing next-generation DNA sequencing data. Genome Res. 2010;20(9):1297–303.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. McCarthy S, Das S, Kretzschmar W, Delaneau O, Wood AR, Teumer A, et al. A reference panel of 64,976 haplotypes for genotype imputation. Nat Genet. 2016;48(10):1279–83.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  43. Welsh S, Peakman T, Sheard S, Almond R. Comparison of DNA quantification methodology used in the DNA extraction protocol for the UK Biobank cohort. BMC Genomics. 2017;18(1):26.

    Article  PubMed  PubMed Central  Google Scholar 

  44. Kent WJ, Sugnet CW, Furey TS, Roskin KM, Pringle TH, Zahler AM, et al. The human genome browser at UCSC. Genome Res. 2002;12(6):996–1006.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  45. Halldorsson BV, Eggertsson HP, Moore KHS, Hauswedell H, Eiriksson O, Ulfarsson MO, et al. The sequences of 150,119 genomes in the UK Biobank. Nature. 2022;607(7920):732–40.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  46. Eggertsson HP, Jonsson H, Kristmundsdottir S, Hjartarson E, Kehr B, Masson G, et al. Graphtyper enables population-scale genotyping using pangenome graphs. Nat Genet. 2017;49(11):1654–60.

    Article  CAS  PubMed  Google Scholar 

  47. Karasik D, Zillikens MC, Hsu YH, Aghdassi A, Akesson K, Amin N, et al. Disentangling the genetics of lean mass. Am J Clin Nutr. 2019;109(2):276–87.

    Article  PubMed  PubMed Central  Google Scholar 

  48. Hubel C, Gaspar HA, Coleman JRI, Finucane H, Purves KL, Hanscombe KB, et al. Genomics of body fat percentage may contribute to sex bias in anorexia nervosa. Am J Med Genet B Neuropsychiatr Genet. 2019;180(6):428–38.

    Article  PubMed  Google Scholar 

  49. Tachmazidou I, Suveges D, Min JL, Ritchie GRS, Steinberg J, Walter K, et al. Whole-genome sequencing coupled to imputation discovers genetic signals for anthropometric traits. Am J Hum Genet. 2017;100(6):865–84.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Tsai MJ, Reppe S, Sato T, Gill R, Wein M, Gautvik K, et al. The musculoskeletal 3D epigenome atlas. American Society of Human Genetics; Oct 18–22; Virtual. ASHG; 2021. p. 325. https://www.ashg.org/wp-content/uploads/2022/01/2021-ASHGMeeting-Abstracts.pdf.

  51. Luo Y, Hitz BC, Gabdank I, Hilton JA, Kagda MS, Lam B, et al. New developments on the Encyclopedia of DNA elements (ENCODE) data portal. Nucleic Acids Res. 2020;48(D1):D882–9.

    Article  CAS  PubMed  Google Scholar 

  52. Ernst J, Kellis M. ChromHMM: automating chromatin-state discovery and characterization. Nat Methods. 2012;9(3):215–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  53. Chen EY, Tan CM, Kou Y, Duan Q, Wang Z, Meirelles GV, et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics. 2013;14: 128.

    Article  PubMed  PubMed Central  Google Scholar 

  54. Kelder T, Pico AR, Hanspers K, van Iersel MP, Evelo C, Conklin BR. Mining biological pathways using WikiPathways web services. PLoS One. 2009;4(7): e6447.

    Article  PubMed  PubMed Central  Google Scholar 

  55. Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27–30.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  56. Gillespie M, Jassal B, Stephan R, Milacic M, Rothfels K, Senff-Ribeiro A, et al. The reactome pathway knowledgebase 2022. Nucleic Acids Res. 2022;50(D1):D687–92.

    Article  CAS  PubMed  Google Scholar 

  57. Ashburner M, Ball CA, Blake JA, Botstein D, Butler H, Cherry JM, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  58. Ward LD, Kellis M. HaploReg: a resource for exploring chromatin states, conservation, and regulatory motif alterations within sets of genetically linked variants. Nucleic Acids Res. 2012;40(Database issue):D930-4.

    Article  CAS  PubMed  Google Scholar 

  59. Rubin GM, Yandell MD, Wortman JR, Gabor Miklos GL, Nelson CR, Hariharan IK, et al. Comparative genomics of the eukaryotes. Science. 2000;287(5461):2204–15.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  60. Kague E, Karasik D. Functional validation of osteoporosis genetic findings using small fish models. Genes (Basel). 2022;13(2):279.

    Article  CAS  PubMed  Google Scholar 

  61. Daya A, Donaka R, Karasik D. Zebrafish models of sarcopenia. Dis Model Mech. 2020;13(3):dmm042689.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  62. Talbot JC, Teets EM, Ratnayake D, Duy PQ, Currie PD, Amacher SL. Muscle precursor cell movements in zebrafish are dynamic and require six family genes. Development. 2019;146(10):dev171421.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  63. Bek JW, Shochat C, De Clercq A, De Saffel H, Boel A, Metz J, et al. Lrp5 mutant and crispant zebrafish faithfully model human osteoporosis, establishing the zebrafish as a platform for CRISPR-based functional screening of osteoporosis candidate genes. J Bone Miner Res. 2021;36(9):1749–64.

    Article  CAS  PubMed  Google Scholar 

  64. Schnorrer F, Schonbauer C, Langer CC, Dietzl G, Novatchkova M, Schernhuber K, et al. Systematic genetic analysis of muscle morphogenesis and function in Drosophila. Nature. 2010;464(7286):287–91.

    Article  CAS  PubMed  Google Scholar 

  65. Chaturvedi D, Reichert H, Gunage RD, VijayRaghavan K. Identification and functional characterization of muscle satellite cells in Drosophila. Elife. 2017;6:6.

    Article  Google Scholar 

  66. Karuppasamy M, English KG, Henry CA, Manzini MC, Parant JM, Wright MA, et al. Standardization of zebrafish drug testing parameters for muscle diseases. Dis Model Mech. 2024;17(1):dmm050339.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  67. Zhang X, Su K, Hsu YH, Crandall CJ, Han Z, Jackson RD, et al. Multi-Ancestry Whole Genome Sequencing Analysis of Lean Body Mass, Mendeley Data, V1. 2025. https://doiorg.publicaciones.saludcastillayleon.es/10.17632/mk32tnmwwt.1.

  68. Kannel WB, Feinleib M, McNamara PM, Garrison RJ, Castelli WP. An investigation of coronary heart disease in families. The Framingham Offspring Study. Am J Epidemiol. 1979;110(3):281–90.

    Article  CAS  PubMed  Google Scholar 

  69. Feinleib M, Kannel WB, Garrison RJ, McNamara PM, Castelli WP. The Framingham Offspring Study. Design and preliminary data. Prev Med. 1975;4(4):518–25.

    Article  CAS  PubMed  Google Scholar 

  70. Roubenoff R, Baumgartner RN, Harris TB, Dallal GE, Hannan MT, Economos CD, et al. Application of bioelectrical impedance analysis to elderly populations. J Gerontol A Biol Sci Med Sci. 1997;52(3):M129–36.

    Article  CAS  PubMed  Google Scholar 

  71. Deng HW, Shen H, Xu FH, Deng HY, Conway T, Zhang HT, et al. Tests of linkage and/or association of genes for vitamin D receptor, osteocalcin, and parathyroid hormone with bone mineral density. J Bone Miner Res. 2002;17(4):678–86.

    Article  CAS  PubMed  Google Scholar 

  72. Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of sociodemographic and health-related characteristics of UK Biobank participants with those of the general population. Am J Epidemiol. 2017;186(9):1026–34.

    Article  PubMed  PubMed Central  Google Scholar 

  73. DePristo MA, Banks E, Poplin R, Garimella KV, Maguire JR, Hartl C, et al. A framework for variation discovery and genotyping using next-generation DNA sequencing data. Nat Genet. 2011;43(5):491–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  74. Mbatchou J, Barnard L, Backman J, Marcketta A, Kosmicki JA, Ziyatdinov A, et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet. 2021;53(7):1097–103.

    Article  CAS  PubMed  Google Scholar 

  75. Gogarten SM, Sofer T, Chen H, Yu C, Brody JA, Thornton TA, et al. Genetic association testing using the GENESIS R/Bioconductor package. Bioinformatics. 2019;35(24):5346–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  76. Zhan X, Hu Y, Li B, Abecasis GR, Liu DJ. RVTESTS: an efficient and comprehensive tool for rare variant association analysis using sequence data. Bioinformatics. 2016;32(9):1423–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  77. Willer CJ, Li Y, Abecasis GR. METAL: fast and efficient meta-analysis of genomewide association scans. Bioinformatics. 2010;26(17):2190–1.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  78. Seifuddin F, Singh K, Suresh A, Judy JT, Chen YC, Chaitankar V, et al. lncRNAKB, a knowledgebase of tissue-specific functional annotation and trait association of long noncoding RNA. Sci Data. 2020;7(1):326.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  79. Voisin S, Jacques M, Landen S, Harvey NR, Haupt LM, Griffiths LR, et al. Meta-analysis of genome-wide DNA methylation and integrative omics of age in human skeletal muscle. J Cachexia Sarcopenia Muscle. 2021;12(4):1064–78.

    Article  PubMed  PubMed Central  Google Scholar 

  80. Consortium EP. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57–74.

    Article  Google Scholar 

  81. Zhou W, Bi W, Zhao Z, Dey KK, Jagadeesh KA, Karczewski KJ, et al. SAIGE-GENE+ improves the efficiency and accuracy of set-based rare variant association tests. Nat Genet. 2022;54(10):1466–9.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  82. Spletter ML, Barz C, Yeroslaviz A, Zhang X, Lemke SB, Bonnard A, et al. A transcriptomics resource reveals a transcriptional transition during ordered sarcomere morphogenesis in flight muscle. Elife. 2018;7:7.

    Article  Google Scholar 

  83. Gargano JW, Martin I, Bhandari P, Grotewiel MS. Rapid iterative negative geotaxis (RING): a new method for assessing age-related locomotor decline in Drosophila. Exp Gerontol. 2005;40(5):386–95.

    Article  PubMed  Google Scholar 

Download references

Acknowledgements

Molecular data for the Trans-Omics in Precision Medicine (TOPMed) program was supported by the National Heart, Lung, and Blood Institute (NHLBI). Genome sequencing for “NHLBI TOPMed: Genetics of Cardiometabolic Health in the Amish” (phs00956.v1.p1) was performed at Broad Institute Genomics Platform (3R01HL121007-01S1). Genome sequencing for “NHLBI TOPMed: Cardiovascular Health Study” (phs001368.v2.p2) was performed at Baylor College of Medicine Human Genome Sequencing Center (HHSN268201600033I), Broad Institute Genomics Platform (HHSN268201600034I). Genome sequencing for “NHLBI TOPMed: Whole Genome Sequencing and Related Phenotypes in the Framingham Heart Study” (phs000974.v1.p1) was performed at the Broad Institute Genomics Platform (3R01HL092577-06S1, 3U54HG003067-12S2). Genome sequencing for “NHLBI TOPMed: San Antonio Family Heart Study” (phs001215.v4.p2) was performed at Illumina (3R01HL113323-03S1, R01HL113322). Genome sequencing for “NHLBI TOPMed: Women’s Health Initiative” (phs001237.v2.p1) was performed at Broad Institute Genomics Platform (HHSN268201500014C). Core support including centralized genomic read mapping and genotype calling, along with variant quality metrics and filtering, was provided by the TOPMed Informatics Research Center (3R01HL-117626-02S1; contract HHSN268201800002I). Core support including phenotype harmonization, data management, sample-identity QC, and general program coordination was provided by the TOPMed Data Coordinating Center (R01HL-120393; U01HL-120393; contract HHSN268201800001I). We gratefully acknowledge the studies and participants who provided biological samples and data for TOPMed. We extend our gratitude to Dr. Joyce van de Leempus for her invaluable critical reading of this manuscript.

Review history

The review history is available as Additional file 4.

Peer review information

Nora Franceschini and Tim Sands were the primary editors of this article and managed its editorial process and peer review in collaboration with the rest of the editorial team.

Funding

This research was made possible with generous partial support from various funding sources, including grants from the National Institutes of Health (NIH) (P30AG028747, R01AR46838, R01AR043351, P20GM109036, R01AR069055, U19AG055373, R01AG061917) and the Framingham Heart Study grant AR041398. Additionally, this research received partial support from the Women’s Health Initiative (WHI) program, which is funded by the National Heart, Lung, and Blood Institute, National Institutes of Health, U.S. Department of Health and Human Services, through contracts 75N92021D00001, 75N92021D00002, 75N92021D00003, 75N92021D00004, and 75N92021D00005. This research was also supported by contracts HHSN268201200036C, HHSN268200800007C, HHSN268201800001C, N01HC55222, N01HC85079, N01HC85080, N01HC85081, N01HC85082, N01HC85083, N01HC85086, and 75N92021D00006 and grants U01HL080295 and U01HL130114 from the National Heart, Lung, and Blood Institute (NHLBI), with additional contribution from the National Institute of Neurological Disorders and Stroke (NINDS). Additional support was provided by R01AG023629 from the National Institute on Aging (NIA). A full list of principal CHS investigators and institutions can be found at CHS-NHLBI.org.

Author information

Author notes

  1. Xiaoyu Zhang, Kuan-Jui Su, Bodhisattwa Banerjee and Ittai Eres co-first authors.

  2. Douglas P. Kiel, Hong-Wen Deng, Ching-Ti Liu and David Karasik co-last authors.

  3. Rebecca D. Jackson is deceased.

Authors and Affiliations

  1. Department of Biostatistics, Boston University School of Public Health, Boston, MA, 02118, USA

    Xiaoyu Zhang, Hanfei Xu & Ching-Ti Liu

  2. Center for Biomedical Informatics and Genomics, Tulane University, New Orleans, LA, 70112, USA

    Kuan-Jui Su, Zhe Luo, Chuan Qiu, Qing Tian, Hui Shen, Lan-Juan Zhao, Xiao Zhang & Hong-Wen Deng

  3. Department of Biochemistry, Larner College of Medicine, University of Vermont, Burlington, VT, 05405, USA

    Bodhisattwa Banerjee

  4. Amgen Inc, South San Francisco, CA, 94080, USA

    Ittai Eres

  5. Hinda and Arthur Marcus Institute for Aging Research, Hebrew SeniorLife, Boston, MA, 02131, USA

    Yi-Hsiang Hsu, Ming-Ju Tsai, Michelle Yau, Douglas P. Kiel & David Karasik

  6. David Geffen School of Medicine, University of California, Los Angeles, CA, 90024, USA

    Carolyn J. Crandall

  7. Azrieli Faculty of Medicine, Bar-Ilan University, 130010, Safed, Israel

    Rajashekar Donaka & David Karasik

  8. Department of Medicine, University of Maryland School of Medicine, Baltimore, MD, 21201, USA

    Zhe Han, Hanhan Liu, Braxton D. Mitchell & May E. Montasser

  9. Department of Medicine, The Ohio State University, Columbus, OH, 43210, USA

    Rebecca D. Jackson

  10. Department of Medicine, University of Washington, Seattle, WA, 98195, USA

    Kerri L. Wiggins

Authors
  1. Xiaoyu Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  2. Kuan-Jui Su
    View author publications

    You can also search for this author inPubMed Google Scholar

  3. Bodhisattwa Banerjee
    View author publications

    You can also search for this author inPubMed Google Scholar

  4. Ittai Eres
    View author publications

    You can also search for this author inPubMed Google Scholar

  5. Yi-Hsiang Hsu
    View author publications

    You can also search for this author inPubMed Google Scholar

  6. Carolyn J. Crandall
    View author publications

    You can also search for this author inPubMed Google Scholar

  7. Rajashekar Donaka
    View author publications

    You can also search for this author inPubMed Google Scholar

  8. Zhe Han
    View author publications

    You can also search for this author inPubMed Google Scholar

  9. Rebecca D. Jackson
    View author publications

    You can also search for this author inPubMed Google Scholar

  10. Hanhan Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  11. Zhe Luo
    View author publications

    You can also search for this author inPubMed Google Scholar

  12. Braxton D. Mitchell
    View author publications

    You can also search for this author inPubMed Google Scholar

  13. Chuan Qiu
    View author publications

    You can also search for this author inPubMed Google Scholar

  14. Qing Tian
    View author publications

    You can also search for this author inPubMed Google Scholar

  15. Hui Shen
    View author publications

    You can also search for this author inPubMed Google Scholar

  16. Ming-Ju Tsai
    View author publications

    You can also search for this author inPubMed Google Scholar

  17. Kerri L. Wiggins
    View author publications

    You can also search for this author inPubMed Google Scholar

  18. Hanfei Xu
    View author publications

    You can also search for this author inPubMed Google Scholar

  19. Michelle Yau
    View author publications

    You can also search for this author inPubMed Google Scholar

  20. Lan-Juan Zhao
    View author publications

    You can also search for this author inPubMed Google Scholar

  21. Xiao Zhang
    View author publications

    You can also search for this author inPubMed Google Scholar

  22. May E. Montasser
    View author publications

    You can also search for this author inPubMed Google Scholar

  23. Douglas P. Kiel
    View author publications

    You can also search for this author inPubMed Google Scholar

  24. Hong-Wen Deng
    View author publications

    You can also search for this author inPubMed Google Scholar

  25. Ching-Ti Liu
    View author publications

    You can also search for this author inPubMed Google Scholar

  26. David Karasik
    View author publications

    You can also search for this author inPubMed Google Scholar

Contributions

The author contributions for this manuscript are as follows: X.Z. and K.J.S. contributed to methodology, software, formal analysis, data curation, visualization, and original draft writing and review; B.B. participated in validation and original draft writing; I.E. contributed to software, validation, formal analysis, data curation, and original draft writing; Y.H.H. contributed to conceptualization, methodology, validation, resources, and formal analysis; C.C. contributed to conceptualization, resources, supervision, writing—review and editing, and project administration; R.D. contributed to validation and visualization; Z.H. participated in validation; R.D.J. contributed to conceptualization; H.L. contributed to validation and supervision; Z.L. participated in data curation; B.M. contributed to resources and supervision; C.Q., L.J.Z., and Q.T. participated in resources and data curation; H.S. contributed to resources, project administration, supervision, and funding acquisition; M.J.T. contributed to validation, formal analysis, and writing—review and editing; K.L.W. and H.X. contributed to methodology and data curation; M.Y. contributed to investigation, writing—original draft, and writing—review and editing; X.Z. contributed to methodology, validation, investigation, and resources; M.M. contributed to conceptualization and project administration; D.P.K. contributed to conceptualization, methodology, investigation, resources, writing—review and editing, supervision, and funding acquisition; H.W.D. contributed to conceptualization, methodology, resources, data curation, writing—review and editing, supervision, and funding acquisition; C.T.L. contributed to conceptualization, methodology, resources, writing—review and editing, and supervision; D.K. contributed to conceptualization, validation, original draft writing, writing—review and editing, and supervision. All authors have reviewed the manuscript and consented to its content.

Corresponding authors

Correspondence to Xiaoyu Zhang, Kuan-Jui Su, Ching-Ti Liu or David Karasik.

Ethics declarations

Ethics approval and consent to participate

All participants provided informed consent, and all studies were approved by the appropriate institutional review boards. Detailed ethics approval information, including the specific institutional review boards and approval statuses for the TOPMed and LOS cohorts, is provided in Additional file 1: SI 1 and SI 2, as well as Additional file 3: Table S2.

Consent for publication

All participants provided informed consent for publication.

Competing interests

The authors declare no competing interests. Dr. Ittai Eres is employed by Amgen Inc and confirms that this employment did not influence the design, conduct, or reporting of this study. All other authors have no financial or non-financial interests that could be perceived as influencing the research reported in this manuscript. Dr. Douglas P. Kiel has received grants to his institution from Radius Health, Amgen, and Solarea Bio. He serves on a Data and Safety Monitoring Committee for Agnovos and has served on Scientific Advisory Boards for Radius Health and Solarea Bio. He receives royalties for publication in UpToDate by Wolters Kluwer.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

13059_2025_3520_MOESM1_ESM.docx

Additional file 1: Supplemental information and methods. Detailed descriptions of additional methods, protocols, and analyses supporting the main study.

13059_2025_3520_MOESM2_ESM.docx

Additional file 2: Supplemental figures. Figures S1 to S8 providing additional visual data and supporting information referenced in the main text.

13059_2025_3520_MOESM3_ESM.xlsx

Additional file 3: Supplemental tables. Contains supplemental tables (S1–S19) providing detailed datasets, statistical results, and additional analyses supporting the findings of this study.

Additional file 4: Review history.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhang, X., Su, KJ., Banerjee, B. et al. Multi-ancestry whole genome sequencing analysis of lean body mass. Genome Biol 26, 106 (2025). https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13059-025-03520-x

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doiorg.publicaciones.saludcastillayleon.es/10.1186/s13059-025-03520-x

Genome Biology

ISSN: 1474-760X

Contact us