Targeted genotype analyses of GWAS-derived lean body mass and handgrip strength-associated single-nucleotide polymorphisms in elite master athletes

Abstract

Recent large genome-wide association studies (GWAS) have independently identified a set of genetic loci associated with lean body mass (LBM) and handgrip strength (HGS). Evaluation of these candidate single-nucleotide polymorphisms (SNPs) may be useful to investigate genetic traits of populations at higher or lower risk of muscle dysfunction. As such, we investigated associations between six SNPs linked to LBM or HGS in a population of elite master athletes (MA) and age-matched controls as a representative population of older individuals with variable maintenance of muscle mass and function. Genomic DNA was isolated from buffy coat samples of 96 individuals [consisting of 48 MA (71 ± 6 yr, age-graded performance 83 ± 9%) and 48 older controls (75 ± 6 yr)]. SNP validation and sample genotyping were conducted using the tetra-primer amplification refractory mutation system (ARMS). For the three SNPs analyzed that were previously associated with LBM (FTO, IRS1, and ADAMTSL3), multinomial logistic regression revealed a significant association of the ADAMTSL3 genotype with %LBM (P < 0.01). For the three HGS-linked SNPs, neither GBF1 nor GLIS1 showed any association with HGS, but for TGFA, multinomial logistic regression revealed a significant association of genotype with HGS (P < 0.05). For ADAMTSL3, there was an enrichment of the effect allele in the MA (P < 0.05, Fisher’s exact test). Collectively, of the six SNPs analyzed, ADAMTSL3 and TGFA showed significant associations with LBM and HGS, respectively. The functional relevance of the ADAMTSL3 SNP in body composition and of TGFA in strength may highlight a genetic component of the elite MA phenotype.

INTRODUCTION

Lean body mass (LBM) plays an important role in metabolic function, mobility, and healthy aging, where progressive declines in LBM and concurrent increases in lipid infiltration can have detrimental impacts related to functional impairments and disability (13, 14, 18, 37). Similarly, declines in muscle strength with aging are associated with impaired quality of life in older adults and increased risk of frailty and hospitalizations (2, 34). Reflecting this, handgrip strength (HGS) is a widely used marker of frailty and a strong predictor of morbidities and survival (21, 38). The heritability of muscle strength has been estimated to be between 30 and 65% (22, 35), with the heritability of the LBM phenotype estimated to be 52–60% (1, 12). To date, few studies have robustly identified candidate genes associated with LBM or HGS on a genome-wide level.

A recent study identified and replicated a set of five loci for total lean body mass (42). Three of these SNPs (near/in genes for IRS1, ADAMTSL3, and VCAN) were also successfully replicated for appendicular lean mass. Further analyses reported that for a subset of these SNPs, LBM-increasing alleles were associated with adverse metabolic profiles [such as the α-ketoglutarate-dependent dioxygenase (FTO) SNP rs9936385], whereas some were associated with metabolic protection [e.g., the rs2287926 SNP associated with the versican (VCAN) gene] (17). Similarly, a number of recent GWAS have reported multiple loci associated with HGS (23, 39). Analyses by Matteini et al. (23) identified one significant genome-wide association of an intergenic SNP located in a chromosomal region that regulates muscle repair and differentiation. In a study by Willems et al. (39), a number of loci out of the 16 SNPs identified were related to genes involved in muscle structure/function; (ACTG1), neurotrophic regulation (TGFA), and excitation-contraction coupling (SLC8A1). Others were identified with less understood roles in muscle function, such as Golgi brefeldin A resistant guanine nucleotide exchange factor 1 (GBF1), a guanine nucleotide exchange factor, and GLIS family zinc finger 1 (GLIS1), a Kruppel-like zinc finger protein that regulates transcription. Thus, further investigation into understanding the roles of these genes in the context of genetic variability of muscle strength is required. Despite the growing number of GWAS linking candidate genetic loci to skeletal muscle-related traits in humans, further validation/replication of these SNPs in independent cohorts has not previously been evaluated, whereas issues surrounding their reproducibility have also been highlighted (11).

Heritable phenotypical traits such as strength and lean mass are undoubtedly associated with physical performance and thus contribute to elite athletic status (6). Specifically, elite master athletes (MA; >65 yr) represent a population in which the effects of age may be addressed independently of the often accompanying disuse (19) and in many cases have displayed greater neuromuscular function than their age-matched inactive counterparts (24, 27, 29). However, there are little data available relating genotype to phenotype in these unique cohorts. In the current study, we first aimed to determine whether associations of SNPs linked to either LBM or HGS in previous GWAS analyses could be replicated in a smaller cohort comprised of a mixed population of elite master athletes (MA; both sprint and endurance) and age-matched nonathlete controls. Second, we aimed to compare allele/genotype frequencies between these two populations to gain further insight into the aforementioned differences in muscular strength and mass between older elite athletes and their age-matched controls. We hypothesized that the population of MA would demonstrates greater enrichments in SNPs associated with higher LBM and/or HGS. To perform targeted genotyping, we used tetra-primer amplification refractory mutation system (ARMS) PCR, which has been reported as a rapid, low-cost, and reliable method for SNP genotyping (26, 40).

MATERIALS AND METHODS

Participants and ethical approval.

The study was conducted in accordance with the Declaration of Helsinki, except for registration in a database. The study was approved by the University Research Ethics Committee and the National Research Ethics Service Committee Northwest (14/NW0275) and (15/NW/0426). All participants provided written, informed consent. Those in the control group (n = 48) were aged 75.3 ± 6.0 yr and were recruited from the local community. The master athletes (n = 48) were aged 70.6 ± 5.9 yr and were recruited from athletic clubs through an advertisement placed in a national athletics magazine and from two national master athletics competitions as part of the wider Vertical Impact of Bone Health in Elderly (VIBE) multiple cohort study (5, 28). All master athletes were actively competing in their respective disciplines, and all completed >5 h/wk of specific training at the time of testing. MAs were classified as sprinters (n = 12) if competing in events <800 m in distance or endurance athletes (n = 36) if competing in events ≥800 m in distance.

The age-graded performance (AGP) of a master athlete allows a comparison of current performance against world record performance in the same discipline, distance, and age group. Mean age-graded performance (AGP) was determined by taking the athlete’s highest-ranked performance within the last year and expressing it as a percentage of the world record for that age and distance. The mean AGP of this athletic cohort was 83.4 ± 8.6%. For example, a 21-min and 20-s 5,000 m for a 70-yr-old man gives an age-graded performance of 83%. All males were chosen for the current analysis to avoid influences of sex-specific hormones.

Dual-energy X-ray absorptiometry scans.

Standing height was measured to the nearest millimeter, and body mass was measured to the nearest 0.1 kg. Whole body, total hip, and lumbar spine dual-energy X-ray absorptiometry (DEXA; Lunar Prodigy Advanced, encore version 10.50.086; GE Healthcare, London, UK) scans were performed while the participant lay supine wearing a light cotton t-shirt to reduce measurement errors due to clothing absorption. Lean mass was taken from results of total body scans and regional analysis of legs and arms. All measurements were recorded after manual adjustment of the regions of interest. Repeat total body scans were performed in eight participants within 1 mo of the first scan. Using these repeat scans, the short-term error for our laboratory was 0.01% for whole body lean mass.

Muscle function.

The investigators provided verbal instructions and a physical demonstration of the muscle function tests. Participants were allowed one practice immediately before the actual assessed trials, which acted as a specific warm-up, and also confirmed that the instructions were understood. In all cases, the muscle function tests were completed between 10 AM and 3 PM.

Hand grip strength was measured using the Jamar dynamometer handle (Sammons Preston, Bolingbrook, IL), as previously described (10). The width of the dynamometer was adjusted for each participant separately. Participants were instructed to stand upright with the arm fully extended along the body, maintaining an ∼5-cm gap between the wrist and the hip or upper leg (so that the hand was not rested against the body). Participants were instructed to squeeze against the handle as hard possible for 3 s. Grip strength was measured three times and recorded in kilograms to the nearest 0.1 kg. For the purpose of this study, the best of three attempts was included in further analysis.

A Leonardo Jump Mechanography Platform (Leonardo Software version 4.2: Novotiec Medical GmbH, Pforzheim, Germany) was used to assess lower limb muscle power during a countermovement vertical jump, as described previously (10). Results for both absolute (W) and relative (W/kg) power were recorded. Briefly, a two-footed countermovement jump was performed starting with feet ∼30 cm apart (slightly narrower than shoulder width) and standing upright on the force plates. Force was sampled at 800 Hz. Participants flexed at the knees before extending as forcefully as possible to take off for the jump. Jumps were performed with a trained research assistant in close proximity to intervene in case of a trip or fall. Each participant repeated the jump sequence three times, with ∼60 s of rest between efforts. The jump with the highest value for power was used for statistical analysis.

Genomic DNA Extraction.

Genomic DNA was extracted from buffy coat samples (200 µL) using the QIAamp blood mini-DNA kit (Qiagen) according to the manufacturer’s instructions. Isolated DNA was quantified on the NanoDrop 2000 (Thermo Fisher Scientific).

SNP selection and primer design.

A set of SNPs were selected, chosen from SNPs previously linked with LBM (42) and HGS in humans (39). SNPs with very low/high-effect allele frequencies (EAFs) in the original GWAS studies (e.g., VCAN, KANSL1, and POLD3) were avoided due to expected difficulties in detecting them in relatively low sample sizes. Primer design was performed using the PRIMER1 program: http://primer1.soton.ac.uk/primer1.html, using the default primer design settings. SNPs that yielded primers with very high guanine-cytosine (GC) content were avoided due to anticipated difficulties during amplification, as well as primer sets with very distinct melting temperatures. A total of 15 SNPs were initially tested for validation; however, technical difficulties meant that a number could not be assessed with the tetra-primer ARMS PCR method, and the final set of six SNPs, three predicted to be associated with LBM and three with HGS, are presented in Table 1.

Table 1.

SNP and primer information for tetra-primer ARMS PCR

SNP	Closest Gene	Allele 1/2	EAF	Primer Sequences (5′–3′)	Annealing Temperature	Ratio of FO/RO/RI/FI
rs2943656	IRS1	A/G	0.38	FO: CTGAGAGCCTGCTCCTTACTCTTGTCTT; RO:	62°C	1:1:3:3
				CGGCATGTTGGAGAGTTACTCTACATGT; FI:
				TTCACCTAAAATTCTCCTCTAAAAACACAG; RI:
				CTCTCTCCATCACCATGGCTTCACCT
rs4842924	ADAMTSL3	T/C	0.52	FO: CAGTTGGAGTACTGAGAATGAGACAGGG;	61°C	2:1:6:2
				RO: AGTCTTAGGACTCAGACTTGCCATCACA; FI:
				GGAAAGGATAAGGATGTTGTGAGCGT; RI:
				GAATAGGCAATAGCTTCCTATGTGAGCG
rs9936385	FTO	T/C	0.61	FO: TGTGTGACCAGCCTCAATAGATTTTATTCA;	62°C	1:1:3:3
				RO: CCATCCTATCAAAAACAGCACTCTCACC; FI:
				TGCATATGAAGAGGGATTTTTTTGCATC; RI:
				TACTGGGAATATGCAGTGAACCACGA
rs958685	TGFA	A/C	0.52	FO: TCCACCCTTAGGAAAAAATGCTTCCTCT;	62°C	1:2:2:6
				RO: TCACATCTTTGTCATGGGACATAGTCCC; FI:
				TTTTTTCATCGGCAGTTTGCAGATACC; RI:
				AGGAGTATCCTTCTTCCACCCACGCT
rs2273555	GBF1	A/G	0.61	FO: CACAACCACAATGTTCGTAAACAGAATG;	61°C	1:1:3:3
				RO: TCTAAAAACTGGGAAAGGAAGCAATGTG;
				FI: TTTCCTAAGTCCTATTTACTGAAAACCAAG;
				RI: ACACTGAAGCCCCACCTAAGGAACGCT
rs4926611	GLIS1	C/T	0.64	FO: GCAGAGCTGGATTTTCAAGAGTCTACCT;	61°C	1:2:3:6
				RO: TTCATCCCTGCTTACCCACTAGAGGTAA; FI:
				TAGAGACACCTGCAACATCCAGCAAAAT; RI:
				CTGAGATTTGCTTTTTAAATTCAGCAGTG

ADAMTSL3, A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3; ARMS, amplification refractory mutation system; EAF, effect allele frequencies; FTO, α-ketoglutarate-dependent dioxygenase; GBF1, Golgi brefeldin A resistant guanine nucleotide exchange factor 1; GLIS1, GLIS family zinc finger 1; IRS1, insulin receptor substrate 1; SNP, single-nucleotide polymorphism; TGFA, transforming growth factor-α.

Tetra-primer ARMS PCR and gel electrophoresis.

Validation of SNP primers and genotyping was performed using the tetra-primer ARMS PCR technique (40). The sequences of primers used for the genotyping of the selected SNPs are shown in Table 1. SNP primers were initially validated and optimized using previously set-out guidelines (26). Initially, amplification was performed using the outer primers only, using a gradient-annealing temperature PCR to determine the optimal annealing temperature for each primer set. Subsequent validation involved incorporating the inner primers in varying amounts to produce detectable bands for each allele-specific amplicon via agarose gel electrophoresis (see below). PCR reactions with a final volume of 18 µL, including 30 ng of genomic DNA, SYBR Select Master Mix (Applied Biosystems), and primers in ratios, are shown in Table 1. Amplification was performed using a Viia 7 real-time PCR machine (Applied Biosystems), using the following cycling conditions: 1 cycle of initial denaturation at 95°C, 2 min; 35 cycles of denaturation at 95°C for 30 s, annealing at 61–62°C (see Table 1 for SNP-specific annealing temperature) for 45s and extension at 72°C for 45s, with a final extension for 5 min at 72°C on standard cycling conditions. PCR products were mixed with 4 µL of gel-loading buffer (Sigma-Aldrich) and 10 µL was electrophoresed on 3% (wt/vol) agarose gels for 120 min at 80 V.

Statistical analyses.

Multinomial logistic regression was performed in R (version 3.6.1) using the nnet package (16) to examine associations between GBF1, GLIS1, and TGFA genotypes and maximal grip strength and to examine associations between IRS1, FTO, and ADAMTSL3 and total lean mass, appendicular lean mass, and percentage body lean mass. Strength of associations were assessed by P values calculated from z values provided from the regression model coefficients and standard errors for each predictor variable. Fisher’s exact test was used for comparison of allele distributions and genotype distributions between MA and control groups, whereas one-way ANOVA was used for multigroup comparisons, with Tukey’s test to correct for multiple comparisons. Comparisons between two groups were made using unpaired t tests. P < 0.05 was taken to be statistically significant. Data were analyzed using GraphPad Prism software version 7.0.

RESULTS

Associations between genotype and functional parameters in MA and nonathletes.

We first aimed to identify any associations of the selected SNP genotypes with lean mass or HGS in a mixed population of older MA and nonathletes (i.e., irrespective of groupings). Within this cohort, total LBM ranged from 36.6 to 69.4 kg, whereas HGS ranged from 20.6 to 54.7 kg. In relation to the SNPs previously linked to HGS [GBF1 (rs2273555; effect allele A), GLIS1 (rs4926611; effect allele C), and transforming growth factor-α (TGFA; rs958685; effect allele A)], there was no significant association of either GBF1 or GLIS1 genotype with HGS (Fig. 1), but with TGFA, there was a significant association between HGS and genotype (mean difference between AA and CC 6.32; 95% CI 0.43-12.1; P < 0.05, Fig. 1A), with the AA genotype (A being the effect allele) having higher HGS. In relation to SNPs that were previously associated with LBM, multinomial logistic regression showed no significant association of total lean mass, %LBM, or appendicular lean mass with insulin receptor substrate 1 (IRS1; rs2943656; effect allele A), FTO (rs9936385; effect allele T), or A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 (ADAMTSL3; rs4842924; effect allele T) genotypes (Figs. 2, 3, and 4). For ADAMTSL3, however, there was a significant association with %LBM (mean difference between TT and CC 5.36, 95% CI 1.38–9.34, P < 0.01; Fig. 4A), where the TT genotype was associated with higher %LBM. Because LBM and HGS are biologically closely related, we also determined whether any of the LBM-associated SNPs were linked to HGS and vice-versa. However, none of the HGS-associated SNPs were significantly associated with LBM (TGFA; β = −4.88, P = 0.305 GLIS1; β = −18.64, P = 0.641, GBF1; β = 2.433, P = 0.354), and none of the LBM-associated SNPs were associated with HGS (FTO: β = −1.716, P = 0.354; IRS1: β = −3.242, P = 0.059; ADAMTSL3: β = −1.432, P = 0.378). There were also no genotype associations of any of the SNPs measured with muscle power measurements [maximum power relative to body weight (P_{max rel})] (TGFA: β = −0.079, P = 0.714; GLIS1: β = −0.002, P = 0.911; GBF1: β = −0.006, P = 0.891; FTO: β = −0.021, P = 0.358; IRS1: β = −0.002, P = 0.905; ADAMTSL3: β = 0.015, P = 0.423).

Fig. 1.

Genotype vs. grip strength for transforming growth factor-α (TGFA; rs958685, effect allele = A), GLIS family zinc finger 1 (GLIS1; rs4926611, effect allele = C), and Golgi brefeldin A resistant guanine nucleotide exchange factor 1 (GBF1; rs2273555, effect allele = A) in a mixed population of older elite athletes (sprint and endurance) and nonathletes. Grip strength according to genotype for TGFA (A), GLIS1 (B), and GBF1 (C) (irrespective of groupings). *P < 0.05 vs. AA (multinomial logistic regression analysis). Ctl, control; MA, master athletes.

Fig. 2.

Genotype vs. total lean mass for A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 (ADAMTSL3; rs4842924, effect allele = T), insulin receptor substrate 1 (IRS1; rs2943656, effect allele = A), and α-ketoglutarate-dependent dioxygenase (FTO; rs9936385, effect allele = T) in a mixed population of older elite athletes (sprint and endurance) and nonathletes. Total lean mass according to genotype for ADAMTSL3 (A), IRS1 (B), and FTO (C) (irrespective of groupings). Ctl, control; MA, master athletes.

Fig. 3.

Genotype vs. appendicular lean mass for A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 (ADAMTSL3; rs4842924, effect allele = T), insulin receptor substrate 1 (IRS1; rs2943656, effect allele = A), and α-ketoglutarate-dependent dioxygenase (FTO; rs9936385, effect allele = T) in a mixed population of older elite athletes (sprint and endurance) and nonathletes. Appendicular lean mass according to genotype for ADAMTSL3 (A), IRS1 (B), and FTO (C) (irrespective of groupings). Ctl, control; MA, master athletes.

Fig. 4.

Genotype vs. %lean mass for A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 (ADAMTSL3; rs4842924; effect allele = T), insulin receptor substrate 1 (IRS1; rs2943656, effect allele = A), and α-ketoglutarate-dependent dioxygenase (FTO; rs9936385, effect allele = T) in a mixed population of older elite athletes (sprint and endurance) and nonathletes. %Lean mass according to genotype for ADAMTSL3 (A), IRS1 (B), and FTO (C) (irrespective of groupings). **P < 0.01 vs. CC (multinomial logistic regression analysis). Ctl, control; MA, master athletes.

Allele frequencies in individuals grouped according to the highest and lowest quartile for % LBM or HGS.

Following on from this, we aimed to determine whether there were any differences in allele frequencies in individuals that had been grouped according to the highest and lowest quartiles for %LBM or HGS. Comparing the upper and lower quartiles for %LBM (irrespective of groupings), there was no difference in allele frequency for the IRS1 or FTO SNPs (Table 2). For ADAMTSL3, comparing the upper and lower quartiles for %LBM (irrespective of groupings), there was an enrichment in the effect allele in the upper quartile for %LBM (P < 0.05; Fisher’s exact test) (Table 2). For TGFA, comparing the upper and lower quartiles for HGS (irrespective of groupings), there was an enrichment in the effect allele in the upper quartile for HGS (P < 0.05; Fisher’s exact test) (Table 2). There were no significant differences in either GBF1 or GLIS1 alleles between the upper and lower quartiles for HGS (Table 2).

Table 2.

Allele frequencies of selected SNPs previously associated with LBM or grip strength in individuals grouped according to the highest and lowest quartile for %LBM or HGS

SNP	Closest Gene	Allele 1/2	Lowest Quartile for %LBM		Highest Quartile for %LBM		SNP	Closest Gene	Allele 1/2	Lowest Quartile for HGS		Highest Quartile for HGS
			Allele 1	Allele 2	Allele 1	Allele 2				Allele 1	Allele 2	Allele 1	Allele 2
rs2943656	IRS1	A/G	28	20	28	20	rs958685	TGFA	A/C	19	29	30	18*
rs4842924	ADAMTSL3	T/C	15	33	27	21*	rs2273555	GBF1	A/G	22	26	21	27
rs9936385	FTO	T/C	34	14	29	19	rs4926611	GLIS1	C/T	28	20	34	14

ADAMTSL3, A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3; FTO, α-ketoglutarate-dependent dioxygenase; IRS1, insulin receptor substrate 1; GBF1, Golgi brefeldin A resistant guanine nucleotide exchange factor 1; GLIS1, GLIS family zinc finger 1; HGS, handgrip strength; %LBM, %lean body mass; SNP, single-nucleotide polymorphism; TGFA, transforming growth factor-α. Frequencies of alleles for 3 lean mass-associated SNPs (IRS-1, FTO, and ADAMTSL3) between the highest and lowest quartile for LBM (irrespective of groupings) and 3 grip strength-associated SNPs (TGFA, GLIS1, and GBF1) between the highest and lowest quartile for HGS (irrespective of groupings).

^*P < 0.05 vs. lowest quartile (Fisher’s exact test).

Allele/genotype distributions for LBM or HGS-associated SNPs in MA versus nonathletes.

In subsequent analyses, we sought to compare allele/genotype distributions for the LBM and HGS-associated SNPs between the elite MA and older nonathlete groups, first comparing participant muscle-related characteristics between MA and control groups. Because multiple group analyses were limited by the relatively low number of available samples from participants in the sprint category (n = 12), sprint and endurance MA were grouped for the majority of our analyses. Whereas total lean mass and appendicular lean mass (ALM) was not different across groups (Fig. 5, A and B), LBM as a percentage of total body weight (%LBM) was significantly lower in controls than MA (P < 0.001 by unpaired t test; Fig. 5C). Likewise, percentage fat mass was significantly higher (P < 0.001 by unpaired t test) in controls than MA (Fig. 5D). HGS and P_{max rel} were no different between MA and controls (Fig. 5, E and F).

Fig. 5.

Phenotype characteristics of older elite athlete (sprint and endurance) and nonathlete (control) populations. Total lean mass (A), appendicular lean mass (ALM; B), %lean mass (C), %fat mass (D), grip strength (E), and maximum power relative to body weight (P_{max rel}; F) in master athletes (MA) and nonathlete controls. ***P < 0.001 (unpaired t test).

Genotype distributions for three SNPs that were previously associated with LBM (IRS1, FTO, and ADAMTSL3) and three SNPs that were previously associated with HGS (TGFA, GBF1, and GLIS1) were analyzed in the 48 MA and 48 older controls. For the SNP associated with the ADAMTLS3 gene, genotype distributions were significantly different between MA and controls (P < 0.05, Fisher’s exact test; Fig. 6). For the SNPs associated with IRS1, FTO, TGFA, GLIS1, and GBF1, there was no difference in genotype frequencies between MA and control groups (Fig. 6). Although analyses focused on the master athletes as a group, compared with nonathlete controls, we also assessed allele distributions for the six SNPs between sprint and endurance MA relative to controls (Table 3). Similar to the genotype distributions between MA and controls, allele distributions for the SNPs associated with IRS1, FTO, TGFA, GLIS1, and GBF1 were not significantly different between groups, whereas for ADAMTSL3, there was an enrichment in the effect allele for both sprint and endurance athletes relative to nonathlete controls [P < 0.05 vs. control (Fisher’s exact test); Table 3].

Fig. 6.

Genotype distributions of selected single-nucleotide polymorphisms (SNPs) previously associated with lean body mass or grip strength in master athlete (MA) and nonathlete [control (Ctrl)] populations. Balloon plot displaying frequencies of genotypes for three lean mass-associated SNPs [insulin receptor substrate 1 (IRS1), α-ketoglutarate-dependent dioxygenase (FTO), and A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3 (ADAMTSL3)] and 3 grip strength-associated SNPs [transforming growth factor-α (TGFA), GLIS family zinc finger 1 (GLIS1), and Golgi brefeldin A resistant guanine nucleotide exchange factor 1 (GBF1)] between elite older athletes and nonathlete controls. *P < 0.05 (Fisher’s exact test).

Table 3.

Allele frequencies of selected SNPs previously associated with lean body mass or grip strength in elite athletes (sprint and endurance) vs. nonathlete controls

SNP	Closest Gene	Allele 1/2	Control (n = 48)		Sprint (n = 12)		Endurance (n = 36)
			Allele 1	Allele 2	Allele 1	Allele 2	Allele 1	Allele 2
rs2943656	IRS1	A/G	57 (59%)	39 (41%)	13 (54%)	11 (46%)	42 (58%)	30 (42%)
rs4842924	ADAMTSL3	T/C	38 (40%)	58 (60%)	15 (62%)*	9 (38%)*	42 (58%)*	30 (42%)*
rs9936385	FTO	T/C	60 (62%)	36 (38%)	10 (42%)	14 (58%)	51 (71%)	21 (29%)
rs958685	TGFA	A/C	51 (53%)	45 (47%)	13 (54%)	11 (46%)	39 (54%)	33 (46%)
rs2273555	GBF1	A/G	44 (46%)	52 (54%)	7 (29%)	17 (71%)	31 (43%)	41 (57%)
rs4926611	GLIS1	C/T	68 (71%)	28 (29%)	15 (62%)	9 (38%)	46 (64%)	26 (36%)

ADAMTSL3, A disintegrin-like and metalloprotease domain with thrombospondin type I motifs-like 3; FTO, α-ketoglutarate-dependent dioxygenase; IRS1, insulin receptor substrate 1; GBF1, Golgi brefeldin A resistant guanine nucleotide exchange factor 1; GLIS1, GLIS family zinc finger 1; SNP, single-nucleotide polymorphism; TGFA, transforming growth factor-α. Frequencies of alleles for 3 lean mass-associated SNPs (IRS1, FTO, and ADAMTSL3) and 3 grip strength-associated SNPs (TGFA, GLIS1, and GBF1) between nonathlete controls and elite athletes (split into sprint and endurance types). n = number of subjects.

^*P < 0.05 vs. control (Fisher’s exact test).

DISCUSSION

Although LBM and HGS represent two highly heritable traits in humans (1, 22, 35), only recently have studies begun to explore the specific genes that contribute to the underlying interindividual variability in skeletal muscle traits such as these (30, 39, 42). Evaluation of these candidate SNPs could prove useful in investigating underlying genetic traits of individuals at variable risk of muscle dysfunction. In the present study, our aim was to determine whether SNPs linked to either LBM or HGS in previous GWAS analyses could be replicated in a smaller cohort composed of elite MA and age-matched controls. We also aimed to determine whether genotype/allele distributions for these SNPs were different between elite MA in comparison with age-matched nonexercising controls as a representative population of older individuals with greater maintenance of muscle mass and function. By comparing allele/genotype frequencies between these two populations using the tetra-primer ARMS technique, we aimed to gain greater insights into the underlying genetic component of the MA muscle phenotype.

We chose to use the tetra-primer ARMS technique as a rapid approach to SNP genotyping, as it provides a cost-effective and accurate methodology (40), but alternative methods are available. The restriction fragment length polymorphism (RFLP) typing method involves restriction endonuclease digestion of PCR products to discriminate between alleles (25), whereas microarray approaches (32) and matrix-assisted laser desorption/ionisation time-of-flight (MALDI-TOF) mass spectrometry (8) allow high-throughput genotyping. We found the tetra-primer ARMS technique robust but requiring substantial optimisation, and some primer sets for SNPs could not be validated, potentially because of the SNPs loci, i.e., in a highly GC-rich region, giving rise to difficulties due to incomplete denaturation of DNA and less than optimal primer annealing (26).

We began by investigating associations between genotype and functional parameters in older MA and nonathletes as a collective cohort. In terms of their predicted associations with either LBM or HGS, out of the six SNPs analyzed, four failed to show any significant association with LBM or HGS (even when those individuals with the highest and lowest quartiles for %LBM or HGS were analyzed). In contrast, the SNP associated with TGFA showed significant associations with HGS, whereas the SNP linked with ADAMTSL3 was associated with LBM (independent of exercise discipline), as predicted by the original GWAS. These findings provide further support to the previous data indicating the potential importance of the TGFA SNP in muscle strength and of ADAMTSL3 in body composition. Interestingly, we found that none of the HGS-associated SNPs were associated with LBM and vice-versa, nor were there any significant associations with P_{max rel}. The reason for this lack of overlap is not clear and requires further investigation of the potential roles of these genes in muscle function. For the SNP associated with TGFA, there was an association between HGS and genotype, with the AA genotype (A being the effect allele promoting increased HGS), having a significantly higher HGS. The consequence of the polymorphism with rs958685 is an intron variant. The potential functional relevance of the TGFA in muscular strength remains to be evaluated, but other intronic SNPs have been shown to be associated with functional elements, including intron splicing enhancers/silencers that regulate alternative splicing events as well as other transcriptional regulatory elements (4). The TGFA gene encodes a growth factor that plays a key role in cellular proliferation, differentiation, and development (33). TGFα also plays a neurotrophic role and promotes neuronal survival during acute injury of motor neurons (15, 20).

A further important finding was that for rs4842924, the SNP related to the ADAMTSL3 gene, the TT genotype was associated with higher %LBM among all volunteers. Initial analyses aimed to replicate the original GWAS (42), which identified SNPs associated with total LBM, with subsequent analysis demonstrating higher associations when adjusting for total fat mass (17). We found instead that for ADAMTSL3 (and other SNPs), there was no association with LBM in either unadjusted or after adjustment for fat mass or for height. We also found no associations of any of the SNPs to appendicular lean mass. There was, however, a significant association of the ADAMTSL3 genotype with LBM as a percentage of whole body mass, demonstrating that it may have importance in terms of body composition. As with TGFA, the consequence of the ADAMTSL3 SNP is an intron variant, and the functional effect (if any) on gene expression is not currently known. Little is understood about the biological functions of ADAMTSL3, but it is a glycoprotein that is related to the ADAMTS family of metalloproteases that may have functions in extracellular matrix regulation (9). The ADAMTSL3 gene has also consistently been linked to height (36) in genome-wide association analyses. Further in vitro experiments will be required to understand the mechanisms underlying ADAMTSL3 gene variants in muscle physiology and relation to LBM in vivo.

We next investigated allele/genotype distributions for LBM or HGS-associated SNPs in MA versus nonathletes. Elite MA represent a unique population of individuals that in general display greater maintenance of neuromuscular function than age-matched inactive populations (24), and although undoubtedly environmental factors play a large role in the MA phenotype (7), there are little conclusive data available related to any underlying genetic components. Whether the high-functioning characteristics of master athletes are more influenced by heritable factors regulating muscle composition/performance or whether the environmental component (i.e., continued high levels of training over the years) are more important for the master athlete phenotype remains to be fully understood. For the present group of individuals studied, whereas total LBM and ALM were not different between MA and controls, %LBM was significantly higher in the MA population. Whereas HGS and P_{max rel} were not different between MA and nonathlete controls, this is likely due to the fact that the majority of the cohort were endurance athletes, which is in line with previous observations with regard to strength differences in endurance versus power MA (24). Although HGS does not always correlate with strength of other functionally important muscle groups, such as the quadriceps (41), it is a useful predictor of a number of health outcomes in middle to older age (3), including all cause mortality (31). In the present study, of the six SNPs measured, five were not different between MA and control; however, for ADAMTSL3, there was an enrichment of the effect allele (T) in the group of MA. Further work investigating these candidate SNPs, and the mechanisms by which they may influence muscle function, could prove useful in understanding the genetic basis of populations with increased/decreased susceptibility of muscle dysfunction (such as frailty and sarcopenia).

Perspectives and Significance

Although there are difficulties associated with studying a cohort, such as that of the MA in terms of gaining sufficient sample numbers, clearly larger MA sample sizes will be needed to explore MA on a genome-wide basis or in a targeted fashion. Indeed, the lack of individuals with the GG genotype for IRS1 in the present study is also a limitation in the context of the relatively small sample size of this study. There is also the potential that the lack of replication for some of the SNPs analyzed in the present study was partly due to the the elite athletes having a different phenotype from those of the general population (as used in the original GWAS analyses). Additionally, effect sizes in the original analyses would be viewed as being small, with standardized β of −0.12 to −0.14 for LBM and 0.13–0.16 for HGS. More work is required to determine the biological significance of these SNPs in LBM and/or muscular strength across different populations of individuals. Nonetheless, in a targeted fashion, we demonstrate that a SNP related to the ADAMTSL3 gene was enriched in elite MA and had significant associations with %LBM. We also confirmed data from previous GWAS of an association of the TGFA SNP with HGS. Future work elucidating the mechanisms by which these gene variants influence muscle mass and function is required to facilitate our understanding of the genetic basis of not only the MA phenotype but also the genetic basis underlying a range of conditions such as frailty and sarcopenia.

GRANTS

This work was supported by funding from the UK Medical Research Council (MRC) as part of the Life Long health and Wellbeing initiative (MR/K025252/1). This work was also supported by the UK MRC (grant no. MR/P021220/1) as part of the MRC-ARUK Centre for Musculoskeletal Aging Research awarded to the Universities of Nottingham and Birmingham and supported by the National Institute of Health Research (NIHR) Nottingham Biomedical Research Centre. The views expressed are those of the authors and not necessarily those of the National Health Service (NHS), the NIHR, or the Department of Health and Social Care.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

J.P., B.E.P., K.S., J.S.M., M.P., and P.J.A. conceived and designed research; H.C. and D.M. performed experiments; H.C., D.M., D.J.W., and M.P. analyzed data; H.C., D.J.W., M.P., and P.J.A. interpreted results of experiments; H.C. and M.P. prepared figures; H.C. and P.J.A. drafted manuscript; H.C., J.P., D.J.W., M.P., and P.J.A. edited and revised manuscript; H.C., J.P., D.M., B.E.P., D.J.W., K.S., J.S.M., M.P., and P.J.A. approved final version of manuscript.