Journal Logo

SPECIAL COMMUNICATIONS: Contrasting Perspectives

DNA Sequence Variations Contribute to Variability in Fitness and Trainability

BOUCHARD, CLAUDE

Author Information
Medicine & Science in Sports & Exercise: August 2019 - Volume 51 - Issue 8 - p 1781-1785
doi: 10.1249/MSS.0000000000001976
  • Free

Genetic differences contribute to human variability in exercise-related traits. In this commentary, the focus is on endurance exercise performance and cardiorespiratory fitness (CRF). The evidence pertaining to other relevant traits (e.g., muscle strength) will not be considered. Three main topics are addressed to provide a brief summary of our current knowledge base.

  • 1. There is a genetic component to cardiorespiratory fitness

Studies in rodents have generated strong data supporting the presence of genetic effects on variability in maximal exercise capacity in sedentary animals. For instance, Lightfoot et al. (1) compared 10 inbred strains of female mice for maximal running performance and found a fourfold difference between the low- and high-performance strains, with a broad sense heritability (the total genetic contribution to a trait) of 0.73 or 73%. The laboratory of Garland (2) showed that mice could be bred for high endurance capacity. In a comprehensive study, Massett (3) compared 24 inbred strains and observed that, in sedentary animals, maximal running capacity ranged from 21 to 42 min between the two extreme strains.

The strongest case for a role of genetic differences in maximal exercise capacity in sedentary animals comes from the selection experiment for low-capacity runners and high-capacity runners over multiple generations starting from an outbred founder population of rats by Koch and Britton (4). After 28 generations of selection, sedentary low-capacity runners and high-capacity runners differed by eightfold in maximal endurance running performance and narrow sense heritability (the genetic variance responsible for the resemblance from one generation to the next, that is, the heritable variance) was of the order of 40%.

The evidence for human studies is just as compelling as that on animal models. Maximal CRF data have been obtained in several twin studies (5–7), and all twin data were recently meta-analyzed by de Geus and collaborators (8). They found a heritability of about 0.70. In the HEalth, RIsk factors, exercise Training And GEnetics (HERITAGE) Family Study, the heritability of CRF in the sedentary state reached about 0.5 in families of European descent (9). Submaximal indicators of CRF have also been examined in twin and family studies in sedentary adults. Coefficients of heritability (× 100) ranging from 48% to 74% were observed in HERITAGE for several indicators of submaximal CRF (10) and 54% to 58% for ventilatory threshold in the sedentary state (11). When submaximal power output (PWC 150) was measured in siblings by descent or by adoption, heritability levels ranged from 0.30 to 0.48 (12). Notably, there is a significant genetic component to stroke volume and cardiac output during submaximal exercise, with heritability coefficients ranging from 0.41 to 0.46 (13).

There is a significant genetic component to variability in CRF in sedentary individuals with a moderate heritability level. The consistency of findings across a wide range of animal and human studies is remarkable.

  • 2. There is a genetic component to the response of cardiorespiratory fitness to an exercise training dose

Studies on the genetics of CRF trainability have each focused on a single dose of exercise. There are no data on the genetics of the response of CRF to incremental doses of exercise or to near maximal or maximal doses. The hypothesis that there is a genetic component to variability in CRF trainability has been confirmed by four complementary lines of evidence.

Massett and collaborators (14) have used intercross breeding between two mouse strains (NZW and 129S1) differing markedly in performance gains following an exercise program and showed that there was a genetic component to trainability. They uncovered loci harboring sequences impacting training response. The same laboratory investigated the response to 4 wk of treadmill running in 148 mice from 24 inbred mouse strains. Running performance improved by 3.4 min (SD, 3.6) on average, with seven strains showing no evidence of performance improvement and five strains improving running time by 5 min and more (3). Heritability reached 0.58 for the change in distance run.

The evidence from rat studies performed by Koch and Britton (4) is even more powerful. Selective breeding was used to test the hypothesis of a genetic component to variability in trainability. Maximal running distance was measured before and after an 8-wk exercise program on the treadmill. In the first generation, genetically heterogeneous rats (N:NIH) exhibited a 140-m gain in running capacity in response to training, with a range from −339 m to +627 m. After 15 generations of selection, “low response trainers” on average experienced a decline of 65 m in maximal running distance with training while “high response trainers” improved by 223 m, with females improving more than males (15). Thus, there was sufficient additive genetic variance in the rat genome to allow response to selection in both positive and negative directions. Moreover, high-response trainer animals responded to high intensity training with a 40% increase in V˙O2max whereas low-response trainer animals failed to improve (16).

A limited number of human studies have dealt with the role of genetic variation in CRF trainability in response to standardized exercise programs. Four of these exercise experiments were performed with pairs of monozygotic (MZ) twins. A first study was undertaken with 10 pairs of MZ twins exposed to an endurance program for 20 wk (17). The second study attempted to replicate the findings of the first experiment based on six pairs of MZ twins exposed to 15 wk of an endurance exercise program (18). The third study focused on a combination of continuous and high-intensity intermittent training in 14 pairs of MZ twins exposed to a 15-wk program (19,20). The fourth study was based on seven pairs of MZ twins who completed a 17-d baseline assessment period during which the daily caloric intake needed to remain weight stable while being sedentary was quantified. This level of individual caloric and macronutrient intake was then clamped for 93 d during which subjects exercised 1 h in the morning and 1 h in the afternoon at low intensity under standardized conditions on cycle ergometers, with a day of rest every 10 d (21). Despite the fact that the exercise doses were different across the four studies, the findings were highly concordant with a substantial degree of within MZ pair resemblance in CRF training response, as revealed by intraclass coefficients ranging from 0.44 to 0.77. These findings were supported and amplified by the results of the HERITAGE Family Study, which showed that the heritability of CRF response to a 20-wk exercise program reached 47% (22).

In summary, the evidence from both animal and human studies is highly concordant in supporting the hypothesis that there is a strong genetic component to variability in CRF trainability with exposure to a given dose of exercise.

  • 3. Why is it so difficult to define the genetic contribution at the molecular level?

It is a challenge to identify the DNA sequence variants accounting for the genetic component of the variance in complex human traits, primarily because the underlying biology of such traits is more complex than once anticipated. One of the lessons of the decade-long research on genomewide association studies (GWAS) of diseases and other complex phenotypes is that the full genotype accounting for the heritability of such traits is composed of very large numbers of DNA variants (23,24).

Despite intensive explorations of the genome in large populations, there is no complex trait as of yet for which we have a complete compendium of all single-nucleotide polymorphisms (SNP) and other DNA polymorphic motifs contributing to the variance. However, we have learned a great deal on the genetic architecture of complex traits as a result of a wave of high-quality GWAS (23). The GWAS-based advances in combination with those from the ENCODE Project (25), the HapMap and 1000 Genomes Project (26), Genotype-Tissue Expression Project (27,28), transgenic and knockout mouse models, computational biology and bioinformatics, and other technologies have given us an unprecedented opportunity to undertake the study of the genetic architecture of complex human traits.

Four biological features represent major hurdles in the effort to dissect the genetic heritability of complex traits, including CRF in the sedentary state and its trainability, in terms of DNA sequence variants and molecular mechanisms. First, it is evident that correlating genotype with phenotype is a more complicated undertaking than once thought. This is even more of a challenge for complex traits as exemplified by experiments performed in yeast (29).

Second, the regulation of gene expression is coordinated but characterized by widely distributed systems of enhancers and repressors among other regulatory sequences. This is one of the cardinal lessons from ENCODE (25). For instance, if one assumes that there are about 21,000 protein coding genes in the human genome, there are about 250 protein binding sites per gene, about 20 enhancer sequences per gene, and three promoter regions per gene, to name but a few important features. Such a high redundancy level provides a mechanism for the smooth regulation of gene expression (30). Moreover, large numbers of SNP occur in CpG islands (multiple dinucleotide sequences of cytosine followed by guanine) and other regulatory motifs impacting gene expression.

Third, redundancy is ubiquitous in metabolic and other pathways and networks. For instance, two or more genes can participate in pathways converging on the same function; two enzymes can be functionally equivalent; there are parallel pathways in substrate metabolism; there are examples of more than one transport mechanism carrying the same molecules.

Fourth, GWAS have established that complex traits are not modulated by a few genes with large effect sizes. Quite the contrary, complex multifactorial traits are influenced by polygenic systems defined by hundreds or thousands of loci (31). These loci are typically characterized by relatively common alleles with small to very small effect sizes. Most of these variants are not located in coding regions and a large fraction of them sits in the noncoding regulatory genome (28). In addition, complex phenotypes are also impacted by low frequency and rare alleles, some of which have larger effect sizes (32,33).

TAKE HOME MESSAGE

The fact that cardiorespiratory fitness in sedentary individuals and its trainability are extraordinarily complex traits cannot be taken as evidence for the absence of a genetic component.

Mouse, rodent, and human experimental genetic studies are concordant: there is a genetic component to cardiorespiratory fitness and its trainability. We now recognize that despite the spectacular advances in our understanding of the human genome, it is a humongous task to define the genomic basis of complex human traits including exercise-related phenotypes.

Pursuing in-depth genetic and molecular studies of exercise traits has the potential to illuminate their underlying biology and how they are regulated at the pathway, system, and whole organism level. The priority should be to expand the search for relevant alleles and to interrogate in depth the data to uncover overarching principles, higher-order drivers, and actionable targets amid the complex molecular and cellular architecture of exercise traits. And yes, we will have to indulge extensively in the sin of “reductionism” for many years if we want this research enterprise to be successful.

RESPONSE TO JOYNER

I agree with my friend, Dr. Joyner, that there is, at present, limited evidence on specific genetic mechanisms contributing to human variation in CRF and its trainability. However, the limited evidence we have is extremely powerful in the sense that it provides unequivocal support for the hypothesis that human genetic differences are playing a substantial role in both phenotypes. His view seems to be that we will recognize the evidence as undeniable only when all genomic variants impacting human variability have been identified and their paths and mechanisms of action have been established. Then he adds another hurdle: these genomic variants have to carry large effect size; otherwise they cannot be taken seriously! This is clearly an extreme position. A more pragmatic view based on animal models and human studies reported to-date would conclude that there is a substantial genetic component to CRF and its trainability even though the details of the genomic architecture and the molecular mechanisms entrained by genomic variants remain unresolved.

However, some scientists and clinicians are impatient and disappointed because all the answers are not in yet. Because our current understanding is limited and rudimentary when it comes to the genomic details, Dr. Joyner concludes that there is no need to invoke genomic differences to account for human variability in CRF and its trainability (34). We should be reminded that the absence of evidence should never be taken as evidence for absence!

Requirements are failing the smell test

Dr. Joyner posits that there are five criteria, defined as being “required,” before accepting a role of genetic variation in CRF. He seems to suggest that there is no need for this genomic, genetic, and molecular stuff for a proper understanding of the biology of CRF and its trainability based on these five criteria. In contrast, I conclude that from a genetic perspective, none of them is truly required.

  • 1. “Identification of potentially causal variants”

Genetic research has the potential to identify DNA variants that are causative of phenotypic differences. There is no objective reason why variants of all effect sizes cannot be causally associated with a trait of interest. A variant with a small effect size contributing to a predisposition to high CRF trainability is part of the causal path from DNA to the trainability phenotype just as is a DNA variant with a large effect size.

  • 2. “Linkage of these variants to deterministic physiological mechanisms”

Yes, for genetic variants associated with the commonly recognized physiological determinants of CRF in the sedentary state, but it is a different story for CRF trainability. In the case of trainability, DNA variants influence a number of pathways (cellular growth, autophagy and apoptosis, angiogenesis, etc.) whose connections with the physiological determinants of CRF may appear weak or remote but are in fact essential for mounting an adaptive response to exercise training, which ultimately will translate into an augmented capacity for oxygen delivery and utilization by working muscles.

  • 3. “Explanation by these variants of more than a small fraction of the phenotypic trait of interest”

This cannot be a valid criterion because it has no basis in reality. We need to evaluate the evidence based on biological reality. Even though there are alleles with large effect sizes impacting complex traits, most of the variance in these traits, including exercise-related traits, are the results of hundreds and thousands of alleles with small effect sizes. The small effect size of an allele is not a reason to negate its importance. The main reason genomic variants could not be associated with CRF phenotypes in two studies cited by Dr. Joyner (35,36) is that they were both statistically underpowered to reveal variants with small effect sizes, an endemic problem in exercise genomic research.

It is important to appreciate that any human genome harbors about 4 to 5 million DNA variants. Among them, about 180 are protein truncating, about 12,000 peptides are sequence altering, and about 600,000 variants reside in promoters, enhancers, transcription factor binding sites and other noncoding regulatory regions. As the latter suggests, variants in protein coding genes are potentially of relevance to exercise biology but DNA sequence variants impacting regulation of gene expression and transcription of various RNA populations are likely to be of even greater importance. At times, variants exhibit large effects, but they typically influence complex traits in a subtle manner. Many of these variants influence gene expression levels in tissues where these genes are normally expressed (27). Moreover, homozygotes for loss of function alleles are of great scientific interest as they are human gene knockouts equivalent to mouse knockouts and are not infrequent in the human genome (37,38).

  • 4. “Demonstration that the gene or pathway is obligatory for an adaptive response”

Small effect size, widely distributed gene expression regulation and biological redundancy are foundational of complex, multifactorial traits. It is not a simple task to define what is obligatory for a trait such as CRF trainability. For instance, a mutation known to cause a particular disease will do so in some individuals, whereas the same mutation may only increase the risk of such a disease in other people. Why is that so? It is the result of multiple interactive or compensatory systems, including gene–gene interactions, environmental and lifestyle agents interacting with alleles, redundancy in gene and pathway functions, epigenetic events and other sources as well. Disqualifying DNA variants with small effect sizes because they are not in a so-called obligatory gene or pathway is not productive.

  • 5. “Applicability of criteria 1 to 4 when a maximal adaptive stimulus has been applied”

Presumably, this requirement does not apply to CRF in the sedentary state, but I also fail to see the rationale for this requirement in the case of CRF trainability. There is no compelling reason why this should be a litmus test for the acceptance of genetic evidence. Submaximal exercise capacity is also an indicator of CRF. What we have learned, thus far, on the genetic basis of trainability in animal models and humans has been derived from studies in which defined submaximal exercise doses were used. Although doses have varied across studies, the findings show consistently that the responsiveness to exercise training stimuli is largely conditional on the genotype of mice, rats, or humans.

Miscellaneous issues

Several other contentious views were expressed by Dr. Joyner, and hopefully they can be addressed in depth at some point in the future. One of them dealt with natural selection and CRF level. There are several examples documenting changes in allele frequency resulting from the selection pressure of an environmental agent, including a body of data on the notion that CRF has been subjected to selection pressure during the evolutionary history of Homo sapiens (39).

The brief comment on height can be misinterpreted. The heritability of height is of the order of 80% of the trait variance. It takes more than 9000 SNP to explain more than 80% of this heritability level. The remainder of the heritability is most likely dependent on low frequency and rare variants with some having an effect size in the range of 2 cm (32). Thus, height has a strong genetic component, which decomposes into a large but finite number of DNA variants.

The citation from Dr. Francis Collins in the Commentary reflects the general thinking on the genetic architecture of common diseases prior to the GWAS era, which provoked a major shift of paradigm. The goal of science is to decrease our level of ignorance, and we should not be surprised if people change their views in light of new evidence. I do not believe that it is productive to confine people in a box based on views expressed in the past when the science base was different. Dr. Collins would undoubtedly summarize the evidence quite differently today in light of the progress made on the genetics of traits, such as type 2 diabetes.

Regarding the difference in muscle slow myosin in single fibers between two MZ twin brothers discordant for level of exercise over 30 yr, large differences in muscle fiber type distributions are observed even among sedentary people with no prior exercise training experience (40). For instance, among sedentary men and women, 25% have less than 35% fiber type 1 fibers in the vastus lateralis but 25% have more than 65% of the same fibers (41). The truth may be more complex than suggested by the case study of a pair of twins.

Finally, the Commentary concludes with a plea to move beyond DNA and genes in exercise biology research! At a time when we enjoy powerful genomic and molecular biology technologies, this sounds as a call to adopt a Lysenko-like approach to exercise biology research. A statement that can be misinterpreted by many.

Concluding statement

We continue to learn a great deal about the genomic basis of complex human traits and their underlying biology. Contrary to expectations, common DNA variants with small effect sizes are of extraordinary importance. The emerging model is one in which alleles at large number of loci modulate the level of a phenotype and the changes in phenotype resulting from an environmental exposure. In this regard, and concordant with the lack of correlation between sedentary CRF and its trainability, the genomic signature underlying CRF in the sedentary state is predicted to be quite different from the genomic profile impacting CRF trainability. Finally, exercise scientists should embrace the new powerful technologies to interrogate the genome, epigenome, transcriptome, proteome, lipidome, and metabolome of all exercise-related traits.

C. B. is partially funded by the John W. Barton Sr. Chair in Genetics and Nutrition, and NIH COBRE grant (NIH 8 P30GM118430-01).

REFERENCES

1. Lightfoot JT, Turner MJ, Debate KA, Kleeberger SR. Interstrain variation in murine aerobic capacity. Med Sci Sports Exerc. 2001;33(12):2053–7.
2. Meek TH, Lonquich BP, Hannon RM, Garland T Jr. Endurance capacity of mice selectively bred for high voluntary wheel running. J Exp Biol. 2009;212(18):2908–17.
3. Avila JJ, Kim SK, Massett MP. Differences in exercise capacity and responses to training in 24 inbred mouse strains. Front Physiol. 2017;8:974.
4. Koch LG, Britton SL. Theoretical and biological evaluation of the link between low exercise capacity and disease risk. Cold Spring Harb Perspect Med. 2018;8(1).
5. Bouchard C, Lesage R, Lortie G, et al. Aerobic performance in brothers, dizygotic and monozygotic twins. Med Sci Sports Exerc. 1986;18(6):639–46.
6. Klissouras V. Heritability of adaptive variation. J Appl Physiol. 1971;31(3):338–44.
7. Maes HH, Beunen GP, Vlietinck RF, et al. Inheritance of physical fitness in 10-yr-old twins and their parents. Med Sci Sports Exerc. 1996;28(12):1479–91.
8. Schutte NM, Nederend I, Hudziak JJ, Bartels M, de Geus EJ. Twin-sibling study and meta-analysis on the heritability of maximal oxygen consumption. Physiol Genomics. 2016;48(3):210–9.
9. Bouchard C, Daw EW, Rice T, et al. Familial resemblance for VO2max in the sedentary state: the HERITAGE family study. Med Sci Sports Exerc. 1998;30(2):252–8.
10. Pérusse L, Gagnon J, Province MA, et al. Familial aggregation of submaximal aerobic performance in the HERITAGE Family Study. Med Sci Sports Exerc. 2001;33(4):597–604.
11. Gaskill SE, Rice T, Bouchard C, et al. Familial resemblance in ventilatory threshold: the HERITAGE Family Study. Med Sci Sports Exerc. 2001;33(11):1832–40.
12. Bouchard C, Lortie G, Simoneau JA, Leblanc C, Thériault G, Tremblay A. Submaximal power output in adopted and biological siblings. Ann Hum Biol. 1984;11(4):303–9.
13. An P, Rice T, Gagnon J, et al. Familial aggregation of stroke volume and cardiac output during submaximal exercise: the HERITAGE Family Study. Int J Sports Med. 2000;21(8):566–72.
14. Massett MP, Avila JJ, Kim SK. Exercise capacity and response to training quantitative trait loci in a NZW X 129S1 intercross and combined cross analysis of inbred mouse strains. PLoS One. 2015;10(12):e0145741.
15. Koch LG, Pollott GE, Britton SL. Selectively bred rat model system for low and high response to exercise training. Physiol Genomics. 2013;45(14):606–14.
16. Wisloff U, Bye A, Stolen T, et al. Blunted cardiomyocyte remodeling response in exercise-resistant rats. J Am Coll Cardiol. 2015;65(13):1378–80.
17. Prud'homme D, Bouchard C, Leblanc C, Landry F, Fontaine E. Sensitivity of maximal aerobic power to training is genotype-dependent. Med Sci Sports Exerc. 1984;16(5):489–93.
18. Hamel P, Simoneau JA, Lortie G, Boulay MR, Bouchard C. Heredity and muscle adaptation to endurance training. Med Sci Sports Exerc. 1986;18(6):690–6.
19. Boulay MR, Lortie G, Simoneau JA, Bouchard C. Sensitivity of maximal aerobic power and capacity to anaerobic training is partly genotype dependent. In: Malina RM, Bouchard C, editors. Sport and Human Genetics. Champaign, IL: Human Kinetics; 1986. pp. 173–81.
20. Simoneau JA, Lortie G, Boulay MR, Marcotte M, Thibault MC, Bouchard C. Inheritance of human skeletal muscle and anaerobic capacity adaptation to high-intensity intermittent training. Int J Sports Med. 1986;7(3):167–71.
21. Bouchard C, Tremblay A, Despres JP, et al. The response to exercise with constant energy intake in identical twins. Obes Res. 1994;2(5):400–10.
22. Bouchard C, An P, Rice T, et al. Familial aggregation of VO(2max) response to exercise training: results from the HERITAGE Family Study. J Appl Physiol (1985). 1999;87(3):1003–8.
23. Manolio TA. In retrospect: a decade of shared genomic associations. Nature. 2017;546(7658):360–1.
24. Visscher PM, Wray NR, Zhang Q, et al. 10 years of GWAS discovery: biology, function, and translation. Am J Hum Genet. 2017;101(1):5–22.
25. ENCODE Project Consortium. An integrated encyclopedia of DNA elements in the human genome. Nature. 2012;489(7414):57–74.
26. 1000 Genomes Project Consortium; Auton A, Brooks LD, et al. A global reference for human genetic variation. Nature. 2015;526(7571):68–74.
27. GTEx Consortium. Human genomics. The genotype-tissue expression (GTEx) pilot analysis: multitissue gene regulation in humans. Science. 2015;348(6235):648–60.
28. Battle A, Brown CD, et al; Consortium GT, Laboratory DA, Coordinating Center -Analysis Working G. Genetic effects on gene expression across human tissues. Nature. 2017;550(7675):204–13.
29. Dowell RD, Ryan O, Jansen A, et al. Genotype to phenotype: a complex problem. Science. 2010;328(5977):469.
30. Osterwalder M, Barozzi I, Tissieres V, et al. Enhancer redundancy provides phenotypic robustness in mammalian development. Nature. 2018;554(7691):239–43.
31. Wood AR, Esko T, Yang J, et al. Defining the role of common variation in the genomic and biological architecture of adult human height. Nat Genet. 2014;46(11):1173–86.
32. Marouli E, Graff M, Medina-Gomez C, et al. Rare and low-frequency coding variants alter human adult height. Nature. 2017;542(7640):186–90.
33. Turcot V, Lu Y, Highland HM, et al. Protein-altering variants associated with body mass index implicate pathways that control energy intake and expenditure in obesity. Nat Genet. 2018;50(1):26–41.
34. Joyner MJ. Limits to the evidence that DNA sequence differences contribute to variability in fitness and trainability. Med Sci Sports Exerc. 2019;51(8):1786–9.
35. Bouchard C, Sarzynski MA, Rice TK, et al. Genomic predictors of the maximal O(2) uptake response to standardized exercise training programs. J Appl Physiol (1985). 2011;110(5):1160–70.
36. Rankinen T, Fuku N, Wolfarth B, et al. No evidence of a common DNA variant profile specific to world class endurance athletes. PLoS One. 2016;11(1):e0147330.
37. Sudmant PH, Rausch T, Gardner EJ, et al. An integrated map of structural variation in 2,504 human genomes. Nature. 2015;526(7571):75–81.
38. MacArthur DG, Balasubramanian S, Frankish A, et al. A systematic survey of loss-of-function variants in human protein-coding genes. Science. 2012;335(6070):823–8.
39. Lieberman DEB, Bramble DM, Raichlen DA, Shea JJ. Brains, Brawn, and the Evolution of Human Endurance Running Capabilities. In: Grine FE, Fleagle JG, Leakey RE, editors. The First Humans: Origin and Early Evolution of the Genus Homo. New York: Springer; 2009. pp. 77–98.
40. Simoneau JA, Bouchard C. Human variation in skeletal muscle fiber-type proportion and enzyme activities. Am J Phys. 1989;257(4 Pt 1):E567–72.
41. Simoneau JA, Bouchard C. Genetic determinism of fiber type proportion in human skeletal muscle. FASEB J. 1995;9(11):1091–5.
Copyright © 2019 by the American College of Sports Medicine