In humans, physical activity declines with age. We tested the hypothesis that genetic background and age interact to determine daily wheel-running physical activity patterns in mice. Five female mice from ten inbred strains (A/J, AKR/J, Balb/cJ, CBA/J, C3H/HeJ, C3Heb/FeJ, C57Bl/6J, C57L/J, DBA/2J, and SWR/J) were studied for 26 wk starting at 10 wk of age. All mice were housed in separate cages, each with a running wheel and magnetic sensor. Throughout the 26-wk period, age-related change in daily duration (P < 0.0001), daily distance (P < 0.0001), and average velocity (P = 0.0003) differed between the inbred strains. Unlike the other strains, SWR/J mice increased their running-wheel activity throughout the 6-mo time period. Broad-sense heritability estimations for the strains across the 26-wk period ranged between 0.410 and 0.855 for the three physical activity phenotypes. Furthermore, the broad-sense heritability estimates for daily running-wheel distance differed across time and suggested an interaction between genetic background and age on physical activity in these inbred mice.
- average daily activity
- body weight
- broad-sense heritability
- inbred mice
an estimated 250,000 deaths per year in the United States can be directly attributed to physical inactivity (2). As individuals age beyond adolescence, the pattern of regular physical activity continues to decrease (4, 12). This extended period of declining physical activity pattern, along with dietary factors, place older adults at an increased risk for cardiovascular disease and/or death (23). Thus understanding possible biological factors that contribute to age-related physical activity decline could aid in the development of treatments that would increase the health of an aging population.
Although it is well accepted that various environmental factors can influence regular physical activity level (24), recent findings suggest that specific physiological systems highly influenced by genetic background may also play a role in regular activity patterns (27). Research with humans (16, 18) and animals (11, 33) indicates that there is genetic control of physical activity. Kaprio et al. (16) and Lauderdale et al. (18) utilized survey analyses with monozygotic and dizygotic twins to investigate the heritability of regular physical activity levels. These authors estimate heritability of daily physical activity from 38 to 62%. Because of significant genetic homology between humans and mice (25), findings concerning genetic influence with mice may have high generalizability with humans. Mice have been bred selectively for low physical activity levels (11, 33). Additionally, Lerman et al. (19) monitored daily wheel running for 2 wk in 5- to 6-mo-old inbred male mice and found a broad-sense heritability ranging from 24 to 59%. However, little is known concerning the influence of age and genetics on the amount of physical activity with aging.
If genetic factors participate in the decline of physical activity with age, a wide range of genetic techniques and methodologies will be available for future research efforts to better understand the decline in physical activity over the lifespan. The current study investigated the influence of genetic factors on the alterations in physical activity level in female inbred mice as they aged from 3 to 9 mo.
Five female mice (5–6 wk old) from 10 inbred strains (A/J, AKR/J, Balb/cJ, CBA/J, C3H/HeJ, C3Heb/FeJ, C57Bl/6J, C57L/J, DBA/2J, and SWR/J) were purchased from Jackson Laboratory (JAX; Bar Harbor, ME). Female mice were used because they are easier to handle, they have been previously shown to spontaneously run farther and faster than male mice (17), and they enable comparison with our previous data on maximal endurance exercise (22). All mice were housed in the university vivarium with 12:12-h light-dark cycles, with room temperatures and relative humidity standardized to 18–22°C and 20–40%, respectively. All mice were provided with water and standard chow (Harland Teklad 8604 Rodent Diet; Madison, WI) ad libitum. All study procedures were approved by the Institutional Animal Care and Use Committee at the University of North Carolina-Charlotte as meeting the guidelines for appropriate treatment of animal subjects as outlined by the United States Department of Agriculture and the Animal Welfare Act.
Physical activity measurement.
Running-wheel activity was used as the measure of physical activity. While the amount of wheel running exercise can be substantially different between individual outbred mice (6), the amount of wheel running activity within selected strains appears to be consistent and repeatable within a generation of animals (11, 21, 33). Past studies have shown a gradual increase in voluntary wheel-running activity in mice from age 3 to 8 wk, with peak activity occurring between 55 and 65 days, after which activity levels begin to decline (33).
At 10 wk of age, all mice were housed individually. Within each mouse cage was a solid-surface running wheel (127-mm diameter; Ware Manufacturing, Phoenix, AZ) interfaced with a magnetic sensor and computer (Sigma Sport BC600, Olney, IL) that counted revolutions of the running wheel and total time the running wheel was turning (19). The running wheels were checked every 24 h to ensure they turned freely, and the total time (min) and the total distance (km) each mouse ran in the past 24 h were recorded. If any wheel appeared to have difficulty in spinning, the wheel was oiled or replaced. Also, running wheels were washed in the university cage washer every 3 wk to maintain cleanliness. Every cleaned wheel was oiled and randomly assigned to a mouse in an effort to control for the influence of each running wheel's performance on daily running-wheel activity. Total daily distance (km) was divided by total daily wheel activity time (min) to calculate the average velocity of exercise (m/min). Data collection was performed daily until the mice were 36 wk old. Body weight (g) was measured weekly.
One-way ANOVAs with repeated measures were performed to compare weekly body weight, daily distance, daily duration, and daily velocity of running-wheel activity. Weekly averages for each running-wheel variable were utilized for all statistical analysis procedures, with F-statistics for the main effects (strain, df = 9; age, df = 25; strain × age, df = 225). Pairwise correlations were performed to investigate the influence of change in body weight on change in running-wheel activity over the study period. The differences in running-wheel activity and body weight between the mean values of the first and last 3 wk of the study period were calculated for each mouse, and a correlation coefficient was determined to assess whether body weight was associated with any of the running-wheel variables. One-way ANOVA with repeated measures was performed to compare broad-sense heritability estimates (see below) between strains (df = 9) at 3-wk intervals. One-way ANOVA with repeated measures was performed to assess differences in broad-sense heritability across the 26-wk study period (df = 8). For all analyses, Shapiro-Wilk W-tests were performed to assess for normality, and power values were calculated with an α-level of 0.05. Student's t-tests were performed to determine whether each broad-sense heritability estimate was different from zero. An α-level of P < 0.05 was used to determine significant effects for each analysis. Post hoc comparisons of between-strain means were performed using Tukey's HSD tests. All statistical analyses were performed using JMP Statistical Analysis software (SAS Institute).
The estimation of broad-sense heritability for each inbred strain and across the 26-wk study period was performed by partitioning the data for each running-wheel phenotype (distance, duration, and velocity) into 3-wk intervals. Running-wheel activity was averaged for each 3-wk interval for each mouse. Mean square of the within-strain comparison was determined separately for each inbred strain for each 3-wk interval, while the mean square of the between-strain comparison was determined from an ANOVA between strains at each 3-wk interval for the estimation of broad-sense heritability. Interclass correlations (rI) and coefficient of genetic determination (g2) are measures of broad-sense heritability and were calculated using methods outlined by Festing (7). Intraclass correlations are defined as the proportion of the total variation that is accounted for by differences between strains and are estimated as follows: rI = (MSB − MSW)/[MSB + (n − 1)MSW], where MSB is the mean square of the between-strain comparison, MSW is the mean square of the within-strain comparison, and n is the number of animals tested per strain (7). The coefficient of genetic determination (g2), while rarely reported in the literature, takes into account the doubling of the additive genetic variance with inbreeding and was calculated as follows: g2 = (MSB − MSW)/[MSB + (2n − 1)MSW], where MSB is the mean square of the between-strain comparison, MSW is the mean square of the within-strain comparison, and n is the number of animals tested per strain (7). The difference in the two heritability estimates lies in the doubling of n in the denominator of g2. The coefficient of genetic determination controls for the additive genetic variance that occurs with inbreeding, and while it results in more conservative heritability estimates, it has been noted to be a better indicator of broad-sense heritability (7). Therefore, all comparisons and expressions of broad-sense heritability are made with g2 results unless noted otherwise in the text. Student’s t-tests (t) were performed to determine whether broad-sense heritability was different from zero.
The estimated percentage of lifespan for each inbred mouse strain at the completion of the study is displayed in Table 1. Forty-five mice completed the 6-mo study period. One A/J mouse died during the 4th wk of the study from unexplained causes, and the data for this animal were not included in the statistical analyses. Four AKR/J mice also died during study weeks 23–25, which corresponded to 85–90% of their expected lifespan (32). Thus the data for these four AKR/J mice are included in the results after adjustment of the study period to correspond specifically to the last week of life for each mouse.
Mean body weights of the 10 strains of mice were significantly different at the start of the physical activity period (P < 0.0001; Fig. 1); AKR/J mice were significantly heavier than A/J, Balb/cJ, C3H/Heb, C57Bl/6J, CBA/J, and DBA/2J mice (normality = 0.001, power > 0.99). Throughout the 26-wk physical activity period, all mice increased in body weight (P < 0.0001). The increase in body weight was significantly greater in the C3Heb/FeJ mice and less in the DBA/2J mice (P < 0.0001, normality = 0.0015, power > 0.99).
Significant differences in all indexes of running-wheel activity were observed between the 10 strains of mice throughout the 26-wk period. When comparing distance run per day (Fig. 2), we found a significant difference between the strains (P < 0.0001, normality < 0.0001, power > 0.99). Overall, daily distance decreased throughout the 26-wk time period (P < 0.0001), although significant interaction (P < 0.0001) suggested a difference between the strains in their daily running distance across the 26-wk study period. Interestingly, C57L/J and SWR/J mice ran farther on a daily basis as they aged compared with the remaining eight strains of mice.
The weekly averages for duration run per day throughout the 26 wk (Fig. 3) were different between the strains (P < 0.0001, normality < 0.0001, power > 0.99), and the duration of running-wheel activity decreased significantly (P < 0.0001) across the 26 wk. A significant interaction of duration across the 26-wk study period (P < 0.0001) was found; a greater duration of activity was found in SWR/J mice compared with the remaining strains of mice except C57L/J.
The average velocity of running-wheel activity per day (Fig. 4) was also significantly different between the strains (P < 0.0001, normality < 0.0001, power > 0.99). Overall, as the mice aged, the average daily velocity of wheel running significantly increased (P < 0.0001), and significant interaction was observed for average daily wheel running velocity across the 26-wk period (P = 0.0003). The average velocity of activity per day was faster for the C57L/J and SWR/J mice compared with the remaining eight strains throughout the 26 wk. From week 5 to week 21, C3H/HeJ mice ran faster than the Balb/cJ, CBA/J, C3Heb/FeJ, C57Bl/6J, and DBA/2J strains.
Correlation analyses with body weight.
No significant correlation was observed between the change in body weight and the change in distance run throughout the 26-wk period (r = −0.138, P = 0.33). A significant correlation was observed between the change in body weight and the change in exercise duration throughout the 26-wk period (r = −0.361, P = 0.01). No significant correlation was found between the change in body weight and the change in average velocity throughout the 26-wk period (r = 0.180, P = 0.22).
Average broad-sense heritability estimates for daily running-wheel distance across the 26-wk study period (Fig. 5A) ranged from 54.6 to 85.5% for interclass correlations (rI) and 41.0 to 76.4% for coefficients of genetic determination (g2). The week 12 broad-sense heritability estimates were found to be significantly lower than the remaining weeks (P = 0.0013, df = 8, normality = 0.0014, power = 0.98) (Table 2). Average broad-sense heritability estimates for daily running-wheel distance between the 10 strains ranged from 46.8 to 87.4% for rI and 31.8 to 78.4% for g2. The heritability estimate for SWR/J mice was significantly lower than for the remaining strains (P < 0.0001, df = 9, normality = 0.0014, power > 0.99). The broad-sense heritability estimates (g2) for distance were different from zero (P < 0.0001, t = 27.446, df = 89).
Average broad-sense heritability estimates for daily running-wheel duration across the 26-wk study period (Fig. 5B) ranged from 59.1 to 81.2% for rI and 45.9 to 71.5% for g2. No significant difference was found between heritability estimates for running-wheel duration across time (P = 0.149, df = 8, normality = 0.083, power = 0.66) (Table 3). Average broad-sense heritability estimates for daily running-wheel duration between the inbred strains ranged from 47.8 to 86.8% for rI and 32.4 to 77.3% for g2. The heritability estimate for duration of running-wheel activity was significantly lower for Balb/cJ, C57Bl/6J, and SWR/J compared with the remaining inbred strains (P < 0.0001, normality = 0.083, power > 0.99). The broad-sense heritability estimates (g2) for duration were different from zero (P < 0.0001, df = 9, t = 26.257, df = 89).
Average broad-sense heritability estimates for daily running-wheel velocity across the 26-wk study period (Fig. 5C) ranged from 57.8 to 78.8% and 42.0 to 67.3% for rI and g2, respectively. The velocity broad-sense heritability estimates were significantly different across time intervals (P = 0.035, df = 8, normality = 0.297, power = 0.83) (Table 4). Average broad-sense heritability estimates for daily running-wheel velocity between the 10 strains ranged from 51.8 to 84.0% and 38.8 to 73.3% for rI and g2, respectively. The average daily velocity heritability estimates for A/J and AKR/J strains were significantly lower than the remaining inbred strains (P = 0.0006, df = 9, normality = 0.297, power = 0.99). Also, the daily velocity heritability estimates for C3H/HeJ and C57Bl/6J strains were significantly higher than the remaining inbred strains. The average daily velocity broad-sense heritability estimates (g2) were different from zero (P < 0.0001, t = 25.270, df = 89).
The current study tested the hypothesis that genetic background participates in the alteration of daily physical activity with aging. Our results demonstrated a significant influence of genetic background on the amount of running-wheel activity performed by the 10 strains studied over the 26-wk period. Additionally, we observed that the amount of influence genetic background exerted on daily activity changed over the 26 wk. Thus the amount of running-wheel activity performed by these mice over the 6-mo study period varied as a function of time and genetic background (i.e., gene × time interaction). Depending on the activity phenotype (i.e., running distance, duration, or velocity), broad-sense heritability estimates (41.0–85.5%) showed a high genetic influence on age-related changes in physical activity. To our knowledge, these are the first data to suggest that genetic background exhibits a highly significant influence on physical activity level which changes as a function of time.
The current study was the first to examine the heritability of age-related alterations in physical activity with the use of a longitudinal model. Epidemiological research with middle-aged and older adults (1, 4, 5), children, and young adults (28) and cross-sectional research with animals (13–15, 20, 35) have consistently observed an age-associated decline in physical activity levels. This decline is commonly believed to be of multifactorial causes, with social, environmental, and physiological influences. Injury, illness, time devoted to childcare, physical labor, inflexible work schedule, residence in unsafe neighborhood, distance from fitness facility, and lack of transportation are typically suggested as important social and environmental factors in the age-related decline in regular activity (for a review, see Ref. 29).
Human studies performed to estimate heritability of regular physical activity levels in adults have utilized interview data and written survey analyses with monozygotic and dizygotic twins to investigate the heritability of regular physical activity levels (16, 18, 30). Their findings suggest a heritability estimate of 38–62% for daily physical activity. In a study by Simonen et al. (30), regression models were used to estimate the influence of genetic and environmental factors on adolescent and adult lifetime exercise. These analyses indicated that heritability accounted for 17 to 51% of the variance in exercise from adolescence to adulthood, respectively. The increase in the influence of genetic background from adolescence to adulthood is similar to our findings, where heritability estimates were lower in the younger age periods and increased into young adulthood. Our heritability estimates are higher than those determined by Simonen et al. (30), which may be explained by the use of inbred strains and highly regulated environmental conditions. Simonen et al. (30) also found that environmental factors influenced 43–46% of the variability in lifetime exercise in their adult population. In a study using only monozygotic twins, Simonen et al. (31) found that environmental factors during childhood significantly influence adulthood exercise patterns. Results of the current study are consistent with Simonen et al. (31) in that a large portion of the physical activity performed at younger ages is more highly influenced by environmental factors when compared with older age periods.
A decline in regular physical activity levels with advancing age is a common finding in animal research (12). However, a majority of research studies have employed a cross-sectional design with male rodents to study age-related differences in physical activity (13–15, 20, 35). Lhotellier and Cohen-Salmon (20) used wheel running in a cross-sectional study design with three inbred strains of mice (Balb/c, DBA/2, and C57BL/6) of ∼21, 57, and 107 wk of age. Similar to our findings, these researchers observed different patterns of age-related declines in physical activity level between the strains. Inasmuch as we attempted to control environmental influences with our longitudinal study design, the differences observed with our data would suggest that genetic background varies in its influence on the age-related changes in physical activity.
Interestingly, while the expected decrease in activity over the 26-wk period was found in seven of the strains, we observed an increase in the daily wheel running activity in SWR/J mice and little change in wheel activity in C3H/HeJ and C57L/J mice. The decline in daily physical activity level with age is similar to findings in humans (3). The lack of a decline in physical activity level by the SWR/J, C3H/HeJ, and C57L/J mice suggests that genetic background may be the primary factor in determining physical activity level with an increase in age. While our data are unique in suggesting that genetic background plays a large role in the determination of daily activity, one may expect an association between physically active inbred strains with increased expression of highly physically functioning systems (i.e., cardiac, skeletal muscle, metabolic, pulmonary, and vascular). Support for this concept comes from Tsao et al. (34), who found that overexpression of the primary glucose transporter isoform in skeletal muscle (GLUT-4) caused a fourfold increase in daily wheel-running activity.
Environment accounted for ∼14.5–59.0% of the variance in daily physical activity performed by the 10 strains across the 26-wk study period. Environmental factors have been commonly accepted as the major influence on daily physical activity level (24). In addition to societal and perceptual factors, other environmental factors that influence the amount of physical activity performed by an individual involve physiological functioning (i.e., cardiac, skeletal muscle, metabolic, pulmonary, and vascular) (26).
All mice involved in the current study were females. Previous research performed on wheel-running activity found female mice to be more active than males (17, 21). In our laboratory, Lightfoot et al. (21) monitored 13 strains from 9 to 15 wk of age and observed that female mice ran farther and faster each day, while the males ran a greater daily duration. Koteja et al. (17) bred mice for high activity levels and found similar differences between the sexes for distance, duration, and running velocity. Lightfoot et al. (21) also found greater heritability estimates with inbred male mice due to smaller variability within the strains. Thus our heritability estimates using female mice from similar inbred strains can be assumed to be somewhat conservative when taking into consideration the expected estimates for an inbred mouse population of both sexes.
As expected (10), body weight increased throughout the 26-wk study period. These body weight alterations were unrelated to the changes observed with distance run or average velocity of exercise from the start to end of the 26-wk study period. However, the alterations in duration of exercise performed throughout the study period were negatively related to the changes observed in body weight with the female mice. Friedman et al. (9) did not observe a correlation between body weight and amount of wheel running with selectively bred house mice. Also, Swallow et al. (33) found no difference in the body weights of sedentary control mice and physically activity mice. Thus the literature and our results are still somewhat ambiguous on what, if any, relationship exists between body weight and daily activity levels.
Of interest was the amount of physical activity performed by strains that were observed previously in our laboratory (22) to exhibit high and low maximal exercise endurance. The A/J, AKR/J, and DBA/2J strains were found to have the lowest maximal exercise endurance when measured at 8–10 wk of age; A/J and AKR/J strains were “aggressively sedentary” (i.e., these strains actively worked to resist forced exercise). In the current study, after 12 wk of age, the A/J, AKR/J, and DBA/2J strains consistently performed the lowest amounts of physical activity compared with the remaining seven strains. We previously found the highest maximal exercise endurance in Balb/cJ mice (22). In the current study, Balb/cJ mice regularly performed moderate to low amounts of physical activity throughout the 6-mo study period. This information is consistent with other literature that suggests a poor relationship between the amount of physical activity being performed and maximal aerobic capacity (19). Therefore, it appears that while maximal endurance exercise can be influenced by regular exercise training, maximal exercise endurance and daily physical activity levels are distinct phenotypes.
Interpretations from the current study must be made with caution. At the end of the study period, the percent lifespan for each strain ranged from 37 to 95% (Table 1), and thus these findings do not reflect relative age. However, the findings do suggest that the amount of influence genetic background has on the amount of physical activity performed varies throughout the first one-half of life. An additional concern for the current study may involve the small number of mice used per strain. Although five mice were utilized with each strain, the power analysis for type I error at 0.05 was ≥0.83 for all ANOVAs, except for the analysis of broad-sense heritability estimates for daily running-wheel duration across the 26-wk study period (power = 0.657). To achieve a power >0.80 for this analysis, a power estimation of 27 additional data points, or data for three additional inbred strains, is necessary across the 26-wk study period. Nevertheless, the use of five mice per inbred strain appears to have delivered adequate statistical power to draw appropriate conclusions.
In conclusion, we found that genetic background contributes significantly to the variance in daily physical activity among female inbred strains of mice. Furthermore, estimates of heritability varied across time within and between strains (i.e., gene × time effects). These findings could have significant impact on developing new methods and approaches for maintaining and increasing daily activity in humans as they age. The results could be useful in improving the screening of those at risk for hypokinetic-related diseases, prevention for those at risk for hypokinesia, and treatment for individuals who are hypokinetic due to genetic background. Also, the appreciation of the increasing influence of genetic background on regular physical activity from adolescence to adulthood implies a decreasing role for environmental factors as young individuals continue to age into adulthood.
This research study was supported by National Institutes of Health Grants AG-022417 (M. J. Turner) and DK-61635 (T. Lightfoot), the University of North Carolina-Charlotte Faculty Development Program (T. Lightfoot), and the National Institute of Environmental Health Sciences (S. R. Kleeberger).
We thank Sherin Salama, Amber Lowe, and Mark Lindley for their assistance with data collection and running-wheel maintenance throughout this project. We acknowledge Dr. B. Harrison for sharing the design of the measurement method of daily physical activity. We also thank the reviewers for insightful comments that resulted in a significantly improved manuscript.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: M. J. Turner, Dept. of Kinesiology, Univ. of North Carolina-Charlotte, Charlotte, NC 28223 (e-mail:)
- Copyright © 2005 the American Physiological Society