Phenotypic and molecular differences between rats selectively bred to voluntarily run high vs. low nightly distances

Abstract

The purpose of the present study was to partially phenotype male and female rats from generations 8–10 (G8–G10) that had been selectively bred to possess low (LVR) vs. high voluntary running (HVR) behavior. Over the first 6 days with wheels, 34-day-old G8 male and female LVRs ran shorter distances (P < 0.001), spent less time running (P < 0.001), and ran slower (P < 0.001) than their G8 male and female HVR counterparts, respectively. HVR and LVR lines consumed similar amounts of standard chow with or without wheels. No inherent difference existed in PGC-1α mRNA in the plantaris and soleus muscles of LVR and HVR nonrunners, although G8 LVR rats inherently possessed less NADH-positive superficial plantaris fibers compared with G8 HVR rats. While day 28 body mass tended to be greater in both sexes of G9–G10 LVR nonrunners vs. G9–G10 HVR nonrunners (P = 0.06), body fat percentage was similar between lines. G9–G10 HVRs had fat mass loss after 6 days of running compared with their prerunning values, while LVR did not lose or gain fat mass during the 6-day voluntary running period. RNA deep sequencing efforts in the nucleus accumbens showed only eight transcripts to be >1.5-fold differentially expressed between lines in HVR and LVR nonrunners. Interestingly, HVRs presented less Oprd1 mRNA, which ties in to potential differences in dopaminergic signaling between lines. This unique animal model provides further evidence as to how exercise may be mechanistically regulated.

approximately 97% of us adults and 92% of adolescents are not meeting US daily physical activity guidelines for 30 and 60 min, respectively (44). The clinical significance of lack of sufficient daily physical activity is an increased prevalence of 35 chronic unhealthy conditions and premature death (5). Motivation, fitness, and genes are all associated with physical activity levels (4). Twin and family studies have shown that genetic factors contribute to variation in reported daily physical activity levels, with identical cotwins showing smaller intraindividual variation than nontwins (see Ref. 4 for references). Animal studies also suggest that brain mechanisms affect the quantity of daily voluntary running (14, 21, 22, 36, 38).

One approach to obtain genetic information on genes and physical activity has been to employ a selective breeding model of physical activity in rodents. According to Rhodes et al. (35), selective breeding is a “powerful alternative” compared with knockout models in studying genotype-phenotype interactions, given that multiple genes contribute to complex phenotypes. Swallow et al. (42) have phenotyped their 10th generation of mice selected for the highest revolutions per day of wheel running, these being coined high voluntary runner or HVR mice. After 60 generations, numerous physiological, behavioral, neurobiological, and gene differences in HVR have been revealed compared with controls (see Ref. 14 for references). Koch and Britton (23) first reported in 2001 a similarly developed selective breeding model of rats that intrinsically possessed high (HCR) vs. low aerobic endurance capacity (LCR) using forced treadmill running to an operationally defined level of exhaustion. After 23 generations, LCRs had numerous greater disease risks, including metabolic syndrome and cardiovascular complications (24), as well as premature aging and a remarkable 28–45% reduction in average lifespan (25) compared with HCR rats. In justifying their rationale for selective breeding, Koch et al. (25) wrote, “This contrasting animal model system may prove to be translationally superior relative to more widely used simplistic models for understanding geriatric biology and medicine.” Nonetheless, neither of these models has selected for the phenotype of low amounts of voluntary running.

In light of over 90% of US residents greater than 12 yr of age failing to meet US guidelines for daily physical activity (44), the present study reports the outcome of our 4-yr selective breeding efforts to develop a rat model that selected for high vs. low voluntary wheel running (HVR and LVR rats, respectively). Our longer term aim is to use the model to understand the neural basis to either be less or more physically active. In the present study, we report phenotypic characteristics that exist in our generation 8–10 (G8–G10) HVR vs. LVR animals, including our criterion variables (i.e., 6-day running distance, time, and speed), as well as skeletal muscle characteristics, body composition, and differentially expressed genes in the nucleus accumbens (NAc) using RNA deep sequencing (RNA-seq). The NAc was assayed due to the fact that this brain area has been investigated as a reward center for naturally motivating activities (i.e., physical activity) as well as drug addiction (20–22, 31). We assessed some of these measurements in rats at 28 days of age (before wheel running), as well as at 34 days of age either 1) after the 6-day running selection period (6-day runners), or 2) in rats that were always housed without running wheels (nonrunners).

MATERIALS AND METHODS

Animals and Selective Breeding

The Institutional Animal Care and Use Committee at the University of Missouri-Columbia approved all animal experiments. Our selective breeding model was performed similar to that of Swallow et al. (42) for HVR mice and that of Koch and Britton (23) who selected HCR and LCR rats. Our founder population consisted of 159 twenty-four-day-old outbred Wistar rats (80 males and 79 females; Charles River Raleigh, Raleigh, NC). These rats were introduced to running wheels (circumference: 0.345 m) at 28 days of age and monitored for 6 days of running using Sigma Sport BC 800 bicycle computers (Cherry Creek Cyclery, Foster Falls, VA). The 26 highest voluntary runners (HVR) (13 males and 13 females) and 26 lowest voluntary runners (LVR) (13 males and 13 females) were chosen from the founder population based on the average running distance on nights 5 and 6, after Garland (Ref. 42, personal communication), and randomly paired for mating for G1 to create 13 founder families for the HVR line and 13 families for the LVR line (Fig. 1).

Fig. 1.

Sum of running distances on days 5 and 6 for each rat in founder Wistar rats. Thirteen males and females that ran the lowest were randomly paired for mating (A), and highest distances were randomly paired (B). Thus 13 high voluntary running (HVR) and 13 low voluntary running (LVR) families were created and are currently being maintained. Subsequent generations were created by a scheduled rotation of breeding within the 13 HVR families and within the 13 LVR families for breeding.

Progeny from the 13 founder families were weaned at 21 days of age and introduced to running wheels at 28 days old, whereby 6-day voluntary running was again monitored for HVR and LVR G1's breeder selection. Rotational breeding using 13 families through G1–G10, as outlined by Koch and Britton (23) after Falconer (12), was performed to the enrich genetic variability of the selection trait. According to Kane et al. (19), replicate lines must be maintained to gain a true estimation of genetic drift. Due to the fact that our resources were extraordinarily limited, we could not determine the role or contribution of genetic drift toward our observed phenotypes given that we have not maintained replicate HVR or LVR lines. Cages were in temperature-controlled animal quarters (21°C) with a 0700–1900 light/1900–0700 dark cycle that was maintained throughout the experimental period, and all animals were provided standard chow (Formulab Diet 5008) and water ad libitum.

Approach for Dissecting Mechanistic Origin of Voluntary Running

Our plan for dissecting out potential characteristics that regulate voluntary running using our selectively bred HVR and LVR lines are presented in Fig. 2. Briefly, after breeding 8–10 generations of HVR and LVR lines, we sought to examine multiple phenotypes between these lines, including hindlimb muscle phenotype, body composition, and mRNA differences in the NAc.

Fig. 2.

Approach for dissecting mechanistic origin of voluntary running.

Skeletal Muscle Phenotyping Procedures of G8 Rats

G8 HVR and LVR rats were phenotyped at 34 days of age under two conditions: 1) those that voluntarily ran during 28–34 days of age (6-day runners), or 2) those that never ran (nonrunners). Like the preceding generations, HVR and LVR G8 6-day runners were weaned at 21 days of age, introduced to running wheels at 28 days of age, and monitored for 6 days of running using bicycle computers. Food (Formulab Diet 5008) and water were provided ad libitum and weighed over the 6-day running period. At 34 days of age between 1700 and 1900 (i.e., up to 2 h before the dark cycle), rats were administered an intraperitoneal (ip) injection of a lethal dose of pentobarbital sodium (60 mg/kg body mass). Skeletal muscles (plantaris and soleus) were extracted and 1) flash frozen in liquid nitrogen and stored at −80°C until RNA analyses, or 2) contralateral muscles were mounted on cork with optimal cutting temperature media, frozen in liquid nitrogen-cooled isopentane, and stored at −80°C until histochemical analyses.

As a part of the current phenotyping endeavors, parents of G8 pups were bred to produce a second litter of offspring that were housed in cages without voluntary running wheels so they never voluntarily ran [termed “nonrunners” to distinguish them from the first litter, which voluntarily ran between 28 and 34 days of age (termed “6-day runners”)]. G8 HVR and LVR nonrunners were also weaned at 21 days of age, whereby they were group housed in conventional rat housing cages without running wheels, and food and water were provided ad libitum. At 34 days of age between 1700 and 1900 (up to 2 h before the dark cycle), rats were similarly administered an ip injection of a lethal dose of pentobarbital sodium (60 mg/kg body mass), and skeletal muscles (plantaris and soleus) were extracted and stored for RNA or histochemical analyses as described above. It should be noted that, due to animal room constraints, HVR and LVR nonrunners were housed four to six animals per cage, and thus daily food intake measurements were averaged for three HVR and three LVR families.

RT-PCR for skeletal muscle mRNA expression.

During RNA isolation, tissues were crushed using a liquid nitrogen-cooled mortar and pestle, and ∼30 mg of powder were placed in TRIzol (Sigma, St. Louis, MO). Tissue lysis was performed at 30 Hz for 1 min using a high-speed shaking apparatus with stainless steel beads (Tissuelyser LT, Qiagen, Valencia, CA). RNA was then isolated from the minced tissue using the TRIzol method, according to manufacturer's instructions, and DNase-treated using DNase I (Thermo Scientific, Glen Burnie, MD), and RNA quality was confirmed on a 1% agarose gel. One microgram of RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription kit (Applied Biosystems, Carlsbad, CA). Twenty-five nanograms of cDNA from each sample were then assayed for the following using gene-specific primers and SYBR green chemistry (Power SYBR green Mastermix, Applied Biosystems): plantaris: GLUT4, peroxisome proliferator-activated receptor-γ coactivator-1α (PGC-1α), hexokinase II (HKII), VEGF, and 18S; soleus: GLUT4, PGC-1α, HKII, CD36, carnitine palmitoyl transferase 1 (CPT1), and 18S. Primer efficiency curves for all genes were generated, and efficiencies ranged between 90 and 110%. mRNA expression values are presented as 2^ΔCT, whereby ΔCT = 18S CT − gene of interest CT, and were normalized to male and female LVRs, respectively. Of note, we initially assayed these mRNAs in 6-day runners from HVR and LVR lines. If differences in an mRNA existed between HVR and LVR 6-day runners, the target was further assayed in the nonrunner HVR and LVR lines to determine whether the mRNA was inherently expressed in a particular line type, regardless of voluntary running.

Histochemistry.

Histochemical staining for NADH-tetrazolium reductase (a marker mitochondrial content) was performed on plantaris sections due to the fact that there is a heterogenous mixture of oxidative and glycolytic fibers in this muscle (3), and NADH-positive fibers are indicative of training status in rats (11, 18). Thus we hypothesized that, if HVR rats were to possess a higher number of NADH-positive fibers, then this may indicate that an increased oxidative capacity of hindlimb fibers could partially be attributed to increased running patterns in these animals. Cryosections were cut 15 μm thick from the midbelly of muscle samples, placed on microscope slides (Fisherbrand Superfrost Plus, Fisher Scientific, Waltham, MA) and stored at −20°C until staining. On the day of NADH staining, slides were removed from −20°C and incubated for 10 min at room temperature in NADH-TR reaction solution per the methods of Allen et al. (2). Slides were then air dried and mounted, examined using an Olympus BX60 photomicroscope at ×20 magnification (Olympus, Melville, NY), and photographed with Spot Insight digital camera (Diagnostic Instruments, Sterling Heights, MI). Fiber counting was performed using Image J software (National Institutes of Health) around the superficial aspect of the plantaris, since this area contains a heterogeneous mixture of glycolytic and oxidative fibers. As reported in previous literature (2), three staining intensities were observed, including larger, lightly stained (fast, glycolytic), larger, moderately stained (fast-oxidative, glycolytic), and smaller, darkly stained (oxidative) fibers. NADH-positive fibers were counted as fibers that were moderately or darkly stained. Roughly 300–500 fibers were counted per animal.

Body Composition Phenotyping Procedures of G9–G10 Rats

G9–G10 HVR and LVR rats were phenotyped for body composition at 28 days of age before running, as well as 34 days of age under two conditions: 1) those that voluntarily ran during 28–34 days of age (6-day runners), or 2) those that never ran (nonrunners). To prevent litter-size effects on body composition, six to eight HVR and LVR litters were culled to 10 pups per family 1 day following birth; of note, one litter had nine pups. During the light cycle, rats were killed in CO₂ chambers, body composition was obtained via dual X-ray absorptiometry calibrated for rats (Hologic, QDR), and omental fat pads were extracted and weighed.

NAc RNA Isolation From G8 HVR and LVR Nonrunners and cDNA Library Preparation for RNA Sequencing

During sacrifices for muscle phenotyping, brains were removed, and NAc tissue was extracted from a subset of seven (4 male, 3 female) G8 HVR nonrunners and eight (4 male, 4 female) G8 LVR nonrunners using a punch tool and brain sectioning apparatus (Braintree Scientific, Braintree, MA). NAc plugs were placed in TRIzol and were stored at −80°C until processing. During tissue processing, samples were lysed in TRIzol using a high-speed shaking apparatus with stainless steel beads (Tissuelyser LT). RNA was subsequently separated into phases using the TRIzol method, according to manufacturer's instructions, and isolated/DNase treated with columns (Macherey-Nagel, Bethlehem, PA). RNA integrity of each sample was checked using BioAnalyzer 2100 automated electrophoresis system (Bio-Rad, Hercules, CA) before cDNA library construction.

cDNA library preparation was performed at the University of Missouri DNA Core using the manufacturer's protocol with reagents supplied in Illumina's TruSeq RNA sample preparation kit v2. Briefly, the poly-A containing mRNA was purified from 2 μg of total RNA, RNA was fragmented, double-stranded cDNA was generated from fragmented RNA, and the index containing sample identifier adapters were ligated to the ends. The final construct of each purified library was evaluated using the BioAnalyzer 2100 automated electrophoresis system, quantified with the Qubit fluorometer using the quant-iT HS dsDNA reagent kit (Invitrogen), and diluted according to Illumina's standard sequencing protocol for sequencing on the HiSeq 2000.

Illumina sequencing of NAc cDNA and data procurement.

RNA-seq procedures occurred at the University of Missouri DNA Core and are described in more detail elsewhere (39). Briefly, following cDNA library construction, samples were loaded into flowcell where clusters of each oligo were replicated. Following this procedure, flowcells were placed in the sequencer, and fluorescently labeled bases were attached to the complementary bases of each sequence. The Illumina Genome Analyzer recorded 50 bp reads. Reads were trimmed to ensure adaptor sequence removal and tiled to a custom reference using NextGENe v1.92 (SoftGenetics, State College, PA).

For RNA-seq data procurement, RNA-seq sequences for each biological sample were trimmed of adapter sequences, and resultant sequences were aligned to a custom database consisting of rat sequences and human homologous sequences using NexGen v2.2. Our decision tree for RNA-seq bioinformatics is presented in Fig. 3, and all analytic procedures were performed using Microsoft Excel v2007. Briefly, reads per million means for the biological replicates within each group were calculated, and a 95% confidence interval was then constructed around the mean number of reads to identify the average number of reads for each database member that was statistically greater than zero. Reads per kilobase per million mapped reads (RPKM) means were then generated for the HVR and LVR 34-day-old nonrunners, and genes whereby HVR or LVR groups means were below a RPKM value of 2.0 were removed from the analysis. Differential gene expression patterns between HVR and LVR nonrunners using RPKM values were subsequently performed, and a t-test for HVR/LVR values greater than ±1.25-fold was executed. False discovery rates were calculated for each gene, and between-group differences were considered to exist at an r value < 0.10. mRNAs that met these thresholds were entered into Ingenuity Pathway Analysis Software (IPA; Ingenuity Systems, Redwood, CA) to examine which NAc gene networks differed between the HVR and LVR lines. While we were interested in testing the hypothesis that dopamine-related transcripts would be differentially expressed within the NAc of HVRs and LVRs, another broader purpose of these RNA-seq efforts was to observe other between-line NAc transcriptomic differences to generate new hypotheses for future experiments using our model. As such, Hawkins et al. (16) suggest that RNA-seq with advanced bioinformatics is an invaluable tool for hypothesis-generating experiments.

Fig. 3.

Decision-making tree for HVR vs. LVR nucleus accumbens (NAc) transcriptome comparisons. FDR, false discovery rate.

Statistics for Muscle and Body Composition Phenotyping

G8 running and feeding data were averaged per family and statistically compared. As an example, 6-day running attributes (running time, running distance, and running pace) were assessed from 24 male HVR 6-day runners, which came from five different families. Due to the fact that an unequal number of rats came from each family (2–7 rats per family in this example), five family averages were calculated for these variables and were statistically compared with corresponding averages from five LVR families. Instances whereby individual rats were used to calculate variable averages due to the fact that an equal representation of rats was used from each family were 1) skeletal muscle attributes (mRNA data and NADH fiber typing); and 2) body composition phenotyping data.

Line (HVR vs. LVR) × sex (female vs. male) × activity (voluntary run vs. no run) statistical comparisons were made using a three-way ANOVA with Holm-Sidak post hoc tests. Exceptions to these statistical methods are stated in figure legends. All variables are expressed as means ± SE. Sigmaplot 12.0 (San Jose, CA) was used for all statistical analyses, and an α value of 0.05 was adopted throughout.

RESULTS

Selective Breeding Separates Voluntary Running Distances

To chronicle the effectiveness with each successive generation of selection for high and low distances of voluntary running, Fig. 4A shows average voluntarily running on days 5 and 6 (selection variable for breeders for next generation) for G1–G9. At G9, voluntary running distance was 8.5-fold greater in male HVR vs. male LVR 6-day runners (9.3 vs. 1.1 km, P < 0.001) (Fig. 4A). Similarly, distance run on days 5 and 6 was 11.0-fold greater in female HVR vs. female LVR 6-day runners (15.4 vs. 1.4 km, P < 0.001). As theorized from recent human epidemiological data, physical inactivity can be the result of fatness rather than its cause (32). We considered that LVRs might have coselected increased body weight that could lower their voluntary running distance with each successive generation. While prerunning (day 28) body weights were either higher or lower in each sex, depending on the generation, this did not seem to be a phenotype that coincidently diverged with running patterns (Fig. 4B).

Fig. 4.

Phenotypic divergence of voluntary running. A: running on days 5 and 6 (criterion selection variable) are presented from generations 1–9 (G1–G9) in HVR and LVR males (left) and females (right). B: body weights (BW) starting at G2 are also presented. Nonbreeding founder Wistars (NonB) were used as a template for comparison. ^a-f different superscripted letters indicates a significant difference from generation to generation (P < 0.05). *Statistical difference from NonB founders. †Statistical difference between HVR and LVR within each generation (P < 0.05). The same rats are shown in A and B. Data are generated from 48–113 rats from 13 families per data point, and data are expressed as means ± SE.

While Running Distances Diverged in G8 While Food Intakes Did Not Differ

Examining 6-day running patterns of G8 male HVR and LVR 6-day runners (n = 24–29 animals from 5 families per line), HVRs ran ∼5× longer (20.6 vs. 4.1 km, P < 0.001), spent ∼4× more time running (594 vs. 143 min, P < 0.001), and ran 19% faster (33.7 vs. 28.4 m/min, P < 0.001) than their LVR counterparts, respectively (Fig. 5, A–C). Likewise, female HVR 6-day runners ran ∼3× longer (23.5 vs. 6.5 km, P < 0.001), spent ∼3× more time running (639 vs. 200 min, P < 0.001), and ran 22% faster (35.7 vs. 29.2 m/min, P < 0.001) than their LVR counterparts, respectively. The difference in running distance was maintained past days 5 and 6. While long-term surveillance of these animals was beyond the realm of the present study, a 4.7-fold higher cumulative running distance occurred over a 42-day period in HVR relative to LVR males [272 km (n = 12) vs. 57 km (n = 5); P < 0.001, data not shown] in G6, showing continued maintenance of the running phenotype with longer durations.

Fig. 5.

Total running distance, total run time running pace, BW gain, and food intake/efficiency in G8 male and female HVRs and LVRs. Male and female HVRs ran longer distances (P < 0.001; A), spent more time running (P < 0.001; B), and ran at a faster pace (P = 0.02–0.03; C) than their LVR counterparts. BWs (D) as well as food intakes and feed efficiency (E) were similar between line types. BWs for each sex were compared with two-way (line × time) ANOVAs and Holm-Sidak post hoc tests. Food intakes and efficiencies were compared with two-way (line × sex) ANOVAs and Holm-Sidak post hoc tests. Data were generated from 4–5 families (24–29 rats) per bar and are expressed as means ± SE.

Since a dose-response relationship exists for voluntary running distance and caloric expenditure (7), the question as to whether greater voluntary running would be associated with greater food intakes was addressed. Examining females, 6-day body weight changes, and feeding patterns of G8 HVR and LVR 6-day runners, both lines gained similar weight from 28–34 days of age (P = 0.15), ate the same amount of food over this period (P = 0.17), and exhibited similar feed efficiencies over this period (P = 0.34) (Fig. 5, D and E). Likewise, no differences in these variables existed between G8 male HVR and LVR 6-day runners (Fig. 5, D and E). Of note, daily food intakes were measured over a 1- to 3-day period around 34 days of age from three G8 HVR and three G8 LVR nonrunner families (14–22 rats/line). There were no differences in food intakes when comparing female HVR and LVR nonrunners (16.9 ± 3.5 vs. 15.7 ± 3.0 g/day, respectively; P = 0.80) and male HVR and LVR nonrunners (17.1 ± 2.0 vs. 18.8 ± 3.4 g/day, respectively; P = 0.68).

Skeletal Muscle PGC-1α mRNA in G8 Male and Female HVR and LVR 6-day Runners and Nonrunners

Since voluntary running is perceived to be a function of exercise capability and motivation (14), some biomarkers of skeletal muscle phenotype related to oxidative capacity were determined. We first tested whether selected metabolic adaptations were present after 6 days of voluntary running, as these rats were then available. Plantaris PGC-1α mRNA (normalized to LVR 6-day runners) was significantly greater in females only after 6 days of voluntary running in HVRs compared with LVRs (P = 0.05; Fig. 6A). Alternatively, normalized to LVR nonrunners, there was also a significant activity effect (P < 0.001), as HVR 6-day runner male and females did present significantly greater plantaris PGC-1α mRNA levels than their nonrunner counterparts (P = 0.003 and P = 0.004, respectively; Fig. 6C). No other mRNA differences existed between HVR and LVR plantaris muscles for GLUT4, VEGF, and HKII in 6-day runners (P > 0.05, Fig. 6A). Soleus PGC-1α mRNA tended to be greater in male HVRs vs. LVR 6-day runners (P = 0.08; Fig. 6B), with no differences for GLUT4, VEGF, CD36, and CPT1 mRNA expression patterns (P > 0.05, Fig. 6B). There was also a significant activity effect (P < 0.001), as HVR 6-day runner males and females did present significantly greater soleus PGC-1α mRNA levels than their nonrunner counterparts (P < 0.001 and P < 0.001, respectively; Fig. 6D).

Fig. 6.

Selected mRNA expression in plantaris and soleus muscles. Six-day runner mRNA expression patterns for plantaris (A) and soleus (B) muscles of male and female HVR and LVR are shown. All values were normalized to 6-day runner LVR, which was adjusted to 1.0. In A, female plantaris proliferator-activated receptor-γ coactivator-1α (PGC-1α) mRNA values were greater in HVR vs. LVR 6-day runners (P = 0.05) with two-way ANOVA (line × sex). In B, male soleus PGC-1α mRNA values tended to be greater in HVR vs. LVR 6-day runner (P = 0.08). PGC-1α mRNA expression patterns within the plantaris (C) and soleus (D) muscles of male and female HVR and LVR 6-day runners and nonrunners are shown. In C, plantaris PGC-1α mRNA values were higher in HVR vs. LVR 6-day runners (P = 0.003 and P = 0.004, respectively). In D, soleus PGC-1α mRNA values were greater in HVR vs. LVR 6-day runner females (P < 0.001 and P < 0.001, respectively). Data were generated from 8–10 rats (5 families, 1–2 rats per family) and are expressed as means ± SE. Unmarked bar values are 1.0 and were used for normalization. Other values are presented above bars.

Given the line-type (HVR vs. LVR) differences in 6-day runners for PGC-1α mRNA expression patterns, the next question became whether these differences were inherent to each line type without running wheels. Thus, we assayed PGC-1α mRNA in nonrunner HVR and LVR muscles to see if this gene was inherently more highly expressed in the HVR line-type without voluntary running activity (Fig. 6, C and D). No inherit differences existed for plantaris PGC-1α mRNA in either sex, although soleus PGC-1α mRNA paradoxically tended to be lower in HVR vs. LVR females (P = 0.06, Fig. 6D). Hence, the skeletal muscle PGC-1α mRNA levels in male HVR vs. LVR 6-day runners (Fig. 6, C and D) was likely an effect of greater running volume instead of HVR vs. LVR type differences.

A three-way ANOVA also revealed there was no line × activity × sex interactive effect regarding soleus PGC-1α mRNA between male and female HVR and LVR 6-day runners vs. nonrunners (P = 0.66, Fig. 6D). There was, however, a significant activity effect (6-day runner vs. nonrunner, P < 0.001), as male and female HVR 6-day runners did have significantly greater soleus PGC-1α mRNA levels than their nonrunner counterparts (P < 0.001 and P < 0.001, respectively). Nonetheless, as with the aforementioned plantaris data, the greater soleus PGC-1α mRNA levels in male HVR vs. LVR 6-day runners (Fig. 6B) was likely an effect of greater running volume instead of HVR vs. LVR-type differences.

NADH Fiber Staining of Superficial Plantaris Fibers in G8 Male and Female HVR and LVR 6-day Runners and Nonrunners

Because of observations regarding running-induced increases in PGC-1α mRNA, percentages of fibers staining with NADH were determined. No line × activity × sex interactive effect regarding plantaris NADH-positive fibers was revealed by three-way ANOVA (P = 0.22). However, a significant activity (6-day runner vs. nonrunner) effect (P < 0.001) was present within both lines. Male and female 6-day runners exhibited a significantly greater proportion of NADH-positive fibers in peripheral fibers of the plantaris muscle than in their nonrunner counterparts [for LVR line (P < 0.001 and P = 0.04, respectively, for each sex) and for HVR line (P = 0.007 and P < 0.001, respectively) (Fig. 7)]. In addition, a line-type effect (LVR vs. HVR, P < 0.001) existed in the plantaris muscle. Female LVR 6-day runners possessed a lower percentage of NADH-positive fibers vs. female HVR 6-day runners (P = 0.01). Furthermore, intrinsic differences were present for both sexes: 1) female LVR nonrunners possessed less NADH-positive fibers vs. female HVR nonrunners (P = 0.02); and 2) male LVR nonrunners possessed fewer NADH+ positive fibers vs. male HVR nonrunners (P = 0.005). Of note, inclusion of all tested animals (i.e., lines, sexes, and activity groups) produced a significant correlation between plantaris PGC-1α mRNA expression patterns vs. NADH-positive fibers (r = 0.365, P = 0.004). Soleus muscles were also stained for subsets of animals. However, given that this is a predominantly high-oxidative, slow-twitch muscle, all fibers stained positive for NADH, and thus NADH analysis was not done.

Fig. 7.

Plantaris muscle NADH fiber typing of peripheral fibers. A: 6-day runners had greater percentage of positive NADH-positive peripheral fibers than respective nonrunners from same litter. *HVR nonrunner > LVR nonrunner. B: cross-sectional views of HVR and LVR nonrunners. Data were generated from 7–10 rats (5 families, 1–2 rats per family) and are expressed as means ± SE.

Body Composition and Tissue Weights in G9–G10 HVR and LVR Rats

After completion of G8, we became aware that rat litter size could potentially cause significant differences in fat mass at 24 days of age (45). Because of G8's experimental design not controlling or recording litter size during mother's nursing, body composition and tissue weights were repeated in G9–G10 pups after culling litter sizes to 10. G9–G10 LVR male (P < 0.01, Fig. 8A) and female (P < 0.001) nonrunners were heavier at day 28 and tended to be heavier at day 34 (P = 0.06 for both sexes). However, there were no differences in body dual-energy X-ray absorptiometry percent body fat between male and female HVR and LVR rats (Fig. 8B), suggesting that we did not select inherently fatter rats coincident with the respective running phenotypes.

Fig. 8.

BW (A), dual-energy X-ray absorptiometry (DXA) body fat percentages (B–D), and omental adipose tissue (E and F) for male and female G9–G10 HVR and LVR 6-day runners and nonrunners. A: at day 28, LVR males weighed more than HVR males (P < 0.001), and LVR females weighed more than HVR females (line above bars, P < 0.001). This became a trend in both sexes at day 34 in LVR and HVR animals that never ran. Asterisks note that 6-day LVR male, HVR male, and LVR female runners weighed less than nonrunners (*P < 0.05; ***P < 0.001). B: there were no differences in body DXA percent body fat between male and female HVR and LVR rats. As with BWs, 6-day LVR male, HVR male, and LVR female runners possessed less %body fat than nonrunners (**P < 0.01; ***P < 0.001). C: both lines and sexes of nonrunners equally gained body fat mass from days 28–34. Importantly, running in 6-day runner LVR males (**P < 0.01) and females (*P < 0.05) prevented gains in fat mass over this period and caused fat mass loss in 6-day runner HVR males (***P < 0.001) and females (P < 0.001). D: combining sexes to increase the n-sizes of LVR (to n = 11) and HVR (to n = 12) runners suggests that longer running distances in the HVR runners caused a significant reduction in fat mass (P = 0.03). E and F: there were no between-line differences in omental fat pad masses of G9–G10 male and female HVR and LVR nonrunners on an absolute (E) or a mg/g BW (F) basis. All HVR and LVR 6-day runners possessed significantly less omental fat mass compared with their sedentary counterparts (*P < 0.05; **P < 0.01; ***P < 0.001). Data were generated from 6–8 families (11–16 animals) per bar for nonrunners and 4–7 families (4–7 animals) per bar for runners and are expressed as means ± SE. While not shown in the figure, HVR females ran a total of 33.5 ± 2.0 km from days 28–34, LVR females ran a total of 5.4 ± 1.7 km, HVR males ran 14.3 ± 3.4 km, and LVR males ran a total of 2.1 ± 0.5 km. NS, nonsignificant.

Nonrunners in both lines and sexes equally gained body fat mass from days 28–34 (Fig. 8C), further supporting the notion that there are no between-line (HVR vs. LVR) differences in adiposity and/or growth rates in fat mass. Importantly, running in 6-day runner LVR males (P < 0.01) and females (P < 0.05) prevented gains in fat mass over this period and caused fat mass loss in 6-day runner HVR males (P < 0.001) and females (P < 0.001), compared with prerunning values. When combining sexes to increase the n-sizes of LVR and HVR runners, longer running distances in the HVR runners caused a significant reduction in fat mass at day 34 from day 28, while LVRs had no gain or loss in the same time period (P = 0.03, Fig. 8D). When examining omental fat pad masses of G9–G10 male and female HVR and LVR nonrunners on a milligram per gram body weight basis, there were no between-line differences (Fig. 8F). As with positive shifts presented in total fat mass with wheel running, all HVR and LVR 6-day runners possessed significantly less omental fat mass compared with their sedentary counterparts (Fig. 8, E and F).

NAc mRNA Differences Between G8 HVR and LVR Nonrunners

Using our previously defined thresholds for RNA-seq analysis (Fig. 3), we determined that 36 NAc mRNAs were differentially expressed between HVR and LVR nonrunners (of 9,889 mRNAs; Fig. 9).

Fig. 9.

Differential NAc mRNA expression patterns between 34-day-old G8 HVR nonrunners (4 male, 3 female) and G8 LVR nonrunners (4 male, 4 female). Thresholds: nonsex biased (P > 0.05 between sexes), reads per kilobase per million mapped reads values > 2.0 for HVRs and LVRs, and r-values < 0.10 of mRNAs that were ±1.25-fold. 36 of 9,889 transcripts met above thresholds (27 upregulated and 9 downregulated in HVR/LVR comparison).

The top associated network as defined by IPA for NAc mRNAs upregulated in HVRs vs. LVRs included “cell morphology, cell death and survival, dermatological diseases and conditions” (11/35 pathway molecules: ↑ARHGAP18, ↑C11orf16, ↑C18orf54, ↑CFL1, ↑ECDHC2, ↑HLA-A, ↑HLA-E, ↑LYN, ↑MLH1, ↑PAFAH1B3, ↑RBM3) (Fig. 10A). The top associated network as defined by IPA for NAc mRNAs downregulated in HVRs vs. LVRs included “nervous system development and function, cell signaling, and molecular transport” (5/35 pathway molecules: ↓CAMKV, ↓OPRD1, ↓REXO4, ↓SLC24A4, ↓TBCD) (Fig. 10B).

Fig. 10.

A: the top-associated network as defined by Ingenuity Pathway Analysis (IPA) for NAc mRNAs upregulated in HVRs vs. LVRs included “cell morphology, cell death and survival, dermatological diseases and conditions” (11/35 pathway molecules: ↑ARHGAP18, ↑C11orf16, ↑C18orf54, ↑CFL1, ↑ECDHC2, ↑HLA-A, ↑HLA-E, ↑LYN, ↑MLH1, ↑PAFAH1B3, ↑RBM3). B: the top-associated network as defined by IPA for NAc mRNAs downregulated in HVRs vs. LVRs included “nervous system development and function, cell signaling, and molecular transport” (5/35 pathway molecules: ↓CAMKV, ↓OPRD1, ↓REXO4, ↓SLC24A4, ↓TBCD, ↓PRRT1).

DISCUSSION

The aim of the present study was to selectively breed rats having both HVR and LVR distances. No model of selective breeding exists, to our knowledge, which has generated rats that display low daily voluntary physical activity levels. Such a model might more closely mimic the large percentage of the US population with low level of daily physical activity (44). As human physical activity is a polygenic condition (29), we and others (25, 35) contend that a selectively bred model of low daily physical activity will be more realistic to the biological basis of human sedentarism than would a transgenic model of lazy behavior produced by knockout or overexpression of a single gene. We also chose to pursue a model utilizing voluntary wheel running to study potential motivational differences for low physical activity given the following commentary by two reports: 1) Leasure and Jones (26) wrote: “…studies suggest that forced and voluntary exercise may not be equivalent in their effects on the brain and behavior”; and 2) Fleshner et al. (13) stated: “. . . forced treadmill running produces equal or greater peripheral metabolic changes [relative to voluntary wheel running] but minimal stress-buffering adaptations.” In the present study, initial phenotyping is reported of our unique LVR line that voluntary runs 20% (male) and 28% (female) of the distance run by the HVR line at G8 (Fig. 5), both derived from the same founder population.

Three translational rationales to public health existed for our development of an animal model that consists of a line of voluntary runners. First, the historical levels of human physical activity have decreased from ∼14,000–20,000 steps/day in the Paleolithic era to ∼5,000 steps/day today (34). Second, current US levels for daily physical activity, assessed with accelerometry, are less than US governmental guidelines, as Troiano and coworkers (44) have reported that ∼48% of children (6–11 yr old) and ∼92% of adolescents (12–19 yr old) are “physically lazy” and do not engage in the US government's recommended >60 min of daily physical activity. In the same report, ∼97% of individuals >20 yr of age did not have >30 min daily physical activity for the US recommended daily physical activity (44). Third, low levels of physical activity is clinically associated with increased risks of 35 unhealthy conditions and premature death (5). We contend that the neurobehavioral basis for low physical activity promises to be an impactful translational research inquiry and believe that our selectively bred LVR rats can potentially be used to further study low motivation for voluntary running and any other phenotype coselected along with this trait.

Given that selective breeding of complex phenotypes, such as voluntary running, coselects other phenotypes (8, 40), a selected few of the potential multiple phenotypes were examined in the present study. Since excess fat can lead to physical inactivity (32), body and fat masses were determined. In this regard, parents of G8 pups were bred to produce a second litter of offspring that were housed in cages without voluntary running wheels so they never voluntarily ran [termed “nonrunners” to distinguish them from the first litter, which voluntarily ran between 28 and 34 days of age (termed “6-day runners”)]. One additional purpose in studying nonrunners was to identify any inherent phenotypes that existed between the LVR vs. HVR lines, which were a result of the selection process and were independent of changes associated with 6 days of running. While LVR nonrunners were inherently heavier than HVR nonrunners in G9–G10, body fat percentages were similar between these animals (Fig. 8), providing evidence that increased fat mass alone was not a factor driving lower voluntary running distances. Interestingly, an increased body weight phenotype in LVR animals was identified, although body fat between lines was similar. Current literature supports an association between an increased body fat and decreases in physical activity. For instance, Swallow et al. (Supplemental Fig. S2 and Table S1 in Ref. 43) report that every 2% increase in visceral adiposity (epididymal and perimetrial) is marked by an approximate 80-m decrement in running to volitional fatigue in 6-mo-old rats. Similarly, male and female LCR rats from Britton's group weighed 92 and 44 g more (39 and 24%), respectively, than HCR males in the 11th generation (43). According to Wisloff et al. (47), body weight explained 7–20% of the variations in the distances run by LCR and HCR females and males during the run to operational exhaustion. In the present study, it should be noted that, in G9–G10 34-day-old nonrunners, LVR females were only 5.5% heavier than HVR females, and LVR males were 5.1% heavier than HVR males. In contrast, pups in both HVR sexes of G6–7 were heavier than LVR pups, despite the substantially greater running phenotype in the former group (Fig. 4B). Hence, together, these findings suggest that body mass is not yet a convincing driving factor in producing the differential running phenotype evident between HVR and LVR rats. This finding is not too surprising given that, while larger body weight differences do exist between Britton's HCR and LCR lines (43), other rodent models of voluntary running are not demarcated by such drastic line-type body weight differences (10, 28).

We posited that declines in voluntary running with artificial selection could also be a result of coselection of lower metabolic capability. In this regard, phenotypic findings for skeletal muscle metabolic indexes were mixed. No intrinsic differences were observed between LVR and HVR for plantaris or soleus muscle PGC-1α mRNA, a transcriptional coactivator that controls the expression of genes involved in oxidative metabolism (27). Furthermore, no differences in GLUT4 (glucose uptake marker) or VEGF (angiogenic marker) mRNAs were found in plantaris or soleus muscles between LVR and HVR rats after 6 days of voluntary running, so intrinsic determinations of these mRNA were not performed in animals never having voluntary running. On the other hand, an intrinsic increase in the percentage of NADH-positive fibers was present in the superficial (mixed fiber) portion of the plantaris muscles of HVR, compared with LVR rats. Taken together, we speculate that any lower intrinsic oxidative status of LVR could have had a marginal contribution to their LVR.

The mesolimbic dopaminergic pathway plays a major role in determining voluntary running motivation (20–22, 46). With regard to our model, our laboratory has previously demonstrated that our HVR but not LVR rats from G4–G5 had voluntary running distances decreased by separate injections of a dopamine D1 receptor agonist and antagonist into the NAc (38). Likewise, we have observed that various dopamine receptor mRNAs (DRD1/DRD2/DRD5) tended to be or were statistically lower in the NAc of G6 LVR 6-day runners compared with G6 HVR 6-day runners (unpublished observations). Therefore, we examined the NAc transcriptome of G8 HVR vs. G8 LVR animals that did not have access to running wheels to test our pre hoc hypothesis, as well as generate future hypotheses with our model. We hypothesized that mRNAs related to dopamine signaling pathways would be inherently differentially expressed between these two lines. Surprisingly, there were very subtle differences between the HVR and LVR NAc transcriptomes, with only eight transcripts being greater than 1.5-fold different between lines (Fig. 9). Hence, this may suggest that subtle alterations in the NAc transcriptome may largely contribute to high vs. low voluntary physical activity patterns. The top associated network, as defined by IPA for NAc mRNAs upregulated in HVRs vs. LVRs, included “cell morphology, cell death, and survival.” The top associated network as defined by IPA for NAc mRNAs downregulated in HVRs vs. LVRs included “nervous system development and function, cell signaling, and molecular transport.” Surprisingly, when examining genes in the dopamine signaling pathway (GO:0007212), adenylate cyclase-activating dopamine receptor pathway (GO:0007191), adenylate cyclase-inhibiting dopamine receptor pathway (GO:0007191), and/or the negative regulation of dopamine receptor signaling pathway (GO:0060160), only the delta opioid receptor (Oprd1) was differentially expressed between HVRs and LVRs (HVR/LVR = −1.3-fold, false discovery rate q-value: 0.065). Interestingly, Greenwood et al. (15) found that 6 wk of voluntary wheel running increased Oprd1 transcript levels in the NAc of rats when comparing them to sedentary-housed counterparts, an effect that was also related to an increase in running motivation when conditioned place preference assays were performed 2 wk after wheel removal. Thus an inherent reduction in opioid signaling in HVRs may fall in line with the hypersensitivity to reward hypothesis, suggesting that HVRs run more than LVRs to experience a similar fulfillment in running reward (30). However, it also remains possible that genes and pathways possibly involved in cell turnover, and not dopamine signaling pathways in the NAc, may be an inherent driver in running phenotypes between HVRs vs. LVRs. This finding is quite intriguing, given that adult neurogenesis, while known to be increased in the dentate gyrus in response to voluntary exercise (33, 48), has been less investigated in the NAc.

According to Dayer et al. (9), “Ongoing neurogenesis in the adult mammalian dentate gyrus and olfactory bulb is generally accepted, but its existence in other adult brain regions is highly controversial.” Notwithstanding, the same report demonstrated that bromodeoxyuridine injections in 9- to 10-wk-old rats revealed that a “vast majority of the new neurons in the striatum” (i.e., bromodeoxyuridine+ cells) existed in the NAc 4–5 wk postinjection. Inta et al. (17) also recently proposed that postnatal neurogenesis can occur in granule neurons that are 1) DRD3 rich, 2) GABAergic, and 3) border the NAc. Therefore, future studies will be needed to determine whether 1) postnatal neurogenesis of DRD3-containing neurons in or near the NAc in HVRs relates to a possible predisposition to be highly motivated to undertake voluntary physical activity, and/or 2) if perturbing one or a number of these genes and/or pathways can increase voluntary wheel running in LVRs, and/or 3) whether amygdala, ventral hippocampus, prefrontal cortex, and/or thalamus are upstream regulatory sites for physical activity that differentially signal the NAc in our HVR vs. LVR rats (Fig. 1 in Ref. 6).

Equally important is the downregulated pathway observed in HVRs, whereby the Oprd1 was a downregulated node in this pathway. Infusion of delta opioid agonists into the NAc has been shown to transiently increase locomotion (41), and Oprd1 have been shown to increase dopaminergic neurotransmission (1). Therefore, it remains possible that HVR and LVR rats still may have an inherently altered NAc dopaminergic signaling networks, respectively.

Short-term exposure to voluntary running had a profound inhibitory effect on growth of adipose tissue mass in juvenile rats. While 28-day-old HVR 6-day runners lost more body fat compared with LVR runner counterparts, LVR 6-day runners prevented gains in total body fat mass, despite their running much less and spending much less time in running wheels (Fig. 8). This finding has important public health ramifications, as it suggests that even minimal physical activity can exhibit profound alterations in body fat percentage in growing rats. Given that these observations were made during a 6-day period of rapid body growth during juvenile states, and in lieu of our laboratory's recent review highlighting the importance of physical activity in young rats (37), as well as our laboratory's recent data in mice (7), the present observation raises the translational question as to how much a sedentary lifestyle in human children is associated with changes at the cellular level that predispose a life course toward obesity. Hence, longer term running studies will unveil how long-term body composition measures are affected in HVR and LVR rats compared with their sedentary counterparts.

It should be noted that the HVR and LVR lines could possess underlying neuro-molecular mechanisms which differentially regulate running phenotypes compared with control (nonselectively bred) rats. Therefore, future studies will be needed to determine whether the NAc transcriptome, for instance, differs between LVR or HVR vs. control, outbred Wistar rats. Likewise, we have assumed that these two lines possess different running motivations based on the large divergence in daily voluntary running patterns. However, more thorough behavioral investigations, for instance conditional placed preference tests as done previously (15), will need to be completed to confirm whether or not physical activity motivations differ between these lines. Nonetheless, this study provides the first comprehensive evidence as to how our unique HVR and LVR rats inherently differ at numerous phenotypic levels.

Perspectives and Significance

While it is fairly well known that physical activity is beneficial for health, almost all US teenagers and adults perform less physical activity than US guidelines recommend. Multiple factors contribute to the motivation to be physically activity, including genes. The unique polygenic animal model of LVR, characterized here, can be employed in future studies, which continue to dissect out differential gene expression patterns between selectively bred LVR and HVR rodents. Eventually, these targets may be able to be manipulated at the gene level to establish more precise cause-and-effect relationships for mechanisms that regulate daily physical activity levels.

GRANTS

Partial funding for this project was obtained from a grant awarded to F. W. Booth by the College of Veterinary Medicine at the University of Missouri, National Institute of Arthritis and Musculoskeletal and Skin Diseases Grant T32-AR048523 (M. D. Roberts), and American Heart Association Grant 11PRE7580074 (J. M. Company). This project was also supported by funds donated through the College of Veterinary Medicine's Development Office.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

Author contributions: M.D.R., J.D.B., J.M.C., L.P.O., A.J.H., C.L.C., J.A.K., T.E.C., M.B., and F.W.B. conception and design of research; M.D.R., J.D.B., J.M.C., L.P.O., A.J.H., R.G.T., K.D.W., C.L.C., J.A.K., and J.A.F. performed experiments; M.D.R., R.G.T., K.D.W., J.A.K., and J.A.F. analyzed data; M.D.R., J.D.B., J.M.C., L.P.O., A.J.H., R.G.T., K.D.W., C.L.C., J.A.F., T.E.C., M.B., and F.W.B. interpreted results of experiments; M.D.R. prepared figures; M.D.R. drafted manuscript; M.D.R., L.P.O., A.J.H., R.G.T., K.D.W., C.L.C., J.A.F., T.E.C., M.B., and F.W.B. approved final version of manuscript; J.D.B., J.M.C., and F.W.B. edited and revised manuscript.