The exercise sex gap and the impact of the estrous cycle on exercise performance in mice

Exercise physiology is different in males and females. Females are poorly studied due to the complexity of the estrous cycle and this bias has created an exercise sex gap. Here, we evaluated the impact of sexual dimorphism and of the estrous cycle on muscle strength and running power of C57BL/6 mice. Like men, male mice were stronger and more powerful than females. Exercise-induced increase of O2 consumption (V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}O2) and CO2 production (V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}CO2) were equal between sexes, indicating that running economy was higher in males. Thermoregulation was also more efficient in males. In females, proestrus increased exercise V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}O2 and V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}CO2 at low running speeds (30–35% female V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}O2max) and estrus worsened thermoregulation. These differences translated into different absolute and relative workloads on the treadmill, even at equal submaximal V̇\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\dot{{\bf{V}}}$$\end{document}O2 and belt speeds. In summary, our results demonstrate the better muscle strength, running power and economy, and exercise-induced thermoregulation of males compared to females. Proestrus and estrus still undermined the running economy and exercise-induced thermoregulation of females, respectively. These results demonstrate an important exercise sex gap in mice.

Although the absolute exercise performance of males was higher, the submaximal comparisons indicated a different conclusion. The ergometric test applied progressive running speeds for males and females through serial acceleration (F 21,310 = 3.2, P < 0.05; Fig. 2E). The treadmill running power in males and females was statistically similar up to 15 m/min (F 21,310 = 3.2, P > 0.05; Fig. 2E), when the relative intensity was 50 ± 3.7% of the maximum power for females, and 35 ± 3.9%% for males. The lower running power of females appeared at speeds 18 → 30 m/ min (F 28,252 = 18.1, P < 0.05; Fig. 2E, gray area). At 30 m/min, the maximum overload of females (100 ± 5.7%) corresponded to a relative overload of males (71 ± 2.2% of maximum). Males reached maximum overload at speeds 39 → 42 m/min (Fig. 2E).
The normalization of exercise performance by body mass eliminates sexual dimorphism. We then normalized the exercise performance by the body mass. This transformation eliminated the sex differences for muscle strength (F 4,32 = 0.78, P > 0.05; Fig. 2D) and running power at speeds 15 → 30 m/min (F 4,32 = 0.63, P > 0.05; Fig. 2G). Males show a better running economy. There were no differences in V  O 2 and V  CO 2 kinetics between sexes. The progressive running speeds of the ergospirometry increased the V  O 2 (F 7,126 = 2.8, P < 0.05; Fig. 3A) and  VCO 2 (F 7,126 = 2.4, P < 0.05; Fig. 3E) of males and females at all comparative intensities (9 → 30 m/min, Fig. 3A and E). Thus, the higher submaximal running power developed at speeds 18 → 30 m/min, associated to the same submaximal V  O 2 , showed a better running economy in males compared to females at ( Fig. 3A and E, gray area).
Exercise-induced thermoregulation is less effective in estrus females. Thermoregulation requires the dissipation of heat produced during exercise. Exercise increased the heat production of males and females (F 7,126 = 264, P < 0.05; Fig. 4A), without influence of the estrous cycle (F 7,94 = 0.32, P > 0.05; Fig. 4B). Environment temperature and humidity did not interfere in the thermography results, since they were similar before and after the exercise test session (Fig. 4C). The thermal image (Fig. 4D) shows a female at rest, with the body and tail heated after a maximum exercise test (Fig. 4E).
The temperature scores ( Fig. 4H and I) reinforced the prominent exercise-induced hyperthermia of females at estrus. Estrus female body heating was larger than that of males and females in other cycles (F 4,43 = 3.3, P < 0.05; Fig. 4H). The tail warming of estrus females was superior to males and females at metestrus after exercise (F 4,43 = 2.3, P < 0.05; Fig. 4H).

Discussion
Sex matters. Sexual dimorphism and the estrous cycle influenced exercise performance and metabolism of mice implying that these factors should be considered in experimental designs and data interpretation involving exercise biology. We showed that males were stronger and more powerful than females at moderate-high intensities of exercise, when evaluating strength and running. Since submaximal and maximum overloads of exercise were different for males and females, but submaximal  VO 2 and  VCO 2 were similar, this means that the running economy of females was lower than that of males. The estrous cycle did not influence muscle strength, but undermined the running economy and exercise-induced thermoregulation.
Size matters. The sex-related exercise differences disappeared after normalization of exercise performance by size (body mass). This had already been described for muscle strength [17][18][19] , but not for running power and economy. However, body size and muscle strength are well-known secondary sexual characteristics, influenced primarily by the anabolic action of the hormone testosterone, a major determinant of sexual dimorphism 20 .
Skeletal muscle mass and strength are lower in females 19,21 . Likewise, normalization of exercise performance by specific muscle mass (rather than body mass) makes sexual dimorphism disappear 19,21 . Male mammals are larger, with larger cross-sectional muscle area 8,10 . Several studies also showed that muscle length (and the length of the long bones) is also higher in male mammals, important for greater tetanic strength of the anterior masseter muscle 8 . Larger levers determine higher torques and muscle strength. Sex is also important for muscle fiber-type composition, especially the myosin IIB gene (fast muscle fiber) 10 . Evidence shows threefold more IIB muscle fibers in the masseter of male mice 8,22 . In addition, testosterone signals hypertrophy in this musculature 20 . Conversely, estrogen decreases muscle contractile force in female mice 23,24 . Thus, muscle strength and running power depends on size and sex: males have large muscles and bones, responsible for great muscle strength; this difference is further amplified by the anabolic effects of testosterone, resulting in larger muscle strength, speed and power.
The testosterone also seems to influence running endurance, but not the running economy. Castration of mouse testicles deplete blood testosterone and impair running wheel endurance (10-30% males with intact gonads) 25 , a model of submaximal physical activity. Testosterone replacement completely reversed this impairment 25 . The antiandrogen Flutamide decreases the treadmill endurance of rats, but does not change  VO 2max and running economy 26 . Here, the exercise-induced submaximal  VO 2 and  VCO 2 up to V  O 2max were similar between sexes, as previously described 11,12,27 . Only one study demonstrated increased female submaximal V  O 2 during treadmill test, which further reinforces the hypothesis of females' worst running economy 27 . These testosterone evidences support the best physical performance (power and endurance) of running male mice, but not the best running economy. On the other hand, estrogen seems to influence  VO 2 and possibly the running economy of mouse. Similar to our results, submaximal  VO 2 was higher in female rats during the estrogen-dominant proestrus at low treadmill speeds 5-12 m/min (6° grade, without acceleration) 17 , which may be considered as a low intensity exercise. We also found these differences at near speeds 9-12 m/min. A possibility is the effect of estradiol in the lung gas-exchange surface area (GSA).  VO 2 is directly proportional to GSA 28,29 ; which increases during proestrus with high estradiol levels 28,29 . Estradiol also increases lung's GSA and V  O 2 in ovariectomized rats 29 . Our results suggest that estrogen can increase V  O 2 during exercise, and worsen the running economy, especially at proestrus.
Exercise-induced hyperthermia is a biological response due to greater muscle activation, mitochondrial uncoupling and proton leak 30 . We now report that sex and the estrous cycle do not modify the calories consumed by exercise, another important variable for running economy; however, our results showed that male thermoregulation was more efficient, since the infrared dissipation of males was more effective. Literature suggests two important points for mouse thermoregulation: body surface area (BSA) and tail dry heat loss. BSA is estimated by the Meeh's formula (BSA = body weight 0.667 ) 25 . The greater body mass of males assists in better heat dissipation during/after exercise. Moreover, tail size seems to be related to thermal stress [14][15][16]31 , with animals that live in warm environments having longer tails 15,32 . Female tails, even longer, warmed up more during exercise than that of males. The tail length of C57BL/6 female mice was similar to that described in female BALB/c mice 15 . Thus, a longer tail length in female mice is suggestive of a required adaptation to compensate for their lower body mass (and area).
Sanchez-Alavez 33 demonstrated that body warming during exercise was higher in female mice at estrous. We saw it in the tails. Progesterone promotes heat conservation and higher body temperatures at rest 34,35 . Bilateral ovariectomy eliminated this estrous-associated change 14,33 . We suggest that this may apply to body temperature of running female mice during estrus, characterized by high progesterone levels. Thus, sex seems to be a crucial factor also for the exercise-induced thermoregulation of mice. Some of our results are similar to those reported in humans, since the physical performance of women is generally lower than in men, in accordance with the exercise gender gap 2,36,37 . The woman's menstrual cycle is divided into three phases: follicular, ovulation and luteal. The follicular phase can be divided into initial and late, corresponding to metestrus and diestrus, respectively. Ovulation corresponds to proestrus, and the luteal phase to estrus. The woman's follicular and luteal menstrual cycle does not seem to influence muscle strength, power, and V  O 2 1,38-40 . Human studies still allow evaluating rate of perceived effort (RPE), which also does not differ in the different menstrual phases 38,41 . However, the differences we found are close to ovulation, virtually impossible to evaluate in women. We demonstrated that the mouse proestrus (or human ovulation) increased  VO 2 and heat production at light exercise.
In summary, our results highlight differences in exercise performance and metabolism between male and female mice. Sex influences size, which appear to be the main factor for mice exercise sex gap. Mouse sexual dimorphism also influenced exercise workload, but not V  O 2 and  VCO 2 , implying a finest running economy in males. Males also presented better thermoregulation after exercise. The estrous cycle played a subtle role in mouse physical performance: proestrus impaired running economy and estrus impaired exercise heat loss. This implies that the impact of the estrous cycle on the performance of females should not be considered a limiting factor for their use in experimental designs. In fact, size is the main factor that should be considered in the construction of experimental designs involving exercising male and female mice. For running, a light-intensity exercise seems similar between the sexes (except proestrus), but the performance of females at moderate-intensity running corresponded to the performance of males at low-moderate intensity; the performance of females at high-intensity running corresponded to the performance of males at moderate-high for males, and male high-intensity running was supra-maximal for females. Failure to consider these differences by measuring only the running speed, as done in most studies, introduces an error to compare performance between sexes. These results are of particular interest to counteract the underrepresentation of females in exercise experimental designs.

Animals.
Male and female C57BL/6 mice (10-12 weeks old) were obtained from Charles River (Barcelona, Spain). Mice were housed under controlled environment (12 h light-dark cycle, lights on at 7:00 AM, and room temperature of 21 ± 1 °C) with ad libitum access to food and water. Animals were housed and handled according to European Union guidelines and the study was approved by the Ethical Committee of the Center for Neuroscience and Cell Biology (University of Coimbra).
The animals were accustomed to the treadmill for 3 days. The open field or grip strength test was performed on the 4 th day in independent groups of animals. Ergospirometry was performed on the 5 th day. All tests were carried out between 9:00 and 17:00 hours in a sound-attenuated and temperature controlled observation room under low-intensity light (≈10 lux), where mice had been habituated for at least 1 hour. The apparatuses were cleaned with 10% ethanol between animals. Within the time window of the tests, we did not record any significant impact of the time of day (morning vs. afternoon) on the treadmill vertical power,  VO 2max and temperature of the tail at rest in either males or females (data not showed).
Vaginal cytology. We evaluated the estrous cycle immediately after the behavioral and exercise experiments, through 4-5 consecutive vaginal lavages (with 40-50 μL of distillated H 2 O) then mounted on gelatinized slides (76 × 26 mm). These procedures lasted no more than 3-5 minutes, and there were no major temporal delays between behavioral experiments and fluid collection for vaginal cytology.
The vaginal smear were desiccated at room temperature and covered with 0.1% crystal violet for 1 min, then twice washed with 1 mL H 2 O and desiccated at room temperature. The slides were mounted with Eukitt medium (Sigma-Aldrich) and evaluated under an optical microscope at 1x, 5x and 20x (Zeiss Axio Imager 2). The characterization of the estrous cycle was performed according to literature 20,42 . Females were categorized for initial (metestrus) or late (diestrus) follicular phase, ovulation (proestrus), or luteal phase (estrus) 20,42 . Open field. The exploration of an open field (38 × 38 cm) was analyzed for 15 min using the ANY-maze ™ video tracking system (Stoelting Co.) 41 .
Grip strength. The animal was hung with its forepaws to the central position of a 300 g metal grid and the grip strength was determined as the weight pushed (in grams) 41 . The computed result was the average of 3 trials, expressed in kgf.
The ergospirometry test was carried out on 5 th day, 48 hours after the last habituation session. The incremental protocol started at 15 cm/s with an increment of 5 cm/s every 2 min, with a constant inclination of 8.7% (5° for the LE8710 model). The exercise test lasted until running exhaustion, defined by the inability of the animal to leave the electrical grid for 5 seconds 43,44 . We estimated the power output for treadmill running based on a standard conversion of the vertical work, body weight and running speed 45,46 . Power is the 1 st derivative of work relative to time (run time at each stage).
Oxygen uptake (  VO 2 ) and carbon dioxide production (  VCO 2 ) were estimated during treadmill running in a metabolic chamber 47 (Gas Analyzer ML206, 23 × 5 × 5 cm, AD Instruments, Harvard) coupled to treadmill. The animals remained in the chamber for 15 min prior to exercise testing. Atmospheric air (≈21% O 2 , ≈0.03% CO 2 ) was renewed at a rate 120 mL/min, using the same sampling rate for the LASER oxygen sensor (Oxigraf X2004, resolution 0.01%) and infrared carbon dioxide sensor (Servomex Model 15050, resolution 0.1%). Heat (calories) was estimated according to the equations of Lusk 48 . Thermal imaging. An infrared (IR) camera (FLiR C2, emissivity of 0.95, FLiR Systems) placed overtop (25 cm height) of a plastic tube (25 cm diameter) was used to acquire a static dorsal thermal image 49 . IR images were taken immediately before and after exercise tests, namely at rest (Fig. 4D) and recovery (REC, Fig. 4E) periods, respectively. IR images were analyzed with FLiR Tools software (Flir, Boston).
Statistics. Data are presented as mean ± Standard Error of the Mean (SEM). A test for normality was performed by Kolmogorov-Smirnov test. For each test, the experimental unit was an individual animal. The frequency of the estrous cycle was assessed using the Kruskal-Wallis test. The role of sex and estrous cycle in the dependent variables body mass, open field, grip strength and vertical power, V  O 2 and  VCO 2 , and body and tail temperature was evaluated using on-way ANOVA. The repeated measures of ANOVA were performed to evaluate the effect of different treadmill speeds, sex and estrous cycle on the vertical power, V  O 2 and  VCO 2 , and heat. The Bonferroni post hoc test was applied for significant F values. The accepted level of significance was p < 0.05. Statistics were performed using Dell Statistica (data analysis software system), version 13. Data availability. The datasets generated and analyzed during the current study are available from the corresponding author on reasonable request.