Autophagy regulates the release of exercise factors and their beneficial effects on spatial memory recall

Abstract

Exercise promotes learning and memory recall as well as rescues cognitive decline associated with aging. The positive effects of exercise are mediated by circulatory factors that predominantly increase Brain Derived Neurotrophic Factor (BDNF) signaling in the hippocampus. Identifying the pathways that regulate the release of the circulatory factors by various tissues during exercise and that mediate hippocampal Mus musculus Bdnf expression will allow us to harness the therapeutic potential of exercise. Here, we report that two weeks of voluntary exercise in male mice activates autophagy in the hippocampus by increasing LC3B protein levels (p = 0.0425) and that autophagy is necessary for exercise-induced spatial learning and memory retention (p < 0.001; exercise + autophagy inhibitor chloroquine CQ versus exercise). We place autophagy downstream of hippocampal BDNF signaling and identify a positive feedback activation between the pathways. We also assess whether the modulation of autophagy outside the nervous system is involved in mediating exercise's effect on learning and memory recall. Indeed, plasma collected from young exercise mice promote spatial learning (p = 0.0446; exercise versus sedentary plasma) and memory retention in aged inactive mice (p = 0.0303; exercise versus sedentary plasma), whereas plasma collected from young exercise mice that received the autophagy inhibitor chloroquine diphosphate failed to do so. We show that the release of exercise factors that reverse the symptoms of aging into the circulation is dependent on the activation of autophagy in young animals. Indeed, we show that the release of the exercise factor, beta-hydroxybutyrate (DBHB), into the circulation, is autophagy-dependent and that DBHB promotes spatial learning and memory formation (p = 0.0005) by inducing hippocampal autophagy (p = 0.0479). These results implicate autophagy in peripheral tissues and in the hippocampus in mediating the effects of exercise on learning and memory recall and identify DBHB as a candidate endogenous exercise factor whose release and positive effects are autophagy-dependent.

Significance statement

Exercise promotes learning and memory formation and reverses the cognitive deficits observed during aging. These effects are mediated by circulatory factors that induce BDNF in the hippocampus. Understanding how these circulatory factors are released and deciphering the molecular mechanisms that mediate the positive effects of exercise will allow us to design therapeutic strategies that can mimic exercise. We describe a dual role for autophagy in mediating the release of the exercise factors into the circulation and in promoting spatial learning and memory recall. Our work identifies beta-hydroxybutyrate as one exercise factor regulated by autophagy and open the field for modulating autophagy as a component of the “exercise pill” and identifying novel factors that are regulated by it.

1. Introduction

Exercise improves learning and memory retention [[1], [2], [3]] and rescues symptoms associated with neurodegeneration [[4], [5], [6], [7]]. The positive changes observed in the brain after exercise are mediated by circulatory factors that are predominantly released by the liver, muscle and bones [8, 9]. These circulatory exercise factors modulate signaling pathways that result in the activation of brain-derived neurotrophic factor (BDNF) expression in the hippocampus [[10], [11], [12], [13]]. Indeed, blocking BDNF signaling in the hippocampus inhibits exercise-induced learning and memory formation [2, 14, 15]. One of the circulatory exercise factors is beta-hydroxybutyrate (DBHB). DBHB which is released by the liver after exercise and acts as a class I histone deacetylase (HDAC) inhibitor in the hippocampus [16]. DBHB activates Bdnf gene transcription by blocking HDAC2/HDAC3 activity at its promoter [16]. FNDC5 and lactate, which are released by the muscle after exercise, also activate hippocampal Bdnf expression thereby promoting learning and memory retention [17, 18]. Even though there has been recent progress in identifying the circulatory factors released after exercise, the pathways regulating their release are not very well described. Moreover, the molecular pathways that are involved in the activation and transmission of BDNF signaling in the hippocampus also remain elusive.

Autophagy is a catabolic process that regulates protein homeostasis and organelle turnover. Autophagy promotes neuronal survival [19] and extends lifespan [20]. Defects in autophagy are observed in neurodegenerative disorders and during aging [[21], [22], [23], [24]]. Autophagy is decreased in old mice and activating it rescues deficits in learning and memory recall [25]. Indeed, pharmacological inhibition of autophagy in the hippocampus disrupts long-term spatial memory retention [26], novel object recognition and contextual fear conditioning memories [25], as well as impairs associative olfactory memory [26]. Coupling between learning-induced translation and autophagy is an essential mechanism of long-term memory [27].

In this work, we endeavored to understand how exercise promotes spatial memory retention. Since autophagy has been implicated in memory recall and since our understanding of how exercise promotes memory recall remains limited, we hypothesized that autophagic pathways in both peripheral tissues and in the hippocampus mediate the effects of exercise on spatial learning and memory retention. We found that short term exercise (2 weeks) promotes spatial learning and memory retention by activating autophagy in the hippocampus of young animals. Indeed, autophagy acts downstream of hippocampal BDNF signaling, while also regulating BDNF levels suggesting the presence of a positive feedback loop between the pathways. We also discovered that exercise promotes the release of circulatory factors in young animals, but not aged animals in an autophagy-dependent manner. Indeed, we identified DBHB as a candidate exercise factor whose release into the circulation is autophagy-dependent. We also discovered that DBHB enhances spatial learning and memory recall by activating autophagy in the hippocampus. Our results suggest that autophagy plays a dual role in promoting the beneficial effects of exercise on the brain. First, it controls the release of factors such as DBHB that activate hippocampal BDNF signaling. Second, it potentiates BDNF signaling in the hippocampus thereby regulating memory retention.

2. Materials and methods

2.1. Animal housing

Young (10-week-old), middle-aged (32-week-old), and old (1 year old) C57BL/6 male mice were individually housed in cages according to the experimental groups. The mice were provided with food and water ad libitum and maintained on a 12-h light-dark cycle. Animal use and care is in accordance with the guidelines and as approved by the Lebanese American University Animal Care and Use Committee (ACUC).

Exercise paradigm: Mice were individually housed in cages and were divided into two groups: sedentary or exercise. The exercise group was provided with free access to running wheels for the duration of fourteen days. Before the start of the behavioral testing, the running wheels were removed from the cages.

Injections: 10-week-old male mice received daily intraperitoneal injections of saline, chloroquine diphosphate (CQ; 50 mg/kg) [28], 7,8 Dihydroxyflavone (7,8-DHF; 5 mg/kg) [29], or chloroquine diphosphate + 7,8 Dihydroxyflavone 15 min prior to the behavioral experiments. 10-week-old male mice received daily intraperitoneal injections of saline, or D-Beta-hydroxybutyrate (DBHB) (100 mg/kg) [30] four days before and during the behavioral experiments.

2.2. Morris Water Maze (MWM)

After the completion of the exercise paradigm, the mice were tested in the MWM as previously described [17, 31]. The MWM assesses spatial learning and memory through testing the ability of the mice to use visual cues placed on the borders of a pool in order to escape the water and reach a hidden platform [31]. During the learning stage (5 days) of the test, the latency of the mice to escape the water, and reach the hidden platform was recorded by the ANY-maze Video Tracking System. During the memory stage of the test, the platform was removed from the pool and the time spent by the mouse in the quadrant that previously contained the platform or the island was also recorded by the ANY-maze Video Tracking System. Each mouse was subjected to three trials from different starting points.

Plasma collection: To assess whether the modulation of autophagy outside the nervous system is involved in mediating exercise's effect on learning and memory recall, plasma donation experiments were conducted.

Plasma Donation from young and middle-aged mice: 10-week-old male mice or 32-week-old male mice that were eventually used for the plasma donation exercised for 14 days and underwent the Morris Water Maze (MWM) for 7 days. The animals were treated with saline or CQ. Sedentary mice were housed in the same conditions and underwent the MWM for 7 days. On the last day of the MWM, the mice were sacrificed, and the trunk blood was collected in EDTA (0.5 M) coated tubes, and centrifuged at 3000 rpm, 4 °C, for 15 min. Pooled plasma was separated from the blood and stored at −80 °C for transfusions into 32-week-old (C57BL/6 J) male mice.

Plasma Transfusions from 10-week-old or 32-week-old mice to 32-week-old mice: Middle-age (32-week-old) C57BL/6 J male mice received intraperitoneal injections of plasma collected from young (10-week-old) or middle-aged (32-week-old) exercise male mice, or sedentary male mice treated with saline or CQ. The plasma was injected in 100 μL increments. Each animal received one injection every three days, for a total of eight injections over the period of twenty two days [32]. On the twenty third day, the mice underwent the MWM.

Tissue collection: At the end of the MWM, animals were sacrificed by decapitation to prevent any effects of anesthesia on metabolite levels in the blood and the brain and as approved by the ACUC. The brain was isolated on dry ice and the hippocampus was dissected and stored at −80° Celsius until use.

Protein Extractions: Total cell proteins from all tissues used in this manuscript were prepared by lysing cells in RIPA-B (1% Triton X-100, 1% SDS, 50 mM Tris-Cl, pH 7.4, 500 mM NaCl and 1 mM EDTA) in the presence of protease inhibitors (Bio-world), the proteasome inhibitor MG-132 (Sigma) and phosphatase inhibitors (Bio-world). The samples were centrifuged for 15 min, at 15,000 rpm and 4 °C and the supernatant was collected [33].

Western blot analysis: Samples were boiled in Laemmli buffer and electrophoresed on Bis-Tris 30% acrylamide gels (Bio-Rad). Proteins were transferred to a PVDF or nitrocellulose membrane (Bio-Rad) using semi-dry TransBlot Turbo Transfer System (Bio-Rad). Nonspecific binding was inhibited by incubation in blocking buffer (BSA/non-fat dried milk and TBS-Tween 20). Antibodies against LC3B (1:1000; Abcam), BDNF (1:500; N-20, Santa Cruz Biotechnology), GAPDH (1:1000; Abcam) and Actin (1:1000; Santa Cruz Biotechnology) were diluted in blocking buffer and the membranes were incubated overnight at 4 °C. Secondary antibodies (BioRad) were used at a 1:5000 dilution followed by incubation for 90 min at room temperature. Finally, proteins were detected by chemiluminescence on ChemiDoc (Bio-Rad) using Clarity Western ECL Substrate (Bio-Rad) or by SuperSignal™ West Femto Maximum Sensitivity Substrate (Thermo Scientific). ImageJ was used for band quantification [34].

DBHB measurements: Plasma DBHB levels were measured using the DBHB Assay kit (MAK041, Sigma) according to the manufacturer's protocol. This is a standard kit used to measure DBHB concentration by a coupled enzyme reaction, which results in a colorimetric (450 nm) product, proportional to the DBHB present in the sample. Using known concentrations (0–10 nmoles) of DBHB, a standard curve was plotted. The background for the assays is the value obtained for the 0 (blank) Beta-Hydroxybutyrate Standard. The background was corrected by subtracting the 0 (blank) value from all readings. The Absorbance at 450 nm for both sedentary and exercise samples was measured. All values obtained were within the linear range of the standard curve and fell between the values obtained for the standards. The number of nmoles was calculated using the equation of the best fit line of the standard curve. The amount of DBHB in ng was calculated using the molecular weight of DBHB and the concentration was calculated by dividing the ng by the volume of the sample.

Statistical Analysis: Sample size was determined based on previous studies published by our lab [16, 17] as well as other labs [35], which used similar experimental paradigms. In addition, the resource equation method [[36], [37], [38], [39]] predicts that a n number of 4 is sufficient to obtain an E value between 10 and 20. E is the degree of freedom of analysis of variance (ANOVA) [36]. Though, this method is based on ANOVA, it is applicable to all animal experiments [36]. Any sample size, which keeps E between 10 and 20 is considered as adequate [[36], [37], [38], [39]]. E can be measured by following formula: E = Total number of animals − Total number of groups {Charan, 2013 #114. In all conducted experiments, the calculated E was between !0–20 and in most cases superior to these values. Unpaired t-test, 1way or 2way ANOVA followed by Tukey post-hoc tests were used to measure statistical significance. P < 0.05 was considered to be statistically significant. *: p < 0.05, **: p < 0.01, ***:p < 0.001 and ****:p < 0.0001. All graphs are presented as mean ± SEM.

3. Results

3.1. Voluntary exercise induces autophagy in the hippocampus of young mice

To assess how voluntary exercise affects autophagy in the hippocampus of mice, we subjected young 10-week-old male mice to a 14-day-voluntary exercise paradigm (Fig. 1A) [40]. This protocol was previously shown to enhance both learning and consolidation of cued conditioned fear [41]. Exercising mice in this paradigm have significantly higher hippocampal LC3B protein levels (p = 0.0425 for exercise versus control, Interaction: (F1, 55) = 6, p = 0.0167 and paradigm (F1, 55) = 2.0, p = 0.1612; treatment (F1, 55) = 0.7, p = 0.4014; 2way ANOVA followed by Tukey's post-hoc test) as measured by Western Blots. This increase was abolished by the brain permeable autophagy inhibitor CQ (p = 0.0427 for exercise versus exercise + CQ, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 1B and C and Supplementary Figure 1). This data is consistent with previously reported increases in brain autophagy after different exercise paradigms [25, [42], [43], [44]]. In order to test whether voluntary exercise induces spatial memory formation by increasing autophagy, we performed MWM experiments. The MWM is a spatial learning task that requires mice to locate a hidden platform in an opaque pool of water using visual cues. Acquisition of spatial learning in both sedentary and exercise mice was observed as reduced latency to reach the hidden platform by day 5. As expected, sedentary control mice acquired spatial learning over the course of the five day training (p = 0.0049 for day2 versus day4 and p = 0.042 for day 2 versus day 5, Interaction: (F12, 1100) = 3.073, p = 0.0003 and days (F4, 1100) = 24.02, p < 0.0001; treatment (F3, 1100) = 100.6, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test). Exercising mice significantly outperformed the control mice (p < 0.0001 from day2-day5 for exercise versus sedentary, Interaction: (F12, 1100) = 3.073, p = 0.0003 and days (F4, 1100) = 24.02, p < 0.0001; treatment (F3, 1100) = 100.6, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 1D). To assess reference memory, we performed a probe trial 24 h after the last training session (day 6), during which the platform was removed. As expected, exercise mice showed significant enhancement of memory recall, as indicated by increased time spent in the target quadrant (p = 0.0016 for exercise versus sedentary Interaction: (F1, 217) = 0.3472, p = 0.5563 and paradigm (F1, 217) = 8.679, p = 0.0036; treatment (F1, 217) = 17.62, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 1E). These results show that the described exercise paradigm promoted spatial learning and memory recall. To assess whether the exercise-mediated activation of autophagy is responsible for enhanced memory formation, we injected exercise mice with the autophagy inhibitor CQ. These mice exhibited significantly worsened learning curves as compared to exercise mice injected with saline (p < 0.001 from day4-day5 for exercise + CQ versus exercise, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 1D). In addition, exercise mice receiving the CQ treatment showed a significant decrease in memory recall as compared to exercise mice receiving saline (p = 0.0423 for exercise versus sedentary, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 1E). Taken together, the data suggests that exercise induces autophagy and in turn enhances spatial learning and memory retention. We were next interested in elucidating the mechanism of action of exercise-induced autophagy. Exercise is known to mediate learning and memory formation by increasing BDNF levels and signaling in the hippocampus [[1], [2], [3], [8], [9], [10], [16], [17], [18], 45, 46]. As a result, we asked whether autophagy induces learning and memory retention by increasing BDNF levels in the hippocampus. As expected, exercise significantly increased hippocampal BDNF levels (p = 0.015, unpaired T-test) (Fig. 1F and G and Supplementary Figure 2). Exercise mice that received CQ showed significantly lower hippocampal BDNF protein levels as compared to exercise mice (p = 0.011, unpaired T-test) (Fig. 1H and I). Our results are consistent with the hypothesis that exercise activates autophagy, which in turn increases hippocampal BDNF levels to mediates spatial learning and memory retention. To test this hypothesis, we predicted that activating the BDNF signaling pathway should bypass the inhibition of autophagy and restore the enhanced learning and memory formation in exercise animals.

Fig. 1.

Voluntary exercise promotes spatial learning and memory retention by activating autophagy (A) The exercise paradigm involves fourteen days of voluntary exercise followed by behavioral training, animal euthanasia and hippocampal isolation. (B) Voluntary exercise for fourteen days significantly increases hippocampal LC3B protein levels and CQ abolishes this effect as measured by Western Blots. Refer to supplementary Fig. 1 for uncropped version of the graph (C) Quantification of the LC3B Western blot. The number of hippocampi used for each group (sed + saline; sed + CQ, Ex; Ex + CQ) is 17, 12, 20 and 10 respectively. *p < 0.05 as measured by 2-way ANOVA followed by Tukey's post-hoc test. (D) Voluntary exercise for fourteen days promotes spatial learning in young animals by decreasing the escape latency in the MWM and CQ treatment abolishes this exercise-mediated effect. Sedentary control mice that received saline significantly acquired spatial learning over the course of the five day training. Exercise mice that received saline showed significantly reduced escape latency compared to sedentary mice, and to exercise mice that received CQ. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The plus (+) sign shows significance between Sed + saline mice (day 2) and Sed + saline mice (day 4 and day 5). The asterisk sign (*) shows significance between Ex + saline mice versus Sed + saline mice. The octothorpe sign (#) shows significance between Ex + saline mice versus exercise Ex + CQ mice. (E) Voluntary exercise for fourteen days promotes memory retention in young animals by increasing the time spent in the target quadrant in the MWM and CQ treatment abolishes this exercise-mediated effect. Exercise mice that received saline spent significantly more time in the target quadrant compared to sedentary mice, and to exercise mice that received CQ. Results are expressed as Mean ± SEM. The number of animals used is: n = 55 Sed + saline, n = 42 Sed + CQ, n = 74 Ex + saline, n = 50 Ex + CQ. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. *p < 0.05, **p < 0.01, and ****p < 0.00001. (F) Representative Western blot image depicting the increase in BDNF protein levels in the hippocampus of exercise mice as compared to mice sedentary mice. Refer to supplementary Fig. 2 for uncropped version of the graph (G) Quantification of the BDNF Western blot. Statistical significance was measured by the unpaired t-test *p < 0.05; p = 0.0212 and df = 2.509. (H) Representative Western blot image depicting the decrease in BDNF protein levels in the hippocampus of exercise mice receiving CQ as compared to exercise mice receiving saline. (I) Quantification of the BDNF Western blot. Statistical significance was measured by the unpaired t-test *p < 0.05; p = 0.011 and df = 4.

3.2. Activation of BDNF signaling does not bypass inhibition of autophagy

In order to understand whether hippocampal BDNF signaling is downstream of exercise-induced autophagy, we tested whether activation of BDNF signaling is sufficient to bypass autophagy inhibition and restore exercise-induced learning and memory retention. We activated BDNF signaling using the bioactive brain permeable high affinity TrkB agonist 7,8-DHF [29] (Fig. 2A). As expected, sedentary mice that received 7,8-DHF exhibited significantly enhanced learning curves as compared to sedentary mice injected with saline (p < 0.0082 on day5 for sedentary + 7,8-DHF versus sedentary, Interaction: (F4, 305) = 3.351, p = 0.0105 and paradigm (F4, 305) = 13.50, p < 0.0001; treatment (F1, 305) = 4.965, p = 0.0266; 2way ANOVA followed by Sidak's multiple comparison test) (Fig. 2B) and showed significant enhancement of memory recall, as indicated by the increased time spent in the target quadrant (p = 0.0235 for sedentary + 7,8-DHF versus sedentary, unpaired T-test,df = 61) (Fig. 2C). In exercise mice, 7,8-DHF treatment did not have significant effects on learning acquisition or memory retention (Fig. 2D and E). This is expected considering that both exercise and 7,8-DHF activate the same hippocampal BDNF pathways. Exercise mice that received CQ showed worsened learning curves (p < 0.0001 on day4–5 for exercise + CQ versus exercise, Interaction: (F12, 770) = 1.655, p = 0.0723 and paradigm (F4, 770) = 9.877, p < 0.0001; treatment (F3, 770) = 39.78, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test) and impaired memory recall (p = 0.0227, (F3, 151) = 4.792, p = 0.0032; one-way ANOVA followed by Tukey's post-hoc test) (Fig. 2D and E). Interestingly, exercise mice receiving the combined CQ and 7,8-DHF treatment also showed significantly worsened learning curves (p = 0.0003 on day 4 and p < 0.0001 on day5 for exercise + CQ+7,8-DHF versus exercise, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 2D) and memory retention as measured by the decreased time spent in the target quadrant (p = 0.0498, one-way ANOVA followed by Tukey's post-hoc test) (Fig. 2E) as compared to exercise mice receiving saline. Our results suggest that activation of BDNF signaling by 7,8-DHF can't bypass the inhibition of autophagy to mediate exercise-induced spatial learning and memory formation. This suggests that BDNF is unlikely to be downstream of exercise-induced autophagy. Rather, our results are consistent with a model by which exercise induces BDNF signaling in the hippocampus of young mice. Activation of BDNF signaling in turn activates autophagy. Finally, activation of autophagy leads to a positive feedback mechanism that increases BDNF expression and signaling and in turn promotes learning and memory formation.

Fig. 2.

Activation of BDNF signaling doesn't bypass the effects of autophagy inhibition on learning and memory formation. (A) The exercise paradigm involves fourteen days of voluntary exercise followed by daily CQ, 7,8-DHF, or 7,8-DHF + CQ injections during seven days of behavioral training, then animal euthanasia and hippocampal isolation. (B) The TRKB agonist 7,8-DHF promotes spatial learning in sedentary mice by decreasing the escape latency in the MWM. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Sidak's multiple comparison test. The asterisk sign (*) shows significance between Sed + saline mice versus Sed+7,8-DHF mice. **p < 0.01 (C) The TRKB agonist 7,8-DHF promotes spatial memory recall in sedentary mice by increasing the time spent in the target quadrant in the MWM. Results are expressed as Mean ± SEM. The number of animals used is n = 55 Sed + saline, n = 8 Sed+7,8-DHF. Statistical significance was measured by unpaired T-test. *p < 0.05, p = 0.0235 and df = 61 (D) The TRKB agonist 7,8-DHF doesn't bypass the negative effect of CQ on spatial learning in exercise mice. Exercise mice that received CQ has worsened learning curve compared to exercise mice that received saline as observed in the increase in escape latency. Mice that received the combined treatment of CQ and 7,8-DHF didn't show enhanced learning as compared to exercise mice that received CQ only as observed by the similar learning curves. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Sidak's multiple comparison test. The asterisk sign (*) shows significance between Ex + saline mice versus Ex + CQ. The octothorpe sign (#) shows significance between Ex + saline mice versus exercise Ex + CQ+7,8-DHF mice. The dollar sign ($) shows significance between Ex + saline mice versus exercise Ex+7,8-DHF mice. **p < 0.01, ***p < 0.001 and ****p < 0.00001. (E) The TRKB agonist 7,8-DHF doesn't bypass the negative effect of CQ on spatial memory retention in exercise mice. Exercise mice that received CQ spent significantly less time in the target quadrant as compared to exercise mice that received saline. Mice that received the combined treatment of CQ and 7,8-DHF didn't show enhanced learning as compared to exercise mice that received CQ only. Results are expressed as Mean ± SEM. Statistical significance was measured by 1way ANOVA followed by Tukey's multiple comparison test. The asterisk sign (*) shows significance versus Ex + saline mice. *p < 0.05.

3.3. Voluntary exercise doesn't induce learning and memory retention in middle-age (32 week old) and old (1 year old) mice

We next tested whether the utilized voluntary exercise paradigm enhances spatial learning and memory retention in middle-age (32-week-old) and old (1-year-old) mice. Hence, we subjected 32-week-old and 1-year-old mice to fourteen days of voluntary wheel exercise. Sedentary 32-week-old and 1-year-old mice served as controls. Fourteen days after voluntary exercise, mice underwent the MWM test to assess hippocampal-dependent spatial learning and memory (Fig. 3A). No enhancement in spatial learning formation and memory retention was observed in both 32-week-old exercise (Fig. 3B and C) mice or 1 year-old exercise mice (Fig. 3D and E) versus their respective sedentary controls. Indeed, 32-week-old and old 1-year-old mice that exercised did not show a significant decrease in the escape latency across the five days of training (Fig. 3B and D), nor did they show any significant increase in the time spent in target quadrant searching for the platform (Fig. 3C and E). Since both middle-age and old mice behaved similarly, we focused the rest of the work on the middle-age group.

Fig. 3.

Voluntary exercise for fourteen days doesn't promote spatial learning and memory retention in middle-age (32-week old) and old (1-year old) mice. (A) 32 week-old or 1 year-old mice were subjected to an exercise paradigm that involves fourteen days of voluntary exercise followed by daily saline injections during seven days of behavioral training, then animal euthanasia and hippocampal isolation. (B) Voluntary exercise for fourteen days doesn't promote spatial learning in middle -age mice. Exercise mice didn't show a significant decrease in escape latency as compared to sedentary mice. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. (C) Voluntary exercise for fourteen days didn't promote memory retention in middle-age mice. Exercise mice spent the same time in the target quadrant as sedentary mice. Results are expressed as Mean ± SEM. The number of animals used is: n = 36 Sed, n = 66 Ex. Statistical significance was measured by the unpaired T-test. P = 0.4756 and df = 100. (D) Voluntary exercise for fourteen days doesn't promote spatial learning in old mice. Exercise mice didn't show a significant decrease in escape latency as compared to sedentary mice. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. (E) Voluntary exercise for fourteen days didn't promote memory retention in old mice. Exercise mice spent the same time in the target quadrant as sedentary mice. Results are expressed as Mean ± SEM. The number of animals used is: n = 6 Sed, n = 7 Ex. Statistical significance was measured by the unpaired T-test. P = 0.5522 and df = 11.

3.4. Plasma from young exercised mice enhances spatial learning and memory retention in middle-aged sedentary mice in an autophagy-dependent manner

We were next interested in understanding why exercise failed to promote spatial learning and memory formation in middle-age mice. Previous work suggests that exercise induces the release of metabolites and proteins from liver, muscle, adipose tissue and bones that mediate its positive effects on learning and memory formation by activating hippocampal BDNF signaling [8, 9]. One hypothesis to explain why exercise didn't promote learning and memory formation in middle-age mice is that the utilized exercise paradigm promotes the release of such factors into the blood of young animals and fails to promote their release into blood in middle-age mice. To test this hypothesis, we assessed whether plasma transfusions from young exercise (10-week-old) mice induce spatial learning and memory formation in middle-age sedentary mice. To perform the plasma transfusions, plasma was collected from 10-week-old exercise or sedentary mice (Fig. 4A). Next, middle-aged sedentary mice received a total of eight intraperitoneal injections of the plasma collected from the young mice over a period of twenty two days. Finally, the sedentary middle-age mice that received the plasma injections underwent behavioral testing (Fig. 4B). Sedentary middle-age mice that received intraperitoneal injections of plasma from young exercise mice showed significantly reduced escape latency as compared to sedentary middle-age mice that received plasma from sedentary young mice (p = 0.0446 on day5 for exercise plasma versus sedentary plasma, Interaction: (F12, 110) = 0.4210, p = 0.9524 and paradigm (F4, 110) = 4.259, p = 0.0030; treatment (F3, 110) = 14.96, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 4C). Moreover, the sedentary middle-age mice that received intraperitoneal injections of plasma from young exercise mice spent significantly more time in the target quadrant as compared to sedentary middle-age mice that received plasma from sedentary young mice (p = 0.0303 for exercise plasma versus sedentary plasma, (F1, 22) = 3.289, p = 0.0834 and paradigm (F1, 22) = 27.49, p < 0.0001; treatment (F1, 22) = 5.975, p = 0.0230; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 4D). Interestingly, when plasma was collected from middle-age exercise or sedentary mice and injected into sedentary middle-age mice, no difference in learning acquisition and memory retention was observed (Fig. 4E and F). These results suggest that exercise indeed promotes the release of factors that promote learning and memory formation into the plasma of young mice, whereas it fails to promote their release in middle-age mice. The loss of these circulating factors is responsible for the inability of exercise to promote learning and memory formation in middle-age mice. We next investigated whether the release of the exercise factors into the circulation is itself autophagy-dependent. For that purpose, plasma was collected from young mice that received CQ during behavioral testing (Fig. 4A) and injected into sedentary middle-age mice (Fig. 4B). Sedentary middle-age mice that received intraperitoneal injections of plasma from young exercise mice that received CQ showed increased escape latency as compared to sedentary middle-age mice that received plasma from exercise young mice (p = 0.0153 on day2 and p = 0.0403 on day 3 for exercise + CQ plasma versus exercise plasma, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 4C). Moreover, the sedentary middle-age mice that received intraperitoneal injections of plasma from young exercise mice spent significantly higher time in the target quadrant as compared to sedentary middle-age mice that received plasma from exercise young mice that received CQ (p = 0.0003 for exercise plasma versus exercise + CQ plasma, 2way ANOVA followed by Tukey's post-hoc test) (Fig. 4D). These results suggest that the release of exercise factors into the circulation is autophagy-dependent. Taken together, our results support roles for autophagy in both peripheral tissues and the brain in promoting exercise's positive effects on spatial learning and memory formation.

Fig. 4.

Plasma of young exercise mice promotes spatial learning and memory retention in middle-age mice in an autophagy-dependent manner. (A) The plasma collection protocol involves: 1) subjecting 10-week-old or 32-week old mice to 14 days of exercise 2) MWM testing, while receiving different treatments, and 3) animal sacrifice and plasma collection. (B) The plasma transfusion protocol involves intraperitoneal injections of 100 μl of collected plasma into sedentary middle age mice every three days for a period of 22 days followed by MWM testing. (C) Plasma collected from young exercise mice promotes spatial learning in middle-age animals by decreasing the escape latency in the MWM. In contrast, plasma collected from young exercise mice that received CQ didn't promote spatial learning in middle-age animals. Middle-age mice that received plasma from young exercise mice showed significantly reduced escape latency as compared to middle-age mice that received plasma from young sedentary mice or young exercise mice treated with CQ. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The n number of animals are 6 for middle-age mice receiving plasma from young sedentary mice/young sedentary mice + CQ and 7 for middle-age mice receiving plasma from young exercise mice/young sedentary mice + CQ. The asterisk sign (*) shows significance between middle-age mice receiving plasma from young Ex + saline mice versus young Sed. The octothorpe sign (#) shows significance between middle-age mice receiving plasma from young Ex + saline mice versus young Ex + CQ. The dollar sign ($) shows significance between middle-age mice receiving plasma from young Ex + saline mice versus young Sed + CQ. *p < 0.05, ^#p < 0.05 and ^$p < 0.05. (D) Plasma collected from young exercise mice promotes spatial memory retention in middle-age animals by increasing the time spent in target quadrant of the MWM. In contrast, plasma collected from young exercise mice that received CQ didn't promote spatial memory retention in middle-age animals. Middle-age mice that received plasma from young exercise mice spent significantly increased time in the target quadrant as compared to middle-age mice that received plasma from young sedentary mice or young exercise mice treated with CQ. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The n number of animals are 6 for middle-age mice receiving plasma from young sedentary mice/young sedentary mice + CQ and 7 for middle-age mice receiving plasma from young exercise mice/young sedentary mice + CQ. The asterisk sign (*) shows significance versus middle-age mice receiving plasma from young Ex + saline mice. *p < 0.05 and ***p < 0.001. (E) Plasma collected from middle-age exercise mice doesn't promote spatial learning in middle-age mice. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The n number of animals are 4 for middle-age mice receiving plasma from middle-age sedentary mice and 7 for middle-age mice receiving plasma from middle-age exercise mice. (F) Plasma collected from middle-age exercise mice doesn't promotes spatial memory retention in middle-age mice. Middle-age mice that received plasma from middle-age exercise mice spent similar times in the target quadrant as compared to middle-age mice that received plasma from middle-age sedentary mice. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The n number of animals are 4 for middle-age mice receiving plasma from middle-age sedentary mice and 7 for middle-age mice receiving plasma from middle-age exercise mice.

3.5. The exercise factor β-hydroxybutyrate (DBHB) promotes spatial learning and memory retention in an autophagy-dependent manner

The ketone body DBHB is an exercise factor that is secreted by the liver, inhibits histone deacetylases such as HDAC2 and HDAC3, and increases the levels of BDNF in vitro and in vivo in the hippocampus of mice [16]. Therefore, we investigated whether DBHB is an exercise factors that promotes spatial learning and memory retention in an autophagy-dependent manner. Young mice that underwent voluntary exercise showed significantly higher levels of DBHB in their plasma as compared to young exercised mice treated with CQ (Fig. 5A) (p = 0.0332, and df = 6, unpaired t-test). Indeed, Western Blots show that DBHB increases LC3B levels in the hippocampus and this increase was abolished by CQ treatment (p = 0.0314 for DBHB versus saline and p = 0.0271 for DBHB versus DBHB + CQ, (F1, 27) = 2.862, p = 0.1022 and paradigm (F1, 27) = 5.934, p = 0.0127; treatment (F1, 27) = 3.265, p = 0.0819; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 5B and Supplementary Figure 3). Consistent with these results, DBHB promoted spatial learning and memory retention (Fig. 5C and D). DBHB had modest effects on spatial learning. Mice that received DBHB had a significantly reduced escape latency on day 2 only (p < 0.0001 on day 2 for DBHB versus saline, (F12, 550) = 1.127, p = 0.3347 and paradigm (F4, 550) = 12.68, p < 0.0001; treatment (F3, 550) = 20.46, p < 0.0001; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 5C). DBHB had a clearer effect on memory retention. Mice that received DBHB spent significantly more time exploring the island as compared to control mice, whereas CQ abolished DBHB's effect on memory formation (p = 0.0005 for DBHB versus saline and p = 0.0479 for DBHB versus DBHB + CQ, (F1, 107) = 4.938, p = 0.0284 and paradigm (F1, 107) = 6.897, p = 0.0099; treatment (F1, 107) = 6.897, p = 0.0099; 2way ANOVA followed by Tukey's post-hoc test) (Fig. 5D). Thus, these results suggest that DBHB is an exercise factor that is released into the circulation in an autophagy-dependent manner. DBHB also induces autophagy in the hippocampus to promote learning and memory formation (Fig. 5E).

Fig. 5.

DBHB promotes spatial memory retention in an autophagy-dependent manner. (A) CQ treatment significantly decreases DBHB levels in the plasma of young exercise mice. The number of animals used for each group is 4. Statistical significance was measured by unpaired t-test *p < 0.05; p = 0.0332 and df = 6. (B) 10 week-old mice were received daily injections of DBHB or CQ or DBHB + CQ four days prior and during seven days of behavioral training followed by animal euthanasia and hippocampal isolation. Representative Western blot image depicting the increase in hippocampal LC3B protein levels of animals that received DBHB as compared to mice that received saline, CQ or DBHB + CQ. Quantification of the LC3B Western blot. Refer to supplementary Fig. 3 for uncropped version of the graph The n number for saline, DBHB, CQ and DBHB + CQ is 10, 10, 5 and 6 respectively. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. ^$$p < 0.01 ***p < 0.001 and ^$$$$p < 0.0001. (C) DBHB promotes spatial learning in young animals by decreasing the escape latency in the MWM. Mice that received DBHB showed significantly reduced escape latency compared to mice that received saline or saline + CQ. Results are expressed as Mean ± SEM. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. The asterisk sign (*) shows significance between mice that received saline versus DBHB. The octothorpe sign ($) shows significance between mice that received DBHB versus saline + CQ. (D) DBHB promotes memory retention in young animals by increasing the time spent on the island in the MWM and CQ treatment abolishes this DBHB-mediated effect. Young mice that received DBHB spent significantly more time in the island as compared to mice that received saline, CQ or DBHB + CQ. Results are expressed as Mean ± SEM. The number of animals used is: n = 52 saline, n = 41 CQ, n = 12 DBHB, n = 6 DBHB + CQ. Statistical significance was measured by Two-way ANOVA followed by Tukey's multiple comparison test. *p < 0.05, ***p < 0.001. (E) Model depicting the different roles of autophagy in mediating exercise's beneficial effect on spatial learning and memory formation.

4. Discussion

Our results provide a link between voluntary exercise and autophagy in the hippocampus and peripheral tissues of young and aging mice. Exercise is a positive modulator for a wide range of central nervous system diseases [47]. Indeed, exercise promotes neurogenesis and learning and memory formation through the action of circulatory blood factors such as DBHB, lactate, irisin and cathepsin-B that converge on the activation of hippocampal Bdnf gene expression and signaling [[16], [17], [18], 32, 48]. These circulatory factors also mediate the beneficial effects of exercise by rescuing symptoms of neurodegeneration as well as promoting resilience to stress and anxiety [16, 35, 49, 50]. The identification of intracellular signals that control exercise-induced memory formation is critical for our understanding of how cognition is regulated by exercise in young, aging and pathological conditions. We showed that exercise upregulates autophagy in the hippocampus of young (10-week-old) mice and that exercise-mediated spatial learning and memory recall are dependent on autophagy in these animals (Fig. 1). Consistent with this observation, exercise was found to activate autophagy in peripheral tissues and in the brain [42] and previous work has hinted to an important role for autophagy in mediating exercise-induced benefits [51, 52]. For example, the activation of autophagy protein expression was shown to be involved in endurance exercise-mediated neurogenesis and synaptic plasticity [53]. Voluntary wheel-running exercise attenuated brain aging in rats through activating miR-130a-mediated autophagy [54] and treadmill exercise decreased β-amyloid deposition by enhancing autophagy-lysosomal activity in APP/PS1 transgenic mice [55].

Our results also reveal that exercise-induced autophagy is downstream of BDNF signaling since TrkB activation could not bypass the inhibition of autophagy (Fig. 2). Interestingly, inhibition of autophagy in exercise animals led to significant decreases in BDNF levels suggesting that a positive feedback interaction between autophagy and hippocampal BDNF signaling may occur (Figs. 1 and 2). How autophagy regulates and is modulated by BDNF signaling may be specific to the environmental signal at hand. Our work suggests that in the context of exercise, both BDNF signaling and autophagy are activated to promote spatial learning and memory recall, whereas previous work established that autophagy is a negative regulator of memory formation under starvation conditions and that it is inhibited by BDNF signaling [56]. Interestingly, like exercise, neuronal stimulation also induces hippocampal autophagy in young mice to mediate contextual fear conditioning memory recall through the formation of dendritic spines and increasing synaptic strength [25]. This suggests that both young age and exercise promote learning and memory formation by activating autophagy [25], whereas in extreme stress conditions autophagy may have negative roles [56]. It is still not understood how exercise mediates an increase in hippocampal autophagy. Indeed, since we modulated autophagy using intraperitoneal injections of CQ, we can't rule out that the effects we are observing are mediated through activation of autophagy inn different brain regions. We have assessed LC3B levels in the cortex in response to exercise and did not detect any significant changes (RK, SFS and JSS, unpublished data). Therefore, additional work is required to identify whether autophagy is activated in other brain regions and where it has a predominant effect in controlling memory formation. Ideally, hippocampal-specific inhibition of autophagy in exercise animals would enhance the conclusions of the manuscript.

Aging is associated with progressive loss of cognitive function and with aberrant autophagy in the brain [57]. Recent studies showed that brain aging may be reversed through systemic administration of circulating blood factors derived from exercised old mice or from young mice [25, 32, 50]. Plasma collected from exercised old mice transferred the cognitive benefits of exercise to inactive old mice through the action of glycosylphosphatidylinositol (GPI)-specific phospholipase D1 (Gpld1), a GPI-degrading enzyme derived from liver. This work implicated a liver-to-brain axis by which circulatory factors mediate the effects of exercise on neurogenesis [32]. In a different context, plasma derived from young animals reversed the age-dependent decline in memory recall by restoring autophagy levels in the hippocampi of old mice [25]. Indeed, osteocalcin, a bone-derived hormone and circulatory factor detected in young plasma was found to be necessary to promote learning and memory in an autophagy-dependent manner. This work implicated a bone-to-brain axis by which circulatory factors mediate the effects of young age on memory recall [25]. We focused on middle-aged mice and tested whether plasma transfusions from young exercised mice induced spatial learning and memory recall in sedentary middle-aged mice. Interestingly, we found that only plasma collected from young exercised mice and not middle-aged exercised mice significantly enhanced spatial learning and memory recall in sedentary middle-aged mice (Fig. 4). While this result may seem inconsistent with previous results showing that plasma from old exercised mice can promote neurogenesis in aged inactive mice, it is important to note that the exercise paradigms used are distinct. In this work, short term exercise (2 weeks of voluntary exercise) was utilized as compared to long-term exercise (6 weeks of voluntary exercise) used in the previous works [32]. Interestingly, we observed that plasma derived from young exercised animals that received CQ failed to transfer the benefits of exercise to sedentary middle-aged mice. This supports a role for autophagy in regulating the identity and pool of the young exercise blood milieu and allows us to distinguish between autophagy's role in peripheral tissues versus the hippocampus in response to exercise. Our results are consistent with a model that exercise induces autophagy in peripheral tissues to regulate the identity and release of the exercise factors into the blood in young animals. Exercise factors, in turn, activate BDNF signaling and autophagy in the hippocampus to promote spatial learning and memory recall (Fig. 5E).

One circulatory exercise factor released by the liver is the ketone body DBHB. Hippocampal DBHB levels increase after exercise. DBHB inhibits histone deacetylases and increases BDNF levels to promote synaptic plasticity (Sleiman et al., 2016). We found that exercise-induced DBHB release into the blood is autophagy dependent (Fig. 5). In addition, we discovered that DBHB increases the hippocampal LC3B levels and enhances spatial learning and memory recall in an autophagy-dependent manner. Thus, our findings are consistent with DBHB serving as one candidate circulatory exercise factor whose release is controlled by autophagy and which mediates the positive effects of exercise on hippocampal-dependent learning and memory recall in an autophagy-dependent manner (Fig. 5E).

In this work, we discovered a dual role for autophagy in mediating the effects of exercise on spatial learning and memory formation. The first role is hippocampal and tightly liked to BDNF signaling; whereas the second role involves the regulation of the identity and concentration of the circulatory exercise factors released by different tissues including the liver. Further work is needed to identify novel circulatory exercise factors that are both autophagy and BDNF-dependent, and that can promote spatial learning and memory formation. This can direct the assembly of an exercise pill that can be utilized for patients afflicted with neurodegenerative diseases or psychiatric diseases.

5. Author contribution statement

Reine Khoury, Joelle Saad, Vanessa Jabre, Litsa Maria Ghayad, Mohamad Khalifeh, Rouba Houbeika, Perla El Ahmad, Amar Mezher, Diala El Masri, Zena Haddad, Fady Eid, Nour Barmo, Patrick Nasrallah: Performed the experiments; Analyzed and interpreted the data.

Sama F. Sleiman, Joseph S. Stephan: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Wrote the paper.

6. Funding statement

This work was supported by grants from the Lebanese American University School of Arts and Sciences and School of Medicine as well as the Lebanese American University Presidential Fellowship to Joseph S. Stephan.

7. Data availability statement

No data was used for the research described in the article.

8. Declaration of interest's statement

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.