Exercise duration differentially effects age-related neuroinflammation and hippocampal neurogenesis

Abstract

The physiological effects of exercise vary as a function of frequency and length. However, research on the duration-dependent effects of exercise has focused primarily on young adults and less is known about the influence of exercise duration in the aged. The current study compared the effects of short-term and long-term running wheel access on hippocampal neurogenesis and neuroimmune markers in aged (19-23 months) male C57BL/6J mice. Aged mice were given 24-hour access to a running wheel for 14 days (short-term) or 51 days (long-term). Groups of non-running aged and young (5 months) mice served as comparison groups to detect age-related differences and effects of exercise. Long-term, but not short-term, exercise increased hippocampal neurogenesis as assessed by number of doublecortin (DCX) positive cells in the granular cell layer. Assessment of cytokines, receptors, and glial-activation markers showed the expected age-related increase compared to young controls. In the aged, exercise as a function of duration regulated select aspects of the neuroimmune profile. For instance, hippocampal expression of interleukin (IL)-10 was increased only following long-term exercise. While in contrast brain levels of IL-6 were reduced by both short- and long-term exercise. Additional findings showed that exercise does not modulate all aspects of age-related neuroinflammation and/or may have differential effects in hippocampal compared to brain samples. Overall, the data indicate that increasing exercise duration produces more robust effects on immune modulation and hippocampal neurogenesis.

Introduction

The pro-cognitive and health benefits of increasing physical activity through exercise have long been appreciated (Cotman et al., 2007). Engaging in regular exercise may be particularly important with aging, as exercise attenuates age-related cognitive decline and deficits in neuroplasticity (Ahlskog et al., 2011; Voss et al., 2013; Cooper et al., 2018). For instance, Erikson et al. (2009) report that higher levels of physical fitness in older adults was associated with increased hippocampal volume and improved spatial memory performance. Research with aged animals parallels these benefits, as exercise enhances learning and memory as well as increases hippocampal neurogenesis in the aged brain (van Praag et al., 2005; Speisman et al., 2013; Littlefield et al., 2015).

Research has established that regulation of neurotrophic and growth factors is a central contributor to the exercise-induced changes in synaptic plasticity and cognitive benefits (Vaynman et al., 2005). However, in the context of aging, attenuating neuroinflammation likely contributes to the positive effects of exercise on brain health (Di Benedetto et al., 2017). Aging negatively affects the regulatory control of the immune system, as both inflammation promoting and resolving functions are impaired in the aged (Kohman, 2012; Norden et al., 2013). The immunocompetent glial cells, microglia and astrocytes, acquire an inflammatory phenotype with normal aging that leads to the development of low-grade chronic neuroinflammation (Dilger et al., 2008; Kohman, 2012). For instance, levels of the proinflammatory cytokines interleukin (IL)-6 and IL-1β are elevated in the aged brain (Ye et al., 1999; Sierra et al., 2007). This age-associated neuroinflammatory state increases susceptibility to cognitive decline, neuroplasticity deficits, and neuropathologies (Lynch et al., 2002; Gemma et al., 2007; Kohman, 2012; Ojo et al., 2015; Soreq et al., 2017; Clarke et al., 2018). Exercise can reduce age-related neuroinflammation, as Speisman et al. (2013) report that exercise attenuated proinflammatory cytokine levels in the aged brain which was associated with memory improvements. Further, microglia from aged mice show attenuated expression of inflammatory-related genes and reduced proliferation following wheel running (Kohman et al., 2012; Kohman et al., 2013). Exercise also normalizes the inflammatory response to an immune challenge, as Barrientos et al. (2011) report that exercise attenuated IL-1β levels in aged rats after an E.coli infection. Collectively, these data demonstrate the immunoregulatory effects of exercise in the aged.

Running behavior and its impacts change with time. For instance, following continual access to a running wheel, rodents gradually increase their distance run, which typically stabilizes after 2-3 weeks of wheel access (Manzanares et al., 2018). This initial variable ramp-up period during the early exercise phase can produce differential effects on physiological and behavioral outcomes relative to longer periods of running. For instance, expression of the transcription factor cAMP response element-binding protein (CREB) as well as levels of phosphorylated CREB are increased during the initial week of wheel access but return to baseline levels at later time-points (Shen et al., 2001; Molteni et al., 2002; Naylor et al., 2005). In contrast, the temporal patterns of exercise-induced increases in phosphorylated mitogen-activated protein kinase (MAPK) and genes in the MAPK pathway show no change during early wheel access but are elevated following 3-4 weeks of wheel access (Shen et al., 2001; Molteni et al., 2002). Similarly, exercise-induced increases in hippocampal protein levels of brain derived neurotrophic factor (BDNF) show duration dependent increases, as BDNF levels continue to rise with longer wheel access in adult rats (Berchtold et al., 2005). Duration dependent effects of exercise are also seen with cognitive enhancements, as two or more weeks of wheel running enhanced performance in an object location memory task in young adult mice, whereas no benefits were seen in mice that ran only 2 days or 1 week (Butler et al., 2019). In sum, select physiological alterations and functional benefits may develop after a select duration of exercise.

The existing data on the duration-dependent effects of exercise has primarily focused on young adults. While exercise has clear benefits in the aged, overall running distance is drastically reduced (Kohman et al., 2012; McMullan et al., 2016). Whether longer durations of exercise are required to obtain beneficial effects in the aged, particularly in the context of neuroinflammation, has yet to be directly assessed. The current study compared the effects of short- and long-term running wheel access on hippocampal neurogenesis and neuroimmune markers in the aged. We hypothesized that longer wheel access would produce greater increases in hippocampal neurogenesis as well as broader effects on the neuroimmune profile in the aged brain.

Experimental Procedures

Subjects and experimental design

In total, the experiment included 40 male C57BL/6J mice that were bred and aged in a University of North Carolina Wilmington (UNCW) facility from breeding stock obtained from the Jackson Laboratory (Bar Harbor, MI). Thirty of the mice were aged (19-23 months) males and the remaining 10 mice were young (5 months) males, samples from these same 40 mice were used in all measurements. Mice were individually housed under a reverse light/dark cycle with free access to food and water throughout the experiment. Aged mice were semi-randomly assigned to an exercise or control condition. Mice in the control condition were individually housed in standard rodent cages. Aged mice in the exercise groups were given 24-hour access to a running wheel (23 cm in diameter, Respironics, Bend, OR) for 14 days (short-term) or 51 days (long-term). Aged mice in the short-term group were housed with wheels at a later time to allow sample collection to occur at the same time for all groups. Wheel rotations were collected in one-minute bins using Vital View software (Respironics, Bend, OR) and converted to km. The length of exercise in the short-term condition was based on prior work that showed cognitive benefits were only seen after two or more weeks of wheel running access (Butler et al., 2019). For the long-term exercise condition, the duration was based on our prior work that reported exercise-induced changes in immune parameters in the aged after similar durations (Kohman et al., 2011; Kohman et al., 2012). Young mice served as a comparison group to detect age-related differences and were therefore all assigned to the control condition resulting in four treatment groups (n=10 per group): Young control, aged control, aged short-term exercise, and aged long-term exercise. Body weight was measured on a weekly basis throughout the experiment. The UNCW Institutional Animal Care and Use Committee (IACUC) approved the experimental methods (protocol # A1516-001) which aligned with the Guide for the Care and Use of Laboratory Animals.

Sample collection

Brains were dissected on ice following decapitation and then hemisected down the interhemispheric fissure. Half of the brain was post-fixed in 4% paraformaldehyde for 24 hours and then stored in a 30% sucrose solution until sectioned on a cryostat (40 μm coronal sections). The hippocampus was dissected from the other half of the brain and snap frozen on dry ice. The remaining brain tissue (i.e., everything except the hippocampus) was snap frozen on dry ice and used for assessment of brain protein levels of cytokines via ELISAs.

ELISA

Brain samples were homogenized in a 1 mM solution of the proteinase inhibitor phenylmethylsulfonyl fluoride (PMSF in 0.1 M phosphate buffer) and supernatant collected after centrifugation (30 minutes at 3500 rpm at 4 °C). Protein levels of IL-6, IL-1β, and IL-10 were assayed via ELISA kits (eBioscience, Santa Clara, CA). The detectable range of the IL-6 ELISA was 15.6-1000 pg/ml, IL-1β ELISA was 15.6-1000 pg/ml, and IL-10 ELISA was 32-2000 pg/ml. Cytokine levels were normalized by total protein concentrations as assessed by a Pierce protein assay (Thermo Scientific, Rockford, IL). Cytokine levels are expressed as picograms of cytokine per microgram (μg) of total protein.

Gene expression

RNA was extracted from the hippocampus samples via the RNeasy Mini Kit (Qiagen, Valencia, CA) and converted to cDNA via High-capacity cDNA reverse transcription kit (Life Technologies Carlsbad, CA). Purity and quantification of the extracted RNA was assessed using a Gen5 Epoch spectrophotometer (BioTek Instruments, Highland Park, VT). Hippocampal samples were assayed for expression of the following target genes using TaqMan™ probe and primer chemistry (Applied Biosystems, Foster City, CA): IL-6 (Mm00446190_m1), IL-1β (Mm00434228_m1), IL-10 (Mm00439614_m1), IL-10 receptor alpha (IL-10ra; Mm00434151_m1), IL-33 (Mm00505403_m1), interleukin 1 receptor-like 1 (IL1rl1; Mm00516117_m1), glial cell line-derived neurotrophic factor (GDNF; Mm00599849_m1), GDNF receptor family alpha 1 (GFRα1; Mm00439086_m1), glial fibrillary acidic protein (GFAP; Mm01253033_m1), and CD86 (Mm00444543_m1). β-Actin (Mm00607939_s1) was used as the endogenous control gene. RT-PCR was conducted with the following cycling conditions: 2 min at 50 °C, 2 min at 95 °C, followed by 40 cycles of 15 sec at 95 °C and 1 min at 60 °C. RT-PCR data were analyzed using the 2^−ΔΔCT method, with a young control as the calibrator.

Immunohistochemistry and image analysis

All hippocampal sections from a one-in-six series were stained for doublecortin (DCX) via immunohistochemistry based on our previously published protocol (Littlefield et al., 2015; Connolly et al., 2020). Coronal sections were first washed with phosphate buffered saline (PBS), incubated in 0.6% hydrogen peroxide solution for 20 minutes, washed again, blocked with 6% normal donkey serum in PBS, and then incubated with a goat anti-DCX antibody (1:1000, C-18, Santa Cruz, SC-8066, Dallas, TX) at 4°C. Following the 72-hour incubation with the primary antibody, hippocampal sections were washed in PBS, submitted to a second blocking step, and then incubated in a solution containing a biotinylated donkey anti-goat secondary antibody (1:250; Jackson ImmunoResearch, 705-065-147, West Grove, PA) for 90 minutes. After washing out the secondary antibody in PBS, sections were incubated in an avidin-biotin complex solution (ABC; Vector, Burlingame, CA) for one hour followed by staining with the diaminobenzidine (DAB) chromogen. Images of the granular cell layer (GCL) of the hippocampus were taken under identical magnification (20X) and lighting conditions. Images from a single selected plane of the z-axis were captured from 10-12 sections per mouse. An estimate of the number of DCX positive cells was measured by counting the number of DCX positive cells in the GCL in ImageJ after applying a threshold to remove background staining. The volume of the GCL was determined for each mouse by summing the outlined trace area (pixels) that was converted to micrometers and multiplied by the section thickness (i.e., 40 μm). The DCX positive cells were divided by the total GCL area (cubic micrometers) to express the data as a density (Clark et al., 2010; Littlefield et al., 2015; Connolly et al., 2020).

Statistical analysis

Wheel running data (i.e., distance run in km) for the first two weeks of wheel access were analyzed by repeated measures ANOVA (i.e., mixed-model), with condition (short- and long-term exercise) as the between-subject factor and day as the within-subject factor. Body weight was analyzed by mixed-model ANOVA with treatment condition (young control, aged control, aged short-term exercise, and aged long-term exercise) as the between-subject factor and week as the within-subject factor. For the mice in the long-term exercise condition, distance run was analyzed with a mixed-model ANOVA with day as the within-subject factor. ELISA, RT-PCR, and DCX data were analyzed with one-way fixed-factor ANOVAs with treatment condition as the between-subject factor. Normality was tested using the Shapiro-Wilk test and equal variance was tested by the F-Test, if data violated one or both assumptions data transformations were employed. Estimated number of DCX positive cells were log transformed to restore normality prior to analysis. A p<0.05 was considered significant. When the overall F value was significant, Fisher’s LSD was used a post hoc test to determine group differences.

Results

Distance run

Analysis of distance run during the first two weeks showed there was a main effect of day (F_13,235=4.52, p<0.0001, see Figure 1), as distance run increased across the first two weeks. There was no difference in distance run between the aged mice in the short-term or long-term exercise conditions (F_1,18=0.37, p=0.54ns, see Figure 1). When evaluating wheel running across the entire length of wheel access for aged mice in the long-term exercise group, there was a main effect of day (F_50,450=4.19, p<0.0001, see Figure 1) that generally showed distance run increased over the first few weeks of wheel access and afterwards became relatively stable during subsequent weeks.

Figure 1. Wheel running distance.

The line graph shows median daily distance run (km) with the first and third quartile shaded. Aged mice in the short-term and long-term exercise groups ran similar distances during the first two weeks of wheel access. Distance run increased during the initial weeks of wheel access and stabilized in the long-term exercise group.

Body weight

Analysis of body weight during the first two weeks of wheel access or control housing showed there was a treatment x week interaction (F_6,72=4.66, p<0.005, see Figure 2). Post hoc testing showed that there was no difference in body weight between the aged control and aged mice in the exercise groups prior to running wheel access, but after 1 week of wheel access aged mice in the exercise groups weighed less than aged controls (p<0.05). Analysis of body weight across the 51 days showed that aged mice in the long-term exercise group maintained lower body weight than aged control mice throughout wheel access (F_16,216=3.81, p<0.0001, see Figure 2). As expected, young control mice weighed less than aged mice in all groups throughout the experiment.

Figure 2. Body weight.

Line graph shows median body weight in grams with the first and third quartile shaded. Body weight did not differ between aged groups at the start of the experiment, but aged mice in the short-term exercise (Aged ST Ex) and long-term exercise (Aged LT Ex) groups weighed less than the aged control (Aged CON) mice beginning one week after wheel access. Young (Young CON) mice weighed less than all aged mice throughout the experiment. Symbols indicate a significant difference between aged control and aged exercise groups (^) and young controls and all aged mice (*).

Hippocampal gene expression

Assessment of immune-related genes within the hippocampus showed that hippocampal expression of IL-10 was increased in aged mice following long-term exercise relative to young and aged control mice (main effect of condition, F_3,36=2.97, p<0.05, see Figure 3A). Short-term exercise did not alter IL-10 expression, nor was there a difference between young and aged control mice. Hippocampal expression of IL-10ra was similar across all conditions (F_3,36=1.67 p=0.18ns, see Figure 3D). For IL-33 expression, there was a main effect of condition (F_3,36=3.28, p<0.05, see Figure 3B). Aged controls and aged mice in the short-term exercise condition showed increased IL-33 expression relative to young controls (p<0.05). Aged mice in the long-term exercise condition did not differ from young or aged control mice for IL-33 expression. Exercise or age had no effect on expression of IL1rl1, the IL-33 receptor, as expression levels did not differ between groups (F_3,36=0.09, p=0.96ns, see Figure 3E). Expression of IL-1β and IL-6 were increased in the aged mice in all conditions relative to young controls (main effects of condition, F_3,36=9.54, p<0.0001; F_3,36=9.68, p<0.0001, respectively, see Figure 3C and 3F). There was no difference in hippocampal IL-1β and IL-6 expression between the aged exercise and aged control mice. Aging increased expression of CD86 in the hippocampus, as aged mice in the exercise and control conditions showed elevated expression relative to young controls (main effect of condition, F_3,36=5.10 p<0.005, see Figure 4A). There were no differences in CD86 expression between the aged exercise and aged control mice. Similarly, expression of GFAP was elevated in the aged mice relative to young control mice and exercise did not modify hippocampal GFAP expression (main effect of condition, F_3,36=8.53 p<0.001, see Figure 4B). Expression of GDNF and the GDNF receptor, GFRα1, did not differ between treatment conditions (F_3,36=1.93, p=0.14ns, F_3,36=0.13, p=0.93ns, respectively, see Figures 4C and 4D). All groups showed similar hippocampal levels of the endogenous control gene β-actin (F_3,36=0.12, p=0.94ns).

Figure 3. Hippocampal cytokine expression.

Box and whisker plots show the median fold change and the 25th and 75th quartiles, with the 5th and 95th percentile shown by the whiskers. Long-term exercise increased hippocampal IL-10 expression in aged mice (A). IL-33 was elevated in aged controls (Aged CON) and aged short-term (aged ST Ex), but not aged long-term (Aged LT Ex), exercise mice relative to young controls (Young CON; B). Aged mice had increased IL-β (C) and IL-6 (F) relative to young controls, but exercise had no effect. There was no difference in expression of IL-1rl1 (E) or IL-10ra (D). Dots represent individual data points within groups. Symbols indicate a significant difference from young control (*) or aged control (^).

Figure 4. Hippocampal expression of glia-related molecules.

Box shows the median fold change and the 25th and 75th quartiles, whiskers show the 5th and 95th percentile. Aged mice in all groups had higher expression of CD86 (A) and GFAP (B) relative to young controls (Young CON). Short-term (Aged ST Ex) and long-term (Aged LT Ex) exercise groups did not differ from aged controls (Aged CON). There were no differences in GDNF (C) or the GDNF receptor GFRα1 (D). Dots represent individual data points within groups. Symbols indicate a significant difference from young control (*).

Brain cytokine levels

There was a significant main effect of condition (F_{3 36}=3.56, p<0.05, see Figure 5A) for IL-6. As expected, aged control mice had higher protein levels of IL-6 compared to young control mice (p<0.05). Exercise reduced IL-6 levels in the aged brain, as aged mice in both the short- and long-term exercise conditions had lower IL-6 levels compared to aged control mice (p<0.05). Levels of IL-1β showed a main effect of treatment (F_3,36=2.97, p<0.05, see Figure 5B) with the aged mice in short-term exercise group having higher IL-1β levels than young controls and aged mice in the long-term exercise condition (p<0.05). For IL-10 there was no difference between the treatment conditions (F_3,36=1.82, p=0.16ns, see Figure 5C).

Figure 5. Brain cytokine levels.

Aged controls (Aged CON) had higher levels of IL-6 relative to young controls (Young CON) and aged mice in the exercise groups (A). Brain levels of IL-1β were elevated in the aged short-term exercise (Aged ST Ex) mice relative to young controls and aged long-term exercise (Aged LT Ex) mice (B). There was no difference in IL-10 levels between the groups (C). Box plots show median protein levels and the 25th and 75th quartiles with the 5th and 95th percentile shown by the whiskers. Dots represent individual data points within groups. Symbols indicate a significant difference from young control (*), aged control (^), or aged short-term exercise (+).

Doublecortin positive cells

For the estimated number of DCX positive cells there was a main effect of condition (F_3,36=101.45, p<0.0001, see Figure 6A) that showed aged mice in the long-term exercise condition had higher numbers of DCX positive cells in the GCL compared to aged control and aged mice in the short-term exercise condition (p<0.05). Young control mice had higher densities of DCX positive cells compared to the aged mice in all groups (p<0.0001). Assessment of the total granular cell layer volume showed that there were no differences between the treatment conditions (F_3,36=1.24, p=0.30ns, see Figure 6B).

Figure 6. Hippocampal neurogenesis.

Assessment of estimated number of DCX positive cells (A) showed that long-term (Aged LT Ex), but not short-term (Aged ST Ex), exercise increased hippocampal neurogenesis relative to aged controls (Aged CON). As expected, aged mice in all groups had fewer DCX positive cells compared to young (Young CON) mice. There was no difference in the volume of the granular cell layer across groups (B). Representative images (C). Plots show the median number of DCX positive cells and the 25th and 75th quartiles by the box and the 5th and 95th percentile by the whiskers with individual data points shown as overlaid dots. Symbols indicate a significant difference from young control (*), aged control (^), or aged short-term exercise (+).

Discussion

The objective of the present study was to compare the physiological effects of short- and long-term exercise in the aged, with an emphasis on the alterations in the neuroimmune profile and hippocampal neurogenesis. The data indicate that brain levels of IL-6 were attenuated by short-term wheel running and these reductions were maintained after long-term wheel running. Comparatively, increases in hippocampal neurogenesis and additional alterations in immune-related molecules were only seen after long-term exercise. However, the immunomodulatory effects of exercise were selective, as not all measures were altered by wheel running. Overall, the findings demonstrate that exercise duration is an important factor, as increasing the length of exercise promotes hippocampal neurogenesis and produces added effects on immune-related molecules in the aged brain.

Comparison of estimated numbers of DCX positive cells across the exercise groups revealed that increases in neurogenesis were only detected in the aged mice with long-term wheel access. These data indicate that a minimum length of exercise may be required to see benefits in hippocampal neurogenesis in the aged. Duration-dependent effects of exercise on neurogenesis are well defined in young adults. For instance, increases in cell proliferation are reported after short-term wheel exposure (e.g., 3 - 12 days) however after longer periods of running (e.g., 32 days or more) proliferation returns to basal levels (van Praag et al., 1999; Kronenberg et al., 2006). In contrast, increases in survival and neuronal differentiation more frequently occur after longer wheel access, though increases have been reported after durations as short as 10 days of wheel access in young adults (van Praag et al., 1999; Kronenberg et al., 2006; Clark et al., 2012). Results from the present experiment indicate that aged mice similarly show duration-dependent increases in neurogenesis, as 14 days of wheel access was insufficient to increase DCX positive cells in the GCL. A prior study that assessed DCX positive cells in 1- and 2-year old mice showed that 10 days of wheel access had no effect in 1-year old but increased DCX positive cells in 2-year old mice (Kronenberg et al., 2006). Data from the 1-year old mice are in alignment with the present results, as short-term exercise did not increase DCX positive cells. Data from the 2-year old mice on the surface appear to contrast with the present data, however, these differences may relate to housing conditions across the experiments as mice in the current study were singly housed, whereas Kronenberg et al. (2006) group housed their mice. Prior work has demonstrated that social versus individual housing can influence the effects of exercise on neurogenesis in young and aged adults (Stranahan et al., 2006; Kannangara et al., 2011). The present data indicate that a longer duration of exercise may be required to detect enhancements in hippocampal neurogenesis.

The present data confirm that voluntary wheel running attenuates central inflammation. Results show that exercise reduced protein levels of IL-6 within the aged brain after engaging in exercise for a short time period, and this reduction was maintained with a longer period of exercise. Additionally, we found modest changes in IL-1β, as brain protein levels were lower in the aged mice in the long-term exercise group compared to those in the short-term exercise group. In contrast, our data showed no effect of exercise at either duration on hippocampal gene expression of IL-6 and IL-1β. The lack of an effect of exercise on age-related basal increases in IL-6 and IL-1β expression in the hippocampus is in accordance with prior reports (Nichol et al., 2008; Barrientos et al., 2011; Gomes da Silva et al., 2013; Littlefield et al., 2017). However, others report exercise-induced reductions in one or both IL-6 and IL-1β (Speisman et al., 2013; Gibbons et al., 2014; Dallagnol et al., 2017). While collectively these data all support anti-inflammatory effects of wheel running, the specific pattern varies. One potential consideration for these seeming different findings is that aged rodents, like humans, are not a homogenous group. Prior work has shown that aging differentially affects memory; with some aged rodents maintaining cognitive function while others show age-related deficits (Foster, 2006; Gerstein et al., 2013). These variations in cognitive status in the aged correlate with alternations in cytokines and chemokines (Scheinert et al., 2015), suggesting heightened neuroinflammation contributes to age-related cognitive deficits. Moreover, while aging is associated with enhanced inflammation, the degree of inflammation varies across aged individuals (Katsel et al., 2009; Bradburn et al., 2017; Walker et al., 2017). Individual differences in the age-related neuroinflammatory profile may determine the response to interventions such as exercise, though more work is needed to confirm this possibility.

The present data indicate duration dependent effects of exercise on hippocampal IL-10 expression, as significant elevations in gene expression were only seen in the long-term exercise group. Treadmill running has also been reported to increase hippocampal IL-10 levels in aged rodents relative to controls though these changes were detected after 10 days of treadmill access (Gomes da Silva et al., 2013). In humans, increases in serum levels of IL-10 were found in physically active aged individuals (Jankord et al., 2004). Interestingly, prior work has reported that after 18 weeks of wheel running access, aged rats show no changes in serum or hippocampal IL-10 levels, potentially indicating the effect dissipates with a longer exercise paradigm (Speisman et al., 2013). In contrast to the present findings, Dallagnol et al. (2017) reported reductions in hippocampal IL-10 in aged mice after 8 weeks of running wheel access. These seemingly differential effects of exercise on IL-10 may need to be considered in the context of the microenvironment. In experiments that reported no change or reductions in IL-10 following exercise also observed reductions in IL-1β (Speisman et al., 2013; Dallagnol et al., 2017). In contrast, experiments that found exercise-induced increases in IL-10 were associated with unchanged IL-1β (Gomes da Silva et al., 2013) in agreement with the present data. Given the regulatory effects of IL-10 on inflammatory cytokines, like IL-1β, (Sawada et al., 1999) one possibility is that exercise promotes IL-10 in the presence of ongoing inflammation whereas induction of IL-10 is unnecessary when inflammation has been resolved. One important note is that reductions in hippocampal protein levels of IL-10 were detected after 10 days of treadmill exercise (Gomes da Silva et al., 2013) whereas in the present study exercise-induced reductions in IL-10 gene expression were only observed in the long-term exercise group. While both forms of exercise have beneficial effects, the behavioral and physiological changes induced by wheel running and treadmill exercise are not identical (Burghardt et al., 2004; Cook et al., 2013; Kim et al., 2020). Potentially, inherent variations in the parameters (e.g., intensity, access length, voluntary or forced) of these exercise protocols may account for the duration-dependent changes in IL-10. Additionally, these differences may relate to assessing protein versus gene expression across the present and Gomes da Silva et al. (2013) study. The present study also found evidence of region-specific changes in IL-10 in response to exercise. While hippocampal expression was increased by long-term exercise, assessment of IL-10 protein levels in brain samples showed a trend for lower IL-10 following short- and long-term exercise, though the difference was not significant. Prior work has shown region-specific changes in cytokine levels in the aged as well as region-specific regulation of cytokines by exercise (Chennaoui et al., 2008; Speisman et al., 2013). Increasing IL-10 in a region characterized by heightened inflammation may represent one mechanism through which exercise regulates neuroinflammation in the aged.

Exercise, whether short- or long-term, had selective effects on neuroimmune-related markers as not all measures showed alterations in response to exercise. For instance, age-related increases in hippocampal gene expression of the astrocyte marker GFAP and the costimulatory molecule CD86 associated with microglia activation were unaffected by exercise. The effects of exercise on astrocytes are mixed, as some report exercise reduces the proportion of GFAP staining or cell numbers (Latimer et al., 2011; He et al., 2017), others report increases (Li et al., 2005; Saur et al., 2014), and some find no differences in agreement with the present data (de Senna et al., 2011; Belaya et al., 2020). Similarly, markers associated with microglia are not consistently reported to be altered by exercise (Mela et al., 2020). However, the lack of an alteration in GFAP and CD86 expression is unlikely to indicate that exercise has no effect on glial cells, as numerous papers report alterations in glia morphology as well as shifts in activation following exercise (Moulin et al., 2007; Kohman et al., 2013; Saur et al., 2014; Littlefield et al., 2015). For instance, several papers report that exercise alters microglia proliferation in both the young and aged brain (Ang et al., 2004; Ehninger et al., 2011; Kohman et al., 2012). Similarly, astrocytes show exercise-induced changes in select morphology measures that may or may not be accompanied by alterations in GFAP (Saur et al., 2014; Belaya et al., 2020). The lack of a change in the current data may reflect the use of global markers that may not capture specific alterations. Moreover, many markers associated with microglia activation are expressed on other cell types that may contribute to the lack of a difference (Magnus et al., 2005). Assessment of IL-33 showed an age-related increase in hippocampal expression of IL-33 relative to young mice. IL-33 can act as an endogenous alarmin that coordinates the inflammatory response as well as engages neuroprotective processes that mediate repair and recovery (Moulin et al., 2007; Gadani et al., 2015; Fairlie-Clarke et al., 2018). The increase in IL-33 hippocampal expression was present in both the aged controls and those in the short-term exercise group. This age-related increase in IL-33 may reflect an attempt to restore tissue homeostasis through engaging repair processes. Prior work has shown IL-33 deficiency results in greater age-related neurodegeneration, indicating its neuroprotective functions in the aged brain (Carlock et al., 2017). Following extended exercise, hippocampal IL-33 gene expression no longer differed from young controls. However, the aged long-term exercise group did not significantly differ from aged controls or those in the short-term exercise condition. These data are inconclusive and may require longer periods of exercise, increasing sample size, or use of alternative methods of IL-33 assessment to dissociate the influence of exercise on IL-33 in the aged hippocampus. Additionally, we assessed hippocampal expression of the neurotrophic factor GDNF and its associated receptor GFRα1. GDNF promotes survival and differentiation of neuron populations during development and in the hippocampus during adulthood (Bonafina et al., 2019). In agreement with the present data, 3-6 months of exercise had no effect on GDNF levels in young adult mice (Faherty et al., 2005; Revilla et al., 2014). In contrast, shorter wheel access of only 14 days increased hippocampal GDNF expression in young adults. Potentially, GDNF may be elevated in response to short periods of exercise in young adults whereas no effect is seen with longer durations. The current data showed no change GDNF or GFRα1 expression with aging or in response to short- or long-term exercise. Prior work indicates that aging may alter induction of GDNF, as young rats show elevated GDNF in response to a lesion whereas aged rats fail to show an increase (Yurek et al., 2001). While exercise may influence GDNF in young adults, under some conditions, aged adults appear to show a differential response to exercise.

One limitation of the present data is the assessment of protein and gene expression changes in different samples. ELISAs were conducted with brain samples (i.e., everything except the hippocampus) and RT-PCR was completed with hippocampal samples. Direct comparison of protein and gene expression in the same samples would have clarified if the observed transcriptional changes resulted in protein differences within a given brain area. An additional limitation is the selective use of males as experimental subjects. Prior work demonstrates that while both males and females acquire a neuroinflammatory profile in response to aging, differences in the nature and severity are seen across the sexes (Berchtold et al., 2008). Similarly, both sexes benefit from engaging in exercise but females generally run greater distances when given running wheel access relative to males (Clark et al., 2011; Barton et al., 2017). Assessment of exercise-induced alterations in immune function show evidence of sex-related differences for select measures. For instance, Bay et al. (2017) report that males and females differed in their acute inflammatory response following administration of a liver carcinogen and that while exercise attenuated inflammation in both sexes the immune parameters, timing, and direction of change varied across the sexes. In aged mice, voluntary running has been recently reported to reduce age-related frailty in both sexes, but females showed reduced frailty scores slightly faster than males (Bisset et al., 2021). Moreover, aged females showed a positive correlation between exercise volume and serum cytokine levels, whereas no association was seen in aged males (Bisset et al., 2021). Our prior work shows that hippocampal microglia from aged males and females show differential response to exercise, as exercise decreased the proportion of microglia positive for major histocompatibility complex (MHC) II in females, but increased MHC II positive cells in males (Kohman et al., 2013). However, exercise attenuated the proportion of CD86 positive microglia selectively in aged males (Kohman et al., 2013). While the immunomodulatory effects of exercise are reported in both sexes, subtle differences have been observed. Whether these differential effects of exercise relate to the well described basal sex differences in immune function are presently unknown (Klein et al., 2016). Future assessment of sex-related differences in the immunoregulatory effects of exercise are needed to determine whether benefits, underlying mechanisms, and response to exercise parameters (e.g., duration, intensity) differ between males and females. Such knowledge may influence the therapeutic application of exercise across the sexes.

While exercise has clear benefits in the aged, few studies have evaluated the parameters of an exercise routine that may be required to observe these positive effects in an aged population. The present data indicate that both short- and long-term exercise can have immunoregulatory effects in the aged, but the extent and potentially stability of these changes increase with a longer duration of exercise. The present experiment demonstrated the exercise-induced alterations observed after short-term exercise are maintained with longer periods of wheel access, and other changes are only detected after long-term exercise. Moreover, increases in hippocampal neurogenesis are only detected after long-term exercise. Given that age-related neuroinflammation contributes to reductions in neurogenesis (Gemma et al., 2007), the requirement of a longer exercise duration to enhance neurogenesis may reflect the time needed to sufficiently reduce inflammation in the aged brain. Many of the benefits of exercise likely result from changes to the microenvironment that support brain health and diminish neuroinflammation. Collectively, the current findings indicate that exercise duration is an important consideration in instituting an exercise routine, as aged mice show greater benefits from engaging in longer periods of exercise.

Highlights.

• Long-, but not short-, term exercise increased hippocampal neurogenesis in the aged.

• Hippocampal IL-10 is increased only after long-term exercise.

• Brain IL-6 levels were attenuated by short- and long-term wheel running.