Given the fact that no effective drugs to treat dementia disorders are available to date, the research focus has more and more shifted toward more promising preventive approaches in recent years. In the last decade, physical inactivity has been repeatedly shown to be one of the major risk factors associated with dementia (Barnes and Yaffe, 2011; Norton et al., 2014). A study by Weuve et al. (2004) was one of the first to show that long-term physical activity enhances cognitive function and lessens cognitive decline in older women (Weuve et al., 2004). Furthermore, a recently published 29-years-long follow-up study with a Finnish twin cohort disclosed that long-term vigorous physical activity during adulthood significantly decreases the risk to develop dementia in later life and leads to a lower mortality rate (Iso-Markku et al., 2015). Importantly, beneficial effects of regular exercise have also been reported in subjects already suffering from mild cognitive impairment (MCI) or dementia (Heyn et al., 2004; Nagamatsu et al., 2013).

Proposed mechanisms disclosing how physical activity lowers the risk of dementia in humans are diverse and include enhanced cerebral blood flow as well as synthesis of neurotransmitters among others (Coelho et al., 2014). Furthermore, it has been reported that higher levels of aerobic exercise are associated with an increased hippocampal volume and improved spatial memory in older subjects (Erickson et al., 2011) and that physical activity reduces hippocampal atrophy in individuals at genetic risk for AD (Smith et al., 2014).

The positive effects of regular exercise on brain health in humans has been tried to be explained by data obtained in experimental animal models. Therefore, rodents are subjected to environmental enrichment (EE) paradigms, which mimic a cognitively and physically active lifestyle (Nithianantharajah and Hannan, 2006). A large amount of recent studies have reported that, e.g., rodent AD models maintained under enriched living conditions show improved performances in hippocampus-dependent cognitive tests such as the Morris water maze (MWM; Polito et al., 2014; Rao et al., 2015; Huttenrauch et al., 2016). Consistent with improved spatial memory performance, exposure to physically stimulating environments has been shown to induce structural and functional changes in various brain regions, with the hippocampus being specifically sensitive to exercise (Cotman and Berchtold, 2007). The effects of physical activity may be partially mediated through the induction of growth factors such as brain derived neurotrophic factor (BDNF), insulin-like growth factor (IGF; Gomez-Pinilla et al., 2002; Cassilhas et al., 2012) or effects on hippocampal neurogenesis (Fabel et al., 2009; Garthe et al., 2016), however, the precise molecular mechanisms through which exercise benefits the brain remain to be elucidated. The ultimate goal would be the development of “enviromimetic” drugs, which mimic or increase the effect on the brain induced by environmental enrichment. Therefore, we aimed on further investigating the molecular changes upon physical activity by performing a transcriptome analysis using a deep sequencing approach in C57BL6 wild-type (WT) mice subjected to prolonged environmental enrichment. Here, we investigated the long-term effect of environmental enrichment lasting from weaning until 12 months of age in C57Bl6/J mice. We demonstrate that continuous mental and physical stimulation improves long-term spatial reference memory and increases the number of neurons in the dentate gyrus (DG). Finally, a deep sequencing analysis identified novel genes that may mediate the beneficial effects of prolonged periods of exercise.

The exploratory and spontaneous locomotor activity of enriched housed WT mice was compared to standard controls in the open field paradigm. No significant differences in the total time of activity (Figure 2B), the total distance traveled (Figure 2C) and the average speed (Figure 2D) were detected between the two groups. The time spent in the center region can be used as a measure of anxiety-related behavior. Mice maintained under EE conditions spent significantly more time in the center region when compared to sedentary controls (p < 0.05; Figure 2A), indicating a loss of anxiety.

Spatial reference memory was assessed in standard and enriched housed WT mice using the MWM task. 12-month-old WT mice housed under SH and EE conditions showed progressively decreased escape latencies over 3 days of cued training as expected (Figure 3A), while no differences in swimming speed were detected (Figure 3D). During the following acquisition training phase, mice housed under enriched conditions showed significantly reduced escape latencies compared to their sedentary controls (Two-way repeated measures ANOVA, p = 0.0187; Figure 3B), indicating improved spatial learning. Again, no differences in swimming speed were detected (Figure 3E). During the final probe trial, both WT SH and EE mice showed a clear preference for the target quadrant, indicating that both groups learned the task. However, mice housed under EE conditions showed a significantly increased time in the target quadrant compared to the sedentary controls (t-test, p = 0.0354; Figure 3C), with again unchanged swimming speed (Figure 3F). Representative occupancy plots confirmed that enriched-housed WT mice focussed their search on the initial platform position, while sedentary mice showed a more random search profile (Figure 3C).

In order to evaluate whether the improved spatial reference memory performance can be attributed to an altered hippocampal neuron number, stereological quantifications of CA1 and DG neuron numbers were performed (Figure 4A). Between SH and EE housed WT mice, no significant difference in the number of CA1 neurons could be detected (WT-SH: 293319 ± 34662; WT-EE: 291751 ± 21897; p > 0.05; Figure 4B). A quantitative analysis of the CA1 volume instead revealed a significant increase of ∼29% in EE compared to SH animals (p = 0.0001; Figure 4C). We next analyzed whether an altered number of DG neurons might contribute to the observed beneficial effects of the EE paradigm. Mice housed under EE conditions harbored ∼15% more DG neurons than sedentary mice (WT-SH: 451001 ± 61864; WT-EE: 517122 ± 30443; p < 0.05; Figure 4D). An analysis of the DG volume also revealed an increase of ∼36% in EE compared to SH mice (p = 0.0015; Figure 4E). In order to evaluate, if an increase in the neurogenesis rate contributed to the altered DG neuron number, DCX-positive neurons were quantified in the SGZ in both EE and sedentary WT mice. No significant differences were detected between WT-SH (985.0 ± 195.4) and WT-EE mice (1027.0 ± 236.2; Figure 5).

In order to analyze if long-term housing under EE conditions has an influence on the brain gene expression profile of WT mice, a deep sequencing analysis on whole brain hemispheres was performed (Figure 6A). Prolonged exposure to enriched conditions led to statistically significant changes in the expression of 137 genes with a log2-fold value of at least 0.38 (corresponding to a change in gene expression of 1.3-fold; Top15 up- or down-regulated genes, Table 1). 103 genes were up-regulated and 34 genes were down-regulated in WT EE mice compared to SH littermate controls (Supplementary Table S1; Figure 6B). A gene ontology analysis of the up-regulated genes using the Database for Annotation, Visualization and Integrated Discovery (DAVID v6.7) bioinformatics package (Huang et al., 2009) revealed significant associations, while only gene ontology categories presenting both p-value < 0.05 and Benjamini adjustment < 0.05 were considered: for the up-regulated genes, an association with the GO terms “transcription” (GO:0006351; Benjamini p-value 1.8E-3), “regulation of transcription” (GO:0006355; Benjamini p-value 4.6E-3), “regulation of RNA metabolic processes” (GO:0051252; Benjamini p-value 6.0E-3) were found, whereas down-regulated genes were associated with GO term “extracellular region” (GO:0005576; Benjamini p-value 2.3E-5) and “extracellular region part” (GO:0044421; Benjamini p-value 2.3E-3).

To validate the differentially expressed genes (DEGs) identified by deep sequencing analysis, both up- and downregulated candidate genes were randomly selected and verified by qRT-PCR. For all selected genes, the qRT-PCR analysis confirmed the expression levels of the deep sequencing results (Figure 6C).

In recent years, great progress has been made in investigating the effect of physical activity on disease progression in preclinical models of Alzheimer’s disease and other neurodegenerative disorders by the use of environmental enrichment paradigms. However, cognitive impairment, dysregulation of neurogenesis and brain atrophy are not only associated with diseases, but also with normal aging. Therefore, the aim of the present study was to elucidate, whether a prolonged environmental enrichment paradigm, providing both cognitive stimulation and increased physical activity (Fabel et al., 2009), has an impact on behavioral performance, hippocampal neuron numbers and cerebral gene expression changes in aging WT mice. The enrichment was started after weaning at the age of 4 weeks and was continued until the age of 12 months. Increased stress levels have been associated with enhanced anxiety phenotypes. Although corticosterone levels have not been assessed in the present study, it has been previously shown that beneficial effects of a 2-months exercise (Cassilhas et al., 2012) or 6-months enrichment period (Cracchiolo et al., 2007) occurred without any changes in blood cytokine or corticosterone levels. Animals allowed to exercise showed a significantly increased time in the center of the open field task, corresponding to a reduced anxiety phenotype and confirming previous data (Benaroya-Milshtein et al., 2004; Kazlauckas et al., 2011). Therefore, we do not expect that altered stress levels affect the behavioral analysis in the present study. Analysis of long-term spatial reference memory using the MWM task revealed a significantly increased performance in the probe trial of enriched compared to sedentary mice. This finding corroborates previous reports describing an improved spatial memory task performance following long-term running in rodent AD models (Garcia-Mesa et al., 2011; Huttenrauch et al., 2016), while related results have been also obtained in WT mice using shorter enrichment protocols (Garthe et al., 2016). However, most available reports use enrichment protocols with much shorter duration of mainly up to 6 months. It has been previously demonstrated that enriched housing increases DG neurogenesis (van Praag et al., 1999; Kobilo et al., 2011), even in aged mice (Kempermann et al., 2002) and that hippocampal neurogenesis is a mechanism critically involved in learning processes (Garthe et al., 2016). On the other hand, strongly decreased overall neurogenesis rates during aging have been also reported (Kuhn et al., 1996; Merkley et al., 2014). In good agreement, the number of new-born doublecortin (DCX)-positive neurons was unaltered between enriched and sedentary WT mice, however, stereological analysis revealed a significantly increased total neuron number in the DG in WT EE, which might reflect an increase in neurogenesis and survival of new born cells in earlier life periods. WT mice maintained under a cognitively and physically stimulating environment displayed a larger hippocampal volume, which suggests that improved cognitive performances are correlated to structural changes in the brain. This is in good agreement with human studies showing that physical activity is associated with an increased volume of the hippocampus (Erickson et al., 2011). Enrichment-induced increases in the CA1 spine density in mice (Rampon et al., 2000), as well as enhanced dendritic arborization and increased dendritic spine density in layer III cortical neurons (Leggio et al., 2005) have been previously reported. In accordance with our finding, significant increases in the CA1 volume, as well as CA1 and DG dendrite length have been demonstrated in enriched mice using Golgi-Cox staining (Faherty et al., 2003). Interestingly, increased numbers of CNPase-positive oligodendrocytes have been detected in CA1 and DG of enriched rats (Zhao et al., 2013) which might also contribute to an overall volume increase in these brain regions. Very recently, chronic imaging of pyramidal cell morphology revealed that exposure to EE results in a 30–40% increase in spine density in 3–4 month-old animals. This could be confirmed in adult animals to a slightly lesser degree (20–30%; Jung and Herms, 2014).

An up-regulation of the gene encoding BDNF has been repeatedly reported (Gomez-Pinilla et al., 2002; Adlard et al., 2005) and is believed to act as a mediator of the efficacy of exercise on synaptic plasticity and cognition (Vaynman et al., 2004). Therefore BDNF served as a control measure in the current study and the BDNF mRNA levels were found to be significantly up-regulated following 11 months of enrichment.

The objective of the transcriptome analysis using deep sequencing was to identify novel genes that might play a role in mediating the beneficial effects of long-term physical activity and environmental enrichment. Changes in anxiety-related behaviors, as reflected in the open field analysis, appear to be regulated by an interconnected system of brain structures including the basolateral amygdala and as lesions in a variety of brain regions including hippocampus, striatum, but also cerebellum and cerebral cortex were shown to impair MWM performance (D’Hooge and De Deyn, 2001), a whole brain RNA-sequencing approach was performed. On the other hand, this approach also includes limitations, as global gene-expression changes cannot be directly related to the morphological alterations detected in the hippocampus and enrichment-specific differences in hippocampal gene expression might be covered or thinned out.

The gene PTGDS encoding prostaglandin D2 synthase was one of the most strongly down-regulated genes that could be identified. It is expressed mainly in astrocytes and oligodendrocytes and acts as a neuromodulator (Urade and Hayaishi, 2000). In vitro experiments have demonstrated that prostaglandin D2 mediates neuronal damage by mimicking the effects of Aβ1-42, leading to microglial activation and neuron loss (Bate et al., 2006). PTGDS has been also identified to be down-regulated in the senescence-accelerated SAMP8 mouse model (Alvarez-Lopez et al., 2013), as well as in the Tg4-42 mouse model of AD, where long-term physical activity led to an amelioration of hippocampal neuron loss and a rescue of behavioral deficits (Huttenrauch et al., 2016).

Several gamma-aminobutyric acid receptor subunits (GABRA2, GABRB2) were found to be up-regulated in mice maintained under enriched environment (EE) conditions. Further, GRIN2B encoding the N-methyl D-Aspartate receptor subtype 2B (NMDA2B, NR2B) is up-regulated following long-term exercise, which could be confirmed by RT-PCR analysis. This finding corroborates earlier studies describing increased NR2B levels in the DG of wheel-running Sprague-Dawley rats (Farmer et al., 2004), as well as in the 3xTg mouse model of AD after 6 months of voluntary exercise (Revilla et al., 2014). Expression levels of the NR2B subunit are selectively decreased in the hippocampus and entorhinal cortex of AD patients compared controls (Bi and Sze, 2002) and an up-regulation of NR2B subunit expression is suggested to enhance synaptic plasticity and memory functions in a broad range of behavioral tasks in rodents (Wang et al., 2014).

Information about effects of physical activity on changes of GABAergic neurotransmission are scarce, however, it has been demonstrated that high anxiety rats subjected to chronic restraint stress showed decreased expression of alpha-2 GABA-A subunits in motor cortex and DG (Wisłowska-Stanek et al., 2013). In contrast, EE mice in the current study show reduced anxiety together with increased GABRA2 expression levels. In good agreement, the N900 mouse strain selectively bred for aggression showed an increased anxiety phenotype together with selective reductions in alpha-2 GABA-A subunit levels in frontal cortex and amygdala (Nehrenberg et al., 2009). The reduced anxiety phenotype together with increased GABRA2 expression observed in the present study supports such an association.

Two of the most strongly up-regulated genes were FOSB and FOSL2. Members of the Fos family can dimerize with proteins of the Jun family to form the transcription factor complex AP-1 (Curran and Franza, 1988). It has been previously demonstrated that the levels of the truncated splice form of FosB, ΔFosB, are potently triggered by long-term physical exercise in the hippocampus and cortex of C57BL/6 mice (Nishijima et al., 2013).

Genes belonging to chaperone families implicated in protein processing within the endoplasmatic reticulum (ER) were also upregulated (CRYAB, HSPA1B, HSPH1). These genes play important roles within ER-associated degradation pathways and are necessary for quality control during protein synthesis by assisting proper folding and protein modifications (Liu and Chang, 2008). HspB5 or α-B-Crystallin (Cryab) has been previously demonstrated to be induced in the brain as a function of short-term exercise in rats (Hu et al., 2009), as well as HSP70 (HspA1B) in the hippocampus and prefrontal cortex following swimming exercise (Liu et al., 2010). We could recently show that a group of similar HSPs were also upregulated upon EE in a novel mouse model for AD, indicating a common mechanism associated with long-term cognitive and physical stimulation (Huttenrauch et al., 2016).

We provide evidence that prolonged physical activity results in reduced anxiety levels, improved learning behavior and an altered gene expression profile in WT mice. This corroborates findings from previous studies using more short enrichment protocols and supports epidemiological data from human aging and AD obtained in retrospective settings.

MH, GS, and OW performed experiments or analyzed data. OW designed the study and wrote the manuscript together with MH. All authors read and approved the final manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.