Use It and Boost It With Physical and Mental Activity

INTRODUCTION

The adage “Use it or lose it” has become a common admonishment, particularly among the elderly who may hope to use life-enriching experiences to protect against memory decline and the accompanying neuropathology associated with neurodegenerative diseases (Churchill et al., 2002). The conventional wisdom that stimulating experiences can benefit brain health is supported by a growing literature that cognitive (Albert et al., 1995; Kramer et al., 1999; Tyas et al., 2007; Basak et al., 2008; Hall et al., 2009) and physical (Yaffe et al., 2001; Kramer and Erickson, 2007; Anderson et al., 2010; Grazina and Massano, 2013; Intlekofer and Cotman, 2013) fitness reduce the risk of Alzheimer’s disease and Parkinson’s disease and may buffer against healthy age-related deficits. Mental and physical workouts can bolster general physiological health, thereby protecting against age-related disorders such as diabetes and cardiovascular disease that in turn can lead to neural and cognitive decline (Fillit et al., 2002).

Cognitive training in healthy older adults can reverse age-related deficits particularly for processes that tap brain regions marked by structural or functional changes with age (Diamond et al., 1975; Ball et al., 2002; Fillit et al., 2002; Hillman et al., 2008; Kramer and Erickson, 2007; Erickson et al., 2009). Cognitive training in youth can also produce improvements in learning and memory; indeed, learning itself is broadly defined by changes in subsequent performance as a result of prior experience. In most cases, assessments of the improvements afforded by prior training are quite selective identifying gains within rather narrow cognitive domains and memory systems tapped by the prior experience (Markowska and Savonenko, 2002; Green and Bavelier, 2008; Kleinknecht et al., 2012). However, there are studies showing that prior training can modulate subsequent learning of different tasks (Buschkuehl et al., 2008) and contexts where the benefits of prior experience generalize broadly, such as enriched environments benefiting many functions including object recognition, cued and contextual conditioning, spatial learning, and motor ability (Greenough et al., 1972; Kolb and Gibb, 1991; Rampon et al., 2000; Dishman et al., 2006; Frick and Benoit, 2010).

Likewise, benefits from aerobic exercise can be gained for brain health throughout the lifespan. Observational studies in adolescents and older adults suggest that individuals with higher levels of physical activity and increased levels of fitness present with better cognitive outcomes and accompanying brain volume savings compared to less fit age-matched individuals (Yaffe et al., 2001; Weuve et al., 2004; Hillman et al., 2008; Erickson et al., 2011). When examined in older adults, those that are more fit or active demonstrate less cognitive decline and cortical volume loss than observed in more sedentary individuals. Findings from young adults have been less consistent perhaps because physical activity benefits those functions most vulnerable to decline with age and health status (Dishman et al., 2006) and most studies assessing young adults examine college students who are relatively healthy.

These physiological and cognitive correlates of fitness may relate to a variety of demographic factors such as nutritional status, comorbidity, and access to health care. However, interventions of regular physical activity with varying duration and frequency, improve overall physiological health across a wide array of model organisms and paradigms (Dishman et al., 2006). Health benefits afforded by exercise are many and include increased aerobic capacity, decreased adiposity and associated body-mass index, tighter regulation of blood glucose and insulin signaling, and increased cardiac output (Anderson et al., 2010; Dishman et al., 2006). Exercise that increases aerobic fitness also improves learning, memory, and other cognitive functions, and increases hippocampal and cortical volume compared to similar measures in sedentary individuals (Kramer and Erickson, 2007; Erickson et al., 2009; Stroth et al., 2009).

Older adults engaging in regular aerobic activity show improvements in working memory, executive function, and visual-spatial processing (Colcombe and Kramer, 2003). Likewise, when exposed to a variety of exercise regimens, aged rodents show improved contextual fear memory, working memory, and spatial memory. In young adult and aged rodents, hippocampus-sensitive learning and memory such as spatial learning in the swim task, contextual fear conditioning, and radial maze performance improve following exercise, particularly after regimens that extend beyond 1–2 weeks (van Praag et al., 1999; Adlard et al., 2004; Albeck et al., 2006; Hopkins et al., 2011). However, exercise periods as short as 1 week can facilitate spatial learning in the swim task (Vaynman et al., 2004) and intermittent exercise given every other day for 4 weeks enhances object recognition memory (Hopkins and Bucci, 2010). Exercise interventions in animal models of brain damage or disease can reverse or slow behavioral deficits and in some cases protect against or reverse degeneration when initiated before damage or at specific delays after damage (Smith and Zigmond, 2003; Will et al., 2004; Woodlee and Schallert, 2006; O’dell et al., 2007). Together, the results support the likelihood that exercise can modulate neural mechanisms that can, in turn, improve cognition particularly in conditions or contexts leading to cognitive impairments such as in aging and neurodegenerative disease.

MULTIPLE MEMORY SYSTEMS APPROACH

Much of what is known about the neural mechanisms of enriching experiences such as exercise and cognitive training focuses on individual brain structures or systems, for example the hippocampus, visual cortex, and prefrontal cortex (Churchill et al., 2002; Raz et al., 2005; Intlekofer and Cotman, 2013). Less known is whether the outcomes of experiences like physical and mental activity that prime subsequent learning are revealed as increased independent function of multiple memory systems, increased competition, increased collaboration, or some permutation of all three.

Our multiple memory systems approach to understand the effects of prior experience on cognition focuses on hippocampus-sensitive and striatum-sensitive learning strategies to test how physical and mental activity modulate different types of cognitive functions that dissociate by the canonical brain systems engaged during learning. Through converging approaches including disrupting performance with functional or structural lesions (Packard et al., 1989; Packard and McGaugh, 1996; White and McDonald, 2002; Chang and Gold, 2003a, 2004; Featherstone and McDonald, 2004; Hallock et al., 2013) and correlating performance with measures of functional (Iaria et al., 2003), chemical (Chang and Gold, 2003b; Gold et al., 2013), and molecular activity (Colombo et al., 2003; Colombo, 2004), the hippocampus is believed to be important for allocentric learning strategies in place tasks and the striatum for egocentric strategies in response tasks (however see Ragozzino et al., 1999 for evidence of collaboration across hippocampus and striatum). While there are undoubtedly numerous brain regions that are engaged during different cognitive tasks, using task-specific dissociations between the hippocampus and striatum and site-specific dissociations between place and response learning has become a testing ground in our laboratory for ideas that probe mechanisms of competition, collaboration, and independence across memory systems.

Such dissociations in place and response learning can be seen as robust shifts in learning strategy associated with fluctuations in circulating levels of ovarian (Korol and Kolo, 2002; Daniel and Lee, 2004; Korol et al., 2004) and adrenal (Sadowski et al., 2009; Schwabe et al., 2010) hormones, as well as ethanol (Matthews et al., 1999). For example, treatment with systemic estrogens improves place learning, yet the same treatments impair response learning (Korol and Kolo, 2002; Zurkovsky et al., 2006, 2007; Pisani et al., 2012). Likewise, exposure to stress, stress hormones, or ethanol shifts male rats from place strategies toward response strategies. Stated in a different way, estrogen deprivation, stress, and ethanol exposure actually impair place learning yet improve striatum-sensitive response learning. Using a memory systems lens, it is possible that the opposing effects of these treatments are due to modulation of interactions between these two memory systems. On the surface, the findings following systemic treatments “look and smell” like competition between the hippocampus and striatum, as shown by the reciprocal improvements and impairments in response and place learning.

If these data were obtained under conditions of hippocampal or striatal lesions, respectively, one might conclude that the hippocampus and striatum compete for cognitive control over learning strategy (White and McDonald, 2002). For example, estradiol treatment may produce better performance on place learning through direct effects on the hippocampus, thereby reducing learning ability in response learning through competitive interactions between the hippocampal and striatal memory systems. The complement might also hold: estradiol may produce reduced functional output of the striatum, lifting the competitive hold on the hippocampus. However, when estradiol was infused into the hippocampus for 2 days, place learning enhancements were observed without any effect on response learning (Zurkovsky et al., 2007). The converse was seen following striatal infusions where response learning impairments were found without changes in place learning (Zurkovsky et al., 2007). Thus, the results reveal independent enhancement and impairment of learning across multiple memory systems rather than a competitive relationship between them.

A similar argument for competition between memory systems could be made for the strategy-shifting effects of other systemic treatments such as exposure to stress, adrenal hormones, or ethanol (Matthews et al., 1999; Sadowski et al., 2009; Schwabe and Wolf, 2009; Schwabe et al., 2010). Analogous direct tests of competition in the shifts from place to response or from place to cued learning seen following stress hormone or ethanol treatments have not yet been made. Our results with estradiol, however, suggest that before making conclusions about how treatments modulate interactions between memory systems direct manipulations of or measurements in each brain region using both the canonical and noncanonical tasks are required.

Within this memory systems framework, we describe evidence here that prior physical and mental activity improve both hippocampal and striatal function. The most parsimonious explanation for these general findings is that prior experience acts without producing competition between these memory systems; direct tests of competition or collaboration are currently underway. Moreover, we build a case for shared mechanisms underlying the experience-induced modulatory effects that involve signaling through brain-derived neurotrophic factor (BDNF). Testing for effects of priming experiences on later learning and their mechanisms in female rats is also of high interest but not yet undertaken as these studies would require additional independent measures of hormone status.

GENERALITY OR SPECIFICITY OF PRIMING

Our driving question is whether both hippocampus-sensitive place learning and striatum-sensitive response learning benefit from prior physical and mental activity. The reasoning behind this is as follows. If priming with physical or cognitive activity induces neural plasticity across a broad range of brain regions, we should expect to see an equally broad range of benefits on cognitive tasks. However, if the brain changes associated with priming are confined only to regions that show certain markers of plasticity such as neurogenesis (van Praag et al., 1999, 2005) then we would expect to see benefits only on tasks that tap those brain regions such as hippocampus and cortex. In this way, the pattern of cognitive enhancements might identify memory systems that are sensitive to modulation by priming and point to candidate neural mechanisms for the cognitive changes. Patterns of dissociations in place and response learning following priming could also suggest interactions and independence between these memory systems.

How the modulatory effects of priming map onto multiple memory systems will likely be, in part, a function of the mechanisms driving the effects. If brain health benefits from exercise, for example, through general physiological and metabolic mechanisms related to fitness (Dishman et al., 2006; Anderson et al., 2010) such as increased VO₂ max (Kramer et al., 2002), anti-inflammation (Cotman et al., 2007), or improved cardiovascular function (Colcombe et al., 2004) including angiogenesis (Isaacs et al., 1992; Black et al., 1990; Anderson et al., 1994), then we might expect exercise-induced improvements to confer quite broadly to many—if not all—brain regions and cognitive functions. However, if the effects of exercise differentiate by brain structures modulated by physical activity, then we would expect to see changes in cognitive function that are constrained by task attribute. Similar constraints on the modulatory effects of prior mental activity may be even more readily seen given that cognitive training likely taps a more selective set of memory systems and neural mechanisms than does exercise on a running wheel (Sirevaag and Greenough, 1987; Black et al., 1990; Isaacs et al., 1992; Klintsova et al., 2004; Bekinschtein et al., 2011).

It is important to note that training paradigms that are more or less selective for a given task or cognitive attribute, based on findings from lesions or measured as changes in functional output, have been well developed and tested under various hormonal and pharmacological contexts (Gold and Korol, 2010). One such task, spontaneous alternation (SA), which takes advantage of a rat’s natural tendency to alternate arm choices and to explore novel arms of a maze, actually results in increased extracellular levels of acetylcholine and lactate in both hippocampus and striatum (Pych et al., 2005; Newman et al., 2011; see Gold et al., 2013) suggesting activation of both structures. Thus, SA, in its various forms, can now be considered a mixed task, though long considered a hippocampal task by many (Ellen and Deloache, 1968; Stevens and Cowey, 1973; Johnson et al., 1977; Lalonde, 2002). Paradigms that require the use of place or response strategies for task solution may dissociate more cleanly vis-α-vis changes in extracellular levels of acetylcholine (Chang and Gold, 2003b; McIntyre et al., 2003; Pych et al., 2005), in extracellular lactate (see Gold et al., 2012), or long-lasting activation of signaling molecules such as pCREB (Colombo et al., 2003). Different behavioral tasks can thus be used to dissociate or associate the priming effects of physical and cognitive activity.

REGULATION OF PLACE AND RESPONSE LEARNING THROUGH PHYSICAL AND MENTAL ACTIVITY

BDNF MEDIATES PRIMING FROM PHYSICAL AND MENTAL ACTIVITY

There are a number of widely accepted mechanisms for how exercise and other enriching experiences promote neural plasticity and improve learning and memory, including augmenting neurogenesis, angiogenesis, and synaptogenesis. One molecule that appears to regulate all of these processes is BDNF, a protein that belongs to a family of growth factors believed to play important roles in neural development and plasticity. BDNF protein levels are relatively high in the rat hippocampus and cortex, but BDNF is also measurable in the striatum and other structures such as the olfactory bulbs (Conner et al., 1997; Ding et al., 2004). Furthermore, BDNF can be released from astrocytes (Parpura and Zorec, 2010) and from neurons either pre- or postsynaptically (Poo, 2001; Horch and Katz, 2002; Edelmann et al., in press). Both pre- and postsynaptic release has been observed for hippocampus (Horch and Katz, 2002) with trophic actions acting both post- and presynaptically, but in the striatum BDNF is thought to signal post-synaptically only through axonal release (Jia et al., 2010; Edelmann et al., in press). These BDNF release patterns across brain regions may produce differences in neural plasticity (Edelmann et al., in press) that, in turn, may promote subtle differences in task-specific responses to priming such as durability of memory.

BDNF messenger RNA and protein are increased in hippocampus (Neeper et al., 1995; Adlard et al., 2004; Klintsova et al., 2004; Molteni et al., 2004; Vaynman et al., 2004; Berchtold et al., 2005) and striatum (Ding et al., 2004; Marais et al., 2009) following regular physical activity and in hippocampus after engaging in learning tasks (Kesslak et al., 1998; Klintsova et al., 2004; Hall et al., 2000; Mizuno et al., 2000; Gomez-Pinilla et al., 2001), thereby making BDNF a viable candidate molecule for mediating our priming effects. We measured hippocampal and striatal concentrations of mature BDNF (mBDNF), the ~14 kDa protein cleaved from the larger proBDNF form that signals primarily through tropomyosin receptor kinase B (TrkB) tyrosine kinase. Brain samples were taken from separate groups of untrained rats that underwent either 3 weeks of exercise (Fig. 3A) or 20 min of SA testing (Fig. 3B). In control rats, mBDNF protein concentrations were almost four times higher in the hippocampus than in the striatum (data not shown). However, voluntary exercise led to elevations in mBDNF that were particularly robust for the striatum (Fig. 3A). Likewise, tissue levels of mBDNF were elevated ~100% in hippocampus and 250% striatum 1 h following SA testing compared to untested controls (Fig. 3B), suggesting that upregulation of BDNF levels, signaling, or both may mediate the priming effects of physical and cognitive activity on both place and response learning.

The role of BDNF in activity-induced plasticity is not a new thought. BDNF is a highly popular molecular target for such brain changes, particularly those observed in the hippocampus. However, what is not clear is whether increased BDNF from physical and mental activity produces the substrates for plasticity that, in turn, lead to cognitive enhancements, or produces increased BDNF signaling itself that is in fact necessary during cognition to produce enhancements, or both. If elevated BDNF produces a brain that is better prepared for subsequent learning and is otherwise unnecessary, then BDNF signaling during learning would not be needed. If, however, BDNF signaling is, or is part of, the mechanism for activity-induced enhanced learning, then blocking signaling selectively during learning would abrogate the observed priming-induced enhancements. Bekinschtein et al. (2011) raise similar issues in their critical analysis of findings pointing to neurogenesis as the substrate for cognitive enhancements from enrichment and exercise, and propose instead BDNF as the mediator of enhancements. In support of this proposition, it has been found that pharmacological, molecular, and immunological blockade of BDNF actions prevents exercise-induced enhancements in cognition (Vaynman et al., 2004; Bekinschtein et al., 2007, 2011; Liu et al., 2009). However, for the most part, the role of BDNF during the priming experience has not been explicitly dissociated from its role during subsequent cognitive testing. For example, intrahippocampal infusions of TrkB immunoglobulin G (IgG) effectively sequestering BDNF during the week of voluntary exercise blocked exercise enhancements of learning in the spatial version of the swim task (Vaynman et al., 2004). Because TrkB-IgG treatment suppressed downstream markers of BDNF signaling when measured after the training period, it is likely that blockade continued beyond the exercise phase and throughout learning. Furthermore, blocking TrkB activation at the end of exercise training but 3 days before training attenuated memory in an inhibitory avoidance task (Liu et al., 2009). With this experimental protocol, it is again unclear whether BDNF blockade was effective because of actions on exercise-related plasticity or on signaling during inhibitory avoidance training. Finally, while there are many tests of BDNF involvement during learning per se, whether similar BDNF-mediated signaling is needed for cognitive priming and is similar between hippocampus and striatum is unknown.

We evaluated whether BDNF signaling in the hippocampus and striatum at the time of learning and memory was necessary to observe the priming effects of physical and mental activity. Just before place or response learning, but after exercise or SA testing, BDNF signaling through its cognate TrkB receptor was blocked. K252a, a tyrosine protein kinase inhibitor known to prevent TrkB phosphorylation and downstream cascades (Tapley et al., 1992; Vaynman et al., 2004), or saline vehicle was infused bilaterally into hippocampus or striatum 30 min before place or response training, respectively. Rats were previously primed with 3 weeks of exercise or with 20 min of SA testing 1 h previously and compared to sedentary controls without running wheels or behaviorally naïve controls without prior alternation testing.

We observed the expected enhancements in place and response learning following exercise (Fig. 4) and following SA (Fig. 5) as seen by reduced trials to reach criterion in both tasks. K252a into the hippocampus blocked place learning enhancements from prior exercise and from cognitive testing. Likewise, K252a into the striatum attenuated the response learning enhancements that resulted from priming with exercise or with cognitive testing. What was striking was that the attenuation of priming effects under TrkB blockade was remarkably similar for place and response learning (compare left and right panels in Figs. 4 and 5) and for the two priming modes, i.e., 3 weeks of exercise and 20 min of SA (compare Figs. 4 and 5). These results suggest that BDNF signaling may represent a shared mechanism for experience-induced plasticity in both memory systems and may reflect a common process that underlies different types of enriching experiences in general.

Given the substantial evidence implicating BDNF in neural plasticity, learning, and memory (Mizuno et al., 2000; Schinder and Poo, 2000; Alonso et al., 2002; Lu, 2003; Lee et al., 2004; Ou and Gean, 2005; Minichiello, 2009; Bekinschtein et al., 2011), we were surprised to find that K252a failed to disrupt place or response learning in sedentary and naïve controls. In fact, there was a nonsignificant trend for K252a to enhance learning in these rats, which may reflect the ability of unopposed or high doses of K252a to activate other cell signaling pathways known to be involved in learning and memory (Roux et al., 2002). Our findings, however, suggest that K252a attenuation of enhanced learning in exercising rats was not a result of simple additive effects of drug impairments and priming-induced enhancements. This raises the possibility that physical and mental exercise create different avenues for neural plasticity such that brains of more fit rats are more likely to signal through BDNF during learning while those of less fit rats use other mechanisms.

Our current working model for the role of BDNF signaling in learning and memory is shown in Figure 6. The model reveals many potential pathways that might contribute to the actions on cognition. Our results to date demonstrate changes in BDNF levels in striatum and hippocampus after cognitive and physical exercise and demonstrate further that signaling at the time of training via TrkB receptors in striatum and hippocampus mediates the effects of exercise on cognition. While the parallels across memory systems are striking, unknown is whether there is divergence of mechanisms downstream from TrkB receptors.

The results observed for both exercise and SA priming suggest that BDNF signaling in the canonical memory systems regulate enhancements in learning resulting from both prior physical and mental activity. Because TrkB signaling was blocked only in the canonical memory system, interactions between hippocampus and striatum, whether competitive or collaborative, were not examined. To test for interactions, effects of hippocampal TrkB blockade on response learning and striatal TrkB blockade on place learning would need to be assessed. However, the results do point to cell signaling pathways that may be broadly involved in cognitive enhancements produced by enriching experiences.

Importantly, blockade did not coincide with either the 3 weeks of exercise or the SA testing, thereby allowing the role of BDNF and its downstream effectors during learning to be teased apart from the actions of BDNF during prior priming. We believe that an important property of priming with exercise and behavioral training is increased BDNF signaling in hippocampus and striatum at the time of learning regardless of the neural plasticity that ensues during the priming period. Moreover, it is clear that one site of action is in the striatum where neurogenesis is not typically found but where BDNF and its cognate receptor are. Thus, while neurogenesis and other processes such as angiogenesis and synaptic plasticity may permit or provide a foundation for plasticity from BDNF signaling, they are not sufficient themselves to support the enhanced place and response learning we observe following prior physical or mental activity. It is possible that those processes produce increased levels of available BDNF that when released facilitate learning.

CONCLUSIONS

The findings obtained with estrogens and with physical and cognitive exercise each exhibit attributes important to understanding multiple memory systems. Showing that enhancement of place learning and impairment of response learning by estrogens are mediated separately by actions at the hippocampus and striatum, respectively, reveals independence of these memory systems rather than competitive interactions between them. Therefore, while competitive interactions are often evident after brain damage, these interactions cannot always be presumed to explain up- and downregulation of learning after manipulations like hormonal status, stress, or aging in the absence of direct tests of the relative contributions of each memory system.

Results obtained with physical and cognitive exercise also reveal independence of, rather than competition between, the systems. Moreover, molecular and pharmacological evidence in both the striatum and hippocampus that BDNF contributes to the enhancing effects of both forms of exercise point to possible independence of actions within both memory systems. However, it is also possible that the simultaneous upregulation of both neural systems does not disrupt the normal balance of contributions of these systems to learning and therefore does not test competition vs. independence of the systems. It is possible too that signaling mechanisms, including some relating BDNF to energy metabolism, downstream of BDNF receptor activation will add even richer information about the bases for independent and interactive processes involving multiple memory systems.