Given that age-related decrements in cognitive function occur heterogeneously in both human and animal model populations, and that cognitive decline occurs on a background of aging-related proteomic alterations, studies pairing behavioral assessments of cognitive function with quantitative proteomic analyses at multiple ages are needed. As the relationship between neurochemistry and behavior has moved to the forefront of neuroscience research, investigations of the effects of spatial learning and memory on the hippocampal proteome have been initiated (Table ). The introduction of additional complexities of study design must be taken into consideration in these investigations to maximize the utility of resultant data. In addition to the intricacies of the proteomic techniques, clearly described animal model and behavioral testing methods are necessary; if these important details are lacking, interpretation of neuroproteomic datasets in a biological context is hindered, ultimately minimizing the impact of the findings. This is illustrated by a potentially influential report describing changes in protein–protein interactions induced by the formation of spatial memory (Nelson et al., 2004
). Rats of undescribed strain, sex, and age were divided into three training groups (Morris Water Maze trained, yoked swim control, and naïve). Following training and demonstration of a spatially focused search pattern in the maze-trained animals, a complex combination of proteomic techniques was used to study differential protein interactions. Hippocampal protein samples were subjected to affinity-based enrichment of protein–protein complexes and depletion of non-interacting proteins. Identification of interacting proteins significantly regulated with spatial learning was performed by separation using 2DE, followed by LC–MS analysis. Of the nearly 250 proteins determined to participate in binding interactions, 32 were statistically increased in protein–protein interaction in the trained versus swim control group (e.g., stathmin, tropomyosin, tubulin, GFAP, spectrin, F-actin capping protein B) while eight decreased their interactions (e.g., actin, complexin 1). Seven proteins with differential binding between swim control and naïve groups were interpreted as indicators of water maze-induced stress. The majority of proteins with spatial memory-induced interaction alterations were classified as synaptic, structural, and signaling proteins. Five of the protein interactions determined to be regulated with spatial memory formation and for which a protein component was identified were selected for pull-down of binding partners. For this work, bait proteins were expressed in bacteria for capture of putative interacting proteins from hippocampal extracts The captured proteins were identified by LC–MS/MS and quantitated by immunoblotting, which demonstrated the following protein–protein interactions: complexin/p25α and complexin/drac1-like protein (decreased with training); CapZ/tubulin, adenylate kinase/actin, and ATP synthase/14-3-3 gamma (increased with training). Interestingly, these changes appeared to be specific to the protein–protein interactions themselves, as total expression of these individual species was not altered.
This study demonstrates the value of combining traditional biochemical and modern proteomic approaches with behavioral testing to provide information regarding neuronal protein–protein interactions that participate in spatial memory formation that could not have been obtained without prior knowledge of at least one binding partner in these interactions. The report, however, is rather technology-driven and lacks information regarding animal strain, age, sex, and behavioral training/testing, as well as a rationale for determining protein–protein interactions in bacteria instead of the hippocampal protein samples from these animals. Although significant confirmatory and control experiments are needed to validate the differential interactions reported with maze training in this study, future studies implementing this approach may provide further insight into the acute effects of spatial learning and memory on the hippocampal proteome and on specific protein–protein interactions that may affect cognitive function, particularly with increasing age.
Using a more traditional proteomic approach, Henninger et al. (2007
) present a well-designed and highly informative study pairing a spatial learning task with 2DE-based quantitation of cytosolic proteins in the rat hippocampus. Forty young-adult male Wistar rats (mean age 3 months) were either trained to locate a hidden escape platform in the Morris Water Maze based on surrounding contextual cues, or were placed in the maze with no platform present to control for exercise and stress in the absence of spatial learning. Maze training was performed twice daily for 7 days, and all animals were sacrificed immediately following the last trial for 2DE quantitation of hydrophilic hippocampal proteins. Protein spots with significantly different expression between maze-trained and control groups were determined to be associated with spatial learning. Nearly 80 unique proteins, representing both training-induced increases and decreases, were identified by MS, and primarily represented metabolic, cytoskeletal, signal transduction, and degradation processes. Enolase, actin-interacting, and proteasome/lysosome subunits were among the subset of proteins increased after spatial learning, while numerous proteins related to synaptic function including NSF, synapsin 2, dynamin, and the 14-3-3 isoforms gamma and epsilon, were decreased. An added strength of this study is the inclusion of immunoblot confirmations to validate neuroproteomic data (i.e., protein identification and quantitation). As the authors point out, many proteins involved in the neuronal processes underlying learning and memory were absent from this analysis, likely because membrane-associated proteins were depleted by the isolation method used, and also because these low-abundance proteins fall below the detection threshold in traditional neuroproteomic techniques. One unfortunate limitation of this work is that proteins determined to be unchanged by spatial training were excluded from identification, which would have added to the growing coverage of the rodent hippocampal proteome and provided insight regarding proteins not necessarily critical for learning and memory.
In a well-designed study of plasticity-induced proteomic alterations, McNair et al. (2006
) demonstrated both temporal and class-specific protein expression pattern changes in ex vivo
hippocampal slices prepared from mice aged 4–8 weeks. Plasticity was simulated both electrophysiologically, using high-frequency stimulation of CA1–CA3 pyramidal cells to induce LTP, and pharmacologically, using glutamate-induced receptor activation. Both acute and chronic alterations in protein expression induced with these models of synaptic plasticity were examined by 2-DIGE and identified by MS and MS/MS. Four hours following glutamate treatment, long-lasting altered expression of 79 proteins was detected, 58 of which were sensitive to NMDA receptor blockade. Many of these proteins were related to glutamate receptor cycling, cytoskeleton regulation, and vesicle trafficking. For example, glutamate stimulation downregulated NSF and upregulated actin, which was confirmed by two-dimensional immunoblotting. More than 80% of hippocampal proteins identified were related to metabolism, cytoplasmic organization and biogenesis, and protein transport and modification. Electrophysiological LTP induction led to altered expression of energy metabolism-related proteins after 10
min, and of cytoplasmic organization- and biogenesis-related proteins after 240
min. LTP-mediated protein changes included calcineurin, VDAC1, MEK1, growth-associated protein 43, synapsin 2, syntaxin 19, and copine 6. Little overlap in proteins with altered expression was detected between the 10 and 240
min time points, although there were similarities between electrophysiological and pharmacological treatment groups. These findings indicate that, in ex vivo
models of hippocampus-dependent learning and memory, distinct temporally regulated proteomic alterations are induced. These changes may represent neuromolecular processes underlying acquisition and retention of spatial memory.
Building on these results, a follow-up study was conducted to examine the hippocampal neuroproteome following environmental enrichment, which has been demonstrated to increase neuronal activity and induce spatial learning and memory formation, on the hippocampal neuroproteome (McNair et al., 2007
). Eight-week old male hooded Lister rats were exposed to an open field (non-enriched) or enriched environment for six weeks, for a final age at sacrifice of 14 weeks (i.e., young-adult). The enriched regimen included toys, running wheels, and novel objects known to stimulate hippocampal plasticity and learning. Neural somata were segregated from their distal processes by dissecting CA1 stratum pyramidale and stratum radiate to allow examination of somatic and dendritic proteomic composition, respectively. Proteomic analyses were performed by 2-DIGE and highly sensitive quadrupole MS. Rats subjected to environmental enrichment exhibited altered expression of numerous proteins in both somatic and dendritic samples, with both inductions and reductions observed compared to non-enriched controls. In total, 50 proteins were identified regardless of expression, approximately 70% of which are involved in energy metabolism, signal transduction, and cytoplasmic organization and biogenesis. In stratum pyramidale, environmental enrichment increased expression of actin, dynamin, rab7, and proteasome subunits, and decreased expression of protein phosphatase 3, synapsin 1, PSD95, dynactin, collapsin response mediator protein 2, destrin, and ubiquitin. Interestingly, many proteins altered with enrichment were identified as housekeeping gene products, indicating that these species may play more complex roles in neuronal function than traditionally thought. In stratum radiate, exposure to the enriched environment was associated with increased phosphoglycerate mutase 1, ATP synthase subunits, VAMP-associated protein B, GFAP, and actin, while GFP58, GNBP alpha inhibiting protein 2, septin 5, cyclin-dependent kinase, and diacylglycerol kinase zeta were decreased. This study design did not control for cognitive or proteomic effects of increased exercise in rats exposed to the enriched environment versus the open field, but it demonstrates two important points. First, exposure to environmental enrichment alters components of the hippocampal proteome, some of which have been previously implicated in learning and memory while others are novel. Second, distinct proteomic changes are induced in CA1 neuronal processes versus cell bodies, indicating that fractionation approaches are necessary for detecting and differentiating these protein expression alterations. Although the findings of this study are quite valuable to the cognitive neuroscience field, additional work examining the effects of environmental manipulation and exercise mature adult and aged rats is needed.
Exercise is known to improve neuronal activity and cognitive function, and exerts neuroprotective effects in aging hippocampus (O'Callaghan et al., 2009
; Gomes da Silva et al., 2010
; Jedrziewski et al., 2010
; Voss et al., 2010
). The effects of exercise on the hippocampal neuroproteome has been evaluated in a noteworthy study of aged male Sprague Dawley rats reported by Chen et al. (2007a
). Sedentary rats were compared voluntarily (running wheel) and involuntarily (treadmill) exercising rats after an 18-month exercise regimen started during early adulthood. At 23 months of age, the animals were sacrificed for 2DE-MS/MS analysis of the hippocampal neuroproteome. Surprisingly, of 74 signaling proteins identified, 15 species were regulated in expression with exercise, and only two of these (protein phosphatase 1 and guanine nucleotide-binding protein 1) were common to both voluntary and involuntary exercise compared to sedentary animals. The 15 differentially expressed proteins were represented in either involuntary exercise versus voluntary exercise or involuntary exercise versus sedentary rats, indicating a major effect of consistent treadmill running throughout adulthood. Interestingly, expression of signaling proteins such as PEBP, 14-3-3 zeta, septin 8, and the early endosome fusion protein EEA1 was significantly different between voluntarily and involuntarily exercising rats, suggesting that voluntary exercise impacts processes underlying learning and memory to a greater extent than involuntary exercise. In agreement with recent findings that exercise improves hippocampus-dependent spatial learning and memory, many of the exercise-regulated proteins identified here have been associated with cognitive function. Additional work pursuing exercise-mediated regulation of memory associated proteins with aging will no doubt increase our understanding of the role of exercise in improved cognition throughout the lifespan.
As illustrated by the comparison of voluntary and involuntary exercise above, implementation of neuroproteomic approaches in behavioral neuroscience research has led to the identification of task-dependent regulation of specific proteins. Differential effects of spatial learning paradigms have been demonstrated in young-adult (10–14 weeks) wild-type mice trained in common behavioral tests of hippocampus-dependent cognitive function (Zheng et al., 2009
). In the Barnes Maze, mice were trained to locate a dark escape chamber hidden below one of 20 holes in a brightly lit circular platform based on fixed visual cues. Mice trained in the Multiple T-Maze learned to locate a goal box containing a food pellet in a maze designed with seven choice points. For both paradigms, yoked controls were included to minimize confounding effects of stress or food deprivation. Following a probe trial to ensure that trained mice had successfully acquired and retained the assigned spatial task, hippocampal proteomic composition was evaluated by 2DE with nano-LC–MS identification of proteins of interest. Several differentially expressed proteins were identified between mice exposed to the different training paradigms. Although most of these proteins are associated with metabolism and have not been previously linked to synaptic mechanisms of learning and memory, several proteins, including dual-specificity protein phosphatase 3, calmodulin, and NSF, are implicated in synaptic signaling, demonstrating the importance of synaptic function and maintenance in cognitive function. The maze-dependent specificity of proteomic profiles described here emphasizes the difficulty of comparing behavioral studies and even between identical rodent strains, and the need for careful reporting of testing paradigms in neuroproteomic evaluations in cognitive neuroscience research. As with age-based neuroproteomic investigations, comparing studies of the hippocampal neuroproteome with models of learning and memory formation is made difficult by study-to-study variation in study design. This is particularly true with regard to the behavioral testing methods used, potential confounding variables such as stress, food deprivation and exercise, and the age and species of study subjects.
It is interesting to note that several proteins implicated in hippocampus-dependent learning and memory by neuroproteomic investigations are differentially expressed with aging. For example, 14-3-3 signaling proteins, which decrease in expression with increasing age (VanGuilder et al., 2010
), are regulated at the levels of protein content and protein–protein interaction following spatial training and exercise (Nelson et al., 2004
; Chen et al., 2007a
; Henninger et al., 2007
). Similarly, members of the family of dihydropyrimidinase-related proteins (also known as collapsin response mediator proteins) are regulated both with aging (Carrette et al., 2006
; Focking et al., 2006
; VanGuilder et al., 2010
) and in models of learning and memory formation (McNair et al., 2006
; Henninger et al., 2007
; Zheng et al., 2009
). Dihydropyrimidinase-related proteins contribute to signal transduction and regulation of microtubule dynamics and cytoskeletal remodeling (Charrier et al., 2003
; Brittain et al., 2009
). Lastly, regulation of protein phosphatases and PEBP (also called hippocampal cholinergic neurostimulating peptide) has been identified in the aged hippocampal neuroproteome (Weinreb et al., 2007
; Freeman et al., 2009
; VanGuilder et al., 2010
). These proteins have been previously implicated in learning and memory, and are among the proteins changed in expression following exercise, environmental enrichment, and training in spatial learning tasks (Chen et al., 2007a
; McNair et al., 2007
; Zheng et al., 2009
), suggesting that age-related dysregulation of these proteins may contribute to deficits of spatial learning and memory.