Brain-Derived Neurotrophic Factor: A Key Molecule for Memory in the Healthy and the Pathological Brain

1The brain derived neurotrophic factor (BDNF) belongs to a family of neurotrophins that have a crucial role in survival and differentiation of neuronal populations during development. In the adult brain, BDNF also maintains high expression levels and regulates both excitatory and inhibitory synaptic transmission and activity-dependent plasticity.

2The expression of BDNF is regulated during transcription and translation, and also by post-translational modifications. The presence of a complex multi-level regulation demonstrates the importance and diversity of BDNF functions. Transcription is controlled by multiple promoters that determine activity-dependent and tissue specific expression. There have been identified at least four BDNF promoters in the rat, each one driving the transcription of mRNAs that contain one of the 8 non-coding exons spliced to the common 30 coding exons, which produce an heterogeneous population of BDNF transcripts. BDNF splicing has been described for several species, including humans, mice, and rats. Additionally, the expression of specific BDNF exons can be regulated by epigenetic mechanisms, suggesting that environmental experiences dynamically influence mature BDNF levels.

3Regarding the pattern of expression of BDNF in the brain, high levels of this molecule have been detected in the hippocampus, amygdala, cerebellum and cerebral cortex in both rodents and humans, with the highest levels found in hippocampal neurons. Lower levels of BDNF have been detected in organs such as the liver, heart, lung, among others. The regulation of each transcript is controlled and/or modulated by factors like neuronal activity, exercise, antidepressants, stress, and hormones such as estrogens.

4Brain derived neurotrophic factor is synthesized as the precursor proBDNF, that can be stored in either dendrites or axons, and undergoes cleavage intra or extracellularly to produce a mature BDNF protein. BDNF is released in an activity dependent manner as a mixture of pro and mature BDNF. Interestingly, BDNF and proBDNF are associated with opposing effects on cellular function, which gives BDNF protein function an additional level of complexity. The proBDNF form is secreted under both pathological and non-pathological conditions. ProBDNF preferentially binds p75 NTR receptor, which facilitates LTD and induces apoptosis. On the other hand, BDNF in its mature form binds specifically to tyrosine kinase receptors (TrkB) and promotes cell survival, facilitates LTP and increases spine complexity. When p75NTR is co-expressed with TrkB receptor it increases neurotrophins binding affinity thereby facilitating ligand discrimination. In this way, proBDNF can be thought as part of a regulatory mechanism of BDNF activity in non-pathological conditions. In addition, the truncated forms of TrkB receptor can act as dominant negative inhibitors of BDNF signaling by internalizing and clearing BDNF from the synapse (; Figure 1).

5Many studies have shown the critical role of BDNF for the regulation of plastic changes in the adult brain, including regulation of the trafficking, phosphorylation and expression levels of NMDARs associated with augmented synaptic strength. Due to its critical role in LTP, BDNF has been postulated to be an essential part of the cellular mechanism supporting memory formation and maintenance by promoting synaptic consolidation. According to this hypothesis, BDNF increases memory storage by favoring changes in spine morphology leading to the stabilization of LTP. BDNF can also increase the number, size and complexity of dendritic spines, probably through unpregulated actin polymerization. Furthermore, BDNF increases neurogenesis through changes in cell survival and proliferation.

6Changes in synaptic connections are thought to support memory storage. There are several lines of evidence that directly link BDNF with learning and memory. For example, BDNF could be a mediator of the plastic changes underlying both spatial and recognition memory processes.

7In this review, we will focus on the role of BDNF in cognitive function in the adult brain under normal and pathological conditions, and evaluate the potential therapeutic actions of BDNF for the treatment of cognitive alterations associated with aging, neuropsychiatric and neurodegenerative diseases. We will particularly focus on the effects of these treatments on mnemonic function.

8In the human BDNF gene, a common single nucleotide polymorphism identified with a Met to Val substitution at codon 66 in the pro-domain of BDNF, called rs6265 or Val66Met polymorphism, affects synaptic targeting of BDNF-containing vesicles and activity-dependent neuronal release of BDNF. Met66 homozygous knock-in mice showed a selective impairment in activity-dependent synaptic plasticity in vitro. Additionally, exogenous application of proBDNF in Val-carriers facilitated LTD and inhibited LTP, while not in Met carriers. Because of its frequency in the human population [up to 30% Met carriers in a European sample] and its association with lower serum levels of BDNF, this single nucleotide polymorphism has been a matter of great interest. This polymorphism has been associated with structural and functional differences in the brain, such as synaptic plasticity and memory performance. Some of these structural changes include volumetric decreases in specific regions -such as the hippocampus, the parahippocampal gyrus, the prefrontal cortex and the amygdala-. The presence of the Met allele is, in turn, associated with poor mnemonic performance on verbal tasks both at short and long delays and also deficits in working and spatial memory tasks. Since most studies relayed mainly on tasks that evaluate item memory, and particularly, verbal memory, the effects over other more complex cognitive functions remain to be tested. did evaluate multiple types of memories, finding only an influence on visuospatial memory, although describes an additional effect of associative memory.

9Besides these correlational studies of Met allele dosage, hippocampal volume/activation and memory performance in healthy subjects, evidence for a role of Val66Met polymorphism in brain structure and function is not conclusive and, even meta-analysis studies show patterns of conflicting results. conducted three separate meta-analyses to determine the effect of the Val66Met polymorphism on declarative memory performance, hippocampal volume and hippocampal activation in humans. They reported that all these measures are reduced in carriers of the Met allele compared to Val homozygotes and this effect cannot be explained by random variables such as age, gender or diagnosis criteria. However, argued that the effect sizes of fMRI data are susceptible to the method used to select the voxels and that the one used in Kambeitz et al. may have led to an inflated estimate of the effect size. In fact, the meta-analysis conducted by did not detect a significant association between the BDNF Val66Met polymorphism and several phenotypes including general cognitive ability, memory, executive function, visual processing skills, and cognitive fluency. They hypothesized alternative scenarios to explain this incongruence in the literature and proposed that instead of being grouped by their behavioral similarity, cognitive phenotypes should be categorized depending on the brain activation pathways engaged. Although an important group of studies could not found an association between Val66Met genotype and memory performance, the results linking BDNF polymorphism and memory are not likely spurious. It is possible that differences in the effect that the BDNF gene exerts across the lifespan and the uncontrolled effects of variables such as gender, stress and physical exercise, known to affect BDNF levels, may have led to a dilution of the effect in certain samples. Furthermore, Met carriers might be able to compensate for the deficiency in BDNF levels. In fact fMRI studies suggest that they present increased medial temporal lobe activity during the engagement of an episodic memory task when compared with non-carriers, and this might “hide” the potential deficits. In conclusion, these studies suggest that presence of the Met allele may confer a disadvantage in cognitive performance, and particularly episodic memory (for a review, see ), but that the effects of BDNF polymorphism may be too complex to be analyzed under the idea of a simple “risk allele.”

10Aging is a major risk for the development of cognitive disorders. Cognitive performance gradually declines with age, probably as a result of cellular and metabolic changes that lead to a progressive reduction in synaptic plasticity in brain regions crucial for cognitive functions. Aging is related to a significant shrinkage of gray matter, an important reduction in the number of synapses and also changes in neuroplasticity-related proteins. The hippocampus is a brain region with high levels of plasticity-related molecules and is particularly sensitive to aging-related cellular alterations that lead to volume reductions. Particularly numerous changes were reported in the dentate gyrus local inhibitory and excitatory circuitry. Furthermore, decreases in adult neurogenesis in the dentate gyrus start emerging at middle age and continue throughout the aging process. Other structures also linked to episodic memory formation, such as the frontal and entorhinal cortices, also show volume reduction and may contribute to the deterioration of cognitive function.

11At the cellular level, changes in hippocampal LTP have also been reported in aged animals, with deficits either related to induction, maintenance and/or expression of LTP depending on the stimulation pattern used for induction. Given that alterations in the thresholds for LTP and LTD in the MTL region have been associated with impairments in long-term memories (for review, see ), age-related cognitive impairment could also be linked to altered LTP function.

12In fact, one of the most consistent plasticity-related deficits associated with aging is the reduction in neurotrophic signaling. BDNF-LTP in the dentate gyrus is impaired in the aged brain, and may be reverted by endogenous BDNF induction by manipulations such as ampakine administration, which can also ameliorate age-associated memory deficits in rodents. Since dysfunction of synaptic plasticity and changes in neuronal activity are associated with worse performance in different cognitive tasks, efforts have been done to link changes in BDNF levels to aging-dependent cognitive decline and to the related alterations in structural and functional integrity of neuronal networks (for review, see ). suggested that age-dependent impairment in cognitive function could be associated with decreases in BDNF expression in the primary regions of the brain affected by aging. Consistent with this idea, the circulating concentrations of BDNF are reduced in aged primates and humans and brain concentration reduced in rats. In aged rodents, the BDNF system is affected at different levels, including reduced transcription, protein synthesis and processing, however, other publications could not find an association between age-related plastic changes and BDNF. The found reductions correlate with hippocampal shrinkage, spatial memory decline and neuronal atrophy. For example, higher BDNF mRNA induction levels were reported after Water Maze task on unimpaired rats in comparison with aged animals. The decrease in BDNF levels observed in aged individuals, was accompanied by a reduction in the expression and/or activation of TrkB receptor and a concomitant increase in the levels of proBDNF and p75NTR, suggesting the presence of additional age-related deficits in BDNF signaling pathway and in the processing of proBDNF to mature BDNF. Additionally, these changes were negatively correlated with performance on the Water Radial Maze. Furthermore, the induction of specific BDNF transcripts after fear conditioning is altered in aged rats, suggesting that aging is not only associated with reduced neurotrophin expression under resting conditions, but might lead to a functional impairment of BDNF in response to a specific task. This is consistent with the aforementioned role of BDNF in activity and experience-dependent structural and functional connectivity changes.

13However, some researchers found no change or even an increase in BDNF associated with aging. This could suggest that the loss of hippocampal BDNF is not part of the mechanisms involved in age-related cognitive decline. However, considering that neuronal cell death is an undoubtedly important element of age-related cognitive impairment, a transient BDNF-related response to neuronal degeneration could be the underlying reason for the increasing levels of hippocampal BDNF reported in some studies. However, since in vivo measurements of BDNF in the brain are not possible, studies in humans rely on inferences about the levels of BDNF on central nervous system. Given that there is evidence that BDNF can cross the blood-brain barrier, these studies assume that serum BDNF is a proxy of BDNF levels in the brain. Consistently, measures of BDNF in the central nervous system (brain BDNF, b-BDNF) correlate with measures of BDNF from the periphery (serum BDNF, sBDNF) with as much as 75% of brain origin BDNF. However, the functional significance of BDNF in the serum is a matter of debate. BDNF is also secreted in several peripheral sites such as platelets, lymphocytes, and skeletal and smooth muscle cells, and also recent studies have questioned the idea of BDNF being able to cross the blood-brain barrier. Another important consideration in these measurements is whether pro and mature forms of BDNF are being measured, because only some assays can differentiate mature BDNF and, considering that they can have opposing effects, these should be taken into account. For this reason, the results of serum studies should be interpreted cautiously, as this could be a potential cause for inconsistencies reported in sBDNF levels between studies. The use of enriched extracellular vesicles of neuronal origin from peripheral blood could provide a novel way to bypass these issues since they could more closely reflect brain changes compared to plasma.

14There are several reasons to think that a decrease in BDNF levels could be detrimental to normal brain functioning, like its role in synaptic plasticity, as described above. Another motive is that BDNF exerts various trophic effects on hippocampal neurons that could help counteract the noxious effects of neuronal cell death. In fact, high levels of BDNF in the hippocampus are related to both survival and differentiation of dentate gyrus progenitor cells in the adult, and low levels of BDNF have been linked to deficient neurogenesis in aged animals. Moreover, BDNF is known to increase with oxidative stress as part of an antioxidant defense during aging. Neuronal loss is an important component of normal aging, however, it does not contribute significantly to learning and memory impairments, suggesting that memory deficits associated with aging are more likely related to alterations in synaptic physiology and aberrant cell signaling, that might contribute to an altered connectivity.

15While aging-related neuroanatomical changes are evident, there is an enormous variability amongst healthy individuals in the pattern of cognitive decline. Memory performance is also partly under genetic control, probably because of age-sensitivity in many cognitive processes. Genetic variance could explain these individual differences in cognitive capacity, especially since heritability of cognitive function increases over the lifespan as brain resources decrease, and can account for as much as one third of the variance in cognitive decline. Both apolipoprotein E (APOE) and BDNF have been associated with variance in cognitive performance in healthy individuals more frequently than other genes, although some results have not been reliably replicated.

16Since BDNF decreases throughout life, it would be interesting to assess the possible associations between the Val/Met polymorphism and age-related cognitive decline. For example, reported poor verbal recall in a sample of healthy elderly Met homozygotes, and found that Met carriers showed greater age-related decline in hippocampal activation during both encoding and retrieval, while other studies also limited to older adults found no impact of age on cognitive tests evaluating learning and memory. Considering these results, this polymorphism could help understand individual differences in cognitive function by genetic-dependent changes in neurotransmitter and neurotrophic factor levels, amongst other factors. However, since reports indicate that both APOE and BDNF polymorphisms accounts for less than 2.3% of the variance, it is fundamental to take into account the complex interplay between associations with other genes and interactions with environmental factors to interpret these results. Many environmental and hormonal factors such as physical exercise, caloric restriction, estrogen levels, and environmental enrichment can influence BDNF levels, making it challenging to link BDNF to age-related memory impairment and hippocampal atrophy. For example, BDNF genotype can modulate the effect of physical exercise on episodic memory performance and brain volume. This is evident by the fact that only Val homozygous benefited from physical exercise with larger MTL volume and hippocampal gray matter, whereas in Met carriers the contrary effect was found. In another study, in a cohort aged 65 or older the strength of the association between incidence of cognitive decline and physical activity increased with the number of Met alleles, suggesting that the Met allele may confer vulnerability to dementia in elders with less physical activity. On the other hand, there are studies reporting a reduced vulnerability of Met carriers to age-related decline in executive function, pointing toward a differential effect of BDNF on cognitive function related to the areas supporting the task.

17The influence of BDNF on cognitive function may change across the lifespan. In fact, the effects of the BDNF Val66Met polymorphism on brain structure and cognitive function were found to differ in an age-dependent manner: while Met carriers showed a reduction in episodic memory performance and hippocampal/parahippocampal volume in samples of around 65 years in comparison to Val carriers, in the elderly (mainly samples of around 75 years) Val/Val individuals had diminished entorhinal cortex thickness, white matter tract integrity, and episodic memory performance. It has been hypothesized that this effect could be related to changes in the level of cleavage, since cleavage molecules such as tPA are known to decline with age which could create a paradoxical effect where greater BDNF secretion would, in fact, lead to cognitive decline. Another explanation could be a decrease in the penetrance of the BDNF genotype across the lifespan, as other factors such as the independent incidence of age-related diseases increase their influence on brain structure and cognition. In fact, there are results that suggest that Met carriers have more preserved frontostratial functions than Val/Val subjects, what may lead to an increase in the use of stratium-dependent mnemonic strategies, that could give a potential advantage on Met carriers that might hinder the original deficits reported in young individuals. This might help explain conflicting results in this regard, since there are studies showing that aged Met-carriers still have deficits in mnemonic performance and they experience a steeper impairment in memory tasks as they age and have diminished performance in remembering neutral faces when compared with Val/Val individuals.

18Aging is normally accompanied by a loss of memory function. Episodic memories are particularly more sensitive to the aging process than procedural or non-declarative memories. Within the limits of the episodic memory domain, some aspects can be more vulnerable to aging than others. For example, associative memory tasks that require binding of multiple pieces of information can be more sensitive to the aging process. This deficit is mostly related to spatial and recognition memory loss. The vulnerability of particular mnemonic processes to aging is probably due each of these functions being supported by distinct brain regions showing differential rates of functional decline with age. Diminished input from the EC to the DG could contribute to deprive the HP from sensory information crucial for discrimination of novel versus familiar stimuli when this stimuli are similar, and in fact reductions in DG-CA3 connectivity are associated with spatial learning decline in old animals. The medial temporal lobe region (MTL), that is thought to support episodic memory function, is particularly vulnerable to cellular alterations that happen during aging and/or pathological dysfunction. Since episodic memory decline is correlated to decreases in hippocampal volume, these changes could explain age related mnemonic deficits. Another important structure for episodic memory in the MTL is the perirhinal cortex (Prh), a region involved in the discrimination of novel and familiar stimuli that is particularly crucial to solve tasks involving ambiguous features. Aging alters discrimination in both in rats and humans increasing a propensity to identify novel stimuli as familiar. This effect was, in some cases, interpreted as a deficit in the ability to bind features of an object, so that decisions are made based on the familiarity of a single component. The consequence is the incapacity to detect novel compositions of familiar features. In fact, complexity and ambiguity of the features was proposed as a determining variable in recognition memory deficits in aged rats, as well as in Prh-lesioned animals using a configural task with complex objects. Several molecular and biochemical alterations have been reported in the Prh of aged animals that could contribute to the associated cognitive deficits. Since exposure to novel objects is related to an increase in BDNF levels in the Prh, and familiarity discrimination in the presence of ambiguous stimuli (and not clearly distinguishable ones) is impaired with BDNF antisense ODNs infusions during a restricted time window after the task, BDNF is an interesting molecular candidate that could help to establish a link between molecular and biochemical alterations and the pattern of both spatial and recognition memory deficits associated with aging. In particular, because optimal cognitive function is linked to efficient neuronal plasticity, these memory deficits might be coupled to alterations in the expression and regulation of plasticity-related proteins such as BDNF, a protein whose expression is both affected in the aging brain and is crucial for memory consolidation and particularly for discrimination of similar memories.

19In correspondence with this idea, decrease in BDNF expression has been associated with neuronal atrophy and death occurring in some neurological disorders. Administration of exogenous BDNF can prevent pathological changes in the nervous system associated with aging [but see for inconsistent results; for review, ], and can rescue both BDNF-induced LTP and spatial memory performance in aged animals. Since BDNF has been linked to synaptic plasticity, neurogenesis, neuronal survival and protection against brain insults, the above results imply the possibility that BDNF could act as a synaptic repair molecule. There are a few evidence that support this idea, for example, acute application of the TrkB agonist 7,8-dihydroxyflavone rescues synaptic plasticity in the hippocampus of aged rats in vitro. Additionally, chronic treatment also prevents age-related impairments in contextual and cued fear conditioning with a simultaneous normalization of the spine levels that normally decrease with age. Furthermore, the Lou/C rat, an animal model of successful aging that presents a preserved cognitive performance across its longer lifespan, showed higher hippocampal BDNF than Wistar rats and a decrease in proBDNF with age. This contrasts with the increase in proBDNF seen in aged Wistar rats. However, the beneficial effect of BDNF on neuroprotection and mnemonic performance in rats decrease as age increases, probably due to additional changes in the processing and signaling pathway. Consistent with the role of BDNF on synaptic plasticity and memory, elderly Lou/C rats never showed short- or long-term memory decline in recognition memory tasks or impaired LTP. However, found that learning-associated cognitive training could increase TrkB receptor expression in aged animals and also increased proBDNF processing both in aged and young rats, indicating that learning leads to a strengthening of BDNF pathway, especially in aged animals where this pathway is affected.

20Reduced levels of BDNF have been reported not only under normal aging conditions but also in pathological conditions including Huntington (HT), Alzheimer’s disease (AD), and Parkinson’s disease. However, the profile of cognitive deficits greatly differs between these pathologies according to the brain regions affected by degeneration. For example, the most profound BDNF deficits are reported in the hippocampus, parietal, entorhinal and frontal cortex for AD and in the striatum and motor cortex for HT. In this section we will focus on AD because it starts mainly as impairment in declarative memories, without affecting other neurological functions. It has been proposed that this feature is related to the degenerative profile of the disease that starts in the hippocampus, parahipocampal cortices and amygdala, but not in primary sensory and motor cortices.

21There is a substantial amount of studies supporting the idea that neurotrophic factors are crucial for the etiology of AD, in particular BDNF. BDNF protein and mRNA levels as well as proBDNF are reduced in the post-mortem brain of AD patients compared with age-matched controls, with no changes in TrkB levels. This reduction was also reported in Mild Cognitive Impairment (MCI), a potentially prodromal stage of AD. Furthermore, reduced circulating levels of BDNF were also found in MCI. BDNF levels are correlated to the severity of the disease and with episodic memory performance in patients, suggesting that these decreases could be related to the pathogenesis of the disease. In conclusion, downregulation of BDNF and proBDNF are thought to be an underlying mechanism related to early AD. However, found that patients in the early stages of AD had significantly higher sBDNF levels than patients in the late stages and also than age-matched controls. This highlights that it is difficult to establish a causal link between BNDF downregulation and the development of this neurodegenerative disease because the pathology is accompanied by a loss of cell density and dendritic spines that could secondarily affect BDNF levels. In this regard, there are also post-mortem and serum level studies that report an increase in BDNF and TrkB concentrations in the hippocampus and parietal cortex of AD patients. This increase may be related to compensatory mechanisms that could contribute to the repair by degradation of β-amyloid. In addition, other potential moderators could contribute to differences and heterogeneity seen in these studies. Differences in diagnostic criteria, stages of the disease, sex and education and the use of pharmacological treatments such as acetylcholinesterase inhibitors or psychotropic medication that are known to raise BDNF levels, or could come from other potential sources outside the CNS such as immune cells.

22Given that synaptic loss is the major correlate of cognitive impairment, much stronger than the presence of plaques or tangles, there is a recent view of AD as a “synaptic pathology”. Aβ monomers are normally generated and secreted at firing synapses, and are not toxic but neuroprotective as they have an active role in synaptic regulation and are crucial for neuronal function. Aβ monomers are one of the many factors that regulate synaptic function and they can activate CREB via the PI3K/AKT pathway, leading to a sustained CREB-regulated transcription and release of BDNF. In this way, BDNF can act as a converging point of many synaptic regulators. In Alzheimer’s disease (AD), neurotoxic β-amyloid (Aβ) oligomers are formed from the self-association of Aβ monomers. These oligomers can promote neurotoxicity through different ways. suggested that the selective vulnerability of memory related areas could be, in fact, a result of specific cellular modifications required for the process of memory consolidation. An important event in AD is the pathogenic Aβ-mediated alterations in the levels of neurotrophic factors (NTFs). Since pathogenic Aβ oligomers cannot can activate PI3K/AKT pathway and induce CREB activation, the increase in the levels of Aβ-oligomers can lead to an impairment in CREB activation in the brain of patients with AD and mouse models of AD. Soluble Aβ oligomers are known to alter signal transduction pathways crucial for learning and memory processes such as CREB-regulated transcription and trafficking of NMDA type of glutamate receptors. Thus, alterations in those pathways could play an important role in the etiology of the disease. Altered levels of BDNF in AD are downstream of Aβ-accumulation and could be related to Aβ–induced dysregulation of CREB transcription. Even if BDNF does not modify Aβ accumulation, it could have an important function in moderating the effects of Aβ on cognitive and structural aspects. BDNF protects against Aβ-mediated toxicity by contributing to its degradation and preventing tau hyperphosphorylation. In this sense, BDNF is expressed by microglial and astroglial cells in the plaque vicinity and seems to protect from neuroinflammation, thereby supporting neuronal survival and preventing apoptosis. On the other side, Aβ down-regulates BDNF mRNA in vitro via reduction of CREB and disrupts retrograde axonal transport of BDNF and conversion of pro-BDNF to mature BDNF. It also interferes with synaptic plasticity mediated by BDNF even at concentrations that do not kill the cells. This downregulation occurs before the appearance of plaques and is linked to memory deficits in AD animal models and in MCI. Tau, a mediator of Aβ-induced toxicity, can significantly downregulate BDNF via transcript IV both in vitro and in vivo by itself. As mentioned before, many studies found that decreases in serum BDNF levels can be detected in individuals with MCI, so it is tempting to speculate that BDNF loss could be involved as an early event in this synaptic dysfunctions. However, the presence of some inconsistencies between studies with MCI patients warns us to be cautious with these speculations. Nevertheless, these results suggest a critical role of BDNF in the regulation of Aβ-amyloid toxicity, suggesting that BDNF dysregulation could contribute to synaptic dysfunction and mnemonic impairment related to AD. This data implies that, although central to the development of AD, changes in BDNF expression could be an effect of earlier functional modifications in other synaptic related proteins. In particular, one of these proteins could be Aβ, that in its monomeric form has a normal physiological role in synaptic plasticity and neuronal survival in the brain and can actually have an active role in these BDNF changes by regulating BDNF transcription and release. In any case, the beneficial effects of BDNF on memory and cognition could reflect its synapse repair features.

23Changes in the cell microenvironment, where a lack of trophic support can lead to a decrease in neuronal survival and proliferative activity, could contribute to the degeneration of specific neuronal subpopulations in pathological conditions. During this period, changes in BDNF levels contribute to age-related hippocampal volume changes, and atrophy associated with pathological conditions. There is evidence of as much as 1–2% annual hippocampal atrophy in the elderly without signs of dementia, while in patients with AD this deterioration goes up to 3–5% per year. In patients with MCI, hippocampal volume is predictive of rapid conversion to dementia, evidencing its importance in the progress of the disease.

24Most studies report that BDNF genotype is not related to the risk of developing AD [but see for evidences of effects present only in women], and Genome Wide Association Studies could not find a relationship between BDNF Val66Met polymorphism and risk of AD. However, some studies do report an increase in the risk for AD in Val carriers. Other studies found an association between Met carriers and greater rates of decline in episodic memory and hippocampal atrophy in patients with MCI, leaving Aβ accumulation unaffected. Although there are certain inconsistencies among the literature, BDNF role in the development of AD seen with Val66Met has been replicated with other BDNF polymorphisms. The lack of consistency between studies could be related to differential effects of BDNF during distinct stages of the disease, with more circulating BDNF in MCI patients, and less in AD patients. Lower BDNF levels may be linked to neuronal death in AD, concealing any effect of the BDNF gene. Since the complexity of the pathological changes stemming from the disease increases as the severity progresses, associations between BDNF Val66Met polymorphism and AD should be more obvious in preclinical stages in which the disease presents almost exclusively subtle alterations in mnemonic performance.

25Neurotrophic factors not only moderate neuronal and synaptic dysfunction but also cognitive decline in AD. Higher sBDNF is associated with a protection against future occurrence of dementia and AD and predictive of slower rates of decline. In the same manner, changes in BDNF levels induced pharmacologically or by aerobic exercise are related to better cognitive function and diminished synaptic dysfunction both in humans at risk of developing AD and in animal models of AD. These effects could be related to the ability of BDNF to prevent lesion-induced neuronal degeneration. According to this idea, post-lesion gene transfer of BDNF partially restored the deficits in learning capacity and synaptic plasticity in an AD model in which BoNTx-induced damage to the entorhinal cortex was used to mimic AD pathology. Neural stem cell transplants or CREB binding protein gene transfers reversed spatial memory deficit via BDNF in AD mouse models, despite widespread Aβ plaque and tau pathology. In a recent study, delivery of BDNF to the entorhinal cortex in amyloid transgenic mice reversed neuronal atrophy and synaptic loss, regulated neuronal signaling, and diminished the related mnemonic deficits without changes in the amyloid plaque load indicating that BDNF can act through amyloid-independent mechanisms to exert its protective effect. Furthermore, 7,8-dihydroflavone (7,8-DHF), Neotrofin (a hypoxanthine derivative that stimulates neurotrophic factor production) and Neuropep-1 (a BDNF modulating peptide) have shown to reverse memory deficits in animal models of AD or even in preclinical trials. In this way, BDNF could mediate the protective effect of exercise and caloric restriction on neurodegeneration. This strengthens the need to develop behavioral interventions that could prevent the risk of developing dementia or slow the progression to dementia in patients with MCI, a path that is currently in progress. Many of these new paths point to lifestyle changes that range from antioxidant diet, environmental enrichment and social interaction to physical or cognitive exercise as potential interventions.

26Chronic stress is a known factor involved in the incidence of AD and cognitive impairment. Structures involved in the control of the physiological status of an organism are susceptible to modulation by chronic stress. In particular, the hippocampus is altered by prolonged exposure to aversive situations. These abnormalities are reflected in deficits in spatial memory tasks and novel object recognition, but also in altered synaptic plasticity processes like suppression of LTP. Chronic stress typically decreases BDNF hippocampal expression, however, when the cause of the stress disappears, the hippocampus shows amelioration of the cognitive and synaptic deficits.

27To this date, a wide variety of strategies have been assessed to reduce the deleterious effects caused by chronic stress. Infusions of BDNF in the rat hippocampus before a chronic restraint stress protocol can protect against the deficits in learning and memory in the MWM and in LTP and shRNA against BDNF before a stress protocol can revert the spatial reference memory deficits during the post-stress-rest period.

28Exercise, is a well-known strategy to increase BDNF brain levels, so it has been proposed as a non-invasive way to mimic the effects of direct BDNF administration over chronic stress. found that the hippocampal BDNF increases in response to exercise after a chronic stress protocol. On the other hand, showed that animals that followed a 30 days swimming training program improved their performance in the T maze after being exposed to chronic stress and this enhancement correlated with upregulation of hippocampal BDNF. Also, described a BDNF-mediated improvement in MWM performance in chronically stressed mice that started treadmill running 12 weeks before the beginning of the stress protocol and continued throughout it.

29investigated the palliative effects of exercise (alone or combined with fluoxetine) on a rat model of post-traumatic stress disorder. This model shows impairments in fear conditioning and extinction, inhibitory avoidance task and location recognition memory. Interestingly, in most tests, the effects of the combined treatment were similar to the ones obtained with exercise alone. They also found, in agreement with the work of, that exercise alone and exercise plus antidepressant enhanced hippocampal BDNF expression, but not antidepressant alone.

30Enriched environment (EE) has been shown to increase BDNF levels in the hippocampus in comparison with standard housing conditions. Thus, EE could be an easy way to promote the systemic and neural recovery from the effects of chronic stress. showed that exposure to EE following 10 days of immobilization (2 h/day) ameliorates spatial memory deficits in a version of the radial arm maze and depressive-like behavior. Recovery seems to be achieved through the modulation of several signaling cascades, including BDNF’s.

31suggested that EE is as effective as Fluoxetine when it occurs after exposure of the animals to a chronic stress protocol, but additional measures of the effectiveness of the stress protocol would be needed to establish the success of the treatment. Interestingly, BDNF levels were increased in the hippocampus of rats that receive either EE or Fluoxetine in comparison with the control group (stressed but without posterior treatment). Considering that chronic stress is linked to depressive-like symptoms, the results obtained using antidepressants do not seem surprising. The mechanisms of antidepressant actions over chronic stress and the putative involvement of BDNF have also been extensively studied but with no consistent results yet. showed that chronic antidepressant treatment reversed depressive-like behavior caused by chronic unpredictable stress-induced and increased BDNF mRNA expression in the granular cell layer of the dorsal hippocampus (independently of exposure to stressors). Using a different stress model, were able to normalize behavioral alterations in mice exposed to a social defeat stress protocol followed by chronic (but not acute) administration of imipramine They proposed a model in which chronic stress induces repression and chronic imipramine induces de-repression of the bdnf gene in the hippocampus through changes in the chromatin structure.

32Some studies are focusing in compounds that have been originally used to treat other diseases but have shown some antidepressant effects in animal models, such as resveratrol. Resveratrol and curcumin, when chronically administrated, prevent the behavioral and biochemical alterations induced by chronic restraint and unpredictable stress, respectively, and those effects seem to be mediated by an increment in the expression of BDNF. show that biperiden alleviates depression-like symptoms induced by chronic unpredictable stress, increasing performance in the sucrose preference, novelty suppression feeding and forced swimming tests. Importantly, these effects were inhibited by pretreatment with the TrkB antagonist K252a.

33Since the evidence suggests that BDNF may drive the recovery from stress-induced effects on the hippocampus, an interesting question emerges: Is BDNF capable of reversing the effects of chronic stress in the presence of the stressor?

34tested the effect of exercise during the exposure to stressors (“protective exercise”). Unlike preventive and therapeutic exercise, no increment on BDNF hippocampal levels was found. On the other hand, explored, in mice, the potential palliative effects of running on chronic stress-related impairments when exercise and stress are co-occurring. They found that the TrkB receptor had higher expression levels in both exercised groups (stressed and non-stressed) compared with both sedentary groups, supporting the hypothesis that the mitigation of the negative consequences of stress by exercise could be mediated by BDNF.

35Depending on the chronic stress protocol (duration and type of stressor), different and even contrasting results have been found. BDNF appears to be an important underlying molecule behind the restitution of a normal cognitive phenotype in animal models of chronic stress. The fact that BDNF could be increased with non-invasive protocols and/or drugs -some of which are used in clinical trials- makes it attractive for human therapies.

36The link between stress, specific genes and the development of psychiatric disorders has been extensively studied, and a causal role has been ascribed of gene-environment interactions in the etiology of many of them. In fact, psychiatric disorders can be defined as clinical entities emerging from the genetic-environmental interaction (for review, ).

37In the last few years, evidence from animal models and clinical studies strongly suggest that dysregulation of neurotrophic factors could play an important role in the etiology of the bipolar disorder (BD), major depressive disorder (MDD), and schizophrenia (SZ). Due to the role of BDNF in neural plasticity, there could be a link between BDNF expression and the cognitive symptoms associated with memory impairments.

38The mnemonic domain is commonly affected in different psychiatric disorders, such as BD, MDD, or SZ. Moreover, studies of post-mortem brain tissue of patients with BD and MDD reported that BDNF levels are decreased in structures involved in memory processes, such as the hippocampus and the prefrontal cortex. In the case of SZ, post-mortem brain tissue analyses have shown more controversial results. While some studies observed an increase in BDNF expression in the prefrontal cortex and the hippocampus, others have shown a decrease in both structures. Although BDNF was originally thought as a viable indicator of pathological brain functioning for early detection of BD, MDD, SZ, or AD, the discriminative power of BDNF as a biomarker is highly limited, since it seems to be a non-specific marker of many neuropsychiatric disorders.

39The BD is a neuropsychiatric disorder that emerges from the interaction between genetic and environmental factors and is characterized by the switching between maniac and depressive episodes (for review, ). It has been proposed that BDNF signaling participates in the physiological effects produced by some pharmacological treatments used for BD. It has been shown that sBDNF decreases in the first episode of unmedicated BD patients and that, after 1-year of pharmacological intervention, sBDNF concentration increases. in addition, there was a a negative correlation between the number of episodes and sBDNF levels. It has been reported that sBDNF positively correlates with the duration of the maniac and depressive episodes. This evidence suggests that episode-related changes in the structure of the brain could be linked to peripheral BDNF concentration. have shown that hippocampal volume is reduced in patients with BD that present Val66Met BDNF polymorphism compared with controls and patients with MDD. Moreover, they proposed a link between the hippocampal volume and the performance in an episodic memory task. Another work has shown that the peripheral BDNF correlates with the performance in episodic memory task in BD patients with the BDNF Val66Met polymorphism. In this line, a recent study has shown that high levels of sBDNF are associated with good cognitive performance, including verbal memory. This evidence suggests that changes in BDNF expression in BD patients could produce structural modifications in the hippocampal formation related to episodic memory impairments. Despite this, most of the studies report alterations in other cognitive domains that are important for a good performance in episodic memory tasks, such as attention and working memory (for review, ). Thus, BDNF dysregulations could be related with the emergence of more complex symptom’s profiles. For this reason, the relationship between BDNF and episodic memory in BD remains unclear.

40Major depressive disorder is one of the most common mood disorders worldwide and is characterized by the absence of pursuit of pleasurable activities and the presence of negative thoughts. Since most common drugs used as antidepressant block the serotonin transporter (SERT), increasing extracellular serotonin in the raphe’s nucleus post-synapses (for review, ), it has been proposed that a misbalance in the serotonergic release could be related to the etiology of the depressive symptoms (for review, ). BDNF regulates the growth and reconstruction of 5-HT containing neuronal terminals in the cortex, and administration of BDNF in the raphe nucleus reduces behaviors related to depressive symptoms in rats. In addition, MDD patients present cognitive decline in different domains, including episodic memory but only recently these deficits have been studied in detail. A large amount of work shows that sBDNF is decreased in MDD. found that patients that recurrently present depressive episodes show lower levels of sBDNF compared with those patients that were cursing their first episode. Interestingly, antidepressant treatment increases sBDNF concentration, but there is no consensus on whether lower sBDNF correlate with poor performance in memory tasks observed in this pathology.

41In the case of the SZ, different studies have shown that the level of sBDNF correlates with cognitive performance in different domains. Despite the lack of consensus on whether basal sBDNF is increased or decreased in SZ patients, some studies have indicated a correlation between memory performance and sBDNF levels. Interestingly, there are evidences that pro-cognitive effects of pharmacological interventions in SZ could be mediated by BDNF. For example, have found that a 12-week chronic treatment with olanzapine produced an increase in BDNF plasma concentration. Moreover, BDNF concentration positively correlated with cognitive performance in a RBANS scale of memory. Not only the pharmacological interventions were effective on the reduction of mnemonic symptoms, different cognitive training protocols were also designed to enhance specific cognitive domains, especially memory. conducted a computerized cognitive training in SZ patients and the patients exposed to this program present higher levels of sBDNF compared to the control group. They observed an enhancement in memory, but a causal link between sBDNF and memory remains unclear.

42Considering all the results mentioned above, it would be tempting to suggest the use of BDNF as a therapeutic target for both age-related and neuropsychiatric-related cognitive dysfunction. This idea has found many difficulties to be put to practice because of the poor brain barrier penetration of BDNF and short half-life on plasma. Moreover, gene therapy and BDNF mimetic strategies came across many negative side effects that led to their abandonment. Clinically plausible alternative approaches could include a natural increase the production of endogenous BDNF. In this sense, epidemiological studies have suggested that a number of lifestyle factors such as physical exercise, diet and social activity and education may reduce the long-term risk of cognitive impairment and dementia, and animal studies are consistent with this idea.

43In particular, the risk of developing AD is highly increased in a lonely person, indicating that social interaction could delay the onset of the disease. Physical activity is another lifestyle factor that could influence the progress of the disease. Recent reports from both epidemiological and interventional studies reinforce the idea of using physical activity as a strategy to increase neuroplasticity in pathological conditions. The influence of behaviors such as exercise and social interaction on learning and memory processes has been thoroughly studied. Researchers have found a relationship between frequent social activity and improved cognitive function. In the same direction, the cognitive improvement due to physical exercise has also been well documented. Physical exercise has shown not only to ameliorate structural changes in the brain, but also to protect against aging-related cognitive decline.

44Considering that lifestyle implementations have the ability to impact in the brain, a central question is how these changes in energy metabolism and social stimuli can impact on the brain structure and interact with synaptic plasticity and molecular systems to improve cognitive function.

45A current model explains the effects of these lifestyle factors in terms of changes in vasculature and neurotrophic and neurotransmitter-support system. Of all these changes, BDNF is the only one that is present in all the aforementioned lifestyle manipulations. BDNF is increased by social interaction with conspecifics in APP/PS1 mice, leading to the reversal of memory deficits. Also, BDNF could be important for the regulation of energy homeostasis, since diminished BDNF levels are associated with disorders of energy metabolism such as obesity and hyperglycemia. In fact, a high fat diet was shown to decrease BDNF levels in the hippocampus and impair learning and memory. Additionally, the increase in BDNF is one of the most consistent changes reported following exercise, as recently discussed in a meta-review. The most robust experiments supporting the fundamental role of BDNF in exercise-induced improvement in cognitive function are the ones in which blockade of BDNF impaired the cognitive improvements induced by exercise. showed that the exercise-induced enhancement of learning in the MWM task was blocked by TrkB-IgG administrated during the exercise period. Furthermore, exercise enabled the acquisition of sub threshold experiences (object location memory task) and this effect was dependent on BDNF. A similar effect was reported by using BDNF siRNA to diminish BDNF function.

46For its practicality, physical activity is the lifestyle change with more potential as a therapeutic/prevention strategy. A bulk of studies have focused on the idea of aerobic exercise as a potential non-pharmacological and low cost treatment to maintain and improve neurocognitive function. A meta-analysis of several longitudinal training studies showed that exercise improved cognitive function regardless of the task type. Recent studies confirmed this effect showing that not only spatial or contextual hippocampal dependent- memory tasks improve with exercise, but also non-spatial memories such as object recognition that are thought to rely more heavily on the Prh than the HP. A single session of cardiovascular exercise benefits long-term memory but does not influence short-term memory. Moreover, exercise can improve memory in aged animals specifically during a restricted time window after the experience, reinforcing the specific role of exercise over the memory consolidation process. Interestingly, many studies found that sex may be an important variable when evaluating the effectiveness of exercise interventions and this is consistent with the sex-specific mechanisms of action of BDNF. The timing of the intervention could also be relevant. In several models of traumatic neurological injury, when interventions are given prior to the damage, induction of BDNF reduced neuronal degeneration and improved cognitive outcome. Although the effects of exercise are somewhat short-lived, some interventions can improve the outcome even when given after the damage. However, the duration of the benefits depends on the age of the subject during the exercise exposure. While adolescent exercise training did not affect BDNF levels immediately after an object recognition task, it did lead to greater BDNF levels in the Prh if the task was done 2 weeks after. In adulthood, exercise increased BDNF levels immediately after the task but this effect was short-lived and lasted less than 2 weeks. These data suggest that exercise could modulate learning related plastic changes in an age-dependent manner. Although the benefits of exercise are related to many growth factors, BDNF is the only one consistently elevated after a few weeks of continuous exercise. This neurotrophin is rapidly induced in the hippocampus and cortical regions, and remains elevated for several days after exercise. In addition, BDNF levels can be rapidly re-induced up to peak levels by a subsequent sub threshold exposure to exercise, even several days after the end of the exercise program.

47Exercise does not influence brain regions uniformly, but affects them in a more selective way, which suggests location-specificity of the molecular pathways involved in exercise-induced plasticity. Interestingly, the effects of exercise on BDNF expression occur in regions related to mnemonic functions such as the anterior hippocampus, cerebellum and frontal cortex, but not others such as the striatum. This is in accordance with previous reports indicating that exercise-induced increases in sBDNF levels are associated with changes in hippocampal volume, which, in turn, correlate with spatial memory performance.

48The positive effects of exercise on plasticity are particularly relevant for the aging population, in which BDNF levels are decreased. Considering that the aging brain is still capable of plasticity, lifestyle related experiences could be a way to recruit plastic processes and counteract the detrimental effects of aging. In aged animals, hippocampal neurogenesis and BDNF levels can increase with exercise. Although these effects are not as robust as those seen in younger animals, the increase in BDNF seems to ameliorate mental deterioration and improve memory function. In fact, long-term exercise programs are able to rescue these cognitive deficits even after the first signs of mnemonic impairment. A recent clinical trial examined the impact of aerobic cardiorespiratory training versus stretching on MCI patients and reported sex-dependent cognitive improvements related with trophic factor and Ab-40 and Ab-42 circulating levels.

49Physical activity is also associated with a lower risk of developing dementia. Many clinical trials point toward improved cognition and reduced incidence of psychiatric symptoms when patients with mild AD received a physical training protocol. However, some studies suggest that environmental enrichment could be more beneficial for cognition than physical exercise alone.

50Coordinative exercise and cognitive training can also induce gray matter plasticity and enhance cognitive functions in older adults. However, the improvement generated by prior training is usually domain-restrictive and only acts over memory systems affected by the previous experience. Nonetheless, there are some reports that show a generalized benefit of prior experience to different tasks and contexts. A combination of physical and cognitive training with control of nutritional and cardiovascular risk factors during a 2-year period led to improved cognitive performance in old adults at risk of developing dementia. In animal studies, environmental enrichment could be seen as a multidomain intervention. It consists of social enrichment, physical exercise and environmental changes and has been shown to increase BDNF levels and enhance learning and memory in different domains such as object recognition, spatial learning and motor abilities. The combination of both sensory enrichment and physical activity has more impact on neuronal plasticity than these elements given independently. This motivated original therapeutical proposals in human research. For example, a novel dancing program with higher cognitive and coordinative demands than previous physical activity programs induced more gray matter increases in an aged group than a traditional sport program with comparable cardiovascular demands.

51One important question regarding the effects of exercise on cognitive function is to establish the mechanisms responsible for the cross talk between cardiovascular/muscle activity and the central nervous system. The muscle higher metabolic rates could lead to the secretion of signaling molecules, that could subsequently upregulate plasticity related gene expression and protect the brain from damage. In this way, elevated plasticity molecules such as BDNF could prime the brain to be better prepared for subsequent changes related to learning or could be selectively secreted in an activity-dependent manner during learning experiences. As a result, exercise could enhance the activity of a general molecular machinery important for learning and memory. In accordance with this view, molecules such as CREB, NMDARs subunits and BDNF are particularly induced following exercise and brain regions important for memory formation such as the hippocampus are selectively influenced by physical activity.

52A potential mechanism for the exercise-related neuroprotective effects of BDNF is via modulation of synaptic and structural plasticity. Plastic changes induced by exercise include increased neurogenesis, greater arborization of neuronal dendrites and synaptogenesis, as well as increased amplitude and reduced threshold for LTP. Since these effects are accompanied by a concurrent increase in BDNF levels, BDNF could be a potential mediator. In addition, increased vascularization accompanied by greater dendritic complexity and neurogenesis could explain the increase in hippocampal volume following exercise. The progressive age-related decline in neurogenesis has been associated with a non-permissive microenvironment with low levels of neurogenesis-promoting factors. However, this microenvironment is still responsive to environmental changes and can be stimulated even at late stages to provide molecular cues for proliferation. Changes in growth factor levels such as BDNF might underlie the decrease in neurogenesis seen as a consequence of disease or aging, and the aged brain retains the capacity to respond to the neurogenesis-stimulating effects of growth factors.

53Exogenous application of BDNF can restore the levels of hippocampal neurogenesis in aged animals. In the same way, exercise-induced increases in neurogenesis are necessary for the physical activity-dependent enhancement in learning and memory. This led to the idea that neurogenesis could be the substrate of this cognitive enhancements mediated by BDNF.

54Another potential beneficial effect of BDNF is its ability to protect neurons from oxidative damage or excitotoxic stress and from Aβ-induced degeneration in animal models of normal and pathological aging. In fact, BDNF is upregulated in response to different kinds of insults to the nervous system. An exercise regime can lessen the accumulation of oxidative cell damage and the dysfunction characteristic of aged animals, and selective suppression of BDNF increases the vulnerability of neurons to excitotoxicity and increases amyloidogenesis.

55Although there should be a clear excitement to establish a systematic exercise program to ameliorate or even prevent symptoms of memory deficits related to aging, psychiatric disorders or diseases, many limitations still exist for this therapeutic line. One of the most important drawbacks of this approach is that the high prevalence of chronic diseases in the aged population affects exercise performance and feasibility, increasing the potential risks of the treatment, especially for high intensity protocols. This directly impacts on the motivation to follow these kinds of treatments, which are known to have low adherence. Programs with lower intensities could be a better choice, since they are still able to impact positively on cognition and neurophysiology in aged subjects without affecting adherence to the treatment. It has been reported that the lack of time is also one of the main reasons for avoiding regular practice of aerobic exercise. Thus, there is interest in developing more efficient training programs involving less time demand but inducing a similar BDNF response. demonstrated, in animals, that the practice of low/moderate intensity aerobic exercise through 2 to 18 months of life increased BDNF in various brain structures like the prefrontal cortex and the hippocampus and reduces the normal decline due aging. Moreover, this protocol improved both novel object recognition and context discrimination capacity. conducted a meta-analysis to determine the impact of acute and regular exercise on BDNF levels in humans. They found that the moderate effect from a single session of exercise was intensified if it was executed after a regular program of exercise. They posit that each episode of exercise results in a “dose” of BDNF activity and that the magnitude of this “dose” can be enhanced over time by regular exercise.

56On the other side, attempts to directly use recombinant BDNF as a therapy have found many methodological limitations. One of the main problems for translating BDNF-based therapies into the clinic is problem of delivery to the brain and the challenge of sustaining the expression for longer intervals since since the recombinant protein has a very short half-life. In this sense, some preclinical studies were oriented to using BDNF fused to cell-penetrating peptides and packed in AAV-constructs and intranasal delivery of these AAV constructs to central nervous system. A10-day-AAV treatment could alleviate depression-like behaviors in mice, and AAV delivery of BDNF striatal neurons induced neurogenesis and increased the lifespan of an animal model of HD. Although beneficial, the use of AAV has been limited by difficulties in biodistribution and the immunogenicity to the virus. An alternative approach is the use of stem cell transplants that can express BDNF and other beneficial factors and can migrate into damaged areas of the brain by their selective tropism to inflammation and apoptosis sites, although they are not permanently integrated into the organism. Mesenchymal stem cell transplantations have shown improvements in behavioral deficits in a murine model of HD, have also slowed the neurodegenerative processes by a diminished atrophy and an increased neurogenesis.

57An alternative possibility to the use of exercise as a ‘natural’ and non-invasive way of increasing BDNF signaling in neuronal networks, is the administration of drugs already available for clinical use. Many medications are capable of impacting on BDNF levels. Memantine and donepezil are a pharmaceutical compound used to alleviate the symptoms of AD that markedly increases BDNF levels in a dose-dependent manner.

58Ampakines are good candidates because they can increase excitatory transmission affecting BDNF levels. In rat hippocampal slice cultures, a very brief clinically tested ampakine treatment produced elevated BDNF protein levels that lasted several days after the exposure. Another therapeutic possibility that has emerged recently is the use of transcranial magnetic stimulation (TMS) to increase BDNF levels. TMS restored the levels of BDNF and TrkB that were normally reduced in aged mice and improved spatial memory.

59In conclusion, several environmental and lifestyle interventions that reduce age-dependent cognitive decline and pathological degeneration can also increase BDNF production, suggesting that BDNF is neuroprotective (Figure 1). Given that cognitive training is a focused approach that selectively acts on sets of memory domains and that drugs are invasive, exercise is still a ‘favorite’ when thinking of potential therapeutic approaches.

60Although BDNF is a key player in synaptic plasticity and memory process, its role in the etiology of cognitive symptoms in pathological conditions remains unclear. BDNF-mediated plastic changes have been proposed as one of the underlying neurobiological substrates of memory consolidation, and changes in BDNF were shown to directly affect memory performance in animal models of neurodegenerative/neuropsychiatric diseases and in normal conditions. In human post-mortem brain tissue, BDNF expression is higher in memory related structures, such as the hippocampus and the amygdala. However, in humans, the relationship between memory and BDNF still remains correlational, mainly because the available techniques do not allow the control of BDNF expression in humans. Despite this difficulty, numerous studies attempted to establish causal relationships between these two factors by analyzing memory performance under conditions that up-regulate or down-regulate BDNF expression. Since brain BDNF expression correlates with sBDNF concentration, this association has been extensively used to study the implication of BDNF in mnemonic functions in humans under normal and pathological conditions. In fact, the concentration of this blood-measurable protein is correlated with the memory impairment in different disorders. Additionally, changes in the trafficking and release of BDNF due to Val66Met polymorphism have also been used to correlate BDNF levels with mnemonic performance and structural changes in memory-related regions in healthy and diseased individuals. It has been shown that interventions such as exercise, chronic administration of fluoxetine, and cognitive training can enhance the sBDNF concentration and correlates with a better performance in memory tasks. The initial idea of using BDNF as a biomarker for neurodegenerative/neuropsychiatric diseases was discarded because changes in BDNF levels are common to many pathological conditions, which underscores its discriminative value and potency. However, considering the data reviewed here, we suggest that BDNF can be thought of as a marker that specifically relates to the occurrence and/or progression of the mnemonic symptoms that are common to many pathological conditions that share deficits in this cognitive domain. Moreover, BDNF was shown to be a shared factor in which converge most of the therapies that have been able to fight these mnemonic symptoms.

61MM was responsible for drafting the manuscript and revising its content. JM and MZ contributed to writing the manuscript and figure design. PB was responsible for the general idea, writing and critically revising, and correcting the manuscript. All authors read and approved the final manuscript, and contributed to the conception of the work.

62The 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.