CREB, cellular excitability, and cognition: implications for aging

Abstract

Humans and laboratory animals display cognitive deficits as they age. However, there are currently no effective therapies available to treat these deficits, as the underlying mechanisms are poorly understood. Studies using pharmacological compounds have found a link between cognitive performance and the intrinsic cellular excitability of CA1 hippocampal neurons. Therefore, it is of great interest to identify molecular regulators that may be influencing both cognition and neuronal excitability, which could be changed with age. One possible regulator is the transcription factor cAMP response element binding-protein (CREB). In young adult animals, manipulation of CREB activity has resulted in modulation of both cognitive performance on behavioral tasks, and neuronal excitability. While evidence is sparse, studies also point to a dysfunction in CREB signaling with aging. We propose that CREB may be a viable therapeutic target for the treatment of age-related cognitive deficits, along with potential experiments to test this hypothesis.

Cognitive and biophysical deficits in aging

Like humans, laboratory animals display impairments in their ability to learn and remember as they age. However, there are currently no treatments for these cognitive deficits, as their underlying cause is not well understood. In laboratory rats, age-related cognitive deficits have been seen in tasks that utilize spatial memory, such as the Barnes circular maze and the Morris water maze, as well as tasks that require complex temporal processing, such as trace associative conditioning [1]–[4]. Along with cognitive impairments, numerous studies have found that aged animals have neurons that are intrinsically less excitable than those from young animals. Specifically, pyramidal cells from area CA1 of dorsal hippocampus exhibit these changes [5]–[9]. One measure of neuronal excitability is the post-burst afterhyperpolarization (AHP). This is a hyperpolarization that occurs after the neuron has fired a burst of action potentials. By bringing the neuron further from action potential threshold, a larger AHP makes a neuron less likely to fire again. After an animal successfully learns a task, the AHPs of their CA1 pyramidal neurons are reduced and the neurons are more excitable [5]. Thus, the reduced intrinsic excitability seen in aged animals is thought to be linked to their cognitive impairments.

Fortunately, several compounds have been found which can increase intrinsic excitability of CA1 pyramidal neurons. These compounds are very promising as when given in vivo, they also improve performance on behavioral tasks. One example is the L-type calcium channel blocker nimodipine. When applied onto neurons in vitro, both young and, especially, aged CA1 pyramidal cells show reduced AHP and spike-frequency accommodation [10]. Furthermore, when nimodipine is given systemically, it facilitates aged animals’ ability to learn trace eyeblink conditioning [11]. Galantamine (Razadyne®), a cholinesterase inhibitor, is another example: it increases excitability of CA1 pyramidal neurons in vitro, and improves acquisition of trace eyeblink conditioning when delivered in vivo to aged animals [12]. In clinical trials, these treatments have proven to be promising options for treatment of various age-related diseases including vascular dementia, chronic cerebrovascular disorder (nimodipine), and Alzheimer’s Disease (Razadyne®) [13]. Nilvadipine, an L-type calcium channel blocker related to nimodipine, is also the focus of a current European clinical trial for treatment of Alzheimer’s disease. These findings suggest a tight link between the intrinsic excitability of CA1 pyramidal neurons, and an animal’s ability to learn and remember. Therefore, it is important to identify the underlying molecular pathways that are disrupted with age, and underlie these changes on both the biophysical and cognitive level.

CREB modulates cognition and excitability in young adult animals

The transcription factor cAMP response element binding-protein (CREB) has been shown to modulate both cognition, and cellular excitability in young adult animals. Long known for its role in memory formation [14]–[17], CREB is activated by phosphorylation at serine133 (pCREB), after which it can dimerize and stimulate transcription of downstream target genes such as BDNF and c-fos [14], [16], [18], [19].

Numerous studies manipulating activity of CREB have demonstrated its crucial role in memory formation [14]–[17], [20], [21], and many comprehensive reviews have been written about it e.g. [22]–[25]. Therefore, this review will highlight a few of the studies. Studies preventing activity of CREB have shown just how important it is for memory formation. Dash et al. were the first to show that blocking CREB function impaired memory [15]. When cAMP-responsive element (CRE) was injected into the nucleus of Aplysia sensory neurons, long-term facilitation normally induced by serotonin was blocked, while short-term facilitation was intact. This showed that by injecting exogenous CRE, CREB’s normal binding site, CREB could no longer bind to CRE sites on DNA and did not give the downstream transcription needed for expression of long-term facilitation. Guzowski and McGaugh demonstrated a similar finding by using an antisense oligodeoxynucleotide against CREB in rats [20]. CREB antisense oligodeoxynucleotides were injected into dorsal hippocampus and caused a 36% reduction in CREB protein levels. When rats were injected prior to training on the Morris water maze, acquisition of the task was not impaired, nor was short-term memory in a retention test carried out 2 hours later. However, a long-term memory test given 48 hours later showed an impairment only in the antisense-injected animals. This indicated that reducing CREB levels impaired long-term memory on a spatial task. A CREB knock-out mouse line was also generated [21]. When these mutants were trained in fear conditioning, they showed very little freezing to context or tone, both in tests of short- and long-term memory. When these mutants were trained on Morris water maze, they had slight impairments in acquisition, and showed no preference for the target quadrant during the probe trial. When long-term potentiation (LTP) experiments were conducted on slices made from these mice, it was found to be small in amplitude, and short-lasting. These results suggest that CREB knock-out results in impairments in memory as well as synaptic plasticity/neuronal excitability. Transgenic mice expressing KCREB, a dominant negative CREB that reduces affinity of normal CREB for DNA, were impaired on both acquisition and retention of the Morris water maze task, and for object recognition [26]. Another transgenic line expressed a mutant CREB where serine 133 was replaced with an alanine to prevent activation of CREB [27]. These mice were impaired in long-term memory in fear conditioning, but short-term memory was left intact. These studies show that reducing CREB activity in different ways result in impairments in memory, typically long-term.

Increasing CREB activity using exogenous compounds, or via transgenic and viral means has been found to facilitate performance on behavioral tasks testing for cognition. When rats received systemic rolipram, a phosphodiesterase inhibitor, for 5 days, hippocampal levels of CREB and CREB phosphorylated at S133 (pCREB) were increased, and memory for fear conditioning was enhanced [28]. Suzuki et al. [29] generated transgenic mice whose CREB had increased affinity for either PKA, which had increased likelihood of being phosphorylated (CREB^Y134F), or for CREB-binding protein (CREB-DIEMDL), CREB’s co-activator. These mice showed facilitation in long-term memory for social recognition, fear conditioning, Morris water maze, and passive avoidance, all the while showing no differences in short-term memory or learning. Another transgenic mouse line expressed constitutively active VP-16 CREB (caCREB) in the amygdala [30]. This resulted in strong memory for contextual and cued fear conditioning. These studies in rats and transgenic mice indicate that activating CREB results in increased transcription of downstream genes, and facilitates cognition in the form of memory.

While transgenic animals are useful tools, since their gene expression is altered from birth, it can be difficult to tease apart effects of the genetic alteration alone, from their potential effects on developmental processes. Thus, viral vector delivery of genes has the advantage of using animals that develop normally. One example used the Sindbis virus to express CREB^Y134F in CA1 hippocampal cells of mice. This resulted in increased contextual fear memory, mirroring previous results found in transgenic mice with the same mutation [31]. In an elegant experiment using HSV to overexpress wild-type CREB, Sekeres et al. [32] demonstrated the necessity and sufficiency of CREB for long-term memory. HSV-CREB was injected into either wild-type mice or CREB knock-out mice, that were then given weak Morris water maze training. In both cases, long-term spatial memory was observed. Josselyn et al. [33] also observed long-term memory with a weak training protocol for fear conditioning (which normally only produces short-term memory) when HSV-CREB was injected into the amygdala of rats. A similar experiment showed HSV-CREB injections into hippocampus improved memory for water cross maze [34]. Most recently, mice injected with HSV-CREB in the retrosplenial cortex also showed improved memory for water maze [35]. This series of experiments indicate that virally-mediated overexpression of CREB is sufficient to facilitate memory in young animals.

CREB’s ability to modulate performance on behavioral tasks is unsurprising, but of particular interest are the recent studies that have reported CREB’s ability to bidirectionally modulate intrinsic neuronal excitability. Expression of caCREB has resulted in increased excitability in many brain areas: CA1 hippocampus [36], locus ceruleus [37], nucleus accumbens [38], and amygdala [39]. Additionally, overexpression of endogenous CREB is sufficient to produce modest increases in excitability of neurons in the amydala [40], [41]. This indicates there is a general mechanism activated by CREB, which works in a universal, non-region specific manner to increase intrinsic excitability of neurons. By expressing dominant negative forms of CREB, excitability can also be reduced [37], [38], [42]. Despite the evidence for CREB’s ability to modulate neuronal excitability, the mechanism of how this occurs is still unknown. Many channels are thought to contribute to the post-burst AHP, including BK, SK2, and SK4. The promoter region for the BK channel encoding gene has been shown to contain binding motifs for CREB [43], suggesting that CREB can directly regulate the transcription of BK channels. However, given that increasing expression of BK channels would enlarge the post-burst AHP [44], this does not explain how CREB might enhance excitability. As for SK2 and SK4, CREB has yet to be shown to directly modulate their transcription, suggesting that their promoters may not contain binding motifs for CREB. Therefore, CREB’s modulation of these channels could be either indirect, via one or more of CREB’s downstream transcriptional targets, or directly by CREB via a non-transcriptional mechanism. Regardless of the mechanism itself, these studies suggest that increases in CREB activity may ameliorate age-related deficits in neuronal excitability and cognition.

CREB in aging – a therapeutic target?

Bach et al. [45] were the first to find evidence that aged mice may have deficits in CREB signaling. Aged mice were impaired in Barnes circular maze performance and in vitro slices had reduced LTP as compared to their young counterparts. Both of these effects were ameliorated by compounds that activated the cAMP/PKA pathway. This pathway is well known to be an upstream activator of CREB via PKA’s kinase activity [14]–[17], so it’s possible that the amelioration seen in this study were a result of increases in CREB activity. Biophysical experiments have demonstrated that learning-induced reductions in the AHP depend on activation of the cAMP/PKA pathway [46]–[49]. Additionally, levels of PKA itself have been found to be decreased in hippocampi of aged rats [50], which could lead to a reduction in CREB activation. Similarly, levels of CaMKIV, another kinase which phosphorylates CREB, is also reduced with age [51]. Furthermore, CREB’s obligate co-activator, CREB-binding protein (CBP), is also reduced with age [52]. Age-related reductions in these crucial activators and partners of CREB could easily lead to insufficient activation of CREB in the aged hippocampus.

Reduced CREB activation is exactly what has been observed in aged animals. Young adult mice and rats exhibit increased pCREB levels following training on tasks such as inhibitory avoidance, contextual fear conditioning, cross-maze training, radial arm maze, and Morris water maze [17], [53]–[59]. Levels of pCREB have been found to be correlated with how well the animal performed on memory tests, where more pCREB is indicative of better performance [60]. In contrast, aged animals who fail to learn or remember a task also do not exhibit the learning-associated increase in pCREB levels. Specifically, immunocytochemistry studies have shown less pCREB in aged animals shortly after contextual fear conditioning [55], inhibitory avoidance [61], and water maze [59]. Western blot experiments have also revealed a reduction in total CREB in aged animals who failed to learn, when measured immediately after water maze training [62]. These results indicate that CREB levels and activity is decreased in aged brains, and suggest increasing CREB activity could ameliorate age-related deficits. Supporting this idea is a parabiosis study bringing together the circulatory systems of young and aged mice [63]. Age-related deficits in cognition and dendritic spine density in the hippocampus were reversed by young plasma, and this effect was dependent on CREB signaling.

While CREB activation in response to behavioral training has been well studied, few have examined basal levels of CREB with age. If basal levels of CREB or pCREB were indicative of cognitive status, they could be used as a biomarker. In western blot experiments on naive animals, Foster et al. [64] showed that total CREB levels are not changed, but that pCREB is decreased in hippocampus of aged (20–24 month old) Fischer 344 rats. Monti et al. [65] showed a different result; while total CREB was also found to be unchanged, pCREB was increased in aged (30 month old) Wistar rats. These studies used protein extracts from the entire hippocampus. Given that dorsal and ventral hippocampus carry out different functions, as well as the major hippocampal sub-regions (CA1, CA3, and DG), it’s possible that by treating all these regions as one will mask subtle differences that may be present. One study did examine pCREB levels in aging using immunohistochemistry. Levels of pCREB were found to drop by middle age, and further decrease into old age [66]. All hippocampal sub-regions showed a decrease in pCREB: CA1 by ~25%, CA3 by ~25%, and DG by ~28%. Unfortunately, the antibody used was not specific only to pCREB, it also recognized related proteins, pCREM and pATF-1, meaning the results cannot be attributed purely to changes to pCREB. Additionally, only looking at pCREB levels do not tell us if the ratio of pCREB/CREB is changing, or if total levels of CREB are reduced. Further, experiments are needed to fully elucidate the age-related changes in basal CREB levels and activation. We recommend a full characterization of age-related changes in basal CREB levels and pCREB in each hippocampal sub-region. Furthermore, examination of CREB levels should be carried out at a time point long after the end of behavioral training to determine if a relationship exists between basal CREB and/or pCREB levels and learning and memory of the task. We expect to see either a decrease in total CREB levels, or a decrease in CREB activity, most likely in CA1 where biophysical deficits have been observed. We also expect that just like in young animals [60], pCREB levels should be related to performance on a behavioral task, most likely on memory performance.

There are also some studies that suggest that increasing CREB levels may not be beneficial. For example, levels of CREB-binding protein were not related to either chronological age, or cognitive status in young and aged rats [67]. Optogenetic activation of A_2A receptors that increase levels of pCREB in the hippocampus impaired, rather than facilitated, performance on the Y-maze task [68]. Finally, age-related working memory deficit, attributed to increased pCREB levels in prelimbic cortex, were rescued by pharmacologically reducing pCREB levels [69]. The conflicting studies presented here, along with incongruous results in how basal CREB activity changes with age, necessitate further work; especially those that examine causal relationship between CREB and age-related cognitive deficits.

We propose that insufficient CREB signaling underlies age-related deficits at the biophysical and cognitive levels. If there is a causal relationship between CREB and the age-related deficits, CREB overexpression should ameliorate these deficits. These experiments could easily be carried out in Fischer 344 x Brown Norway F1 hybrid rats, which are already well-characterized to have age-related cognitive and biophysical deficits [4], [70]. To test for rescue after the onset of these deficits, as opposed to a preventative approach, viral vectors are an ideal tool. In our hands, expression of caCREB can be achieved in young rats using a viral vector. In these young rats, CA1 neurons expressing caCREB have reduced AHP amplitude (Fig. 1), replicating previous findings with caCREB expression in young mice [36]. However, caCREB must be used with caution, as expression for more than 10 or 11 days can lead to impairment on behavioral tasks [71]. To avoid the toxicity associated with caCREB, we recommend overexpression of wild-type, non-mutated CREB, which has been shown to be sufficient to increase neuronal excitability [32]–[34], [40]. Given the tight link between neuronal excitability and cognition identified by pharmacological manipulations, we expect CREB to also ameliorate age-related cognitive deficits. To detect the potential improvements in behavior with CREB overexpression, behavioral tests such as Morris water maze and trace eye blink conditioning can be used, as both of these tasks are highly sensitive to age-related cognitive deficits [4], [72], [73]. In order to achieve long-lasting overexpression of CREB through the duration of training for these tasks, we recommend using adeno-associated viruses (AAV) which have been shown to stably express for more than 15 months [74], [75]. In this proposed set of experiments, young and aged animals can be first injected with AAV into CA1 of hippocampus, and given time for virus to express. The animals would then be trained on one or more behavioral tasks to detect cognitive deficits or amelioration of such in the aged animals. Two weeks following the end of behavioral training, animals would then be euthanized to detect any changes to basal measurements of cellular excitability, such as the AHP. If dysfunctional CREB activation does indeed underlie age-related cognitive deficits, we would expect to see an amelioration of age-related cognitive deficits, most likely in the domain of long-term memory, alongside a reduction in the AHP from infected CA1 neurons.

Figure 1.

Constitutively-active CREB expression reduces peak and slow AHP evoked with 15 suprathreshold current injections at 50Hz in CA1 pyramidal neurons from young adult rats. Whole-cell current clamp recording experiments were conducted 2 weeks after the injection of a lentiviral vector to express VP16-CREB (caCREB) in CA1 of the hippocampus. The caCREB sequence was downstream from a hSyn promoter, followed by an EGFP reporter (downstream of its own hSyn promoter), to allow for visualization of infected cells. a) Example stitched 10x confocal image of EGFP expression 2 weeks post-injection of lentivirus into dorsal CA1 area. Summed z-stack of infected CA1 pyramidal neurons (inset). Left scale bar 250 μm, right scale bar 100 μm. b) Peak AHP amplitudes measured from caCREB-infected cells (green: n = 11) are smaller than that of neighboring uninfected control cells (blue: n = 10) (unpaired t-test: * p < 0.05). c) Slow AHP amplitudes measured from caCREB-infected cells are also smaller than that of neighboring uninfected control cells (** p < 0.01). d) Representative traces from AHP recordings made from caCREB-infected and control cells, arrow indicates 1 s time point where slow AHP was measured. Bars in b and c represent mean ± S.E.M.

Concluding Remarks

CREB is one of the most well studied proteins that are important for memory formation in young adult animals. Other proteins, such as early growth response protein 1 (EGR-1) and brain-derived neurotrophic factor (BDNF), have also been demonstrated to be important for memory formation in young adult animals. However, like CREB, much is yet to be discovered about the aging-related changes in their function and/or activity. Moreover, it is possible that there is an interaction of those proteins, such as CREB and BDNF, in a signaling cascade that ultimately leads to successful memory formation. Hence, altering the activity of only CREB may not fully rescue the aging-related cognitive deficits. However, given the impact of altering just the CREB levels on memory formation in young adults, we hypothesize that increasing CREB activity will ameliorate the cognitive-deficits observed in normal aged subjects.

Highlights.

• Biophysical and cognitive deficits are linked in aging

• CREB modulates biophysics and cognition in young adult animals

• CREB dysfunction may underlie biophysical and cognitive changes in aging

• CREB may be a therapeutic target for age-related cognitive deficits

Abbreviations

AHP

afterhyperpolarization

BDNF

brain-derived neurotrophic factor

caCREB

constitutively-active CREB

CRE

cAMP response element

CREB

cAMP response element binding-protein

EGR-1

early growth response protein 1

LTP

long-term potentiation