PDE4A5 signalling impairs hippocampal synaptic plasticity and long-term memory

Posted comment on ´Compartmentalized PDE4A5 Signaling Impairs Hippocampal Synaptic Plasticity and Long-Term Memory` by B.Y. R Havekes, A.J. Park, R.E. Tolentino, V.M Bruinenberg, J.C. Tudor, Y. Lee, R.T. Hansen, L.A. Guercio, E. Linton, S.R. Neves-Zaph, P. Meerlo, G.S. Baillie, M.D. Houslay and T. Abel and published in Journal of Neuroscience 24 Aug 2016 36(34) 8936 – doi 10.1523/JNEUROSCI.0248-16.2016


Havekes and colleagues investigated the link between the binding of specific compartmentalized cAMP-specific phosphodiesterase 4 (PDE4) isoforms in mouse excitatory hippocampal neurons and cognitive changes associated with some neurological disorders. Expression levels of PDE4 isoforms are known to be altered in traumatic brain injury, autism, schizophrenia, bipolar disorder for example as well as being affected by ECT and antidepressant treatment. It is also known that the PDE4 isoforms exert their influence on cognitive capability by binding via their N terminals to specific protein complexes and affecting degradation of cAMP in specific intracellular compartments. In order to investigate the effect of the PDE4 isoforms in the hippocampal cells, Havekes and colleagues altered PDE4A5 and PDE4A1 expression in mice and performed various in vivo cognitive tests eg. object–place recognition task, fear-conditioning task, open field task, and zero maze task and several in vitro tests on cultured cells such as electrophysiology and fluorescence resonance energy transfer sensory imaging.

Havekes and colleagues found that virally induced PDE4A5 expression was observed in the excitatory neurons in hippocampus, but not in the astrocytes. This expression led to increased PDE4 activity in the hippocampus inducing reduced cAMP levels in this area, but not in the prefrontal cortex or cerebellum. The cAMP effect was not overall, but specific for certain intracellular compartments. The increased PDE4A5 protein levels were found not to alter basal synaptic transmission in the Schaffer collateral-CA1 pathway, but decreased synaptic potentiation.

They also investigated the link between PDE4A5 level and long-term context-shock associations and found that selective overexpression of PDE4A5 attentuated long-term memory. Increased protein levels did not affect freezing levels during training, but decreased freezing levels were observed when the mice were re-exposed after the conditioning training period. This result was explained by short-term memories not needing cAMP signaling whereas long-term memories did. Hence, PDE4A was said to lead to impairment of hippocampal plasticity resulting in long-term memory problems. Repeating the context shock associations with tone-cued fear conditioning instead (a process that uses the amygdala region rather than the hippocampus due to the fear element of the electric shock) found similar freezing levels under all conditions. Therefore, it was concluded that tone-cued fear conditioning is not affected by PDE4A5 levels in the hippocampus.

The authors also looked at the effect of PDE4A5 levels on performance of the object-location memory task in mice. They found that mice expressing eGFP or PDE4A5 reduced exploratory behaviour during training as they learnt the locations of objects. After learning, eGFP mice could remember the locations, but mice overexpressing the PDE4A5 protein demonstrated reduced memory and explored all of the objects to the same extent. In the case of the novel object recognition task, mice with both eGFP and PDE4A5 over-expression demonstrated the same exploration of novel objects showing that they could determine novel objects from familiar ones. An investigation of cAMP responses using the ICUE3 biosensor in hippocampal neurons expressing a control vector and full-length PDE4A5 found that baseline FRET responses were not affected by the overexpression. The attenuated forskolin-mediated FRET response could be normalized by application with the PDE inhibitor IBMX which suggested that the decrease in FRET response was due to the overexpression of PDE4A5 and not a result of nonspecific alterations in PDE/cAMP signaling.

It is known that the N terminal of the PDE4 isoforms is important for PDE4 binding to complex groups and Havekes and colleagues investigated if PDE4A5 also requires the N terminal for the context-shock results. The PDE4 isoform was truncated at the N terminal at 303bp and no impairment of long-term memory was found in this test. A repetition of the object-place memory test also found no difference between the eGFP and PDE4A5 over-expression animals. The investigation of cAMP responses using ICUE3 biosensor in hippocampal neurons expressing a control vector and full-length PDE4A5 which led to the attenuated forskolin-mediated FRET response which could be normalized by the application of the PDE inhibitor IBMX was also not observed with the truncated version. These investigations supported the view that the N terminal was important for the placement of the PDE4 isoform in the cell. Using fluorescent imaging, Havekes and colleagues found there was different intracellular distribution between the full and truncated versions. The full length form was found in discrete perinucleur areas and the dendritic compartments, whereas the truncated version was found predominately only in the former.

Havekes and colleagues also investigated another PDE4 isoform that of the PDE4A1. They found differences between PDE4A5 and PDE4A1 with the PDE4A5 isoform being membrane associated, whereas 4A1 was located in the Golgi. The investigators also found that overexpression of PDE4A1 produced no change in memory when tested using the object-location memory test. Hence, it was suggested that PDE4A1 does not target protein complexes critical for the formation of object location memories and that the two 4A5 and 4A1 isoforms affect different cellular compartments.

With the link between PDE4A5, its overexpression, cAMP increase and cognitive disorders being established, the authors concluded their article by suggesting that instigating N terminal changes would produce an alternative method of regulating the PDE4A5 cellular level. This method would be welcomed as an alternative to using the broad PDE4 inhibitors which cause such undesirable side effects such as diarrheoa and emesis.


What makes Havekes and colleagues article interesting is that it investigates indirectly the role of cyclic adenosine monophosphate (cAMP) in hippocampal cells and memory and perhaps gives an indication of one of the elements required in the process of ´switching off` an active cell once the synaptic stimulation is over. The article looks at the binding of cAMP-specific phosphodiesterase 4 (PDE4) isoforms to specific proteins in identified compartments of the post-synaptic regions of excitatory neurons in the mouse hippocampus. The authors found that expression of the PDE 4 gene leads to production of the protein and its subsequent specific binding to intracellular proteins results in a reduction in cellular cAMP level. Further investigation by the authors showed that this was a specific effect to one isoform of the PDE4 protein (the A5) and binding required a functional N terminal. Negative effects on cognition were attributed to this N terminal binding such as interaction with beta-arrestins, a molecular element critical for learning and memory and association with certain proteins containing the SH3 domain such as src tyrosyl kinase family, the inhibition of which also leads to memory defects.

From a biochemical point of view, PDE4A5 provides a tool by which cAMP functioning within the synaptic area can be investigated. Cyclic AMP is a multifunctional second messenger and its production from adenylate cyclase within the neuron is linked with the opening of chloride ion channels, protein kinase (PK) activation and gene transcription (eg. CREB phosphorylation). Therefore, if cAMP levels are reduced then either the PDE4A5 protein reduces cAMP production by binding directly to the adenylate cyclase enzyme (AC) and eliciting conformational changes that prevent the enzyme from working, or it increases the level at which the cAMP formed by a normal acting AC is degraded. Since the PDE4A5 protein is described as a phosphodiesterase (breakdown of cAMP to AMP) then the latter seems to be how this protein functions in the normal cell. Therefore, it can be said that if the level of this common second messenger is reduced on PDE4A5 binding then the protein is likely to play a role in the ´switching off` mechanisms of the neuronal cell after stimulation (e.g. in hyperpolarization for example). The question is which natural cAMP dependent neuronal functions is PDE4A5 likely to have an effect on?

Havekes and colleagues found in their study that PDE4A5 binding was perinucleur, dendritic and inter-compartmentalised. Therefore, the known role of cAMP in chloride ion channel functioning can be ruled out as a location for the PDE4A5 effect since chloride ion channels are situated on the cell membrane surface. Under normal activation, cAMP would increase the opening of the chloride channels to aid hyperpolarization and this is linked with GABA binding. The hippocampal CA3 area contains GABA interneurons and increased GABA receptor binding in this area is linked with fear memory which correlates with the observation that increased PDE4A5 expression results in anxiety and emotional memory changes. Hence, an increase in GABA binding leading to increased long-term depression of the relevant interneurons may mean that hyperexcitability of the CA1 area may occur. This would be consistent with the cognitive effects observed. However, since a membrane effect is not attributed to the PDE4A5 action then an influence on chloride channel opening by affecting cAMP level can probably be ruled out and it can be assumed that the cognitive effects observed with increased PDE4 expression come from other factors.

It is more likely that the PDE4A5 protein instigates its effect on cognition by influencing the performance of the various protein kinases involved in neuronal functioning. Cyclic AMP activates protein kinases by altering the enzyme`s quartenary structure and therefore, reduced cAMP would reduce the level of functioning protein kinases within the cell. For example, in the presence of PDE4A5 binding there would possibly be reduced activation of calcium calmodulin protein kinase which leads to decreased phosphorylation of the synapsin proteins and synaptogamin synaptic vesicular proteins. This would result in for example less vesicular transport in the synapse leading to less release and degradation of neurotransmitters and lower receptor trafficking. Therefore, it may be suggested that this could be a pathway by which the ´switching off ` of the synapse post-stimulation might occur.

A similar rationale could also be applied to the actin binding protein, girdin, which is one of the proteins responsible for the neuron`s actin-based cytoskeleton. This protein interacts with Src tyrosyl kinase which acts on the NR2B subunit of the NMDA receptor in the hippocampus. This type of glutamate receptor is linked to normal neuronal functioning after stimulation and long-term potentiation of the area. Therefore, a reduction in cAMP level induced by the PDE4A5 binding could lead to an effect on the actin cytoskeleton of the pre- and post-synaptic areas resulting in less vesicular transport and trafficking of proteins, receptors etc. as well as an effect on the very receptors that are linked with long-term potentiation and memory. A more direct influence of cAMP on the NMDA receptor also comes from its effect on post-synaptic protein kinase A (PKA) activation. This enzyme would normally phosphorylate a particular residue of the GluN2 subunit of the NMDA receptor and this subunit has been found to be critical for correct synaptic targeting of the receptor. Therefore, a reduced level of cellular cAMP would mean less protein kinase A phosphorylation of the subunit and lower NMDA receptor numbers at the cell membrane. It is likely that in this case long-term potentiation would not occur and this would result in lower or non-existent memory formation. Therefore, PDE4A5 binding would reduce neuronal functioning after stimulation and this effect would mean binding is located in the neuronal dendrites.

Havekes and colleagues also found that PDE4A5 binding was located in the perinuclear region of the cell and this could be explained by decreased PKA functioning, too. In this case, the PKA phosphorylates the cAMP response binding protein (CREB) which binds to the DNA. Activation of this protein results in changes in gene transcription, eg. nuclear factors such as Bdnf. There is evidence of CREB involvement in PDE4A5 binding and hence, reduced cAMP levels could ultimately affect the amount of gene transcription occurring at the nuclear level.

Therefore, it appears that PDE4A5 could be involved in the ´switching off` of the active neuronal cell and it is likely that this effect is brought about by the reduced cAMP level influencing protein kinase activity at both the perinuclear and dendritic locations. Since there is less known about the mechanisms involved in ´rebalancing` the cells after firing in readiness for the next firing stimulus, identification of elements such as PDE4A5 helps to elucidate the process. This is important because it may be possible in the case of cognitive disorders which involve the hyperexcitability of areas that manipulation of such an element can induce the cell to ´switch` off  thus returning the area to its correct firing level and restoring appropriate cognitive function.

Since we`re talking about the topic…………………..

……if PDE4A5 function is linked with protein kinase activity then can we assume that use of a PK inhibitor such as staurosporine would have no additional effect on cell functioning and cAMP level if PDE4A5 gene expression was increased?

…….can we assume that the administration of entomidate which effects GABA receptor binding and hyperpolarization through chloride ion channel opening confirms the non-involvement of cAMP at chloride ion channels in the presence of increased PDE4A5 expression?

….is it possible that investigation of neuronal activity of schizophrenic sufferers who are reported to have disrupted N terminal binding of PDE4A5 would demonstrate unusual protein kinase functioning and that further investigation of the areas and particular protein kinases would elucidate exactly where the PDE4A5 works?


This entry was posted in hippocampus, long-term memory, PDE4A5, Uncategorized and tagged , , . Bookmark the permalink.