Half a decade ago, the cyclic nucleotides cyclic adenosine 3′, 5′-monophosphate (cAMP) and cyclic guanosine 3′, 5′-monophosphate (cGMP) were identified as key second messenger molecules that mediate the intracellular effects of many signals known as “first messengers,” such as hormones or neurotransmitters. cAMP and cGMP signaling pathways regulate a vast number of physiological processes, including cell proliferation and differentiation, gene expression, apoptosis, and several metabolic processes, such as insulin secretion, glycogen synthesis, or lipogenesis. After their discovery, many years elapsed before cyclic nucleotides signaling proved to be a selective and effective process to modulate biological pathways. Since then, it has become clear that signaling by cyclic nucleotides modulates a countless number of biological functions, thus requiring a thigh control of their intracellular concentrations. This is achieved not only at the level of the synthesis of cAMP and cGMP, by adenylyl cyclases and guanylyl cyclases, respectively, but also at the level of degradation (Maurice et al. 2014). In 1958, Sundherland and colleagues reported that heart extracts contained an enzyme – a phosphodiesterase (PDE) – that hydrolyzed cAMP to AMP and thus reducing or terminating its action. Additionally, they found that the activity of PDE was blocked by methylxanthines, such as caffeine. Thus, the hydrolytic activity of PDEs provides the primary mechanism to counteract the synthesis of the cyclic nucleotide by the cyclases, reducing or terminating the second messenger signal.
PDEs belong to a superfamily of enzymes encoded by 21 genes and grouped in 11 distinct families (PDE1–11), each with unique properties in terms of their subcellular localization, substrate specificity, affinity of binding, enzymatic kinetics, and regulatory properties. It is now becoming clear that the intracellular distribution of the PDEs is likely to be a key regulator of local concentrations of cyclic nucleotides, thus modulating the amplitude and duration of the signal at spatially defined intracellular sites. Among the PDE families, phosphodiesterase 2 (PDE2A) is a cyclic GMP-stimulated phosphodiesterase that hydrolyzes both cAMP and cGMP with low affinity. The distinctive feature of this PDE is the stimulation or inhibition of cAMP hydrolysis by low or high concentrations of cGMP binding to one of its domains, thus allowing a cross-talk between cGMP- and cAMP-mediated signaling. Additionally, given that PDE2A is expressed in several different tissues, for example, brain, heart, liver, and adrenal gland, it is likely to be involved in the regulation of a wide variety of processes. In support of its important role, similar expression patterns have been found across several mammalian species from human to dogs and rodents (Stephenson et al. 2009).
Biochemistry and Structure
The catalytic domain of PDE2A shares the compact helical structure common to all PDEs. The active site, which is buried largely inside the catalytic domain, has two divalent metal ions (Zn2+ and Mg2+). Interestingly, an invariant glutamine residue in the catalytic domain (Gln859 in the human PDE2A sequence) functions as the key determinant for nucleotide specificity of PDEs in general. The surrounding residues anchor the glutamine residue in different orientations for cAMP and for cGMP: in cAMP-specific PDEs, this residue adopts a conformation that allows hydrogen bonding to the adenine base; in cGMP-specific PDEs, the glutamine residue orientation promotes the formation of H-bonds with the guanine base; in PDE2A, this residue can toggle between these two orientations conferring dual-specificity to the enzyme (Zhang et al. 2004). Nevertheless, other residues in the catalytic site, apart from the invariant Gln, may play an additional role in determining substrate specificity.
Function and Regulation
PDE2A is a dual substrate enzyme able to hydrolyze both cAMP and cGMP with approximately equal affinity (Km of 30 and 10 μM, respectively) and efficacy (Vmax of 120 and 123 μmol min−1 mg−1). Specifically, PDE2A hydrolyzes the cyclic phosphate ring that is unique to cAMP/cGMP by insertion of a solvent-derived hydroxyl at the phosphorous atom in the six-member cyclic phosphate ring. This reaction produces either 5′-AMP or 5′-GMP, both inactive as a second messenger in cyclic nucleotides pathways (Francis et al. 2009). A special feature of PDE2A is that it displays positive cooperativity for both cyclic nucleotides, that is, both cAMP and cGMP regulate PDE2A enzymatic activity. This positive cooperativity is due to the binding of nucleotides to the GAF-B domain (allosteric site), which induces a conformational change in the protein and increases the catalytic activity (Martins et al. 1982). Although the affinity of GAF-B for cGMP is 10 times greater than for cAMP, usually cAMP levels within cells are much higher than cGMP. Indeed, low concentrations of cGMP stimulate cAMP hydrolysis through binding to GAF-B, since GAF-A does not bind cGMP. This allosteric stimulation of cAMP hydrolysis by cGMP is biologically more important than inhibition of cAMP hydrolysis via competitive binding to the catalytic site, although cAMP hydrolysis can be inhibited in vitro by high concentrations of cGMP. On the other hand, cAMP is able to stimulate cGMP hydrolysis but only at low concentrations of cGMP. Thus, allosteric interaction with both cyclic nucleotides can enhance the catalytic activity of PDE2A. Such a negative feedback mechanism allows a cross-talk between cAMP and cGMP signaling pathways (Francis et al. 2009). The crystal structure of a dimerized PDE2A without cyclic nucleotides suggests that in this state the catalytic domain predominantly adopts a closed conformation in which the H- and M-loops of the catalytic domain (residues 702–723 and 830–856 of human PDE2A, respectively) are folded in at the dimer interface, preventing substrate access to the catalytic site. cGMP binding to GAF-B induces a conformation change between GAF-B and catalytic domain causing the catalytic domain to move apart and adopting an open conformation. This structural change allows the H-loop to swing open, allowing substrate access to the catalytic site (Fig. 1c, d) (Pandit et al. 2009).
Mechanisms other than allosteric regulation by cGMP may also be important in PDE2A function. AIP, a component of the aryl hydrocarbon receptor (AhR) complex, was shown to be a PDE2A interactor. Binding of PDE2A to AIP inhibits the nuclear translocation of AhR in hepatocytes. Thus, cyclic nucleotide signaling has a role in AhR trafficking and PDE2A, through biding to AIP and may regulate AhR mobility in the cytosol. There is so far no evidence of posttranslational modifications that regulate PDE2A activity (Azevedo et al. 2014). There is also little knowledge regarding regulation of PDE2A gene expression or its turnover in cells. For example, PDE2A expression increases after treatment with vascular endothelial growth factor or tumor necrosis factor-α in human umbilical vein endothelial cells. However, the pathway that mediates this regulation of PDE2A remains unknown.
The ability of PDE2A to hydrolyze a confined pool of cAMP is supported by a number of studies. For example, while a rise in cAMP due to PDE3 and PDE4 inhibition results in hypertrophy, increasing cAMP levels via PDE2A inhibition counteracts hypertrophy. PDE2A inhibition involves the generation of a local pool of cAMP and activation of PKA type II subset, leading to phosphorylation of the transcription factor nuclear factor of activated T cells (NFAT) (Zoccarato et al. 2015). In line with this model, increased expression of PDE2A has been reported to be associated with heart failure in several animal models as well as in humans (Aye et al. 2012), where excessive hydrolysis of a local pool of cAMP may result in reduced phosphorylation of NFAT and consequent activation of pro-hypertrophic genes, possibly contributing to the myocardial remodeling associated with heart failure.
PDE2A-mediated cAMP hydrolysis has also been associated with promotion of cardiac fibroblast to myofibroblast (MyoCF) conversion, another process contributing to cardiac remodeling. Although cGMP activates PDE2A, cGMP-elevating stimuli surprisingly antagonized PDE2-induced MyoCF phenotype (Vettel et al. 2014).
Since PDE2A is expressed in a wide variety of human tissues, such as brain, neurons, heart, liver, kidney, lung, adrenal gland, platelets, endothelial cells, or macrophages, other biological functions include memory improvement, thrombin-induced activation of human platelets, excitation/contraction coupling, monocytes to macrophages differentiation, and endothelial barrier function (Zaccolo and Movsesian 2007; Azevedo et al. 2014). As suggested by several studies, the functional outcome downstream of PDE2A is unlikely to rely exclusively on the activity of this enzyme as cyclic nucleotides levels are regulated by a complex network where multiple PDEs contribute simultaneously to regulate a specific pool of second messenger (Zaccolo and Movsesian 2007). Additionally, other enzymes such as cAMP/cGMP-dependent kinases, phosphatases, and cyclases, together with PDEs, form a complex signaling system that allows compartmentalization of cyclic nucleotides signaling in order to generate specific intracellular responses to different stimuli.
Genetics and Subcellular Localization
A single PDE2A gene (chromosome 11q13.4) gives rise to multiple PDE2A variants due to alternative splicing and use of alternative initiation codons: the soluble PDE2A1; the membrane-associated PDE2A2 and PDE2A3 (Russwurm et al. 2009); in addition, the sequence for a PDE2A4 variant has been deposited (NCBI Reference Sequence: NM_001146209.2); however, no further characterization of this isoform is available. All PDE2A isoforms share the same C-terminal sequence (including GAF-A, GAF-B, and catalytic domains) although each isoform has a unique N-terminal sequence (Azevedo et al. 2014). These isoforms have been cloned from different tissues/organisms: bovine adrenal (PDE2A1), rat brain (PDE2A2), and human brain (PDE2A3) tissues. PDE2A1 still remains as the only splice variant whose amino acid sequence was determined by sequencing the purified protein. The splice variants 2 and 3 were cloned from a rat brain and bovine/human cDNA libraries, respectively. However, it still remains unclear which PDE2A isoform is expressed as a protein in a given tissue or species since reliable isoform-specific antibodies have not been developed yet (Russwurm et al. 2009). Nevertheless, there are no known differences in kinetics behavior among the PDE2A isoforms.
The N-terminal region of each PDE2A isoforms differs in its hydrophobicity, making it likely to determine the subcellular localization of the different isoforms. The PDE2A1 variant, which has the most hydrophilic N-terminus, was originally isolated from the soluble fraction of bovine adrenal gland. N-terminal PDE2A fragments fused to a fluorescent protein were shown to be sufficient to mediate subcellular targeting (Acin-Perez et al. 2011). According to this study, PDE2A1, PDE2A2, and PDE2A3 localize in the cytosol, mitochondria, and plasma membrane, respectively. The N-terminus of PDE2A3 features a motif for N-myristoylation at Gly2 that may regulates its plasma membrane targeting (Russwurm et al. 2009).
Pharmacology and Perspectives
PDE2A-specific inhibitors provide a powerful tool for investigating cyclic nucleotides signaling and, more specifically, for dissecting the role played by this dual substrate phosphodiesterase in the complex cyclic nucleotide signaling pathway. For many years, the only PDE2A inhibitor available has been EHNA, which however lacks the high selectivity and potency for conclusive pharmacological evaluation since it also acts as a potent adenosine deaminase inhibitor (Fig. 3a) (Gomez and Breitenbucher 2013). More recently, Bayer has disclosed several potent and more selective PDE2A inhibitors, including BAY-60-7550 (Fig. 3a) (Boess et al. 2004). BAY-60-7550 inhibits the hydrolytic activity of purified PDE2A (IC50 = 0.002 μM) and is selective with respect to other PDE isoforms (50-fold versus PDE1C, >100 fold selectivity over other PDE isoforms). Although none of the inhibitors that have been developed so far has been tested in humans, animals’ studies have suggested promising effects of these compounds, especially BAY-60-7550, in improving learning ability (Boess et al. 2004). Furthermore, BAY-60-7550 produced antidepressant-like effects in stressed mice, supporting a potential role for PDE2A in cognitive impairment related to stress. Although the poor pharmacokinetics properties of BAY-60-7550 probably preclude its use in clinical studies, the compound has been an essential tool to study the role of PDE2A in cyclic nucleotides signaling and to test the potential therapeutic use of PDE2A inhibition (Gomez and Breitenbucher 2013).
In 2013, the crystal structure of PDE2A bound to BAY-60-7550 was published, along with the PDE2A apo structure (Fig. 3a). This showed that BAY-60-7550 binds to the catalytic site of the enzyme and interacts with the conserved glutamate residue Gln589. When comparing both structures, it appears that binding of BAY-60-7550 induces a conformational change of the enzyme revealing a hydrophobic pocket which can accommodate the propylphenyl group of the compounds. This may contribute to the high selectivity of BAY-60-7550 for PDE2A over other members of the PDE family since this hydrophobic pocket was not found in any other published PDE-inhibitor complex (Zhu et al. 2013). So far, there is only one PDE2A inhibitor under clinical study (phase I), developed by Pfizer, suggested for migraine (Maurice et al. 2014).
More recently, PDE2A was characterized in human malignant melanoma pseudomyxoma peritonei cells which have mutations in PDE2A2. These PDE2A mutations were associated with downregulation of cyclin A and induction of G2/M arrest, thus raising the possibility that PDE2A2 might be a future target in treating malignant melanoma (Morita et al. 2013).
Although no known diseases have been firmly associated with PDE2A dysfunction so far, PDE2A knockout mice have been reported to be embryonically lethal at E17-E18 (Stephenson et al. 2009).
Recently, a promising tool based on optogenetics (biological technique that uses light to control cells in a living tissue) has been developed. A light-activated phosphodiesterase (LAPD) has been engineered which is light sensitive and catalytically active since it has the effector module from human PDE2A. Upon red-light absorption, LAPD upregulates hydrolysis of both cAMP and cGMP up to sixfold. On the other hand, far-red light can be used to downregulate its activity. This new tool can be used in living organisms and has the potential to unveil new functions of PDE2A (Gasser et al. 2014).
PDE2A is a 105 kDa homodimer that exists in particulate and soluble forms, depending on one of the four splicing variants. Full-length PDE2A includes an N-terminal-sequence, which differs between different variants, two GAF domains that regulate dimerization and catalytic activity in response to cGMP and one C-terminal catalytic domain. PDE2A is an enzyme that has the ability to terminate cyclic nucleotide signaling by catalyzing the hydrolysis of both cAMP and cGMP with similar rates with positively cooperative kinetics. For instance, once cGMP binds to the GAF domain, PDE2A undergoes a conformation change, and its catalytic activity increases. Thus, a unique property of PDE2A is its ability to link cAMP and cGMP signaling pathways, that is, PDE2A mediates negative crosstalk between the two cyclic nucleotides.
A unique PDE2A gene gives rise to multiple PDE2A variants due to alternative splicing, and each isoform has a different subcellular localization as a result of different hydrophilicities of its N-terminus. As a consequence, it is likely that the different isoforms may have different functions within the cell. Development of improved and isoform-specific antibodies could help defining isoform-specific effects.
PDE2A is expressed in multiple tissues and cell types, such as brain, heart, platelets, endothelial cells, and macrophages. Thus, it can potentially regulate a variety of different physiological processes and functions. However, to date PDE2A inhibitors have only been used as research tools, rather than in clinical studies. Nonetheless, since the discovery in 2002 of the first potent and selective PDE2A inhibitor, BAY-60-7550, there is now significant preclinical information for the potential role of PDE2A inhibitors in the treatment of pathological conditions, including disorders with a cognitive component. Since PDE2A is a potential pharmaceutical target widely spread among different tissues and cell types, overcoming possible side effects when treating one disease condition may become a challenging task – thus, the need for developing compounds with improved features.
- Vettel C, Lämmle S, Ewens S, Cervirgen C, Emons J, Ongherth A, et al. PDE2-mediated cAMP hydrolysis accelerates cardiac fibroblast to myofibroblast conversion and is antagonized by exogenous activation of cGMP signaling pathways. Am J Physiol Heart Circ Physiol. 2014;306(8):H1246–52.PubMedCrossRefGoogle Scholar