Cyclic Nucleotide Phosphodiesterases: important signaling modulators and therapeutic targets

Abstract

By catalyzing hydrolysis of cAMP and cGMP, cyclic nucleotide phosphodiesterases are critical regulators of their intracellular concentrations and their biological effects. Since these intracellular second messengers control many cellular homeostatic processes, dysregulation of their signals and signaling pathways initiate or modulate pathophysiological pathways related to various disease states, including erectile dysfunction, pulmonary hypertension, acute refractory cardiac failure, intermittent claudication, chronic obstructive pulmonary disease, and psoriasis. Alterations in expression of PDEs and PDE-gene mutations (especially mutations in PDE6, PDE8B, PDE11A and PDE4) have been implicated in various diseases and cancer pathologies. PDEs also play important role in formation and function of multi-molecular signaling/regulatory complexes called signalosomes. At specific intracellular locations, individual PDEs, together with pathway-specific signaling molecules, regulators, and effectors, are incorporated into specific signalosomes, where they facilitate and regulate compartmentalization of cyclic nucleotide signaling pathways and specific cellular functions. Currently, only a limited number of PDE inhibitors (PDE3, PDE4, PDE5 inhibitors) are used in clinical practice. Future paths to novel drug discovery include the crystal structure-based design approach, which has resulted in generation of more effective family-selective inhibitors, as well as burgeoning development of strategies to alter compartmentalized cyclic nucleotide signaling pathways by selectively targeting individual PDEs and their signalosome partners.

Introduction

Cyclic nucleotide phosphodiesterases (PDEs) catalyze the hydrolytic destruction of the intracellular second messengers, cyclic adenosine monophosphate (cAMP) and cyclic guanosine monophosphate (cGMP), to produce 5’-AMP and 5’-GMP, respectively. These second messengers are important regulators of intracellular signal transduction, particularly those arising via activation of extracellular membrane receptor proteins, which lead to activation adenylyl and guanylyl cyclases and intracellular signaling pathways that regulate many critical physiological processes, including cell proliferation and apoptosis, myocardial contractility, vascular and airway smooth muscle relaxation, reproduction, immune/inflammatory responses, visual transduction, secretion, bone development, etc.. Conversely, dysruptions in these signaling pathways play key roles in the pathogenesis and symptomatic presentations of many clinically relevant disease states, including obesity, diabetes, heart failure, pulmonary hypertension, erectile dysfunction, chronic obstructive pulmonary disease (COPD), arthritis, certain forms of retinal degeneration, and cancer, etc. (Bender and Beavo 2006;Francis et al. 2011;Manganiello et al. 2007;Tsai and Kass 2009;Maurice et al. 2014a).

As depicted in Fig.1, intracellular cAMP and cGMP levels are controlled by adenylyl and guanylyl cyclases, which catalyze their synthesis, and by PDEs, which catalyze their hydrolysis. Cyclic nucleotide-regulated signal transduction can be initiated through several mechanisms, i.e., by cyclic nucleotide-induced activation of cAMP- and cGMP-protein kinases (PKA and PKG, respectively), with subsequent phosphorylation and regulation of downstream effectors, or by binding to and activation of specific cyclic nucleotide binding proteins which directly mediate cyclic nucleotide actions, e.g., Epac 1 and Epac 2 (cAMP-activated guanine-nucleotide-exchange proteins), cyclic nucleotide-gated (CNG) ion channels, and several PDEs (especially PDEs2, 5, 6) which contain allosteric cyclic nucleotide-binding sites in addition to their catalytic sites (Bender and Beavo 2006;Francis et al. 2011;Manganiello et al. 2007;Tsai and Kass 2009). cAMP-binding directly activates Epac proteins, resulting in activation of Rap1 (small GTPase) and diverseRap1-signaling pathways (Bos 2006;de Rooij et al. 1998). CNG channels, which belong to the superfamily of voltage gated ion channels, are relatively highly expressed in sensory and olfactory neurons where they mediate chemosensory transduction (Biel and Michalakis 2009).

Figure 1. Cyclic-nucleotide mediated signal transduction of phosphodiesterases.

The PDE superfamily contains 11 highly regulated and structurally related gene families. Some PDE's are specific for cAMP (PDE-4, -7, -8), some are cGMP specific (PDE-5, -6, -9), and some exhibit mixed specificity (PDE-1, -2, -3, -10, -11). Individual families contain closely-related subfamilies, e.g., PDE1A, PDE1B, PDE1C. Cyclic nucleotides are generated through activation of cyclases in response to a variety of regulatory signals or hydrolyzed by phosphodiesterases. Activation of cAMP- and cGMP-dependent protein kinases: protein kinase A (PKA) and protein kinase G (PKG) leads to the phosphorylation of specific substrate proteins that initiate and regulate cyclic nucleotide-dependent and kinase-mediated signaling pathways and their biological effects. Some effects of cyclic nucleotides are mediated by binding to and interaction with specific proteins such as cAMP-GEF, which activate RAP1-pathways. CNG ion channels activated by cAMP and cGMP are important cellular switches and transduce changes in intracellular cyclic nucleotides into changes in membrane potential. PDE's can be activated by binding of cAMP or cGMP to non catalytic sites.

In catalyzing the hydrolysis of cAMP and cGMP, PDEs play crucial roles in controlling their intracellular concentrations, the compartmentalization of their signaling pathways, and their downstream physiological effects. PDEs constitute a large and diverse gene superfamily which contains 11 distinct gene families (PDEs 1-11) (Fig. 1), classified according to their affinities for cAMP and cGMP, inhibitor sensitivities, responses to specific effectors, mechanisms of regulation, and in their subcellular targeting and recruitment into specific, multimolecular regulatory complexes or “signalosomes” (Francis et al. 2011;Houslay et al. 2007;Maurice et al. 2014a). Most families contain more than one gene (e.g., PDE1A, PDE1B, PDE1C).The PDE superfamily encodes many proteins, containing close to 100 distinct PDE isoforms produced via alternative mRNA splicing or transcriptional and translational processing. Based on their substrate specificities and selectivities, PDE families can be grouped into three broad categories: some are specific for cAMP hydrolysis (PDE4, PDE7, and PDE8); others, cGMP-specific (PDE5, PDE6, and PDE9); and some families exhibit dual specificity (PDE1, PDE2, PDE3, PDE10, and PDE11), hydrolyzing both cAMP and cGMP with different affinities (Fig.1). In addition, pharmaceutical companies have developed specific and potent family-selective inhibitors for almost all the PDE families (Table1) (Francis et al. 2011). It is of interest that, in the 1970s and 1980s, basic biochemical techniques (e.g., gel filtration and ion exchange chromatography of tissue extracts), coupled with analysis of catalytic properties of partially purified proteins (e.g., specificities and affinities for cGMP and/or cAMP as well as different sensitivities to available inhibitors and effectors such as calcium/ calmodulin) identified five distinct PDEs, now recognized as PDEs1-5 (Keravis et al. 1980;Lugnier et al. 1986;Thompson et al. 1979;Thompson and Appleman 1971;Weishaar et al. 1985;Yamamoto et al. 1983). Thus, the molecular diversity of the PDE family was recognized before the tools of molecular genetics and biology identified the PDE super gene family (PDEs 1-11) and the large number (~100) of protein isoforms generated by the PDE genes (Bender and Beavo 2006;Francis et al. 2011;Manganiello et al. 2007;Tsai and Kass 2009).

Table-1

Family	Substrate/Regulation	Inhibitors	Functions of family inhibitors
PDE1	cAMP/cGMP (Ca2+/Calmodulin Stimulated)	Vinpocetine, SCH51866 8-MeoM-IBMX IC86340 ITI-214	•inhibitors block catecholamine-induced cardiac hypertrophy in cultured myocytes and by in vivo perfusion (Miller et al. 2009), block VSMC proliferation (Nagel et al. 2006), reduce pathological vascular remodeling (Cai et al. 2011) •inhibitors are being studied as cognition enhancers (Reneerkens et al. 2009) and as Therapy for schizophrenia (News and analysis 2011) •inhibitors are used as anti-cancer and anti-inflammatory agents (Jeon et al. 2010; Medina 2010;Shimizu et al. 2009)
PDE2	cAMP/cGMP (cGMP-Stimulated)	EHNA, BAY-60-7550 PDP,Oxindole	•inhibitors induce L-type Ca2+ current (Ica) in myocytes (Rivet-Bastide et al. 1997), increase memory functions and neuronal plasticity (Boess et al. 2004), provide therapeutic benefits in sepsis and ARDS (Seybold et al. 2005) •inhibit cell cycle progression of endothelial cells, block cell proliferation and pathological angiogenesis (Favot et al. 2003)
PDE3	cAMP (cGMP-Inhibited)	Cilostazol, Cilostamide, K-134, milrinone	•inhibitors for treatment of intermittent claudication (Stevens et al. 2012), stroke (Lee et al. 2013b), heart failure (Movsesian et al. 2009), improved left ventricular relaxation with accelerated SR Ca(2+) ATPase activity (Yano et al. 2000), antithrombotic therapy (Yoshida et al. 2012), and inhibition of oocyte maturation (Masciarelli et al. 2004) • inhibitors block proliferation of human malignant melanoma, squamous cell carcinoma, and human HSG cells (Murata et al. 2002)
PDE4	cAMP(cGMP-insensitive)	Rolipram, Roflumilast, Cilomilast, Etazolate, Ibudilast, Apremilast	•inhibitors are FDA-approved for COPD (Rabe 2011) and psoriatic arthritis (Wittmann & Helliwell 2013) •inhibitors used for Alzheimer's (Vellas et al. 2011), memory improvement, neuroprotection, inflammation, (Barad et al. 1998;Chen et al. 2007;Oger et al. 2005;Reneerkens et al. 2009), multiple sclerosis (Mangas et al. 2010), rheumatoid arthritis (Jacque & Van den Bosch 2013), and murine systemic lupus erythematosus (Keravis & Lugnier 2012) •inhibitors suppressed the growth of cancer cells (lung, colon, brain, pancreatic, and CLL) (Lin et al. 2013;Pullamsetti et al. 2013)
PDE5	cGMP	Sildenafil (Viagra), Tadalafil (Cialis), Vardenafil (Levitra) Avanafil Dipyridamole Zaprinast	•inhibitors used for erectile dysfunction (Yuan et al. 2013), pulmonary hypertension (Oudiz et al. 2012) and benign prostatic hypertrophy(Gacci et al. 2014) •inhibitors are being studied for diabetes, atherosclerosis and metabolic syndrome (Francis & Corbin 2011;Lugnier 2011) and lower urinary urinary tract symptoms (LUTS) (Uckert & Oelke 2011) •inhibitors suppress the growth of breast cancer cells (Tinsley et al. 2011)
PDE6	cGMP (transducin activated)	Zaprinast, Sildenafil, Vardenafil, Tadalafil	• most PDE5 inhibitors weakly inhibit PDE6. PDE6 isozymes are highly expressed in photoreceptors •mutations in PDE6 are associated with hereditary eye diseases (Huang et al. 1995; Muradov et al. 2003)
PDE7	cAMP (Rolipram Insensitive)	Thiadiazoles, BRL 50481, IC242,BMS, ASB6165	•inhibitors are antiinflammatory and neuroprotective (Safavi et al. 2013). •PDE7B expression is significantly increased in CLL cells and inhibition of PDE7 promoted apoptosis of cancer cells (Zhang et al. 2008a)
PDE8	cAMP(Rolipram-insensitive, IBMX-insensitive	PF-04957325	• inhibitors are used to inhibit steroidogenesis in the adrenal gland (Tsai et al. 2011) •polymorphisms in PDE8 are associated with developing prostate, testicular and adenocortical tumors(Horvath et al. 2008)
PDE9	cGMP (IBMX Insensitive)	BAY-73-6691, PF-04447943, SCH51866	•Inhibitors enhance cognition as PDE9 influences NO-cGMP signaling in rodent brain (Hutson et al. 2011;Reneerkens et al. 2009) •enzyme expression is increased in breast tumors (Karami-Tehrani et al. 2012)
PDE10	cGMP-sensitive, cAMP-selective	Papaverine, TP-10,MP-10	•use of inhibitors represent a novel approach for treatment of psychosis as PDE10 families contribute to overall brain function by regulating cGMP signaling (Kelly & Brandon 2009;Siuciak et al. 2006)
PDE11	cGMP-sensitive, dual spec	None selective	•Polymorphisms in the genes encoding PDE11 have been associated with developing prostate, testicular and adenocortical tumors (Faucz et al. 2011;Horvath et al. 2009)

Intracellular pools of cAMP and cGMP and their signaling pathways and biological responses are temporally, spatially, and functionally compartmentalized (Bender and Beavo 2006;Francis et al. 2011;Keravis and Lugnier 2012;Manganiello et al. 2007;Tsai & Kass 2009). PDEs are not only critical regulators of the amplitude, duration, and termination of intracellular cyclic nucleotide signals, but PDEs also play an important role in this exquisite compartmentalization (Conti et al. 2014;Houslay 2010;Maurice et al. 2014a). Early studies, for example, indicated that, in perfused rat hearts, although both epinephrine and PGE1 increased cAMP and PKA activity, they activated different pools of PKA, with only epinephrine increasing phosphorylase activity and myocardial contractility (Hayes et al. 1980). With respect to PDE localization/compartmentation, early studies indicated that, in rat adipocytes, insulin-induced activation of a membrane associated PDE3, not cytosolic PDEs, was critical in its antilipolytic action (Elks and Manganiello 1985;Manganiello and Vaughan 1973). More recent studies using fluorescence resonance energy transfer (FRET)-imaging or biosensors, such as EPAC-based cAMP sensors and cyclic nucleotide-sensitive ion channels, have verified the temporal, spatial, and functional compartmentation of cyclic nucleotides and their signaling pathways (Francis et al. 2011;Jurevicius and Fischmeister 1996;Zaccolo et al. 2000).

Compartmentalization of specific sets of cAMP signal transducers and effector systems permits activation and regulation of individual signaling pathways, as well as propagation of specific downstream signals and biological responses At the molecular level, specific PDEs or subsets of PDEs contribute to this compartmentalization by their permanent or reversible recruitment to different subcellular locations where they are incorporated into different macromolecular cyclic nucleotide signaling complexes or signalosomes (summarized in Table-2). In these signalosomes PDEs assume specific functional roles by regulating distinct cyclic nucleotide signaling pathways at these different subcellular locations (Francis et al. 2011;Maurice et al 2014a;Maurice et al. 2014b). Mechanistically, individual PDEs, in an isoform-specific manner, are targeted to different subcellular locations via their interactions with cellular structural elements and organelles such as lipid rafts and caveolae, with molecular scaffolds such as A Kinase Anchoring Proteins (AKAPs), β-arrestin, or RACK1 (Receptor for activated C kinase), or with other adaptor complexes or regulatory partners, including EPACs. These binding interactions are highly regulated and reversible. These compartmentalized signalosomes contain PDEs together with pathway-specific regulators and effectors, as well as different proportions of cyclic nucleotide effectors, including adenylyl cyclases, cyclic nucleotide-dependent protein kinases (PKA and PKG) and other kinases, Epacs (guanine nucleotide exchange proteins activated by cAMP), phosphoprotein phosphatases, A-Kinase anchoring proteins (AKAPs) and/or other scaffold proteins (Table 2 represents a partial list of PDE-containing signalosomes, with scaffold proteins, known interacting partners, and presumed function.) (Francis et al. 2011;Houslay et al. 2007;Keravis and Lugnier 2010;Maurice et al. 2014a;Maurice et al. 2014b;Tsai and Kass 2009). Incorporation of individual PDEs into specific signalosomes regulates specificity of signal transduction in two ways: by tight temporal and spatial regulation of cyclic nucleotide concentrations and gradients within specific compartments and by preventing diffusion of signals into neighboring compartments.

Table 2.

PDE	Scaffold/Docking proteins	Signaling partners	Function
PDE1		1. (Ca2+)calmodulin	1. Sperm function (Lefievre et al. 2002)
PDE2	1. SPHKAP	1. PKA/PKG	1. cardiac adrenergic response (Aye et al. 2012)
PDE3A	1. AKAP18	1. SERCA2, PKA-R, PP2A, PLN	1. SR Ca2+ uptake and myocardial function(Beca et al. 2013)
	2. BIG	2. PKA-R	2. ARF function(Puxeddu et al. 2009)
	3. 14-3-3	3. PKC	3. Platelet inhibition(Hunter et al. 2009)
	4. PI3Kγ	4. PKA, β-AR, Cav1.2, PLB	4. arrhythmia (Ghigo et al. 2012)
		5. PKAc, Cdc25C, PP2A	5. meiosis and female fertility (Shen et al. 2010)
		6. CFTR	6. submucosal gland secretion (Penmatsa et al. 2010)
PDE3B	1. IRS,14-3-3	1. PI3K, HSP90, PKB	1. Insulin-induced signaling complex, cAMP pools and lipolysis. (Ahmad et al. 2007)
	2. 14-3-3, Cav-1	2. PKA-R, HSP90, Cav-1, PL, PP2A, β3-AR	2. β3-AR -and insulin-signaling complexes. (Ahmad et al. 2009)
	3. IR	3. PI3K, IRS	3. Insulin/PI3K activation of PDE3B (Rondinone et al. 2000)
	4.14-3-3	4. PKA, PKB	4. 14-3-3 binding regulate PDE3 activity (Palmer et al. 2007)
	5. PI3Kγ	5. MAPK, PKB, PKA	5. inflammation, cardiac contractility (Patrucco et al. 2004; Perino et al. 2011)
	6. PI3Kγ	6. Epac	6. angiogenesis, cell adhesion (Wilson et al. 2011).
PDE4	1. Disc1	1. PDE4D9, ATF4, D1-R	1. neuronal functions(Soda et al. 2013)
	2. Disc1	2. PDE4A5, MK2, PKA and AIP	2. immune, inflammatory, and stress responses(MacKenzie et al. 2011)
	5. Yotiao	5. I(Ks), PDE4D3, PKA, PP1, KCNQ1, AC	5. Cardiac I(Ks) Potassium Channel and arrhythmia (Terrenoire et al. 2009)
	6. AKAP79	6. PDE4D, β-arrestin, β2-AR, ERK, PKA	6. β2AR activation of ERK (Lynch et al. 2005)
	7. AKAP95, AKAP149, MTG	7. PDE4A, PDE7A, PKA-RII	7. activation of T-cells (Asirvatham et al. 2004)
	8. mAKAP	8. PDE4D, PKA, RyR2	8. myocardial contractility and arrhythmia (Dodge et al. 2001; Lehnart et al. 2005)
	11. PI3K	11. PKB, PDE4	11. activation of T-cells (Bjorgo et al. 2010)
	12. RACK-1, β-arrestin	12. PDE4D5, ERK	12. β2-AR signaling (Bolger et al. 2006)
	13. Myomegalin	13. PDE4D	13. Targeting to Golgi/centrosome (Verde et al. 2001)
	14. Spectrin	14. PDE4D4	14. EC gap formation (Creighton et al. 2008)
	15. Ndel1	15. PDE4D3, PKA	15. neuronal development (Collins et al. 2008)
	16. SHANK	16. PDE4D, CFTR, PKA	16. membrane ion transport (Lee et al. 2007)
		17. PKA,PDE4D, Epac1, VE-Cad	17. VEC permeability and adhesion(Rampersad et al. 2010)
		18. PDE4A5, XAP2	18. attenuation of PDE activity (Bolger et al. 2003)
		19. PDE4B, LTCC	19. myocyte Ca++ release(Leroy et al. 2011)
		20. HSP20, PKA, β-AR	20. myocyte hypertrophy(Sin et al. 2011)
		21. PDE4D, SERCA2, PLN	21. cardiac contractility (Beca et al. 2011)
PDE5		1. IP(3)R1, PKG, IRAG	1. Platelet function (Wilson et al. 2008)
PDE6	1. β-arrestin	1. PDE6, transducin	1. Photoreceptor activation (Matte et al. 2012;Wensel 2008)
PDE7	1. MTG	1. PKA	1. T cell activation(Asirvatham et al. 2004)
PDE8	1. AKAP-LBC	1. PDE8A, Raf1, PKA, AKAP-Lbc, MEK, ERK	1. PDE8A binds Raf1 and exerts inhibitory effect on Raf1 and Erk signaling (Brown et al. 2013;Smith et al. 2010)

Because of the critical importance of cyclic nucleotide signaling in health and disease and the critical roles that PDEs play in regulating cyclic nucleotide signaling pathways and their biological effects, the mammalian PDE superfamily has been a major target for drug discovery (Bender and Beavo 2006;Francis et al. 2011;Keravis and Lugnier 2012;Maurice et al. 2014a;Tsai and Kass 2009). Currently, however, only a small number of PDE inhibitors have been approved as therapeutics for a limited number of disease states, including erectile dysfunction (ED), benign prostatic hyperplasia (BPH), chronic obstructive pulmonary disease (COPD), pulmonary hypertension, intermittent claudication, heart failure, psoriatic arthritis, etc. Recent advances in PDE biology, especially the experimental confirmation of the signalosome concept, have fostered new interest in targeting PDEs. There is increased understanding not only that the extensive molecular diversity of PDEs has been harnessed to regulation of many specific, compartmentalized signaling pathways and physiological and pathophysiological processes, but also that these compartmentalized PDEs and their signalosome partners present unique therapeutic targets, and that targeting them could generate less toxic therapeutics with greater specificity and efficacy. There is considerable hope and expectation for generating these therapeutics because of these remarkable advances in both PDE biology and structural biology, especially the computational design approach to novel drugs, as well as the development of increasingly sophisticated strategies to target specific PDEs and their signalosome partners (Houslay 2010;Keravis and Lugnier 2010;Lee et al. 2013a).

This review first discusses structure/function information that is important in PDE regulation, in their incorporation into and function in signalosomes, and in design of specific PDE inhibitors. We then discuss the signalosome “concept”, and present selected examples of the role of PDE-containing signalosomes in regulation of the cyclic nucleotide signaling pathways and biological effects in specialized subcellular compartments. The signalosome concept supports the notion that not only individual PDEs play specific functional roles in regulating specific, compartmentalized cyclic nucleotide signaling pathways, but that they also are potential targets for new therapeutics. We also review some physiological and pathophysiological functions of different PDEs, and the current status of clinically-approved PDE inhibitors, as well as potential therapeutic possibilities in the development of new family specific PDE inhibitors as well as pharmacologic modulators and disruptors which could target individual PDEs and their signalosome partners.

Structure/Function analysis of PDEs

As depicted in Fig. 2, PDEs are modular proteins that have a common structural organization, with a conserved catalytic core (250-300 amino acids) in the C-terminal portion of the PDE molecules which separates divergent N-terminal regulatory regions and targeting domains from hydrophilic C-termini (Ahmad et al. 2012;Bender and Beavo 2006;Francis et al. 2011).

Figure 2. Domain organization of different PDE gene families.

The N-terminal part of the PDE gene family (shown in black) contains targeting domains responsible for localization of PDE isoforms to specific subcellular sites of and signalosomes. The N-terminal regulatory domain (shown in green) contains PDE family-specific subdomains [allosteric ligand binding sites, i.e. GAF domain for PDE2,5,6,10,11; Ca+2/Calmodulin-binding site for PDE1; transmembrane domain in PDE3; upstream conserved regions (UCRs) in PDE4; per-arnt-sim (PAS) and the Y-homologous receiver domain (REC) in PDE8]. N-terminal hydrophobic regions (NHRs) in PDE3 are important in regulating subcellular localization and formation of compartmentalized signalosomes. The conserved catalytic domain (shown in blue) is located in the carboxy-terminal part of the PDEs. PDE3 catalytic domain contains a unique 44-amino-acid insert (shown in red).

Catalytic domain

The PDE catalytic core is more highly conserved among members of the same family (>80% amino acid identities) than between different gene families (~25-40%). It contains a histidine-rich PDE signature sequence motif [HD(X2)H(X4)N] and two concensus metal-binding domains (HX₃HX_24-26E) which are common to all PDEs and critical for catalytic function (Francis et al. 1994;Zhang et al. 2002). Site directed mutagenesis of these critical histidines abrogates activity of several PDE isoforms. Zn²⁺ is thought to bind to one of the two metal-binding sites site in most PDEs (perhaps not to PDE3). Although Mg⁺⁺ is generally considered as the second metal ligand, Mn⁺⁺ and Co⁺⁺ also support PDE catalytic activity (Francis et al. 2011).

Analyses of the crystal structures of the catalytic domains of different PDE families (PDEs 1-5, and 7-10) indicate that core catalytic domains exhibit a uniform conformation, with primary differences at their NH2- or COOH terminal ends (Ke and Wang 2007;Liu et al. 2008;Wang et al. 2008). The catalytic cores have a common compact structure consisting of three subdomains containing 15-17 α-helices. The active site forms a hydrophobic pocket which contains two metal binding sites, highly conserved histidines and aspartates, and an invariant glutamine that forms critical hydrogen bonds with substrates. Co-crystallization of PDE-inhibitor complexes suggests that the invariant glutamine and a highly conserved phenylalanine (tryptophan in PDE11) are also required for inhibitor binding (Ke and Wang 2007). Thus the catalytic pocket contains structural elements responsible for both non-selective binding of inhibitors and for the binding/hydrolysis of cAMP and cGMP. In addition to these common structural elements, the catalytic core also contains specific sequences responsible for family-specific inhibitor binding and substrate affinities The PDE3 catalytic domain contains a 44 aa insert, which is unique to the PDE3 family but differs in PDE3A and PDE3B. This insert not only distinguishes PDE3 from the other PDE gene families, but also differentiates the catalytic domains of PDE3A and PDE3B subfamilies (Degerman et al. 1997).

The design of PDE inhibitors has been an active area of drug development for more than thirty years. Although methylxanthines inhibit almost all PDEs, family-specific and family-selective inhibitors, i.e., drugs that target individual PDE families with 10-100 fold greater potency than other PDE families, are available for almost all PDE families (Table 1). These family-specific inhibitors are important for both basic research and clinical applications (Francis et al. 2011;Keravis and Lugnier 2012;Maurice et al. 2014a). Many of these family-specific inhibitors (Table 1) were developed using “classical” medicinal chemical approaches, involving screening of extensive molecular libraries to discover lead compounds which were then systematically modified and derivatized to produce potent and family-selective competitive inhibitors which interfered with binding/hydrolysis of cyclic nucleotide substrates at the active site. Currently, a structure-based design approach to drug development utilizes typographical information from crystal structures of PDE catalytic regions and inhibitor/PDE co-crystal complexes together with virtual screening and computational modeling and design. The structure-based design approach was recently used to develop potent and family-selective PDE4 (Burgin et al. 2010;Card et al. 2005), PDE9A (Claffey et al. 2012;Meng et al. 2012), and PDE10A (Ho et al. 2012;Malamas et al. 2011) inhibitors.

Regulatory region

As depicted in Fig. 2, N-terminal regulatory regions of PDE molecules are highly divergent. They define specific properties and characteristics of individual PDE isoforms, and contain structural determinants and specific amino acid sequences that allow different PDEs to be incorporated into compartmentalized signalosomes and targeted to specific intracellular locations, to respond selectively to specific regulatory signals, and to interact and communicate with different signaling molecules. Many PDEs exist as asymmetric homo- or heterodimers; N-terminal regulatory regions include dimerization domains and autoinhibitory modules (e.g., in PDEs 1, 4, and 5). As also indicated in Fig.2, they contain sites and domains that are subject to different types of modification (e.g., sites in PDEs 1, 3, 4, 5, 10, and 11 for phosphorylation by various protein kinases), or sites that interact with allosteric ligands (e.g., cGMP binding sites in GAF domains), specific effectors (e.g., Ca2+/calmodulin), protein partners [e.g., Raf1 (Fig. 3)] or molecular scaffolds (e.g., A kinase anchoring proteins (AKAPs) (Fig. 4), β-arrestin and caveolin), and thereby regulate catalytic activity, protein-protein interactions, and/or subcellular compartmentation and localization (Table 2). N-terminal regulatory regions of some PDE families contain unique subdomains, e.g., GAF, UCR, PAS, NHR, Ca2+/calmodulin binding domains (Fig. 2), which confer specialized regulatory properties to these PDE families,

Figure 3. PDE8A regulates Raf-1 and Erk signaling network.

Raf is a key activator of Erk signaling pathways. AKAP-LBC positions PKA for a preferred role in cAMP/PKA-induced phosphorylation of S259-Raf1, which inhibits its kinase activity, and consequently, inhibits activation of erk and erk-signaling. Raf-1 promotes assembly of the PDE8A/Raf-1 signalosome, leading to PKA-induced phosphorylation and activation of PDE8A, thus allowing PDE8A to reduce cAMP and thus limit PKA-mediated phosphorylation of Raf-1 at S259 and limit inhibition of Erk signaling.

Figure 4. AKAPs: Scaffolds for PDE-containing signalosomes.

The subcellular localization of different AKAPs is shown. Compartmentalization of cyclic nucleotide signaling pathways involves the specific subcellular targeting/tethering of scaffolding molecules (i.e. AKAPs), effectors of cyclic nucleotides (phosphodiesterases, protein kinase A, protein kinase G, protein phosphatses, and substrates), and other signaling molecules.

Regulatory region: GAF domains

The regulatory regions (schematically represented in Fig. 2), of five PDE families (PDEs 2, 5, 6, 10, and 11) contain two homologous GAF domains (domains found in certain proteins, i.e., cGMP binding PDEs, Anabena adenylyl cyclase, and Fh1A, an Escherichia coli transcriptional regulator) which are arranged in tandem in their regulatory regions. GAF domains (150-200 amino acids) function as ligand-binding domains or facilitators of protein-protein interactions. (Gross-Langenhoff et al. 2006;Heikaus et al. 2008;Wu et al. 2004). Crystal structure of the PDE2A regulatory region indicated that its GAF-A domain is involved in dimerization and GAF-B domain in cGMP-binding (Martinez et al. 2002). Cyclic nucleotide binding to GAF domains in PDEs2, 5, 6 allosterically activates their catalytic activities (Francis et al. 2002;Wu et al. 2004;Zhang et al. 2008b). Phosphorylation of PDE5 is preferentially mediated by PKG and is facilitated by allosteric binding of cGMP or PDE5 inhibitors to the GAF-A domain (Bessay et al. 2007). GAF domain structures of PDE2, -5, -6, -10, -11, reveal differences in dimerization properties and highlight the structural diversity of GAF domain-containing PDEs. (Heikaus et al. 2009;Pandit et al. 2009).

Regulatory region: UCR and PAS domains

The N-terminal portion of PDE4 isoforms contain upstream conserved regions (UCR1 and UCR2) (Fig. 2). In fact, PDE4 splice variants are categorized into three groups based on the presence or absence of UCR domains, with “long forms” containing both UCR1 and UCR2, “short forms”, UCR2, and “super short forms”, truncated UCR2 domains. Variants lacking both UCR1 and UCR2 are catalytically inactive (Houslay et al. 2005). Studies suggest that UCR1 and UCR2 form a interacting module that regulates PDE4 dimerization and catalytic activity, and also allow interactions with heterologous regulatory partners (Beard et al. 2000;Francis et al. 2011;Richter and Conti 2002). UCR2 exhibits an autoinhibitory role; PKA-induced phosphorylation of a serine residue in UCR1 relieves the autoinhibition and activates long form PDE4 splice variants by presumably disrupting the interaction of UCR2 with the catalytic site. The presence of UCR1 also specifies the response to ERK-induced phosphorylation of the C-terminal end of the catalytic domain, with ERK-phosphorylation activating PDE4 short forms and inhibiting long forms (which contain UCR1) (Bender and Beavo 2006;MacKenzie et al. 2000). These differential phosphorylations allow feedback loops between PKA and ERK signaling. ERK-induced inhibition of PDE4 activity leads to localized increases in cAMP, leading to activation of PKA. Activated PKA then phosphorylates/activates long form PDE4, relieving it from ERK-mediated inhibition. Thus, UCR1 and UCR2 play a key role in defining the functional response of PDE4D long and short isoenzymes to PKA and ERK phosphorylation (MacKenzie et al. 2000;Oki et al. 2000).

Crystallographic studies have identified three distinct conformations of PDE4 homodimers (Burgin et al. 2010;Francis et al. 2011). In one conformation, a UCR2-based α-helix from one monomer folds over the catalytic site to inhibit cAMP hydrolysis by the second monomer, and interactions between UCR2 and certain PDE4 inhibitors, e.g. RS25344, increased inhibitor affinity at the active site. Because these allosteric interactions between UCR2 and the catalytic site are unique for different PDE4 subfamilies, there is interest in designing subfamily selective inhibitors or allosteric modulators. One such PDE4D-selective allosteric inhibitor, D159687, was developed and shown to have potent anti-inflammatory effects in cell- and animal-models, and also to improve memory and cognition in mice, with minimal emesis side effects, suggesting that this type of PDE4D selective inhibitor could have utility in inflammatory conditions such as COPD, asthma, psoriasis and arthritis, as well as improve cognitive function and memory and treat depression (Burgin et al. 2010;Gurney et al. 2011)). In another PDE4 conformation, a C-terminal regulatory helix (Control Region 3 [CR3]) folds across the active site, limits access to substrate, and also modulates inhibitor interactions at the active site. Since a single amino acid polymorphism in CR3 distinguishes PDE4B from PDE4D, analysis of co-crystal structures of PDE4B and PDE4D with inhibitors has allowed structure-based design of selective PDE4B inhibitors (Fox et al. 2014).

The N-terminal regulatory region of PDE8 isoforms contain Per-Arnt-Sim domains (PAS domains) [an acronym for three eukaryotic proteins Period clock protein, Aryl hydrocarbon receptor nuclear tranlocator (Arnt), and Single minded protein that contain these sequences] (Wang et al. 2001;Wu and Wang 2004). Association of PDE8A1 with IκB proteins, through interaction between the PAS domain of PDE8A1 and the ankyrin repeat domains of IκB proteins, activates PDE8A1 (Wu and Wang 2004).

Regulatory region: Calcium/calmodulin binding domain

PDE1 isoforms (PDE1A, PDE1B, PDE1C), which are widely expressed in mammalian tissues, are the only mammalian PDEs that are regulated by allosteric interactions with the calcium calmodulin complex (Ca²⁺-CaM) (Fig. 2). PDE1 isoforms have similar structural arrangement consisting of two N-terminal CaM-binding domains which are separated from the catalytic site by an autoinhibitory domain. Binding of Ca2+ to free calmodulin (CaM) activates the Ca²⁺-CaM complex which associates with PDE1 isoforms, relieves autoinhibitory constraints, and activates PDE1. Binding of Ca²⁺-CaM to PDE1 isoforms increases Vmax up to 10-fold, with little effect on Km for substrate (Bender and Beavo 2006). Phosphorylation of PDE1 isozymes by PKA or Ca²⁺-CaM protein kinase results in a decrease in the affinity of isozymes for Ca²⁺-CaM (Hashimoto et al. 1989). PDE1 activity is determined by intracellular Ca²⁺ levels (Francis et al. 2010).The PDE1 family provides an important mechanism for cross talk between calcium and cyclic nucleotide signaling pathways (Bender and Beavo 2006).

Regulatory region:N-terminal hydrophobic domains (NHR1 and NHR2)

As indicated in Fig. 2, the N-terminal portions of PDE3A and PDE3B include two domains, NHR1 (amino acids 1-300) which contains a large hydrophobic domain with 5-6 predicted transmembrane helices, and NHR2 (aa 300-500), with a smaller hydrophobic region of ~50 aa (Conti and Beavo 2007;Degerman et al. 1997;Hambleton et al. 2005;Taira et al. 1993). The amino acid sequences of both NHR1 and NHR2 differ considerably between PDE3A and PDE3B. Downstream of NHR1 are consensus phosphorylation sites for cAMP-dependent protein kinase (PKA), protein kinase B (PKB), and PKC (Hambelton et al. 2005;Hunter et al. 2009;Manganiello et al. 2007). Subcellular fractionation of Flag-tagged WT and N-terminal truncations of PDE3A and PDE3B and immunofluorescence localization of PDE3B in 3T3-L1 adipocytes suggested that structural determinants responsible for membrane association are localized in NHR1 domain while NHR2 seemed to be important for efficient membrane association and targeting (Shakur et al. 2000). There are three different variants of PDE3A (PDE3A1-3) which are products of the single PDE3A gene (Hambleton et al. 2005;Manganiello et al. 2007). The three PDE3A isoforms have identical sequences except for N-terminal sequence deletions of different lengths. These N-terminal differences predict differential phosphorylation by PKA, PKB and PKC, and affect their intracellular distribution. PDE3A1, which contains NHR1 and NHR2, is found exclusively in cardiac membranes, PDE3A2, which contains part of NHR1 and NHR2, is found in both cardiac cytosol and membrane fractions, and PDE3A3, which lacks NHR1 and NHR2, is primarily cytosolic, All three PDE3A isoforms have very similar catalytic activities and sensitivities to PDE3 inhibitors (Hambleton et al. 2005).

PDEs and compartmentalization of cyclic nucleotide signaling: the “signalosome” hypothesis

Intracellular cyclic nucleotides and their signaling pathways are physically and functionally compartmentalized. Various imaging approaches, using FRET and cyclic nucleotide biosensors, have confirmed that, in these temporally-regulated and spatially-constrained microdomains, individual PDEs or subsets of PDEs, via their incorporation/recruitment into distinct and differentially-localized signalosomes, play an important role in generation, regulation, turnover, and diffusion of cyclic nucleotide gradients, and in the transduction and modulation of their signals and downstream effects (Francis et al. 2011;Houslay et al. 2007;Keravis and Lugnier 2010;Kritzer et al. 2012;Maurice et al. 2014a). The signalosome concept has been a very exciting and important development in the PDE field, because it has linked the large number of PDE isoforms to compartmentalized regulation of specific cyclic nucleotide signaling pathways and downstream biological processes, and thereby dramatically delineated specific functional roles for many individual PDE isoforms. Table 2 lists a number of PDE-containing signalosomes, their interacting partners, and functions subscribed to these signalosomes. The incorporation of different types of PDEs into specific signalosomes (i.e., PDEs from different PDE families and subfamilies or different splice, transcriptional or translational variants from the same family), with their distinct intrinsic characteristics and regulatory properties, contributes to the specificity and diversity of compartmentalized cyclic nucleotide signaling and allows integration and crosstalk between cyclic nucleotide signals and other signaling networks and systems (Dodge et al. 2001;Dodge-Kafka et al. 2010;Houslay et al. 2007;Kritzer et al. 2012;Stangherlin et al. 2011).

As an example of the latter, Raf-MEK-ERK mitogen-activated protein kinase cascades are key signaling pathways involved in the regulation of normal cell proliferation, survival and differentiation. Increased expression of this pathway has been associated with cancer. cAMP negatively regulates Extracellular-signal regulated protein kinase (ERK) or mitogen-activated protein kinase (MAPK) through Raf-dependent signaling, with Raf-1 being the initiating kinase in the kinase cascade that activates MAPK. Phosphorylation of Raf-1 on S259 by PKA inhibits Raf-1 kinase activity and its downstream signaling, resulting in inhibition of ERK. Binding of AKAP with PKA and PDE8A forms a signaling complex/module or signalosome in which PDE8A interacts with Raf-1 and protects it from inhibitory phosphorylation by PKA (Brown et al. 2013). As seen in Fig. 3, cAMP activates PKA, phosphorylates and activates PDE8A1, thus reducing cAMP and thereby downregulating PKA as a feed back inhibitor of ERK.

All PDE4 family isoforms (PDE4A, PDE4B, PDE4C, PDE4D) contain highly conserved catalytic domains. A striking characteristic of the PDE4 family, however, are the many unique PDE4 N-terminal splice variants which encode proteins with different lengths of N-terminal truncations. Due to the presence of “so-called” upstream conserved regions 1 and 2 (UCR1 and UCR2) in the PDE4 N-terminal regions (Fig. 2), PDE4 isoforms are ideally designed for intracellular targeting to different subcellular compartments, to molecular scaffolds including AKAPS and β-arrestin, and to many interacting partners, includung RACKS (receptor for activated protein kinase C 1), myomegalin, immunophilin XAP2, EPACs, spectrin, disrupted in schizophrenia (DISC1), and others (Table-2). In these signalosomes, PDE4 isoforms determine local cAMP gradients, and, together with sequestered PKA and EPAC partners, gate activation of spatially segregated, localized cyclic nucleotide signaling pathways. The classical molecular basis for compartmentalization of cAMP signaling involves anchoring of PKA isoforms at specific intracellular sites via AKAP scaffolding proteins (Fig. 4). These proteins serve multiple functions. They target PKA to different subcellular locations and position PKA molecules in close proximity to their substrates. Importantly AKAPs also organize formation of localized, unique signalosomes consisting of different proportions of cyclases, kinases, kinase substrates, phosphatases, PDE (especially PDE4) isoforms, and other proteins, including EPACs. These signalosomes are self-modulatory, with cAMP-induced activation of PKA resulting in phosphorylation of PKA substrates/targets and also phosphorylation/activation of PDE4 isoforms, which leads to decreased cAMP in the immediate vicinity of the signalosome, thus limiting the extent of PKA activation and returning PKA to its basal state (Fig. 4, Table 2) (Dodge-Kafka et al. 2010;Kritzer et al. 2012;Manganiello et al. 2007).

The ability of different PDE4 isozymes modulate intracellular cAMP gradients and compartmentalized activation of substrates and biological responses requires that these PDE-containing signalosomes each can exhibit selective anchoring in specific cellular microdomains, where they regulate discrete, largely non-overlapping cAMP pools. Recent studies strongly suggest, for example, that specific PDEs are functionally coupled with discrete pools of adenylyl cyclase/G-protein-coupled receptors that respond to specific hormones. For example, in rodent cardiomyocytes, stimulation of β1AR induces strong inotropic and lusitropic responses by increasing a discrete pool of cAMP, which is regulated by PDE4 isoforms and which activates and phosphorylates key regulators of the excitation/contraction cycle, whereas β2AR-mediated signals are regulated by multiple PDEs, including PDE3, PDE3 and PDE4 (Nikolaev et al. 2006). Although PDE2 accounts for a minor fraction of PDE in neonatal rat cardiomyocytes, it seems to be specifically involved in the hydrolysis of cAMP induced by BAR stimulation. Since PDE2 is stimulated by cGMP, it mediates inhibitory effects of cGMP on BAR-induced increases in cAMP and contractility (Stangherlin et al. 2011).

In regulating myocardial contractility in rodents, different PDE3 and PDE4 isoforms are associated with AKAP-based signalosomes thought to modulate effects of cAMP on L-type and ryanodine-sensitive Ca2+ channels (RyR2) as well as SERCA2 (Beca et al. 2013;Kerfant et al. 2007;Lehnart et al. 2005;Leroy et al. 2011). PDE3 inhibitors, such as milrinone, induce myocardial contractility via cAMP/PKA-catalyzed phosphorylation of phospholamban (PLB) and consequent activation of sarcoplasmic reticulum (SR) Ca²⁺ ATPase (SERCA2) and Ca²+ uptake into the SR (Movsesian et al. 2011;Tsai and Kass 2009). Studies with PDE3A and PDE3B-KO mice indicated that inotropic responses to PDE3 inhibitors were preserved in PDE3B-KO, but not in PDE3A-KO, mice (Sun et al. 2007), and that PDE3A, not PDE3B, is the milrinone-sensitive PDE3 isoform that regulates basal Ca++ transients and contractility (Beca et al. 2013). Targeted to the SR, an AKAP18δ-based signalosome, containing PDE3A, SERCA2, PKA, PP2A, phospholamban and other proteins, was reported to regulate cAMP/PKA phosphorylation of phospholamban and subsequent activation of SERCA2 and uptake of Ca²⁺ during diastole (Beca et al. 2013;Lygren et al. 2007); a similar AKAP-based signalosome, containing human PDE3A appears to regulate SERCA2 and Ca++ uptake into human SR (Ahmad F et al unpublished). As a component of an AKAP16/AKAP18-organized signalosome, PDE4B is tethered to L-type Ca⁺⁺ channels (LTCC) and colocalized with the LTCC along T-tubules where it seems to be involved in the regulation of cAMP/PKA-induced Ca⁺⁺ uptake, and protection against ventricular arrythmias (Lehnart et al. 2005;Leroy et al. 2011). An AKAP6 (mAKAP)-based signalosome tethered/targeted PDE4D to RyR2 channels where it modulated PKA-induced phosphorylation of RyR2 and thus regulated Ca⁺⁺-induced Ca⁺⁺ release from the SR. By limiting β-adrenoreceptor (β-AR)-induced phosphorylation of RyR2, PDE4D may protect against proposed pro-arrythmogenic effects of hyper-phosphorylated RyR2 (Lehnart et al. 2005). PDE4D-knockout (KO) mice exhibit an age-dependent progressive cardiomyopathy, with dysregulation of Ca²⁺ transients and increased incidence of arrhythmias, associated with cAMP/PKA-induced hyperphosphorylation of the RyR2, which was likely due to the absence of PDE4D3 from a signalosome containing a mAKAP/PKA/PDE4D/RyR2 complex (Lehnart et al. 2005). Thus, as components of different AKAP-based signalosomes, PDE4B, PDE4D and PDE3A each regulate different phases of the calcium-mediated excitation–contraction coupling cycle and offer protection against arrhythmias (Bers 2002). In addition, following βAR-agonist stimulation, PDE4D5 preferentially associates with β-arrestin and the complex is recruited to the βAR in the vicinity of plasma membranes (PM), where it is involved in the regulation of B-receptor signaling and trafficking (Houslay et al. 2007;Kritzer et al. 2012).

In cardiomyocytes, a mAKAP-organized signalosome that is located at the nuclear envelope also modulates cAMP-mediated hypertrophic responses. mAKAP provides the scaffold for signalosomes that contain PDE4D3, PKA, Epac1, PP2A, adenylyl cyclase, and an ERK signaling module (Bauman et al. 2007;Dhanasekaran et al. 2007;Houslay et al. 2007;Kritzer et al. 2012). This mAKAP-based signalosome, in which PP2A-induces dephosphorylation/ inactivation of PDE4D3 (Dodge-Kafka et al. 2010) and PKA-induces phosphorylation/ activation (Dodge et al. 2001), coordinates the integration of PKA- and Epac1/ERK- based signaling complexes (Dodge-Kafka et al. 2005). ERK regulates a cAMP feedback loop. Anchored PKA in the signaling complex stimulates PDE4D3 to reduce cAMP concentrations, whereas the ERK module suppresses PDE4D3 (Dodge-Kafka et al. 2005;Kritzer et al. 2012).

The subcellular localization of different PDEs and their incorporation into signalosomes is a common theme in PDE biology (Table-2). In human platelets, PDE5, PKG1B, and inositol-1,4,5-triphosphate receptor type-1 (IP3(3)R1) are integral components of a signalosome that regulates effects of cGMP on IP(3)R1-mediated Ca²⁺ release (Wilson et al. 2008). In platelets. PDE3A is constitutively associated with leptin receptor, and modulates effects of leptin on platelet activation (Elbatarny and Maurice 2005). In adipocytes, spatial segregation of PDE3B-containing signalosomes and the lipolytic machinery allows activated PDE3B to selectively mediate the inhibitory effects of insulin on cAMP/PKA signaling, phosphorylation of hormone sensitive lipase (HSL), and lipolysis; this compartmentalization most likely explains the observed inhibition of insulin's antilipolytic effects by PDE3 inhibitors but not by PDE4 inhibitors (Elks and Manganiello 1985;Zmuda-Trzebiatowska et al. 2006). In cells that regulate energy metabolism, including adipocytes, hepatocytes, and pancreatic β-cells, PDE3s are rapidly activated by in response to agents that increase cAMP, and also to insulin, IGF-1, leptin, and IL-4 (Degerman et al. 2004). In adipocytes, activation of PDE3B by insulin or a β3-receptor agonist, CL316,243 (CL), involves the reversible assembly of distinct signalosomes that contain phosphorylated PDE3B and different sets of signaling effectors thought to be involved in activation of PDE3B by insulin or cAMP/PKA (Ahmad et al. 2007;Ahmad et al. 2009;Nilsson et al. 2006). In adipocytes and FDCP2 myeloid cells, insulin and IGF-1 were found to activate PDE3B via wortmannin-sensitive, PI3K-dependent signals that activated PKB, which phosphorylated/activated PDE3B (Degerman et al. 2004). This paradigm, i.e., growth factors/cytokine-induced phosohorylation/activation of PDE3 via PI3-K/PKB activation, provides a mechanism for counter-regulatory effects of insulin, IGF-1, and leptin in inhibiting certain cAMP-mediated processes in different cellular environments, including lipolysis, glycogenolysis, pancreatic insulin secretion, and feeding behaviour (Tables 1, 2) (Manganiello et al. 2007). Activation of PDE3A by leptin in platelets and by insulin/IGF-1 in Xenopus oocytes are important in reducing cAMP and mediating the stimulatory effects of insulin and IGF-1 on oocyte maturation and of leptin on platelet activation (Tables 1, 2). PDE3 inhibitors can attenuate or block these effects (Manganiello et al. 2007).

The signalosome concept has not only significantly increased understanding of the compartmentalization of cyclic nucleotide signaling and functional roles of individual PDEs, but has also increased the number of potential therapeutic targets. Just as disruption/dysregulation of compartmentalized signaling pathways and signalosomes may be associated with disease states, targeting these compartmentalized PDE-containing signalosomes could occasion development of effective, more selective, and safer therapeutics that might have widespread clinical utility.

PDE Functions: Implications regarding pathophysiology of disease and therapeutics

Historically, in cells containing multiple PDEs, family-specific PDE inhibitors (Table 1) were initially utilized to pharmacologically define roles of specific PDEs in spatial and/or functional compartmentalization of specific cyclic nucleotide signaling pathways and discrete cellular functions (Tables 1, 2). For example, PDE3 inhibitors, not PDE4 inhibitors, blocked the anti-lipolytic effects of insulin, which were also blocked in adipocytes from PDE3B KO mice (Choi et al. 2006;Elks & Manganiello 1985). In cultured renal mesangial cells, effects of PDE3 and PDE4 inhibitors indicated that PDE3 and PDE4 each selectively regulated functionally distinct and compartmentalized cAMP pools that controlled cell growth and generation of reactive oxygen species, respectively (Chini et al. 1997).

The utility of family-selective inhibitors is limited, however, because current family-selective inhibitors inhibit all family members in all cells in which they are expressed, and do not distinguish between subfamily members (e.g., PDE1 inhibitors do not distinguish between PDE1A, PDE1B, PDE1C). More pertinent information has been gathered in studies with single PDE gene KO mice, in studies in which individual PDEs were targeted for knockdown with small interfering RNAs (siRNAs), or studies demonstrating association of human diseases with specific PDE mutations. These types of studies, especially with KO mice, do indicate that the ability of specific PDEs to regulate discrete cAMP/cGMP signaling pathways and actions is genetically determined and nonredundant.

For example, PDE3 inhibitors, not PDE4 or 5 inhibitors, inhibit oocyte maturation in cultured mammalian oocytes, and administration of PDE3 inhibitors blocked pregnancy in intact rodents (Masciarelli et al. 2004;Norris et al. 2009;Wiersma et al. 1998). Female PDE3A KO, not female PDE3B, KO mice are sterile, most likely due to inhibitory effects of increased oocyte cAMP on meiotic progression and maturation of oocytes and, consequently, their competency for fertilization (Masciarelli et al. 2004). These studies offered proof of principle that oocyte PDE3A could be a contraceptive therapeutic target.

Cultured aortic vascular smooth muscle cells (VSMC) from PDE3A KO mice, but not PDE3B KO or WT mice, exhibited a markedly reduced growth response to serum or the mitogen, PDGF, due to dysregulation of cAMP/PKA- and MAP- signaling pathways and alterations in critical cell cycle regulatory proteins, including p21 and p53. siRNA-induced knockdown of PDE3A in PDE3B KO VSMC inhibited the mitogenic response to PDGF in the PDE3B VSMC (Begum et al. 2011). Similarly, knockdown of PDE1C in VSMC from pulmonary arteries from patients with pulmonary hypertension increased cAMP content and inhibited cell growth (Murray et al. 2007). These knockdown studies suggested that targeting PDE3A and/or PDE1C might reduce the pathological vascular remodeling observed in post-stenting and post-angioplasty restenosis, atherosclerosis, pulmonary hypertension, and other cardiovascular disorders (Begum et al. 2011;Chan and Yan 2011).

Compared with their wild-type (WT) counterparts, PDE3B KO mice and PDE4B KO mice were leaner, with lower fat pad weights and smaller adipocytes, and with reduced inflammation in white adipose tissue (WAT) (Choi et al. 2006;Zhang et al. 2009). PDE3B-KO WAT exhibited phenotypic characteristics of brown adipose tissue (Guirguis et al. 2013). These data suggest a role for both PDE3 and PDE4 in regulation of energy homeostasis, and are of particular interest, because some of the beneficial metabolic effects of polyphenol resveratrol, at least in rodents, may be mediated by inhibition of PDE3 and PDE4 (Park et al. 2012).

The PDE4 family (PDE4A, PDE4B, PDE4C, PDE4D) is the largest gene family in the mammalian PDE superfamily. Different PDE4 splicing variants (>25) are targeted to discrete subcellular compartments, where hormones activate them in a temporally- and spatially-dependent manner (Conti et al. 2014). Many studies, especially those with PDE4 KO mice, have demonstrated that individual PDE4 subtypes play complementary, not redundant, roles in the control of cell functions (Ariga et al. 2004;Beca et al. 2011;Hansen et al. 2000;Jin et al. 2007;Jin and Conti 2002;Peter et al. 2007) For example, PDE4 isoforms are relatively highly expressed in immune/inflammatory cells, and PDE4 inhibitors are potent anti-inflammatory agents in many model systems, including asthma, arthritis, and experimental autoimmune encephalomyelitis (EAE, a T-cell driven animal model of multiple sclerosis) (Torphy 1998). During T-cell activation, PDE4 isoforms, especially PDE4B, are incorporated, in a phosphatidyl3-kinase-dependent manner, into a PKB/β-arrestin signalosome and recruited into lipid rafts, where, by modulating compartmentalized cAMP levels, they regulate the cytokine environment important for T-cell activation/differentiation (Bjorgo et al. 2011). In monocytes/macrophages, upregulation of PDE4B, most likely the PDE4B2 splice variant, is required to relieve cAMP inhibition of tumor necrosis factor (TNF) production and toll like receptor signaling (Jin et al. 2007). Studies with PDE4B and PDE4D KO mice have dissected the specific contributions of PDE4B and PDE4D to regulation of the immune/inflammatory response (Houslay et al. 2005;Jin et al. 2007). With respect to allergic airway reactivity, PDE4B deletion inhibited neutrophil chemotaxis and recruitment of inflammatory cells to the lung, inhibited production of Th2-type cytokines and TNFα, and inhibited proliferation of T-cells in lymph nodes, but did not reduce methacholine–induced smooth muscle contraction. PDE4D ablation reduced the airway smooth muscle contractile response, and recruitment of inflammatory cells to the lung and neutrophil chemotaxis, without altering other aspects of the inflammatory response (Houslay et al. 2005;Jin et al. 2007). These types of studies clearly supported the development of pan-PDE4 inhibitors, and PDE4B-selective inhibitors, as anti-inflammatory agents, and a pan-PDE4 inhibitor, roflumilast, has been approved for treatment of chronic obstructive pulmonary disease (Rabe 2011). Some uncertainties remain, however. Although expression of PDE4B2 is increased in the brainstem and spinal cord in mice with EAE, and the PDE4 inhibitor, rolipram, reduced clinical signs of the disease, PDE4B KO mice unexpectedly developed EAE faster than WT mice (Sanabra et al. 2013). Furthermore, a small clinical study of effects of rolipram in patients with multiple sclerosis (MS) was prematurely terminated because of side effects (nausea, emesis) and an unexpected increase in brain inflammation in the MS patients, even though rolipram administration reduced inflammation markers in circulating lymphocytes (Bielekova et al. 2009). The reason for this striking discrepancy and failure of PDE4 inhibition to reduce brain inflammation in human MS is not understood.

In preclinical studies, use of both mouse genetic models as well as PDE inhibitors has contributed to our understanding of how the different PDE families contribute uniquely to overall brain function and how their inhibition might correct cognitive or behavioural deficits and confer neuroprotective benefits (Kelly and Brandon 2009;Richter et al. 2013). Targeted inactivation of PDE1B alters locomotion activity and DARPP-32 phosphorylation in response to dopamine agonists and display impaired spatial learning (Reed et al. 2002). In PDE1c knockout mice olfactory sensory neurons (OSNs) show reduced sensitivity and attenuated adaptation to repeated stimulation, (Cygnar and Zhao 2009). PDE4 inhibitors, which were initially observed to exhibit antidepressant actions in humans (Wachtel 1983), improve cognition and long-term memory in healthy rodents as well as rodent models of human genetic diseases (Richter et al. 2013). Multiple studies, especially those with PDE4 KO mice, have pointed to inhibition of PDE4D and not other PDE4 isoforms as most likely responsible for the emetic effects (Robichaud et al. 2002) as well as the antidepressant and pro-cognitive and memory-enhancing responses to pan-PDE4 inhibitors. In intact rodents, newly developed PDE4D selective inhibitors produce pro-cognitive and memory-enhancing effects similar to rolipram, a “classical” pan-PDE4 inhibitor. In addition, PDE4D KO mice exhibit antidepressive behaviour, and improved memory and cognitive function, but showed reduced responses to rolipram and other PDE4 inhibitors and antidepressant drugs. Studies in PDE4B KO preclinical models implicate this enzyme in anxiety-related behaviour. (Burgin et al. 2010;Houslay et al. 2005;Kelly and Brandon 2009;Richter et al. 2013). Phosphodiesterase 10A (PDE10A), which hydrolyzes both cAMP and cGMP with high affinity, is relatively highly expressed in dopaminoreceptive medium spiny neurons of the striatum, and regulates cyclic nucleotide- activation of these neurons (Siuciak et al. 2006). In rodent model systems, PDE10A KO mice exhibited increased social interactions, suggesting that cAMP signaling in striatal medium spiny neurons might be involved in regulating social behavior (Sano et al. 2008). Administration of family-selective PDE10A inhibitors have also been reported to produce antipsychotic and pro-cognitive effects, and to increase sociality in rodents (Ho et al. 2012;Kelly and Brandon 2009;Malamas et al. 2011). In addition to their anti-inflammatory actions and positive effects on cognition and memory-enhancement, PDE4 inhibitors have been reported to exhibit neuroprotective or neuroregenerative effects in rodents following ischemia-induced stroke, administration of B-amyloid peptides or spinal cord injury (Kelly and Brandon 2009;Richter et al. 2013). Also, in a transgenic R6/2 mouse model of Huntington's disease, inhibition of PDE4 or PDE10A alleviated motor and behavioural deficits (DeMarch et al. 2008;Giampa et al. 2010). Although these studies have increased interest in developing PDE inhibitors as therapeutics for basal ganglia disorders, it should be noted that chronic administration of high concentrations of rolipram produced harmful effects related to cognition and memory in some animal model systems (Richter et al. 2013)

In some instances, a specific PDE serves as a crucial effector system and regulates a unique cellular function. PDE6 is highly expressed in photoreceptors and pineal gland (Morin et al. 2001;Ridge et al. 2003). In retinal rods and cones, light induced activation of photoreceptor PDE6 results in hydrolysis of cGMP, alterations in Ca++ current, and initiation of visual signal transduction (Burns and Arshavsky 2005). Mutations in human PDE6α, β, or γ subunits are associated with hereditary eye diseases, including night blindness and retinitis pigmentosa (Gal et al. 1994;Huang et al. 1995). Because chronic elevations of cGMP in photoreceptors leads to cell death and progressive retinal degeneration (Davis et al. 2008), certain PDE5 inhibitors (for example, sildenafil and vardenafil), which can inhibit PDE6, are contraindicated in patients with retinal degeneration (Azzouni and Abu 2011). Mutations in AILP1 (the gene encoding aryl hydrocarbon receptor-interacting protein-like 1), a chaperone of PDE6A, allow proteolytic destruction of PDE6A and are associated with Leber congenital amaurosis type 4, a severe form of childhood blindness (Ramamurthy et al. 2004).

PDE4D mutations have been reported in patients with acrodysostosis (a rare autosomal-dominant condition characterized by facial dysostosis, severe brachydactyly, and short stature) (Lee et al. 2012). Similarly, human PDE4D haplotypes and single-nucleotide polymorphisms (SNPs) have been correlated with ischaemic stroke (Gretarsdottir et al. 2003) and with responses to short-acting bronchodilators in pediatric asthma (Labuda et al. 2011), whereas PDE4B SNPs and decreased expression of PDE4B are associated with schizophrenia (Fatemi et al. 2008). In addition, DISC1 (disrupted in schizophrenia homolog 1), a known risk factor for schizophrenia, and PDE4B interact physically in cells to influence the catalytic activity of PDE4B (Table 2) (Clapcote et al. 2007;Millar et al. 2007). Together, these observations support the notion that PDE4B may be involved in schizophrenia.

PDEs and Cancer

Evidence suggests that impairment in the generation of cyclic nucleotides and/or overexpression of PDEs are implicated in various cancer pathologies. Inhibition of selected isoforms of PDEs could provide antitumor therapy by regulating the intracellular levels of compartmentalized cAMP and cGMP and thus inhibit cell growth and migration, and/or induce apoptosis in target tumor cells (Table 1) (Savai et al. 2010). For example, in Chronic lymphocytic leukemia (CLL) and some malignant carcinoma cells, cAMP levels were significantly decreased due to overexpression of PDE7 (Zhang et al. 2008a) or PDE4 (Marko et al. 1998), respectively. In fact, in CLL cells, overexpression of PDE7 was associated with poor prognosis, and selective PDE7 inhibitors increased cAMP-signaling, leading to increased apoptosis and inhibition of proliferation of CLL cells (Zhang et al. 2008a;Zhang et al. 2011). In 11 different types of primary human tumor samples, expression of PDE4D was up-regulated compared with corresponding nontransformed tissues. Targeting of endogenous PDE4D with shRNAs or a specific PDE4 inhibitor caused apoptosis and growth inhibition in multiple types of cancer cells, but not normal cells, and re-expression of PDE4D increased tumor cell growth (Lin et al. 2013). Thus, these studies suggest that PDE7 and PDE4D might serve as biomarkers/prognostic indicators as well as therapeutic targets in certain tumors. Similarly, elevated cGMP levels, via activation of cGMP/PKG signaling, inhibited tumor cell growth and migration (Deguchi et al. 2004;Pitari et al. 2001;Shailubhai et al. 2000). Although expression of PDE6, which specifically hydrolyzes cGMP, was recently reported to be elevated in human breast cancer cell lines (Dong et al. 2013), in other human breast cancer cells, inhibition of PDE5, another cGMP-specific PDE, by siRNA-induced PDE5 knockdown or inhibition with the drug, sundilac, promoted apoptosis (Tinsley et al. 2011).

On the other hand, although PDE inhibitors can inhibit growth of certain tumors, cAMP can also promote hyperplastic changes and growth of tumors (Almeida and Stratakis 2011). PDE8 regulates cAMP-mediated steroidogenesis in adrenocortical tissues and Leydig testicular cells, and plays an important role in regulating the sensitivity to luteinizing hormone for testosterone production (Vasta et al. 2006). PDE8B KO mice exhibit increased expression of several steroidogenic proteins and elevated urinary corticosterone levels (Tsai et al. 2011). PDE8 may be a potential therapeutic target for the treatment of several different adrenal diseases (Tsai e al. 2011). Mutations in human PDE8B are associated with development of micronodular adrenal hyperplasia, adrenocortical tumours, and clinical signs of Cushing's disease (Rothenbuhler et al. 2012). One mutation in human PDE8B (His350Pro), from a patient with severe adrenal hyperplasia, impaired PDE8B catalytic activity, and its expression in Hela cells resulted in increased cAMP-signaling, suggesting a role for cAMP in development of adrenal hyprerplasia. Such PDE8B mutations may not directly cause adrenal hyperplasia, however, since PDE8B KO mice do not develop adrenal tumors (Tsai and Beavo 2011). Inactivating PDE11A gene mutations are also associated with the development of adrenal hyperplasia and Cushing syndrome (Libe et al. 2008), and with Carney complex (CNC). CNC is caused by germline mutations in the alpha regulatory subunit of PKA (PRKARIA) and is associated with endocrine tumors, including nodular adrenal hyperplasia and adrenal and testicular tumors (Levy et al. 2011;Libe et al. 2011). PDE11A mutations may also play a role in susceptibility to prostate cancer (Faucz et al. 2011) and testicular germ cell tumors (TGCT) (Horvath et al. 2009).

PDEs as possible targets in Oral Cancer

Some reports have suggested that PDEs may provide new targets for the treatment of oral cancer. For example, PDE1 is a pharmacological and specific target of differentiation-inducing factor-1 (DIF-1), an antitumor agent isolated from Dictyostelium discoideum (Shimizu et al. 2004). In human oral malignant melanoma MAA cells, inhibition of PDE1 by Vinpocetin was reported inhibit proliferation (Shimizu et al. 2009). PDE2A regulates cell cycle progression in human oral malignant melanoma PMP cells by modulating cyclin A expression (Morita et al. 2013); targeting PDE2A by the PDE2 inhibitor, EHNA, or siRNA-induced knockdown inhibited PMP cell proliferation and invasion (Hiramoto et al. 2014). Two cisplatin-resistant oral squamous cell carcinoma cell lines, Sa-3R and H-1R cells, exhibited increased expression of PDE3B compared with parental cell lines (Yamano et al. 2010), and, as assessed by PDE3B immunohistochemical staining, PDE3B expression levels were significantly higher in the tumors unresponsive to cisplatin than in responsive tumors (Yamano et al. 2010). Furthermore, the antitumor growth effect of the combination of a PDE3-specific inhibitor (cilostazol) and cisplatin was also greater than with either cilostazol or cisplatin alone, with a significant increase in the number of apoptotic and cell growth-suppressive cancer cells in cisplatin-resistant Sa-R cells (Uzawa et al. 2013). In human neoplastic submandibular gland intercalated duct HSG cell lines, PDE3A and PDE3B mRNAs were expressed and a PDE3-specific inhibitor (cilostamide) inhibited proliferation of these cells (Murata et al. 2001). However, in other tumors the role of PDE3 is not clear. PDE3-specific inhibitors (cilostamide and trequinsin) did not inhibit the proliferation of human oral malignant melanoma HMG cells, and in human oral osteosarcoma HOSM-1 cells neither PDE3A or PDE3B mRNAs were expressed (Murata et al. 2002).

PDEs and bone metabolism

Bone is mineralized osseous tissue that carries out mechanical and metabolic functions in the skeleton. Maintenance of normal bone mass depends on the balance between osteoblast bone formation and osteoclast bone destruction (Boyle et al. 2003;Duong and Rodan 2001). A number of studies suggest that PDE inhibitors promote bone formation via stimulation of osteoblast differentiation/formation (Kinoshita et al. 2000;Tokuhara et al. 2010;Wakabayashi et al. 2002) and/or inhibition of bone resorption (Noh et al. 2009;Park and Yim 2007;Yao et al. 2007), and thus could possibly provide novel therapeutics for osteoporosis (Epstein 2012). cAMP and cGMP have dramatic effects on osteoblasts. Cyclic AMP derivatives or cAMP generating agents such as PTH and forskolin inhibit ERK2 activation by bFGF and PDGF-BB, and thus inhibit osteoblast proliferation through attenuation of MAP kinase pathways and through effects on BMP signaling pathways and on Cbfa1 (Chaudhary and Avioli 1998;Rodan and Noda 1991). PTH and its active N-terminal peptide, which acts through PTH receptor 1 (PTHR1), signal via cAMP/PKA and Ca2+/PKC pathways (Aghaloo et al. 2006;Datta and Abou-Samra 2009;Nakao et al. 2009). Currently, Forteo (N-terminal PTH peptide, 1-34 amino acid) is FDA-approved for anabolic treatment of osteoporosis (Canalis 2010;Hodsman et al. 2005). PTH works via PKA-dependent transactivation of cAMP-stimulated core binding factor alpha1 (Cbfa1)-regulated expression of osteoblast specific genes such as osteopontin, collagen type I, osteocalcin, and sialoproteins (Rodan & Noda 1991). cGMP has also been show to regulate differentiation of osteoblasts (Broderick et al. 2007;Hagiwara et al. 1996;Wimalawansa 2008). Prkg2 knockout mice, deficient in type II cGMP-dependent protein kinase (PKG), exhibit a dwarfism phenotype, with impaired endochondral ossification (Miyazawa et al. 2002). Studies with PDE-inhibitors which activate cAMP signaling pathways showed that PDE4-selective inhibitors, rolipram and XT-611 (Horiuchi et al. 2002;Wakabayashi et al. 2002;Yamagami et al. 2003), and the non specific PDE inhibitor, pentoxifylline (Horiuchi et al. 2004), potentiated bone formation by BMP-stimulated osteoblast differentiation.

Osteoclasts, which play a role in the development of osteoporosis, are highly motile cells that carry out bone resorption (Chellaiah et al. 2003). In osteoclasts, Rho-A, which is critical for osteoclast motility and function can be directly phosphorylated and inactivated by PKA (Chellaiah et al. 2000;Chellaiah et al. 2003;Nakamura et al. 2007). Thus, PDE inhibitors, which increase cAMP/PKA signaling, also inhibit osteoclast motility and function (Noh et al. 2009;Park and Yim 2007;Yao et al. 2007).

In osteoporosis, bone mineral density (BMD) is reduced, bone microarchitecture deteriorates, and the amount and variety of non-collagenous proteins in bone is altered (Epstein 2012). PDE inhibtors could play a potential role in treatment of osteoporosis, especially glucocorticoid-induced osteoporosis in post menopausal women (Saag 2003;Teitelbaum et al. 2011). Approximately 50% of the chronic glucocorticoid users will develop fractures due to bone loss (Lukert and Raisz 1990). In glucocorticoid-induced osteoporosis, viability and functions of osteoblasts are inhibited, which substantially differs from other forms of osteoporosis (Teitelbaum et al. 2011). Since there is a considerable cross-talk between glucocorticoid and cAMP signaling pathways (Lerner and Epstein 2006;Tiwari et al. 2005), PDE inhibitors, by acting to both promote bone formation and inhibit bone resorption might fill a critical need as therapeutics for glucocorticoid-induced osteoporosis.

PDEs and salivary gland function

Salivary glands can be classified as exocrine glands whose secretions protect and facilitate the function of all oral and upper gastrointestinal tract tissues. The vast majority of cells in these glands are epithelial and of two broad types: acinar and duct. Acinar cells secrete a primary fluid that contains ~85% of the secreted proteins and duct cells secrete remaining ~15% of salivary proteins found in saliva (Perez et al. 2010).The expression and secretion of salivary gland proteins are regulated by cAMP (Zhang and Martinez 1999). Different PDE isoforms are present in different glandular epithelial cells. For example, PDE3A mRNA was expressed in isolated rat submandibular acini cells, but not (or at much lower concentration) in duct cells, and PDE3B mRNA was not detected in acini or duct cells (Shimizu et al. 2006).

With respect to specific roles of individual PDEs in salivary gland function, isoproterenol-induced amylase release, but not constitutive amylase release, was enhanced in the presence of a PDE4-specific inhibitor (rolipram) in mouse, rat and rabbit parotid acinar cells (Satoh et al. 2009), and a PDE3-specific inhibitor (cilostamide) increased apomucin gene expression in isolated rat submandibular acini cells (Shimizu et al. 2006). The ability of PDEs to regulate the expression and secretion of salivary gland proteins suggests possible new targets for treatment of oral disease related to hyposalivation.

Clinical Applications of PDE inhibitors

The methylxanthines, theophylline, caffeine, and isobutylemethylxanthine (IBMX) inhibit almost all PDE's except PDE8 and 9, and are classified as non-selective PDE inhibitors. Before they were recognized as PDE inhibitors, however, xanthine derivatives, especially theophylline and caffeine, were used in clinical situations as bronchodilators (acute asthma) and as cardiotonic agents and diuretics (Schudt et al. 2011). Currently, in certain instances, theoplylline is used as adjunct therapy in asthma, COPD, emphysema and chronic bronchitis, and is being evaluated in corticosteroid-resistant asthma and COPD (ClinicalTrials.gov Identifier: NTC00241631) (Barnes 2013). Xanthine derivatives are used sparingly and with concern, however, because of their narrow risk/benefit profile. One xanthine deravitive, pentoxifylline (Trental: Sanofi) is used to treat intermittent claudication. Another non-selective PDE inhibitor, dipyridamole (Persantine) blocks hydrolysis of cAMP and cAMP in platelets and thus inhibits platelet activation/aggregation. It is used to prevent post-surgical thromboembolism and stroke, and it can be used for myocardial stress tests (Schaper 2005).

PDE family-selective inhibitors (Table1) were developed to replace non-specific PDE inhibitors, with the aim of improving selectivity, potency and efficacy for inhibition of the individual PDE families, for both research purposes as well as treatment of diseases with more specificity and less toxicity. These potent, family-selective PDE inhibitors have provided, and will continue to provide, important pharmacological tools for characterizing the distribution and functional role of individual PDEs in cells and tissues, under normal conditions or in pathological states.

Currently only PDE3, PDE4, and PDE5 family-selective inhibitors have received regulatory approval for clinical applications. These currently available drugs and other family-selective PDE inhibitors under development (Table 1) are in clinical trials, which can be accessed in clinicaltrials.gov website (Maurice et al. 2014a).

PDE3 inhibitors

Cilostazol, a PDE3-selective inhibitor, is widely used to treat intermittent claudication, which is a peripheral vascular disease, characterized by ischemia-induced cramping and pain in the legs (Chi et al. 2008;Liu et al. 2011). Cilostazol reduces symptomatology via cAMP-induced vasodilatation, as well as cAMP-mediated inhibition of vascular inflammation and of platelet aggregation /activation (Kambayashi and Liu 2007).

Although PDE3 inhibitors were developed to improve hemodynamic responses in patients with heart failure, in clinical trials, chronic administration of the PDE3 inhibitor milrinone increased mortality, most likely due to arrhythmias and cardiac arrest (Packer et al. 1991). Milrinone is approved, however, for acute treatment of adults awaiting heart transplants and adults with refractory, decompensated heart failure (Movsesian et al. 2011).

PDE4 inhibitors

PDE4 isoforms are relatively highly expressed in cells involved in immunoflammatory responses, and PDE4 inhibitors, by increasing cAMP-signaling, act as potent anti-inflammatory agent in a variety of cellular and animal preclinical models. There has been considerable effort to develop PDE4 inhibitors as therapeutics for important pulmonary inflammatory diseases, especially asthma, allergies rhinitis, chronic obstructive pulmonary disease (COPD), idiopathic pulmonary fibroins, etc. (Torphy 1998). COPD is a major, world-wide public health problem and leading cause of death in affected individuals who smoke. COPD is multi-systemic and clinically heterogeneous. It is characterized by progressive airway inflammation and remodeling, leading to airway obstruction and a relentless decline in expiratory airflow and lung function. COPD is also associated with cardiovascular, metabolic, and musculoskeletal co-morbidities. Recently a relativity new PDE4 inhibitor, roflumilast, has been approved for use in COPD, especially to reduce exacerbations in subsets of COPD patients with bronchitis, and as add-on therapy to patients receiving inhaled corticosteroids, or standard bronchodilators such as long-acting β2-adrenergic agonists (LABA) or anti-cholinergic drugs (Fabbri et al. 2009;Giembycz and Newton 2014;Rabe 2011).

Because PDE4 isoforms exhibit widespread tissue distribution, perhaps more than any other PDE family, inhibition of PDE4 enzymes alters cAMP-signaling in many cells. In addition to these beneficial effects, however, pan-PDE4 inhibitors also elicit dose- limiting side effects, especially nausea and emesis, primarily brought about by inhibition of PDE4 [perhaps PDE4D more than other PDE4 subfamilies (Robichaud et al. 2002) in the brain and gut (Tenor et al. 2011). These side effects have limited development and broad use of PDE4 inhibitors as therapeutics. In addidition, older PDE4D KO mice develop arrhythmias and heart failure (Lehnart et al. 2005), and inhibition of PDE4 can also induce arrythmias in isolated human atrial muscle strips (Leroy et al. 2011). Clinical trials with roflumiliast, however, did not report increases in serious cardiovascular events, indicating that the approved drug doses for roflumilast are not cardiotoxtic (Rabe 2011). Because of the clinical promise of PDE4 inhibitors, pharmaceutical interest has shifted forwards developing subfamily specific inhibitors, i.e., PDE4B vs PDE4D, or other novel approaches to improve the risk/benefit profile of PDE4 inhibitors.

PDE5 inhibitors

Male erectile dysfunction is a world wide public health problem that is often associated with clinically serious co-morbidities such as cardiovascular disease, diabetes, atherosclerosis, and metabolic syndrome (Francis and Corbin 2011). PDE5 enzymes are highly specific for hydrolyzing cGMP. During male sexual arousal, nitric oxide (NO)-induced increases in cGMP lead to relaxation of vascular smooth muscle in the corpus cavernosum, and, consequently, to penile erection. Since PDE5 is relatively highly expressed in the corpus cavernosum, PDE5 inhibitors can prevent the destruction of cGMP and potentiate effects of cGMP on penile erection, and improve function in normal men and especially in men with erectile dysfunction related to deficiencies in NO/cGMP-signaling (Francis and Corbin 2011). Orally administered and relatively safe, PDE5 inhibitors, sildenafil (Viagra:Pfizer), vardenafil (Levitra:Bayer/ Glaxo SmithKline) and long-acting tadalafil (Cialis:Lilly) have been enormously successful as therapy for erectile dysfunction. Adverse effects, including headache, flushing, nasal congestion, and backache are relatively mild. Because sildenafil and vardenafil inhibit PDE6, their use is sometimes accompanied by brief and reversible visual disturbances. PDE5 inhibitors are not recommended for patients with retinitis pigmentosa because increases in cGMP are associated with retinal degeneration in preclinical studies (Azzouni & Abu 2011;Davis et al. 2008).

PDE5 is relatively highly expressed vascular smooth muscle. In the pulmonary vascular bed, pulmonary hypertension results in increased mean pulmonary arterial pressure and pulmonary arterial resistance secondary to vasoconstriction and hypertrophy of small pulmonary arteries. Inhibitors of PDE5 increase cGMP/PKG-signaling, which results in pulmonary vasodilation and inhibition of vascular hypertrophy and pathological remodeling (Ghofrani et al. 2006). PDE5 inhibitor formulations for pulmonary hypertension, i.e. sildenafil (Revatio), tadalafil (Adcirca), and vardenafil, are FDA-approved as part of the therapeutic armamentarium for pulmonary hypertension (Archer and Michelakis 2009). It is because of the expression of PDE5 in vascular smooth muscle that use of nitrates is not compatible with PDE5 inhibitors. Simultaneous administration of nitrates or NO-producing drugs and PDE5 inhibitors could result in increased cGMP in the vascular network, leading to widespread vasorelaxation and a precipitous drop in blood pressure, and possibly death.

PDE inhibitors: “Off target” effects

Several factors have limited the clinical success of currently available family-selective PDE inhibitors. Unfortunately these family-selective drugs are targeted to PDE catalytic sites, and therefore inhibit all members of the same family, even in undesirable or “off target” locations, and produce dose-limiting side effects, e.g., pan-PDE4 inhibitors inhibit PDE4, perhaps PDE4D (Robichaud et al. 2002), in the emetic center of the brain, causing emesis, a major side effect which negatively impacts widespread clinical use of pan-PDE4 inhibitors (Tenor et al. 2011). In addition some “off target” effects occur via inhibition (albeit with less potency and efficacy) of multiple PDE families as well as via interactions with non-PDE proteins.

Some “off target” effects may be beneficial. Sildenafil, which is highly selective for PDE5, also inhibits PDE1 with 40 fold less selectivity (Bischoff 2004). Since PDE1 is more highly expressed than PDE5 in human myocardium, and since inhibitors of PDE1 in rodent models blocks development of cardiac hypertrophy and vascular remodeling (Miller et al. 2009), it has been suggested that some of the cardioprotective effects of sildenafil may be related to inhibition of PDE1, and that targeting of both PDE1 and PDE5 may be useful in treating cardiac failure (Movsesian and Kukreja 2011). Use of the PDE3 inhibitor, cilostazol, as therapy for intermittent claudication is not associated the cardiotoxic effects associated with chronic administration of the PDE3 inhibitor milrinone. This improved safety profile of cilostazol may be related to its inhibition of adenosine uptake in the myocardium (by mechanisms independent of inhibition of PDE3) (Kambayashi and Liu 2007;Shakur et al. 2002).

New applications for approved drugs, new drugs, novel approaches to target PDEs

Despite the problems with currently available inhibitors, several important factors have been responsible for the continued interest in developing PDE inhibitors as therapeutics. One primary factor relates to the worldwide success of PDE5 inhibitors as treatment for erectile dysfunction and pulmonary hypertension. In addition, advances in structural biology, in crystal structure-based drug design and computational modeling of potential drug candidates, and in the experimental verification of the signalosome hypothesis have also been important factors in stimulating interest in developing novel PDE inhibitors. The signalosome concept has brought about the realization that as components of compartmentalized signalosomes, individual PDEs modulate specific signaling pathways and biological responses, and that targeting these signalosomes could allow treatment of a wide range of decreases with greater specificity and less toxicity. The signalosome concept has also been a crucial driving force in the development of novel approaches to targeting individual PDEs (Houslay 2010;Keravis and Lugnier 2012;Lee et al. 2013a;Maurice et al. 2014a).

PDE inhibitors: clinical studies

Many clinical studies (with NTC numbers available), utilizing PDE inhibitors, are listed on the clinicaltrials.gov website (Maurice et al. 2014a). In some trials, currently approved drugs are being studied either in defined subsets of patients or specific clinical phenotypes in diseases for which PDE inhibitors have been approved, or as novel drugs in new clinical situations. Similarly, novel drugs are targeting novel diseases or diseases targeted by currently approved PDE inhibitors.

As for novel applications of PDE3 inhibitors, citostazol has been reported to attenuate post angioplasty restenosis and reduce the progression of atherosclerosis in diabetic patients (Katakami et al. 2010;Tamhane et al. 2009), most likely due to its inhibitory effects on vascular smooth muscle proliferation and vascular wall inflammation (Liu et al. 2011).

Because of their anti-inflammatory actions, PDE4 inhibitors are being considered for a range of inflammation-related diseases, including arthritis, systemic lupus erythematosis, inflammatory bowel disease, and type 1-diabetes (Page and Spina 2011;Wittmann and Helliwell 2013). The PDE4 inhibitor, apremilast (Otezia: Celgene), was recently FDA-approved for treatment of psoriatic arthritis (FitzGerald 2014;Wittmann & Helliwell 2013), and GSK256066 (administered by inhalation) was reported to reduce responses to inhaled allergen is a small asthma trial (Singh et al. 2010). In addition recent findings suggest that the beneficial effects of resveratrol, a polyphenol found in red wine, in preventing obesity and type 2 diabetes in rodents may be mediated via its inhibition of PDE3 and PDE4, and subsequent activation of cAMP/PKA-and AMPK-signaling (Park et al. 2012). In fact, during clinical trials with roflumiliast, subjects reported weight loss, and in a small number of newly diagnosed type 2 diabetics, roflumiliast was reported to reduce blood glucose and glycated hemoglobin (Rabe 2011;Wouters et al. 2012). Roflumilast is currently being studied in obese/diabetic patients (сf Clinictrails.gov and (Maurice et al. 2014a)).

The PDE5 inhibitors, avanafil (Stendra:Vivus), a rapidly aching PDE5 inhibitor, and tadalafil (Cialis), have recently been approved for treatment of erectile dysfunction and benign prostatic hyperplasia (BPH), respectively (Carson et al. 2014;Wang et al. 2014). Sildenafil has been reported to prevent cardiac dysfunction and cardiomyapathy in the mdx mouse model of Duchenne muscular dystrophy (Adamo et al. 2010), and there are ongoing clinical trials designed to test effectiveness of PDE5 inhibitors in Becker and Duchenne muscular dystrophies, heart failure and other cardiovascular disorders, type2 diabetes, and other disease states (Francis and Corbin 2011;Kukreja et al. 2011;Lugnier 2011;Percival et al. 2011;Tsai and Kass 2009) (cf Clinictrails.gov and (Maurice et al. 2014a).

PDE4 and PDE10 inhibitors, because of their beneficial effects in preclinical models of various neuropsychiatric disorders, are being studied as potential cognition- and memory-enhancing and neuroprotective agents (Blokland et al. 2012;Schmidt 2010). They are being tested as therapeutics for cognitive and behavioural deficits in dementia, schizophrenia and Alzheimer's disease, as well as in the neurodegenerative Huntington's disease (Bollen and Prickaerts 2012;Kehler and Nielsen 2011;Richter et al. 2013) (cf Clinictrails.gov and (Maurice et al. 2014a)).

Novel targeting approaches

Targeting multiple PDEs

The multiple clinical phenotypes that are hallmarks of many complex human diseases such as COPD, metabolic syndrome, obesity, type 2 diabetics, and asthma, reflect deregulation of multiple signaling networks. As components of individual signalosomes, individual PDEs regulate cycle nucleotide gradients in specific signaling compartments, and therefore targeting multiple PDEs in different compartments and different cells and tissues might provide more effective therapy in complex diseases. Targeting multiple PDEs might also produce synergistic responses, so that drugs could be administered at lower dosages and with fewer side effects. Multiple PDEs can be targeted with multiple drugs, drugs with a single pharmacophore of mixed specificity/selectivity, or drugs with two distinct pharmacophores connected by a chemical linker (Giembycz and Maurice 2014;Giembycz and Newton 2011). For example, cell-based studies (Walz et al. 2007), studies with PDE3B KO mice (Choi et al. 2006), and studies in transgenic mice overexpressing PDE3B in pancreatic β-cells (Harndahl et al. 2002) clearly establish a central role for PDE3B in regulation of insulin secretion. However, inhibition of PDE8B (Dov et al. 2008) and PDE10A (Nawrocki et al. 2014) also modulate insulin secretion. Recent studies in a murine model of diet-induced obesity have indicated that adiposity and weight gain was reduced and insulin sensitivity was improved in PDE10A KO mice or in WT mice treated with a PDE10 inhibitor (Nawrocki et al. 2014). Targeting these PDEs might provide information as to the mechanisns for integration of multiple signaling networks in β-cell function as well as therapy for obesity and diabetes. Resveratrol, a non-selective PDE inhibitor, may provide proof of principle for this therapeutic approach, since many of its beneficial metabolic effects as treatment for obesity and diabetics in rodents may be related to its activation of cAMP- and AMPK-signaling pathways via inhibition of PDE3 and PDE4 (Park et al. 2012).

Currently the design/discovery of dual inhibitors has primarily reflected the desire to maximize the potential of the powerful anti-inflammatory actions of PDE4 inhibitors (Giembycz and Newton 2011). The phenotypes of pulmonary inflammation-related disorders, asthma and COPD, include airway hyperreactivity and airway remodeling (thickening, hyperplasia and hypertrophy). PDE1 inhibitors block smooth muscle proliferation, and the dual PDE1 /PDE4 inhibitor, KF 1954, was reported to reduce both remolding and inflammation in an asthma murine model (Kita et al. 2009). PDE3 inhibitors have been reported to enhance the anti-inflammatory effects of PDE4 inhibitors in rodent macrophages, and act as bronchodilators and vasorelaxants in humans. Thus simultaneously targeting PDE3 and PDE4 could bring about improved clinical potency and efficacy in asthma and COPD, due to bronchodilation and enhanced anti-inflammatory responses (Banner and Press 2009;Giembycz and Newton 2011;Page and Spina 2011). In fact, a dual PDE3/PDE4 inhibitor (RPL554: Verona Pharma), with IC50 values of 400pM and 1.2μM for PDE3 and PDE4, respectively, has been evaluated in small clinical trials in patients with allergic rhinitis, asthma, and COPD. These studies suggest that acute administration (inhalation) of RPL554 improves lung function with bronchodilator and anti-inflammatory responses, and with lack of gastro-intestinal side effects of “classical PDE4 inhibitors (Calzetta et al. 2013;Franciosi et al. 2013;Page and Spina 2012). However, the long-term effects of administration of this compound, especially given the cardiotoxic effects of milrinone in patients with heart failure, must be seriously considered.

Additional “multi-modal” approaches, especially for pulmonary diseases, involve development of inhaled PDE4 and PDE3/PDE4 inhibitors, inhaled antisense oligonucleotides for knockdown of individual PDE4 isoforms, or utilization of combinations of PDE4 inhibitors (roflumilast) and inhaled corticosteroids, LABAs such as salmeterol, or anticholinergics such as the muscarinic receptor antagonist tiotroprium (approved as combination therapy for COPD) (Banner and Press 2009;Giembycz and Newton 2011;Page and Spina 2011). It is of interest that bivalent drugs/ligands, i.e., single molecules containing two pharmacophores that engage two different targets, have been developed; one compound exhibits both PDE4 inhibitor activity and β2-adrenergic agonist activity. Another bivalent ligand has been reported to both inhibit PDE4 and act as an anti-cholinergic agent (by interactions with the muscarinic receptor) (Giembycz & Maurice 2014).

Allosteric modulators

As discussed above, in PDE4 isoforms, UCR2 regions (Fig2) and CR3 regions (control Region3, a helical segment in the C-terminal region), fold over and “cap” the catalytic site, where they interact with certain classes of PDE4 inhibitors and increase their affinities at the catalytic site. Allosteric modulators are drugs that disrupt these interactions and partially inhibit PDE4 activity so as to maintain some basal cAMP signaling and reduce toxicity (this is rather controversial at this time). Because the structural determinants and amino acid sequences in the UCR2 and CR3 domains that form the “cap” differ in PDE4B and PDE4D, Burgin and associates have taken advantage of the polymorphisms and generated specific modulators (RS25344, DI59153) for PDE4D with anti-inflammatory, pro-cognitive, and anti-depressant actions (without significant emesis) in rodent models (Burgin et al. 2010;Gurney et al. 2011). Further studies have shown that a single polymorphism in the CR3 helix (Leu 674 in PDE4B1 verses Gln594 in PDE4D) which interacts with a 2-arylpyrimidine PDE4 inhibitor (A-33) is responsible for the70-80 fold subtype selectivity of A-33 for PDE4B over PDE4D (Fox et al. 2014). Furthermore, studies with PDE4 KO mice suggested that, in a mouse model of pulmonary allergic hypersensitivity, deletion of PDE4B resulted in a phenotype which demonstrated many of the anti-inflammatory responses ascribed to pan-PDE4 inhibitors (Houslay et al. 2005;Jin et al. 2007). Since other studies with PDE4 KO mice indicated that inhibition of PDE4D, not PDE4B, is associated with the emetic response of pan-PDE4 inhibitors (not studied in human patients) (Robichaud et al. 2002), development of PDE4B –selective inhibitors might provide efficacious anti-inflammatory agents, with fewer side effects than pan-PDE4 inhibitors, for use in pulmonary inflammatory diseases such as asthma, and in other inflammation-related diseases as well.

The X-ray crystal structure of a truncated PDE2 molecule that contained N-terminal GAF domains and catalytic domains suggest that dimerization holds the catalytic domain in an orientation that restricts access of substrate to the catalytic sites (Pandit et al. 2009). cGMP binding induces a conformational change in the GAF-B domain and the adjoining linker region connecting it to the catalytic domain. Movement of the linker region is proposed to alter the orientation of the catalytic domain and relieve autoinhibitory constraints (Pandit et al. 2009). Compounds that block allosteric activation of PDE5 by binding to the PDE5 GAF domain have been described, and serve as proof of principle that the structural determinants that allow allosteric modulation of other PDEs might be similarly targeted (Schultz et al. 2011).

Taken together, these studies suggest possibly effective alternative strategies to designing subtype- selective PDE inhibitors, in addition to the crystal structure/computational design approach to catalytic site inhibitors.

Signalosome disruptors

As discussed above many individual PDE isoforms perform non-redundant functions via their incorporation/recruitment into macromolecular signaling complexes or signalosomes. PDE-containing signalosomes, via protein/protein or protein/lipid interactions, are tethered/localized to distinct subcellular signaling domains/compartments where they regulate local cyclic nucleotide gradients, signaling pathways, and physiological/pathophysiological events. It is thought that rather than directly inhibiting PDE catalytic activity, displacement of PDEs from signalosomes or disrupting the signalosomes could affect compartmentalized, physiologically relevant cyclic nucleotide signaling pathways. Signalosome destabilizers would have more selective, specific, and compartmentalized effects as opposed to more global responses to catalytic site inhibitors. These approaches, however, require both knowledge of the interaction domains of PDEs and their signalosome partners, as well as novel targeting strategies.

The heat shock protein HSP20 serves a cardioprotective function in blocking the hypertrophic response to β-adrenergic stimulation in rodent models or cultured cadriomyocytes (Fan et al. 2006). A specific signalosome, containing a PDE4D/HSP20 signaling complex, regulates PKA-induced phosphorylation of HSP20, and modulates βAR-induced hypertrophic responses. Exposure of cultured cardiomyocytes to a cell-permeable peptide, designed to disrupt interactions between the PDE4 catalytic domain and HSP60, disrupted the complex, increased PKA-induced phosphorylation of HSP60, and attenuated the hypertrophic response (Fan et al. 2006;Sin et al. 2011). Recent studies also suggest that, in human aortic endothelial cells (HAEC), an Epac1-based signalosome that contains an Epac1/PDE3B complex regulates PI3-Kγ activity and cAMP-dependent regulation of HAEC adhesion, spreading, as well as tubule formation and angiogenesis (Wilson et al. 2011). The N-terminal domain of PDE3B was required for incorporation of PDE3B into these Epac-based signalosomes (Raymond et al. 2007). In these HAEC, a cell-permeable peptide designed to interfere with Epac1/PDE3B interactions produced effects similar to those elicited by cilostamide (specific PDE3 inhibitor) or siRNA-induced knockdown of PDE3B, i.e., Epac1/R-Ras-dependent activation of PI3-Kγ (Wilson et al. 2011). These and other studies provide proof of principle regarding use of signalosome disruptors to target individual PDEs and modulate compartmentalized cAMP-signaling pathways and biological effects (Keravis and Lugnier 2012;Lee et al. 2013a;Maurice et al. 2014a). They also present a significant challenge to design effective targeting methodologies in the intact organism.

Summary and conclusions

PDEs modulate myriad physiological and pathophysiological processes via their regulation of the critical intracellular second messengers, cAMP and cGMP. Formation of PDE-containing signalosomes, through protein/protein interactions with regulatory partners and adaptor, anchoring, and/or scaffold proteins, contribute to the specificity of cAMP-and cGMP-mediated signal transduction. Formation of these multimolecular signaling/regulatory complexes involves recruitment of specific PDEs and specific signaling molecules into compartmentalized signaling networks, in which spatial constraints and organization allow tight local control of cAMP and cGMP signals and their temporal transduction along specific pathways. At this time, PDE inhibitors are used to treat diseases which reflect dysregulation of these signals and signaling networks, e.g., erectile dysfunction, pulmonary hypertension. Emerging understanding of the functional roles of multiple PDE isozymes in these compartmentalized signaling networks and of the molecular forces involved in signalosome assembly, together with increasing ability to target these localized signalosomes and affect these compartmentalized networks and localized cyclic nucleotide gradients, will provide a platform for the PDE superfamily to emerge as an enticing target to treat major diseases with next generation inhibitors and modulators.

Abbreviations

AC

Adenylyl cyclase

ARDS

Acute respiratory distress syndrome

ATF

activating transcription factor 4

AIP

aryl hydrocarbon receptor-interacting protein

AKAP

A-Kinase Anchoring Proteins (AKAPs)

βAR

β-adrenergic receptor

BAY60-7550

(2-[(3,4-dimethoxyphenyl)methyl]-7-[(1R)-1-hydroxyethyl]-4 phenylbutyl]-5-methyl-imidazo[5,1-f][1,2,4]triazin-4(1H)-one)

BK(Ca)

Ca(2+)-activated K(+) channels

BPH

benign prostatic hyperplasia

Ca(v)1.2

L-type calcium channel

ICa

cardiac L-type Ca2+ current

Cav

Caveolin

cAMP

cyclic AMP

cGMP

cyclic GMP

CL

CL316243

cGKII

cGMP-dependent protein kinase II

CNG

cGMP-gated cation channel

CLL

B-cell chronic lymphocytic leukemia

COPD

Chronic obstructive pulmonary disease

DISC1

Disrupted-In-Schizophrenia 1

D1-R

dopamine-D1 receptor

DARPP-32

dopamine-and cAMP-regulated phosphoprotein of Mr 32 kDa

EC

Endothelial cell

EHNA

(erythro-9-(2-hydroxy-3-nonyl)adenine)

EPAC

Exchange protein directly activated by cyclic AMP

FRET

fluorescence resonance energy transfer

GSK-3

glycogen synthase kinase-3

GCAP

guanylate cyclase activating proteins

HAEC

human aortic endothelial cells

HSG

human neoplastic submandibular gland intercalated duct cells

HSL

hormone sensitive lipase

IR

Insulin receptor

IP3R

insoitol-1,4,5-triphosphate receptor

IRAG

IP₃R1-associated PKG-substrate protein

KCNQ

IKs, potassium channel subunit

LTCC

L-type Ca⁺⁺ channels

LUTS

lower urinary tract symptoms

MAPKAPK2

MAPK-activated protein kinase2

mAKAP

muscle-specific A-kinase anchoring protein

MDR

multi drug resistance

MRP7

ATP-binding cassette C10, multidrug resistance protein 7

MTG

myeloid translocation gene

NCKX2

Na(+)/Ca(2+), K(+) exchanger

NIHL

Noise-induced hearing loss

Ndel1

Nuclear distribution element-like protein

PDE

phosphodiesterase

PDP

(9-(6-Phenyl-2-oxohex-3-yl)-2-(3,4-dimethoxybenzyl)-purin-6-one)

PKA

cAMP-dependent protein kinase

PKA-RII

PKA-regulatory subunit

PIP3

phosphatidylinositol-(3,4,5)-triphosphate

PI3-Kγ

phosphoinositide-3-kinaseγ

PLB

phospholamban

PMCA4

plasma membrane Ca(2+)/calmodulin-dependent ATPase 4

RACK-1

receptors for activated C-kinase-1

RyR

ryanodine receptor

SPHKAP

sphingosine kinase type 1 interacting protein

SR

sarcoplasmic reticulum

SERCA2

SR calcium ATPase type 2a

SMC

smooth muscle cell

UCR

Upstream conserved region

VEC

Vascular endothelial cell

VE-Cad

vascular endothelial cadherin

VSMC

vascular smooth muscle cells