Phosphodiesterases: Evolving Concepts and Implications for Human Therapeutics

Abstract

Phosphodiesterases (PDEs) are a superfamily of enzymes that hydrolyze cyclic nucleotides. While the 11 PDE subfamilies share common features, key differences confer signaling specificity. The differences include substrate selectivity, enzymatic activity regulation, tissue expression, and subcellular localization. Selective inhibitors of each subfamily have elucidated the protean role of PDEs on normal cell function. PDEs are also linked to diseases, some of which affect the immune, cardiac, and vascular systems. Selective PDE inhibitors are clinically used to treat these specific disorders. Ongoing preclinical studies and clinical trials are likely to lead to the approval of additional PDE-targeting drugs for therapy in human disease. In this review, we discuss the structure and function of PDEs and examine current and evolving therapeutic uses of PDE inhibitors, highlighting their mechanisms and innovative applications that could further leverage this crucial family of enzymes in clinical settings.

INTRODUCTION

Phosphodiesterases (PDEs) are a large family of ubiquitously expressed proteins (1) that were first described by Sutherland & Rall in 1958 (2). PDEs hydrolyze the second messengers cyclic adenosine and guanosine monophosphate (cAMP and cGMP, respectively), converting them into AMP and GMP, respectively. These cyclic nucleotides are synthesized by corresponding adenylyl and guanylyl cyclase (AC and GC, respectively), which are activated in response to various G protein–coupled receptor (GPCR) agonists, hormones, or nitric oxide (NO). By fine-tuning cyclic nucleotide–dependent signaling, classically conferred by the actions of protein kinases A and G (PKA and PKG, respectively), PDEs regulate many intracellular biochemical pathways central to cellular function, homeostasis, and stress responses. Early characterization of PDEs revealed that their enzymatic activity was inhibited by methylxanthines (e.g., caffeine) (2), suggesting that PDEs may be important in human diseases. Sutherland was awarded the 1971 Nobel Prize for his discovery that cAMP was a common intermediary in cellular responses to certain hormones and neurotransmitters. There are currently 35 PDE inhibitors approved by various regulatory agencies, including the US Food and Drug Administration (FDA), for the broad treatment of human diseases, with 14 additional agents in ongoing clinical investigations (3).

STRUCTURAL AND FUNCTIONAL CONCEPTS UNDERLYING PDEs

PDE-mediated cell signaling regulation is dependent upon the synthesis of cyclic nucleotides. cAMP synthesis is coupled to stimulatory GPCRs, with peptides and neurotransmitters generally serving as GPCR ligands (4). Upon ligand binding, the activated stimulatory G (G_s) protein stimulates one (or more) of the ten ACs to generate cAMP (5). Elevation of cAMP in turn activates PKA, exchange protein activated by cAMP (Epac), hyperpolarization-activated cyclic nucleotide–gated channels (HCNs), or Popeye domain–containing proteins (6). cGMP is generated by GC-1 or GC-A/B. The soluble GC-1, activated by NO, is located in the cytoplasm and plasma membrane (7). Natriuretic peptides (NPs) binding to their cognate receptors activate GC-A/B, which are physically coupled to plasma membrane NP receptors (8). cGMP in turn activates cGMP-dependent kinase (cGK, also known as PKG) and cGMP-gated ion channels (9).

The PDEs are a superfamily consisting of 11 main submembers (PDE1–11). They are transcribed from 21 genes that generate more than 100 different isoforms and splice variants. PDEs vary in their affinities for the substrates cAMP and cGMP. PDE1, 2, 3, 10, and 11 hydrolyze both. On the other hand, PDE4, 7, and 8 are cAMP specific, while PDE5, 6, and 9 are cGMP specific (1, 8). PDEs further differ in their enzymatic kinetics, tissue expression, and intracellular compartmentalization (8), as summarized in Table 1.

Table 1.

PDE tissue, cell expression, and compartments

PDE family (references)	Isoforms and splice variants	Substrate specificity	Main tissue expression	Compartment(s)
PDE1 (10–13)	PDE1A	cGMP > cAMP	Vasculature, kidney, heart	Nucleus, cytosol
	PDE1B	cGMP > cAMP	Lymphocytes, brain	Cytosol
	PDE1C	cGMP = cAMP	Vasculature, heart	Cytosol, plasma membrane
PDE2 (14–16)	PDE2A1 PDE2A2 PDE2A3	cGMP = cAMP	Brain, heart, kidney	Cytosol, SR, myofilament
PDE3 (17–20)	PDE3A1 PDE3A2 PDE3A3	cAMP > cGMP	Heart, vasculature, platelets	Cytosol, SR, nucleus
	PDE3B	cAMP > cGMP	Adipose, liver, pancreas	Plasma membrane (T-tubules)
PDE4 (21–27)	PDE4A	cAMP	Peripheral blood cells, heart, brain	Cytosol
	PDE4B	cAMP	Peripheral blood cells, heart, brain, germ cells	Plasma membrane
	PDE4C	cAMP	Brain	Cilia
	PDE4D3 PDE4D5 PDE4D8 PDE4D9	cAMP	Peripheral blood cells, heart, liver, brain, germ cells	Plasma membrane (T-tubules), SR, cytosol
PDE5 (28–30)	PDE5A	cGMP	Heart, lungs, penis	Cytosol, myofilament
PDE6 (31)	PDE6D	cGMP	Eye	Plasma membrane
PDE7 (32–34)	PDE7A1 PDE7A2 PDE7A3	cAMP	Heart, skeletal muscle, spleen, peripheral blood cells	Cytosol, plasma membrane
	PDE7B	cAMP	Brain, peripheral blood cells, heart, kidney, stomach, thyroid, bladder	Unknown
PDE8 (35, 36)	PDE8A1 PDE8A2	cAMP	Testis, spleen, colon, heart	Cytosol
	PDE8B1 PDE8B2	cAMP	Heart	Plasma membrane
PDE9 (37, 38)	PDE9A	cGMP	Heart, liver, adipose, brain, bladder	Cytosol, mitochondria, plasma membrane, nucleus
PDE10 (39, 40)	PDE10A	cAMP = cGMP	Brain, heart	Plasma membrane
PDE11 (41)	PDE11A1 PDE11A2 PDE11A3 PDE11A4	cAMP = cGMP	Brain, eye	Nucleus, cytosol, plasma membrane

Abbreviations: cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; PDE, phosphodiesterase; SR, sarcoplasmic reticulum.

The C-terminal catalytic domain is well conserved across PDE submembers. This region includes the cyclic nucleotide binding site, which is also the primary target for most pharmacologic PDE inhibitors (Figure 1). Crystallographic studies of the C terminus based on PDE1, 4, and 5 revealed that key chemical interactions within a hydrophobic pocket determine substrate recognition in a mechanism known as a glutamine switch (42). This mechanism postulates that hydrophobic residues within the pocket domain permit or restrict the ability of an invariant glutamine (Q) to switch sterically. This movement is critical in the formation of two hydrogen bonds with the purine moiety of one or both cyclic nucleotides, serving as a mechanistic determinant of substrate specificity. In cGMP-specific PDE5A (Figure 1a), as an example, Q775 stabilizes the invariant Q817 to strongly favor two hydrogen bonds between Q817 and the hydrolysis enzymatic product, GMP (42). The glutamine switch hypothesis then states that hydrogen bond formations between Q775 and Q817, and between Q817 and cGMP, render PDE5A highly cGMP selective.

Figure 1.

Substrate recognition in the catalytic domain of PDEs. (a) Visualization of the hydrophobic pocket within the PDE5A catalytic domain. The nucleotide-recognizing glutamines are highlighted in red, and the blue dashed circle shows the clamp mechanism important in substrate specificity. The invariant glutamine Q817 is stabilized by Q775 in a particular orientation as to form two hydrogen bonds with GMP/sildenafil. (b) The invariant glutamine Q817 is supported by Q775 to be fixed in a cGMP-, but not cAMP-, recognizing orientation. Mutations of either, however, showed losses in cGMP recognition, with no increases in cAMP recognition, challenging the glutamine switch hypothesis. (c) The structure of PDE4D2 is shown, where the invariant glutamine Q369 forms a single bond with cAMP, contrary to the two predicted with AMP binding. This is due to the different orientation of AMP versus cAMP that propels the adenine moiety of AMP farther away from Q369 to form an H bond. (d) A summary schematic of specific domains found in the variable N-terminal region of PDEs. Abbreviations: CaM, calmodulin; cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; GAF, cGMP-dependent PDE, Anabaena AC, Escherichia coli FhlA; NHR, N-terminal hydrophobic region; PAS, Per-Arnt-Sim; PDE, phosphodiesterase; Q, glutamine; UCR, upstream conserved region. Panel a adapted from Reference 42, figure 1g, with permission from Elsevier; panel b adapted from Reference 43, figure 2 (CC BY 4.0); and panel c adapted from Reference 44, figure 2c, with permission from Elsevier.

This hypothesis, however, has been challenged on several grounds. One is that mutating the invariant glutamine Q817 or the surrounding Q775 in PDE5A leads to a loss of cGMP recognition but without increasing affinity for cAMP (43) (Figure 1b). Disrupting this bond would, in theory, render PDE5A a dual-substrate enzyme. Furthermore, a PDE4D2 crystallography study with the substrate cAMP versus the product AMP [the latter was used to resolve the original structure (42)] identified only one hydrogen bond between the invariant Q369 and cAMP (Figure 1c).This finding contrasts with the prediction of the glutamine switch model, in that two hydrogen bonds between Q369 and the substrate would confer cAMP specificity to PDE4 (44). In the same study, a comparison of PDE4 against the dual-substrate PDE10A (with 70-fold preference for cAMP over cGMP) revealed that the two enzymes likely have distinct means of substrate recognition. The body of evidence included distinct substrate orientation and hydrogen bond formation patterns for each of the PDEs. In summary, while an invariant glutamine is important in substrate affinity, it does not serve as a deterministic mechanism for substrate recognition in all PDE subfamily members. Nevertheless, the hydrophobic pocket is critical since PDE inhibitors act as false substrates, as shown in Figure 1a by the structure of sildenafil binding and inhibiting PDE5A.

The variable N-terminal regulatory region confers functional and subcellular localization specificity to PDE families (Figure 1d). Depending on the PDE, this domain includes calmodulin (CaM) binding, GAF (name derived from cGMP-dependent PDE, Anabaena AC, Escherichia coli FhlA), N-terminal hydrophobic regions (NHRs), upstream conserved regions (UCRs), and Per-Arnt-Sim (PAS). PDE7 is the only family member without any known regulatory domains (32). The PDE1 family uniquely contains a calcium/CaM binding domain that is required for its activation (6). PDE2, 5, 6, 10, and 11 all contain GAF domains that conformationally regulate PDE activity by allosteric, noncatalytic binding of one or the other cyclic nucleotide (45). For instance, cGMP binding to the GAF-B domain in PDE2A activates cAMP hydrolysis, whereas cGMP binding to GAF-A in PDE5A activates cGMP hydrolysis. NHR domains allow membrane targeting and are only present in PDE3,where its splice variants are marked by the presence of one or more NHR1 and NHR2 (46). UCRs are exclusively found in PDE4 isoforms and distinguish long, short, and supershort variants of this large family (1). Functionally, UCRs help to mediate PDE4 oligomerization and binding to other partners, such as the disrupted in schizophrenia 1 protein (1, 47). A PAS domain is unique to PDE8 and plays a role in protein-protein interactions (48). PDE9A contains the seven-residue nuclear localization signal pat7 motif (49), though its enzymatic function is not restricted only to the nucleus. In addition to these domains, PDEs are subject to posttranslational modifications such as phosphorylation and myristoylation that can further regulate the enzymatic function of the PDEs (9).

COMPARTMENTALIZED CYCLIC NUCLEOTIDE SIGNALING

Studies over the past decade have provided new insights into how PDEs control signaling cascades in specific subcellular domains. We discuss the most relevant and recent studies and direct readers to other excellent reviews for additional information (9, 11, 50, 51). Four concepts (Figure 2a–d) are helpful in illustrating compartmentalized PDE-mediated cell signaling pathways: (a) Signalosomes confine PDEs spatially, (b) this enables cyclic nucleotide generators or transporters to transduce different physiologic stimuli with distinct functional outcomes; (c) cross talk between cAMP and cGMP intracellularly and between cells suggests a wider impact of localized signaling; and (d) pathologic alterations can disrupt compartmentalized signaling.

Figure 2.

Conceptualizations of compartmentalized cyclic nucleotide signaling. (a) The scaffold protein AKAP18 forms a signalosome with SERCA2A, PKA, and PDE3A at the SR, where it mediates β₁AR signaling. (b) Compartments of cGMP signaling downstream of different stimulants are shown. Shaded areas represent cGMP pools. (c) NO-induced cGMP (circles) production in fibroblasts translocates to cardiomyocytes via Cx43 and inhibits PDE3 to increase cAMP levels. (d) Mislocalization of β₃AR from T-tubules to the general sarcolemmal membrane from healthy to failing cardiomyocytes alters cGMP and cAMP signaling. Abbreviations: AKAP, A-kinase anchoring protein; ANP, atrial natriuretic peptide; BNP, brain natriuretic peptide; AR, adrenergic receptor; cAMP, cyclic adenosine monophosphate; cGMP, cyclic guanosine monophosphate; CNP, C-type natriuretic peptide; Cx43, connexin 43; GC, guanylyl cyclase; GC-A/B, guanylyl cyclase-A/B; NO, nitric oxide; PDE, phosphodiesterase; PKA, protein kinase A; SERCA2A, SR ATPase; SR, sarcoplasmic reticulum. Figure adapted from images created with BioRender.com.

Compartment-specific macromolecular signalosomes spatially confine PDE signaling. cAMP-hydrolyzing PDEs form such signalosomes through scaffolds, classically including A-kinase anchoring proteins (AKAPs) (52). There are more than 50 members of this family of scaffolding proteins, and each forms a complex with a select cAMP-hydrolyzing PDE, PKA, GPCR, and/or additional proteins at a given intracellular domain. While many substrates might theoretically be phosphorylated by PKA,the PKA-AKAP-PDE signalosome narrows the interactome to local substrates. Table 2 lists known AKAP-PDE signalosomes and their cellular localization. Figure 2a illustrates AKAP18δ in complex with PDE3A, PKA, and the sarcoplasmic reticulum (SR) proteins SR ATPase (SERCA) and phospholamban (53). A dual role of AKAP18δ in complex with the CaM-dependent protein kinase II has also been reported (54).

Table 2.

AKAPs mediate cAMP-hydrolyzing PDE signaling

Cellular compartment	AKAP(s)	Signalosomes	Reference(s)
Plasma membrane	AKAP5, AKAP9, AKAP18a, AKAP18b, AKAP79, AKAP150	β-AR, L-type calcium channels, potassium channels	61–66
Actin cytoskeleton	AKAP250 (AKAP12)	β2-AR signaling, PDE4	67
Cytoskeleton	Gravin, Ezrin	Not reported	68
Sarcoplasmic reticulum	AKAP18δ	SERCA2A, CamKII	53, 54
Cytosol	SKIP, GSKIP	Not reported	69–73
Mitochondria	D-AKAP1	Not reported	74
Nucleus	AKAP95 (AKAP8)	cAMP, PDE4D5, PKA	75
	AKAP6 (mAKAP)	PDE4D3, PKA, PP2A, Epac1, ERK	76, 77
	Pericentrin, AKAP350	Not reported	78, 79

Abbreviations: β-AR, beta adrenergic receptor; AKAP, A-kinase anchoring protein; CaMKII, calcium-calmodulin dependent protein kinase II; cAMP, cyclic adenosine monophosphate; Epac, exchange protein directly activated by cAMP; ERK, extracellular signal regulated protein kinase; GSKIP, GSK3β interaction protein; mAKAP, muscle-specific AKAP; PDE, phosphodiesterase; PKA, protein kinase A; PP2A, protein phosphatase 2A; SERCA2A, sarcoplasmic reticulum ATPase pump; SKIP, sphingosine kinase interacting protein.

GPCRs have distinct receptor-associated cAMP gradients at the plasma and intracellular membranes. These gradients were experimentally demonstrated using biosensors that detect cAMP levels at fixed distances from specific receptors. In human embryonic kidney cells, glucagon-like peptide 1 receptor stimulation establishes a local concentration gradient of cAMP distinct from that generated by β₂-adrenergic agonism (55). In specialized cells such as cardiomyocytes, differential modulation of cAMP at the myofilament, caveolae, and junctional SR has been demonstrated (15,56–58). Computational simulations support the concept that regulating localized cAMP pools can exert particular functional responses, such as increasing cardiomyocyte contractility (56, 57). GPCRs are also found in intracellular organelles where internalized agonists can induce cAMP signaling (Figure 2a). Recent studies using optogenetic activation of AC at the sarcolemmal/junctional SR versus the bulk SR revealed that selective β₁-adrenergic receptor (β₁AR) agonism in the latter domain can specifically control the kinetics of heart relaxation by increasing PKA phosphorylation of phospholamban. In contrast, the strength of heart contraction is determined by GPCR agonism at the sarcolemmal/junctional SR through PKA phosphorylation of the ryanodine receptor (59). A similar mechanism has been shown to activate PKA specifically at the SR (60).

The cGMP/PDE/PKG signaling system does not have an identified anchoring protein akin to AKAPs. Nevertheless, its compartmentalization is largely determined by the initiating molecule or peptide that activates the GC production of cGMP (Figure 2b). Synthesis or uptake of NO activates GC-1 to generate cGMP, and PDE1A, 2A, and 5A regulate this pool of cytosolic cGMP (10, 80).In a negative feedback,PDE5 suppresses NO/cGMP signaling. Given the role of this signaling in vascular relaxation, inhibiting the enzyme prolongs the effects of NO to reduce vascular tone (81). In the heart, augmenting NO/cGMP signaling by PDE5A inhibition counters pathological chamber remodeling (such as hypertrophy and fibrosis) and also reduces positive inotropic effects from β-adrenergic stimulation (82).

NPs couple to GC-A/B to also leverage cGMP/PKG signaling. Atrial or brain natriuretic peptide (ANP or BNP) couples to GC-A. In cardiomyocytes, PDE2, 3, and 9 can regulate GC-A signaling. PDE2 and 3 regulation controls calcium homeostasis, which is important in excitation-contraction coupling (15). In contrast, PDE9A localization at mitochondria appears to have a specific influence over metabolism, particularly in fatty acid oxidation (37).The C-type NP (CNP) couples to GC-B, and this pool of cGMP is regulated by PDE1 (10), 2, and 3 (15). In the heart, CNP, but not ANP/BNP, induces lusitropic and negative inotropic changes by targeted signaling in the SR and myofilament domains (15). Specific domain-targeted cGMP biosensors show that PDE2 and 3 overlap in their modulation of CNP-coupled cGMP levels at the SR and myofilament (15). In mice, PDE1-mediated cGMP regulation could only be detected in response to CNP/GC-B but not to NO/GC-1 stimulation. Nevertheless, PDE1 inhibition functionally enhanced cardiac relaxation effects from CNP as well as NO (10).

While PDE-regulated compartmentalization of cyclic nucleotide-mediated signaling translates to selective responses, changes in the level of one nucleotide type can modulate that of the other. Such cross talk between cAMP and cGMP is particularly facilitated by dual-substrate enzymes PDE2 and 3. When levels of cGMP rise, cGMP binding to the PDE2 GAF domain will allosterically activate PDE2 hydrolysis of cAMP. Another example is cGMP binding to the catalytic domain of PDE3, which competitively inhibits cAMP hydrolysis. Thus, increased cGMP can lower or activate PKA by PDE2 activation or PDE3 inhibition, respectively. Physiologically, CNP-triggered cGMP production activates PKA signaling via a PDE3-dependent process and is associated with both β₁- and β₂AR stimulation in healthy and failing myocardium (83). A recent preclinical study demonstrates how cGMP-cAMP cross talk may induce muscle atrophy associated with heart failure (HF) (84). In combination with CNP, skeletal muscle–derived musculin potentiates cardiomyocyte cGMP and cAMP production. The latter increases phospholamban PKA phosphorylation in a PDE3-dependent manner that does not require PKG activation. This CNP/cGMP/cAMP/PKA cross talk also appears to be cardioprotective in a pressure overload (transaortic constriction surgery) model in the mouse (84).These results show how peripheral tissue signaling can regulate compartmentalized cyclic nucleotide signaling pathways in a paracrine manner to promote the development of heart disease.

Paracrine signaling that creates cGMP-cAMP cross talk among different cell types has also been described for NO/cGMP regulation of cAMP/PKA. In contrast to the negative inotropic effects of NO at high concentrations (micromolar range), low concentrations (nanomolar range) of NO induce a positive inotropic effect in cardiomyocytes via a PKA-dependent mechanism (85). At low concentrations, NO-induced cGMP inhibits PDE3 to activate cAMP/PKA signaling (82). There are controversies about the source of cGMP, however, as some studies have failed to find that NO induces cGMP synthesis in cardiomyocytes (86). An alternative proposal bypasses the requirement for cardiomyocyte NO/GC-1 activation (Figure 2c).This proposal suggests that the gap junction protein connexin 43 enables transfer of cGMP from fibroblasts to cardiomyocytes, which in turn inhibits PDE3 to increase PKA phosphorylation of the myofilament proteins (e.g., troponin I) (87).

The ratio between the two cyclic nucleotides also impacts their relative roles in ciliated cells. In ciliated cells, the balance between cAMP and cGMP determines neuronal cell asymmetry, or polarity (88), and cyst growth (27). This balance is critical in neuronal migration and cyst formation. In cortical interneurons, an increased cGMP/cAMP ratio in the cilia promotes a cell polarity change, whereas a reduced ratio maintains basal polarity. Manipulation of either cyclic nucleotide in the centrosome, on the other hand, affects cell movement but not polarity. While PDE4C regulates ciliary cAMP signaling in epithelial cells (27), whether the enzyme would also play a critical role in interneuron polarity has not been elucidated.

Cyclic nucleotides can also be extruded from cells into the extracellular space, where they generate intracellular signaling changes that engage PDEs. For example, extracellular cAMP can generate a positive feedback cycle that amplifies prosurvival signaling in cardiomyocytes. The feedback cycle begins by the translocation of cAMP via multiple drug resistant 4 (MRP4, ABCC4) transporters. Surface-bound ecto-5′-nucleotidases CD39 and CD73 catabolize the pool of extracellular cAMP to form adenosine (89). Adenosine can then bind to its cognate receptor to stimulate intracellular cAMP production to complete a feedback cycle. Adenosine signaling is antiapoptotic in ischemic heart disease (90). Adenosine 2 receptor (A₂R) agonists reduce cellular susceptibility to hydrogen peroxide–induced cell death in vitro, mechanistically requiring MRP4 (91). Exogenously increasing extracellular cAMP decreases cardiac injury and contractility impairments in a mouse model of ischemic heart disease (91). cAMP is also neuroprotective in acute ischemic stroke by acting on multiple cell types to reduce inflammation (92), indicating wider implications for extracellular cAMP signaling. PDE1C coimmunoprecipitates with A₂Rs in cardiomyocytes (93), suggesting that adenosine and PDE1C signaling pathways may work synergistically.

Pathologic alterations can disrupt compartmentalized cell signaling. PDE5A regulates the NO-derived cGMP pool in healthy cardiomyocytes, but this selectivity switches to NP-derived cGMP in murine hearts subjected to pathological pressure overload (80). In this setting, PDE5A relocalizes from Z-disks to a more diffused cytosolic distribution. This changes PDE5 regulation of cGMP that is derived from NO to that of NP (80). Other components of cyclic nucleotide signaling, such as GPCRs (58, 94, 95), GCs (96), and protein kinases (80), have also been shown to relocalize to disrupt nanodomain regulation. Specific perturbations are described for the specialized membrane invaginations in cardiomyocytes called T-tubules. T-tubules are structural hubs for protein mediators of excitation-contraction coupling, including β₃ARs. In diseased hearts, β₃ARs mislocalize as a result of reduced T-tubular structural integrity (detubulation). Detubulation alters the cGMP and cAMP signaling pathways, so that neither remains confined to T-tubules. These changes specifically abrogate PDE2-mediated cGMP-cAMP cross talk between the β₃AR-associated cGMP pool and cAMP (58).The result of this uncoupling is that cAMP levels, normally kept low because of cGMP activation of PDE2, become diffused and elevated (Figure 2d). Similarly, detubulation also disrupts β₂AR/PDE3 signaling (94, 95).

Collectively, PDEs critically maintain distinct cell signaling pathways downstream of cAMP and cGMP. The cross talk between cAMP and cGMP via PDE2 and 3 is another mediator of cell signaling. Pathologic perturbations to many of these aspects are now known. Therapeutics that leverage either or both cell signaling pathways are effective in treating human diseases. While the initial development of PDE inhibitors as therapeutics preceded much of this molecular insight, such findings are likely to shape future advances.

THERAPEUTIC IMPLICATIONS: CURRENT AND FUTURE CONSIDERATIONS

Clinical findings substantiate the essential roles PDEs play in human health. For example, a mutation in PDE1C is associated with hearing loss (97), while individuals with a specific PDE11A polymorphism are more susceptible to mood disorders and less likely to respond to antidepressants (98). Many studies have also reported pathologically enhanced functions of PDEs, spawning efforts to develop selective PDE inhibitors for human disease. Currently in the United States, ten different PDE-specific inhibitors are FDA approved to treat nine different diseases (Figure 3). What follows is a review of these uses, grouped by the PDEs and cyclic nucleotides primarily impacted.

Figure 3.

Depiction of the human body, showing systems or organs where phosphodiesterase (PDE)-specific inhibitors are currently approved by the US Food and Drug Administration for use. The targeted PDE is listed, followed by the names of approved drugs underneath. Figure adapted from images created with BioRender.com.

THERAPEUTICS LEVERAGING cAMP REGULATION

PDE4: Inflammatory Diseases

PDE4 inhibitors exert an anti-inflammatory effect by increasing regulatory T cells (99) while reducing proinflammatory effector T cells (100). Current FDA-approved PDE4 inhibitors include roflumilast, apremilast, and crisaborole. Orally administered roflumilast is effective against refractory and/or severe forms of chronic obstructive pulmonary disease (101), while topical roflumilast is used for plaque psoriasis. Apremilast is an oral drug for plaque psoriasis, psoriatic arthritis, and oral ulcers associated with Behcet’s disease. Crisaborole is topically applied to treat atopic dermatitis (102).

Several pathways can mediate the beneficial effects of PDE4 inhibitors in these inflammatory disorders (Figure 4). In the skin (see the left side of Figure 4), PDE4 inhibition suppresses excessive keratinocyte regeneration to reduce epidermal thickness (103). PDE4 inhibition exerts antiproliferative effects by regulating cell cycle progression in keratinocytes. PDE4 inhibition increases expression of the cell cycle inhibitor p21. Additionally, PDE4 inhibition limits nuclear translocation of nuclear factor κB p65 subunit to decrease transcript levels of cyclin D1, a key regulator of the cell cycle (104). In the adaptive immune response (see the center of Figure 4), interleukin-2 (IL-2) receptor binding transduces T cell activation to promote cell differentiation and proliferation. PDE4B2 promotes IL-2 production, while high cAMP levels (e.g., via PDE4 inhibition) diminish this response (99, 105). In airway smooth muscle cells (see the right side of Figure 4), PDE4 inhibition exerts an anti-inflammatory effect through the immune suppressor mitogen-activated protein kinase phosphatase 1 (MKP1), whose promoter contains a cAMP response element (106). PDE4 inhibition potentiates β₂AR-stimulated cAMP levels to increase MKP1 messenger RNA (mRNA) and protein levels. MKP1 promotes the degradation of proinflammatory transcripts by dephosphorylating mitogen-activated protein kinases.PDE4 inhibition also promotes bronchodilation by reducing bronchial smooth muscle cell contraction (107).

Figure 4.

Mechanisms of PDE4 inhibitors in immune diseases. Indicated uses and effects are listed for three different PDE4 inhibitors in different formulations. The cell type targets and known mechanisms are summarized. Abbreviations: β₂AR, β₂-adrenergic receptor; cAMP, cyclic adenosine monophosphate; CRE, cAMP response element; IL-2, interleukin-2; MAPK, mitogen-activated protein kinase; MKP1, mitogen-activated protein kinase phosphatase 1; NF-κB, nuclear factor κB; PDE, phosphodiesterase; Teff, effector T; Treg, regulatory T.

These therapies nevertheless have some limitations in that they have systemic side effects such as diarrhea, emesis, and headache, and relatively weaker potency as compared to alternative treatments (102).Side effects from these PDE4 inhibitors occur largely due to drug actions on multiple other tissues and/or lack of specificity for isoforms within the subfamily. Such intrinsic limitations are common to PDE inhibitors as they are designed to competitively block substrate binding to the hydrophobic pocket within the catalytic domain (42) and have no means of targeting specific systems. Importantly, the three FDA-approved PDE4 drugs do not discriminate among the major PDE4A–D isoforms (108).

New inhibitors that can preferentially target PDE4B or have limited ability to cross the blood-brain barrier are under investigation in recognition of the distinct roles of PDE4 isoforms.PDE4B is generally considered the major anti-inflammatory regulator (109, 110), while brain PDE4D in nonadrenergic neurons is likely to induce emesis (111). Difamilast, a PDE4B-selective inhibitor, has shown strong efficacy in patients with atopic dermatitis (112), with substantiated results in a mouse model of the disease (113). ME3183 is another potent PDE4 (including PDE4B) inhibitor that has limited penetration to the central nervous system at usual pharmacologic doses (114),and it is being evaluated to treat plaque psoriasis (NCT05268016).

Additional efforts are evaluating the efficacy of targeting dual or different PDEs: Dual PDE4/7 targeting may have synergistic effects as the two enzymes play overlapping roles in T cell IL-2 production and proliferation (115). This approach, however, has had mixed results regarding efficacy in various cell lines and preclinical models (116). Nevertheless, the dual inhibitor YM-393059 has shown a lower potential emetic effect, as assessed by α2-adrenoceptor agonist–induced sleep duration in mice—animals that do not have an emetic response (117). Evidence also supports a role for cAMP-hydrolyzing PDE8A in the innate immune response (32). In contrast to the antiproliferative effects induced by PDE4 inhibition, that of PDE8A limits T cell motility and adhesion, revealing a potentially new immune modulatory approach (118).

PDE4: Potential Therapeutics in Alzheimer’s Disease and Cancer

Cerebral neuronal physiology (e.g., neuroplasticity) and pathophysiology [amyloid-β peptide plaque formation in Alzheimer’s disease (AD)] rely on cAMP signaling (119) to shape memory, learning, and decision making. Clinical findings have suggested that PDE4 inhibition may provide a means to counter dementia in AD. Pathologic increases in PDE4D1, D3, D5, and D8 mRNA levels in the middle temporal gyrus were observed in patients with AD (26). Of these, PDE4D1 and 4D3 levels have been associated with increased amyloid-β plaque formation (26). The PDE4 inhibitor roflumilast improved memory performance in otherwise healthy elderly subjects (120) and was extended to AD patients in two Phase II trials (NCT00362024 and NCT03817684). The first was completed in 2007, but the results were not disclosed. Results are pending for the second study, which was completed in 2023. The latter study tested the investigational agent BPN14770, an allosteric, potent PDE4D-selective modulator that by design inhibits 80–90% of enzymatic activity. BPN14770 was 500–3,000 times less likely than rolipram to be emetic in several preclinical models, including canines and primates (121), and did not induce significant adverse effects in adults with fragile X syndrome (122). Recently, it was shown that long forms of PDE4D, specifically D3, D5, D7, and D9, regulate neuronal plasticity while localizing to particular cellular compartments (123).Future design of inhibitors that are PDE4D splice variant–specific could offer groundbreaking innovation toward targeted treatment of neurocognitive disease.

Oncogenic mutations have been found in multiple genes that encode proteins of cAMP signaling pathways, and PDEs may be potential targets for cancer treatment (124). PDE4 subfamily members, especially 4B and 4D, have been implicated in multiple cancer types, including those affecting the hematologic, skin, or central nervous system. Multiple tumor cell lines and human cancerous tissue show increased PDE4 isoform expression levels (125). A recent study, for example, reported that increased tumor PDE4D expression is associated with worse outcomes in pancreatic cancer patients (126). In a mouse model of pancreatic cancer, inhibiting PDE4D was antioncogenic by inhibiting mammalian target of rapamycin complex (mTORC1) signaling via PKA phosphorylation of Raptor (126). The general PDE4 inhibitor roflumilast and the PDE4D-specific inhibitor GEBR-7b were both effective in these mice. Antioncogenic efficacy of GEBR-7b has also been reported in organoids and mice with estrogen receptor–positive breast cancer that is resistant to standard of care (endocrine therapy, chemotherapy, and cell cycle inhibitors). Although still early in development, these findings demonstrate that targeting PDE4D may be effective at inhibiting tumor growth across different cancer types.

THERAPEUTICS LEVERAGING cGMP REGULATION

PDE5: Circulatory System Diseases

A serendipitous observation in patients with ischemic chest pain suggested that PDE5 inhibition might enhance penile erections (127). PDE5 inhibition did not prove efficacious for ischemic heart disease, but it would go on to become successful in treating erectile dysfunction (ED). The FDA approved the use of PDE5 inhibitors for ED starting in 1998 for sildenafil, and tadalafil five years later, followed by vardenafil and avanafil. A meta-analysis of 48 clinical trials showed that the efficacy of these PDE5 inhibitors is uniform across age groups in treated men (128).Sildenafil and tadalafil have the greatest therapeutic efficacy and fewer adverse side effects (129). Patients with underlying cardiac conduction abnormalities or those taking PDE3 inhibitors, however, should not receive vardenafil to avoid risks in QT interval prolongation (130). Sildenafil and tadalafil are also widely used to treat pulmonary arterial hypertension (PAH) and have become a mainstay of therapy. PAH is associated with impaired endogenous NO production in the pulmonary vascular bed (131),and treatment with PDE5 inhibitors reduced pulmonary vascular resistance to improve clinical symptoms.

PDE5 is highly expressed in smooth muscle cells and blood vessels, particularly those in the pulmonary vasculature and corpus cavernosum (81). In vascular smooth muscle cells of systemic vessels, PDE5 regulates cGMP levels coupled to NO stimulation. NO diffusion into vascular smooth muscle increases GC-1 generation of cGMP and PKG activation. The smooth muscle relaxes upon PKG activation by a reduction in intracellular calcium or in myofilament sensitivity to calcium (132). However, this vasodilatory effect is countered by PDE5 since cGMP binding to GAF A domains and PKG phosphorylation of PDE5 at serine 92 activate the enzyme’s hydrolytic activity (81). PDE5 inhibition thus prolongs the vasomotor and signaling actions of NO, enhancing its vasodilatory effects. In the pulmonary vasculature, however, PDE5 regulates NP/GC-A/B signaling pathways. This regulation is specific to the pulmonary versus systemic vessels (133).This is of relevance to PAH, given that NP-mediated signaling assumes a greater role over NO/GC-1 in hypoxic conditions (134). Pulmonary-specific combined PDE5/NP signaling enhancement may therefore have potential additional benefits in the lung.

Potential Therapeutics in Heart Failure

In the heart, cGMP-hydrolyzing PDE5A and 9A mediate a variety of effects that are nearly exclusively linked to activating PKG signaling (Table 3). Localized to distinct domains, these PDEs regulate different compartmentalized cell signaling pathways in myocytes: PDE5A regulates the NO-coupled cGMP pool, and PDE9A regulates that coupled to NP signaling. PDE5A localizes to the sarcomere Z-disk region but can take on a more diffused cytosolic distribution in conditions such as HF, as modeled in mouse hearts subjected to chronic pressure overload (80). In cardiomyocytes, PDE9A is found in mitochondrial subfractions and localizes to the organelle based on gold-labeled immunostaining in transmission electron microscopy (37). PDE9A also colocalizes with SERCA2A (28).

Table 3.

Cardiac PDEs and their relevant mechanisms in heart failure

PDE	Function	Compartment(s)	Signaling	Reference(s)
PDE5	Structure	Myofilament; membrane	NO/GC-1/cGMP/PKG/ calcineurin/NFAT/TRPC6 Amphiphysin II (BIN1)	28, 80, 135, 142, 148
	Protein quality control and autophagy	Lysosome; proteasome	PKG/TSC2/mTORC1 PKG/CHIP/Hsc70	29, 30, 144
PDE9	Structure	SR	NP/GC-A/B/cGMP/PKG/PLN/ SERCA2A	156
	Metabolism	Mitochondria	NP/GC-A/B/cGMP	37
		Nucleus	PPARα	37
PDE3	Contractility	Plasma membrane; SR	cAMP/PKA/LTCC cAMP/PKA/PLN/SERCA2A	13, 20, 157
	Relaxation	SR	cAMP/PKA/PLN/SERCA2A	20, 157
PDE1	Contractility	Plasma membrane	cAMP/PKA/LTCC	13
	Relaxation	Caveolin-associated membrane; cytosolic	NO/GC-1/cGMP CNP/GC-B/cGMP	10

Abbreviations: BIN, bridging integrator 1; cAMP, cyclic adenosine monophosphate; CHIP, carboxyl terminus of Hsc70-interacting protein; CNP, C-type natriuretic peptide; cGMP, cyclic guanosine monophosphate; GC, guanylyl cyclase; Hsc70, heat shock cognate 71-kDa protein; LTCC, L-type calcium channel; mTORCl, mammalian target of rapamycin complex 1; NFAT, nuclear factor of activated T cells; NO, nitric oxide; NP, natriuretic peptide; PDE, phosphodiesterase; PKA, protein kinase A; PKG, cGMP-stimulated protein kinase; PLN, phospholamban; PPARa, peroxisome proliferator–activated receptor α; SR, sarcoplasmic reticulum; TRPC, transient receptor potential cation channel subfamily C, member 6; TSC2, tuberous sclerosis complex subunit 2.

PDE5 inhibition with sildenafil protects the mouse heart against pathologic remodeling and dysfunction in an NO- and PKG oxidation–dependent manner (28, 135). Specific cardiac effects of PDE5 inhibition include antihypertrophic (136, 137), antifibrotic (138), enhanced proteostasis (29), and proautophagic (30, 139, 140) responses. These findings have been attributed to PKG activation and phosphorylation of regulator of signaling proteins 2 and 4 (141), transient receptor potential channel canonical type 6 (142, 143), tuberous sclerosis type 2 protein (30, 144), and carboxyl terminus of Hsc70-interacting protein (CHIP, or Stub1) (29). Additionally, influences of PDE5 inhibition on blunting adrenergic responsiveness have been reported in mice, dogs, and humans, where they were linked to NO-dependent signaling (145–147). In a sheep model of tachypacing-induced cardiomyopathy, PDE5 inhibition improved T-tubular structure, which was associated with improved compartmentalized adrenergic signaling (148).

Clinical evaluation of PDE5 inhibitors in patient groups with different forms of HF, however, has not found consistent efficacy. PDE5 inhibition in patients with heart failure with reduced ejection fraction (HFrEF) yielded mixed results, though early studies in HFrEF patients with type 2 pulmonary hypertension looked promising (81). Longer-term, adequately powered controlled studies in HFrEF with or without type 2 pulmonary hypertension have either yet to be performed or yielded inconsistent results. No clinical improvements were observed in additional trials evaluating PDE5 inhibitors in patients with heart failure with preserved ejection fraction (HFpEF) (149) or Duchenne’s muscular dystrophy (150). To date, the HFpEF study remains the largest placebo-controlled, chronic-therapy trial testing PDE5 inhibition in any form of HF.

NO-related signaling is often depressed in HF in part due to oxidative stress (151), whereas cGMP/PKG signaling associated with NP stimulation is often increased. Accordingly, the impact of PDE5 inhibitors in HF may be limited, whereas PDE9 inhibitors might be more effective (Table 3). Preclinical studies found that PDE9 inhibition conferred antihypertrophic and antifibrotic effects in a pressure-overload mouse model (28) and improved mitochondrial fatty acid oxidation while promoting lipolysis in a cardiometabolic syndrome/obesity model (37). Improvement in myocardial compliance has additionally been reported (152). Evidence supporting the selectivity of PDE9 for NP-dependent signaling in the heart was first observed when PDE9 inhibition remained effective in countering pressure-overload pathobiology in mice despite inhibiting NO synthase. This finding contrasted with PDE5 inhibition that was effective so long as NO synthase was not inhibited (28). In a mouse model of cardiometabolic syndrome, PDE9 inhibition reduced diet-induced obesity by lowering total body and brown fat, while increasing mitochondrial fatty acid oxidation and lipolysis in a peroxisome proliferator-activated receptor alpha (PPARα)-dependent manner. Interestingly, this only occurred in male and ovariectomized female mice (37). The lack of impact in female mice with intact ovaries may be due to redundant control of PPARα signaling by estrogen. These results expand the potential therapeutic utility of PDE9 inhibition to include obesity, a major comorbidity found in many HF patients, although potential sex-dependent dimorphism should be considered as trials are pursued. Several subacute sheep HF studies (several days of tachypacing) have explored PDE9 inhibitory effects on cardiorenal and circulatory function and reported synergy between PDE9 inhibition and NP coadministration or neprilysin inhibition (153–155). A Phase IIa trial evaluating the PDE9 inhibitor CRD-740 in HF patients recently reported that the drug successfully elevated cGMP levels in treated patients. Phase IIb multinational trials for HFrEF (NCT06215911) and HFpEF (NCT06215586) are underway.

PDE5/PDE9: Potential Therapeutics in Cancer

For patients with esophageal adenocarcinoma, survival is inversely correlated with esophageal PDE5A expression level (158). Adenocarcinoma development is in part driven by microenvironment changes induced by tumor-associated fibroblasts. In a mouse model of the disease, PDE5 inhibition with vardenafil or tadalafil combined with chemotherapy reduced tumor mass and suppressed transdifferentiation of cancer-associated fibroblast (158). Fibroblasts are the predominant mesenchymal connective tissue cell type (128), and fibroblast-specific PDE inhibition may offer a conceptual case study to treat diseases with underlying maladaptive changes in tissue repair. Increased PDE5A expression is also predictive of lower survival in patients with colorectal cancer (CRC), with in vitro and in vivo studies in preclinical models showing that PDE5 inhibition induces cancer cell apoptosis while inhibiting cell growth (159). In fact, use of sildenafil reduces the risk of cancer in men who are diagnosed with benign colorectal neoplasm (129). A similar reduction in cancer risk was also found in men with ED who were prescribed PDE5 inhibitors (129). Reduced PKG expression in CRC (160) might blunt benefits from PDE5 inhibition, and yet this approach remains of interest (161). Finally, PDE5A and PDE9A expression are increased in human breast carcinoma (162), and PDE9 inhibition is pro-apoptotic as shown in cancer cell lines (40). However, the antioncogenic potentials of PDE9 inhibitors have yet to be evaluated in vivo.

THERAPEUTICS LEVERAGING cAMP/cGMP REGULATION

PDE3A: Intermittent Claudication

PDE3 is a dual PDE with 4–10 times greater hydrolytic activity for cAMP compared to cGMP. All three approved PDE3 inhibitors (cilostazol, amrinone, and milrinone) target both the PDE3A and 3B isoforms. However, their clinically recognized effects are largely thought to be mediated by PDE3A, which is expressed in platelets, smooth muscle cells, and the myocardium. PDE3B regulates lipolysis, but its therapeutic potential for diabetes has not yet been tested (163).PDE3A-mediated cAMP signaling regulates platelet function and vascular tone (164). Ligands such as adenosine and certain prostaglandins induce PDE3-dependent platelet activation by acting upon prothrombotic thromboxane A2, phospholipase D1, and cyclooxygenases (165). PDE3 regulates vascular tone via cAMP regulation, in part through Epac activation (166).

Intermittent claudication is a peripheral arterial disease that restricts blood flow to muscles when they are active. The resulting supply/demand mismatch causes pain and impairs muscle function during ambulation (167). Cilostazol (FDA approved in 1999) can abate these symptoms by targeting PDE3 to diminish platelet activation and to promote vasodilation. Inhibition of cGMP-hydrolyzing PDE5A, which is also highly expressed in platelets, does not show efficacy in claudication patients (168). This suggests that although PDE3 is a dual-substrate enzyme, the therapeutic effects are likely mediated by cAMP. Due to the potential adverse cardiovascular effects and modest efficacy, however, cilostazol use is limited to patients who remain symptomatic despite lifestyle modifications (169). Cilostazol may have specific applications in patients with coronary artery disease who are refractory to dual antiplatelet therapy (aspirin and clopidogrel). Platelet reactivity is increased in these refractory patients, who are more susceptible to adverse ischemic events. In a randomized study, cilostazol suppressed platelet reactivity following 30 days of treatment (170). Aspirin, which is commonly used as an antiplatelet medication, induces overexpression of MRP4, the nucleotide transporter that additionally extrudes aspirin. Besides inhibiting PDE3, cilostazol reduces platelet activation by antagonizing MRP4 function to increase intracellular cAMP levels (171). Nevertheless, cilostazol, is not commonly used for this patient population.

PDE3A: Acute Decompensated Heart Failure

PDE3 is a predominant microsomal cAMP hydrolyzer in larger mammalian (e.g., rabbit, dog, and human) hearts (172, 173).Unlike cilostazol, which elevates cAMP levels primarily in smooth muscle cells and platelets, the principal effects of inotropic PDE3 inhibitors (amrinone and milrinone) are to increase cardiomyocyte and smooth muscle cell cAMP levels (174). The notable inotropic effects of amrinone in HF patients and canine and feline models without inciting arrhythmias (175, 176) led to FDA approval of the drug in 1984. Milrinone is 20 times more potent than amrinone, acutely lowers afterload and preload to enhance hemodynamics, and has positive inotropic effects in patients with severe HF (177, 178). Sustained use (37 days) in patients with severe systolic HF led to persistent improvements in hemodynamics and cardiac function (178).

However, significant side effects from milrinone emerged, including increased ventricular arrhythmias and, in some instances, sudden cardiac death (178). Subsequent evaluation of the long-term efficacy and safety of milrinone in severe HF found worsened mortality in the treatment group (179). These findings were reminiscent of the sudden cardiac deaths and complex electrophysiologic effects caused by the drug in dogs with HF (180, 181).Years later, the efficacy of PDE3 inhibition using another drug, enoximone, was evaluated in HF patients who were receiving concurrent β-AR blockers. The proposed hypothesis was that excessive cAMP signaling associated with PDE3 inhibition would be dampened by concomitant β-adrenergic blockade (182).The study also used lower doses of the PDE3 inhibitor enoximone than had been previously tested. However, there were no clinical improvements in this study (182). To date, PDE3 inhibitors are used to treat acute decompensated systolic HF and as palliative or bridging treatment for advanced HF patients.

Selective inhibition of PDE3A may offer a safer means yet to increase cardiac contractility (Table 3). PDE3A modulates contractility by regulating cAMP-mediated signaling at the SR by binding to phospholamban and SERCA2A (20). Deletion of PDE3A but not PDE3B blocks the inotropic effects of milrinone in mice (157, 164). To overcome the challenges of designing an inhibitor that can discriminate among PDE3 isoforms and splice variants, a small peptide that disrupts PDE3A-SERCA2A binding has been developed. The disruptor peptide works as a SERCA2Aactivator—independent of PKA activity and even in the absence of phospholamban, the endogenous inhibitory regulatory protein. Delivering the peptide to the myocardium improved heart contractility and reduced mortality in a murine model of HF (183). Peptide disruptors have also been designed to target PDE4 in multiple diseases, including hypertrophic cardiomyopathy (184). While further clinical testing awaits, these recent studies showcase peptide disruptors as novel therapeutic tools.

Potential Therapeutics: PDE1 and Heart Failure

Lenrispodun—a highly specific, bioavailable PDE1 inhibitor (185)—has vasodilating and inotropic effects in rabbits, dogs with HF, and HFpEF patients (186, 187) (Table 3). PDE1 inhibition may hold promise in HF treatment because its downstream effects on PKA activation and calcium homeostasis are different compared to PDE3 inhibition. Unlike the inotropic and lusitropic effects from PDE3 inhibition, those from PDE1 inhibition persist despite concomitant β₁-adrenergic blockade (186). PDE1C coimmunoprecipitates and colocalizes with adenosine type 2 receptors (93), and blocking the adenosine 2B receptor abrogates the inotropic effects of PDE1 inhibition (186). PDE1 inhibition increases L-type calcium channel activation, akin to that of PDE3.However,whereas PDE3 inhibition also augments SR calcium reuptake and release, this is minimal with PDE1 inhibition (13). This difference correlated with fewer spontaneous calcium release events and associated arrythmias in cardiomyocytes subjected to PDE1 compared PDE3 inhibition (13).

Whereas PDE3A expression decreases in HF (188), PDE1C expression increases in dilated cardiomyopathy (186). PDE1 inhibition or deletion is also beneficial by opposing cell apoptosis, hypertrophy, and fibroblast transactivation in mouse models of HF (189–191).Together with contractile and lusitropic effects, these diverse effects opposing pathologic structural remodeling of the heart may reflect PDE1 regulation of both cAMP and cGMP. The contractile effects are in part mediated via L-type calcium channel activation in a PKA-dependent manner (13). In mice, cGMP production downstream of either NP or NO stimulation mediates the lusitropic effects of PDE1 inhibition (10). cGMP signaling is also important for antiapoptotic effects, whereas both cGMP and cAMP help to protect against fibrosis and hypertrophy (189–191). Species-dependent differences in PDE1 cell signaling, however, may in part explain the multiple effects of PDE1 inhibition. The inotropic effects, for instance, are more prominently observed in larger mammals that primarily express the cAMP-favoring PDE1C isoform (13). In contrast, mouse and rat hearts express mostly PDE1A,which favors cGMP hydrolysis. The combined effects of PDE1 inhibition may nevertheless offer critical advantages when compared to PDE3 inhibition. Lenrispodun was well tolerated in HFrEF patients in a small Phase Ib/II study (187). While active evaluation of lenrispodun for HF is in a hiatus for now, the drug is being tested for potential benefit in patients with Parkinson’s disease (NCT05766813).

PDE10A: Potential Therapeutics in Cardio-oncology

The dual cAMP- and cGMP-hydrolyzing PDE10 has emerged as a modulator of chemotherapy efficacy and associated cardiotoxicity. Selective inhibition or genetic deletion of PDE10 enhances the efficacy of chemotherapy in multiple cancer cell lines. Concurrently, PDE10 inhibition limits cardiotoxicity in a mouse model of ovarian cancer (192). Because PDE10A expression is particularly high in striatal medium spiny neurons (MSNs), several inhibitors targeting PDE10 have been evaluated for their antipsychotic efficacy in schizophrenia. While these reagents have not shown clinical benefit (see below), the drug appears to be safe (193). Whether PDE10 inhibition may have utility as cardio-oncologic therapeutics remains to be tested.

Potential Therapeutics: Multiple PDEs in Neurological and Psychiatric Disorders

PDE10A expression in the brain is exclusively found in the postsynaptic membranes of MSNs, with high levels in indirect versus direct pathway cells (40). Striatal MSNs process dopamine-encoded signaling (194).Dopaminergic circuitry affects human behavior and decision making, and multiple lines of evidence have shown that dopamine levels or neuronal sensitivity to dopamine is increased in individuals with schizophrenia or during psychosis (194). The role of PDE10A in this circuitry is complex because of the multitiered nature of neuronal circuitry. Dopamine release can increase or decrease cAMP production by acting on D₁ and D₂ receptors, respectively expressed in direct- and indirect-pathway MSNs. By stimulating NO production, dopamine can also increase cGMP levels, which would be hydrolyzed by PDE10A. NO can have a localized yet pervasive effect through its free diffusion across MSNs. Antipsychotic drugs, which antagonize dopamine D₂ receptors, are the standard-of-care therapy to reduce positive symptoms in schizophrenia and are thought to increase cAMP levels in the indirect MSNs (194). Early studies showed that PDE10 inhibition mimicked these antipsychotics by increasing cAMP levels. However, clinical trials have failed to show efficacy of PDE10 inhibition compared to antipsychotics in patients with schizophrenia (193).

PDE1B is also highly expressed in the striatum, where the isoform’s expression pattern closely follows that of D₁ receptors (195). Dual inhibition of PDE1B and 10A in a rat model of schizophrenia showed comparative or superior improvement (amelioration of negative symptoms and improved recognition memory) compared to an antipsychotic (179). Multiple PDE targeting may therefore offer advantages with improved therapeutic efficacy.

The nonselective PDE inhibitor ibudilast (MN-166,targeting PDE3,4,10,and 11) is approved in Japan and Korea for asthma (119). Recent trials of ibudilast in patients with multiple sclerosis (a chronic autoimmune disease with variable neurological features) and amyotrophic lateral sclerosis (ALS, a fatal disease that leads to progressive loss of control over voluntary muscles) showed promising outcomes. Ibudilast treatment slowed thalamic region brain atrophy, progression of neurodegeneration, and the onset of disease in patients with multiple sclerosis (196). Initial trials of ibudilast lacked generalized efficacy for ALS patients, owing to small cohort size (197) or differences in disease history between groups (198). Subsequent analysis, however, succeeded in identifying a subgroup of patients in whom ibudilast may be effective (199). Based on these findings, ALS patients with upper limb or bulbar onset and a disease history of 18 months or less will be given ibudilast in conjunction with the standard riluzole (glutamate blocker) treatment (NCT04057898) (199).

CONCLUSION

Numerous physiologic responses rely on PDE regulation of cyclic nucleotide-mediated signaling pathways. PDE inhibitors are used to treat a broad array of clinical diseases, and ongoing studies may expand this to HF, cancer, and neurologic diseases. The dominant strategy has been to block the catalytic site of cAMP or cGMP nucleotide hydrolysis, which provided selectivity between family groups but not among isoforms. Different isoforms localize to varying compartments with distinct signaling features, implying that the full potentials of PDE pharmacology have yet to be realized. Allosteric modifiers of PDEs, disruption of protein-protein binding to dislodge a PDE from its engaged signalosome, and multiple PDE–targeting therapies are new directions that may improve both the breadth and promise of PDE modulation for human disease.