Phosphodiesterase inhibitors as emerging therapeutics for skeletal disorders: A comprehensive review of mechanisms and repurposing potential

Abstract

Skeletal disorders such as osteoporosis, osteoarthritis, and rheumatoid arthritis represent major global health burdens with limited therapeutic innovation. Inhibitors of phosphodiesterases (PDEs), enzymes that regulate the intracellular levels of the cyclic nucleotides cAMP and cGMP, are increasingly recognized as potential medications that can improve the health of bone and cartilage. Although they have been used in clinics for decades, their function in bone or cartilage remains unclear.

In preclinical models, inhibitors that target PDE3 (e.g., cilostazol, milrinone), PDE4 (roflumilast, apremilast), and PDE5 (e.g., sildenafil, avanafil, vardenafil) exhibit promising anabolic, anticatabolic, and anti-inflammatory effects on skeletal tissues. Broad-spectrum inhibitors such as pentoxifylline and dipyridamole also demonstrate dual benefits in terms of bone regeneration and joint preservation.

This review examines the mechanistic basis and therapeutic potential of clinically approved PDE inhibitors—originally developed for cardiovascular, neurological, and inflammatory conditions—for skeletal applications. Understanding the mechanism of action of PDE inhibitors can facilitate their translation into the clinic, help with their application in combined therapies, or minimize their potential adverse effects. This approach offers a cost-effective and viable path toward novel therapies for musculoskeletal health.

Highlights

• •There are limited therapeutic options for the treatment of skeletal disorders.
• •Inhibitors of phosphodiesterases (PDEs) could be repurposed for this objective.
• •Drugs targeting PDEs 3, 4 and 5 had positive effect in animal models.
• •The mechanism is complex and not fully understood.

1. Introduction

Skeletal disorders such as osteoporosis, osteoarthritis (OA) and rheumatoid arthritis (RA) are leading causes of pain, disability, and healthcare burden worldwide. Despite the availability of antiresorptives, synthetic anabolics, and joint replacement surgery, there is a constant effort for another treatment methods. The development of new drugs is a lengthy and costly process with no guarantee of success. These limitations have intensified interest in repurposing agents with established safety profiles, pharmacology, and manufacturability. Among the most attractive candidates related to the health of the skeletal system are the phosphodiesterase (PDE) inhibitors, which have been used in clinical practice for decades (Baillie et al., 2019).

PDEs regulate the intracellular levels of 3′,5′-cyclic adenosine monophosphate (cAMP) and 3′,5′-cyclic guanosine monophosphate (cGMP), which participate in signaling pathways that control bone and cartilage homeostasis. For example, cAMP activates protein kinase A (PKA), which modulates the activity of RUNX family transcription factor 2 (RUNX2), a master transcription factor that regulates osteoblast differentiation (Selvamurugan et al., 2009; Komori, 2024). cGMP, on the other hand, activates protein kinase GII during skeletogenesis, which mediates osteoblast mechanotransduction and PKGII null mice display dwarfism (Pfeifer et al., 1996; Rangaswami et al., 2009). Thus, manipulating cAMP/cGMP stability through PDE inhibition can directly affect the cellular signaling pathways involved in skeleton formation and homeostasis. Importantly, the same pathways also regulate catabolic processes in bone and cartilage (Houard et al., 2013), suggesting that the suppression of PDEs could exert dual anabolic and anticatabolic effects in age-related skeletal disorders.

Most PDE inhibitors approved for clinical practice are used as therapeutics in the context of the central nervous system, cardiovascular system, reproduction, cancer and metabolic disorders (Baillie et al., 2019). However, increasing evidence suggests their potential utility in musculoskeletal medicine (Porwal et al., 2021). For example, the cAMP level in the synovial fluid negatively correlates with the inflammatory burden of the joint, and cAMP enhances anti-inflammatory responses in patients with RA and OA (Morovic-Vergles et al., 2008). Thus, the cAMP level contributes to joint health in a manner that is more complex than simple regulation of RUNX2 activity.

In this review, we summarize how cAMP/cGMP are involved in cell signaling that affects bone formation, which PDEs can regulate their level, and how PDE inhibitors affect bone growth and cartilage health. On the basis of the results from different cell and animal models, we suggest a general mechanism by which specific classes of inhibitors affect bone and cartilage. Finally, we suggest which PDE inhibitors could be repurposed for the treatment of osteoporosis, RA, and other skeletal conditions.

2. Signaling pathways involved in bone formation

Skeletogenesis is complex process involving interplay of numerous signaling pathways, which is in detail described elsewhere (Zhu et al., 2024). Here we briefly outline the main processes and list the pathways that can be affected by cAMP or cGMP signaling.

Human bones develop through two primary mechanisms, intramembranous and endochondral ossification, both of which are initiated by the condensation of mesenchymal progenitor cells.

The majority of human bones are created through the process of endochondral ossification, where mesenchymal condensates first differentiate into chondrocytes, forming a cartilage scaffold that is later replaced by bone (Berendsen and Olsen, 2015). This process requires the transcription factor SOX9 for differentiation into the chondrogenic lineage and bone morphogenetic protein (BMP) receptors 1A and 1B for chondrogenesis and maintenance of SOX9 expression (Kronenberg, 2003; Berendsen and Olsen, 2015). Proliferating chondrocytes exit the cell cycle and differentiate into prehypertrophic chondrocytes, which are characterized by the expression of Indian Hedgehog (IHH) and parathyroid hormone (PTH)/PTH-related peptide 1 receptor (PTH1R). SOX9 is further required for maintaining chondrocyte identity but must be downregulated for progression to hypertrophy (Ikegami et al., 2011). The RUNX2 and RUNX3 transcription factors are necessary for chondrocyte maturation and indirectly function to increase chondrocyte proliferation by promoting IHH expression (Fujita et al., 2004). The balance between chondrocyte proliferation and hypertrophy is regulated by interactions between several signaling pathways, including IHH, PTH-related peptide (PTHrP), BMP, WNT, and fibroblast growth factor (FGF)(Ornitz and Marie, 2015). FGF signaling has vital role in bone development as missense mutations either in numerous FGFs or their receptors (FGFRs) can cause various congenital bone diseases in humans, including chondrodysplasia syndromes, craniosynostosis syndromes and syndromes with dysregulated phosphate metabolism (Su et al., 2014). Binding of FGFs to FGFRs causes activation of MAPK, phosphatidylinositol 3-kinase/protein kinase B (PI3K/PKB) and phospholipase C γ (PLCγ) signaling (Su et al., 2014). Apart from FGF signaling, chondrocyte maturation into hypertrophy is also modulated by calcium/calmodulin-dependent protein kinase II (CAMKII), which increases RUNX2 and β-catenin activity in response to WNT and PTHrP pathways stimulation (Li et al., 2011; Ornitz and Marie, 2015). Chondrocyte hypertrophy generates an expansive force that results in bone elongation (Berendsen and Olsen, 2015). Hypertrophic chondrocytes also begin to mineralize their extracellular matrix and have the capacity to differentiate into osteoblasts within the primary spongiosa (Kronenberg, 2003; Berendsen and Olsen, 2015). Through the expression of vascular endothelial growth factors (VEGFs), hypertrophic chondrocytes attract endothelial cells, osteoprogenitor cells, and osteoclasts, which remodel and organize the chondrogenic matrix to form the primary ossification center and eventually bone (Liu and Olsen, 2014).

In the process of intramembranous ossification, mesenchymal cells differentiate directly into osteoblasts to form flat bones of the skull and parts of the clavicle. The early commitment of mesenchymal stem cells to osteoblasts requires the expression of RUNX2, a master transcription factor that regulates the expression of key genes in osteoblasts, such as type I collagen, osteopontin (OP), osteocalcin (OC), and transforming growth factor β (TGF-β) (Ornitz and Marie, 2015). The regulation of intramembranous bone formation involves several factors, including TGF-βs, BMPs, FGFs, and WNTs, all of which have been shown to control cell differentiation and survival in a spatiotemporal manner (Kozhemyakina et al., 2015).

Apart from classical cell signaling pathways, physical force also governs proper development and maintenance of the skeletal system. The sensing of mechanical load is mediated by Piezo cationic channels, which are expressed in osteoblasts and chondrocytes and which are essential for proper bone growth and homeostasis (Lee et al., 2014; Zhou et al., 2020; Hendrickx et al., 2021; Qin et al., 2021). Piezo channels act partially through the transcriptional co-activator with PDZ-binding motif (TAZ) and yes-associated protein (YAP). At the level of YAP/TAZ, the perception of mechanical stress crosstalk with Wnt, TGF-β and BMP signaling to drive osteogenic differentiation (Wei et al., 2020; Li et al., 2024).

Throughout life, bone homeostasis is maintained by a dynamic balance between osteoblast-driven formation and osteoclast-mediated resorption, where receptor activator of nuclear factor κB (RANK)/osteoprotegerin (OPG) signaling plays a pivotal role (Wu et al., 2024). RANK is a transmembrane protein expressed on the surface of osteoclasts and their precursors, and it can be activated upon interaction with the RANK ligand (RANKL), which is exposed on osteoblasts. This interaction activates downstream signal transduction cascades, including mitogen-activated protein kinase (MAPK) and nuclear factor (NF)-κB, resulting in the amplification of NFAT transcription factors. NFATs then promote osteoclastogenesis and eventually bone resorption. The interaction of RANK with RANKL also activates reverse signaling in osteoblasts and promotes the RUNX2 osteogenic program. OPG serves as a decoy receptor for RANKL produced by osteoblasts, which blocks the interaction between RANKL and RANK and thus prevents the formation of osteoclasts and the subsequent breakdown of bone (Wu et al., 2024).

With aging, this balance is disrupted: osteoblastic activity decreases, whereas osteoclastic bone resorption further stimulated by inflammatory signaling prevails (Majidinia et al., 2018). In addition, estrogen supports OPG expression, suggesting a connection between estrogen postmenopause dropdown and the development of osteoporosis due to decreased OPG levels (Wu et al., 2024).

All these steps that coordinate bone formation, growth, and remodeling can be directly or indirectly affected by signaling pathways that use cAMP/cGMP as secondary messengers.

2.1. cAMP signaling in bone

In the skeletal system, cAMP can be produced by soluble adenylyl cyclases (ACs), which are activated by bicarbonate (Wiggins et al., 2018), or by membrane ACs downstream of G-protein coupled receptors (GPCRs), such as calcitonin receptor or prostaglandin E2 (PGE2) receptors (Martin, 2021). However, the most relevant source of cAMP in the context of bone is the signaling mediated by the GPCR PTH1R.

PTH1R is the sole receptor for PTH and PTHrP, two peptide hormones that are key regulators of calcium homeostasis and chondrocyte proliferation and differentiation, respectively (Gensure et al., 2005). PTH1R is essential for proper development of the skeletal system, as inactivating mutations result in a perinatal lethal disorder called Blomstrand's chondrodysplasia, which is characterized by short limbs, advanced dentition, and osteosclerosis (Karaplis et al., 1994; Jobert et al., 1998; Zhang et al., 1998). Ligand binding to the receptor activates AC through the G protein, which leads to an increase in the intracellular level of cAMP. cAMP then activates mainly exchange proteins directly activated by cAMP (EPAC)(de Rooij et al., 1998) and PKA, which, among other proteins, phosphorylate and thus activate RUNX2 and cAMP response element-binding protein (CREB) to regulate the osteogenic gene program (Datta and Abou-Samra, 2009; Selvamurugan et al., 2009). CREB is a key mediator of PTH1R/cAMP/PKA signaling in chondrocytes, as its dominant negative variant in chondrocytes leads to short-term dwarfism due to a delay in chondrocyte hypertrophy (Long et al., 2001).

cAMP/PKA signaling also modulates other signaling pathways important for bone development. cAMP can enhance BMP signaling through PKA-CREB and MAP-kinase phosphatase 1 (MKP1)-dependent mechanisms (Ohta et al., 2008; Ghayor et al., 2009). PKA activates an essential regulator of terminal osteoblast differentiation, activating transcription factor 4 (ATF4)(Yang et al., 2004), which interacts with β-catenin and together regulates the expression of terminal osteoblast markers such as osteocalcin (Yu et al., 2013). Stimulation of the Wnt pathway has also a positive effect on bone formation through the suppression of osteoclastogenesis. In osteoclasts, Wnt3A stimulation leads to the activation of canonical β-catenin signaling as well as cAMP accumulation and subsequent PKA activation, which is essential for the suppression of osteoclast differentiation (Weivoda et al., 2016).

Whereas Wnt signaling is potentiated by cAMP/PKA, the Hedgehog pathway is primarily blocked by cAMP/PKA (Yang et al., 2015). During osteogenesis, cAMP/PKA restricts bone formation to the skeleton by inhibiting Hedgehog signaling in mesenchymal progenitor cells (Regard et al., 2013). In mature osteoblasts, Hedgehog signaling controls PTHrP and RANKL expression and thus regulates bone formation and resorption (Mak et al., 2008).

PTH/PTH1R induces two branches of intracellular signaling. Besides AC activation, it initiates PLCγ/Src, which phosphorylates and so stabilizes YAP, resulting in enhanced bone marrow stromal cell differentiation into bone (Monaci et al., 2025). Conversely, cAMP/PKA mediated YAP phosphorylation inhibits YAP activity (Yu et al., 2012; Kim et al., 2013). During osteogenic differentiation, YAP activity is controlled also by pH through cAMP/PKA (Tao et al., 2016).

In addition, cAMP production and induction of cAMP-responsive genes is enhanced by Piezo1 activation, whereas loss of Piezo1 results in increased expression of Pde4a and lower cAMP level, as was described in mouse calvarial cells (Li et al., 2022). In chondrocytes, the cAMP-PIEZO1 interaction is reversed; cAMP elevation inhibits the PIEZO1 channel but mitigates PIEZO1-induced apoptosis of human chondrocytes (Lawrence et al., 2017).

2.2. cGMP signaling in bone

cGMP can be generated either by soluble guanylyl cyclases, which are activated by NO or carbon monoxide (Derbyshire and Marletta, 2009), or by particulate guanylyl cyclases, which serve as transmembrane receptors for natriuretic peptides (Misono et al., 2005; Kuhn, 2009). C-type natriuretic peptide (CNP), which acts via natriuretic peptide receptor B (NPRB), is the most important natriuretic peptide for growth plate cartilage, as disruption of CNP signaling leads to dwarfism, whereas overexpression of CNP results in skeletal overgrowth (Weir et al., 1996; Chusho et al., 2001; Yasoda et al., 2004). NPRB-produced cGMP acts chiefly via PKGI or II, which phosphorylate a number of intracellular proteins including glycogen synthase kinase (GSK)-3β, negative regulator of WNT signaling, or RAF, a component of MAPK signaling (Kawasaki et al., 2008; Kamemura et al., 2017; Hofmann, 2020). In mice, the loss of PkgI in osteoblasts leads to decreased osteogenic activity and poor fracture healing, and the general loss of PkgII impairs endochondral ossification and causes dwarfism (Pfeifer et al., 1996; Schall et al., 2020).

CNP/NPRB/cGMP signaling is the major antagonist of FGF signaling, at the level of the MAPK kinase RAF via PKGII-mediated inhibitory phosphorylation (Krejci et al., 2005). Identification of this regulatory loop have spurred the development of CNP analogs, such as vosoritide, as potential therapies for achondroplasia, the most common form of human dwarfism, which is caused by an activating mutation in FGFR3 (Yasoda et al., 2004; Lorget et al., 2012).

cAMP and cGMP are involved in the regulation of virtually all signaling pathways that contribute to the formation of skeletal tissues (Fig. 1). Their proper utilization in a particular pathway is ensured by PDEs, which restrict cAMP/cGMP to specific signaling pockets and prevent their unintended action over the cell (Maurice et al., 2014). Although there may be some functional redundancy among isoenzymes, most PDE isoforms play specific physiological roles, and some are relevant for skeletal tissues.

Fig. 1.

Simplified overview how cAMP/cGMP affects cartilage and bone. cAMP is produced by adenyl cyclase (AC) and activates mainly protein kinase A (PKA) and cGMP is synthesized by guanylyl cyclases (GC) and activates protein kinase G (PKG), which stimulate (black lines) or interfere with (red lines) signaling of major cell signaling pathways involved in formation and homeostasis of skeletal tissues. ATP, Adenosine Triphosphate; BMP, Bone Morphogenic Protein; BMPR, BMP receptor; CNP, C-type natriuretic peptide; FGF, Fibroblast Growth Factor; FGFR, FGF receptor; FZD, Frizzled; GTP, Guanosine Triphosphate; Hh, Hedgehog; NO, nitric oxide; NPRB, Natriuretic peptide receptor B; PTCH, patched; PTH, parathyroid hormone; PTHrP, PTH related peptide; PTH1R, PTH/PTHrP 1 receptor; c-GC, soluble guanylyl cyclase.

3. PDEs regulate cAMP/cGMP intracellular levels

PDEs are the only enzymes that breakdown the phosphodiester bond in cyclic nucleotides and thus regulate the availability of cAMP/cGMP for downstream signaling. In humans, PDEs are produced from 21 genes, but owing to the use of multiple promoters or alternative splicing, there can be more than one hundred different PDE isoforms (Francis et al., 2011; Azevedo et al., 2014). Due to this variability, each type of cell can hold its own set of PDEs in specific signaling microdomains where they maintain the proper level of cyclic nucleotides (Maurice et al., 2014). This compartmentalization ensures that the same secondary messengers can be used by multiple signaling cascades in one cell at the same time.

On the basis of their substrate specificity or homology of the catalytic C-terminal domain, PDEs are divided into 11 families (PDE1–11), and each family comprises one or more genes named in alphabetic order (e.g., PDE4A-D) (Baillie et al., 2019). Families of PDE4/7/8 selectively hydrolyze cAMP, whereas PDE5/6/9 hydrolyze cGMP. PDE1/2/3/10/11 exhibit dual specificity, acting on both cAMP and cGMP with varying affinities, depending on the isoform (Conti and Beavo, 2007).

3.1. PDEs expressed in bone and cartilage

There are only a few direct studies on PDE expression in the skeletal system and even fewer showing how PDE expression changes during development or with aging. Yet a comprehensive picture can be obtained from publicly available data on the global gene expression profiles of cell cultures and skeletal tissues.

Human and mouse growth plate chondrocytes, as well as chicken limb bud-derived developing chondrocytes, concomitantly express PDEs 1B, 3B, 4B, 4D, 5A, 7A, 8A and 10A (Li et al., 2017; Takács et al., 2023; Kawabe et al., 2025). Two studies collectively reported the presence of transcripts for PDEs 1A, 1B, 4A, 4B, 4C, 5A, 7A, 8A and 8B in human articular cartilage (Wang et al., 2009; Dunn et al., 2016). Grogan et al. reported relatively high expression of PDE3B in the surface zone of human and bovine articular cartilage, where chondroprogenitors reside, whereas PDE7A was expressed in the deep zone (Grogan et al., 2013). A comparison of healthy and damaged human knee OA cartilage revealed that PDE10A was significantly upregulated in damaged cartilage and that PDE3B, PDE7B, PDE4C and PDE1A were significantly downregulated (Dunn et al., 2016). In addition, exercise increases the expression of Pde3A, Pde3B and Pde10a in the articular cartilage of rats (Blazek et al., 2016).

Expression profiling of whole bone or cultivated murine osteocytes revealed that the transcripts of Pde2a, 5a, and 4d were among the most abundant, whereas Pde3a, 4b, 7b, 4a, 1b, 10a, 3b, 8a, 7a, 1a, and 9a presented moderate to negligible expression (Kim et al., 2020; Youlten et al., 2021).

Cultured human osteoclasts expressed PDEs 1B, 2A, 3B, 4A, 4B, 4D, 7A, 8A and 8B (Hansen et al., 2024).

Notably, Pde5a expression in 40-wk-old mice was significantly greater than that in young mice, suggesting that PDE5A could be targeted in older individuals to prevent bone loss (Kim et al., 2020). In the damaged cartilage of OA patients, PDE10A was significantly upregulated in the healthy part of the joint (Dunn et al., 2016), and in developing murine osteoblasts, it was significantly upregulated by PTH/PTHrP (Mosca et al., 2023).

Taken together, these expression patterns indicate that PDE families 1, 2, 3, 4, 5, 7, 8 and 10 are present and relevant for the skeletal system, whereas genes from PDE families 6, 9 and 11 are not expressed. Therefore, specific functions and interactions with signaling pathways will be further discussed only for genes from relevant PDE families.

3.1.1. Phosphodiesterase 1

Three subfamily genes, PDE1A, B and C, encode dual-specificity PDEs. Whereas PDE1A and PDE1B preferentially hydrolyze cGMP, PDE1C hydrolyzes cAMP and cGMP with similar K_m values (Loughney et al., 1996). The N-terminal region contains two Ca²⁺/calmodulin-binding domains and two phosphorylation sites that modulate the biochemical activity of these enzymes. Whereas Ca²⁺/calmodulin binding stimulates PDE1 enzymatic activity, the phosphorylation of PDE1A1 and PDE1A2 by PKA and the phosphorylation of PDE1B1 by CaMKII decrease PDE1B1 sensitivity to Ca²⁺/calmodulin and thus reduce the PDE1-mediated breakdown of cyclic nucleotides (Zhao et al., 1997). PDE1A was found to be upregulated by TGF-β and was implicated in myofibroblast formation, as well as pathological cardiomyocyte hypertrophy, blood pressure regulation and renal health (Miller et al., 2009; Zhou et al., 2010; Wang et al., 2017). PDE1B is expressed in multiple regions of the central nervous system, including dopaminergic circuits, where it regulates dopamine-linked signaling and synaptic plasticity, but it is also expressed in macrophages and T cells without known functions (Reed et al., 2002; Bender et al., 2004). Inhibition or loss of PDE1C1 reduces smooth muscle cell proliferation and neointimal hyperplasia and provides a protective effect against the development of heart failure, presumably through PKA and PI3K/PKB signaling (Cai et al., 2015; Knight et al., 2016).

Although PDE1 genes are also expressed in osteoblasts/osteocytes and chondrocytes, they were never functionally connected to the skeletal system.

3.1.2. Phosphodiesterase 2

The single PDE2A gene produces three known PDE2A isoforms that localize differentially to the cytosol, mitochondria, and cellular membranes (Guellich et al., 2014). All PDE2A isoforms can hydrolyze both cGMP and cAMP with similar maximal rates and relatively high K_m values, but cGMP binding to the N-terminal GAF domain allosterically stimulates cAMP hydrolysis, which then occurs at a 10-fold higher rate (Martinez et al., 2002). In cardiomyocytes, PDE2A degrades cGMP and modulates local cAMP pools, thereby fine-tuning β-adrenergic and other signaling pathways, and Pde2a^−/− mice exhibit severe cardiac malformations (Assenza et al., 2018). PDE2A suppresses Wnt/β-catenin signaling by inhibiting cAMP accumulation and GSK-3β phosphorylation in gliomas (Li et al., 2021) and controls cGMP-dependent Notch signaling in endothelial cells (Carlantoni et al., 2024). In human osteosarcoma cells, PDE2A regulates cell migration and proliferation (Murata et al., 2019).

3.1.3. Phosphodiesterase 3

The PDE3 subfamily of genes comprises PDE3A and PDE3B, which have high affinities for both cAMP and cGMP, although cGMP binding competitively inhibits cAMP hydrolysis (Degerman et al., 1997). PDE3s are also activated by PKA and PKB, which can phosphorylate their N-terminus (Shakur et al., 2001; Zmuda-Trzebiatowska et al., 2007).

PDE3A is known to regulate platelet activation, vascular smooth muscle proliferation, cardiomyocyte contractility, and oocyte maturation, as Pde3a^−/− female mice are infertile (Masciarelli et al., 2004; Sun et al., 2007; Begum et al., 2011). Mutations in PDE3A that potentiate PKA-mediated PDE3A activation are also linked to irresponsiveness to PTH/PTHrP signaling in humans, which results in pronounced acceleration of chondrocyte maturation, brachydactyly and short stature (Elli et al., 2018; Reyes et al., 2020). PDE3B is a major PDE in adipose tissue, the liver, and the pancreas, where it regulates insulin secretion and energy metabolism (Shakur et al., 2001). In addition, Pde3b^−/− mice generated by the International Mouse Phenotyping Consortium presented increased lengths of tibias (mousephenotype.org), a phenotype similar to CNP overexpression and FGFR3 loss (Liu and Lefebvre, 2015; Li et al., 2017). This finding is in line with recent discovery that Pde3b inhibition can increase bone outgrowth in mice via a mechanism previously described by same group for CNP (Miyazaki et al., 2022; Kawabe et al., 2025).

Therefore, both PDE3 genes function in growth plate chondrocytes, with PDE3A regulating primarily cAMP levels induced by PTH1R and PDE3B degrading cGMP produced by NPRB.

3.1.4. Phosphodiesterase 4

Four highly similar subfamily genes, PDE4A to —D, encode cAMP-specific rather ubiquitously expressed PDEs in multiple splice variants (Engels et al., 1994; Engels et al., 1995). Each variant is located in specific cell microdomains, and all of them can be recruited to β-arrestin1/2 to reduce cAMP activity in the subcellular compartment(s) enriched in β-receptor/G-protein-mediated signaling (Perry et al., 2002; Bender and Beavo, 2006). PDE4s can be phosphorylated by MAPK kinases, which inhibits their activity, but PKA phosphorylation can overcome this restriction to stimulate cAMP hydrolysis. This mechanism serves to locally regulate the cAMP level, where MAPK signaling increases the cAMP concentration, which in turn activates PKA, which phosphorylates PDE4, resulting in a return of cAMP to a lower level (Houslay and Adams, 2003). PDE4A was found to play a role in anxiety regulation, emotional memory, and T-cell activation and, together with PDE4D, in cardiomyocyte contractility (Hansen et al., 2014; Mika et al., 2019; Schmetterer et al., 2019). PDE4B is known to be involved in regulating inflammation in pulmonary and kidney fibrosis, in immune and dendritic cells, and in synoviocytes and intervertebral chondrocytes (Xu et al., 2020; Su et al., 2022; Xu et al., 2025a; Xu et al., 2025b). PDE4C is not linked with the skeletal system, but PDE4D is. Missense mutations affecting PKA phosphorylation sites in the PDE4D amino terminus cause unregulated activation of PDE4D and lead to acrodysostosis, a rare congenital malformation syndrome that involves shortening of the interphalangeal joints of the hands and feet, increased bone age, intrauterine growth retardation, juvenile arthritis, short stature and intellectual disability (Lee et al., 2012; Linglart et al., 2012). The mechanism is very similar to that of PDE3A (Reyes et al., 2020). As such, PDE4D was identified as one of several loci affecting bone mineral density (Reneland et al., 2005; Greenbaum et al., 2022).

3.1.5. Phosphodiesterase 5

PDE5A is the only member of the cGMP-specific PDE5 subfamily whose activity is controlled by PKA and PKG through phosphorylation of its N-terminus (Zoraghi et al., 2005). PDE5A regulates vascular smooth muscle relaxation and vasodilation, mainly through controlling cGMP levels via NO-induced signaling, which is used to treat erectile dysfunction and chronic pulmonary hypertension (Sáenz de Tejada et al., 2004; Barnett and Machado, 2006). PDE5A can also affect the localization of endothelial NO synthase (NOS3) and thus regulate NO signaling (Gebska et al., 2011). In osteoblasts and osteocytes, PDE5A controls PKG signaling by downregulating NO-induced cGMP (Kim et al., 2021).

3.1.6. Phosphodiesterase 7

Two subfamily genes, PDE7A and PDE7B, encode high-affinity cAMP-specific PDEs widely expressed in the whole body, including the brain, heart, lungs, muscles and immune cells (Chen et al., 2021). PDE7A contributes to working memory and spatial learning, whereas PDE7B was found to be an important regulator of cancer cell proliferation (Liu et al., 2023; Du et al., 2024). PDE7A regulates PKA activity not only by hydrolyzing cAMP but also via direct interaction with PKA (Han et al., 2006). The inhibition of PDE7, especially in combination with PDE3-targeting drugs, is considered a treatment for some autoimmune diseases or selected respiratory syndromes (Szczypka, 2020). In osteoblasts, the inhibition or silencing of PDE7 genes upregulated the expression of osteogenic markers such as osteocalcin (Pekkinen et al., 2008).

3.1.7. Phosphodiesterase 8

PDE8A and PDE8B are the only subfamily genes, highly expressed in steroidogenic cells, such as Leydig cells and adrenocortical cells, where they regulate basal steroidogenesis by controlling different cAMP pools (Leal, 2021; Vasta et al., 2006). PDE8A is also highly expressed in chondrocytes, and its deficiency leads to increased body size and abnormal bone structure (mousephenotype.org)(Liu and Lefebvre, 2015; Li et al., 2017), whereas Pde8b-deficient male mice are infertile (Vasta et al., 2006; Leal et al., 2021).

PDE8A can be directly phosphorylated by focal adhesion kinase (FAK) after mechanical stimulus in an osteocyte-like cell line, which potentiates its activity and downregulates the expression of cAMP/PKA target genes (Papaioannou et al., 2024).

3.1.8. Phosphodiesterase 10

The sole gene PDE10A hydrolyzes both cAMP and cGMP (Fujishige et al., 1999). High affinity for cAMP inhibits cGMP hydrolysis, making this enzyme a cAMP-inhibited dual-substrate PDE (Omori and Kotera, 2007). PDE10A is highly expressed in the striatum of the brain together with PDE4s, but their actions are distinct from each other (Nishi et al., 2008). Pde10a knockout mice have lower body weights than their wild-type siblings, with female animals being more affected than males (Siuciak et al., 2006). Consistent with the differences in body weight, PDE10A is important for the regulation of the energy balance of brown and white fat cells as well as insulin resistance (Nawrocki et al., 2014; Hankir et al., 2016). Genetic deletion and pharmacological inhibition of Pde10a protects mice from diet-induced obesity and insulin resistance (Nawrocki et al., 2014). PDE10A was also suggested to be a modulator of osteogenic differentiation as well as mechanotransduction in bone marrow-derived mesenchymal stromal cells, with a further unclarified mechanism (Müller-Deubert et al., 2020).

4. PDE inhibitors and their effects on bone and cartilage

Because of huge complexity and overlapping expression, the classic tools of molecular genetics were barely used to determine the function of individual PDE isoforms. Instead, the inhibitors that block the catalytic function of PDEs were preferred but most of them target multiple PDE isoforms. Consequently, straight evidence regarding the function of distinct PDE isoforms is mostly missing and function of PDEs can only be deduced indirectly, based on their expression in skeletal system, what is known about them in general and how their inhibitors affect the skeletal system in growth, and pathology.

Currently, there are 32 PDE inhibitors approved for various therapeutic uses in Europe, the USA or Asia (Table 1). None of them, nor any of the 14 inhibitors that were in clinical trials in 2022 (Bondarev et al., 2022), are dedicated to the treatment of any musculoskeletal condition. However, dozens of preclinical studies have shown, mostly on rodent models, that PDE inhibitors have positive effects on the health of bone and cartilage, although the precise mechanism(s) are not yet fully understood (Table 2).

Table 1.

Overview of PDE inhibitors approved for clinical use and treatment of indicated syndromes in Europe (EMA, European Medicines Agency), the USA (FDA, Food and Drug Administration) or Asia.

Compound (popular trade names)	Indication	PDE target	Activity	Approval date (USA, Europe and Asia markets)
Vinpocetine (Cavinton)	Cerebral vascular disorders and memory impairment	1 cAMP/cGMP	Increase cAMP>cGMP	USA as an over-the-counter dietary supplement, some countries in Europe (for example, Spain, 1997) and Asia (for example, India, 2002)
Cilostazol (Pletal, Ekistol)	Peripheral arterial disease (PAD), Intermittent claudication	3 cAMP/cGMP	Increase cAMP>cGMP	FDA (1999), some countries in Europe (for example, UK, 2000) and Asia (for example, South Korea, 1990)
Milrinone (Primacor, Corotrope)	Decompensated cardiac failure	3 cAMP/cGMP	Increase cAMP>cGMP. Increase the intracellular level of cAMP to cause vasodilation.	FDA (1987), EMA (2016), Asia (for example, Japan, 1996)
Amrinone (Inamrinone, Inocor)	Decompensated cardiac failure	3 cAMP/cGMP	Increases cAMP by preventing its breakdown	FDA (1984), some countries in Asia (for example, India, 1988)
Enoximone (Perfan)	Decompensated cardiac failure	3, (4) cAMP/cGMP	Increases intracellular levels of cAMP by inhibiting cGMP-inhibited PDE	Some countries in Europe (for example, France, 1987)
Olprinone (Coretec)	Heart failure	3 cAMP/cGMP	Increased intracellular cAMP concentrations and subsequent activation of protein kinase A (PKA)	Japan (1996)
Pimobendan (Acardi)	Heart failure	3 cAMP/cGMP	Increase cAMP/cGMP	Japan (1994)
Anagrelide (Agrylin, Xagrid)	Thrombocythemia	3 cAMP/cGMP	Increase cAMP>cGMP	FDA (1997), EMA (2004), some countries in Asia (for example, South Korea, 2004)
Roflumilast (Daliresp, Daxas)	Chronic obstructive pulmonary disease (COPD)	4 cAMP	cAMP	FDA (2011), EMA (2010), some countries in Asia (for example, India, 2014)
Apremilast (Otezla)	Psoriatic arthritis (PA), Psoriasis	4 cAMP	cAMP	FDA (2014), EMA (2014), some countries in Asia (for example, Japan, 2016)
Crisaborole (Eucrisa)	Decompensated cardiac failure, Atopic dermatitis (AD), Moderate atopic dermatitis (patients >2 years old)	4 cAMP	cAMP	FDA (2016)
Drotaverine (No-Spa, Doverin)	Functional bowel disorders; pain caused by smooth muscle spasm	4 cAMP	cAMP	Some countries in Europe (for example, Hungary, 1963) and Asia (for example, China, 1999)
Sildenafil (Viagra, Revatio)	Erectile dysfunction (ED), Pulmonary arterial hypertension (PAH)	5 cGMP	cGMP	FDA (1998), EMA (1998), Asia (for example, Japan, 1999)/FDA (2014), EMA (2005), Asia (for example, Japan, 2008)
Vardenafil (Levitra, Staxyn, Vivanza)	Erectile dysfunction (ED)	5 cGMP	cGMP	FDA (2003), EMA (2003), some countries in Asia (for example, Japan, 2004)
Tadalafil (Cialis, Adcirca)	Erectile dysfunction (ED), Benign prostatic hyperplasia (BPH), Pulmonary arterial hypertension (PAH)	5 cGMP	cGMP	FDA (2003), EMA (2002), some countries in Asia (for example, India, 2003)/FDA (2009), EMA (2008), some countries in Asia (for example, India, 2009)
Avanafil (Stendra, Spedra)	Erectile dysfunction (ED)	5 cGMP	cGMP	FDA (2012), EMA (2013), some countries in Asia (for example, South Korea, 2011)
Udenafil (Zydena)	Erectile dysfunction and hypertension	5 cGMP	cGMP	Some countries in Asia (for example, South Korea, 2005)
Mirodenafil (Mvix)	Erectile dysfunction	5 cGMP	cGMP	South Korea (2007)
Papaverine (Pavabid, Pavagen)	Visceral spasm, vasospasm and erectile dysfunction	10A cAMP/cGMP	Increase cAMP/cGMP	FDA, Europe (for example, Hungary, 1933), Asia (for example, Japan, 1953)
Theophylline (Theolair, Slo-Bid, Theo 24)	Chronic obstructive pulmonary disease (COPD), Asthma	3, 4, 7 cAMP/cGMP	Increase cAMP/cGMP	FDA (1937), Europe (for example, Spain, 1922), Asia (for example, India, 1969)
Aminophylline (Phyllocontin)	Asthma and bronchoconstriction	3, 4, 7 cAMP/cGMP	Increase cAMP/cGMP	FDA (1940), Europe (for example, Hungary, 1935), Asia (for example, India, 1950)
Choline theophyllinate (Oxtriphylline, Choledyl)	Asthma and bronchoconstriction	3, 4, 7 cAMP/cGMP	Increase cAMP/cGMP	FDA (1981), Europe (only in Greece, 2003)
Dyphylline (Dilor, Lufyllin)	Asthma and bronchoconstriction	3, 4, 7 cAMP/cGMP	Increase cAMP/cGMP	FDA (1951), some countries in Europe (for example, Spain, 1968) and Asia (for example, Japan, 1952)
Ibudilast (Ketas, Pinatos, Eyevinal)	Asthma and dizziness related to cerebral infarction, Allergic conjunctivitis	10A, 4, 11, 3 cAMP/cGMP	Increase cAMP/cGMP	Asia (Japan, 1989; South Korea, 1998; China, 2003)
Tofisopam (Emandaxin, Grandaxin)	Anxiety	4, 10, 3, 2 cAMP/cGMP	Increase cAMP/cGMP	Some countries in Europe (for example, Hungary, 1974) and Asia (for example, Japan, 1985)
Pentoxifylline (Trental, Pentoxil)	Peripheral arterial disease (PAD), Intermittent claudication	4, 5 cAMP/cGMP	Increase cAMP/cGMP	FDA (1984), some countries in Europe (for example, Spain, 1978) and Asia (for example, India, 1975)
Dipyridamole (Persantine)	Postoperative thromboembolic prophylaxis	8, 1, 3, 2 cAMP/cGMP	Increase cAMP/cGMP	FDA (1961), some countries in Europe (for example, Spain, 1986) and Asia (for example, India, 1964)
Acefylline	Bronchodilator and cardiac stimulant, Emphysema	Nonselective	peptidylarginine deiminase (PAD) activator, inhibits cAMP isoenzymes	Not specific separate approval found by FDA, Included in European Pharmacopoeia; used in formulations but no explicitly approval report. ASIA (India, Japan, Vietnam) Approved and used for respiratory conditions including asthma and COPD.
Doxofylline	Asthma, COPD, and Bronchospasm	4, 2 cAMP/cGMP	Increase cAMP/cGMP	Orphan drug designation granted in 2014, but no full approved by FDA, Approved in various European countries including Italy from 1988. ASIA (India, Phillippines, South Korea, Mexico) Approved and used for asthma and COPD.
Flavoxate hydrochloride (Rec-7-0040; DW61, Urispas, Genurin, Bladderon, Bladuril)	Antispasmodic agent and muscarinic mAChR antagonist. Moderate calcium antagonistic activity and local anesthetic effect. Urinary incontinence.	Nonselective	Acts as a direct antagonist of muscarinic acetylcholine receptor	FDA approved (generic of Urispas) in 2003.Clinical trial NCT00992238 in USA and NCT00440739 in Thailand. Approved (Genurin) in Italy, (Bladderon) in Japan, (Bladuril) in Chile.
Mebeverine hydrochloride (Colofac, Aurobeverine)	Treatment of irritable bowel syndrome. To relieve spasm of smooth muscle. Musculotropic agent that potently blocks intestinal peristalsis.	Nonspecific	Directly blocks voltage-operated sodium channels and inhibits intracellular calcium accumulation	Not FDA approved as a standalone drug. Approved in Malta, Northen Ireland. Approved in UK. Marketed in India.
Alverine citrate (NSC 35459)	Irritable Bowel Syndrome (IBS). Functional gastrointestinal disorders. Vasodilator, relieves spasm of smooth muscle.	Nonselective	5-HT1A receptor antagonist	Not approved by FDA. In clinical trials with not provided locations (China, Hungary, Mexico). Approved in UK, Ireland, France and others across Europe. Approved in India, Japan, Korea, Thailand, Singapore.

Table 2.

Overview of the effect of PDE inhibitors on preclinical models and in clinical trials. Where more than one model was used for the same drug, only the model closer to clinic is listed. BMSC, Bone Marrow derived Stem Cells; OVX, ovariectomy; DMM, Dislocation of Medial Meniscus; LPS, Lipopolysaccharide.

Drug	Subject	Modeled situation	Effect	Reference
Amrinone	Mouse calvarial bone	Osteoporosis	Inhibition of bone resorption (200 μM)	(Krieger and Stern, 1982; Krieger et al., 1986)
Anagrelide	Mouse metatarsal bones	Achondroplasia	Improved bone outgrowth (10 μM)	(Kawabe et al., 2025)
Apremilast	ATDC5 chondrocyte cell line	OA	Suppressed cell senescence (1 μM for 7 days)	(Wang et al., 2021)
	Human T/C-28a2 chondrocytes		Chondrocyte protection from inflammation (2 μM for 12 h)	(Zhang et al., 2020)
	Clinical trial	RA	No significant effect in patients with active RA (30 mg, twice a day for a year)	(Genovese et al., 2015)
Avanafil	Rats	Osteoporosis	Reduction of oxidative stress and bone atrophy (10 mg/kg/day, daily for 30 days)	(Huyut et al., 2018)
Cilostazol	Mice	Achondroplasia	Improved bone outgrowth in 3 weeks old mice (10 mg/kg/day, 28× in 4 weeks)	(Kawabe et al., 2025)
		Fracture healing	Promotion of blood vessel formation and bone regeneration (30 mg/kg/day, daily in 2–10 weeks)	(Menger et al., 2023b) (Menger et al., 2024b)
	Rats intraarticularly injected by mono-iodoacetate	Osteoarthritis	prevention of cartilage destruction (30 mg/kg/day, 28× in 4 weeks)	(Lee et al., 2008)
	Human chondrocytes		Prevention of the degradation of collagen type II (10 μM)	(Wang et al., 2014)
	6 weeks old OVX mice		Reduced bone loss by inhibition of osteoclastogenesis (0.5 mg/kg/day, daily for 8 weeks)	(Ke et al., 2015)
	Rat articular chondrocytes	Chondrocyte viability	Induction of cellular senescence and resistance to etoposide-induced apoptosis (≥2 μM)	(Kim et al., 2012)
	Clinical trial	RA	May be beneficial, no clear conclusion (50 mg twice daily for 3 months)	(Eldadamony et al., 2025)
Dipyridamole	Mice	Bone formation	Enhanced bone regeneration (8 weeks treatment)	(Mediero et al., 2015)
			Counteracted tenofovir-induced bone Resorption (25 mg/day, daily for 4 weeks)	(Conesa-Buendía et al., 2019)
	Sheep		Improved the calvarial bone regeneration (soaked matrix, 100 μM)	(Bekisz et al., 2018)
	Rabbit		Increased bone volume (soaked matrix, ≥10 μM)	(Witek et al., 2019)
Ibudilast	Mice injected with LPS	Fracture healing	Improved bone healing (4 mg/kg/day, daily for 3 days)	(Chang et al., 2021)
	Mice injected with collagen	RA	Inhibition of disease progression (10 mg/kg, every other day for 10 days)	(Clanchy and Williams, 2019)
Milrinone	Mouse metatarsal bones	Achondroplasia	Improved bone outgrowth (10 μM)	(Kawabe et al., 2025)
Olprinone	Mouse metatarsal bones		Improved bone outgrowth (10 μM)	(Kawabe et al., 2025)
Papaverine	Rats	Tendon rupture	Higher tensile strength (30 mg at the site of injury)	(Can et al., 2024)
	Human mesenchymal stem cells	Bone differentiation	Diminished osteogenic differentiation (10 μM)	(Müller-Deubert et al., 2020)
	Mesenchymal stem cells		No significant effect (concentration not stated)	(Silva et al., 2021)
Pentoxifylline	Osteogenic cell lines	Bone formation	Improved chondro/osteogenic differentiation (500 μg/ml for six days)	(Tsutsumimoto et al., 2002)
	Mice		Enlarged and more calcified ossicles (200 mg/kg/day, daily for 3 weeks)	(Horiuchi et al., 2004)
			Enhanced bone formation in WT (≥100 mg/kg/day, daily for 5 weeks)	(Kinoshita et al., 2000)
			Increased bone formation and bone mass (50 mg/kg/day, daily for 12 weeks)	(Roser-Page et al., 2022)
			Enlarged and more calcified ossicles (≥50 mg/kg/day, daily for 3 weeks)	(Horiuchi et al., 2001)
	Rats	Fracture healing	Improved bone regeneration (50 mg/kg/day, daily for 1 week)	(MalekiGorji et al., 2020)
			Improved fracture healing (50 mg/kg/day, daily for 30 days)	(Atalay et al., 2015)
			Improved bone formation (25 mg/kg/day, daily for 8 weeks)	(Çakmak et al., 2015)
	13 months old OVX Rats	Osteopenia	Restored healthy bone properties (25 mg/kg/day, daily for 12 weeks)	(Pal et al., 2019b)
	Newborn rats	Bone formation	Improved bone growth after phototherapy (50 mg/kg, twice daily for 7 days)	(Atabek et al., 2007)
	Rabbit	Osteopenia	Restored healthy bone properties (12.5 mg/kg/day, daily for 4 months)	(Pal et al., 2019a)
	Dogs	OA	Mitigated inflammation (10 mg/kg, daily for 60 days)	(Parlak et al., 2025)
	Clinical trial	RA	Decreased the pain severity score (1200 mg/day for 1 month)	(Dubost et al., 1997)
			Significant diminution in number of tender and swollen joints	(Maksymowych et al., 1995)
	Case study		Complete recovery from arthritis (300 mg/day)	(Ishii et al., 1997)
Roflumilast	MH7A FLS cells	RA	Reduced inflammatory response	(Zhong et al., 2021)
	Mice derived osteoclast	Osteopetrosis	Rescue of osteoclast activity (≥50 nM)	(Hong et al., 2024)
	Mice		No significant effect	(Alam et al., 2024)
	Rats	Intervertebral disc degeneration	(5 mg/kg/day, daily for 4 weeks)	(Xu et al., 2025b)
Rolipram	Mice	Bone formation	Enhanced bone formation in WT (≥10 mg/kg/day, daily for 5 weeks)	(Kinoshita et al., 2000)
			Increased trabecular bone volume (20 mg/kg/day, daily for 4 weeks)	(Bonnet et al., 2007)
	Mice derived osteoclast	Osteopetrosis	Rescue of osteoclast activity (≥100 nM)	(Hong et al., 2024)
	6 months old OVX rats	Osteopenia	Prevented bone loss (≥0.3 mg/kg/day, daily for 60 days)	(Yao et al., 2007)
	Mice injected with collagen	RA	Amelioration of arthritis, suppression of TNF-α production and inhibition of Thl Activity (>10 mg/kg/day, two times daily for 10 days)	(Ross et al., 1997)
	Rats injected with streptococcal cell walls		Attenuation of arthritis score, reduction of edema (≥3 mg/kg/day, daily for 5 days)	(Laemont et al., 1999)
	Rats injected with Mycobacterium		Abrogated oedema formation, inhibited hyperalgesia, reduced tissue destruction (3 mg/kg/day, daily for 5 days)	(Francischi et al., 2000)
Sildenafil	Mice	Fracture healing	Accelerated fracture healing, enhanced bone formation (5 mg/kg/day, daily for 5 weeks)	(Histing et al., 2011)
			Accelerated fracture healing (5 mg/kg/day, daily for 5 weeks)	(Histing et al., 2011)
			Improved bone regeneration (5 mg/kg/day, daily for 10 weeks)	(Menger et al., 2023a)
	Old mice		Delayed bone remodeling with less osteoclasts (5 mg/kg/day, daily for 5 weeks)	(Menger et al., 2024a)
	Rats		Accelerated fracture healing (5 mg/kg/day, daily for 5 weeks)	(Toğral et al., 2015)
			Improved bone regeneration (10 mg/kg/day, daily for 1 week)	(MalekiGorji et al., 2020)
	5 months old OVX mice	Osteoporosis	Inhibited bone loss (10 mg/kg/day, daily for 1 month)	(Hu et al., 2025)
		Osteopenia	Increased bone strength and vascularity (≥6 mg/kg/day, daily for 6 weeks)	(Pal et al., 2020)
Tadalafil	Mice	Bone health	Anabolic and antiresorptive actions on the skeleton (2 mg/kg daily, for 6 weeks)	(Kim et al., 2020)
		Bone formation	Reduced bone mass (≥45 mg/kg/day, daily for 2 month)	(Gong et al., 2014)
	Rats	Fracture healing	Accelerated fracture healing (1 mg/kg/day, daily for 5 weeks)	(Toğral et al., 2015)
	8 months old OVX rats	Osteoporosis	Increased bone mass, reduction of bone resorption (10 mg/kg/day, daily for 2 months)	(Alp et al., 2017)
	Rabbits	Fracture healing	Improved bone regeneration (matrix soaked with 30 mg of tadalafil for 4 weeks)	(Soufdoost et al., 2019)
Theophylline	Rats	Osteopenia	Diminished bone regeneration, osteoblast apoptosis (≥25 mg/kg/day, daily for 12 days)	(Pal et al., 2016)
	Rats injected with Freund's Adjuvant	RA	Decreased arthritis-index, paw volume and ankle diameter (≥10 mg/kg/day, daily for 14 days)	(Pal et al., 2015)
			Anti-arthritic effect (20 mg/kg/day, daily for 14 days)	(Gaafar et al., 2018)
			Decreased arthritis-index, paw volume and ankle diameter (≥15 mg/kg/day, daily for 11 days)	(Gomaa et al., 2009)
Tofisopam	Mice	Bone formation	Increased trabecular bone volume (10 mg/kg/day, daily for 4 weeks)	(Bonnet et al., 2007)
Udenafil	8 months old OVX rats	Osteoporosis	Increased bone mass, reduction of bone resorption (10 mg/kg/day, daily for 2 months)	(Alp et al., 2017)
Vardenafil	Mice	Bone health	Anabolic and antiresorptive actions on the skeleton (10 mg/kg daily, for 6 weeks)	(Kim et al., 2020)
	5 months old OVX mice	Osteopenia	Increased bone strength and vascularity (≥2.5 mg/kg/day, daily for 6 weeks)	(Pal et al., 2020)
	8 months old OVX rats	Osteoporosis	Increased bone mass, reduction of bone resorption (10 mg/kg/day, daily for 2 months)	(Alp et al., 2017)
	Rats	Fracture healing	Improved bone regeneration (5 mg/kg/day, daily for 42 days)	(Atcı et al., 2021)
Vinpocetine	BMSCs	Bone differentiation	No effect in low concentration (0.17 μM), impaired in high concentrations (≥5 μM)	(Yıldırım and Sezer, 2021)
	DMM mice	Osteoarthritis	Improved cartilage degeneration, subchondral remodeling, synovitis, and ECM degradation (≥5 mg/mg/day, 16× in 8 weeks)	(Wang et al., 2024)
	10 weeks old OVX mice	Osteoporosis	Inhibition of OVX-induced bone loss by inhibition of osteoclastogenesis (≥2.5 mg/mg/day, 40× in 8 weeks)	(Zhu et al., 2020)
Zaprinast	Rats	Osteoporosis	Reduction of oxidative stress and bone atrophy (10 mg/kg/day, daily for 30 days)	(Huyut et al., 2018)

Some studies indicate that the mechanism involves the suppression of osteoclastogenesis (Park and Yim, 2007; Yao et al., 2007), whereas others suggest the promotion of osteoblast formation, either directly or through the potentiation of other effectors, such as PGE2, PTH, and BMP signaling (Kinoshita et al., 2000; Horiuchi et al., 2001; Horiuchi et al., 2002; Tsutsumimoto et al., 2002; Wakabayashi et al., 2002; Horiuchi et al., 2004; Tokuhara et al., 2010). The following sections summarize the current state of knowledge about the effects of individual marketed PDE inhibitors on bone or cartilage, with a special focus on the mechanism of action.

4.1. PDE1 inhibitors

Vinpocetine, a known inhibitor of PDE1, attenuates ovariectomy-induced bone loss at a dose of 2.5 mg/kg for 8 weeks, presumably through the suppression of osteoclastogenesis via a reduction in NF-κB, MAPK, and PKB signaling in osteoclasts (Zhu et al., 2020). However, vinpocetine can inhibit NF-κB signaling without affecting PDE1 (Sitges et al., 2005; Jeon et al., 2010). In the tert-butyl hydroperoxide-induced OA model, vinpocetine was able to protect against osteoarthritis by inhibiting ferroptosis and extracellular matrix degradation via activation of the cAMP/cGMP-independent Nrf2/GPX4 pathway (Wang et al., 2024). A clinically relevant dose of vinpocetine (62 ng/ml) does not affect bone marrow-derived stem cell differentiation into osteoblasts, but high concentrations impair this process (Yıldırım and Sezer, 2021). Overall, these data suggest that vinpocetine acts through different mechanisms than PDE1 inhibition and that its utilization is controversial.

4.2. PDE3 inhibitors

Cilostazol and milrinone are the most common PDE3 inhibitors used to treat intermittent claudication and congestive heart failure, respectively (Baillie et al., 2019). Early studies suggested that cilostazol has a protective effect on rodent articular chondrocytes (Lee et al., 2008; Kim et al., 2012; Wang et al., 2014). However, they did not provide any mechanism (Lee et al., 2008; Kim et al., 2012), or they suggested that cAMP lowered the production of the collagen-degrading enzyme matrix metalloprotease 13 (MMP-13) through suppressed STAT1 phosphorylation (Wang et al., 2014). STAT1 was previously identified as an important negative regulator of cartilage growth (Sahni et al., 2001). Cilostazol has already been tested in clinical trials as an adjuvant therapy for RA, but the results are not convincing (Eldadamony et al., 2025). Kawabe et al. recently showed that cilostazol, milrinone, olprinone and anagrelide increase cGMP levels in growth plate chondrocytes, likely via PDE3B inhibition, supporting bone growth in the same manner as the authors previously described for CNP (Miyazaki et al., 2022; Kawabe et al., 2025). Mechanistically, CNP treatment or PDE3 inhibition increased PKG activity, which stimulated the opening of Big Potassium (BK) channels. Subsequent K⁺ influx facilitates the opening of transient receptor potential cation channels, subfamily M member 7 (TRPM7), which physiologically, spontaneously and intermittently open to generate intracellular Ca²⁺ fluctuations. BK-stimulated Ca²⁺ entry further activated CaMKII. This cascade stimulates extracellular matrix synthesis and chondrocyte function, both of which are essential for cartilage health and repair (Kawabe et al., 2025). In addition to PKG, PKA can phosphorylate and activate BK channels (Calderone, 2002; Kyle and Braun, 2014). This mechanism differs from what has been previously described by other groups, according to which CNP stimulates the proliferation of growth plate chondrocytes by antagonizing the FGFR3-induced MAPK signaling pathway (Yasoda et al., 2004; Krejci et al., 2005). Both mechanisms are likely correct, as both start with increased PKG activity, which, in one case, phosphorylates and thus inactivates c-RAF kinase, a part of the MAPK signaling pathway, or, in other cases, stimulates the opening of BK channels. Further research is needed to clarify the exact mechanism of CNP and cilostazol action on growth plate chondrocytes.

In a mouse model of atrophic nonunion, daily oral administration of cilostazol significantly increased callus bone formation, improved biomechanical strength, and amplified angiogenesis by increasing VEGF expression (Menger et al., 2023b). In osteoblasts, cilostazol activates the PI3K pathway and RUNX2 expression, both of which are pivotal for osteoblast differentiation and fracture repair (Menger et al., 2024b). Furthermore, in ovariectomized mouse models mimicking postmenopausal osteoporosis, the administration of cilostazol led to the attenuation of bone loss, presumably through a decrease in osteoclast activity and a reduction in oxidative stress markers. In vitro, cilostazol inhibited osteoclast differentiation via PKA-mediated phosphorylation of the transcription factor NFAT2, which suppressed NF-κB pathway activation, which is important for osteoclast maturation (Ke et al., 2015).

Amrinone in early studies inhibited stimulated bone resorption in neonatal mouse calvaria and fetal rat limb bone cultures (Krieger et al., 1986). Research has shown that amrinone inhibits the release of Ca²⁺ from neonatal mouse calvaria stimulated by parathyroid hormone (PTH), 1,25-dihydroxyvitamin D3, or prostaglandin E2 but does not affect cAMP levels in bone (Krieger and Stern, 1982). These findings suggest that its inhibitory effect on bone resorption might be mediated via mechanisms other than direct modulation of cAMP signaling, likely through the inhibition of osteoclast activity, similar to cilostazol and milrinone.

There is no published effect of enoximone, pimobendane or anagrelide on cells that form bone and cartilage.

4.3. PDE4 inhibitors

Rolipram has been shown to suppress bone loss or promote bone formation in mice by itself or in combination with BMP ligands 2 and 4 (Kinoshita et al., 2000; Horiuchi et al., 2002; Park and Yim, 2007; Yao et al., 2007; Tokuhara et al., 2010). In addition, rolipram has been reported to suppress TNF-α and other proinflammatory cytokines, effectively reducing joint inflammation and bone and cartilage destruction in arthritis animal models (Ross et al., 1997; Laemont et al., 1999; Francischi et al., 2000). On the other hand, rolipram stimulated osteoclast formation in vitro through the PKA/p38/MAPK pathway and thus reduced high bone mass in mouse explants of autosomal dominant osteopetrosis type II (Cho et al., 2004; Hong et al., 2024). However, rolipram has been withdrawn from the market because of its narrow therapeutic window and severe gastrointestinal adverse effects, including nausea and vomiting (Scott et al., 1991; Bielekova et al., 2009), so it cannot be repurposed for bone therapy.

Roflumilast had similar positive impact on osteoclast activity in vitro as rolipram (Hong et al., 2024), however in vivo application had no conclusive effect on reducing high bone mass in mouse model of autosomal dominant osteopetrosis type II (Alam et al., 2024). On the other hand, it was able to suppress IL-18-induced inflammation in fibroblast-like synoviocytes and regulate the expression of metalloproteinases MMP-3 and MMP-9, which are involved in cartilage degradation, indicating potential chondroprotective effects (Zhong et al., 2021). In a rat model of intervertebral disc degeneration, roflumilast inhibited Pde4b, so it was able to reverse pathological cartilage degradation. The mechanism may involve the suppression of Yap-induced expression of long-chain fatty acid-CoA ligase 4 (ACSL4), a positive regulator of an iron-dependent form of cell death, ferroptosis (Xu et al., 2025b).

Like roflumilast, apremilast also suppressed TNF-β/IL-18/IL1-β-induced cartilage degradation, but the exact mechanism is not clear (Tenor et al., 2002; Zhao et al., 2020; Wang et al., 2021). Apremilast-mediated inhibition of inflammatory cytokines such as TNF-α and IL-1β has also been described for psoriatic arthritis (Chen et al., 2018a; Li et al., 2018). As PDE4 inhibitors are commonly used for managing disorders such as psoriatic arthritis, which shares an inflammatory pathology with cartilage degradation, apremilast can work primarily through the general suppression of inflammation. In addition, apremilast has a direct effect on SOX9 production in chondrocytes, likely through the cAMP/PKA/CREB stimulatory axis (Zhang et al., 2020). Though, apremilast did not demonstrate significant effect against RA and clinical trial was terminated in phase II (Genovese et al., 2015).

The PDE4 inhibitors crisaborole and drotaverine have not been reported to be associated with bone or cartilage.

4.4. PDE5 inhibitors

PDE5A was identified as the most abundant cGMP-specific PDE in mouse bone; interestingly, its expression increases with age (Kim et al., 2020). Sildenafil, tadalafil, udenafil and vardenafil are selective PDE5 inhibitors widely used for treating erectile dysfunction or pulmonary hypertension (Wang et al., 2012; Warli et al., 2023). In addition, they have promising effects on bone repair and regeneration in various animal models of fracture healing (Histing et al., 2011; Toğral et al., 2015; Soufdoost et al., 2019; Kim et al., 2020; MalekiGorji et al., 2020; Menger et al., 2023a; Menger et al., 2024a), osteopenia (Pal et al., 2020) and osteoporosis (Huyut et al., 2018). The primary mechanism likely involves the cGMP–PKGII pathway, which stimulates MAPK and Wnt signaling (Rangaswami et al., 2009; Alp et al., 2017; Pal et al., 2020), although BMP and other pathways have also been implicated (Huyut et al., 2022). For example, sildenafil and vardenafil have been reported to enhance TGF-β signaling, although the underlying mechanism remains unclear (Atcı et al., 2021; Hu et al., 2025). Together with their direct effects on osteogenic differentiation during fracture healing (Kim et al., 2020; Hu et al., 2025), both drugs increase VEGF production by osteoblasts, thereby improving angiogenesis at the site of injury (Histing et al., 2011; Pal et al., 2020). The effect of PDE5 inhibitors on bone is dose dependent: while approximately 2 mg/kg/day tadalafil is beneficial (Kim et al., 2020), very high doses, such as 45–75 mg/kg/day, have the opposite effect and reduce bone mass (Gong et al., 2014). The suggested mechanism involves PKGII, which phosphorylates and activates GSK-3β, which in turn destabilizes β-catenin, a key component of canonical Wnt signaling, leading to reduced osteoblast differentiation (Gong et al., 2014).

Overall, sildenafil and vardenafil exert anabolic effects by stimulating osteoblast activity, promoting vascularization, and inhibiting osteoclastogenesis, although sildenafil may produce stronger responses in certain models (Alp et al., 2017; Kim et al., 2020).

Avanafil, a highly selective PDE5 inhibitor, has been shown—along with udenafil—to improve bone growth, increase mineral density, and suppress osteoclast activity in rat models of osteoporosis via the NO/cGMP/PKG pathway (Alp et al., 2017; Huyut et al., 2018). Avanafil was also effective against glucocorticoid-induced bone loss in rats, likely through activation of the same signaling cascade (Huyut et al., 2018).

Mirodenafil, another potent PDE5 inhibitor used clinically for erectile dysfunction, has not yet been evaluated in the context of bone or cartilage.

4.5. PDE10 inhibitors

The PDE10 A inhibitor papaverine demonstrated positive biomechanical and histopathological effects on tendon repair in rats (Can et al., 2024) but also reduced osteogenic differentiation and mineralization in vitro (Müller-Deubert et al., 2020). Additionally, papaverine treatment did not significantly affect the reconstitution of parietal bone from mesenchymal stem cells in a rat model (Silva et al., 2021), suggesting that its application for the treatment of musculoskeletal disorders cannot be recommended.

4.6. Broad-spectrum inhibitors

In contrast to previously discussed drugs, theophylline—an inhibitor of PDE3, PDE4, and PDE7—has detrimental effects on bone. It promotes calcium loss and induces osteoblast apoptosis by increasing reactive oxygen species through elevated cAMP (Fortenbery et al., 1990; Pal et al., 2016), resulting in impaired fracture healing in adult rat models (Pal et al., 2016). This is clinically relevant for patients with chronic obstructive pulmonary disease who often receive theophylline; however, vitamin D supplementation can mitigate these adverse effects (Pal et al., 2016). Interestingly, theophylline exerts anti-inflammatory and antiarthritic effects on cartilage, reducing inflammatory infiltration, pannus formation, and cartilage erosion in arthritis models (Pal et al., 2015; Gaafar et al., 2018), an effect further enhanced by low-dose nitric oxide donors (Gomaa et al., 2009).

Aminophylline, another nonselective PDE inhibitor, alleviates collagen-induced rheumatoid arthritis in rats by increasing cAMP and PKA activity and inhibiting MAPK signaling (Zhang et al., 2008). Its use during pregnancy, however, has been linked to metabolic bone disease associated with prematurity (Chen et al., 2018b).

Ibudilast, which targets PDE3, PDE4, PDE10A, and PDE11, reversed the lipopolysaccharide-induced inhibition of osteoblast differentiation while reducing osteoclast activation, which resulted in improved bone healing in murine femoral defect models (Chang et al., 2021). It also has strong immunomodulatory effects in RA, reducing the levels of proinflammatory cytokines such as TNF-α and IL-12, as well as the levels of chemokines in synovial fibroblasts; inhibits the response of interleukin-17-producing T-helper (Th17) cells; and enhances the efficacy of TNF inhibitors in vitro and in vivo (Clanchy and Williams, 2019).

Tofisopam, a broad-spectrum PDE inhibitor used clinically as an anxiolytic and antidepressant agent, stimulates bone formation in preclinical studies, increasing serum osteocalcin and improving trabecular bone volume and thickness (Bonnet et al., 2007). These anabolic effects are attributed to the inhibitory effects of the compound on PDE2 and PDE4, which enhances BMP signaling, thereby promoting osteoblast activity and bone matrix deposition. Notably, tofisopam was found to have a more pronounced effect on bone microarchitecture than selective PDE4 inhibitors did, highlighting its potential advantage as a modulator of skeletal anabolism. In addition to effects on growing bone, the ability of tofisopam to potentiate BMP signaling also implicates a role in bone regeneration, as the BMP pathway is essential for bone repair processes, including callus formation and remodeling (Bonnet et al., 2007).

Pentoxifylline has protective and regenerative effects across skeletal contexts. It preserves growth plate integrity after phototherapy-induced injury (Atabek et al., 2007), supports fracture healing and callus formation (Atalay et al., 2015; MalekiGorji et al., 2020), and improves bone strength while enhancing VEGF-mediated angiogenesis (Kinoshita et al., 2000; Çakmak et al., 2015) and osteoblast differentiation (Horiuchi et al., 2001; Tsutsumimoto et al., 2002; Horiuchi et al., 2004). Mechanistically, it elevates cAMP, which augments BMP signaling, leading to osteoblast recruitment and bone matrix formation, while simultaneously suppressing osteocalcin expression in a cAMP-independent manner (Tsutsumimoto et al., 2002). In osteoporosis models, pentoxifylline restores bone mass and strength without increasing resorption, aided by improved vascularity (Pal et al., 2019a; Pal et al., 2019b; Roser-Page et al., 2022). In OA, it improves joint perfusion, attenuates inflammation, and supports tissue repair (MalekiGorji et al., 2020; Parlak et al., 2025), whereas in RA patients, it reduces synovial fibroblast proliferation, joint inflammation and structural damage (Maksymowych et al., 1995; Dubost et al., 1997; Ishii et al., 1997).

Dipyridamole promotes bone regeneration and osteogenesis of growing pediatric bone by increasing the level of extracellular adenosine, activating adenosine receptor A2A to stimulate osteoblast proliferation and differentiation (Mediero et al., 2015; Bekisz et al., 2018; Witek et al., 2019). It also protects against osteoporotic bone loss by inhibiting osteoclast differentiation, improving the balance between bone formation and resorption (Conesa-Buendía et al., 2019).

There is no scientific evidence of the effects of flavoxate hydrochloride, dyphylline, acefylline, doxofylline, or choline theophyllinate on bone or cartilage.

5. Function of PDEs in bone and cartilage

Based on the PDE expression and general function, animal gene knock-out studies, and indirect evidence obtained with the inhibitors, a generalized picture about the role of individual PDE genes in bone metabolism could be drawn (Fig. 2).

Fig. 2.

Overview of major PDEs action in growing and inflamed bone and cartilage. In growth plate cartilage, PDE3A, PDE3B, PDE4D and likely PDE8A negatively regulate signaling pathways that are stimulating (black) or inhibiting (red) transition of chondrocytes during longitudinal bone growth. PDE5A negatively regulates osteoblast activity, presumably through interference with Wnt and YAP/TAZ mechanosensing pathways. PDE8A and PDE10A also affect growth and health of skeletal system by further unclarified manner. PDE4D and PDE4A are involved in join inflammation in aged RA patients.

No strong evidence suggests PDE1, 2 or 7 have some role in bone or cartilage metabolism.

PDE3 genes, PDE3B in particular, showed to be important regulators of growth plate chondrocyte maturation. PDE3A is likely involved in reduction of cAMP rise mediated by PTH1R and doesn't have major impact of skeletal growth (Elli et al., 2018). On contrary, PDE3B is degrading cGMP induced by NO or NPRB signaling, attenuating mechanical and CNP induced bone growth (Wang et al., 2011; Miyazaki et al., 2022; Kawabe et al., 2025).

Two PDE4 genes, 4A and 4B, are involved in synoviocytes inflammation which is linked with arthritis (Crilly et al., 2011). In addition, PDE4B regulates inflammation in chondrocytes (Xu et al., 2025a; Xu et al., 2025b). PDE4D lowers cAMP induced by PTH1R signaling in chondrocytes of growth plate cartilage (Lee et al., 2012; Linglart et al., 2012).

PDE5A controls PKG signaling by downregulating NO-induced cGMP in osteoblasts and osteocytes, leading to lower induction of pro-osteogenic BMP, Wnt/β-catenin target genes (Ramdani et al., 2018; Kim et al., 2021; Huyut et al., 2022). It also stimulates osteoclasts, presumably in NO dependent manner (Kim et al., 2020; Kim et al., 2021; Menger et al., 2024a).

Of the PDE8 gene family, PDE8A is present in both chondrocytes and osteocytes and is deficiency has impact on skeletal growth without further clarification (mousephenotype.org). PDE8's activity is negatively regulated by FAK in response to mechanical stress (Papaioannou et al., 2024).

Inhibition of PDE10 improves mechanical properties of tendons, suggesting role of PDE10 A in that part of skeletal system (Can et al., 2024). In addition, mechanical stretching upregulates PDE10A, which supports the expression of mechanoresponsive genes (Müller-Deubert et al., 2020).

6. Summary and therapeutic outlook

Complete mechanism by which PDE inhibitors affect bone and cartilage is still unclear, as indicated by the multiple modes of action of cilostazol (Yasoda et al., 2004; Krejci et al., 2005; Wang et al., 2014; Kawabe et al., 2025), contradictory effects on WNT signaling in osteoblasts induced by different tadalafil doses (Gong et al., 2014; Kim et al., 2020), or detrimental effects on bone caused by the PDE3, PDE4 and PDE7 inhibitor theophylline (Pal et al., 2016).

However, a strong body of evidence from animal models suggests that PDE inhibitors are promising candidates for repurposing in the treatment of fracture healing and age-related and developmental skeletal disorders. Among 32 inhibitors that are currently approved for clinical use, 20 have some positive effects on the skeletal system. In particular, inhibitors that target PDE3, PDE4 and PDE5 offer exciting prospects.

PDE4 inhibitors (apremilast and roflumilast) are most applicable for cartilage protection in inflammatory joint diseases such as OA and RA. Their effect is probably both direct and indirect. In immune cells, they suppress the production of proinflammatory cytokines (TNF-α, IL-1β, and IL-17) and thus indirectly protect cartilage from degradation (Tenor et al., 2002; Chen et al., 2018a; Wang et al., 2021; Su et al., 2022; Xu et al., 2025b). In chondrocytes, they induce SOX9 production through the cAMP/PKA/CREB axis, which directly nourishes the chondrogenic program (Zhang et al., 2020). On the basis of the expression profile and mutant phenotype, PDE4D and 4B isoforms are likely the main targets of these inhibitors in cartilage (Lee et al., 2012; Linglart et al., 2012; Su et al., 2022).

PDE5 inhibitors (e.g., sildenafil, tadalafil, vardenafil, and avanafil) show strong efficacy in treating osteopenia, osteoporosis, and fracture healing. Acting via the NO/cGMP/PKGII axis, they stimulate MAPK and WNT signaling in osteoblasts, promote their differentiation, and enhance VEGF-mediated angiogenesis, accelerating bone regeneration (Rangaswami et al., 2009; Histing et al., 2011; Alp et al., 2017; Huyut et al., 2022; Menger et al., 2023a). Therefore, the combination of PDE5 inhibitors with exogenous sources of NO in bone could further enhance bone anabolic processes (Wimalawansa, 2010).

PDE3 inhibitors (e.g., cilostazol and milrinone) may be used to treat growth disorders, osteoporosis or fracture repair. By increasing cGMP and activating PKG, they suppress FGFR3/MAPK signaling and trigger the BK/TRPM7/CaMKII cascade, supporting growth plate expansion and bone growth (Yasoda et al., 2004; Krejci et al., 2005; Kawabe et al., 2025). In osteoblasts, PDE3 blockade enhances PI3K–RUNX2 signaling, thereby promoting bone formation (Menger et al., 2024b), whereas in preosteoclasts, PKA-mediated phosphorylation of the transcription factor NFAT2 reduces osteoclast formation (Ke et al., 2015). This inhibition likely affects both PDE3 isoforms, as mutations in any of them are linked to the skeletal phenotype (Elli et al., 2018; Reyes et al., 2020; Kawabe et al., 2025). As PDE inhibitors are applied orally, PDE3-targeting drugs might represent an interesting alternative or enhancement to current achondroplasia therapy, which is based on daily injections of a CNP analog.

Broad-spectrum inhibitors such as pentoxifylline and dipyridamole exhibit dual anabolic and anti-inflammatory effects, making them suitable for treating osteoporosis, OA, and RA. Their application can be combined with bisphosphate, BMP ligands or the synthetic parathyroid hormones teriparatide or abaloparatide to further enhance their impact on bone health (Horiuchi et al., 2001; Horiuchi et al., 2004).

Although how PDE inhibitors work is still not completely clear, we already know enough to consider them as viable therapeutic options for the treatment of RA and osteoporosis or to improve fracture healing. However, their integration into clinical practice will require time and further research.

CRediT authorship contribution statement

Vlad-Constantin Ursachi: Writing – review & editing, Writing – original draft, Validation, Investigation. Jan Horak: Writing – review & editing, Writing – original draft, Investigation. Bohumil Fafilek: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

No data was used for the research described in the article.