There is currently great interest in attempting to deploy inhibitors specific for phosphodiesterase-4 (PDE4) to treat conditions associated with inflammation, such as asthma and chronic obstructive pulmonary disease (COPD) (Maurice et al., 2003; Spina, 2004; Jeffery, 2005; Lipworth, 2005). The development of such PDE4-selective, active site-directed inhibitors has recently been furthered by the elucidation of the three-dimensional structure of the PDE4 catalytic unit and that of other PDEs, which form the 11-member PDE super-family (Xu et al., 2000; Card et al., 2005). However, not only has an increasing range of PDE4 inhibitor scaffolds become available but also the range of potential therapeutic uses for such compounds has broadened.

The article by Tilley and Maurice (2005) in this issue of Molecular Pharmacology highlights an exciting new aspect of “PDE4-ology”, which relates to a potential role for PDE4 inhibitors in adjunctive pharmacotherapy after percutaneous coronary interventions. In this case, the target cells in question are vascular smooth muscle cells (VSMC), which can undergo a phenotypic change from a “normal”, “contractile/quiescent” state to an “activated”, “synthetic” state. Various types of vascular injury can effect this transition, including, in the case of percutaneous coronary intervention, both inflation of the intracoronary balloon and the metal of the stent itself. The activation of proliferative capacities of VSMC causes vessel wall remodeling associated with reduced contractile capacity and thickening of vessels. Inappropriate migration of VSMC into the intimal layer and the generation of extracellular matrix disrupts normal vessel functioning are characteristic of atherosclerosis and can promote postangioplasty “in-stent” restenosis. Provision of a means of addressing these clinically important problems would thus have considerable therapeutic potential. Pertinent to this, a number of investigators have shown that increasing cAMP levels in activated VSMC can profoundly inhibit their proliferation and migration as well as the synthesis of extracellular matrix proteins (Vadiveloo et al., 1997; Kronemann et al., 1999; Hayashi et al., 2000; Koyama et al., 2001; Pelletier et al., 2005). On this basis, one might expect activated smooth muscle cells to deploy a mechanism that would protect them from potential inhibitory effects of cAMP. In their study, Tilley and Maurice (2005) identify just such a mechanism, namely the induction of specific PDE4 isoforms, which serve to desensitize the susceptibility of activated VSMC to the inhibitory action of cAMP.

PDE4 enzymes specifically hydrolyze cAMP, compared with cGMP, and are regulated through phosphorylation by both protein kinase A (PKA) (Alvarez et al., 1995; MacKenzie et al., 2002) and ERK (Hoffmann et al., 1999; Baillie et al., 2000, 2001; MacKenzie et al., 2000). PDE4s also provide a key component that underpins compartmentalized cAMP signaling in cells (Mongillo et al., 2004) because of their ability to target to specific intracellular sites and to interact with various signaling scaffold proteins (Conti et al., 2003; Houslay and Adams, 2003). Four genes (A/B/C/D) encode some 20 different PDE4 isoforms, each of which is characterized by a unique N-terminal region whose key role is to confer intracellular targeting to specific sites and signaling complexes (Conti et al., 2003; Houslay and Adams, 2003). The isoforms encoded by each PDE4 gene are separated into long, short, and supershort groupings based upon the presence/absence of upstream conserved (UCR) domains located between the isoform-specific N-terminal region and the catalytic unit (Fig. 1). Long isoforms have such domains, called UCR1 and UCR2, whereas short isoforms have only UCR2 and supershort isoforms merely a truncated UCR2. These UCR domains exert a regulatory role; UCR1 provides a site for stimulatory phosphorylation by PKA (Sette and Conti, 1996; MacKenzie et al., 2002). In addition, upon ERK phosphorylation of the PDE4 catalytic unit, the paired UCR1/UCR2 module orchestrates an inhibitory action in long forms, whereas the lone UCR2 orchestrates a stimulatory action in short forms, with little discernible effect seen in supershort forms (Baillie et al., 2000; MacKenzie et al., 2000).

The distinct modes of intracellular targeting and recruitment to scaffold complexes, coupled with differences in regulation by PKA and ERK, mean that the pattern of PDE4 isoform expression in a cell can be expected to have profound effects on both compartmentalized cAMP signaling and cross talk with the ERK pathway (Houslay and Kolch, 2000; Houslay and Adams, 2003; Baillie and Houslay, 2005). In this regard, Tilley and Maurice (2005) have made the unusual observation that the short PDE4D1 and supershort PDE4D2 isoforms are specifically induced in activated VSMC. Such a remodeling of the PDE4 profile has interesting consequences for the cross-talk with the growth-promoting ERK signaling pathway (Fig. 1), which is activated in proliferating VSMC, because the short PDE4D1 isoform will be activated by proliferative signals involving ERK (Baillie et al., 2000). This mechanism will serve to further depress potentially inhibitory levels of cAMP and, in so doing, further promote the proliferative response. It is interesting that a cognate remodeling has been described upon monocyte-to-macrophage differentiation (Shepherd et al., 2004). In monocytes, ERK-inhibited long PDE4D isoforms predominate, whereas in macrophages, ERK-activated short PDE4B2 isoform predominates. Thus, in macrophages, activation of ERK by proinflammatory agents serves to depress inhibitory cAMP levels, whereas in monocytes, a transient, inhibitory increase in cAMP occurs because of an initial ERK-mediated inhibition of PDE4 long isoforms, which generates a rise in cAMP that causes PKA to subsequently phosphorylate and activate the long isoform (Fig. 1) (Baillie et al., 2000, 2001; MacKenzie et al., 2000).

PDE4D1/2 isoforms have a very restricted pattern of expression, making Tilley and Maurice's observations particularly interesting; they identified increased histone-3-acetylation, which applies to histones associated with smooth muscle-specific promoters upon differentiation (Cao et al., 2005), as playing a key role in regulating the functioning of the PDE4D1/2 intronic promoters in VSMC. It is intriguing that a decrease in histone deacetylase activity has been observed in COPD (Ito et al., 2005), a pathological condition in which airway remodeling occurs and PDE4 inhibitors have therapeutic potential (Jeffery, 2005; Lipworth, 2005). Nothing is known, however, about changes in PDE4 expression profile in pulmonary VSMC of COPD patients, so it would seem pertinent to address this, especially because a change in the macrophage PDE4 profile of patients with COPD has been noted (Barber et al., 2004).

A hallmark of atherosclerosis is the proliferation and migration of VSMC. In this regard, the provocative proposal has been made (Gretarsdottir et al., 2003) that the PDE4D gene may be linked to large vessel (e.g., carotid) occlusive stroke but not to strokes that occur via occlusion of small vessels. PDE4 genes are extremely large and encompass approximately 50 kilobases of sequence, making them difficult to analyze: seven exons encode the catalytic unit, eight exons encode UCR1/2 together with their linker regions, and individual 5′ exons encode the extreme N-terminal regions unique to each isoform (Houslay and Adams, 2003). The first linkage study, which was done on a cohort of patients from Iceland, identified association of stroke with a noncoding region of the PDE4D gene (Gretarsdottir et al., 2003). This is well upstream from the intronic PDE4D1/2 promoter (Vicini and Conti, 1997) and has been suggested (Gretarsdottir et al., 2003; Gulcher et al., 2005) to be associated with a putative promoter for the long PDE4D7 isoform. The mutations associated with stroke predisposition were proposed to lead to enhanced PDE4D7 expression (Gretarsdottir et al., 2003). If PDE4D7 were to be up-regulated in VSMC the prediction would be that enhanced VSMC proliferation and migration into the intima with concomitant lesion formation would ensue. Indeed, RhoA activity is intimately linked with cell migration (Nobes and Hall, 1999), and increased RhoA activity is invariably seen in various models of hypertension, including VSMC from spontaneously stroke prone rats (Moriki et al., 2004). Because PDE4 inhibition and increased cAMP levels have been shown to inhibit Rho activity and cell migration (Fleming et al., 2004; Netherton and Maurice, 2005), this may indicate a further link. If PDE4D7 was up-regulated in macrophages, then the resultant reduction in cAMP levels might be predicted to enhance the susceptibility of macrophages to activation in atherosclerotic plaques, influencing atherogenesis and plaque stability.

Intronic promoters for a number of specific PDE4 isoforms (Vicini and Conti, 1997; Rena et al., 2001; Le Jeune et al., 2002; Wallace et al., 2005), including PDE4D1 (Vicini and Conti, 1997), have been identified and defined by first cloning, in frame, 1500 base pairs or so of sequence located immediately 5′ to the ATG start codon. A similar strategy applied to PDE4D7, to encompass the presumed association sites, would allow rigorous evaluation of mutations suggested to either predispose or protect from large vessel and cardiogenic stroke. However, because PDE4D7 is the most 5′ isoform encoded by the PDE4D gene, this will not be done quite so readily. This is because three exons, rather than one, encode the unique N-terminal region of PDE4D7. The additional complexity in splicing associated with PDE4D7 may therefore underpin a purported susceptibility to altered expression as a consequence of mutations identified in the Icelandic study. A functional promoter analysis of PDE4D7 becomes all the more pressing in view of recent studies on European populations (Bevan et al., 2005; Lohmussaar et al., 2005) that have failed to identify such a clear-cut connection between the PDE4D7 markers and stroke, as identified in the Icelandic study (Gretarsdottir et al., 2003).

One of these new studies (Bevan et al., 2005), however, did record the possibility of an association of PDE4D with cardioembolic stroke and, although it failed to show that PDE4D was a major risk factor for ischemic stroke, the authors could not exclude a connection. The strong correlation seen in the Icelandic study (Gretarsdottir et al., 2003) may result from a founder effect within the restricted gene pool of this small population. Notwithstanding this, such studies and those of Tilley and Maurice (2005) suggest an urgent need for promoter analyses to be performed on PDE4D7 and PDE4D1/2 isoforms in VSMC and macrophages so as to appreciate regulation both in resting cells and also those activated by various challenges. These are an essential prerequisite to any further evaluation involving transgenic analyses. It should also be noted that Maurice and coworkers have demonstrated that PDE4D3 expression can be regulated by ERK action through changes in message stability (Liu et al., 2000). These data suggest the possibility that mutations identified in the Icelandic study could either lead to direct alterations in PDE4D7 transcript stability or exert indirect effects by changing regulatory processes associated with transcript stability.

It will be particularly interesting to delineate the functional consequences of up-regulating PDE4D1, PDE4D2, and PDE4D7 in activated VSMC. Given that PDE4 enzymes provide a major component that underpins compartmentalized cAMP signaling in cells, such an evaluation should probe for a role associated with the key differences between activated and resting VSMC. These features include proliferation, for which there are a wealth of data indicating various potential points for inhibitory intervention by cAMP (Balmanno et al., 2003); migration, where PDE4 inhibition and elevation of cAMP levels in a number of cell types seems to exert its action by an inhibitory effect on Rho-regulated processes (Fleming et al., 2004; Netherton and Maurice, 2005); cell shape, where cAMP-mediated inhibition of Rac plays a role (Pelletier et al., 2005) and, finally, regarding extracellular matrix generation, where PDE4 inhibitors exert an inhibitory effect (Maurice et al., 2003). However, little is known about the functional role of these isoforms regarding each of the processes.

To undertake these studies, it will be necessary to inactivate them selectively, which has not been done so far. However, the identical catalytic unit of isoforms from any one PDE4 subfamily, coupled with the similarity of catalytic units between subfamilies, has meant that no highly specific, isoform-discriminatory inhibitors are available.

Until recently, the only way of addressing the functional significance of specific PDE4 isoforms has been to overexpress them in cells. However, our understanding of the role of specific PDE4 subfamilies and isoforms dramatically changed with the advent of gene knockout approaches (Hansen et al., 2000; Jin and Conti, 2002; Ariga et al., 2004) and, crucially, the development of both dominant-negative (Baillie et al., 2003; McCahill et al., 2005) and siRNA-mediated knockdown strategies (Lynch et al., 2005) to discern the functional role of individual PDE4 isoforms. Thus siRNA-mediated knockdown can be used to target either entire PDE4 subfamilies or specific isoforms, dependent upon probe design (Lynch et al., 2005), whereas dominant-negative approaches exploit the targeting of specific isoforms by overexpressing a catalytic inactive isoform to displace the tethered endogenous active species from the functionally relevant site in the cell (Baillie et al., 2003; Baillie and Houslay, 2005).

Is it feasible to try to employ such strategies to isolate the role of PDE4 isoforms in activated VSMC and thereby to determine their potential as specific targets for selective therapeutic intervention? PDE4D7 is a long isoform that contains both UCR1 and UCR2 together with a unique 91-amino acid N-terminal region (Fig. 1). Thus, as used to assess PDE4D5 and PDE4D3 functioning (Baillie et al., 2003; Baillie and Houslay, 2005), the unique 5′ region of PDE4D7 makes it eminently suitable for siRNA intervention, and its N-terminal and UCR targeting regions indicate potential for a dominant-negative approach. PDE4D1 is a classic short form; it lacks UCR1 but has both the 80-amino acid UCR2 together with a unique 45-amino acid N-terminal region, again making it suitable for both approaches (Fig. 1). PDE4D2 is a supershort isoform that has just a truncated UCR2 that lacks the N-terminal first 34 amino acids of full-length UCR2 (Fig. 1). However, it is curious in that, unlike “classic” supershort forms such as PDE4A1 (Sullivan et al., 1998), PDE4D2 lacks any unique N-terminal region. This means that there is no possibility of using a siRNA-mediated approach to target PDE4D2 specifically, and because a dominant-negative approach may well interfere with other isoforms, assessment of the role of PDE4D2 will be extremely challenging.

The ability of changes in cAMP to alter VSMC and monocyte/macrophage functioning, the alterations in PDE4 occurring in such cells upon activation/differentiation and proven action of PDE4-selective inhibitors all argue for further analysis of the role of PDE4 isoforms in these cell systems. This will indicate whether PDE4 can be exploited either diagnostically or therapeutically in certain forms of stroke. Here Tilley and Maurice (2005) propose a novel therapeutic use of PDE4 inhibitors that might be further refined by targeting action to the PDE4D1/2 short isoforms. Exploiting dominant-negative and siRNA technologies may provide the required proof of principle to test this notion before progressing along the rocky road of attempting to develop a PDE4D1/2-selective inhibitor. It is worth noting, however, that the Ca²⁺-activated PDE1C isoform has been shown to be up-regulated in proliferating VSMC (Rybalkin et al., 2002) and that a PDE1-selective inhibitor can attenuate VSMC proliferation (Phillips et al., 2005). Because PDE1 and PDE4 invariably control different pools of cAMP in cells, a comparative evaluation of PDE1C with PDE4D1, PDE4D2 and PDE4D7 might prove highly informative in terms of identifying regulatory mechanisms as well as defining appropriate therapeutic targets and strategies and, possibly, diagnostic indicators.